Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Jan 1.
Published in final edited form as: Schizophr Res. 2022 Dec 31;251:82–90. doi: 10.1016/j.schres.2022.12.004

Means of Suicide Among Adults with Schizophrenia Across the Life Span

Natalie Bareis a, Mark Olfson a,b, Tobias Gerhard c, Stephanie Rolin a, T Scott Stroup a
PMCID: PMC9872522  NIHMSID: NIHMS1860017  PMID: 36592524

Abstract

Background and Hypotheses:

Adults with schizophrenia have increased risk of suicide with highest risk among younger adults. We investigated whether means of suicide among these adults were different from the general population.

Study Design:

This retrospective longitudinal analysis used the National Death Index to characterize means of suicide among 4 cohorts of Medicare patients with schizophrenia (2007–2016) by age: 18 to 34, 35 to 44, 45 to 54, and aged 55+ years. Means of suicide were categorized by age at death and sex. Adjusted hazard ratios were calculated for common means. Mortality rates per 100,000 person-years were estimated by age group stratified by sex, and standardized to the general population by age, sex, and race-ethnicity using standardized mortality ratios.

Study Results:

668,836 adults were included with 2,218 suicide decedents: 1,444 men and 774 women. The most common means of suicide was poisoning (36.8%), with a significant sex difference by means: 55.9% of women died by poisoning, 13.8% by firearms, 11.0% by hanging and 9.4% by jumping, while among men suicide by poisoning (26.6%), firearms (25.5%), and hanging (24.2%) were similar, followed by jumping (12.0%). Suicide by poisoning among the schizophrenia cohort was 10 times that of the general population, while suicide by firearm was twice that of the general population.

Conclusions:

Means of suicide differed for patients with schizophrenia compared to the general population: poisoning was the most common means among men and women with schizophrenia, while firearms accounted for over half of all suicides in the general U.S. population.

Keywords: schizophrenia, suicide, psychiatric epidemiology

1. Introduction

Suicide prevention is an important public health goal for the treatment of people with schizophrenia (Laursen et al., 2014; Lu et al., 2019). Individuals with schizophrenia in the United States have a lifetime risk of suicide of approximately 4.9%, representing an important cause of premature death in this population (Hor and Taylor, 2010; Palmer et al., 2005). Among adults with schizophrenia, the suicide rate is about 4.5 times higher than the general population (Olfson et al., 2021). Unlike the general population, the highest risk of suicide is among younger adults, with highest risk in the first decade after initial diagnosis (Sher and Kahn, 2019) and declines with increased age for both men and women, perhaps reflecting healthy survivor bias (Laursen et al., 2007; Moride and Abenhaim, 1994). Because of the young age of people who die by suicide, it is the means of death associated with the highest mean potential years of life lost (Suicide Deaths in the United States, 2021; Olfson et al., 2015). However, it is unknown if means of suicide among adults with schizophrenia are different from the general population or if the means vary across the life span and by sex.

Identifying common means of suicide can focus means reduction efforts to reduce overall suicide risk (Lubin et al., 2010), as restricting access to lethal means has been one of the most successful suicide prevention strategies around the world (Wu et al., 2021). Within the general US population, firearms, suffocation (including hanging), and poisoning (including drug overdose) account for over 90% of suicide deaths (Suicide Deaths in the United States, 2021; Web-based Injury Statistics Query and Reporting System (WISQARS), 2021). In men in 2020 the most common means were firearms (58.1%), suffocation (26.3%), and poisoning (7.9%).

However, during that same period, in women firearms (33.8%), suffocation (27.4%), and poisoning (29.3%) were similarly common (Web-based Injury Statistics Query and Reporting System (WISQARS), 2021). Reducing access to firearms has been a successful method for reducing risk of suicide not only by firearms (Shenassa et al., 2004) but also total suicide risk, and has been particularly effective with youth (Lubin et al., 2010). In addition, multiple methods used to address access to toxic medications have been effective at reducing suicide by poisoning. These include limiting the size of packs of over-the-counter analgesics and limiting the number of pills dispensed by pharmacies to short-term supplies (Hawton et al., 2004). In some parts of the world, reducing access to organophosphate pesticides has been associated with reduced suicide rates (Bonvoisin et al., 2020). Barriers such as fences or safety nets on bridges and other high places used for jumping have been found to reduce the number of suicides by jumping (Okolie et al., 2020). However, it has been difficult to identify effective deterrents for suicide by hanging in the community due to the accessibility of ligatures and suspension points, and hanging has been increasing over time in all age groups (Gunnell et al., 2005).

Previous studies of the risk and means of suicide among people with schizophrenia and without have used relatively small sample sizes (fewer than 100 cases (Feigelman et al., 2016; Palmer et al., 2005; Shields et al., 2007)), that were not nationally representative (Bhatia et al., 2006; Carpenter et al., 2021), often conducted outside the US (Bonvoisin et al., 2020; Hor and Taylor, 2010; Ishii et al., 2014; Lubin et al., 2010; Wu et al., 2021), with samples that are relatively old (Lubin et al., 2010; Shenassa et al., 2004), or focused on one means of suicide (Brown et al., 2018; Miller et al., 2020). To our knowledge this is the largest study of a national sample of adults with schizophrenia to identify specific means of suicide in the US stratified by age group and sex. The aims of these analyses were to examine means of suicide in a nationally representative sample of adult Medicare recipients with schizophrenia, risk of suicide by these means in this cohort, and to compare these means to those in the general US population. Understanding differences in means and rates of suicide in individuals with schizophrenia compared to the general population, and the extent to which they vary by sex and across the life span, can inform public health strategies and clinical services designed to reduce suicide risk in individuals with schizophrenia.

2. Material and Methods

2.1. Sample

These retrospective longitudinal analyses were performed with a 50% random nationally representative sample of Medicare Fee-For-Service patients with Part D prescription drug coverage from January 1, 2007, to December 31, 2016, obtained from the Centers for Medicaid and Medicare Services (CMS). Eligibility for Medicare services is limited to individuals aged 65 or older or any age including younger individuals with disabilities as determined by the Social Security Administration. These data were linked to National Death Index data, which provides dates and underlying cause of death information for individuals who died during the study period. Total US resident population death information was obtained from the 2007–2016 Centers for Disease Control and Prevention WONDER Underlying Cause of Death database (CDC WONDER. Underlying Cause of Death, 1999–2020, 2021). The New York State Psychiatric Institute and Rutgers University Institutional Review Boards approved this study and deemed that informed consent was not required for use of this retrospective, deidentified data.

2.2. Cohort Construction

Using these Medicare data to create the cohort, we included patients with schizophrenia, defined as a patient receiving a diagnosis of a schizophrenia-spectrum disorder (International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), code 295.X or International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), codes F20.X and F25.X) during two or more outpatient visits within 365 days of each other or one or more inpatient visits (Lurie et al., 1992). The date on which patients met criteria for schizophrenia may not coincide with the first observed diagnosis of schizophrenia in the study data. The start of follow-up (i.e., index date) was the first date that patients met criteria for schizophrenia and had 12 months of continuous Medicare enrollment. Patients 18 years of age or older on their index date were included in the sample. The sample was broken down into 4 age cohorts at index (18–34, 35–44, 45–54, 55+ years). Patients were censored at either their date of death, loss of Medicare enrollment, age at entrance into a new age cohort, five years following their index date, or right censored on December 31, 2016, whichever came first. Censoring at 5 years following index addresses concerns that the accuracy of comorbidity measurements collected prior to the index date erodes over time and that beyond 5 years, they are likely less clinically meaningful.

2.3. Study outcomes

The study outcomes were means of death by suicide. We grouped the means into commonly used categories including poisoning (ICD-10-CM codes X60–64, X66–69), firearms (X72-X74), hanging (X70), jumping (X80-X81), cutting or piercing (X78), drowning (X71) and other means that included burning (X76), car crash (X82), intentional self-harm (Y87.0) and other (X83-X84) (Conner et al., 2019). A detailed breakdown of each ICD-10-CM means of suicide is included in Supplemental Table 1.

The final cohort included 668,836 patients with 2,997,308 years of follow-up and included 2,218 patients with death by suicide (X60-X84 and Y87.0) listed as cause of death in the linked National Death Index data (Olfson et al., 2021).

2.4. Sociodemographic and clinical characteristics

Characteristics were identified for the total sample, by 4 age cohorts (18–34, 35–44, 45–54, and 55+ years), and stratified by sex (female and male). Demographic characteristics included race-ethnicity (defined as non-Hispanic Black, non-Hispanic White, Hispanic and an Other category that included Native American or Alaskan, Asian, Native Hawaiian, other Pacific Islander, more than one race, unknown and missing race-ethnicity) and census region of the United States (South, Midwest, West and Northeast) (2010 Census Regions and Divisions of the United States, 2010) to address the geographic variation in availability of different means of suicide (Shrira and Christenfeld, 2010). Variables representing clinical comorbidities were based on diagnoses from claims in all treatment settings including outpatient mental health or physical health visits and included psychiatric diagnoses and indicators of general medical health that have significant impacts on the health of individuals with schizophrenia. In addition, we included variables representing inpatient admissions and emergency department visits, given their strong association with suicide risk (Ahmedani et al., 2014; Goldman-Mellor et al., 2019; Olfson et al., 2016). Comorbid clinical diagnoses and service use were ascertained 12 months prior to patients’ index date (not immediately prior to death) and included any ICD code indication (in any position) of these conditions for at least one claim regardless of service type. Psychiatric diagnoses included depressive, anxiety, alcohol use, and drug use disorders (excluding nicotine), suicidal ideation, and suicide attempt or self-injury (i.e., deliberate self-harm). General health indicators included sleep disorders and chronic pain (see Supplemental Table 2 for ICD-10-CM codes used to define the diagnoses). Patients’ Elixhauser Comorbidity Index score, a commonly used claims-based measure to quantify overall burden of comorbid diagnoses, was calculated using these comorbid clinical diagnoses excluding alcohol abuse, drug abuse, psychosis, and depression comorbidities from the score (see Supplemental Table 4 for diagnoses algorithms used to calculate Elixhauser Comorbidity scores) (Quan et al., 2005; van Walraven et al., 2009). Service use measures included any inpatient hospitalizations or emergency department (ED) visits and two 3-level variables were separately classified by principal diagnosis codes as mental health, substance use, or nonbehavioral health inpatient admissions and ED visits (see Supplemental Table 2 for ICD-10-CM codes used to define service use classification).

2.5. Statistical Analyses

Data analyses were performed from June 1 to December 31, 2021. Using the full sample of patients with schizophrenia (N=668,836) and the subsample that died by suicide (N=2,218), we examined the sociodemographic and clinical characteristics of the samples and then stratified the suicide decedents by sex to determine whether there were sex differences in these characteristics. We then determined the means of suicide for the total sample of decedents, then stratified by age groups at time of death (18–34, 35–44, 45–54, and 55 and over) and finally by sex.

We then examined the sociodemographic and clinical characteristics of the decedents who used the 4 most common means of suicide. Separate Cox proportional hazards regression models were fit for each of the means and adjusted for covariates to account for variation found by age, sex, comorbidities, and service use. The covariates included the four age cohorts, sex, race-ethnicity, region of the United States, psychiatric diagnoses, indicators of general medical health, and service use characteristics. Any inpatient hospitalizations or ED visits were analyzed in separate regressions from the two 3-level mental health, substance use, or nonbehavioral health classified inpatient and ED visits due to co-linearity. Time until suicide or censoring during follow-up were the dependent variables and the demographic and clinical characteristics were the independent variables of interest.

Finally, we calculated the person-level age-, sex-, and race-ethnicity-adjusted suicide standardized mortality ratios (SMRs) with 95% confidence intervals (CIs). The observed suicide deaths during follow-up for each means of suicide were compared to the expected annual suicide rates for those means derived from the general US population using 2007–2016 CDC WONDER data (CDC WONDER. Underlying Cause of Death, 1999–2020, 2021). Suicide SMRs were calculated for those means for the total sample, by age group and stratified by sex. For the SMR analysis, each suicide death was assigned to the age group in which it occurred, therefore calculations of person-years were censored at either their date of death or right censored on December 31, 2016, whichever came first, and not at loss of Medicare fee-for-service enrollment, age of entering a different cohort, or 5 years from ascertainment.

3. Results

Table 1 details characteristics of the total sample of patients with schizophrenia at index, or the time of cohort entry, and of the suicide decedents. The mean age of the total sample was 53.1 years, with the majority male, non-Hispanic White and over one third lived in the South. Among the total sample, the most common comorbid mental health conditions were depressive disorders while chronic pain was the most common comorbid physical condition. Few patients had a history of suicidal ideation or deliberate self-harm. The majority of acute health service use was an ED visit for nonbehavioral health.

Table 1.

Characteristics of the total sample, suicide decedents and decedents by sex in the Medicare program

Total Sample Suicide Decedents
Total Sex
Female Male Differencec Between Sexes (p-value)
N (%) 668,836 2,218 774 (34.90) 1444 (65.10)
Age (M, SD) 53.12 (16.26) 44.11 (12.39)d 46.78 (12.61)d 42.68 (12.03)d <0.01
Age Group (N, %) <0.01
 18–34 95,195 (14.23) 434 (19.57)d 110 (14.21)d 324 (22.44)d
 35–44 112,283 (16.79) 480 (21.64)d 142 (18.35)d 338 (23.41)d
 45–54 166,939 (24.96) 691 (31.15)d 256 (33.07)d 435 (30.12)d
 55+ 294,419 (44.02) 613 (27.64)d 266 (34.37)d 347 (24.03)d
Sex (N, %)
 Female 317,892 (47.53)
 Male 350,943 (52.47)
Race (N, %) 0.26
 Black 159,903 (23.91) 176 (7.94) 50 (6.46) 126 (8.73)
 White 423,744 (63.36) 1,781 (80.30) 635 (82.04) 1,146 (79.36)
 Hispanic 59,150 (8.84) 143 (6.45) 47 (6.07) 96 (6.65)
 Othera 26,039 (3.89) 118 (5.32) 42 (5.43) 76 (5.26)
Region (N, %) * 0.82
 South 240,642 (36.02) 706 (31.83) 248 (32.04) 458 (31.72)
 Midwest 160,523 (24.02) 562 (25.34) 187 (24.16) 375 (25.97)
 West 113,287 (16.95) 487 (21.96) 173 (22.35) 314 (21.75)
 Northeast 153,705 (23.00) 463 (20.87) 166 (21.45) 297 (20.57)
Comorbid diagnoses b
 Depressive (N, %) 277,406 (41.5) 1,211 (54.60) 507 (65.50) 704 (48.75) <0.01
 Anxiety (N, %) 170,587 (25.5) 753 (33.95) 324 (41.86) 429 (29.71) <0.01
 Alcohol Use (N, %) 52,758 (7.9) 329 (14.83) 87 (11.24) 242 (16.76) <0.01
 Drug Use (N, %) 129,071 (19.3) 864 (38.95) 287 (37.08) 577 (39.96) 0.19
 Sleep disorders (N, %) 90,416 (13.5) 396 (17.85) 171 (22.09) 225 (15.58) <0.01
 Chronic pain (N, %) 288,465 (43.1) 920 (41.48) 402 (51.94) 518 (35.87) <0.01
 Suicidal Ideation (N, %) 64,565 (9.7) 532 (23.99) 205 (26.49) 327 (22.65) 0.04
 Suicide attempt or self-injury (N, %) 10,704 (1.6) 206 (9.29) 85 (10.98) 121 (8.38) 0.04
 Elixhauser Comorbidity Score (N, %) <0.01
  0 402,633 (60.20) 1,260 (56.81) 360 (46.51) 900 (62.33)
  1–2 125,484 (18.76) 605 (27.28) 241 (31.14) 364 (25.21)
  3+ 140,719 (21.04) 353 (15.92) 173 (22.35) 180 (12.47)
Use of Acute Health Services b
 Any inpatient admission (N, %) 321,548 (48.1) 1,330 (59.96) 514 (66.41) 816 (56.51) <0.01
  Mental Health (N, %) 191,026 (28.6) 1,080 (48.69) 408 (52.71) 672 (46.54) <0.01
  Substance Use (N, %) 12,753 (1.9) 96 (4.33) 26 (3.36) 70 (4.85) 0.10
  Nonbehavioral health (N, %) 189,623 (28.4) 593 (26.74) 259 (33.46) 334 (23.13) <0.01
 Any ED visit (N, %) 409,104 (61.2) 1,590 (71.69) 621 (80.23) 969 (67.11) <0.01
  Mental Health (N, %) 173,919 (26.0) 965 (43.51) 373 (48.19) 592 (41.00) <0.01
  Substance use (N, %) 18,818 (2.8) 131 (5.91) 40 (5.17) 91 (6.30) 0.28
  Nonbehavioral health (N, %) 351,950 (52.6) 1,278 (57.62) 517 (66.80) 761 (52.70) <0.01
Open in a new tab

Data are from US Medicare 2007–2016 files.

*

Region is missing for 679 non-decedents

a.

Includes Native American or Alaskan, Asian, Native Hawaiian, other Pacific Islander, more than 1 race, unknown, missing race/ethnicity.

b.

Measured 12 months before baseline.

c.

F-test for continuous and χ2 for categorical variables. An α=0.05 was used for all statistical tests, with significance indicated in Bold.

d.

Age at death

The suicide decedents were younger, with an average age of 44.1 years at time of death, and similarly a majority were non-Hispanic White men and almost one-third lived in the South. During the 12 months before index, over half of the decedents had depressive disorders and a higher percentage of almost all comorbidities than the total sample. During the 12 months before index, one-quarter of decedents had suicidal ideation and 9.3% had deliberate self-harm. Almost half of the decedents had an inpatient admission or ED visit for mental health during the 12 months prior to index. In addition, differences between the sexes were significant for all characteristics except race, region, drug use, and hospitalizations and ED visits for substance use 12 months prior to index.

3.1. Means of suicide in the sample of Medicare decedents

Supplemental Table 3 shows the means of death for the suicide decedents. The highest percentage of suicides were by poisoning (36.8%) with drug poisonings accounting for over 90% of all poisoning deaths. Among decedents aged 18–34 years, the highest percentage of suicides were similarly common by hanging, poisoning and firearms. Poisoning was the most common means of suicide for all other age groups, followed by firearms, and hanging. The fraction of deaths by suicide that were by hanging and jumping decreased with age, while the less common means of cutting/piercing and drowning increased with age.

3.2. Sex differences in means of suicide in the sample of Medicare decedents

There were significant differences in means of suicide by sex (Figure 1, χ2=201.13, df=6, p<0.01). Among women, over half of deaths by suicide were due to poisoning, while the percentage of suicide deaths by poisoning, firearms and hanging were similar in men (Figure 1). Means of suicide also varied by age within each sex (women: χ2=31.39, df=18, p=0.03; men: χ2=39.70, df=18, p<0.01). The highest percentage of death by suicide in women across all age groups was by poisoning, and the percentage of suicide by firearms remained nearly constant as age increased. Among the youngest cohort of men, death by hanging had the highest percentage, while men ages 35–54 years had the highest percentage by poisoning. The highest percentage of suicides in the oldest cohort of men was by firearms.

Figure 1.

Figure 1.

Open in a new tab

Means of Suicide for Women and Men Diagnosed with Schizophrenia

3.3. Relative Risk of different means of suicide in the full sample

Results from the Cox proportional hazards regression models are in Table 2. Risk of suicide by drug poisoning was higher in women than men and for patients living in the West compared to the South. Risk of suicide by drug poisoning was lower for patients in the oldest age cohort compared to the youngest cohort, and non-Hispanic Blacks and Hispanics compared to non-Hispanic Whites. Patients with most of the comorbid diagnoses had increased risk of drug poisoning, with the greatest risk in patients with deliberate self-harm and drug use disorders. Patients with inpatient admissions for both mental health and non-mental health and any ED visits had higher risks of suicide by drug poisoning compared to those without inpatient admissions for those purposes and compared to no ED visits. However, risk of suicide by other poisoning means (i.e., solid, liquid or gas) was higher in patients with sleep disorders, among patients with deliberate self-harm, and those with inpatient admissions for mental health. Risk was lower for individuals with suicidal ideation.

Table 2.

Adjusted Hazard Ratios of Suicide Risk by the Most Common Means Among Adults with Schizophrenia in the Medicare Program*

Poison Firearmg Hanging Jumpingh
Druge Solid, Liquid or Gasf
Age Group (ref=18–34)
 35–44 1.23 (0.99–1.53) 0.62 (0.29–1.33) 0.68 (0.52–0.88) 0.69 (0.54–0.89) 0.59 (0.42–0.83)
 45–54 1.08 (0.87–1.34) 0.57 (0.27–1.18) 0.62 (0.48–0.80) 0.54 (0.42–0.70) 0.58 (0.42–0.81)
 55+ 0.58 (0.45–0.74) 0.47 (0.21–1.03) 0.50 (0.38–0.66) 0.33 (0.24–0.45) 0.30 (0.19–0.45)
Sex (ref=male)
 Female 1.47 (1.27–1.71) 0.60 (0.34–1.06) 0.36 (0.29–0.45) 0.35 (0.28–0.45) 0.68 (0.51–0.90)
Race (ref=white)
 Black 0.23 (0.17–0.30) 0.40 (0.16–1.02) 0.25 (0.18–0.34) 0.28 (0.19–0.40) 0.36 (0.23–0.56)
 Hispanic 0.61 (0.45–0.82) 0.69 (0.25–1.95) 0.22 (0.12–0.39) 0.79 (0.57–1.11) 0.73 (0.46–1.17)
 Othera 0.89 (0.62–1.28) 1.73 (0.67–4.44) 0.66 (0.41–1.08) 1.12 (0.75–1.65) 1.47 (0.93–2.33)
Region (ref=South)
 Midwest 1.03 (0.86–1.25) 1.51 (0.74–3.08) 0.64 (0.51–0.80) 1.13 (0.87–1.48) 1.11 (0.75–1.64)
 West 1.31 (1.07–1.61) 1.67 (0.77–3.62) 0.69 (0.53–0.89) 1.68 (1.29–2.19) 1.95 (1.33–2.86)
 Northeast 0.92 (0.75–1.13) 1.41 (0.67–2.98) 0.32 (0.24–0.43) 1.12 (0.86–1.48) 2.09 (1.47–2.97)
Comorbid diagnoses b
 Depressive (ref=No) 1.65 (1.38–1.96) 1.53 (0.83–2.82) 1.33 (1.08–1.64) 1.21 (0.97–1.51) 0.93 (0.69–1.26)
 Anxiety (ref=No) 1.30 (1.10–1.53) 1.35 (0.73–2.49) 1.18 (0.95–1.47) 1.18 (0.94–1.49) 0.86 (0.62–1.20)
 Alcohol Use (ref=No) 1.01 (0.80–1.26) 1.43 (0.61–3.34) 0.99 (0.71–1.36) 1.19 (0.89–1.61) 1.43 (0.96–2.12)
 Drug Use (ref=No) 2.08 (1.73–2.49) 0.65 (0.31–1.39) 1.48 (1.17–1.88) 1.52 (1.20–1.94) 1.45 (1.05–1.99)
 Sleep disorders (ref=No) 1.24 (1.04–1.48) 1.96 (1.04–3.69) 1.25 (0.97–1.61) 0.95 (0.71–1.26) 0.79 (0.52–1.21)
 Chronic pain (ref=No) 1.12 (0.96–1.32) 0.77 (0.43–1.39) 1.10 (0.90–1.35) 0.85 (0.68–1.06) 0.72 (0.53–0.98)
 Suicidal Ideation (ref=No) 1.30 (1.06–1.59) 0.41 (0.17–0.98) 1.24 (0.92–1.67) 1.78 (1.35–2.36) 1.58 (1.09–2.30)
 Suicide attempt or self-injury(ref=No) 2.51 (1.97–3.21) 9.06 (4.02–20.43) 2.27 (1.50–3.44) 2.41 (1.66–3.52) 2.38 (1.38–4.10)
 Elixhauser Comorbidity Index (ref=0)
  1–2 1.02 (0.82–1.26) 0.82 (0.37–1.81) 0.70 (0.52–0.94) 0.78 (0.59–1.04) 1.04 (0.71–1.52)
  3+ 0.77 (0.59–1.00) 0.74 (0.28–1.95) 0.43 (0.29–0.63) 0.55 (0.37–0.82) 0.84 (0.50–1.42)
Use of Acute Health Services b
 Any inpatient admission (ref=No)c 1.05 (0.81–1.37) 2.22 (0.94–5.21) 1.37 (1.03–1.83) 1.46 (1.08–1.97) 1.57 (1.05–2.34)
  Mental Health (ref=No)d 1.34 (1.06–1.69) 2.34 (1.08–5.09) 1.19 (0.88–1.61) 1.43 (1.05–1.95) 1.62 (1.06–2.46)
  Substance Use (ref=No)d 1.24 (0.86–1.77) 0.22 (0.03–1.82) 1.04 (0.57–1.92) 0.69 (0.38–1.27) 0.25 (0.08–0.85)
  Nonbehavioral health (ref=No)d 1.29 (1.05–1.59) 1.63 (0.74–3.58) 1.21 (0.90–1.64) 1.19 (0.88–1.62) 0.97 (0.63–1.48)
 Any ED visit (ref=No)c 1.34 (1.08–1.67) 1.22 (0.57–2.62) 1.15 (0.90–1.46) 1.25 (0.96–1.63) 1.32 (0.93–1.87)
  Mental Health (ref=No)d 0.96 (0.78–1.18) 0.68 (0.32–1.46) 1.10 (0.83–1.45) 1.13 (0.85–1.51) 1.46 (0.99–2.15)
  Substance use (ref=No)d 0.93 (0.67–1.29) 3.13 (0.97–10.09) 0.58 (0.32–1.02) 0.87 (0.55–1.39) 0.98 (0.51–1.88)
  Nonbehavioral health (ref=No)d 1.13 (0.94–1.36) 1.17 (0.60–2.26) 1.08 (0.87–1.34) 1.06 (0.85–1.33) 0.89 (0.66–1.20)
Open in a new tab
*

All covariates were included in the models. An α=0.05 was used for all statistical tests, with significance indicated in Bold.

a.

Includes Native American or Alaskan, Asian, Native Hawaiian, other Pacific Islander, more than 1 race, unknown, missing race/ethnicity

b.

Measure 12 months before index

c.

Results from separate fully adjusted models of any inpatient admission or any ED visit only.

d.

Results from separate fully adjusted models of the 3-level variables classified as mental health, substance use, and non-behavioral health visits.

e.

X60-X64;

f.

X66-X69;

g.

X72-X74;

h.

X80-X81

Risk of suicide by firearms decreased with age, was lower in women compared to men, and for non-Hispanic Blacks and Hispanics compared to non-Hispanic Whites. Patients living in all regions outside of the South compared to patients living in the South, as well as those with higher comorbidity scores had lower risks of suicide by firearms. Only comorbid depression, drug use and deliberate self-harm were associated with higher risk of suicide by firearm, as was any inpatient admission.

Similarly, risk of suicide by hanging was lower in women compared to men, and for non-Hispanic Blacks compared to non-Hispanic Whites. However, patients living in the West had higher risks of suicide by hanging compared to the South. Patients with drug use, suicidal ideation and deliberate self-harm were each at higher risk than patients without those comorbidities, and risk was higher in patients with any inpatient admission and admissions for mental health compared to none.

Finally, risk of suicide by jumping also decreased with age, was lower in women compared to men, and for non-Hispanic Blacks compared to non-Hispanic Whites. Patients living in the West and Northeast had higher risk of suicide by jumping compared to the South. Patients with drug use disorders, suicidal ideation and deliberate self-harm also had higher risk of suicide by jumping. Any inpatient and mental health inpatient admissions had higher risk of suicide by hanging. Patients with chronic pain and inpatient admissions for substance use have lower risk of suicide by hanging.

3.4. Means of suicide of the Medicare decedents relative to the general US population

Results from the standardized mortality ratios (SMRs) in Table 3 and Figure 2 revealed large differences between the schizophrenia sample and the general population in their rates for the different means of suicide. In this schizophrenia sample, the highest suicide rate per 100,000 person-years was for poisoning, nearly 10.5 times that of the general US population. In comparison, suicide by firearms, the most common means in the general population, was twice as common in this sample of decedents with schizophrenia. Suicide by jumping was the highest compared to the general population, approximately 15 times more common among this schizophrenia cohort. Across the age groups, suicide by jumping had the highest SMR in this sample. The suicide rate per 100,000 person-years generally declined with age as did the SMRs for all means. The highest SMR for poisoning was seen in the youngest cohort, ages 18–34 years.

Table 3.

Decedent Suicide Means Standardized Mortality Ratios, Full Sample, by Age Cohorts and by Sex

Total 18–34 35–44 45–54 55+
Method O E SMR O E SMR O E SMR O E SMR O E SMR
Poisoning 27.26 2.60 10.48 (7.47–16.65) 36.32 1.51 24.05 (17.81–32.24) 38.81 3.12 12.44 (9.16–16.35) 34.58 4.34 7.97 (5.81–10.68) 17.22 2.74 6.29 (4.18–9.48)
Firearm 15.85 7.98 1.99 (1.24–2.94) 33.05 6.27 5.27 (3.89–7.21) 21.28 7.08 3.01 (2.06–4.36) 16.60 8.94 1.86 (1.19–2.77) 9.85 10.30 0.96 (0.53–1.53)
Hanging 14.51 3.90 3.72 (2.35–5.70) 39.60 4.89 8.10 (6.02–10.67) 19.82 4.83 4.10 (2.69–5.89) 15.48 4.56 3.40 (2.17–5.15) 6.73 2.32 2.90 (1.49–5.03)
Jumping 8.21 0.52 15.79 (9.22–27.74) 20.62 0.63 32.73 (21.89–47.10) 12.10 0.58 20.86 (13.20–33.94) 8.86 0.61 14.52 (7.86–23.64) 3.83 0.39 9.82 (4.16–18.52)
Cutting/Piercing 3.44 0.28 12.29 (5.80–25.80) 2.95 0.15 19.67 (7.27–37.14) 3.55 0.30 11.83 (5.41–24.08) 4.74 0.45 10.53 (4.89–19.48) 2.76 0.35 7.89 (3.11–15.92)
Othera 3.04 0.26 11.69 (6.24–27.79) 6.87 0.24 28.63 (14.39–48.62) 2.92 0.29 10.07 (3.76–19.21) 4.12 0.34 12.12 (6.48–25.79) 1.63 0.23 7.09 (2.69–16.04)
Drowning 1.70 0.16 10.63 (3.87–23.05) 2.62 0.14 18.71 (7.79–39.80) 1.67 0.17 9.82 (3.64–21.70) 1.87 0.20 9.35 (3.09–18.44) 1.42 0.18 7.89 (3.44–20.49)
Female
Poisoning 31.53 2.52 12.51 (9.01–16.99) 51.59 1.25 41.27 (31.76–52.74) 47.42 2.95 16.07 (12.29–20.80) 45.27 4.11 11.01 (8.40–14.37) 20.00 2.56 7.81 (4.91–11.85)
Firearm 7.79 2.15 3.62 (1.91–6.07) 18.95 1.58 11.99 (7.73–17.23) 10.85 2.34 4.64 (2.65–7.30) 9.12 3.03 3.01 (1.81–5.20) 5.22 2.08 2.51 (1.35–4.92)
Hanging 6.19 1.52 4.07 (2.27–7.68) 20.00 1.94 10.31 (6.47–15.64) 10.85 1.81 5.99 (3.43–9.44 9.43 1.74 5.42 (3.16–9.06) 2.17 0.93 2.33 (1.17–5.99)
Jumping 5.32 0.27 19.70 (10.42–37.93) 15.79 0.27 58.48 (36.68–87.00) 7.43 0.29 25.62 (14.19–45.03) 7.55 0.34 22.21 (12.10–38.41) 2.68 0.23 11.65 (4.74–24.22)
Cutting/Piercing 2.04 0.11 18.55 (9.91–50.65) 1.05 0.05 21.00 (12.37–73.78) 1.71 0.10 17.10 (6.19–36.89) 2.83 0.17 16.65 (6.41–32.77) 1.91 0.13 14.69 (4.76–28.38)
Othera 2.11 0.14 15.07 (7.79–39.8) 6.32 0.11 57.45 (31.40–106.10) 1.71 0.15 11.40 (4.12–24.59) 4.40 0.20 22.00 (11.01–43.84) 0.76 0.12 6.33 (2.02–0.00)
Drowning 1.38 0.12 11.50 (5.16–30.74) 2.11 0.08 26.37 (13.62–69.65) 1.14 0.12 9.50 (5.16–30.74) 1.89 0.16 11.81 (3.87–23.06) 1.15 0.14 8.21 (4.42–26.35)
Male
Poisoning 23.63 2.94 8.04 (5.50–11.33) 29.46 1.76 16.74 (11.97–22.99) 33.84 3.29 10.29 (7.41–13.73) 27.56 4.58 6.02 (4.24–8.32) 13.74 2.96 4.64 (2.84–7.08)
Firearm 22.65 14.73 1.54 (1.04–2.18) 39.44 10.82 3.65 (2.72–4.82) 27.27 11.86 2.30 (1.64–3.21) 21.55 15.03 1.43 (0.97–2.06) 15.66 20.06 0.78 (0.49–1.17)
Hanging 21.54 6.52 3.30 (2.24–4.74) 48.47 7.75 6.25 (4.79–8.07) 24.97 7.87 3.17 (2.16–4.39) 19.48 7.47 2.61 (1.74–3.81) 12.46 3.97 3.14 (1.93–4.96)
Jumping 10.65 0.82 12.99 (7.56–20.83) 22.81 0.98 23.28 (15.69–32.76) 14.78 0.87 16.99 (10.51–25.55) 9.74 0.88 11.07 (6.24–17.91) 5.27 0.58 9.09 (4.85–17.66)
Cutting/Piercing 4.62 0.50 9.24 (4.40–17.53) 3.80 0.26 14.62 (6.24–27.79) 4.60 0.51 9.02 (4.32–17.19) 6.01 0.74 8.12 (4.67–15.77) 3.83 0.61 6.28 (2.66–11.84)
Othera 3.82 0.40 9.55 (4.06–18.06) 7.13 0.37 19.27 (11.12–35.30) 3.61 0.43 8.40 (3.78–16.80) 3.94 0.49 8.04 (3.31–14.74) 2.72 0.36 7.56 (3.03–15.48)
Drowning 1.97 0.22 8.96 (2.81–16.77) 2.85 0.20 14.25 (5.45–27.86) 1.97 0.22 8.96 (2.81–16.77) 1.86 0.24 7.75 (2.58–15.37) 1.76 0.23 7.65 (2.69–16.04)
Open in a new tab

O: Observed; E: Expected; SMR: Standardized Mortality Ratio

a.

Includes X76 (Burning), X82 (Car crash), X83 (Other specified), X84 (Unspecified), Y87.0 (Intentional self-harm)

Figure 2.

Figure 2.

Open in a new tab

Standardized Mortality Ratios for Means of Suicide by Sex and Across Age Groups

Women had the highest rates of suicide per 100,000 person-years and the highest rate from poisoning, which was 12.5 times that of the general US population. The highest rates of suicide by all means were in the youngest cohort, ages 18–34 years. Their rate of poisoning per 100,000 was 41.3 times that of the general population. Jumping had the highest SMR in this cohort, 58.5 times that of the general population. Although in general the SMRs for women declined with age, they were high for all means except firearms and hanging, which were similar to men and the general population.

In men, the overall SMRs for the various means were smaller than in women, with SMRs ranging between 1.5 (firearms) to 13.0 (jumping) times that in the general population. The highest SMRs were also in the youngest cohort, ages 18–34 years, where the highest rate of suicide by jumping was over 23.3 times that of the general population. As in women, the SMRs for men declined with age except for hanging which remained relatively constant with increasing age from 2.6 to 6.3 times the general population.

4. Discussion

In this study of suicide among patients diagnosed with schizophrenia; poisoning was the most common means of suicide. This was followed by firearms, hanging, and jumping. The risk of suicide by drug poisoning had different predictors than other poisoning means and was the only means with higher risk for women. The risk of suicide by other poisoning was high for patients with prior deliberate self-harm but lower for those with suicidal ideation. Predictors of risk were similar for the other means: reduced risk as age increased, for men, and for non-Hispanic Blacks; increased risk for patients with comorbid drug use, deliberate self-harm, inpatient admissions and living in the West compared to the South except by firearms.

The suicide rate by poisoning was over 10 times that of the general US population. Suicide by jumping was almost 16 times more common than in the general population. By comparison suicide by firearms was double the rate in the general population. Similar to previous findings, as patients aged the rates for most of the means of suicide declined (Olfson et al., 2021).

Sex differences were striking in our schizophrenia cohort. Poisoning accounted for over half of suicides among women but in men poisoning, firearms, and hanging were similarly common. In addition, the SMRs were much larger in women than in men, largely because suicide among women in the general population is far less common than in men (Hedegaard et al., 2021) while this sex difference is greatly attenuated among people diagnosed with schizophrenia.

In addition to having a much higher rate of suicide than the general population, patients diagnosed with schizophrenia have a markedly different pattern of means of suicide. Poisoning, mostly due to drug overdose, was the main means of suicide in this schizophrenia cohort. In post hoc analyses, we explored which types of drugs were reported as the primary cause of drug poisoning deaths in our sample of decedents and determined risk of death by these different drug poisoning means and found that 60.0% were non-psychotropic prescription medications and 26.6% were psychotropics (Supplemental Table 6 and 7). Women had the highest risk of poisoning by non-psychotropic prescription medications and patients with prior deliberate self-injury had the highest risk of suicide by psychotropics. It has been suggested that self-poisoning is an often impulsive (Mann et al., 2021) means of suicide and involves medications readily available in households (Hunt et al., 2010). Therefore, limiting access to lethal amounts of prescribed and over-the-counter drugs may be an important aspect of means restriction in this patient population (Hawton, 2007; Miller et al., 2020).

This study adds to the literature suggesting that patients with schizophrenia use less common means of suicide such as poisoning or jumping (Docherty et al., 2021; Harkavy-Friedman et al., 1999; Heila et al., 1999; Hunt et al., 2010). The reasons for these differences remain unclear, however they may be related to the presence of active psychotic rather than affective symptoms (Docherty et al., 2021). Adults with psychotic disorders who survive suicide attempts by jumping from heights have reported delusional beliefs at the time of the suicide attempt (Nielssen et al., 2010). Further research is needed to understand the intent of suicide attempters with schizophrenia that could add clarity for next steps in suicide prevention.

Findings in this study must be viewed considering its limitations. The diagnosis of schizophrenia cannot be confirmed although we used a previously validated method to identify patients with schizophrenia (Lurie et al., 1992). The accuracy of the comorbid general medical and mental health disorders as well as suicidal ideation in particular (Claassen and Larkin, 2005), cannot be confirmed in these claims data. Although death by suicide cannot be verified, the National Death Index has been found to have 90% concordance with other methods that identify death by suicide (i.e., hospital, autopsy, law enforcement, and medical examiner records) (Moyer et al., 1989). This study was limited to patients enrolled for at least one year in Medicare, which may lead to selection bias excluding patients with shorter Medicare enrollment or those with other or no health insurance. Because the Medicare system is primarily a health care program for individuals older than 65 years, results may be more representative of older patients with schizophrenia and does not include time after first diagnosis when risk of suicide is greatest in this population. However, Medicare includes younger individuals with disabilities including schizophrenia. Using the nationally representative Medical Expenditure Panel Survey (MEPS), Khaykin, et al., (2010) found that 46% of adults with schizophrenia had Medicare coverage of whom 93% were under age 65 (Khaykin et al., 2010). Our sample may be affected by survivor bias because we are reporting the risk of future suicide deaths in a cohort of individuals that excludes those who died by suicide prior to enrolling in Medicare, and reports indicate up to 60% of individuals die from their first suicide attempt (Bostwick et al., 2016). Finally, these data are from 2006–2017 and may not reflect contemporary patterns in means of suicide among adults with schizophrenia.

5. Conclusions

The high rate of suicide among people diagnosed with schizophrenia and their distinctive pattern of means of suicide highlights the importance of continued research to inform suicide prevention efforts in individuals with schizophrenia. Due to the high rates of suicide by poisoning and specifically by drug overdose in this sample, restricting access to lethal medications may be effective. Effectiveness studies in the UK showing that reducing the quantity of dispensed and over-the-counter medications (Hawton et al., 2004) and the introduction of blister packs was followed by decreased rates of suicide by poisoning, supports adopting these methods as policy changes to enact in the United States (Blakey, 2020). In this sample, risk of poisoning was related to a recent history of suicidal ideation, suicide attempt, inpatient admissions for both mental and non-mental health, and emergency department visits. This is consistent with previous studies in which some individuals present in the ED for non-psychiatric reasons shortly before their death (Claassen and Larkin, 2005), and underscores the need for active follow-up after a suicide-related crisis, inpatient admission, or ED visit (Mann et al., 2021). Evidence-based methods including multi-level approaches (van der Feltz-Cornelis et al., 2011) such as the Zero Suicide model, with risk screening, assessment, and continued therapeutic engagement may be an effective approach to suicide prevention care in this patient population (Brodsky et al., 2018).

Supplementary Material

1

Acknowledgements:

This study was supported by the NIMH P50 ALACRITY Research Center, Optimizing and Personalizing Interventions for Schizophrenia Across the Lifespan, supplement P50MH11584302S2, Identifying age-related antecedents to suicide in schizophrenia. Dr. Bareis is supported by K23MH129628 and by L30MH131131. Dr. Rolin is supported by K23MH126312 and by L30MH120711.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References:

  1. U.S. Department of Commerce Economics and Statistics Administration U.S. Census Bureau, 2010. 2010 Census Regions and Divisions of the United States. https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf (accessed July 1 2021).
  2. Ahmedani BK, et al. , 2014. Health care contacts in the year before suicide death. J Gen Intern Med 29(6), 870–877. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bhatia T, et al. , 2006. Differing correlates for suicide attempts among patients with schizophrenia or schizoaffective disorder in India and USA. Schizophr Res 86(1–3), 208–214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Office of Disease Prevention and Health Promotion. U.S. Department of Health and Human Services, 2020. A Comprehensive Approach to Preventing Suicide: The Role of Law and Policy and the Social Determinants of Health. https://www.healthypeople.gov/sites/default/files/MHMD_Law_%26_Health_Policy_Webinar_052620.pdf (accessed January 2 2022).
  5. Bonvoisin T, et al. , 2020. Suicide by pesticide poisoning in India: a review of pesticide regulations and their impact on suicide trends. BMC Public Health 20(1), 251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bostwick JM, et al. , 2016. Suicide Attempt as a Risk Factor for Completed Suicide: Even More Lethal Than We Knew. Am J Psychiatry 173(11), 1094–1100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brodsky BS, et al. , 2018. The Zero Suicide Model: Applying Evidence-Based Suicide Prevention Practices to Clinical Care. Front Psychiatry 9, 33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brown TL, et al. , 2018. Access to Psychotropic Medication via Prescription Is Associated With Choice of Psychotropic Medication as Suicide Method: A Retrospective Study of 27,876 Suicide Attempts. J Clin Psychiatry 79(6). [DOI] [PubMed] [Google Scholar]
  9. Carpenter JE, et al. , 2021. A Stepwise Approach for Preventing Suicide by Lethal Poisoning. Fed Pract 38(2), 62–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Centers for Disease Control and Prevention, 2021. CDC WONDER. Underlying Cause of Death, 1999–2020. https://wonder.cdc.gov/ucd-icd10.html (accessed July 1 2021).
  11. Claassen CA, Larkin GL, 2005. Occult suicidality in an emergency department population. Br J Psychiatry 186, 352–353. [DOI] [PubMed] [Google Scholar]
  12. Conner A, et al. , 2019. Suicide Case-Fatality Rates in the United States, 2007 to 2014: A Nationwide Population-Based Study. Ann Intern Med 171(12), 885–895. [DOI] [PubMed] [Google Scholar]
  13. Docherty AR, et al. , 2021. Suicide and Psychosis: Results From a Population-Based Cohort of Suicide Death (N = 4380). Schizophr Bull. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Feigelman W, et al. , 2016. Investigating Correlates of Suicide Among Male Youth: Questioning the Close Affinity Between Suicide Attempts and Deaths. Suicide Life Threat Behav 46(2), 191–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Goldman-Mellor S, et al. , 2019. Association of Suicide and Other Mortality With Emergency Department Presentation. JAMA Netw Open 2(12), e1917571. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gunnell D, et al. , 2005. The epidemiology and prevention of suicide by hanging: a systematic review. International journal of epidemiology 34(2), 433–442. [DOI] [PubMed] [Google Scholar]
  17. Harkavy-Friedman JM, et al. , 1999. Suicidal behavior in schizophrenia: characteristics of individuals who had and had not attempted suicide. Am J Psychiatry 156(8), 1276–1278. [DOI] [PubMed] [Google Scholar]
  18. Hawton K, 2007. Restricting Access to Methods of Suicide: Rationale and Evaluation of this Approach to Suicide Prevention. Crisis 28, 4–9.17555027 [Google Scholar]
  19. Hawton K, et al. , 2004. UK legislation on analgesic packs: before and after study of long term effect on poisonings. BMJ 329(7474), 1076. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. National Center for Health Statistics, 2021. Suicide Mortality in the United States, 1999–2019. NCHS Data Brief, no 398. https://www.cdc.gov/nchs/data/databriefs/db398-H.pdf (accessed January 2 2022).
  21. Heila H, et al. , 1999. Life events and completed suicide in schizophrenia: a comparison of suicide victims with and without schizophrenia. Schizophr Bull 25(3), 519–531. [DOI] [PubMed] [Google Scholar]
  22. Hor K, Taylor M, 2010. Suicide and schizophrenia: a systematic review of rates and risk factors. J Psychopharmacol 24(4 Suppl), 81–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hunt IM, et al. , 2010. Method of suicide in the mentally ill: a national clinical survey. Suicide Life Threat Behav 40(1), 22–34. [DOI] [PubMed] [Google Scholar]
  24. Ishii T, et al. , 2014. Characteristics of attempted suicide by patients with schizophrenia compared with those with mood disorders: a case-controlled study in northern Japan. PLoS One 9(5), e96272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Khaykin E, et al. , 2010. Health insurance coverage among persons with schizophrenia in the United States. Psychiatr Serv 61(8), 830–834. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Laursen TM, et al. , 2007. Increased mortality among patients admitted with major psychiatric disorders: a register-based study comparing mortality in unipolar depressive disorder, bipolar affective disorder, schizoaffective disorder, and schizophrenia. J Clin Psychiatry 68(6), 899–907. [DOI] [PubMed] [Google Scholar]
  27. Laursen TM, et al. , 2014. Excess early mortality in schizophrenia. Annu Rev Clin Psychol 10, 425–448. [DOI] [PubMed] [Google Scholar]
  28. Lu L, et al. , 2019. Prevalence of suicide attempts in individuals with schizophrenia: a meta-analysis of observational studies. Epidemiol Psychiatr Sci 29, e39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lubin G, et al. , 2010. Decrease in suicide rates after a change of policy reducing access to firearms in adolescents: a naturalistic epidemiological study. Suicide Life Threat Behav 40(5), 421–424. [DOI] [PubMed] [Google Scholar]
  30. Lurie N, et al. , 1992. Accuracy of diagnoses of schizophrenia in Medicaid claims. Hosp Community Psychiatry 43(1), 69–71. [DOI] [PubMed] [Google Scholar]
  31. Mann JJ, et al. , 2021. Improving Suicide Prevention Through Evidence-Based Strategies: A Systematic Review. Am J Psychiatry 178(7), 611–624. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Miller TR, et al. , 2020. Incidence and Lethality of Suicidal Overdoses by Drug Class. JAMA Netw Open 3(3), e200607. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Moride Y, Abenhaim L, 1994. Evidence of the depletion of susceptibles effect in non-experimental pharmacoepidemiologic research. J Clin Epidemiol 47(7), 731–737. [DOI] [PubMed] [Google Scholar]
  34. Moyer LA, et al. , 1989. Validity of death certificates for injury-related causes of death. Am J Epidemiol 130(5), 1024–1032. [DOI] [PubMed] [Google Scholar]
  35. Nielssen O, et al. , 2010. Suicide attempts by jumping and psychotic illness. Aust N Z J Psychiatry 44(6), 568–573. [DOI] [PubMed] [Google Scholar]
  36. Okolie C, et al. , 2020. Means restriction for the prevention of suicide by jumping. Cochrane Database Syst Rev 2, CD013543. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Olfson M, et al. , 2015. Premature Mortality Among Adults With Schizophrenia in the United States. JAMA psychiatry 72(12), 1172–1181. [DOI] [PubMed] [Google Scholar]
  38. Olfson M, et al. , 2021. Suicide Risk in Medicare Patients With Schizophrenia Across the Life Span. JAMA Psychiatry 78(8), 876–885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Olfson M, et al. , 2016. Short-term Suicide Risk After Psychiatric Hospital Discharge. JAMA psychiatry 73(11), 1119–1126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Palmer BA, et al. , 2005. The lifetime risk of suicide in schizophrenia: a reexamination. Arch Gen Psychiatry 62(3), 247–253. [DOI] [PubMed] [Google Scholar]
  41. Quan H, et al. , 2005. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 43(11), 1130–1139. [DOI] [PubMed] [Google Scholar]
  42. Shenassa ED, et al. , 2004. Safer storage of firearms at home and risk of suicide: a study of protective factors in a nationally representative sample. J Epidemiol Community Health 58(10), 841–848. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Sher L, Kahn RS, 2019. Suicide in Schizophrenia: An Educational Overview. Medicina (Kaunas) 55(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Shields LB, et al. , 2007. Schizophrenia and suicide: a 10-year review of Kentucky medical examiner cases. J Forensic Sci 52(4), 930–937. [DOI] [PubMed] [Google Scholar]
  45. Shrira I, Christenfeld N, 2010. Disentangling the person and the place as explanations for regional differences in suicide. Suicide Life Threat Behav 40(3), 287–297. [DOI] [PubMed] [Google Scholar]
  46. Suicide Prevention Resource Center, University of Oklahoma Health Sciences Center, 2021. Suicide Deaths in the United States. https://sprc.org/scope/means-suicide (accessed October 4 2021).
  47. van der Feltz-Cornelis CM, et al. , 2011. Best practice elements of multilevel suicide prevention strategies: a review of systematic reviews. Crisis 32(6), 319–333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. van Walraven C, et al. , 2009. A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care 47(6), 626–633. [DOI] [PubMed] [Google Scholar]
  49. Centers for Disease Control and Prevention, National Center for Injury Prevention and Control, 2021. Web-based Injury Statistics Query and Reporting System (WISQARS). www.cdc.gov/injury/wisqars (accessed October 1 2021).
  50. Wu Y, et al. , 2021. Sex-specific and age-specific suicide mortality by method in 58 countries between 2000 and 2015. Inj Prev 27(1), 61–70. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1860017-supplement-1.doc (296.5KB, doc)

RESOURCES