UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Standardizing our perinatal language to facilitate data sharing Massey, Kiran Angelina 2008

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

Item Metadata


24-ubc_2009_spring_massey_kiran.pdf [ 2.46MB ]
JSON: 24-1.0066857.json
JSON-LD: 24-1.0066857-ld.json
RDF/XML (Pretty): 24-1.0066857-rdf.xml
RDF/JSON: 24-1.0066857-rdf.json
Turtle: 24-1.0066857-turtle.txt
N-Triples: 24-1.0066857-rdf-ntriples.txt
Original Record: 24-1.0066857-source.json
Full Text

Full Text

STANDARDIZING OUR PERINATAL LANGUAGE TO FACILIATE DATA SHARING by Kiran Angelina Massey B.Sc., Queen’s University, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Reproductive and Developmental Science) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2008 © Kiran Angelina Massey, 2008  ABSTRACT Our ultimate goal as obstetric and neonatal care providers is to improve care for mothers and their babies. Continuous quality improvement (CQI) involves iterative cycles of practice change and audit of ongoing clinical care identifying practices that are associated with good outcomes. A vital prerequisite to this evidence based medicine is data collection.  In Canada, much of the country is covered by separate fragmented silos known as regional reproductive care databases or perinatal health programs. A more centralized system which includes collaborative efforts is required. Moving in this direction would serve many purposes: efficiency, economy in the setting of limited resources and shrinking budgets and lastly, interaction among data collection agencies. This interaction may facilitate translation and transfer of knowledge to care-givers and patients. There are however many barriers towards such collaborative efforts including privacy, ownership and the standardization of both digital technologies and semantics.  After thoroughly examining the current existing perinatal data collection among Perinatal Health Programs (PHPs), and the Canadian Perinatal Network (CPN) database, it was evident that there is little standardization of definitions. This serves as one of the most important barriers towards data sharing.  ii  To communicate effectively and share data, researchers and clinicians alike must construct a common perinatal language. Communicative tools and programs such as SNOMED CT® offer a potential solution, but still require much work due to their infancy. A standardized perinatal language would not only lay the definitional foundation in women’s health and obstetrics but also serve as a major contribution towards a universal electronic health record.  iii  TABLE OF CONTENTS Abstract…………………………………………………………………………………..  ii  Table of contents………………………………………………………………………..  iv  List of tables……………………………………………………………………………..  vi  List of figures…………………………………………………………………………….  vii  List of abbreviations…………………………………………………………………….  viii  Acknowledgements …………………………………………………………………….  ix  Dedication…………………………………………………………………………….....  xi  Co-authorship statement………………………………………………………………  xiii  Chapter 1: Introduction…………………………………………………………………  1  1.1 Obstetrics: An historical leader in knowledge brokering…………….….  1  1.2 Existing data collection in Canada……………………………………..….  2  1.2.1 Research projects (opportunistic databases) …………….……  3  1.2.2 Provincial perinatal health programs…………….…………….  4  1.2.3 National networks in maternal and newborn care…………….  5  1.3 Continuous quality improvement: The ‘PDSA’ approach…………….…  6  1.4 The advantages of building networks …………….…………….…………  9  1.4.1 Efficiency…………….…………….…………….…….…  9  1.4.2 Electronic records…………….…………….…………….………  10  1.4.3 Knowledge interaction…………….…………….…………….…  10  1.5 Objective…………….…………….…………….…………….…………….  11  1.6 Bibliography…………….…………….…………….…………….…………  12  Chapter 2: First Manuscript …………….…………….…………….…………….……  14  2.1 Challenges of building a database with convergence & collaboration in mind…………….…………….…………….…………….…………….…………  14  2.1.1 Geography.…………….…………….…….…………….………  14  2.1.2 Standardization of semantics……….…….…….…….…….…  15  2.1.3 Standardization of digital exchange of information..…………  20  2.1.4 Ownership of the data.…………….…………….…….………….  21  2.1.5 Privacy and ethics.…………….…………….…….…………….  22 Privacy and confidentiality…….……………….………  22  iv Security…….……………….……………….……………  23 Research ethics boards…….……………….…………  26  2.1.6 Stable funding…….……………….……………….……………… 2.2 Bibliography…….……………….……………….……………….…………  27 29  Chapter 3: Second Manuscript …….……………….……………….……………….…  31  3.1 A current landscape of provincial perinatal data collection in Canada…  31  3.1.1 Introduction……….…………………….…………………….……  31  3.1.2 Methods……….…………………….…………………….………  32  3.1.3 Results……….…………………….…………………….…………  34  3.1.4 Discussion……….…………………….…………………….……  37  3.1.5 Conclusion……….…………………….…………………….……  40  3.2 Tables from results……….…………………….…………………….……  42  3.3 Bibliography……….…………………….…………………….……………  57  Chapter 4: Third Manuscript ……….…………………….…………………….………  58  4.1 What is SNOMED CT® and why should the ISSHP care? ……….……  58  4.2 Bibliography……….…………………….…………………….……………  65  Chapter 5: Conclusion……….…………………….…………………….………………  66  5.1 Databases as the foundation of evidence based medicine……….……  66  5.2 Knowledge interaction ……….…………………….…………………….…  67  5.3 Semantics integration ……….…………………….…………………….…  71  5.3.1 Semantic discrepancies ……….…………………….…………  71  5.3.2 Preserving our perinatal knowledge ……….……………………  75  5.4 Future direction: The Electronic Health Record ……….…………………  76  5.5 Final thoughts (summary) ……….…………………….……………………  80  5.6 Bibliography……….…………………….…………………….……………  83  Appendices……….…………………….…………………….…………………….…… Appendix A: Canadian Perinatal Network Collaborative Group Members…  85 85  v  LIST OF TABLES  3.1 Canadian Perinatal Health Programs ….……………………….………………....  42  3.2 Information collected by PHPs, CPN and the CMDS: Demographics………….  43  3.3 Information collected by PHPs, CPN and the CMDS: Past medical/obstetric/surgical history………………………….…………………………....  44  3.4 Information collected by PHPs, CPN and the CMDS: Current pregnancy……..  46  3.5 Information collected by PHPs, CPN and the CMDS: Maternal interventions and complications…….………………………………………………..….  48  3.6 Information collected by PHPs, CPN and the CMDS: Fetal complications and interventions…….………………………………………………………..…………….....  51  3.7 Information collected by PHPs, CPN and the CMDS: Labour and delivery…....  52  3.8 Information collected by PHPs, CPN and the CMDS: Perinatal outcomes….....  54  vi  LIST OF FIGURES  1.1 PDSA Cycle (‘Plan’, ‘Do’, ‘Study’, ‘Act’) …….………………………………..…....  7  2.1 SNOMED CT® and the example of pre-eclampsia…….………………………...  19  4.1 The hierarchies of SNOMED CT®…….…………………………….……………...  60  4.2 A SNOMED CT® map for the concept of ‘Gestational Hypertension’…….…….  62  4.3 A potential map for the concept of ‘Pre-eclampsia’…….………………………...  63  5.1 Model of knowledge interaction…….……………………………………………....  68  5.2 BCPHP PDSA cycle informing a CPN cycle…….………………………………...  70  5.3 CPN PDSA cycle informing a BHPHP cycle…….………………………………...  71  5.4 Point-to-point strategy (‘Cognitive heterogeneity’) …….……………………….....  72  5.5 Rule based linking (‘Naming heterogeneity’) …….………..……………………...  74  5.7 Data exchange that decreases flow of personal identifiers…….………………..  79  vii  LIST OF ABBREVIATIONS APHP  Alberta Provincial Perinatal Health Program  Atlee  Nova Scotia Atlee Perinatal Database  BCPHP  British Columbia Perinatal Health Program  CAPSNet  Canadian Perinatal Surgery Network  CDSR  Cochrane Database of Systematic Reviews  CIHI  Canadian Institute for Health Information  CMDS  Canadian Minimal Dataset  CNN  Canadian Neonatal Network  COACH  Canada’s Health Informatics Association  CPN  Canadian Perinatal Network  CPPC  Canadian Perinatal Programs Coalition  CSA  Canadian Standards Association  EHR  Electronic Health Record  FAN  Fetal Alert Network  ICD  International Classification of Diseases  IHTSDO  International Health Terminology Standards Development Organization  NHS  National Health Services  NIDAY  NIDAY Perinatal Database of Ontario  NLPPP  Newfoundland and Labrador Provincial Perinatal Program  PDSA  Plan Do Study Act  PHI  Personal Health Information  PHP  Perinatal health program  RCP  Reproductive Care Programs  REB  Research Ethic Boards  SNOMED CT®  Systematized NOmenclature of MEDicine Clinical Terms  SWOPP  Southwestern Ontario Perinatal Partnership  TPU  Tertiary perinatal units  T  viii  ACKNOWLEDGEMENTS I offer my sincerest gratitude to the faculty, staff and fellow students at the UBC. I owe particular thanks to my supervisor Dr. Laura Magee whose intelligence, curiosity and passion for learning has inspired me as a life long learner. Her patience and guidance has made an everlasting impression on me both personally and as a scholar while her kindness and compassion has made her a personal mentor to me for years to come.  Many thanks to Roshni Nair and my thesis committee: Sheryll Dale and Drs. Peter von Dadelszen, John Mark Ansermino and Robert Liston who always made time from their busy schedules to provide answers to my many questions. I truly appreciate your guidance and encouragement throughout this project.  I also appreciate my funding agency, Interdisciplinary Women’s Reproductive Health Research, who has been a financial supporter of this research. In particular Drs. Wendy Robinson and Patricia Janssen have provided many opportunities for myself and all trainees to explore research from a variety of facets enriching and expanding our desire for research.  To the MFM research group of the ‘B4 corridor’, thank you! Your encouragement and openness to share gave me support daily. You have offered so much both  ix  academically and personally from day one. I have learned a great deal from you all individually and as a group that I will take with me where ever I go.  Lastly to my dearest friends and family miles away and the new support group I have made here in Vancouver. Without your love, encouragement and endless support this dream would have never come to fruition. In particular I would lastly like to thank my parents, Charles and Jhaneeta Massey who have given me infinite and unconditional support throughout the years. As I end this journey to begin a new one I know it is solely because of your love, sacrifices and dedication for me to succeed and be happy.  x  DEDICATION I dedicate this thesis to all of you who not only made this journey possible but unforgettable, even on the rainy days.  Thank you.  xi  CO-AUTHORSHIP STATEMENT  Together my committee consisting of clinicians, researchers, reproductive care coordinators and informaticians and I developed a project that would collaborate on all of our interests. The design of the research touches and converse all of these fields.  I personally conducted all my own research and literatures reviews. I also did all the data analysis which involved contacting all provincial perinatal health programs.  I predominately wrote all the manuscripts which were all edited by my supervisor, Dr. Laura Magee. All manuscripts were reviewed by the coauthors and were submitted to the editors and journals by myself.  I personally wrote this thesis in its entirety.  xii  CHAPTER 1: INTRODUCTION  1.1 Obstetrics: an historical leader in knowledge brokering Over the past few decades, obstetrics has been a leader in the dissemination of medical knowledge. In 1979, when Archie Cochrane proposed, “a critical summary, adapted periodically, of all relevant randomized controlled trials (RCT)”(1), it was obstetrics that first took action. A registry was developed of controlled trials of interventions during pregnancy and early infancy(2). Thus was born the Oxford Database of Perinatal Trials, otherwise known as the “Odd Pot” (ODPT). ODPT served as a resource for reviews of interventions in maternal and neonatal care, and an important tool used by those involved in quality of care promotion(3). In 1993, with the influence of the worldwide web and advanced software, the ODPT became an electronic publication known as the Cochrane Collaboration Pregnancy and Childbirth Database (CCPC). Archie Cochrane also urged specialty fields to arrange significant summaries of RCT data in the 1972 publication Effectiveness and Efficiency: Random Reflections on Health Series (1). It was obstetrics that again answered the call by publishing the first edition of A Guide to Effective Care in Pregnancy and Childbirth (1989) (4) which summarized RCT’s regarding maternal and infant care in order to better understand health practices and set policies(5). A Guide to Effective Care in Pregnancy and Childbirth (6) is still used today as an effective resource among health care providers and their patients.  Systematic reviews of RCT analyses later extended into many other health and medical fields to eventually form the Cochrane Database of Systematic Reviews (CDSR) in 1994.  1  The rest is, as they say, history as this accomplishment proved to be an enormous step forward not only for obstetrics but for medicine as a whole. Having the CDSR available on delivery suites is considered to be an important quality of care resource when assessments are made of academic obstetrics and gynecology services and programs. No application for research funding can be made without reference to the relevant Cochrane review.  The CDSR is based on the idea that the randomized controlled trial is the least biased form of information about interventions that improve maternal and perinatal outcomes. However, the many limitations of relying solely on randomized trials have become more obvious over time. First, for many of the questions posed in maternal/infant care, there are insufficient trials to guide clinical practice. Second, trials are onerous to mount and very expensive to conduct. As such, they must be reserved for the most important of research questions. Finally, there are other questions that are simply unethical to study by using a randomized trial design. Continuous quality improvement, in particular the ‘Plan, Do, Study, Act’ (PDSA) (discussed below) approach, offers an alternative to the randomized control trial approach. Regardless of the approach, quality improvement is dependent on the ability to collect data.  1.2 Existing data collection in Canada With advancements in database technology, we are accumulating vast quantities of data at a record pace and from a variety of sources; research projects, provincial perinatal health programs and national networks.  2  1.2.1 Research projects (opportunistic databases) Many researchers have developed databases for local as well as multicentre, collaborative studies with the goal of collecting information for a special purpose or specific project. Each database is designed and built to fulfill specific goals and objectives within the cost and time constraints of the project. Each database is usually active only for the duration of the study. A wide variety of programs and digital formats may be used even by the same researchers, depending on the expertise and longevity of employment of other team members, particularly trainees. For the few that are digitally compatible and employ the same software, the definitions of the data fields become the ultimate barrier, as they too have a great deal of discrepancy.  As local funding sources come to an end and researchers move on to other projects, the maintenance of these databases does not usually continue, and valuable data may be lost. Stewardship is often delegated to temporary residents or fellows who inevitability leave the project and move on. Often when the principal investigator is subsequently contacted, data cannot be found or retrieved in a timely manner (if at all). The ineffectiveness of contacting authors for missing information is a well known issue that many reviewers encounter(7). This is a problem that could be potentially avoided if the lifespan of the database were dependent on an organization(s) or network(s), rather than on an individual(s).  In obstetrics, special purpose databases collect very detailed information about the intervention(s) of interest, and the potential confounders of the relationship between the intervention and the outcomes. In addition, they gather much of the general information that is also collected by perinatal health programs (see below). It follows that there is much  3  duplication of effort and data collection, particularly with respect to patient demographics, diagnoses and basic procedures. By working in isolation, as researchers so often do, inefficiency, error and expense often result.  1.2.2 Provincial perinatal health programs (PHP) In Canada, there are many perinatal health programs (PHP) that collect data on maternal and newborn outcomes, for all deliveries. These programs have been set up to monitor geographical disparities in perinatal health and determinants of adverse outcomes (i.e. low socioeconomic status). Health outcomes may be improved at the population level by reducing health disparities(8), be they cultural, geographic, monetary (in terms of income), educational, ethnic, or medical (e.g., differences in admission or access to procedures). Much of collected information is used for regional reporting to policy makers and government; however, this is only a portion of the larger mandate of the PHPs which also includes peer review, education, guideline development, etc.  Perinatal health programs in Canada are located: in Alberta [APHP (Alberta Provincial Perinatal Program)], British Columbia [BCPHP (BC Perinatal Health Program)], Newfoundland and Labrador [NLPPP (Newfoundland and Labrador Provincial Perinatal Program)], Nova Scotia (Nova Scotia Atlee Perinatal Database), Ontario (NIDAY Perinatal Database), and Prince Edward Island. Data are collected on maternal demographics, pregnancy characteristics, details of labour and delivery, and initial evaluation of the newborn. Data are collected via a variety of methods; data may be entered directly into the electronic database and periodically returned to the provincial database, data may be collected using paper forms that are sent to the coordinating center, or collected on site periodically by a data abstractor from the central site. Data on maternal mortality and morbidity are often obtained by linkage with national discharge abstract databases (DAD).  4  However, there is insufficient information collected about interventions to allow for identification of those practices associated with the best outcomes for mothers and babies. For example, few reproductive care programs have information about maternal transport, outpatient surveillance, or the use of specific tocolytics, antibiotics, or antihypertensive agents.  1.2.3 National networks in maternal and newborn care The Canadian Neonatal Network (CNN) The CNN was established by Dr. SK Lee in 1995 in response to the critical need for new knowledge in maternal and newborn care. Data are collected across Canada in the context of routine clinical care, and are entered into a computerized database. This network provides an established infrastructure for collection of high quality standardized data for research and thereby reduces the cost.  The CNN database has been used to show that obstetric practice may influence neonatal outcomes. For example, Synnes et al showed that mode of delivery was associated with the risk of intraventricular haemorrhage among babies admitted to NICU(9). Also, Lee et al determined that the most cost-effective strategy for screening for retinopathy of prematurity is routine screening only of infants with birthweight <1200g(10).  With the success of the CNN, we sought to develop a network aimed at improving neonatal and maternal outcomes via perinatal care. From this idea, The Canadian Perinatal Network was born.  5  The Canadian Perinatal Network (CPN) The CPN is made up of Canadian researchers across Canada who collaborate on research issues relating to perinatal health. The inaugural project of the CPN is to identify, in the setting of threatened very preterm births at 22-28+6 weeks’ gestation, interventions that are related to good/poor maternal and perinatal outcomes and/or greater resource use. This requires collection of detailed information about obstetric practices in Canada’s 24 tertiary perinatal units (TPU). CPN links with the well-established national CNN (Canadian Neonatal Network) and CAPSNet (Canadian Perinatal Surgery Network).  The CPN was borne of a desire to generate new knowledge in the area of obstetric care, by using a continuous quality improvement approach to study clinical practice. This approach is identifies those practices that are associated with good outcomes for mothers and babies, correcting for potential confounders of that relationship. The CPN approach is predicated on the Plan, Do, Study, Act (PDSA) cycle (see below) and is designed to improve the quality of care (11).  1.3 Continuous quality improvement: The ‘PDSA’ approach Continuous quality improvement via the PDSA (Plan, Do, Study, Act) approach offers an alternative to the randomized control trial.  6  Figure 1.1 PDSA Cycle (Plan, Do, Study, Act) Modified from Langley (12)  The approach tests a hypothesis or a theory for healthcare improvement (‘Plan’), implements the plan with a study protocol and data collection (‘Do’), examines and summarizes the data collected (‘Study’), and then determines the change and iteration for further steps (‘Act’)(13). Thereafter, possibilities include: a new iterative cycle, testing of an intervention in a randomized controlled trial, or translation of new knowledge into clinical practice.  Most are familiar with the ‘Plan’ and ‘Do’ components, for example, a knowledge audit, in which all potential sources of information are explored. However, there is often ‘starvation in the face of plenty’, because data are everywhere yet are not easily accessible. For example, a questionnaire was recently sent to all site investigators in the Canadian Perinatal Network (CPN). We inquired about basic information and characteristics about their tertiary perinatal unit. We found that some of the largest centers in the country were having trouble finding basic information such as number of antepartum beds or the percent of deliveries at a given gestational age. This held true for some investigators who tried, in vain, to obtain the information directly from Department Heads. Clearly, these data are collected by the facility,  7  but they are not readily accessible. In contrast, preparation of a recent grant application required knowledge about neonatal intensive care unit admissions among babies of hypertensive mothers; the Canadian Neonatal Network database was contacted, and the co-ordinator was able to provide the data in 48hr by electronic analysis. In terms of the ‘Do’ component, readers will be very familiar with hospital guidelines for investigation and/or management of patients.  Unfortunately, some will also be familiar with the lack of data collection that follows the ‘Plan’ and ‘Do’ steps. That is to say that we often do not know if most guidelines are being implemented, and if they are implemented, whether or not the outcomes of interest are being changed. In contrast, the formal PDSA cycle ensures objectives are set and alternative solutions are considered, and forces measurement to take place(14). When used, the PDSA cycle encourages evidence-based solutions to develop knowledge and achieve the highest quality of care as issues are assessed and changes are monitored and re-evaluated.  PDSA research builds knowledge(15). For example, at Birmingham Children’s Hospital, it was observed that incoming emergency and elective patients were waiting for assessment until outgoing patients had vacated their beds. A proposed solution to decrease wait times was to create a discharge lounge where patients who were being discharged could wait for their prescriptions or paperwork(14). The ‘Plan’ was a written proposal for the discharge lounge with an operation protocol, followed by the ‘Do’ step of providing a facility in which to test the idea. The actual ‘Study’ component included measuring the waiting times of the patients. The results showed that beds were freed up earlier, children were less upset about being in the hospital, and that there was general improvement in patient and family  8  satisfaction(14). As a result, a longer test was scheduled which produced similar results as the initial test. With multiple PDSA testing of the same concept at different facilities, it was found that the lounge was mainly used by patients waiting for their take-home drugs. This created a new piece of knowledge to be addressed by more PDSA cycles. This would not normally have been highlighted if it were not for multiple cycles and fine adjustments to the PDSA cycle.  PDSA research can also build networks. PDSA cycles should be not only iterative (in that knowledge learned from one cycle informs another cycle), but these cycles should be interactive. Data sharing among various data agencies creates more new knowledge which would otherwise be impossible individually.  1.4 The advantages of building networks Data collection and sharing permits answers to clinical questions. Whether its specific research questions (opportunistic databases), geographical trends and policies (provincial perinatal health programs) or best practice approaches (national networks) these agencies provide valuable information for a variety of purposes. Since many of these networks collect similar or identical information, there is potential for collaborative efforts. These efforts may results in enhanced efficiency, preparation for electronic records, and knowledge interaction.  1.4.1 Efficiency Basic patient information such as patient name, date of birth and estimated due date (EDD), are some examples of the many fields that are common to most systems. Duplicate data not only are time consuming but may also lead to error. With the convergence of these  9  systems, duplicate data are entered once, reducing error, saving the data abstractor time and as a result reduces cost.  1.4.2 Electronic records The interest in electronic health records (EHR) is growing exponentially. In the United Kingdom, the National Health Services (NHS) aims to have 60,000,000 patients with a centralized electronic medical record by 2010. In Canada, the Alberta Netcare is used to record results of patient’s diagnostic tests which can be used for pharmacutical services(16).  Of the various electronic patient record systems that emerge in hospitals the networks which will be the most valuable will undoubtedly be the ones that can adapt to the electronic record, and link to other networks. This transition may allow for real time data collection within the databases and a smooth flow of information sharing between systems.  1.4.3 Knowledge interaction Networks which are based on a model of ‘knowledge interaction’ are the key to improving our health care system. Knowledge is generated by a ‘Plan, Do, Study, Act (PDSA)’ continuous quality improvement approach. Interaction among and between networks enhances the generation of new and more difficult clinical questions that we are better able to answer. The translation of this new knowledge into clinical practice and ensuring that it affects the intended population is critical to making positive changes for patients.  10  1.5 Objective This is an investigation of potential limiting factors involved in the convergence and collaboration of various perinatal database networks. To accomplish this objective, three different manuscripts were written and submitted. Chapter 2 (Manuscript 1) discusses the possible challenges organizations will encounter upon data sharing. These challenges set the foundation of limitations that must be addressed and resolved before networks can initiate interactive efforts. Of these challenges, semantic interoperability proves to be a significant rate limiting step and is further explored in the following two manuscripts.  Concentrating on semantic discrepancies, Chapter 3 (Manuscript 2) examines existing perinatal data collection by Perinatal Health Programs (PHPs). These data fields and definitions were compared to CPN and the Canadian Minimal Dataset, proposed as a common dataset by the Canadian Perinatal Programs Coalition of Canadian PHPs. This manuscript is used to show that not only are there differing data fields, but of the similar data fields that are collected, the definitions of these concepts are different.  Lastly, Chapter 4 (Manuscript 3) explores the role of a communicative tool used in the harmonization of data field definitions. SNOMED CT® [The Systematized NOmenclature of MEDicine Clinical Terms (SNOMED CT®)] creates a standardized language permitting streamlined exchange of data. In particular this introduction to SNOMED CT®, the future medical universal language, explains the need for special interest groups and subspecialty involvement with its implementation.  Together these three manuscripts will not only form a foundation for potential knowledge interaction for various networks and organizations such as PHPs and CPN but will also explore standardized communication for the future and the electronic health record.  11  1.6 Bibliography (1) Cochrane AL. Effectiveness and Efficiency: random reflections on health services. London: Nuffield Provincial Hospitals Trust, 1972. (2) Starr M and Chalmers I. The evolution of The Cochrane Library, 1988-2003. www.update-software.com/history/clibhist.htm . 2003. (3) Grant A, Chalmers I. Register of randomised controlled trials in perinatal medicine. Lancet. 1981;1(100). (4) Enkin M, Keirse M, Chalmers I. A Guide to Effective Care in Pregnancy and Childbirth. 1 ed. London: Oxford Medical Publications, 1989. (5) Childbirth Connection. About This Book:A Guide to Effective Care in Pregnancy and Childbirth. http://childbirthconnection.org/article.asp?ck=10014 . 2006. (6) Enkin M, Keirse M, Neilson J et al. A Guide to Effective Care in Pregnancy and Childbirth. 3 ed. London: Oxford University Press, 2000. (7) Khoshdel A, Attia J, Carnery SL. Basic concepts in meta-analysis: a primer for clinicians. International Journal of Clinical Practice. 2006;60(10):1287-1294. (8) Shoultz J, Fongwa M, Tanner B et al. Reducing health disparities by improving quality of care: lessons learned from culturally diverse women. Journal of Nursing Care Quality. 2006;21(1):86-92. (9) Synnes AR, Chien LY, Peliowski A et al. Variations in intraventricular hemorrhage incidence rates among Canadian neonatal intensive care units. J Pediatr. 2001;138(4):525-531. (10) Lee SK, Normand C, McMillan D et al. Evidence for changing guidelines for routine screening for retinopathy of prematurity. Arch Pediatr Adolesc Med. 2001;155(3):387-395. (11) De Lusignan S, Wells S., Shaw A. et al. A knowledge audit of the managers of primary care organizations: top priority is how to use routinely collected clinical data for quality improvement. Medical Informatics & the Internet in Medicine. 2005;30(1):69-80. (12) Langley GL, Nolan KM, Nolan TM et al. The Improvement Guide. A Practical Approach to Enhancing Organizational Performance. San Francisco: Jossey-Bass, 1996. (13) Speroff T, O'Connor GT. Study designs for PDSA quality improvement research. Qual Manag Health Care. 2004;13(1):17-32. (14) Walley P, Gowland B. Completing the circle: from PD to PDSA. Int J Health Care Qual Assur Inc Leadersh Health Serv. 2004;17(6):349-358.  12  (15) Speroff T, O'Connor GT. Study designs for PDSA quality improvement research. Qual Manag Health Care. 2004;13(1):17-32. (16) Electronic health record. Wikipedia: The Free Encyclopedia on Line . 2007. Wikimedia Foundation Inc.  13  CHAPTER 2: FIRST MANUSCRIPT 2.1 Challenges of Building a Database with Convergence & Collaboration in Mind 1 We were acutely aware that the CPN database must not reinvent ‘the wheel’, but capitalize on existing infrastructure and definitions. Here, we review the issues that needed to be addressed in trying to build the CPN database and meet these goals.  2.1.1 Geography We used the CNN as our model. However, the NICU is very geographically defined. Babies reside there until discharge home or to another unit. However, in obstetrics, patients may be managed as outpatients, inpatients (and then admitted to various wards), go to a wide variety of laboratories and ultrasound departments, and be seen by a wide variety of consultants, each of whom may keep their own patient record. This highlighted for us the many places from where information must be gathered.  In response to these concerns, as a first step, we chose to limit the geographical scope of CPN and focus on inpatients, admitted to one of Canada’s tertiary perinatal units.  A version of this chapter has been published as a book chapter in Medical Informatics in Obstetrics and Gynecology Edited by David Parry and Emma Parry New Zealand 2008. Massey KA, Morris TJ, Liston RM, von Dadelszen P, Ansermino JM, and Magee LA on behalf of the Canadian Perinatal Network Collaborative Group (Appendix) and the British Columbia Perinatal Health Program. “Building Knowledge in Maternal and Infant care” in Medical Informatics in Obstetrics and Gynecology. Auckland. November 2008  1  14  2.1.2 Standardization of semantics Before systems can connect to each other and before we can even begin to think about hardware, software, and programming, we need to be speaking the same language. In essence, databases and the future electronic health record are about concepts and their definitions first and foremost, and not about computers, cables, and internet connections.  Standardization of terminology is perhaps the least obvious but most important challenge for data sharing. Data terms must be universally understood (i.e., the definitions must be the same). In creating CPN, it was clear very early on that there is no standardization in the published literature for most obstetric or neonatal terms in common use. For example, perinatal mortality is defined differently by different reproductive care programmes between Canadian provinces (e.g., WHO definition by birth at ≥28 weeks vs. birth at ≥20 weeks or ≥500g). What constitutes reduced biological growth potential is variably defined, ranging from birth weight <2500g, to birth weight <3rd centile for gestational age and gender (1), which is further complicated by the skewness of birthweight data remote from term (i.e., as these babies were not normal, or they would not have been born).  The CPN terminology has relied on a strict classification system. The CPN terms have been drawn from the proposed CPPC ‘Minimal Dataset’, and the CNN database manual. However, many of terms (variables) were chosen from among a variety of definitions in the published literature.  In Canada and internationally, the most common health care coding system used by hospitals is another strict classification system called the International Classification of  15  Diseases (ICD). The ICD codes were defined by the World Health Organization (WHO), and ICD-10-CA is an enhanced version of ICD-10 which was recently developed by Canadian Institute for Health Information (CIHI) in 2006 to include classification of morbidity. ICD-10CA includes codes for disease and related health problems as well as a number of conditions and situations that represent other risk factors to health, such as occupational and environmental factors, lifestyle and psycho-social circumstances. The terms, gestational hypertension, placenta previa and maternal blood transfusions can be coded and grouped into categories, which allows for aggregation and retrieval of information.  There are problems with a system of strict classification. • First, the details of novel significance can be lost. For example, in pregnancy, women with pre-existing hypertension due to one of a number of endocrine causes (such as thryrotoxicosis or hyperaldosteronism) receive the same code, ‘Pre-existing hypertension, unspecified, code O10.9). • Second, healthcare is a dynamic environment. As researchers and clinicians learn new and better procedures and interventions, categories of coding systems change and evolve. For example, in 1993, ICD-9 795.8 was ‘positive serological or viral culture findings for human immunodeficiency virus (HIV)’. By 1994, the same code had been deleted, and then re-introduced in 2006 but as ‘abnormal tumour markers’. You can see that it would be difficult at best to compare data between versions of a classification system! • Third, at present, data abstraction and coding is usually done by a trained clerk (with limited content knowledge) from information that can be deciphered from the medical record. This may be improved by electronic health records that will allow for coding of  16  information at the bedside by the clinician. His/her content (‘domain’) knowledge and better understanding of the specific situation will facilitate a higher level of data reliability. • Finally, perhaps the most interesting limitation that current coding systems encounter is the inability for codes to form ‘relationships’ with other codes. For example, a code for gestational hypertension is actually a form of hypertension, and women with gestational hypertension are at increased risk of long-term hypertension, but in ICD-10, these are completely different codes. Using the example of the secondary causes of hypertension (above), the thyrotoxicosis might be coded in addition to the pre-existing hypertension, unspecified, but there is nothing that links them together, so one doesn’t know if they are related. Although these relationships may not seem critical to coding terms for easier retrieval, they are absolutely necessary in order for us to use this systems to ask and answer questions and generate knowledge.  Coding has been used by industry for many years, most commonly evident as the ubiquitous ‘barcode’ however, in building the CPN database, we have recognized that there is an alternative to our use of this strict classification system.  The Systematized Nomenclature of Medicine (SNOMED) Clinical Terms (CT®) The future of universal health care is the semantic harmonization promised by SNOMED®. The International Health Terminology Standards Development Organization (IHTSDO) supports the effort to produce standardized global clinical terminology and governs and owns SNOMED-CT® intellectual property(2). Currently, the IHTSDO has nine countries as charter members: Australia, Canada, Denmark, Lithuania, New Zealand, Sweden, The  17  Netherlands, United Kingdom, and the United States. Access to SNOMED-CT® is provided to clinicians, researchers, and administrators in these countries.  In SNOMED-CT®, human-readable terms are assigned computer-readable codes. This clinical terminology is not only a way of coding that preserves specific medical information, but it allows for different systems to easily and accurately exchange information. In essence, SNOMED-CT® is our new universal ‘language’.  The basic elements of SNOMED-CT® include concepts, descriptions, hierarchies and relationships. A concept is a single clinical meaning which has its own concept code and a human readable name attached to it [e.g., pre-eclampsia (SNOMED concept ID)]. As illustrated by Figure 3, all concepts are organized into hierarchies (e.g., clinical findings and disorders, procedures, or observable entities), each of which contains sub-hierarchies (not shown in Figure 3). Each concept falls under one or more hierarchies. In Figure 3, preeclampsia falls under the clinical findings and disorders hierarchy; however, pre-eclampsia could be linked to other hierarchies such as ‘laboratory’ with low platelets being the linking concept. Each concept also has a number of human readable descriptions, such as pregnancy induced hypertension, pre eclampsia, and high blood pressure which would be associated with the gestational hypertension concept. Lastly, but most importantly, SNOMED-CT® allows concepts to be linked through relationships. Relationships are a powerful and innovative feature that will allow for clinicians and researches to build complex concepts with multiple terms to describe a specific situation. Figure 3 illustrates SNOMEDCT® using ‘pre-eclampsia’ as an example.  18  GH → Gestational hypertension PIH → Pregnancy induced hypertension sBP → Systolic blood pressure dBP → Diastolic blood pressure  HIERARCHIES (Components) GH with Proteinuria  DESCRIPTIONS (Synonyms)  Toxaemia  Clinical Findings and Disorders  PIH Procedures  MAIN CONCEPT  PRE-ECLAMPSIA  Laboratory  Staging and Scales:  RELATIONSHIPS Situation with Explicit Context  Obesity ASSOCIATED CONCEPTS  Hypertension  IgA Nepropathy  Figure 2.1 SNOMED-CT® and the example of pre-eclampsia (Massey KA, Magee LA, von Dadelszen P 2008)  Essentially, SNOMED-CT® has the ability to take a human written or spoken sentence, such as “a fetus has a gestational age of 24 weeks, is a male, and has a mother with eclampsia”, and convert it into a series of codes which represent the description, the specific concept and the relationships. Whether this sentence is dictated in Spanish or Russian, it would be coded exactly the same in SNOMED-CT®, enabling someone in Japan to understand the diagnosis, procedure and clinical findings. As such, SNOMED-CT® has international capability and will become the universal language of medical care. In fact, any health informatics system that is marketed in Canada will soon have to be based on SNOMED-  19  CT® terminology because it provides a consistent way to index, store, retrieve, and aggregate clinical data across specialties and sites of care and will form the backbone of the electronic health record, for quality of patient care and research.  2.1.3 Standardization of digital exchange of information Databases can be created using many different platforms, applications, interfaces and communication protocols. Interoperability is commonly limited to expensive, specifically designed interfaces that are limited in the amount of data exchange and are not adaptive to changes in individual systems. Digital exchange of information between and among networks can be expensive, inaccurate (e.g., if probabilistic), and sometimes, impossible. However, with the future wave of electronic health records, it is absolutely crucial that existing systems be made digitally compatible. Clinical information should not only be accessible for care of the individual patient, but should provide knowledge for outcome improvements.  The CPN database shares the same digital technology as the CNN database. These are, in essence, the same database. CPN patients whose babies are admitted to neonatal intensive care, have the CNN number of the baby recorded in the database; thereafter, all neonatal outcome information can be obtained from CNN without any addition data collection by CPN. This digital integration has been a logical extension of other work within our unit in which all of our projects must have databases built with the same digital technology.  20  2.1.4 Ownership of the data As the Government of Canada Task Force for the National Consultation on Access to Scientific Research Data (NCASRD) points out(3), historically, research has been driven by individual researchers motivated by peer recognition and institutional reward, with collaborative team efforts aimed at improving the public good often coming second. It is not surprising then that there is little motivation for researchers to contribute to data-sharing initiatives once data are no longer linked to the individual. In addition, researchers involved in collaborative projects may also feel personally responsible for data that they individually collect at their centres, or may have concerns that their own research endeavors will be slowed down if they join larger multi-centre research efforts(4).  Local sites often ‘own’ their data; they have free access to it and there is also a reporting function. However, in our database access to national data will require application to the CPN Steering Committee.  Changing the current mind set of well established organizations ultimately requires a paradigm shift away from competitive research. At the core of this issue is the idea of data ownership, which includes everything from intellectual ownership (for example, who can perform secondary data analysis, or who is entitled to authorship on publications), to where the data files will physically be stored(4). It is necessary to attend to these issues by creating inter-institutional committees so that specific guidelines can address, on a practical level, how data will be shared. (At present, a motivator for participation in the CPN is the fact that local data are owned and housed with the local site.) In the future, committees may choose to do away with the term “ownership” all-together, choosing instead the less emotionally-  21  charged term “stewardship”(4). Although this may seem like nothing more than semantics, it may be one small step toward a less individualistic research culture, driven by leaders and owners rather than collaborators. Universities could certainly facilitate this by recognizing collaborative research activities in performance reviews of faculty and Departments. Funding agencies are moving in the right direction. The Canadian Institutes of Health Research (CIHR) recognize the importance of collaborative research, and now offer team grants aimed at recognizing the achievements of groups of researchers, not just individuals. In the future, negotiation, open communication, and compromise will all be key elements in successfully dealing with the issues of data ownership, proprietary concerns, and authorship(5).  2.1.5 Privacy and Ethics The sensitivity of personal health information is perhaps second only to financial information(6). Since 1988, COACH (Canada’s Health Informatics Association) has been providing Healthcare Professionals and Healthcare Organizations with the guidance necessary to ensure that the security, privacy, and confidentiality of Personal Health Information (PHI) are upheld. The newly-released 2006 COACH edition reflects our increased use of information technology in the healthcare industry, in particular the risk that is introduced by the remote capabilities and access speed that are offered by electronic records(7). Privacy and confidentiality Privacy is the fundamental right of every individual to dictate how their personal information is used and exchanged. Confidentiality is the organizational obligation to protect the personal information that is entrusted to it.  22  Canada made a commitment (as did other countries) to uphold an individual’s right to privacy when we joined the global community in signing the OECD (Organization for Economic Co-operation and Development) Guidelines on the Protection of Privacy and Transborder Flow of Personal Data. These guidelines were used by the Canadian Standards Association (CSA) as the basis for the Model Code for the Protection of Personal Information. The CSA guidelines balance the individual right to privacy with the information requirements of private organizations, and are intended to assist organizations in developing appropriate policies and practices specific to their circumstances(8). Security Security is the process by which the information is protected(9). The mechanisms by which we prevent the unauthorized access, use and disclosure of PHI are known as security safeguards(10). Healthcare organizations should have security policies in place that are based on a national or international standard.  As electronic storage of PHI becomes increasingly straightforward, so does access to this information. As such, access control is becoming a fundamental component of data security. Preventing the unauthorized access to PHI has its challenges, not only from a technical standpoint (the implementation of user authentication systems, to password management systems, to securing remote or mobile access connections), but also from a management standpoint (deciding who should have access rights). As is so often the case, the latter may be the most complex part to tackle. Take for example the common situation where one individual performs multiple roles within a healthcare organization, each with different  23  associated levels of access privilege, as might be true for a physician who is both a clinician and a researcher. In this situation, a role-based access control model needs to be implemented(11). Or the case where temporary access may be required for transient workers such as volunteers or students, which would necessitate regular and timely removal of access privileges when they leave. Finally, as information systems become increasingly complex, organizations often find that it is necessary to bring in third party specialists for system maintenance, and these individuals must still meet all the same privacy requirements as in-house users, complete with a contractual agreement that reflects this(12).  Advances in electronic databases bring with them not only data access challenges, but also those relating how to safely and efficiently store and archive data. We have moved well beyond the secure cabinet in a locked room. To enhance security, data should be stored on centralized servers, not on local workstations. In addition, to avoid increased security risk, servers should only be accessed using equipment maintained by specialized organizations. This can result in additional financial and time requirements, especially for data collection initiatives that span more than one health region, province/state/county/country, and each jurisdiction having its own policies, procedures and legislation. When electronic data are archived, special considerations must also be made to account for the risk of data corruption, and the possibility of the eventual inaccessibility of data due to outdated storage technology(13).  E-health applications, mobile electronic devices, and Internet communications also introduce new security risks that require appropriate safeguards. If mobile electronic devices such as PDAs, laptops and removable media (such as memory sticks, DVDs and flash  24  cards) are being used to access PHI, special security measures such as frequent data backups and special encryption procedures must be in place to mitigate the increased risk of loss of, physical damage to, or theft of these devices. Many e-health applications are also becoming increasingly dependent on Internet communications, as outlined in detail elsewhere(14). The general principles to keep in mind when considering Internet data transfers is that appropriate encryption methods are used to ensure that confidentiality is maintained, the information is not altered during the transfer, and that the sender of the information is always traceable(15).  Information system security measures need to be taken at every phase, from the inception of the system, to its development, administration and use(16). When the information system will be administered across healthcare organizations, the added challenge of having to meet the highly-variable policies and legal requirements of each organization is introduced. Not only is the security of PHI dependent on technology and policies, but more importantly it is dependent on users who employ the technology to abide by the policies in a securityconscious manner(17). While outside hackers are a concern, internal security breaches are probably more of a threat to the security of PHI, due more often to a lack of education and understanding than the express intent to breach security. It is the user who must not share their password with others, not log on to a computer and then leave it unattended, not remove computer equipment from secured areas, and not let security incidents go unreported.  The CPN relies on data access measures (i.e., username and password protected access), with storage of the database on a secure hospital network.  25  The implementation of any new information technology introduces new risk. It is important to identify and classify these risks based on their likelihood of occurring and the impact it would have if they did. By doing this, we are able to balance the benefit the technology provides with the risk that it introduces. It is becoming more common for jurisdictions to legislate risk assessment in the form of Privacy Impact Assessments and Threat and Risk Assessments to identify threats to privacy and security respectively, as well as safeguards against these risks. As technology is constantly being updated, so must we update these assessments(18). Research Ethics Boards Not only must healthcare and research institutions adhere to whatever regional and/or national privacy legislation exists, but they must also have an ethics committee in place to review how PHI is accounted for within the organization. Research Ethics Boards (REBs) in particular review all proposed research projects that involve the secondary use of data (i.e., data not used for direct clinical care administration). During the review process, REBs consider the purpose for the data collection, what type of patient consent process is necessary, and what limits should be placed on the collection, use, disclosure, and retention of the data. There are numerous challenges that we as health care professionals face, from the REB application process to following board recommendations and privacy laws.  In Canada, there is wide jurisdictional variability in how privacy legislation is interpreted(19), making national multi-centre research initiatives difficult to conduct. In the case of multicentre collaborative projects, the same project must be reviewed by a separate REB in each institution where data will be collected. Some of the inefficiencies of this system could be  26  avoided if Canada had a centralized REB application process. Not only is the application process itself a challenge, but so is the variation in each board’s interpretation of existing privacy legislation. Perhaps this is best demonstrated by the variability in patient consent requirements for the secondary use of personal health information. For example, whether or not (or in what format) patient consent will be required for the same multi-centre project may vary, with one centre requiring no patient consent in any form, another requiring that the patient has the option of opting out, and yet another stipulating that express (written or oral) consent must be obtained(20). These decisions may have a direct impact on outcomes.  In summary, balancing the need to protect patient privacy with the need to share information is not a simple task. As a community of healthcare providers, we have a responsibility to circulate information to ensure the best possible treatments for patients, while at the same time protecting the patient’s legal right to privacy. It can take significant time and resources to translate ethical and legal guidelines into concrete organizational policies and procedures, especially in light of the wide jurisdictional variations in legislative requirements in areas such as patient consent, safeguards, and risk assessment.  2.1.6 Stable funding This challenge reflects the reluctance of funding agencies to fund databases, as ongoing endeavors. As such, the national neonatal and paediatric networks (i.e., CNN and CAPSNet) as well as CPN, have chosen to focus on specific research projects to obtain initial funding. CPN has chosen to focus on threatened very preterm birth (i.e., threatened birth before 29 weeks’ gestation). This is both a major public health concern worldwide, and recognized to hold the greatest potential for improvement of outcomes for mothers and  27  babies. Initially, CPN is focusing on the management of the major spontaneous and iatrogenic causes of threatened very preterm birth at 220-286 weeks: spontaneous preterm labour, preterm pre-labour rupture of membranes (PPROM), gestational hypertension, intrauterine fetal growth restriction, and antepartum haemorrhage.  In order to demonstrate the need for, and justify, stable funding, we must be able to demonstrate that we have a core dataset onto which can be clipped additional data collection for specific research studies. We must also work with regional surveillance databases funded by government or health authorities; these regional databases also afford us the opportunity for knowledge interaction.  28  2.2 Bibliography (1) Kramer MS, Platt RW, Wen SW et al. A new and improved population-based Canadian reference for birth weight for gestational age. Pediatrics. 2001;108(2):E35. (2) SNOMED. www.update-software.com/history/clibhist.htm . 2007. (3) Strong DF and Leach PB. National Concultation on Access to Scientific Research Data: Final Report. 31-1-0005. Ottawa, Canada, Canada Institure for Scientific and Technical Information. (4) Carey TS, Howard DL, Goldmon M et al. Developing Effective Interuniversity Partnerships and Community-Based Research to Address Health Disparities. Academic Medicine. 2005;80(11):1039-1045. (5) Love D, Luis M, Paita C, Custer W. Data Sharing and Dissemination Strategies for Fostering Competition in Health Care. Health Servies Research. 2001;36(1):277290. (6) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (7) Croll PR, Croll J. Investigating risk exposure in e-health systems. International Journal of Medical Informatics. 2007;76:460-465. (8) Canadian Standards Association. Canadian Standards Association: Privacy Code. http://www.csa.ca/standards/privacy/code/Default.asp?language=english . 2007. (9) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (10) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (11) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (12) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (13) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (14) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (15) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada.  29  (16) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (17) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (18) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (19) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (20) COACH: Canada's Health Informatics Association. Guidelines for the Protection of Health Information. 16-12-0007. Toronto, Canada. (21) Menzies J, Magee LA, Li J et al. Instituting surveillance guidelines and adverse outcomes in preeclampsia. Obstet Gynecol. 2007;110(1):121-127.  30  CHAPTER 3: SECOND MANUSCRIPT 3.1 A Current Landscape of Provincial Perinatal Data Collection in Canada 1  3.1.1 Introduction In Canada, more than 350,000 babies are born each year. Preterm birth complicates 7.6% of births, with variations of ±15% among provinces(1). Despite improvements in prenatal care, preterm birth is still the most important cause of perinatal mortality and morbidity, and is recognized to hold the greatest potential for improvement of outcomes.  While it is certain that obstetric practice can influence neonatal outcomes, there are many controversies around the care of women at risk for very preterm delivery. The inaugural project of the Canadian Perinatal Network (CPN) is to identify, in the setting of threatened very preterm births at 22+0-28+6 weeks’ gestation, interventions that are related to good/poor maternal and perinatal outcomes and/or greater resource use. This requires collection of detailed information about obstetric practices which CPN eventually hopes to collect from all 24 tertiary perinatal units (TPUs) in Canada. CPN links with the well-established national CNN (Canadian Neonatal Network) and CAPSNet (Canadian Perinatal Surgery Network).  Many TPUs collect data as part of a provincial perinatal health program (PHP). These programs have been set up to monitor geographical disparities in perinatal health and  1  A version of this chapter has been accepted and is in press. Massey KA, Magee LA, Dale ,Claydon J, Morris TJ, von Dadelszen P, Liston RM and Ansermino JM on behalf of the Canadian Perinatal Network Collaborative Group (Appendix A) and the British Columbia Perinatal Health Program. A Current Landscape of Provincial Perinatal Data Collection in Canada. Journal of Obstetrics and Gynaecology of Canada 2008.  31  determinants of adverse outcomes (i.e. low socioeconomic status). Much of this information is used for regional reporting to government and policy makers; however, this is only a portion of the larger mandate of the PHPs. The mandate also includes peer review, education, guideline development, etc. We analyzed the breadth of data collection by Canadian PHPs in the hope of harmonizing data collection with that of CPN. Presented here is a detailed analysis of existing data collection by Canadian PHPs.  3.1.2 Methods Information about PHPs was gathered from websites, provincial reports, and one-on-one communication with co-ordinators. With the assistance of representatives at the BCPHP, we contacted the co-ordinators from existing PHPs, then requested and received electronic manuals for their databases. From these manuals, each data field and its definition were recorded to form a master table. This table was sent to the co-ordinators from each of the provincial PHPs to confirm accuracy of the PHP data collection. This included clarifying which data fields were collected from the medical record and which were obtained from the Canadian Institute for Health Information (CIHI) database, which includes demographics, ICD-10 diagnoses, and procedures.  The information in the table was organized into the following categories: demographics, past medical/obstetric/surgical history, current pregnancy, maternal complications and interventions, fetal complications and interventions, labour and delivery, and perinatal outcomes.  32  For the purposes of comparison, our analysis includes data fields and definitions from the Canadian Minimal Dataset (CMDS) proposed by the Canadian Perinatal Programs Coalition (CPPC) (2) and the CPN Database (May 2008 version) (3).  The Canadian Perinatal Programs Coalition Database Committee was created in 1988 and aims to serve as a vehicle for the exchange of ideas and information between professionals involved in the PHPs. At the moment, each province with a PHP collects its own data, but the CPPC is in the process of determining how and what could be shared nationally while still satisfying each province’s needs and privacy concerns. The CPPC has recognized both the variability in the information collected and the lack of standardized definitions. The CPPC has proposed a CMDS, and the provinces are currently comparing their respective databases to the CMDS. Each will make a commitment as to when they will be in compliance with the CMDS(2).  Although there are a variety of regional databases, for example SWOPP (Southwestern Ontario Perinatal Partnership), we included only provincial perinatal initiatives. It is also important to note that there are other provincial networks, such as FAN (Fetal Alert Network) in Ontario, which focus on birth defects and prenatal screening. Manitoba, New Brunswick, Northwest Territories, Nunavut, Quebec, and Saskatchewan were not included because they do not have a functional reproductive care database, although some are in the process of establishing one.  33  3.1.3 Results Since 2004, PHPs have, collectively, covered two thirds of births in Canada with perinatal data collection beginning as far back as 1988 (Atlee). The six programs used in this analysis are listed together with their characteristics in Table 1; this includes Nova Scotia Atlee Perinatal Database (Nova Scotia)(4), PEI (Prince Edward Island) Reproductive Care Program (PEI), NLPPP (Newfoundland and Labrador Provincial Perinatal Program)(5), NIDAY Perinatal Database (Ontario)(6), APHP (Alberta Provincial Perinatal Health Program)(7), and the BCPHP (British Columbia Perinatal Health Program). The data from Whitehourse (Yukon) are collected using the BCPHP database platform but are housed in their own separate territorial database (8).  Table 2 (Demographics) shows that while all programs collect maternal and infant personal identifiers, only two document ethnicity: NIDAY, which only defines aboriginal status and Atlee which is a non mandatory field. All PHPs gather socioeconomic status (SES) and body mass index (BMI) while lone parent status is collected by four PHPs. In comparison, CPN collects most maternal and infant identifiers, ethnicity (not mandatory) and other demographics listed in Table 2.  Table 3 (Past medical/obstetric/surgical history) shows most PHPs document previous diabetes mellitus, miscarriage/terminations, perinatal death, or Caesarean section. While the data fields are similar, many definitions are different. For example, pre-existing hypertension is defined by CPN as ‘a blood pressure of 140/90 mmHg or greater at least twice before pregnancy or before 20 weeks’ gestation’; Atlee defines pre-existing hypertension as ‘documentation that would indicate hypertension was a diagnosis or present before pregnancy’ (4), while the NLPPP uses only the ICD-10-CA code (O10). Most PHPs that collect parity also differ in the style of collection; PEI and NLPPP calculate parity as  34  [gravidity – (elective terminations+ miscarriages)]. BCPHP records parity>=1 if previous term+preterm >0 or if previous vaginal deliveries+previous caesarean deliveries>0 or living>0. Other PHPs such as Atlee and CPN calculate parity by totalling the previous number of stillbirths and live births, while the APHP collects parity as Gravida - Abortions - 1 (for the current pregnancy). Previous stillbirths are collected by most PHPs and are consistently defined as ‘the complete expulsion or extraction from its mother after at least 20 completed weeks gestation or after attaining a weight of at least 500g, of a product of conception in which, after the expulsion or extraction, there is no breathing, cardiac activity, pulsation of the umbilical cord, or unmistakable movement of voluntary muscle’. The only PHP that uses a different definition of stillbirth is Niday, which defines stillbirth as greater than 20 weeks, though the Niday system will allow births of 18-20 weeks. Preterm birth is documented by all PHPs, CPN, and the CMDS; however, its definition varies: most include babies born <37 weeks gestation (CPN, PEI, NLPPP, BCPHP), while others such as Atlee use LMP, ultrasound dating and infant clinical assessment to define preterm birth as desired (although < 37 weeks is the most common definition). Few PHPs document other previous medical complications (e.g. gestational diabetes mellitus or gestational hypertension). This pattern of data collection is reflected by CMDS. CPN collects more detailed previous medical and obstetric complications relevant to the major determinant of very preterm birth.  Table 4 (Current pregnancy) shows limited reporting on most variables by all programs. All PHP’s, CMDS and CPN collect information regarding expected date of confinement and multiple pregnancies. Although lifestyle choices such as smoking, and alcohol and illicit drug use are documented by all PHPs, they differ in the data fields and the amount of detail they collect. For example, most collect the number of cigarettes per day in pregnancy, but only two PHPs (Niday and PEI) and the CMDS specify whether or not there was smoking before or after 20 weeks. Moreover, only 2 PHPs (PEI and NLPPP) and the CMDS document  35  exposure to environmental tobacco and only the CMDS collects nicotine replacement therapy. The same trend applies with alcohol; while all PHPs, the CMDS, and CPN collect alcohol or illicit drug use in the pregnancy, only 2 PHPs, the CMDS, and CPN collect the amount of alcohol used (eg. binging, >3 drinks/day) while Atlee only collects alcohol abuse. Only 3 PHPs, the CMDS, and CPN specify which drugs (e.g. opiates, cannabis, cocaine, hallucinogens, stimulants, solvents, etc.). Many PHPs collect information on the nature of prenatal care and type of provider, as does the CMDS, however this is rarely collected by CPN. Fewer PHPs (BCPHP and PEI) collect Rh immunoglobulin (given antenatally or postpartum).  Table 5 (Maternal complications and interventions) shows that few interventions other than maternal transport, blood transfusion, and corticosteroids are collected by the PHPs consistent with the CMDS. These interventions are the focus of CPN data collection, which focuses on a high risk population as opposed to normal pregnancies. For example, while antibiotic use is collected by three PHPs and CPN, the PHPs only document if antibiotics were given antepartum, intrapartum and postpartum; they do not collect type or dosing all of which are captured by CPN. Also, while MgSO4 is collected by CPN in detail (type, route, indication, timing and date), Atlee and PEI only capture these data if they are used as a generic antihypertensive, anticonvulsant or to stop preterm labour.  While all PHPs, the CMDS, and CPN collect corticosteroids, they all capture them slightly differently. The PHPs collect yes/no for corticosteroids. PEI, Atlee, Niday, MDS, and CPN also identify whether there was either a complete or partial course. Some PHPs collect type of corticosteroid; Atlee and PEI collect dexamethasone or betamethasone, while CPN collects all types. The CMDS and CPN also collect timing, time of first dose and whether administration was antepartum, intrapartum, or postpartum. In addition, CPN collects  36  indication for drug initiation and route of administration. Most PHPs gather information on gestational hypertension, gestational diabetes mellitus, antepartum haemorrhage, group B streptococcus (GBS), chorioamnionitis, hepatitis B, HIV and urinary tract infections (UTIs) in the current pregnancy.  Table 6 (Fetal complications and interventions) demonstrates that half of the PHPs collect basic fetal complications, none of which are proposed in the CMDS. These complications and interventions are the focus of CPN data collection.  Table 7 (Labour and delivery) shows that basic variables are well-documented by all PHPs, although details such as induction of labour and delivery modes may differ. Neonatal data collected are listed in Table 8. All programs collect neonatal mortality and major morbidities.  3.1.4 Discussion The current landscape of Canadian provincial perinatal data collection covers a wide variety of information; this basic surveillance allows for sufficient information to oversee trends in birth outcomes and in interventions. The factors that are collected most frequently by provincial PHPs are population based data: demographics (both maternal and neonatal, including personal identifiers), past obstetrical history, maternal lifestyle, labour and delivery, and basic neonatal outcomes.  Ethnicity (Table 1) is documented by only two PHPs and CPN (non mandatory), This deemphasis reflects varying perceptions about ethnicity and continued debate about its significance with respect to health. Today, ethnicity is commonly defined as self-  37  identification with a culture (if any) of which individuals consider themselves to be a part. As such, ethnicity becomes subjective, making it almost impossible to relate ethnicity data as collected to specific genetic risks that correlate with ethnicity defined by genetic polymorphisms. As our population become more ethnically heterogeneous, ethnicity becomes less of a discernible predictor of health status. Rather, the cultural implications of ethnicity become more significant (9). Even if it were advantageous to use ethnicity as an indicator, large databases do not have the facility to collect this information, due to the complexities annotated above. Although the paucity of ethnicity data in the various Canadian PHP databases appears justified, we do need variables that denote socioeconomic status and lifestyle, such as beliefs, diet, and support.  Including personal identifiers in databases allows for confirmation of data, updates, corrections or improvements and/or definitive linkage to other provincial or national databases. In this analysis, all of the PHPs, the CMDS, and CPN allow for this. On the other hand, a patient’s health number or name allows for linkage to other identifiers, which could lead to other information and networks such as laboratory results at other facilities or other provincial or national networks. In this case, only PEI, NLPPP, Atlee, BCPHP, CMDS, and CPN collect this information.  By only including the direct variables collected by the programs, the current study is limited by not accounting for the potential linkages that databases could have with other resources for additional information using personal identifiers. For example, PHPs could potentially link a patient to other databases to receive laboratory results, prescriptions filled by pregnant women, or details of newborn screening and immunization. This could also allow for longitudinal linkages that would connect to educational or environmental databases for long term outcomes.  38  Our analysis shows that the CPN database was needed to collect detailed information on interventions. Within most PHPs there are insufficient data with respect to obstetric (and neonatal) practices to enable the identification of practices associated with good/poor maternal or perinatal outcomes. For example, using PHP data to determine the optimal type of antibiotic for PPROM. Specifically, no PHPs have detailed information about outpatient surveillance programs (Table 4), while the use of MgSO4 is only specifically named in two PHPs (PEI and Atlee). This is despite the use of MgSO4 being a grade I-A recommendation in the Canadian (Magee 2008), US (NHBPEP 2000), and Australasian (ASSHP 2000) guidelines for management of the hypertensive disorders of pregnancy, and, therefore, an auditable standard for provincial, national, and international benchmarking. When databases label MgSO4 as an antihypertensive it means that the use of an effective intervention cannot be assessed in that jurisdiction.  The variability between PHP databases is recognized by the CPPC, which proposed the CMDS as a first step towards harmonization, time and budgetary constraints allowing. While many PHPs collect similar information, many collect information in various ways (e.g. direct versus calculated parity) or have different definitions (e.g. for preterm delivery or pre-existing hypertension).  It must be recognized that no database can be all things to all people. The PHPs aim to monitor trends over time, evaluate guidelines, identify issues requiring further attention and for epidemiological and other research. Identifying geographical clusters of risk factors for targeting health care actions is a powerful tool for effective interventions(10), as PHP population cohorts represent the general maternity population. The PHPs should also ensure that there is equal emphasis on both determinants of health and interventions, such  39  as the grade I-A recommendations outlined by clinical practice guidelines. The PHPs would then be in a position to audit and validate recommendations in larger jurisdictions. Examples of grade IA recommendations that PHPs cannot report on are the CMDS’s lack of prenatal diagnosis or routine prenatal bloodwork. According to Summers et al. all pregnant women in Canada should be offered prenatal screening and second trimester ultrasound for dating growth and anomalies(11). Similarly, Keenan-Lindsay and Yudin (12) have advised that all women should be offered HIV screening at their first prenatal visit (I-A); however, APHP and NLPPP do not collect this information. Moreover, the use of MgSO4 as prophylaxis against, and treatment of, eclampsia in women with severe preeclampsia (I-A) (13) is only collected by Atlee and PEI (as a yes/no question). Every PHP collects information about antenatal corticosteroid use.  The challenge begins now that researchers and PHPs have started to become interested in inter-provincial and national comparisons, especially with the electronic health record (EHR) on the horizon. While the shift to electronic health records (EHRs) will be the main stimulus driving standardization of data definitions, PHPs will influence what data fields are collected. As it stands, most PHPs influence, revise, and create the antenatal, perinatal and labour and delivery forms. As such, electronic vendors will be seeking advice and expertise from the PHPs as to which data to include in EHRs.  3.1.5 Conclusion The variable language(s) spoken across perinatal databases identifies a great challenge that all local, provincial and national organizations will eventually have to overcome. The future direction is a standardized language on which all databases and the electronic health  40  record are based. Standardized definitions allow for data sharing supplemented by specific, focused data collection. The CPPC, through the CMDS, has started to tackle this issue. Converging with the new language of the EHR, SNOMED CT ®, will be the next step.  41  3.2 Tables from results Table 3.1 Canadian Perinatal Health Programs Variables  RCPNS (NS) 1973  PEI 1984  NLPPP (NL) 1979  NIDAY (ON) 1997  APHP (AB) 1992  BCPHP (BC) ∗ 1988  Database established Version used in this analysis  1988  1990  2001  1997  1992  1994  Apr 2007  2007  Oct 2007  Nov 2006  Oct 2007  Apr 2008  Population  All pregnanci es † and births in NS  All pregnanci es†¶  All births in ON║  All births All births in AB in BC  Number of Deliveries/yr in database Website  8500 (2006)  1400 (2005)  All pregnan cies† and births in 2 health authoriti es § 2800 (2002)  125 000 (2006/07)  45 200 (2006)  42 000 (2007)  http://rcp. nshealth.c a/ Yes  http://ww w.gov.pe. ca/health No  http://w ww.nlpp p.ca/ Yes  https://ww w.nidaydat abase.com No  http://ww w.aphp.c a/ No  http://ww w.bcphp. ca/ Yes  Program established  CIHI data included? RCPNS (Reproductive Care Program of Nova Scotia), NS (Nova Scotia), NSAPD (Nova Scotia Atlee Perinatal Database) PEI (Prince Edward Island), NL (Newfoundland), ON (Ontario), AB (Alberta), BC (British Columbia)  ∗  The B.C Perinatal Database is also used by the Yukon territories This only includes pregnancies reported after 20 weeks gestation ‡ Includes babies also born to Nova Scotia residents ¶ Includes babies also born to Prince Edward Island residents § This includes Eastern Health and Labrador-Grenfell Health ║ In 2006/07 95% of births in Ontario †  42  Table 3.2 Demographics Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  Maternal personal identifiers (one or more)  6  √  √  Hospital chart number 6  √  Provincial health number 4 ∗  √  Mother’s first/last name 3  √ √  Mother’s date of birth 6  √  √  †  4  Infant(s) personal identifiers (one or more)  6  √  √  Hospital chart number 5  √  √  Other  √  Provincial health number 4 Infant’s first/last name 3  √  Date of birth 6 Ethnicity  2  Language Spoken Socioeconomic status  √ ‡  √ √¶  1 §  6  √  Lone parent  4  √  √  BMI (calculated from prepregnancy height and weight)  6  √  √  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), BMI (body mass index),  ∗  This includes NLPPP, PEI, Atlee and BCPHP Some programs collect information about street address, city, health region, province, country of residence (as opposed to place of birth). ‡ Niday collects information on first national status (1st Natio, Metis, Inuit), while Atlee collects non mandatory information about the following race/ethnicities: Acadian , African Canadian, Asian, Caucasian, First Nations, Hispanic, Jewish, Mediterranean, Middle Eastern, Québécois, Other. ¶ CPN collects information on broad categories but the information is not mandatory § SES is defined by education, occupation and/or postal code †  43  Table 3.3 Past medical/obstetric/surgical history Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  Pre-existing Hypertension  5  √  √  Diabetes mellitus  5  √  √  Thromboembolism  2  √  Thrombophilia  2  √  Other ∗  4  Previous blood transfusion  1  MEDICAL HISTORY  OBSTETRIC/GYNAECOLOGICAL HISTORY  Gravidity  4  √  √  Parity  5  √  √  √  √  Reported directly 1 Calculated † 4 Spontaneous abortions  2  √  √  Elective terminations  2  √  √  ‘Abortions’ (spontaneous and therapeutic)  5  √  √  Uterine structural abnormalities  3  √  Prior obstetric history Previous congenital anomalies 2 Previous stillbirth 5  √  Previous neonatal death 5  √  √  ∗  Atlee, PEI and BCPHP include data from CIHI for ICD codes which reports on “disease” in the following systems: pulmonary, heart, renal, endocrine, and gastrointestinal, in addition to blood dyscrasias, neurological illness, neoplasms, psychiatric illness.  44  Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  Previous preterm delivery 6  √  √  Previous low birthweight 5 infant†  √  √  Previous GH 1  √  Previous PPROM 0  √  Previous GDM 3  √  Previous Caesarean section 6  √  √  Previous postpartum 3 depression Date of last pregnancy  0  √  Date of last delivery  1  √  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), D&E (dilatation and curettage), GDM (gestational diabetes mellitus), GH (gestational hypertension), PPROM (preterm pre-labour rupture of membranes)  45  Table 3.4 Current pregnancy Variables  N PHP’s that collect CMDS the information (n=6)  CPN  Prenatal care Date of first visit 5  √  No. visits in pregnancy 2  √  Attendance at classes/education 4  √  Antenatal care provider  4  √  3  √  Type  ∗  Reproductive assistance 3 Expected date of confinement  6  √  According to LMP 4  √  According to ultrasound 1 Details of first trimester ultrasound †  1  Multiple pregnancy  6  Type of twins 2  √ √  √  √ √  PRENATAL DIAGNOSIS Maternal serum screening  4  √  Nuchal translucency  2  Amniocentesis  3  √  Chorionic villous sampling  3  √  Uterine artery Doppler velocimetry  1  √  Rubella immune  2  √  Blood type  2  √  Rh type  1  Third Trimester Maternal hemoglobin  2  ROUTINE PRENATALS  ∗  This includes; Family Physician, Midwife, Nurse Practitioner, Registered Nurse, Obstetrician, Other, or None † Atlee Database reports on crown/rump length or biparietal diameter, head circumference, abdominal circumference, and femur length.  46  Variables  N PHP’s that collect CMDS the information (n=6)  Maternal “antibody conditions” ‡  3  CPN  LIFESTYLE ¶ Weight gain in pregnancy §  5  √  √  Smoking  6  √  √  Alcohol in pregnancy  6  √  √  Illicit drug use  6  √  √  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), BP (blood pressure), LMP (last menstrual period)  ‡  These are reported only by the Atlee Database, and comprise a mixture of allo- (to many red blood cell antigens including Rh, and “PL-A1 Platelet antigen negative”), auto-antibodies (ANA, antiDNA, anti-SSA and-SSB), and thrombophilias (i.e., Factor V Leiden mutation, anticardiolipin, lupus anticoagulant.  ¶ §  Defined as non heritable factors that may affect pregnancy. Calculated from pre-pregnancy weight and either pre-delivery weight or admission weight.  47  Table 3.5 Maternal interventions and complications Variables  N PHP’s that collect the CMDS information (n=6)  CPN  GENERAL INTERVENTIONS Transport Reason for transport 1 Transferring hospital 4  √  √  Receiving hospital 3  √  √  Expectant management  0  √  Outpatient surveillance program ∗  0  √  Blood transfusion  4  √  √  Drug therapy Preconceptual folate 2 Prevwentative therapies 0 Corticosteroids 6  √ √  √  Antihypertensives 3  √  MgSO4 † 2  √  Anticonvulsants 1  √  Tocolytics 3  √  Antibiotics 3  √  Other (list) 3  √  ICU admission  1  √  Self-measurement of BP  0  √  Fetal fibronectin testing 0  √  Preterm labour ⊥ PPROM ⊥  0  √  Short cervix ⊥  0  √  ∗  Maternal outpatient intensive surveillance care: antepartum home care, obstetrical day unit visit(s), or other (specify the type of care received, e.g. visits from home care nurse). Routine antenatal care, including outpatient clinic visits, is NOT intensive outpatient surveillance † Listed as an antihypertensive drug in BCPHP therefore not distinguishable  48  Variables  N PHP’s that collect the CMDS information (n=6)  CPN  Prolapsing membranes  0  √  Cervical surveillance 0  √  Cervical cerclage GH ⊥  ‡  2  √  5  √  √  HELLP syndrome 3  √  √  Eclampsia 3  √  √  Highest systolic BP 1  √  Highest diastolic BP 1  √  Worse dipstick proteinuria 0 antenatally  √  24hr urine collection 1 proteinuria (g/d)  √  Antepartum haemorrhage*  4  √  √  Due to placental previa 3  √  √  Due to placental abruption 3  √  √  Due to another cause 4  √  GDM  5  √  Fever  2  √  GBS infection or colonization 5  √  √  Chorioamnionitis 4  √  √  Infections Obstetric  Other ¶ Hepatitis A 2 ‡  Atlee and PEI collect this as coded data from CIHI (double check). CPN collects the following cerclage information: Timing (elective or rescue, before or during admission), Type [McDonald (transvaginal), Shirodkar (transvaginal), transabdominal, or unknown], and Removal [PPROM diagnosed, when APH diagnosed, with onset of labour, electively at 36 weeks, or at or after delivery].  ¶  Infections are collected by the BCPHP are postpartum wound infections, blood culture agents and other agents. These are coded so may not necessarily be coded as a standard. (Hep B, HIV testing and UTIs are standard with BCPHP).  49  Variables  N PHP’s that collect the CMDS information (n=6)  CPN  Hepatitis B 4  √  √  Hepatitis C 3  √  √  HIV 5  √  √  Herpes simplex genitalis 3  √  √  Varicella zoster 2  √  CMV 2  √  Tuberculosis 1 UTI 4  √  √  §  3  √  √  Other (list) 3  √  √  STD Abdominal inflammation/surgery  3  Coagulopathy  3  √  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), MgSO4 (Magnesium Sulfate) , BP (blood pressure), ICU (intensive care unit), PPROM (preterm pre-labour rupture of membranes), GH (gestational hypertension), HELLP (hemolysis, elevated liver enzyme, low platelet) syndrome, HIV (human immunodeficiency virus), GDM (gestational diabetes mellitus), GBS (Group B Streptococcus), CMV (Cytomegaalovirus), STD (sexually transmitted disease), UTI (urinary tract infection) ⊥ These are indications for enrolment in CPN and dates of onset and hospital admission are recorded.  §  Chlamdia, gonorrhea, syphilis, bacterial vaginosis, human papillomavirus  50  Table 3.6 Fetal complications and interventions Variables  N PHP’s that collect the CMDS information (n=6)  CPN  3  √  Antenatal surveillance ∗  4  √  Polyhydramnios  4  √  Amnioreduction 2  √  IUGR ⊥  Oligohydramnios  4  √  Amnioinfusion 3  √  Other procedures †  3  √  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), IUGR (intrauterine fetal growth restriction) ⊥ These are indications for enrolment in CPN and dates of onset and hospital admission are recorded.  ∗  Includes one or more of; cardiotocography, amniotic fluid volume, biophysical profile, umbilical artery Doppler velocimetry, middle cerebral artery Doppler, ultrasonographic estimated fetal weight, felt lung volume (with PPROM) by ultrasound or MRI. † Includes one or more of; amnion septostomy, bladder shunt, cord occlusion, intrauterine transfusion, laser surgery, multi-fetal reduction, pleuroamnio shunt, open fetal surgery, fetal blood transfusion, fetal drainage, fetal stent placement, cordocentesis etc  51  Table 3.7 Labour and delivery Variables  N PHP’s that CMDS collect the information (n=6)  CPN  Length of stay ∗  5  √  Admission number  5  Induction of labour Indication 6 Method (list) 6  √ √  √  √  √  N induction attempts 2 Labour augmentation  6 Method (list) 4  ROM Spontaneous/artificial 3  √  Date/time 5  √  √  6  √  √  Fetal presentation 1st stage of labour  Duration of 1st stage 5 st  Rate of dilation for 1 stage General tract and perineal trauma nd  2 stage of labour duration  √  2 5  √  6  rd  3 stage of labour duration Fetal surveillance in labour  ‡  †  √  4 ¶  Scalp pH 2 Meconium staining 4  √  Shoulder dystocia  3  √  ANALGESIA/ANAESTHESIA  6  √  √  ∗  Calculated by Day of discharge-day of admission and/or date and time of discharge - date and time of birth.. This refers to the delivery admission except for CPN which also prefers to other antepartum admissions. Atlee and NLPPP also have the capability to collect antepartum admissions prior to deliver admissions † Rate = (Cervical full dilation- dilation at admission)/ (time at full dilation-time at admission) ‡ Episiotomy, Laceration, Cervical tear ¶ External/internal and intermittent/continuous FHR monitoring, auscultation of FHR, scalp pH  52  Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  Analgesia Type None 6  √  √  Narcotics (includes PCA pump) 5  √  √  Nitrous oxide 3  √  √  Local anaesthetic (e.g., pudendal 5 block)  √  √  Regional (epidural, spinal, or 5 spinal-epidural)  √  √  General 5  √  √  6  √  √  Unknown 2  √  √  Other  §  Maternal complications of anaesthesia  3  Mode of delivery  6  √  √  Vaginal 6  √  √  Forceps/vacuum 5  √  √  √  √  VBAC 5 Caesarean section 6 Indications for Caesarean section 3 Type 6  √ √  √  Cervical dilatation at C/S 3 Intrapartum blood loss  3  Type of care-giver who delivered ║  5  √  Mother’s discharge disposition (home, transfer, unk) alive/dead  4  √  √  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), C/S (Caesarean section), FHR (fetal heart rate), PCA (Patient Controlled Analgesia) Pump, VBAC (vaginal birth after Caesarean section)  §  Includes non pharmalogical methods such as Transcutaenous Nerve stimulation, Hypnosis, Acupuncture, Aromatherapy, New Age, Deep Breathing/ Meditation ║ GP, obstetrician, RN, midwife, nurse practitioner, resident, other MD, paramedic, at home, surgeon, no attendant  53  Table 3.8 Perinatal outcomes Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  OUTCOME Termination of pregnancy 0  √  Stillbirth 6  √  √  Live birth 6  √  √  Neonatal death 6  √  √  Age at death ∗ 3  √  √  Autopsy 3  √  Cause 3  √  √  6  √  √  Meconium aspiration  3  √  Fracture  3  √  pH 5  √  √  Base excess 5  √  √  For neonatal death  Gestational age NEONATAL  Cord blood gases  pCO2 2  √  Apgars 1 minute 6  √  5 minutes 6  √  10 minutes 4  √  Vitamin K administered  2  Resuscitation/Ventilation  6  Admission to hospital  3  Gender  5  Rh type  1  ∗  √  √  √  √  Can be calculated by Date of death – Date of birth  54  Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  Blood group  1  Birthweight 5  √  √  Length 4  √  Head circumference 6  √  Anthropomorphic measurements  Major Congenital anomaly †  4  √  Chromosomal abnormality  3  √  Neonatologist  2  Admission to NICU  6  √  √  TRIPS score 1 SNAP II score 1 Length of stay (days) ‡ 5  √  Mechanical ventilation & 5 mode  √  Total parenteral nutrition 3  √  Transfusion 2  √  Complications of 4 prematurity ¶ Procedures §  √  3  √ √  Complications of 3 procedures Central venous catheter(s) 1 Infant medications  2  Other neonatal complications ║  4  √ √ √  √  †  Specific list from Niday and CMD includes: Anencephaly, Spina Bifida Meningocele, myelomeni ngocele, Hydrocephaly, Cleft lip, Cleft palate, Down syndrome, Neural Tube defect, CNS, GI,Renal, Respiratory, Cardiovaacular, Musculoskeletal, Other. Others (BCPHP, PEI and Atlee) are coded, while APHP is Y/N. CPN has a field to enter the data ‡ Calculated from Discharge date- Admission date ¶ BPD (bronchopulmonary dysplasia) or CLD (chronic lung disease), ROP (retinopathy of prematurity), severe IVH (intraventricular haemorrhage), NEC (necrotizing enterocolitis), HIE (hypoxic-ischemic encephalopathy)/convulsions/seizures, neonatal sepsis § Laparotomy, thoracotomy, craniotomy, ECMO (extracorporeal membrane oxygenation).  55  Variables  N PHP’s that collect the information (n=6)  CMDS  CPN  Infant feeding Mother’s intention to 5 breastfeed Early breast contact 2 Infant feeding (type) 5  √  Reason for substitute 2 Jaundice/phototherapy  3  Newborn screening & type  1  Immunizations  1  Discharge disposition Details of transfer ∗∗ 4  √  √  Needed home oxygen 2 Neonatal Follow up clinic  1  PHP (Perinatal Health Program), CMDS (Canadian Minimal Data Set), CPN (Canadian Perinatal Network), C/S (Caesarean section), ECMO (extracorporeal membrane oxygenation), FHR (fetal heart rate), NICU (neonatal intensive care unit), SNAP (score for neonatal acute physiology), TRIPS (temperatory, respiratory status, systolic BP, response to noxious stimuli)  ║  Includes: fetal malnutrition/soft tissue wasting, patent ductus arteriosus, pulmonary hypertension of newborn, respiratory distress syndrome, birth asphyxia sequelae, neoplasms  ∗∗  Included at least two of the following; date, time, destination and reason for transfer  56  3.3 Bibliography (1) Canadian Perinatal Health Report. 2000. Ottawa, Health Canada. (2) Canadian Perinatal Programs Coalition (CPPC). Canadian Perinatal Programs Coalition (CPPC) Terms of Reference. 2004. (3) Canadian Perinatal Network. http://www.cpn-rpc.org/ . 2006. (4) Nova Scotia Atlee Perinatal Database. http://rcp.nshealth.ca/rcp_3347.html . 2008. (5) Newfoundland and Labrador Provincial Perinatal Program. http://www.nlppp.ca/perinataldatabase.htm . 2008. (6) Niday Perinatal Database. https://www.nidaydatabase.com/info/guide_definitions.shtml . 2008. (7) Alberta Perinatal Health Program. http://www.aphp.ca/ . 2007. (8) B.C. Perinatal Health Program. http://www.bcphp.ca/ . 2008. (9) Callister P, Didham R, Potter D, Blakely T. Measuring ethnicity in New Zealand: developing tools for health outcomes analysis. Ethn Health. 2007;12(4):299-320. (10) Clark RC, DeMarco ML. Development of an information management system using a strategic planning process. Top Health Inf Manage. 2001;22(2):44-51. (11) Summers AM, Langlois S, Wyatt P, Wilson RD. Prenatal screening for fetal aneuploidy. J Obstet Gynaecol Can. 2007;29(2):146-179. (12) Keenan-Lindsay L, Yudin MH, Boucher M et al. HIV screening in pregnancy. J Obstet Gynaecol Can. 2006;28(12):1103-1112. (13) Magee L.A, Helewa M, Moutquin J.M vDP. Diagnosis, Evaluation and Management of the Hypertensive Disordersof Pregnancy. Journal of Obstetrics and Gynaecology Canada. 2008;30(3):1-48.  57  CHAPTER 4: THIRD MANUSCRIPT 4.1 What is SNOMED CT® and why should the ISSHP care? 1 While there are currently many electronic health record (EHR) vendors, each with their own electronic system and rules, it remains that there is only one electronic language that will be a key player of the success of electronic health. The Systematized NOmenclature of MEDicine Clinical Terms (SNOMED CT®) is the future universal language of the EHR. It is owned and governed by the International Health Terminology Standards Development Organization (IHTSDO®) which contains 9 charter members (Australia, Canada, Denmark, Lithuania, Sweden, the Netherlands, New Zealand, United Kingdom and United States). These countries participate in the ongoing development of SNOMED CT®(1) and are granted free access to the database. These countries also have very active memberships within the ISSHP. IHTSDO residents of these countries have free access to SNOMED CT® and are expected to contribute to the ongoing development of SNOMED CT®. As it stands in Canada today, SNOMED CT® is available to the Standards Collaborative Members of Infoway. In the meantime, Canada has set up a Canadian National Product Center (NPC) to provide management, distribution, support, and surveillance of the database.  SNOMED CT® was initially a joint initiative between the National Health System in the UK and the College of American Pathologists. This dialog resulted in a universal medical dialect with sufficient detail to have utility for by clinicians in all healthcare disciplines and settings. SNOMED CT® allows human readable terms to be given computer readable codes and descriptions. This essentially permits synchronized standardized nomenclature. This is available not only for the EHR but also for any computerized entry systems, prescribing 1 A version of this chapter has been accepted for publication. Massey KA, Ansermino JM von Dadelszen P, Morris TJ (1), Liston RM, Magee LA. What is SNOMED CT® and why should the ISSHP care? Hypertension in Pregnancy 2008.  58  order entry, laboratory results, result reporting, genetic databases, surgical procedures etc. This clinical terminology supersedes and it is expected to replace existing classification systems such as the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) as classified by the World Health Organization (WHO). In particular, SNOMED CT® distinguishes itself from all other classifications systems by its ability to form a web of knowledge around a single concept. This may involve various descriptions, definitions, qualifiers or relationships to other concepts (e.g. gestational hypertension and preeclampsia).  As it constantly grows and continues to develop, SNOMED CT® contains more than 357,000 concepts with unique meanings and formal logic-based definitions organized into hierarchies (Figure 4.1) such as body structure (i.e. mitral valve), event (e.g., pregnancy), clinical findings, and disease (gestational hypertension), etc (2). Each concept also has many ‘descriptors’ (synonyms) as well as a unique meaning and formal logic-based definition(1). ‘Relationships’ link the concepts together. Along with anatomical definitions, SNOMED-CT® contains standards for all diagnoses, laboratory results, and non-laboratory interventions and procedures.  59  Figure 4.1 The hierarchies of SNOMED CT ®  At present, there is no special interest group for the hypertensive disorders of pregnancy (HDP) within the SNOMED CT® initiative. We believe that members of the ISSHP, and others interested in the HDP, should take a leadership role in this regard for a number of reasons.  60  To access variability of ‘gestational hypertension’ terms we looked to the current version of SNOMED CT® for a list of current concepts available within the hypertensive disorders of pregnancy. Unfortunately, of the little information that does exist in this subspecialty, some of it was clinically incorrect. Specifically the concepts and hierarchies around the HDP are not in total alignment with currently agreed concepts in the area. For example, gestational hypertension was found under the concept of ‘hypertension AND/OR vomiting complicating pregnancy childbirth AND/OR puerperium’. Listed as descriptors (synonyms) of gestational hypertension were both appropriate (e.g., ‘pregnancy-induced hypertension’) and inappropriate terms (e.g., ‘vomiting complicating pregnancy’). As subtypes of gestational hypertension, there were types of ‘chronic hypertension’ and ‘renal hypertension’, as well as ‘toxemia of pregnancy’. This means that the latter was listed as a separate entity from ‘preeclampsia’ (Figure 4.2).  61  Figure 4.2 A SNOMED CT® map for the concept of ‘Gestational Hypertension’  A common misconception of SNOMED CT® is that it will house quantitative criteria or threshold measurements (such as blood pressure); however, these fall outside the current realm of SNOMED CT® definitions. Rather than standardizing clinical implementations the goal is to standardize the language as a means to communicate.  The challenge of SNOMED CT® for many is to define concepts using relationships instead of clinical criteria. For example, pre-eclampsia uses a measurement of proteinuria (≥0.3 g/d) and gestational hypertension as a standard clinical definition. To solve this in SNOMED  62  CT® one might propose ‘proteinuria in pregnancy’ as a new concept which would then be linked to ‘pre-elampsia’ and to ‘gestational hypertension’ to form relationships that define pre-eclampsia (Figure 4.3).  Figure 4.3 A potential map for the concept of ‘Pre-eclampsia’  Even with its concerns of inappropriate terms and challenge of defining relationships, SNOMED CT® still proves to be extremely important and relevant to our clinical world. Making our health care world universal is an enormously long and difficult task; SNOMED CT® represents a small step in the right direction. Without ensuring that we are all speaking the same language, the EHR will merely be an easier retrieval and storage device. The more we ignore this problem the farther we are from making the EHR a daily reality. Since  63  all vendors in the charter countries will eventually be required to use SNOMED CT® as an essential language it’s imperative that subspecialties understand and participate.  Specifically it is in the best interest of the subspecialty clinicians and researchers to become contributors and developers in the SNOMED CT® standard. This dynamic process is further stimulated by the fact that existing data should frequently be reassessed and re-evaluated by experts in the field. How we as clinicians and researchers think of the HDP must be reflected in SNOMED CT® terminology in order for the electronic health record to be relevant, and to ensure that we use the same language in the HDP. Leaders in subspecialty fields have an influential role in development of relevant SNOMED CT® terminology to facilitate comparison between studies, aggregation of studies by meta-analysis, and benchmarking between centers within and between countries.  In the future, SNOMED-CT® will form the basis of the electronic health record, clinicians and researchers in all fields must take an interest in this new terminology. Since there is currently no special interest group within the IHTDSO for obstetrics or for the HDP, we invite interested members to contact us at the Canadian Perinatal Network project e-mail address (cpn@cw.bc.ca).  64  4.2 Bibliography (1) Canada Health Infoway. About SNOMED CT. http://sl.infowayinforoute.ca/content/dispPage.asp?cw_page=snomedct_e . 2006. (2) IHTSDO®. International Health Terminology Standards Development Organisation. http://www.ihtsdo.org . 2008.  65  CHAPTER 5: CONCLUSION 5.1 Databases as the foundation of evidence based medicine Effective clinical practitioners need to be well educated and prepared to use evidence based research about optimal practices(1). This relationship is clear: practitioners that use clinical evidence experience improvements in patient care(2). Clinical information is commonly accessible though reports, peer reviewed papers and systematic reviews via organizations such as the NHS Center for the Reviews and Dissemination as well as the Cochrane Collaboration. However, the very basis of evidence based medicine always stems from some form of collected data. As such, data collection agencies, whether they are regional or national, play an integral role in shaping the way we practice medicine. These organizations are needed to constantly create and evaluate the information reaching the clinicians and ensure that it is correct, updated and accessible (3). Electronic access to such health data can significantly improve information seeking and sharing which is vital to the limited time available in busy clinical settings(4).  As is the case with many aspects of health care, most databases have been developed at different times with different software and general framework. It would be completely inefficient to simply discard existing systems or databases and replace them with new ones(5). A more efficient use of time would be to introduce techniques that aid in system integration allowing for collaborative efforts.  Collaborative efforts in health care at the clinical level can be seen through multidisciplinary patient care teams who are well suited to manage the increasing complexity of patient conditions and treatments(6). The sharing of information among and between teams is  66  commonly seen as directly affecting patient care. This is also the case with data sharing. As such, more and more data collection agencies should be inquiring about information seeking to make their system more efficient and effective. Reddy and Spence define collaborative information seeking (CIS) as “an information access activity related to a specific problem solving activity …directly and/or through texts as information …either in a specific workplace setting or in a more open community or environment”(6).  Data collected via perinatal health programs offers a medium for collaboration and a flow of information. Moreover, the potential ways that these networks can interact create a wealth of information allowing us to not only pose more detailed questions but also to be able to accurately answer them. This process is referred to as knowledge interaction.  5.2 Knowledge interaction Quality improvement involves sharing knowledge within and between organizations whether they be hospitals, networks, or regions (national and international).  There continue to be attempts to build organizations committed to building knowledge and sharing it. In 2003, The National Health Service (NHS), the United Kingdom’s publicly funded health care system, introduced the “National Knowledge Service” (NKS) as a step towards the knowledge management program. There are three types of knowledge that the NKS concentrates on: knowledge from research (evidence), knowledge from collected and audited data (statistics) and knowledge from the experience of patients and clinicians(7). The basic idea of the NKS is to use the leading and latest medical knowledge to guide health care professionals and patients in decision making(8). The benefits to the quality of  67  health care derived from already existing resources, research, data and experience has a much larger and global impact when organized and combined with each other. This reduces the most common problems in our current system including: unknown variables in clinical practises, errors and waste of data, inappropriate care, unsatisfactory patient experiences, and failure to implement new knowledge and technology(7).  Knowledge interaction involves interactive PDSA cycles between different data agencies. This flow of information allows for different and more difficult clinical questions to be addressed. This concept of knowledge interaction led us to look at a model of interaction between CPN, our national academic database, and the British Columbia Perinatal Health Program (BCPHP), our provincial (regional) reproductive care program (Figure 5.1).  PDSA: Plan, Do, Study, Act knowledge flow  Regional Network  PDSA  PDSA  PDSA  PDSA  Literature  National Network  Figure 5.1 Model of knowledge interaction (9)  68  In this way, PDSA cycles occur (and interact) in both tertiary perinatal units with high risk women (CPN), and community hospitals that serve the low risk, general obstetric population (BCPHP). The following figures and descriptions are examples of various situations where knowledge interaction would prove to be not only applicable but also effective.  BCPHP PDSA cycle informing a CPN cycle (Figure 5.2) Maternal weights in pregnancy have been increasing over time, and so have relevant pregnancy complications (e.g., Caesarean section, and post-operative wound infections). The BCPHP monitors such parameters, and they could, for example, take an idea about intervention to CPN. This initiative might target obese women following a complicated pregnancy, and attempt to alter diet and lifestyle. If successful in effecting weight loss, and in improving pregnancy outcome, this initiative could be fed back to BCPHP for wider implementation.  69  PDSA: Plan, Do, Study, Act knowledge flow  Regional Network  PDSA  PDSA  PDSA  PDSA  Literature  National Network  Figure 5.2 BCPHP PDSA cycle informing a CPN cycle (9)  CPN PDSA cycle informing a BHPHP cycle (Figure 5.3) One question that CPN addressing is whether beta-lactam antibiotics are associated with more necrotizing enterocolitis, a serious bowel problem in preterm babies, than are macrolide antibiotics, when these antibiotics are given for PPROM. If CPN finds that one antibiotic is superior, then a policy for that antibiotic’s administration can be implemented provincially, with ongoing audit of the neonatal outcomes of interest.  70  PDSA: Plan, Do, Study, Act knowledge flow  Regional Network  PDSA  PDSA  PDSA  PDSA  Literature  National Network  Figure 5.3 CPN PDSA cycle informing a BHPHP cycle (9)  5.3 Semantic integration 5.3.1 Semantic discrepancies As demonstrated by Chapter 2 (Manuscript 1), semantic interoperability proves to be one of the greatest challenges of data sharing. Before any communication can be made between and among health jurisdictions, it is important to recognize that there is a great deal of semantic discrepancy among perinatal health programs (Manuscript 2). The complexity of semantic discrepancies stems from two factors: language and data. While conclusion may be drawn from data based on the collection of facts, facts can only be communicated through language(10). Together form the basis of semantic heterogeneity within databases.  71  Semantic heterogeneity is formally defined as a fact having multiple descriptions within various networks. This seems to be the most significant and rate limiting issue of data sharing.  Bishr describes two different forms of semantic heterogeneity; cognitive and naming(11). ‘Cognitive heterogeneity’ occurs when individuals from the same jurisdiction share similar facts expressed through language, which may not translate to other agencies through the use of data. In other words, the same facts (shared concepts) have different definitions(11). This form of heterogeneity proves to be more difficult to resolve (than naming heterogeneity) since it requires a common nomenclature or definitions. The form of semantic integration used to solve cognitive heterogeneity is called peer-to-peer or point-to-point. This involves direct communication and negotiation between all participating systems (12). Since every system is directly connected to another (Figure 5.4) point-to-point can prove to be extremely difficult, especially in health care as it requires full agreement upon an official definition.  Figure 5.4 Point-to-point strategy to integrate systems. Modified from Wangler (5)  72  The second form of semantic heterogeneity described by Bishr is ‘naming heterogeneity’ which is the easier of the two to resolve(11). This involves one concept with one definition having many names or terms used to address it. Another way to describe this is when two different concepts have the same name(10). Naming heterogeneity can be resolved using simple linking and mapping techniques. Rule based links are often another solution. These are based on centralized repository, as is the case with Population Data B.C. This organization has started working on a widespread data warehouse. Here, data is funneled from various disciplines and organizations. These include things such as PharmaNet which tracks all prescription drugs dispensed in the province, and the Early Development Instrument which is a measure of child school readiness (13). Population Data B.C aims to not only organize the data but to allow for data linkage providing semantic interoperability. When using rule based links, unidentified data is often stored in a type of data warehouse which is available for extraction(14). Data linking or integration is often controlled at the provider level with a of process manager who handles requests and retrieval of information between the systems(15). The process manager visualizes and facilitates communication and conveys information to all participants in the system (Figure 5.5). This reduces the complexity by decreasing the amount of direct interface connections and requires only one source point to change (the message to the process manager) if there is a change to one of the systems(5).  73  Figure 5.5 Rule based linking  All of these efforts move toward a universal solution. With the expansion of surveillance systems and databases, we have become experts in data collection; however, we need to work towards making optimal use of such. To put it simply, we have a wealth of knowledge but few tools for communicating and sharing them effectively. As it stands we have multiple terms with the same definition, as well as single terms representing different independent ideas. It is imperative that we standardize our dialect in order to discontinue repeating the cycle of discrepancies that currently riddles our clinical knowledge.  Effective communication allows comparison of data from many sources, as well as collaboration among various organizations. There have been many efforts to move towards one single communicative architecture that facilitates interoperability; one such is SNOMED CT®. Once we have built a language, fine tuning our semantics will allow us to have a complete description of a concept and, therefore, a better understanding of our perinatal language  74  5.3.2 Preserving our perinatal knowledge through semantic integration As previously mentioned, the clinical medical field is one of most semantically challenged areas as the challenge is two fold. First is the challenge to agree to the standardization of the language and ensure that the same nomenclature is used. Second is the challenge that standardized clinical interpretations are needed to ensure quality of care(16). Here, a certain degree of conformity is needed for integration into a standardized system. The latter of the two limitations is strongly applicable and of concern to the world of perinatal health.  Although semantic interoperability in perinatal language has been recognized by few individual researchers, Schuurman is one of these few who have made a difference. Her interest is in the social connotations of women’s reproductive roles due to the events of pregnancy. This led her to discover large discrepancies in the definition of stillbirth at 20 weeks gestation and miscarriage occurring <20 weeks gestation. She found that in B.C due to the stigma of stillbirths, most doctors record fetal demise just beyond 20 weeks as miscarriages not stillbirths, while Alberta has reported strict standards that have increased numbers of stillborns(17). Moreover, she found that in other areas of Canada fetal death is associated with fetal demise although there is no differentiation between stillbirths, miscarriage or therapeutic abortions. To tackle these issues of naming heterogeneity she uses technology to link similar concepts (i.e. still birth, miscarriage and fetal demise) between smaller databases in provincial jurisdictions based on their relationships.  Before we resort to cross mapping terms, early integration is optimally preferred. As new networks frequently emerge regionally, provincially and nationally, early efforts should be taken to integrate their language and definitions with other similar programs. Such is the  75  case with CPN. After surveying of the various definitions among programs it became clear that CPN did not want to simply re-invent the wheel when creating its definitions. Instead, CPN semantically integrated its data fields with both its sister network, the Canadian Neonatal Network (CNN) as well as the Canadian Minimal Dataset (CMDS) proposed by the Canadian Perinatal Programs Coalition (CPPC)(18).  These are important steps towards acknowledging and recognizing our semantic differences and the problems that this creates. In Canada, with few and finite perinatal health programs data integration via merging and mapping concept relationships may be a possible solution. However, this is only an acute solution. If we truly want to tackle the problem with some long term influence we should be seeking larger initiatives such as SNOMED CT®. Participating in such endeavors will be a step in the right direction for the future of perinatal health, something that all programs express interest in. Semantic interoperability is the precursor to creating a sustainable, interactive and fully functional electronic health system.  5.4 Future direction: The Electronic Health Record At present, in the course of clinical care, data are collected, but not in a way that facilitates knowledge generation and improves care. Since patients are treated in a variety of settings, by a variety of healthcare professional teams, the first goal of an EHR will be to provide for retrieval and storage; this has been achieved in many settings. However, with standardization of terminology, we can move beyond this to interpretation of the data and generation of knowledge. SNOMED-CT® will be an integral part of the new language that  76  healthcare providers learn during their early training and that forms the basis of EHR data screens.  The adoption of technology within the overall health system has been notably slower than other industries. Interest in its use, however, is growing exponentially. In the United Kingdom, the National Health Services (NHS) aims to have 60,000,000 patients with a centralized EHR by 2010, while in Canada the goal is to provide EHR to fifty percent of the population by the same year. This goal was set by Canada Health Infoway, a “not-for-profit corporation mandated to accelerate the modern systems of information technology” (19) after receiving $500 million from the Canadian government in 2001 followed by an additional $600 million in 2003.  Canada Health Infoway and its partners aim to deliver the electronic heath record (EHR) and as a result, provide timely access to accurate information, enhance disease management, and enable long term sustainability. Built on patient and provider registries, the EHR will facilitate sharing of clinical information allowing for easier identification of patient problems, laboratory test results, clinical reports, immunization history, and medication profiles, while avoiding duplicate history taking, investigations, treatments, and data collection(19). The systems likely to be the most valuable will be adaptable to changing needs within the organization, and will be able to be linked to other systems.  The United Kingdom is currently a leader in the switch to the electronic health record; however, the major concern at the moment is safeguarding electronic data. One instance of mishandling identifiable data occurred in October 2007 when two computer discs were lost  77  that contained the personal details of 25 million parents and families in the UK receiving child benefits(20). During this ‘child benefit scandal’ the discs, owned by Her Majesty's Revenue and Customs, were sent by junior staff from HM Revenue and Customs (HMRC) to the National Audit Office (NAO). The data were sent as unrecorded internal mail on October 18 and by October 24 the NAO complained to the HMRC that the data had not yet been received(21). It was not until early November that senior officials at the HMRC were informed of the loss. The lost data, which was believed to affect approximately 25 million people in the UK, included names, addresses and dates of birth of children, as well as the National Insurance numbers and bank details of their parents(22). The blame was initially placed on junior officials from the government; however, upon closer review, many believe that the problem lay with senior management and the flow of identifiable data between centres. This is based on the fact that the NAO had actually requested that identifiable bank information and details be stripped from the data they were to receive. Unfortunately, this did not occur since the HMRC claimed the removal of such data would be "too costly and complicated"(23).  Episodes like this have caused concern among local centers sharing detailed identifiable data with the NHS. Potential breaches of security and privacy failures that could potentially arise increase greatly when sharing large amounts of data from multiple sources and locations. In an editorial by McGilchrist et al.(24) he identifies the concept of ‘pseudonymisation’. This concept allows the combination of anonymised data such as laboratory reports and diagnosis, but excludes linkages to any other personal identifiers such as personal health number, date of birth and postal code. The anonymised data could be accessed via a new common identifier, such as a study or research code, that links different records to the same patient whose personal information may otherwise only be  78  accessed with a personal health number. For example if a pregnant woman with diabetes arrives at Site A and her doctor at that site wishes to get her results from her eye appointment at the ophthalmologist (site B), he may send Site B her “code” (Figure 5.7). The main concept is that any data within the site is secure and may include personal identifiers; however, any data accessed from outside the site requires a code and does not include personal identifiers.  Figure 5.7 Data exchange that decreases flow of personal identifiers(9)  79  Many measures have been taken to avoid security breaches. These includes the reconfiguration of all personal and company computers to prevent downloading of data on to removable media which could only be reactivated with the approval of a senior manager for a specific purpose(25). As it stands now authorized system users and management are the only ones with access privileges to personal health information. As such there needs to be no malicious intent, inappropriate use, incompetence or mistakes (26). Continuing to give personal health information requires a great deal of trust from the patients and if their private personal information is displaced or mishandled, then the confidence and trust in such agencies is diminished.  Ultimately the challenge that prevails is balancing the need for health information with the need for security. As it stands, there is an imperative need to proceed with the EHR as many believe that the benefits greatly outweigh the security risks. As Flegel suggests in his editorial, an electronic patient record could inform a physician about a patient’s history in a timely matter so they do not prescribe a drug that is harmful to the patient(27). It seems as though the situation has become counterproductive by allowing the potential risks to impede progression. At some point we must recognize that our concerns with privacy and security may actually do more harm than good.  5.5 Final thoughts (summary) Continuous quality improvement involves iterative cycles of practice, change, and audit, of ongoing clinical care. An obvious prerequisite to this is the need for ongoing data collection about interventions and outcomes, as well as demographics, pregnancy characteristics, and neonatal care that may affect the intervention-outcome relationship.  80  In Canada (as in some other developed countries), much of the country is covered by regional reproductive care databases that monitor geographical trends and disparities in health outcomes. As such, there is little information about interventions, especially outside the period of labour and delivery. Also, there is no standardization of definitions, and efforts to produce a ‘minimal dataset’ have not yet yielded agreement, even after many years of work.  A more comprehensive system is required. The rate-limiting step in this process of data acquisition and knowledge generation/translation will be semantics/definitions, not computers and cables. Moving in this direction, particularly with respect to standardization of definitions, would lay the foundation for the electronic health record, which cannot build its foundation that is our current definitional structure in women’s health and obstetrics. Efficiency and economy aside, the EHR is needed for patient safety, however, concerns about privacy and access remains major barriers.  The next critical step is standardization of our scientific database language. SNOMED CT® will underpin these efforts. Research groups and subspecialty clinicians are encouraged to contact the International Health Terminology Standards Development Organization (IHTSDO) for information about who to contact in their own country in order to access the SNOMED CT ® database. By doing so, one can get a better understanding of what currently exists in the database, and what remains to be clinically described. The step in standardizing our medical terminology is the collaboration of the clinical experts in the field,  81  not solely the onus of informaticians and system administrators. Together we can build an effective system improving care for women and their babies everywhere.  82  5.6 Bibliography (1) Prescott K, Lloyd M, Douglas HR et al. Promoting clinically effective practice: general practitioners' awareness of sources of research evidence. Fam Pract. 1997;14(4):320-323. (2) Magrabi F, Westbrook JI, Coiera EW. What factors are associated with the integration of evidence retrieval technology into routine general practice settings? Int J Med Inform. 2007;76(10):701-709. (3) Prescott K, Lloyd M, Douglas HR et al. Promoting clinically effective practice: general practitioners' awareness of sources of research evidence. Fam Pract. 1997;14(4):320-323. (4) Magrabi F, Westbrook JI, Coiera EW. What factors are associated with the integration of evidence retrieval technology into routine general practice settings? Int J Med Inform. 2007;76(10):701-709. (5) Wangler B, Åhlfeldt R.M, Perjons E. Process Oriented Information Systems Architectures in Healthcare. Health Informatics Journal. 2003;9(4):253-265. (6) Reddy M.C, Spence P.R. Collaborative information seeking: A field study of multidisciplinary patient care team. Information Processing and Management. 2008;44:242-255. (7) Muir Gray JA. Best Current Evidence: Concepts and Plans. 5-28. 2006. National Health Services. (8) De Lusignan S, Wells S., Shaw A. et al. A knowledge audit of the managers of primary care organizations: top priority is how to use routinely collected clinical data for quality improvement. Medical Informatics & the Internet in Medicine. 2005;30(1):69-80. (9) Massey KA, Morris TJ, Liston RM et al. Building Knowledge in Maternal and Infant Care. In: Parry D, Parry E, editors. Medical Informatics in Obstetrics and Gynecology. Auckland: Medical Information Science Reference, 2008. (10) Yaser Bishr. Overcoming the semantic and other barriers to GIS interoperability. International Journal of Geographical Information Science. 1998;12(4):299-314. (11) Bishr Y. Overcoming the semantic and other barriers to GIS interoperability. International Journal of Geographical Information Science. 1998;12(4):299-314. (12) Schuurman N, Leszczynski A. A method to map heterogeneity between near but non-equivalent semantic attributes in multiple health data registries. Health Informatics J. 2008;14(1):39-57. (13) Population Data BC. Population Data BC. http://www.popdata.bc.ca . 2008.  83  (14) Arzt NH. The new alphabet soup: models of data integration, part 2. J Healthc Inf Manag. 2006;20(2):9-11. (15) Schuurman N, Leszczynski A. A method to map heterogeneity between near but non-equivalent semantic attributes in multiple health data registries. Health Informatics J. 2008;14(1):39-57. (16) Schuurman N, Leszczynski A. A method to map heterogeneity between near but non-equivalent semantic attributes in multiple health data registries. Health Informatics J. 2008;14(1):39-57. (17) Schuurman N, Leszczynski A. A method to map heterogeneity between near but non-equivalent semantic attributes in multiple health data registries. Health Informatics J. 2008;14(1):39-57. (18) Canadian Perinatal Programs Coalition (CPPC). Canadian Perinatal Programs Coalition (CPPC) Terms of Reference. 2004. (19) Canada Health Infoway. 2015: Advancing the Next Generation of Health Care in Canada. 2006. (20) Brown apologises for records loss. BBC 2007 Nov 21. (21) Wikipedia contributors. 2007 UK child benefit data scandal. Wikipedia, The Free Encyclopedia . 29-2-2008. (22) Wikipedia contributors. 2007 UK child benefit data scandal. Wikipedia, The Free Encyclopedia . 29-2-2008. (23) Wikipedia contributors. 2007 UK child benefit data scandal. Wikipedia, The Free Encyclopedia . 29-2-2008. (24) mcgilchrist M, Sullivan F, Kalra D. Assuring the confidentiality of shared electronic health records. British Medical Journal. 2007;335:1223-1224. (25) Parliamentary reporter. Three million records lost in another government data scandal. Computing 2007 Dec 18. (26) mcgilchrist M, Sullivan F, Kalra D. Assuring the confidentiality of shared electronic health records. British Medical Journal. 2007;335:1223-1224. (27) Flegel K, Hebert PC, Stanbrook MB et al. Getting to the electronic medical record. CMAJ. 2008;178(5):531, 533.  84  Co-Investigators (CIHR)  Steering Committee Member  Collaborator (Site Investigators)  Appendix A: Canadian Perinatal Network Collaborative Group Members  Principal Investigator (CIHR)  APPENDICES  Victoria Allen  IWK Health Centre, Halifax NS  9  9  9  Mark Ansermino  BC Children’s Hospital, Vancouver BC  9  9  François Audibert  Hôpital Sainte-Justine, Montréal QC  Jon Barrett  Women’s College Hospital, Toronto ON  Updated: July 28, 2008  9 9  9 9  9  9  Emmanuel Bujold  Centre Hospitalier de l'Université Laval (CHUL), Québec QC  Craig Burym  St. Boniface General Hospital, Winnipeg MB Winnipeg Health Science Centre, Winnipeg MB  9  George Carson  Regina General Hospital, Regina SK  9  Joan Crane  Women's Health Program, Eastern Health, St. John's NF  Jerome Dansereau  Victoria General Hospital, Victoria BC  Nestor Demianczuk  Royal Alexandra Hospital, Edmonton AB  Duncan Farquharson  Royal Columbian Hospital, New Westminster BC  9  Rob Gratton (pending)  Saint Joseph’s Health Centre, London ON  9  Shoo Lee  iCARE, University of Alberta, Edmonton AB  Robert Liston  BC Women’s Hospital, Vancouver BC  9  9  9  Laura Magee  BC Women’s Hospital, Vancouver BC  9  9  9  Angela Mallozzi (pending)  Royal Victoria Hospital, Montréal QC  9  Sarah McDonald (pending)  McMaster University Medical Centre, Hamilton ON  9  Jean-Marie Moutquin  Centre Hôspitalier Universitairé de Sherbrooke, Sherbrooke QC  Femi Olatunbosun  Royal University Hospital, Saskatoon SK  9  Jean-Charles Pasquier  Centre Hôspitalier Universitairé de Sherbrooke, Sherbrooke QC  9  Bruno Piedboeuf  Centre Hospitalier de l'Université Laval, Québec QC  Frank Sanderson (pending)  Saint John Regional Hospital, Saint John NB  Graeme Smith  Kingston General Hospital, Kingston ON  9  9  9  Peter von Dadelszen  BC Women’s Hospital, Vancouver BC  9  9  9  Mark Walker  The Ottawa Hospital, Ottawa ON  9  9  9  Wendy Whittle  Mount Sinai Hospital, Toronto ON  9  9  Liz Whynot  BC Women’s Hospital & Health Centre, Vancouver BC  9  Stephen Wood  Foothills Medical Centre, Calgary AB  9  9 9  9  9  9  9  9  9  9 9  9  85  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items