This cross-sectional study investigates the substances detected at the time of death in poisoning- and nonpoisoning-related suicide.
Key Points
Question
What are the substances detected at the time of death in poisoning and nonpoisoning suicides?
Findings
In this cross-sectional study, data were extracted from postmortem toxicology reports from 13 664 suicide deaths in Australia. Psychotropic medicine usage was higher in poisoning than nonpoisoning suicides, especially for antidepressants, benzodiazepines, and opioids; alcohol was almost equally prevalent, whereas cannabinoids and amphetamines were more commonly detected in nonpoisonings.
Meaning
Substance misuse is a risk factor for all suicide, and substances overrepresented in poisonings should be prescribed cautiously.
Abstract
Importance
Determining the association between drug use and suicide is complicated but can help to inform targeted suicide prevention strategies.
Objective
To examine the substances prevalent in poisoning- and nonpoisoning-related suicides in Australia.
Design, Setting, and Participants
This was a multiple-year, cross-sectional study of suicides from July 2013 to October 2019 in Australia with toxicology data available in a national coronial database. The cause of death was classified as poisoning related if any type of poisoning was determined by the coroner to contribute to the cause of death. Prevalence ratios (PRs) were calculated to compare substance detection in poisoning- vs nonpoisoning-related suicides. Data were analyzed from October 2021 to April 2023.
Exposures
All substances detected in decedents at the time of death according to toxicology reports were recorded.
Main Outcome(s) and Measure(s)
The most common individual substances and substance classes were identified. From these, blood concentrations of substances of interest were analyzed, and the most commonly occurring combinations of substance classes were listed.
Results
Toxicology was performed on 13 664 suicide decedents (median [IQR] age, 44 [31-57] years; 10 350 male [76%]). From these, 3397 (25%) were poisoning-related suicides (median [IQR] age, 50 [38-63] years; 2124 male [63%]). The remainder were classified as nonpoisoning-related suicides (median [IQR] age, 42 [29-55] years; 8226 male [80%]). PRs for common medicine classes being detected in poisoning-related suicides compared with nonpoisoning-related suicides were as follows: antidepressants (PR, 1.63; 95% CI, 1.54-1.73), benzodiazepines (PR, 2.01; 95% CI, 1.90-2.13), nonopioid analgesics/anti-inflammatory drugs (PR, 1.88; 95% CI, 1.78-2.00), and opioids (PR, 2.72; 95% CI, 2.58-2.87). Alcohol (as ethanol ≥0.03 g/100 mL) was almost equally prevalent in poisoning- and nonpoisoning-related deaths (PR, 1.07; 95% CI, 1.01-1.14), whereas amphetamines (PR, 0.68; 95% CI, 0.61-0.77) and cannabinoids (PR, 0.67; 95% CI, 0.60-0.74) were detected more often in nonpoisoning-related suicides. Combinations of multiple sedative agents in poisoning-related suicides were common.
Conclusions and Relevance
Both poisoning- and nonpoisoning-related suicide deaths featured a high prevalence of psychotropic medicines or potential intoxication, which suggests the association of suicide with poor mental health and substance misuse. Findings suggest that substances with a high involvement in poisoning-related suicides should be prescribed cautiously, including antidepressants that are toxic in overdose, sedatives, opioids, and potentially lethal combinations.
Introduction
There are complex individual- and population-level factors involved in suicide.1 This includes drug use, where both licit and illicit use is strongly implicated in suicide. Misuse of alcohol, addictive substances, and prescription medicines are well-recognized risk factors for suicide and can also be used as a means for suicide.1 Conversely, appropriate use of medicines for mental and other chronic health conditions is essential for suicide prevention.2
Commonly detected medicine classes in poisoning suicides in Canada, the US, Europe, New Zealand, and Australia include hypnosedatives, tricyclic antidepressants, benzodiazepines, neuroleptics, and over-the-counter medicines.3,4,5,6,7,8 Opioids, both prescribed and illicit, remain a concern in high-income countries.4,8 In contrast, pesticides often predominate in poisoning suicides in lower-income countries.8,9
Nationwide surveillance of suicide and self-harm is essential for detecting the emergence of new methods, including toxic substances used, to allow targeted means restriction and priority antidote development.10 Indeed, means restriction to reduce public access to poisons is a population-level intervention associated with a decrease in suicide deaths by poisoning without a corresponding increase in other methods of suicide.11
Means restriction interventions should target highly toxic substances most frequently implicated in fatal overdoses.11,12 Case fatality studies, which compare lethal poisonings relative to poisoning events, provide strong evidence to identify agents highly lethal in overdose and toxic combinations.7,12,13 A complementary study design that does not require a denominator of nonfatal cases is to determine substances most commonly detected in poisoning compared with nonpoisoning suicides. This is because substances can be detected at autopsy even when they did not contribute to death (in both poisoning and nonpoisoning suicides).14 Substances more often detected in poisoning suicides can add further weight to evidence from case fatality studies. In addition, identifying substances in nonpoisoning suicides sheds light on medicine use at the time of suicide more generally.
In Australia, there are an average of 3000 deaths by suicide every year.15 Almost two-thirds of intentional poisoning deaths in Australia involved 2 or more substance classes combined.6 Poisoning deaths involving multiple substances are more difficult to interpret as not every substance is necessarily contributory to the death.14,16 Substance combinations can also contribute to death by synergistic toxicity, and concentrations do not have to be at toxic levels to cause poisoning.6,14
We aimed to characterize substances detected in suicides in Australia. Specifically, we aimed to (1) identify the most common substance classes and individual substances detected at autopsy and their prevalence in poisoning- and nonpoisoning-related suicides, (2) analyze concentrations of selected substances of interest in poisoning- and nonpoisoning-related suicides, and (3) characterize substance class combinations detected in poisoning- and nonpoisoning-related suicides.
Methods
Study Design
This was a multiple-year, cross-sectional study of suicides from 2013 to 2019 in Australia. Data were collected after the death of each decedent. This study represents 1 aim of the larger Australian Suicide Prevention using Health Linked data (ASHLi) study, which involves data collected retrospectively at a population level from government and health administrative databases in Australia. The ASHLi study has been approved by the following ethics committees: (1) Victorian Department of Justice and Community Safety Human Research Ethics Committee, (2) Western Australian Coronial Ethics Committee, (3) Australian Institute of Health and Welfare, and (4) New South Wales Population & Health Services Research Ethics Committee.17 A waiver of consent for the use of the coronial data in this study was granted by the human research ethics committee. We investigated all available closed coronial cases in people 10 years or older in Australia who were deemed by the coroner as having an intentional death due to external cause(s). There were 13 decedents with undetermined intent who were not included in this data extraction. Due to administrative requirements where other databases were involved, data were available between July 1, 2013, and October 10, 2019.17 More details surrounding the wider ASHLi study have been described in detail elsewhere.17 The current study uses coronial data only.
Details on the number of cases extracted each year are available in eFigure 1 and eTable 6 in Supplement 1. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)18 reporting guidelines (eAppendix in Supplement 1).
Data Source
The National Coronial Information System (NCIS) is the official database of reportable deaths in Australia and New Zealand. The NCIS contains reports from coroners and forensic specialists, including toxicology findings where this procedure has been performed and the cause(s) of death as determined by the coroner.
Data Extraction and Preparation
We extracted and coded all substances detected at time of death and if available, the concentration and specimen sample site (eg, postmortem femoral blood, urine) for each decedent. We categorized each decedent with a positive toxicology finding as single or multiple agents, based on whether 1 or multiple substances were detected. We excluded substances if they were metabolites detected in conjunction with their parent substance (eTable 1 in Supplement 1). We only considered alcohol as a detected substance if present in concentrations of 0.03 g/100 mL or greater. Low levels of ethanol, typically less than 0.03 g/100 mL, may be formed during postmortem decomposition (eMethods in Supplement 1).19,20
Cause of Death Data
We searched data fields in the NCIS to determine whether a poisoning was involved in the method of death. These fields are used by the coroner, where cause of death is a free-text field, and classifications for the mechanism and object of injury are coded categorical fields. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes are externally supplied by the Australian Bureau of Statistics. We searched these fields to identify poisoning suicides. We categorized a suicide as poisoning related if poisoning was mentioned in any data field to contribute to the death to any extent. If poisoning was mentioned in the cause of death fields among other means of suicide, such as hanging, this was still considered a poisoning-related suicide. Poisoning-related suicides were divided into medicine- or drug-poisoning suicides and nontherapeutic gas– or chemical-poisoning suicides.
Statistical Analysis
To address our first aim, we identified the most common substances and substance classes detected. We calculated the prevalence ratio (PR), which is similar to relative risk, for each substance to estimate how often a substance was detected in poisoning-related suicides compared with nonpoisoning-related suicides (eMethods in Supplement 1).21 We also calculated the PR for substance classes. We used a modified Poisson regression model to calculate age- and sex-adjusted PR (aPR) for substances and classes.22,23 Age and sex were grouped into categorical variables.22,23
To address the second aim, we analyzed substance concentrations. We chose the most common substances and grouped them according to their indication or unit of measurement. We listed the median concentration of each substance in poisoning- and nonpoisoning-related suicides. We used 90th percentile (P90) to represent the relatively higher concentration threshold for each substance and calculated the percentage of observations above P90 that were categorized as poisoning-related suicides. We graphed substance concentrations in poisoning- and nonpoisoning-related suicides on a log (base 10) scale on violin plots, using kernel density estimation.24,25,26
Kernel density estimation shows the probability density, or relative probability, of each value occurring in the distribution.25,26 The kernel function applies a weighted average to sections of data and smooths the appearance of the violin or curve.26 This creates a visual representation of the distribution of data.
To address the third study aim, we concatenated the substance classes detected in each decedent to create a list of all existing substance class combinations. The 30 most common combinations, stratified by poisoning- and nonpoisoning-related suicides, were presented on an UpSet plot.27,28
An UpSet plot displays mutually exclusive groupings of data sets on a bar chart. This is read with a corresponding matrix table in position of the axis. The matrix indicates which variable (substance class) is represented in each data set by coloring the circle for the same variable if present. When a combination of multiple variables are present, this is further highlighted with a line to connect the colored circles. Data were analyzed from October 2021 to April 2023 using SAS software, version 9.3 (SAS Institute) and Microsoft Excel (Microsoft Corporation).
Results
Sample Characteristics
There were 14 330 cases of suicide decedents available from the NCIS that met inclusion criteria for the ASHLi study.17 Toxicology was performed on 13 664 decedents (median [IQR] age, 44 [31-57] years; 10 350 male [76%]; 3314 female [24%]). From these, 3397 (25%) were poisoning-related suicides (median [IQR] age, 50 [38-63] years; 2124 male [63%]; 1273 female [37%]). The remainder were classified as nonpoisoning-related suicides (median [IQR] age, 42 [29-55] years; 8226 male [80%]; 2041 female [20%]). There were similar demographics for the groups with and without toxicology performed (eTable 2 in Supplement 1).
One-quarter of the included suicides (3397 [24.9%]) were poisoning related. Mixed methods involving both poisoning and nonpoisoning methods of death (eg, poisoning and hanging) occurred in 633 decedents. Medicine or drugs, including illicit drugs, were involved in 2132 poisoning-related suicides (1096 male [51%]/1036 female [49%]) (eTable 3 in Supplement 1). Nontherapeutic gases or chemicals were involved in 1265 poisoning-related suicides (1028 male [81%]/237 female [19%]).
There were 10 267 nonpoisoning-related suicides (8226 male [80%]/2041 female [20%]). Other demographic factors including age, urbanicity, relationship status, and employment status are in eTable 3 in Supplement 1.
Toxicology Findings
Substances and Substance Classes Detected at Autopsy
Table 1 shows the most common substances detected postmortem with unadjusted PRs and PRs adjusted for age and sex. Substances with the highest (unadjusted) PRs to be detected in poisoning- compared with nonpoisoning-related suicides included pentobarbitone, carbon monoxide, and oxycodone (PR range, 2.92-4.23). Delta-9-tetrahydrocannabinol (THC; PR, 0.66; 95% CI, 0.59-0.74) and methylamphetamine (PR, 0.65; 95% CI, 0.57-0.74) were more commonly detected in nonpoisoning suicides.
Table 1. Frequency of the Most Common Substances Detected in Descending Ordera.
| Substance detected | Cause of suicide death | PR (95% CI) to be detected in poisoning- compared with nonpoisoning-related suicide | aPR ratio (95% CI), adjusted for age and sex | ||||
|---|---|---|---|---|---|---|---|
| Nonpoisoning-related suicide | Poisoning-related suicide | Total | |||||
| Single agent | Multiple agent | Single agent | Multiple agent | ||||
| Alcoholb | 937 | 2254 | 111 | 1016 | 4318 | 1.07 (1.01-1.14) | 1.19 (1.12-1.26) |
| Diazepam | 88 | 1374 | 7 | 880 | 2349 | 1.70 (1.60-1.81) | 1.53 (1.44-1.63) |
| Paracetamol | 116 | 1249 | 14 | 928 | 2307 | 1.89 (1.78-2.01) | 1.56 (1.47-1.66) |
| Delta-9-tetrahydrocannabinol | 265 | 1113 | 9 | 275 | 1662 | 0.66 (0.59-0.74) | 0.85 (0.76-0.96) |
| Methylamphetamine | 167 | 892 | 10 | 202 | 1271 | 0.65 (0.57-0.74) | 0.83 (0.73-0.94) |
| Codeine | 8 | 490 | <6c | 588 | 1086 | 2.42 (2.27-2.58) | 2.14 (2.00-2.28) |
| Mirtazapine | 73 | 611 | <6 | 326 | 1010 | 1.33 (1.21-1.46) | 1.14 (1.04-1.25) |
| Carbon monoxide | 28 | 222 | 105 | 640 | 995 | 3.58 (3.40-3.76) | 3.58 (3.39-3.78) |
| Citalopram | 105 | 570 | <6 | 246 | 921 | 1.08 (0.97-1.21) | 1.00 (0.90-1.12) |
| Quetiapine | 29 | 408 | 6 | 389 | 832 | 2.03 (1.88-2.19) | 1.82 (1.68-1.97) |
| Oxycodone | 20 | 239 | <6 | 495 | 754 | 2.92 (2.75-3.10) | 2.32 (2.17-2.47) |
| Temazepamd | 52 | 315 | 8 | 272 | 647 | 1.81 (1.65-1.98) | 1.37 (1.24-1.50) |
| Venlafaxine | 42 | 364 | <6 | 198 | 604 | 1.34 (1.19-1.51) | 1.14 (1.02-1.29) |
| Sertraline | 56 | 358 | <6 | 152 | 566 | 1.08 (0.94-1.25) | 1.01 (0.89-1.16) |
| Desmethylvenlafaxined | 40 | 338 | <6 | 174 | 552 | 1.28 (1.13-1.46) | 1.18 (1.05-1.34) |
| Doxylamine | 42 | 303 | <6 | 200 | 545 | 1.51 (1.34-1.69) | 1.31 (1.17-1.46) |
| Ibuprofen | 27 | 293 | 0 | 201 | 521 | 1.59 (1.42-1.78) | 1.47 (1.31-1.64) |
| Olanzapine | 30 | 334 | <6 | 155 | 519 | 1.21 (1.06-1.39) | 1.13 (0.98-1.29) |
| Amitriptyline | 18 | 165 | 7 | 304 | 494 | 2.69 (2.49-2.89) | 2.16 (2.00-2.32) |
| Metoclopramide | 10 | 151 | <6 | 273 | 434 | 2.66 (2.46-2.88) | 2.08 (1.92-2.25) |
| Fluoxetine | 38 | 229 | 0 | 140 | 407 | 1.40 (1.22-1.61) | 1.29 (1.13-1.48) |
| Caffeine | 27 | 238 | 0 | 114 | 379 | 1.22 (1.04-1.42) | 1.17 (1.00-1.35) |
| Morphined | <6 | 225 | <6 | 126 | 351 | 1.46 (1.27-1.69) | 1.31 (1.15-1.49) |
| Tramadol | <6 | 150 | <6 | 192 | 342 | 2.33 (2.11-2.57) | 1.99 (1.81-2.20) |
| Midazolam | 9 | 250 | <6 | 88 | 347 | 1.02 (0.85-1.23) | 1.01 (0.85-1.18) |
| Alprazolam | 6 | 161 | <6 | 137 | 304 | 1.85 (1.63-2.10) | 1.70 (1.50-1.93) |
| Lidocaine/lignocaine | <6 | 198 | <6 | 91 | 289 | 1.27 (1.07-1.51) | 1.22 (1.05-1.43) |
| Duloxetine | 19 | 144 | <6 | 119 | 282 | 1.72 (1.50-1.98) | 1.50 (1.32-1.71) |
| Oxazepamd | 11 | 112 | <6 | 115 | 238 | 1.98 (1.73-2.26) | 1.43 (1.26-1.63) |
| Pentobarbitone | 0 | <6 | 20 | 196 | 216 | 4.23 (4.10-4.36) | 3.10 (2.90-3.32) |
| 7-Aminoclonazepamd | <6 | 109 | 0 | 103 | 212 | 1.98 (1.72-2.29) | 1.77 (1.53-2.04) |
| Amlodipine | 16 | 110 | <6 | 81 | 207 | 1.59 (1.34-1.89) | 1.32 (1.11-1.56) |
| Promethazine | <6 | 85 | <6 | 115 | 200 | 2.36 (2.09-2.67) | 1.94 (1.71-2.19) |
| Nordiazepamd | 11 | 133 | 0 | 52 | 196 | 1.07 (0.84-1.35) | 0.97 (0.77-1.22) |
| Quinine | 16 | 109 | 0 | 64 | 189 | 1.37 (1.12-1.67) | 1.12 (0.92-1.36) |
| Metoprolol | <6 | 96 | 0 | 85 | 181 | 1.91 (1.63-2.24) | 1.44 (1.23-1.68) |
| Salicylic acid | 15 | 93 | 0 | 78 | 186 | 1.70 (1.43-2.02) | 1.38 (1.17-1.64) |
| Metformin | <6 | 100 | 0 | 71 | 171 | 1.68 (1.41-2.02) | 1.35 (1.12-1.63) |
| Methadone | <6 | 109 | <6 | 61 | 170 | 1.45 (1.18-1.78) | 1.45 (1.19-1.77) |
| Zopiclone | <6 | 72 | 0 | 98 | 170 | 2.36 (2.07-2.69) | 1.96 (1.71-2.26) |
| Zolpidem | 8 | 63 | 0 | 100 | 171 | 2.39 (2.10-2.72) | 1.90 (1.66-2.17) |
| Fentanyl | <6 | 85 | <6 | 77 | 162 | 1.93 (1.64-2.28) | 1.64 (1.38-1.93) |
| Irbesartan | 12 | 94 | <6 | 54 | 160 | 1.36 (1.10-1.70) | 0.99 (0.80-1.22) |
| Telmisartan | 13 | 80 | <6 | 62 | 155 | 1.62 (1.33-1.97) | 1.26 (1.04-1.53) |
| Pregabalin | <6 | 46 | 0 | 108 | 154 | 2.88 (2.59-3.21) | 2.25 (2.02-2.51) |
| Propranolol | <6 | 47 | 6 | 101 | 154 | 2.85 (2.56-3.18) | 2.33 (2.06-2.62) |
| Nicotine/cotinine | <6 | 115 | 0 | 39 | 154 | 1.02 (0.78-1.34) | 1.09 (0.84-1.42) |
| Ketamine | 6 | 112 | 0 | 36 | 154 | 0.94 (0.70-1.25) | 0.86 (0.65-1.14) |
| Valproic acid | <6 | 80 | <6 | 63 | 143 | 1.79 (1.48-2.15) | 1.69 (1.40-2.04) |
| Acetone | 23 | 67 | 0 | 48 | 138 | 1.40 (1.12-1.77) | 1.35 (1.09-1.67) |
| Paliperidone | 27 | 90 | 0 | 19 | 136 | 0.56 (0.37-0.85) | 0.60 (0.39-0.91) |
| Chlorpheniramine | <6 | 85 | 0 | 47 | 132 | 1.44 (1.14-1.81) | 1.32 (1.06-1.64) |
| 11-Nor-delta-9-carboxy- tetrahydrocannabinold | 18 | 86 | 0 | 28 | 132 | 0.85 (0.61-1.19) | 1.04 (0.75-1.45) |
| Nitrazepam | <6 | 50 | <6 | 80 | 130 | 2.51 (2.18-2.89) | 1.81 (1.56-2.10) |
| Lamotrigine | <6 | 64 | 0 | 63 | 127 | 2.01 (1.69-2.41) | 1.73 (1.45-2.05) |
| Lorazepam | <6 | 64 | 0 | 55 | 119 | 1.87 (1.54-2.28) | 1.66 (1.36-2.02) |
| Atenolol | 10 | 64 | 0 | 46 | 120 | 1.55 (1.23-1.95) | 1.11 (0.88-1.40) |
| Benzoylecgonine | <6 | 98 | 0 | 18 | 116 | 0.62 (0.41-0.95) | 0.87 (0.57-1.33) |
| Paroxetine | 7 | 72 | 0 | 37 | 116 | 1.29 (0.98-1.68) | 1.14 (0.87-1.49) |
| 7-Aminonitrazepam | 0 | 44 | 0 | 65 | 109 | 2.43 (2.07-2.84) | 1.77 (1.50-2.09) |
| Nortriptylined,e | <6 | 35 | 0 | 32 | 67 | 1.93 (1.50-2.48) | 1.58 (1.22-2.04) |
| Unduplicated count of decedents with any substance detected | 2728 | 5722 | 389 | 2891 | 11 730 | NA | NA |
Abbreviations: aPR, adjusted prevalence ratio; NA, not applicable; PR, prevalence ratio.
The substance categories are not mutually exclusive (ie, multiple toxicology findings from 1 person can contribute to multiple substance lines).
Alcohol is only included if ethanol level is greater than or equal to 0.03 g/100 mL.
Values less than 6 are shown as <6 to comply with ethical approvals; the total, prevalence ratio calculations, and adjusted prevalence ratio (from modified Poisson regression model) have treated these <6 values as 0.
Metabolites were excluded if they were found in the same decedent with a parent drug (specifically: desmethylvenlafaxine if found with venlafaxine; morphine if found with codeine; oxazepam if found with diazepam or temazepam; temazepam if found with diazepam; nordiazepam if found with diazepam; 7-aminoclonazepam if found with clonazepam; 11-nor-delta-9-carboxy-tetrahydrocannabinol if found with delta-9-tetrahydrocannabinol; 7-aminonitrazepam if found with nitrazepam; and nortriptyline if found with amitriptyline).
Nortriptyline is not the next most common substance but has been included as an additional row because it was included in our subsequent concentration analysis.
Substances including citalopram, nordiazepam, sertraline, and midazolam were equally prevalent in poisoning- and nonpoisoning-related suicides (with 95% CI for PR crossing 1). Alcohol was almost equally prevalent in poisoning- and nonpoisoning-related suicides (PR, 1.07; 95% CI, 1.01-1.14). There was a median (IQR) of 4 (2-6) substances detected in poisoning-related suicides and 2 (1-3) substances detected in nonpoisoning-related suicides (Table 2).
Table 2. Number of Substances Detected in Decedents Where Toxicology Was Performed in Poisoning- and Nonpoisoning-Related Suicides.
| No. of substances detecteda | No. (%) | Total | |
|---|---|---|---|
| Nonpoisoning-related death | Poisoning-related death | ||
| 0 | 1817 (94) | 117 (6)b | 1934 |
| 1 | 2728 (88) | 389 (12) | 3117 |
| 2 | 2234 (80) | 546 (20) | 2780 |
| 3 | 1496 (74) | 534 (26) | 2030 |
| 4 | 888 (63) | 519 (37) | 1407 |
| 5 | 499 (54) | 420 (46) | 919 |
| 6 | 291 (50) | 287 (50) | 578 |
| 7 | 174 (43) | 230 (57) | 404 |
| 8 | 84 (35) | 154 (65) | 238 |
| 9 | 30 (27) | 83 (73) | 113 |
| 10 | 13 (23) | 43 (77) | 56 |
| >10 | 13 (15) | 75 (85) | 88 |
| Total | 10 267 (75) | 3397 (25) | 13 664 |
| Mean (SD) | 2.14 (1.85) | 4.09 (2.60) | 2.62 (2.23)c |
| Median (IQR) | 2 (1-3) | 4 (2-6) | 2 (1-4)c |
The count of substances excludes alcohol (when ethanol <0.03 g/100 mL) and any metabolites that were present in conjunction with their parent drug (exclusion rules for metabolites in eTable 1 in Supplement 1).
The coroner may still rule a death as poisoning-related when substances are not detected through standard toxicology testing based on circumstantial or other evidence. Failures to detect a specific substance in toxicology can be due to delays in testing, poor quality samples, or lack of existing technology.
Overall value (ie, for the 13 664 decedents).
Table 329 shows the most common substance classes detected postmortem. The list of substances in each substance class is available in eTable 5 in Supplement 1. Antidepressants, alcohol, benzodiazepines, nonopioid analgesics/anti-inflammatory drugs, opioids, and cannabinoids were the most common classes. PRs for common medicine classes being detected in poisoning-related suicides compared with nonpoisoning-related suicides were as follows: antidepressants (PR, 1.63; 95% CI, 1.54-1.73), benzodiazepines (PR, 2.01; 95% CI, 1.90-2.13), nonopioid analgesics/anti-inflammatory drugs (PR, 1.88; 95% CI, 1.78-2.00), and opioids (PR, 2.72; 95% CI, 2.58-2.87). The highest PRs for substance classes to be detected in poisoning-related suicides were reported for nontherapeutic gases and chemicals, opioids, antiepileptics, central nervous system agents, gastrointestinal medicines, and sedatives (z drugs; PR range, 2.36-3.10) (Table 3).29 Cannabinoids, amphetamine-type substances, and stimulants were detected more often in nonpoisoning-related suicides.
Table 3. Frequency of All Substance Classes Detected in Descending Order.
| Substance classa | Nonpoisoning-related suicide, No. (%) | Poisoning-related suicide, No. (%) | Total | PR (95% CI) to be detected in poisoning compared with nonpoisoning suicide | aPR (95% CI), adjusted for age and sex |
|---|---|---|---|---|---|
| Antidepressant | 3105 (67) | 1557 (33) | 4662 | 1.63 (1.54-1.73) | 1.40 (1.32-1.48) |
| Alcoholb | 3191 (74) | 1127 (26) | 4318 | 1.07 (1.01-1.14) | 1.19 (1.12-1.26) |
| Benzodiazepine | 2606 (62) | 1605 (38) | 4211 | 2.01 (1.90-2.13) | 1.71 (1.61-1.81) |
| Nonopioid analgesic/ anti-inflammatories |
1748 (61) | 1139 (39) | 2887 | 1.88 (1.78-2.00) | 1.57 (1.48-1.67) |
| Opioid | 1274 (49) | 1323 (51) | 2597 | 2.72 (2.58-2.87) | 2.35 (2.22-2.48) |
| Cannabinoid | 1628 (83) | 344 (17) | 1972 | 0.67 (0.60-0.74) | 0.87 (0.78-0.96) |
| Antipsychotic | 1134 (63) | 655 (37) | 1789 | 1.59 (1.48-1.70) | 1.48 (1.38-1.58) |
| Nontherapeutic gases and chemicals | 542 (37) | 922 (63) | 1464 | 3.10 (2.94-3.27) | 3.04 (2.88-3.21) |
| Amphetamine | 1184 (82) | 252 (18) | 1436 | 0.68 (0.61-0.77) | 0.87 (0.78-0.98) |
| Cardiovascular | 769 (58) | 566 (42) | 1335 | 1.85 (1.72-1.98) | 1.44 (1.34-1.55) |
| Antihistamine | 549 (59) | 376 (41) | 925 | 1.71 (1.58-1.86) | 1.50 (1.38-1.62) |
| Antiepileptic | 359 (40) | 534 (60) | 893 | 2.67 (2.51-2.84) | 2.22 (2.08-2.36) |
| Gastrointestinal (including antiemetics) | 230 (41) | 329 (59) | 559 | 2.51 (2.33-2.71) | 1.98 (1.84-2.14) |
| Anesthetic | 320 (71) | 128 (29) | 448 | 1.16 (0.99-1.34) | 1.11 (0.96-1.27) |
| Sedative (z drugs) | 144 (43) | 190 (57) | 334 | 2.36 (2.14-2.60) | 1.93 (1.74-2.13) |
| Anti-infective | 193 (59) | 132 (41) | 325 | 1.66 (1.45-1.90) | 1.31 (1.16-1.49) |
| Hypoglycemic | 151 (55) | 122 (45) | 273 | 1.83 (1.60-2.09) | 1.45 (1.26-1.66) |
| Cough and cold medicine | 160 (67) | 78 (33) | 238 | 1.33 (1.10-1.59) | 1.33 (1.12-1.59) |
| Stimulant | 163 (80) | 41 (20) | 204 | 0.81 (0.61-1.06) | 1.01 (0.77-1.32) |
| Other central nervous system agents | 57 (35) | 106 (65) | 163 | 2.67 (2.37-3.00) | 2.25 (2.00-2.54) |
| Miscellaneous | 67 (54) | 56 (46) | 123 | 1.85 (1.52-2.24) | 1.73 (1.43-2.09) |
Abbreviations: aPR, adjusted prevalence ratio; PR, prevalence ratio.
The substance class profile categories are not mutually exclusive (ie, multiple toxicology findings from one person can contribute to multiple substance class lines). Multiple substances may be included in a single substance class. Substance classes were defined by the Anatomical Therapeutic Chemical Classifications according to the anatomic main group (Anatomical Therapeutic Chemical first level).29 We used the pharmacological subgroup (Anatomical Therapeutic Chemical third level) to distinguish psychotropic classes of interest.29
Alcohol is only included if ethanol level is greater than or equal to 0.03 g/100 mL. Other types of alcohol (eg, methanol, isopropyl alcohol) were classified as nontherapeutic gases and chemicals.
Alcohol (as ethanol ≥0.03 g/100 mL) was almost equally prevalent in poisoning- and nonpoisoning-related deaths (PR, 1.07; 95% CI, 1.01-1.14), whereas amphetamines (PR, 0.68; 95% CI, 0.61-0.77) and cannabinoids (PR, 0.67; 95% CI, 0.60-0.74) were detected more often in nonpoisoning-related suicides.
Concentrations of the Selected Substances
Table 4 shows the median and P90 values for concentrations of substances detected postmortem. Median concentration was higher in the poisoning-related suicide subgroup than in the nonpoisoning-related subgroup for most substances, except for midazolam (0.04 mg/L vs 0.08 mg/L), ibuprofen (5 mg/L vs 5 mg/L), THC (0.007 mg/L vs 0.008 mg/L), and alcohol (0.12 g/100 mL vs 0.13 g/100 mL) (Table 4). Substances with the highest rate of observations above P90 being poisoning-related included amitriptyline and nortriptyline (100%), tramadol (100%), quetiapine (98%), oxycodone (96%), and codeine (95%).
Table 4. Median and 90th Percentile (P90) Values for Concentrations of the Substances of Interest, by Classa.
| Substance detected | Median concentration (nonpoisoning-related suicides), mg/L | Median concentration (poisoning-related suicides), mg/L | P90 concentration (all suicides), mg/L | Percentage of observations above P90 that were poisoning related, % |
|---|---|---|---|---|
| Antidepressants and antipsychotics | ||||
| Amitriptyline | 0.06 | 0.785 | 3.6 | 100 |
| Nortriptylinea | 0.055 | 0.7 | 3.4 | 100 |
| Desmethylvenlafaxineb | 0.4 | 0.51 | 1.7 | 79 |
| Venlafaxine | 0.4 | 1.2 | 2.6 | 92 |
| Citalopram | 0.15 | 0.34 | 0.8 | 70 |
| Fluoxetine | 0.3 | 0.42 | 1.3 | 63 |
| Mirtazapine | 0.09 | 0.2 | 0.5 | 85 |
| Sertraline | 0.18 | 0.25 | 0.7 | 39 |
| Olanzapine | 0.08 | 0.17 | 0.34 | 73 |
| Quetiapine | 0.05 | 0.77 | 4.7 | 98 |
| Benzodiazepines and doxylamine | ||||
| Alprazolam | 0.03 | 0.07 | 0.22 | 47 |
| Diazepam | 0.09 | 0.14 | 0.5 | 51 |
| Midazolam | 0.08 | 0.04 | 0.6 | 11 |
| Oxazepamb | 0.12 | 0.38 | 1.8 | 67 |
| Temazepamb | 0.05 | 0.2 | 0.8 | 72 |
| Doxylamine | 0.09 | 0.3 | 1 | 58 |
| Nonopioid analgesics and opioids | ||||
| Ibuprofen | 5 | 5 | 15.5 | 57 |
| Paracetamol | 5 | 10 | 56 | 76 |
| Codeinec | 0.06 | 0.19 | 1.3 | 95 |
| Morphineb,c | 0.12 | 0.2 | 1.1 | 85 |
| Oxycodone | 0.04 | 0.22 | 0.97 | 96 |
| Tramadol | 0.47 | 1.2 | 8.6 | 100 |
| Illicit drugs and nontherapeutic substances | ||||
| Delta-9-tetrahydrocannabinol | 0.008 | 0.007 | 0.034 | 16 |
| Methylamphetamine | 0.16 | 0.2 | 0.77 | 28 |
| Alcohold | 0.13 g/100 mL | 0.12 g/100 mL | 0.24 g/100 mL | 21 |
| Carbon monoxide | 5.5 g/100 mL | 11 g/100 mL | 77 g/100 mL | 98 |
| Carbon monoxide (unpreserved blood) | 5 g/100 mL | 61 g/100 mL | 85 g/100 mL | 100 |
Results are from postmortem femoral blood, with the exception of unpreserved blood for carbon monoxide. P90 may be interpreted with the context that 25% of all suicides were poisoning related.
Metabolites were excluded if they were found in the same decedent with a parent drug (specifically: nortriptyline if found with amitriptyline; desmethylvenlafaxine if found with venlafaxine; morphine if found with codeine; oxazepam if found with diazepam or temazepam; and temazepam if found with diazepam).
Measurements taken for the unconjugated (free) forms of codeine and morphine.
Alcohol is only included if ethanol level is greater than or equal to 0.03 g/100 mL.
We visualized the distribution of concentrations for each substance (eFigures 2-7 in Supplement 1). Concentrations were generally higher in the poisoning-suicide subgroup for amitriptyline, nortriptyline, desmethylvenlafaxine, and venlafaxine (eFigure 2 in Supplement 1), quetiapine (eFigure 3 in Supplement 1), oxazepam and temazepam (eFigure 4 in Supplement 1), and codeine, oxycodone, paracetamol, and tramadol (eFigure 5 in Supplement 1). This difference was also observed for carbon monoxide when measurements were taken from unpreserved blood (eFigure 7 in Supplement 1). For illicit drugs, both THC and methylamphetamine showed similar concentrations in poisoning-related and nonpoisoning-related suicides (eFigure 6 in Supplement 1). The violin plots for most substances were unimodal with a symmetric appearance. However, this was a log (base 10) scale and represents that the concentration values were positively skewed with most measurements occurring at lower values.
Substance Class Combinations
The 30 most common substance class profiles and combinations are shown in the UpSet plot (Figure).27,28 This represents 45% (4601) of nonpoisoning-related suicides and 25% (836) of poisoning-related suicides. The presence of nontherapeutic gases and chemicals, opioids, or opioids in combination, were associated with a higher proportion of poisoning-related suicides. Profiles involving 1 or multiple psychotropics were common, with the usage of an antidepressant plus benzodiazepine being more common than the usage of a benzodiazepine alone.
Figure. UpSet Plot27,28 of the 30 Most Common Substance Class Profiles Detected In Descending Order.
A, The horizontal bar chart indicates the total number of cases in which each substance class was detected. B, UpSet plot. The vertical bar chart indicates how many times each substance class profile (single or multiple classes) was detected. Substance classes listed in the matrix are indicated as being present in each set if the circle for the corresponding class was filled in. Multiple filled-in circles connected by a line indicates a combination with multiple substance classes being present. The counts of each substance class profile used to make this figure are available in eTable 4 in Supplement 1. A small number of anti-inflammatory medicines for immune conditions were included in the nonopioid analgesic class.
aValues less than 6 are hidden to comply with ethical approvals, with some poisoning-related suicide subgroups being less than 6; in this case, the intersection size shows the total from nonpoisoning-related suicides only, whereas percentage as poisoning-related suicides is also hidden.
Discussion
Our cross-sectional study has provided a comprehensive overview of all substances detected at autopsy in suicides in Australia, 2013 to 2019. Taken together, our data on PRs and concentrations suggest several potential targets for means restriction and/or more cautious prescribing to at-risk individuals. These include oxycodone, codeine, amitriptyline, pentobarbitone, pregabalin, propranolol, quetiapine, venlafaxine, and tramadol; all of which have either high PRs and/or a high proportion of concentrations above the P90 in poisoning suicides. Detailed toxicologic information such as that presented here is not routinely included in suicide surveillance systems, which limits the potential to select effective substances for means restriction.
Our study has concentration information to suggest that many medicines are present at lower levels. This implies background use or presence as an innocent bystander when present at the time of suicide. However, medicines can still have contributory or psychomodulating effects at low and therapeutic concentrations.
Poisoning-related suicides involving medicines were more likely to involve opioids and sedating psychotropics, including analgesics, anxiolytics, and antidepressants. Notably for antidepressants, tricyclic antidepressants (TCAs) were more likely to be associated with poisoning-related suicides, especially at high concentrations, whereas selective serotonin reuptake inhibitors (SSRIs) were distributed similarly among poisoning- and nonpoisoning-related suicides. These findings are consistent with case fatality studies, although we cannot prove a direct causation between substances present and fatal toxicity.7,12,13,30 In poisoning suicides with nontherapeutic gases or chemicals, alcohol or antidepressants were most commonly codetected.11,31
Nonpoisoning-related suicides were more likely to contain amphetamines, cannabinoids, and SSRIs at higher concentrations. Illicit drugs are often associated with violent, nonpoisoning suicide methods.32 The presence of amphetamines and stimulants may be a confounding factor as they counter respiratory depressants and could reduce the lethality of some polydrug mixes, depending on the effects of each substance. Alcohol can also exacerbate suicide risk during a mental health crisis, and this was relevant for both poisoning- and nonpoisoning-related suicides.20 This further emphasizes the importance of targeting alcohol exposure as a modifiable risk factor for suicide, as well as considering how alcohol can compound the lethality of suicide methods.20 Medical marijuana has been legal in Australia since 2016, with specific access criteria for prescriptions, whereas recreational marijuana was illegal during the study period.33
Demographic factors are known to influence suicide risk and choice of suicide method.30 aPRs for age and sex made the differences in PR less apparent or moved aPR closer to 1, which may reflect that female individuals choose poisoning as a method of suicide more commonly than male individuals and the varied use of medicines and substances throughout the life span.30
Specific substances and medicine profiles should be monitored (especially when these are commonly used for poisoning), are toxic at high concentrations, or have additive toxicity when misused intentionally.11 Up-scheduling medicines to prescription only and legislating stricter therapeutic indications can reduce access to toxic medicines. On a patient level, real-time prescription monitoring has been implemented in many parts of the world and should be used by health care practitioners to help reduce problematic medicine misuse.34 Patients with identified suicide risks should be prescribed the less toxic agents as first-line treatment (ie, SSRIs in preference to TCAs) or treatments such as long-acting antipsychotic depot injections, which remove the means for self-poisoning. Other interventions include safe storage, packaging (eg, blister strips), and disposal of unnecessary/unwanted medicines to reduce the lethality of impulsive overdose suicides.35,36
Strengths and Limitations
A strength of our study is the robust rules for metabolite coding, which remove all possible metabolites if found with a parent drug. However, this assumes that multiple medicines on the same metabolite pathway were not concomitantly prescribed or taken together. Thus, we may be underestimating the involvement of certain drugs that may be taken together, eg, temazepam and diazepam. An additional strength is that our study can distinguish exact agents. This is in comparison to studies that rely on ICD-10 coding, which combines many drugs of interest into unhelpful classes with differing toxicity (eg, fentanyl and tramadol share an ICD-10 code). Knowing which substances have been taken (rather than broader classes) provides useful information for targeting means restriction.
This study has some limitations. There were specific substances that were not routinely analyzed. Medicines including lithium and pregabalin were not routinely tested in every state or territory, whereas substances such as nitrogen gas could not be analyzed postmortem. These testing limitations explain why some poisoning-related suicides (n = 117) had no substances detected. Different substances remain in the body for different durations after exposure. For example, a regular cannabinoid user can test positive for approximately 2 to 3 weeks after the last use.32 Conversely, medicines such as buprenorphine and olanzapine break down rapidly after death and can fall below the detection limit.37 Concentrations can also decrease after poisoning occurs, eg, for acetaminophen/paracetamol where death from hepatotoxicity is delayed and concentrations can become undetectable by the time of death.38,39
We relied on using coroners’ judgment to code a suicide as poisoning related. It is possible that substances that are not lethal in overdose have not been coded as poisoning related, even if they were taken with suicidal intent. Importantly, a key strength is that we looked at all drugs present in toxicologic findings, rather than the coroners’ list of drugs determined to have caused the death. This minimized the potential impact of coroners not coding a substance in their findings.
We did not have a control population that was exposed to the same medicines and did not die by suicide. Causal links between substances and suicidality should not be interpreted from these data, especially for psychotropic medicines where the association between medicines for mental health conditions and suicide is complicated.2,40 Drugs that appeared frequently in the suicide cohort are also commonly used in the general population, especially over-the-counter medicines.
Concentrations in single-agent poisoning-related suicides are predictive of the concentration threshold that would be solely responsible for a poisoning death.14 However, in compliance with our ethics approval, we could not report these concentrations if the subgroup count was less than 6. This small sample size of single-agent overdoses is also a limitation reported in other postmortem toxicology studies.14,16
Conclusions
In this cross-sectional study, one-quarter of Australian suicides were poisoning related. Both poisoning- and nonpoisoning-related suicides reported a high prevalence of psychotropic medicines and/or potential intoxication, and the former emphasizes the association between mental health conditions and suicide. Findings suggest that substances frequently involved in poisoning-related suicides should be prescribed cautiously or be monitored, especially in combinations, including TCAs, sedatives, opioids, and nontherapeutic gases.
eMethods
eTable 1. Metabolite and Parent Drug Coding Rules
eTable 2. Comparison of Age and Sex in Toxicology Performed and Toxicology Not Performed
eTable 3. Comparison of Demographic Variables Based on Poisoning or Nonpoisoning-Related Suicides Where Toxicology Was Performed
eTable 4. Frequency of the 30 Most Common Substance Classes Profiles and Combinations in Descending Order, Used To Create the Upset Plot
eTable 5. List of Substances in Each Substance Class
eFigure 1. Comparison of the Total Number of Annual Suicide Cases Reported in Australia vs the Number of Annual Suicide Cases Included in the Ashli Study
eTable 6. Table From the Ashli Protocol, Published by Chitty et al 2020, BMJ Open
eFigure 2. Violin Plot (Log Base 10 Scale) of Concentrations of Antidepressants Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 3. Violin Plot (Log Base 10 Scale) of Concentrations of Antipsychotics Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 4. Violin Plot (Log Base 10 Scale) of Concentrations of Benzodiazepines and Doxylamine Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 5. Violin Plots (Log Base 10 Scale) of Concentrations of Nonopioid Analgesics and Opioids Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 6. Violin Plot (Log Base 10 Scale) of Concentrations of Illicit Drugs Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 7. Violin Plot (Log Base 10 Scale) of Concentrations of Carbon Monoxide (Detected in Femoral and Unpreserved Blood) and Alcohol (as Ethanol≥0.03 g/100 mL) in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eAppendix. STROBE Statement
eReferences
Data Sharing Statement
References
- 1.World Health Organization . Suicide: key facts. Accessed August 24, 2022. https://www.who.int/news-room/fact-sheets/detail/suicide
- 2.Zalsman G, Hawton K, Wasserman D, et al. Suicide prevention strategies revisited: 10-year systematic review. Lancet Psychiatry. 2016;3(7):646-659. doi: 10.1016/S2215-0366(16)30030-X [DOI] [PubMed] [Google Scholar]
- 3.Sinyor M, Howlett A, Cheung AH, Schaffer A. Substances used in completed suicide by overdose in Toronto: an observational study of coroner’s data. Can J Psychiatry. 2012;57(3):184-191. doi: 10.1177/070674371205700308 [DOI] [PubMed] [Google Scholar]
- 4.Toxicology testing and results for suicide victims—13 states, 2004. JAMA. 2007;297(4):355-356. doi: 10.1001/jama.297.4.355 [DOI] [PubMed] [Google Scholar]
- 5.Fountain JS, Reith DM, Tomlin AM, Smith AJ, Tilyard MW. Deaths by poisoning in New Zealand, 2008-2013. Clin Toxicol (Phila). 2019;57(11):1087-1094. doi: 10.1080/15563650.2019.1582777 [DOI] [PubMed] [Google Scholar]
- 6.Chrzanowska A, Man N, Darke S, et al. Unintentional and intentional drug poisoning deaths, Australia, 2012-2016: drug pattern profile and demographic characteristics. Drug Alcohol Depend. 2021;229(Pt B):109112-109112. doi: 10.1016/j.drugalcdep.2021.109112 [DOI] [PubMed] [Google Scholar]
- 7.Brett J, Wylie CE, Raubenheimer J, Isbister GK, Buckley NA. The relative lethal toxicity of pharmaceutical and illicit substances: a 16-year study of the Greater Newcastle Hunter Area, Australia. Br J Clin Pharmacol. 2019;85(9):2098-2107. doi: 10.1111/bcp.14019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Albano GD, Malta G, La Spina C, et al. Toxicological findings of self-poisoning suicidal deaths: a systematic review by countries. Toxics. 2022;10(11):654. doi: 10.3390/toxics10110654 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Valipour R, Shekari A, Setareh M, Soltaninejad K. Pattern of suicide methods and postmortem toxicological findings in suicide-related deaths: a retrospective 7-year forensic-based study in Iran. Am J Forensic Med Pathol. 2021;42(1):23-29. doi: 10.1097/PAF.0000000000000607 [DOI] [PubMed] [Google Scholar]
- 10.World Health Organization . LIVE LIFE: an implementation guide for suicide prevention in countries. Accessed April 18, 2023. https://www.who.int/publications/i/item/9789240026629
- 11.Lim JS, Buckley NA, Chitty KM, Moles RJ, Cairns R. Association between means restriction of poison and method-specific suicide rates: a systematic review. JAMA Health Forum. 2021;2(10):e213042-e213042. doi: 10.1001/jamahealthforum.2021.3042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Miller TR, Swedler DI, Lawrence BA, et al. Incidence and lethality of suicidal overdoses by drug class. JAMA Netw Open. 2020;3(3):e200607-e200607. doi: 10.1001/jamanetworkopen.2020.0607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Daly C, Griffin E, Corcoran P, et al. A national case fatality study of drugs taken in intentional overdose. Int J Drug Policy. 2020;76:102609. doi: 10.1016/j.drugpo.2019.102609 [DOI] [PubMed] [Google Scholar]
- 14.Jones AW, Holmgren A, Ahlner J. Postmortem concentrations of drugs determined in femoral blood in single-drug fatalities compared with multidrug poisoning deaths. Forensic Sci Int. 2016;267:96-103. doi: 10.1016/j.forsciint.2016.08.015 [DOI] [PubMed] [Google Scholar]
- 15.Australian Institute of Health and Welfare . Deaths by suicide over time. Accessed May 24, 2022. https://www.aihw.gov.au/suicide-self-harm-monitoring/data/deaths-by-suicide-in-australia/suicide-deaths-over-time
- 16.Lefrancois E, Reymond N, Thomas A, Lardi C, Fracasso T, Augsburger M. Summary statistics for drugs and alcohol concentration recovered in postmortem femoral blood in Western Switzerland. Forensic Sci Int. 2021;325:110883. doi: 10.1016/j.forsciint.2021.110883 [DOI] [PubMed] [Google Scholar]
- 17.Chitty KM, Schumann JL, Schaffer A, et al. Australian Suicide Prevention Using Health-Linked Data (ASHLi): protocol for a population-based case series study. BMJ Open. 2020;10(5):e038181. doi: 10.1136/bmjopen-2020-038181 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Geulayov G, Ferrey A, Casey D, et al. Relative toxicity of benzodiazepines and hypnotics commonly used for self-poisoning: an epidemiological study of fatal toxicity and case fatality. J Psychopharmacol. 2018;32(6):654-662. doi: 10.1177/0269881118754734 [DOI] [PubMed] [Google Scholar]
- 19.Kugelberg FC, Jones AW. Interpreting results of ethanol analysis in postmortem specimens: a review of the literature. Forensic Sci Int. 2007;165(1):10-29. doi: 10.1016/j.forsciint.2006.05.004 [DOI] [PubMed] [Google Scholar]
- 20.Chong DG, Buckley NA, Schumann JL, Chitty KM. Acute alcohol use in Australian coronial suicide cases, 2010-2015. Drug Alcohol Depend. 2020;212:108066. doi: 10.1016/j.drugalcdep.2020.108066 [DOI] [PubMed] [Google Scholar]
- 21.Boston University School of Public Health . Confidence intervals: summary module. Accessed October 18, 2022. https://sphweb.bumc.bu.edu/otlt/mph-modules/bs/bs704_confidence_intervals/BS704_Confidence_Intervals11.html
- 22.Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159(7):702-706. doi: 10.1093/aje/kwh090 [DOI] [PubMed] [Google Scholar]
- 23.Fang J. Using SAS procedures FREQ, GENMOD, LOGISTIC, and PHREG to estimate adjusted relative risks—a case study. Accessed April 11, 2023. https://support.sas.com/resources/papers/proceedings11/345-2011.pdf
- 24.Matange S. SAS violin plots. Accessed April 11, 2023. https://blogs.sas.com/content/graphicallyspeaking/2012/10/30/violin-plots
- 25.Hintze JL, Nelson RD. Violin plots: a box plot-density trace synergism. Am Stat. 1998;52(2):181-184. doi: 10.1080/00031305.1998.10480559 [DOI] [Google Scholar]
- 26.Härdle W, Werwatz A, Müller M, Sperlich S.. Nonparametric and Semiparametric Models. Springer Berlin Heidelberg; 2004. [Google Scholar]
- 27.Sarkar D. UpSet plot using GTL. Accessed December 15, 2021. https://blogs.sas.com/content/graphicallyspeaking/2021/02/08/upset-plot-using-gtl
- 28.Lex A, Gehlenborg N, Strobelt H, Vuillemot R, Pfister H. UpSet: visualization of intersecting sets. IEEE Trans Vis Comput Graph. 2014;20(12):1983-1992. doi: 10.1109/TVCG.2014.2346248 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.World Health Organization . Anatomical Therapeutic Chemical (ATC) classification. Accessed October 18, 2022. https://www.who.int/tools/atc-ddd-toolkit/atc-classification
- 30.Spicer RS, Miller TR. Suicide acts in 8 states: incidence and case fatality rates by demographics and method. Am J Public Health. 2000;90(12):1885-1891. doi: 10.2105/AJPH.90.12.1885 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Burnett ACR, Chen NA, McGillivray L, Larsen ME, Torok M. Surveillance of suicide deaths involving gases in Australia using the National Coronial Information System, 2006 to 2017. Aust N Z J Public Health. 2021;45(3):242-247. doi: 10.1111/1753-6405.13087 [DOI] [PubMed] [Google Scholar]
- 32.Drummer OH. Postmortem toxicology of drugs of abuse. Forensic Sci Int. 2004;142(2-3):101-113. doi: 10.1016/j.forsciint.2004.02.013 [DOI] [PubMed] [Google Scholar]
- 33.Arnold JCNT, Nation T, McGregor IS. Prescribing medicinal cannabis. Aust Prescr. 2020;43(5):152-159. doi: 10.18773/austprescr.2020.052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Dobbin M, Liew DF. Real-time prescription monitoring: helping people at risk of harm. Aust Prescr. 2020;43(5):164-167. doi: 10.18773/austprescr.2020.050 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Buckley NA, Newby DA, Dawson AH, Whyte IM. The effect of the introduction of safety packaging for carbamazepine on toxicity in overdose in adults. Pharmacoepidemiol Drug Saf. 1995;4(6):351-354. doi: 10.1002/pds.2630040606 [DOI] [Google Scholar]
- 36.Barber CW, Miller MJ. Reducing a suicidal person’s access to lethal means of suicide: a research agenda. Am J Prev Med. 2014;47(3)(suppl 2):S264-S272. doi: 10.1016/j.amepre.2014.05.028 [DOI] [PubMed] [Google Scholar]
- 37.Peters FT, Steuer AE. Antemortem and postmortem influences on drug concentrations and metabolite patterns in postmortem specimens. WIREs Forensic Sci. 2019;1(1):e1297. doi: 10.1002/wfs2.1297 [DOI] [Google Scholar]
- 38.Baselt RC. Disposition of Toxic Drugs and Chemicals in Man. 8th ed. Biomedical Publications; 2008. [Google Scholar]
- 39.Bateman DN. Paracetamol poisoning: beyond the nomogram. Br J Clin Pharmacol. 2015;80(1):45-50. doi: 10.1111/bcp.12604 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Sinyor M, Knipe D, Borges G, et al. ; International COVID-19 Suicide Prevention Research Collaboration . Suicide risk and prevention during the COVID-19 pandemic: 1 year on. Arch Suicide Res. 2022;26(4):1944-1949. doi: 10.1080/13811118.2021.1955784 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eMethods
eTable 1. Metabolite and Parent Drug Coding Rules
eTable 2. Comparison of Age and Sex in Toxicology Performed and Toxicology Not Performed
eTable 3. Comparison of Demographic Variables Based on Poisoning or Nonpoisoning-Related Suicides Where Toxicology Was Performed
eTable 4. Frequency of the 30 Most Common Substance Classes Profiles and Combinations in Descending Order, Used To Create the Upset Plot
eTable 5. List of Substances in Each Substance Class
eFigure 1. Comparison of the Total Number of Annual Suicide Cases Reported in Australia vs the Number of Annual Suicide Cases Included in the Ashli Study
eTable 6. Table From the Ashli Protocol, Published by Chitty et al 2020, BMJ Open
eFigure 2. Violin Plot (Log Base 10 Scale) of Concentrations of Antidepressants Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 3. Violin Plot (Log Base 10 Scale) of Concentrations of Antipsychotics Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 4. Violin Plot (Log Base 10 Scale) of Concentrations of Benzodiazepines and Doxylamine Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 5. Violin Plots (Log Base 10 Scale) of Concentrations of Nonopioid Analgesics and Opioids Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 6. Violin Plot (Log Base 10 Scale) of Concentrations of Illicit Drugs Detected in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eFigure 7. Violin Plot (Log Base 10 Scale) of Concentrations of Carbon Monoxide (Detected in Femoral and Unpreserved Blood) and Alcohol (as Ethanol≥0.03 g/100 mL) in Femoral Blood, in Poisoning and Nonpoisoning-Related Suicide Subgroups
eAppendix. STROBE Statement
eReferences
Data Sharing Statement