Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaluating the quality of clinical trials using a Clinical Trial Evaluation System (CTES) Franciosi, Luigi Giuseppe 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1998-0155.pdf [ 14.19MB ]
Metadata
JSON: 831-1.0099241.json
JSON-LD: 831-1.0099241-ld.json
RDF/XML (Pretty): 831-1.0099241-rdf.xml
RDF/JSON: 831-1.0099241-rdf.json
Turtle: 831-1.0099241-turtle.txt
N-Triples: 831-1.0099241-rdf-ntriples.txt
Original Record: 831-1.0099241-source.json
Full Text
831-1.0099241-fulltext.txt
Citation
831-1.0099241.ris

Full Text

Evaluating the Quality of Clinical Trials Using A Clinical Trial Evaluation System (CTES) by Luigi Giuseppe Franciosi B.Sc, University of British Columbia, 1993  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES Department of Pharmacology & Therapeutics We accept this thesis as conforming  THE UNIVERSITY OF BRITISH COLUMBIA April 1998 © Luigi G. Franciosi, 1998  In  presenting  degree freely  at  this  the  thesis  in  partial  fulfilment  of  University  of  British  Columbia,  I agree  available for  copying  of  department publication  this or of  reference  thesis by  this  for  his thesis  and study. scholarly  or for  her  Department The University of British Vancouver, Canada  DE-6 (2/88)  Columbia  purposes  gain  requirements that  agree  may  representatives.  financial  permission.  I further  the  It  shall not  be is  that  the  permission  granted  allowed  an  advanced  Library shall  by  understood be  for  for  the that  without  make  it  extensive  head  of  my  copying  or  my  written  II  Abstract A clinical trial is commonly defined as a human experiment that has the potential to improve the quality of health care.  It is also an argument. In the realm of medical  investigations, a clinical trial is classified as an experimental analytic study because it can be designed and implemented in a manner that allows investigators to argue on the basis of objective, unbiased evidence. Currently, there is a need for such evidence by medical decision-makers. The scientific basis of this 'evidence-based medicine' relies upon clinical trials, in particular, randomised controlled trials (RCTs). These trials are regarded as the most reliable trials for evaluating treatments.  However, the best  available evidence is usually something less than a RCT. Between the years of 1993 and 1997, I assisted in the design and implementation of six clinical trials. I wanted to evaluate the quality of my trials, but I realised that many scales and checklists available were only concerned with assessing a trial's design and analysis. Little consideration was made about a trial's implementation. Thus, based on my intimate knowledge of clinical trials, I developed a Clinical Trial Evaluation System (CTES) that allowed me to assess the amount my trials, or any clinical trial, deviated from a randomised controlled trial done according to "Good Clinical Practices". The evaluation involved determining whether or not items relevant to the trial's question, design, statistics, ethics, and standard operating procedures were considered.  For each of my trials, I derived a  clinical trial score that I then turned into a % deviation from the best possible trial. All of the trials deviated for many reasons. The strengths of CTES are the assessment of a trial's quality following its completion, the comparison of trial scores and % deviations, the determination of deviation patterns within and amongst trials, and the explanations  Ill  for w h y deviations occurred from the best possible trial. A n o t h e r strength is the better planning of s u b s e q u e n t clinical trials.  Future d e v e l o p m e n t s of the system will involve  the refinement of C T E S , the determination of interrater reliability, the estimation of time for completion, and the creation of relational database for evaluating and planning clinical trial protocols.  I believe that students, clinicians, and evidence-based medicine  organisations can optimise the quality of their trials by using the Clinical Trial Evaluation System.  iv  Table of Contents Abstract  ii  Table of Contents  iv  List of Tables  xvi  List of Figures.  xvii  Abbreviations  xviii  Acknowledgements  xix  PART I.  THE QUALITY OF CLINICAL TRIALS  1. What is a clinical trial?  1  2. Where do clinical trials fall in the classification of medical investigations?  2  3. What is the purpose of clinical trials?  6  4. Why are randomised controlled trials preferred?  8  5. Why is evidence from clinical trials poor?  9  6. What methods are available for assessing the quality of clinical trials and what is their major limitation?  11  PART II.  THE DEVELOPMENT OF A CLINICAL TRIAL EVALUATION SYSTEM (CTES) FOR EVALUATING THE QUALITY OF CLINICAL TRIALS  7. What is a definition of quality?  13  8. What are Good Clinical Practices (GCPs)?  13  9. What are the principal practices for designing and implementing the best possible clinical trial?  15  9.1 Define and state the primary question in advance 9.1.1 What is involved defining a question? 9.1.1.1 Identifying the specific problem that is need of a solution 9.1.1.1.1 Past research as a source of problems 9.1.1.1.2 Personal experience as a source of problems 9.1.1.1.3 Theories as a source of problems 9.1.1.2 Reviewing the relevant literature 9.1.1.3 Making a clear and concise statement of the problem in the form of a question 9.1.1.4 Considering the importance of the question 9.1.1.5 Refining the question to make it answerable 9.1.1.5.1 Selecting a primary response variable 9.1.1.5.1.1 Understanding the response variable in relation to other types of variables 9.1.1.5.1.2 Knowing how a response variable can be measured 9.1.1.5.1.2.1 nominal measurement scale 9.1.1.5.1.2.2 ordinal measurement scale 9.1.1.5.1.2.3 interval measurement scale 9.1.1.5.1.2.4 ratio measurement scale 9.1.1.5.1.3 Considering the problems with selecting and measuring variables 9.1.1.5.1.3.1 appropriateness of the variable 9.1.1.5.1.3.2 experimenter and subject bias 9.1.1.5.2 Specifying the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and the possible direction and magnitude of the relationship or difference that is to be detect 9.1.2 Why state the question in advance? 9.1.2.1 Considering the question's feasibility 9.1.2.1.1 time 9.1.2.1.2 subject population 9.1.2.1.3 expertise 9.1.2.1.4 expense 9.1.2.1.5 facilities and equipment 9.1.2.1.6 ethical implications 9.1.2.2 Formulating and stating a hypothesis 9.2 Choose a design that is appropriate for the primary question 9.2.1 Sampling and screening subjects from a population  15 16 16 16 16 17 17 17 18 18 19 19 20 20 20 21 21 21 21 21  22 23 23 24 24 24 25 25 25 26 28 29  vi 9.2.1.1 Identifying and describing the population that the primary question addresses 9.2.1.2 Considering the method of sampling subjects from the population 9.2.1.2.1 Idealised sampling procedures: the relevance of a random sample 9.2.1.2.2 Nonprobability sampling procedures: a sample of convenience 9.2.1.3 Determining how the subjects are to be recruited 9.2.1.4 Screening subjects 9.2.1.4.1 Determining the eligibility criteria 9.2.1.4.2 Considering the types of subjects pre- and postscreening 9.2.1.4.2.1 Refusers 9.2.1.4.2.2 Nonqualifiers 9.2.1.4.2.3 Dropouts 9.2.1.4.2.4 Discontinuers 9.2.1.4.2.4.1 Monitoring subject compliance 9.2.1.5 Considering issues related to generalisability 9.2.1.5.1 Threats to the selection process 9.2.1.5.2 Threats to the trial setting and conditions 9.2.1.6 Determining the appropriate sample size 9.2.3 Choosing an experimental design 9.2.3.1 Parallel group design 9.2.3.1.1 Defining the control and test treatments 9.2.3.1.1.1 Understanding the word 'control' 9.2.3.1.1.2 Defining the control treatment.. 9.2.3.1.1.2.1 A standard treatment 9.2.3.1.1.2.2 A placebo.. 9.2.3.1.1.2.2.1 The 'placebo effect' 9.2.3.1.1.2.3 No treatment 9.2.3.1.1.3 Defining the test treatment(s) 9.2.3.1.1.4 Example of control and test treatment groups 9.2.4 Randomising eligible subjects to treatments 9.2.4.1 Methods of generating random assignment 9.2.4.2 Assessing the baseline characteristics of all subjects 9.2.5 Blinding all parties in the trial as much as possible 9.2.5.1 Types of Blinding 9.2.5.1.1 Double blinding 9.2.5.1.2 Single blinding 9.2.5.1.3 Blinding not possible?; the blinding of an independent observer 9.2.5.1.4 Assessing the adequacy of blinding  29 29 29 30 31 31 31 32 32 32 32 32 33 33 34 35 35 36 38 38 38 39 39 39 40 40 40 41 41 41 42 42 43 43 43 44 44  vii 9.2.6 Considering the potential confounding factors in the clinical trial 9.2.6.1 9.2.6.2 9.2.6.3 9.2.6.4 9.2.6.5 9.2.6.6 9.2.6.7 9.2.6.8  History Maturation Testing Instrumentation Statistical regression Selection of subjects Mortality Interaction between the selection process and other confounders 9.2.6.9 Maintenance of treatment conditions over time  44 45 45 46 46 47 47 48 48 49  9.3 Use appropriate statistics  50  9.3.1 What are statistics?  50  9.3.2 Considering the use of descriptive statistics  52  9.3.2.1 Measures of central tendency 9.3.2.1.1 Mean 9.3.2.1.2 Mode 9.3.2.1.3 Median 9.3.2.2 Measures of variability 9.3.2.2.1 Variance 9.3.2.2.2 Standard deviation 9.3.2.2.3 Coefficient of variation 9.3.2.3 Correlation 9.3.2.4 Regression 9.3.2.4.1 Linear regression 9.3.2.4.2 Non-linear regression 9.3.3 Considering the use of inferential statistics 9.3.3.1 Formulating and stating the null and alternative hypothesis 9.3.3.2 Specifying an alpha and beta level 9.3.3.3 Considering statistical power 9.3.3.3.1 Conducting a pilot study and/or a review of the literature to determine the possible variance 9.3.3.4 Choosing a statistical test appropriate for the question 9.3.3.4.1 Parametric statistical tests 9.3.3.4.1.1 Types of tests and their assumptions 9.3.3.4.1.2 Assessing the normality of the data 9.3.3.4.1.3 Using transformations to normalise the data 9.3.3.4.2 Nonparametric tests 9.3.3.4.2.1 Types of tests and their assumptions  53 53 54 54 55 55 55 57 57 57 58 58 59 59 61 62 62 62 63 63 64 64 64 64  viii 9.3.3.4.3 Considering one tailed versus two tailed statistical tests 9.3.3.5 Considering the use of a confidence interval 9.3.3.6 Considering the meanings of 'statistical' and 'clinical' significance 9.3.3.7 Considering the issues related to chance 9.3.3.8 Considering statistical methods to deal with missing values 9.3.3.9 Considering statistical methods to deal with outliers 9.3.3.10 Considering the use of an interim analysis 9.4 Consider the ethical issues  65 67 68 68 69 70 70 71  9.4.1 Considering the potential risks and benefits for the subjects involved?  71  9.4.2 Considering the appropriateness of the experimental design and statistics?  72  9.4.3 Considering any "conflict of interest" regarding the treatment(s) being investigated  73  9.4.4 Considering the early termination of the clinical trial  73  9.4.5 Considering issues related to subject consent  74  9.4.6 Considering issues related to subject confidentiality  75  9.4.7 Reviewing and referencing the Declaration of Helsinki for the protection of trial subjects 9.4.8 Seeking the opinion of an ethics committee  75 76  9.5 Have standard operating procedures in place  77  9.5.1 What are Standard Operating Procedures (SOPs)?  77  9.5.2 Procedures for delegating responsibilities  78  9.5.3 Procedures for developing and storing principal clinical trial documents  78  9.5.4 Procedures for protecting trial subjects  81  9.5.5 Procedures for monitoring trial events  82  9.5.6 Procedures for defining and collecting adverse events  83  ix 9.5.7 Procedures for developing and handling trial treatments  84  9.5.8 Procedures for data management  85  9.5.9 Procedures for audit and inspection of a clinical trial  85  9.5.10 Procedures for the collection and management of clinical laboratory data  86  9.5.11 Miscellaneous procedures  86  10. What items do I consider relevant when evaluating the quality of clinical trials?  87  10.1 Items relevant to the primary question  88  10.2 Items relevant to the design  89  10.3 Items relevant to the statistics  90  10.4 Items relevant to the ethics  91  10.5 Items relevant to the standard operating procedures 10.6 The clinical trial score (CTS) and the % deviation from the best possible trial  92  PART III.  93  THE USE OF THE CLINICAL TRIAL EVALUATION SYSTEM TO ASSESS THE QUALITY OF CLINICAL TRIALS  11. The Clinical Evaluation of the Controlled Environment Treatment (CET) System - A Medical Device for Burn Patients  95  11.1 Introduction  95  11.2 Methods 11.2.1 Stage One  97 97  11.2.2 Stage Two  98  11.3 Results of the CTES assessment 11.3.1 Items relevant to the primary question for the CET trial 11.3.1.1 A discussion about the CET trial's questions  101 101 102  X  11.3.2 Items relevant to the design for the CET trial 11.3.2.1 A discussion about the CET trial's design  105 106  11.3.3 Items relevant to the statistics for the CET trial 11.3.3.1 A discussion about the CET trial's statistics  108 109  11.3.4 Items relevant to the ethics for the CET trial 11.3.4.1 A discussion about the CET trial's ethics  110 110  11.3.5 Items relevant to the standard operating procedures for the CET trial 11.3.5.1 A discussion about the CET trial's standard operating procedures 11.3.6 The clinical trial score (CTS) and the % deviation for the CET trial 11.4 Discussion about the CTS and the % deviation from the best possible clinical trial  112 113 114 114  12. The Analgesic Efficacy of Intraarticular Morphine (MOR) Following Shoulder Surgery  115  12.1 Introduction  115  12.2 Methods  116  12.3 Results of the CTES assessment  118  12.3.1 Items relevant to the primary question for the MOR trial 12.3.1.1 A discussion about the MOR trial's question 12.3.2 Items relevant to the design for the MOR trial 12.3.2.1 A discussion about the MOR trial's design  118 119 120 121  12.3.3 Items relevant to the statistics for the MOR trial 12.3.3.1 A discussion about the MOR trial's statistics  122 123  12.3.4 Items relevant to the ethics for the MOR trial 12.3.4.1 A discussion about the MOR trial's ethics  127 127  12.3.5 Items relevant to the standard operating procedures for the MOR trial 12.3.5.1 A discussion about the MOR trial's standard operating procedures  128 129  xi 12.3.6 The clinical trial score (CTS) and the % deviation for the MOR trial 12.4 A discussion about the CTS and the % deviation from the best possible trial 13. The Clinical Safety Assessment of the Novel Drug, TSB330  130  130 131  13.1 Introduction  131  13.2 Methods  131  13.3 Results of the CTES assessment  136  13.3.1 Items relevant to the primary question for the 330 trial 13.3.1.1 A discussion about the 330 trial's question  136 137  13.3.2 Items relevant to the design for the 330 trial 13.3.2.1 A discussion about the 330 trial's design  138 139  13.3.3 Items relevant to the statistics for the 330 trial 13.3.3.1 A discussion about the 330 trial's statistics  140 141  13.3.4 Items relevant to the ethics for the 330 trial 13.3.4.1 A discussion about the 330 trial's ethics  142 142  13.3.5 Items relevant to the standard operating procedures for the 330 trial 13.3.5.1 A discussion about the 330 trial's standard operating procedures  144  13.3.6 The clinical trial score (CTS) and the % deviation for the 330 trial  145  13.4 A discussion about the CTS and the % deviation from the best possible trial  143  145  14. The Cost Evaluation of the Novel General Anaesthetic, Sevoflurane, (SEV) in Patients Undergoing Arthroscopic Menisectomy  146  14.1 Introduction  146  14.2 Methods  147  xii 14.3 Results of the CTES assessment  150  14.3.1 Items relevant to the primary question for the SEV trial 14.3.1.1 A discussion about the SEV trial's question  150 151  14.3.2 Items relevant to the design for the SEV trial 14.3.2.1 A discussion about the SEV trial's design  152 153  14.3.3 Items relevant to the statistics for the SEV trial 14.3.3.1 A discussion about the SEV trial's statistics  154 155  14.3.4 Items relevant to the ethics for the S E V trial 14.3.4.1 A discussion about the SEV trial's ethics  156 156  14.3.5 Items relevant to the standard operating procedures for the SEV trial 14.3.5.1 A discussion about the SEV trial's standard operating procedures  158  14.3.6 The clinical trial score (CTS) and the % deviation for the SEV trial  159  14.4 A discussion about the CTS and the % deviation from the best possible trial  157  159  15. The Clinical Evaluation of Intradermal TSB430 as a Novel Peripheral Analgesic  160  15.1 Introduction  160  15.2 Methods  160  15.3 Results of the CTES assessment  164  15.3.1 Items relevant to the primary question for the 430 trial 15.3.1.1 A discussion about the 430 trial's question 15.3.2 Items relevant to the design for the 430 trial 15.3.2.1 A discussion about the 430 trial's design  164 165 167 168  15.3.3 Items relevant to the statistics for the 430 trial 15.3.3.1 A discussion about the 430 trial's statistics  169 170  15.3.4 Items relevant to the ethics for the 430 trial 15.3.4.1 A discussion about the 430 trial's ethics  171 171  xiii 15.3.5 Items relevant to the standard operating procedures for the 430 trial 15.3.5.1 A discussion about the 430 trial's standard operating procedures 15.3.6 The clinical trial score (CTS) and the % deviation for the 430 trial 15.4 A discussion about the CTS and the % deviation from the best possible trial  172 173 174 174  16. A Clinical Trial of a Local Anaesthetic (ANA) Technique in Patients Undergoing Lower Limb Surgery  175  16.1 Introduction  175  16.2 Methods  177  16.3 Results of the CTES assessment  183  16.3.1 Items relevant to the primary question for the ANA trial 16.3.1.1 A discussion about the ANA trial's question 16.3.2 Items relevant to the design for the ANA trial 16.3.2.1 A discussion about the ANA trial's design  181 182 183 184  16.3.3 Items relevant to the statistics for the ANA trial 16.3.3.1 A discussion about the ANA trial's statistics  185 186  16.3.4 Items relevant to the ethics for the ANA trial 16.3.4.1 A discussion about the ANA trial's ethics  187 187  16.3.5 Items relevant to the standard operating procedures for the ANA trial 16.3.5.1 A discussion about the ANA trial's standard operating procedures  189  16.3.6 The clinical trial score (CTS) and the % deviation for the ANA trial  190  16.4 A discussion about the CTS and the % deviation from the best possible clinical trial  188  190  xiv  PART IV.  STRENGTHS, FUTURE DEVELOPMENTS AND POTENTIAL USERS OF THE CLINICAL TRIAL EVALUATION SYSTEM (CTES)  17. Strengths of CTES  191  17.1 Assessment of a trial's quality following its completion  191  17.2 Comparison of trial scores and % deviations  191  17.3 Determination of deviation patterns within and amongst trials  192  17.4 Explanations for why deviations occurred from the best possible trial  203  17.5 Better planning of subsequent trials  203  18. Future Developments of CTES  203  18.1 Refinement of the Clinical Trial Evaluation System...  203  18.2 Determination of interrater-reliability  204  18.3 Estimation of time for completion  205  18.4 Creation of a relational database for evaluating and planning clinical trial protocols  205  19. Potential users of CTES  206  19.1 Students  206  19.2 Clinicians  206  19.3 Evidence-based organisations  206  BIBLIOGRAPHY  207  XV  APPENDICES  219  Appendix 1 Clinical Pharmacology Research Organisation (CPRO)  220  Appendix 2  Experimental Designs in Clinical Research  221  Appendix 3  Twelve Concepts of Control in Clinical Research  229  Appendix 4  Statistical Tests in Clinical Research  234  Appendix 5  Declaration of Helsinki  236  Appendix 6  Protocol for the Clinical Evaluation of CET in Burns  239  Appendix 7  Results of the Statistical Analysis for the Analgesic Efficacy of Intraarticular Morphine Following Shoulder Surgery  241  Appendix 8  Clinical Research Protocol: Phase I Evaluation of TSB330  248  Appendix 9  Clinical Research Protocol: Evaluation of the Cost of Sevoflurane Compared to Isoflurane for General Anaesthesia in Arthroscopic Menisectomy  262  Appendix 10  Appendix 11  Clinical Research Protocol: Evaluation of Intradermal TSB430 as a Peripheral Analgesic in Subjects with Capsaicin-lnduced Pain  270  Results of the Statistical Analysis for the Evaluation of Intradermal TSB430 as a Peripheral Analgesic in Subjects with Capsaicin-lnduced Pain  279  xvi List of Tables Table  Page  1  A clinical trial is both an experiment and an argument  2  2  Categories of clinical trials and their objectives  37  3  A frequency distribution of heart rates obtained from a sample of 20 subjects Roles and responsibilities of key participants in the implementation of a clinical trial  52  4 5 6  The dose fractions of the novel drug TSB330 that were to be administered  79 132  A parallel-group design that was not considered for the TSB330 trial due to cost  139  7  The demographics of 330 trial subjects  141  8  The mean total cost per patient for the sevoflurane and isoflurane groups  155  9  The Latin Square design used for the TSB430 trial  162  10  Mean pain scores with capsaicin at 1.5 minutes for the 430 trial  170  11  A summary of scores and % deviations for all my clinical trials  192  12  Patterns of deviation within and amongst trial questions  194  13  Patterns of deviation within and amongst trial designs  196  14  Patterns of deviation within and amongst trial statistics  198  15 16  Patterns of deviation within and amongst trial ethics Patterns of deviation within and amongst trial standard operating procedures A factorial design in which two treatments or factors are compared simultaneously to each other and with a placebo control  200  Control expressed as three distinct concepts  229  17 18  201 226  xvii List of Figures Figure  Page  1  Classification of medical investigations  3  2  The standard normal curve  53  3  The mean, mode and median of a sample distribution  54  4  The populations for null and alternative hypotheses  60  5  One-tailed test  66  6  Two tailed test  67  7  A patient undergoing CET treatment  96  8  The CET feedback system  96  9  The frequency distribution of control group vales for the MOR trial  125  10  The frequency distribution of control group values following a log transformation for the MOR trial  125  The frequency distribution of intraarticular morphine group values for the MOR trial  126  11 12  The frequency distribution of intraarticular morphine group values following log transformation for the MOR trial  126  13  The critical path analysis for the implementation of the TSB330 trial  135  14  The injection sites on the volar surface of a subject's nondominant forearm for the TSB430 trial The reduction of capsaicin pain with intradermal TSB430 in six  161  subjects  166  16  The basic two-period crossover design  223  17  A graphical illustration of Treatment A and B rendered significantly different using a sequential design  225  15  XVIII  Abbreviations AEs/AE  Adverse Event(s)  ANA  Anaesthetic; A clinical trial of a local anaesthetic technique in patients undergoing lower limb surgery  CET  Controlled Environment Treatment; The Clinical Evaluation of CET in Burns  CRFs/CRF  Case Report Form(s)  CTES  Clinical Trial Evaluation System  CTS  Clinical Trial Score  FDA  Food and Drug Administration, USA  GCPs/GCP  Good Clinical Practice(s)  GMPs/GMP  Good Manufacturing Practice(s)  i.a.  intraarticular; morphine is inject into a joint  MOR  Morphine; The analgesic efficacy of intraarticular morphine following shoulder surgery  PACU  Post Anaesthesia Care Unit  PCA  Patient Controlled Analgesia  SAEs/SAE  Serious Adverse Event(s)  SEV  Sevoflurane; The cost evaluation of novel general anaesthetic Sevoflurane in patients undergoing arthroscopic menisectomy  SOPs/SOP  Standard Operating Procedure(s)  TBSA  Total Body Surface Area  UBC  University of British Columbia  VAS  Visual Analogue Scale  330  TSB330; The Clinical Safety Assessment of the Novel Drug TSB330  430  TSB430; The Clinical Evaluation of Intradermal TSB430 as a Novel Peripheral Analgesic  xix Acknowledgements I am indebted to Drs. Bernard MacLeod and David Quastel for allowing me the opportunity to study under their guidance. I am very grateful to Drs. Michael Walker and Allen Bain for their valuable advice, moral and financial support during my first years as a graduate student. I would like to thank Drs. Don Zawrowny and Joel Singer of the Canadian HIV Trials Network, and Mrs. Ruth Milner and Dr. Keith Chambers of the VGH Clinical Epidemiology and Evaluation Research Centre for their personal training. I also appreciate the enthusiasm, helpful suggestions and support of Dr. Jeff Roberts, Ms. Tanya Mah, Mr. Alireza Azmudeh, Mr. Noam Butterfield, Dr. Stephen Schwarz, Dr. Troy Sarich, and Dr. AN Nekooeian during my years as a graduate student.  I would like to thank all the members of the Clinical Pharmacology Research Organisation (CPRO) and the Department of Pharmacology & Therapeutics, in particular, Drs. Morley Sutter and David Godin for taking the time to discuss interesting topics in science like the meaning of the word 'hypothesis' and distinguishing empirical science from superstitious pseudoscience.  I would also like to extend my gratitude and thanks to The University of British Columbia for its financial support by awarding me a University Graduate Fellowship and a Faculty of Medicine Summer Studentship this past year.  Most importantly, I would like to thank my family for their support and encouragement in helping me through all my endeavours.  L FRANCIOSI  1  PART I. THE QUALITY OF CLINICAL TRIALS  1.  WHAT IS A CLINICAL TRIAL? A clinical trial is both an experiment and an argument.  A clinical trial is a planned, controlled and ethical human experiment that has the potential for improving a patient's quality of life and developing a health care system that is better managed (Altman, 1991; Friedman et al., 1996; Lawrence, 1991). But is it just an experiment? At the beginning of one his lectures on experimental design and statistics, Dr. Michael J. A. Walker made the comment that an experiment is not simply a test or a trial but also an argument! When a person conducts an experiment, he or she does so in order to demonstrate, refute or explain to an audience their claim or conclusion. Hence, a clinical trial is more than an experiment, it is also an argument that consists of premises leading to a conclusion (Copi, 1972; Mason et al., 1989).  As an experiment and an argument, a clinical trial has a hypothesis as its major premise, a number of direct observations as its minor premises, and a scientific conclusion as its argumentative conclusion (Table 1). In order for a clinical trial to operate as both, it must test whether the hypothesis is supported by the experimental observations. If the clinical trial is designed and implemented appropriately, then the  L. FRANCIOSI  2  observations will lead logically to a conclusion. (Copi, 1972; Mason et al., 1989; Soccio era/., 1992).  A CLINICAL TRIAL IS BOTH AN EXPERIMENT AND AN ARGUMENT An Experiment has:  An Argument has:  Example Statement  Hypothesis  Major Premise  Based on experience, Drug A reduces blood pressure more than Drug B  Direct Observations  Minor Premises  In a controlled setting, Drug B subjects were observed to have blood pressures that were 25% lower than Drug A subjects  Scientific Conclusion  Argumentative Conclusion  Drug A does not reduce blood pressure more than Drug B  Table 1. A clinical trial is both an experiment and an argument.  2.  WHERE DO CLINICAL TRIALS FALL IN THE CLASSIFICATION OF MEDICAL INVESTIGATIONS? Clinical trials are classified  as experimental  analytic  studies.  A medical investigation is an inquiry into, or a study of, the prevention, diagnosis, alleviation, or curing of disease (Paikeday, 1981). There are two main categories of studies: descriptive and analytical (Figure 1). Descriptive studies are concerned with "describing the general characteristics of the distribution of a disease, particularly in relation to person, place, and time" (Hennekens et al., 1987). These studies can be  L FRANCIOSI  3  to c o '•4—»  co "w  CD >  o  tu E c o »  CO o >^ CO  to _co  o  <D 3  o>  L. FRANCIOSI  done on a whole population or on a single individual.  4  In population studies, the  objective is to use "data from entire populations to compare disease frequencies between different groups during the same period of time or in the same population at different points in time" (Hennekens et al., 1987).  Case reports or case series are  detailed descriptions of specific characteristics in a single patient or in a series of patients, respectively (Essex-Sorlie, 1995). individuals is the cross-sectional survey.  Another type of descriptive study of  This looks for an association between a  disease and its potential causes by studying the characteristics of a group of individuals at one point in time (Essex-Sorlie, 1995; Hennekens et al., 1987). Most descriptive studies can be done easily and quickly due to readily available information on individuals; hence, they are relatively inexpensive to perform (Hennekens et al., 1987). The primary usefulness of descriptive studies is in the formulation of hypotheses which can then be tested using analytical studies (Hennekens et al., 1987; Mitchael et al., 1984).  Analytical studies are subclassified as either observational or experimental (EssexSorlie, 1995; Hennekens etal., 1987). In observational studies, the clinical investigator observes which individuals are exposed or not exposed to a certain factor and then, he or she notes which individuals developed a particular outcome of interest (Hennekens et al., 1987; Mitchael et al., 1984). One type of observational study is the retrospective case-control. In this study, one group of patients who has a disease and another group who does not have it (i.e., a control group) are selected. They are then compared to determine the proportion of patients in each group that were exposed to some previous factor, for example, the consumption of red wine. If this factor is more common in the  L. FRANCIOSI  5  diseased patients, then this information would be regarded as evidence that the factor caused the disease (Hennekens et al., 1987). A second type of observational study is the cohort, which can be done either prospectively or retrospectively. In a prospective cohort, two groups of individuals are assembled: one group is exposed to a factor and the other group is unexposed. In such a study, the disease has not yet occurred in these individuals; thus, they are followed at regular time intervals to learn about the development of the disease or to determine the outcome of interest (Hennekens et al., 1987; Essex-Sorlie, 1995). When a cohort is done retrospectively, the investigators look back in time by reviewing the medical records of individuals in either exposed and unexposed groups. In other words, the study occurs after the disease or outcome of interest has occurred (Hennekens et al., 1987; Essex-Sorlie, 1995).  Another form of analytical study is the experimental study, or clinical trial, in which individuals are selected and assigned to receive a certain treatment (Hennekens et al., 1987; Essex-Sorlie, 1995). The treatment (or intervention) that is being investigated may be a "prophylactic, a diagnostic or therapeutic agent, a device, a regimen, a procedure, or a control" (Friedman et al., 1996).  The individuals are evaluated  prospectively to determine whether the treatment has produced some outcome of interest (Friedman era/., 1996; Hennekens etal., 1987; Meinert et al., 1986).  Clinical trials are subclassified as either primary or secondary preventative trials (Hennekens et al., 1987). In primary prevention trials, a medical treatment is evaluated in individuals that are usually healthy, but are at risk of developing a disease (Hennekens et al., 1987).  In secondary prevention trials (or therapeutic trials), the  L. FRANCIOSI  6  individuals are patients who already have a particular disease and the evaluation involves the determination of the ability of a treatment to "diminish symptoms, prevent recurrence, or decrease the risk of death from that disease" (Hennekens et al., 1987). Usually, the evaluation performed in either a preventive or therapeutic clinical trial involves an explicit comparison of groups (Hennekens et al., 1987). For example, if a clinical investigator has a question about a relationship or difference between two treatments, then he or she can select a number of individuals, assign them to be in either treatment group, and then compare the groups by controlling and systematically testing them in an objective, unbiased manner (Altman, 1991; Hennekens etal., 1987).  3.  WHAT IS THE PURPOSE OF CLINICAL TRIALS?  Clinical trials are designed and implemented to evaluate the safety, the efficacy, the efficiency or the effectiveness of medical treatments. Clinical trials provide evidence that can be used to make decisions concerning the care of individual patients and the control of health care costs.  Clinical trials are designed and implemented to evaluate the safety, the efficacy, the efficiency or the effectiveness of medical treatments (Feinstein, 1977, 1985; Friedman etal., 1996; Spilker, 1991).  The evaluation of 'safety' is an important part of most clinical trials, in particular when the medical treatments are drugs (Spilker, 1984, 1991). A complete definition of safety is difficult to find, but it usually involves an evaluation of the dose tolerance, the dose frequency, and the duration of exposure to a treatment (Spilker, 1984, 1991).  L FRANCIOSI  7  Evaluating the 'efficacy' of a treatment involves determining if it actually does something, i.e. Does it work?  This means comparing the treatment to either no  treatment or what I will later describe as a 'placebo' control (Feinstein, 1985).  When a clinical trial evaluates the 'efficiency' of a medical treatment, the efficacy of the treatment is compared to the efficacy of a standard treatment. This usually entails asking the question:  Does the treatment work better than the standard? (Feinstein,  1985).  A clinical trial that assesses a treatment's 'effectiveness' involves determining the extent to which the treatment achieves its intended purpose (Stedman, 1995). In other words, the impact of the treatment on society is evaluated in terms of convenience, comfort and cost (Feinstein, 1985).  The results from clinical trials represent evidence that can be used by either medical professionals or policymakers when they are faced with decisions concerning the care of individual patients and the control of health care costs.  The use of evidence in  medical decision-making is commonly referred to as 'evidence-based (Sackett et al., 1995; Sackett et al., 1996).  medicine'  Those who practise evidence-based  medicine may also use evidence from other types of medical investigations or from statistical analyses such as a meta-analyses which combine the results of similar clinical trials to base conclusions on a greater number of individuals (Dollery, 1996; Petitti, 1994; Sox et al., 1988; Sackett era/., 1991). However, proponents of evidencebased medicine prefer evidence from individual clinical trials, in particular, trials that  L. FRANCIOSI  8  have been designed and implemented in a randomised and controlled fashion (Sackett etal., 1995; Sackett et al., 1996).  4.  WHY ARE RANDOMISED CONTROLLED TRIALS PREFERRED?  Randomised controlled clinical trials are preferred because designed to control bias(es) that can influence medical  they are specifically decision-making.  It is well recognised by proponents of evidence-based medicine that many decisionmakers are using little or no evidence (Sackett et al., 1995; Smith, 1991). Physicians have little time to critically appraise and assimilate evidence from the medical literature (Davidoff et al., 1995; Pick-up et al., 1983; Sackett  al., 1995).  Evidence that is  available, either in the form of textbooks or journal articles, is often "hard to find, disorganised, outdated, and/or simply wrong" (Sackett et al., 1991).  Hence, many medical decisions may be influenced by misleading information or personal bias for certain therapies over others (Sackett et al., 1996).  In the Oxford  English Dictionary, the word bias is defined as "a swaying influence, impulse, or weight; anything which turns a man [or woman] to a particular course, or gives the direction to his [or her] measures."  Physicians may make therapeutic decisions on the basis of  experiences they had with previous patients or the consultations they had with either colleagues or drug company representatives (Dollery, 1996; Good, 1995; Sackett et al., 1995; Seale et al., 1994). They may also make decisions based on what the previous generation of physicians taught them in medical school (Seale er al., 1994; Smith, 1991). Some therapies have been taught and practised for years or even centuries without any scientific evidence to support their use (Rawlins, 1990). Past examples  L. F R A N C I O S I  include radical mastectomies  9  for breast cancer and blood letting for reversing  'hyperkinetic circulation' in patients with pneumonia.  Such biases can be controlled by the design and implementation of randomised controlled trials (Friedman et al, 1996; Sackett et al., 1996). In a British Medical Journal (BMJ) article, the editor also acknowledges this and goes further to say that one of the causes of wide variations in medical practise is the poverty of solid scientific evidence from clinical trials (Smith, 1991).  5.  WHY IS EVIDENCE FROM CLINICAL TRIALS POOR?  Many medical interventions still need to be evaluated using randomised controlled trials. Clinical trials that are conducted still have some deficiencies in their design and/or implementation because their investigators are poorly trained in the fundamentals.  In the BMJ article, the editor goes on to discuss that many medical treatments have yet to be evaluated using randomised controlled trials. He writes: "Where is the wisdom we have lost in knowledge, and where," asked T S Eliot, "is the knowledge we have lost in information?" There are perhaps 30,000 biomedical journals in the world, and they have grown steadily by 7% a year since the seventeenth century. Yet only 15% of medical interventions are supported by solid scientific evidence, David Eddy, a professor of health policy and management at Duke University, North Carolina, told a conference in Manchester last week. This is partly because only 1% of the articles in medical journals are scientifically sound and partly because many treatments have never been assessed at all. "If," said Professor Eddy, "it is true, as the total quality management gurus tell us, that 'every defect is a treasure' then we are sitting on King Solomon's mine." (Smith, 1991) 12  23  L. FRANCIOSI  10  His figure of 15% was determined in 1983 by assessing all currently used diagnostic and therapeutic procedures on the market and giving each an equal weighting in the final percentage (Greenhalgh, 1996; Smith, 1991). Another exact study was done in 1990 with a result of 21%; thus, more than three-quarters of treatments on the market still have to be evaluated (Greenhalgh, 1996). Thus, the best available evidence is usually something less than a randomised controlled clinical trial (Smith, 1991).  However, physicians think otherwise. In two recent surveys, physicians believe that just over 80% of the medical treatments that they actually use on their patients are evidence-based (Gill et al., 1996; Greenhalgh, 1996; Ellis et al., 1995). Does this imply that more medical treatments have been assessed using randomised controlled trials? A proponent of evidence-based medicine believes that the surveys were biased and that the earlier figures are more realistic (Greenhalgh, 1996).  What about the quality of randomised controlled trials that are designed and implemented?  The quality of clinical trials, in general, has improved over the years  (Friedman et al., 1996). However, "there is still considerable doubt that major progress has occurred" (Friedman et al., 1996). Clinical trials, including randomised controlled trials, still have deficiencies in design, conduct, analysis, presentation, or interpretation of the results because their investigators have a poor understanding  of the  fundamentals (Friedman et al., 1996; Schulz et al., 1995). This leads to the question: How can investigators assess and improve the quality of their clinical trials?  L. FRANCIOSI  6.  11  WHAT METHODS ARE AVAILABLE FOR ASSESSING THE QUALITY OF CLINICAL TRIALS AND WHAT IS THEIR MAJOR LIMITATION?  Scales and checklists have been developed to assess the quality of clinical trials. Most evaluate a trial's design and statistical analysis. Little consideration is made about assessing a trial's implementation.  In the clinical trial literature, two common methods are available to assess trial quality: scales and checklists (Moher et al., 1995; Spilker, 1991). They are usually made up of a selected number items that evaluate the quality of a trial's published report (Moher et al., 1995; Spilker, 1991).  In a scale, each item may be scored for its presence or absence, assigned a weighted score that depends on the its relative importance, or considered in terms of the rater's degree of satisfaction (Moher et al., 1995; Spilker, 1991).  Scores from separate  categories of items are added to derive a total composite score or index of quality (Spilker, 1991). In a checklist, no numerical scores are given to the items. Instead, the results of the checklist as well as the evaluator's comments are used to obtain a qualitative estimate of the overall quality (Moher et al., 1995).  In reading a review of these scales and checklists, I realised that developers of these methods were mainly concerned with evaluating a trial's reported design and statistical analysis (Moher et al., 1995). Little consideration was made about assessing a trial's implementation. If this assessment was included, then it would reveal to investigators what important items they may have forgotten, not known about, or ignored while conducting the clinical trial. This would allow them to consider the relevant items when  L. FRANCIOSI  12  planning their next clinical trial. The result would be an improvement in the quality of their clinical trials.  This lack of completeness was a concern. I used some of these scales and checklists to assess my trials, and I was getting a variety of results. Like the authors of the review article, I felt that developers of these scales and checklists did not really define the construct 'quality' (Moher et al., 1995; 1996). Thus, based on my intimate knowledge of my trials, I developed a Clinical Trial Evaluation System (CTES) for assessing the quality of clinical trials.  L. FRANCIOSI  13  PART II. THE DEVELOPMENT OF A CLINICAL TRIAL EVALUATION SYSTEM (CTES) FOR EVALUATING THE QUALITY OF CLINICAL TRIALS  7. WHAT IS A DEFINITION OF QUALITY? Quality can be defined as the best possible trial or the randomised controlled trial that is designed and implemented according to Good Clinical Practices.  There is agreement amongst clinical trial experts that the randomised controlled trial is the best clinical trial for minimising or avoiding potential biases in treatment evaluations (Altman, 1991; Chalmers et al., 1981; Friedman era/., 1996; Moher et al., 1995, 1996; Pocock, 1983; Sackett et al., 1991; Sackett et al., 1995; Sackett et al., 1996; Spilker, 1991). It can be designed and implemented according to the highest standards. These standards are commonly known as Good Clinical Practices (Allen et al., 1991; Hvidberg, 1994; Morice, 1991; Regnier, 1990; Shenfield etai, 1991). Thus, a definition of quality can be the best possible trial or the randomised controlled trial done according to Good Clinical Practices.  8. WHAT ARE GOOD CLINICAL PRACTICES? Good Clinical practices are a set of standards for the design, planning, conduct, analysis and assessment of clinical trials. Their purpose is to protect the rights of human subjects, to ensure the collection of accurate and consistent data, and to prevent fraud by those who conduct clinical trials.  L. FRANCIOSI  14  Good Clinical Practices (GCPs) are "a series of guidelines which specify points required for clinical trials to be acceptable to regulatory authorities" (Morice, 1991). In particular, they are "a set of standards for the design, planning, conduct, analysis and assessment of clinical trials" (Shenfield etal., 1991).  The purpose of GCPs is to protect the rights of human subjects, to ensure the collection of accurate and consistent data, and to prevent fraud by those who conduct clinical trials. (Allen et al., 1991; Hvidberg, 1994; Morice, 1991). GCPs evolved from Good Manufacturing Practices (1963) and Good Laboratory Practices (1978), which are codes of conduct that were first introduced by the United States to address the fraudulent practices that occurred during the manufacturing of pharmaceuticals and the conduct of laboratory research, respectively (Shenfield et al., 1991). Prior to the 1960s, clinical drug trials were widely considered "second rank research" since their "scientific basis was dubious and their conduct often sloppy" (Hvidberg, 1994). During the 1960s and 1970s, it became apparent to the US Food and Drug Administration (FDA) "that the conduct of clinical drug trials may have caused violation of human rights and may have been subject to scientific misconduct or even fraud" (Hvidberg, 1994). This forced the FDA to introduce a code of Good Clinical Practices in 1977 (Hvidberg, 1994; Shenfield et al., 1991).  These practices were regarded as "a set of principles and not legal  requirements"; they were a "standard of practise which was acceptable to the FDA" (Morice, 1991). Other regulatory authorities around the world adopted this code with some slight differences (Allen etal., 1991; Morice, 1991).  L FRANCIOSI  9. WHAT A R E THE PRINCIPAL PRACTICES FOR IMPLEMENTING THE BEST POSSIBLE CLINICAL TRIAL?  DESIGNING  15  AND  The principal GCPs are: define and state a primary question in advance, choose a design that is appropriate for the question posed, use proper statistics, consider the ethical issues involved, and have standard operating procedures in place to implement the clinical trial.  In reviewing the standards by which clinical trials are considered 'acceptable' in Europe (e.g., Bennett, 1994; Bohaychuk et al., 1993; Cohen et al., 1995; Spriet et al., 1985, 1992) as well as the FDA guidelines expressed in textbooks and manuals on the design and implementation of trials (e.g., Friedman et al., 1996; Sayre, 1994; Shapiro et al., 1983; Spilker, 1991), there are five common Good Clinical Practices. They are: 1. Define and state the primary question in advance; 2. Choose a design that is appropriate for the question posed; 3. Use appropriate statistics; 4. Consider the ethical issues; 5. Have standard operating procedures in place.  9.1 DEFINE AND STATE THE PRIMARY QUESTION IN ADVANCE A good clinical practice in the design and implementation of the best possible trial is to clearly define and state the primary question, well in advance (Bohaychuk et al., 1993; Carpenter, 1993; DeAngelis, 1990; Friedman et al., 1996; Hamilton, 1974; Lawrence, 1991; Portney et al., 1993; Shapiro et al., 1983). To follow this practice, the investigator should consider what is involved in defining the primary question and why it is necessary to state it in advance.  L. FRANCIOSI  16  9.1.1 WHAT IS INVOLVED IN DEFINING A PRIMARY QUESTION? Defining a primary question involves identifying a specific problem that needs a solution, reviewing the relevant literature, making a clear and concise statement of the problem in the form of a question, considering the question's importance, and refining the question to make it answerable.  9.1.1.1 Identifying  a specific problem that is need of a solution  Before any clinical trial is designed and implemented, a specific problem that needs a solution should be identified by the investigator (Portney et al., 1993). involves considering where problems arise.  This task  Three sources of problems are past  research, personal experience, and theories (Christensen, 1991; Portney et al., 1993). 9.1.1.1.1 Past research Past research can indicate to investigators what problems have already been investigated and what problems need to be investigated in an area of interest (Christensen, 1991; Portney et al., 1993). Problems can arise from inconsistencies or limitations that occurred in the methodology of past trials. Problems can also arise out of past research from descriptive studies, which "document trends, patterns or characteristics in data" (Portney et al., 1993). 9.1.1.1.2 Personal experience In clinical practice, problems can develop when the clinician is presented with some dilemma or when there may be some aspect of practise that requires documentation (Portney et al., 1993). In particular, problems can arise out of the need to "objectively document the effects of treatments that are based in tradition, authority, or trial and error" (Portney et al., 1993). Such treatments are based on "assumptions that need to  L. FRANCIOSI  17  be challenged and tested" (Portney et al., 1993). By doing so, the objective evidence obtained from a trial can be presented to others and can be used in future decisionmaking (Portney etal., 1993). 9.1.1.1.3 Theories  Theories are groups of interrelated concepts, definitions, or propositions that are logically organised to give meaning to a complex collection of individual facts and observations (Christensen, 1991; Portney et al., 1993).  Problems can arise when  theories are used to "develop broad generalisations from observation or to verify explanations by testing them in practice" (Portney et al., 1993). 9.1.1.2 Reviewing the relevant literature  Once the problem is identified, a review of the literature should be conducted to determine the current state of knowledge about it (Cohen et al., 1995; Portney et al, 1993). This process usually involves searching through books, journals or computer databases.  The search may reveal that a solution already exists for the problem.  Thus, the investigator should consider whether the problem is still worth pursuing (Carpenter, 1993; Christensen, 1991; Portney etal., 1993). 9.1.1.3 Making a clear and concise statement of the problem in form of a question  If the results of literature review indicate that no solution exists for the problem, then the investigator should make a clear and concise statement of the problem. This involves formulating and stating an interrogative sentence, or question (Christensen, 1991; DeAngelis, 1990; Friedman et ai, 1996; Portney et al., 1993).  This presents the  problem in a manner that minimises its interpretation and distortion (Christensen,  L. FRANCIOSI  18  1991). Sometimes investigators may pose an "objective", an "aim" or a "purpose" in place of a question (Christensen, 1991; Portney et al., 1993). Each kind of statement is acceptable.  However, investigators should be aware that these statements "do not  necessarily communicate the problem to be investigated" (Christensen, 1991). 9.1.1.4 Considering the importance of the question  When the question is stated, there should be some consideration of its clinical importance (Cohen etal., 1995; Lawrence, 1991; Portney etal., 1993). In other words, does the question pass the "so what" test? The results of a question's investigation should have the potential to impact existing treatment, theory, or policies related to practice (Portney et al., 1993).  By considering the question's importance, it will  determine its worthiness for investigation as well as aid in the process of refining the question (Carpenter, 1993; Cohen etal., 1995; Lawrence, 1991; Portney etal., 1993). 9.1.1.5 Refining the question to make it answerable  Whether the question pertains to establishing a relationship between drug dose and a therapeutic response or it entails resolving a difference (or a lack of a difference) between two or more treatment groups, the question should be refined so that it is specific enough to provide the relevant information that is being sought (Cohen et al., 1995; DeAngelis, 1990; Feinstein, 1995; Goodwin, 1995; Hennekens et al., 1987; Portney et al., 1993; Shapiro et al., 1983).  This process involves selecting an  appropriate response variable and then specifying it along with the subject population of interest, the treatments to be investigated, the time for follow-up, and if possible, the  L. FRANCIOSI  19  direction and magnitude of the relationship or difference that is to be detected (Cohen etal., 1995; Friedman etal., 1996; Selwyen, 1996; Shuster, 1990). 9.1.1.5.1  Selecting a primary r e s p o n s e variable to m e a s u r e  When the importance of a question is considered, the investigator should define the question in terms of a primary response variable that is both clinically meaningful and relevant (DeAngelis, 1990; Friedman et al., 1996). If the investigator decides to use more than one response variable, then the possibility exists that a significant result may occur by chance alone (Friedman et al., 1996). Interpretation can also become difficult when several variables give inconsistent results (Friedman et al., 1996). Investigators would have to consider which one of the variables is important and explain why the others give conflicting results (Friedman et al., 1996). "Unless they have made the determination of the relative importance prior to data collection, their explanations are likely to be unconvincing" (Friedman et al., 1996).  Selecting a primary response  variable involves understanding what it is in relation to other types of variables, knowing how it can be measured, and considering the problems associated with selecting and measuring variables. 9.1.1.5.1.1  Understanding  the response  variable  in relation to other types of  variables  A variable is defined as "any characteristic of an organism, environment, or experimental situation that can vary from one organism to another, from one environment to another, or from one experimental situation to another" (Christensen, 1991). In a clinical trial, the investigator "directly varies some factor (or factors), holds all else constant, and then observes the results of the systematic variability" (Goodwin, 1995). The factors under the control of the investigator are called the independent  L FRANCIOSI  20  variables, the putatively constant variables are referred to as extraneous or confounding variables, and the behaviours or outcomes measured are called the dependent variables or response variables (Goodwin, 1995). In the clinical setting, examples of independent variables may be age, time, drug dose, or exposure level (Friedman et al., 1996).  Confounding variables may or not be known (Goodwin, 1995).  variables may be total mortality, death from a specific cause,  a  Response laboratory  measurement, the amount of drug consumed, or the cost of a treatment (Friedman et al., 1996).  9.1.1.5.1.2  Knowing  how a response  variable can be  measured  In order to measure a response variable, the investigator should use a proper measurement scale (Breakwell et al., 1995; Goodwin, 1995; Portney et al., 1993). There are four possible measurement scales: nominal, ordinal, interval and ratio. 9.1.1.5.1.2.1 Nominal Measurement Scale  A nominal scale is scale where "objects or people are assigned to a name, number, letter or symbol according to some criterion" (Portney et al., 1993). For example, in a university exam, each student may be assigned to a seat number. The resultant list of student names and their respective seat numbers represents a nominal scale. 9.1.1.5.1.2.2 Ordinal Measurement Scale  An ordinal scale is a scale in which categories are rank ordered (Breakwell et al., 1995). For example, students may be ranked according to the letter grades they received on the exam.  L. FRANCIOSI  21  9.1.1.5.1.2.3 Interval Measurement Scale  An interval scale is like an ordinal scale except it has equal distances or intervals between the events that are ordered (Goodwin, 1995; Portney et al., 1993). This scale may be the time of day when the exam began (ex. at 9:00AM) to the time it ended (ex. at 10:00AM). 9.1.1.5.1.2.4  Ratio Measurement Scale  A ratio scale is an interval scale that has an absolute zero value, or a complete absence of what is being measured . Again using the exam as an example, the total amount of time required by students to complete the exam is a ratio scale.  One student may  decide not to write the exam (i.e., time is zero) or another student may write and finish the exam in half time that it took the last student to hand in his or her exam. 9.1.1.5.1.3  Considering  the problems  associated  with selecting  and measuring  variables  9.1.1.5.1.3.1 Appropriateness of the Variable  Problems may occur when an inappropriate or unreliable variable is selected. (Goodwin, 1995; Mason et al., 1989). For example, an inappropriate primary response variable is blood pressure for assessing the efficacy of an antifungal drug. 9.1.1.5.1.3.2  Experimenter and Subject Bias  Other problems may be related to the generalisations made about the relationships between the response variable and the independent variable (Mason et al., 1989). For instance, if the investigator is not blind to the trial treatments, then he or she may collect the measurements in such a way that their personal bias may influence the results. This bias would be referred to as 'experimenter bias'. The same is also true for the  L. FRANCIOSI  22  subjects participating; they may produce a 'subject bias' in the results by telling the investigator what he or she may be pleased to hear.  This is particularly true if a  behavioural response or even a subjective response like pain is being measured (Mason etal., 1989). 9.1.1.5.1.4 Specifying the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and the possible direction and magnitude of the relationship or difference that is to be detected  A specific question should include: the primary response variable, the population of interest, the treatments in terms of dose and frequency, and the period of time for follow-up (Cohen et al., 1995, Friedman et al., 1996; Selwyen, 1996; Shuster, 1990). Many questions (for example, '7s drug A better than drug B") are too general and do not specify what important information will be derived upon its investigation (DeAngelis, 1990; Friedman et al., 1996; Selwyen, 1996). Thus, making the question specific (for example, "In population W, is drug A at daily dose X more efficacious in reducing Z over a period of time T than drug B at daily dose Y?")  will likely lead to a more  informative answer (Friedman et al., 1996).  If possible, a specific question should also specify the direction and magnitude of the relationship or difference that is to be detected (Friedman et al., 1996; Shuster, 1990). In the previous example, the direction was a reduction of Z in favour of drug A, but the magnitude of this reduction was not stated. Some clinically relevant effect sizes may be 20%, 50%, or 80% (Cohen, 1988).  L. FRANCIOSI  23  Investigators should also realise that if the question is made too specific, then the generalisability of the results to other populations may be difficult (DeAngelis, 1990). For example, if the question indicates that a young population with a particular disease is going to be investigated, then it may be difficult to generalise the results to an elderly population who may also have the same disease.  9.1.2 WHY STATE THE QUESTION IN ADVANCE? Stating the question in advance is essential because it allows the investigator to consider the question's feasibility and to formulate and state a scientific hypothesis. In the process, the investigator can begin planning the design and implementation of the clinical trial.  Once the primary question is defined, it should be stated prior to starting the trial (DeAngelis, 1990; Friedman et al., 1996; Portney et al., 1993). One reason is that it allows the investigator to consider whether or not it is feasible to pursue the question. If the question is feasible, the investigator can then also formulate and state a scientific hypothesis. By performing these activities, it enables the investigator to begin planning the design and implementation of the clinical trial (Christensen, 1991; Cohen et al., 1995; DeAngelis, 1990; Feinstein, 1985; Friedman etal., 1996; Portney etal., 1993). 9.1.2.1 Considering the question's feasibility  Considering a question's feasibility mainly involves evaluating whether or not there are enough resources to investigate it. The investigator should assess the question in terms of time, subject population, expertise, expense, facilities, equipment, and ethical implications (Christensen, 1991; Cohen etal., 1995; DeAngelis, 1990; Friedman etal.,  L. FRANCIOSI  24  1996; Portney et al., 1993). He or she can do this by asking themselves a number of questions. 9.1.2.1.1 Time Can the question be answered within a realistic period of time? Can a timetable be developed?  Performing a pilot study may be useful in estimating time requirements  (Portney et al., 1993). Is the question still relevant once the trial is completed? New advances may occur during a lengthy trial that the question or answer may no longer be relevant in clinical practise (Friedman etal., 1996). 9.1.2.1.2 Subject Population Is there a sufficient number of subjects available? Healthy persons are usually more available than patients (Cohen et al., 1995). The investigator should not overestimate the number of subjects that can be recruited (Cohen et al., 1995). The question will only be answered if there is an initial assessment of subject recruitment (Cohen et al., 1995; Portney etal., 1993) 9.1.2.1.3 Expertise Is the person asking the question, i.e. the investigator, experienced enough to carry out the clinical trial? Are the clinical trial personnel (i.e., physicians, nurses, research assistants, monitors, computer programmers, statisticians, pharmacists, secretaries, data typists, lab technicians) qualified to assist the investigator in answering the question? Is training necessary? Is there enough time available for training? Is there an immediate need for consultants and advisors for technical and statistical assistance? (Christensen, 1991; Cohen etal., 1995; Portney etal., 1993)  L. FRANCIOSI  25  9.1.2.1.4 Expense  What are the budgetary requirements for the question? This involves considering the costs associated with personnel, hospital accommodation, treatments, lab assays, disposables, travel, stationary, and capital equipment (Cohen et al., 1995).  Some  clinical trials may cost anywhere between $3000 and $15,000 US per subject (Hennekens etal., 1987). 9.1.2.1.5 Facilities and Equipment  What type of space is necessary to answer the question? available?  Does the facility have the proper equipment and instruments for  investigating the question?  Is the equipment available? Is the facility a teaching  hospital? Is there easy access to emergency staff? organisation?  Is there enough space  Is the facility a research  Is it adequately equipped to handle emergencies?  Is it within a few  minutes drive of a hospital? Is there easy access for an ambulance? Are the corridors or doorways of the facility wide enough for a bed or stretcher to pass? (Cohen et al., 1995; Portney etal., 1993) 9.1.2.1.6 Ethical Implications  Does the question pose any serious ethical issues? Are the rights of subjects being protected? (Portney et al., 1993)  The investigator must ensure that the question  satisfies the six ethical principles of research: "(1) the research is necessary to accomplish an important goal for the benefit of society; (2) the information cannot be obtained in any other way; (3) the foreseeable benefits outweigh the possible risks to the subjects, and the risks themselves are not serious; (4) only qualified investigators conduct the study, conforming to a scientifically valid study design; (5) the subjects  L FRANCIOSI  26  selected are both necessary and appropriate to obtaining the necessary information; and (6) each subject voluntarily consents to participate in the research" (Bankowski et al., 1993). 9.1.2.2 Formulating  and stating a scientific  hypothesis  If the question is feasible to pursue, the investigator should then formulate and state a scientific hypothesis (Christensen, 1991; Friedman et al., 1996; McConway, 1994; Portney etal., 1993).  A scientific hypothesis is nothing more than "an informed guess about the world" (McConway, 1994). In simple terms, it may be regarded as a possible answer to the question being asked (Christensen, 1991; Portney etal., 1993). Sometimes, a scientific hypothesis is a proposed explanation for a causal observation of events (Christensen, 1991; Copi, 1972). Such a hypothesis is usually relevant to the fact that it is intended to explain, testable if some observable fact is deducible from it, compatible with previously established hypotheses, more powerful than other hypotheses in allowing predictions or explanations to be made from it (i.e., has a greater number of observable facts deducible from it), and simple enough to explain or fit the facts together (McConway, 1994; Copi, 1972).  When a scientific hypothesis is formulated, it is put forward tentatively and is more or less probable on the basis of available facts or evidence gathered from a review of the literature (Christensen, 1991; Copi, 1972; Portney et al., 1993). Its statement also aids  L. FRANCIOSI  27  the investigator in determining how the data is going to be analysed and interpreted following the completion of the clinical trial (Portney etal., 1993).  During data analysis, a statistical hypothesis, or 'null hypothesis' is formulated. This hypothesis differs from the scientific hypothesis in that it expresses no relationship or no difference between the independent and dependent variables (Portney et al., 1993). In other words, it is the opposite of the scientific hypothesis (or the 'alternative hypothesis'). The reason for a having a null hypothesis in statistics is because it is the actual hypothesis that is being tested in a clinical trial (Christensen, 1991). In order to obtain support for the scientific hypothesis, evidence must be collected that enables the investigator to reject the null hypothesis (Christensen, 1991).  A scientific hypothesis "cannot be proven but only disproven" (Sutter, 1996; McConway, 1994). Investigators may only design and implement clinical trials "that have a chance of supporting their 'pet' hypothesis and neglect to carry out those that have a chance of disproving it" (McConway, 1994).  By focusing on disproof, the investigators are  protected "from their inevitable vested interest in the outcome of their experiments" and it "encourages the ideal objective detachment investigation" (McConway, 1994).  required by scientific methods of  L. FRANCIOSI  28  9.2 CHOOSE A DESIGN THAT IS APPROPRIATE FOR THE QUESTION POSED  When the primary question is defined and stated in advance, the investigator should choose an overall design that is appropriate for it (Bennett, 1994; Bohaychuk et al., 1993; Cohen et al., 1995; Friedman et al., 1996; Sayre, 1994; Shapiro et al., 1983; Sprietef a/., 1985, 1992). In order to accomplish this, a number of considerations have to be taken into account.  The population that the question addresses has to be identified and described to allow the sampling and the screening of potential subjects according to defined eligibility criteria. Since the primary variable should have already been chosen, the investigator can calculate the appropriate numbers required in each trial group; this process establishes the size of the sample to be taken.  An experimental design has to be  considered to address the specific question being asked. In the best possible trial, a parallel group design is the most appropriate.  With such a design, control and test  treatments need to be defined. Upon considering the type of design, eligible subjects have to be randomised to treatments, thus creating control and test groups which have to be assessed for baseline characteristics. The concern is that subjects in each group should adequately represent the population of interest.  When subjects are being  randomised, some method of blinding (i.e. single blind, double-blind) should be considered to avoid selection bias. Throughout the design process, there must be consideration of generalisability and confounding to direct decision-making.  L. FRANCIOSI  29  9.2.1 SAMPLING AND SCREENING SUBJECTS FROM A POPULATION  9.2.1.1 Identifying and describing the population that the primary question addresses  In clinical trials, a population refers to the members of a clear set or class of subjects with the condition or characteristics of interest (Friedman et al., 1996; Payton, 1994). In the best possible trial, a population is identified and then described in order to "confirm the appropriateness of the sample taken" and to "know to which groups the results may pertain" (Lang et al., 1997). "The description should include relevant demographic, diagnostic (including the stage of disease), prognostic, and comorbid factors" (Lang et al., 1997). As well, "major subgroups of interest in the population should be identified" (Lang et al., 1997). Once the population is adequately described, subjects should be sampled from the population and then screened according to predetermined eligibility criteria (Friedman et al., 1996). 9.2.1.2 Considering the method of sampling subjects from the population  A sample is composed of those subjects selected from a population for investigation. This sample is also referred to as the 'study' population (Friedman et ai, 1996; Payton, 1994). There are two types of sampling procedures: idealised (random sample) and nonprobability (convenience sample). 9.2.1.2.1 Idealized sampling procedures: the relevance of a random sample In the 'ideal' clinical trial (not to be confused with the 'best possible' trial), it is important to ensure that the groups of subjects being compared are the same in all relevant respects beforehand (Finney, 1980; Hill etal., 1991; Zolman, 1993; Ludbrook, 1994a; Rosenthal et ai, 1984). This entails identifying the distributing factors or characteristics  L. FRANCIOSI  30  of the world's population of subjects and then equalising them in the groups being evaluated, i.e. the control and treatment group. identify all these factors precisely.  However, it may be impossible to  Thus, in the ideal clinical trial, the method of  randomly sampling is used to overcome this problem.  Based on probability, the sample is considered random only when every subject in the world's population of subjects is given an equal and independent opportunity to participate in the clinical trial. Practically, this would mean using a table of random numbers, a computer random number generator or simply putting the name of each subject in a hat and drawing out the required number of subjects (Goodwin, 1995; Hill et al., 1991; Payton, 1994; Zolman, 1993).  In doing so, the sample would be  representative of the world's population of subjects, and thus, would not be biased in any fashion when the results are generalised (Knapp et al., 1992). A random sample also represents the ideal sample for using inferential (parametric) statistics to draw valid conclusions about the world's population of subjects (Altman, 1991a; Lumley et al., 1994). 9.2.1.2.2 Nonprobability sampling procedures: a sample of convenience In the best possible trial, a sample of convenience is obtained. It is a method that is based not on probability but rather on 'availability' (Goodwin, 1995; Hulley et al., 1988). The investigator requests subjects from an accessible population (e.g. a hospital) who meet the general requirements of the clinical trial. Why not a random sample? It is impractical for an investigator to sample the world's population of subjects because of the cost and time involved (Altman, 1991; Friedman etal., 1996).  L FRANCIOSI  31  9.2.1.3 Determining how the subjects are to be recruited  Subjects may be recruited by sending a mailed letter, making a telephone call, or using advertisements in form of posters, television, and radio (Friedman et ai, 1996; Spilker, 1991). Successful recruitment of clinical trial subjects usually involves developing a careful plan with multiple strategies, establishing interim goals, and preparing to devote the necessary effort (Friedman et al., 1996). 9.2.1.4 Screening Subjects for eligibility  9.2.1.4.1 Determining the eligibility criteria  The study population is defined by a number of inclusion and exclusion criteria (Friedman et al., 1996). These criteria, which are stated in advance, "relate to the safety and the anticipated effect of the intervention" (Friedman et ai, 1996). Generally, the criteria are concerned with subjects who have the potential to benefit from the treatment and can provide the investigator the best possibility to detect a difference (Friedman et ai, 1996). Eligibility criteria are also concern with the possible adverse events that may occur with the treatment being investigated (Friedman et ai, 1996). Thus, any subject for whom the treatment is known to be harmful is not admitted into the clinical trial. Other criteria may relate to the subject's disease status or their ability to comply with the trial's procedures or requests for follow-up visits (Friedman et ai, 1996).  L. FRANCIOSI  9.2.1.4.2  32  Considering the types of subjects pre- and postscreening  9.2.1.4.2.1  Refusers  When potential subjects are asked to participate in the trial, there may be a few who are not willing to be screened or may not want to sign the informed consent form (Friedman et al., 1996; Spilker, 1991).  These persons are called 'refusers'.  The  number of refusers, their physical description, and if possible, their reasons for not participating should be recorded by the investigator (Hennekens et al., 1987; Spilker, 1991). 9.2.1.4.2.2  Nonqualifiers  Upon being screened, certain subjects may be labelled as 'nonqualifiers' because they did not meet the eligibility criteria (Friedman et al., 1996; Spilker, 1991). The number of nonqualifiers and their baseline characteristics should be recorded (Hennekens et al., 1987; Spilker, 1991) 9.2.14.2.3  Dropouts  Once eligible subjects have started in the clinical trial, they may be either unwilling to continue in the trial or simply lost to follow-up (Friedman et al., 1996; Spilker, 1991). They are now called 'dropouts'.  Again, the number of dropouts, their baseline  characteristics, and the reasons for dropping out should be recorded (Hennekens et al., 1987; Spilker, 1991) 9.2.1.4.2.4  Discontinuers  Eligible subjects may be discontinued or withdrawn by the investigator because they are not complying with protocol requirements, they are no longer eligible for participation or  L. FRANCIOSI  33  they had a (serious) adverse event (Friedman et al., 1996; Spilker, 1991). The number of discontinuers, their baseline characteristics, and the reasons for being discontinued should be recorded (Hennekens etal,, 1987; Spilker, 1991). 9.2.1.4.2.4.1 Monitoring subject compliance  The investigator should consider monitoring a subject's compliance with the trial's requirements (Friedman et al., 1996; Cohen et al., 1995; Spilker, 1991).  Some  examples of adherence-monitoring techniques are counting the number of pills, measuring physiological responses, (Friedman et al., 1996).  performing interviews, or giving out diaries  Such monitoring enables the investigator to consider the  possible influence of non-compliance in the trial's results (Friedman et al., 1996). 9.2.1.5 Considering the issues related to generalisability of the results  The reason that refusers, nonqualifiers, dropouts, and discontinuers are recorded is that this information allows the investigator to determine the presence of any differences between participants and nonparticipants (Hennekens etal., 1987; Spilker, 1991).  It also helps the investigator decide whether the results among the trial  participants are generaliseable to the population that they were sampled from (Hennekens et al., 1987).  The concept of generalisability is related to the trial's external validity (Goodwin, 1995; Mason et ai, 1989). This type of validity is concerned with the reliability or robustness of the clinical trial results (i.e., if the clinical trial was repeated a number of times, are the results going to be the same?) (Brinberg et ai, 1982).  The highest degree of  external validity is achieved when the investigator is able to extend his or her results to  L. FRANCIOSI  34  the other populations of subjects (e.g. from males to females), to other environments (e.g. from the laboratory to real life settings or situations), and to other times (e.g. from 1964 to 1997) (Goodwin, 1995). Thus, it is dependent on the characteristics of the subjects selected for the trial and the setting (conditions) that subjects had to endure (Mason et al., 1989). The investigator should consider what the potential threats are to a trial's external validity, in particular, factors that affect the trial's selection process and its setting. 9.2.1.5.1  T h r e a t s to t h e selection process  A specific threat to the generalisation of the trial's results may be some aspect of the selection process acting in combination with some other influence (Mason et al., 1989). This interaction may occur when subjects are selected based on "their tendency to give a certain kind of reaction to the treatments being investigated in the clinical trial" (Mason et al., 1989). This may be the case if highly motivated subjects are used because they may comply with the treatments better than the general population (Mason et al., 1989).  Another type interaction may occur between testing the subjects before entry into the trial and then selecting them to be in a particular treatment group (Mason et al., 1989). For instance, if hypertensive patients are given an antihypertensive drug before being randomised to either the antihypertensive or control, they may produce results that are only generaliseable to situations where a pre-test of the antihypertensive drug is given. Hypertensive patients that are not pre-tested may not produce the same results (Mason efa/., 1989).  L. FRANCIOSI  35  Another possible interaction may occur between the selection process and maturation (Mason et al., 1989).  For example, if growing children are administered a treatment  that takes m o n t h s or years to produce an effect, then the results m a y not  be  generaliseable to adults or to children at different stages of development.  9.2.1.5.2 Threats to the trial setting or conditions T h e generalisation of results may be influenced by the uniqueness of the setting w h e r e the trial is being implemented (Mason et al., 1989). T h e research protocol may require samples or m e a s u r e m e n t s not normally d o n e in clinical practise. T h e results of such a trial may be limited to the clinical trial situation.  9.2.1.6 Determining the appropriate sample size T h e investigator should determine the n u m b e r of subjects to be sampled for the trial by using mathematical equations (Bennett, 1994; Carpenter, 1993; Friedman etal., 1996; Lerman, 1996; Shuster, 1990).  T h e calculation of s a m p l e size is based on a primary  response variable and on adequate levels of statistical significance and power (to be discussed later) (Bennett, 1994; Carpenter, 1993; Friedman etal., 1996; Lerman, 1996; Shuster, 1990).  S a m p l e size equations can be found in most statistics textbooks or  literature reviews on s a m p l e size calculation (Friedman et al., 1996; Lerman, 1996; Shuster, 1990).  If the investigator calculates a sample size or c h o o s e s a n u m b e r  arbitrarily, he or she should provide a rationale or clinical justification for the s a m p l e size determined (Bennett, 1994).  L FRANCIOSI  36  9.2.2 CHOOSING AN EXPERIMENTAL DESIGN An experimental design is a strategy used by the investigator to control extraneous variables other than the primary response variable, and to allow for the generalisability of the results (Christensen, 1991).  It is dependent on whether the question is about  determining relationships or resolving differences between variables (Friedman et al., 1996; Goodwin, 1995; Lumley et al., 1994). Clinical trials, in particular drug trials, have been categorised to address each kind of question (Table 2).  When a primary question concerns evaluating the existence and magnitude of a relationship between two or more variables, the relationship may be either an association or one of cause and effect (Hennekens et al., 1987). Such questions are frequently asked by epidemiologists when evaluating relationships between disease and exposure (Hennekens et al., 1987). They may also be asked by pharmacologists when they are determining whether drug X at dose Y is toxic or if drug Z causes a dosedependent decrease in some measured response W.  Such questions are usually  investigated in Phase I and II clinical trials (Table 2) (Altman, 1991; Johnson et al., 1977).  In Phase I clinical trials, the questions are concerned with safety, in particular, determining the harmful effects that a treatment may have in humans (Altman, 1991; George, 1980; Johnson et al., 1977).  The test treatment is given to normal healthy  volunteers (or sometimes patients) to determine if there is any relationship between the doses used and any toxic side effects observed.  These type of trials can be  •  L. FRANCIOSI  37  randomised controlled trials, however, generally they are not because of safety and economic reasons. The investigator and volunteers know the treatment being given, and thus the trial results may be influenced by bias.  Hence, Phase I trials may be  referred to as uncontrolled and 'open-labelled' trials.  Phase II clinical trials, on the other hand, are trials that are normally conducted under randomised and controlled conditions. Their objective is to establish a test treatment's 'efficacy' or ability to produce a therapeutic effect in a selected group of individuals. Another way of putting it is that the trial's question concerns determining whether a relationship exists, for example, between the dose of the drug used and its observed therapeutic effect.  Categories  Objectives  Phase 1  Determination of safety  Phase II  Evaluation of efficacy  Phase III  Evaluation of efficiency or effectiveness  Phase IV  Postmarketing surveillance  Table 2. Categories of clinical trials and their objectives.  L. FRANCIOSI  38  When the primary question concerns determining the existence and magnitude of a difference or a lack of difference (equivalency) between variables, a Phase III clinical trial is conducted (Table 2). These trials, which are normally randomised controlled trials, use a sufficient number of subjects to provide a definitive assessment of the test treatment's efficiency and effectiveness (Altman, 1991; Johnson et al., 1977).  Phase IV clinical trials are extensions of Phase I, Phase II, and Phase III trials. They are concerned with evaluating either type of question after the test treatment has been marketed to the entire subject population.  Even though each category of clinical trial may be designed a number of other ways (Appendix 2), the 'parallel group design' is preferred. This type of experimental design is the basis of the randomised controlled trial or the best possible trial (Altman, 1991). 9.2.2.1 Parallel Group  Design  In a parallel group design, two (or more) groups of subjects are studied at the same time (Altman, 1991; Johnson et al., 1977). This design involves defining a control and test treatment(s), selecting a method of randomising eligible subjects to treatments, determining the potential degree of blinding that is possible, and lastly, considering the potential known or unknown confounders that the design is suppose to control. 9.2.2.1.1 Defining the control and test treatments 9.2.2.1.1.1  Understanding  the word  control  The word "control" was derived from "counter-roll", a master list that was used to check and correct other lists (Rosenthal et al., 1984). In clinical research, it may be expressed  L. FRANCIOSI  39  as three distinct ideas: regulation, comparison, and control period (Appendix 3) (Feinstein, 1973). Most commonly, 'control' refers to regulating all aspects of the trial except the one or more aspects that are being studied. This attempt is also referred to as the "control of confounding variables". Such variables can confound or confuse the interpretation of the results if not controlled. 9.2.2.1.1.2  Defining  the control  treatment  In a clinical trial, there are three possible types of controls or control groups: a standard treatment, a placebo, or no treatment. 9.2.2.1.1.2.1 A standard treatment  One type of control group may be made up of subjects who are given a therapy that is already used as standard medical practise.  These subjects are then compared to  another group of patients who are given a test treatment which is to be investigated. By comparing the two groups of patients, it provides a means of determining if the experimental therapy has more or less the same therapeutic effect as the standard therapy being used. 9.2.2.1.1.2.2 A placebo  Another type of control group may involve subjects that are given a placebo or an inert substance that is identical in appearance and taste to the test treatment, but has no therapeutic effect.  This blinds subjects and clinical trial staff to the identity of the  treatments. In this way, an unbiased assessment of the test treatment's efficacy can be obtained.  L. FRANCIOSI  40  9.2.2.1.1.2.2.1 The placebo effect  In a group given a placebo, a beneficial effect or side effect observed could be due to the belief that they had taken the test treatment. This is known as the 'placebo effect'. Some studies have suggested that a placebo effect may occur in 30% of subjects. 9.2.2.1.1.2.3 No treatment  The third type of control group may involve subjects being given no treatment. Such a control is used to compare the physical reaction of subjects to that of test treatment being given in the other group. However, this control is not normally used because of poor subject compliance, a high dropout rate, and for ethical reasons. 9.2.2.1.1.3 Defining the test treatment(s)  When the primary question is determined, the investigators already have in mind what type or class of treatment that they want to investigate.  This test treatment (or  intervention) may be a drug, a procedure, or a lifestyle modification. The investigator should define the treatment so that issues such as the potential benefits of the intervention can be weighed against potential the risks, its standardisation and stability over the course of the trial can be considered, and its availability in terms of ethical and/or regulatory approval can be determined. Its definition also aids with the logistics of other design aspects (Friedman et al., 1996).  A test treatment may be compared to another form or forms of itself. Thus, it can be defined in more than one way.  Examples include differing doses of phenobarbital,  different routes of administration of morphine, and the short term and the long term effects of aspirin.  L. FRANCIOSI  41  9.2.2.1.2 An example of control and test treatment groups  An example of control and test treatment groups is found in the story of how the ancient Egyptians discovered citron, a large lemon-like fruit, to be an antidote for poison (Jones, 1964; Rosenthal et al., 1984). The story is told as follows: It seems that a magistrate had sentenced some convicted criminals to be executed by being exposed to poisonous snakes. It was later reported to him, however, that none of the criminals had died despite the care in carrying out the sentence. Inquiring into the matter, he learned that the criminals, just before being bitten by the snakes, had been given citron to eat by an old woman who took pity on them. The magistrate hypothesized that it must have been the citron that had saved them, and he had the criminals divided into pairs in order to test his hypothesis. Citron was fed to one of each pair and not to the other. When the criminals were again exposed to the poisonous snakes, the ones who had eaten the citron suffered no harm while the untreated "controls" died instantly. (Rosenthal et al., 1984)  9.2.3 RANDOMISING ELIGIBLE SUBJECTS TO TREATMENTS 9.2.3.1 Methods for generating random assignment  When the sample is taken, another important issue is to ensure that the assignment of individuals to either the control or test treatment group is done in an unbiased way (independent of patient characteristics). Hence, the methods of random chance are used, i.e. using random number tables, simple random number generation, picking names out of hats (Altman, 1991a; Friedman et al., 1996; Goodwin, 1995; Hill et al., 1991).  L FRANCIOSI  42  9.2.3.2 Assessing the baseline characteristics of subjects  Following randomisation and prior to the administration of treatment, it is possible that the characteristics of the control group may differ from those of the test treatment group. In other words, randomisation may produce imbalanced groups at baseline.  Thus, the investigator should compare baseline characteristics, particularly, if the number of subjects per group is small. This assessment should include a comparison of  "risk  or  prognostic  factors,  pertinent  demographic  characteristics, and medical history" (Friedman et ai, 1996).  and  socio-economic  Such data can be  obtained from interviews, questionnaires, physical examinations, and lab tests.  If an imbalance exists, then the investigator can stratify or subgroup on the basis of these factors at the time of randomisation or during the statistical analysis of data (Altman, 1991b; Friedman etal., 1996; Guyatt ef ai, 1995a). 9.2.4 BLINDING ALL PARTIES IN THE TRIAL AS MUCH A S POSSIBLE In trials of antianginal drugs, an American doctor by the name of Harry Gold introduced a novel technique that was later became known as 'blinding' (Gold et ai, 1937). It involved telling neither the patient nor the physician what treatment was being given. This was done to avoid the potential bias that occurred from the knowing or unknowing communication of a physician's preference for a treatment to the patient. The patient's outcome was influenced to the extent that the patient knowingly or unknowingly performed in a manner that met the physician's expectations. This simple participation  L. F R A N C I O S I  43  and interaction between the two most often rendered the results biased for the preferred treatment.  Blinding was also developed to deal with biases related to selecting subjects and evaluating treatments whose effects were subjective (e.g., analgesics and the measurement of pain relief) (Rawlins, 1990). 9.2.4.1 Types of Blinding  9.2.4.1.1  D o u b l e blinding  The main parties in a clinical trial are the subjects, the investigator(s) evaluating the primary question, and the rest of the clinical trial personnel. After subjects have been randomly assigned to their respective groups, the way subjects and the investigator respond and observe during the trial may be affected by the knowledge of which treatment is being given and in turn favour or disfavour the use of that treatment. As already mentioned, a way to overcome this bias is to have neither the subject nor the investigator know what the treatment is being given. This technique is referred to as 'double blinding'. 9.2.4.1.2  Single-Blinding  If the subject does not know which treatment is being given, but the investigator knows due to safety or practical reasons, then this is known as single blinding.  L. FRANCIOSI  44  9.2.4.1.3 Blinding impossible; the blinding of an independent observer However, there are occasions when it is impossible to blind all parties in a clinical trial. This is often the case in trials of surgery. One method of overcoming this is to have someone outside the trial, i.e. a blinded independent observer perform the evaluations. 9.2.4.2 Assessing the adequacy of blinding  Prior to disclosing the identity of the treatments, the investigator should consider assessing the adequacy of blinding since the degree of blinding has a profound effect on data collection and interpretation (Friedman et ai, 1996; Spilker, 1991). A method of estimating the adequacy of blinding is to ask the subjects and the clinical staff to guess which treatment group the subject was in (Friedman et al, 1996). If fewer than half the subjects in each group correctly guess the treatment, then the investigator should suspect that all parties in the trial knew what was given but were not telling the truth (Friedman etai, 1996). 9.2.5  CONSIDERING THE POTENTIAL CLINICAL TRIAL  CONFOUNDING  FACTORS  IN T H E  The investigator should consider the degree to which the clinical trial is free of extraneous variables or confounders (Campbell et ai, 1963; Goodwin, 1995; Mason et ai, 1978; Mason et ai, 1989). This process is also referred to as considering the 'internal validity' of the trial. In an internally valid clinical trial, the results, as measured by the primary variable, are directly associated with the treatment(s) being investigated and not the result of a confounder (Goodwin, 1995). confounders:  history, maturation, testing,  There are different types of  instrumentation,  statistical regression,  selection of patients, mortality, interaction with the selection process and other  L. FRANCIOSI  45  confounders, and the maintenance of treatment conditions over time (Campbell et al., 1963; Goodwin, 1995; Mason etal., 1989). 9.2.5.1  History  As a confounder, history may manifest itself as a specific, uncontrolled event which occurs sometime before or during the conduct of the clinical trial (Goodwin, 1995; Mason et al., 1989). More specifically, it is an event that might occur in the time span between the measurements of the primary variable (Mason et al., 1978). Thus, the longer this time span is, the greater the possibility that such an event can happen in the trial setting (Mason et al., 1986). For example, if one subject in the test treatment group develops symptoms of the flu three days into the trial and then three days later, another subject ingests a substance that interacts with the test treatment's metabolism, then these events may nullify or enhance the observed effect between the control and test treatment groups. 9.2.5.2  Maturation  Maturation may become a threat when the subjects in the trial undergo some change over time within the treatment conditions (Campbell et al., 1963).  For instance, if  babies were used as subjects for a five year trial, then they will grow during its conduct. If subjects are examined every ten minutes for 72 hours, then may become tired or hungry. These physical, physiological, or psychological changes which are unrelated to the treatment being investigated may account for the observed results (Mason et al., 1989).  L. FRANCIOSI 9.2.5.3  46  Testing  Confounding may affect the results when subjects are tested or measured a number of times before, during and after the trial (Goodwin, 1995; Campbell et al., 1963). This confounding action is also known as 'repeated testing.'  Any observed differences  between the trial groups may explained by simple practice, boredom, and/or fatigue acquired during the continued testing (Mason etal., 1989). 9.2.5.4  Instrumentation  During the course of the trial, any changes that occur in the measuring instruments or devices can influence the outcome of the trial (Mason et al., 1989). For example, the misuse, calibration or servicing of a heart monitor during an antiarrhythmic trial may confound the results. Confounding may occur with a written examination that is taken before and after a trial (Goodwin, 1995). The level of difficulty may change from the first examination to the next, and thus, any improvement or decline in subject performance may be attributed to this confounding. Another problem may occur when the measuring device is a person making the observations (Goodwin, 1995, Mason et al., 1989).  In other words, an observer may get better or worse at measuring a  particular subject response, thus making the later observations different from the earlier ones. Problems such a practise, boredom or fatigue may account for this confounding (Mason et al., 1989). If more than one observer is measuring a subjective response, then any variation between observers may also affect the trial outcome (Goodwin, 1995, Mason et al., 1989).  L FRANCIOSI  47  9.2.5.5 Statistical Regression  Statistical Regression is another confounder that is evident when subjects  are  "assigned to extreme groups on the basis of a relatively unreliable pre-test" (Mason et al., 1989). For example, suppose subjects have their blood pressures measured and then on the basis of their readings, they get assigned to either a high blood pressure group or a low blood pressure group. Each group then undergoes a different treatment. Subjects who are in the high blood pressure group may average a lower blood pressure on the second reading regardless of what treatment they are given. The same scenario is also possible for subjects in the low blood pressure group; they may average a higher blood pressure on the second reading. This regression of pressures is often in the direction of the mean blood pressure from the normal distribution. Hence, the use of the terms 'statistical regression' or 'regression towards the mean' (Bland et al., 1994bd; Goodwin, 1995). Any observed difference or lack of difference between treatments may be explained by this type of confounding.  9.2.5.6 Selection of subjects  The method of selection, as seen in the previous example, is another possible threat to the internal validity of a clinical trial (Goodwin, 1995).  Subjects in one or more  treatment groups may be systematically different from subjects in the control group. This 'selection bias' may be related to age, sex, weight, or any other characteristic that might modify the magnitude of the effects seen in the trial (Campbell et al., 1963; Hennekens etal., 1987). For this reason, randomisation is used (Mason etal., 1989).  L FRANCIOSI  48  9.2.5.7 Mortality  When a clinical trial occurs over an extended period of time, subjects may move away, lose interest, or become uncomfortable such that they may decide to exercise their right to 'drop out' of the trial (Goodwin, 1995; Mason et ai, 1989).  There is also the  possibility that subjects may die during the course of the trial (Goodwin, 1995). If any of these events were to occur, they would be characterised as 'experimental attrition' or a loss of subjects from the trial population (Goodwin, 1995). If death occurs, then it is specifically referred as 'experimental mortality' (Campbell et al., 1963). Even though the subject groups may be equated by randomisation before the start of the trial, this type of confounding may render the groups of subjects subsequently incomparable. This may bias the trial results in favour of the treatment of interest (Mason et al., 1989). 9.2.5.8 Interaction with the selection process and other confounders  Up to now, confounding factors have been considered as individual threats to the internal validity of a clinical trial. The potential also exists that some interaction of the above threats may occur.  The most common interactions involve the process of  selection (Mason et ai, 1989). For example, one group of subjects may be more ill than another group since they were chosen from a different setting. Thus, it may take the test treatment longer to produce an effect and in turn mislead the investigator into thinking that the observed effects were due to the treatment rather than the interaction between the selection process and maturation (Mason et ai, 1989).  Another type of interaction that may occur is between selection and instrumentation (Mason et ai, 1989). For example, a group of selected subjects may be unable to read  L. FRANCIOSI  49  a discomfort questionnaire because they are illiterate or English is not their first language. 9.2.5.9 Maintenance  of treatment conditions over time  Another possible threat to a trial's internal validity may involve the inability to maintain the same trial conditions over time, even after randomising subjects to their respective treatments (Mason et al., 1989). Some examples that may explain the observed results include: subjects speaking to each other about their responses, subjects exchanging trial medicines, subjects having relatives bring food and beverages other than those prescribed, subjects not complying with their treatment or follow-up schedules, and physicians prescribing alternative treatments due to their ethical concerns about giving 'placebo'.  L. FRANCIOSI  9.3  50  USE APPROPRIATE STATISTICS  When the investigator has selected a design for the question, he or she should determine the appropriate statistics before conducting  the clinical  trial  (Altman,  1991a; Bennett, 1994; Bohaychuk et al., 1993; Cohen et al., 1995; Friedman et al., 1996; Hayes, 1982; Hennekens et al., 1987; Knapp et al., 1992; Mason et al., 1989; Pace, 1992; Sayre, 1994; Shapiro etal., 1983; Spilker, 1991; Spriet etal., 1985, 1992). This process involves considering the meaning of statistics, understanding their purpose, and determining which statistics or statistical tests will be used to analyse the data.  9.3.1 WHAT ARE STATISTICS? Statistics  represent the measured characteristics of a sample from which conclusions about the population are made  Statistics are the "measured characteristics of a sample" (Portney et al., 1993). They are collected and tabulated in a manner that allows the investigator to draw conclusions from them (Pace, 1992).  In particular, he or she can use them to estimate the  parameters of a population (Armitage et al., 1994; Essex-Sorlie, 1995; Hopkins et al., 1987; Lumleyefa/., 1994; Portney et al., 1993).  L. FRANCIOSI  51  When a sample is taken from a population, there are three problems that the investigator should be aware of: 1. "how to ensure the validity of the sample, i.e. that it is truly representative of the population from which it is drawn so that the results taken from the sample can be applied to the population as a whole" (Lumley etal., 1994) 2. "how to assess the reliability of the sample results, i.e. how similar would the result be from other samples drawn from the same population" (Lumley et al., 1994) 3. "how to assess whether the effects found between or within samples actually exist or not in the population from which the samples are drawn" (Lumley et al., 1994) Statistics address these problems by enabling the investigator to: 1. "be reasonably sure that a sample is likely to be representative of the population from which it is drawn" (Lumley et al., 1994) 2. "quantify the errors involved in generalizing from a sample to a population" (Lumley etal., 1994) 3. "decide whether to accept or reject any given hypothesis" (Lumley et al., 1994) There are two kinds of statistics that the investigator should consider: descriptive and inferential.  L. FRANCIOSI  52  9.3.2 CONSIDERING THE USE OF DESCRIPTIVE STATISTICS Descriptive statistics take a large amount of the measurements accumulated during the clinical trial and turns them into a small set of numbers that can be more easily understood (Goodwin, 1995). This involves organising a set of readings into a table and then creating a picture or graph called a frequency distribution (Table 3). Heart Rate (beats/minute)  Frequency  57  1  58  3  59  2  60  5  61  4  62  3  63  1  64  1  Frequency as asterisks (frequency or s a m p l e distribution)  **** *  Table 3. A frequency distribution (asterisks) of heart rates obtained from a sample of 20 subjects. This sample distribution is an estimate of the population distribution or 'standard normal curve' (Figure 2). From such a distribution, the measures of the central tendency and the variability of the data can be calculated (Altman, 1995; Goodwin, 1995; Knapp et al., 1992).  L. F R A N C I O S I  53  Figure 2. The standard normal curve.  9.3.2.1 Measures of Central Tendency  The investigator can determine where values tend to cluster in a sample or population distribution by calculating a mean, mode or median. 9.3.2.1.1 Mean  The most commonly used measure of central tendency is the mean or average (Figure 3) (Goldstein, 1963; Lumley et al., 1994). It is found by adding the readings together and dividing by the total number of readings as shown below. THR mean = — — n where HR = heart rate and n = the number of measurements  L. FRANCIOSI  54  9.3.2.1.2 Mode The mode is the most frequent reading in a set of measurements (Figure 3). It enables the investigator to get a rough estimate of the sample data's central tendency by easily inspecting it (Goldstein, 1963; Lumley et al., 1994). Hence, in Table 3, the mode is 60 beats/minute. 9.3.2.1.3 Median The median is the measure in the exact middle of a set of measurements (Figure 3). It is used when one or two of the readings taken are very different from the rest of measurements (i.e., the sample distribution is skewed). In this situation, the calculation of mean may give a distorted view of the typical measurement (Goldstein, 1963; Lumley etal., 1994).  Positively skewed distribution of a sample  mean mode  Symmetrical distribution of a sample after its measurements underwent a log or square root transformation  mean median mode  Negatively skewed distribution of a sample  mean mode  Figure 3. The mean, mode and median of a sample distribution (untransformed and transformed).  L. FRANCIOSI  55  9.3.2.2 Measures of Variability (Dispersion)  The investigator can also determine the variability (i.e., how much the values are spread out or dispersed) in a sample distribution by calculating a variance, a standard deviation or a coefficient of variation. 9.3.2.2.1  Variance  One measure of variability is the variance, s (Goldstein, 1963; Lumley et al., 1994). It 2  represents an estimate of the population variance ( a ) or what the dispersion of heart 2  rate measurements may be around the population mean (Figure 2).  It can be  calculated from the formula: g 2  _ X(HR-HR)  2  n-1  S  where HR = measured heart rate and HR = mean heart rate 9.3.2.2.2 S t a n d a r d deviation  The most commonly used measure of variability is the standard deviation, s.  It is  calculated as the square root of the sample variance (Goldstein, 1963). For a sample of heart rates, the standard deviation is a measure of the average amount that the heart rates in the sample distribution deviated from the mean heart rate. This is also known as "summing the squares" (i.e., Z(HR- HR) ) as seen below. The reason for having a 2  square root sign in the formula is to have the variance estimate in the same measurement units as the mean heart rate.  L FRANCIOSI  56  The standard deviation usually accompanies the mean when describing a sample (Goldstein, 1963; Goodwin, 1995; Lumley et al., 1994). This is done to standardise the measurements in way that both the amount of the measurement deviates from the mean and its relative position from the mean can be determined (Figure 2).  For  example, if the calculated mean heart rate of the sample is 60.3 beats/minute, then another subject who provides a reading of 62 beats/minute can be determined to be +1.7 beats/minute above the mean (62 - 60.3 = +1.7). How extreme this value is from the mean depends how dispersed the other readings are around the mean (Lumley et al., 1997).  Since the standard deviation of the sample is 1.5 beats/minute, this  indicates that the reading of 62 beats/minute is +1.7/1.5 or +1.1 standard deviation units above the mean. This process is known as "standardizing" or "transforming to z scores" or "standard scores" (Lumley et al., 1994). This expresses the "deviations from the mean in standard units, instead of the original units" of beats/minute (Lumley et al., 1994).  "This conversion enables the position of a subject on one variable to be  compared with its position on a second variable, even though the units of the measurement for the two variables are different.  Two important properties of the  distribution of standard scores are that its mean is always zero and its standard deviation is always equal to one" (Figure 2) (Lumley et al., 1994).  If we were to measure the entire population, this concept of standardization would also apply. As illustrated in Figure 2, the area under the normal distribution would also indicate that two-thirds of all heart rates measured (68%, or 34% + 34%) in the population would fall within a single standard deviation on either side of the mean. Furthermore, 95% of all heart rates would fall within two standard deviations on either  L. FRANCIOSI  57  side of the mean. Heart rates that would fall beyond +2 standard deviations would be considered rare; they would occur only 5% of the time. However, if these rare events did occur, then they would be described as 'statistically significant'. 9.3.2.2.3 Coefficient of variation  Another measure of variability is the coefficient of variation (CV). As seen below, it can be calculated by taking the standard deviation (stdev) of the distribution and dividing it by the mean. It is usually expressed as a percentage. It is a more useful measure than standard deviation when comparing the variability of two or more sets of data with different units of measurement (Goldstein, 1963; Lang era/., 1997). cv  =**?L  mean 9.3.2.3 Correlation  Another descriptive statistic, correlation, is used to determine the existence of an association between two variables and to assess the degree by which the variables covary (Altman, 1980e; Bland etal., 1994a; Goodwin, 1995; Guyatt etal., 1995c; Lang et al., 1997). This process involves the use of a scatterplot and the determination of a statistic called the "coefficient of correlation". This coefficient may also be called the Pearson's r or Spearman's rho depending on the scale that the data were measured on (Appendix 4) (Lumley et al., 1994). 9.3.2.4 Regression Analysis  If the investigator wants to estimate the numerical relationship between variables, i.e., make a prediction, then he or she may conduct a regression analysis of the linear or  L. FRANCIOSI  58  non-linear form (Appendix 4) (Altman, 1980e; Bland et al., 1994a; Goodwin, 1995; Guyatt et al., 1995c; Lang et al., 1997).  This analysis is used for questions that  evaluate any cause and effect relationship between an independent x variable and a dependent y variable (Lang et al., 1997). 9.3.2.4.1  Linear regression  Linear regression involves creating a scatterplot with the independent variable as the xaxis and dependent variable as the y-axis (Gad et al., 1986; Goodwin, 1995; Lang et al., 1997). If the investigator assumes that the relationship between the variables is linear, then the points on the scatterplot should be in a straight line. This line is also called the 'regression line'. The investigator can predict a y value from the equation of the straight line, y = mx + b (where m is the slope and b is the y-intercept). 9.3.2.4.2 Non-linear regression  When the numerical relationship between variables is non-linear, the investigator can predict y values on the basis of a non-linear regression analysis (Bowen et al., 1995; Gad etal., 1986; Johnson, 1992; Lang etal., 1997; Motulsky etal., 1987). An example of non-linear regression is a plot of drug doses in log units on the x-axis and the corresponding measured responses on the y-axis. The investigator assumes that the relationship between the two variables appears as a sigmoid curve (Bowen et al., 1995). Thus, he or she can use the logistic equation of y = x /(1+x) for a sigmoid curve n  to fit the data and then predict response values.  n  L FRANCIOSI  9.3.3  59  CONSIDERING INFERENTIAL STATISTICS  When the investigator has selected a measure of central tendency and variability, for example, a mean and standard deviation, he or she can use of inferential statistics to determine if the relationship or difference to be detected happened by chance. This involves formulating and stating a null and alternative hypothesis, specifying an alpha (a) and beta (P) level, considering statistical power, selecting an appropriate statistical test for the primary question, considering the use of a confidence interval, distinguishing between clinical and statistical significance. Other concerns such as considering issues related to chance, selecting statistical methods to deal with missing values and outliers, and considering a possible interim analysis should be thought about while using inferential statistics. 9.3.3.1 Formulating and stating the null and alternative hypothesis  As mentioned in section 9.1.2.2, the formulation and statement of a scientific hypothesis enables the investigator to perform 'hypothesis testing' (Christensen, 1991; Guyatt et al., 1995a; McConway, 1994). He or she makes the assumption that there is no difference or relationship between the variables being investigated (Goodwin, 1995; Knapp et al., 1992). This assumption is called the null hypothesis (null = nothing); it is symbolised as H . The hypothesis that is being investigated in the clinical trial is the 0  scientific hypothesis, or the alternative hypothesis that is denoted by HA. Each hypothesis represents a possible population from which a sample is drawn (Figure 4). There are four possible outcomes from testing a null hypothesis, H : 0  L FRANCIOSI  60  1. H is rejected (i.e., the sample was drawn from the H Population) 0  A  2. H is not rejected (i.e., the sample was drawn from the H Population) 0  0  3. H is falsely rejected (i.e., the sample was really drawn from the H Population; this occurrence happened by chance and is described either as an alpha (a) error or Type I error) 0  0  4. H is not rejected when it should be rejected (i.e., the sample was really drawn from H Population; this occurrence happening by chance is described either as a beta (P) error or Type II error). 0  A  Null Hypothesis H  Alternative Hypothesis, H  5% of H Population 0  a is 1.65 Standard Deviations from n (i.e., this is a one-tailed test)  Possible Conclusions:  0  Do not reject H ; conclude sample was drawn from H Population 0  o  Reject H ; conclude sample was drawn from H. Population 0  Figure 4. The populations for the null and alternative hypotheses.  L. FRANCIOSI  61  9.3.3.2 Specifying an alpha and beta level  To minimise the chance of making alpha and beta errors, the investigator sets alpha and beta levels of statistical significance (Pace, 1992). In this way, the investigator can be assured that H can only be rejected with some degree of confidence. 0  By convention, alpha is set at 0.05 (a=0.05), but it can be set at other levels such as a=0.01  (Goodwin, 1995; Knapp et al., 1992).  However, a small a will increase the  chances that a beta ((3) error will occur. If H is rejected when alpha equals 0.05, it 0  means that the probability is 1 in 20 that the effect or outcome of interest is the result of chance. In other words, if the trial were repeated 2 0 times, there is a high probability that one of the trials may falsely reject the null hypothesis.  The reason for choosing 0.05 relates to the characteristics of the normal curve (Goodwin, 1995).  In the distribution of heart rates, the probability is low, 5 % or less,  that a given heart rate will be more than two standard deviations from the mean. Such an event is rare. Similarly, when two samples are compared, a question may be: what is the probability that an observed difference between the sample means is a real difference? If the probability is 5% or 0.05 for such an event, then a probability that is less than or equal to alpha means rejecting H and concluding that the difference 0  observed is a statistically significant. In other words, one of the samples was drawn from the H population. A  L FRANCIOSI  62  By convention, beta is set at 0.20 (p=0.20), but it can be set at other levels such as p=0.10 (Goodwin, 1995; Knappefa/., 1992). A beta of 0.20 means that the investigator will allow up to 20% (or lower) chance probability of committing a beta error (Goodwin, 1995; Lerman, 1996; Pace, 1992). In the process, he or she has also decided that they want 80% statistical 'power' ((1 - (3)x100%) to detect any difference that may exist between two samples (Figure 4). 9.3.3.3 Considering statistical power  In order for the investigator to maximise this statistical power, he or she can raise a , reduce any variability amongst and between the trial groups, increase the sample size, or ensure that the difference between the groups is large enough that it can be detected (Altman, 1980c; Altman, 1991a; Pace, 1992). Although all these options should be considered, in many cases, the practical solution is increasing the sample size (Lerman, 1996).  Formulas for the sample size calculation are available in most statistics  textbooks or literature reviews (Friedman etal., 1996; Lerman, 1996; Shuster, 1990).  9.3.3.3.1  Conducting a pilot study and/or a review of the literature to determine the possible  variance  The investigator usually undertakes a pilot study or literature search to determine the possible variance (standard deviation) in the clinical trial.  This is to predict an  appropriate sample size (Friedman etal., 1996; Lerman, 1996). 9.3.3.4 Choosing a statistical test appropriate for the primary question  As mentioned previously, hypothesis testing involves rejecting or not rejecting the null hypothesis. The steps are: stating the null and alternative hypotheses, setting an alpha  L. FRANCIOSI  63  and beta level, selecting a statistical test for the primary question, checking to see if the measured variables violate the test's assumptions, calculating a test statistic, finding the critical value from tables of the test's theoretical sampling distribution, and then comparing the test statistic with the critical value. If the test statistic exceeds the critical value, the investigator can reject the null hypothesis (Goodwin, 1995; Knapp et al., 1992) . It is important that the investigator select an statistical test that is appropriate not only for the question, but also for the type of primary variable that is being measured (Lumley et al., 1994; Pace, 1992). Thus, the investigator should determine whether to use a parametric test, a non-parametric test, or both on this basis (Appendix 4). 9.3.3.4.1 Parametric statistical tests Parametric statistical tests are a class of tests that are used to analyse data that conform to a known distribution (Lang et ai, 1997). 9.3.3.4.1.1 Types of tests and their assumptions  There are variety of parametric tests available (Appendix 4). Some of the more common ones are correlation, regression, t-tests, and ANOVA (Altman, 1991a; Altman, 1996; Lumley et ai, 1994). Each test has unique restrictions on how they can be used, however, they all assume that random samples are taken, that the samples came from normally distributed populations, and that variances of populations are equal (Gad et ai, 1986). The majority of these tests are said to be 'robust' because they work even when the data do not entirely fulfil all basic assumptions generally considered to be relevant to them (Loos, 1995).  L. FRANCIOSI  9.3.3.4.1.2  Assessing  64  the normality of the data  If the data is known to be normally distributed and has outlying values, the investigator should assess its normality (Altman, 1980d; Altman, 1991a; Armitage et al., 1994). Some examples of normality tests are:  Normal plots, Kolmogorov-Smirnov test,  Lilliefors modified Kolmogorov-Smirnov test, and Shapiro-Wilk test (Altman, 1991a; Armitage et al., 1994). Many of these tests are too tedious to do by hand and would normally be done using statistics programs like SAS, JMP-IN, PRISM or NCSS (Gad et al., 1986). 9.3.3.4.1.3  Using transformations  to normalize  the data  When the data has been assessed as non-normal, there a number of transformations that the investigator can use to normalise the data (Altman, 1980d; Armitage et al., 1994; Bland, 1995a; Bland et al., 1996abcd).  Two examples are logarithmic and  square root transformations. When data are transformed, the investigator should again test for normality (Armitage et al., 1994). 9.3.3.4.2 Non-parametric tests  Nonparametric tests are used to analyse data that do not meet the meet the assumptions of parametric tests. 9.3.3.4.2.1  Types of tests and their basic  assumptions  There are number of commonly used nonparametric tests (Appendix 4). Some of these are the Spearman Rank Correlation, Mann-Whitney U test, Wilcoxon Matched Pairs Signed Ranks Test, Friedman Two-Way Analysis (Lumley et al., 1994). These test are easier to learn, quicker to do, and 95% as powerful as parametric statistics (Loos,  L. FRANCIOSI  65  1995). These tests do not assume any distribution, in particular, the normal distribution (Gad etal., 1986; Sprent, 1989). 9.3.3.4.3 C o n s i d e r i n g o n e tailed versus t w o tailed statistical test  The investigator has the option to select which tail of the normal distribution he or she is interested in (Bland et al., 1994c; Lumley et al., 1992; Knapp et al., 1992; Pace, 1992). If alpha is set at 0.05, this may mean that he or she has accepted that the chance that the test will produce an alpha error 5% of time at one end of normal distribution (onetailed) or 2.5% of time on each end (i.e., two-tailed). The investigator may choose to perform a one tailed test if he or she is confident that the relationship or difference is in one direction (Bland etal., 1994c; Lumley etal., 1992; Knapp etal., 1992; Pace, 1992). Words that may describe this difference, as stated in the question and hypothesis are "more than", "less than", "reduce", "enhance", "increase", or "decrease" (Portney et al., 1993). In a one-tailed test, an alpha of 5% may correspond to either -1.65 or +1.65 standard deviations from the population mean, u, , depending on which end of the 0  distribution that the investigator is interested in (Figure 5).  L. FRANCIOSI  / \  a = 0.05  66  \ J  'M  a corresponds to +1.65 standard deviations from u 0  Figure 5. One-tailed test.  If the investigator is not sure in which direction that the difference is going in, or if he or she wishes to show no difference between groups (equivalency), then the investigator can be conservative and select a two-tailed test (Figure 6) (Bland et al., 1994c; Lumley et al., 1992; Knapp et al., 1992; Pace, 1992). In this test, an alpha of 5% may correspond to +1.96 standard deviations from the population mean, p. (since 95% of 0  measurements fall within ±1.96 standard deviations).  L. FRANCIOSI  67  Figure 6. Two-tailed test.  9.3.3.5 Considering  the use of a confidence  interval  The investigator should consider the use of a confidence interval (Altman, 1980f; Berry, 1986, 1988; Bland etal., 1996d; Borenstein, 1994; Guyatt etal, 1995b; Simon, 1986;). The use of such an interval enables the investigator to define a range of values within which the true population parameter is likely to fall with 95% or greater confidence. This confidence interval is created from the estimate (e.g. sample mean), the test statistic (e.g. t test = 1.96 for 95% confidence) and the sample-to-sample variation in form of a standard error of the estimate (standard deviation divided by the square root of the sample size) (Altman, 1991a).  L. FRANCIOSI  68  9.3.3.6 Considering the meanings of the statistical and clinical significance  The meanings of 'statistical significance' and the 'clinical significance' should be considered by the investigator prior to conducting the clinical trial (Altman, 1980g; Armitage et al., 1994, Berry, 1986, 1988; Hennekens et al., 1987; Lang et al., 1997). As already mentioned, statistical significance means the rejection of null hypothesis based on a level of alpha (Lang et al., 1997). On the other hand, clinical significance means the "clinically significant difference" or the difference in treatment outcomes that would make clinicians or policymakers adopt or promote the better treatment (Detsky, 1986). The meanings of each should be considered because: a statistically significant difference may be clinically important; the difference may be statistically significant but unclear as to whether it is clinically significant; the difference may be statistically significant, but too small to be clinically important, the difference may not be statistically significant but may be large enough to be clinically important; or another possibility is that the difference may not be both statistically and clinically significant (Armitage et al., 1994). 9.3.3.7 Consider the issues related to chance  A primary concern for the investigator should be: "Will I use the statistics properly such that I will draw the appropriate conclusions from them?" (Goodwin, 1995).  Such a  consideration represents another form of validity known as statistical conclusion validity (Goodwin, 1995; Guyatt et al., 1995a; Mason et al., 1989). This validity is not only concerned with whether or not the trial's sample size and power are both proper, but also with the possibility that the investigator might go 'fishing' for a significant result by repeated statistical testing. When is the case, there is a chance that the investigator  L. FRANCIOSI  69  will commit a Type I error (Bland etal., 1995b; Guyatt et al., 1995a; Mason etal., 1989). Another issue related to chance is random variation. If the measures the investigator chose are unreliable, random variation may occur which would increase the likelihood of missing a significant result. In other words, there is chance that the investigator will commit a Type II error (Mason era/., 1989). 9.3.3.8 Statistical methods for dealing with missing  values  During the conduct of a clinical trial, there may be instances where subjects dropout or are withdrawn from the trial before the relevant response is measured (Altman, 1980d; Altman, 1991a; Armitage et al., 1994). In the best possible trial, these subjects are not omitted from the analysis; the investigator must include them in order to avoid biasing the results. This is also known as "intent(ion)-to-treat" approach where all randomised subjects are analysed. An investigator should consider statistical methods that deal with 'missing values' (Altman, 1991a; Armitage et al., 1994; Friedman etal., 1996; Little etal., 1987).  For some experimental designs such as the Latin Square, there are mathematical equations that can be used to insert values (Armitage et al., 1994; Little et al., 1987). Another method is the Last Value Carried Forward (LVCF) where the investigator brings forward the last available response on each subject (Armitage et al., 1994).  Other  approaches include arbitrarily inserting a poor measurement or randomly choosing a value from the possible range of measurements already collected (Armitage et al., 1994; Crepeau, 1983; Little et al., 1987). All these methods should also be considered  L FRANCIOSI  70  by the investigator when a small number of values are accidentally lost or accidentally forgotten to be measured (Armitage et al., 1994). 9.3.3.9 Statistical methods for dealing with outliers  The results of a clinical trial might be positively or negatively skewed because of outlying values (Figure 3). When this is the case, there a number of methods for dealing with these 'outliers'.  One method is to transform the data as previously  discussed. Another way is to use Chauvenet's criterion which assesses whether or not the outlying values happened by chance (i.e. are they part of the same population from which the sample of values was drawn?) (Barnett et al., 1978; Gad et al., 1986). If they are not, the values are dropped from the data set and the results of the test reported (Lang et al., 1997). Another method is to simply analyse and present the data with and without the outliers (Friedman etal., 1996). 9.3.3.10 Considering the use of an interim analysis  The investigator should also consider the possibility of performing an interim analysis on the data (Lang etal., 1997; Spilker, 1991; TFWGAESC, 1994). This data analysis is performed either informally or formally before the final statistical analysis for several reasons: A) Ethical: stopping a clinical trial so as to not withhold "a better treatment from patients longer than necessary", B) Financial: stopping a clinical trial to avoid spending more money on it if it "cannot demonstrate desired differences", and C) Practical: being "assured that the clinical trial is progressing as planned" (Spilker, 1991).  L. FRANCIOSI  71  9.4 CONSIDER THE ETHICAL ISSUES  Before, during and after defining the question and selecting the appropriate design and statistics, the investigator should consider the ethical issues involved (Bennett, 1994; Bohaychuk et al., 1993; Cohen et al., 1995; Friedman et al., 1996; Lawrence, 1991; Sayre, 1994; Shapiro et al., 1983; Spilker, 1991; Spriet et al., 1985, 1992).  These  issues are concerned with: the potential risks and benefits of the treatment, the appropriateness of the design and statistics, the investigator's conflict of interest regarding the treatment being investigated, the termination of a trial when the optimal treatment is found, the consent and confidentiality of subjects, the importance of Declaration of Helsinki, and the opinion of an ethics committee.  9.4.1 CONSIDER THE POTENTIAL RISKS AND BENEFITS FOR THE SUBJECT INVOLVED To properly conduct a clinical trial, an investigator must consider the potential risks and benefits for the subjects involved (Lawrence, 1991; Spriet et al., 1985).  A risk is  defined as "the probability that a given patient will experience a deleterious reaction to the specific treatment under study or, more generally, that something bad will happen." (Spilker, 1991). The investigator must assess whether or not the proposed treatment is safe or unlikely to do harm to the subjects (Hill, 1963). From the perspective of the subject, the term 'benefit' refers to an 'improvement' in a treatment's efficacy and subject's quality of life, which can be either predictable (almost definite), probable (>50%), possible (<50%), or unlikely (<5%) (Spilker, 1991).  L FRANCIOSI  72  Most often, this consideration involves the investigator weighing the advantages against the disadvantages of the proposed treatment (Pochin, 1981). In particular, he or she performs a risk-benefit analysis that addresses the question: "What is the value of giving a specific treatment to a specific patient at a specific time under a defined set of conditions?" (Spilker, 1991). 9.4.2 CONSIDERING STATISTICS  THE APPROPRIATENESS  OF  THE  DESIGN AND  During or after the selection of a trial design and statistics, the investigator must think about whether or not they are appropriate from ethical standpoint (Altman, 1980a; Altman, 1980b; Altman, 1991a; Friedman etal., 1996; Hill, 1963; Spilker, 1991). This involves considering a number of questions: •  Is the type of subject that is to be entered into the trial appropriate?  •  Is the selected number of subjects sufficient to detect an important treatment difference?  •  Would all subjects agree to being randomly allocated to different treatments?  •  Would each subject have an acceptable chance of receiving the new treatment?  •  Would all subjects agree to being given a placebo treatment or no treatment as a control?  •  Would it be proper for the investigator not to know what treatment is being administered to a subject?  •  Does the trial have enough statistical power to detect an important treatment difference?  •  Are the statistical tests and methods to be used in trial appropriate enough that statistical expertise is not required?  The answers to many of these questions may not be black and white, but they need to be addressed in order to avoid:  L. FRANCIOSI  1. "misuse of patients by exposing them to unjustified inconvenience" (Altman, 1991a)  73  risk and  2. "misuse of resources, including the researchers' time, which could be better employed on more valuable activities" (Altman, 1991a) 3. "the consequences of publishing misleading results, which may include the carrying out of unnecessary further work" (Altman, 1991a)  9.4.3 CONSIDERING ANY "CONFLICT OF TREATMENT(S) BEING INVESTIGATED  INTEREST"  REGARDING THE  In a clinical trial, the investigator should consider whether or not there is any conflict of interest "regarding the treatment being investigated (CSA/CEJA, 1990; Lawrence, 1991; Rothman, 1993; Thompson, 1993). A conflict of interest is "a set of conditions in which professional judgement concerning a primary interest (such as a patient's welfare or the validity of research) tends to be unduly influenced by a secondary interest" (Thompson, 1993).  A secondary interest may be a financial gain, a preference for family and  friends, or a desire for prestige and power (CSA/CEJA, 1990; Rothman, 1993; Thompson, 1993). A conflict of interest itself does not mean that the investigator did something wrong; "it merely refers to a setting in which factors exist that might influence one's conduct" (Rothman, 1993). This influence may lead to the increased possibility that work conducted by the investigator is biased (Rothman, 1993). When investigators "do not take reasonable precautions to avoid situations of conflict or do not observe the rules regulating such conflicts, they have acted unethically" (Thompson, 1993). 9.4.4 CONSIDERING THE EARLY TERMINATION OF THE CLINICAL TRIAL The investigator should consider the possibility of discontinuing the clinical trial early for a number of reasons ((Bankowski, 1992; Friedman et al., 1996; Lawrence, 1991;  L. FRANCIOSI  Spilker, 1991; TFWGAESC, 1994).  74  Subjects may experience (serious) adverse  reactions during the trial that it may create an unacceptable risk for other subjects to continue (Spilker, 1991; TFWGAESC, 1994).  There may not be enough subjects  enrolled in the trial (Friedman et al., 1996; Spilker, 1991). There may be failures or changes in the way data is collected and stored that render the trial unacceptable (Friedman et al., 1996). It is possible that the result of an interim analysis may identify an optimal treatment, which raises ethical concerns for those subjects who are not receiving it (Lawrence, 1991; Spilker, 1991). "If early termination for ethical issues is considered possible at the outset of a clinical trial, then criteria should be established to define conditions under which the trial would be terminated early" (Spilker, 1991). 9.4.5 CONSIDERING ISSUES RELATED TO SUBJECT CONSENT It is important during the planning of the clinical trial to also consider the issues related to subject consent (Bankowski, 1992; Friedman era/., 1996; Hill, 1963; Spilker, 1991). A major issue for the investigator to consider is: Would a subject consent to be in his or her trial if it is designed in this manner? Such a consideration may lead to the rejection and acceptance of his or her proposed design. If subjects agree to participate in the clinical trial, then the investigator must get verbal and/or signed consent. Sometimes, subjects may leave it up to the investigator to decide whether or not they should participate in the trial (Lidz et al., 1983). It is important that the investigator be aware of this vulnerability when obtaining consent.  If a subject no longer wants to participate,  then he or she should have the right to withdraw at anytime without any change to their medical care (Bankowski, 1992).  L. FRANCIOSI  75  9.4.6 CONSIDER THE ISSUE OF SUBJECT CONFIDENTIALITY The possibility exists that an investigator may display slides or photos of his or her subjects to the public without their permission (Spriet et al., 1992). Clinical trial records may contain information related to a subject's state of health which he or she may not want to have disclosed.  Thus, these issues of subject confidentiality need to be  considered by the investigator (Bankowski, 1992). This confidentiality of the subject also needs to be weighed against maintaining a subjects identity in cases where longterm follow-ups are needed or in verifying the authenticity of data collected. Thus, subject initials, hospital number, or social insurance number may be sufficient to trace subjects when they or documents related to their participation in the clinical trial are necessary (Spilker, 1991). 9.4.7 REVIEWING AND REFERENCING THE DECLARATION OF HELSINKI FOR THE PROTECTION OF TRIAL SUBJECTS In considering consent and confidentiality of subjects, there is a formal document that investigators should review and reference if they are going to conduct research on subjects (Bankowski, 1992; Vollmann et al., 1996). The Declaration of Helsinki, which was created in 1964 and then revised in the 1970s and 1980s, forms the accepted basis for clinical trial ethics (Appendix 5).  Its basic principle is that if a subject is  randomised to a treatment or if a treatment is given to a subject to advance medical knowledge, then informed consent must be obtained by the investigator (Bankowski, 1992).  The document considers and discusses: adherence to scientific principles, the  formulation of a protocol, proper qualifications for conducting a trial, weighing the risks against the benefits, obtaining informed subject consent, giving subjects the right to  L. FRANCIOSI  76  withdrawal from the trial at anytime; maintaining subject confidentiality, discontinuing the trial if the hazards outweigh the potential benefits, and publishing accurate results (Appendix 5). 9.4.8 SEEKING THE OPINION OF AN ETHICS COMMITTEE Another basic principle of the Declaration of Helsinki is that the investigator seek the opinion of an institutional review board or ethics committee (Bankowski, 1992; Bennett, 1994).  This board or committee, which is commonly affiliated with the hospital or  university where the trial is to be conducted, is responsible for evaluating the protocol in terms of the potential risks and benefits that subjects will have to endure and forjudging whether or not there is an adequate informed consent form (Pace, 1992; Spriet et al., 1992). An ethics committee normally does not consider issues related to experimental design or statistics (Altman, 1991a; Johnson et al., 1977). However, according to Dr. Morley Sutter, a member of the UBC Ethics Committee, no risk is worth taking if there is no question or significance in what an investigator is planning to do (Personal Communication, M. C. Sutter).  L. FRANCIOSI  77  9.5 HAVE STANDARD OPERATING PROCEDURES IN P L A C E  A good clinical trial practise is to have standard operating procedures (SOPs) in place to implement the clinical trial (Bennett, 1994; Bohaychuk et al., 1993; Cohen et al., 1995; Sayre, 1994; Spriet etal., 1992).  9.5.1 WHAT ARE STANDARD OPERATING PROCEDURES (SOPs)? SOPs are a set of detailed instructions which indicate how a clinical trial is to be implemented from start to finish. They are said to be "standard" because they can be applied to many clinical trials  These procedures are a set of detailed instructions which indicate how a clinical trial is to be implemented from start to finish (Bennett, 1994; Cohen et al., 1995; Sayre, 1994). They are said to be "standard" because they can be applied to many clinical trials (Cohen et al., 1995). When written and complied into a manual, SOPs can then be used as a training aide for new clinical trial staff as well as a reference to assure both accountability and consistency in every aspect of a trial's conduct (Sayre, 1994). In this case, SOPs encompass the important aspects of 'quality assurance' (i.e. a planned series of actions necessary to satisfy the requirements of regulatory authorities like the US Food and Drug Administration and the Canadian Health Protection Branch for registering new treatments) (Bennett, 1994; Sayre, 1994; Spriet etal., 1992). They also address 'quality control' by monitoring the conduct of the trial to detect and eliminate any unsatisfactory performance (Bennett, 1994; Spriet et al., 1992). A manual of SOPs contains an up-to-date list of procedures for: delegating responsibilities for those parties  L. FRANCIOSI  78  involved in the clinical trial; creating, finalising, and archiving principle trial documents; obtaining ethics approval from an institutional review board; monitoring trial events; defining and collecting adverse events; preparing test medication; managing clinical and laboratory data; auditing and inspecting the trial; and deciding miscellaneous aspects of the clinical trial (Bohaychuk et al., 1993; Sayre, 1994; Spielman, 1991; Spriet etal., 1992). 9.5.2 PROCEDURES FOR DELEGATING RESPONSIBILITIES The first section of a SOP manual contains a clear statement of why the clinical trial is being conducted (Spriet et al., 1992). When the trial is being implemented, there are three main parties involved: the sponsor, the investigator, and the clinical monitor (Morice et al., 1991). The sponsor delegates responsibilities to each party in advance; the first SOPs of this section instruct and clarify what their roles are at any time during the trial's conduct (Morice et al., 1991; Spielman, 1992; Spilker, 1991). Some of these SOPs/responsibilities will vary from trial-to-trial, but those most commonly encountered are described in Table 4. At the end of this section, the sponsor usually states either as a standard  operating procedure, a policy or a responsibility how the principal  investigator will be chosen for the clinical trial (Spriet et al., 1992). 9.5.3 PROCEDURES FOR DEVELOPING AND STORING PRINCIPLE CLINICAL TRIAL DOCUMENTS The next section of an SOP manual contains procedures which instruct the sponsor on how to create, finalise, distribute and maintain key documents for the implementation of the clinical trial (Spielman, 1992; Spriet etal., 1992).  L. FRANCIOSI  Key Participants in the Implementation of a Clinical Trial  79  Roles and Responsibilities selects the investigator, and insure that he or she is appropriately qualified, has sufficient experience, and has time to devote to the clinical trial  SPONSOR:  ensures that the trial can take place in the location designated  An organisation or an individual that initiates, manages, and/or finances the clinical trial; the individual may be the principal investigator.  ensures that an adequate population of subjects is available fore the investigator to complete the trial within a realistic time. draws up the protocol for the trial, which must be designed so that it is sufficiently sensitive to measure the putative effects of the drug under investigation. signs a formal agreement with sponsor  PRINCIPAL INVESTIGATOR:  ensures that ethics committee approval is obtained  An individual, usually a physician, who is qualified to assume the responsibility for conducting the clinical trial.  ensures that each subject entered into the trial has properly completed a consent form ensures proper collection and entry of data into the case report forms agrees on a cost of the trial at the onset of the trial oversees the progress of the trial and ensure adherence to the protocol  CLINICAL MONITOR: An individual designated by the sponsor to organise, initiate and review the progress of the clinical trial.  performs pre-trial visits to all parties engaged in the conduct of the trial, pre-determines laboratory procedures performs random checks of reference samples during assays visits the investigator during the trial to check raw data ensures correct filing out of case report forms  Table 4 . Roles and Responsibilities of key participants in the implementation of a clinical trial  L. FRANCIOSI  80  The first SOPs concern the development of an investigator's brochure by the sponsor and then its subsequent distribution to the principal investigator and other clinical staff in the trial (Cohen et al., 1995). This document contains information that is required "prior to the onset of a clinical trial including chemical and pharmaceutical data, toxicological, pharmacokinetic and pharmacodynamic data in animals and the results of earlier clinical trials" (Bennett, 1994).  The next principal document that is developed and given to an investigator according to written SOPs is a protocol (Cohen etal., 1995; Iberef al., 1987; Meinert, 1986; Pocock, 1983; Spilker, 1984; 1991). It is "the written mechanism that describes how the clinical trial design will be implemented" (Spilker, 1991). Generally, it contains the following headings:  Background and Objectives, Clinical Trial Design, Informed Consent,  Subject Inclusion and Exclusion Criteria, Baseline Evaluation, Study Treatment Details and Schedule(s), Subject Outcome Evaluation, Subject Follow-up Evaluation, Statistical Analysis, Interim Analysis, Adverse Event Collection and Reporting, Case Report Forms and  Data  Handling, Trial  Monitoring,  Protocol Deviations, and Administrative  Responsibilities (Pocock, 1983; Spilker, 1984). An important SOP stated in this section is that all amendments made to the protocol be signed, dated, and recorded on the final protocol (Spielman, 1992; Spriet etal., 1992).  The next SOPs are concerned with the establishment and distribution of case report forms (CRFs) in the clinical trial (Cohen et al., 1995; Spilker et al., 1991). A CRF represents "a record of the data and other information on each subject in the trial as  L. FRANCIOSI  81  defined by the protocol" (Bennett, 1994). SOPs instruct that this data be recorded in an accurate and verifiable manner (Spielman, 1992; Spriet etal., 1992).  Another set of SOPs are also written for the creation of the clinical trial's final report. SOPs usually instruct that the report be written by the sponsor according to procedures which ensure that it is complete and comprehensive (Spielman, 1992; Spilker, 1991; Spriet et al., 1992). This documents includes "a description of experimental (including statistical) methods and materials, a presentation and evaluation of the results, statistical analyses, and a critical statistical and clinical appraisal" (Bennett, 1994).  The final SOPs of this section concern where, how and how long all trial documents are to be archived following the completion of the trial; common SOPs require that documents be stored at the clinical trial site in a locked cabinet for 15 years (Cohen et al., 1995; Spilker, 1991). 9.5.4 PROCEDURES FOR PROTECTING TRIAL SUBJECTS In this section of the manual, there is a standard operating procedure which instructs the principal investigator to submit the clinical trial protocol to an ethics committee or institutional review board for approval (Bennett, 1994; Bankowski et al., 1993; Cohen et al., 1995; Spielman, 1992; Spilker, 1991; Spriet etal., 1992). This committee or board, which is commonly affiliated with the hospital or university where the trial is to be conducted, is responsible for reviewing and evaluating the protocol for safety in terms of the potential risks and benefits that subjects will to endure (Spriet ef al., 1992). Other SOPs describe how written informed consent is to be obtained from all subjects and  L FRANCIOSI  82  how information pertaining to their involvement is to be kept confidential (Bankowski et al., 1993; Spilker, 1991). Also in this section are SOPs which instruct the sponsor to register any new medication with government regulatory authorities as well as to submit a protocol to them for their approval of the trial's conduct in the country (Spilker, 1991; Spriet etal., 1992). 9.5.5 PROCEDURES FOR MONITORING TRIAL EVENTS Standard operating procedures in this section describe: how data is to be collected within a set period of time by the principal investigator; and how and when it is to be checked by the clinical monitor to ensure that it is complete, reliable, and compliant with what was stated in the protocol (Cohen et al., 1995; FDA, 1988; Spielman, 1992; Spilker, 1991; Spriet etal., 1992).  Typically, SOPs instruct the clinical monitor to make an initial contact either by an informal phone conversation or short meeting with the principal investigator (Spielman, 1992). According SOPs, the monitor discusses the test medication and the rationale for conducting the trial and obtains information concerning the institution's ethics committee (Cohen etal., 1995; FDA, 1988; Spielman, 1992).  A formal pre-trial visit is then conducted to ensure that the trial site is adequate, and that the clinical trial design is presented and developed with the principal investigator and clinical trial site staff (Cohen et al., 1995; FDA, 1988; Spielman, 1992).  Some  SOPs at this visit include reviewing the investigator's responsibilities, obtaining up-todate curriculum vitas from the investigator and staff, reviewing a proposed protocol and  L. FRANCIOSI  83  sample case report forms, performing an on-site examination of equipment and facilities, and discussing issues concerning regulatory compliance and the preliminary trial budget with the investigator (Spielman, 1992).  Then, the clinical monitor is required to conduct a study initiation visit to finalise and to instruct the investigator and staff on the matters of the previous visit and to ensure that ethics and regulatory approval for the clinical trial has been obtained (FDA, 1988; Spielman, 1992).  The monitor then performs routine scheduled visits according to  SOPs to ensure subject safety, data quality, and compliance with the protocol and the regulatory bodies (Cohen etal., 1995; FDA, 1988; Spielman, 1992).  Then, standard operating procedures require that the monitor perform a final visit or trial close-out (Cohen etal., 1995; FDA, 1988; Spielman, 1992; Spilker, 1991). SOPs at this visit include:  retrieving all used and unused trial medication (drug accountability),  reviewing any protocol deviations and recording the reasons why, and ensuring that all trial data are complete and filed appropriately. 9.5.6 PROCEDURES FOR DEFINING, ADVERSE EVENTS (AEs)  COLLECTING  AND REPORTING  OF  Procedures are also written for defining, collecting and reporting any adverse events that occur during the trial's conduct (Cohen et al., 1995; Friedman et al., 1996; Spielman, 1992; Spilker, 1991; Spriet et al., 1993). Commonly, adverse events are any undesirable experiences occurring to a subject, during a clinical trial, whether or not they are causally related to the test medication being investigated (Bennett, 1994). If the subject is 'rechallenged' with the test medication (according to an SOP) and the  L. FRANCIOSI  84  experience is determined to be related, then there are SOPs which define the event as an adverse reaction (Bennett, 1994; Spilker, 1991). If the event is fatal, life threatening, disabling, or requires prolonged hospitalisation, then it is defined as a serious adverse event (SAE) (Bennett, 1994). All these events are then reported, according to a SOP, to the ethics committee as well as to the regulatory authorities so that physicians make well informed decisions if the test medication goes on the market (Cohen et al., 1995; Spielman, 1992; Spilker, 1991; Spriet etal., 1992).  9.5.7 PROCEDURES FOR DEVELOPING AND HANDLING TRIAL TREATMENTS In many countries, the manufacturing of test medication is subjected to regulations on Good Manufacturing Practices (GMPs) (Willig et al., 1992). These practices are code of practice that manufactures must follow to ensure that the medication is of high purity and properly formulated for human-use (Bennett, 1994). SOPs are written to ensure that GMPs have been followed, including the packaging and labelling of the medication according to the needs of the clinical trial, i.e. for randomisation and double blinding purposes (Spriet et al., 1992). Other SOPs in this section ensure that a expiry date is determined and clearly stated on all test medication used during the trial (Willig et al., 1992). There are also procedures which account for all medication used during the trial; this is to ensure that none of the medication is misplaced and/or mistakenly used on subjects in other trials or with other illnesses at the trial site. This was also referred to previously as 'drug accountability' (Spielman, 1992).  L. FRANCIOSI  85  9.5.8 PROCEDURES FOR DATA MANAGEMENT When the clinical trial is ongoing or near completion, the data that is accumulated is very extensive and most often difficult to analyse (Spielman, 1992). Thus, to make the process easier, there are standard operating procedures outlined in this section of the manual which instruct how data is to be translated from the CRFs into a format that can be checked by a computer (Spielman, 1992; Spilker, 1991). Normally, these SOPs require that the data in the CRFs be coded, i.e. giving a number or a short descriptive word(s) for each piece of data (Spielman, 1992). Then, these coded data are entered twice into the computer, by two different data entry operators. The two datasets are then compared by the computer and any discrepancies requires that one data operator recheck the CRF and make the necessary changes. The most common discrepancies that SOPs address are those associated with missing data and the logical errors that happen when laboratory values are outside of normal range (Spielman, 1992; Spriet et al., 1992). 9.5.9 PROCEDURES FOR THE AUDIT AND INSPECTION OF A CLINICAL TRIAL Standard operating procedures in this section are written to verify that all SOPs developed for the clinical have been successfully followed (Spriet et al., 1992). This usually entails that an independent 'audit' and 'inspection' be performed by the sponsor and regulatory authorities, respectively. This audit and/or inspection mainly addresses quality assurance, i.e. the investigator and sponsor are checked to see if their responsibilities have been fulfilled, and the trial documents, computer and laboratory systems are also verified for performance and accuracy (Spilker, 1991; Spriet et al., 1992).  L. FRANCIOSI  9.5.10  86  PROCEDURES FOR THE COLLECTION AND MANAGEMENT OF CLINICAL LABORATORY DATA  Laboratory tests are commonly used in clinical trials to either assess the efficacy of a test medication or to evaluate its safety and tolerability (Cohen et al., 1995; Spielman, 1992; Spilker, 1991). Another use is to screen subjects for possible entry into the trial (Spriet et al., 1992).  Standard operating procedures also are in place to tell  laboratory/clinical staff:  how these tests are to be calibrated and done step-by-step;  how much blood or urine samples are needed; how samples are to be collected, and labelled; how samples are to be shipped from the trial site and how they are to be received at the laboratory (Cohen et al., 1995; Spriet et al., 1992; Spilker, 1991). If more than one trial centre is involved, then SOPs require that one central laboratory be used; this enables the same system of quality assurance and quality control to be implemented for all assays (Spriet et al., 1992). 9.5.11 MISCELLANEOUS PROCEDURES A standard operating procedure manual ends with a number of SOPs which indicate how a multiple-centre trial will be co-ordinated (if applicable), how a clinical trial will be terminated  in case of any serious adverse events or poor subject/investigator  compliance with the protocol, how any documentation lost during the trial will be managed, how all parties in the trial are to be insured, and how and where results will be published (Cohen etal., 1995; Spielman, 199; Spilker, 1991; Spriet etal., 1992).  L FRANCIOSI  10. WHAT ITEMS DO I CONSIDER QUALITY OF CLINICAL TRIALS?  RELEVANT WHEN  87  EVALUATING THE  Each of the five Good Clinical Practices has a number of items that I consider relevant based on my knowledge of clinical trials (Section 10.1 to 10.5). I have turned each item into a question and assessed for its presence or absence. A yes answer is given a score of 1 and a no answer a score of 0. At the end of each G C P section, the scores are added. A similar calculation is made when all GCPs are scored. This final score is referred to as the Clinical Trial Score or CTS (Section 10.6). A percent deviation from the best possible trial is then calculated.  When using this Clinical Trial Evaluation System (CTES), an item may be considered in an implicit manner (i.e., the item is thought about but is not formally written down) or in an explicit manner (i.e., the item is thought about and then formally written down).  L. FRANCIOSI  10.1 Items considered relevant to the primary question ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there a primary question?  2.  Was the question formally stated in advance? (i.e., was it stated in a protocol prior to conducting the trial?)  3.  W a s a specific problem identified? (i.e., from past research, personal experience, or theory?)  4.  Was a review of the literature conducted?  5.  Was the question's clinical importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  6.  Was the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  7.  W a s an appropriate primary response variable chosen?  8.  W a s the time needed to investigate the question considered?  9.  Was the availability of subjects considered?  10. Was the expertise of the investigator and clinical trial staff considered? 11. W a s the expense considered? 12. Were the adequacy considered?  and availability  of  facilities  and  equipment  13. Were the question's ethical implications considered? 14. W a s a scientific hypothesis formulated? 15. W a s the scientific hypothesis formally stated in advance?  TOTAL ITEMS CONSIDERED (Score out of 15)  L. FRANCIOSI  10.2 Items considered relevant to the design ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  2.  W a s the importance of a random sample considered? (i.e., it being the 'ideal' sample for generalising the results and using statistics)  3.  Was there a method to recruit subjects?  4.  Were the inclusion criteria defined?  5.  Were the exclusion criteria defined?  6.  Was there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  7.  Was there consideration for recording nonqualifiers? (i.e., subjects who do not meet the eligibility criteria)  8.  W a s there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  9.  W a s there consideration for recording discontinuers? (i.e., subjects who are discontinued by the investigator due to noncompliance or a (serious) adverse event)  10. Was there consideration for monitoring subject compliance? 11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?) 12. Was a sample size calculation performed? 13. Was there a rationale or clinical justification for determining the sample size? 14. Was there a defined control treatment? (i.e., a standard, placebo, or no treatment?) 15. Was there a defined test treatment(s)? 16. Was there a method to randomly allocate subjects to treatments? 17. Was there consideration for assessing the baseline characteristics? 18. Was the potential degree of blindness considered? 19. Was there consideration for assessing the adequacy of blinding? 20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  TOTAL ITEMS CONSIDERED (Score out of 20)  89  L FRANCIOSI  10.3 Items considered relevant to the statistics ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was a measure of central tendency considered? (i.e., a mean, mode or median)  2.  Was a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  3.  Was a correlation and/or a form of regression analysis considered?  4.  Was a null hypothesis formally stated in advance?  5.  Was an alternative hypothesis formally stated in advance?  6.  Was an alpha level formally specified in advance?  7.  Was a beta level formally specified in advance?  8.  Was there consideration about statistical power before and after the trial?  9.  W a s a pilot study or a review of the literature conducted to determine the possible variance?  10. W a s a parametric and/or nonparametric statistical test considered for the question? 11. W a s the statistical test and its assumptions appropriate for the question asked? 12. Was the direction of the test considered? (i.e., one-tailed versus two-tailed?) 13. W a s there consideration for assessing the normality of the data? 14. Were data transformations considered? 15. Was the use of a confidence interval considered? 16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance? 17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?) 18. Was there a statistical method in place to deal with missing data? 19. Was there a statistical method in place to deal with outliers? 20. W a s there consideration for an interim analysis?  TOTAL ITEMS CONSIDERED (Score out of 20)  L. FRANCIOSI  10.4 Items considered relevant to the ethics ANSWERS:  YES = Score of 1  NO = Score of 0  1. W a s there consideration about the potential risks involved? 2.  W a s there consideration about the potential clinical benefits involved?  3.  W a s the appropriateness of the experimental design considered?  4.  W a s the appropriateness of the statistics considered?  5. W a s there consideration about any "conflict of interest" regarding the treatment(s) being investigated? 6.  W a s early termination of the clinical trial considered?  7. Were issues related to subject consent considered? 8.  Were issues related to subject confidentiality considered?  9. W a s the opinion of an ethics committee sought? 10. W a s there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  L. FRANCIOSI  10.5 Items relevant to standard operating procedures ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was there a manual of standard operating procedures?  2.  Were responsibilities delegated to all clinical trial participants/staff?  3.  Was there an investigator's brochure (or a collection of relevant literature)?  4.  W a s there a protocol?  5.  Were there case report forms?  6.  W a s there a final report?  7.  W a s there documented approval from an ethics committee and/or regulatory authority?  8.  W a s there an approved consent form?  9.  W a s there a means of maintaining subject confidentiality?  10. W a s there any monitoring of the trial? 11. Were there definitions for an adverse event and a serious adverse event? 12. Were there methods for collecting and reporting any (serious) adverse event? 13. Were there a method for preparing treatments for the clinical trial? 14. Were there methods for managing the collected data? 15. Was there an audit and/or inspection planned for the purpose of quality assurance? 16. Were there methods for collecting and managing clinical laboratory data? 17. W a s there consideration about how the trial will be terminated? 18. Was there consideration about how any documentation lost during the trial would be managed? 19. Was there any consideration or formal statement about insurance and liability? 20. W a s there any consideration about publishing the results of the clinical trial?  TOTAL ITEMS CONSIDERED (Score out of 20)  92  L FRANCIOSI  10.6 The Clinical Trial Score (CTS) and % deviation from the best possible trial  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items) Choose an appropriate design for the question (20 items) Use appropriate statistics (20 items) Consider the ethical issues (10 items) Have standard operating procedures in place (20 items) CLINICAL TRIAL SCORE (CTS) (85 items) DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85)) x 100)  %  L. FRANCIOSI  94  PART III. THE USE OF THE CLINICAL TRIAL EVALUATION SYSTEM TO A S S E S S THE QUALITY OF CLINICAL TRIALS  After developing the Clinical Trial Evaluation System, I used it to assess the quality of six clinical trials that I helped conduct between the years of 1993 and 1997 at the University of British Columbia. One of the first trials was an investigation of a medical device called Controlled Environment Treatment (CET) System, evaluating its safety, efficacy and utility in a population of burn patients. Three of the trials were conducted in the field of anaesthesia.  One of them investigated the effects of intraarticular  morphine on pain following shoulder surgery.  Another evaluated the costs of two  general anaesthetics, sevoflurane and isoflurane in patients undergoing knee surgery. The third trial was an investigation of a local anaesthetic procedure to determine if it produces better pain relief than narcotics in patients also undergoing knee surgery. There was also a phase I trial in which a new drug, TSB330 was administered to healthy male volunteers for the first time to assess its safety.  Finally, there was a  clinical trial that evaluated whether or not TSB430, an antiemetic and tranquilliser, produced local analgesia in subjects with capsaicin-induced pain. For each of these trials, the background and the methods considered in their design and implementation will be described. Following this description, the results of the CTES assessment and a brief discussion about the deviations that occurred from the best possible trial will be presented.  L. FRANCIOSI  95  11. The Clinical Evaluation of the Controlled Environment Treatment (CET) System - A Medical Device for Burn Patients  11.1 Introduction In September of 1993, I was introduced to my first clinical trial, which involved the evaluation of a medical device called The Controlled Environment Treatment System or CET. The device uses sterile, tempered air as a dressing medium for chronic wounds. This "Air Bandage" was intended to improve wound healing and reduce infection by both eliminating the daily bandage changes of conventional therapy and controlling the environment at the wound site.  The CET device had three components: a computerised blower unit that sat underneath a hospital bed, a sterile plastic disposable unit which contained the affected limb, and a flexible hose that delivered the treated air from the blower unit to the disposable unit (Figure 7).  Its microprocessor-controlled feedback system, which was stored in the blower unit, worked in a manner illustrated in Figure 8.  Clinical staff would set the blower's  microprocessor to certain known parameters of temperature, humidity, and applied pressure that are applicable for the patient's wound condition (Burgess et al., 1977; Hunt et al., 1987; Kegel, 1976; Kostyuchenok et al., 1988; Lashchevker et al., 1987; Morris et al., 1979; Redhead et al., 1978; Sologub et al., 1983; Troup, 1980).  The  microprocessor would then measure and control these parameters at the wound site by  L.  FRANCIOSI  96  Figure 7. A patient undergoing CET Treatment. The CET system has a computerised blower unit underneath the bed, a sterile plastic disposable unit with the large leg wound, and a flexible hose connecting the two units.  Figure 8.  The CET feedback system.  L. FRANCIOSI  using feedback sensors in the disposable unit.  97  The current measurements of  temperature, humidity, and applied pressure would be displayed on the blower's control console. If the staff felt that the wound's progress was not satisfactory, then they would change the settings. The blower also had the capability of being programmed, i.e., a nomogram could used to adjust parameters when certain stages of wound healing were reached.  As seen in Figure 7, the disposable unit was transparent. This allowed clinical staff to assess the wound's progress without removing the bandage.  It also enabled staff to  palpate or touch the affected limb without protective clothing since the inside of the Air Bandage was considered sterile.  11.2 Methods The company that developed this technology provided a number of CET units to associated UBC researchers.  I assisted these researchers in implementing a clinical  trial to investigate this new therapy. A protocol was already in place that required the trial to be done in two stages (Appendix 6).  11.2.1 Stage  One  In the first stage, five patients were to be entered into the trial in order to evaluate the number of dressing changes that occurred in patients with comparable burns on both limbs. The timing of eschar (dead tissue) formation, and the isolation capabilities of the CET disposable unit were also to be determined.  L. FRANCIOSI  98  After obtaining informed consent from the patient, wound culture swabs and photographs of the patient's burned areas were taken. Next, the nurse in charge was to perform a bath with hybitane soap and then put a gauze dressing with silver sulfadiazine, an antibacterial cream, to cover the burn on one limb; this was the conventional treatment. The CET Air Bandage was to be placed on the other affected limb.  The patient was to have daily bathes and dressing changes with silver  sulfadiazine on one side while the Air Bandage, which had predetermined settings of temperature (32°C), humidity (45-44%), and pressure (15 mmHg) was not to be changed for at least 6 days. Depending on the patient's condition, culture swabs from both sites and blood analyses were to be done.  The number of dressing changes as well as any Air Bandages that needed to be changed were to be recorded. Any problems isolating the limb into the Air Bandage were also to be recorded. An independent consultant was to be hired to evaluate the formation of eschar from photographs. This assessment was to be done in a blinded fashion, i.e. the consultant was not to be told which limb had which treatment. 11.2.2  Stage Two  After the results of the first phase were obtained, the second stage of the clinical trial was to be conducted. The objective of this phase was to perform repetitive quantitative bacteriology and cytology from the burns in order to demonstrate that there was no harmful change in the formation of eschar and wound healing when CET treatment was performed on patients. As well, an estimation of the effectiveness of the CET treatment was also to be determined.  L. FRANCIOSI  T h e plan w a s to perform this phase in a blinded and randomised fashion.  99  Forty  patients were to be selected and randomised to receive either the novel C E T treatment or the conventional treatment.  For entry into the trial, patients had to be: (a) adult  patients (over 18 years old) with second and third degree burns, f r o m 5 to 2 0 % Total Body Surface A r e a (TBSA); a n d (b) patients had to have o n e b u r n e d limb in order to allow for the placement of the C E T Air Bandage.  Patients were to be excluded: (a) if  they were unable to give informed consent; and (b) if they had any serious pre-existing illness, for e x a m p l e , vascular insufficiency, insulin d e p e n d e n t diabetes, cancer, and poor immunity.  Photography and w o u n d cultures were again to be taken according to a routine schedule.  T h e C E T micro-processor w a s to be p r o g r a m m e d in order to monitor the  patients' healing progression and log statistical information for later statistical evaluation of the therapies.  Finally, the results of the bacteriology, cytology, and any w o u n d healing evident from the photography for both groups of patients w e r e t h e n to be a s s e s s e d using a blinded, independent consultant (i.e. the consultant w a s not to be told the treatments that the patients received). Statistical analyses were to be d o n e using analysis of variance and its derivatives w h e r e v e r it w a s appropriate.  T h e results of the entire clinical trial were  then to be presented at the annual meeting of the A m e r i c a n Burn Association.  L. FRANCIOSI  100  I and five other UBC students were asked to assist in conducting both stages of the clinical trial. One of our responsibilities was to develop a number of case report forms that would enable us to capture all information collected during trial's conduct.  L. FRANCIOSI  11.3 Results of the CTES Assessment 11.3.1 Items  considered  relevant  ANSWERS:  to the primary  question  Y E S = Score of 1  for the CET trial  NO = Score of 0  1.  W a s there a primary question?  2.  W a s the question formally stated in advance? protocol prior to conducting the clinical trial?)  3.  Was a specific problem identified? experience, or theory?)  4.  W a s a review of the literature conducted?  1  5.  W a s the question's importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  1  6.  W a s the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  0  7.  W a s an appropriate primary response variable chosen?  0  8.  W a s the time needed to investigate the question considered?  0  9.  W a s the availability of subjects considered?  1  1 (i.e., was it stated in a  (i.e., from past research, personal  0 1  10. W a s the expertise of the investigator and clinical trial staff considered?  1  11. W a s the expense considered?  1  12. Were the adequacy considered?  and availability  of facilities  and  equipment  1  13. Were the question's ethical implications considered?  1  14. W a s a scientific hypothesis formulated?  1  15. Was the scientific hypothesis formally stated in advance?  0  TOTAL ITEMS CONSIDERED (Score out of 15)  10  L. FRANCIOSI  102  11.3.1.1 A discussion about the CET trial's questions  In the clinical evaluation of the Controlled Environment Treatment System (CET), there was no clearly stated primary question.  Looking at the protocol in Appendix 6,  the  clinical trial was to be done in two stages. In the first stage, the protocol stated that there were three criteria that were to be evaluated: the estimation of dressing changes, the evaluation of the timing of eschar formation, and the evaluation of the isolation capabilities. In the second stage, one objective was to demonstrate that CET did not affect the formation of eschar and wound healing by doing bacteriology and cytology measurements.  Another objective was to determine if there was a difference in  effectiveness between CET and the conventional treatment of burns. It appears that the principal investigators had questions concerning the safety, the utility, and the effectiveness of the CET medical device. Unfortunately, it was not clear which question was of primary importance, i.e., safety, utilisation or effectiveness?  None of the questions were specific. No primary response variables were specified. The variables relevant to safety were unspecified. Blood pressure, body temperatures and microbiology culture swabs were taken, but it was clear which was important. The response  variables for the determination  of efficacy were wound size, pain,  bacteriology. But the relative importance of each and how to measure them was not specified.  At times, the number of bandage changes appeared to be the most  important variable while at other times, patient preference was stressed.  L. FRANCIOSI  103  Specific questions could have been: "In patients with second and third degree burns on one leg, are there any adverse events associated with the use of the Controlled Environment Treatment System?" or "Do patients with second and third degree burns on one arm who receive CET have fewer bandage changes than patients who do not?" Equivalency or evaluated reductions in morphine consumption and infection rates between those patients who receive CET and those who do not are appropriate questions.  During the conduct of the trial, it was difficult to approach any of the questions because there was no schedule. Each investigator was a practising physician. Meetings were often held around their schedules. These meetings were often not pertinent to the clinical trial.  The staff that were to assist in the conducting the clinical trial kept  changing. It was difficult to collect consistent data, because the new staff had little or no training.  The issue of subject availability was considered, however, it was not comprehensively considered before the trial.  Most of the available burn victims were ineligible to  participate in the clinical trial. On average, there were ten eligible subjects available a year. If the trial was conducted on 45 patients, it would take 4.5 years to complete! This resulted in the use of other types of patients, decreasing the power of the trial.  While the physicians were expert clinicians, there was a lack of expertise in the conduct of a medical device trial.  None were experts in the scientific evaluation of wound  L. FRANCIOSI  healing or wound management.  104  Technical and/or statistical expertise was not sought  by the investigators.  In the budget for the clinical trial, the money available was used to develop the CET medical device. Money was not allotted for the trial design, statistics or personnel.  L. FRANCIOSI  11.3.2  Items  considered  relevant  ANSWERS:  to the design  for the CET trial  YES = Score of 1  NO = Score of 0  1.  Was the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  2.  Was the importance of a random sample considered? (i.e. it being the 'ideal' sample for generalising the results and using statistics?)  3.  Was there a method to recruit subjects?  1  4.  Were the inclusion criteria defined?  1  5.  Were the exclusion criteria defined?  1  6.  Was there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  0  7.  Was there consideration for recording nonqualifiers? the eligibility criteria)  0  8.  Was there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  1  9.  Was there consideration for recording discontinuers? (i.e., subjects who are withdrawn by the investigator due to noncompliance or a (serious) adverse event)  0  (i.e., subjects who do not meet  1  10. Was there consideration for monitoring subject compliance?  1  11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?)  1  12. Was a sample size calculation performed?  0  13. Was there a rationale or clinical justification for determining the sample size?  1  14. Was there a defined control treatment? (i.e., a standard, placebo, or no treatment?)  1  15. Was there a defined test treatment(s)?  1  16. Was there a method to randomly allocate subjects to treatments?  1  17. Was there consideration for assessing the baseline characteristics?  1  18. Was the potential degree of blindness considered?  1  19. Was there consideration for assessing the adequacy of blinding?  0  20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  0  TOTAL ITEMS CONSIDERED (Score out of 20)  14  1  L. FRANCIOSI  106  11.3.2.1 A discussion about the CET trial's design  The lack of a stated primary question rendered the design of the CET trial inadequate according to Good Clinical Practice. The protocol encompassed two designs. The first stage of the trial to be conducted according to a crossover design and the second stage according to a parallel-group design.  But because of time, lack of supervision of  students, lack of eligible burn patients and little money to perform the trial, a case series design ultimately resulted.  The patient population that needed to be investigated could have been better described in the protocol. A clear statement of the rationale for patient choice and how CET could have benefited this population would have improved the trial. The first patient was a person with peripheral vascular disease although this was not included in the population to be investigated.  The investigators overestimated the number of eligible burn  patients. If statistics on patient availability had been obtained in the planning stage, this would not have occurred.  No record of patients who were unwilling to participate or patients that were not eligible to participate was kept.  Such a record would have helped determine if the CET  technology was appropriate for the patient population selected.  Even though two  withdrawals were recorded, there was no mechanism for recording dropouts or discontinuers in the protocol although I recorded this information as the trial progressed.  Patients dropped out for reasons which had not been anticipated, for  example, two patients' religious beliefs mandated that they be bathed daily.  L. FRANCIOSI  There is no indication of how the sample size was determined.  107  There was no  calculation of sample size based on either a pilot study and/or the literature.  The  investigators believed the numbers chosen were both statistically and medically significant based upon their clinical experience (Appendix 6).  A blinded evaluation of the data was part of the protocol.  However, a blinded  consultant was not hired as planned because the data was incomplete. undergoing conventional therapy were not evaluated.  Patients  Thus, there was no formal  control group in the second stage.  Issues related to confounding were not considered in the trial design. In retrospect, confounders proved to be history, instrumentation, and maintenance of treatment conditions over time. The patients' treatment history varied, some patients were given conventional treatment prior to being given CET. This confounds the analysis because the patient's impression will be influenced by prior experience with conventional bandaging whereas all patients did not have the same experience. Even though not specified in the protocol, humidity and temperature readings were used to ensure that settings of 32°C and 45% humidity were consistent with readings in the Air Bandage. It was later discovered that the meter was not calibrated and that the readings were off by 20-30% resulting in undesired temperature and humidity settings.  This discovery  confounded results particularly in comparison of wounds of earlier and later patients.  L. FRANCIOSI 11.3.3 Items considered relevant to the statistics for the CET trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s a measure of central tendency considered? (i.e., a mean, mode or median)  1  2.  W a s a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  1  3.  Was a correlation and/or a form of regression analysis considered?  1  4.  W a s a null hypothesis formally stated in advance?  0  5.  Was an alternative hypothesis formally stated in advance?  0  6.  Was an alpha level formally specified in advance?  1  7.  W a s a beta level formally specified in advance?  0  8.  W a s there consideration for statistical power before and after the trial?  0  9.  W a s a pilot study or a review of the literature conducted to determine the possible variance?  0  10. Was a parametric and/or nonparametric statistical test considered for the question?  1  11. W a s the statistical test and its assumptions appropriate for the question asked?  0  12. W a s the direction of the test considered? (i.e., one-tailed versus two-tailed?)  1  13. W a s there consideration for assessing the normality of the data?  0  14. Were data transformations considered?  0  15. Was the use of a confidence interval considered?  0  16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance?  1  17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?)  1  18. Was there a statistical method in place to deal with missing data?  0  19. Was there a statistical method in place to deal with outliers?  0  20. W a s there consideration for an interim analysis?  1  TOTAL ITEMS CONSIDERED (Score out of 20)  9  1  L. F R A N C I O S I  109  11.3.3.1 A discussion about the CET trial's statistics  More than half of the recommended statistical questions were considered in the protocol. Again, the lack of a stated primary question made it impossible to select appropriate statistics.  The references to statistics in the protocol were the use of ANOVA and its derivatives. However, they were not done. Statistical advise prior to commencing the clinical trial would have been beneficial.  L. F R A N C I O S I  11.3.4 Items considered relevant to the ethics for the CET trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was there consideration about the potential risks involved?  1  2.  Was there consideration about the potential clinical benefits involved?  1  3.  Was the appropriateness of the experimental design considered?  1  4.  Was the appropriateness of the statistics considered?  0  5.  Was there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  1  6.  Was early termination of the clinical trial considered?  1  7.  Were issues related to subject consent considered?  1  8.  Were issues related to subject confidentiality considered?  1  9.  Was the opinion of an ethics committee sought?  1  10. Was there a formal statement indicating that the principles of the - Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  0  8  11.3.4.1 A discussion about the CET trial's ethics  Consideration of the ethical implications of the questions were not recorded. The appropriateness of the design and statistics, or the importance of the Declaration of Helsinki were not formally considered.  An ethical issue was the use of CET for nonresearch purposes.  Ineligible patients  requested the CET device for treatment on many occasions because the physicians and nurses were convinced of the benefits of the avoidance of bandage changes. This was a common occurrence in the paediatric units. One of the investigators felt that it  110  L FRANCIOSI  111  was unethical to treat patients with a technology that was not yet shown to be safe and effective from clinical trials. Thus, he had the trial halted until appropriate funds and properly trained staff were available to conduct trials.  L. FRANCIOSI  112  11.3.5 Items relevant to standard operating procedures for the CET trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there a manual of standard operating procedures?  0  2.  Were responsibilities delegated to all clinical trial participants/personnel?  0  3.  Was there an investigator's brochure (or a collection of relevant literature)?  1  4.  Was there a protocol?  1  5.  Were there case report forms?  1  6.  Was there a final report?  1  7.  Was there documented approval from an ethics committee and/or regulatory authority?  8.  Was there an approved consent form?  1  9.  W a s there a means of maintaining subject confidentiality?  1  1  10. Was there any monitoring of the trial?  0  11. Were there definitions for an adverse event and a serious adverse event?  0  12. Were there methods for collecting and reporting any (serious) adverse event?  0  13. Were there a method for preparing treatments for the clinical trial?  1  14. Were there methods for managing the collected data?  0  15. Was there an audit and/or inspection planned for the purpose of quality assurance?  0  16. Were there methods for collecting and managing clinical laboratory data?  0  17. Was there consideration about how the trial will be terminated?  0  18. W a s there consideration about how any documentation lost during the trial would be managed?  0  19. Was there any consideration or statement about insurance and liability?  0  20. W a s there any consideration about publishing the results of the clinical trial?  1  TOTAL ITEMS CONSIDERED (Score out of 20)  9  L. FRANCIOSI  113  11.3.5.1 A discussion about the CET trial's standard operating procedures  There was no standard operating procedures manual for this trial.  Such a manual  would have been valuable in performing the trial. During the conduct of the trial, the students were not sure what data to collect since the question changed with time. This uncertainty was seen in the information collected in the case report forms. Information that was relevant to both stages was being collected simultaneously. The lay-out of the case reports forms kept changing and the amount of information being collected was also less. Patients other than those meeting the inclusion criteria in the protocol were allowed to participate. A set of standard operating procedures would have avoided the inconsistencies in data collection, improved the supervision of personnel, and prevented the deviations from the protocol.  L. FRANCIOSI  11.3.6  The Clinical Trial Score for the CET trial.  (CTS) and % deviation  from  the best possible  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items)  10  Choose an appropriate design for the question (20 items)  14  Use appropriate statistics (20 items)  9  Consider the ethical issues (10 items)  8  Have standard operating procedures in place (20 items)  9  CLINICAL TRIAL SCORE (CTS) (85 items)  50  DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85))x 100%)  41%  114  trial  11.4 A discussion about the CTS and the % deviation from the best possible trial The Clinical Trial Score of 50 indicates that the trial deviated from the best possible trial by 41%. Looking at the scores obtained for each good clinical practice, the pattern of deviation in this clinical trial was: Ethics(20%) < Design(30%) < Question(33%) < Statistics(55%) = SOPs(55%) The quality of this clinical trial could have been improved if we had a mechanism to evaluate the initial proposal. The suggested Clinical Trial Score provides a tool to avoid these errors in the future.  L. FRANCIOSI  115  12. The Analgesic Efficacy of Intraarticular Morphine (MOR) following Shoulder Surgery  12.1 Introduction In January 1994, researchers at UBC considered the relevance of peripheral opioid analgesia. It had been known for some time that mu, delta and kappa opioid receptors possibly existed in peripheral tissues (Joris et al., 1987; Stein et al., 1989). The resident had read a number of publications, in which the authors attempted to demonstrate the analgesic efficacy of intraarticular morphine. The results of these trials were conflicting (Heard etal., 1992; Raja etal., 1992; Stein etal., 1991).  One clinical trial reported the analgesic effect of intraarticular morphine following knee arthroscopy (Stein et al., 1991). In a double-blind manner, the investigators randomly assigned forty-two patients to four treatments: 1 mg morphine intraarticularly and saline intravenously, saline intraarticularly and 1 mg morphine intravenously, 0.5 mg morphine intraarticularly and saline intravenously, and the last treatment, 1 mg morphine and 0.1 mg naloxone intraarticularly and saline intravenously.  They reported a significant  decrease in pain and analgesic consumption in the group that received 1 mg of morphine intraarticularly and no difference with the administration of the lesser dose of morphine or the intravenous administration. They also found the addition of naloxone, an opioid receptor antagonist, reversed the analgesic effect of the intraarticular morphine. However, the results were questionable because of their use of historical  L. FRANCIOSI 116 controls, the differences in time course associated with rescue medications that they used, and their problem of multiple testing over a period of 24 hours.  Another group used a larger study population (139 patients), all of whom also underwent knee arthroscopy (Heard et al., 1992). They found no difference with the use of 0.6 mg morphine intraarticularly as compared to a saline control group. The only difference noted was the morphine group had a longer time to their first request for analgesic.  12.2 Methods To determine  if intraarticular morphine could produce peripheral analgesia, a  randomised, controlled double-blind clinical trial was designed to be conducted at Vancouver Hospital, UBC site.  For the trial population, 40 patients that were  undergoing shoulder surgery were selected. (They considered these type of patients 'appropriate' for the trial because they experienced a significant amount of postoperative joint pain that usually required one to two days of hospitalisation). If, after obtaining informed consent, any of these patients had systemic illnesses, known allergies to the study medications, or were less than 16 to great than 50 years of age, they were to be excluded from the trial.  At the time of surgery, each patient was to be randomly assigned to one of the two groups (20 patients per group) by the sealed envelope method.  Then, a standard  general anaesthetic consisting of propofol induction and laryngeal mask management of the airway or the addition of succinylcholine was to be applied followed by intubation.  L. FRANCIOSI  117  Maintenance of anaesthesia was to be carried out with nitrous oxide, isoflurane and oxygen. During the procedure, no morphine or opioid was to be used. On closing the shoulder capsule, the patient was then to be given an intraarticular injection of 10 ml of trial solution by the anaesthetist. The medical staff and patient were all to be blinded to the trial solutions. The patients were to receive either 10 ml of 5 mg of preservativefree morphine in 0.9% saline or 10 ml of 0.9% saline alone. As closing continued, all patients were to be loaded with 0.2 mg/kg of morphine intravenously by the anaesthetist.  The patient's arm was then to be placed into a sling and upon emerging  from the anaesthesia, the patient was to be taken to the recovery room.  All patients were then to receive morphine in 2 mg intravenous aliquots as needed in the recovery room. The patient was to be on a PCA device for twenty four hours postoperative. The PCA was to be set for 2 mg morphine intravenous bolus with a lockout time of 6 minutes. The accumulated dose at six hours was to be recorded. The VAS was a secondary measure that used to be evaluate pain in the recovery room and at 6 hours postoperatively.  The trial was scheduled to begin in April 1994. All recording of data and administration of study medications were to be done mainly by a medical resident. When the trial was completed, I was to review the resident's final report and to assist in the preparation of a manuscript for submission to a peer-reviewed journal.  L. FRANCIOSI  12.3 Results of the CTES assessment 12.3.1  Items  considered  relevant  ANSWERS:  to the primary  question  YES = Score of 1  for the MOR  trial  NO = Score of 0  1.  W a s there a primary question?  2.  W a s the question formally stated in advance? protocol prior to conducting the clinical trial?)  3.  W a s a specific problem identified? experience, or theory?)  4.  W a s a review of the literature conducted?  1  5.  W a s the question's importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  1  6.  W a s the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  0  7.  W a s an appropriate primary response variable chosen?  1  8.  Was the time needed to investigate the question considered?  1  9.  Was the availability of subjects considered?  1  1 (i.e., was it stated in a  (i.e., from past research, personal  1 1  10. W a s the expertise of the investigator and clinical trial staff considered?  1  11. Was the expense considered?  1  12. Were the adequacy considered?  and availability  of  facilities  and  equipment  1  13. Were the question's ethical implications considered?  1  14. W a s a scientific hypothesis formulated?  1  15. W a s the scientific hypothesis formally stated in advance?  0  TOTAL ITEMS CONSIDERED (Score out of 15)  13  118  L. FRANCIOSI  119  12.3.1.1 A discussion about the MOR trial's question  The CTES assessment indicates that thirteen out of fifteen items relevant to the primary question were considered.  The primary question of this clinical trial was "Does intraarticular morphine peripheral  analgesia  following shoulder  surgery?"  produce  However, this question was not  specific enough. It could have read: "In patients undergoing shoulder surgery, is 5 mg  of  morphine  consumption  intraarticularly  more  efficacious  in  reducing  in the first 6 hours following surgery than a saline  morphine  placebo?"  Based on a review of the literature, the hypothesis that intraarticular morphine produces analgesia by binding to peripheral opioid receptors was formulated. However, this was not stated in advance.  L. FRANCIOSI 1 2 0  12.3.2 Items considered relevant to the design for the MOR trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  1  2.  W a s the importance of a random sample considered? (i.e. it being the 'ideal' sample for generalising the results and using statistics?)  1  3.  W a s there a method to recruit subjects?  1  4.  Were the inclusion criteria defined?  1  5.  Were the exclusion criteria defined?  1  6.  W a s there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  0  7.  Was there consideration for recording nonqualifiers? (i.e., subjects who do not meet the eligibility criteria)  0  8.  W a s there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  0  9.  W a s there consideration for recording discontinuers? (i.e., subjects who are withdrawn by the investigator due to noncompliance or a (serious) adverse event)  1  10. Was there consideration for monitoring subject compliance?  1  11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?)  0  12. Was a sample size calculation performed?  0  13. Was there a rationale or clinical justification for determining the sample size?  0  14. Was there a defined control treatment? (i.e., a standard, placebo, or no treatment?)  1  15. Was there a defined test treatment(s)?  1  16. Was there a method to randomly allocate subjects to treatments?  1  17. Was there consideration for assessing the baseline characteristics?  1  18. Was the potential degree of blindness considered?  1  19. W a s there consideration for assessing the adequacy of blinding?  0  20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  0  TOTAL ITEMS CONSIDERED (Score out of 20)  12  L FRANCIOSI  121  12.3.2.1 A discussion about the MOR trial's design  Twelve out of twenty items relevant to trial design were considered. population was not described.  The patient  Refusers or nonqualifiers were not recorded.  Issues  related to generalisability and confounding were not considered. The reasons for the choice of sample size was not considered. The adequacy of blinding in this trial was not assessed.  L. FRANCIOSI  12.3.3 Items  considered  relevant  ANSWERS:  to the statistics  for the MOR trial  YES = Score of 1  NO = Score of 0  1.  Was a measure of central tendency considered? (i.e., a mean, mode or median)  1  2.  Was a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  1  3.  Was a correlation and/or a form of regression analysis considered?  1  4.  W a s a null hypothesis formally stated in advance?  0  5.  W a s an alternative hypothesis formally stated in advance?  0  6.  Was an alpha level formally specified in advance?  1  7.  Was a beta level formally specified in advance?  0  8.  W a s there consideration for statistical power before and after the trial?  0  9.  W a s a pilot study or a review of the literature conducted to determine the possible variance?  0  10. Was a parametric and/or nonparametric statistical test considered for the question?  1  11. Was the statistical test and its assumptions appropriate for the question asked?  1  12. Was the direction of the test considered? (i.e., one-tailed versus two-tailed?)  1  13. W a s there consideration for assessing the normality of the data?  0  14. Were data transformations considered?  0  15. W a s the use of a confidence interval considered?  0  16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance?  1  17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?)  1  18. Was there a statistical method in place to deal with missing data?  0  19. W a s there a statistical method in place to deal with outliers?  0  20. W a s there consideration for an interim analysis?  0  TOTAL ITEMS CONSIDERED (Score out of 20)  9  1  L. FRANCIOSI  123  12.3.3.1 A discussion about the MOR trial's statistics  The results of CTES assessment indicate that eleven out of the 20 items relevant to statistics were not considered.  Even though a two-tailed t-test was appropriately selected for this question, it was not performed properly.  No significant difference between groups (p=0.08) was found,  however, the data from discontinued patients were not included in the test, introducing a possible bias in the results (Appendix 7).  The normality of the data as well as the possibility of outliers were not considered until after the trial was completed. Upon analysing the data, I found that a log transformation of the control group values (Figures 9 and 10) and the morphine group values (Figures 11 and 12) made the data more normally distributed. When I performed a two-tailed test on the transformed data, I came to the same conclusion, no significant difference exists between groups (p=0.08). However, when I eliminated the outliers from the raw and transformed data according Chauveanet's criterion, I found that re-testing the data produced a significant result (p«0.02). This finding raised the question: Does the use of intraarticular  morphine  produce pain rather than pain relief?  Unsure of this, I  performed a retrospective power analysis according to Cohen et al., 1988 and discovered that with the outliers, the trial only had 41% power to detect a difference between the two groups and without the outliers, the trial only had 62% power (Appendix 7).  Each power value is the same whether or not the data was log  L FRANCIOSI  transformed.  124  The initial analysis specified that the beta for this trial was 0.10, or in  other words, 90% power to detect a significant difference between groups.  When I considered the 41% power figure in terms of sample size, I found that 60 patients per group were needed to detect a difference! (Cohen et al., 1988).  If a  literature review or pilot study had been done, the sample size would have been chosen to detect a difference.  (In the medical literature, there are investigators who report  nonsignificant results, but fail to consider the possibility that their trials may be underpowered as a result of their small sample size (e.g., Freiman etal., 1978)).  L. FRANCIOSI  125  Frequency Distribution of Raw Control Group Values  --  1111 MiiiiiiiiiiiiiR ^iilillliilll  ••Hi ••H  -  ••Mill  • • • • • 1 1 1  1  Less than 0.2  1  0.2 to 0.4  1  0.4 to 0.6  1  0.6 to 0.8  1  Greater than 0.8  m g / k g m o r p h i n e at 6 h o u r s  Figure 9. The frequency distribution of control group values for the MOR trial. The distribution of these values is said to be 'positively skewed' or 'skewed to the right'  Frequency Distribution of LOG Transformed Control Group Values 7  T  6 5  g 4 V 3  S  3 2 + 1 • 0 Less than-0.8  -0.8 to-0.6  -0.6 to-0.4  -0.4 to-0.2  Greater than -0.2  Log (mg/kg morphine at 6 hours)  Figure 10. The frequency distribution of control values following a log transformation for the MOR trial. This distribution appears to approximate the standard normal curve in Figure 2.  L. F R A N C I O S I  126  Frequency Distribution of Raw IA Morphine Group Values 8 7 6 5  c 3 2  LL.  4 3, 2 1 0 Less than 0.25  0.25 to 0.6  0.6 to 0.9  0.9 to 1.3  Greater than 1.3  mg/kg morphine at 6 hours with 5 mg IA  Figure 1 1 . The frequency distribution of the intraarticular morphine group values for the MOR trial. This distribution is also positively skewed.  Frequency Distribution of LOG Transformed IA Morphine Group Values  IT c  3  o  3 tr £ 2  Less than-0.6  -0.6 to-0.4  -0.4 to-0.2  -0.2 to 0.01  Greater than 0.01  Log (mg/kg morphine at 6 hours with 5 mg IA)  Figure 1 2 . The frequency distribution of the intraarticular morphine group values following a log transformation for the MOR trial. This distribution also appears to approximate the standard normal curve, but is more platykurtic (middle-heavy).  L. FRANCIOSI  127  12.3.4 Items considered relevant to the ethics for the MOR trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there consideration about the potential risks involved?  1  2.  W a s there consideration about the potential clinical benefits involved?  1  3.  W a s the appropriateness of the experimental design considered?  1  4.  W a s the appropriateness of the statistics considered?  0  5.  W a s there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  1  6.  W a s early termination of the clinical trial considered?  0  7.  Were issues related to subject consent considered?  1  8.  Were issues related to subject confidentiality considered?  1  9.  W a s the opinion of an ethics committee sought?  1  10. W a s there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  0  7  12.3.4.1 A discussion about the MOR trial's ethics  As the CTES assessment indicates, thirty percent of items were not considered. A sample size calculation was not performed and the statistics were improperly used. Items 6 and 10 were not considered.  L. FRANCIOSI  128  12.3.5 Items relevant to standard operating procedures for the MOR trial  ANSWERS:  Y E S = Score of 1  NO = Score of 0  1.  W a s there a manual of standard operating procedures?  0  2.  Were responsibilities delegated to all clinical trial participants/personnel?  1  3.  Was there an investigator's brochure (or a collection of relevant literature)?  1  4.  Was there a protocol?  1  5.  Were there case report forms?  0  6.  Was there a final report?  1  7.  Was there documented approval from an ethics committee and/or regulatory authority?  1  8.  Was there an approved consent form?  1  9.  Was there a means of maintaining subject confidentiality?  1  10. Was there any monitoring of the trial?  0  11. Were there definitions for an adverse event and a serious adverse event?  0  12. Were there methods for collecting and reporting any (serious) adverse event?  0  13. Were there a method for preparing treatments for the clinical trial?  1  14. Were there methods for managing the collected data?  1  15. W a s there an audit and/or inspection planned for the purpose of quality assurance?  0  16. Were there methods for collecting and managing clinical laboratory data?  1  17. Was there consideration about how the trial will be terminated?  0  18. W a s there consideration about how any documentation lost during the trial would be managed?  0  19. W a s there any consideration or statement about insurance and liability?  1  20. Was there any consideration about publishing the results of the clinical trial?  1  TOTAL ITEMS CONSIDERED (Score out of 20)  12  L FRANCIOSI  1  12.3.5.1 A discussion about the MOR trial's standard operating procedures  In this clinical trial, there was no manual of standard operating procedures. Thus, large number of items relevant to SOPs were not considered.  12.3.6  L. FRANCIOSI  130  The Clinical Trial Score (CTS) and % deviation from the best possible for the MOR trial  trial  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items)  13  Choose an appropriate design for the question (20 items)  12  Use appropriate statistics (20 items)  9  Consider the ethical issues (10 items)  7  Have standard operating procedures in place (20 items)  12  CLINICAL TRIAL SCORE (CTS) (85 items)  53  DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85))x 100%)  38%  12.4 A discussion about the CTS and the % deviation from the best possible trial The Clinical Trial Score of 53 indicates that the trial deviated from the best possible trial by 38%. Looking at the scores obtained for each good clinical practice, the pattern of deviation in this clinical trial: Question(13%) < Ethics(30%) < Design(40%) = SOPs(40%) < Statistics(55%) It is evident that the quality of this clinical trial could have been improved if the CTES assessment had been available in the planning stages.  L. FRANCIOSI  131  13. The Clinical Safety Assessment of the Novel Drug, TSB330  13.1 Introduction In the summer and fall of 1996, I participated in a co-operative program in which I as a student helped design and implement a clinical safety trial for a local pharmaceutical company. TSB330, a novel drug with desirable analgesic and cardiovascular effects, was planned to be tested outside of North America.  As outlined in the protocol (Appendix 8), the trial was to be conducted on 18 healthy, male volunteers, with the primary objective of evaluating the safety of TSB330. The group of individuals selected for this trial were to receive ascending doses of TSB330 by the intravenous route and then monitored for 48 hours for any electrocardiograph changes. To further establish safety and tolerance of the drug, its pharmacokinetics were to be determined by taking blood samples at predetermined times. 13.2 Methods The trial was to be divided into two stages: Stage 1, low doses of TSB330 in 12 volunteers and Stage 2, high doses in 6 volunteers (Table 5). There were to be five groups, with a specified number of volunteers per group. Each volunteer was to be given no more than three doses, each on a separate visit. The first and second stage of the trial were to be done separately at two different hospitals, designated Site 1 and Site 2, respectively.  L. FRANCIOSI  TSB330 Stage 1: Group 1 Group 2 Group 3 Group 4 Stage 2: Group 5  First Hospital Visit  Second Hospital Visit  Third Hospital Visit  Number of Volunteers  1/1000 1/500 1/250 1/100  1/500 1/250 1/100 1/50  1/250 1/100 1/50 1/20  3 3 3 3  1/50  1/20  1/10  6  132  Table 5. The dose fractions of the novel drug TSB330 that were to be administered. These fractions were based on a predicted safe dose of 0.203 mg/kg; the highest dose to be administered was 1/10 the safe dose or 0.0203 mg/kg. Because TSB330 was inactive enantiomer of a Kappa agonist, its potential effects as a narcotic had to be investigated. Subjects in the second stage were to be evaluated for these particular effects.  The healthy individuals that were to participate in the trial had to have certain entry criteria: they had to be male with an age between 18 and 35; they had to have normal histories, physical examinations, and routine laboratory test results; they had to have a weight and height that was ±10% of the acceptable average of the normal population; and they had to provide informed written consent. The volunteers had to be subjected to a 30 minute interview for a mental health evaluation and to understand their personal reasons for participating in the trial. A list of criteria that were to be used to exclude volunteers is presented in Appendix 8.  L. FRANCIOSI  133  Again, as illustrated in Table 5, each subject that was to receive three doses of TSB330. Only one dose was to be administered per week. Each subsequent dose was to be two times the concentration of the last, and was to be administered providing the previous dose was clinically safe.  The safe dose was determined to be 0.203 mg/kg, on the basis of animal studies performed prior to the trial. The maximum dose to be given was to be 1/10 the safe dose or 0.0203 mg/kg. The drug at any dose was to be administered by a slow bolus infusion over 15 minutes.  The safety of TSB330 was to be assessed  by clinical observation, physical  examination, electrocardiogram, vital signs (including blood pressure and heart rate) and clinical laboratory evaluation at specified times. All adverse events were to be recorded. There were also criteria to determine when to stop an infusion of TSB330 during the trial (Appendix 8).  A number of clinical staff were to be recruited by a principal investigator.  This  investigator was to give each person a responsibility to implement this clinical trial (Appendix 8). In order to help the company co-ordinate the trial from Vancouver, the company's director of clinical development and myself were to perform a critical path analysis (Cohen etal., 1995; Clelend etal., 1983; Lockyer, 1984; Rosenau, 1981). The results of this analysis is illustrated in Figure 13; the initials in the diagram represent the following people: S S for Senior Scientist, LF for Luigi Franciosi, DCD for Director of  L. FRANCIOSI  134  Clinical Development, CS for Company Secretary, LA for Laboratory Assistant, and CC for Company Chemist.  Another person included in the critical path was a clinical monitor working for a contract research organisation. The company hired him to be the key person who gathers all legal documents and assures that the trial is initiated, conducted, and finished appropriately.  As evident from Figure 13, one of my responsibilities was to create and help finalise case report forms (CRFs) for both trial sites. These CRFs were to allow clinical staff and myself to capture all data accumulated during each volunteer visit. At the time, the trial was scheduled to begin around October 1996.  SS/LF fly to Centre 1  Confirm pickup at airport in foreign country  L. FRANCIOSI 135  a: *  CL C  H  K  S E  S E  & E  3  L. FRANCIOSI  13.3 Results of the CTES assessment 13.3.1  Items  considered  relevant  ANSWERS:  to the primary  question  Y E S = Score of 1  for the TSB330  trial  NO = Score of 0  1.  Was there a primary question?  2.  W a s the question formally stated in advance? protocol prior to conducting the clinical trial?)  3.  W a s a specific problem identified? experience, or theory?)  4.  W a s a review of the literature conducted?  1  5.  W a s the question's importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  1  6.  W a s the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  0  7.  W a s an appropriate primary response variable chosen?  0  8.  W a s the time needed to investigate the question considered?  1  9.  W a s the availability of subjects considered?  1  1 (i.e., was it stated in a  (i.e., from past research, personal  1  1  10. W a s the expertise of the investigator and clinical trial staff considered?  1  11. W a s the expense considered?  1  12. Were the adequacy considered?  and availability  of facilities  and  equipment  1  13. Were the question's ethical implications considered?  1  14. W a s a scientific hypothesis formulated?  1  15. W a s the scientific hypothesis formally stated in advance?  0  TOTAL ITEMS CONSIDERED (Score out of 15)  12  136  L. FRANCIOSI  13.3.1.1 A discussion about the 330 trial's  137  question  The items that the 330 trial did not consider were the question's specificity, the use of a primary response variable, and the formulation and statement of a hypothesis.  The primary question of the 330 trial was:  "Is  TSB330  safe  enough. A more appropriate question might have asked, "In healthy there  TSB330 washout  health  male  The CTES assessment suggests that this question was not specific  volunteers?"  are  in  any given  recorded as separate  adverse  events  15 minute  associated  slow  bolus  with infusions  male  ascending with  volunteers, doses  a seven  of day  period?"  Another deviation from the best possible trial was the lack of an appropriate primary response variable.  In most trials of safety, it is difficult to determine a single,  appropriate response variable for assessing a treatment's safety (Spilker, 1991). We used a number of variables such as blood pressure, heart rate and E C G to evaluate whether or not TSB330 was safe.  We did formulate a scientific hypothesis that TSB330 was safe, based on animal studies, but we did not formally state this in the protocol.  L. FRANCIOSI  13.3.2  Items  considered  relevant  ANSWERS:  to the design  for the 330 trial  YES = Score of 1  NO = Score of 0  1.  Was the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  1  2.  W a s the importance of a random sample considered? (i.e. it being the 'ideal' sample for generalising the results and using statistics?)  1  3.  Was there a method to recruit subjects?  1  4.  Were the inclusion criteria defined?  1  5.  Were the exclusion criteria defined?  1  6.  Was there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  0  7.  Was there consideration for recording nonqualifiers? the eligibility criteria)  0  8.  W a s there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  1  9.  Was there consideration for recording discontinuers? (i.e., subjects who are withdrawn by the investigator due to noncompliance or a (serious) adverse event)  1  (i.e., subjects who do not meet  10. Was there consideration for monitoring subject compliance?  1  11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?)  1  12. Was a sample size calculation performed?  0  13. W a s there a rationale or clinical justification for determining the sample size?  1  14. W a s there a defined control treatment? (i.e., a standard, placebo, or no treatment?)  0  15. W a s there a defined test treatment(s)?  1  16. W a s there a method to randomly allocate subjects to treatments?  0  17. Was there consideration for assessing the baseline characteristics?  1  18. W a s the potential degree of blindness considered?  1  19. Was there consideration for assessing the adequacy of blinding?  0  20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  0  TOTAL ITEMS CONSIDERED (Score out of 20)  13  138  L. FRANCIOSI  139  13.3.2.1 A discussion about the 330 trial's design  As the results of CTES assessment indicate, less than half of the items relevant to the design were not considered. The reason for the deviations was the lack of a parallel group design. During the planning stages of this trial, we did consider a number of designs, in particular the one presented in Table 6 (O'Grady et al., 1997). However, we realised we did not have the resources for a randomised, double-blind clinical trial. We opted instead for a clinical trial design without any blinding, placebo or randomisation. Thus, the influence of bias in the trial results could have been possible.  Week \  1 2 3 4 5 6  No. Subjects on Placebo 1 1 1 1 1 1  Number of Subjects who have received each treatment (where X is the first dose) X 2X 4X 8X 16X 1 5 -  -  -  4 -  1 4 -  1 4 -  1 4  Table 6. One of the parallel-group designs that was considered for the TSB330 trial, but was not done due to insufficient resources.  L FRANCIOSI  13.3.3 Items considered relevant to the statistics for the 330 trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was a measure of central tendency considered? (i.e., a mean, mode or median)  1  2.  W a s a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  1  3.  Was a correlation and/or a form of regression analysis considered?  1  4.  Was a null hypothesis formally stated in advance?  0  5.  W a s an alternative hypothesis formally stated in advance?  0  6.  Was an alpha level formally specified in advance?  1  7.  W a s a beta level formally specified in advance?  0  8.  Was there consideration for statistical power before and after the trial?  1  9.  Was a pilot study or a review of the literature conducted to determine the possible variance?  1  10. Was a parametric and/or nonparametric statistical test considered for the question?  0  11. Was the statistical test and its assumptions appropriate for the question asked?  0  12. W a s the direction of the test considered? (i.e., one-tailed versus two-tailed?)  0  13. Was there consideration for assessing the normality of the data?  0  14. Were data transformations considered?  0  15. Was the use of a confidence interval considered?  0  16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance?  1  17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?)  1  18. Was there a statistical method in place to deal with missing data?  0  19. Was there a statistical method in place to deal with outliers?  0  20. Was there consideration for an interim analysis?  0  TOTAL ITEMS CONSIDERED (Score out of 20)  8  140  L. FRANCIOSI  13.3.3.1 A discussion about the 330 trial's  141  statistics  Since there was no parallel group design in this clinical trial, many items relevant to statistics were not considered.  Descriptive statistics such as a mean and standard  deviation were used for assessing baseline characteristics and summarising subject demographics (Table 7). We also graphed the plasma concentration of two high doses of TSB330 over time for a non-linear regression analysis.  Volunteer  Number  Age (years)  Height (cm)  Weight (kg)  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  24 21 23 23 26 21 20 22 24 36 35 32 39 36 25 33 29 30  172 175 186 180 155 170 175 168 165 158 174 170 166 180 170 178 181 175  84 72 102 62 54 76 68 58 69 60 70 74 75 82 56 75 76 90  Mean  27.7  172.1  72.3  Standard Deviation  6.1  7.9  12.3  Table 7. The demographics of 330 trial subjects.  L. FRANCIOSI  142  13.3.4 Items considered relevant to the ethics for the 330 trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there consideration about the potential risks involved?  1  2.  W a s there consideration about the potential clinical benefits involved?  1  3.  W a s the appropriateness of the experimental design considered?  1  4.  W a s the appropriateness of the statistics considered?  1  5.  W a s there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  1  6.  W a s early termination of the clinical trial considered?  7.  Were issues related to subject consent considered?  8.  Were issues related to subject confidentiality considered?  9.  Was the opinion of an ethics committee sought?  1  10. W a s there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  10  13.3.4.1 A discussion about the 330 trial's ethics  All items relevant to ethics were considered.  In the planning stages, we did not  consider if the investigators were in any conflict of interest with TSB330.  We later  learned that the investigators were both on the ethics committee that approved the use of this treatment. They did deal with this potential 'conflict of interest' by disclosing to other committee members their financial and academic interests in investigating TSB330.  L. FRANCIOSI  13.3.5  Items  relevant  to standard  ANSWERS:  operating  procedures  Y E S = Score of 1  143  for the 330 trial  NO = Score of 0  1.  Was there a manual of standard operating procedures?  0  2.  Were responsibilities delegated to all clinical trial participants/personnel?  1  3.  W a s there an investigator's brochure (or a collection of relevant literature)?  1  4.  W a s there a protocol?  1  5.  Were there case report forms?  1  6.  W a s there a final report?  1  7.  W a s there documented approval from an ethics committee and/or regulatory authority?  8.  W a s there an approved consent form?  1  9.  W a s there a means of maintaining subject confidentiality?  1  1  10. Was there any monitoring of the trial?  1  11. Were there definitions for an adverse event and a serious adverse event?  1  12. Were there methods for collecting and reporting any (serious) adverse event?  1  13. Were there a method for preparing treatments for the clinical trial?  1  14. Were there methods for managing the collected data? 15. Was there an audit and/or inspection planned for the purpose of quality assurance?  1  16. Were there methods for collecting and managing clinical laboratory data?  1  17. Was there consideration about how the trial will be terminated?  1  18. Was there consideration about how any documentation lost during the trial would be managed? 19. Was there any consideration or statement about insurance and liability?  1  20. Was there any consideration about publishing the results of the clinical trial?  1  TOTAL ITEMS CONSIDERED (Score out of 20)  17  L. FRANCIOSI  144  13.3.5.1 A discussion about the 330 trial's standard operating procedures  Although a manual of standard operating procedures (SOPs) did not exist, almost all items were considered in this section. The investigators and the monitor did not ensure that the data collected in the case report forms were complete.  This resulted in a  noticeable difference in the quality of data captured into the CRFs between the two sites.  If a manual of standard operating procedures had existed, then the observed  inconsistencies in the data collection would have not occurred.  L. FRANCIOSI  13.3.6  145  The Clinical Trial Score (CTS) and % deviation from the best possible trial for the 330 trial  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items)  12  Choose an appropriate design for the question (20 items)  13  Use appropriate statistics (20 items)  8  Consider the ethical issues (10 items)  10  Have standard operating procedures in place (20 items)  17  CLINICAL TRIAL SCORE (CTS) (85 items) DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85))x 100%)  60  29%  13.4 A discussion about the CTS and the % deviation from the best possible trial The Clinical Trial Score of 60 indicates that the trial deviated from the best possible trial by 29%. Looking at the scores obtained for each good clinical practice, the pattern of deviation in this clinical trial was: Ethics(0%) < SOPs(15%) < Question(20%) < Design(35%) < Statistics(60%) If the Clinical Trial Evaluation System had been used on the proposed protocol, it would have revealed weaknesses in design and statistics resulting in an improved trial.  L. FRANCIOSI  146  14. The Cost Evaluation of the Novel General Anaesthetic, Sevoflurane (SEV), in Patients undergoing Arthroscopic Menisectomy  14.1  Introduction  In the spring of 1996, a clinical investigator and his fellow anaesthetists at Vancouver Hospital were given the opportunity to use a new general anaesthetic  called  sevoflurane. Marketed by Abbott Laboratories, the use of this anaesthetic had some major advantages. It did not produce any of the lung irritation seen with other general anaesthetics, and it had rapid induction and awakening times (Doi et al., 1993; Lerman et al., 1994).. In one study, it was demonstrated to have an awakening time that was approximately half of the conventionally-used anaesthetic, isoflurane for comparable surgical procedures (Frink et al., 1992). Because of this early recovery, sevoflurane had the potential to permit for rapid patient assessment and discharge from the operating room and the postanaesthesia care unit. This, in turn, could result in an associated reduction in costs (Brown, 1995). Hence, sevoflurane could save money in the long term.  They priced sevoflurane at $300.00 for every 250 ml bottle, in  comparison to isoflurane which was $57.49 for every 100 ml bottle.  Before switching to this new anaesthetic, the investigator wanted to verify the potential cost savings.  In July 1996, he and I began planning a pharmacoeconomic trial of  sevoflurane at Vancouver Hospital, UBC site (Appendix 9). At the time, we agreed that if a trial were to be conducted, the published results would not be generalised to any other institution except this hospital.  One primary question was to be considered:  L. FRANCIOSI  147  Was there a cost difference between a sevoflurane anaesthetic and an isoflurane anaesthetic in a specific population of patients at Vancouver Hospital, UBC site? We also wanted to establish the effect of sevoflurane and isoflurane on readiness for discharge from the recovery room. 14.2  Methods  The trial was to be done in a prospective and randomised fashion (Appendix 9). The participating anaesthetist was to administer the inhalational agent as clinically indicated. To minimise the variance due to individual anaesthetists, the patients were to be randomly assigned in lines of 2, to either group. The number of cases per day was anticipated to be more than 2 and less than 4.  We decided that the most suitable patients to be used for this clinical trial were those undergoing an arthroscopic menisectomy, a surgical procedure commonly done to treat sport injuries.  Its primary purpose was to remove torn semilunar knee cartilages  (menisci) that would otherwise not repair on its own. Thus, the patients were very comparable since they were usually young, healthy, and athletic. In order to obtain them for the trial, they were to be recruited by their orthopaedic surgeon. Each patient was to be sent a letter two weeks prior to their surgery in order to inform them again about the objectives of the trial. Informed consent was to be obtained either at the surgeon's office or on the day of surgery by one of us.  In terms of sample size, 40 patients (20/group) were to be enrolled; this figure is based on the rationale that if we did not detect a difference at 40, then the variance would so  L. FRANCIOSI  large that a larger trial would have to be conducted.  148  These patients were to be  randomly assigned to receive either sevoflurane or isoflurane. But in order to qualify for the trial, they had to undergo arthroscopic menisectomy only at Vancouver Hospital, UBC Site.  Patients were to be excluded if they were less than 18 or greater than 45  years of age; had any illness other than surgical indication; were ASA III or greater; had any major systemic illness, such as diabetes or rheumatoid arthritis; had taken any central nervous system medication or any medication unrelated to the surgical procedure; had a bleeding tendency or allergy to medications; were unable to provide informed consent; had postoperative intraarticular drainage; and were female patients who were pregnant.  We were to reserve one anaesthetic machine for the trial. On it, there were to be three vaporisers, one standard isoflurane, one trial isoflurane and one trial sevoflurane. The machine had to be set at a three litres per minute flow rate for all anaesthetics used; this was to minimise the variance between anaesthetists and make the two groups more comparable.  The amount of isoflurane or sevoflurane used for each patient was to be determined using the anaesthetic  machine's computer program.  It recorded the inspired  concentration of the inhaled agent over time and the volume of fresh gas flow. This information was later to be retrieved by downloading it on to a 3.5 inch computer disc, and then exporting it to a Microsoft Excel spreadsheet.  L. FRANCIOSI  149  The cost of the agent for each patient was then determined from the percentage of the total cost of the agent established from the total volume of agent used in all patients. Thus, the number of bottles of isoflurane and sevoflurane used over the trial period also needed to be recorded. The type and amount of adjunct drugs administered in the Post Anaesthesia Room (PAR) were also to be recorded. To establish the total cost for each patient and the variance in the cost between patients in each group, the cost of the adjunct drugs was to be added to the cost of the inhaled agent. In this way, the overall mean cost and variance would be compared between groups.  Each patient was also to be evaluated for discharge from PAR, every 15 minutes; the earliest time that discharge criterion were met was to be recorded.  As well, the  readiness for discharge from the hospital, the incidence of nausea and vomiting seen after surgery were also to be noted.  In order to collect all this information, we had to develop a number of case report forms for the trial. At the time, a fellow student, who wanted to learn more about clinical trials, offered to assist in the design of these forms. The investigator and I later decided, that he would also share some of my responsibilities in implementing the clinical trial, in particular, assisting me in preparing and submitting the trial protocol to both the UBC Ethics Committee and the Department of Research Services of Vancouver Hospital for approval. He was also to help me record the data when the trial was scheduled to begin in the fall.  L. FRANCIOSI  14.3  Results of the CTES assessment  14.3.1  Items  considered  relevant  ANSWERS:  to the primary  question  YES = Score of 1  for the SEV trial  NO = Score of 0  1.  W a s there a primary question?  2.  Was the question formally stated in advance? protocol prior to conducting the clinical trial?)  3.  Was a specific problem identified? experience, or theory?)  4.  W a s a review of the literature conducted?  1  5.  W a s the question's importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  1  6.  W a s the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  0  7.  W a s an appropriate primary response variable chosen?  1  8.  W a s the time needed to investigate the question considered?  1  9.  W a s the availability of subjects considered?  1  1 (i.e., was it stated in a  (i.e., from past research, personal  1 1  10. W a s the expertise of the investigator and clinical trial staff considered?  1  11. W a s the expense considered?  1  12. Were the adequacy considered?  and availability  of facilities  and  equipment  1  13. Were the question's ethical implications considered?  1  14. W a s a scientific hypothesis formulated?  1  15. W a s the scientific hypothesis formally stated in advance?  1  TOTAL ITEMS CONSIDERED (Score out of 15)  14  150  L. FRANCIOSI  14.3.1.1  151  A discussion about the SEVtrial's question  As the results of the CTES assessment indicate, we considered fourteen of the fifteen items relevant to the primary question. Again, like many of the previous trials, the question in this trial was not specific enough. The primary question was "Is there a difference general  in  cost  between  a sevoflurane  of Anaesthesia  and the Formulary  there  difference  anaesthetic on  cost  when  arthroscopic  3Uminute  flow  anaesthetic  It could have read: "From the perspective  anaesthetic?"  a  general  Committee  between  a  sevoflurane  anaesthetists  are allowed  mensiectomy  patients  rate?"  This  at Vancouver  question  and  to practice  with  the  clearly  isoflurane  of the Hospital,  a  Department UBC site, is  isoflurane  as they  exception  and  and  general  normally that  thoroughly  they  would use  defines  a  the  pharmacoeconomic problem that is of interest to anaesthetists and to members of the formulary committee.  It also implies that a cost-minimisation analysis is to be  conducted (Bootman et al., 1989; Drummond et al., 1987; Drummond et al., 1991; Einarson etal., 1997; Jolicoeur et al., 1992; McNiece, 1996).  L. FRANCIOSI  14.3.2  Items  considered  ANSWERS:  relevant  to the design  for the SEV trial  Y E S = Score of 1  NO = Score of 0  1.  Was the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  2.  Was the importance of a random sample considered? (i.e. it being the 'ideal' sample for generalising the results and using statistics?)  3.  W a s there a method to recruit subjects?  1  4.  Were the inclusion criteria defined?  1  5.  Were the exclusion criteria defined?  1  6.  Was there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  0  7.  Was there consideration for recording nonqualifiers? the eligibility criteria)  0  8.  Was there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  1  9.  Was there consideration for recording discontinuers? (i.e., subjects who are withdrawn by the investigator due to noncompliance or a (serious) adverse event)  1  (i.e., subjects who do not meet  10. Was there consideration for monitoring subject compliance?  1  1  11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?) 12. W a s a sample size calculation performed? 13. W a s there a rationale or clinical justification for determining the sample size?  1  14. W a s there a defined control treatment? (i.e., a standard, placebo, or no treatment?)  1  15. Was there a defined test treatment(s)?  1  16. Was there a method to randomly allocate subjects to treatments?  1  17. Was there consideration for assessing the baseline characteristics?  1  18. W a s the potential degree of blindness considered?  1  19. W a s there consideration for assessing the adequacy of blinding?  0  20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  1  TOTAL ITEMS CONSIDERED (Score out of 20)  16  152  L. FRANCIOSI  14.3.2.1  A discussion about the SEVtrial's  153  design  As the CTES assessment indicates, we failed to consider four of the twenty items relevant to design. Like many of the previous trials, refusers and nonqualifiers were not recorded. We did perform a pilot study and had reviewed the literature to determine the possible variance, but we did not calculate a sample size. Our rationale for not doing this was that if we could not pick up a cost difference in 40 patients (i.e., 20 patients/group), then the variance in cost would be so great that a larger sample size would be necessary. No assessment of the adequacy of blinding was performed.  L FRANCIOSI  14.3.3  Items  considered  ANSWERS:  relevant  to the statistics  Y E S = Score of 1  for the SEV trial  NO = Score of 0  1.  Was a measure of central tendency considered? (i.e., a mean, mode or median)  2.  Was a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  3.  Was a correlation and/or a form of regression analysis considered?  4.  Was a null hypothesis formally stated in advance?  5.  Was an alternative hypothesis formally stated in advance?  1  6.  W a s an alpha level formally specified in advance?  1  7.  Was a beta level formally specified in advance?  8.  Was there consideration for statistical power before and after the trial?  1  9.  W a s a pilot study or a review of the literature conducted to determine the possible variance?  1  1  1  10. Was a parametric and/or nonparametric statistical test considered for the question? 11. W a s the statistical test and its assumptions appropriate for the question asked?  1  12. W a s the direction of the test considered? (i.e., one-tailed versus two-tailed?)  1  13. W a s there consideration for assessing the normality of the data?  1  14. Were data transformations considered?  1  15. Was the use of a confidence interval considered?  1  16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance? 17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?)  1  18. W a s there a statistical method in place to deal with missing data?  0  19. Was there a statistical method in place to deal with outliers?  1  20. W a s there consideration for an interim analysis?  0  TOTAL ITEMS CONSIDERED (Score out of 20)  16  1  L. FRANCIOSI  155  14.3.3.1 A discussion about the SEVtrial's statistics  Most of the items relevant to statistics were considered. The results of a two-tailed ttest indicated that there was a significant cost difference between a sevoflurane general anaesthetic and a isoflurane general anaesthetic (p<0.01) (Table 8). (The total cost per patient was determined as a percentage of the group cost; it was based on the total cost of the inspired agent, the total cost of intraoperative drugs and the total cost of postoperative drugs).  Sevoflurane Group  Isoflurane group  (n=20)  (n=20)  $38.10  $23.87  ±$10.13 ($27.97 - $48.23)  +$6.59 ($17.28-$30.46)  ±$4.44 ($33.66 - $42.54)  ±$2.89 ($20.98 - $26.76)  Mean total cost / patient Standard deviation 95% Confidence Interval for the mean  Table 8. The mean total cost per patient for the Sevoflurane and Isoflurane groups.  In a retrospective power analysis, done according to Cohen et al., 1988, we had more than enough power (>99.5%) to pick up a difference with 20 patients per group. In order to perform these statistical analyses, the null hypothesis and beta level were specified after the trial was completed. We also did not plan how we going to deal with missing data when performing these analyses. As well, we did not think about having an intermin analysis for this clinical trial.  L. FRANCIOSI  14.3.4  Items  considered  ANSWERS:  relevant  to the ethics  for the SEV trial  YES = Score of 1  NO = Score of 0  1.  Was there consideration about the potential risks involved?  1  2.  Was there consideration about the potential clinical benefits involved?  1  3.  Was the appropriateness of the experimental design considered?  0  4.  Was the appropriateness of the statistics considered?  1  5.  Was there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  0  6.  Was early termination of the clinical trial considered?  0  7.  Were issues related to subject consent considered?  1  8.  Were issues related to subject confidentiality considered?  1  9.  Was the opinion of an ethics committee sought?  1  10. Was there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  0  6  TOTAL ITEMS CONSIDERED (Score out of 10)  14.3.4.1  156  A discussion about the SEV trial's ethics  As the assessment indicates, four items were not considered. calculate the number of patients that were minimally  required  Since we did not  per group, we did not  really consider appropriateness of our design. Items 5, 6, and 10 were not considered because we did not know about their relevance in clinical trials at the time.  L. FRANCIOSI  14.3.5  Items  relevant  to standard  ANSWERS:  operating  procedures  Y E S = Score of 1  157  for the SEV trial  NO = Score of 0  1.  Was there a manual of standard operating procedures?  0  2.  Were responsibilities delegated to all clinical trial participants/personnel?  1  3.  Was there an investigator's brochure (or a collection of relevant literature)?  1  4.  W a s there a protocol?  1  5.  Were there case report forms?  1  6.  Was there a final report?  1  7.  Was there documented approval from an ethics committee and/or regulatory authority?  8.  Was there an approved consent form?  1  9.  Was there a means of maintaining subject confidentiality?  1  1  10. Was there any monitoring of the trial?  0  11. Were there definitions for an adverse event and a serious adverse event?  0  12. Were there methods for collecting and reporting any (serious) adverse event?  1  13. Were there a method for preparing treatments for the clinical trial?  1  14. Were there methods for managing the collected data?  1  15. W a s there an audit and/or inspection planned for the purpose of quality assurance?  0  16. Were there methods for collecting and managing clinical laboratory data?  1  17. Was there consideration about how the trial will be terminated?  0  18. W a s there consideration about how any documentation lost during the trial would be managed?  0  19. W a s there any consideration or statement about insurance and liability?  0  20. Was there any consideration about publishing the results of the clinical trial?  1  TOTAL If EMSGONSIDERED (Score out of 20)  13  L. FRANCIOSI  158  14.3.5.1 A discussion about the SEV trial's standard operating procedures More than half of the items relevant to standard operating procedures were considered. We did not have the knowledge to create our own manual of standard operating procedures, but if we did have one in place, we would have considered all of the outstanding items.  L. FRANCIOSI  14.3.6  The Clinical for the SEV  Trial Score trial  (CTS) and % deviation  from  the best possible  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items)  14  Choose an appropriate design for the question (20 items)  16  Use appropriate statistics (20 items)  16  Consider the ethical issues (10 items)  6  Have standard operating procedures in place (20 items)  13  CLINICAL TRIAL SCORE (CTS) (85 items)  65  DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85))x100%)  24%  159  trial  14.4 A discussion about the CTS and the % deviation from the best possible trial The Clinical Trial Score of 65 indicates that the trial deviated from the best possible trial by 24%. Looking at the scores obtained for each good clinical practice, the pattern of deviation in this clinical trial was: Question(7%) < Design(20%) = Statistics(20%) < SOPs(35%) < Ethics(40%) The quality of this clinical trial could have been improved if we had focused more on its SOPs and ethics during the planning stages.  L. FRANCIOSI  160  15. The Clinical Evaluation of Intradermal TSB430 as A Novel Peripheral Analgesic  15.1  Introduction  Aside from my clinical work, I was also given the opportunity to do some basic research in the area of pain. Researchers had discovered a series of compounds which acted in a novel manner to provide local analgesia. These compounds, which act on the fine pain nerve endings were termed selective nociceptor blockers or 'nociblockers'. In the animal studies that I helped conduct, TSB430, an antiemetic and tranquilizer, was shown to have nociblocker properties.  Taking into account that TSB430 was in common clinical use and had limited toxicity, the researchers were interested in testing TSB430 in humans in order to repeat the results seen in animal studies. In the summer and fall of 1995, I worked with them to design and implement a clinical trial with the specific objective of determining if intradermal TSB430 did produce peripheral analgesia in healthy subjects that had capsaicin induced pain (Appendix 10). 15.2  Methods  The trial was to be done in a randomised, double-blind fashion, comparing four doses of TSB430 with a saline control and a positive control (lidocaine).  It was to be  conducted on a total of 6 subjects. After obtaining informed consent, subjects were to be screened in order to determine if they qualified for the trial.  The only inclusion  criterion that was to be used was that all subjects had to be over the age of 18.  L. FRANCIOSI  161  Subjects that were to be excluded were those taking medication for a pre-existing illness and were unable to give informed consent. As well, students in the laboratory, like myself, could not be subjects.  The drugs and the concentrations that were to be tested by us were: 30, 100, 300, 1000 u,g/ml of TSB430; 1000 u.g/ml of lidocaine; and 0.9% normal saline.  Each of these  solutions was to be given a letter in order to blind both the subjects and the person injecting to the contents of the syringe. Capsaicin, the active agent of hot peppers, was to be injected in a concentration of 1 ng/10 uJ. Because it was of plant origin and the risk for an allergic reaction during intradermal injections was minimal, it was to be used as the pain producing substance for the trial. It also was also chosen for its ability to stimulate nerve endings; the same endings that TSB430 was to block.  In order to perform the trial, we decided to have six areas on the volar surface of the non-dominant forearm to serve as our injection sites for the six corresponding test solutions A, B, C, D, E, and F. These sites were to be in 3 rows of 2 from the proximal to the distal areas of the forearm (Figure 14).  proximal medial  a  P  y  8  8  <(>  lateral  distal  Figure 14. The injection sites on the volar surface of a subject's non-dominant forearm for the TSB430 clinical trial. Each site was to be represented by a Greek letter.  L. FRANCIOSI  162  All the test solutions were to be injected intradermal^/ using a v o l u m e of fifty microlitres and at a m i n i m u m interval of five minutes between injections. T h e injections were also to be d o n e in the s e q u e n c e a,  (3, y, 5, s, and <> j, so that the timing and order of the test  solutions w e r e to be altered systematically according a Latin square design (Table 9).  Injection site 1 A  2 B  Subjects 3 4 C D  5 E  6 F  B  C  D  E  F  A  E F  E  A  A B  A B  s  D E F  F  5  C D  <!>  F  A  B  C  a 3 Y  Table 9.  T h e Latin square design used for the T S B 4 3 0 trial. time and order effects of the test solutions being injected.  C D  B C D E  This design alters any  For example, as can be seen in Table 9, Subject 2 w a s to receive his first injection B at site a, then injection C at site p, and etc.  At one and half minutes after a test solution w a s to be injected, 10 u.l of the capsaicin solution w a s to be injected into the middle of the wheal area of the previous injection. A pain intensity rating w a s then immediately to be obtained. Tests of pin prick sensation in the wheal area were to be done at 1, 2, 3, 4, and 5 minutes after injection of the test solution.  The pain m e a s u r e m e n t s were to be m a d e as follows: (a) Pain produced by intradermal injection of capsaicin w a s to be evaluated by asking the subject to rate his or her pain on a 0 to 10 integer scale, immediately; 0 indicated no pain while 10 represented the  L. FRANCIOSI  163  worst pain imaginable; (b) for standardisation, to establish a modulus of pain prior to each injection, a heated (50°C) brass rod was to be placed on the uninjected volar surface of the forearm for five seconds and the subject told to regard this as a five on the integer scale. This was to be done prior to the injection of the test substance and also the capsaicin. At five minutes, the 50°C brass rod was to be placed beside the wheal and the subject told to regard this as a five on the integer scale. Then, the rod was to be placed directly on the wheal and the subject asked to evaluate the pain on the 0 to 10 integer scale compared to the standard; and (c) at 1, 2, 3, 4, and 5 minutes postinjection, a pin prick test was to be performed in the wheal area. The subject was to be asked whether or not he or she felt any sharp pain in the wheal area after a pin prick. The responses, YES or NO, were to be recorded.  In terms of statistical evaluation, the integer value of the pain reported at VA and 5 minutes, and the pin prick responses recorded over 5 minutes for each treatment were to be compared between treatment groups. Analysis of variance was to be done for the Latin square design according to D. C. Montgomery in his book Design and Analysis of Experiments. As well, a correlation analysis and a repeated measures ANOVA were to be performed on the results obtained. An alpha level of 0.05 was to be used for all analyses performed after the trial.  Volunteer recruitment, the preparation of drug solutions, and the development of case report forms were my responsibilities. Both investigators agreed that all injections were to be done by Dr. X and that the trial would take place in November 1995.  L. F R A N C I O S I  15.3  Results of the C T E S a s s e s s m e n t  15.3.1  Items considered  relevant to the primary question for the 430 trial  ANSWERS:  Y E S = Score of 1  NO = Score of 0  1.  Was there a primary question?  2.  Was the question formally stated in advance? protocol prior to conducting the clinical trial?)  3.  Was a specific problem identified? (i.e., from past research, personal experience, or theory?)  1  4.  Was a review of the literature conducted?  1  5.  Was the question's importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  1  6.  Was the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  0  7.  Was an appropriate primary response variable chosen?  1  8.  Was the time needed to investigate the question considered?  1  9.  Was the availability of subjects considered?  1  1 (i.e., was it stated in a  1  10. Was the expertise of the investigator and clinical trial staff considered?  1  11. Was the expense considered?  1  12. Were the adequacy considered?  and availability of  facilities  and equipment  1  13. Were the question's ethical implications considered?  1  14. Was a scientific hypothesis formulated?  1  15. Was the scientific hypothesis formally stated in advance?  1  T O T A L ITEMS C O N S I D E R E D (Score out of 15)  14  164  L. FRANCIOSI  15.3.1.1  165  A discussion about the 430 trial's question  Prior to the designing and implementing the trial, we had a specific problem: we needed to demonstrate that our compounds selectively blocked pain receptors or nociceptors in humans. Thus, we asked the primary question: "Does local  analgesia  in healthy  subjects  with  capsaicin  intradermal induced  TSB430  pain?"  produce  According to  the results of CTES assessment, this question was not specific enough. The question could have been: intradermals,  healthy  subjects  is there a reduction  300, or 1000 ug/ml scale?"  "In  given  in a subject's  of intradermal  TSB430  50ul  of  1ng/10ul  pain with 50ul of either as measured  capsaicin 0, 30. 100.  by a standard  pain  In this question, the independent variables (i.e., concentrations of TSB430)  as well as the method to measure the primary outcome are described. It would have enabled us to know exactly how the trial was to be conducted.  The more general  question could have been interpreted in different ways.  When the trial was conducted on November 1, 1995, the results indicated that as subjects were being given increasing concentrations of TSB430, the amount of pain they experienced decreased (Figure 15). Thus, we had evidence which supported our primary hypothesis that TSB430 nociceptors.  reduces  capsaicin  pain  by selectively  blocking  L. F R A N C I O S I  INTRADERMAL TSB430: P a i n s c o r e s w i t h c a p s a i c i n a t 1.5  W O R S T PAIN  10  MEAN±SEM  n=6 subjects  minutes  P<0.05  8 +  PAIN SCORE  NO PAIN  o 0 ug/ml TSB430 (SALINE)  30 ug/ml TSB430  100ug/ml TSB430  300 ug/ml TSB430  1OOOug/ml TSB430  Figure 15. The reduction of capsaicin pain with intradermal TSB430 in six subjects  166  L. FRANCIOSI 15.3.2 Items considered relevant to the design for the 430 trial ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  2.  Was the importance of a random sample considered? (i.e. it being the 'ideal' sample for generalising the results and using statistics?)  3.  Was there a method to recruit subjects?  4.  Were the inclusion criteria defined?  5.  Were the exclusion criteria defined?  1  6.  Was there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  0  7.  Was there consideration for recording nonqualifiers? the eligibility criteria)  0  8.  Was there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  1  9.  Was there consideration for recording discontinuers? (i.e., subjects who are withdrawn by the investigator due to noncompliance or a (serious) adverse event)  1  1  1  (i.e., subjects who do not meet  10. Was there consideration for monitoring subject compliance?  1  11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?)  1  12. Was a sample size calculation performed?  1  13. Was there a rationale or clinical justification for determining the sample size?  1  14. Was there a defined control treatment? (i.e., a standard, placebo, or no treatment?)  1  15. Was there a defined test treatment(s)?  1  16. Was there a method to randomly allocate subjects to treatments?  1  17. Was there consideration for assessing the baseline characteristics?  1  18. Was the potential degree of blindness considered?  1  19. Was there consideration for assessing the adequacy of blinding?  0  20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  1  TOTAL ITEMS CONSIDERED (Score out of 20)  17  167  L. FRANCIOSI  15.3.2.1  168  A discussion about the 430 trial's design  As the results of CTES assessment indicate, the majority of items were considered. Again,, the  recording of refusers  and  nonqualifiers, the  monitoring of subject  compliance, and the assessment of blinding were not considered in this trial.  L. FRANCIOSI  15.3.3  Items  considered  ANSWERS:  relevant  to the statistics  YES = Score of 1  for the 430 trial  NO = Score of 0  1.  Was a measure of central tendency considered? (i.e., a mean, mode or median)  2.  Was a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  3.  Was a correlation and/or a form of regression analysis considered?  4.  Was a null hypothesis formally stated in advance?  5.  W a s an alternative hypothesis formally stated in advance?  1  6.  Was an alpha level formally specified in advance?  1  7.  W a s a beta level formally specified in advance?  1  8.  Was there consideration for statistical power before and after the trial?  1  9.  Was a pilot study or a review of the literature conducted to determine the possible variance?  1  1  1  10. Was a parametric and/or nonparametric statistical test considered for the question?  1  11. Was the statistical test and its assumptions appropriate for the question asked?  1  12. Was the direction of the test considered? (i.e., one-tailed versus two-tailed?)  1  13. W a s there consideration for assessing the normality of the data?  1  14. Were data transformations considered?  1  15. Was the use of a confidence interval considered?  1  16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance? 17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?)  1  18. Was there a statistical method in place to deal with missing data? 19. W a s there a statistical method in place to deal with outliers?  1  20. Was there consideration for an interim analysis?  TOTAL ITEMS CONSIDERED (Score out of 20)  18  1  L. FRANCIOSI  170  15.3.3.1 A discussion about the 430 trial's statistics  Most items relevant to statistics were considered.  Some aspects of the statistical  analysis such as formulating and stating null hypothesis were done after the trial was conducted (Appendix 11). Following the trial, we decided to conduct two-tailed t-tests to look at differences between saline and TSB330 pain scores (Table 10).  Subjects SALINE 30 ng/ml TSB430 100 ng/ml TSB430 300 ng/ml TSB430 1000 ng/ml TSB430 1000 u-g/ml Lidocaine  1  2  3  4  5  6  MEAN  SEM  7 9 1 1 0 1  10 7 8 0 0 6  7 5 4 0 0 1  0 2 2 2 0 7  8 3 4 3 0 1  5 4 3 0 1 0  6.17 5.00 3.67 1.00 0.17 2.67  1.40 1.06 0.99 0.52 0.17 1.23  P-Values* Non-significant Non-significant p < 0.01 p < 0.002 0.010 > p > 0.05  P by t test versus saline (two-tailed)  Table 10. Mean Pain Scores with Capsaicin at 1.5 minutes for the 430 trial.  L. FRANCIOSI  171  15.3.4 Items considered relevant to the ethics for the 430 trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there consideration about the potential risks involved?  1  2.  W a s there consideration about the potential clinical benefits involved?  1  3.  W a s the appropriateness of the experimental design considered?  1  4.  W a s the appropriateness of the statistics considered?  1  5.  W a s there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  1  6.  W a s early termination of the clinical trial considered?  1  7.  Were issues related to subject consent considered?  1  8.  Were issues related to subject confidentiality considered?  1  9.  W a s the opinion of an ethics committee sought?  1  10. Was there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  15.3.4.1  0  9  A discussion about the 430 trial's ethics  Almost all items relevant to ethics were considered.  In this trial, the Declaration of  Helsinki was not considered. During the trial's design and implementation, we did not discuss the issue of conflict of interest. Later, it appeared to me that the researchers had an vested interest in the trial's outcome, but they were not in a 'conflict of interest' because they took steps to conduct the trial in a double-blind, placebo-controlled randomised fashion.  L. FRANCIOSI 172 15.3.5  Items relevant to standard operating procedures for the 430 trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there a manual of standard operating procedures?  0  2.  Were responsibilities delegated to all clinical trial participants/personnel?  1  3.  W a s there an investigator's brochure (or a collection of relevant literature)?  1  4.  Was there a protocol?  1  5.  Were there case report forms?  1  6.  Was there a final report?  1  7.  Was there documented approval from an ethics committee and/or regulatory authority?  8.  Was there an approved consent form?  1  9.  Was there a means of maintaining subject confidentiality?  1  10. Was there any monitoring of the trial?  1  1  11. Were there definitions for an adverse event and a serious adverse event? 12. Were there methods for collecting and reporting any (serious) adverse event?  1  13. Were there a method for preparing treatments for the clinical trial?  1  14. Were there methods for managing the collected data?  1  15. Was there an audit and/or inspection planned for the purpose of quality assurance?  0  16. Were there methods for collecting and managing clinical laboratory data?  0  17. Was there consideration about how the trial will be terminated?  1  18. W a s there consideration about how any documentation lost during the trial would be managed?  0  19. Was there any consideration or statement about insurance and liability?  1  20. W a s there any consideration about publishing the results of the clinical trial?  1  TOTAL ITEMS CONSIDERED (Score out of 20)  15  L. FRANCIOSI  173  15.3.5.1 A discussion about the 430 trial's standard operating procedures  We did not consider five out of the 20 items because of the number of reasons.  If a  manual of standard operating procedures had existed, then I would have: defined an adverse event, asked a fellow graduate student to audit the trial for assuring quality, determined how clinical laboratory data would be collected and managed in the case of an adverse event, and considered how any lost documentation would have been handled.  L. FRANCIOSI  15.3.6  The Clinical Trial Score for the 430 trial  (CTS) and % deviation  from  the best possible  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items)  14  Choose an appropriate design for the question (20 items)  17  Use appropriate statistics (20 items)  18  Consider the ethical issues (10 items)  9  Have standard operating procedures in place (20 items)  15  CLINICAL TRIAL SCORE (CTS) (85 items) DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85))x 100%)  174  trial  73  14%  15.4 A discussion about the CTS and the % deviation from the best possible trial The Clinical Trial Score of 73 indicates that the trial deviated from the best possible trial by 14%. Looking at the scores obtained for each good clinical practice, the pattern of deviation in this clinical was: Question(7%) < Ethics(10%) = Statistics(10%) < Design(15%) < SOPs(25%) The quality of this clinical trial could have been improved if we had focused more on its SOPs.  L. FRANCIOSI  175  16. A Clinical Trial of a Local Anaesthetic (ANA) Technique in Patients Undergoing Lower Limb Surgery  16.1  Introduction  In the spring of 1995, a practising anaesthetist and I discussed a clinical question about whether the introduction of a local anaesthetic technique would manage a patient's postoperative pain to the extent that he or she could go home the same day. In other words, can the use of this technique turn a patient undergoing surgery at Vancouver Hospital from an inpatient to an outpatient?  The surgical procedure that he was considering was done routinely at hospital for lower limb injuries.  The postoperative pain that followed surgery was treated with opioid  narcotic drugs, in particular morphine that was delivered intravenously using a patientcontrolled analgesia (PCA) device (Highgenboten, 1992; Ferrante, 1991). The use of morphine had a number of unwanted side effects like nausea, vomiting, constipation, and urinary retention.  A side effect that occurred infrequently, but needed to be  monitored hourly during its use was respiratory depression, or the unexpected slowing and/or stopping of the patient's ability to breathe.  These limitations associated with  morphine use, including an overnight stay for the PCA device, suggested that better pain management was needed.  At the time, the anaesthetist and I decided that we would perform a pilot study at the Vancouver Hospital, UBC site, to begin addressing the clinical question.  Our  L. F R A N C I O S I  176  preliminary findings indicated that the technique was effective in terms of reducing surgical pain and minimising the use of PCA morphine.  Following the pilot study, we decided to design a clinical trial in which the following primary question would be asked: between performing technique  "Is there  the technique  with a saline placebo  a significant difference in pain relief  with a local anaesthetic  control?"  and performing  the  Taking into account that one group was  to receive saline, the question had to be evaluated with some pain relief in mind. This forced us to make the primary question more specific: "In patients undergoing surgery,  is  consumed technigue placebo  there a significant in the  first 24 hours  with a local anaesthetic control?"  difference following  in the  total amount  surgery  and performing  between  of  morphine  performing  the technigue  knee  the  with a saline  A secondary question that would be asked was: Is there a  significant difference in the incidence of side effects (or adverse events) between the two groups?  The overall hypothesis for the trial was that the local anaesthetic  technique would provide better pain relief with fewer side effects than with conventional PCA morphine. If the results were positive, then the surgical procedure would be done on an outpatient basis.  In order to finance and implement the clinical trial, we approached a multinational drug company for their assistance.  A clinical trial protocol and a number of case report  forms were to be developed by the company with the agreement that they were not to be disclosed to the public.  L. FRANCIOSI  16.2  177  Methods  The clinical trial, which was to be done at Vancouver Hospital, UBC site, had a randomised, placebo-controlled, double-blind, parallel-group design. We already had determined from a sample size calculation that a total of 44 patients (22 per group) had to be enrolled in order to have enough statistical power to resolve a difference between the two groups; the calculation was based on the pilot study results. On the day of surgery, patients had to be asked for informed consent and then had to be screened according to a number of entry criteria. The inclusion criteria that were to be used were as follows: •  Male or female scheduled for in-patient lower limb surgery at the Vancouver Hospital, UBC site  •  Aged 19 to 45 years  •  ASA Class I (American Society of Anesthesiologists Patient Class I; Defined as a patient that has no organic, physiologic, biochemical, or psychiatric disturbance. The pathologic process for which the operation is to be performed is localised and does not entail a systemic disturbance)  •  Provide written informed consent  The exclusion criteria that were to be used were as follows: •  A known history of allergy, sensitivity or any other form of reaction to local anaesthetics of the amide type, acetaminophen or narcotics  •  Suspected inability to comply with study procedures, including language difficulties or medical history and/or concomitant disease, as judged by the investigator  •  Suspected significant alcohol, drug or medication abuse, as judged by the principal investigator  •  Receiving regular treatment with analgesics, sedatives or any other medication with central nervous system effects  •  Tendency to bleed  L. FRANCIOSI  178  •  Women who are pregnant or who are not practising medically acceptable contraception  •  Previous inclusion in this study  •  Participation in clinical studies during this study or in the 14 days prior to admission to this study  Patients who met all of the inclusion criteria and none of the exclusion criteria were to be randomised to receive the technique with either a local anaesthetic or saline. Prior to surgery, all patients were to receive the technique by the same anaesthetist.  Following surgery, acetaminophen with codeine was to be administered, initially, for relief of postoperative pain. The patient was to be connected to a PCA machine and intravenous morphine started if pain relief was inadequate or if the patient's pain at rest was greater than 50 mm on a 100 mm visual analogue scale (VAS). Assessments of postoperative pain were to be recorded at specified times starting one hour after arrival at the post anaesthesia care unit (PACU). Adverse events were to be recorded during hospitalisation. Adverse events reported by the patient (i.e. nausea, vomiting, etc.) were to be recorded during the 24 hour postoperative treatment period.  Any  concomitant medication (i.e. medication unrelated to the clinical trial investigation) was to be recorded from the administration of technique until hospital discharge.  At the end of the clinical trial, the primary variable, total morphine consumption, was to be analysed statistically by using analysis of variance and confidence intervals for the difference in morphine consumption between treatments. If the statistical assumptions did not hold, an equivalent non-parametric analysis was to be performed (i.e., Wilcoxon  L. FRANCIOSI  179  rank sum test with adjustments for ties and corresponding confidence intervals). Adverse events like nausea and vomiting were to be tabulated per treatment group. All statistical tests were to be two-sided, confidence intervals were to be constructed with 95% confidence and any comparison was to be declared significant if p value was less or equal to 0.05.  A computer generated randomisation list was to be produced by a sponsoring drug company. Prior to surgery, patients were to be randomised to receive the technique with either a local anaesthetic or saline with equal probability. Patients were to be assigned consecutive numbers as they were randomised into the trial.  To prevent  imbalances in the number of patients, treatments were to be evenly blocked. Patients not valid for the trial were to be replaced using the next available patient number.  Double-blindness was to be maintained by means of the identical appearance of the local anaesthetic and saline vials. The randomisation list was to be held by the drug company. The randomisation list was only to be available to those people responsible for drug packaging, until all editing and validation of the CRFs had been completed by the company.  Individual treatment code envelopes were to be provided by the  company indicating the treatment allocation for each randomised patient. One set was to be held by the company. A second set was to be kept by the principal investigator at the hospital in a safe but accessible place. On completion of the trial, the company's clinical monitor was to ensure the return of the code envelopes and document the reason for any opened envelopes.  L. FRANCIOSI  180  Initial contact and recruitment of patients, the performance of technique, and the recording of all data into the CRFs were to be done by the principal investigator and the coinvestigator. They had to meet with the clinical monitor on a routine basis to see that all data recorded were correct and legible. When the trial was to be completed in the Fall of 1996, my responsibility was to assist the investigators in publishing the results of trial in a peer-reviewed anaesthesia journal.  L FRANCIOSI  16.3  Results of the CTES assessment  16.3.1  Items considered  relevant to the primary question for the ANA trial  ANSWERS:  YES = Score of 1  1.  W a s there a primary question?  2.  W a s the question formally stated in advance? protocol prior to conducting the clinical trial?)  3.  W a s a specific problem identified? experience, or theory?)  NO = Score of 0 1 (i.e., was it stated in a  1  (i.e., from past research, personal 1  4.  W a s a review of the literature conducted?  1  5.  W a s the question's importance considered? (i.e., its impact on existing treatment, theory or policies related to practise)  1  6.  Was the question specific? (i.e., was the primary response variable, the population of interest, the treatments (dose and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference to be detected specified?)  '  7.  W a s an appropriate primary response variable chosen?  1  8.  W a s the time needed to investigate the question considered?  1  9.  W a s the availability of subjects considered?  1  10. W a s the expertise of the investigator and clinical trial staff considered?  1  11. W a s the expense considered?  1  12. Were the adequacy considered?  and availability  of facilities  and equipment  1  13. Were the question's ethical implications considered?  1  14. Was a scientific hypothesis formulated?  1  15. Was the scientific hypothesis formally stated in advance?  1  TOTAL ITEMS CONSIDERED (Score out of 15)  15  181  L. FRANCIOSI  182  16.3.1.1 A discussion about the ANA trial's question  As indicated by the results of the CTES assessment, the investigator and I properly considered the nature and feasibility of the primary question that we asked before the clinical trial was designed and implemented.  L. FRANCIOSI  16.3.2  Items considered relevant to the design for the ANA trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s the population to be investigated described? (i.e., in terms of demographics, diagnostic (including stage of the disease), prognostic, and comorbid factors?)  2.  Was the importance of a random sample considered? (i.e. it being the 'ideal' sample for generalising the results and using statistics?)  3.  Was there a method to recruit subjects?  1  4.  Were the inclusion criteria defined?  1  5.  Were the exclusion criteria defined?  1  6.  Was there consideration for recording refusers? (i.e., subjects who do not want to be screened or may not want to sign the consent form)  0  7.  Was there consideration for recording nonqualifiers? the eligibility criteria)  0  8.  W a s there consideration for recording dropouts? (i.e., subjects who are either unwilling to continue after starting in the trial or lost to follow-up)  1  9.  W a s there consideration for recording discontinuers? (i.e., subjects who are withdrawn by the investigator due to noncompliance or a (serious) adverse event)  1  (i.e., subjects who do not meet  10. Was there consideration for monitoring subject compliance?  1  1  11. Were issues related to generalisability considered? (i.e., threats to the selection process and the clinical trial setting?) 12. Was a sample size calculation performed?  1  13. Was there a rationale or clinical justification for determining the sample size?  1  14. Was there a defined control treatment? (i.e., a standard, placebo, or no treatment?)  1  15. Was there a defined test treatment(s)?  1  16. Was there a method to randomly allocate subjects to treatments?  1  17. Was there consideration for assessing the baseline characteristics?  1  18. Was the potential degree of blindness considered?  1  19. Was there consideration for assessing the adequacy of blinding?  0  20. Were the potential confounders in a clinical trial considered? (i.e., history, maturation, testing, instrumentation, statistical regression, selection of subjects, mortality, the selection process interacting with other threats, and maintenance of treatment conditions over time?)  1  TOTAL ITEMS CONSIDERED (Score out of 20)  17  183  L. FRANCIOSI  184  16.3.2.1 A discussion about the ANA trial's design  In this clinical trial, we considered 85% of the items relevant to design. The recording of refusers and nonqualifiers as well as the assessment of the adequacy of blinding were not considered.  L FRANCIOSI  16.3.3  Items considered relevant to the statistics for the ANA trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was a measure of central tendency considered? (i.e., a mean, mode or median)  1  2.  Was a measure of variability considered? (i.e., a variance, standard deviation or coefficient of variation)  1  3.  Was a correlation and/or a form of regression analysis considered?  1  4.  Was a null hypothesis formally stated in advance?  1  5.  Was an alternative hypothesis formally stated in advance?  1  6.  Was an alpha level formally specified in advance?  1  7.  Was a beta level formally specified in advance?  1  8.  Was there consideration for statistical power before and after the trial?  1  9.  Was a pilot study or a review of the literature conducted to determine the possible variance?  1  10. Was a parametric and/or nonparametric statistical test considered for the question?  1  11. Was the statistical test and its assumptions appropriate for the question asked?  1  12. Was the direction of the test considered? (i.e., one-tailed versus two-tailed?)  1  13. Was there consideration for assessing the normality of the data?  1  14. Were data transformations considered?  1  15. Was the use of a confidence interval considered?  1  16. Were the meanings of 'clinical significance' and 'statistical significance' considered in advance?  1  17. Were issues related to chance considered? (i.e., repeated statistical testing and random variation?) 18. Was there a statistical method in place to deal with missing data?  1  19. Was there a statistical method in place to deal with outliers?  1  20. Was there consideration for an interim analysis?  1  TOTAL ITEMS CONSIDERED (Score out of 20)  20  1  L. FRANCIOSI  186  16.3.3.1 A discussion about the ANA trial's statistics  All of the items relevant to statistics were considered because we had the assistance of the drug company's statistician.  L. FRANCIOSI  16.3.4  Items considered  relevant to the ethics for the ANA trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  W a s there consideration about the potential risks involved?  1  2.  W a s there consideration about the potential clinical benefits involved?  1  3.  Was the appropriateness of the experimental design considered?  1  4.  Was the appropriateness of the statistics considered?  1  5.  W a s there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  1  6.  W a s early termination of the clinical trial considered?  1  7.  Were issues related to subject consent considered?  1  8.  Were issues related to subject confidentiality considered?  1  9.  W a s the opinion of an ethics committee sought?  1  10. W a s there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  16.3.4.1  187  1  10  A discussion about the ANA trial's ethics  All of the items relevant to ethics were considered because of our knowledge of the practices as well as the drug company's policies, which ensure that all of their trials are conducted in an ethical manner.  L. FRANCIOSI  188  16.3.5 Items relevant to standard operating procedures for the ANA trial  ANSWERS:  YES = Score of 1  NO = Score of 0  1.  Was there a manual of standard operating procedures?  1  2.  Were responsibilities delegated to all clinical trial participants/personnel?  1  3.  Was there an investigator's brochure (or a collection of relevant literature)?  4.  Was there a protocol?  5.  Were there case report forms?  6.  Was there a final report?  7.  Was there documented approval from an ethics committee and/or regulatory authority?  8.  Was there an approved consent form?  9.  Was there a means of maintaining subject confidentiality?  1  1  10. W a s there any monitoring of the trial? 11. Were there definitions for an adverse event and a serious adverse event? 12. Were there methods for collecting and reporting any (serious) adverse event? 13. Were there a method for preparing treatments for the clinical trial? 14. Were there methods for managing the collected data?  1  15. W a s there an audit and/or inspection planned for the purpose of quality assurance?  1  16. Were there methods for collecting and managing clinical laboratory data? 17. W a s there consideration about how the trial will be terminated?  1  18. Was there consideration about how any documentation lost during the trial would be managed?  1  19. W a s there any consideration or statement about insurance and liability?  1  20. Was there any consideration about publishing the results of the clinical trial?  TOTAL ITEMS CONSIDERED (Score out of 20)  1  20  L. FRANCIOSI  189  16.3.5.1 A discussion about the ANA trial's standard operating procedures  The drug company had a manual of standard operating procedures.  As the CTES  assessment suggests, all of the items relevant to SOPs were considered in this clinical trial.  16.3.6  L. FRANCIOSI  190  The Clinical Trial Score (CTS) and % deviation from the best possible for the ANA trial  trial  GOOD CLINICAL PRACTICES  Score  Define and state a primary question well in advance (15 items)  15  Choose an appropriate design for the question (20 items)  17  Use appropriate statistics (20 items)  20  Consider the ethical issues (10 items)  10  Have standard operating procedures in place (20 items)  20  CLINICAL TRIAL SCORE (CTS) (85 items) DEVIATION FROM THE BEST POSSIBLE TRIAL ((1-(CTS/85)) x 100%)  16.4  82  4%  A discussion about the CTS and the % deviation from the best possible  trial The Clinical Trial Score of 82 indicates that the trial deviated from the best possible trial by only 4%. Looking at the scores obtained for each good clinical practice, the pattern of deviation in this clinical trial was: Question(0%) = Statistics(0%) = Ethics(0%) = SOPs(0%) < Design(15%) This clinical trial could have been the best possible trial if we had considered the three items relevant to trial design during the planning stages of this trial.  L. FRANCIOSI  191  PART IV. STRENGTHS, FUTURE DEVELOPMENTS AND POTENTIAL USERS OF THE CLINICAL TRIAL EVALUATION SYSTEM (CTES)  17.  17.1  Strengths of CTES  Assessment of a trial's quality following its completion  As I have already demonstrated, an investigator can use the Clinical Trial Evaluation System to evaluate the quality of their clinical trial following its completion.  It is  assumed that the investigator is using CTES for the first time and that he or she will assess the trial based on their informal and formal considerations made during its planning.  In other words, a review of both their previous decision making and final  clinical trial protocol is necessary. The results of the CTES assessment will indicate to the investigators what their best possible trial was under the circumstances and what they could have done to improve its quality. 17.2  Comparison of trial scores and % deviations  If an investigator begins to use CTES on a number of clinical trials, he or she can begin comparing their scores and % deviations. For example, I helped design and implement the SEV and 430 trials with the same investigator. As illustrated in Table 11, the % deviations obtained for each Good Clinical Practice (GCP) are 24% and 14%, respectively. He and I also designed the ANA trial, but it was implemented by a drug  L. FRANCIOSI  GOOD CLINICAL PRACTICES  192  CET  MOR  330  SEV  430  ANA  Define and state a primary question well in advance (15 items)  10  13  12  14  14  15  Choose an appropriate design for the question (20 items)  14  12  13  16  17  17  Use appropriate statistics (20 items)  9  9  8  16  18  20  Consider the ethical issues (10 items)  8  7  10  6  9  10  Have standard operating procedures in place (20 items)  9  12  17  13  15  20  CLINICAL TRIAL SCORE (CTS) (85 items)  50  53  60  65  73  82  DEVIATION FROM THE BEST POSSIBLE TRIAL (1-(CTS/85))x 100)  41%  38%  29%  24%  14%  4%  Table 11. A summary of scores and % deviations for all my clinical trials  company. There is a striking improvement in G C P scores when either the SEV or the 430 trial is compared to the ANA trial.  However, it is important to note that it is  inappropriate to assume that, for example, the CET trial (CTS=50) is more than half the quality of the ANA trial (CTS=82). To compare Clinical Trial Scores, the investigator must also compare their patterns of deviation to further establish their quality.  17.3  Determination of deviation patterns within and amongst trials  As I have already shown for each of my clinical trials, a pattern of deviation can be  L. FRANCIOSI  193  determined by looking at the % items that investigator did not consider or %  deviation for each GCP. For example, the patterns of deviation for my trials were:  CET: Ethics(20%) < Design(30%) < Question(33%) < Statistics(55%) = SOPs(55%)  MOR: Question(13%) < Ethics(30%) < Design(40%) = SOPs(40%) < Statistics(55%) 330:  Ethics(0%) < SOPs(15%) < Question(20%) < Design(35%) < Statistics(60%)  SEV: Question(7%) < Design(20%) = Statistics(20%) < SOPs(35%) < Ethics(40%) 430:  Question(7%) < Ethics(10%) = Statistics(10%)< Design(15%) < SOPs(25%) ANA: Question (0%) = Statistics (0%) = Ethics (0%) = SOPs (0%) < Design (15%)  Each pattern of deviation provides the investigator with an indication of where improvements in trial quality could have occurred.  As mentioned previously, the  investigator can use these patterns to compare the quality of his or her trials. This is particularly useful if he or she wants to know what G C P items were not considered amongst his or her trials. I will demonstrate this with my clinical trials.  L. FRANCIOSI  Items  194  CET  MOR  330  SEV  430  ANA  1.  Was there a primary question?  1  •1  1  1  1  1  2.  Was the question formally stated in advance?  0  1  1  1  1  1  3.  Was a specific problem identified from personal experience, past research or theories?  1  1  1  1  1  1  4.  Was a review of the literature conducted?  1  1  1  1  1  1  5.  Was the question's clinical importance considered?  1  1  1  1  1  1  6.  Was the question specific? (i.e., Was the primary response variable, the population of interest, the treatments (does and frequency), the period of time for follow-up, and if possible, the direction and magnitude of the relationship or difference specified?)  0  0  0  0  0  1  Was an appropriate primary response variable chosen?  0  1  0  1  1  1  8.  Was the time needed to investigate the question considered?  0  1  1  1  1  1  9.  Was the availability of subjects considered?  1  1  1  1  1  1  10. Was the expertise of the investigator and clinical trial staff considered?  1  1  1  1  1  1  11. Was the expense considered?  1  1  1  1  1  1  12. Were the adequacy and availability of facilities and equipment considered?  1  1  1  1  1  1  13. Were the question's ethical implications considered?  1  1  1  1  1  1  14. Was a scientific hypothesis formulated?  1  1  1  1  1  1  15. Was the scientific hypothesis formally stated in advance?  0  0  0  1  1  1  10  13  12  14  14  15  7.  TOTAL ITEMS CONSIDERED (Score out of 15)  Table 12. Patterns of deviation within and amongst trial questions  L. FRANCIOSI  195  In reviewing Table 12, two deviation patterns can be observed with respect to clinical trial questions.  1. A pattern is seen with the CET trial. Since it had no stated primary question, four items were not considered.  2. Five out of six clinical trials did not have a specific question (item 6).  L. FRANCIOSI  Items 1. 2. 3.  196  CET  MOR  330  SEV  430  ANA  W a s the population to be investigated described?  1  1  1  1  1  1  W a s the importance of a random sample considered?  1  1  1  1  1  1  1  1  1  1  W a s there a method to recruit subjects?  1  4.  Were the inclusion criteria defined?  1  1  1  1  1  1  5.  Were the exclusion criteria defined?  1  1  1  1  1  1  6.  Was there consideration for recording refusers?  0  0  0  0  0  0  7.  Was there consideration for recording nonqualifiers?  0  0  0  0  0  0  8.  W a s there consideration for recording dropouts?  1  0  1  1  1  1  9.  W a s there consideration for recording discontinuers?  0  1  1  1  1  1  10. W a s there consideration for monitoring subject compliance?  1  1  1  1  1  1  11. Were issues related to generalisability considered?  1  0  1  1  1  1  12. Was a sample size calculation performed?  0  13. W a s there a rationale or clinical justification for determining the sample size?  1  0  1  1  1  1  14. W a s there a defined control treatment?  1  1  0  1  1  1  15. W a s there a defined test treatment(s)?  1  1  1  1  1  1  16. Was there a method to randomly allocate subjects to treatments?  1  1  0  1  1  1  17. Was there consideration for assessing the baseline characteristics?  1  1  1  1  1  1  18. W a s the potential degree of blindness considered?  1  1  1  1  1  1  19. Was there consideration for assessing adequacy of blinding?  0  0  0  0  20. Were the potential confounders in a clinical trial considered?  0  0  0  14  12  13  TOTAL ITEMS CONSIDERED (Score out of 20)  liii^  Table 13. Patterns of deviation within and amongst trial designs  0  0 1  16  17  17  L. FRANCIOSI  197  In reviewing Table 13, a number of deviation patterns can be observed with respect to trial designs.  1. There are two distinct groups of trials. The CET, MOR, and 330 trials considered less items than the SEV, 430, and ANA trials.  2. The CET, MOR, and 330 trials did not consider potential confounders (item 20).  3. The 330 trial had no control group or randomisation method (items 14 and 16, respectively).  4. Four out of six trials did not consider calculating a sample size (item 12).  5. All of the trials did not consider the recording of refusers and nonqualifiers, and the assessment of the adequacy of trial blinding (items 6, 7, and 19, respectively).  L. FRANCIOSI  Items  198  CET  MOR  330  SEV  430  ANA  1.  Was a measure of central tendency considered?  1  1  1  1  1  1  2.  Was a measure of variability considered?  1  1  1  1  1  1  3.  Was a correlation and/or a form of regression analysis considered?  1  1  1  1  1  0  0  0  0  0  0  1  1  4. Was a null hypothesis formally stated 0 in advance? 5. Was an alternative hypothesis 0 formally stated in advance?  1  6.  Was an alpha level formally specified in advance?  1  1  1  1  1  7.  Was a beta level formally specified in 0 advance?  0  0  0  7  1  8. Was there consideration for statistical0 power before and after the trial ?  0  1  1  9. Was a pilot study or a review of the literature conducted for variance?  0  0  1  1  1  1  10. Was a parametric and/or nonparametric test considered for the question?  1  1  0  1  1  1  11. Was the statistical test and its assumptions appropriate for question?  0  1  0  1  1  1  12. Was the direction of the test considered?  1  1  0  1  1  13. Was there consideration for assessing 0 the normality of the data? 14. Were data transformations 0 considered?  0  0  1  7  0  0  1  1  15. Was the use of a confidence interval 0 considered?  0  0  1  1  16. Were 'clinical' and 'statistical significance' considered in advance?  1  1  1  1  1  17. Were issues of chance considered?  1  1  1  1  1  o  0  0  0  1  1  :  18. Was there a statistical method in place 0. to deal with missing data? 19. Was there a statistical method in place to deal with outliers? 20. Was there consideration for an interim 1 •analysis? TOTAL ITEMS CONSIDERED (Score out of 20)  9  0  0  0  1  9  8  16  18  Table 14. Patterns of deviation within and amongst trial statistics  1  1  1 1  1  1  20  L. FRANCIOSI  199  In reviewing Table 14, a number of deviation patterns can be observed with respect to trial statistics. 1. Again, there are two distinct groups of trials.  The CET, MOR, and 330 trials  considered less items than the SEV, 430, and ANA trials.  2. a) The CET, MOR, and 330 trials did not consider formally stating a null hypothesis, alternative  hypothesis  or  beta  level  (item  4,  b) There was also no consideration for assessing  5,  and  7,  respectively).  normality, choosing data  transformations, or using confidence intervals (items 13, 14, and 15 respectively). c) These trials did not have statistical methods in place to deal with missing data or outliers (items 18 and 19, respectively).  L. FRANCIOSI  200  Items  CET  MOR  330  SEV  430  ANA  1.  Was there consideration about the potential risks involved?  1  1  1  1  1  1  2.  Was there consideration about the potential clinical benefits involved?  1  1  1  3.  Was the appropriateness of the design considered?  1  1  0  1  4.  Was the appropriateness of the statistics considered?  0  1  1  1  5.  Was there consideration about any "conflict of interest" regarding the treatment(s) being investigated?  1  1  1  0  6.  Was early termination of the clinical trial considered?  1  0  0  7.  Were issues related to subject consent considered?  1  1  1  8.  Were issues related to subject confidentiality considered?  1  1  1  9.  Was the opinion of an ethics committee sought?  1  1  1  0  0  1  8  7  10  10. Was there a formal statement indicating that the principles of the Declaration of Helsinki were complied with?  TOTAL ITEMS CONSIDERED (Score out of 10)  ttEftl t  6  1  1  0  «  9  10  Table 15. Patterns of deviation within and amongst trial ethics  In reviewing Table 15, most of the clinical trials did not have a consideration or formal statement indicating that the principles of the Declaration of Helsinki were complied with.  In the SEV trial, there was no consideration for items 3, 5 and 6. Thus, it had the lowest number of items considered.  L. FRANCIOSI  Items  201  CET  MOR  330  SEV  430  ANA  0  0  0  0  7  1  1  1.  Was there a manual of standard operating procedures?  0  2.  Were responsibilities delegated to all clinical trial participants/personnel?  0  1  1  3.  W a s there an investigator's brochure (or a collection of relevant literature)?  1  1  1  1  4.  W a s there a protocol?  1  1  1  1  1  1  5.  Were there case report forms?  1  0  1  1  1  1  6.  W a s there a final report?  1  •  1  1  1  1  7.  W a s there documented approval from an ethics committee and/or regulatory authority?  1  1  1  1  1  1  1  1  1  1  1  1  1  10. Was there any monitoring of the trial?  0  0  1  0  7  11. Were there definitions for an adverse event and a serious adverse event?  0  0  1  0  7  12. Were there methods for collecting and reporting any (serious) adverse event?  0  0  1  1  1  1  13. W a s there a method for preparing treatments for the clinical trial?  1  1  1  1  1  1  14. Were there methods for managing the collected data?  0  1  0  1  1  1  15. Was there an audit and/or inspection planned for quality assurance?  0  0  1  0  0  7  16. Were there methods for collecting and managing clinical laboratory data?  0  1  1  1  1  17. Was there consideration about how the trial will be terminated?  0  0  1  0  7  18. Was there consideration of how lost documentation would be managed?  0  0  0  0  0  7  19. Was there any consideration or statement about insurance & liability?  0  1  1  0  1  1  20. Was there any consideration about publishing the results of the clinical trial?  1  1  1  1  1  1  9  12  17  13  15  20  8.  Was there an approved consent form?  9.  W a s there a means of maintaining subject confidentiality?  TOTAL ITEMS CONSIDERED (Score out of 20)  1  1 1  1 1  Table 16. Patterns of deviation within and amongst trial standard operating procedures  L FRANCIOSI  202  In reviewing Table 16, a number of deviation patterns can be observed with respect to trial standard operating procedures. 1. Five out of six trials did not have a manual of standard operating procedures. This can probably account for the lack of consideration of items 10, 11, 15, 17 and 18 in most trials. The ANA trial, which had an SOP manual, considered all items.  2. In the 330, 430, and ANA trials, there was consideration for trial monitoring. This may also account for their close scores and distinctiveness from the rest of the trials.  L. FRANCIOSI  17.4  203  Explanations for why deviations occurred from the best possible trial  Explanations for why deviations occurred vary within and amongst clinical trials. The two most common reasons in my trials were: 1. Lack of a clearly defined question 2. Lack of proper planning 17.5  Better planning of subsequent trials  Based on the knowledge gained from previous patterns of deviation, an investigator is now able to better plan any subsequent trials. If an investigator has never used the Clinical Trial Evaluation System, he or she is able to use it as a checklist to optimise the quality of their next clinical trial.  18. Future Developments of CTES  18.1  Refinement of the Clinical Trial Evaluation System  Some items in certain GCP sections could be weighted more than others based on their importance. For example, a method of randomisation may be given a score of 2 rather than a score of 1 when it has been considered.  Items in the primary question and  ethics sections may be valued more since the total number items are less than other sections.  L. FRANCIOSI  204  This system may also be improved by the removal of items that overlap in content with others. For example, a primary question can represent the scientific hypothesis that is being tested. Hence, items pertaining to the formulation and statement of a hypothesis are not necessary.  Another refinement of the Clinical Trial Evaluation System may be the replacement of the Clinical Trial Scores with a yes/no system that evaluates the presence and absence of items in a clinical trial protocol. Even though there is no evidence at this time, the possibility exists that CTES may be used inappropriately as a ratio scale when comparing total scores of clinical trials.  By removing the scoring system, the total  evaluation of a clinical trial protocol would require the user of CTES to consider the implication of each question for that particular trial. Thus, a comparison of two or more clinical trial protocols in a similar subject area could be readily done based on a weighed decision made through the Clinical Trial Evaluation System.  18.2  Determination of interrater reliability  If there is more than one person assessing the quality of a clinical trial, then it would be anticipated that the results of the Clinical Trial Evaluation System would be consistent (Moher et al., 1995). Thus, future developments will involve determining the interrater reliability of the system with more clinical trials. Unfortunately, CTES cannot be used for rating published trials because such reports lack the necessary details for answering the system's questions.  L. FRANCIOSI  18.3  205  Estimation of mean time for completion  Like any checklist or scale, the amount of time require to complete CTES is important (Moher et al., 1995). Current estimates suggest a 15 to 20 minute completion time, however, more raters and clinical trials are needed in order to estimate a mean time.  18.4 Creation of a relational database for evaluating and planning clinical trial protocols As already mentioned, an investigator can review his or her results of previous CTES assessments in order to optimise the quality of future trials. It is anticipated that that an investigator will begin accumulating CTES results in the form of a database. Thus, future developments of the system will focus on the creation of relational database that can be used as a guide for planning and evaluating clinical trial protocols.  The  computer program will consist of a question on an MS Windows screen with a yes or no box and a comment section. There may be a subsection for a total score and then a final summary section for the Clinical Trial Score and % deviation from the best possible trial. The program will also have screens for item definition, item references, and HELP.  It will be linked to database containing original protocols that can be  reviewed or re-assessed by raters. This relational database will form the basis a larger database that will be used to refine the Clinical Trial Evaluation System.  L FRANCIOSI  206  19. Potential users of CTES  19.1  Students  Students who are being trained in the design and implementation of clinical trials or who already have experience can use CTES to determine what their best possible trial could have been under the circumstances. 19.2  Clinicians  The degree of deviation in each of my trials was related to the inherent characteristics of the question being addressed.  An improvement in trial quality could have been  possible if the clinical investigators and I had planned the trials better. Thus, whether or not a clinician is experienced in clinical trials, the Clinical Trial Evaluation System can be used by the him or her to determine their previous deviations and plan their next clinical trials to optimise their quality. 19.3  Evidence-based organisations  Evidence-based  organisations  such  as  the  Clinical  Pharmacology  Research  Organisation (CPRO) (Appendix 1) can use CTES to monitor and to improve the quality of their clinical trials.  CPRO already has a committee of clinical trial experts that  reviews research questions posed by clinical investigators and then provides them with necessary training and student assistance.  Both investigators and students in CPRO  can use this system to ensure both consistency and accountability in their clinical trial work.  L. FRANCIOSI  207  BIBLIOGRAPHY  Allen, M. E., Vandenburg, M. J.: Good Clinical Practice: rules, regulations and their impact on the investigator. Br. J. Clin. Pharmac. 32: 463-465, 1991. Altman, D. G.: Statistics and ethics in medical research: Misuse of statistics is unethical. Br. Med. J. 281: 1182-1184, 1980a. Altman, D. G.: Statistics and ethics in medical research: Study Designs. Br. Med. J. 281: 1267-1269, 1980b. Altman, D. G.: Statistics and ethics in medical research: III How large a sample? Br. Med. J. 281: 1336-1338, 1980c. Altman, D. G.: Statistics and ethics in medical research: Collecting and screening data. Br. Med. J. 281: 1399-1401, 1980d. Altman, D. G.: Statistics and ethics in medical research: Analysing data. Br. Med. J. 281: 1463-1475, 1980e. Altman, D. G.: Statistics and ethics in medical research: VI Presentation of results. Br. Med. J. 281: 1542-1544, 19801 Altman, D. G.: Statistics and ethics in medical research: VII Interpreting results. Br.  Med. J. 281: 1612-1614, 1980g.  Altman, D. G. (ed.): Practical Statistics for Medical Research. London: Chapman and Hall, 1991a. Altman, D. G.: Randomisation: Essential for reducing bias. Br. Med. J. 302: 14811482, 1991b. Altman, D. G.: The normal distribution. Br. Med. J. 310: 298, 1995. Altman, D. G.: Comparing several groups using analysis of variance. Br. Med. J. 312: 1472-1473, 1996. Armitage, P.: Sequential Medical Trials, Second Edition. Oxford: Blackwell Scientific Publications, 1975.  L FRANCIOSI  208  Armitage, P., Berry, G.: Statistical Methods in Medical Research, Third Edition. Oxford: Blackwell Scientific Publications, 1994. Bankowski Z., Levine, R. J.: Ethics and Research on Human Subjects: International Guidelines. Geneva: Council for International Organisations of Medical Sciences (CIOMS), 1993. Barnett, V., Lewis, T.: Outliers in Statistical Data. Chichester: John Wiley & Sons, Ltd., 1978. Bennet, P. (ed.): Good Clinical Practice and Ethics in European Drug Research. Bath: Bath University Press, 1994. Berry, G.: Statistical significance and confidence intervals. The Medical Journal of Australia 144: 618-619, 1986. Berry, G.: Statistical significance and confidence intervals. Clinical Practice 42(11): 465-468, 1988.  The British Journal of  Bland, J. M., Altman, D. G.: Correlation, regression, and repeated data. Br. Med. J. 308: 896, 1994a. Bland, J. M., Altman, D. G.: Regression towards the mean. Br. Med. J. 308: 1499, 1994b. Bland, J. M., Altman, D. G.: One and two sided tests of significance. Br. Med. J. 309: 248, 1994c. Bland, J. M., Altman, D. G.: Some examples of regression towards the mean. Br. Med. J. 309: 780, 1994d. Bland, M.: An Introduction to Medical Statistics, Second Edition. New York: Oxford University Press, 1995a. Bland, J. M., Altman, D. G.: Multiple significance tests. Br. Med. J. 310: 170, 1995b. Bland, J. M., Altman, D. G.: Logarithms. Br. Med. J. 312: 700,1996a. Bland, J. M., Altman, D. G.: Transforming data. Br. Med. J. 312: 700,1996b. Bland, J. M., Altman, D. G.: The use of transformation when comparing two means. Br. Med. J. 312: 1153, 1996c. Bland, J. M., Altman, D. G.: Transformations, means, and confidence intervals. Br. Med. J. 312: 1079, 1996d.  L FRANCIOSI  209  Bohaychuk, W., Ball, G.: Good Clinical Research Practices: Standard Operating Procedures for Investigators, Second Edition. Hampshire: Good Clinical Research Practices, 1993. Bootman, J. L . , Larson, L. N., McGhan, W. F., Townsend, R. J.: Pharmacoeconomic Research and Clinical Trials: Concepts and Issues. DCIP. The Annals of Pharmacotherapy 23: 693-697, 1989. Borenstein, M.: The Case for Confidence Intervals in Controlled Clinical Trials. Controlled Clinical Trials 15:411-428, 1994. Bowen, W. P., Jerman, J. C : Nonlinear regression using spreadsheets. TiPS 16: 413-417, 1995. Breakwell, G. M., Hammond, S., Fife-Schaw, C : Research Methods in Psychology. London: Sage Publications, 1995. Bringberg, D., Kidder L.H. (eds.): Forms of Validity in Research. Jossey-Bass Publishers, 1982.  San Francisco:  Brown, B., Jr.: Sevoflurane: Introduction and Overview. Anesth. Analg. 81: S1-S3 1995. Burgess, E. M., Pedegana, L. R.: Controlled Environment Treatment for Limb Surgery and Trauma (A Preliminary Report). Bull. Pros. Res. 10(28): 16-57,1977. Campbell, D. T., Stanley, J. C : Experimental and Quasi-Experimental Designs for Research. Chicago: Rand McNally College Publishing Company, 1963. Carpenter, L. M.: Is the Study Worth Doing? Lancet 342: 221-223,1993. Cleland, D. I., King, W. R.: System Analysis and Project Management, Third Edition. New York: McGraw-Hill, Inc., 1983. Cohen, J.: Statistical Power Analysis for the Behavioral Sciences, Second Edition. Hillsdale: Lawrence Erlbaum Associates, Inc., 1988. Cohen, A., Posner, J., (eds.): A Guide to Clinical Drug Research. Dordrecht:: Kulwer Academic Publishers, 1995. Copi, I. M. (ed.): Introduction to Logic, Fourth Edition. Company, 1972.  New York: The MacMillan  Christensen, L. B.: Experimental Methodology, Fifth Edition. Boston: Allyn and Bacon, 1991.  L. FRANCIOSI  210  Crepeau, H.: An Examination of the Missing Data Problem in Experiments on the Effect of Drugs on Coronary Ligation in Rats. Vancouver: Helene Crepeau, 1983. (CSA/CEJA) Council on Scientific Affairs and Council on Ethical and Judicial Affairs: Conflicts of Interest in Medical Center/Industry Research Relationships. JAMA 263(20): 2790-2793, 1990. Daly, L. E., Bourke, G. J., McGilvray, J.: Interpretation and Uses of Medical Statistics, Fourth Edition. Oxford: Blackwell Scientific Publications, 1991. Davidoff, F., Haynes, B., Sackett D., Smith R.: Evidence based medicine: a new journal to help doctors identify the information they need. BMJ 310: 1085-1086, 1995. DeAngelis, C : An Introduction to Clinical Research. Press, 1990  New York:  Oxford University  Detsky, A. S., Sackett, D. L..: Establishing Therapeutic Equivalency: What is a Clinically Significant Difference? Arch. Intern. Med. 146: 861-862,1986. Detsky, A. S., Sackett, D. L..: When Was a 'Negative' Clinical Trial Big Enough? How Many Patients You Needed Depends on What You Found. Arch. Intern. Med. 145: 709-712, 1985. Doi, M., Ikeda, K.: Airway irritation produced by volatile anesthetics during brief inhalation: comparison of halothane, enflurane, isoflurane, and sevoflurane. Can. J. Anaesth. 40: 122-126,1993. Dollery, C : Clinical pharmacology: future prospects for the discipline. Br. J. Clin. Pharmacol. 42: 137-141, 1996. Drummond, M. F., Stoddart, G. L., Torrance, G. W.: Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford University Press, 1987. Drummond, M. F., Davies, L.: Economic Analysis Alongside Clinical Trials: Revisiting the Methodological Issues. International Journal of Technology Assessment in Healthcare 7(4): 561-573,1991. Einarson, T. R., Shear, N. H., Oh, P. I.: Models for Pharmacoeconomic Analysis. Can. J. Clin. Pharmacol. 4(1): 25-29, 1997. Ellis, J., Mulligan I., Rowe, J., Sackett, D. L.: Inpatient general medicine is evidencebased. A team, Nuffield Department of Clinical Medicine. Lancet 345: 407410, 1995. Essex-Sorlie, D.: Medical Biostatistics & Epidemiology. East Norwalk: Appleton & Lange, 1995.  L. FRANCIOSI  211  Feinstein, A. R.: Clinical Biostatistics XIX. Ambiguity and abuse in the twelve concepts of control. Clin. Pharmacol. Ther. 14(1): 112-122,1973. Feinstein, A. R.: Clinical Biostatistics. St. Louis: C. V. Mosby Company, 1977. Feinstein, A. R.: Clinical Epidemiology: The Architecture of Clinical Research. Philadelphia: W. B. Saunders Company, 1985. Ferrante, M. F.: Patient controlled analgesia. In: Max, M., Max, R., Laska, E., (eds.): Advances in pain research and therapy, Volume 18. New York: Raven Press, 1991. Finney, D. J.: Statistics for Biologists. London: Chapman and Hall, 1980. Freiman, J. A., Chalmers, T. C , Smith H., Kueker, R. R.: The Importance of Beta, The Type II Error and Sample Size in the Design and Interpretation of the Randomised Control Trial. NEJM 299: 690-694,1978. Friedman, L. M., Furberg, C. D., DeMets, D. L.: Fundamentals of clinical trials, Third Edition. St. Louis: Mosby-Year Book, 1996. Frink, E. J., Jr., Malan, T. P., Atlas, M.: Clinical comparison of sevoflurane and isoflurane in healthy patients. Anesth. Analg. 74: 241-245,1992. Gad, S. C , Weil, C. S.: Statistics and Experimental Design For Toxicologists. Caldwell: Telford Press, 1986 Gehlbach, S. H.: Interpreting the Medical Literature: A Clinician's Guide. Toronto: D. C. Heath and Company, 1982. George, C. F.: Clinical Pharmacology: Drug Development. Brit. Med. J. 281: 13971399, 1980. Gill, P., Dowell, A. C , Neal, R. D., Smith, N., Heywood, P., Wilson, A. F.: Evidencebased general practice: a retrospective study of interventions in one training practice. BMJ 312: 812-21,1996. Gold, H., Kwit, N. T., Otto, H.: The xanthines (theobromine and aminophylline) in the treatment of cardiac pain. JAMA 108: 2173-2179,1937. Goldstein, G.: A Clinician's Guide to Research Design. Chicago: Nelson-Hall, 1980. Good, M. D.: Cultural Studies of Biomedicine: An Agenda for Research. Med. 41(4): 461-473, 1995.  Soc. Sci.  Goodwin, C. J.: Research in Psychology: Method and Design. New York: John Wiley & Sons, Inc., 1995.  L. FRANCIOSI  212  Greenhalgh, T.: "Is my practice evidence-based?" Should be answered in qualitative as well as quantitative terms. BMJ 313: 957-958,1996. Guyatt, G., Jaeschke, R., Heddle, N., Cook, D., Shannon, H., Walter, S.: Basic Statistics for Clinicians: 1. Hypothesis Testing. Can. Med. Assoc. J. 152(1) 27-32, 1995a. Guyatt, G., Jaeschke, R., Heddle, N., Cook, D., Shannon, H., Walter, S.: Basic Statistics for Clinicians: 2. Interpreting Study Results: Confidence Intervals. Can. Med. Assoc. J. 152(2) 169-173, 1995b. Guyatt, G., Jaeschke, R., Heddle, N., Cook, D., Shannon, H., Walter, S.: Basic Statistics for Clinicians: 4. Correlation and Regression. Can. Med. Assoc. J. 152(4) 497-504, 1995c. Hamilton, M.: Lectures on the Methodology of Clinical Research. London: Churchill Livingstone, 1974 Hayes, A. W.: Principles and Methods of Toxicology. New York: Raven Press, 1982. Heard, S. O., Edwards, W. T.: Analgesic effect of intraarticular bupivacaine or morphine after arthroscopic knee surgery: A randomised, prospective, doubleblind study. Anesth. Analg. 74: 822-826,1992. Hennekens, C. H., Buring, J. E., Mayrent, S. L. (ed.): Epidemiology in Medicine, First Edition. Boston: Little, Brown and Company, 1987. Highgenboten, C. L.: Outpatient anterior cruciate ligament reconstruction and patientcontrolled analgesia. JAMA 268: 3432,1992. Hill, A. B.: Medical Ethics and Controlled Trials. Br. Med. J. 1043-1049, 1963. Hopkins, K. D., Glass, G. V., Hopkins, B. R.: Basic Statistics for the Behavioral Sciences, Second Edition. Boston: Allyn and Bacon, 1987. Hulley, S. B., Cummings, S. R.: Designing Clinical Research: Approach. Baltimore: Williams & Wilkins, 1988.  An Epidemiologic  Hunt, T. K., Rabkin, J. M.: Local Heat Increases Blood Flow and Oxygen Tension in Wounds. Arch. Surg. 122: 221-225,1987. Hvidberg, E. F.: An Historical Overview and Actual Status of Good Clinical Practice. Drug Information Journal. 28: 1089-1092,1994. Iber, F. L., Riley, W. A., Murray, P. J.: Conducting Clinical Trials. New York: Plenum Publishing, 1987.  L. FRANCIOSI  213  Johnson, M. L : Why, When, and How Biochemists Should Use Least Squares. Anal. Biochem. 206: 215-225, 1992. Johnson, F. N., Johnson, S. (eds.): Publications, 1977.  Clinical Trials.  London: Blackwell Scientific  Jolicoeur, L. M., Jones-Grizzle, A. J., Boyer, J. G.: Guidelines for performing a pharmacoeconomic analysis. Am. J. Hosp. Pharm. 49: 1741-1747, 1992. Jones, F. P.: Experimental method in antiquity. American Psychologist. 1964.  19: 419,  Joris, J. L., Dubner, R., Hargreaves, K. M.: A target for opioids released during stress and inflammation? Anesth. Analg. 66: 1277-1281,1987. Kalso, E., Tramer, M. R., Carrol, D., McQuay, H. J., Moore, R. A.: Pain relief from intra-articular morphine after knee surgery: a qualitative systematic review. Pain 71: 127-134, 1997. Kegel, B.: Controlled Environment Treatment (CET) for Patients with Below-Knee Amputations. Physical Therapy 56(12): 1366-1371,1976. Knapp, R. G., Miller, III, M. C : Clinical Epidemiology and Biostatistics. Baltimore: Williams and Wilkins, 1992 Kostyuchenok, B. M., Kerimov, M. M., Matasov, V. M., Titova, M. I., Muzykant, L. I., Goncharova, Z. G.: Effect of Controlled Abacterial Medium on the Course of Wound Process in Patients with Chronic Venous Insufficiency of Lower Limbs Complicated by Trophic Ulcer. Khirurgiia 91-95,1988. Lang, T. A., Secic, M.: How to Report Statistics in Medicine: Annotated Guidelines for Authors, Editors, and Reviewers. Philadelphia: American College of Physicians, 1997. Lashchevker, V. M., Sorochenko, L. K., Georgiyeva, G. V., Marilova, S. B.: The Treatment of Extensive Burns in Children in the Aerotherapeutic Installations. Khirurgiia 51-53, 1987. Lawrence, W., Jr.: Some Problems with Clinical Trials: James Ewing Lecture. Arch. Surg. 126: 370-378, 1991. Lerman, J.: Study design in clinical research: sample size estimation and power. Can. J.Anaesth. 43(2): 184-191,1996. Lerman, J., Sikich, N., Kleinman, S., Yentis, S.: The pharmacology of sevoflurane in infants and children. Anesthesiology 80: 814-824,1994.  L FRANCIOSI  214  Little, R. J. A., Rubin, D. B.: Statistical Analysis with Missing Data. New York: John Wiley & Sons, Inc., 1987. Lockyer, K.: Critical Path Analysis and other Project Network Techniques, Fourth Edition. Bath: The Pitman Press, 1984. Loos, F.: Research Foundations for Psychology and the Behavioral Sciences. York: HarperCollins, 1995.  New  Ludbrook, J.: Advantages of Permuatation (Randomization) Tests in Clinical and Experimental Pharmacology and Physiology. Clinical and Experimental Pharmacology and Physiology. 21: 673-686, 1994a. Ludbrook, J.: Repeated measurements and multiple comparisons in cardiovascular research. Cardiovascular Research. 28: 303-311, 1994b. Lumley, J. S. P., Benjamin, W.: Research: Some Ground Rules. New York: Oxford University Press, 1994. Makuch, R. W., Johnson, M. F.: Some Issues in the Design and Interpretation of 'Negative' Clinical Studies. Arch. Intern. Med. 146: 986-989, 1986. Mason, E. J., Bramble, W. J.: Understanding and Conducting Research: Applications in Education and the Behavioral Sciences. New York: McGraw-Hill, Inc., 1978. Mason, E. J., Bialer, M.: The Logical Structure and Validity of Experimental Design in Pharmacokinetics and Clinical Pharmacology. Biopharmaceutics and Drug Disposition. 10: 331-351,1989. McConway, K.: Studying Health and Disease. 1994.  Philadelphia: Open University Press,  McNiece, W. L.: Drug Cost Analysis in Anaesthesia. Anesthesiology Clinics of North America 14(3): 573-590, 1996. Meinert, C. L.: Clinical Trials: Design, Conduct, and Analysis. New York: Oxford University Press, 1986. Mitchael, M., Boyce, W. T., Wilcox, A. J.: Biomedical Bestiary: An Epidemiologic Guide to Flaws and Fallacies in the Medical Literature. Boston: Little, Brown and Company, 1984. Moher, D., Jadad, A. R., Nichol, G., Penman, M., Tugwell, P., Walsh, S.: Assessing the Quality of Randomized Controlled Trials: An Annotated Bibliography of Scales and Checklists. Controlled Clinical Trials 16: 62-73,1995. Montgomery, D. C : Design and Analysis of Experiments, Second Edition. Toronto: John Wiley & Sons, 1984.  L. FRANCIOSI  Morice, A.: Good Clinical Practice and the clinical pharmacologist. Pharmac. 32: 529-530,1991.  215  Br. J. Clin.  Morris, A. M., Morrison, G. W.: Controlled environment therapy in the management of lymphoedema: a case report. Br. J. Surg. 66: 785-786, 1979. Motulsky, H. J., Ransnas, L. A.: Fitting curves to data using nonlinear regression: a practical and nonmathematical review. FASEB J 1: 365-374,1987. O'Grady, J., Joubert, P. H. (eds.): C R C Handbook of Phase l/ll Clinical Trials. Boca Raton: CRC Press, Inc., 1997. Pace, N. L : Research Design and Statistics. In: Barashi, P., Cullen, B., Stoelting, R.: Clinical Anaesthesia, Second Edition. Philadelphia: J. B. Lippincott Company, 1992. Paikeday, T. M.: Collins English Dictionary: Canadian Edition. Don Mills: Wm Collins Sons & Company Ltd., 1981. Parkinson, C , McAuslane, N., Lumley, C , Walker, S. (eds.): The Timing of Toxicological Studies to Support Clinical Trials. Dordrecht: Kluwer Academic Publishers, 1994. Payton, O. D.: Research: The Validation of Clinical Practice, Edition 3. Philadelphia: F. A. Davis Company, 1994. Petitti, D. B.: Meta-Analysis, Decision Analysis, and Cost-Effectiveness Analysis: Methods for Quantitative Synthesis in Medicine. New York: Oxford University Press, 1994. Pochin, E. E.: Risk Benefit in Medicine. In: Cavalla, J. F. (ed.): Risk-Benefit Analyses in Drug Research. London: MTP Press Ltd., 1981. Pick-up, A. J., Mee, L. J., Hedley A. J.: The general practitioner in continuing education. J. R. Coll. Gen. Pract. 33: 486-490, 1983. Pocock, S. J.: Clinical Trials: A Practical Approach. Chichester: John Wiley & Sons Ltd., 1983. Raja, S. N., Dickstein, R. E., Johnson, C. A.: Comparison of postoperative analgesic effects of intraarticular bupivacaine and morphine following arthroscopic knee surgery. Anesthesiology 77: 1143-1147,1992. Rawlins, M. D.: Development of rational practice of therapeutics. 729-733, 1990.  Br. Med. J. 301:  L. FRANCIOSI  216  Redhead, R. G., Snowdon, C : A new approach to the management of wounds of the extremities; Controlled environment and its derivatives. Prosthetics and Orthotics International 2: 148-156,1978. Regnier, B.: Good Clinical Practice. Eur. J. Clin. Microbiol. Infect. Dis. 9(7): 519-522, 1990. Rosenau, M. D., Jr.: Successful Project Management: A Step by Step Approach with Practical Examples. California: Wadsworth, Inc., 1981. Rosenthal, R., Rosnow, R. L : Essentials of Behaviorial Research: Methods and Data Analysis. New York: McGraw-Hill, Inc., 1984. Rothman, K. J.: Conflict of Interest: 269(21): 2782-2784, 1993.  The New McCarthyism in Science.  JAMA  Sackett, D. L..: How to read clinical journals: V: To distinguish useful from useless or even harmful therapy. CMAJ 124:1156-1162,1981. Sackett, D. L.., Haynes, R. B., Guyatt, G. H., Tugwell, P.: Clinical Epidemiology: A Basic Science for Clinical Medicine, Second Edition. Boston: Little, Brown, 1991. Sackett, D. L., Rosenberg, W. M. C : The need for evidence-based medicine. Journal of the Royal Society of Medicine. 88: 620-624,1995. Sackett, D. L , Rosenberg, W. M., Gray, J. A. M., Haynes, R. B., Richardson, W. S.: Evidence-based medicine: what it is and what it isn't. BMJ 312: 71-72, 1996. Sayre, J. E.: G C P Quality Audit Manual, Second Edition. Buffalo Grove: Interpharm Press, Inc., 1994. Schulz, K. F., Chalmers, I., Hayes, R. J., Altman, D. G.: Empirical Evidence of Bias: Dimensions of Methodological Quality Associated with Estimates of Treatment Effects in Controlled Trials. JAMA 273: 408-412,1995. Seale, C , Pattison, S. (eds.): Medical Knowledge: Philadelphia: Open University Press, 1994.  Doubt and Certainty.  Selwyn, M. R.: Principles of Experimental Design for the Life Sciences. Boca Raton: CRC Press, Inc., 1996. Senn, S.: Cross-over Trials in Clinical Research. Chichester: John Wiley & Sons, Ltd., 1993. Shapiro, S. H., Louis, T. A. (eds.): Clinical Trials: Issues and Approaches. New York: Marcel Dekker, Inc., 1983.  L. FRANCIOSI  217  Shenfield, G. M., Hirshorn, J. E.: Good Clinical Practice in the investigation of pharmaceutical products. The Medical Journal of Australia. 154: 631-636, 1991. Shuster, J. J., CRC Handbook of Sample Size Guidelines for Clinical Trials. Boca Raton: CRC Press, Inc., 1990. Simpson, J. A., Weiner, E. S. C. (eds.): The Oxford English Dictionary, Second Edition. Oxford: Clarendon Press, 1989. Simon, R.: Confidence Intervals for Reporting Results of Clinical Trials. Ann. Intern. Med. 105: 429-435, 1986. Smith, R.: Where is the wisdom...? The poverty of medical evidence. Br. Med. J. 303: 798-799, 1991. Soccio, D. J., Barry, V. E.: Practical Logic: An Antidote for Uncritical Thinking, Fourth Edition. Fort Worth: Holt, Rinehart and Winston, Inc., 1992. Sologub, V. K., Yakovlev, G. B., Muzykant, L. I., Kaema, R. I.: The Course of the Wound Process in the Treatment of Burnt Patients in Controlled Abacterial Medium. Khirurgiia 91-95, 1983. Sox, Jr., H. C , Blatt, M. A., Higgins, M. C , Marton, K. I.: Medical Decision Making. Boston: Butterworth Publishers, 1988. Spielman, K. S. (ed.): The Barnett Self Instructional CRA Training Series. Waltham: Parexel International Corporation, 1992. Spilker, B.: Guide to Clinical Studies and Developing Protocols. Press, 1984.  New York: Raven  Spilker, B.: Guide to Clinical Studies and Developing Protocols. Press, 1984.  New York: Raven  Spilker, B.: Guide to Clinical Trials. New York: Raven Press, 1991. Spilker, B., Schoenfelder, J.: Data Collection Forms in Clinical Trials. New York: Raven Press, 1991. Sprent, P.: Applied Nonparametric Statistical Methods. Ltd., 1989.  London: Chapman and Hall,  Spriet, A., Dupin-Spriet, T.: Good practice of clinical trials. Paris: Karger, 1992. Spriet, A., Simon, P.: Methodology of clinical drug trials. Paris: Karger, 1985.  L. FRANCIOSI  218  Stedman, T. L : Stedman's Medical Dictionary, 26th Edition. Baltimore: Williams & Wilkins, 1995. Stein, C , Millan, M. J.: Peripheral opioid receptors mediating antinociception in inflammation: Evidence for involvement of mu, delta, and kappa receptors. J. Pharm. Exper. Ther. 248(3): 1269-1275,1989. Stein, C , Comisel, K.: Analgesic effect of intraarticular morphine after arthroscopic knee surgery. NEJM 325(16): 1132-1126,1991. Sutter, M. C : Assigning Causation in Disease: Beyond Koch's Prospectives in Biology and Medicine. 39(4): 581-592, 1996.  Postulates.  (TFWGAESC) Task Force of the Working Group on Arrhythmias of the European Society of Cardiology: The early termination of clinical trials: causes, consequences, and control. European Heart Journal 15: 721-738,1994. Thompson, D. F.: Understanding Financial Conflicts of Interest. NEJM 329(8): 573576, 1993. Troup, I. M.: Controlled environment treatment (CET): The use of a new concept of wound environment in amputation surgery and other conditions of the extremities. Prosthetics and Orthotics International 4: 15-28,1980. Vollmann, J., Winau, R.: Nuremberg Doctor's Trials: Informed consent in human experimentation before the Nuremberg code. BMJ 313: 1445-1449,1996. Wall, P. D., Melzack, R.: Textbook of Pain. New York: Churchill Livingstone, 1994. Webster, N.: Webster's New Universal Unabridged Dictionary, 2nd Edition. New York: Dorset and Baber, 1983. Whitehead, J.: The Design and Analysis of Sequential Clinical Trials. New York: Ellis Horwood Ltd., 1983. Willig, S. H., Stoker, J. R.: Good Manufacturing Practices for Pharmaceuticals: A Plan for Total Quality Control, Third Edition. New York: Marcel Dekker, Inc., 1992. Woods, J. R., Williams, J. G., Tavel, M.: The Two-Period Crossover Design in Medical Research. Annals of Internal Medicine. 110(7): 560-566,1989. Zolman, J. F.: Biostatistics: Experimental Design and Statistical Inference. New York: Oxford University Press, 1993.  L. FRANCIOSI  219  APPENDICES  Appendix 1 Clinical Pharmacology Research Organisation (CPRO)  220  Appendix 2  Experimental Designs in Clinical Research  221  Appendix 3  Twelve Concepts of Control in Clinical Research  229  Appendix 4  Statistical Tests in Clinical Research  234  Appendix 5  Declaration of Helsinki  236  Appendix 6  Protocol for the Clinical Evaluation of CET in Burns  239  Appendix 7  Results of the Statistical Analysis for the Analgesic Efficacy of Intraarticular Morphine Following Shoulder Surgery  241  Appendix 8  Clinical Research Protocol: Phase I Evaluation of TSB330  248  Appendix 9  Clinical Research Protocol: Evaluation of the Cost of Sevoflurane Compared to Isoflurane for General Anaesthesia in Arthroscopic Menisectomy  262  Appendix 10  Appendix 11  Clinical Research Protocol: Evaluation of Intradermal TSB430 as a Peripheral Analgesic in Subjects with Capsaicin-lnduced Pain  270  Results of the Statistical Analysis for the Evaluation of Intradermal TSB430 as a Peripheral Analgesic in Subjects with Capsaicin-lnduced Pain  279  L. FRANCIOSI  APPENDIX 1  C L I N I C A L  P H A R M A C O L O G Y  R E S E A R C H  O R G A N I Z A T I O N  D e p a r t m e n t s of A n a e s t h e s i a and P h a r m a c o l o g y & T h e r a p e u t i c s , U B C O V E R V I E W  STUDENT INVOLVEMENT  CLINICIAN INVOLVEMENT  - Literature Searches - Study Design - Data Acquisition & Analysis  - Research Proposals - Study Design - Principal Investigator  LOGISTICS FINANCING PERSONNEL  CLINICIAN BENEFITS  STUDENT-BENEFITS  - Realization of Research Ideas Through Collaboration with Research Experts & Students  Education in Clinical Research Exposure to Medical Specialties Exposure to Industry & Academia  LONG TERM GOALS • To Increase the Amount and Quality of Clinical Research Performed Within the Faculty of Medicine -To Improveithe Level of GliriicaliResearchCu.n'ding - To Make Health Care Delivery More Efficient and Effective - • - To Develpp,.CIinical Research Coordinators Am'ong GraduateStudents ;  220  L. FRANCIOSI  221  APPENDIX 2 EXPERIMENTAL DESIGNS IN CLINICAL RESEARCH  There are a number of experimental designs used in clinical research. These designs are: parallel-group, crossover, matched-pairs, sequential, factorial, adaptive, zelen and repeated mesures. Parallel Group Designs  The simplest design is called the parallel group design in which two different groups of patients are studied at the same time (Altman, 1991a; Johnson et al., 1977). The type of comparisons made with this design are as follows: •  a comparison between the treatment which the trial has been set up to test and the complete absence of treatment;  •  a comparison between the test treatment and another treatment known, or believed, to be without therapeutic effect (sometimes called a placebo treatment);  •  a comparison between the test treatment and another treatment that was established in Phase II trials to have therapeutic efficacy  •  a comparison between one form of treatment and another form - or forms - of the same treatment (ex. In Phase I and II trials, different dose levels of the same drug, different routes of administration, different regimes of treatment could be compared after their completion)  •  a comparison between the short term and the long term effects of the same treatment.  L. FRANCIOSI  222  Using a parallel group design may require a large number of patients and considerable expense. Thus, depending on type of therapy being tested, an alternative to this design is the cross-over design in which fewer patients are needed to get the same number of precise measurements. It is a design that can lead to considerable saving in resources (Senn, 1993). Cross-over Designs  In a cross-over design, patients are given sequences of treatments with the object of studying differences between individual treatments (Altman, 1991; Senn, 1993). Here, the method of randomisation is used to determine the order in which the treatments are to be received. Usually each patient undergoes a minimum of two periods/treatments as illustrated in Figure 16 (Woods et al., 1989). This design represents a within-patient rather than between-patient  comparison of the parallel group design, and it is  particularly used if the number of patients that are available to participate is small. However, there are two main disadvantages to the use of this design. Patients may drop out after the first treatment, and so not receive the next treatment; thus, making the trial incomplete. There may be a carry-over of the treatment effect from one period to the next, so that the results obtained during the second treatment are affected by what happened in the first period. However, a washout period or a time when no treatment is given is sometimes introduced between the treatment groups to try to eliminate the carry-over effects.  L. FRANCIOSI  223  Patient Sample  RANDOMISATION  PERIOD 1  PERIOD 2  Sequence 1 (Group 1)  Sequence 2 (Group 2)  Treatment A  Treatment B  CROSS-OVER  CROSS-OVER  1  1  Treatment B  Treatment A  Figure 16. The basic two-period crossover design. Patients are assigned to group 1 to  receive treatment A in period 1 and treatment B in period 2. Patients in group 2 receive the treatments in reverse order. Matched Paired Comparisons Design  The paired comparisons design is variant of the crossover design in which different treatments are evaluated in the same patients at the same time (Altman, 1991). This design is used for treatments that can be given independently to matching parts of the anatomy, such as limbs or eyes. It has all the advantages of the crossover design but none of the disadvantages, thus, making it a very powerful to use.  The nearest  equivalent to this within-patient design is the matched pair design, where pairs of patients are matched for age, sex or other characteristics which will minimise the  L. FRANCIOSI  224  variation between them, in turn, making them very comparable. The two treatments are then allocated to the pair of patients at random. Sequential Designs  This is a unique design in that the clinical trial is stopped when there is a statistically significant difference or a lack of difference between treatment groups (Altman, 1980b; Altman, 1991; Armitage, 1975; Armitage et al., 1994; Whitehead, 1983). The theory that underlies this design involves much advanced mathematical statistics. Figure 17 illustrates in a simple manner how this design works. According to the design plan, a certain number of patients are allocated to control and treatment groups and then asked whether or not they experienced some event of interest, for example nausea. The next patients are entered into the trial, for example, at rate of five per week. Then at the end of a set period of time, for example, every month, two statistics Z and V are calculated; Z represents a measure of the observed difference accumulated between groups whereas V indicates the amount of information about the true difference contained in Z. When these are plotted over time, what is obtained is called a sample path line. When this line crosses the boundary of a region indicating either a significant or a nonsignificant difference in responses between groups, the trial immediately stops. (These boundaries for regions A and B are also known as 'stopping boundaries'; prior to crossing, the line sits in what is called the triangular or continuation region). As is evident with use of this design, the clinical trial may require less patients and less time in comparison to a parallel group trial that has a fixed sample size and length. However, this design is only useful when the difference in the effectiveness of the two treatments is anticipated to be large. In terms of statistical analysis, the data from a  L FRANCIOSI  225  sequential trial is analysed after each patient's results become available. Hence, there are problems with blinding participants in the trial. But overall this experimental design is considered good and should be used whenever possible (Altman, 1991).  Figure 17. A graphical illustration of Treatments A and B rendered significantly different using a sequential design.  Factorial Designs  In this design, two (or more) independent factors, for example, treatments A and B, are simultaneously compared with each other and with a placebo control (Altman, 1991; Goodwin, 1995). Patients are divided into four groups who receive the placebo control treatment, A only, B only, and both A and B (Table 17). The design allows also the  L. F R A N C I O S I  226  investigation of the interaction (or synergy) between A and B. In this w a y two primary questions are being a n s w e r e d :  Is treatment A better than Placebo?; Is treatment B  better than placebo control? A third question concerning the interaction of A and B can also be a n s w e r e d : W h a t is t h e effect of treatment A in presence a n d a b s e n c e of treatment B? However, s o m e disadvantages may include: a poor patient compliance (i.e. patients may only want take one drug), and/or an unexpected drug interaction that m a y produce toxicity.  Treatment B  No  TreatmentB  Treatment A  No Treatment A  A and B together  B only  (Group 1)  (Group 2)  A only  No A and B; Placebo only (Group 4)  (Group 3)  Table 17.  A factorial design in which t w o treatments or factors are c o m p a r e d simultaneously to each other and with a placebo control  Adaptive and Zelen Designs Adaptive designs have been created to deal with t h e ethical d i l e m m a of randomising patients to a treatment that m a y be unbeneficial (Altman, 1991a).  T h u s , this design  causes t h e n u m b e r of subjects getting t h e inferior treatment to diminish as t h e trial proceeds. A patient's treatment d e p e n d s to s o m e extent o n t h e o u t c o m e o f treatment in previous patients in the trial. However, there are practical difficulties with this design in that the results f r o m each patient need to be immediately k n o w n .  L  FRANCIOSI  227  With a Zelen's design, the problems of getting informed consent from patients is avoided (Altman, 1991a).  In this design, patients are randomly assigned to two  therapies like the ideal trial.  However, half of the patients are given the standard  therapy, and then, they are treated as if they were not in the clinical trial. The other patients are offered the experimental therapy, but they are given the option to choose the standard therapy instead. When the trial is over, the two groups of patients are analysed as if they were originally randomised regardless of which treatment those in the second group chose. Thus, a disadvantage with design is that it is not known in advance how many patients in the second group will turn down the experimental therapy. As well, it may be unethical to not tell the first group of patients that they are in a trial. Repeated Measures Design Clinical trials may be characterised by repeated measurements of the same type, made on each of a number of patients at successive points in time (Armitage et al., 1994). Ideally, when two groups of patients are compared as in parallel group design, the resulting difference or lack of a difference between them would be due to each patient's contribution to the group total of a single measurement taken at one point in time. (These designs are called between-patients designs since all of the variation among the measurements  obtained  is based  on  individual differences  between  patients  (Rosenthal et al., 1984). Patients in each group are also said to be 'nested' within their treatment condition. In other words, they all have been observed once under a single treatment condition of the trial. A treatment condition may be a dose of drug or a type  L. FRANCIOSI  228  of intelligence test, for example). Depending on the specificity of the clinical trial's primary question, it may be necessary to take more than one measurement in patients and/or to administer two or more treatment conditions to the same patients over time. This type of design is known as a repeated measures design. When more than one measure or treatment is conducted, patients can also serve as their own controls. (Hence, when this is the case, patients are said to be 'crossed' by treatment conditions rather than 'nested' within them). A major disadvantage to the use of this design is that can produce a false answer for the question being asked (Ludbrook, 1994b). This may occur if the sequence of measurements and/or treatments are not evenly spaced in time; if the order in which treatments are given affects each others measured response; and/or if the number of measurements taken is extremely large (i.e. greater than g(n -1) where g is the number of treatments, and n is the number of patients).  One example of a repeated measures design may be taking measurements of respiratory function at three successive hours in asthmatic patients that were randomly assigned to receive different doses of a bronchodilator. These patients would then be compared over time to determine the effects of the different doses on respiratory function (Armitage etal., 1994).  L FRANCIOSI  229  APPENDIX 3 TWELVE CONCEPTS OF CONTROL IN CLINICAL RESEARCH  In clinical research, there are twelve concepts of control that should be considered (Feinstein, 1973).  The word control is usually expressed as three distinct ideas:  regulation, comparison, and the control period (Table 18).  THREE DISTINCT IDEAS OF CONTROL  CONDITIONS  Control of a Manoeuvre REGULATION  Quality Control Control of the Disease Process Control of the Environment The Type of Control Manoeuvre Size of the Control group  COMPARISON  Control of Strata Case-control The Control Value Stabilisation Period  CONTROL PERIOD  Qualification Period Washout Period  Table 18. Control expressed as three distinct ideas.  L. FRANCIOSI  The idea of regulating a  230  process  A) The control of the manoeuvre  In an experiment, the investigator "controls", governs, decides, assigns or chooses the manoeuvre (i.e. the treatment) that each person investigated will be exposed. This is not the case in an observational study where the manoeuvre is self-selected and selfreferred by patients. By controlling the assignment of the manoeuvre, the investigator can make the assignment of randomisation, thereby allow subsequent application of statistical inference based on random allocation and therefore avoid possible bias that can influence assignment. B) Quality Control  This kind of control involves determining how well a process is being performed by using statistical techniques. For a product, the assessment depends on measuring the size, shape, or some dimensional property of the finished product. If the final products have similar values for this measurement, the results are regard as consistent, and the process as having high quality. The product could also be a specimen or a laboratory measure. C) Control of the Disease process  This use of the word control refers to the adequacy with which a treatment produces certain desired effects in the activity of a patient's disease.  Some examples are  controlling blood sugar in a diabetic with insulin, producing a remission in a cancer patient with methotrexate, or reducing the disease activity in an arthritic patient with gold therapy.  L. FRANCIOSI  231  D) Control of the environment.  This refers to the investigators ability to govern the environment in which the research manoeuvres are administered.  (In reality, the clinical investigator can't control the  environment and therefore must content with the possible bias that may be caused with migration or loss of patients by non-compliance with either the prescribed manoeuvres or the schedule of the follow-up examination). The idea of  comparison  A) The control manoeuvre (or group)  As already described above, the control manoeuvre (or group) refers to the manoeuvre that is compared with the principal manoeuvre, e.g., the saline injection versus the ACTH; the placebo versus the active drug; the high dose versus the low dose; nonsmoking versus smoking. (The choice of an appropriate control is dependent on its potency (dosage or procedures with which the manoeuvres are administered), its relativity with which other comparators are chosen, its internal accompaniment provided by ancillary ingredients, the anaesthesia or concomitants of surgery, its external accompaniment provided by the persons or place where it is being administered, and its concurrency with which it precedes, parallels, or follows the principal manoeuvre). B) Size of the control group  The "control group" consists of subjects who receive the comparative (or control) manoeuvre.  The groups should have the same number of subjects so that the  possibility exists for producing 'statistically significant' results; if the groups do not have  L. FRANCIOSI  232  the same numbers, then the investigator has a scientific obligation to describe his or her reasons or justifications for the inequalities. C) Control of strata  The "control" of strata means stratifying patients according to characteristics such as age and sex that are present at baseline before the manoeuvre is imposed. This is done to achieve similarity or prevent an 'imbalance' in the groups compared. In doing so, stratification reduces bias that may occur when the manoeuvres are not randomly assigned and the randomisation does not produce equal proportions of important strata in people assigned to compared manoeuvres. D) Case-control  As discussed previously, the case-control is when the investigator looks back in time from the disease to a possible cause. In this study, the control group is disease free whereas the case group has the disease of interest. A control patient is matched to a case according to their characteristics such as age and sex. E) Control Value  The control value is the baseline value that is taken in a before and after pair of values; it is the change that occurs that is of interest. The control value is not compared to an after value; it cannot substitute for the control provided by a comparative manoeuvre. To reach a conclusion, a control manoeuvre, not merely a control value may be needed.  L FRANCIOSI  The control  233  period:  A) Stabilisation period:  The stabilisation period is a time interval that is used to allow the observed values of data to stabilise into the result that will be called the baseline or control value. B) Qualification period  The qualification period is a time interval that is used to test a person's eligibility or qualification for admission into a clinical trial. There may be bias in results because the qualification for compliance may not be representative of the population C) Washout period:  The washout period is the time period that is often necessary between successive treatments administered in a crossover trial. Problems may arise if the period is too short, or omitted when necessary.  The frequency of the washout may also be a  concern; if the primary variable is based on the a subjective measurement, then it is difficult to decide whether a placebo or no treatment should be used during washout.  L. FRANCIOSI  APPENDIX 4 STATISTICAL TESTS IN CLINICAL RESEARCH  Statistical Tests for a Relationship between Variables  234  L FRANCIOSI APPENDIX 4 (CONTINUED)  CO  235  L. FRANCIOSI  APPENDIX 5 DECLARATION OF HELSINKI  236  L FRANCIOSI  237  APPENDIX 5 (CONTINUED) Declaration of Helsinki  therapeutic value to the person subjected to the research. Recommendations guiding physicians Special caution must be exercised in in biomedical research involving the conduct of research which may human subjects affect the environment and the welfare of animals used for research must be Adopted by the 18th World Medical respected. Assembly. Helsinki. Finland. June Because it is essential that the results 1964. and amended by the 29th Worldof laboratory experiments be applied to Medical Assembly. Tokyo. Japan human beings to further scientific October 1975. 35th World Medical knowledge and to help suffering Assembly. Venice. Italy. October 1983 humanity, the World Medical and the 41st World Medical Assembly.Association has prepared the following Hong Kong. September 1989. recommendations as a guide to every physician in biomedical research involving human SUBJECTS. They should be kept under review in the Introduction future. It must be stressed that the standards as drafted are only a guide to It is the mission of the physician to physicians all over the world. safeguard the health of the people. His Physicians are not relieved from or her knowledge and conscience are criminal. civil and ethical dedicated to the fulfilment of this responsibilities under the laws of their mission. own countries. The Declaration of Geneva of The World Medical Association binds the physician with the words. "The health of my subject will be my first I. Basic principles consideration.'' and the International Code of Medical Ethics declares that. 1. Biomedical research involving "A physician shall act only in the human subjects must conform to subject's interest when providing generally accepted scientific principles medical care which might have the and should be based on adequately effect of weakening the physical and performed laboratory and animal mental condition of the subject." experimentation and on a thorough The purpose of biomedical research knowledge of the scientific literature. involving human subjects must be to 2. The design and performance of improve diagnostic, therapeutic and each experimental procedure involving prophylactic procedures and the human subjects should be clearly understanding of the etiology and formulated in an experimental protocol pathogenesis of disease. which should be transmitted for In current medical practice most consideration, comment and guidance diagnostic, therapeutic or prophylactic to a specially appointed committee procedures involve hazards. This independent of the investigator and the applies especially to biomedical sponsor provided that this independent research. committee is in conformity with the laws and regulations of the country in Medical progress is based on which the research experiment is research which ultimately must rest in performed. part on experimentation involving human subjects. 3. Biomedical research involving In the field of biomedical research a human subjects should be conducted fundamental distinction must be only by scientifically qualified persons recognised between medical research in and under the supervision of a which the aim is essentially diagnostic clinically competent medical person. or therapeutic for a subject and The responsibility for the human medical research, the essential object of subject must always rest with a which is purely scientific and without medically qualified person and never implying direct diagnostic or rest on the subject of the research, even  though the subject has given his or her consent 4. Biomedical research involving human subjects cannot legitimately be carried out unless the importance of the objective is in proportion . to the inherent risk to the subject 5. Every biomedical research project involving human subjects should be preceded by careful assessment of predictable risks in comparison with foreseeable benefits to the subject or to others. Concern for the interests of the subject must always prevail over the interests of science and society. 6. The right of the research subject to safeguard his or her integrity must always be respected. Every precaution should be taken to respect the privacy of the subject and to minimise the impact of the study on the subject's physical and mental integrity and on the personality of the subject. 7. Physicians should abstain from engaging in research projects involving human subjects unless they are satisfied that the hazards involved are believed to be predictable. Physicians should cease any investigation if the hazards are found to outweigh the potential benefits. 8. In publication of the results of his or her research, the physician is obliged to preserve the accuracy of the results. Reports of experimentation not in accordance with the principles laid down in this Declaration should not be accepted for publication. 9. In any research on human beings, each potential subject must be adequately informed of the aims, methods, anticipated benefits and potential hazards of the study and the discomfort it may entail. He or she should be informed that he or she is at liberty to abstain from participation in the study and that he or she is free to withdraw his or her consent to participation at any time. The physician should then obtain the subject's freely-given informed consent preferably in writing. 10. When obtaining informed consent for the research project the physician should be particularly cautious if the subject is in a dependent relationship to him or her or may  L FRANCIOSI  A P P E N D I X 5 (CONTINUED) consent under duress. In that case the informed consent should be obtained by a physician who is not engaged in the investigation and who is completely independent of _ this official relationship. 1 1 . In case of legal incompetence, informed consent should be obtained from the legal guardian in accordance with national legislation. Where physical or mental incapacity makes it' impossible to obtain informed consent, or when the subject is a minor, permission from the responsible relative replaces that of the subject in accordance with national legislation. Whenever the minor child is in fact able to give a consent, the minor's consent must be obtained in addition to the consent of the minor's legal guardian. 12. The research protocol should always contain a statement of the ethical considerations involved and should indicate that the principles enunciated in the present Declaration are complied with. II. Medical research combined with professional care (clinical research) 1. In the treatment of the sick person, the ph> sician must be free to use a new diagnostic and therapeutic measure, if in his or her judgement it offers hope of saving life, reestablishing health or alleviating suffering. 2. The potential benefits, hazards and discomfort of a new method should be weighed against the advantages of the best current diagnostic and therapeutic method. 3. In any medical study, every subject including those of a control group, if any • should be assured of the best proven diagnostic and therapeutic method. 4. The refusal of the subject to participate in a study must never interfere with the physician-subject relationship. 5. If the physician considers it essential not to obtain informed consent, the specific reasons for this proposal should be stated in the experimental  protocol for transmission to the independent committee (1,2). 6. The physician can combine medical research with professional care, the objective being the acquisition of new medical knowledge, only to the extent that medical research is justified by its potential diagnostic or therapeutic, value for the subject III. Non-therapeutic biomedical research involving human subjects (Non-clinical biomedical research) 1. In the purely scientific application of medical research carried out on a human being, it is the duty of the doctor to remain the protector of the life and health of that person on whom biomedical research is being carried out 2. The subjects should be volunteers either healthy persons or subjects for whom the experimental design is not related to the subject's illness. 3. The investigator or the investigating team should discontinue the research if in his/her or their judgement it may, if continued, be harmful to the individual. 4. In research on man, the interest of science and society should never take precedence over considerations related to the well-being of the subject.  238  L FRANCIOSI  239  APPENDIX 6 Clinical Trial Protocol for the: EVALUATION OF CONTROLLED ENVIRONMENT TREATMENT (CET) IN BURNS  Stage One 5 patients Criteria: Estimation of dressing  changes  Evaluation of the timing of a escar formation Evaluation of isolation  capabilities  After the first evaluation of lesions, done by on-call surgeon or resident at Emergency Room, the patient will be admitted to the burn unit ward. Prior to the beginning of a local treatment, wound culture swabs will be taken, as well as photographs of the burned areas at admission. Trunk and face are excluded from the research project. A nurse in charge will perform the first bath with hybitan soap and will put the first dressing with silver sulfadizin to cover a lesions on one limb, at the same time CET bag will be placed on the other limb (the same anatomical level). Silver sulfadiazine cream will be applied with a sterile, gloved hand and/or sterile tongue blade to a thickness of 1/16 inch. After admission patients will have daily bath unless it will be postponed, and daily dressing change with silver sulfadiazine on one side. CET bag will not be changed during first 6 days with settings T0= 32C, pressure not exceeding 15mm Hg, Humidity 45-55%, unless patient's local or general conditions will indicate otherwise. Culture swabs will be taken from both sites according to routine patient care schedule. Laboratory will do a complete blood work according to hospital standards, patients condition and decision of Burn unit director. Any additional analysis depending on the patients condition like blood gases, hematocrete, etc. will be taken on the base of a current patient condition and decision of Burn unit director or resident. None of the above are specific for CET study. For circumferential burns with risk of compartment syndrome patient will be taken to O.R. upon availability of O.R. time and escarotomy should be done by on call surgeon within first 24 hours or as soon as indicated. Regarding escar formation a concrete issues in the process of evaluation are: a blinded researcher conclusion regarding demarcation line, size and condition of a escar, a condition and size of a lesions, photographs from the same distance with the same light will be taken every day. a timing for debridement. Such factors as: availability of donors sites, patient well-being, delays relating to the hospital and not to the patient will be noted in the patients chart.(sometimes there would be a delay in timing of patient take in OR not depending upon surgeons decision). Culture swabs will be also taken right before any operation from both sites.(like routine procedure) A number of dressing change will be recorded as well as a number of a bags used for treatment of given patient. W e do not expect any additional load on nurse at burn unit ward.  L. FRANCIOSI  240  Stage T w o  Upon completion the primary evaluation of CET treatment modality on 5 patients, more intensive clinical study will be conducted according to a simple assessment of bacterial wound invasion (vice versa to wound healing). This level will include repetetive quantitative bacteriology and cytology from the lesions in order to demonstrate that there is no harm in formation of escar and wound healing in CET treatment modality. All expenses for additional bacteriological and cytological work will be paid directly to appropriate departments. Patient Population For Trial Forty individual patients will be randomized into treatment or control groups for a estimation of the effectiveness of burn treatment modality. Randomization will be in 10 lines of 4 patients per line. Patients will be identified only by an assigned number in reporting the results. Any patient may refuse to participate in the trial, or withdraw AT ANY TIME from this trial without any consequences to their care. Trial Entrance Criteria The trial will have the following entrance criteria: (a) Adult patient (over 18 years old) with second and third degree burns, from 5 to 20% TBSA (b) Patients' wounds/burns must conform to a distribution limited to sites to which uncomplicated CET compartments can be applied (limbs). Trial Exclusions The trials will have the following exclusion criteria: (a) Patients who are unable or inability to give informed consent (b) Patients with serious pre-existing illness, i.e. vascular insufficiency, insulin dependent diabetic, cancer, immunocompromized situations, etc. STATISTICAL EVALUATION: A blinded random design will be used. Analysis of Variance, ANCOVA and MANCOVA will be used as appropriate. The CET device's internal micro-processor shall be programmed and used to monitor patient healing progression and log statistical information which will be sourced in the statistical evaluation of the therapies. RESULTS The medical results of the trial will be presented at American Burn Association conference and published in the burn treatment literature. We believe the 40 patient size of the trial of the project is statistically and medically relevant. Notwithstanding this observation, the limited nature of this initial trial suggests that a more extensive multi-point trial will become appropriate if the results of the research are promising. The dynamics of the physical stages of wound healing and implications for other kinds of trauma could also be a subject of scientific articles made possible by this trial and the unique technical characteristic (the wound can be visually observed, palpated and monitored to a higher degree than previously possible) by the use of the CET Device.  L. FRANCIOSI  APPENDIX 7 RESULTS OF THE STATISTICAL ANALYSIS FOR THE THE ANALGESIC EFFICACY OF INTRAARTICULAR MORPHINE FOLLOWING SHOULDER SURGERY  241  L FRANCIOSI  APPENDIX 7 (CONTINUED) e .' a 0  n —  ca «° E "3,n•0  m ae> y> ID ro 0 ci 6 6  D  in s n a r a c o o o c o c M i o r o T - v n c o CM ^ to in in co n i q a n M o n ^ N o O T ^ d d d o d d d d r d d d  00 rg d  cn r>- m T - m T - T t T - N S O i o i o o i n CN CM n i n 1 0 ^ <D c o t o c n c N C M c n c M c o c M Td d d d d o d d d d d d d d d r  "S «=s! ffl o g , £ Si  0 0 GO M 0 T-  0.58 0.38 0.09 | | |  Morphine at 6 hour (mg/kq)  |  M 1*  19.39 15.73 3.71  |  Q. 3 — '" S I ? E <o &  | | |  6 *  H _ .Si  9.67 13.29 3.13  I m T - m m co'A. N i n i n i n ^ N v m m  ^ O ^ " o > r a . h~ 03 in 0 c j co N r*. m r*n  ^ ' • ( o J r M O ) ^ " cnr^co^cotofOcocn 00 cn 0  1  co CM  1  0  1  0  coco  cu, s  o  1  T - CM CM cn m CM e o ^ ^ ^ c n c n c n i n s CO CN CO T - CM CO C N C N C O C O C O C M T - C O C O  30.7 11.8 2.79  I  5  co  13.99 2.40 0.57  PAR OR Morphine Loading Dose Morphine Loading Dose (mg) (mq)  CL  77.5 12.6 2.98  a>  I  <° r; CM CM  CD  I  c  m  c  Z  c ,5 « a  c  •  z < u. S  STD. DEV. SEM  O  £  242  L FRANCIOSI 243 APPENDIX 7 (CONTINUED)  3 O . C  _ , ^  *-*  < 0  io«nno o n oin un <r o n v n n n - O O O O O O oooo'oooooo  co £ E  O T C0(  x: •— a. i— o 5  L I  N N  O N N  » N  " ! S  I  M  O  (  t-Ni-  M 0  N  <  O  O  ^  C  F?  O  in>-i-T-  0. o Q c o o , CO  CD c 'sz Or o "5 CD in o Q  C M  O O O O CO o  o o ~ ooo CO  CM  o  CNmir>*-'*cor~co<n, «- i - C M  o  2 0)  c  Q. k. O  .a crt 5«1 1  0 ( I D  0  (  0  3 c b c n S(Di ^ 2 o> £ ^(— c o d > !oo i «CO g fO) c e oo o^  E m 3 CJ  £coio^-oomi-eocofcocMr^oooT-oo»oo  tE  CO  e  c o z  CL  3  s  o o  k-  e  oo  0 > O T - C M C O ^ U 1 C O ^ C O O ) O ^ C M C O 1 - C M C N C M C M C M C M C M C M C M C M C O C O C O C O  S  K I CO  C O C O  S  L FRANCIOSI  244  APPENDIX 7 (CONTINUED) R A W Values  Log Transformed Values  with outliers (bold)  with outliers (bold)  (mg/Kg) Morph 6hr + 5 mg ia 1 0.349 2 0.247 3 0.470 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Mean Stdev SEM  Control 0.348 1.047  0.538 0.426 0.853 0.300 0.198 0.388 0.355 0.343 0.457 0.545  1.633  0.581 0.506 0.683 0.890 0.676 0.977 0.324 0.261 1.029 0.311 0.449 0.280 0.800 1.200 0.648 0.381  log Morph 6hr + 5mg -0.457 -0.607 -0.328 0.213  0.342 0.371 0.338 0.235 0.165 0.454 0.247  -0.236 -0.296 -0.166 -0.051 -0.170 -0.010 -0.489 -0.583 0.012 -0.507 -0.347 -0.554 -0.097 0.079 -0.255 0.247  0.058  0.058  0.925  0.090  0.020  -0.269 -0.371 -0.069 -0.523 -0.704 -0.411 -0.449 -0.464 -0.340 -0.263 -0.034 -0.465 -0.431 -0.471 -0.628 -0.783  -0.395 0.215 0.051  Power Analysis: Power Analysis: sigma prime 0.320915455 sigma prime ES 0.604319009 ES Power 41% Power TWO TAILED; AIpha==0.05  log Control -0.459  0.23178763 0.604283317 41%  TWO TAILED; Alpha=0.05  OUTLIERS - Rejection line Morph 6hr + 5 mg ia (mg/Kg) Mean Stdev % of ND for rejection] mean+/-rejection line  0.648  Control Group low  high  0.001  1.295  0.381  0.247  9 0.647  9  low  high  0.033  0.875  0.421  [log Morph 6hr + 5mg Mean Stdev % of ND for rejection mean+/-rejection line  0.454  -0.255 0.247 9 0.421  log Control low  high  -0.676  0.166  -0.395 0.215 9 0.366  low  high  -0.761  -0.029  L. FRANCIOSI  APPENDIX 7 (CONTINUED)  Raw Values with Outliers t-Test: Two-Sample Assuming Equal Variances  Mean Variance Observations Pooled Variance Hypothesized Mean Difference df tStat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail  Morph 6hr + 5mg ia Control 0.648106533 0.454171223 0.144810391 0.061163067 18 18 0.102986729 0 34 1.812957026 0.039337763 1.690923455 0.078675527 2.032243174  Log Transformed Values with Outliers t-Test: Two-Sample Assuming Equal Variances  Mean Variance Observations Pooled Variance Hypothesized Mean Difference df tStat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail  log Morph 6hr + 5mg log control -0.255197295 -0.395262693 0.061200397 0.046250613 18 18 0.053725505 0 34 1.81284995 0.039346199 1.690923455 0.078692398 2.032243174  245  L. FRANCIOSI 246 APPENDIX 7 (CONTINUED) RAW Values without outliers Morph 6hr + 5 mg ia (mg / Kg)  Log Transformed Values without outliers  Control  log Morphine 6hr + 5mg  log Control  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 MEAN STDEV  0.349 0.247 0.470 0.581 0.506 0.683 0.890 0.676 0.977 0.324 0.261 1.029 0.311 0.449 0.280 0.800 1.200  0.348 0.538 0.426 0.853 0.300 0.198 0.388 0.355 0.343 0.457 0.545 0.342 0.371 0.338 0.235 0.165  -0.457 -0.607 -0.328 -0.236 -0.296 -0.166 -0.051 -0.170 -0.010 -0.489 -0.583 0.012 -0.507 -0.347 -0.554 -0.097 0.079  -0.459 -0.269 -0.371 -0.069 -0.523 -0.704 -0.411 -0.449 -0.464 -0.340 -0.263 -0.465 -0.431 -0.471 -0.628 -0.783  0.590 0.299  0.388 0.163  SEM  0.073  0.041  -0.283 0.225 0.055  -0.444 0.172 0.043  Power Analysis: sigma prime ES  Power  0.24095 0.840298 62%  TWO TAILED; Alpha= 0.05  Power Analysis: sigma prime ES  Power TWO TAILED; Alpha=0.05  0.200287486 0.804164528 62%  L. FRANCIOSI  APPENDIX 7 (CONTINUED)  Raw Values without Outliers t-Test: Two-Sample Assuming Equal Variances  Morph 6hr + 5mg  Control  0.38771724 0.590186959 0.026425973 0.089687871 16 17 0.059077275 0 31 2.391538266 0.011515718 1.695518677 0.023031437 2.039514584  Mean Variance Observations Pooled Variance Hypothesized Mean Difference df tStat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail  Log Transformed Values without Outliers t-Test: Two-Sample Assuming Equal Variances  log Morph 6hr + 5mg Mean Variance Observations Pooled Variance Hypothesized Mean Difference df tStat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail  tog Control  -0.282725833 -0.443789925 0.050516524 0.02971363 17 16 0.040450608 0 31 2.299130828 0.014198972 1.695518677 0.028397944 2.039514584  L  FRANCIOSI  APPENDIX 8  CLINICAL RESEARCH PROTOCOL P H A S E I CLINICAL E V A L U A T I O N O F  Sponsor:  TSB330  DRUG COMPANY X Vancouver, CANADA  Study Sites: SITE 1:  SITE 2:  Dr. G (Principal Investigator)  Dr. E (Sub-Investigator)  MS Clinical Pharmacology Unit  FUC Clinical Pharmacology Unit  Department of Pharmacology Faculty of Medical Sciences PO BOX XXXX CITYX COUNTRYB  Department of Pharmacology NM street Postal Code XXXX CITY Y COUNTRY B  Study Monitor:  Mr. C, CRO COMPANY  Protocol Number:  TSB961016  Protocol Date:  16 October 1996  Emerqencv Contacts:  CITY X, COUNTRY B: CITY Y, COUNTRY B: Vancouver, Canada:  ph. ph. ph. Ph. ph. Ph.  xxx-xxx-xxx-xxx xxx-xxx-xxx-xxx xxx-xxx-xxx-xxx xxx-xxx-xxx-xxx xxx-xxx-xxx-xxx xxx-xxx-xxx-xxx  248  L FRANCIOSI  1.  249  OBJECTIVES 1.1  Primary Objective This trial is intended to demonstrate the safety of TSB330 in healthy male volunteers.  1.2  Secondary Objectives Pharmacokinetic data have already been obtained in animals. A secondary objective of this trial will be to determine the pharmacokinetics in man.  2.  BACKGROUND Confidential. This protocol is for a Phase I trial, in which 18 healthy, male volunteers will receive ascending doses of TSB330 by the intravenous route and will be monitored for 48 hours for electrocardiographic changes, with blood sampling at predetermined times to determine pharmacokinetics and to further establish drug safety and tolerance.  3.  TRIAL PLAN 3.1  Type of Trial This will be a prospective phase I open trial which will use seven doses of TSB330 in normal healthy volunteers; no more than three doses will be administered to one volunteer (depending on safety and tolerance). The trial will be divided into two stages: Stage 1 (Low doses of TSB330) with 12 volunteers and Stage 2 (High doses of TSB330) with 6 volunteers.  Stage 1: Group 1 Group 2 Group 3 Group 4 Stage 2: Group 5  First Hospital Stay  Second Hospital Stay  Third Hospital Stay  Number of Volunteers  1/1000 1/500 1/250 1/100  1/500 1/250 1/100 1/50  1/250 1/100 1/50 1/20  3 3 3 3  1/50  1/20  1/10  6  L FRANCIOSI  3.2  250  Trial Sites/Investigator This trial will be conducted at two centers:  SITE 1:  SITE 2:  Dr. G (Principal Investigator)  Dr. E (Sub-Investigator)  MS Clinical Pharmacology Unit Department of Pharmacology Faculty of Medical Sciences PO BOX XXXX CITYX COUNTRY B  FUC Clinical Pharmacology Unit Department of Pharmacology NM street Postal Code XXXX CITY Y COUNTRY B  3.3  Study Personnel and their Responsibilities  SITE 1: Name  Title  Duties  Dr. G  Principal Investigator  Directing the clinical trial at the Site 1  Dr. R  Study coordinator  Volunteer screening; Assisting in completion of Case Report Forms  Dr. A  Cardiologist  Administering infusion of TSB330; Monitoring safety of volunteer during infusion  Mr. C  Clinical Research Associate  Monitoring of the clinical trial to ensure quality  L. FRANCIOSI  251  SITE 2: Name  Title  Dr. E  Sub-Investigator  Dr. O  Study Coordinator  Dr. F  Cardiologist  Mr. C  3.4  Clinical Research Associate  Duties Directing the clinical trial at the Site 2 Volunteer screening; Assisting in completion of Case Report Forms Administering infusion of TSB330; Monitoring safety of volunteer during infusion Monitoring of the clinical trial to ensure quality  Trial Population Normal, healthy volunteers will be offered participation into this trial.  3.4.1  Trial Size A total of 18 volunteers will be given TSB330.  3.4.2  Inclusion Criteria  •  Males  •  Age: 18-35  •  Have normal histories, physical examinations, and routine laboratory test results  •  Weight and height of all participating males are ± 10% of the acceptable average of the normal population  •  Informed written consent Once evaluated as being healthy, the volunteers will be subjected to one interview for 30 minutes for a mental health evaluation, as well as to understand the emotional condition for the participation in the investigation.  3.4.3  Exclusion Criteria  •  Women  •  Males of age < 18 or > 35 years  •  Volunteers that have taken any medication for at least 2 weeks before initiation of the trial, in particular, aspirin  •  Volunteers positive for Hepatitis B, Hepatitis C or HIV (Confidential)  •  Inability to give informed consent  L FRANCIOSI  •  252  The results of the laboratory tests are not within the limits considered normal (within the mean ±10%) unless the investigator considers that they are not clinically significant  •  Participating in other experimental drug studies within 3 months of this study will be excluded  •  A history of drug and/or alcohol use, or has ingested alcohol within 48 hrs prior to the clinical study  •  Hospitalized for any reason within the 8 weeks after the start of the study  •  A history of hepatic disease, renal disease, cardiac disease, hematopoietic disease, or episodes of epilepsy  •  A history of cardiac disease or symptoms of cardiac origin  •  Hypotension or hypertension of any etiology requiring pharmacological treatment  •  Myocardial infarction, angina pectoris and/or congestive heart insufficiency  •  Back flow or evidence of mitral valve prolapse detected during clinical examinations and/or ECG Bi-doppler tests  •  An arterial pressure below 90/50 mmHg (in the supine position) and/or a heart rate of 55 beats/min during the clinical examination  •  A fall in arterial systolic pressure greater than 20 mmHg in the orthostatic position and an increasing heart rate of 20 beats/min after standing for 2 minutes  •  Loss of more than 450 ml of blood within three months prior to the study  •  A history of significant hypersensitivity to any kind of drug  •  Any other condition which would make it impossible for participation  Trial Medication Details  3.5.1  Treatment Assignment Each volunteer will receive three doses of TSB330. Only one dose will be administered to the volunteer in one week. Each subsequent dose will be two times the concentration of the last, and will be administered providing the previous dose was clinically safe. The safe dose has been determined, on the basis of animal studies, to be 0.5 pmole/kg (0.203 mg/kg). The maximum dose to be given to humans will be 1/10 of the safe dose or 0.05 pmole/kg (0.0203 mg/kg). The drug will be administered by a slow bolus infusion over 15 minutes.  L. FRANCIOSI  3.5.2.  253  Dose Calculations  For a 70 kg volunteer : Code  Dose (mg/kg)  Total Dose (mg)  ml of saline to be removed from a 50 ml saline bottle  ml of 0.04 mg/ml stock solution to make up the 50 ml saline/drug mixture  Resultant 50 ml solution made from stock (mg/ml)  volume (ml) to be infused using 0.5 (ml/kg)  flow rate over 15 minutes (ml/min.)  0.014 0.028 0.056 0.14  0.5 1.0 2.0 5.0  0.5 1.0 2.0 5.0  0.0004 0.0008 0.0016 0.004  35 35 35 35  2.3 2.3 2.3 2.3  0.28 0.7 1.4  10.0 25.0 -  10.0 25.0 -  0.008 0.02 0.04  35 35 35  2.3 2.3 2.3  Stage 1 study 1/1000 1/500 1/250 1/100  0.0002 0.0004 0.008 0.002  Stage 2 study 1/50 1/20 1/10  0.004 0.01 0.02 3.5.3.  Trial Medication Supplies TSB330 (molecular weight: 406 grams/mole) will be supplied as a sterilized solution in a vial containing 2 mis, at a concentration of 1.0 mg/ml, for each volunteer. The trial medication will be directly supplied by Drug Company X.  3.6  Dietary Restrictions The volunteers will be first hospitalized the day before the start of the study and if they decide to enter, they will be assisted and cared for by medical and nursing staff. They will all receive a general hospital diet until 2300 hours (11 PM) when they will begin to fast until the clinical previsit test at 0700 hours (7 AM) of the following day. The volunteers will remain fasting from solid food for 4 hours after an infusion of TSB330 at which time they will be allowed to break and receive a general diet.  3.7  Safety Assessment Procedures The safety of TSB330 will be assessed by clinical observation, physical examination, electrocardiogram, vital signs (including blood pressure and heart rate) and clinical laboratory evaluation. Physical examinations, electrocardiograms, and clinical laboratory evaluations will be performed before dose administration, 1 to 6 hours after dosing, and at study closing in the supine position. Vital signs will be performed before dosing and at 0, 0.5, 1, 2, 4, 8, and 24 hours post-dose. All symptoms or adverse experiences will be recorded after dose administration.  3.8  Specimen Collection Twelve, 10 ml heparinized serum plasma samples will be collected at the time 0, 5 min, 10 min, 15 min, 30 min, 1 hr, 2 hr, 4 hr, 12 hr, 24 hr, and urine samples will be taken for a period of 24 hrs (intervals of 0-2 hrs, 2-4 hrs, 4-8 hrs, 8-12 hrs, 12-24 hrs). TSB330  L. FRANCIOSI  254  plasma and urine concentrations will be determined by high pressure chromatography (HPLC) and the characterization of metabolites through chromatography coupled to a mass spectrometer (LC-MS/MS).  liquid liquid  Narcotic Effects Evaluation (Stage 2) During Stage 2, measurements of the following pharmacodynamic parameters will be made by the volunteer and/or the observer at baseline (within 30 minutes prior to dosing), 10 min, 30 min, 1 hr, 4 hr, and 12 hr post-dose: a.  Mood (measured by VAS)  b.  Sedation (measured by VAS) VAS (Visual Analog Scale) measurements will be on a 10 cm scale anchored on one end as Worst Mood and the other end as Best Mood or Asleep and Awake for sedation.  c.  Respiratory Rate - Breaths/minute  d.  Pupil Size - measured by a pupilometer (pupil gauge). Only the left eye will be measured at all times. In an assessment, the volunteer will be asked to focus on an object across the room. The observer will place the pupil gauge on the lateral aspect of the volunteer's left eye, move the gauge up or down to locate the circle on the gauge corresponding to the size of the pupil, select one of the millimeter circles on the gauge and record that number. Each assessment (for each volunteer) will be done in similar light conditions, i.e. preferably in a darkened room.  e.  Specific Drug Effect Questionnaire (measured by VAS). The questionnaire consists of 10 items to be rated by the volunteer and observer. The items are related to signs of opiate agonist drugs. The volunteer will rate each item by placing a vertical mark along a 10 cm visual analog scale (VAS) anchored on one end by 'not at all' and at the other end by 'an awful lot'. The observer will rate each item by placing a vertical mark along a 10 cm VAS anchored at one end by 'not at all' and the other end by 'extremely'. Measurements to be made by volunteer will be obtained under the supervision of a clinical assistant at each time point indicated above. The clinical assistant should take precautions to prevent the volunteers from disclosing their ratings to other volunteers. The results of this questionnaire will not be considered adverse experiences, per se. Adverse experiences will be those spontaneously reported by the volunteer. Volunteer questions: 1)  How much of the effects of the drug do you feel?...On the VAS scale, not at all or an awful lot?  2)  How itchy does your skin feel?...On the VAS scale, not at all or an awful lot?  3)  How relaxed do you feel?...On the VAS scale, not at all or an awful lot?  4)  How sleepy do you feel?...On the VAS scale, not at all or an awful lot?  5)  How drunk do you feel?...On the VAS scale, not at all or an awful lot?  6)  How nervous do you feel?...On the VAS scale, not at all or an awful lot?  7)  Are you full of energy?...On the VAS scale, not at all or an awful lot?  L. FRANCIOSI  8)  How much of a need to talk do you feel?...On the VAS scale, not at all or an awful lot?  9)  How sick do you feel?...On the VAS scale, not at all or an awful lot?  10)  How dizzy do you feel?...On the VAS scale, not at all or an awful lot?  255  Observer's questions: How much does the volunteer:  f.  1)  show any drug effect?...On the VAS scale, not at all or extremely?  2)  scratch?...On the VAS scale, not at all or extremely?  3)  look relaxed?...On the VAS scale, not at all or extremely?  4)  look drunk?...On the VAS scale, not at all or extremely?  5)  look nervous?...On the VAS scale, not at all or extremely?  6)  talk?...On the VAS scale, not at all or extremely?  7)  show any sign of vomiting?...On the VAS scale, not at all or extremely?  8)  look confused?...On the VAS scale, not at all or extremely?  9)  look restless?...On the VAS scale, not at all or extremely?  10)  show any sign of perspiring?...On the VAS scale, not at all or extremely?  Elicited Opioid Effects - The presence or absence of the following adverse signs will be noted at baseline and between 0 - 0.5 hours, 0.5 - 1 hours, 12 hours, and 2 - 4 hours: nausea somnolence vomiting dizziness pruritus constipation  If any of these opioid effects are present they will be rated using the following scale: 0 1 2 3  - None - Awareness of symptom but easily tolerated - Discomfort - interferes with usual activity - Incapacitating - inability to do usual activity  The use of this "elicited opioid effect" questionnaire is independent of the Adverse Reaction case report form. As such, their use is mutually exclusive and independent. 3.10  Dropouts and Trial Completion For any volunteer who decides to drop out of the trial, their reasons will be recorded and reported to DRUG COMPANY X Completion of the trial according to protocol will require compliance and adherence to the trial schedule.  L. FRANCIOSI 4.  TRIAL FLOWSHEET / TIMELINE  256  L. FRANCIOSI  257  L FRANCIOSI  258  PROBLEM MANAGEMENT  5.1  Infusion of TSB330 must be stopped if:  •  Bradycardia occurs during infusion at less than 55 beats per minute  •  Blood pressure falls more than 20 mmHg systolic and/or diastolic  •  Significant PR prolongation occurs *  •  ECG diagnostic criteria are not met  •  QRS widens more than 15% from control value *  •  QT interval is prolonged by more than 10% *  •  ST segment elevation is seen *  •  Any abnormal waves in ECG  •  Feelings that are intolerable by volunteers  (*) Please note that standard values for PR, QRS, QT, and ST time intervals can be found in Ferrer, M.: Electrocardiographic Notebook. 4th ed. New York, Futura Publishing Co., 1973. If standard values from the Brazilian population exist, please use these values for this trial. 5.2  Emergency Contacts  For the purposes of routine trial management, the following study monitor should be contacted:  Study Monitor:  Mr. C CRO COMPANY. ph.: XXX-XXX-XXXX  For the purposes of urgent consultation, in the event that the trial monitor is unavailable, the emergency contacts are:  Investigator: (Site 1)  Dr. G ph.: X X X - X X X - X X X - X X X ph.: XXX - XXX - XXX - XXX ph.: XXX-XXX-XXX-XXX  (Home) (Office) (Home FAX)  L. FRANCIOSI Investigator: (Site 2)  Dr. ph. ph. ph. ph.  DRUG COMPANY X:  XXX XXX XXX XXX  -  XXX XXX XXX XXX  -  XXX XXX XXX XXX  -  XXX XXX XXX XXX  Dr. M Senior Scientist ph.:X-XXX-XXX-XXXX ph.:X-XXX-XXX-XXXX ph.: X - XXX - XXX-XXXX ph.:X-XXX-XXX-XXXX  259  (Home) (Office) (Office) (FAX)  (Home) (Nortran Office) (Personal Office) (FAX)  Adverse Experiences Details of all adverse experiences (AEs) encountered during the trial will be reported in the Case Report Forms (CRFs) - regardless of perceived relationship to the investigational new drug (TSB330). Each experience will be assessed by the investigator as to its potential relationship to investigational new drug. Appropriate action will be taken by the investigator, if indicated, and reported. 5.3.1  Serious, Unexpected or Alarming AEs Any volunteer experiencing a serious, unexpected or alarming clinical AE will immediately be withdrawn from the trial medication, if appropriate. Detailed follow-up and other appropriate measures will be instituted and documented. When possible, plasma sample(s) should be obtained, for determination of the level of the investigational new drug. The sponsors will be notified, by telephone, of all such serious, unexpected or alarming AEs regardless of presumed causal relationship to investigational new drug. The investigator must be prepared to supply the sponsors with following information, at the time of first reporting of severe, unexpected or alarming AEs: a)  Study Code  b)  volunteer's initials and code number  c)  date of infusion  d)  type of AE, and time and date of onset  e)  details of AE, including severity and progression  f)  date and apparent cause of death, if applicable  g)  the investigator's opinion regarding potential relationship of AE to the investigational new drug, and rationale supporting this assessment  L. FRANCIOSI  5.3.2.  260  Follow-up on Serious, Unexpected or Alarming AEs Periodic reporting of the volunteer's subsequent course must be provided, until its resolution or until otherwise notified by the sponsors. These various reports and the completed CRF are to be provided to DRUG COMPANY X as expeditiously as possible.  5.4  Protocol Adherence The investigator must sign a Trial Agreement Letter confirming full agreement with the terms of the approved protocol and with any subsequent approved amendment(s) to such protocol. It will be the investigator's responsibility to ensure that the trial is conducted according to the protocol, and that all efforts are made to ascertain that the trial participants are complying with the requests made of them for the trial. Any subsequent protocol amendments must be agreed to by DRUG COMPANY X and the investigator.  ADMINISTRATIVE 6.1  Review and Consent Requirements 6.1.1  Regulatory Approval DRUG COMPANY X will supply all background data necessary for submission to the regulatory bodies.  6.1.2  Informed Consent The investigator will explain the nature of the trial, and the risks which can reasonably be foreseen, to each participant prior to entry into the trial. It should be made clear that the participant is free to withdraw from the trial at anytime for any reason, without jeopardizing any subsequent treatment by the investigator. It should also be made clear that all medical information relevant to volunteers will be kept confidential. Written informed consent will be obtained from each participant, prior to entry into the trial.  6.2  Observers Dr. M, Mr. H and Mr. Lui Franciosi will visit the trial center and observe the technical and data recording aspects of the trial.  6.3  Data Recording Case Report Forms (CRFs) will be used to collect the necessary data for each volunteer. These are to be kept as up-to-date as possible, to allow prompt review by the study monitor and trial observer. The investigator must provide a triplicate copy of each completed CRF. A CRF will be completed for all volunteers receiving the investigational new drug TSB330. Although designated trial staff may enter data into the CRFs, it is the responsibility of the investigator to ensure that the recorded data are accurate and complete. Each set of  L. FRANCIOSI  261  CRFs will contain a signature page which must be signed by the technician on duty signifying review of, agreement with and responsibility for the data contained therein. Corrections of any errors or omissions in the entries must be initialed and dated; the original entry should be crossed out but remain legible. All entries are to be in black ink, and no correction fluid is to be used. No blank sections should be left on the CRFs; missing data should be identified as: N.D. N. Av. N. Ap. 6.4  = (not done) = (not available) = (not applicable)  as appropriate.  Record Keeping The investigator must maintain or have access to separate independent source records (see Section 6.2) of the data contained in the CRF (i.e. progress notes, laboratory reports, medication records, x-rays). These source documents should made available to the study monitor and trial observers during the trial period, to resolve any problems during CRF review or following completion of the trial. Following completion of the trial, all independent records and copy of each participant's signed consent form and completed CRF must be retained by the investigator for 15 years.  6.5  Monitoring Monitoring of the trial will be carried out by Mr. C in order to ensure that the trial is carried out according to highest standards. It must be standard practice that the investigator be in contact on a regular basis to discuss the trial's progress, to review CRFs - reference being made to source documents where necessary - and to ascertain that written consents are being obtained and drug inventories maintained, as required by the protocol. The investigator will be observed in the process of conducting the trial. Completed CRFs or CRF sections will also be turned over to CRO COMPANY and DRUG COMPANY X at appropriate and confirmed times.  6.6  Termination of the Trial The maximum number of volunteers to be studied is as specified in Section 3.3.1. DRUG COMPANY X reserves the right to discontinue the trial at any time, if deemed necessary, for medical or administrative reasons. Should such action be required, reimbursement will be made for all reasonable expenses already incurred by the investigator.  6.7  Trial Confidentiality The investigator shall agree to the following terms of confidentiality: a)  All unpublished information given to the investigator by DRUG COMPANY X shall be kept strictly confidential and shall not be published or disclosed to any third party without the prior written consent of DRUG COMPANY X.  b)  Details of any planned public disclosure (e.g. text of any lecture, or copy of an abstract or manuscript for publication) arising from the trial will be supplied to the DRUG COMPANY X at least four weeks before presentation or submission. This is not intended to limit the rights of the investigator to publish such information, but rather to enable the sponsors to make constructive comments about the text or manuscript.  L FRANCIOSI  APPENDIX 9  CLINICAL RESEARCH PROTOCOL EVALUATION OF THE COST OF SEVOFLURANE COMPARED TO ISOFLURANE FOR GENERAL ANAESTHESIA IN ARTHROSCOPIC MENISECTOMY  Investigators:  Dr. X, M.D. Mr. Franciosi, B.Sc. Mr. A, B.Sc.  Protocol Number:  XXXXX961213  Protocol Design:  Prospective; Randomized.  Protocol Date:  13 December 1996 (FINAL)  262  L. FRANCIOSI  1.  263  OBJECTIVES 1.1  Primary Objective To compare the drug costs associated with sevoflurane anaesthesia compared to isoflurane anaesthesia.  1.2  Secondary Objectives To establish the effect of sevoflurane and isoflurane on readiness for discharge from the recovery room.  2.  BACKGROUND Sevoflurane has recently been introduced as an inhalational agent that may reduce recovery time. W e propose to establish the cost of a sevoflurane anasesthetic in a specific type of surgery with a number of different anaesthetists, practising as they would normally.  3.  STUDY PLAN 3.1  Type of Study This will be a prospective, randomized study. The administering anaesthetist will administer the inhalational agent as clinically indicated. To minimize the variance due to individual anaethetists, the patients will be randomly assigned in lines of 2, to either group. The number of cases per day is anticipated to be more than 2 and less than 4.  3.2  Investigator This study will be conducted at one center: direction of the principal investigator.  3.3  Vancouver Hospital UBC site under the  Study Population Suitable patients undergoing arthroscopic menisectomy are to be offered participation in this study. Patients will be recriuted in the surgeons office or in the daycare. Informed consent will be obtained. 3.3.1  Study Size A total of 40 patients will be randomly assigned to receive either sevoflurane or isoflurane.  3.3.2  Inclusion Criteria •  3.3.3  Patients undergoing arthroscopic menisectomy at Vancouver Hospital UBC Site.  Exclusion Criteria • • • •  Age: < 18 or > 45 yr. Any illness other than surgical indication Inability to give informed consent Postoperative intraarticular drainage  L FRANCIOSI  3.4  264  Study Medication Details 3.4.1  Treatment Assignment During the randomized treatment period, patients will receive either sevoflurane or isoflurane as determined for 20 lines of 2 patients. The hospital patient identification label will be placed on the patient study record.  3.4.2  Anaesthetic Management Anaesthetic management will be determined by the anaesthesiologist except that the patient will be assigned to one of two groups, recieving either isoflurane or sevoflurane. It is anticipated that the anesthetic will generally consist of a short acting narcotic, propofol induction, laryngeal mask and nitrous oxide, oxygen and inhaltional agent. The surgeon will administer bupivacaine into the knee as is standard practice in our institution.  3.4.3.  Study Medication Supplies Isoflurane and sevoflurane will be administered from the vaporizors on the OHMEDA CD anaesthetic machine equipped with version 4.XX software.  3.5  Cost Assessment Procedures 5.1  Outline Patients will be randomly assigned to either: Group S (sevoflurane) or Group I (isoflurane) The study monitor will fill out a standard form which identifies the patient by study identification number and record the number of ampules or ml of each adjuvant drug.  On the study anaesthetic machine (in OR 5) there will be three vapourizers, one standard isoflurane, one study isoflurane and one study sevoflurane. The volume of isoflurane and and sevoflurane will be recorded over the study period. The volume remaining in the vaporizer at the end of the study will be determined and subtracted from the total. The costs of the total drugs will be determined from the costs of the inhalational and adjuvant drugs. Thes costs will be compared. The cost of anaesthetic adjuvant drugs admininstered in the PAR will be deterimed. The incidence of nausea and vomiting will be noted. The patient will be evaluated for fitness, for discharge from PAR, every 15 minutes. The earliest time that discharge criterion are met will be noted. Readiness for discharge from the hospital will be determined and noted. The amount of isoflurane used for each patient will be detrmined using the CD modulus version 4.xx. In this version sevoflurane is included for infrared analysis. The inspired concentration of the inhaled agent will be recorded on floppy disc. This data will be exported to Excel as an ASCII file. The inspired flow volume will be recorded on the anaesthetic record by the anaesthetist and  L. FRANCIOSI  265  transferred to the spreadsheet by the study monitor. The integral of the time, fresh gas flow volume and inspired concentration will be determined to give the isoflurane and sevoflurane used. The percent of inhaled agent used in each case will be established. The cost of the agent for each patient will be determined from the percentage of the total cost of the agent established from the total volume of agent used.in all patients. The cost of the inhaled agent will be added to the cost of the adjuvant agents to establish the total cost for each paitent and the variance in the cost between patients in each group. This mean cost and variance will be compared between groups.  Dropouts and Study Completion For any patient who decides to drop out of the study, their reasons will be recorded and reported to XXXXX.  L. FRANCIOSI  STUDY FLOWSHEET  Obtain consent  Assign patient to predetermined group  Group S (sevoflurane)  Group I (isoflurane)  i  Standard anaesthetic  Arthroscopic menisectomy  4-  Transfer to PAR Transfer the adjuvant drug use from the anaesthetic record to a spreadsheet  Transfer the case data from the anaesthetic machine to a spreadsheet  266  L FRANCIOSI  267  PROBLEM MANAGEMENT 5.1  Emergency Contacts For the purposes of routine study management, the study monitor should be contacted. For the purposes of urgent consultation, the emergency contacts may be reached at:  Contact # 1 :  Contact #2:  5.2  Dr. X Principal Investigator Direct dial: (604)  xxx-xxxx  Mr. Franciosi Direct dial:  xxx-xxxx  (604)  Adverse Experiences Details of all adverse experiences (AEs) encountered during the study will be reported in the case report forms (CRFs)-regardless of perceived relationship to study medication. Each experience will be assessed by the investigator as to its potential relationship to study medication. Appropriate action will be taken by XXXXX, if indicated, and reported. 5.2.1  Serious, Unexpected or Alarming AEs Any patient experiencing a serious, unexpected or alarming clinical AE will immediately be withdrawn from the study, if appropriate. Detailed follow-up and other appropriate measures will be instituted and documented.  XXXXX will be prepared to supply the following information, at the time of first reporting of severe, unexpected or alarming AEs: a)  protocol title and number  b)  patient's initials and study code number  c)  patient's gender and date of birth  d)  date of study medication administration  e)  condition(s) being treated with study and/or concomitant medication(s)  f)  type of AE, and time and date of onset  g)  details of AE, including severity and progression  h)  XXXXX's opinion regarding potential relationship of AE to study medication, and rationale supporting this assessment  L. FRANCIOSI  6.  268  ADMINISTRATIVE 6.1  Review and Consent Requirements 6.1.1.  Informed Consent XXXXX staff will explain the nature of the study, and the risks which can reasonably be foreseen, to each participant prior to entry into the study. It should be made clear that the participant is free to withdraw from the study at anytime for any reason or will be withdrawin by the investigator during general anaesthesia if necessary, without jeopardizing any subsequent treatment by the investigator. It should also be made clear that all medical information relevant to patients will be kept confidential. Written informed consent will be obtained from each participant, prior to entry into the study. The patient will receive a copy of the signed consent form; the original is to be retained by XXXXX. The consent form used for this study will be written by the UBC Human Experimental Ethics Committee.  6.2  Observers Properly trained and qualified technicians will assist in the technical and data recording aspects of the study.  6.3  Data Recording A CRF will be completed for all patients . Although designated study staff may enter data into the CRFs, it is the responsibility of XXXXX to ensure that the recorded data are accurate and complete. Each set of CRFs will contain a signature page which must be signed by the anaesthetist signifying review of, agreement with and responsibility for the data contained therein. Corrections of any errors or omissions in the entries must be initialed and dated; the original entry should be crossed out but remain legible. All entries are to be in BLACK ink, and no correction fluid is to be used. No blank sections should be left on the CRFs; missing data should be identified as: N.D. = ( n o t done) N. Av. = (not available) N. Ap. = (not applicable)  6.4  as appropriate.  Study Supplies 6.4.1  Study Medications  Sevoflurane and isoflurane are available at the hospital.  7.  STATISTICAL ANALYSIS 7.1  Cost Evaluation 7.1.1  The cost of adjuvant drugs and the inhalational agent will be calculated from the drugs used. The amount used at induction versus maintenance will be analyzed by integrating the concentration, minute volume over the total anaesthetic time.  7.1.2.  The cost of adjuvant drugs in PAR will be calculated.  L FRANCIOSI  7.2  Discharge Evaluation A standard discharge criterion will be determined every 15 minutes after entry to PAR. The time to readiness for discharge from PAR will be compared .  8. 1.  REFERENCES Aldrete, J & Kroulik, D. A postanesthetic recovery score. Anesth Anlag, 49: 6, Nov/Dec, p.926.  269  L. FRANCIOSI  APPENDIX 10  CLINICAL RESEARCH PROTOCOL EVALUATION OF INTRADERMAL TSB430 AS A PERIPHERAL ANALGESIC IN SUBJECTS WITH CAPSAICIN-INDUCED PAIN  Investigators:  Dr. X, M.D. D. Y, M.D. Mr. Franciosi, B.Sc.  Sponsor:  DRUG COMPANY X Vancouver, BC  Protocol Number:  TSB951031b  Protocol Design:  Randomized and Double-blind  Protocol Date: This Version:  26 October 1995 31 October 1995  270  L. FRANCIOSI  1.  271  OBJECTIVE 1.1  Primary Objective This study is intended to test the concept of selective blockade of peripheral nociceptors. Local peripheral analgesia without central effects is the desired outcome.  2.  BACKGROUND There is a need for analgesics without the side effects of those presently available. Narcotics have central actions, including respiratory depression, addiction potential and sedation. NSAIDs are of limited efficacy and produce significant bleeding tendencies. A series of compounds have been discovered which act in a novel manner to provide analgesia. These compounds, which act on the fine nerve afferents are termed selective nociceptor blockers, or nociblockers. One of these blockers is TSB430. TSB430 is in common clinical use as an anti-emetic and tranquilizer and is known to have limited toxicity. Because of its actions within the CNS, it is unlikely to be clinically useful as a nociblocker. However, it is a safe agent for human testing; from other studies it is clear that nociceptor blockade is a common feature of a variety of drugs with varying specific actions. In this study, healthy subjects will be injected intradermal^ with a TSB430 followed later by capsaicin, a known pain producing substance for intradermal injections, in order to demonstrate nociceptor blockade.  3.  STUDY PLAN 3.1  Type of Study This will be a randomized, double-blind study, comparing four doses of TSB430 with a saline control and a positive control (lidocaine).  3.2  Investigator This study will be conducted at one center under the direction of Dr. X, principal investigator.  3.3  Study Population Suitable subjects are to be offered participation in this study. Subjects will be enlisted on the day of the study in which all intradermal injections will be made. Informed consent will be obtained. 3.3.1  Study Size A total of 6 subjects will have all injections in their forearms assigned according to a Latin Square Design.  3.3.2  Inclusion Criteria Subjects over the age of 18.  L. FRANCIOSI  3.3.3  272  Exclusion Criteria Subjects taking medication for a pre-existing illness Inability to give informed consent Students in the laboratory of Dr. X or Dr. Y)  3.4  Study Medication Details 3.4.1  Treatment Assignment The six treatments to which each subject will be exposed will be randomized for each subject (Latin Square Design).  3.4.3.  Study Medication Supplies Study medication will consist of appropriately labeled vials, each containing TSB430, saline, or lidocaine as determined by the randomization table.  3.5  Procedures Fifty microliters ( u l ) of a test substance will be injected intradermally at a site on the volar surface of the non-dominant forearm. At one and half minutes postinjection, 10 LIL of capsaicin solution will be injected into the middle of the wheal area of the previous injection. A pain intensity rating will then be immediately obtained. Pin prick tests in the wheal area will be done at 1, 2, 3, 4, and 5 minutes after injection of the test substance. In total, each subject will receive 12 injections, four of which will be four different concentrations of TSB430, one injection of saline, one injection of lidocaine with one injection of capsaicin for each treatment and control. 3.5.1.  3.5.2.  Pain Measurement (a)  Pain produced by intradermal injection of capsaicin will be evaluated on a 0 to 10 integer scale, immediately.  (b)  For standardization, to establish a modulus of pain prior to each injection, a heated (50 ° C ) brass rod will be placed on the uninjected volar surface of the forearm for five seconds and the subject told to regard this as a five on the integer scale. This will be done prior to the injection of the test substance and also the capsaicin. At five minutes, the 5 0 ° C brass rod will be placed beside the wheal and the subject told to regard this as a five on the integer scale. The rod will then be placed directly on the wheal and the subject will be asked to evaluate the pain on the 0 to 10 integer scale compared to the standard.  (c)  At 1, 2, 3, 4, and 5 minutes postinjection, a pin prick test will be performed in the wheal area. The subject will be asked whether or not he felt any sharp pain in the wheal area after a pin prick. The responses, YES or NO, will be recorded.  Drug Dosage Capsaicin:  1 ng (2.8 pmol) in 10 LIL  TSB430:  30, 100, 300, 1000 L i g / m L (1.5, 5, 15, and 50  Lidocaine: Saline:  Ltg, each in a volume of 50 u l ) 1000 u.g/mL (50 ng/50 LIL injection) Sterile normal saline (0.9%)  L. FRANCIOSI  3.5.3.  273  Injection Sites Six areas in 3 rows of 2 from proximal to distal forearm (3.75 cm apart). These sites will be :  proximal a y  medial  p 5  lateral  (j)  S  distal 3.5.4.  Injection Sequence Injections will be made at minimum intervals of five minutes, in the sequence a, (3, y, 8, C, and <|>, so that timing of A, B, C, D, E, and F are altered systematically according to the Latin Square Design.  Example:  Subject 2 will receive his first injection B at site a , then injection C at site p\ and etc.  Injection site  a P  Y 5 8  Subjects  1  2  3  4  5  6  A B C D E F  B C D E F A  C D E F A B  D E F A B C  E F A B C D  F A B C D E  Study Completion Completion of the study according to protocol will require compliance with the study medication regimen, as well as adherence to the study schedule.  L. FRANCIOSI  274  STUDY FLOWSHEET  Obtain consent Fill out Appendix C - Medical/Illnesses Report Subject's nondominant forearm assigned to 6 treatments and injection sites (according to the Latin Square Design) STEP 1:  Establish a pain modulus on forearm (standard 50 X temperature probe with a pain score = 5/10).  STEP 2:  Perform intradermal injection of test substance at the a position.  STEP 3:  One minute and twenty seconds postinjection, establish a pain modulus beside the wheal (standard 50 °C temperature probe with a pain score = 5/10).  STEP 4:  One and a half minutes postinjection, inject 10 u.L of capsaicin solution into the middle of the wheal of the previous injection; obtain an immediate pain rating on a 0 to 10 integer scale (no pain to 10, worst pain imaginable) and enter score into the case report form (Appendix A).  STEP 5:  At 1, 2, 3, 4, and 5 minutes postinjection, perform a pin prick test in the wheal area. Ask the subject if he/she felt the sharp prick of the pin and record a Y E S or NO answer in Appendix A.  STEP 6:  At five minutes, place the 50°C brass rod beside the wheal and tell subject to regard this as a five on the integer scale. Then place the rod directly on the wheal and obtain an immediate pain rating on the 0 to 10 integer scale. Enter score in Appendix A.  STEP 7:  Inject next test substance at the P position.  STEP 8:  Repeat STEPS 1 to 7 for injections sites y to <)>.  Any adverse events occurring from any of the injections are to be entered into Appendix B.  L. FRANCIOSI  275  PROBLEM MANAGEMENT 5.1  Emergency Contacts For the purposes of routine study management, the study monitor should be contacted. For the purposes of urgent consultation, in the event that the study monitor is unavailable, the backup emergency contacts may be contacted. Study Monitor:  5.2  Mr. Lui Franciosi Clinical Research Assistant Direct dial: (604) Message: (604)  XXX-XXXX XXX-XXXX  Back-up Contact # 1 :  Dr. X Principal Investigator / Anaesthesia Direct dial: (604) XXX-XXXX  Back-up Contact #2:  Dr. Y Direct dial:  (604)  XXX-XXXX  Adverse Experiences Details of all adverse experiences (AEs) encountered during the study will be reported in the CRFs-regardless of perceived relationship to study medication (Appendix B). Each experience will be assessed by the investigator as to its potential relationship to study medication. Appropriate action will be taken by DRUG COMPANY X, if indicated, and reported.  DRUG COMPANY X will be prepared to supply the sponsors with the following information, at the time of first reporting of severe, unexpected or alarming AEs:  5.3  a)  protocol title and number  b)  subject's initials and study code number  c)  subject's date of birth  d)  date of study medication administration  e)  type of AE, and time and date of onset  f)  details of AE, including severity and  g)  DRUG COMPANY X's opinion regarding potential relationship of AE to study medication, and rationale supporting this assessment  progression  Protocol Adherence DRUG COMPANY X will be expected to sign a Study Agreement Letter confirming full agreement with the terms of the approved protocol and with any subsequent approved amendment(s) to such protocol. It will be DRUG COMPANY X's responsibility to ensure that the study is conducted according to the protocol, and that all efforts are made to ascertain that the study participants are complying with the requests made of them for the study.  L. FRANCIOSI  276  Any subsequent protocol amendments must be agreed to by the INVESTIGATORS and DRUG COMPANY X.  ADMINISTRATIVE 6.1  Consent Requirements The study will be conducted according to the 1989 Drugs Directorate Guidelines (Conduct 1  of Clinical Investigations) issued by the Health Protection Branch . 6.1.1  Informed Consent DRUG COMPANY X staff will explain the nature of the study, and the risks which can reasonably be foreseen, to each participant prior to entry into the study. It should be made clear that the participant is free to withdraw from the study at anytime for any reason, without jeopardizing any subsequent treatment by the investigator. It should also be made clear that all medical information relevant to subjects will be kept confidential. Written informed consent will be obtained from each participant, prior to entry into the study. The subject will receive a copy of the signed consent form; the original is to be retained by DRUG COMPANY X.  6.2  Observers Properly trained and qualified technicians will assist in the technical and data recording aspects of the study.  6.3  Data Recording Case Report Forms (CRFs) will be used to collect the necessary data for each subject (Appendix A, B and C). These are to be kept as up-to-date as possible, to allow prompt review by any study monitor. DRUG COMPANY X will provide a copy of each completed CRF; INVESTIGATORS will retain the original. A CRF will be completed for all subjects receiving treatment or control, if part of the study. Although designated study staff may enter data into the CRFs, it is the responsibility of DRUG COMPANY X to ensure that the recorded data are accurate and complete. Each set of CRFs will contain a signature page which must be signed by the technician on duty signifying review of, agreement with and responsibility for the data contained therein. Individual pages are to be initialed by the designated study monitor. Corrections of any errors or omissions in the entries must be initialed and dated; the original entry should be crossed out but remain legible. All entries are to be in BLACK ink, and no correction fluid is to be used. No blank sections should be left on the CRFs; missing data should be identified as: N.D. = (not done) N. Av. = (not available) N. Ap. = (not applicable) as appropriate.  6.4  Record Keeping Following completion of the study, all independent records and copy of each participant's signed consent form and completed CRF must be retained by DRUG COMPANY X, for possible future reference.  L. FRANCIOSI  6.5  277  Study Supplies 6.5.1  Study Medications TSB430 and lidocaine will be purchased from the hospital pharmacy. Capsaicin will be bought from Research Biochemicals International.  6.6  Termination of the Study The maximum number of subjects to be studied is as specified in Section 3.3.1. INVESTIGATORS reserves the right to discontinue the study at any time, if deemed necessary for medical or administrative reasons. Should such action be required, reimbursement will be made for all reasonable expenses already incurred by DRUG COMPANY X.  6.7  Study Confidentiality DRUG COMPANY X shall agree to the following terms of confidentiality: a)  All unpublished information given to DRUG COMPANY X by INVESTIGATORS shall be kept strictly confidential and shall not be published or disclosed to any third party without the prior written consent of INVESTIGATORS.  b)  Details of any planned public disclosure (e.g. text of any lecture, or copy of an abstract or manuscript for publication) arising from the study will be supplied to the INVESTIGATORS at least four weeks before presentation or submission. This is not intended to limit the rights of DRUG COMPANY X to publish such information, but rather to enable the sponsors to make constructive comments about the text or manuscript.  STATISTICAL ANALYSIS  7.1  Efficacy Evaluation The integer value of the pain reported at 1 % and 5 minutes, and the pin prick responses recorded over 5 minutes for each treatment will be compared between treatment groups. Analysis of variance will be done for the Latin Square Design according to Design and Analysis of Experiments, D.C. Montgomery . Correlation analysis and repeated measures ANOVA will also be performed on the results obtained from the TSB430 concentrations used. An alpha level of 0.05 will be used for all analysis performed in this study. 2  7.2  Safety Evaluation All reported adverse experiences (AEs) will be tabulated.  L. FRANCIOSI  278  APPENDICES Confidential. REFERENCES 1.  Drugs Directorate Guidelines: Conduct of Clinical Investigations, Health Protection Branch, Health and Welfare Canada, 1989, Ottawa, Canada..  2.  Montgomery DC. Design and Analysis of Experiments, 2nd ed. 1984, Toronto, Canada.  John Wiley & Sons Inc.,  L. FRANCIOSI  APPENDIX 11 RESULTS OF STATISTICAL ANALYSES FOR THE EVALUATION OF INTRADERMAL TSB430 A S A PERIPHERAL ANALGESIC IN SUBJECTS WITH CAPSAICIN-INDUCED PAIN  (A1): Pain Scores with Capsaicin At 1.5 minutes (Latin Square ANOVA) Subjects  Injection Site  1  2  3  6  Row Sum  7  7  0  4 7  5  a  4  0  25  P  9  0  1  2  3  5  20  Y  0  6  4  2  8  4  24  5  1  8  0  0  3  1  13  £  1  0  7  2  0  0  10  <> f  1  10  5  0  1  3  20  Column Sum  19  31  17  13  19  13  112  HQ:  The mean pain score of subjects is the same for each of the six treatments.  HA:  The mean pain score of subjects is not the same for each of the six treatments.  Analysis of Variance for Latin Square: Summary By Row Row Row Row Row Row Row  Count  Sum  Mean  6 6 6 6 6 6  25 20 24 13 10 20  4.17 3.33 4.00 2.17 1.67 3.33  Count  Sum  Mean  6 6 6 6 6 6  19 31 17 13 19 13  3.17 5.17 2.83 2.17 3.17 2.17  1 2 3 4 5 6  Summary By Column Column Column Column Column Column Column  1 2 3 4 5 6  279  L. FRANCIOSI  Mean Pain Scores with Capsaicin at 1.5 minutes Subjects 1  2  3  SALINE  7  10  30 ng/ml TSB430  9  7  100 ng/ml TSB430  1  4  5  6  MEAN  SEM  7  0  5  2  8  5  6.17  1.40  3  4  5.00  1.06  8  4  2  4  3  3.67  0.99  300 ng/ml TSB430  1  0  0  2  3  0  1.00  0.52  1000 ng/ml TSB430  0  0  0  0  0  1  0.17  0.17  1000 ng/ml Lidocaine  1  6  1  7  1  0  2.67  1.23  '  Summary by Treatment: Count  Sum  Mean  0.90% Saline  6  37  6.17  30 Mg/ml TSB430  6  30  5.00  100 ng/mlTSB430 300 ng/ml TSB430  6  22  3.67  6  6  1.00  1000 ng/ml TSB430  6  1  0.17  1000 /ug/ml TSB430  6  16  2.67  SS  DF  ANOVA Results: Source of Variation  MS  F  Total Rows Columns Treatments  335.56 29.89 36.56 159.22  35 5 5 5  5.98 7.31 31.84  Error  109.89  20  5.49  Fcrit(5,20)  one-tailed:  2.71  two-tailed:  3.29  a  = 0.05  1.09 1.33 5.80  Non-significant Non-significant Significant (0.005 >p> 0.002 two  Correlation of Capsaicin Pain Scores with Doses of TSB430: I  r =-0.6225 (p < 0.001)  Reject H Q  280  L. FRANCIOSI  281  (A2): Pain Scores with Capsaicin At 1.5 minutes (Repeated Measures ANOVA of TSB430 Concentrations with Capsaicin Pain Scores) Treatments Pain Scores  Saline  Subject 1  7  30 ng/ml TSB430 9  Subject 2  10  Subject 3  7  Subject 4  0  Subject 5 Subject 6 TOTAL  100 ng/ml TSB430 1  300 ng/ml TSB430 1  1000 ng/ml TSB430 0  TOTAL  7  8  0  0  25  5  4  0  0  16  2  2  2  0  6  8  3  4  3  0  18  5  4  3  0  1  13  37  30  22  6  1  96  18  Ho:  The mean pain score of subjects is the same for each of the five treatments.  H/\:  The mean pain score of subjects is not the same for each of the five treatments.  N= SUMSQR Pain Scores =  30 596  GRAND TOTAL = Correction =  96 307.20  TOTAL SS = TRECETENT SS =  288.80 157.80  SUBJECTS SS =  39.60  REMAINDER SS =  91.40  ANOVA Source of Variation  (5 treatments (columns) X 6 subjects (rows)) (7 + 9 + 1 +....+3 + 0 + 1 ) (Sum of all rows or columns) 2  2  2  2  2  ((Grand total) /N - Correction) (SUMSQR Pain Scores - Correction) 2  ( ( 3 7 + 3 0 + 2 2 + 6 + 1 )/6 subjects - Correction) 2  2  2  2  2  ( ( 1 8 + 2 5 + 1 6 + 6 + 1 8 + 1 3 )/5 treatments Correction) (Total SS - Treatment SS - Subjects SS) 2  2  2  2  2  SS  DF  MS  Total Treatment  288.80 157.80  29 4  39.45  Subjects. Remainder  39.60 91.40  5 20  7.92 4.57  Fcrit(4,20)  one tailed:  2.87  two-tailed:  3.51  a=0.05  2  •  2  F 8.63  Significant (p < 0.001 two-tailed)  Reject H Q  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099241/manifest

Comment

Related Items