The suicide rates in the deliberate self-poisoning (DSP) vs control groups were 278 per 100 000 person-years and 7 per 100 000 person-years, respectively (hazard ratio, 41.96; 95% CI, 27.75-63.44).
The mortality rates in the 2 groups were 1107 per 100 000 person-years and 237 per 100 000 person-years, respectively (hazard ratio, 5.55; 95% CI, 5.12-6.02).
eTable 1. External Cause of Death for Cases.
eTable 2. External Cause of Death for Controls.
Finkelstein Y, Macdonald EM, Hollands S, Sivilotti MLA, Hutson JR, Mamdani MM, Koren G, Juurlink DN, for the Canadian Drug Safety and Effectiveness Research Network (CDSERN). Risk of Suicide Following Deliberate Self-poisoning. JAMA Psychiatry. 2015;72(6):570-575. doi:10.1001/jamapsychiatry.2014.3188
Suicide is the tenth leading cause of death in the United States, and its rate has risen by 16% in the past decade. Deliberate self-poisoning is the leading method of attempted suicide. Unlike more violent methods, which are almost universally fatal, survival following self-poisoning is common, providing an opportunity for secondary prevention. However, the long-term risk of suicide following a first episode of self-poisoning is unknown.
To determine the risk of suicide and mortality from other causes following a first self-poisoning episode.
Design, Setting, and Participants
Population-based cohort study using multiple linked health care databases. We identified all individuals with a first self-poisoning episode in Ontario, Canada, from April 1, 2002, through December 31, 2010, and followed up all surviving participants until December 31, 2011, or death, whichever occurred first. For each individual with a deliberate self-poisoning episode, we randomly selected 1 control from the same population with no such history, matched for age (within 3 months), sex, and calendar year.
Main Outcomes and Measures
The primary analysis examined the risk of suicide following discharge after self-poisoning. The secondary analyses explored factors associated with suicide and examined the risk of death caused by accidents or any other cause.
We identified 65 784 patients (18 482 [28.1%] younger than 20 years) who were discharged after a first self-poisoning episode. During a median follow-up of 5.3 years (interquartile range, 3.1-7.6 years), 4176 died, including 976 (23.4%) by suicide. The risk of suicide following self-poisoning was markedly increased relative to controls (hazard ratio, 41.96; 95% CI, 27.75-63.44), corresponding to a suicide rate of 278 vs 7 per 100 000 person-years, respectively. The median time from hospital discharge to completed suicide was 585 days (interquartile range, 147-1301 days). Older age, male sex, multiple intervening self-poisoning episodes, higher socioeconomic status, depression, and recent psychiatric care were strongly associated with suicide. Patients with a self-poisoning episode were also more likely to die because of accidents (hazard ratio, 10.45; 95% CI, 8.10-13.47) and all causes combined (hazard ratio, 5.55; 95% CI, 5.12-6.02).
Conclusions and Relevance
A first self-poisoning episode is a strong predictor of subsequent suicide and premature death. Most suicides occur long after the index poisoning, emphasizing the importance of longitudinal, sustained secondary prevention initiatives.
Suicide is a global public health concern, claiming more than 1 million lives each year.1 In the United States, suicide rates have increased by more than 16% during the past decade, suggesting that current prevention strategies have limited effect. Suicide accounts for more than 38 000 deaths2 and approximately 1.5 million years of life lost annually3 and is the tenth leading cause of death in the United States, ahead of traffic fatalities, liver disease, and septicemia.4,5 In Canada, suicide is the second leading cause of death in individuals aged 15 to 35 years, comparable with rates in the United States (10.8 and 12.7 per 100 000 persons, respectively).6,7 Worldwide, there are approximately 20 attempts for every completed suicide.8 Aside from the loss of life, suicide can have devastating consequences for the family, friends, and community of the deceased.4
Deliberate self-poisoning is the most common method of attempted suicide, accounting for 85% to 95% of suicide-related hospital admissions.9 Although the intent of lethality varies widely, self-poisoning accounts for roughly 1 in 300 visits to North American emergency departments and generates substantial costs to the health care system.10 Surviving patients are at risk for repeated suicide attempts using more violent means, which are frequently fatal.11- 13 However, the magnitude and temporal course of this risk have not been well characterized.
Limited understanding of the risk factors and motivation for self-harm makes suicide prevention one of the most challenging areas in clinical practice,14 leading some experts to suggest that suicide prevention is an elusive public health goal.15 Screening instruments for suicide often fail to predict completed suicide,16 and optimal follow-up recommendations for at-risk patients have not been established. Yet health care professionals, psychiatrists in particular, are often called on to predict and, by extension, prevent suicide, particularly in patients who have previously harmed themselves.17 Evidence regarding risk factors for suicide after a self-harm episode is sparse.
Because little is known about the long-term outcomes of patients who have self-poisoned, attention has traditionally focused on the immediate postattempt period. We sought to determine the risk of suicide and other causes of mortality following a first self-poisoning episode in a large population over a decade.
Using multiple linked health care databases, we conducted a population-based cohort study of all residents of Ontario, Canada’s most populous province. Its roughly 13 million residents have access to universal health care, including physicians’ services and hospital care. The study was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre and the requirement for patient consent was waived.
We identified all emergency department visits and hospitalizations for deliberate self-poisoning using the National Ambulatory Case Reporting System and the Canadian Institute for Health Information Discharge Abstract Database, respectively. We obtained physician claims data from the Ontario Health Insurance Plan database and demographic information from the Registered Persons Database. We ascertained vital statistics using the Ontario Registrar General-Death database, which contains the manner and cause of death of all fatalities up to December 31, 2011. For Ontarians who die of unnatural causes, including suicide, a coroner’s investigation determines the official cause of death and is registered in this database. These databases were linked in an anonymous fashion using encrypted identifiers and are routinely used to study health outcomes at the population level.18- 21
We identified all individuals who attended an emergency department or were admitted to the hospital for a first episode of deliberate self-poisoning in Ontario between April 1, 2002, and December 31, 2010. All surviving participants were followed up until death or the end of the study period (December 31, 2011), whichever occurred first. Volition of self-poisoning was determined using external cause-of-injury codes (codes X60-X69) in the International Classification of Diseases, Tenth Revision. To focus exclusively on the first episode of self-poisoning, we restricted our analysis to patients with no previous hospitalization or emergency department visits for intentional poisoning (International Classification of Diseases, Ninth Revision, codes E950-E952) in the preceding 10 years. The discharge date for the first self-poisoning episode served as the cohort entry date for all analyses.
Patients who were given a psychiatric diagnosis had at least 1 of the following: (1) a hospital admission, (2) an emergency department visit, or (3) an outpatient physician claim for each diagnosis of interest. For each individual with a deliberate self-poisoning episode, we identified potential controls among Ontarians of the same age (within 3 months) and sex who had no history of deliberate self-poisoning on the case’s hospital discharge date. From this pool, we randomly selected 1 control for each case. These individuals constituted the reference group.
The primary outcome was suicide following discharge from the hospital after deliberate self-poisoning. We ascertained the suicide rates in the self-poisoning and control cohorts and conducted secondary analyses of accidental death and all-cause mortality. We used Kaplan-Meier curves to characterize the timing of the suicide and all-cause mortality following self-poisoning and used a multivariable Cox proportional hazards regression model to explore the potential risk factors for suicide. A second self-poisoning episode was treated as a time-varying covariate in the model to reflect the potential change in the risk of suicide over time depending on repeated episodes.
All analyses were performed using SAS, version 9.3 (SAS Institute), and a 2-tailed type I error rate of P = .05 as the threshold for statistical significance.
During the 10-year study period, we identified 72 509 individuals who presented to the hospital with deliberate self-poisoning. We excluded 6047 (8.3%) with a previous episode of self-poisoning in the preceding 10 years as well as 128 non-Ontarians and 26 individuals with incomplete data. Of the remaining 66 308 patients, 28 732 (43.3%) were hospitalized, most (18 151 [63.2%]) in critical care units. Overall, 524 patients (0.7%) died in the hospital following their first self-poisoning episode.
Our remaining analyses focused on the 65 784 patients who survived until hospital discharge. The median age at discharge following a first self-poisoning episode was 32 years (interquartile range, 20-45 years) and almost two-thirds (61.7%) were women. These patients were followed up for a median of 5.3 years (interquartile range, 3.1-7.6 years) and a cumulative follow-up of 349 904 person-years. The characteristics of the cohort are shown in Table 1.
Of the 65 784 patients discharged following self-poisoning, 4176 died during follow-up, including 976 (23.4% of all deaths) who committed suicide. The risk of suicide following self-poisoning was more than 40-fold higher than among population-based controls (hazard ratio, 41.96; 95% CI, 27.75-63.44), corresponding to a suicide rate of 278 per 100 000 person-years vs 7 per 100 000 person-years, respectively (Table 2 and Figure 1). The median time from hospital discharge to completed suicide was 585 days (interquartile range, 147-1301 days). Of the 976 suicide deaths, 107 were teenagers, representing most of the 188 fatalities in this age group, with a median time of 789 days (interquartile range, 388-1529 days) between the first self-poisoning episode and suicide. Of 107 suicides by teenagers following a self-poisoning episode, 21 (19.6%) eventually committed suicide by overdose. In contrast, only 1 suicide and 34 total deaths were recorded in the respective teenage control group. Overall, in 40.9% (n = 399) of completed suicides following a first self-poisoning episode, overdose was the eventual mechanism (eTable 1 in the Supplement). Most (371 [93.0%]) died in the prehospital setting.
The strongest predictor of eventual suicide following deliberate self-poisoning was advanced age, with risk increasing in a monotonic fashion across each decade of adult life (P < .001) (Table 3). Other risk factors included male sex, multiple intervening self-poisoning episodes, higher socioeconomic status, a formal diagnosis of depression, and psychiatric care in the year preceding the first self-poisoning episode. A diagnosis of alcohol dependence was not independently associated with suicide (Table 3).
All-cause mortality following self-poisoning was 1107 per 100 000 person-years compared with 237 per 100 000 person-years in the control group (hazard ratio, 5.55; 95% CI, 5.12-6.02) (Figure 2). Nearly half (1885 [45.1%]) of all deaths following self-poisoning resulted from suicides, accidents, or undetermined intent (Table 2). Of note, the risk of accidental death was more than considerably higher among self-poisoning survivors relative to controls (hazard ratio, 10.45; 95% CI, 8.10-13.47) (Table 2), as was the risk of death caused by natural progression of disease (Table 2). eTables 1 and 2 in the Supplement detail the specific mechanisms of suicidal and accidental death in both cohorts.
In this population-based study, we found that a first hospital presentation for self-poisoning is a powerful predictor of subsequent suicide and premature death. The interval from the first self-poisoning episode to suicide was longer than 18 months in most cases. While nearly all individuals survive their first hospital presentation for self-poisoning, the risk of subsequent suicide was more than 40-fold higher in these patients than among population-based controls. Suicide accounted for one-fourth of all fatalities in this group during the following decade, suggesting a major opportunity for secondary prevention. Older age, male sex, multiple self-poisoning episodes, higher income, depression, and prior psychiatric care were strong predictors of suicide following a first self-poisoning episode. A recent review22 of suicide studies not restricted to self-poisoning reported that older age, suicidal ideation, and previous suicide attempts were the strongest predictors of completed suicide, while living alone, male sex, and alcohol abuse were weaker predictors.
We also found that self-poisoning was associated with a markedly increased risk of accidental death, which, together with suicide, accounted for nearly half of all deaths in the following decade. We suspect that some accidental deaths or deaths of indeterminate intent are, in fact, suicides that were not classified as such by investigating coroners in the absence of definitive proof of intent.23,24 Examples include falls from height or motor vehicle collisions without a suicide note. In addition, we observed an increased rate of death caused by progression of underlying disease in the self-poisoning cohort (Table 3). Previous research suggests that social deprivation and underuse of medical care contribute to this observation in patients who survive self-poisoning25 and other types of self-harm,26 but various unmeasured determinants of health may also play a role.
Our finding that suicide risk persists long after the initial self-poisoning episode is of particular importance because it emphasizes the need for longitudinal surveillance and sustained secondary prevention initiatives in this high-risk population. Recent evidence suggests that ongoing communication with patients following self-poisoning substantially reduces the subsequent risk of attempted and completed suicide.27,28 In contrast, short-term interventions have been largely unsuccessful at reducing suicide rates, particularly when guided by patient reports of suicidal intent.29 Self-reported suicide motivation is often unreliable, as evidenced by the high rate of suicide among psychiatric patients shortly after hospital discharge.30,31 Conversely, the predictors of suicide identified in our study are objective. Although some have been previously documented, those predictors can facilitate risk stratification following a first self-poisoning episode to better target prevention efforts.
Existing publications regarding the temporality and risk of suicide following deliberate self-harm are encumbered by factors that limit the inferences that can be drawn, including limited follow-up (frequently up to 1 year)32 in many studies, lack of a defined inception cohort, and inclusion of small (eg, single institution)25 or highly selective populations30,31,33 (eg, psychiatric inpatients). Our study addresses many of those limitations by examining a large population of patients for up to 10 years following a first self-poisoning episode compared with a matched reference group from the general population. The findings establish that the rates of suicide and premature mortality following self-poisoning are considerably higher than previously appreciated.25,32 In a single-center study from Nottingham, England,13 912 patients with deliberate self-poisoning were observed for up to 16 years after their initial presentation. After 1 year and 10 years, respectively, 5 (0.5%) and 26 individuals (2.9%) had committed suicide.
Some limitations of our study merit emphasis. Our data sources do not permit a detailed exposition of the circumstances of each self-poisoning event or individual motivations of subsequent suicide. Some self-poisoning episodes may have been impulsive and not of fatal intent, particularly in younger people. As noted, the coroner’s determination of the manner of death likely underestimates suicide, in part because of the stigma associated with suicide and the financial consequences for the family. Therefore, the true incidence of suicide following self-poisoning is likely even higher than our study suggests.
We found that a first episode of deliberate self-poisoning is an exceedingly strong predictor of subsequent suicide and premature death. Suicide accounts for roughly 1 in 4 fatalities in the decade following self-poisoning. Many additional deaths are assigned an undetermined intent, suggesting that the true rate of subsequent suicide is even higher. Most suicides occur long after the index episode. Our findings regarding long-term suicide risk emphasize the need for sustained and focused prevention efforts, particularly in patients with objective risk factors.
Submitted for Publication: July 23, 2014; final revision received November 6, 2014; accepted December 22, 2014.
Corresponding Author: Yaron Finkelstein, MD, ABCP(Dip), Department of Pediatrics, The Hospital for Sick Children, University of Toronto, 555 University Ave, Toronto, ON M5G 1X8, Canada (firstname.lastname@example.org).
Published Online: April 1, 2015. doi:10.1001/jamapsychiatry.2014.3188.
Author Contributions: Drs Finkelstein and Juurlink had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Finkelstein, Macdonald, Hollands, Sivilotti, Hutson, Mamdani, Juurlink.
Drafting of the manuscript: Finkelstein, Macdonald, Hollands, Sivilotti, Hutson, Juurlink.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Finkelstein, Macdonald, Hollands, Hutson, Mamdani, Koren.
Obtained funding: Juurlink.
Administrative, technical, or material support: Finkelstein, Macdonald, Hutson, Juurlink.
Study supervision: Finkelstein, Sivilotti, Juurlink.
Conflict of Interest Disclosures: Dr Mamdani reports serving on advisory boards and/or receiving honoraria from AstraZeneca, Bristol-Myers Squibb, Eli Lilly and Company, GlaxoSmithKline, Hoffman–La Roche, Novartis AG, Novo Nordisk, and Pfizer. No other disclosures were reported.
Funding/Support: This study was supported by an Emerging Teams grant from the Canadian Institutes of Health Research, grant 92247 from the Canadian Drug Safety and Effectiveness Research Network (Drs Juurlink and Mamdani), and the Institute for Clinical Evaluative Sciences, a nonprofit research institute sponsored by the Ontario Ministry of Health and Long-Term Care.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Group Information: The Canadian Drug Safety and Effectiveness Research Network (CDSERN) group members were Colin R. Dormuth, ScD, Therapeutics Initiative, Pharmacology and Therapeutics, and Department of Anesthesiology, University of British Columbia, Vancouver, Canada; Collette B. Raymond, PharmD, MSc, Department of Community Health Sciences and Manitoba Center for Health Policy, University of Manitoba, Winnipeg, Canada; Anita Kozyrskj, PhD, Department of Pediatrics, University of Alberta, Edmonton, Canada; Yola Moride, PhD, Faculty of Pharmacy, Université de Montréal, Montréal, Canada; and Michael Paterson MSc, Chronic Disease and Pharmacotherapy, Institute for Clinical Evaluative Sciences, Toronto, Canada.
Disclaimer: The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding sources. No endorsement by the Institute for Clinical Evaluative Sciences or the Ontario Ministry of Health and Long-Term Care is intended or should be inferred.
Additional Contributions: Brogan Inc, Ottawa, provided the Drug Product and Therapeutic Class Database.