Suicide and Life‐Threatening Behavior 1

© 2020 The American Association of Suicidology

DOI: 10.1111/sltb.12627

Prevalence and Characteristics Associated with

Chronic Noncancer Pain in Suicide Decedents:
A National Study
GABRIELLE CAMPBELL , SHANE DARKE, LOUISA DEGENHARDT, HARRIET TOWNSEND,
GREGORY CARTER, BRIAN DRAPER, MICHAEL FARRELL, JOHAN DUFLOU AND JULIA LAPPIN

Abstract
Objective: The aims were to estimate the prevalence of CNCP in suicide decedents,
and compare sociodemographic and clinical characteristics of people who die by
suicide (i) with and without a history of CNCP and (ii) among decedents with
CNCP who are younger (<65 years) and older (65 + years).
Method: We examined all closed cases of intentional deaths in Australia in 2014,
utilizing the National Coronial Information System.
Results: We identified 2,590 closed cases of intentional deaths in Australia in 2014
in decedents over 18 years of age. CNCP was identified in 14.6% of cases.
Decedents with CNCP were more likely to be older, have more mental health and
physical health problems, and fewer relationship problems, and were more likely to
die by poisoning from drugs, compared with decedents without CNCP.
Comparisons of older and younger decedents with CNCP found that compared to
younger (<65 years) decedents with CNCP, older decedents (65 + years) were less
likely to have mental health problems.
Conclusions: This is the first national study to examine the characteristics of suicide
deaths with a focus on people with CNCP. Primary care physicians should be aware of
the increased risk for suicide in people living with CNCP, and it may be useful for
clinicians to screen for CNCP among those presenting with suicidal behaviors.

GABRIELLE CAMPBELL, National Drug and University of Newcastle, Newcastle, NSW,

Alcohol Research Centre, UNSW, Sydney, NSW, Australia and Calvary Mater Newcastle, Edith St
Australia and School of Health and Sport Sciences, Waratah, NSW, Australia; BRIAN DRAPER, Centre
University of the Sunshine Coast, Sunshine Coast, for Healthy Brain Ageing, School of Psychiatry,
QLD, Australia; SHANE DARKE, HARRIET University of New South Wales, Sydney, NSW,
TOWNSEND, AND MICHAEL FARRELL, National Australia and Dementia Centre for Research
Drug and Alcohol Research Centre, UNSW, Collaboration, University of New South Wales,
Sydney, NSW, Australia; LOUISA DEGENHARDT, Sydney, NSW, Australia; JOHAN DUFLOU, Sydney
National Drug and Alcohol Research Centre, Medical School, University of Sydney, Sydney,
UNSW, Sydney, NSW, Australia, School of NSW, Australia; JULIA LAPPIN, National Drug and
Population and Global Health, University of Alcohol Research Centre, UNSW, Sydney, NSW,
Melbourne, Melbourne, VIC, Australia, Murdoch Australia and School of Psychiatry, University of
Children’s Research Institute, Parkville, VIC, New South Wales, Sydney, NSW, Australia.
Australia and Department of Global Health, Address correspondence to G. Campbell,
School of Public Health, University of School of Health and Sport Sciences, University of
Washington, Seattle, WA, USA; GREGORY the Sunshine Coast, Sunshine Coast, USCLocked
CARTER, Centre for Brain and Mental Health Bag 4 Maroochydore, QLD 4558, Australia;
Research, Faculty of Health and Medicine, E‐mail: gcampbell@usc.edu.au
2 PREVALENCE AND CHARACTERISTICS

INTRODUCTION Joiner’s interpersonal theory (IPT) of

suicide (Joiner, 2007) provides some explana-
Suicide remains a significant cause of pre- tions of the association between suicide and
ventable mortality worldwide and is ranked in chronic pain. The impact of chronic pain on
the top ten leading causes, by age‐standard- an individual’s employment, physical and
ized mortality rates, in several regions glob- mental functioning and quality of life, and the
ally (Naghavi, 2019). In the United States, reliance on for assistance with everyday tasks
nearly 45,000 lives were lost to suicide in is prevalent in this population. These are con-
2016 and several States have observed a 30% sistent with the perceived burdensomeness
increase in suicide rates since 1999 (CDC, and thwarted belongingness in Joiner’s theory
2018). In Australia, from 2016 to 2017, the (Joiner, 2007). The integrated motivational–
number of suicide deaths increased by 9.1% volitional model of suicidal behavior
(Australian Bureau of Statistics, 2018). Pre- (O’Connor & Kirtley, 2018) provides further
liminary Australian 2017 data suggest that evidence, proposing that increased physical
suicide rates were among the highest pain tolerance, as experienced by people with
recorded in a decade (Australian Bureau of chronic pain, can lead to acquired capability
Statistics, 2018). Importantly, people who die to act on suicidal thoughts.
by suicide are not a homogenous group Although there have been some recent
(Clapperton et al., 2018). Rather, patterns, advances in knowledge of suicide‐related
trends, and key risk factors for suicide change behaviors and CNCP, much less is known
over time. It is crucial that emerging risk fac- about the prevalence and associations of
tors are investigated and identified in order to CNCP with suicide mortality (Cochran et al.,
optimize effective prevention strategies. 2014; Ilgen et al., 2016; Petrosky et al., 2018).
One such risk factor is chronic non- To date, no studies have specifically examined
cancer pain (CNCP), that is, pain that persists the prevalence of CNCP among deaths by
for longer than three months (Merskey & suicide at the population level. Only a minor-
Bogduk, 1994). CNCP is estimated to affect ity of people who engage in suicide‐related
approximately one‐fifth to one‐third of the behaviors, such as ideation and attempts, will
world’s population (Campbell et al., 2015a; die by suicide (Carroll et al., 2014). By exam-
Cimas et al., 2018; Dahlhamer et al., 2018; ining suicide deaths, we can obtain a greater
Jackson et al., 2016). Prevalence is expected understanding of the prevalence of the prob-
to increase in the coming decades, reflecting lem and understand the factors that may
the increasing number of older adults living interact with CNCP to increase the likeli-
with chronic health problems in these popula- hood of suicide. This has the potential to
tions (Blyth et al., 2001; Johannes et al., uncover possible opportunities to intervene
2010). People living with CNCP often expe- and target suicide prevention in the CNCP
rience complex comorbidities including mul- population. The current study examined all
tiple pain conditions and other physical suicide deaths in Australia in 2014 in order to:
health and mental health issues. Emerging
research has linked CNCP as a risk factor for (1) determine the proportion of com-
suicide‐related behaviors, that is, ideations, pleted suicide cases where there is a
planning, and attempts (Calati et al., 2015; history of CNCP;
Campbell et al., 2015a; Ilgen et al., 2013). (2) compare the sociodemographic and
People living with CNCP are 2–3 times more clinical characteristics of people
likely to engage in suicide‐related behaviors who die by suicide with and without
than the general population (Tang & Crane, a history of CNCP; and
2006). Up to 60% of those who report life- (3) compare sociodemographic and
time and past 12‐month suicide‐related clinical characteristics and circum-
behaviors have a history of chronic pain stances of death between younger
(Campbell et al., 2016). and older people (below and of/
CAMPBELL ET AL. 3

above 65 years) with a history of manner of death, in people aged 18 years and
CNCP. over, from the January 1, 2014, until the
December 31, 2014, were examined. The year
2014 was selected as it was the most recent
METHOD year that would have the greatest number of
closed cases to be investigated. Data extrac-
The National Coronial Information tion occurred from August 2017 to February
System (NCIS) is an electronic database of 2018. During this time, the proportion of
information containing case details from coro- closed (and therefore eligible) cases was 88%.
nial files of all Australian states and territories A sample of 2708 suicide cases was identified
from July 1, 2000 (with the exception of from the NCIS database. Cases where the
Queensland where files date from January 1, decedent was under the age of 18 years
2001). The NCIS is managed by the Victorian (n = 81) were excluded. Cases were also
Department of Justice and has been found to excluded if they were still under investigation
have more detailed information recorded than at the time of data extraction (n = 37) result-
other information systems, such as the Aus- ing in a final sample of 2590.
tralian Bureau of Statistics National Deaths
Index data (Bugeja et al., 2016). Outcome Variables
The coronial files consist of documenta-
tion including coroners’ findings, autopsy and Coronial files were examined to deter-
toxicology reports, and police narratives. Cause mine mentions of chronic pain. The presence
of death is ascertained by a forensic pathologist of chronic pain was coded if there were men-
and noted in the autopsy and coroner’s report. tions of chronic pain in any of the coronial
In Australia, the criteria for reporting a death files, such as police narratives, findings, or
vary between jurisdictions. In general, a death autopsy reports. Related terms such as arthri-
is reportable to the coroner where the person tis, fibromyalgia, back or neck pain, and
dies unexpectedly and the cause of death is migraines were included. Coronial files,
unknown; the person died in a violent and including police narratives, and autopsy,
unnatural manner; the person died during, or coronial, and toxicology reports, were manu-
as a result of anesthetic; the person was “held in ally reviewed by research assistants. Weekly
care” or in custody immediately before death; a discussion by the team assisted with coding
medical practitioner has been unable to issue a ambiguous cases. Cases that included men-
death certificate stating the cause of death; or tions of cancer (n = 124) were excluded from
the identity of the decedent is unknown. the CNCP group and included in the non‐
CNCP group, because the focus for this study
Ethics was CNCP.

This study was funded through a Soci- Independent Variables

ety for Mental Health Research early career
researcher grant, and ethical approval was Routinely collected data in the NCIS
obtained through UNSW ethics (HC17024), include age, sex, country of birth, employ-
NCIS (MO388), Department of Justice and ment status, usual occupation, marital status,
Regulation Human Research Ethics Com- location and activity at the time of incident,
mittee (CF/17/4103), and the Coroners mechanism and object or substance involve-
Court of Western Australia (EC12‐2017). ment at the time of death, and medical cause
of death and intent. The most frequent mode
Case Identification of suicide was categorized into seven groups
based on ICD 10 codes from the underlying
All closed cases where suicide (inten- cause of death. These included hanging (ICD
tional self‐harm) was determined to be the 10 code X70), poisoning by drugs (ICD 10
4 PREVALENCE AND CHARACTERISTICS

codes X60‐X65), carbon monoxide poisoning Toxicology reports were inspected to

(ICD 10 code X67), firearms (ICD 10 codes identify substances that were present in the
X72‐X74), jumping (ICD 10 code X80), and blood at the time of death: alcohol, cannabis,
sharp object (ICD 10 code X78). All other amphetamines, opioids, nonopioid anal-
underlying causes of death were coded into an gesics, benzodiazepines, antidepressants,
“other” category. This included drowning antipsychotics, and pentobarbitone.
(X71), smoke/fire or explosives (X75‐X76),
moving object (X81), motor vehicle crash Data Analysis
(X82), and unspecified means (X83‐X84). The
decedent’s postcode was used to determine All analyses were conducted using
whether the decedent resided in a major city, STATA version 15. Frequencies and propor-
regional or remote center. These were based tions and 95% confidence intervals (CI) are
on the Australian Bureau of Statistics remote- presented for sociodemographic and clinical
ness classifications. Years of potential life lost characteristics. For age at death, mean and
(YLL) was calculated by multiplying the num- standard deviations are reported as the data
ber of deaths by the remaining life expectancy were normally distributed. Comparisons were
for each decedent. made using logistic regression and reported
Data on mental health mentions and as odds ratios and 95% CI are presented. Due
other physical health conditions were recorded to multiple comparisons, we used a p‐value
into a free text field and were later recoded into significance cutoff of .001.
the following categories: Physical health free
text was recoded into musculoskeletal, cardio- Secondary Analysis
vascular, respiratory, diabetes, cancer, and
liver disease. Mental health free text was Additional analyses were conducted to
recoded into depression, anxiety, schizophre- examine sociodemographic, clinical, and cir-
nia/psychosis, bipolar affective disorder, per- cumstantial differences, on a reduced set of
sonality disorders, and posttraumatic stress the most common variables, between younger
disorder. These were classified based on speci- people (<65 years) with and without chronic
fic mentions of the conditions in the coronial pain and older people (65 years and older)
reports. Mentions of illicit drug and alcohol with and without chronic pain to examine
use were also coded. The total numbers of whether there were any potential informative
mental and physical health problems were cal- associations (see Appendices 1 and 2).
culated, and a binary variable was created to
identify three or more mental and physical
health conditions documented for each case. RESULTS
Coronial files were also searched for
information pertaining to common risk fac- Characteristics of Suicide Deaths in
tors for suicide. These included mentions of Australia in 2014
recent job loss, relationship problems (part-
ner, friends, or family), recent loss of children A number of 2,590 closed cases of inten-
(due to death, custody arrangements, or tional self‐harm were identified. The mean age
divorce), grief, and recent financial difficul- of decedents was in the mid‐forties
ties. These data were collected to compare (46.4 years), and three‐quarters were male
whether there were any differences in known (76.2%). Over one‐third were married or de
risk factors for people living with CNCP, facto (36.9%), employed (38.4%), and living in
compared to people without CNCP. Data rural or remote areas (35.0%) (Table 1). The
were also extracted for whether the decedent mean YPLL due to suicide was 36.5 (SD 16.1).
had left a note or a message and whether a Mentions of mental health problems
previous suicide attempt was mentioned in were common (65.3%) with depression the
the coronial files. most prevalent, mentioned in over half of
CAMPBELL ET AL. 5

cases. One in ten decedents had mentions of n = 379) of cases and contributed a mean of
three or more mental health problems. Physi- 27.5 (SD 15.4) YPLL. Less than 3% of cases
cal health problems were identified in over reported three or more physical health prob-
one‐fifth of cases (22.4%), the most common lems (Table 1).
being cardiovascular disease (9.8%). CNCP Among all suicide deaths, the most
was identified in 14.6% (95% CI 13.3–16.0, common risk factors identified were

TABLE 1
Characteristics of People Who Died Via Suicide According to History of Chronic Noncancer Pain,
Australia, 2014
People without a People with a Bivariate comparisons
recorded history recorded history
Total of CNCP of CNCP
n = 2590 n = 2,211 n = 379
% (95% CI) % (95% CI) % (95% CI) OR (95% CI) p value

Mean age (SD) 46.4 (17.49) 44.7 (16.9) 56.6 (17.6) 11.9 (10.02–13.72) <.0001
Male 76.2 (74.5–77.8) 77.0 (75.2–78.7) 71.2(66.5–75.6) 1.35 (1.06–1.72) .015
Married/de facto 36.9 (35.1–38.8) 36.5 (34.5–38.5) 39.8 (35.0–44.9) 1.15 (0.92–1.44) .207
Employed 38.4 (26.5–40.3) 40.1 (38.0–42.1) 28.5 (24.2–33.3) 0.60 (0.47–0.76) <.0001
Rural/remote 35.0 (33.7–37.7) 35.3 (33.4–37.4) 36.4 (31.7–41.4) 1.05 (0.84–1.32) .681
Mental health
Any mental health 65.3 (63.4–67.1) 63.6 (61.6–65.6) 74.9 (70.3–79.1) 1.71 (1.34–2.19) <.0001
illness
Depression 56.0 (54.1–57.9) 54.0 (52.0–56.1) 67.3 (62.4–71.8) 1.75 (1.39–2.20) <.0001
Anxiety 15.2 (13.9–16.6) 14.7 (13.3–16.2) 18.5 (14.9–22.7) 1.31 (1.00–1.75) .057
Schizophrenia/ 8.2 (7.2–9.3) 8.3 (7.2–9.5) 7.7 (5.4–10.8) 0.92 (0.61–1.38) .682
psychosis
Bipolar 4.9 (4.1 5.8) 4.8 (4.0–5.8) 5.0 (3.2–7.7) 1.04 (0.63–1.71) .884
Personality 3.7 (3.0–4.5) 3.7 (3.0–4.6) 3.4 (2.0–5.8) 0.92 (0.51–1.67) .790
disorder
PTSD 2.6 (2.0–3.3) 2.4 (1.8–3.1) 4.0 (2.4–6.5) 1.71 (0.95–3.07) .072
Alcohol use 9.2 (8.1–10.3) 9.0 (7.9–10.3) 10.0 (7.4–13.5) 1.13 (0.78–1.62) .522
problems
Substance use 7.7 (6.7–8.8) 7.6 (6.6–8.8) 7.9 (5.6–11.1) 1.04 (0.69–1.56) .854
problems
3 or more MH 10.0 (8.9–11.3) 9.8 (8.6–11.1) 11.6 (8.7–15.3) 1.21 (0.86–1.71) .271
problems
Physical health
Any physical 22.4 (20.9–24.1) 17.8 (16.3–19.5) 49.3 (44.3–54.4) 4.49 (3.57–5.65) <.0001
health illness
Musculoskeletal 8.7 (7.7–9.9) 2.9 (2.3–3.7) 42.5 (37.6–47.5) 24.4 (17.7–33.6) <.0001
Cardiovascular 9.8 (8.8–11.1) 7.4 (6.4–8.6) 24.3 (20.2–28.9) 4.00 (3.01–5.31) <.0001
disease
Respiratory 4.0 (3.3–4.8) 2.9 (2.3–3.7) 10.0 (7.4–13.5) 3.68 (2.43–5.58) <.0001
disease
Diabetes 3.9 (3.2–4.7) 3.3 (2.6–4.1) 7.6 (5.4–10.8) 2.46 (1.58–3.84) <.0001
Liver disease 2.6 (2.0–3.3) 2.2 (1.7–2.9) 4.7 (3.0–7.4) 2.20 (1.28–3.82) .005
3 or more physical 2.6 (2.1–3.3) 1.6 (1.1–2.2) 8.7 (6.2–12.0) 5.9 (3.6–9.7) <.0001
health problems

Bold is significant at p < .001

6 PREVALENCE AND CHARACTERISTICS

relationship problems (24.5%) followed by and one‐third (32.2%, n = 122) 65 years or

financial difficulties (14.9%) (Table 2). One‐ over. Compared with those younger than 65,
quarter of decedents had made a previous sui- older decedents were less likely to have identi-
cide attempt (26.0%) and more than half left a fied mental health problems, more likely to
note (58.6%). The most common method have physical health problems, and more
used was hanging (55.2%), followed by over- likely pentobarbitone present, though this
dose from poisoning (14.5%). The most com- was not significant at the conserva-
monly detected drug present was alcohol, tive < 0.001 level (Table 3).
identified in over one‐third of cases (37.0%),
followed by benzodiazepines (24.9%) and Comparisons between Younger (<65
antidepressants (24.1%). Years) and Older (65 Years and Over)
Decedents With and Without CNCP
Characteristics Associated with Chronic
Pain in Suicide Deaths Additional analyses were performed
comparing decedents younger and older with
Compared with decedents without and without chronic pain. Among younger
CNCP, decedents with CNCP were older people (<65 years), decedents with CNCP
and less likely to be employed. Decedents were significantly more likely to have any
with CNCP were also more likely to have mental health or physical health problems
mental health problems, particularly depres- (see Appendix 1). Younger people with
sion (67.3% vs 54.0%) and more likely to have CNCP were also less likely to die by hanging
physical health problems than those without and more likely to die by poisoning and have
CNCP (Table 1). more pharmaceuticals present in their toxi-
Compared with decedents without cology.
CNCP, decedents with CNCP were less Among older decedents (65 years and
likely to be recorded as having relation- greater) with and without CNCP, there were
ship problems. CNCP decedents were less very few significant (see Appendix 2). Dece-
likely to die by hanging and had more dents with CNCP were more likely to have
than three times the odds of dying by physical health problems recorded, more
drug poisoning (OR 3.48, 95% CI 2.70– likely to die by poisoning, and more likely to
4.48). With the exception of alcohol, there have nonopioid analgesics present in their
were marked differences in the toxicology toxicology, compared to decedents without
profiles of decedents with and without CNCP.
CNCP. Decedents with CNCP were sig-
nificantly more likely to have opioids (OR
3.42, 95% CI 2.69–4.35) and nonopioid DISCUSSION
analgesics present (OR 3.02. 95% CI
2.36–3.87). Although overall use was rare To our knowledge, this is the first
(n = 50, 1.9%), decedents with CNCP had study to examine the characteristics of suicide
over four times the odds of having pento- deaths with a focus on people with CNCP in
barbitone present in the toxicology (OR a national population. Overall, we identified
4.05, 95% CI 2.28–7.21) than those 2,590 closed cases of suicide deaths in people
without. over 18 years of age in Australia, among
whom approximately one in seven decedents
Sociodemographic, Clinical Characteristics had a documented history of CNCP. Unem-
and Circumstances: Comparisons between ployment, mental health, and physical health
Younger and Older Decedents with CNCP problems were common in both but were
more common among decedents with CNCP
Among decedents with CNCP, two‐ than those without. CNCP suicide deaths
thirds (67.8% n = 257) were under 65 years were proportionally more likely to occur by
CAMPBELL ET AL. 7

TABLE 2
Risk Factors, Suicide Method, and Toxicology among People Who Died Via Suicide According to History
of Chronic Noncancer Pain, Australia, 2014
People without a People with a Comparisons
recorded history recorded history
Total of CNCP of CNCP
n = 2590 n = 2,211 n = 379
Recent risk factors % (95% CI) % (95% CI) % (95% CI) OR (95% CI) p value

Job loss 6.1 (5.2–7.1) 5.9 (5.0–6.9) 7.4 (5.1–10.5) 1.27 (0.84–1.95) .258
Relationship 24.5 (22.9–26.2) 26.0 (24.2–27.8) 16.1 (12.7–20.2) 0.55 (0.41–0.73) <.0001
problems
Loss of children 7.6 (6.6–8.7) 7.9 (6.9–9.2) 5.3 (3.4–8.1) 0.64 (0.40–1.04) .070
Financial difficulties 14.9 (13.6–16.4) 14.6 (13.2–16.1) 16.9 (13.4–21.0) 1.19 (0.89–1.59) .251
Previous attempt 26.0 (24.4–27.7) 25.3 (23.6–27.2) 29.8 (25.4–34.6) 1.25 (0.98–1.59) .069
Left note 58.6 (56.7–60.5) 57.6 (55.5–59.7) 64.6 (59.7–69.3) 1.34 (1.07–1.69) .010
Method
Hanging 55.2 (53.3–57.1) 57.7 (55.6–59.8) 40.6 (35.8–45.7) 0.50 (0.40–0.63) <.0001
Poisoning—drugs 14.6 (13.2–16.0) 11.6 (10.4–13.0) 31.7 (27.2–36.5) 3.52 (2.73–4.53) <.0001
Carbon monoxide 6.3 (5.4–7.3) 6.2 (5.2–7.2) 6.8 (4.7–9.9) 1.12 (0.73–1.74) .599
Firearm 6.0 (5.1–7.0) 5.9 (5.0–6.9) 6.6 (4.5–9.6) 1.13 (0.73–1.76) .587
Jumping 4.2 (3.5–5.0) 4.5 (3.7–5.4) 2.3 (1.2–4.5) 0.52 (0.26–1.04) .063
Sharp object 2.7 (2.2–3.4) 2.7 (2.1–3.5) 2.9 (1.6–5.2) 1.07 (0.56–2.06) .835
Othera 14.4 (13.1–15.8) 14.6 (13.2–16.2) 12.9 (9.9–16.7) 0.86 (0.63–1.19) .377
Toxicology—drugs detected
Alcohol 37.0 (35.1–38.9) 37.1 (35.2–39.2) 35.9 (31.2–40.9) 0.95 (0.75–1.19) .630
Cannabis 10.7 (9.5–11.9) 11.2 (10.0–12.6) 7.4 (5.1–10.5) 0.63 (0.42–0.95) .027
Amphetamines 6.6 (5.7–7.7) 7.3 (6.2–8.4) 2.9 (1.6–5.2) 0.38 (0.20–0.71) .002
Opioids 17.3 (15.9–18.8) 14.1 (12.7–15.6) 35.9 (31.2–40.9) 3.42 (2.69–4.35) <.0001
Nonopioid analgesic 16.1 (14.8–17.6) 13.4 (12.1–14.9) 31.9 (27.4–36.8) 3.02 (2.36–3.87) <.0001
Benzodiazepines 24.9 (23.2–26.6) 22.4 (20.7–24.2) 39.1 (34.2–44.1) 2.22 (1.76–2.79) <.0001
Pentobarbitone 1.9 (1.5–2.5) 1.4 (1.0–1.9) 5.3 (3.4–8.1) 4.05 (2.28–7.21) <.0001
Antidepressant 24.1 (22.5–25.8) 22.5 (20.8–24.3) 33.2 (28.7–38.2) 1.71 (1.35–2.17) <.0001
Antipsychotics 9.3 (8.3–10.5) 8.8 (7.7–10.0) 12.7 (9.7–16.4) 1.51 (1.07–2.11) .017
a
Other includes drowning, vehicles, pesticides, moving object, not specified, fire, and explosives.
(ICD 10 codes X66, X68, X69, X71, X75 X79, X81–X84)
Bold is significant at p < .001

intentional overdose with opioids, benzodi- 2018). Our study, like that of Petrosky et al.,
azepines, antidepressants, and nonopioid (2018), relied on coronial data and therefore
analgesics. police narratives and coronial findings to
CNCP is a prevalent problem, and sui- identify CNCP in suicide. It is likely that our
cidality is common among people living with estimates of CNCP are therefore less than
CNCP (Campbell et al., 2015a). This study true prevalence.
demonstrates that CNCP is also importantly CNCP has a major impact on the indi-
associated with suicide deaths. We identified vidual in terms of quality of life, mental health
that 14.6% of all adult deaths by suicide have and physical health status, relationships, and
a documented history of CNCP. This figure employment (Blyth et al., 2001; Breivik et al.,
is higher than those reported elsewhere: A 2013; Campbell et al., 2015b). We found that,
recent study from the United States identified compared with people who died by suicide
chronic pain in 8.8% of cases (Petrosky et al., without CNCP, people with CNCP were less
8 PREVALENCE AND CHARACTERISTICS

TABLE 3
Characteristics, circumstances, and toxicology comparisons of people less than 65 years and 65 and over
with CNCP
A. People with a B. People with B vs A (ref) Comparisons
recorded history a recorded history
of CNCP younger of CNCP 65 years
than 65 years and older
n = 257 n = 122
% (95% CI) % (95% CI) OR (95% CI) p value

Male 72.8 (70.0–77.9) 68.0 (59.1–75.8) 0.79 (0.50–1.27) .342

Rural/remote 38.4 (32.6–44.6) 32.2 (24.4–41.2) 0.76 (0.48–1.20) .244
Any mental health illness 83.3 (78.2–87.4) 57.4 (48.3–66.0) 0.27 (0.17–0.44) <.0001
Three or more mental 15.6 (11.6–20.6) 3.3 (1.2–8.5) 0.18 (0.06–0.53) .002
health illnesses
Depression 73.5 (67.8–78.6) 54.0 (45.1–62.8) 0.42 (0.27–0.66) <.0001
Anxiety 21.8 (17.1–27.3) 11.5 (6.9–18.6) 0.28 (0.21–0.37) .017
Substance use 11.3 (7.9–15.8) 0.8 (0.1–0.6) 0.06 (0.09–0.48) .008
Alcohol use 13.2 (9.6–18.0) 3.3 (1.2–8.5) 0.22 (0.08–0.64) .005
Any previous attempt 35.4 (29.8–41.5) 18.0 (12.1–26.0) 0.40 (0.24–0.68) .001
Intention reported 61.9 (55.7–67.6) 70.5 (61.7–78.0) 1.47 (0.93–2.34) .102
Physical health
Three or more physical 5.1 (2.9–8.5) 16.4 (10.7–24.2) 3.68 (1.76–7.68) .001
health illnesses
Musculoskeletal 34.2 (28.7–40.3) 59.8 (50.8–68.2) 2.86 (1.83–4.46) <.0001
Cardiovascular 11.3 (7.9–15.8) 51.6 (42.7–60.5) 8.40 (4.97–14.19) <.0001
Respiratory 7.0 (4.4–10.9) 16.4 (10.8–24.2) 2.60 (1.32–5.13) .006
Method
Hanging 43.9 (37.9–50.1) 33.6 (25.7–42.6) 0.64 (0.41–1.01) .056
Poisoning—drugs 31.1 (25.7–37.1) 32.8 (25.0–41.7) 1.08 (0.68–1.71) .746
Carbon monoxide 6.2 (3.8–9.9) 8.2 (4.4–14.7) 1.34 (0.59–3.06) .479
Firearm 5.4 (3.2–9.0) 9.0 (5.0–15.7) 1.72 (0.76–3.90) .195
Jumping 3.5 (1.8–6.6) 0 – –
Sharp object 0.8 (0.2–3.1) 7.4 (3.8–13.7) 10.2 (2.16–47.8) .003
Other* 11.7 (8.2–16.2) 15.6 (10.1–23.3) 1.39 (0.75–2.59) .292
Toxicology—drugs detected
Alcohol 40.9 (35.0–47.1) 25.4 (18.4–34.0) 0.49 (0.31–0.79) .004
Cannabis 10.9 (7.6–15.4) 0 – –
Opioids 37.4 (31.6–43.4) 32.8 (25.0–41.7) 0.82 (0.52–1.29) .387
Nonopioid analgesic 27.6 (22.5–33.5) 41.0 (32.6–50.0) 1.81 (1.16–2.86) .010
Benzodiazepines 37.3 (31.6–43.5) 42.6 (34.0–51.7) 1.25 (0.80–1.93) .326
Pentobarbitone 3.1 (1.6–6.1) 9.8 (5.6–16.6) 3.40 (1.35–8.54) .009
Antidepressant 35.8 (30.1–41.9) 27.9 (20.5–36.6) 0.69 (0.43–1.11) .127
Antipsychotics 15.6 (11.6–20.6) 6.6 (3.3–12.7) 0.38 (0.17–0.84) .017
a
Other includes drowning, vehicles, pesticides, moving object, not specified, fire, and explosives.
(ICD 10 codes X66, X68, X69, X71, X75 X79, X81–X84).
Bold is significant at p < .001

likely to be employed, more likely to have may be understood by the fact that the
mentions of any mental health problems par- deceased with CNCP were more likely to be
ticularly depression, and more likely to have older and therefore more likely to have physi-
other physical health comorbidities. This cal health problems and be retired.
CAMPBELL ET AL. 9

Nonetheless, one in three decedents Increasing our knowledge of the char-

with CNCP was less than 65 years old. Pre- acteristics of people with CNCP who die by
vious research has found that younger people suicide is essential to create opportunities for
living with CNCP have more complex clini- intervention. Reducing access to means has
cal profiles than older people with CNCP been found to be one of the most effective
(Campbell et al., 2015b). This was supported measures to reduce suicide deaths. In the cur-
in the current study. Deceased with CNCP rent study, we found that decedents with
who were under 65 years were significantly CNCP were proportionally more likely to die
more likely to have mental health problems by intentional overdose than by hanging.
identified compared with those 65 years or Hanging is traditionally the most common
older. As identification of mental health method for suicide in Australia (Australian
problems in older people is often a challenge Bureau of Statistics, 2018). As we identified,
in primary care and for families, it is unclear people with CNCP were also more likely to
whether these findings are an accurate repre- have other physical health problems and were
sentation of what happens (Andreas et al., more likely to have medicines, including opi-
2017). Considering that CNCP can be a life- oids, nonopioid analgesics, and benzodi-
time condition, it is important that the men- azepines, present in their toxicology reports.
tal health of people living with CNCP is With multiple physical health problems
assessed and treated and that access to suit- comes exposure to multiple medicines. The
able care is made available and affordable. potential for dependence and misuse of these
Additionally, we identified that both drugs is well known (Darke et al., 2019;
younger and older decedents with CNCP Lader, 2011). Furthermore, there is increas-
were more likely to have physical health ing potential for people to gain access to
problems recorded, over and above their potentially lethal medications through online
CNCP. The importance of physical health markets including the Dark Web (Oberhaus,
as risk factor for suicide should be investi- 2018). Concerningly, we found that people
gated further. with CNCP had four times the odds of having
In the overall sample, relationship pentobarbitone present in their toxicology at
problems were a common risk factor for sui- the time of death. Pentobarbitone is a short‐
cide, reported in approximately one‐quarter acting barbiturate central nervous system
of cases. Consistent with previous US depressant. Internationally, it has garnered
research (Petrosky et al., 2018), relationship attention due to advocacy through euthanasia
problems were less likely to be reported groups, (Solbeck et al., 2019) and currently, it
among Australian cases of suicide with is classified as a highly restricted and con-
CNCP. This finding supports the conclu- trolled Schedule 8 drug in Australia.
sions of the US study (Petrosky et al., 2018), Although restriction of access to poten-
that other stressors (including CNCP) may tially lethal means has been found to be an
be more important in this group. effective measure in reducing suicide behav-
Approximately one in three of all iors (Hawton et al., 2013; Mann et al., 2005)
decedents had made a previous attempt. among people living with CNCP with multi-
While a previous attempt was not signifi- ple physical and mental health problems,
cantly more common among those with there is a need to balance the need for effec-
CNCP, this finding suggests that there tive pain management and the ease of access
may be opportunities for intervention to other available effective treatments.
among people with CNCP and suicidal Although increased screening of suicidal
behaviors. Screening for CNCP should be behaviors in primary care has been recom-
conducted among those presenting to care mended (O’Connor et al. 2013), it is impor-
with suicidal behaviors. Similarly, enquiry tant to acknowledge that only a minority of
about suicidal behaviors should be made in people with suicidal ideation will make a sui-
those with CNCP. cide attempt and even fewer will die by suicide
10 PREVALENCE AND CHARACTERISTICS

(Carroll et al., 2014). Importantly, the predic- NCIS has been recognized as a unique data
tive value of suicidal ideation for subsequent source that provides reliable and high‐quality
suicide is significantly lower in nonpsychiatric detailed information (Bugeja et al., 2016), sev-
cohorts such as primary health care or general eral limitations should be acknowledged. The
population samples than among people who current study was based on coronial files and
receive psychiatric care (McHugh et al., not on clinical diagnoses. Reliance on inves-
2019). Increasing access to other treatment tigative reports means that the results pre-
options, especially pain management pro- sented here are likely to underrepresent the
grams, would be preferable to simply restrict- true numbers of people experiencing CNCP.
ing access to necessary medications in people We are limited to details that are provided by
with suicidal ideation. Indeed, it has been sug- police, family, and friends. We do not know
gested that the restriction of medication, with any specific pain factors such as pain duration,
no alternative effective pain management, cause of pain, pain severity, and impact of
might actually lead to increased suicide rates pain on everyday life. Additionally, it is diffi-
in people living with CNCP (Ilgen, 2018). cult to ascertain from coronial files the exact
With an aging population and risk factors that were directly related to an
increases in CNCP, where there is no simple individual’s suicide. Cases were also restricted
effective treatment, it is important to increase to those coded as having deliberate intention.
access to specialist supportive services such as Cases of undetermined intent were thus not
multidisciplinary pain treatment programs included.
that may assist patients in managing their
CNCP. Pain management programs com-
prising a multidisciplinary team of health pro- CONCLUSION
fessionals, including psychologists,
physiotherapists, and pain specialists, are the The current study was the first national
best approach for managing CNCP (Kamper study to examine CNCP among suicide
et al., 2015). One of the challenges in provid- deaths in Australia. Acknowledgment and
ing effective treatment is accessibility to these understanding of CNCP as a risk factor for
multidisciplinary pain management programs suicide are crucial. Primary care physicians
which, unlike opioid medication, are largely should be aware of the increased risk for sui-
not subsidized by the health system. Access cide in people living with CNCP, and clini-
and affordability of these programs are essen- cians should screen for CNCP among those
tial in order to address the multitude of prob- presenting with suicidal behaviors. Although
lems experienced by this population. In restriction of access to means is an effective
addition, suicide rates are 1.7 higher in rural/ suicide prevention strategy, it is important
remote areas (AIHW, 2019). We found that that patients with suicidal ideation receive
just under 40% of younger people who die by targeted and tailored support, rather than
suicide with CNCP lived in rural or remote simply having access to their medicines
areas. Access to multidisciplinary pain ser- restricted. In order to address CNCP and sui-
vices for people in rural and remote areas in cide, it is important that both affordability
Australia remains limited and is a crucial tar- and access to pain management programs
get for services improvement (Hogg et al., become a priority.
2012). This study was funded by the Society
for Mental Health Research Early Career
Limitations Research Fellow grant. GC is supported by
National Health and Medical Research
One of the major strengths of the cur- Council fellowships (#1119992). The
rent study was that it included all closed sui- National Drug and Alcohol Research Centre
cide cases in people over the age of 18 years at University of New South Wales, Australia,
in Australia in 2014. Although the use of is supported by funding from the Australian
CAMPBELL ET AL. 11

Government, under the Substance Misuse Benckiser/Indivior for a study of opioid sub-
Prevention and Service Improvements Grant stitution therapy uptake among chronic non-
Fund. GC has received an investigator‐initi- cancer pain patients. This is not related to the
ated untied educational grants from Reckitt current study.

REFERENCES

AIHW (2019). Rural & remote health. (2015b). The Pain and Opioids IN Treatment
Retrieved September 5, 2018, from https://www.ai study: Characteristics of a cohort using opioids to
hw.gov.au/reports/rural‐health/rural‐remote‐health/ manage chronic non‐cancer pain. Pain, 156, 231–
contents/deaths‐remoteness. 242.
ANDREAS, S., SCHULZ, H., VOLKERT, J., CARROLL, R., METCALFE, C., & GUNNELL,
DEHOUST, M., SEHNER, S., SULING, A., ET AL. D. J. P. O. (2014). Hospital presenting self‐harm
(2017). Prevalence of mental disorders in elderly and risk of fatal and non‐fatal repetition: systematic
people: The European MentDis_ICF65+ study. review and meta‐analysis. PLoS ONE, 9, e89944.
British Journal of Psychiatry, 210, 125–131. CDC (2018). CDC vital signs. Retrieved
Australian Bureau of Statistics (2018). Cause November 14, 2018, from https://www.cdc.gov/
of death: Australia, 2017. Retrieved November 5, vitalsigns/pdf/vs‐0618‐suicide‐H.pdf.
2018, from http://www.abs.gov.au/ausstats/abs@. CIMAS, M., AYALA, A., SANZ, B., AGULLÓ‐
nsf/Lookup/by%20Subject/3303.0~2017~Main% TOMÁS, M. S., ESCOBAR, A., & FORJAZ, M. J. (2018).
20Features~Intentional%20self‐harm,%20key% Chronic musculoskeletal pain in European older
20characteristics~3. adults: Cross‐national and gender differences.
BLYTH, F. M., MARCH, L. M., BRNABIC, A. J.M., European Journal of Pain, 22, 333–345.
JORM, L. R., WILLIAMSON, M., & COUSINS, M. J. CLAPPERTON, A., BUGEJA, L., NEWSTEAD, S.,
(2001). Chronic pain in Australia: A prevalence & PIRKIS, J. (2018). Identifying typologies of persons
study. Pain, 89, 127–134. who died by suicide: Characterizing suicide in
BREIVIK, H., EISENBERG, E., & O’BRIEN, T. Victoria, Australia. Archives of Suicide Research, 1–16.
(2013). The individual and societal burden of COCHRAN, B. N., FLENTJE, A., HECK, N.
chronic pain in Europe: The case for strategic pri- C., VAN DEN BOS, J., PERLMAN, D., TORRES, J.,
oritisation and action to improve knowledge and ET AL. (2014). Factors predicting development
availability of appropriate care. BMC Public Health, of opioid use disorders among individuals who
13, 1229. receive an initial opioid prescription: Mathemat-
BUGEJA, L., IBRAHIM, J. E., FERRAH, N., ical modeling using a database of commercially‐
MURPHY, B., WILLOUGHBY, M., & RANSON, D. insured individuals. Drug and Alcohol Dependence,
(2016). The utility of medico‐legal databases for 138, 202–208.
public health research: a systematic review of peer‐ DAHLHAMER, J., LUCAS, J., ZELAYA, C.,
reviewed publications using the National Coronial NAHIN, R., MACKEY, S., DEBAR, L., ET AL. (2018).
Information System. Health Research Policy and Sys- Prevalence of chronic pain and high‐impact
tems, 14, 28. chronic pain among adults ‐ United States, 2016.
CALATI, R., BAKHIYI, C. L., ARTERO, S., MMWR. Morbidity and Mortality Weekly Report, 67,
ILGEN, M., & COURTET, P. (2015). The impact of 1001–1006.
physical pain on suicidal thoughts and behaviors: DARKE, S., LAPPIN, J., & FARRELL, M.
Meta‐analyses. Journal of Psychiatric Research, 71, (2019). The clinician’s guide to illicit drugs and health.
16–32. London: Silverback Publishing.
CAMPBELL, G., BRUNO, R., DARKE, S., HAWTON, K., BERGEN, H., SIMKIN, S.,
SHAND, F., HALL, W., FARRELL, M., ET AL. (2016). DODD, S., POCOCK, P., BERNAL, W., ET AL. (2013).
Prevalence and correlates of suicidal thoughts and Long term effect of reduced pack sizes of paraceta-
suicide attempts in people prescribed pharmaceuti- mol on poisoning deaths and liver transplant activ-
cal opioids for chronic pain. The Clinical Journal of ity in England and Wales: Interrupted time series
Pain, 32, 292–301. analyses. BMJ, 346, f403.
CAMPBELL, G., DARKE, S., BRUNO, R., & HOGG, M. N., GIBSON, S., HELOU, A.,
DEGEHARDT, L. (2015a). The prevalence and corre- DEGABRIELE, J., & FARRELL, M. J. (2012). Waiting
lates of chronic pain and suicidality in a nationally in pain: A systematic investigation into the provi-
representative sample. Australian & New Zealand sion of persistent pain services in Australia. Medical
Journal of Psychiatry, 49, 803–811. Journal of Australia, 196, 386–390.
CAMPBELL, G., NIELSEN, S., BRUNO, R., ILGEN, M. (2018). Pain, opioids, and suicide
LINTZERIS, N., COHEN, M., HALL, W., ET AL. mortality in the united statespain, opioids, and
12 PREVALENCE AND CHARACTERISTICS

suicide mortality in the United States. Annals of MERSKEY, H., & BOGDUK, N. (1994). Task
Internal Medicine, 169, 498–499. Force on Taxonomy of the International Associa-
ILGEN, M. A., BOHNERT, A. S., GANOCZY, tion for the Study of Pain. In Pain IAftSo (Ed.),
D., BAIR, M. J., MCCARTHY, J. F., & BLOW, F. C. Classification of chronic pain: descriptions of chronic pain
(2016). Opioid dose and risk of suicide. Pain, 157, syndromes and definition of pain terms. Seattle: IASP.
1079–1084. NAGHAVI, M. (2019). Global, regional, and
ILGEN, M., KLEINBERG, F., IGNACIO, R., national burden of suicide mortality 1990 to 2016:
BOHNERT, A. S., VALENSTEIN, M., MCCARTHY, J. systematic analysis for the Global Burden of Dis-
F., ET AL. (2013). Noncancer pain conditions and ease Study 2016. BMJ, 364, l94.
risk of suicide. JAMA Psychiatry, 70, 692–697. OBERHAUS, D. (2018). Rise of the DIY death
JACKSON, T., THOMAS, S., STABILE, V., machines. Vice.
SHOTWELL, M., HAN, X., & MCQUEEN, K. (2016). O’CONNOR, E., GAYNES, B., BURDA, B. U.,
A systematic review and meta‐analysis of the global SOH, C., & WHITLOCK, E. P. (2013). U.S. Preven-
burden of chronic pain without clear etiology in tive Services Task Force Evidence Syntheses, formerly
low‐ and middle‐income countries: trends in Systematic Evidence Reviews. Screening for Suicide
heterogeneous data and a proposal for new assess- Risk in Primary Care: A systematic evidence review for
ment methods. Anesthesia & Analgesia, 123, 739– the U.S. Preventive Services Task Force. Rockville,
748. MD: Agency for Healthcare Research and Quality
JOHANNES, C. B., LE, T. K., ZHOU, X., JOHN- (US).
STON, J. A., & DWORKIN, R. H. (2010). The preva- O’CONNOR, R. C., & KIRTLEY, O. J. (2018).
lence of chronic pain in United States adults: The integrated motivational‐volitional model of
Results of an Internet‐based survey. The Journal of suicidal behaviour. Philosophical Transactions of the
Pain, 11, 1230–1239. Royal Society B: Biological Sciences, 373, 20170268.
JOINER, T. (2007). Why people die by suicide. PETROSKY, E., HARPAZ, R., FOWLER, K. A.,
Cambridge: Harvard University Press. BOHM, M. K., HELMICK, C. G., YUAN, K., ET AL.
KAMPER, S. J., APELDOORN, A. T., CHIAR- (2018). Chronic pain among suicide decedents,
OTTO, A., SMEETS, R. J., OSTELO, R. W., GUZMAN, 2003 to 2014: findings from the national violent
J., ET AL. (2015). Multidisciplinary biopsychosocial death reporting system. Annals of Internal Medicine,
rehabilitation for chronic low back pain: Cochrane 169, 448.
systematic review and meta‐analysis. BMJ, 350, SOLBECK, P., SNOWDON, V., RAJAGOPALAN,
h444. A., & JHIRAD, R. (2019). Suicide by fatal pentobar-
LADER, M. (2011). Benzodiazepines revis- bital intoxication in Ontario, Canada, from 2012
ited–will we ever learn? Addiction, 106, 2086–2109. to 2015. Journal of Forensic Sciences, 64, 309–313.
MANN, J., APTER, A., BERTOLOTE, J., BEAU- TANG, N. K., & CRANE, C. (2006). Suicidal-
TRAIS, A., CURRIER, D., HAAS, A., ET AL. (2005). Sui- ity in chronic pain: a review of the prevalence, risk
cide prevention strategies: A systematic review. factors and psychological links. Psychological Medi-
JAMA, 294, 2064–2074. cine, 36, 575–586.
MCHUGH, C. M., CORDEROY, A., RYAN, C.
J., HICKIE, I. B., & LARGE, M. M. (2019). Associa- Manuscript Received: July 2, 2019
tion between suicidal ideation and suicide: Meta‐ Revision Accepted: December 9, 2019
analyses of odds ratios, sensitivity, specificity and
positive predictive value. BJPsych Open, 5, e18.
CAMPBELL ET AL. 13

APPENDIX 1

Comparison of younger people (<65 years) with and without history of chronic non-
cancer, Australia, 2014

Bivariate comparisons
Younger people
(< 65 years old) Younger people
without a with a recorded
recorded history of
history of CNCP CNCP
n = 1,928 n = 257
% (95% CI) % (95% CI) OR (95% CI) p value

Mean age (SD) 40.2 (12.5) 46.6 (10.7) 6.48 (4.87–8.08) <.0001
Male 76.7 (74.8–78.5) 72.8 (67.0–77.9) 1.23 (0.92–1.65) .163
Married/de facto 34.8 (32.7–36.9). 38.5 (32.7–44.6) 1.18 (0.89–1.53) .235
Employed 44.6 (42.4–46.8) 39.7 (33.7–45.8) 0.82 (0.63–1.07) .136
Rural/remote 34.7 (32.6–36.8) 38.4 (32.6–44.6) 1.18 (0.90–1.54) .236
Mental health
Any mental health 65.4 (63.2–67.4) 83.3 (78.2–87.4) 2.64 (1.88–3.71) <.001
illness
Depression 55.3 (53.1–57.5) 73.5 (67.8–78.6) 2.25 (1.68–3.00) <.0001
Anxiety 15.4 (13.9–17.1) 21.8 (17.1–27.3) 1.53 (1.11–2.11) .009
3 or more MH 10.6 (9.3–12.0) 15.6 (11.6–20.6) 1.56 (1.08–2.25) .018
problems
Physical health
Any physical health 12.9 (11.5–14.5) 38.9 (33.1–45.0) 4.29 (3.23–5.70) <.0001
illness
3 or more physical 0.5 (0.3–0.9) 5.1 (3.0–8.5) 10.2 (4.43–25.55) <.0001
health problems
Method of suicide
Hanging 61.0 (58.8–63.2) 44.0 (38.0–50.1) 0.50 (0.39–0.65) <.0001
Poisoning—drugs 11.0 (9.6–12.5) 30.7 (25.4–36.7) 3.59 (2.66–4.85) <.0001
Toxicology—drugs detected
Alcohol 39.5 (37.4 (41.7) 40.9 (35.0–47.0) 1.06 (0.81–1.38) .682
Opioids 13.5 (12.1–15.1) 37.4 (31.6–43.5) 3.81 (2.87–5.06) <.0001
Nonopioid analgesic 12.0 (10.7–13.6) 27.6 (22.5–33.4) 2.79 (2.06–3.79) <.0001
Benzodiazepines 21.3 (19.5–23.2) 37.4 (31.6–43.5) 2.21 (1.68–2.91) <.0001
Antidepressant 22.0 (20.2–23.9) 35.8 (30.1–41.9) 1.98 (1.50–2.61) <.0001
Antipsychotics 9.0 (7.8–10.3) 15.6 (11.6–20.6) 1.87 (1.29–2.71) .001

Bold is significant at p < .001

14 PREVALENCE AND CHARACTERISTICS

APPENDIX 2

Comparison of older people (65 years and over) with and without history of chronic non-
cancer, Australia, 2014

Bivariate comparisons
Older people
Older people (65+) with a
(65 + years old) recorded
without a recorded history of
history of CNCP CNCP
n = 283 n = 122
% (95% CI) % (95% CI) OR (95% CI) p value

Mean age (SD) 75.6 (8.0) 77.5 (8.5) 1.88 (0.15–3.62) .034
Male 79.2 (74.0–83.5) 68.0 (59.2–75.8) 1.78 (1.11–2.87) .017
Married/de facto 48.1 (42.3–53.9) 42.6 (34.1–51.6) 0.80 (0.52–1.23) .315
Employed 9.2 (6.3–13.2) 4.9 (2.2–10.6) 0.51 (0.20–1.28) .150
Rural/remote 39.9 (34.4–45.8) 32.2 (24.4–41.1) 0.72 (0.46–1.21) .144
Mental health
Any mental health 51.6 (45.7–57.4) 57.4 (48.4–65.9) 1.26 (0.82–1.94) .285
illness
Depression 45.6 (39.8–51.4) 54.1 (45.1–62.8) 1.41 (0.92–2.15) .116
Anxiety 9.5 (6.6–13.6) 11.5 (6.9–18.5) 1.23 (0.62–2.43) .554
3 or more MH 4.2 (2.4–7.3) 3.3 (1.2–8.5) 0.77 (0.24–2.42) .649
problems
Physical health
Any physical health 51.2 (45.4–57.0) 71.3 (62.6–78.7) 2.36 (1.50–3.73) <.0001
illness
3 or more physical 8.8 (6.0–12.8) 16.4 (10.8–24.1) 2.02 (1.08–3.80) .029
health problems
Method
Hanging 35.3 (30.0–41.1) 33.6 (25.7–42.5) 0.93 (0.59–1.45) .738
Poisoning—drugs 15.9 (12.1–20.7) 32.8 (25.0–41.7) 2.58 (1.57–4.23) <.0001
Toxicology—drugs detected
Alcohol 21.2 (16.8–26.4) 25.4 (18.4–34.0) 1.27 (0.77–2.08) .352
Opioids 17.7 (13.6–22.6) 32.8 (25.0–41.7) 2.273 (1.40–3.70) .001
Nonopioid analgesic 23.0 (18.4–28.3) 41.0 (32.5–50.0) 2.33 (1.45–3.67) <.0001
Benzodiazepines 30.4 (25.3–36.0) 42.6 (34.1–51.6) 1.70 (1.10–2.64) .018
Antidepressant 26.1 (21.3–31.6) 27.9 (20.6–36.6) 1.09 (0.68–1.76) .719
Antipsychotics 7.4 (4.9–11.1) 6.6 (3.3–12.6) 0.88 (0.38–2.04) .757

Bold is significant at p < .001