Suicide by chemical overdose in New Zealand has been said to be both increasing1 and decreasing,2 with this discrepancy explained at least in part by different definitions of these deaths. This paper utilised the Chemical Injury Surveillance System definition, which included inhalation or ingestion of chemicals and self-immolation or dousing with petrol or other hydrocarbons followed by lighting one's self on fire (Ministry of Health and the Institute for Environmental Science and Research, Ltd (ESR) as described in McDowell et al1).This is the first New Zealand report of suicide deaths attributed to chemical substances where the presence of the suspected chemicals was verified by toxicologists with access to all records (including Pathology reports when applicable) reviewing detailed toxicology reports for each case.Choice of method for committing suicide is influenced by availability of material.3-7 Although little is known about risk factors for chemical suicide in New Zealand, intentional overdoses of tricyclic antidepressants (TCAs) are the most common fatal agent among prescription medicines in New Zealand.1,8,9By looking at this specific suicide method it may be possible to identify a population at risk of using toxic materials. This population may differ significantly from those being targeted by the Suicide Prevention Strategy.10 Verification of toxicology data may also provide new information on specific toxic substances requiring restricted access in New Zealand.Methods According to the 1988 Coroner's Act, all deaths in New Zealand "Without known cause, suicide or unnatural or violent" must be reported to the Coroner. All deaths resulting from violent, unnatural, unexpected, or suspicious causes from January 2001 to November 2005 and given a final verdict by November 2005 were reviewed. Suicide deaths were those deemed "Intentional, with the intent of taking one's own life" by the Coroner.11 Those determined by the coroner as "Unintentional or Undetermined" were not included. Deaths where the main cause of death was non-chemical (for example by hanging and firearms, etc.) were used to compare with those that were caused by chemical over-exposure. Chemical suicide is defined as any suicide resulting from over-exposure to a chemical. This includes carbon monoxide poisoning from car exhaust, overdose of therapeutic or recreational substances, ingestion of flammable liquids, solvents, pesticides or cyanide or self-immolation. The primary substance causing each death was initially recorded from the conclusion of the ESR toxicology report as obtained through the Coronial file. When toxicological data was not available (such as a death resulting from flaming petrol), the primary substance was taken from the Coroner's ruling on cause of death. This study did not have access to decedents' prior medical or prescription histories. It does not include non-fatal suicide attempts. The data collection method has been described in more detail previously.12 Toxicology results for primary substance resulting in death were reviewed by an ESR toxicologist (D. Kappatos) and verified according to laboratory records. Incidents were defined as resulting from a primary substance (where the primary substance could be identified), from multiple chemicals (polydrug overdose) if these were identified or "substance not determined" if none was identified. Cases that could not be verified using toxicology data were excluded from further analysis in the specific substance part of this research. Ethnicity was assigned in hierarchical order according to the Ministry of Health Ethnicity Data Protocol.13 Persons identified in the coronial database with more than one ethnicity were classified hierarchically as Māori (if Māori was one of the ethnicities); Pacific Island; Asian; other groups (except New Zealand European); and New Zealand European. Due to a time lag between events and completion of inquest/submission of records, some deaths occurring within this time period (approximately 11% according to ESR experience1) were likely to be absent. For this reason, age-standardised population rates were calculated for the three years 2001, 2002 and 2003. Population data for New Zealand was taken from the Census Year 2001, Usually Resident Population.14 SAS v9.1.3 software was used to compute statistics for comparison of chemical suicide rates by gender, age, ethnicity and employment category. The Cochran-Mantel-Haenszel correlation was used to derive relative risks. Significant differences between age groups was determined using a two-sample t-test for comparing means, while differences between genders, ethnicity and employment category were determined using the Pearson chi-square test for binomial proportions. Stata v11.1 software was used to conduct multivariate logistic regression of binary outcomes (chemical versus other suicide methods). Age and gender differences by chemical category were analysed using one-way ANOVA and t-test for independent samples with unequal variance. Results During the period 2001-2003, suicide was the ninth most common cause of death in New Zealand, accounting for 2% of all deaths.15 Among non-medical causes, suicide was second only to accidental injury, largely traffic accidents. The Ministry of Justice Coronial Services Office reported all deaths to ESR from the period 2001 to November 2005, after the completion of investigations and determination of intent of the deceased. As of November 2005 there were 2261 deaths reported as suicide to ESR. Of these, approximately 10% (219 cases) occurred in the years prior to 2001—leaving a total of 2042 cases for the relevant period (see Table 1). Table 1. All suicides in New Zealand 2001 to 2005 as recorded in November 2005 Study group Cases N (%) Average age (SD) Range Waged (%) Female (%) Māori (%) All cases occurring 2001 to 2005 2042 (100) 40.6 (17.9) 11-95 48.0 24.3 17.0 Deaths by non-chemical methods 1399 (69) 38.8a (18.0) 11-95 47.7 21.7 21.7d Deaths by chemical methods 643 (31) 44.4 (17.2) 13-92 47.1 29.9 6.7 Carbon monoxide (CO) overdose 418 (20) 42.7b (16.4) 13-91 55.5c 20.4 7.2 Other chemical methods 225 (11) 47.7 (18.0) 14-92 35.6 47.6 5.8 a. Younger than victims using chemical methods, p<0.0001 b. Younger than victims using non-CO chemical methods, p<0.001 c. More likely to be waged than victims using non-CO chemical methods, p<0.001 d. More likely to be Maori than victims using chemical methods, p<0.001 Table 1 outlines characteristics of intentional deaths during this period. The majority (69%, 1399 cases) of suicides resulted from non-chemical methods such as hanging and firearms. The remaining 643 deaths, or 31% of total suicides, resulted from chemical methods, with carbon monoxide overdose being the most common. Age—Victims who used chemical methods were five years older, on average, than those who used other methods. Age differences between chemical and non-chemical suicide were largest for Māori victims: 36 versus 29 years old, respectively (p<0.0005). Data for Pacific Island Peoples and Asian victims were too sparse to test for age differences by choice of modality. Ethnicity—When broken down by ethnicity, age-standardised suicide rates based on complete data from 2001 to 2003 show that the chemical suicide rate is considerably higher among New Zealand Europeans than all other ethnic groups (Figure 1). The suicide rate for Māori using non-chemical methods is disproportionately high, accounting for one in five of all cases of non-chemical suicide. However, this is not reflected in self-poisoning suicides, where Māori cases represent one in fifteen cases. Gender—Males outnumber females by 3:1 in all suicides. In chemical suicides, however, gender differences are more marked; carbon monoxide deaths are far more common for males than females (5:1) and non-carbon monoxide chemical deaths are nearly equal between males and females (118 males vs 107 females). Logistic regression of chemical suicide versus non-chemical means—Among all suicides, the strongest association with using chemical methods was protective Māori ethnicity, representing a risk reduction of more than 3-fold odds. Being female was also significantly associated with chemical modality, with females 1.6 times more likely to use chemicals compared with males (see Table 2). Figure 1. Age-standardised annual suicide rates for 2001-2003 by ethnicity, chemical versus non-chemical methods Table 2. Logistic regression of chemical versus non-chemical suicide N=2042 Variable Odds Ratio (95% confidence interval) P value Māori ethnicity Waged employment Female Age in years 0.31 (0.22 - 0.44) 1.19 (0.97 - 1.45) 1.64 (1.31 - 2.05) 1.01 (1.01 -1.02) 0.0001 0.07 0.0001 0.0001 Being in waged employment was positively related to chemical over-exposure compared tononchemical suicide modalities in logistic regression, although this was not statistically significant. This is primarily due to the large representation of waged victims (56%) using carbon monoxide poisoning, compared to the 36% of victims using other chemical means who were employed at the time of death. Specific substances not including carbon monoxide—One in 10 suicides (225/2042 cases) that occurred between 2001 and 2005 were the result of chemical overdose or self-immolation/burning (six events). Poisonings were the result of more than one substance in half of cases (51%) where two or more substances were detected. In approximately one in 10 cases (19 overall), more than one substance contributed to the overall death: these are listed as polydrug exposures in Table 3. Table 3. Chemicals used in suicide events, 2001 to 2005 Category of chemical substance Cases recorded from Coronial Database (N) Average age of decedents (years) Percent female (%) Cases excluded after toxicology verification (N) Antidepressants Sedatives and relaxants Analgesics Other drugs Industrial substances Polydrug overdoses Substances not determined 69 15 44 29 42 19 6 41.1* 67.9* 49.6 48.5 50.8 45.9 40.5 51% 73%* 50% 45% 26%* 58% 50% 1 2 7 2 11 2 4 Total 225 47.7 48% 29 *Statistically significant difference from all other categories using t-test for independent samples with unequal variance. Table 3 above lists the chemical categories identified from Coronial files and later verified by ESR toxicology laboratory data for 225 chemical suicides. A total of 87% of coronial records were confirmed using toxicology data. In 29 cases, chemical suicides could not be verified and these are listed below: No toxicology data and no specific chemicals from the death scene (four cases) Paracetamol overdoses (five of six cases in this dataset) sometimes lack forensic toxicology tests because overdose is confirmed by blood test at hospital admission. Paracetamol overdose is typically prolonged with characteristic symptoms. Post-mortem pathology results of the deceased reveal paracetamol-induced liver damage as further confirmation of the cause of death, rather than post-mortem toxicology results. Evidence of specific chemicals found at the death scene is sometimes recorded without toxicology data for verification. In this dataset we encountered the following examples: intentional fire using hydrocarbon solvent (eight cases), multiple chemicals (two cases) and single cases each of amitriptyline, caustic soda, clonazepam, codeine, ethylene glycol, glyphosate, insulin, meprobamate, methamphetamine and morphine or heroin. All 29 cases lacking toxicological verification were excluded from further analysis. Antidepressants were the largest single category of non-carbon monoxide (non-CO) chemicals used for suicides, particularly TCAs, which accounted for 95% of all antidepressant suicides and 33% of all non-CO chemical suicides. People using antidepressants were younger on average (41 years of age) than all other people using non-CO chemicals (51 years of age). As shown in Table 3, deaths involving sedatives and relaxants occurred among people who were older and more likely to be female, compared to all other categories of chemicals used. Conversely, those who used industrial chemicals were more likely to be male than all other chemical categories. Table 4. Chemical suicide events verified for analysis after toxicology review

Suicide by chemical overdose in New Zealand has been said to be both increasing1 and decreasing,2 with this discrepancy explained at least in part by different definitions of these deaths. This paper utilised the Chemical Injury Surveillance System definition, which included inhalation or ingestion of chemicals and self-immolation or dousing with petrol or other hydrocarbons followed by lighting one's self on fire (Ministry of Health and the Institute for Environmental Science and Research, Ltd (ESR) as described in McDowell et al1).This is the first New Zealand report of suicide deaths attributed to chemical substances where the presence of the suspected chemicals was verified by toxicologists with access to all records (including Pathology reports when applicable) reviewing detailed toxicology reports for each case.Choice of method for committing suicide is influenced by availability of material.3-7 Although little is known about risk factors for chemical suicide in New Zealand, intentional overdoses of tricyclic antidepressants (TCAs) are the most common fatal agent among prescription medicines in New Zealand.1,8,9By looking at this specific suicide method it may be possible to identify a population at risk of using toxic materials. This population may differ significantly from those being targeted by the Suicide Prevention Strategy.10 Verification of toxicology data may also provide new information on specific toxic substances requiring restricted access in New Zealand.Methods According to the 1988 Coroner's Act, all deaths in New Zealand "Without known cause, suicide or unnatural or violent" must be reported to the Coroner. All deaths resulting from violent, unnatural, unexpected, or suspicious causes from January 2001 to November 2005 and given a final verdict by November 2005 were reviewed. Suicide deaths were those deemed "Intentional, with the intent of taking one's own life" by the Coroner.11 Those determined by the coroner as "Unintentional or Undetermined" were not included. Deaths where the main cause of death was non-chemical (for example by hanging and firearms, etc.) were used to compare with those that were caused by chemical over-exposure. Chemical suicide is defined as any suicide resulting from over-exposure to a chemical. This includes carbon monoxide poisoning from car exhaust, overdose of therapeutic or recreational substances, ingestion of flammable liquids, solvents, pesticides or cyanide or self-immolation. The primary substance causing each death was initially recorded from the conclusion of the ESR toxicology report as obtained through the Coronial file. When toxicological data was not available (such as a death resulting from flaming petrol), the primary substance was taken from the Coroner's ruling on cause of death. This study did not have access to decedents' prior medical or prescription histories. It does not include non-fatal suicide attempts. The data collection method has been described in more detail previously.12 Toxicology results for primary substance resulting in death were reviewed by an ESR toxicologist (D. Kappatos) and verified according to laboratory records. Incidents were defined as resulting from a primary substance (where the primary substance could be identified), from multiple chemicals (polydrug overdose) if these were identified or "substance not determined" if none was identified. Cases that could not be verified using toxicology data were excluded from further analysis in the specific substance part of this research. Ethnicity was assigned in hierarchical order according to the Ministry of Health Ethnicity Data Protocol.13 Persons identified in the coronial database with more than one ethnicity were classified hierarchically as Māori (if Māori was one of the ethnicities); Pacific Island; Asian; other groups (except New Zealand European); and New Zealand European. Due to a time lag between events and completion of inquest/submission of records, some deaths occurring within this time period (approximately 11% according to ESR experience1) were likely to be absent. For this reason, age-standardised population rates were calculated for the three years 2001, 2002 and 2003. Population data for New Zealand was taken from the Census Year 2001, Usually Resident Population.14 SAS v9.1.3 software was used to compute statistics for comparison of chemical suicide rates by gender, age, ethnicity and employment category. The Cochran-Mantel-Haenszel correlation was used to derive relative risks. Significant differences between age groups was determined using a two-sample t-test for comparing means, while differences between genders, ethnicity and employment category were determined using the Pearson chi-square test for binomial proportions. Stata v11.1 software was used to conduct multivariate logistic regression of binary outcomes (chemical versus other suicide methods). Age and gender differences by chemical category were analysed using one-way ANOVA and t-test for independent samples with unequal variance. Results During the period 2001-2003, suicide was the ninth most common cause of death in New Zealand, accounting for 2% of all deaths.15 Among non-medical causes, suicide was second only to accidental injury, largely traffic accidents. The Ministry of Justice Coronial Services Office reported all deaths to ESR from the period 2001 to November 2005, after the completion of investigations and determination of intent of the deceased. As of November 2005 there were 2261 deaths reported as suicide to ESR. Of these, approximately 10% (219 cases) occurred in the years prior to 2001—leaving a total of 2042 cases for the relevant period (see Table 1). Table 1. All suicides in New Zealand 2001 to 2005 as recorded in November 2005 Study group Cases N (%) Average age (SD) Range Waged (%) Female (%) Māori (%) All cases occurring 2001 to 2005 2042 (100) 40.6 (17.9) 11-95 48.0 24.3 17.0 Deaths by non-chemical methods 1399 (69) 38.8a (18.0) 11-95 47.7 21.7 21.7d Deaths by chemical methods 643 (31) 44.4 (17.2) 13-92 47.1 29.9 6.7 Carbon monoxide (CO) overdose 418 (20) 42.7b (16.4) 13-91 55.5c 20.4 7.2 Other chemical methods 225 (11) 47.7 (18.0) 14-92 35.6 47.6 5.8 a. Younger than victims using chemical methods, p<0.0001 b. Younger than victims using non-CO chemical methods, p<0.001 c. More likely to be waged than victims using non-CO chemical methods, p<0.001 d. More likely to be Maori than victims using chemical methods, p<0.001 Table 1 outlines characteristics of intentional deaths during this period. The majority (69%, 1399 cases) of suicides resulted from non-chemical methods such as hanging and firearms. The remaining 643 deaths, or 31% of total suicides, resulted from chemical methods, with carbon monoxide overdose being the most common. Age—Victims who used chemical methods were five years older, on average, than those who used other methods. Age differences between chemical and non-chemical suicide were largest for Māori victims: 36 versus 29 years old, respectively (p<0.0005). Data for Pacific Island Peoples and Asian victims were too sparse to test for age differences by choice of modality. Ethnicity—When broken down by ethnicity, age-standardised suicide rates based on complete data from 2001 to 2003 show that the chemical suicide rate is considerably higher among New Zealand Europeans than all other ethnic groups (Figure 1). The suicide rate for Māori using non-chemical methods is disproportionately high, accounting for one in five of all cases of non-chemical suicide. However, this is not reflected in self-poisoning suicides, where Māori cases represent one in fifteen cases. Gender—Males outnumber females by 3:1 in all suicides. In chemical suicides, however, gender differences are more marked; carbon monoxide deaths are far more common for males than females (5:1) and non-carbon monoxide chemical deaths are nearly equal between males and females (118 males vs 107 females). Logistic regression of chemical suicide versus non-chemical means—Among all suicides, the strongest association with using chemical methods was protective Māori ethnicity, representing a risk reduction of more than 3-fold odds. Being female was also significantly associated with chemical modality, with females 1.6 times more likely to use chemicals compared with males (see Table 2). Figure 1. Age-standardised annual suicide rates for 2001-2003 by ethnicity, chemical versus non-chemical methods Table 2. Logistic regression of chemical versus non-chemical suicide N=2042 Variable Odds Ratio (95% confidence interval) P value Māori ethnicity Waged employment Female Age in years 0.31 (0.22 - 0.44) 1.19 (0.97 - 1.45) 1.64 (1.31 - 2.05) 1.01 (1.01 -1.02) 0.0001 0.07 0.0001 0.0001 Being in waged employment was positively related to chemical over-exposure compared tononchemical suicide modalities in logistic regression, although this was not statistically significant. This is primarily due to the large representation of waged victims (56%) using carbon monoxide poisoning, compared to the 36% of victims using other chemical means who were employed at the time of death. Specific substances not including carbon monoxide—One in 10 suicides (225/2042 cases) that occurred between 2001 and 2005 were the result of chemical overdose or self-immolation/burning (six events). Poisonings were the result of more than one substance in half of cases (51%) where two or more substances were detected. In approximately one in 10 cases (19 overall), more than one substance contributed to the overall death: these are listed as polydrug exposures in Table 3. Table 3. Chemicals used in suicide events, 2001 to 2005 Category of chemical substance Cases recorded from Coronial Database (N) Average age of decedents (years) Percent female (%) Cases excluded after toxicology verification (N) Antidepressants Sedatives and relaxants Analgesics Other drugs Industrial substances Polydrug overdoses Substances not determined 69 15 44 29 42 19 6 41.1* 67.9* 49.6 48.5 50.8 45.9 40.5 51% 73%* 50% 45% 26%* 58% 50% 1 2 7 2 11 2 4 Total 225 47.7 48% 29 *Statistically significant difference from all other categories using t-test for independent samples with unequal variance. Table 3 above lists the chemical categories identified from Coronial files and later verified by ESR toxicology laboratory data for 225 chemical suicides. A total of 87% of coronial records were confirmed using toxicology data. In 29 cases, chemical suicides could not be verified and these are listed below: No toxicology data and no specific chemicals from the death scene (four cases) Paracetamol overdoses (five of six cases in this dataset) sometimes lack forensic toxicology tests because overdose is confirmed by blood test at hospital admission. Paracetamol overdose is typically prolonged with characteristic symptoms. Post-mortem pathology results of the deceased reveal paracetamol-induced liver damage as further confirmation of the cause of death, rather than post-mortem toxicology results. Evidence of specific chemicals found at the death scene is sometimes recorded without toxicology data for verification. In this dataset we encountered the following examples: intentional fire using hydrocarbon solvent (eight cases), multiple chemicals (two cases) and single cases each of amitriptyline, caustic soda, clonazepam, codeine, ethylene glycol, glyphosate, insulin, meprobamate, methamphetamine and morphine or heroin. All 29 cases lacking toxicological verification were excluded from further analysis. Antidepressants were the largest single category of non-carbon monoxide (non-CO) chemicals used for suicides, particularly TCAs, which accounted for 95% of all antidepressant suicides and 33% of all non-CO chemical suicides. People using antidepressants were younger on average (41 years of age) than all other people using non-CO chemicals (51 years of age). As shown in Table 3, deaths involving sedatives and relaxants occurred among people who were older and more likely to be female, compared to all other categories of chemicals used. Conversely, those who used industrial chemicals were more likely to be male than all other chemical categories. Table 4. Chemical suicide events verified for analysis after toxicology review