Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Adverse drug events associated with the antidotes for toxic alcohol poisoning : a comparison of ethanol… Lepik, Katherine Jean 2007

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-ubc_2007-0160a.pdf [ 4.79MB ]
Metadata
JSON: 831-1.0100871.json
JSON-LD: 831-1.0100871-ld.json
RDF/XML (Pretty): 831-1.0100871-rdf.xml
RDF/JSON: 831-1.0100871-rdf.json
Turtle: 831-1.0100871-turtle.txt
N-Triples: 831-1.0100871-rdf-ntriples.txt
Original Record: 831-1.0100871-source.json
Full Text
831-1.0100871-fulltext.txt
Citation
831-1.0100871.ris

Full Text

ADVERSE DRUG EVENTS ASSOCIATED WITH THE ANTIDOTES FOR TOXIC ALCOHOL POISONING: A COMPARISON OF ETHANOL AND FOMEPIZOLE by Katherine Jean Lepik B.Sc.(Pharm.), University of British Columbia, 1983 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T OF THE R E Q U I R E M E N T S FOR THE D E G R E E OF M A S T E R OF SCIENCE in The Faculty of Graduate Studies Health Care and Epidemiology T H E UNIVERSITY OF BRITISH C O L U M B I A Apri l 2007 © Katherine Jean Lepik, 2007 ABSTRACT Background: The toxic alcohols methanol and ethylene glycol (antifreeze) can cause severe morbidity or death i f ingested. Two antidotes are available for treatment: ethanol, which is inexpensive but has many adverse effects and fomepizole, which is expensive but allegedly safer. There are no comparative studies of these antidotes. Study objectives were to characterize the adverse drug events (ADEs) associated with ethanol and fomepizole and to compare the A D E rates between antidotes. Methods: This retrospective cohort study included patients >13 years hospitalized for toxic alcohol poisoning and treated with at least one dose of antidote between 1996-2005. Charts from ten B.C. hospitals were double reviewed by abstractors. Potential ADEs were evaluated by a consensus panel of three toxicologists. The primary outcome was any (i.e. at least one) A D E . Each poisoning incident contributed treatment-days from antidote start until either the onset of the first A D E or completion of antidote treatment. Time to event curves were generated by the Kaplan-Meier method. Cox regression was used to model the association between A D E rate and antidote type, using ethanol as the reference group. Results: After exclusions, there were 130 ethanol and 42 fomepizole treated incidents. Central nervous system ADEs occurred in 48% of ethanol and 2% of fomepizole treated incidents, cardiovascular ADEs in 7% ethanol and 1% fomepizole, gastrointestinal ADEs in 9% ethanol and 7% fomepizole. Ethanol was associated with minor hypoglycemia in 4%, phlebitis in 4% and diuresis in 6% of cases. The proportion of incidents with any A D E was 57% for ethanol and 12% for fomepizole. The unadjusted A D E rate (95% CI) per antidote treatment-day was 0.93 (0.87, 0.98) for ethanol and 0.13 (0.02, 0.24) for fomepizole. A significantly higher A D E rate for ethanol was evident within two hours after antidote start and was sustained throughout treatment. The Cox regression model adjusted for baseline severity of illness gave a rate ratio of 0.17 (0.07, 0.42), showing a six-fold reduction in A D E rate with fomepizole relative to ethanol. Discussion: Ethanol-related central nervous system effects accounted for most of the difference between treatment groups. The results suggest that fomepizole produces fewer ADEs than ethanol. u T A B L E OF CONTENTS ABSTRACT ii T A B L E OF CONTENTS Hi LIST OF TABLES. . vi LIST OF FIGURES viii LIST OF ABBREVIATIONS ix ACKNOWLEDGEMENTS x 1 INTRODUCTION 1 1.1 Overview 1 1.2 Burden of illness of toxic alcohol poisoning 2 1.3 Toxicology and clinical management of toxic alcohol poisoning 3 1.3.1 Mechanism of toxicity, clinical effects 3 1.3.2 Diagnosis .....4 1.3.3 General approach to treatment 5 1.3.4 Ethanol : 6 1.3.4.1 Ethanol historical perspective and pharmacology 6 1.3.4.2 Ethanol dose and administration 7 1.3.4.3 Ethanol adverse effects 7 1.3.4.4 Available ethanol products .• 9 1.3.5 Fomepizole 9 1.3.5.1 Fomepizole historical perspective and pharmacology 9 1.3.5.2 Fomepizole dose and administration 9 1.3.5.3 Fomepizole adverse effects 11 1.3.5.4 Available fomepizole products 11 1.3.6 Which antidote? The place of ethanol and fomepizole in therapy 11 1.4 Evaluating safety: The study of adverse drug events 14 1.4.1 Overview and definitions of drug safety terminology 14 1.4.1.1 Adverse drug reaction 14 1.4.1.2 Adverse drug effect... 14 1.4.1.3 Serious adverse drug reaction 15 1.4.1.4 Adverse event 15 1.4.1.5 Adverse drug event 15 1.4.1.6 Medication error 16 i i i 1.4.1.7 Drug related problem 17 1.4.2 Causality assessment of adverse drug events 17 1.4.2.1 Expert opinion 18 1.4.2.2 Algorithms 18 1.4.2.3 Bayesian approach 20 1.4.2.4 Adverse drug event causality assessment in patients with toxic alcohol poisoning 21 1.5 Thesis theme, objectives and hypothesis 23 1.6 References 24 2 ADVERSE DRUG EVENTS ASSOCIATED WITH T H E ANTIDOTES FOR TOXIC A L C O H O L POISONING: A COMPARISON OF ETHANOL AND FOMEPIZOLE... . . . . 31 2.1 Introduction 31 2.1.1 Background : 31 2.1.2 Importance 32 2.1.3 Goals of this investigation 33 2.2 Methods 34 2.2.1 Study design 34 2.2.2 Setting 34 2.2.3 Selection of subjects 34 2.2.4 Data collection and processing 35 2.2.5 Methods of measurement 36 2.2.6 Outcome measures and treatment group assignment 38 2.2.7 Primary data analysis 38 2.2.8 Sensitivity analyses 40 2.3 Results 41 2.3.1 Characteristics of study subjects 41 2.3.2 Main results .43 2.3.2.1 A D E evaluation and inter-rater agreement 43 2.3.2.2 Summary of antidote ADEs 45 2.3.2.3 Onset time of ADEs 47 2.3.2.4 Comparison of ADEs between treatment groups 48 2.3.3 Sensitivity analysis 49 2.4 Limitations 51 2.5 Discussion 52 2.6 References 54 iv 3 DISCUSSION 57 3.1 Clinical and health policy implications 57 3.1.1 Choice of antidote 57 3.1.2 Antidote stocking decisions 58 3.2 Ongoing research and future study 59 3.3 References 60 APPENDICES 61 Appendix A : New onset symptoms during antidote therapy.... 61 Appendix B : Examination of possible biases influencing study results 63 B . l Overview 63 B.2 Caregiver bias in symptom documentation 64 B.3 Influence of previous reports of antidote adverse reactions on reviewer assessment.... 65 B.4 Influence of serum ethanol level in identifying incidents with high antidote dose 66 B.5 Comparison of panel and blinded reviewer A D E evaluations 67 B . 6 Conclusion 72 Appendix C: Serious antidote-related adverse drug events (ADEs) 73 C. l Definition of serious antidote-related A D E s 73 C. 2 Case summaries of serious antidote-related A D E s 74 C.2.1 Serious ethanol-related A D E s 74 C. 2.2 Serious fomepizole-related A D E s 79 Appendix D: Cox proportional hazard regression model 80 D. l Covariate selection 80 D.2 Model testing for proportionality 81 D. 2.1 Any A D E model 81 D.2.2 Severe A D E model 84 Appendix E: Certificates of ethical approval 87 References for Appendices 93 v LIST OF TABLES Table 1.1 B .C . Drug & Poison Information Centre intravenous (IV) ethanol dosage guidelines? Table 1.2 Manufacturer's dosage recommendations for fomepizole (Antizol™) 10 Table 1.3 B .C . Drug & Poison Information Centre dosage recommendations for fomepizole. 10 Table 2.1 Clinical and demographic characteristics of toxic alcohol poisoning incidents, by antidote treatment group 42 Table 2.2 Therapeutic interventions for toxic alcohol poisoning incidents, by antidote treatment group 43 Table 2.3 Proportion of toxic alcohol poisoning incidents with antidote-related adverse drug events (ADEs), determined by consensus agreement of expert reviewers 45 Table 2.4 Association between antidote-related adverse drug event (ADE) rate and antidote type in toxic alcohol poisoning incidents 49 Table 2.5 Sensitivity analyses for any antidote-related adverse drug event (any ADE) in toxic alcohol poisoning incidents 50 Table A . l Symptoms excluded from review by expert panel 61 Table A.2 Toxic alcohol poisoning incidents with new onset symptom(s) during antidote therapy and proportion of incidents assessed by expert panel as having ADE(s) 62 Table B.l Previously reported adverse reactions to ethanol and fomepizole 65 Table B.2 New onset symptoms during antidote therapy: classified by whether of not the symptom is known to be a common adverse reaction associated with the antidote 66 Table B.3 Case characteristics relevant for the causality assessment of potential antidote-related ADEs 68 Table B.4 Inter-rater agreement between expert reviewers on independent evaluation of any ADE 69 Table B.5 Panel decision regarding any ADE, all reviewed incidents: characteristics associated with yes/ no decision 70 Table B.6 Case characteristics associated with outcome of any ADE: consensus panel and blinded reviewer evaluations 71 vi Table B.7 Panel and blinded reviewer decision regarding any ADE, subset of 78 reviewed incidents 72 Table D. l Univariate analyses of potential predictor variables in the Cox regression model... 80 Table D.2 Test for significant slope in Schoenfeld residuals plot, any ADE model 81 Table D.3 Test for significant slope in Schoenfeld residuals plot, severe ADE model 85 vii LIST OF FIGURES Figure 1.1 Mechanism of methanol and ethylene glycol toxicity and therapeutic interventions 3 Figure 1.2 Relationship between adverse drug reactions and medication errors 16 Figure 2.1 Time to onset of first antidote-related adverse drug event (any ADE) in toxic alcohol poisoning incidents 47 Figure 2.2 Time to onset of first antidote-related adverse drug event (severe ADE) in toxic alcohol poisoning incidents 48 Figure B.l Comparison of consensus panel and reviewer 4 evaluation of antidote ADEs in the blinded and unblinded subsets of incidents 69 Figure D. l Survival curves for adjustment variable A P A C H E II score (3 categories), any ADE model 82 Figure D.2 Scatter plots of Schoenfeld residuals, any ADE model 83 Figure D.3 Survival curves for adjustment variable pre-treatment agitation, severe ADE model 84 Figure D.4 Survival curves for adjustment variable ingestion of ethanol pre-hospital, severe ADE model 85 Figure D.5 Scatter plots of Schoenfeld residuals, severe ADE model 86 LIST OF ABBREVIATIONS A P A C H E II acute physiologic and chronic health evaluation II score BP blood pressure CNS central nervous system DPIC B C Drug & Poison Information Centre ED emergency department F female GCS Glasgow coma scale HCO3- serum bicarbonate H R heart rate ICU intensive care unit IV intravenous M male M A P mean arterial pressure R R respiratory rate n/a not available ix ACKNOWLEDGEMENTS This project would not have been possible without the support of many friends, colleagues and mentors. In particular, I would like to thank my supervisor Dr. Adrian Levy for encouraging me to pursue a Master of Science degree and to strive for excellence. Many thanks to supervisory committee members Mr. Derek Daws for facilitating financial support for my research and for accommodating my studies through a flexible work schedule, to Dr. Roy Purssell for his ongoing support and encouragement from the beginning of this project and to Dr. Boris Sobolev for his thoughtful guidance and emphasis on practical approaches to problem solving. I greatly appreciate the enormous commitment of co-investigators Dr. Gunnar Erhardt, Dr. Christopher DeWitt and Dr. James Kennedy and research assistant Dr. Jane Brignall. Many thanks to my family, friends and colleagues for their patience and support during all stages of this project. Funding from the Michael Smith Foundation for Health Research and the B.C. Drug and Poison Information Centre is gratefully acknowledged. x 1 Introduction 1 INTRODUCTION 1.1 Overview Methanol and ethylene glycol are found in automotive antifreeze. When ingested, these toxic alcohols cause severe metabolic acidosis, visual impairment (from methanol) or kidney failure (from ethylene glycol) and death unless treatment is initiated within hours of ingestion. Treatment includes the use of an antidote to prevent formation of toxic metabolites. For many decades, ethanol was the only antidote available for toxic alcohol poisoning. Although inexpensive, ethanol is potentially toxic and is difficult to administer. In 2001 a new antidote, fomepizole, was introduced to the Canadian market. Fomepizole is considered relatively safe and is easy to administer, but costs $1000-$4000 per treatment course. To date, there have been no clinical trials or observational studies which compare the safety, efficacy or effectiveness of these two antidotes in poisoned patients. The purpose of this study was to estimate the incidence of adverse drug events associated with the therapeutic use of ethanol and fomepizole for toxic alcohol poisoning and to compare the adverse event rates associated with each antidote. The following review of the literature summarizes the epidemiology, toxicology and treatment of toxic alcohol poisoning, with emphasis on antidote therapy, and introduces the terminology and methods used in medication safety research. 1 1 Introduction 1.2 Burden of illness of toxic alcohol poisoning Poisoning is an important cause of intentional and unintentional injury. In 2003, poisoning resulted in approximately 300 fatalities, 5000 hospital admissions and 5200 hospital emergency department visits in British Columbia (B.C.) and cost an estimated $53.5 million in direct healthcare costs.1 Non-pharmaceutical toxins such as solvents, corrosives and carbon monoxide account for a small proportion of poisonings, but can produce severe injury and disability.2 The toxic alcohols methanol and ethylene glycol are the principle ingredients of automotive antifreeze products and are among the most common causes of hospitalization for poisoning by non-pharmaceuticals.2 A retrospective study of toxic alcohol poisoning in B.C. found 201 hospital admissions for suspected methanol and ethylene glycol poisoning province-wide between Apri l 1996 and March 2001.3 The medical chart was reviewed for 118 of these cases and hospital separation abstracts were available for the remaining 83 hospitalizations. Adjustment for estimated exclusions in the 83 cases without medical charts gave an average of 38 toxic alcohol- related hospitalizations per year in B.C. , or 1.0 /100,000 population per year. Combined data from B.C. coroner's records and in-hospital deaths provide an estimate of five deaths per year from toxic alcohol poisoning, or 0.14 /100,000 population per year. The majority of the hospitalizations (89%) and all of the pre-hospital deaths involved adults or adolescents. For the 118 cases where reason for exposure was known, fewer than 20% involved unintentional exposure or substance abuse, the remainder were cases of attempted suicide or unknown intent. Patients hospitalized for suspected toxic alcohol poisoning require substantial use of healthcare resources. O f the 201 hospitalizations, 38% required transfer to a larger health care facility, with the average transfer distance ranging from 45 km in the lower mainland to 640 km in the north.3 One third of the transfers required emergent use of air ambulance. Among the 118 cases with medical chart review data, 109 (92%) received antidote therapy, 79 (67%) required hemodialysis, 78 (66%) spent at least one day in the intensive care unit, and the mean duration of hospital stay was five days for methanol and 14 days for ethylene glycol poisoning. 3 ' 4 2 1 Introduction 1.3 Toxicology and clinical management of toxic alcohol poisoning 1.3.1 Mechanism of toxicity, clinical effects The parent compounds methanol and ethylene glycol have a low order of toxicity and their clinical effects are limited to symptoms of intoxication or gastrointestinal irritation at high doses.5"8 The major toxic effects are the result of metabolic byproducts which accumulate as the toxic alcohols are broken down within the body. The main elimination pathway for both methanol and ethylene glycol begins with metabolism by the enzyme alcohol dehydrogenase (ADH), followed by one or more additional metabolic steps, see Figure 1.1. The principle toxic metabolites are formic acid (formate anion) from methanol and glycolic acid (glycolate anion) from ethylene glycol. Figure 1.1. Mechanism of Methanol and ethylene glycol toxicity and therapeutic interventions3 © BC Drug & Poison Information Centre 2005, used with permission Methanol or Ethylene Glycol (minor toxicity) metabolism via enzyme alcohol dehydrogenase Toxic Acid Metabolites formate (from methanol) glycolate (from ethylene glycol) accumulation (slow elimination) Clinical Effects: - metabolic acidosis - formate: visual impairment, brain injury, - glycolate: kidney damage, brain injury - death "Simplified mechanism omits several intermediate steps ANTIDOTE inhibits alcohol dehydrogenase & prevents metabolite formation Toxic Acid Metabolites T sodium bicarbonate corrects metabolic acidosis HEMODIALYSIS enhances elimination of methanol, ethylene glycol and their toxic metabolites The minimum toxic doses of the toxic alcohols are not clearly established. Ingestion of greater than 5-10 mL of 100% methanol or ethylene glycol is potentially harmful to an adult and 3 1 Introduction warrants assessment at a health care facility, while greater than 30 mL methanol or 60 mL ethylene glycol is potentially fatal i f untreated.9 Following ingestion of methanol or ethylene glycol, symptoms resembling ethanol intoxication may be apparent within two hours. There is a latent period of approximately 2-6 hours for ethylene glycol or 4-12 hours for methanol, during which time the toxic metabolites accumulate to a concentration capable of causing metabolic acidosis.6"13 Worsening acidosis is associated with a progressive deterioration in clinical condition including nausea, vomiting, rapid, laboured breathing and altered mental status progressing to coma, seizures, and death. Severe methanol poisoning may cause permanent or partially reversible visual impairment, while ethylene glycol may cause acute renal (kidney) failure which requires weeks of hemodialysis support. Survivors of severe toxic alcohol poisoning may have permanent neurologic injury. Prognosis is best predicted by the serum concentration of formate or glycolate. 1 4 - 2 1 Neither of these metabolites are measured in clinical laboratories, but both correlate well with arterial pH or serum bicarbonate, such that the more severe the metabolic acidosis, the greater the risk of fatal outcome or sequelae. Any factors which permit further accumulation of acid metabolites, such as delayed presentation to hospital or delayed initiation of treatment, increase the risk of a poor outcome. 1.3.2 Diagnosis Diagnosis of toxic alcohol poisoning is made on the basis of a history of exposure and/ or the presence of characteristic clinical symptoms and laboratory test results. Other medical conditions which may resemble toxic alcohol poisoning include alcoholic or diabetic ketoacidosis, lactic acidosis and salicylate poisoning. A quantitative serum methanol or ethylene glycol concentration measured by gas chromatography provides a definitive diagnosis, but few hospitals have timely access to this assay. In B.C. , only three hospital laboratories can measure toxic alcohol levels: two in the greater Vancouver area and one in Victoria. 2 2 Turn around time for test results is approximately four hours in hospitals with on site gas chromatography, but two to three days for hospitals in the interior and northern regions of B . C . 3 4 1 Introduction As a result of the difficulty making a definitive diagnosis and the high risk associated with delaying treatment, antidote therapy is usually started whenever there is a strong suspicion of toxic alcohol exposure. Antidote may therefore be given to patients who are eventually found not have toxic alcohol poisoning. It is difficult to quantify the proportion of patients who receive antidote but do not require it, because they cannot be readily identified through a search of hospital separation records. Most of these cases are either treated as outpatients in the emergency department, or have a hospital admission that does not include toxic alcohol poisoning among the diagnostic codes. 1.3.3 General approach to treatment Treatment of toxic alcohol poisoning requires correction of the metabolic acidosis and prevention of further toxic metabolite accumulation. The three central components of treatment are intravenous sodium bicarbonate, antidote and hemodialysis (see Figure 1.1). Two antidotes are currently available for toxic alcohol poisoning: ethanol and fomepizole. Both act by competitively inhibiting A D H to prevent metabolism of methanol and ethylene glycol to formate and glycolate, respectively. Once the major metabolic pathway has been blocked, the toxic alcohols are slowly excreted via the urine and lungs. 2 3" 2 9 During antidote therapy the elimination half-life is prolonged to approximately 50 hours for methanol and 19 hours for ethylene glycol (longer in renal failure), therefore complete elimination may take many days in a severely poisoned patient. Antidote treatment does not correct the toxic effects of metabolites which have already formed. Prompt administration of sodium bicarbonate is required for treatment of metabolic acidosis. Hemodialysis enhances elimination of the parent toxic alcohols and their metabolites and is used both to correct severe metabolic acidosis and to shorten the treatment course. 6 ' 8 ' 1 5 ' 2 1 ' 2 3 ' 3 0 - 3 5 Patients who develop renal failure secondary to ethylene glycol poisoning require hemodialysis support until renal function recovers. Antidote treatment is indicated for patients with a combination of exposure history, clinical symptoms and laboratory test results suggestive of toxic alcohol poisoning, or a confirmed toxic alcohol serum concentration of 20 mg/dL (equivalent to 6 mmol/L methanol or 3 mmol/L 5 1 Introduction ethylene glycol). 6" 8' 1 0 - 1 3 The toxic alcohol serum level thresholds are arbitrary and based on historical convention, and some authors have suggested they are too conservative.12' 3 6 Hemodialysis is indicated for patients with persistent metabolic acidosis, serious symptoms and/ or renal failure. 1 0 ' 1 1 There is lack of consensus regarding the threshold for dialysis for the sole purpose of enhancing toxic alcohol elimination in a clinically well patient. Historically, a toxic alcohol concentration greater than 50 mg/dL (15 mmol/L methanol or 8 mmol/L ethylene glycol) has been considered indication for dialysis. 1 0 ' 1 1 Recently, it has been suggested that patients can be managed safely with fomepizole alone i f they do not have metabolic acidosis or renal impairment; however, this approach would result in a more prolonged treatment course . 1 1 ' 1 3 ' 3 7 - 3 9 Antidote treatment, with or without hemodialysis is continued until acidosis is corrected, symptoms have resolved and serum toxic alcohol concentration is <20 mg/dL (<6 mmol/L methanol or < 3 mmol/L ethylene glycol). 1 0 ' 1 1 As with the threshold for starting therapy, the threshold for ending therapy is also arbitrary, and some references suggest higher or lower toxic alcohol concentrations as the endpoint. 1 2 ' 1 3 ' 4 0 In B .C . a typical treatment course for methanol or ethylene glycol poisoning is 24-36 hours of antidote therapy with one or more sessions of hemodialysis to enhance toxin removal. 3 ' 4 1 1.3.4 Ethanol3 © B.C. Drug & Poison Information Centre 2005, adapted with permission 1.3.4.1 Ethanol historical perspective and pharmacology Ethanol acts as a competitive inhibitor of A D H because it has a much greater affinity for the enzyme than either methanol or ethylene glycol. 8 ' " ' 4 2 ' 4 3 Experimental data indicate that as little as a 1:1 molar ratio of ethanol to methanol wi l l inhibit >90% of methanol metabolism, while an even lower ratio may be sufficient for ethylene glycol. It has been suggested that a 1:4 ratio of serum ethanol to serum toxic alcohol concentration should usually provide adequate antidotal protection. 8 ' 4 3 The true "therapeutic range" for ethanol is therefore dependent on the toxic alcohol serum concentration. The antidotal properties of ethanol were first recognized in the 1940s when Oluf Roe observed that patients who co-ingested both ethanol and methanol suffered a less severe clinical course than those who ingested methanol alone. 4 4 ' 4 5 Roe suggested that a blood ethanol concentration 6 1 Introduction of 100 mg/dL (22 mmol/L) might be beneficial. Six decades later, this arbitrary figure (or a range from 100-150 mg/dL; 22-32 mmol/L) remains the conventional "therapeutic" serum ethanol concentration for treatment of toxic alcohol poisoning. 1 0' u ' 4 2 ' 4 3 ' 4 6 ' 4 7 In perspective, this therapeutic range is higher than the Canadian legal limit for driving under the influence of alcohol (blood ethanol 80 mg/dL, equivalent to 19 mmol/L serum ethanol)8' 4 8 and would be expected to produce symptoms of intoxication in the average adult. 1.3.4.2 Ethanol dose and administration A variety of ethanol dosage guidelines have been published. These typically present a range of infusion rates based on the patient's weight, usual ethanol drinking habits and use of concurrent hemodialysis. 5 ' " ' 1 2 ' 4 0 ' 4 2 ' 4 3 ' 4 7 ' 4 9 ' 5 0 These rates are suggested starting points and must be adjusted based on serum ethanol concentrations. The recommended loading dose ranges from 7.5-10 mL/kg of a 10% solution (or equivalent) with initiation of the maintenance infusion either following or given concurrently with the loading dose.5' " ' 1 2> 4 0 ' 4 2> 4 3 ' 4 7- 4 9> 5 0 Table 1.1 shows the dose recommended by the B.C. Drug and Poison Information Centre (DPIC). Table 1.1 B.C. Drug & Poison Information Centre intravenous (IV) ethanol dosage guidelines9 © BC Drug & Poison Information Centre 2005, used with permission Treatment Patient Characteristics Dose of IV ethanol as a 10% v/v in D5W solution Instructions Loading dose Al l patients* 7.5 mL/kg Infuse over 60 minutes Maintenance dose without hemodialysis Average adult** 1.0 mL/kg/hr then adjust Begin after completion of loading dose; adjust rate to maintain serum ethanol 100-150 mg/dL (22-32 mmol/L) Heavy drinker 1.5 mL/kg/hr then adjust Maintenance dose during hemodialysis Average adult** 2.0 mL/kg/hr then adjust Heavy drinker 3.0 mL/kg/hr then adjust * omit or reduce loading dose if patient has an elevated serum ethanol level prior to treatment ** children, non-drinkers and patients with liver disease may require a lower dose than the average adult 1.3.4.3 Ethanol adverse effects There are no prospective studies evaluating the safety of ethanol as an antidote, therefore much of our understanding of ethanol toxicity is extrapolated from reports of acute alcohol intoxication or chronic substance abuse.42 Ethanol is known to produce central nervous system effects ranging from mild drowsiness or euphoria to coma with respiratory depression. Severe intoxication may result in hypotension, tachycardia, cardiac dysrhythmias or acute myocardial 7 1 Introduction infarction. Hypoglycemia may occur, with children and malnourished patients being at greatest r i sk ." ' 4 2 Ethanol intravenous solutions are hypertonic and may result in vein irritation. 1 1 ' 4 3 There is considerable individual variation in susceptibility to the toxic effects of ethanol. In the naive drinker, serum ethanol levels as low as 20 mg/dL (5 mmol/L) may result in euphoria and disinhibition, while serum levels of 100-200 mg/dL (22-44 mmol/L) may be associated with symptoms of marked intoxication including loss of coordination, confusion, aggressive behavior or stupor.8' " Serum concentrations > 200 mg/dL (> 44 mmol/L) may result in coma, loss of airway protection or hypotension in inexperienced drinkers.8 Persons who chronically abuse may tolerate much higher serum concentrations than ethanol-naive subjects.8 Despite the well documented toxicity of ethanol when used recreationally, the medical literature contains few reports of adverse events associated with the therapeutic use of ethanol as an antidote. Anecdotal reports have described respiratory arrest and aggressive behaviour attributed to ethanol during treatment of toxic alcohol poisoning. 5 1 ' 5 2 Two published retrospective case series have addressed ethanol safety issues. 5 3 ' 5 4 Roy et. al. performed a retrospective review of 60 children <18 years who received ethanol treatment for methanol poisoning at two Quebec pediatric hospitals.54 Six patients (10%) were described as "more drowsy" after receiving ethanol, one experienced gastritis, one experienced a drop in blood pressure and none developed symptoms of hypoglycemia. It should be noted that 75% of the patients in this series were five years of age or younger and most (63%) were eventually found to have clinically insignificant methanol exposures which did not require treatment. The low ethanol adverse event rate reported by Roy et. al. may not be representative of ethanol toxicity in adult patients, who are more likely to be seriously poisoned and receive more aggressive treatment. Lister et. al. performed a retrospective review of 27 adult patients with toxic alcohol poisoning treated with ethanol and hemodialysis in an Ottawa hospital.5 3 Although the authors concluded that ethanol was a safe and effective antidote, the only measure of antidote safety addressed in this case series was hypoglycemia. There were no episodes of hypoglycemia; however, in the 8 1 Introduction absence of data regarding other symptoms such as central nervous system effects, it is not possible to evaluate safety. 1.3.4.4 Available ethanol products In North America, there are no pharmaceutical ethanol products licensed for the use of as an antidote. Hospitals use a variety of sources of ethanol for antidote administration. Alcoholic beverages are usually used for oral doses. Intravenous solutions of 5-10% ethanol concentration are typically prepared from ampoules of a sterile, 100% (dehydrated) ethanol product which is intended for use as a sclerosing agent.55 One 10 mL ampoule cost $9.00 in 2006. A community hospital requires an inventory of 12-18 ampoules ($108-$ 162), while a secondary or tertiary care facility requires 30-60 ($270- $540). 5 5 A 24 hour ethanol treatment course for a dialyzed patient costs approximately $300 excluding laboratory monitoring costs.3 1.3.5 Fomepizole3 © B.C. Drug & Poison Information Centre 2005, adapted with permission 1.3.5.1 Fomepizole historical perspective and pharmacology Fomepizole is a potent inhibitor of A D H , with in vitro affinity for the enzyme 500-1000 times greater than that of ethanol and approximately 5000-10,000 times that of methanol or ethylene g lyco l . 3 3 ' 5 6 Experimental methanol poisoning in monkeys showed that intramuscular fomepizole effectively blocked formate accumulation at serum fomepizole concentrations >10 umol/L. 5 7 Human volunteer studies have evaluated safety and pharmacokinetics in non-poisoned subjects.58"60 The first cases of fomepizole use in human ethylene glycol poisoning were reported in the 1980s 6 I ' 6 2 , followed by two open label clinical trials involving a total of 30 patients with toxic alcohol poisoning. 2 4 ' 2 5 In addition to these studies, there are numerous case ~K1 "KQ ft"\ 11 reports describing use of fomepizole in adults and children. 1.3.5.2 Fomepizole dose and administration A variety of fomepizole dosage regimens have been used, all of which use fixed milligram per kilogram doses given as an intermittent intravenous infusion. Fomepizole serum concentrations are not monitored clinically and are not required for dose adjustment. The regimen recommended by the North American manufacturer (Antizol™ brand) is shown in Table 1.2. Some patients 9 1 Introduction have been treated with lower doses. 6 1 ' 6 2 , 7 1 ' 7 2 ' 7 6 ' 7 9 There are differences between the North American and European dosage regimens. For example, the North American product recommends increasing the maintenance dose to 15 mg/ kg after 48 hours of therapy based on limited information from a volunteer study. 6 0 In Europe, the maintenance dose is not automatically increased, and is often tapered at the end of therapy.33 DPIC has developed a modified dosing schedule based on the literature and use experience (Table 1.3).9 This regimen usually supplies a total fomepizole dose similar to that provided by the Antizol™ regimen, but is simplified to facilitate pre-approval of doses under the DPIC's fomepizole stock replacement program (see below). 4 1 Table 1.2 Manufacturer's dosage recommendations for fomepizole* (Antizol™) Dose in patients not requiring hemodialysis41 Loading dose 15 mg/kg Maintenance dose 10 mg/kg every 12 hours for the first 4 doses (i.e. 48 hr from the start of therapy) then 15 mg/kg every 12 hours until treatment endpoint Dose in patients requiring hemodialysis'" At the beginning of hemodialysis: if < 6 hr since last dose, do not administer dose if > 6 hr since last dose, administer next dose During hemodialysis: dose every 4 hr At the end of hemodialysis: Time between last dose and end of hemodialysis: < 1 hr do not administer dose 1-3 hr administer 50% of the next dose > 3 hr administer the next dose After hemodialysis resume dosing every 12 hours until endpoint * all doses given by IV intermittent infusion Table 1.3 B.C. Drug & Poison Information Centre dosage recommendations for fomepizole9 © BC Drug & Poison Information Centre 2005, adapted with permission Dose in patients not requiring hemodialysis* Loading dose 15 mg/kg, to a maximum of 1500 mg in patients weighing < 150 kg Maintenance dose 10 mg/kg every 12 hours Dose in patients requiring hemodialysis* Loading dose (dose 1) 15 mg/kg, to a maximum of 1500 mg in patients weighing <150 kg. Begin hemodialysis as soon as possible for patients with significant acidosis or within ~8 hours for clinically stable patients. During hemodialysis (dose 2) 10 mg/ kg given 4 hours after the start of hemodialysis. Continue dialysis until endpoint At the end of hemodialysis (dose 3) 10 mg/kg given immediately after the end of hemodialysis Additional doses are usually not required during or after subsequent hemodialysis sessions * all doses given by IV intermittent infusion. The DPIC guidelines provide the same dose as the Antizol™ product monograph in most instances. The DPIC regimen may provide a lower total dose if: the patient weighs more than 100 kg, has a delay of > 6 hours between the first dose and hemodialysis, receives dialysis for > 8 hours, or receives fomepizole therapy for > 48 hours. 10 1 Introduction 1.3.5.3 Fomepizole adverse effects Fomepizole is well tolerated in doses of 10 to 15 mg/kg. 7 8 The most commonly reported adverse effects include headache (14%), nausea (11%), dizziness (6%) drowsiness (6%) and unpleasant smell or taste (6%). Events infrequently associated with fomepizole therapy include tachycardia, bradycardia, seizures, rash, elevated transaminases, eosinophilia; however, the causal relationship is uncertain. 2 4 ' 2 5 ' 6 0 ' 7 8 In volunteer studies, single doses of 50-100 mg/kg resulted in symptoms of intoxication (slurred speech, dizziness, visual disturbances) and nausea in most subjects.58 1.3.5.4 Available fomepizole products Fomepizole is the only drug to receive US Food and Drug Administration (1999) and Health Canada (2000) approval for use as an antidote for toxic alcohol poisoning. 8 0 In Canada, the brand Antizol™ (Jazz Pharmaceuticals, formerly Orphan Medical) was marketed in 2001 and became available in B.C. hospitals in January 2002. Each 1500 mg vial of fomepizole costs $1000. A community hospital requires an inventory of one vial, while four vials are recommended for secondary and tertiary care hospitals with hemodialysis facilities.5 5 A 24 hour fomepizole treatment course for a dialyzed patient costs approximately $2000. 3 ' 4 1 B.C. DPIC has offered a fomepizole ^tock-replacement program for B .C . hospitals since the introduction of this antidote. Participating hospitals purchase initial inventory.41 If fomepizole is used in consultation with a Poison Control Centre toxicologist, DPIC wil l replace the antidote stock free of charge. As of December 2005, 48 patients had been treated with fomepizole. 1.3.6 Which antidote? The place of ethanol and fomepizole in therapy Since its introduction to the market, fomepizole has been promoted as.the preferred antidote for toxic alcohol poisoning.8 1 Antizol™ marketing materials and opinion leaders in clinical toxicology list the benefits of fomepizole over ethanol as "proven efficacy", fewer adverse effects, regulatory approval as an antidote, and a weight- based dosage regimen which does not c i n 11 S I 8*7 require titration based on serum drug levels. ' ' ' ' Some authors have also suggested that the acquisition cost of fomepizole could be offset by the cost savings that result from not having to monitor serum ethanol levels, being able to treat the patient on a medical ward rather than in 11 1 Introduction the intensive care unit (ICU), not requiring an infusion pump for a continuous intravenous infusion, having fewer antidote-related adverse events and, in some cases, obviating the need for hemodialysis. 3 3 ' 8 3" 8 5 Despite the above claims, there is limited evidence to conclusively prove the superiority of fomepizole over ethanol. There are no clinical trials or observational studies which directly compare the two antidotes. The one trial which enabled fomepizole to become licensed as a pharmaceutical product enrolled 30 patients (19 ethylene glycol and 11 methanol poisoning) in an un-blinded study with no comparison group. 2 4 ' 2 5 The limited safety data related to the use of ethanol as an antidote prohibits quantitative comparison of the adverse effect profiles of ethanol and fomepizole. The claims of potential cost savings associated with fomepizole, such as preventing ICU admission or preventing need for hemodialysis, remain unsubstantiated. The two economic analyses performed to date, as yet unpublished in the peer-reviewed literature, invoke assumptions that may not be representative of typical clinical practice (e.g. assuming serum ethanol levels are measured every two hours throughout antidote therapy). 8 4 ' 8 5 Acceptance of fomepizole has been slow. In B.C. , fewer than 20% of acute care hospitals stocked fomepizole three years after the introduction of this antidote. Physicians have demonstrated reluctance to prescribe fomepizole even when it is available. DPIC is aware of 17 cases of toxic alcohol poisoning between 2002-2004 in which the physician expressed a preference to either initiate or continue ethanol therapy despite the availability of fomepizole within the hospital.4 1 Twelve cases were eventually treated with fomepizole while five were treated only with ethanol. DPIC is not routinely consulted in all poisoning cases, thus these figures likely underestimate the true frequency of physician selection of ethanol antidote despite fomepizole availability. Stated reasons for prescribing ethanol instead of fomepizole included: lack of awareness that fomepizole was available in the hospital, prescriber preference to use a familiar therapy rather than a new antidote, relative ease of access to ethanol (e.g. "the fomepizole is locked in the pharmacy and the pharmacist wil l have to be called back to dispense it") and concern about fomepizole acquisition cost, coupled with the belief that fomepizole may not provide sufficient 12 1 Introduction advantage over ethanol to warrant the greater cost. The decision to choose ethanol over fomepizole was not necessarily related to the severity of illness. In eight of 17 cases, the patient was severely i l l , including one fatal case and four cases who developed renal failure. Physician preference for ethanol was also noted in some cases despite poor patient tolerance of ethanol or difficulty with ethanol administration. . This pattern of practice is not restricted to healthcare professionals within B.C. Mycyk et. al. described hospital compliance with Illinois Poison Center recommendations to initiate fomepizole therapy.86 Within a two year period (2001-2002), the poison center recommended fomepizole for 61 cases of suspected toxic alcohol poisoning. Only thirty nine (64%) of these cases eventually received fomepizole. If the poison center staff had suggested ethanol as a second line alternative to fomepizole, the physician was more likely prescribe ethanol compared to cases where only fomepizole was recommended (odds ratio 5.2, Cl95o/o 1.9, 14.1). Reasons given for not prescribing fomepizole were very similar to the B .C . experience, and included lack of familiarity, lack of pharmacist availability to dispense fomepizole, inability to locate fomepizole within the hospital and concern about the high cost. The authors of the two Canadian case series of ethanol antidote use both concluded that ethanol therapy is effective and/or safe. Lister et. al. express the opinion that "the higher acquisition cost of fomepizole is a significant consideration, especially in view of limited data to suggest its superiority over ethanol" and affirm that IV ethanol combined with hemodialysis remains "the preferred therapeutic approach" for toxic alcohol poisoning at their institution. 5 3 Roy et. al. conclude that "with appropriate monitoring in a controlled environment such as a pediatric intensive care unit, ethanol therapy does not carry as many risks as currently believed". 5 4 Health care professionals are divided on the role of fomepizole and ethanol in the management of toxic alcohol poisoning. In particular, there are different interpretations of the available data regarding antidote safety. It is imperative to obtain comprehensive, comparable safety data for both ethanol and fomepizole in order to make an informed decision regarding the place of these antidotes in therapy. 13 1 Introduction 1.4 Evaluating safety: The study of adverse drug events 1.4.1 Overview and definitions of drug safety terminology One means of evaluating the safety of a medication is to consider the frequency of harmful effects resulting from with medication use. Despite the importance of this field of study, there are no universally accepted definitions for the terms used to describe drug safety.87"91 A particular term may have several different meanings, while several different terms may be used to describe the same concept. The following is a review of the terminology used to describe medication safety, and an explanation of the terms used in this study. 1.4.1.1 Adverse drug reaction The classical definition of an adverse drug reaction was developed by the World Health Organization (WHO) in 1972: " A response to a drug which is noxious and unintended, and which occurs at doses normally used in man for the prophylaxis, diagnosis, or therapy of disease, or for the modification of physiological function."9 2 The utility of this definition is limited by the explicit exclusion of the harmful effects associated with unintentional or intentional overdose, including those resulting from medication error or lack of adherence to a prescribed drug regimen. Health Canada's Adverse Drug Reaction Monitoring Program uses a much broader definition for the purpose of eliciting reports of adverse reactions: "An adverse reaction (AR) is a harmful and unintended response to a health product. This includes any undesirable patient effect suspected to be associated with health product use. Unintended effect, health product abuse, overdose, interaction (including drug-drug, and drug-food interactions) and unusual lack of therapeutic efficacy are all considered to be reportable A R s . " 9 3 Some authors suggest that the term adverse drug reaction should be broadened to include any symptom attributed to a drug, regardless of dose, while others suggest that it should be narrowed to exclude trivial symptoms which are annoying but not clinically significant.91 1.4.1.2 Adverse drug effect The term adverse drug effect is usually synonymous with an adverse drug reaction. The distinction is that a "reaction" is from the point of view of the patient while an "effect" is from the point of view of the drug. 9 1 14 1 Introduction 1.4.1.3 Serious adverse drug reaction Serious adverse drug reactions (or effects) are defined by the W H O as any untoward medical occurrence that at any dose:9 4 • Requires inpatient hospitalization or prolongation of existing hospitalization • Results in persistent or significant disability/incapacity • Is life-threatening • Results in death or, as per Health Canada, requires a significant intervention to prevent one of the above outcomes.93 As a regulatory term, "serious" is not synonymous with "severe". Severe refers to the intensity of the symptom and could describe an event of minor medical significance, such as a severe headache.94 Serious refers to a clinical outcome. A l l serious events are severe, but the converse is not necessarily true. 1.4.1.4 Adverse event The term "adverse event" is a regulatory term, usually applied in the context of pharmacovigilance (drug safety monitoring programs). The WHO defines an adverse event as "Any untoward medical occurrence that may present during treatment with a pharmaceutical 92 product but which does not necessarily have a causal relationship with this treatment." A n adverse event is therefore distinguished from an adverse drug reaction in that there is no presumption of a causal connection between the drug and the clinical phenomenon. This term is often used in clinical trials of new drugs, when the safety profile of the medication is unknown. Although the terms "adverse event" and "adverse drug event" are sometimes used interchangeably, Aronson et. al. point out that the very inclusion of the word "drug" implies the suspicion of a causal link with a drug. 9 1 1.4.1.5 Adverse drug event In the patient safety literature, the term "adverse drug event" is typically used to provide a more comprehensive definition of harm resulting from any use of a drug. 8 7 This definition incorporates "adverse drug reactions" associated with normal use of a drug (WHO definition) and harm associated with medication errors, overdoses, dosage reductions or drug 15 1 Introduction discontinuation. Figure 1.2 depicts the relationship between adverse drug "events", "reactions" and medication errors under this interpretation. Figure 1.2 Relationship between adverse drug reactions and medication errors The area of each section may not reflect the proportion of incidents occurring in each category. 1.4.1.6 Medication error The National Coordinating Council-Medication Error Reporting and Prevention uses the following definition: 9 5 ' 9 6 " A medication error is any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the health care professional, patient, or consumer. Such events may be related to professional practice, health care products, procedures, and systems, including prescribing; order communication; product labeling, packaging, and nomenclature; compounding; dispensing; distribution; administration; education; monitoring; and use." Medication errors which result in harm to the patient are adverse drug events. Errors which do not cause harm, but have the potential to do so are potential adverse drug events. See Figure 1.2 Adverse Drug Reactions Adverse Drug Events Potential Adverse Drug Events 16 1 Introduction 1.4.1.7 Drug related problem: A n even broader term, "drug related problem", describes an "event or circumstance involving drug treatment that actually or potentially interferes with the patient's experiencing an optimum outcome of medical care".97 In addition to adverse drug reactions, drug related problems may include failure to prescribe a drug when it is required or use of a drug when it is not indicated, selection of the wrong drug, incorrect dose and drug interactions. 9 0 ' 9 7 These concepts extend beyond patient safety to the provision of optimal care. 1.4.1.7Adverse drug events: terminology used in this study The objectives of this study are most closely aligned with evaluation of patient safety rather than pharmacovigilance or provision of pharmaceutical care, therefore it is most appropriate to use the terminology of the patient safety literature. In this study the term adverse drug event (ADE) wil l be used to describe any harm to the patient associated with the use of the antidotes ethanol or fomepizole, whether the harm is a result of the inherent toxicity of the drug during normal clinical use (an adverse drug reaction), or the result of inappropriate administration (a medication error). 1.4.2 Causa l i t y assessment of adverse d r u g events The identification of an "adverse drug event" implies a suspected causal association between the drug and a clinical phenomenon. A variety of methods have been proposed to provide qualitative assessment of the probability that the drug is responsible for the event. At best, these tools provide a standardized approach to causality assessment that can decrease disagreement between assessors and provide a framework for consistent classification of likelihood. 9 8 However, none of these methods can determine with certainty whether or not an event was really due to the drug. Five criteria are common to all methods of evaluating the causal relationship between drug administration and an adverse clinical event; 8 7 ' 9 9 ' 1 0 0 • Temporal relationship: Did the onset or the clinical event occur after the initiation of drug therapy, within a plausible time frame? • Previous reports: Are there previously documented reports which link the drug and the clinical event? 17 1 Introduction • Competing causes: Are there other plausible causes of the clinical event, including other drugs or disease? • Dechallenge: Did the clinical event resolve with drug discontinuation (or dose reduction)? • Rechallenge: Did the clinical event recur with drug re-administration, after the drug had been stopped? 1.4.2.1 Expert opinion Although there is no gold standard for determining the true association between a drug and a clinical event, the consensus agreement of an expert panel is the traditional method and is the standard against which other methods are validated. 9 9 ' 1 0 1 The assessors use a process of "global introspection" to consider all the relevant factors in the case, weigh their relative importance and arrive at a conclusion regarding causality. 1 0 2 This form of clinical judgment has been likened to the degree of proof required in law to decide innocence or guilt, and is similar to the process of arriving at a diagnosis in clinical practice. 1 0 3 Expert opinion has been criticized as resulting in poor inter-rater agreement and lack of transparency in how the assessors arrive at their decision. 1 0 2 ' 1 0 4 ' 1 0 5 "Poor" agreement was defined as 50% unanimous agreement between three reviewers in one study and 40-60% agreement between three reviewers in another study. 1 0 4 ' 1 0 5 The W H O pharmacovigilance system relies on the global introspection method for causality assessment, but provides guidance notes to assessors in order to standardize reporting. 1 0 0 Six causality categories are defined, with increasingly stringent assessment criteria required as the level of "certainty" increases. For example, a "possible" causal link requires only a reasonable temporal relationship between drug intake and the event, while a "probable" rating also requires that the event was unlikely due to disease or other drugs and that there is documentation of symptom resolution with drug withdrawal. The WHO guidelines use no numerical scoring and assessors are given considerable latitude to weigh the available evidence. 1.4.2.2 Algorithms Numerous algorithms have been developed for the purpose of operationalizing causality assessment.9 9'1 0 3 Algorithms pose a series of questions which incorporate the five principles of 18 1 Introduction causality assessment. The best known tools are those designed for use in a general clinical practice setting, such as the 56 step Kramer algorithm or the 10 question Naranjo algorithm.1 0 6" 1 0 8 Some tools have been developed for use in specialty settings (e.g. in-house algorithms for pharmaceutical companies) or to evaluate specific types of drug-related events. Algorithms may assign a numerical score for each response, or guide the user through a decision table analysis. The final score is categorized as a qualitative probability rating ranging from high ("certain", "definite") to low ("doubtful", "unlikely") probability that the drug caused the symptom in question. 9 9 ' 1 0 3 Evaluation of algorithms focuses on reproducibility (do different observers reach the same conclusion using this method?) and validity (does this method result in the right answer?).1 0 1 Attempts to validate the results of algorithm-based causality assessment against expert opinion are often methodologically flawed. Some studies used the same group of reviewers to first assess cases by unguided expert opinion, then review the same cases using an algori thm. 1 0 8 ' 1 0 9 The high (>80%) agreement between the algorithm and expert opinion may have been influenced by recall despite the use of a "washout" period between evaluations, particularly as only 30-60 cases were used in these studies. Other studies have compared expert panel decision with algorithm results from a single reviewer.9 9' 1 0 6 In one study, results showed 71% agreement between the panel and algorithm 1 0 6, while another found that agreement ranged from 45-61%, depending on causality category (best agreement for "probable").99 These findings may not be representative of the algorithm performance in clinical practice when used by reviewers with diverse backgrounds. Due to the lack of a true gold standard for assessing validity, the performance of an algorithm is more commonly evaluated on reproducibility of results. Several early studies concluded that algorithms were superior to expert panel decision because they produced better inter-rater agreement. The Kramer algorithm resulted in inter-rater kappa scores of 0.57 compared to 0.26 for unguided expert opinion, while reviewers using the Naranjo algorithm had kappa scores of 0.69- 0.86 compared to 0.21-0.37 with unguided opin ion . 1 0 8 ' 1 0 9 Both evaluations used the same "crossover" design described in the above validity studies, always testing the algorithm second. Subsequent studies have shown variable performance of algorithms in various health care 19 1 Introduction settings. Kane-Gill et. al. tested the Naranjo algorithm for causality assessment of suspected adverse drug events in patients in the intensive care unit. Four independent reviewers evaluated 142 suspected adverse drug events in patients admitted to the intensive care unit. Inter-rater agreement was poor (kappa 0.14- 0.33) and the authors concluded that the "Naranjo criteria need modification for use in ICU to improve reliability, validity and clinical usefulness."110 Some studies have tested for agreement between results obtained with different causality algorithms. A simulation study showed that concordance between several algorithms was only slightly better than by chance alone. 1 1 1 Another found 64-67% agreement between three algorithms (weighted kappa 0.28-0.48). 1 1 2 Hutchinson et. al. pointed out that in order to avoid disagreement between reviewers, algorithm questions may be "posed with false precision" and focus on the most objective information, rather than the information which might be most important for a particular case. 1 0 1 Others have noted that algorithms are inflexible and do not incorporate the wide variety of information that may be available regarding the case. 1 1 3 The scoring methods are arbitrary and the weighting does not permit a particularly strong piece of evidence in one category to exert sufficient influence over the final score. It has been suggested that the most useful thing about algorithms is that they focus the attention of the reviewer on important details of adverse drug event assessment.111 1.4.2.3 Bayesian approach The "Bayesian Adverse Reactions Diagnostic Instrument" (BARDI) is the only causality assessment tool which provides a quantitative estimate of the probability that a drug was responsible for a suspected adverse drug event." 3" 1 1 5 B A R D I aims to answer the question: In this patient, who is treated with drug X for disease Y , what is the probability that a particular symptom (e.g. headache) is due to drug X ? " First, the prior odds of the adverse drug event are obtained from existing clinical trial and epidemiologic data: headache has been reported in 15% of the patients treated with drug X . Next, the prior odds are modified by additional information for all possible competing causes: the likelihood ratios for headache caused by disease Y , headache caused by co-administered drugs. The typical time of onset of the symptom onset is also considered: when is a drug X-related headache most likely to begin, compared to the typical onset time of disease Y-related headache? The output of the model is the posterior odds, 20 1 Introduction converted to the probability that this patient's headache was caused by drug X . The closer to 1.0 the posterior probability, the more likely that drug X is responsible for the headache. The advantage of B A R D I is that it can be customized for a particular drug, has flexible weightings able to respond to strong evidence, is transparent, reproducible and resistant to rater bias and is able to simultaneously evaluate multiple drugs. 1 1 3 - 1 1 5 The disadvantage is that B A R D I requires high quality data in order to accurately estimate the prior odds and likelihood, ratios required for the model. If all the necessary information is available, the drug-specific model is very labour intensive to develop. B A R D I is impractical in many clinical or research settings and is not feasible i f there is inadequate background information regarding the important variables. 1.4.2.4 Adverse drug event causality assessment in patients with toxic alcohol poisoning Investigating antidote-related adverse drug events in patients with toxic alcohol poisoning presents a methodological challenge because it may be difficult to observe the components considered essential for causality assessment. • Temporal relationship: May be confounded by inaccurate record keeping which misrepresents the temporal relationship between initiation of drug therapy and symptom onset. • Previous reports: When comparing an old drug with a new drug, there may be an imbalance in the availability of previously reported information, with more information available regarding the older drug. • Competing causes: It is anticipated that most study subjects wi l l have disease and/ or treatment related competing causes for suspected adverse drug events. • Dechallenge: Confirmation of symptom resolution with antidote discontinuation may be confounded by concurrent interventions (which could either account for symptom resolution or symptom continuation) or incomplete record keeping. • Rechallenge: There are limited opportunities to observe rechallenge in this population. Antidote therapy is typically short (24-36 hours) and is perceived to be essential, therefore it may be continued despite poor patient tolerance. Antidote discontinuation and re-introduction is uncommon. 21 1 Introduction In the case of the antidotes for toxic alcohol poisoning, the paucity of clinical trial and observational data make it impossible to use B A R D I . Existing algorithms designed for one-size-fits-all causality assessment (including the W H O guidance notes) are not suitable for the specialized circumstances surrounding this patient population and treatment. Developing and validating a specialized algorithm is beyond the scope of this study. Expert panel evaluation, despite the known limitations, has withstood the test of time and remains the reference standard for causality assessment. This study wil l use an expert panel of physicians with extensive training and experience in clinical toxicology, including experience with administration of both antidotes. Rather than using decision rules to guide evaluations, panelists wi l l be trained in the basic principles of causality assessment prior to case evaluation, and wil l rank cases on a qualitative probability scale. 22 1 Introduction 1.5 Thesis theme, objectives and hypothesis The theme of this thesis is antidote-related safety. The study objectives are to estimate the incidence, type and severity of adverse drug events attributed to the antidotes ethanol and fomepizole and determine whether the adverse drug event rate differs between the two antidotes. The hypothesis is that patients treated with fomepizole wil l have a lower antidote-related adverse event rate than those treated with ethanol. 23 1 Introduction 1.6 References 1. Pike I, Han G, Kinney J, et al. The economic cost of poisoning in British Columbia, Canada. Paper presented at: North American Congress of Clinical Toxicology, 2006; San Francisco. 2. H D W . Injury hospitalizations 2000/2001. Health Data Warehouse BC Ministry of Health Planning, BC Ministry of Health Services. Available at: https://webapps.hnet.bc.ca/hdw/db/hdw/index.htm. Accessed February 26, 2007. 3. Lepik K J , Daws DE, Purssell R A . Review of methanol and ethylene glycol poisoning in British Columbia 1996-2001. Vancouver: B C Drug & Poison Information Centre; September 1 2005. 4. Lepik K J , Levy A R , Daws D E , et al. Duration of hospital stay for treatment of methanol and ethylene glycol poisoning in British Columbia (abstract). Canadian Journal of Clinical Pharmacology. 2005;12(l):e74. 5. Ford M D , McMartin K . Ethylene glycol and methanol. In: Ford M D , Delaney K A , Ling L J , et al., eds. Clinical Toxicology. Philadelphia: W.B.Saunders Co; 2001:757-767. 6. Jolliff H A , Sivilotti M L . Ethylene glycol. In: Dart R C , ed. Medical Toxicology. 3 ed. Philadelphia: Lippincott Williams & Wilkins; 2004:1223-1230. 7. Sharma A N . Toxic alcohols. In: Goldfrank L R , Flomenbaum N E , Lewin N A , et al., eds. Goldfrank's Toxicologic Emergencies. 7 ed. New York: McGraw-Hill ; 2002:980-990. 8. Sivilotti M L . Ethanol, Isopropanol and Methanol. In: Dart R C , ed. Medical Toxicology. 3 ed. Philadelphia: Lippincott Williams & Wilkins; 2004:1211-1230. 9. DPIC. Treatment guidelines for methanol and ethylene glycol poisoning. Vancouver, B C : B C Drug & Poison Information Centre; Prepared by: K J Lepik; Reviewed by: JR Brubacher, C DeWitt, G Erhardt, M Friesen, R Gair, L Hayes, J Kennedy, D A Kent, D Leong, R L i , R A Purssell; September 19 2005. . 10. Barceloux D G , Bond GR, Krenzelok EP, et al. American Academy of Clinical Toxicology practice guidelines on the treatment of methanol poisoning. Journal of Toxicology - Clinical Toxicology. 2002;40(4):415-446. 11. Barceloux D G , Krenzelok EP, Olson K , et al. American Academy of Clinical Toxicology Practice Guidelines on the Treatment of Ethylene Glycol Poisoning. Journal of Toxicology - Clinical Toxicology. 1999;37(5):537-560. 12. Jacobsen D. Ethylene glycol and other glycols. In: Brent J, Wallace K L , Burkhart K K , et al., eds. Critical Care Toxicology: diagnosis and management of the critically poisoned patient. Philadelphia: Elsevier Mosby; 2005. 13. Jacobsen D, McMartin K . Methanol and formaldehyde poisoning. In: Brent J, Wallace K L , Burkhart K K , et al., eds. Critical Care Toxicology: diagnosis and management of the critically poisoned patient. Philadelphia: Elsevier Mosby; 2005. 14. Mahieu P, Hassoun A , Lauwerys R. Predictors of methanol intoxication with unfavourable outcome. Hum Toxicol. Mar 1989;8(2): 135-137. 15. Osterloh JD, Pond S M , Grady S, et al. Serum formate concentrations in methanol intoxication as a criterion for hemodialysis. Ann Intern Med. Feb 1986;104(2):200-203. 16. Anderson TJ, Shuaib A , Becker WJ. Methanol poisoning: Factors associated with neurologic complications. Can J Neurol Sci. 1989;16(4):432-435. 24 1 Introduction 17. L iu JJ, Daya M R , Carrasquillo O, et al. Prognostic factors in patients with methanol poisoning. Journal of Toxicology - Clinical Toxicology. 1998;36(3):175-181. 18. Hovda K E , Hunderi O H , Tafjord A B , et al. Methanol outbreak in Norway 2002-2005: epidemiology, clinical features and prognostic signs. Journal of Internal Medicine. 2005;258:181-190. 19. Porter W H , Rutter PW, Bush B A , et al. Ethylene glycol toxicity: the role of serum glycolic acid in hemodialysis. Journal of Toxicology - Clinical Toxicology. 2001;39(6):607-615. 20. Karlson-Stiber C, Persson H . Ethylene glycol poisoning: experiences from an epidemic in Sweden. Journal of Toxicology - Clinical Toxicology. 1992;30(4):565-574. 21. Jacobsen D, Ovrebo S, Ostborg J, et al. Glycolate causes the acidosis in ethylene glycol poisoning and is effectively removed by hemodialysis. Acta Medica Scandinavica. 1984;216(4):409-416. 22. DPIC. Survey of hospital laboratory resources in British Columbia. Vancouver, B C : B.C. Drug & Poison Information Centre; (update 2002) 2001. 23. Jacobsen D, Webb R, Collins TD, et al. Methanol and formate kinetics in late diagnosed methanol intoxication. Medical Toxicology & Adverse Drug Experience. Sep-Oct 1988;3(5):418-423. 24. Brent J, McMartin K , Phillips S, et al. Fomepizole for the treatment of methanol poisoning. N Engl J Med. Feb 8 2001;344(6):424-429. 25. Brent J, McMartin K , Phillips S, et al. Fomepizole for the treatment of ethylene glycol poisoning. NEngl J Med. Mar 18 1999;340(ll):832-838. 26. Sivilotti M L , Burns M J , McMartin K E , et al. Toxicokinetics of ethylene glycol during fomepizole therapy: implications for management. For the Methylpyrazole for Toxic Alcohols Study Group.[see comment]. Ann Emerg Med. Aug 2000;36(2): 114-125. 27. Jacobsen D, Hewlett TP, Webb R, et al. Ethylene glycol intoxication: evaluation of kinetics and crystalluria. American Journal of Medicine. Jan 1988;84(1): 145-152. 28. Cheng JT, Beysolow TD, Kaul B, et al. Clearance of ethylene glycol by kidneys and hemodialysis. Journal of Toxicology - Clinical Toxicology. 1987;25(l-2):95-108. 29. Palatnick W, Redman L W , Sitar DS, et al. Methanol half-life during ethanol administration: implications for management of methanol poisoning. Ann Emerg Med. Aug 1995;26(2):202-207. 30. Yip L , Jacobsen D. Endogenous formate elimination and total body clearance during hemodialysis. Journal of Toxicology - Clinical Toxicology. 2003;41(3):257-258. 31. Moreau C L , Kerns W, 2nd, Tomaszewski C A , et al. Glycolate kinetics and hemodialysis clearance in ethylene glycol poisoning. M E T A Study Group. Journal of Toxicology -Clinical Toxicology. 1998;36(7):659-666. 32. Burns A B , Bailie GR, Eisele G, et al. Use of pharmacokinetics to determine the duration of dialysis in management of methanol poisoning. American Journal of Emergency Medicine. Sep 1998;16(5):538-540. 33. Megarbane B , Borron SW, Baud F. Current recommendations for treatment of severe toxic alcohol poisonings. Intensive Care Medicine. 2005;31:189-195. 34. Kerns W, 2nd, Tomaszewski C, McMartin K , et al. Formate kinetics in methanol poisoning, [see comment]. Journal of Toxicology - Clinical Toxicology. 2002;40(2):137-143. 25 1 Introduction 35. Hantson P, Haufroid V , Wallemacq P. Formate kinetics in methanol poisoning. Hum Exp Toxicol. 2005;24:55-59. 36. Kostic M A , Dart RC. Rethinking the toxic methanol level. Journal of Toxicology -Clinical Toxicology. 2003;41(6):793-800. 37. Caravati E M , Heileson H L , Jones M . Treatment of severe pediatric ethylene glycol intoxication without hemodialysis. Journal of Toxicology - Clinical Toxicology. 2004;42(3):255-259. 38. Boyer EW, Mejia M , Woolf A , et al. Severe ethylene glycol ingestion treated without hemodialysis. Pediatrics. Jan 2001;107(1):172-173. 39. Borron SW, Megarbane B , Baud FJ. Fomepizole in treatment of uncomplicated ethylene glycol poisoning.[see comment]. Lancet. Sep 4 1999;354(9181):831. 40. Olson Ke. Poisoning & Drug Overdose. 4th ed. New York: Lange Medical Books: McGraw-Hil l ; 2004. 41. Lepik K J , Daws DE. BC Drug and Poison Information Centre fomepizole program report 2002-2004. Vancouver: B C Drug & Poison Information Centre; September 1 2005. 42. Seifert S. Ethanol. In: Dart RC, ed. Medical Toxicology. 3 ed. Philadelphia: Lippincott Williams & Wilkins; 2004:196-199. 43. Howland M A . Antidotes in Depth: Ethanol. In: Goldfrank LR, Flomenbaum N E , Lewin N A , et al., eds. Goldfrank's Toxicologic Emergencies. 7 ed. New York: McGraw-Hil l ; 2002:995-998. 44. Roe O. Clinical investigations of methyl alcohol poisoning with special reference to the pathogenesis and treatment of amblyopia. Acta Medica Scandinavica. 1943;113(6):558-608. 45. Roe O. Methanol poisoning: its clinical course, pathogenesis and treatment. Acta Medica Scandinavica. 1946;Suppl 182:1-253. 46. Jacobsen D, McMartin K . Antidotes for methanol and ethylene glycol poisoning. Journal of Toxicology - Clinical Toxicology. 1997;35(2): 127-143. 47. Kent D, Willis G, Lepik K , eds. Poison Management Manual. 4th ed. Vancouver: B C Drug & Poison Information Centre; 1997. 48. Criminal Code, Canada 253 (b). Motor Vehicles, Vessels and Aircraft, Operation while Impaired. Canadian Legal Information Institute. Available at: http://www.canlii.org/ca/sta/c-46/sec253.html. Accessed September 22, 2005. 49. Ethylene glycol: Poisindex Management. Thomson MICROMEDEX. 50. Cobaugh D. Ethanol. In: Brent J, Wallace K L , Burkhart K K , et al., eds. Critical Care Toxicology: diagnosis and management of the critically poisoned patient. Philadelphia: Elsevier Mosby; 2005. 51. Kowalczyk M , Halvorsen S, Ovrebo S, et al. Ethanol treatment in ethylene glycol poisoned patients. Veterinary & Human Toxicology. Aug 1998;40(4):225-228. 52. Meyer RJ, Beard M E , Ardagh M W , et al. Methanol poisoning. New Zealand Medical Journal. Jan 28 2000; 113( 1102): 11 -13. 53. Lister D, Tierney M , Dickinson G. Effectiveness of IV ethanol therapy combined with hemodialysis in the treatment of methanol and ethylene glycol poisoning. Canadian Journal of Hospital Pharmacy. 2005;58(3): 142-147. 26 1 Introduction 54. Roy M , Bailey B , Chalut D, et al. What are the adverse effects of ethanol used as an antidote in the treatment of suspected methanol poisoning in children? Journal of Toxicology - Clinical Toxicology. 2003;41 (2): 155-161. 55. DPIC. Antidote stocking guidelines for B. C. hospitals. Vancouver: B C Drug & Poison Information Centre; Prepared by: K J Lepik. Reviewed by: JR Brubacher, D Daws, C DeWitt, G Erhardt, M Friesen, R Gair, L Hayes, J Kennedy, D A Kent, D Leong, R L i , R A Purssell, M O Weins.; June 12 2006. 56. McMartin K . Fomepizole. In: Brent J, Wallace K L , Burkhart K K , et al., eds. Critical Care Toxicology: diagnosis and management of the critically poisoned patient. Philadelphia: Elsevier Mosby; 2005. 57. McMartin K E , Hedstrom K G , Tolf BR, et al. Studies on the metabolic interactions between 4-methylpyrazole and methanol using the monkey as an animal model. Arch Biochem Biophys. Feb 1980;199(2):606-614. 58. Jacobsen D, Sebastian CS, Blomstrand R, et al. 4-Methylpyrazole: a controlled study of safety in healthy human subjects after single, ascending doses. Alcohol Clin Exp Res. Aug 1988;12(4):516-522. 59. Jacobsen D, Barron SK, Sebastian CS, et al. Non-linear kinetics of 4-methylpyrazole in healthy human subjects. Eur J Clin Pharmacol. 1989;37(6):599-604. 60. Jacobsen D, Sebastian CS, Barron SK, et al. Effects of 4-methylpyrazole, methanol/ethylene glycol antidote, in healthy humans. Journal of Emergency Medicine. Jul-Aug 1990;8(4):455-461. 61. Baud FJ, Bismuth C, Gamier R, et al. 4-Methylpyrazole may be an alternative to ethanol therapy for ethylene glycol intoxication in man. Journal of Toxicology - Clinical Toxicology. 1986;24(6):463-483. 62. Baud FJ, Galliot M , Astier A , et al. Treatment of ethylene glycol poisoning with intravenous 4-methylpyrazole. N Engl J Med. Jul 14 1988;319(2):97-100. 63. Lushine K A , Harris CR, Holger JS. Methanol ingestion: prevention of toxic sequelae after massive ingestion. Journal of Emergency Medicine. May 2003;24(4):433-436. 64. Velez LI , Kulstad E, Shepherd G, et al. Inhalational methanol toxicity in pregnancy treated twice with fomepizole. Veterinary & Human Toxicology. Feb 2003;45(l):28-30. 65. Essama Mbia JJ, Guerit J M , Haufroid V , et al. Fomepizole therapy for reversal of visual impairment after methanol poisoning: a case documented by visual evoked potentials investigation. American Journal of Ophthalmology. Dec 2002;134(6):914-916. 66. McMurray M , Carty D, Toffelmire E B . Predicting methanol clearance during hemodialysis when direct measurement is not available. CAANT Journal - Canadian Association of Nephrology Nurses & Technicians. Jan-Mar 2002; 12(l):29-38; quiz 32-23. 67. Sivilotti M L , Burns M J , Aaron C K , et al. Reversal of severe methanol-induced visual impairment: no evidence of retinal toxicity due to fomepizole. Journal of Toxicology -Clinical Toxicology. 2001;39(6):627-631. 68. Brown M J , Shannon M W , Woolf A , et al. Childhood methanol ingestion treated with fomepizole and hemodialysis. Pediatrics. Oct 2001;108(4):E77. 69. Hantson P, Wallemacq P, Brau M , et al. Two cases of acute methanol poisoning partially treated by oral 4-methylpyrazole. Intensive Care Medicine. May 1999;25(5):528-531. 70. Burns M J , Graudins A , Aaron C K , et al. Treatment of methanol poisoning with intravenous 4-methylpyrazole. Ann Emerg Med. Dec 1997;30(6):829-832. 27 1 Introduction 71. Megarbane B , Borron SW, Trout H , et al. Treatment of acute methanol poisoning with fomepizole. Intensive Care Medicine. Aug 2001;27(8): 1370-1378. 72. Bekka R, Borron SW, Astier A , et al. Treatment of methanol and isopropanol poisoning with intravenous fomepizole. Journal of Toxicology Clinical Toxicology. 2001;39(1):59-67. 73. Benitez JG, Swanson-Biearman B , Krenzelok EP. Nystagmus secondary to fomepizole administration in a pediatric patient. Journal of Toxicology - Clinical Toxicology. 2000;38(7):795-798. 74. Baum CR, Langman C B , Oker E E , et al. Fomepizole treatment of ethylene glycol poisoning in an infant. Pediatrics. Dec 2000;106(6):1489-1491. 75. Harry P, Jobard E, Briand M , et al. Ethylene glycol poisoning in a child treated with 4-methylpyrazole. Pediatrics. Sep 1998;102(3):E31. 76. Hantson P, Hassoun A , Mahieu P. Ethylene glycol poisoning treated by intravenous 4-methylpyrazole. Intensive Care Medicine. Jul 1998;24(7):736-739. 77. McMartin K E , Heath A . Treatment of ethylene glycol poisoning with intravenous 4-methylvyV?LZO\Q. NEngl J Med. Jan 12 1989;320(2):125. 78. Antizol R (fomepizole) injection product monograph. Jazz Pharmaceuticals. Available at: http://antizol.info/monograph.htm. Accessed March 2, 2007. 79. Harry P, Turcant A , Bouachour G, et al. Efficacy of 4-methylpyrazole in ethylene glycol poisoning: clinical and toxicokinetic aspects. Hum Exp Toxicol. Jan 1994; 13(1):61 -64. 80. Notices of compliance: prescription pharmaceuticals for human use. January 1 -December 31, 2000. Health Canada: Therapeutic Products Directorate. Available at: http://www.hc-sc.gc.ca/hpfb-dgpsa/tpd-dpt/noc/2000/pre2000et_e.txt. Accessed February 12, 2005. 81. Antizol R (fomepizole) injection- why stock? Jazz Pharmaceuticals. Available at: http://antizol.info/whystock.htm. Accessed March 2, 2007. 82. Casavant M J . Fomepizole in the treatment of poisoning. Pediatrics. Jan 2001;107(1):170. 83. Brent J. Antidotes and alcohols: has fomepizole made ethanol an obsolete therapy? American College of Medical Toxicology. Available at: http://www.ijmt.net/ijmt/l_l/l_l_2.html. Accessed January 29th, 2007. 84. Marraffa J, Stork C, Medicis J. Cost-effectiveness of fomepizole versus ethanol in the management of acute ethylene glycol exposure. (Abstract 128). Clin Toxicol. 2005;43(6):691. 85. Shiew C, Dargan P, Greene S, et al. A n economic analysis: is fomepizole really more expensive than ethanol for treatment of ethylene glycol poisoning? (Abstract 127). Clin Toxicol. 2005;43(6):690. 86. Mycyk M , Des Lauriers C, Metz J, et al. Compliance with poison center fomepizole recommendations is suboptimal in cases of toxic alcohol poisoning. American Journal of Therapeutics. 2006;13:485-489. 87. Nebeker J, Barach P, Samore M . Clarifying adverse drug events: a clinician's guide to terminology, documentation and reporting. Ann Intern Med. 2004;140:795-801. 88. Morimoto T, Gandhi T, Seger A , et al. Adverse drug events and medication errors: detection and classification methods. Quality and Safety in Health Care. 2004; 13:306-314. 89. Gharaibeh M N , Greenberg HE, Waldman SA. Adverse drug reactions: a review. Drug Inf J. 1998;32:323-338. 28 1 Introduction 90. van M i l JWF, Westerlund T, Hersberger K E , et al. Drug-related problem classification systems. Annals of Pharmacotherapy. 2004;38:859-867. 91. Aronson J, Ferner R. Clarification of terminology in drug safety. Drug Saf. 2005;28(10):851-870. 92. W H O - U M C . Side effect- adverse reaction. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Available at: http://www.who-umc.org/DynPage.aspx?id=22676. Accessed March 3, 2007. 93. C A D R M P . Canadian adverse drug reaction monitoring program (CADRMP) guidelines for the voluntary reporting of suspected adverse reactions to health products by health professionals and consumers. Health Canada: Marketed Health Products Directorate. 2006-05. Available at: http://www.hc-sc.gc.ca/dhp-mps/alt_formats/hpfb-dgpsa/pdf/medeff/ar-ei_guide-ldir_e.pdf. Accessed March 3, 2007. 94. W H O U M C . Definitions: Serious adverse reaction. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Apri l 13, 2006. Available at: http://www.who-umc.org/DynPage.aspx?id=22680. Accessed Jan 11, 2007. 95. N C C _ M E R P . Medication error definition. National Coordinating Council for Medication Error Reporting and Prevention. Available at: http://www.nccmerp.org/aboutMedErrors.html. Accessed May 9, 2005. 96. N C C M E R P . N C C M E R P taxonomy of medication errors. National Coordinating Council for Medication Error Reporting and Prevention. Available at: http://www.nccmerp.org/pdf/taxo2001-07-31 .pdf. 97. Hepler C, Strand L M . Opportunities and responsibilities in pharmaceutical care. Am J HospPharm. 1990;47:533-543. 98. W H O - U M C . The use of the W H O - U M C system for standardised case causality assessment. Available at: http://www.who-umc.org/pdfs/Causality.pdf. Accessed Sept 24, 2005. 99. Macedo A F , Marques FB, Ribeiro CF, et al. Causality assessment of adverse drug reactions: comparison of the results obtained from published decisional algorithms and from the evaluations of an expert panel, according to different levels of imputability. Journal of Clinical Pharmacy and Therapeutics. Apr 2003;28(2): 137-143. 100. W H O - U M C . Causality Assessment of Suspected Adverse Reactions. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Available at: http://www.who-umc.org/DynPage.aspx?id=22682. Accessed Feb 16, 2007. 101. Hutchinson T A , Lane D A . Assessing Methods for Causality Assessment of Suspected Adverse Drug-Reactions. J Clin Epidemiol. 1989;42(1):5-16. 102. Kramer M S . Assessing causality of adverse drug reactions: global introspection and its limitations. DruglnfJ. 1986;20:433-437. 103. Stephens M D B . The diagnosis of adverse medical events associated with drug tretment. Adverse Drug Reactions and Acute Poisoning Reviews. 1987;1:1-35. 104. Karch FE, Smith C L , Kerzner B , et al. Adverse Drug-Reactions - Matter of Opinion. Clin Pharmacol Ther. 1976;19(5):489-492. 105. Blanc S, Leuenberger P, Berger JP, et al. Judgments of Trained Observers on Adverse Drug-Reactions. Clin Pharmacol Ther. 1979;25(5):493-498. 29 1 Introduction 106. Karen FE, Lasagna L . Toward Operational Identification of Adverse Drug-Reactions. Clin Pharmacol Ther. 1977;21(3):247-254. 107. Kramer M S , Leventhal J M , Hutchinson T A , et al. A n algorithm for the operational assessment of adverse drug reactions. I. Background, description, and instructions for use. Jama. Aug 17 1979;242(7):623-632. 108. Naranjo C A , Busto U , Sellers E M , et al. A Method for Estimating the Probability of Adverse Drug-Reactions. Clin Pharmacol Ther. 1981;30(2):239-245. 109. Hutchinson T A , Leventhal J M , Kramer M S , et al. A n algorithm for the operational assessment of adverse drug reactions. II. Demonstration of reproducibility and validity. Jama. Aug 17 1979;242(7):633-638. 110. Kane-Gill SL, Kirisci L , Pathak DS. Are the Naranjo criteria reliable and valid for determination of adverse drug reactions in the intensive care unit? Annals of Pharmacotherapy. Nov 2005;39(11): 1823-1827. 111. Pere JC, Begaud B, Haramburu F, et al. Computerized Comparison of 6 Adverse Drug Reaction Assessment Procedures. Clin Pharmacol Ther. Oct 1986;40(4):451-461. 112. Michel DJ, Knodel L C . Comparison of three algorithms used to evaluate adverse drug reactions. Am JHosp Pharm. 1986;43:1709-1714. 113. Lane D A , Kramer M S , Hutchinson T A , et al. The causality assessment of adverse drug reactions using a Bayesian approach. Pharmaceutical Medicine. 1987;2(3):265-283. 114. Hutchinson T A . Bayesian assessment of adverse drug reactions, [comment]. Cmaj. Nov 28 2000;163(11):1463-1464. 115. Lanctot K L , Naranjo C A . Comparison of the Bayesian approach and a simple algorithm for assessment of adverse drug events. Clin Pharmacol Ther. Dec 1995;58(6):692-698. 30 2 Antidote ADEs 2 ADVERSE DRUG EVENTS ASSOCIATED WITH T H E ANTIDOTES FOR TOXIC A L C O H O L POISONING: A COMPARISON OF ETHANOL AND FOMEPIZOLE 1 2.1 Introduction 2.1.1 Background Methanol and ethylene glycol are found in automotive antifreeze and some solvent mixtures. When ingested, these toxic alcohols are converted to metabolites which can cause metabolic acidosis, renal failure (ethylene glycol), blindness (methanol) and death. The American Association of Poison Control Centers' national poisoning and exposure database recorded 1195 methanol and 2298 ethylene glycol exposures treated in U.S. health care facilities in 2005.1 Treatment of toxic alcohol poisoning includes supportive care and administration of either ethanol or fomepizole as an antidote to block production of toxic metabolites. Ethanol has been used as an antidote for methanol poisoning since the 1940s. 2 ' 3 Although there are no clinical trials investigating its use for toxic alcohol poisoning, the weight of evidence suggests that it is effective when administered orally or as a continuous intravenous infusion titrated to an ethanol serum concentration of 100 to 150 mg/dL (22 to 32 mmol/L). 2" 8 The acute toxicity of ethanol as a substance of abuse is well documented and includes central nervous system depression, cardiovascular effects and hypoglycemia; however, there is limited information regarding ethanol toxicity when it is used as an antidote.4'9' 1 0 Anecdotal reports describe respiratory arrest and aggressive behaviour attributed to ethanol."' 1 2 Roy et. al. found minimal evidence of ethanol antidote toxicity in 60 children treated for methanol poisoning. 1 3 Lister et. al. found no evidence of hypoglycemia in 27 adults treated with ethanol and hemodialysis, but did not evaluate other adverse reactions.14 Fomepizole was first marketed in North America in 1998 based on evidence from volunteer studies, case reports and an unblinded, single arm clinical trial demonstrating its effectiveness as an antidote.15"18. Fomepizole is given as a fixed milligram per kilogram dose administered by 1 A version of this manuscript will be submitted to Annals of Emergency Medicine. For the purpose of this thesis, the candidate is the sole author. Co-authors will contribute only after thesis submission. 31 2 Antidote ADEs intermittent intravenous infusion and does not require serum level monitoring. 1 9 The most commonly reported adverse reactions include headache, nausea, dizziness, increased drowsiness and unpleasant taste, with rare reports of seizure, cardiovascular effects and elevation of hepatic transaminases.19 This safety profile is a compilation of events reported in a variety of controlled settings (volunteer studies, clinical trial), and may not be representative of the adverse events occurring in clinical practice. 2.1.2 Importance Fomepizole is promoted as the preferred antidote for toxic alcohol poisoning on the grounds that is easier to administer, has fewer adverse effects than ethanol and is the only marketed antidote approved by US and Canadian regulatory agencies for treatment of toxic alcohol poisoning. 7 ' 1 0 ' 2 0 ' 2 1 American Association of Poison Control Center statistics indicate that while fomepizole use increased and ethanol use declined between 2000 and 2005, both antidotes were still prescribed. 1 ' 2 2 The major deterrent to the widespread acceptance of fomepizole is its relatively high acquisition cost of approximately $1000-$4000 per treatment course.15' 2 3 Recent publications suggest that ethanol may be preferred for use in combination with hemodialysis, or for treatment of "borderline" ethylene glycol poisoning when a short treatment course is anticipated.14' 2 4 Clinicians wil l often initially prescribe ethanol for suspected toxic alcohol poisoning and switch to fomepizole after the diagnosis has been confirmed, or when a supply of fomepizole becomes available. 2 5 ' 2 6 To date, there are no clinical trials or observational studies which compare the safety and effectiveness of ethanol and fomepizole in poisoned patients. The purpose of this study is to address unanswered questions regarding the comparative safety of ethanol and fomepizole by investigating the occurrence of adverse drug events (ADEs) associated with the clinical use of these antidotes for toxic alcohol poisoning. In this study, the term "adverse drug event" includes both adverse drug reactions, which are a result of the inherent toxicity of a drug during normal clinical use and harm resulting from incorrect usage (medication 32 2 Antidote ADEs 2.1.3 Goals of this investigation The objectives of this study are to: 1) Estimate the proportion of incidents with suspected toxic alcohol poisoning which experience ADEs attributed to ethanol or fomepizole, and characterize the type and severity of these ADEs. 2) Compare the adverse event rates in ethanol and fomepizole treated patients, using the primary outcome of any ADE and the adjusted rate ratio (RR) as the measure of effect size. The secondary outcome is a comparison of the event rates of any severe ADE. 33 2 Antidote ADEs 2.2 Methods 2.2.1 Study design This was a retrospective cohort study in which the exposure was the antidote treatment and the outcome was the occurrence of an antidote-related A D E . The target population included all patients who received antidote for treatment of suspected toxic alcohol poisoning; however, the available population was limited to patients who could be identified through a search of hospital administrative databases. The study sample was drawn from a convenience sample of hospitals identified through British Columbia (B.C.) Drug and Poison Information Centre records as having used the antidotes fomepizole and ethanol during the study period. The centre has provided approval for the use of fomepizole through an antidote inventory replacement program since 2002, and therefore maintains records of fomepizole use throughout the province. 2.2.2 Setting The sample was drawn from hospital admissions between January 1 s t 1996 and December 31 s t 2005. This study period spanned six years before and four years after the availability of fomepizole in B .C . In order to represent diverse practice settings, ten study hospitals were selected in nine different cities throughout the province. Two hospitals were tertiary care teaching facilities, four were secondary, care regional facilities and four were urban community hospitals. A l l hospitals had emergency departments (EDs) and intensive care units (ICUs). The tertiary and secondary care facilities had hemodialysis capability, while the community hospitals did not. Ethical approval was obtained from the five regional research and ethics boards governing the study hospitals. 2.2.3 Selection of subjects Potential study subjects were identified by searching each hospital's electronic health records database for admissions with the following International Classification of Disease (ICD) codes in any diagnostic field: methanol poisoning ICD-9 980.1, ICD-10 T51.1, ethylene glycol poisoning: ICD-9 982.8, ICD-10 T52.8 and T51.8. The ICD codes for ethylene glycol are non-specific and also capture admissions for poisoning by other organic solvents and alcohols. 34 2 Antidote ADEs A poisoning "incident" was included i f the patient was >13 years of age at the time of hospital admission and received at least one dose of antidote (ethanol or fomepizole) for treatment of suspected methanol or ethylene glycol poisoning. Admissions were excluded i f there was no antidote treatment or no suspicion of toxic alcohol poisoning. The records of patients transferred between hospitals were matched such that this occurrence was counted as one incident. A patient with repeated toxic alcohol exposures could contribute more than one incident during the study period. 2.2.4 Data collection and processing Information was collected from medical charts by two abstracters: a pharmacist with A A P C C Poison Information Specialist certification (KL) and a physician research assistant (JB). Abstractors used standardized paper data collection forms to record patient characteristics, toxin exposure history, symptoms, vital signs, Glasgow Coma Scale (GCS), laboratory test results and treatments (antidotes, other medications, procedures). Incidents were assigned a study identification number; unique patient identifiers were not recorded. Detailed information was recorded for all symptoms and treatments from the time of hospital admission to completion of antidote therapy. A synopsis of the clinical course was recorded from completion of antidote until recovery or death. A l l mandatory variables had a dedicated field for data entry, with operational definitions printed on the forms. Narrative information describing patient symptoms was abstracted from nursing and physician notes verbatim. If the chart contained conflicting accounts of an event, all versions were recorded. A l l charts were reviewed by two abstracters. During an initial training period, both abstractors independently reviewed charts using separate data collection forms and refined coding rules. Once both abstracters were using a consistent approach, one would act as the primary reviewer and the other performed a second review to verify accuracy and completeness. Abstracters conferred to reach agreement on ambiguous charting or conflicting interpretation. Although the abstracters were aware of the study objectives, the data collection process did not permit them to make a judgment of symptom causality, limiting potential bias from this source. 35 2 Antidote ADEs 2.2.5 Methods of Measurement The abstracted data were entered into a Microsoft Access™ database by one investigator (KL) . Data verification was done using range tests for extreme values. A n Acute Physiologic and Chronic Health Evaluation II ( A P A C H E II) score was calculated for each incident to provide a measure of baseline severity of illness. A P A C H E II is a composite score of 14 variables, and is a validated scale for stratification of I C U patients by severity of i l lness . 3 1 , 3 2 The pre-antidote baseline was defined as the most abnormal value of each variable measured between hospital admission and the start of antidote, to a maximum of 24 hours after ICU admission. The severity of each symptom was classified as minor, moderate or severe, according to the Poisoning Severity Score. 3 3 This standardized, validated scale for poisoning morbidity was developed jointly by the International Programme On Chemical Safety and the European Association of Poisons Centres and Clinical Toxicologists. A total Poisoning Severity Score was assigned for case outcome based on the most severe individual symptom at any time between hospital admission and recovery or death.33 A symptom was identified for review as a potential A D E i f it was first documented during antidote treatment, i f a pre-existing symptom worsened by at least one severity category during antidote treatment, or i f the hospital caregiver or patient attributed the symptom to the antidote. Selected laboratory test abnormalities with onset during antidote therapy were also identified as potential ADEs: hypoglycemia, defined as blood glucose < 72 mg/dL (< 4.0 mmol/L), abnormal liver function tests or worsening metabolic acidosis (possible antidote failure). The following symptoms were excluded from review a priori, due to their almost certain association with a non-antidote cause: renal impairment in ethylene glycol poisoning, visual impairment in methanol poisoning, infection, complications from hemodialysis catheter placement or heparinization (e.g. bleeding, bruising). Ethanol withdrawal symptoms were excluded i f ethanol abuse/ dependence was a known pre-existing condition. Abstracted records were scanned three times to ensure that potential ADEs were identified consistently for all incidents. 36 2 Antidote ADEs Four physicians (RP, C D , GE, JK) acted as expert reviewers to evaluate whether the new onset symptoms during antidote treatment were antidote-related ADEs . The reviewers included three emergency medicine physicians with board certification as medical toxicologists and one internist/ clinical pharmacologist. A l l were medical consultants for the B.C. poison control centre, with an average of 15 years each (range 5-22 years) consulting experience. Prior to distribution of study materials, reviewers met to discuss criteria for defining ADEs. Reviewers were provided with an information package summarizing adverse reactions associated with ethanol, fomepizole and hemodialysis. Each reviewer received computer-generated abstracts of poisoning incidents, blinded to hospital site, year of admission and repeated admission status of the patient. Reviewers selected one of the four categories "definite", "probable", "possible", or "doubtful" to best describe the probability that a potential A D E was antidote-related. Symptoms designated definite or probable were classified as ADEs. Three reviewers (reviewers 1-3) independently evaluated potential ADEs , then met to reach consensus on classification. A symptom was designated an antidote related A D E i f all three reviewers classified it as definite or probable in the independent review, or agreed that it was an A D E in the subsequent consensus meetings. Upon completion of A D E classification, incidents were categorized as having: • no ADE: no symptoms attributed to the antidote • any ADE: one or more antidote-related symptoms of any severity • severe ADE: (a subset of any A D E ) one or more severe antidote-related symptoms • serious ADE: (a subset of severe A D E ) one or more antidote-related symptoms consistent with the World Health Organization criteria for a serious event, defined below. A serious A D E was defined as one which prolonged hospitalization, resulted in persistent or significant disability, was life threatening, resulted in death, or required a significant intervention to prevent one of these outcomes. 3 4 ' 3 5 To evaluate the possibility of reviewer bias resulting from knowledge of antidote treatment, a fourth reviewer (reviewer 4) independently evaluated the potential A D E symptoms, but did not participate in the consensus discussions. In two thirds of the incidents, this reviewer was blinded 37 2 Antidote ADEs to antidote treatment group, while one third remained un-blinded for the purpose of comparison. The abstracted information for the blinded incidents showed the start and completion times of "antidote" therapy, but suppressed dose and timing of dose adjustments. Data on serum ethanol concentration and serum osmolality were masked after the start of antidote therapy. For each incident, the blinded reviewer was asked to guess the identity the antidote type (ethanol, fomepizole, both or uncertain). 2.2.6 Outcome measure and treatment group assignment The primary outcome was the dichotomous variable any ADE (reference level no ADE) and the secondary outcome measure was severe ADE. The unit of analysis was the poisoning incident. Incidents were classified as either ethanol or fomepizole treated. Incidents initially treated with ethanol then switched to fomepizole were classified as fomepizole-treated and baseline variables were adjusted to account for any worsening during ethanol therapy. Ethanol-related symptoms with onset before fomepizole start time contributed to the baseline severity of illness, but were not counted as ADEs. Incidents were excluded from analysis i f there were ethanol-related symptoms with onset after the start of fomepizole, or i f it was impossible to determine which antidote was responsible for an A D E . Antidote start time was the beginning of the first antidote dose, excluding initial ethanol use in the fomepizole group. In ethanol-treated patients, antidote completion was defined as the discontinuation time of a continuous intravenous infusion or 2 hours after last oral dose. In fomepizole-treated patients, antidote completion time was one dosage interval after the last antidote dose (four hours during hemodialysis or 12 hours without hemodialysis).1 9 2.2.7 Primary data analysis A l l data summaries and analyses were performed using Microsoft Access 2000™ and S-Plus for Windows™ v. 7.0. Baseline characteristics, treatments before and during antidote therapy and case outcome were summarized by treatment group. Categorical variables were expressed as number and proportion of incidents and continuous variables were reported as mean + standard deviation (sd). Baseline comparisons between treatment groups used the Chi-square test for categorical variables, t-test for continuous variables with normal distribution and Wilcoxon rank-38 2 Antidote ADEs sum for continuous variables with skewed distribution. A two sided p-value <0.05 identified a statistically significant difference between treatment groups. Antidote ADEs were summarized by treatment group according to body system and symptom severity. Results were expressed as the proportion (95% confidence interval, Cl95<>/0) of the incidents experiencing each symptom and the proportion ( 0 9 5 % ) of incidents with any, severe or serious ADE. Inter-rater agreement between expert reviewers was evaluated through kappa scores calculated for reviewer pairs. 3 6 Percent agreement was calculated for the independent evaluations of any ADE between reviewer pairs and for unanimous agreement between all reviewers. Time to any ADE or severe ADE was calculated from antidote start time to onset of the first symptom classified as an A D E . Time to event curves were generated using the product limit (Kaplan-Meier) method and the probability of experiencing an antidote-related A D E was determined for several representative time points.3 7 For the purpose of calculating the A D E rates, each incident contributed antidote treatment hours from antidote start until either the onset of the first A D E or until completion of antidote treatment (if there was no ADE) . A D E rate per person-day was calculated by dividing the total number of incidents with any A D E by the sum of treatment hours/ 24. The unadjusted rate ratio (RR) was calculated by dividing the A D E rate for fomepizole by the A D E rate for ethanol, such that a value less than 1.0 would indicate a relatively lower event rate in the fomepizole group. Rates and RRs for severe ADEs were calculated by the same method. Cox proportional hazard regression was used to model the association between any ADE and antidote treatment group, using ethanol as the reference category.37' 3 8 Potential adjustment variables from baseline clinical and treatment variables were systematically tested in the model i f they showed a significant association with A D E in univariate Cox models. Covariates were retained i f they significantly improved model fit (Chi-square test) relative to the reduced model or i f there was a strong clinical rationale for inclusion. Exponentiation of the regression coefficient for the treatment group provided an estimate of the A D E R R adjusted for confounders. 39 2 Antidote ADEs 2.2.8 Sensitivity analyses Four sensitivity analyses were planned to examine biases which could influence the strength of association between the primary outcome any ADE and antidote treatment group. 1) Reviewer bias during panel discussion: Reviewers used a consensus process to resolve disagreements regarding symptom classification. To test for bias introduced during panel discussions, the outcome was restricted to ADEs identified by unanimous independent agreement of the reviewers 1-3 (i.e. before the consensus discussions). 2) Reviewer bias due to knowledge of treatment group: Outcome was limited to the A D E assessments of the reviewer 4, including both blinded and unblinded subsets. 3) Repeated measures of the same patients: The inclusion of multiple poisoning incidents for the same patient could introduce bias i f the repeat admissions differed from the rest of the sample in their susceptibility to antidote ADEs . This was evaluated by restricting the sample to one incident (the first in the series) per patient per treatment group. 4) Selective antidote prescribing (confounding by indication): Prior to 2002, ethanol was the only available antidote. After 2002, the case mix in each treatment group may have been influenced by selective prescribing of fomepizole based on case characteristics. This was tested by restricting the analysis to incidents admitted during the period when both ethanol and fomepizole were available (2002-2005). 5) Antidote treatment crossover: Patients who experience an adverse effect from one antidote may be more likely to be switched to a second antidote. This treatment crossover could influence the occurrence or observation of ADEs with the second antidote. It was anticipated that crossover would most likely involve a switch from ethanol to fomepizole. The possible influence on study results was tested by excluding incidents with treatment crossover. 40 2 Antidote ADEs 2.3 Results 2.3.1 Characteristics of study subjects Hospital charts were requested for 228 potential poisoning incidents, five of which could not be retrieved. Of the 223 reviewed incidents, 28 were excluded because there was no suspicion of toxic alcohol exposure and 21 were excluded because they did not receive either antidote. The study sample included 174 incidents of suspected toxic alcohol poisoning, 130 treated with ethanol and 44 with fomepizole. The majority of the ethanol-treated incidents (86%) occurred between January 1996 and December 2001, prior to the availability of fomepizole. Table 2.1 summarizes the baseline characteristics of poisoning incidents by treatment group. The ethanol group had more patients with repeated admissions, with an average of 1.3 incidents per patient versus 1.1 in the fomepizole group. Four patients contributed incidents to both treatment groups. Most patients with multiple admissions ingested methanol, resulting in a higher proportion of methanol exposures in the ethanol treatment group. The fomepizole group had a higher proportion of incidents with co-ingestion of pharmaceuticals, minor baseline hypertension and/ or tachycardia, and higher mean A P A C H E II scores than the ethanol group. A l l other baseline clinical variables were similar between treatment groups (p >0.05). Fifteen fomepizole incidents (34%) received ethanol for a mean of 5.3 ± 5.1 hours before switching to fomepizole. None were switched from fomepizole to ethanol. As shown in Table 2.2, a higher proportion of the fomepizole group received hemodialysis in combination with antidote. The use of other treatments was similar for both antidote groups. The two treatment groups had similar outcomes. The ethanol incidents had final Poisoning Severity Scores of no effect or minor effect in 16%, moderate or severe effect in 77% and death in 6%, while the fomepizole group had 16% no effect-minor effect, 72% moderate-severe and 11%) fatal. Patients with ethylene glycol poisoning developed acute renal failure in 25% of ethanol-treated and 16% of fomepizole-treated incidents. Visual sequelae secondary to methanol poisoning occurred in 6% of ethanol-treated and 9% of fomepizole-treated incidents. There was one instance of neurologic sequelae in each treatment group. 41 2 Antidote ADEs Table 2.1 Clinical and demographic characteristics of toxic alcohol poisoning incidents, by antidote treatment group Ethanol Fomepizole P-130 incidents 44 incidents value Patient and Exposure Characteristics Number of patients 97 41 Age on admission (yearsj mean ± sd 37+14 39+ 18 0.42 Sex male n (% of incidents) 89 (68) 33 (75) 0.53 Ethanol abuse history yes 60 (46) 20 (45) 0.92 Toxin type Methanol 69 (53) 11 (25) Ethylene glycol 55 (42) 31 (70) 0.003 Not toxic alcohol (final diagnosis) 6 ( 5) 2 ( 5 ) Toxin exposure reason Intentional (includes undetermined cause) 107 (82) 38 (86) Unintentional (includes substance abuse) 23 (18) 6 (14) 0.70 Co-ingestants (an incident may have more than one co-ingestant) Ethanol (excluding antidote) 32 (25) 14 (32) 0.46 Pharmaceutical 1 includes illicit drugs) 15 (12) 12 (27) 0.02 Non-pharmaceutical 11 ( 8) 2 ( 5) 0.60 Presentation delay mean + sd 9.5 ± 14.1 9.7+10.7 0.34 time (hr) between toxin exposure and hospital admission Peak serum methanol or ethylene glycol concentration not measured 15 (12) 1 ( 2) below treatment threshold (< 20 mg/dL)* 21 (16) 5 (11) above treatment threshold (> 20 mg/dL)* 94 (72) 38 (86) 0.12 Baseline Clinical Condition APACHE II score mean ± sd 8.3 + 8.3 11.3 + 9.0 0.02 Cardiovascular minor-moderate heart rate 40-59 or 110-179, MAP 50-69 or 33 (25) 21 (48) 0.01 110-159; conduction disturbance severe heart rate <40 or >180, MAP <50 or > 160, 5 ( 4) 3 ( 7) <0.66 symptomatic dysrhythmia, cardiac arrest Central Nervous System (CNS) CNS depression minor-moderate GCS 9-14 (drowsiness - stupor) 43 (33) 19(43) 0.30 severe GCS3-8 (coma) 32 (25) 13 (30) 0.66 CNS excitation -minor-moderate restlessness - agitation, single seizure .17 (13) 8 (18) 0.56 severe violent, combative, multiple seizures 16 (12) 5 (11) 0.92 Headache 13 (10) 2 ( 5) 0.42 Gastrointestinal nausea, vomiting, abdominal pain 34 (26) 7 (16) 0.24 Hepatic elevated hepatic enzymes, INR, bilirubin 25 (19) 9 (20) 0.97 Hypoglycemia blood glucose <72 mg/dL* 2 ( 2) 2 ( 5) 0.57 Metabolic Acidosis minor-moderate pH 7.15-7.32 or H C 0 3 10-20 mmol/L 41 (32) 12 (27) 0.73 severe pH <7.15 or H C 0 3 < 10 mmol/L 41 (32) 18 (41) 0.34 Ophthalmic papiledema, visual impairment 20 (15) 3 ( 7) 0.23 Renal impairment: serum creatinine >2.25 mg/dL 7 ( 5) 4 ( 9) 0.61 Data summary, comparison: Categorical: number (%) of incidents, Chi-square test. Continuous: mean + sd, t-test if normally distributed, Wilcoxon rank sum if not normally distributed. Abbreviations: GCS: Glasgow coma score, MAP: mean arterial pressure, APACHE II: acute physiologic and chronic health evaluation II (a summary score of severity of illness), HC03: serum bicarbonate. *SI unit equivalents: methanol 20 mg/dL = 6 mmol/L, ethylene glycol 20 mg/dL= 3 mmol/L; blood glucose 72 mg/dL = 4.0 mmol/L; serum creatinine 2.25 mg/dL =200u.mol/L 42 2 Antidote ADEs Table 2.2 Therapeutic interventions for toxic alcohol poisoning incidents, by antidote treatment group Treatment Type Ethanol Fomepizole p-" 130 incidents 44 incidents value Antidote Therapy Antidote treatment duration (hr) mean + sd 31.2 + 24.3 22.9+ 10.1 0.22 Other Treatments Before Antidote Intubation + ventilation n (% of incidents) 30 (23) 8 (18) 0.63 Sedative medication * At least one dose 40 (31) 20(45) 0.11 # doses, if sedated mean + sd 3.3 + 2.2 4.2 + 3.5 0.44 Physical restraint 14 (11) 6 (14) 0.81 Cardiovascular medication (at least one dose) 9 ( 7 ) 5 (11) 0.53 Other Treatments During Antidote Intubation + ventilation 39 (30) 10 (23) 0.46 Sedative medication during antidote therapy At least one dose 93 (72) 30 (68) 0.82 # doses, if sedated mean±sd 12.2 ±20.9 7.8 ± 9.0 0.35 Physical restraint 36 (28) 8 (18) 0.29 Cardiovascular medication (at least one dose) 15 (12) 7 (16) 0.62 Hemodialysis (toxin clearing) at least one session 89 (68) 39 (89) 0.02 #hours, if dialyzed mean + sd 9.4 + 4.7 . ' 9.1 ±3.4 • 0.70 Data summary, comparison: Categorical: number (%) of incidents, Chi-square test. Continuous: mean ± sd, t-test if normally distributed, Wilcoxon rank sum if not normally distributed. Sedative medications: benzodiazepines, barbiturates, general anesthetics, sedating antihistamines, antipsychotics. Cardiovascular medications: pressors/ inotropes, antihypertensives, antidysrhythmics, atropine, diuretics, plasma expanders, vasopressin. Hemodialysis includes sessions for removal of toxin(s) and correction of acidosis; excludes renal support 2.3.2 Main results 2.3.2.1 ADE evaluation and inter-rater agreement One hundred and forty two of the 174 incidents (82%) had new onset symptoms during antidote therapy, including 114/130 (88%) of the ethanol and 28/44 (64%) of the fomepizole-treated groups. [Refer to Appendix A for a summary of the symptoms flagged for review by the expert panel.] Of the reviewed incidents 74/114 (65%) ethanol and 5/28 (18%) fomepizole incidents were assessed as having any ADE. A n evaluation of the case-related information associated with reviewer acceptance or rejection of symptoms as antidote ADEs indicated that the experts used similar evaluation criteria for A D E assessment regardless of antidote type (Appendix B). 43 2 Antidote ADEs Inter-rater agreement was evaluated for the independent reviews (prior to consensus discussions) for the 135 incidents with complete reviewer data. For reviewers 1-3, percent agreement between reviewer pairs was 73-76%, and kappa scores were 0.41-0.49 (moderate agreement according to Landis & Koch classification of strength of agreement).39 There was unanimous, independent agreement between reviewers 1-3 on the outcome of any ADE for 63% of the reviewed incidents. There was greater variability between each of the three panelists and the fourth independent reviewer, with 57-78% percent agreement between reviewer pairs and kappa scores of 0.26- 0.55 (fair to moderate agreement). There were no significant differences between inter-rater agreement in the blinded and un-blinded subsets of incidents (Appendix B). Comparison of the final consensus panel decision of any ADE to reviewer 4's evaluations showed 78% agreement and kappa score of 0.57 for the un-blinded incidents and 71% agreement, kappa 0.45 for the blinded subset (both moderate agreement). Reviewer 4 was less likely to designate a symptom as an antidote A D E than the panel decision in both the blinded and un-blinded subsets, but the difference was more pronounced when he was blinded to the treatment group. The potential influence of bias on reviewer decision is examined in more detail in Appendix B . It appears that the loss of important dose-related information with the blinding process accounted for much of the difference between the panel and blinded reviewer decision. Reviewer 4 correctly identified the antidote treatment group in 51% of the blinded incidents; however, a higher proportion of the incidents evaluated as having any ADE were correctly identified (82%) compared to those with no A D E (39% correct). Ethanol was correctly identified as the antidote more often than fomepizole. Symptoms excluded from A D E evaluation accounted for less than 20% of the new symptoms during antidote therapy and were present in a similar proportion of incidents in each treatment group (Appendix A) . Four of the incidents treated with both ethanol and fomepizole had ethanol-associated ADEs, but no additional symptoms during fomepizole treatment. These incidents were retained in the fomepizole group coded as no ADE. Two incidents were excluded from analysis because it was not possible to determine which antidote was responsible for a suspected A D E . 44 2 Antidote ADEs 2.3.2.2 Summary of Antidote ADEs Table 2.3 summarizes the proportion of incidents with any, severe and serious ADEs and provides a breakdown of ADEs by body system. Case summaries are provided for all incidents with serious ADEs, see Appendix C. Table 2.3 Proportion of toxic alcohol poisoning incidents with antidote-related adverse drug events (ADEs), determined by consensus agreement of expert reviewers n (% of incidents; CI 9 5 %) Ethanol Fomepizole Antidote-Related ADEs 130 incidents 42 incidents Any ADE > 1 ADE, any symptom severity 74 (57; 48,65) 5 (12; 2,22) Severe ADE > 1 ADE of "severe" symptom intensity 26 (20; 13,27) 2 ( 5; 0,11) Serious ADE > 1 potentially life threatening ADE 11 ( 8; 4,13) 1 ( 2; 0, 7) Antidote ADEs by Symptom Category Cardiovascular, any 7 ( 5; 2, 9) 1 (2; 0, 7) minor-moderate heart rate 40-59 or 110-179, M A P 50-69 or 3 ( 2; 0, 5) 0 110-159; conduction disturbance severe heart rate <40 or >180, MAP <50 or > 160, 4 ( 3; 0, 6) 1 (2; 0, 7) symptomatic dysrhythmia, cardiac arrest Central Nervous System (CNS), any 63 (48; 40,57) 1 (2; 0, 7) CNS depression, any 55 (42; 34,51) 1 (2; 0,7) minor-moderate GCS 9-14 (drowsiness - stupor) 41 (32; 24,40) 0 severe GCS 3-8 (coma) 14 ( i i ; 5,16) 1 (2; 0,7) CNS excitation, any 23 (18; 11,24) 0 minor-moderate restlessness - agitation, single seizure 13 (10; 5,15) 0 severe violent, combative, multiple seizures 10 ( 8; 3,12) 0 Headache 4 ( 3; 0, 6) 0 Gastrointestinal nausea, vomiting, epigastric/ abdominal pain 12 ( 9; 4,14) 3 (7; 0,15) Hepatic elevated hepatic enzymes, INR, bilirubin 0 0 Hypoglycemia blood glucose <72 mg/dL (< 4 mmol/L) 5 ( 4; 1. 7) 0 Metabolic Acidosis 0 0 Worsening acidosis during antidote (suspected antidote failure) 0 0 Other Antidote ADEs, any 8 ( '6; 2,10) 0 Polyuria (diuresis) 3 ( 2; 0, 5) 0 Phlebitis (antidote intravenous line) 5 ( 4; 1, 7) 0 Abbreviations: GCS: Glasgow coma score, MAP: mean arterial pressure, APACHE II: acute physiologic and chronic health evaluation II (a summary score of severity of illness), Cl95%: 95% confidence interval Central nervous system (CNS) symptoms accounted for the majority of ADEs in the ethanol group. O f the fourteen incidents with severe ethanol-induced CNS depression (i.e. coma), eight experienced serious events involving compromised airway or breathing which required 45 2 Antidote ADEs intervention. Ten incidents experienced severe ethanol-related agitation or violent behavior which required pharmacologic and physical and restraint. No serious events could be directly attributed to agitation; however, one patient suffered a life-threatening arterial tear during hemodialysis catheter insertion which was likely secondary to his agitated state during the procedure. The single CNS adverse event attributed to fomepizole involved transient re-sedation from GCS 13 to GCS 6 after administration of an excessive dose. The patient remained intubated before, during and after this event. Cardiovascular ADEs accounted for less than five percent of all antidote-associated symptoms, but contributed to clinically important morbidity. Three of the four incidents with severe ethanol-related hypotension had serious events requiring bolus doses of intravenous fluid, plasma expander, pressors or other interventions. One critically i l l patient experienced a dramatic drop in blood pressure and had a cardiac arrest during the ethanol loading infusion. He was resuscitated but his blood pressure dropped again when ethanol was restarted. There was one serious fomepizole-related cardiovascular A D E . The patient experienced hypotensive bradycardia immediately following infusion of the first and second fomepizole doses, both given during hemodialysis. In the first episode, mean arterial pressure fell to 33 and heart rate to 29 but both corrected promptly with atropine and plasma expander. The second episode was less severe (see Appendix C). Minor gastrointestinal ADEs were noted in both ethanol and fomepizole-treated patients. Neither group had altered hepatic function tests attributed to antidote. The ADEs hypoglycemia, phlebitis, headache and polyuria were noted in a small proportion of ethanol-treated incidents (<5% each) but were absent in the fomepizole group. Most of these events were of minor clinical importance. Four of the five incidents with "hypoglycemia" had blood glucose of 58-68 mg/dL (3.2-3.8 mmol/L) and one had a low of 49 mg/dL (2.7 mmol/L) but none were symptomatic. One of the five incidents with ethanol-related phlebitis developed a markedly swollen hand, but the others had only localized pain and inflammation. Polyuria, possibly due to osmotic diuresis, responded to ethanol dose adjustment. 46 2 Antidote ADEs 2.3.2.3 Onset time of ADEs Figures 2.1 and 2.2 show the cumulative probability of experiencing any ADE or a severe ADE respectively. The significantly higher adverse event rate in the ethanol-treated incidents was evident soon after the start of antidote and was sustained throughout the treatment course. By two hours after antidote start, the probability of having any ADE was 35% for ethanol and 5% for fomepizole-treated incidents, while the probability of having a severe ADE was 16% for ethanol and 2% for fomepizole at the two hour mark. Figure 2.1 Time to onset of first antidote-related adverse drug event (any ADE) in toxic alcohol poisoning incidents L U Q < < 15 -O o CL > _ E 0 0 d Hour Cumulative probability of Any ADE (Cl95%) Ethanol Fomepizole 1 0.25 (0.17, 0.32) 0.02 (0.00,0.07) 2 0.34 (0.25,0.42) 0.05 (0.00,0.11) 6 0.45 (0.36, 0.53) 0.10 (0.00,0.18) 24 0.58 (0.48, 0.67) 0.13 (0.02,0.23) 48 0.67 (0.54,0.76) 0.13 (0.02,0.23) Ethanol 1 18 24 30 36 Elapsed time (hours) after start of antidote 42 48 47 2 Antidote ADEs Figure 2.2 Time to onset of first severe antidote-related adverse drug event (severe ADE) in toxic alcohol poisoning incidents oo d CD d CL d Hour Cumulative probability of Severe ADE (Cl95%) Ethanol Fomepizole 1 0.11 (0.05,0.16) 0.02 (0.00,0.07) 2 0.16 (0.09,0.22) 0.02 (0.00,0.07) 6 0.21 (0.13,0.29) 0.02 (0.00,0.07) 24 0.25 (0.16,0.33) 0.06 (0.00,0.14) 48 0.25 (0.16,0.34) 0.06 (0.00,0.14) Ethanol Fomepizole 24 30 36 42 48 Elapsed time (hours) after start of antidote 2.3.2.4 Comparison of ADEs between treatment groups The unadjusted R R for any ADE showed a seven-fold reduction in the adverse event rate in the fomepizole treatment group relative to ethanol (Table 2.4). This effect was attenuated to a six-fold reduction following adjustment for potential confounders using Cox proportional hazard regression. The final model included the A P A C H E II score as an adjustment variable, categorized to represent minor, moderate and severe illness. The most severe category of initial illness was associated with a 2.6-fold decrease in the risk of experiencing an antidote A D E relative to the least severe illness category. There was a four-fold reduction in the unadjusted risk of experiencing a severe ADE in the fomepizole treatment group relative to ethanol treatment. Adjustment for the presence of pre-treatment agitation and ethanol ingestion pre-hospital increased the treatment effect to a nearly six-fold difference between groups (Table 2.4). 48 2 Antidote ADEs The presence of either of these confounders was associated with more than a two-fold increase in the risk of experiencing an A D E . Table 2.4 Association between antidote-related adverse drug event (ADE) rate and antidote type in toxic alcohol poisoning incidents Covariate, category levels ADE rate per person-day of antidote treatment Unadjusted RR(CI9S%)* Adjusted RR(CI 9 5 %)** Any antidote-related ADE Antidote ethanol (reference) 0.93 (0.87,0.98) 1.0 1.0 fomepizole 0.13 (0.02,0.24) 0.14(0.06,0.36) 0.17(0.07, 0.42) APACHE II 0-2 (reference) 1.0 >2-8 0.71 (0.43, 1.19) >8-30 0.38(0.21,0.67) Severe antidote-related ADE Antidote ethanol (reference) 0.20 (0.13,0.27) 1.0 1.0 fomepizole 0.05 (0.00,0.12) 0.25 (0.06,1.07) 0.18 (0.04,0.76) Pre-Treatment no (reference) 1.0 Agitation yes 2.42 (1.13, 5.18) Ethanol no (reference) 1.0 Coingestant yes 2.67(1.26,5.66) Abbreviations: RR: Rate Ratio, CI9 5%: 95% confidence interval. •"Unadjusted RR: ADE rate in fomepizole-treated divided by ADE rate in ethanol-treated **Adjusted RR is calculated by exponentiation (ep) of the Cox proportional hazard regression model coefficients (P) for each predictor variable. Interpretation: the risk of experiencing an antidote ADE relative to the reference level of the variable, when all other covariates are held constant. A l l covariates included in the models complied with the assumption of proportionality over time required for the Cox model (Appendix D). The R-squared values for the final models were 0.21 for any ADE and 0.10 for severe ADE, which suggests that these models explained less than one quarter of the sources of variability in antidote A D E rates. 2.3.3 Sensitivity analyses The results of the five planned sensitivity analyses are presented in Table 2.5 and demonstrate that the results of the primary analysis were robust to the application of more stringent criteria for defining any ADE and to modifications of case mix. The treatment effect was maintained, and in some instances was even magnified, and the confidence intervals around the relative rate estimates did not include 1.0. 49 2 Antidote ADEs Table 2.5 Sensitivity analyses for any antidote - related adverse drug event (any ADE) in toxic alcohol poisoning incidents Any ADE rate per person-day of antidote treatment Adjusted A n a , y s l S Ethanol Fomepizole RR(CI 9 5o / o)* Primary Outcome Any ADE, consensus panel agreement 0.93 (0.87,0.98) 0.13 (0.02,0.24) 0.17(0.07,0.42) (reviewers 1, 2 & 3) Vary criteria for identifying Any ADE 1) Unanimous agreement on independent 0.60 (0.50,0.70) 0.03 (0.00,0.07) 0.05 (0.01,0.36) evaluation (reviewers 1, 2 & 3 pre-consensus) 2) Reviewer 4 independent review 0.44 (0.35,0.53) 0.05 (0.00,0.12) 0.14(0.03,0.57) (66% blinded to treatment group, 34% un-blinded) Vary sample selection 3) Restrict sample to one incident per patient 0.87 (0 79, 0.96) 0.14 (0 03, 0.26) 0 18(0 07, 0.45) per treatment group (first admission) 4) Restrict sample to 2002-2005 admissions 1.00 (1 00, 1.00) 0.13 (0 02, 0.24) 0 13(0 04, 0.46) (after fomepizole availability) 5) Restrict sample to incidents with one antidote 0.93 (0 87, 0.98) 0.16 (0 02, 0.30) 0 19(0 07, 0.51) (exclude incidents with treatment crossover) Abbreviations: RR: Rate Ratio, CI9504: 95% confidence interval. Number of incidents: Analyses 1 and 2 n= 130 ethanol, 42 fomepizole. Analysis 3: n= 97 ethanol, 39 fomepizole. Analysis 4: n= 18 ethanol, 42 fomepizole. Analysis 5: n= 130 ethanol, 29 fomepizole *adjusted RR, Cox proportional hazard regression model ADE~ Antidote + APACHE II as per table 2.4 50 2 Antidote ADEs 2.4 Limitations Study limitations include the retrospective design, which is susceptible to both missing information and misclassification (e.g. incorrect documentation of symptom onset times in the medical chart). There is no gold standard for definitively identifying an adverse drug event. 2 9 ' 4 0 ' 4 1 Evaluation of symptom causality by an expert panel is traditionally used as the standard for validation of other methods such as algorithms and was therefore chosen as the assessment tool in this study; however, the process of causality assessment is inherently subjective.40' 4 2 - 4 4 This study is unique in the use of a blinded review to evaluate potential biases influencing the conclusions of the expert panel. While this exercise provided insight into the decision-making process, its utility was limited by the inability to completely mask the identity of the antidote and the loss of other important information required for causality assessment (e.g. antidote dose). The study subjects were adults and adolescents, most of whom were treated with a short (approximately 24 hour) course of antidote with hemodialysis. The results may not be generalizable to children, or to patients who receive a more prolonged course of antidote without hemodialysis. 51 2 Antidote ADEs 2.5 Discussion This is the first study to compare the safety of ethanol and fomepizole when used as antidotes for toxic alcohol poisoning and represents one of the largest studies of these antidotes to date. The results confirm that fomepizole is associated with a significantly lower adverse event rate than ethanol, showing a six-fold reduction in the adjusted risk of any adverse event and similar reduction in the adjusted risk of a severe event. Ethanol-related CNS toxicity was largely responsible the difference in A D E rates between treatment groups. Severe antidote-related symptoms developed in 20% of ethanol-treated incidents. Nearly half of these incidents (8% of the ethanol-treated group) had potentially life-threatening ADEs which required intervention such as intubation or hemodynamic support. While severe ADEs such as violence or coma without respiratory depression were not immediately life-threatening, patients with these symptoms required intensive management by hospital staff. Some of the symptoms allegedly associated with ethanol therapy, such as hypoglycemia and phlebitis were seen infrequently and did not result in significant morbidity. The low incidence of hypoglycemia confirms the observations of others and suggests that ethanol antidote is unlikely to cause clinically significant hypoglycemia when administered with a source of glucose. 1 3 ' 1 4 The episode of serious hypotensive bradycardia attributed to fomepizole responded promptly to treatment; however, this event serves as a reminder that fomepizole is a new therapy and our knowledge of its adverse effects is still evolving through post-marketing surveillance. Healthcare professionals should continue to observe and report unexpected reactions to new drugs: In most cases, the onset of antidote-related toxicity was rapid and the advantage of fomepizole over ethanol was apparent within two hours after the first dose of antidote. This observation has important implications for case management. Antidote-related toxicity is most likely to begin while the patient is in the ED, therefore the choice of antidote could have a substantial impact on the degree of patient monitoring and support required of ED personnel. Patients with toxic alcohol poisoning often require transfer to another hospital, which typically occurs within a few hours after presentation to the ED. The intoxicating effects of ethanol can place both the patient 52 2 Antidote ADEs and attending health care professionals at risk during transport. One patient in this study developed severe agitation, declining level of consciousness and depressed respirations secondary to ethanol during transport in a small aircraft. The timing of antidote-related ADEs also suggests that the practice of starting a patient on ethanol then switching to fomepizole after confirmation of diagnosis puts the patient at increased risk for experiencing an ethanol-related A D E . In the study sample, the average duration of ethanol therapy prior to switching to fomepizole was five hours and approximately half of the patients treated with both antidotes experienced an ethanol-related adverse event. In conclusion, fomepizole was associated with a lower risk of any ADE and a lower risk of severe ADEs than ethanol when used as an antidote for toxic alcohol poisoning. This treatment advantage was apparent shortly after onset of therapy and was maintained throughout the treatment course. Areas for future study include ethanol and fomepizole-related medication errors and economic evaluation. 53 2, Antidote ADEs 2.6 References 1. Lai M W , Klein-Schwartz W, Rodgers C G , et al. 2005 annual report of the American association of poison control centers' national poisoning and exposure database. Clin Toxicol. 2006;44(6):803-932. 2. Roe O. Methanol poisoning: its clinical course, pathogenesis and treatment. Acta Medica Scandinavica. 1946;Suppl 182:1-253. 3. Roe O. Clinical investigations of methyl alcohol poisoning with special reference to the pathogenesis and treatment of amblyopia. Acta Medica Scandinavica. 1943; 113(6):558-608. 4. Seifert S. Ethanol. In: Dart R C , ed. Medical Toxicology. 3 ed. Philadelphia: Lippincott Williams & Wilkins; 2004:196-199. 5. Sharma A N . Toxic alcohols. In: Goldfrank L R , Flomenbaum N E , Lewin N A , et al., eds. Gold/rank's Toxicologic Emergencies. 7 ed. New York: McGraw-Hill ; 2002:980-990. 6. Jolliff H A , Sivilotti M L . Ethylene glycol. In: Dart R C , ed. Medical Toxicology. 3 ed. Philadelphia: Lippincott Williams & Wilkins; 2004:1223-1230. 7. Ford M D , McMartin K . Ethylene glycol and methanol. In: Ford M D , Delaney K A , Ling LJ , et al., eds. Clinical Toxicology. Philadelphia: W.B.Saunders Co; 2001:757-767. 8. Howland M A . Antidotes in Depth: Ethanol. In: Goldfrank LR, Flomenbaum N E , Lewin N A , et al., eds. Gold/rank's Toxicologic Emergencies. 6 ed. New York: Appleton & Lange; 1998. 9. Sivilotti M L . Ethanol, Isopropanol and Methanol. In: Dart R C , ed. Medical Toxicology. 3 ed. Philadelphia: Lippincott Williams & Wilkins; 2004:1211-1230. 10. Barceloux D G , Krenzelok EP, Olson K , et al. American Academy of Clinical Toxicology Practice Guidelines on the Treatment of Ethylene Glycol Poisoning. Journal of Toxicology - Clinical Toxicology. 1999;37(5):537-560. 11. Kowalczyk M , Halvorsen S, Ovrebo S, et al. Ethanol treatment in ethylene glycol poisoned patients. Veterinary & Human Toxicology. Aug 1998;40(4):225-228. 12. Meyer RJ, Beard M E , Ardagh M W , et al. Methanol poisoning. New Zealand Medical Journal. Jan 28 2000;113(1102):11-13. 13. Roy M , Bailey B , Chalut D, et al. What are the adverse effects of ethanol used as an antidote in the treatment of suspected methanol poisoning in children? Journal of Toxicology - Clinical Toxicology. 2003 ;41 (2): 15 5-161. 14. Lister D, Tierney M , Dickinson G. Effectiveness of IV ethanol therapy combined with hemodialysis in the treatment of methanol and ethylene glycol poisoning. Canadian Journal of Hospital Pharmacy. 2005;58(3): 142-147. 15. Brent J, McMartin K , Phillips S, et al. Fomepizole for the treatment of ethylene glycol poisoning. NEngl J Med. Mar 18 1999;340(11):832-838. 16. Brent J, McMartin K , Phillips S, et al. Fomepizole for the treatment of methanol poisoning. N Engl J Med. Feb 8 2001;344(6):424-429. 17. Jacobsen D, Sebastian CS, Blomstrand R, et al. 4-Methylpyrazole: a controlled study of safety in healthy human subjects after single, ascending doses. Alcohol Clin Exp Res. Aug 1988;12(4):516-522. 18. Baud FJ, Galliot M , Astier A , et al. Treatment of ethylene glycol poisoning with intravenous 4-methylpyrazole. N Engl J Med. Jul 14 1988;319(2):97-100. 54 2 Antidote ADEs 19. Antizol (fomepizole) injection product monograph. Jazz Pharmaceuticals. Available at: http://antizol.info/monograph.htm. Accessed March 2, 2007. 20. Brent J. Antidotes and alcohols: has fomepizole made ethanol an obsolete therapy? American College of Medical Toxicology. Available at: http://www.ijmt.net/ijmt/l_l/l_l_2.html. Accessed January 29th, 2007. 21. Barceloux D G , Bond GR, Krenzelok EP, et al. American Academy of Clinical Toxicology practice guidelines on the treatment of methanol poisoning. Journal of Toxicology - Clinical Toxicology. 2002;40(4):415-446. 22. Litovitz T, Klein-Schwartz W, White S M , et al. 2000 Annual report of the American Association of Poison Control Centers Toxic Exposure Surveillance System. American Journal of Emergency Medicine. 2001;19(5):337-395. 23. Megarbane B , Borron SW, Trout H , et al. Treatment of acute methanol poisoning with fomepizole. Intensive Care Medicine. Aug 2001;27(8):1370-1378. 24. Jacobsen D. Ethylene glycol and other glycols. In: Brent J, Wallace K L , Burkhart K K , et al., eds. Critical Care Toxicology: diagnosis and management of the critically poisoned patient. Philadelphia: Elsevier Mosby; 2005. 25. Hovda K E , Hunderi O H , Tafjord A B , et al. Methanol outbreak in Norway 2002-2005: epidemiology, clinical features and prognostic signs. Journal of Internal Medicine. 2005;258:181-190. 26. Mycyk M , Des Lauriers C, Metz J, et al. Compliance with poison center fomepizole recommendations is suboptimal in cases of toxic alcohol poisoning. American Journal of Therapeutics. 2006;13:485-489. 27. W H O - U M C . Side effect- adverse reaction. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Available at: http://www.who-umc.org/DynPage.aspx?id=22676. Accessed March 3, 2007. 28. Aronson J, Ferner R. Clarification of terminology in drug safety. Drug Saf. 2005;28(10):851-870. 29. Nebeker J, Barach P, Samore M . Clarifying adverse drug events: a clinician's guide to terminology, documentation and reporting. Ann Intern Med. 2004;140:795-801. 30. Morimoto T, Gandhi T, Seger A , et al. Adverse drug events and medication errors: detection and classification methods. Quality and Safety in Health Care. 2004; 13:306-314. • 31. Knaus W, Draper E, Wagner D, et al. A P A C H E II: A severity of disease classification system. Crit Care Med. 1985;13(10):818-829. 32. Marandon J, Pezzullo J. Scoring systems for ICU and surgical patients: A P A C H E II. Societe Frangaise d'Anesthesie et de Reanimation. Available at: http://www.sfar.org/scores2/apache22.html. 2006. 33. Persson H E , Sjoberg G K , Haines JA, et al. Poisoning Severity Score. Grading of acute poisoning. Journal of Toxicology - Clinical Toxicology. 1998;36(3):205-209. 34. W H O U M C . Definitions: Serious adverse reaction. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Apri l 13, 2006. Available at: http://www.who-umc.org/DynPage.aspx?id=22680. Accessed Jan 11, 2007. 35. C A D R M P . Canadian adverse drug reaction monitoring program ( C A D R M P ) guidelines for the voluntary reporting of suspected adverse reactions to health products by health professionals and consumers. Health Canada: Marketed Health Products Directorate. 55 2 Antidote ADEs 2006-05. Available at: http://www.hc-sc.gc.ca/dhp-mps/alt_formats/hpfb-dgpsa/pdf/medeff/ar-ei_guide-ldir_e.pdf. Accessed March 3, 2007. 36. Szklo M , Nieto FJ. Epidemiology. Beyond the basics. Gaithersburg, M D : Aspen Publishers, Inc.; 2000. 37. Selvin S. Practical Biostatistical Methods. Belmont: Duxbury Press; 1995. 38. Cox DR. Regression models and life tables. Journal of the Royal Statistical Society Series B (Methodology). 1972;34(2): 187-220. 39. Landis JR, Koch G G . The measurement of observer agreement for categorical data. Biometrics. 1977;33(1): 159-174. 40. Macedo A F , Marques F B , Ribeiro CF, et al. Causality assessment of adverse drug reactions: comparison of the results obtained from published decisional algorithms and from the evaluations of an expert panel, according to different levels of imputability. Journal of Clinical Pharmacy and Therapeutics. Apr 2003;28(2): 137-143. 41. W H O - U M C . Causality Assessment of Suspected Adverse Reactions. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Available at: http://www.who-umc.org/DynPage.aspx?id=22682. Accessed Feb 16, 2007. 42. Hutchinson T A , Lane D A . Assessing Methods for Causality Assessment of Suspected Adverse Drug-Reactions. J Clin Epidemiol. 1989;42(1):5-16. 43. Karch FE, Smith C L , Kerzner B , et al. Adverse Drug-Reactions - Matter of Opinion. Clin Pharmacol Ther. 1976; 19(5):489-492. 44. Kramer M S . Assessing causality of adverse drug reactions: global introspection and its limitations. DruglnfJ. 1986;20:433-437. 56 3 Discussion 3 DISCUSSION 3.1 Clinical implications The results of this study show a substantially lower rate of antidote-related ADEs in patients with toxic alcohol poisoning treated with fomepizole compared to those treated with ethanol, with ethanol-related central nervous system effects accounting for the majority of the difference. This finding is consistent with the low frequency of fomepizole-related adverse drug events in pre-marketing studies and the known intoxicating properties of ethanol at the serum concentrations used therapeutically for toxic alcohol poisoning. 1 ' 2 The majority of patients who experienced an antidote-related A D E had the onset of the first event within the first few hours of therapy. These findings inform clinical decision making and help clarify the place of fomepizole in therapy. 3.1.1 Choice of antidote There is insufficient evidence to determine whether fomepizole is superior to ethanol in terms of improved survival or reduced toxin-related sequelae (blindness, renal failure). In theory, both antidotes should be equally efficacious in blocking production of toxic metabolites i f administered in adequate doses. Until there is a sufficient body of evidence to reveal any difference in effectiveness, the choice between the two antidotes must be based primarily on their comparative safety and ease of administration. In September 2005, B.C. DPIC released treatment guidelines for the management of toxic alcohol poisoning which designated fomepizole as the antidote of first choice for toxic alcohol poisoning.3 This recommendation was based on the results of a pilot study regarding ethanol safety, DPIC fomepizole program data, extensive literature review and consensus agreement of a panel of DPIC pharmacists, medical toxicologists and nurses. The study presented in Chapter 2 was undertaken to provide additional information which would either confirm this decision, or lead to modification of the treatment guidelines. The results of this study do support the choice of fomepizole as the preferred antidote for toxic alcohol poisoning on the basis of a more favourable adverse effect profile. The rapid onset of antidote-related ADEs suggests that, in order to prevent ethanol-related ADEs and take full advantage of the advantages of fomepizole, it is most logical to initiate therapy with fomepizole 57 3 Discussion whenever antidote is required, rather than start with ethanol and switch to fomepizole after confirming the diagnosis obtaining a fomepizole supply. 3.1.2 An t ido te stock ing decisions If fomepizole is to be used as a first line antidote, it must be available in acute care hospitals throughout the province. When fomepizole (Antizol™) was first introduced to the Canadian market, it had a three year shelf life with no credit available for expired stock. It was therefore considered a prohibitively expensive inventory item for small health care facilities which infrequently treat cases of toxic alcohol poisoning.4 The shelf life was subsequently extended to four years and the cost associated with carrying fomepizole in inventory has been reduced through the introduction of a credit policy for expired stock (i.e. the Canadian distributor credits the hospital for the full price of expired fomepizole provided the outdated product is returned according to the conditions).5 These changes make it feasible for small health care facilities to stock a costly antidote which might only be used infrequently. At the time of writing, DPIC continues to offer free stock replacement for fomepizole used in consultation with a toxicologist. The B.C. Drug and Poison Information Centre (DPIC) publishes antidote stocking guidelines for B .C . hospitals. In response to the results of a pilot study regarding ethanol safety, DPIC fomepizole program data, the preliminary findings of this study (Chapter 2) and the distributor's credit return policy, the June 2006 version of these guidelines recommend that all B .C . acute care hospitals carry an inventory of fomepizole appropriate for their needs (one to four via ls) . 4 ' 6 ' 7 As of January 2007, 66 (73%) of B C acute care hospitals surveyed stocked at least one vial of fomepizole (data on file, DPIC). 58 3 Discussion 3.2 Ongoing and future study The study reported here (Chapter 2) is one component of a more comprehensive investigation of the safety of ethanol and fomepizole. The second phase of the study wil l characterize and compare medication errors associated with these antidotes. Harm associated with medication errors is potentially preventable or ameliorable. Results of the medication error study wil l inform the development of educational materials to alert health care professionals about common sources of error and suggest preventive measures. Discovering sources of error which lead to wastage or excessive doses of fomepizole are of particular interest, given the high cost of this antidote. A n economic evaluation would provide clarity regarding the most cost-effective way to use the available therapies for toxic alcohol poisoning. A previous attempt to compare the cost of ethanol and fomepizole-related ADEs was hampered by lack of data.8 A similar evaluation using the data from this study would be of interest. There is still uncertainty whether uncomplicated cases of ethylene glycol poisoning are best treated with fomepizole alone or with antidote in combination with hemodialysis. 2 ' 9 A comparison of these two therapeutic approaches would be a logical focus of a future economic evaluation. 59 3 Discussion 3.3 References 1. Barceloux D G , Bond GR, Krenzelok EP, et al. American Academy of Clinical Toxicology practice guidelines on the treatment of methanol poisoning. Journal of Toxicology - Clinical Toxicology. 2002;40(4):415-446. 2. Barceloux D G , Krenzelok EP, Olson K , et al. American Academy of Clinical Toxicology Practice Guidelines on the Treatment of Ethylene Glycol Poisoning. Journal of Toxicology - Clinical Toxicology. 1999;37(5):537-560. 3. DPIC. Treatment guidelines for methanol and ethylene glycol poisoning. Vancouver, B C : B C Drug & Poison Information Centre; Prepared by: K J Lepik; Reviewed by: JR Brubacher, C DeWitt, G Erhardt, M Friesen, R Gair, L Hayes, J Kennedy, D A Kent, D Leong, R L i , R A Purssell; September 19 2005.. 4. Lepik K J , Daws DE. BC Drug and Poison Information Centre fomepizole program report 2002-2004. Vancouver: B C Drug & Poison Information Centre; September 1 2005. 5. Memo: Paladin Labs credit return policy for Antizol. Paladin Labs Inc., Montreal QC; 2005. 6. DPIC. Antidote stocking guidelines for B.C. hospitals. Vancouver: B C Drug & Poison Information Centre; Prepared by: K J Lepik. Reviewed by: JR Brubacher, D Daws, C DeWitt, G Erhardt, M Friesen, R Gair, L Hayes, J Kennedy, D A Kent, D Leong, R L i , R A Purssell, M O Weins.; June 12 2006. 7. Lepik K J , Purssell R A , Levy A R , et al. Adverse reactions to ethanol antidote for toxic alcohol poisoning: a multi-centre retrospective review (abstract). Clinical Toxicology. 2005;43(6):657-658. 8. Marraffa J, Stork C, Medicis J. Cost-effectiveness of fomepizole versus ethanol in the management of acute ethylene glycol exposure. (Abstract 128). Clin Toxicol. 2005;43(6):691. 9. Borron SW, Megarbane B , Baud FJ. Fomepizole in treatment of uncomplicated ethylene glycol poisoning.[see comment]. Lancet. Sep 4 1999;354(9181):831. 60 Appendix A APPENDICES Appendix A: New Onset Symptoms During Antidote Therapy A l l incidents with new onset symptoms during antidote therapy were reviewed by the expert panel, with the exception of symptoms excluded a priori (see Chapter 2, Methods). Table A . l summarizes the symptoms excluded from consideration as potential antidote ADEs . Table A . l Symptoms excluded from review by expert panel number (% of incidents) Symptom Category Ethanol Fomepizole n=130 n= 42 Renal impairment: creatinine >2.25 mg/dL (>200pmol/L) 10 (8) 3 (7) Visual impairment or blurred vision 12 (9) 1 (2) Infection 3 (2) 3 (7) Hemodialysis procedural complication 4 (3) 1 (2) Ethanol withdrawal symptoms 4 (3) . 2 (5) Table A.2 summarizes the proportion of incidents with new onset symptoms during antidote therapy and shows the proportion which were classified as ADEs by the expert reviewers. For incidents with new cardiovascular and gastrointestinal symptoms, the proportion of symptoms classified as ADEs was similar for both ethanol and fomepizole treated patients. Reviewers classified a significantly higher proportion of the incidents with new CNS symptoms as antidote ADEs in the ethanol group (65%) compared to the fomepizole group (7%). Possible explanations for this discrepancy are explored in Appendix B . The fomepizole group had no new onset symptoms or laboratory test abnormalities related to hepatic function, low blood glucose, worsening acidosis or other symptoms. In the ethanol group, most of the hepatic abnormalities recorded after the start of antidote (Table A.2) were actually the first recorded hepatic function tests during the hospital admission and likely reflected baseline status. None were designated as antidote ADEs. Three ethanol treated incidents had a minor drop in arterial pH or serum bicarbonate after the start of antidote therapy; however, alternate causes could explain these changes and none were considered to have antidote failure. 61 Appendix A Table A.2 Toxic alcohol poisoning incidents with new onset symptom(s) during antidote therapy and proportion of incidents assessed by expert panel as having ADE(s) # incidents with an ADE / New onset symptom during antidote treatment, by body system # incidents with a new symptom (%) Ethanol Fomepizole P-value* Cardiovascular 11 27 (26) 1/5 (20) 0.999 minor-moderate heart rate 40-59 or 110-179, MAP 50-69 or 3/22 (14) 0/1 ( 0) *na 110-159; conduction disturbance severe heart rate <40 or > 180, MAP <50 or > 160, 4 /5 (80) 1/4 (25) na symptomatic dysrhythmia, cardiac arrest na Central Nervous System (CNS) 63/ 97 (65) 1/15 ( 7) <0.001 CNS depression 55/ 73 (75) ' 1/6 (17) 0.007 minor-moderate GCS 9-14 (drowsiness - stupor) 41/48 (85) 0/3 ( 0) na severe GCS 3-8 (coma) 14/25 (56) 1/3 (33) na CNS excitation • 23/50 (50) 0/8 ( 0) 0.017 minor-moderate restlessness - agitation, single seizure 13/33 (39) 0/5 ( 0) na severe violent, combative, multiple seizures 10/17 (59) 0/3 ( 0) na Headache 4/10 (40) 0/3 ( 0) 0.497 Gastrointestinal nausea, vomiting, epigastric/ abdominal 12/33 (36) 3/11 (27) 0.722 pain Hepatic elevated hepatic enzymes, INR, bilirubin 0/ 14 ( 0) 0/0 ( 0) na Hypoglycemia blood glucose <72 mg/dL* 5/9 (56) 0/0 ( 0) na Metabolic Acidosis Worsening acidosis during antidote 0/3 ( 0) 0/0 ( 0) na (suspected antidote failure) Other Antidote ADEs 8/13 (62) 0/0 ( 0) na Polyuria (diuresis) 3/5 (60) 0/0 ( 0) na Phlebitis (antidote IV line) 5/8 (63) 0/0 ( 0) na *Fisher's Exact test, na: comparisons were not made for symptom severity subcategories or if one of the treatment groups had no new onset symptom in a symptom category. 62 Appendix B Appendix B: Examination Of Possible Biases Influencing Study Results B.l Overview The study results show a large difference in rate of any antidote-related A D E between ethanol and fomepizole treated incidents (adjusted rate ratio 0.17). This difference between treatment groups could represent a real difference in occurrence of antidote-related symptoms. However, it is important to identify and, i f possible, quantify sources of bias in order to enhance the validity of the results and understand the potential weaknesses. There are four areas in which bias could inflate the observed A D E rate in ethanol treated poisoning incidents: 1) Health care providers may be more likely to observe and record a symptom if there is an expectation that it wi l l occur. The intoxicating properties of ethanol are well known, while fomepizole is promoted as having few side effects on the central nervous system (CNS). This could result in more meticulous documentation of CNS symptoms in ethanol treated patients. 2) In the causality assessment of adverse drug reactions, previous reports of a reaction are weighed as positive evidence of an A D E . The toxicity of ethanol is well documented in the substance abuse literature, whereas fomepizole is a relatively new drug and its known toxicity profile is still evolving. Expert reviewers may have been more likely to classify a symptom as an A D E i f it had been previously linked to the antidote. 3) Dose-related toxicity is an important consideration when evaluating whether or not a symptom is an A D E . Administration of an excessive ethanol dose may be determined from the administered milliliter per kilogram dose and the serum ethanol level, while an excessive fomepizole dose can only be identified by the administered milligram per kilogram dose. Reviewers potentially had more opportunity to identify ADEs associated with high doses of ethanol than fomepizole. 4) Reviewers were aware of the study hypothesis that fomepizole has a lower A D E rate than ethanol. Despite efforts to be impartial, the reviewers may have been more likely to attribute ADEs to ethanol than fomepizole. 63 Appendix B These potential biases result from an unavoidable imbalance in the available information regarding ethanol and fomepizole. The following analyses look for evidence of these biases and evaluate the impact of these biases on the study results. B.2 Caregiver bias in symptom documentation Incidents with a new onset symptom during antidote therapy were scanned for documentation that either a hospital caregiver or the patient had expressed the opinion that one or more symptoms were caused by the antidote. This information was routinely recorded during data collection. Incidents for which the caregiver-attributed symptom was the only symptom determining the outcome any ADE were identified. The proportion of reviewed incidents with any caregiver-attributed symptom, and the proportion of all incidents with any ADE resulting from a caregiver-attributed symptoms were calculated to determine the impact of caregiver documentation on the study outcome any ADE. Hospital caregivers attributed symptoms to the antidote in 29/111 (26%) of the ethanol-treated and 2/23 (9%) of fomepizole-treated incidents reviewed by the panel. CNS symptoms accounted for the majority of symptoms attributed to ethanol (28/29 incidents) only one of which was a severe symptom (coma). The remainder were minor to moderate CNS depression or "intoxication". Health care professionals attributed one episode of hypotensive bradycardia and one case of transient coma to fomepizole. A total of 21/130 (16%) of ethanol-treated incidents and 2/42 (5%) of fomepizole incidents had an outcome of any ADE in which the only symptom contributing to this outcome was caregiver-attributed to the antidote. The primary outcome of any ADE may be influenced by an excess of caregiver-attributed mild-moderate CNS symptoms in the ethanol treatment group; however, the secondary analysis of severe ADE is unlikely to be affected. Documentation of patient symptoms in the medical chart is a routine component of care. While caregiver opinion might influence observation or documentation of minor or "borderline" symptoms, it is unlikely that a healthcare professional would fail to notice or record a severe symptom such as coma or violent behavior. The large reduction in severe ADE rate in the fomepizole group relative to ethanol (four-fold reduction unadjusted, nearly six-fold reduction adjusted) suggests that the study results 64 Appendix B were robust to the potential influence of caregiver bias in recording ethanol-related CNS symptoms. B.3 Influence of previous reports of antidote adverse reactions on reviewer assessment Previously reported antidote-related adverse reactions were classified according to World Health Organization categories of common (>1%), uncommon (>0.1% and <1%) and rare (<0.1%) with frequency estimated from current toxicology texts and product information.1 Ethanol adverse reactions included information extrapolated from literature on acute intoxication. Table B . l shows the classification of symptoms documented in this study.sample. For each antidote, the proportion of incidents with new onset symptoms during antidote therapy, the proportion of incidents classified as having an antidote A D E by the expert panel and the proportion of total incidents with any ADE were stratified according to whether or not there was at least one symptom previously documented as an adverse reaction. Table B.l Previously reported adverse reactions to ethanol and fomepizole Estimated frequency* Ethanol Fomepizole Common Very common >1/10 Common > 1/100 and < 1/10 "intoxication", minor- severe CNS depression, respiratory depression, minor- severe agitation, headache, cardiovascular symptoms (except cardiac arrest), gastrointestinal symptoms, hypoglycemia, polyuria, phlebitis, hepatic function test abnormalities Uncommon or rare seizure, cardiac arrest, pruritis, Uncommon > 1/1,000 and < 1/100 muscle pain, dystonia, coughing Rare < 1/1000 headache, nausea, vomiting intoxication, minor-moderate CNS depression, minor- moderate agitation coma, seizure, cardiovascular symptoms * Based on World Health Organization categories.' Table includes only symptoms seen in this study sample, not the complete range of previously reported adverse reactions Results are summarized in Table B.2. As expected, the ethanol-treated incidents had a higher proportion of incidents with symptoms previously documented as drug-related toxicity. In ethanol-treated patients, reviewers were more likely to classify common symptoms as antidote ADEs, but were more willing to classify uncommon symptoms as ADEs in fomepizole treated patients. The final outcome of any ADE was defined exclusively by previously reported ADEs in the ethanol group, but almost half the fomepizole incidents with any ADE had symptoms that are considered uncommon or rare. Thus, although there was a potential information asymmetry 65 Appendix B which theoretically made it easier for reviewers to identify ethanol ADEs , the expert panel demonstrated a greater willingness to attribute uncommon or rare adverse reactions to fomepizole than to ethanol. If anything, this would have attenuated, rather than magnified the observed treatment effect. Table B.2 New onset symptoms during antidote therapy: classified by whether or not the symptom is known to be a common adverse reaction associated with the antidote Ethanol Fomepizole Reviewed incidents with new onset symptoms * n/ n incidents (%) n/ n incidents (%) Common 107/111 (96) 16/23 (70) Uncommon/ Rare 4/111 ( 4) 7/23 (30) Proportion of reviewed incidents classified as any ADE ** Common 74/107 (69) 3/16 (19) Uncommon/ Rare 0 / 4 ( 0 ) 2/ 7 (29) Impact on final outcome: Proportion of incidents with any ADE** Common 74/130 (57) 3/42 ( 7) Uncommon/ Rare 0 /130 ( 0) 2/42 ( 5) * Classified by the most common new onset symptom during antidote therapy, as per Table A2.1 definitions. **Classified by most common symptom contributing to the outcome any ADE (any adverse drug event) B.4 Influence of serum ethanol level in identifying incidents with high antidote dose Incidents with new onset symptoms during antidote therapy were scanned for evidence of a higher than usual antidote dose or a high serum ethanol concentration. High dose thresholds were defined as ethanol 10% infusion or equivalent dose >10 mL/kg load, > 2 mL/kg/hr without hemodialysis, >4 mL/kg/hr with hemodialysis or serum ethanol >184 mg/dL (> 40 mmol/L) during ethanol therapy, and fomepizole >15 mg/kg load, >10 mg/kg maintenance and/ or a shorter than the recommended interval between doses. Incidents for which the high dose could be identified only by serum ethanol concentration were distinguished from those with a high administered dose. Among the ethanol-treated incidents with new onset symptoms, a total of 27/ 111 (24%) received a high dose: 19/111 (17%) could only be identified through knowledge of the ethanol serum level while 8/111 (7%) had an administered dose higher than the generally accepted guidelines. Two of the 23 reviewed fomepizole incidents (9%) received a high administered dose. Most of reviewed incidents with high dose antidote were classified as having any ADE (89% of ethanol 66 Appendix B and 100% of fomepizole), including 17/19 (89%) of the incidents identified by ethanol serum concentration alone. The maximum impact of the additional dosage information provided by serum ethanol concentrations was therefore 17/130 (13%) of ethanol incidents. As discussed in section A2.4, below, the blinded review largely eliminated the influence of serum ethanol dose on reviewer decision. The sensitivity analysis restricting the outcome to reviewer 4's outcome evaluation (Chapter 2, Table 2.5) addresses the potential bias introduced through the additional information provided by serum ethanol levels. The treatment effect was maintained in this sensitivity analysis, which suggests the treatment effect was robust to the potential impact of additional information from antidote serum concentrations. B.5 Comparison of panel and blinded reviewer ADE evaluations. As discussed in Chapter 2 methods, the primary outcome of any ADE was evaluated by a panel of three reviewers. In order to test whether the reviewer assessment was influenced by knowledge of treatment group, reviewer 4 independently evaluated antidote ADEs in a subset of poisoning incidents with masked treatment group. Kappa scores and percent agreement were calculated for reviewer pairs and the results of final A D E assessment were compared between the consensus panel and fourth reviewer. A list of case characteristics relevant for the classification of adverse drug events was developed based on published criteria for A D E causality evaluation.2"5 Categories were refined until they adequately captured the full spectrum of information provided by the abstracted poisoning incidents (see Table B.3 for definitions). A l l evaluations of case characteristics were performed by an investigator (KL) who was not a member of the expert review panel. If expert reviewers applied the same criteria for evaluating new onset symptoms associated with ethanol and fomepizole therapy, the case characteristics associated with acceptance or rejection of symptoms as ADEs should be similar for both antidote treatment groups. Likewise, the characteristics should be the same for evaluations by the blinded reviewer and un-blinded consensus panel, with the exception of factors directly affected by the blinding process (e.g. dose). A significant difference in case characteristics between treatment groups or between blinded and un-blinded subsets could identify possible sources of reviewer bias. 67 Appendix B Table B.3 Case characteristics relevant for the causality assessment of potential antidote-related ADEs Case Characteristic Definition Close temporal relationship Close temporal relationship (< 2 hours) between start of antidote ( or dose change or measurement of high serum ethanol level) and symptom onset. Dechallenge or rechallenge Dechallenge: close temporal relationship between antidote discontinuation and symptom resolution. Rechallenge: recurrence of symptom with re-administration of antidote (e.g. antidote re-started after being held, or an additional bolus dose or intermittent infusion) Caregiver attributed Health care provider notes in the hospital chart attribute the symptom to the antidote, or quote the patient attributing the symptom to the antidote, (e.g. "patient is drowsy from the alcohol infusion"). This variable was routinely collected during data extraction and expert reviewers were aware of this information. Previously documented antidote adverse effect Alternate cause: Clinical Condition The symptom is a previously documented adverse effect associated with the antidote, as per standard toxicology references and/ or drug monograph list. Excludes symptoms which are considered to be rare or have a questionable association with the antidote. Presence of one or more conditions or concurrent treatments which could readily explain the presence of the symptom. The sub- categories clinical condition, high toxic alcohol level and other treatment are not mutually exclusive. "Readily" means that the alternate cause is well documented as having the potential to cause the symptom in question. Overall severity of illness (e.g. profound metabolic acidosis) or trajectory of clinical course (e.g. level of consciousness steadily declining before antidote) could account for the symptom High toxic alcohol level Other treatment High antidote dose High serum toxic alcohol level (threshold > 50 mmol/L based on clinical experience), or evidence of a rising serum toxic alcohol level around the time of symptom onset Another treatment could be responsible for the symptom. E.g. sedative drugs. Antidote dose (mg/kg or infusion rate) is higher than usual recommended dose or serum ethanol level is substantially greater than recommended therapeutic range >184 mg/dL (> 40 mmol/L); this threshold was selected because it would usually produce significant intoxication in the average adult) Missing Data Important information is regarding time course of events, doses, concurrent treatments, etc. is missing from the abstracted hospital record. (Does NOT apply to masked information in the blinded review) 68 Appendix B Table B.4 shows results of inter-rater agreement between reviewer pairs, assessed on independent A D E evaluations. The kappa scores were not significantly different (confidence intervals overlap) between any of the reviewer pairs, but there was greater variability between the reviewer 4 and the panelists (reviewers 1-3) than between the three panel members. Figure B . l shows the results of the consensus panel and reviewer 4 evaluation of antidote ADEs in the blinded and un-blinded subsets of incidents. In the un-blinded subset of 59 incidents, the panel assessed 61% (49, 73) as having any A D E , while the fourth reviewer found any A D E in 46% (33, 58). In the blinded subset of 76 incidents, the panel found 55% (44, 66) with any A D E versus the 29% (19, 40) by the blinded reviewer. There was a greater discrepancy between the panel and fourth reviewer for the blinded subset. Table B.4 Inter-rater agreement between expert reviewers on independent evaluation of Any ADE Three main panelists n=135 incidents reviewer number % agreement kappa (95%CI) 1 &3 76% 0.49 (0.32, 0.66) 1 &2 76% 0.47 (0.30,0.63) 2&3 73% 0.41 (0.25, 0.57) Three main panelists with blinded reviewer, UN BLINDED subset n=58 incidents reviewer number 1&4 2&4 3&4 % agreement 78% 64% 74% kappa (95%CI) 0.55 (0.30,0.81) 0.29(0.08,0.50) 0.49(0.24,0.73) Three main panelists with blinded reviewer, BLINDED subset n= 77 incidents reviewer number 1&4 2&4 3&4 % agreement kappa (95%C1) 61% 0.28 (0.10,0.47) 57% 0.26 (0.10, 0.42) 70% 0.44 (0.24, 0.63) Figure B . l Comparison of consensus panel and reviewer 4 evaluation of antidote ADEs in the blinded and unblinded subsets of incidents Reviewer # 4, blinded Reviewer # 4, un-blinded ADE yes ADE no ADE yes ADE no Consensus ADE yes 22 21 43 Consensus ADE yes 25 11 panel ADE no 1 32 33 panel ADE no 2 21 23 53 76 27 32 36 23 59 Case characteristics associated with expert panel evaluation of ADEs are summarized in Table A2.5. Incidents with new onset symptoms during antidote therapy were more likely to be evaluated as having any ADE i f there was a close temporal relationship between symptom onset 69 Appendix B and antidote start (or dose change), i f the symptom(s) had been previously documented as an antidote related adverse effect, i f a hospital caregiver had attributed the symptom to the antidote, or i f the dose or serum level of antidote was higher than the usual recommended dose. De-challenge and re-challenge did strongly influence the decision to assess a symptom as an A D E when present; however, these characteristics were present in very few incidents (8/135, 6%). Incidents with new onset symptoms assessed as having no ADE were significantly more likely to have an alternate cause for the symptom, particularly those related to the clinical condition of the patient or other treatments. Few incidents had missing data, but incomplete information was also associated with assessment of no ADE. Table B.5 Panel decision regarding any ADE, all reviewed incidents: characteristics associated with yes/ no decision Number (%) of incidents Fishers exact Case Characteristic ADE "yes" ADE "no" p value n= 79 n=55 Close temporal relationship 49 (62) 8 (15) <0.001 Dechallenge or rechallenge 8 (10) 0 ( 0) 0.021 Caregiver Attributed 26 (33) 5 ( 9) 0.002 Previously documented ADE 77 (97) 46 (84) 0.007 Any known alternate cause 34 (43) 47 (85) <0.001 High antidote dose 26 (33) 3 ( 5) <0.001 Missing Data 3 ( 4) 8 (15) 0.050 There were no significant differences in the case characteristics between the fomepizole and ethanol groups for symptoms rejected as antidote ADEs (Table B.6, top panel). Most incidents had obvious confounders which presented alternate causes for the symptom and minimal evidence to implicate the antidote. Case characteristics were also similar between treatment groups for the symptoms accepted as antidote ADEs (Table B.6, bottom panel), with the exception of previously documented adverse reactions (lower in the fomepizole group) and alternate causes (higher in the fomepizole group). These results suggests that the expert panel applied the same criteria for A D E classification when reviewing either antidote, but may have compensated for the relatively limited use experience with fomepizole by showing a greater willingness to accept a previously unreported A D E or consider an A D E in the presence of a confounder. 70 Appendix B Table B.6 Case characteristics associated with outcome of any ADE: consensus panel and blinded reviewer evaluations Number (%) of incidents Fisher's exact Case Characteristic ADE "yes" ADE "no" p value 3 member consensus panel n=43 n = 33 Close temporal relationship 25 (58) 2 ( 6) <0.001 Dechallenge or rechallenge 5 (12) 0 ( 0 ) 0.065 Caregiver Attributed 16 (37) 3 ( 9) 0.007 Previously documented ADE 41 (95) 28 (85) 0.229 Any known alternate cause 23 (53) 27 (82) 0.014 High antidote dose • 10 (23) 2 ( 6) 0.058 Missing Data 1 ( 2) 1 ( 3) 0.999 Reviewer 4, blinded to antidote n=23 n= 53 Close temporal relationship 16 (70) 11(21) <0.001 Dechallenge or rechallenge 4 (17) 1 ( 2) 0.027 Caregiver Attributed 14 (61) 5 ( 9) <0.001 Previously documented ADE 22 (96) 47 (89) 0.668 Any known alternate cause 4 (17) 46 (87) <0.001 High antidote dose 4 (17) 8 (15) 0.999 Missing Data 0 ( 0) 2 (4) 0.999 Table B.7 shows that, with the exception of antidote dose (masked in the blinded review), there did not appear to be any differences in the characteristics of cases accepted or rejected as antidote ADEs by the consensus panel and blinded reviewer 4. The panel was somewhat more willing to accept ADEs despite the presence of confounders than the blinded reviewer. Knowledge of antidote dose may have permitted the consensus panel to weigh the relative importance of alternate cause of the symptom against the effects of a high antidote dose. Note that there were very few incidents with dechallenge/ rechallenge or missing data, therefore statistical comparison of these factors is not meaningful. The similarities between the panel and blinded reviewer suggest that knowledge of antidote type did not lead to a systematic difference in the way that reviewers classified antidote ADEs. Limitations of this evaluation include the small number of blinded incidents, the use of only one blinded reviewer rather than a panel of blinded experts, and the inability to "blind" antidote treatment group without the loss of important information necessary for A D E evaluation. 71 Appendix B Table B.7 Panel and blinded reviewer decision regarding any ADE, subset of 78 reviewed incidents Number (%) of incidents Fishers exact Case Characteristic ADE "yes" ADE "no" p value Panel evaluations (antidote type known) n= 43 n= 33 Close temporal relationship 25 (58) 2 ( 6) <0.001 Dechallenge or rechallenge 5 (12) 0 ( 0) 0.065 Caregiver Attributed 16 (37) 3 ( 9) 0.007 Previously documented ADE 41 (95) 27 (82) 0.071 Any known alternate cause 23 (53) . 27 (82) 0.014 High antidote dose 11 (26) 2 ( 6) 0.032 Missing Data 1 ( 2) 3 ( 9) 0.031 4th reviewer evaluations (antidote type BLINDED) n= 22 n= 54 Close temporal relationship 15 (68) 12 (22) <0.001 Dechallenge or rechallenge 4 (18) 1 ( 2) 0.023 Caregiver Attributed 13 (59) 6 (11) <0.001 Previously documented ADE 21 (95) 47 (87) 0.425 Any known alternate cause 4 (18) 46 (85) <0.001 High antidote dose 5 (23) 8 (15) 0.54 Missing Data 0 ( 0) 4 (17) 0.317 B.6 Conclusion In conclusion, there was some evidence that caregiver bias may have inflated the documentation of minor to moderate CNS symptoms in the ethanol group; however, the secondary outcome of any severe A D E showed a strong treatment effect (Chapter 2 Table 2.4) and should be robust to caregiver opinion. Knowledge of high serum ethanol concentrations may have influenced reviewers to accept more ethanol-related symptoms as ADEs, but the sensitivity analysis incorporating the blinded review maintained a strong treatment effect despite loss of this information (Chapter 2 Table 2.5). Presence of a previously reported adverse reaction did not appear to influence reviewers to accept ethanol ADEs. It is possible that reviewers may have been biased to accept fomepizole ADEs with less supporting evidence than they required for ethanol ADEs , which would have attenuated the treatment effect. It appears that any biases influencing the A D E rates for each antidote are relatively minor, and had a limited influence over the study outcome. 72 Appendix C Appendix C: Serious Antidote-Related Adverse Drug Events (ADEs) C . l Definition serious antidote-related ADEs The World Health Organization definition of a serious adverse drug event or reaction is: any untoward medical occurrence that at any dose:6 • Requires inpatient hospitalization or prolongation of existing hospitalization • Results in persistent or significant disability/incapacity • Is life-threatening • Results in death or, as per Health Canada, requires a significant intervention to prevent one of the above.7 In this study, criteria for serious ethanol or fomepizole-related adverse drug events (ADEs) are: 1) Serious central nervous system (CNS) A D E : • CNS depression with documented loss or airway protection, apneic episodes, respiratory arrest or an intervention suggestive of compromised airway/ breathing e.g. intubation, insertion of oral airway, maintaining jaw thrust, administering oxygen. • CNS depression resulting in prolongation of hospital stay, i.e. clinically significant antidote-related CNS depression beyond recovery from the acute poisoning and associated complications (toxin eliminated, acid base status returned to normal, not dialysis dependent, no complications such as aspiration pneumonia). • Severe agitation resulting in a potentially life threatening injury or endangerment (e.g. the patient is injured as a result of the agitation or intervention to restrain the patient, the patient escapes from hospital while toxic alcohol level is potentially lethal). The following are not serious CNS ADEs: • CNS depression without documented compromise of airway/ breathing and which does not prolong the hospital stay. Rationale: transient CNS depression (including coma) alone is not potentially life threatening in a hospitalized patient. • Agitation/ combative/ uncooperative behaviour, no matter how disruptive or annoying, which does not result in physical harm to the patient. 73 Appendix C 2) Serious cardiovascular A D E : • Cardiac arrest • Clinically significant increase or decrease in blood pressure or heart rate or dysrhythmia which requires intervention such as administration of pressors, antihypertensives, atropine, plasma expanders, cardioversion, or resuscitative measures. 3) Serious metabolic A D E : • Severe hypoglycemia: blood glucose < 30 mg/dL (< 1.7 mmol/L) • Antidote failure which results in clinically significant worsening of toxicity C.2 Case summaries of serious antidote-related adverse drug events (ADEs) Abbreviations used in case summaries: A P A C H E II acute physiologic and chronic health evaluation II score BP blood pressure GCS Glasgow coma scale H C 0 3 " serum bicarbonate H R heart rate ICU intensive care unit M A P mean arterial pressure n/a not available R R respiratory rate Osmole gap (OG) was calculated using the equation: OG= measured serum osmolality - [(2*sodium)+(urea)+(glucose)+(ethanol)] with all variables in mmol/L. C.2.1 serious ethanol-related ADEs ID 13 Antidote: ethanol ADE type: CNS Patient Information: 33-year-old female ingested ~ 350 mL of radiator antifreeze (ethylene glycol) and presented to hospital approximately 7 hours after exposure. Baseline (pre-antidote): GCS 15 (drowsy), H R 94, BP 100/69, M A P 79, RR 20, peak serum ethylene glycol 397 mg/dL (64 mmol/L), lowest H C 0 3 " 11 mmol/L, A P A C H E II score 4. Description: Loading dose 320 mL 10% IV ethanol infused over 20 minutes. By the end of the loading dose, pre-treatment drowsiness had deteriorated to coma (GCS 3). No other sedatives given. Intervention: Rapid sequence intubation & mechanical ventilation 45 minutes after onset of coma. Resolution: Continued ethanol, hemodialysis. Required sedation while intubated. Extubated after dialysis. Survived without sequelae. 74 Appendix C ID 74 Antidote: ethanol ADE type: CNS Patient Information: 37-year-old male ingested "1/2 bottle" of radiator antifreeze (ethylene glycol) and presented to hospital approximately 9.5 hours after exposure. Baseline (pre-antidote): GCS 14, HR 115, BP 142/80, M A P 101, RR 26, peak ethylene glycol 682 mg/dL (110 mmol/L), lowest H C C V 6 mmol/L, A P A C H E II score 8. Description: Initially calm but became agitated, therefore given 1 mg lorazepam 1.5 hours before antidote started. Loading dose of 420 mL 10% IV ethanol infused over 30 minutes. 15 minutes after infusion began, patient was comatose GCS 3, no response to nail bed pressure; mouth breathing. Intervention: Given oxygen by face mask, but not intubated. Resolution: Coma lightened to GCS 8 within 2 hours. Continued ethanol, hemodialysis, intermittent sedation for agitation. Developed acute renal failure with eventual recovery. ID 123 Antidote: ethanol ADE type: CNS Patient Information: 51-year-old male weighing 75 kg unintentionally ingested 1 mouthful of radiator antifreeze (ethylene glycol) and presented to hospital approximately 45 minutes after exposure. Baseline (pre-antidote): GCS 15, H R n/a, BP 120/74, M A P 89, RR n/a, peak ethylene glycol 12 mg/dL (1.9 mmol/L), lowest H C C V 39 mmol/L, A P A C H E II score 4. Description: Minor, unintentional exposure with prompt presentation to hospital. IV ethanol 10% infusion 681 mL load over 1 hour then 150 mL/hr for 13 hours. Symptoms of intoxication first documented 2.5 hours after antidote start, deteriorated to coma with intermittent apneic spells and decreased respiratory rate, drop in oxygen saturation to 70%. Symptoms associated with rising serum ethanol. Maximum recorded 193 mg/dL (42 mmol/L). No other sedatives. Intervention: Given oxygen, admitted to ICU, not intubated. Resolution: Recovered without sequelae after ethanol discontinued. ID 134 Antidote: ethanol ADE type: CNS Patient Information: 53-year-old male found down on sidewalk. A n antifreeze (ethylene glycol) container was in the vicinity but, in retrospect, he likely had not ingested any. Baseline (pre-antidote): GCS 4, H R 88, BP 136/88, M A P 104, RR 6, peak osmole gap 16 (ethylene glycol level not measured), blood gases pre-intubation pH 7.29, pC02 58, HCO3" 27, A P A C H E II score 16. Description: On admission, the patient was comatose with respiratory depression associated with serum ethanol of 4001 mg/dL (87 mmol/L). He was intubated and ventilated, admitted to ; ICU and treated with ethanol IV 10% 83 mL/hr infusion for 17 hours for presumption of toxic alcohol exposure. Serum ethanol level remained highly elevated during IV ethanol treatment. Final diagnosis was ethanol intoxication. Intervention: Required intubation and mechanical ventilation throughout IV ethanol administration. Considered a serious A D E due to prolongation of hospital stay: Administration of ethanol infusion resulted in persistently high serum ethanol: serum ethanol 286 mg/dL, 258 mg/ dL ( 62 mmol/L, 56 mmol/L) during therapy. The patient required continued support with mechanical ventilation. 75 Appendix C Resolution: The patient was extubated 3 hours after IV ethanol discontinuation and awoke promptly. He denied toxic alcohol exposure. ID 138 Antidote: ethanol ADE type: CNS Patient Information: 20-year-old female weighing 57 kg ingested approximately 450 mL of windshield washer antifreeze (methanol) and presented to hospital approximately 4 hours after exposure. Baseline (pre-antidote): GCS 15 (drowsy), H R 118, BP 110/75, M A P 87, RR 18, peak methanol 109 mg/dL (34 mmol/L), lowest pH 7.25, A P A C H E II score 11. Description: 750 mL of 5% ethanol IV of over 1 hour, then 100 mL/hr . Shortly after completion of the loading dose, patient became comatose (GCS 3) with shallow respirations. No other sedatives given. Intervention: Oral airway inserted & tolerated x 5 minutes; maintenance infusion was held for 30 minutes, then resumed at the same dose. Resolution: Level of consciousness began to lighten within 1 hour, fully awake.within 4 hours. Survived without sequelae. Lengthy hospital stay due to chronic underlying health problem. ID 160 Antidote: ethanol ADE type: CNS Patient Information: 52-year-old male ingested approximately 1000 mL of radiator antifreeze (ethylene glycol) and presented to hospital approximately 1 hour after exposure. Baseline (pre-antidote): GCS 15 ("intoxicated", due to high ethylene glycol level but ethanol negative), H R 80, BP 130/78, M A P 95, R R 20, peak ethylene glycol 701 mg/dL (113 mmol/L), no pre-treatment measurement of pH or serum bicarbonate, A P A C H E II score 2 Description: Given oral ethanol equivalent to 500 mL of 40% ethanol (liquor), ingested quickly. IV infusion 10% ethanol 50 mL/hr was started 1 hour later. Increasingly drowsy after oral loading dose, progressing to stupor, loss of airway protection, periods of apnea during transport between hospitals (no other sedation had been given at this point). Also intermittently combative and uncooperative. Serum ethanol level was 276 mg/dL (60 mmol/L) at the destination hospital. Intervention: Required stimulation (rubbing jaw) to breath. Oral airway inserted, tolerated for 4 minutes. Resolution: Level of consciousness improved slowly, fully awake within ~ 12 hours. Treated with hemodialysis & ethanol. Recovered without sequelae. ID 162 Antidote: ethanol ADE type: CNS Patient Information: 53-year-old male weighing 62 kg. Likely no toxic alcohol exposure. Baseline (pre-antidote): GCS 15, H R 106, BP 133/68, M A P 90, RR 20, peak serum methanol 11 mg/dL (3.6 mmol/L); possibly a lab artifact, lowest pH 7.37 mmol/L, HCO3" 10 mmol/L, A P A C H E II score 2. Serum ethanol was not measured on admission, but was 32 mg/dL (7 mmol/L) six hours after arrival at ED; pre-antidote. Description: alcoholic patient with heavy consumption of ethanol beverages 24 hr prior to admission, vomiting 1-2 days. High osmole gap of 52 (NOT corrected for unmeasured ethanol) and anion gap (29) raised suspicion of toxic alcohol poisoning, but the patient denied exposure. Probable alcoholic ketoacidosis. 1000 mL loading dose of 10% ethanol IV followed by 100 mL/hr was tolerated, but serum ethanol was undetectable after 3 hr of therapy. The patient was given a second 1000 mL loading dose, maintenance infusion was increased to 200 mL/hr. CNS depression noted 2 hours later; 76 Appendix C progressive decline to coma (GCS 6) and loss of gag reflex associated with rising ethanol level first 313 mg/dL (68 mmol/L) then 507 mg/dL (110 mmol/L). Intervention: transfer from medical ward to ICU, 45 degree elevation of head of bed, not intubated Resolution: Began to wake up 3 hours after ethanol infusion was discontinued. Peak "methanol" level was below treatment threshold, therefore 2 day ethanol infusion also prolonged hospital stay. ID 173 Antidote: ethanol ADE type: CNS Patient Information: 43-year-old male weighing 95 kg ingested 1000 mL (allegedly) of radiator antifreeze (ethylene glycol) and presented to hospital approximately 1 hour after exposure. Baseline (pre-antidote): GCS 15, H R 48, BP 132/82, M A P 99, R R , peak ethylene glycol 106 mg/dL (17 mmol/L), lowest pre-treatment pH 7.14, H C 0 3 " 6 mmol/L, A P A C H E II score 9 Description: History of exposure unknown therefore delayed diagnosis. Patient became more acidotic during hospital stay. Very agitated/ combative on admission; 10 mg of haloperidol given 6 hours before antidote started (remained awake, no other sedatives). Antidote initiated 7 hours after admission: 500 mL IV loading dose of 10% ethanol. Coma onset 15 minutes after start of antidote loading dose. Intervention: Rapid sequence intubation 60 minutes after onset of coma. Resolution: Remained intubated with sedation until toxin clearing hemodialysis and antidote completed, then extubated. Required additional dialysis for renal support, with eventual recovery. ID 28 Antidote: ethanol ADE type: Cardiovascular Patient Information: 57-year-old male weighing 75 kg ingested an unknown quantity of radiator antifreeze (ethylene glycol) and presented to hospital approximately 7 hours after exposure. Also co-ingested ethanol. Baseline (pre-antidote): GCS 6, HR 125, BP 105/65, M A P 78, R R 28, peak ethylene glycol 211 mg/dL (34 mmol/L), serum ethanol 147 mg/dL (32 mmol/L), lowest H C C V 7.39 mmol/L, H C 0 3 " 16 mmol/L, A P A C H E II score 17 Description: Found unresponsive, admission serum ethanol level was measured 2.5 hr before antidote started. BP stable pre-antidote. Received 2 L IV saline before antidote. IV ethanol 10% infusion started at 245 mL/hr (3.3 mL/kg/hr); there was no loading dose, presumably because of high serum ethanol on admission. BP dropped sharply within 30 minutes after the start of the ethanol infusion (BP 76/56, M A P 62) and continued to decline ( M A P 55) before dialysis started. Further decline (BP 80/30, M A P 47) during dialysis. Intervention: Plasma expander before and during dialysis, dopamine and norepinephrine started during dialysis. Resolution: Hypotension corrected with pressors; dose slowly tapered over 36 hours and d/c at same time as antidote. Recovered without sequelae. 77 Appendix C ID 62 Antidote: ethanol ADE type: Cardiovascular Patient Information: 63-year-old male weighing 80 kg was on an ethanol drinking binge for 2-3 days prior to admission, denied toxic alcohol exposure. Baseline (pre-antidote): GCS 15, H R 112, BP 120/80 , M A P 93, RR 36, admission ethanol 226 mg/dL (49 mmol/L); both methanol and ethylene glycol levels negative, lowest pH 6.66, HCO3" 3 mmol/L, A P A C H E II score 18 Description: Alcoholic patient, brought to hospital due to coffee ground emesis. Minimal monitoring in the first few hours in hospital. Suspicion of possible toxic alcohol exposure due to metabolic acidosis with high anion gap, renal impairment. Serum ethanol declined to 28 mg/dL (6 mmol/L) prior to treatment. BP was stable, M A P in 90's until ~ 1 hour prior to the start of antidote, when it dropped to 100/45 (MAP 63). A 545 mL loading dose of 10% ethanol was initiated, set to infuse over 30 minutes. Ten minutes after the start of antidote, BP dropped to 83/36 ( M A P 52), H R 61, followed by ventricular tachycardia, cardiac arrest near the end of the loading dose. After resuscitation, the antidote loading dose was re-started and the BP fell again to 73/20 ( M A P 38) within 20 minutes. Intervention: Initial resuscitation with atropine 1 mg, cardioversion, epinephrine, lidocaine, intubation and ventilation. Second drop in BP (after antidote restarted) was treated with dopamine and norepinephrine infusion. Resolution: Sufficiently stabilized to tolerate hemodialysis and completion of antidote therapy. Remained critically i l l . Died on the third hospital day. Final diagnosis alcoholic ketoacidosis, sepsis, multi-organ failure. ID 82 Antidote: ethanol ADE type: Cardiovascular Patient Information: 52-year-old male weighing 100 kg ingested an unknown quantity of radiator antifreeze (ethylene glycol) and presented to hospital approximately 7.5 hours after exposure. Baseline (pre-antidote): GCS 3, HR 70, BP 215/115 , M A P 148, R R , peak ethylene glycol 552 mg/dL (89 mmol/L), peak osmole gap 134, lowest pH 7.11, A P A C H E II score 21 Description: Delayed diagnosis due to no history of exposure. Awake (GCS 15), but behaviour "odd" on admission. Rapidly deteriorated; comatose with apneic spells, intubated, in Initially hypertensive (known underlying medical condition) M A P 148-132. Treated with one dose of 5 mg IV metoprolol, given 2.5 hours before start of antidote and 17 hours before the antidote A D E . Treated with 750 mL 10% IV ethanol loading dose followed by 200 mL/ hr. Maintenance infusion rate progressively increased from 250-400 mL/hr during hemodialysis and serum ethanol level rose steadily. Approx 15 hours after the start of antidote and -12 hours after the start of dialysis, the patient became hypotensive 60/32 ( M A P 41). This A D E coincided with a peak serum ethanol 48 mmol/L measured shortly before the lowest recorded BP. At the time of A D E onset, osmole gap had declined to 34 and acidosis had resolved. Intervention: BP corrected within 45 minutes after initiation of dopamine infusion. Ethanol infusion rate was decreased. Resolution: Dopamine continued for 12 hours. Developed renal failure with eventual recovery. 78 Appendix C C.2.2 Serious fomepizoie-reiated ADEs ID 44 Antidote: fomepizole ADE type: Cardiovascular Patient Information: 59-year-old male weighing 53 kg ingested an unknown quantity of radiator antifreeze (ethylene glycol) which he had mistaken for juice and mixed with ethanol beverage. Presented to hospital approximately 10 hours after exposure. Baseline (pre-antidote): GCS 7, H R 70, BP 160/100 , M A P 120, R R 16, peak ethylene glycol 118 mg/dL (19 mmol/L), ethanol undetectable, lowest pH 7.17, A P A C H E II score 22. Description: Delayed diagnosis due to no history of exposure. Pre-treatment vital signs were stable at HR 88-120, M A P 108-123. Hemodialysis and the fomepizole (19 mg/kg) loading dose began within 5 minutes of each other. Immediately after completion of the 30 minute fomepizole infusion, the patient developed hypotensive bradycardia (HR 29, M A P 46) which corrected to baseline with 1 mg atropine. Hemodialysis continued uneventfully until the second fomepizole dose (10 mg/kg) 4 hours later. Upon completion of the infusion H R transiently dropped to 48, M A P to 67. Intervention: first episode responded promptly tol mg atropine, plasma expander; patient was intubated and ventilated during resuscitation. Second episode resolved rapidly; small bolus of plasma expander (primarily given for decreased urine output). Resolution: Third dose of fomepizole (10 mg/kg) was given post-dialysis and was well tolerated. Patient developed acute renal failure but eventually recovered. 79 Appendix D Appendix D: Cox Proportional Hazard Regression Model D.l Covariate selection A l l potential predictor variables were systematically tested for an association with the primary outcome of any ADE or secondary outcome severe ADE (Table D. 1). Table D.l Univariate analyses of potential predictor variables in the Cox regression model Covariate (reference category) RR from exponentiated regression coefficient ( 0 9 5 % ) *Any ADE~ •Severe ADE~ Main predictor variable model model Antidote: fomepizole (ethanol) 0.15 (0.06, 0.37) 0.22 (0.05, 0.91) Patient and Exposure Age: (continuous variable, years= 0) 0.995 (0.98,1.01) 1.02 (0.99, 1.04) Age category: > 50 years (<50 years) 0.89 (0.52, 1.52) 2.16(1.01,4.61) Sex: male (female) 0.82 (0.52, 1.31) 1.05 (0.46,2.4) Ethanol abuse history: yes (no) 0.86 (0.55, 1.34) 1.33 (0.63,2.8) Toxin type: ethylene glycol (methanol or not toxic alcohol) 0.65 (0.41, 1.02) 1.38 (0.65, 2.91) Toxin exposure reason: unintentional (intentional) 1.3(0.73,2.19) 1.02 (0.39,2.67) Co-ingestants: present (absent) Ethanol 1.13 (0.69, 1.80) 2.55 (1.21, 5.35) Pharmaceutical 0.98 (0.53, 1.81) 1.23 (0.47,3.25) Non-pharmaceutical 0.71 (0.29, 1.76) 0.43 (0.09,3.17) Presentation delay: (continuous variable, hours =0) 0.99(0.97, 1.01) 0.97(0.92,1.01) Peak serum methanol or ethylene glycol concentration: above 0.75 (0.45, 1.25) 0.78 (0.33, 1.84) treatment threshold (below threshold or not measured) Baseline Clinical Condition (pre antidote) summary score APACHE II score: (integer score, lowest=0) 0.94 (0.91, 0.97) 0.99 (0.95, 1.03) APACHE II three category 0-2 (reference) >2-8 0.57 (0.35, 0.95) 0.51 (0.20, 1.30) >8-30 0.31 (0.17, 0.54) 0.69(0.29, 1.63) Baseline Clinical Condition(pre antidote)- by system Cardiovascular symptoms: present (absent) 0.82 (0.50, 1.29) 0.88 (0.40, 1.94) CNS depression: present (absent) 0.97 (0.2, 1.53) 0.75 (0.36, 1.57) CNS -agitation, excluding seizures: present (absent) 1.00 (0.58, 1.71) 2.53 (1.20, 5.41) Headache: present (absent) . 0.67(0.27,1.66) n/a** Gastrointestinal symptoms: present (absent) 0.775 (0.78, 2.13) 1.31 (0.58,2.98) Hepatic abnormality: present (absent) 1.4 (0.84, 2.35) 1.34 (0.57,3.16) Metabolic Acidosis: present (absent) 0.91 (0.57, 1.44) 0.70(0.33, 1.45) Visual impairment: present (absent) 1.24 (0.67,2.29) 1.01 (0.35,2.91) Renal impairment: present (absent) 1.99 (0.92,4.34) 2.06 (0.62, 6.84) Other Treatments Hemodialysis: at least one session (none) 0.70 (0.44, 1.14) 0.70 (0.32,1.56) Sedation pre-antidote: at least one dose (none) 0.57 (0.34, 0.95) 0.94(0.68, 1.30) Intubation pre- antidote: yes (no) 0.28 (0.14, 0.59) 0.40 (0.12, 1.31) Physical restraint pre-antidote: yes (no) 1.05 (0.52, 2.1) ' 2.83(1.2,6.67) Cardiac med pre-antidote: at least one dose (none) n/a** 0.32 (0.42, 4.56) * Abbreviations: any or severe ADE antidote-related adverse drug event. RR: Rate ratio. CI95o/o: 95% confidence interval. APACHE II: acute physiologic and chronic health evaluation. **n/a= unable to test; convergence failure. 80 Appendix D Potential covariates were systematically tested in the model i f they showed clear separation between strata in a Kaplan Meier curve of cumulative A D E rate (not shown) or i f the confidence interval around the rate ratio obtained from a univariate analysis did not include 1.0 (Table D . l ) . The final models, coefficients and overall model fit are discussed in Chapter 2. D.2 Model testing- tests for proportionality The fundamental assumption of the Cox proportional hazard model is that proportionality between predictor variable categories remains constant over time. This assumption was tested for each covariate through visual inspection of univariate Kaplan Meier plots and plots of Schoenfeld residuals, and through testing for significant slope of the Schoenfeld residual plots. The Kaplan Meier curves for the primary predictor variables any A D E or severe A D E are shown in Chapter 2, Figures 2.1 and 2.2. D.2.1 Any ADE model Survival curves for all categories of A P A C H E II (Figure D . l ) and Antidote treatment category (Chapter 2, Figure 2.1) appear proportional over time. The lines do not converge or cross. Plots of Schoenfeld residuals (Figure D.2) are approximately linear, which supports the assumption of proportionality over time. Table D.2 tests for significant slope in these plots. A l l p-values are >0.05, which suggests there is no significant association between each coefficient and time (ie supports the assumption of proportionality over time). variable rho Chi squared p value Antidote 0.00928 0.00663 0.935 APACHE 2-8 -0.06350 0.31767 0.573 APACHE 8-30 -0.15215 1.79193 0.181 GLOBAL NA 1.79330 0.616 81 Appendix D Figure D.l Survival curve for adjustment variable APACHE II score (3 categories) any ADE model A P A C H E 0-2 A P A C H E 2-8 A P A C H E 8-30 f-• 4 2-8 8-30 exp(coef) exp(-coef) lower95 upper95 0.573 1.74 0.346 0.950 0.305 3.27 0.171 0.544 40 60 80 100 Elapsed time (hours) after start of antidote 82 Appendix D Figure D.2 Scatter plots of Schoenfeld residuals, any ADE model Schoenfeld residuals Any ADE, APACHE 2-8 o c o o o O O O O c p 3 O o O O O Q ° OO C 0.27 0.55 0.79 1.5 2.4 5.3 18 33 Time Top panel: APACHE score category 2-Schoenfeld residuals Any ADE, APACHE 8-30 CO O O CO O CO o o o o o c o o o o o o o o o O O 0 3<fe o o o 0.27 0.55 0.79 1.5 2.4 5.3 Time 18 33 Center panel: APACHE score category 8-30 Schoenfeld residuals Any ADE, Tx Category 113 U 10 Bottom panel: Antidote treatment category 83 Appendix D D.2.2 Severe ADE model Survival curves for all categories of pre-treatment agitation (Figure D.3, D.4), pre-hospital ingestion of ethanol (Figure D.4) and antidote treatment category (Chapter 2, Figure 2.2) appear proportional over time. The lines do not converge or cross. Plots of Schoenfeld residuals (Figure D.5) are approximately linear, which supports the assumption of proportionality over time. Note that the plot for the covariate pre-treatment agitation does show a trend towards a downward slope, but this is not sufficiently pronounced to invalidate model assumptions. Table D.3 tests for significant slope in these plots. A l l p-values are >0.05, which suggests there is no significant association between each coefficient and time. Conclusion: both models satisfy the proportionality requirements of a Cox proportional hazard model. Figure D.3 Survival curves for adjustment variable pre-treatment agitation, severe ADE model co d co d d CM d o d 40 60 80 Elapsed time (hours) after start of antidote 84 Appendix D Figure D.4 Survival curves for adjustment variable ingestion of ethanol pre-hospital, severe ADE model 40 60 Elapsed time (hours) after start of antidote Table D.3 Test for significant slope in Schoenfeld residuals plot, severe ADE model variable rho Chi squared p value Antidote 0.0335 0.0310 0.860 Pre- treatment Agitation -0.2316 1.5074 0.220 Ethanol co-ingestion 0.0462 0.0599 0.807 (pre- hospital) GLOBAL N A 1.5454 0.672 85 Appendix D Figure D.5 Scatter plots of Schoenfeld residuals, severe ADE model Schoenfeld residuals Severe ADE TxCat Top panel: Antidote treatment category Schoenfeld residuals Severe ADE Agitation o o Q -O o 1 I [ 1 1 1 1 1 1 0.26 0.57 1.2 2.5 3.4 4.7 14 22 Center panel: Pre-treatment agitation Schoenfeld residuals Severe ADE EtOH pre HCF " ~ i r -b-... o , .o - 'o ~1 1 0.26 0.57 1.2 2.5 3.4 . 4.7 14 22 Bottom panel: Pre-hospital ingestion of ethanol 86 Appendix E Certificates Of Ethical Approval References: Appendices References 1. W H O - U M C . Frequency of adverse drug reactions. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Available at: http://www.who-umc.org/DynPage.aspx?id=22684. Accessed Feb 22, 2007. 2. W H O - U M C . Causality Assessment of Suspected Adverse Reactions. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Available at: http://www.who-umc.org/DynPage.aspx?id=22682. Accessed Feb 16, 2007. 3. Naranjo C A , Busto U , Sellers E M , et al. A Method for Estimating the Probability of Adverse Drug-Reactions. Clin Pharmacol Ther. 1981 ;30(2):239-245. 4. Nebeker J, Barach P, Samore M . Clarifying adverse drug events: a clinician's guide to . terminology, documentation and reporting. Ann Intern Med. 2004;140:795-801. 5. Morimoto T, Gandhi T, Seger A , et al. Adverse drug events and medication errors: detection and classification methods. Quality and Safety in Health Care. 2004; 13:306-314. 6. W H O _ U M C . Definitions: Serious adverse reaction. World Health Organization Collaborating Centre for International Drug Monitoring: Uppsala Monitoring Centre. Apr i l 13, 2006. Available at: http://www.who-umc.org/DynPage.aspx?id=22680. Accessed Jan 11, 2007. 7. C A D R M P . Canadian adverse drug reaction monitoring program ( C A D R M P ) guidelines for the voluntary reporting of suspected adverse reactions to health products by health professionals and consumers. Health Canada: Marketed Health Products Directorate. 2006-05. Available at: http://www.hc-sc.gc.ca/dhp-mps/alt_formats/hpfb-dgpsa/pdf/medeff/ar-ei_guide-ldir_e.pdf. Accessed March 3, 2007. 93 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0100871/manifest

Comment

Related Items