Government of CanadaPublic Health Agency of Canada / Agence de santé publique du Canada
   
Skip all navigation -accesskey z Skip to sidemenu -accesskey x Skip to main menu -accesskey m  
Français Contact Us Help Search Canada Site
PHAC Home Centres Publications Guidelines A-Z Index
Child Health Adult Health Seniors Health Surveillance Health Canada
   



Volume 19, No.4 - 2000

 [Table of Contents] 

  

Public Health Agency of Canada (PHAC)

Surveillance of Drug Overdose Deaths Using Medical Examiner Data

Christiane Poulin, Jonathan Stein and John Butt

 

Abstract

This paper describes the epidemiology of drug overdose deaths investigated by the medical examiner in one of the cities participating in the Canadian Community Epidemiology Network on Drug Use and assesses the quality of the information obtained from medical examiner charts with respect to drug overdose deaths, for surveillance purposes. Information was abstracted from medical examiner charts of all deaths involving drugs from 1993 to 1995 in Halifax, Nova Scotia. During these three years, 636 deaths from all causes were investigated by the medical examiner. Of the 42 overdose deaths, 47.6% were suicides. Ethanol was detected in 47.8% of overdose deaths, and 61.9% of all overdose deaths involved psychotropic medications. Two deaths were attributed to an illicit drug (cocaine). An independent review performed by a toxicologist and a medical examiner revealed poor overall agreement concerning overdose as a cause of death (Kappa coefficient: 0.27). In conclusion, the average crude mortality rate due to drug overdose in Halifax from 1993 to 1995 was 4.1 deaths per 100,000 population. Potential threats to the quality of data were the lack of standardization concerning toxicological testing and the definition of drug overdose.

Key words: Canada; drug overdose death; epidemiology; surveillance

Introduction

Death due to overdose is one of the most dire consequences of drug abuse. Overdose deaths can be intentional or unintentional, and they can result from both licit and illicit drug abuse. Drugs commonly implicated in overdose deaths are alcohol, psychoactive medications, analgesics, illicit drugs such as cocaine and heroin, and multiple drugs taken concomitantly.

In Vancouver, the crude mortality rate due to overdose of an illicit drug increased eightfold among males over four years, from 2 per 100,000 in 1989 to 16 per 100,000 in 1993.1 This was considered an epidemic, and the evidence suggested that young or naive users were at high risk of overdose because of especially pure heroin.1 Toronto also experienced a rise in heroin deaths such that, in 1992, the crude mortality rate due to heroin overdose was 1.4 deaths per 100,000.2 Elsewhere in Canada, little is known about the epidemiology of deaths attributed to licit and illicit drugs.

Drug overdose deaths are medico-legal cases investigated by medical examiners or coroners. Although medical examiner charts are recognized as a key source of information for monitoring such deaths, the accuracy of official statistics on drug mortality remains uncertain.3-6 Numerous sources of bias have been identified. Selection bias may exist because the deaths of persons not well known to a doctor or with little access to health care are more likely to be investigated by a medical examiner than are deaths of persons known to the medical system.7 Medical examiners can be pressured by insurance companies to classify a death as a suicide, by the deceased's beneficiary to classify it as accidental, or by family members to report a less stigmatizing cause of death.5,7,8 In addition, the experience, residence and religion of medical examiners can influence their investigative and reporting decisions.9

Establishing death due to overdose is a medico-legal decision that can be based on a wide assortment of evidence: autopsy; toxicological testing; description of circumstances obtained from police, witnesses, persons close to the deceased or who have knowledge of the deceased; suicide notes; medical and psychiatric history including medication and substance abuse; and legal antecedents including those pertaining to alcohol and other drugs. None of these elements of evidence can be said to be necessary or sufficient for a death to be attributed to a drug overdose. Rather, the medical examiner orders, reviews and assesses the collective evidence and arrives at a medico-legal decision based on his/her expert judgement. In particular, a death may be validly attributed to overdose even in the absence of toxicological testing. By contrast, even when toxicological testing reveals the presence of one or more drugs, a death may nonetheless validly be declared as not due to overdose.

The objectives of the present study were to describe the epidemiology of drug overdose deaths in Halifax and to assess the quality of the information obtained from medical examiner charts with respect to drug overdose deaths, for surveillance purposes. Halifax, Nova Scotia, is one of the sentinel cities participating in the Canadian Community Epidemiology Network on Drug Use (CCENDU), a national surveillance system on substance abuse.


Setting

The systems for the investigation of deaths vary across provincial jurisdictions.12 In Nova Scotia, the 1989 Fatality Injuries Act requires medical examiners to investigate and determine the cause of death in cases of death due to violence, undue means, culpable negligence or undetermined cause, death in jail or prison, or death in a place or under circumstances requiring an inquest by statute. The Chief Medical Examiner is a medical pathologist appointed by the Governor-in-Council. For the three-year period (1993-1995) of the present study, Halifax was served by the Chief Medical Examiner as well as four medical examiners working on a contract basis. In 1995, the Regional Municipality of Halifax (Halifax) had a population of 342,771 persons.13


Methods

The present study recognizes two circumstances concerning death and drugs: death due to overdose, and death where drugs are implicated or detected through toxicological testing but where the final medico-legal decision is not death due to overdose.

This study used a case series of all charts from the Office of the Chief Medical Examiner for deaths investigated in Halifax from 1 January 1993 to 31 December 1995. All charts were reviewed manually in order to identify all cases of death with a final medico-legal disposition of death due to overdose and all cases where a toxicology profile had been obtained, whether or not the results were positive. Only cases where the deceased both resided and died in Halifax were included. Residents of Halifax who died elsewhere were excluded because of insufficient documentation.

The complete medical examiner chart regarding death due to overdose includes a medical examiner report, the results of a toxicology profile if one has been ordered and possibly a psychological autopsy in cases of suicide. The medical examiner defines the cause of death, and the report provides details about the manner and circumstances of death and the results of a medico-legal autopsy. In Halifax, an overdose is classified as intentional only if established conclusively through a note written by the deceased. In our study, cases with a final medico-legal disposition of death due to overdose were abstracted as to sex, age at death, cause of death, manner of death and toxicology results.

In Halifax, toxicology profiles are obtained according to the circumstances of the individual case. Criteria for requesting toxicology profiles are not standardized or explicit. In general, specimens are collected from blood, urine and/or vitreous humour. Toxicological analysis is performed at the provincial laboratory except in criminal cases when analysis is performed at the RCMP forensic laboratory. Initially, blood and urine specimens undergo drug screening techniques. A positive screening test is followed by a confirmatory analysis to identify and quantify the drug present. The provincial laboratory reports only those results that exceed a specific concentration (for example, 10 mg/dl for alcohol). The presence of a given drug in toxicological testing does not exclude the presence of one or more other drugs.

Toxicological testing represents objective evidence that may provide a useful confirmatory component in the surveillance of death due to overdose, for example, as a potential means of identifying false positive cases. However, toxicological testing is not necessarily ordered as part of the investigative process of death due to overdose. Furthermore, in cases where multiple drugs are detected or where levels are not flagrantly in the toxic levels, the decision to attribute a death to drug overdose may hinge on a knowledge of drug interactions and metabolism. In such cases, the expert opinion of a toxicologist may become germane.

Therefore, as part of our study, a toxicologist and a medical examiner each were given a list of the age, sex and toxicology results of all deaths with a final medico-legal disposition of drug overdose, and each independently classified the deaths as either due to overdose or as indeterminate/not consistent with overdose. Agreement between the toxicologist and the medical examiner was assessed using the Kappa coefficient.14

Deaths where toxicological testing is performed but which are not ultimately considered as due to overdose do not necessarily have the same level of documentation as do those with a final disposition of death due to overdose. Thus, it may not be possible to retrospectively identify false negative cases of death due to overdose. Nonetheless, one would expect that, as a group, deaths not considered to be due to overdose would have a different overall mix of toxicology profiles than that of the group of deaths due to overdose. Therefore, the two groups of deaths (overdose and not overdose) were compared in two ways. First, differences in the median number of drugs detected on toxicological testing, which was not normally distributed, were compared using the non-parametric Kruskal-Wallis test. Second, the proportions of deaths where specific categories of drugs were found were compared using the chi-squared test.

The chi-squared test or Fisher's exact test was used to compare the numbers of (i) drug overdose deaths, (ii) cases undergoing toxicological testing and (iii) cases of positive toxicology, as proportions of the total number of deaths investigated in each of the three years. The Fisher's exact test was used to compare the number of cases of suicide among deaths due to overdose during the three-year period, according to sex. Differences in median age at death, which was not normally distributed, were tested using the non-parametric Kruskal-Wallis test. EpiInfo Version 6 was used for data management and analysis.15


Results

From 1993 to 1995, 636 deaths of persons who resided and died in Halifax were investigated by a medical examiner (Table 1). Forty-two (6.6%) deaths had a final medico-legal disposition of being due to drug overdose. There was no significant difference in the proportions of overdose deaths over the three years (p=0.67). Therefore, the average annual crude rate of mortality due to drug overdose in Halifax was 4.1 per 100,000 population (14 ÷ 342,771 x 100,000) from 1993 to 1995.

During the same time period, toxicological testing was performed in 292 (45.9%) of the deaths investigated by a medical examiner. A significant increase in the proportion of cases undergoing toxicology tests occurred between 1994 and 1995 (p=0.002) [Table 1]. Cases of positive toxicology were found in 189 (64.7%) of the deaths tested over the three years, with significant differences in the annual proportions from year to year (p=0.034).

Of the 42 cases of death due to overdose, 52% were male. The age at death ranged from 22 to 73 years; the median age at death was 45.6 with no statistical difference over the three-year period (p=0.11). Twenty (47.6%) overdose deaths were suicides, nineteen (45.2%) were unclassified or undetermined and three (7.1%) were unintentional. Suicide was recorded as the manner of death in a larger proportion among females than among males (65% vs 32.8%, p<0.03).

Table 2 lists the drugs implicated and/or detected in the 42 overdose deaths. Toxicological tests were performed on 38 of these cases. Testing was not performed in the remaining four cases where the individuals were admitted to hospital in critical condition. Those four deaths were attributed to overdoses of insulin, valproic acid, verapamil and carbamapezine.

Ethanol was the drug most frequently implicated and/or detected (47.6%) in cases of death due to overdose. Illicit drugs (cocaine and cannabis) were detected in five overdose deaths; however, only two of these deaths had a final medico-legal disposition of death due to overdose of an illicit drug (cocaine). Viewed another way, 61.9% of all deaths due to overdose in Halifax from 1993 to 1995 involved psychotropic medications often prescribed in the treatment of mental health disorders (antidepressants, benzodiazepines, antipsychotics, hypnotics and sedatives).

During the study period, 38 of the 292 cases that underwent toxicological testing received a final disposition of death due to overdose and 254 cases did not. The number of drugs detected among overdose deaths was significantly greater than that detected among non-overdose deaths (overdose: median 2, 25th and 75th percentiles 2 and 3; non-overdose: median 1, 25th and 75th percentiles 0 and 1) (p<0.01). As well, with the exception of alcohol and cannabis, the proportions of deaths in which 11 specific categories of drugs were implicated were significantly greater for overdose deaths than for deaths not due to overdose (Table 2). We conclude that, at the group level, the toxicological profile of drug overdose deaths was significantly different from that where death was designated as not due to overdose.

Finally, the determination of death (overdose or indeterminate/not consistent with overdose) from the 38 toxicology results, independently assessed by a toxicologist and a medical examiner, revealed poor overall agreement between the two reviewers (Kappa coefficient of 0.27).

 

TABLE 1

Deaths investigated by a medical examiner, Halifax, 1993-1995

 

1993

1994

1995

3-year period

Number of deaths investigated

221

201

214

636

Number of (%a) deaths with toxicology profiles

87

(41.2%)

86

(42.8%)

119

(55.6%)

292

(45.9%)

Number of (%b) deaths with positive toxicology

61

(70.1%)

46

(53.5%)

82

(68.9%)

189

(64.7%)

Number of (%a) overdose deaths

12

(5.4%)

14

(7.0%)

16

(7.5%)

42

(6.7%)



a Percent of deaths investigated
b Percent of deaths with toxicology profiles

   

Discussion

Based on medical examiner investigations as reported, the average annual crude rate of mortality due to drug overdose in Halifax was 4.1 per 100,000 population from 1993 to 1995. The rate of mortality due to drug overdose, for all drugs combined, appears to be low in Halifax. The vast majority of drug overdose deaths in Halifax were due to licit substances, primarily alcohol and prescription psychotropic medications.

Few drug overdose deaths in Halifax involved illicit drugs, and the rate of deaths actually attributed to an illicit drug was 0.2 deaths per 100,000 population from 1993 to 1995. In contrast, Vancouver recorded 22 deaths per 100,000 population involving heroin, cocaine and other illicit drugs in 1995, and Toronto's mortality rate for heroin as the sole lethal drug was 1.45 deaths per 100,000 population.10,16 The finding of a low mortality rate for illicit drug overdose in Halifax is corroborated by other population-level indicators collected by the CCENDU.10,11 Halifax has low per capita numbers of cocaine- and heroin-related law enforcement charges and hospital separations, as compared with Vancouver, Toronto and Montreal.

The most fundamental problem with the quality of the data in the present study was the lack of an explicit definition of drug mortality. According to Shai (1994), definitions used by various American institutions range from the broad concept of "drug-induced deaths," which includes deaths from both dependent and non-dependent drugs, legal and illegal drug use, as well as poisoning from medically prescribed and other drugs, to a narrowly defined "drug dependence."5 Shai defined drug mortality as "deaths due to psychoactive drugs, legal or illegal, through natural causes (chronic or acute narcotism) or accidental or purposive overdose."5

Another major potential source of bias in the Halifax data was the apparent lack of uniform methods and interpretive criteria in the medical examiner investigations conducted by a total of five examiners over the three-year period. Toxicological testing was requested without explicit guidelines and was performed on less than half of the deaths investigated.

Toxicology analysis is considered an essental adjunct to anatomical diagnosis in medical examiner cases possibly related to drug abuse.3,5,6,17,18 Generally, a positive history of drug or alcohol abuse, or the presence of items associated with alcohol or drug abuse at the scene, leads to the requestioning of toxicology tests.5 Although universal toxicological screening could potentially serve to improve case ascertainment, Tormey et al. (1989) stated that such a strategy would be time consuming, costly and inefficient.18 Furthermore, universal screening could lead to the identification and potential misclassification of cases of therapeutic and subtherapeutic drug levels of questionable toxicological significance (false positives).18 According to Jammehdiabadi and Tierney (1991), a thorough history and clinical assessment of the patient who overdosed take precedence over analytical screens in determining what drugs are involved in suspected overdose.19

In our study, the poor degree of agreement between the toxicologist and the medical examiner, who independently reviewed the toxicology results, exemplifies the difficulty of judging cause of death, especially where multiple drugs are detected or the levels are not flagrantly in the toxic range. However, our finding of a significant difference in toxicology profiles of the two groups of deaths (i.e. those with a final medico-legal disposition of death due to drug overdose and those without) suggests remarkable consistency within each group and a strong difference between the two groups. Our study might have been strengthened by reviewing the medical examiner charts of the latter group in order to identify cases of false negativity. However, even that procedure would have been fraught with ambiguity because of information bias from variability in wording and completeness of records and from the inter-examiner variation in certification judgements.5,7,9

In conclusion, numerous studies have emphasized that suicide, drug overdose deaths, lethal poisonings and injury deaths generally are underreported, whatever the surveillance measure.3,5-7,20-22 Indeed, our own study reveals potential sources of bias and inconsistency concerning medical examiner data on drug-related deaths. Nonetheless, given that medical examiner reports provide valuable information not easily available elsewhere, the CCENDU has adopted drug overdose deaths as investigated by medical examiners to be one of its key indicators for surveillance purposes. The CCENDU's real challenge now is to work toward the improvement of medical examiner information in Canada as an accurate and reliable source of surveillance data.

TABLE 2

Drugs implicated and/or detected in overdose deaths and drugs detected
in non-overdose deaths where toxicology testing was performed,
Halifax, 1993-1995

 

Overdose deaths (n=42*)

Other deaths with toxicology (n=254)

x2 test

Drugs

Cases

(%)

Cases

(%)

p-value**

Ethanol

20

(47.6 )

98

(38.6 )

0.27

Other alcohola

3

(7.1 )

1

(0.3 )

<0.001

Antidepressantsb

16

(38.1 )

5

(2.0 )

<0.001

Medications available without prescriptionc

12

(28.6 )

28

(11.0 )

0.002

Narcotic analgesicsd

10

(23.8 )

11

(4.3 )

<0.001

Benzodiazepines

8

(19.0 )

21

(8.3 )

0.03

Drugs for major psychiatric or neurologic disorderse

5

(11.9 )

9

(3.5 )

0.02

Hypnotics and sedativesf

5

(11.9 )

4

(1.6 )

<0.001

Other prescription medicationsg

4

(9.5 )

0

( 0 )

<0.001

Cocaine

3

(7.1 )

3

(1.2 )

0.01

Cannabis

2

(4.8 )

6

(2.4 )

0.37

Miscellaneoush

3

(7.1 )

17

(6.7 )

N/A


* Toxicological testing was performed for 38 overdose deaths. In the 4 cases where it was not done, overdose was substantiated through other evidence.
** Chi-squared test compares overdose and non-overdose deaths for each specific drug category. It was not performed for the category "Miscellaneous."
a Methanol, isopropanol
b Amitriptyline, doxepin, impramine, nortriptyline, sertraline, fluoxetine
c Salicylate, acetaminophen, diphenhydramine
d Codeine, meperidine, morphine, methadone
e Methotrimeprazine, trifuloperazine, carbamapezine, valproic acid
f Chloral hydrate, butalbital, phenobarbital, zopiclone
g Orphenadrine, methocarbamol, verapamil, insulin
h Overdose deaths: toluene, hydrocarbon, cyanide; non-overdose deaths: atropine, lidocaine, ranitidine, ketamine

   

Acknowledgements

We thank doctors Ian Salathiel and Albert Fraser for independently classifying a group of toxicology profiles. This research was funded by Health Canada through the National Health Research and Development Program (project no 6606-6022-703).


References

    1. Task Force Into Illicit Narcotic Overdose Deaths in British Columbia. Report of the Task Force Into Illicit Narcotic Overdose Deaths in British Columbia. British Columbia: Ministry of Attorney General, 1994.

    2. Metro Toronto Research Group on Drug Use. Drug use in Metropolitan Toronto 1994.

    3. Dijkhuis H, Zerling C, Parrish G, Bennett T, Kemper CG. Medical examiner data in injury surveillance: a comparison with death certificates. Am J Epidemiol 1994;139:637-43.

    4. Rutternber AJ, Luke JL. Heroin-related deaths: new epidemiologic insights. Science 1984;226:14-20.

    5. Shai D. Problems of accuracy in official statistics on drug-related deaths. Int J Addictions 1994;29:1801-11.

    6. Soslow AR, Woolf AD. Reliability of data sources for poisoning deaths in Massachusetts. Am J Emerg Med 1992;10:124-7.

    7. Graitcer PL, Williams WW, Finton RJ, Goodman RA, Thacker SV, Hanzlick R. An evaluation of the use of medical examiner data for epidemiologic surveillance. Am J Public Health 1987;77:1212-4.

    8. Girela E, Lachica E, Pounder D. Death certification of problem drinkers. Med Sci Law 1992;32:233-6.

    9. Jarvis GK, Boldt M, Butt J. Medical examiners and manner of death. Suicide and Life-Threatening Behaviour 1991;21:115-33.

    10. Poulin C. The Canadian Community Epidemiology Network on Drug Use. Inaugural national report. NHRDP Project No 6606-6022-703. Ottawa: Canadian Centre on Substance Abuse, 1997.

    11. Poulin C, Fralick P, Whynot E, El-Guebaly N, Kennedy D, Bernstein J, et al. The epidemiology of cocaine and opiate use in urban Canada. Can J Public Health 1998;89:234-8.

    12. Combs DL, Parrish RG, Ing R. Death investigation in the United States and Canada, 1995. Atlanta: Centers for Disease Control and Prevention, 1995.

    13. Statistics Canada. Annual demographic statistics, 1995. Ottawa, 1995; cat 91-213-XPB.

    14. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960;20:37-46. As cited in: Kelsey JL, Thompson WD, Evans AS. Methods in observational epidemiology. New York: Oxford University Press, 1986:288-93.

    15. Dean AG, Dean JA, Coulombier D, Brendel KA, Smith DC, Burton AH, et al. Epi Info, Version 6: a word processing, database, and statistics program for epidemiology on microcomputers. Atlanta: Centers for Disease Control and Prevention, 1994.

    16. Metro Toronto Research Group on Drug Use. Fax on drugs. 1997 Apr 29;1-3.

    17. Conroy C, Russel JC. Medical examiner/coroner records: uses and limitations in occupational injury epidemiologic research. J Forensic Sciences 1990; 35:932-7.

    18. Tormey WP, Crosse H, Pierce A. Chemical toxicology for the coroner. Human Toxicol 1989;8:375-9.

    19. Jammehdiabadi M, Tierney M. Impact of toxicology screens in the diagnosis of a suspected overdose: salicylates, tricyclic antidepressants, and benzodiazepines. Vet Hum Toxicol 1991;33:40-3.

    20. Blanc PD, Jones MR, Olson KR. Surveillance of poisoning and drug overdose through hospital discharge coding, poison control center reporting, and the Drug Abuse Warning Network. Am J Emerg Med 1993;11:14-9.

    21. Cooper PN, Milroy CM. The coroner's system and under-reporting of suicide. Med Sci Law 1995;35:319-26.

    22. Pollock DA, Holmgreen P, Lui KJ, Kirk ML. Discrepancies in the reported frequency of cocaine-related deaths, United States, 1983 through 1988. JAMA 1991;266:2233-7.


Author References
Christiane Poulin, Associate Professor, Department of Community Health and Epidemiology, Faculty of Medicine, Dalhousie University, 5849 University Avenue, Halifax, Nova Scotia   B3H 4H7; E-mail: Christiane.Poulin@dal.ca
Jonathan Stein, University of Toronto, Toronto, Ontario
John Butt, Chief Medical Examiner of Nova Scotia, Halifax, Nova Scotia

 

[Previous][Table of Contents] [Next]

 
Last Updated: 2002-10-22 Top
Important Notices