Characteristics of U.S. Suicide Decedents in 2005–2010 Who Had Received Mental Health Treatment

We obtained 2005–2010 data from the NVDRS, which captures details on violent deaths among the deaths registered within each of 18 states (Alaska, Colorado, Georgia, Kentucky, Maryland, Massachusetts, Michigan, New Jersey, New Mexico, North Carolina, Ohio, Oklahoma, Oregon, Rhode Island, South Carolina, Utah, Virginia, and Wisconsin). Data for Michigan and Ohio were available for only 2010. Data sources for NVDRS include death certificates, law enforcement reports, and coroner and medical examiner reports; these sources are used to more comprehensively describe each violent incident. Suicide deaths are identified according to the manner of death recorded in the various data sources. State abstractors follow a strict coding manual to ensure consistent reporting and reconcile any differences across the data sources (6).

Adult suicide decedents who received mental health treatment within two months before death were compared with suicide decedents who were not known to have received mental health treatment shortly before death. Because help-seeking behavior among adolescents differs from behavior among adults, only decedents over age 18 were considered (7). Treatment was defined as seeing a psychiatrist, psychologist, general medical doctor, therapist, or other counselor for a mental health or substance misuse problem; receiving a prescription for a psychiatric medication; attending anger management classes; or residing in an inpatient or halfway house facility for mental health problems (6).

To qualify as suicide by poisoning, a substance had to be ingested and deemed coresponsible for the death. Drugs on the scene that were not ingested were not counted (6). Suicides by poisoning were coded as poisoning involving commonly prescribed substances if one or more of the substances used in the act was technically a controlled substance that would require a prescription (6).

Suicide decedents were compared with respect to sociodemographic characteristics, health- and stress-related characteristics, and the suicide method involved. Logistic regression was used to calculate odds ratios for receiving treatment before suicide for all above characteristics. We adjusted comparisons for age, sex, race-ethnicity, and history of suicide attempt. History of suicide attempt was adjusted only for age, sex, and race-ethnicity. Because of multiple testing, we set the level of statistical significance to ≤.001. We performed analyses using PASW Statistics 18. This study was determined to be exempt from human subjects review by the Institutional Review Board of the Centers for Disease Control and Prevention.

Of the 57,877 suicides among persons >18 years of age recorded in NVDRS between 2005 and 2010, 16,471 (28.5%) had received treatment within two months of suicide. Of those who did not receive treatment in the two months before suicide (N=41,406), 3,198 (7.7%) had received mental health treatment in the past. Being male (adjusted odds ratio [AOR]=.47), race-ethnicity other than non-Hispanic white (AORs=.61–.73), and being ages 19–49 (AOR=.69–.91) or ≥70 (AOR=.63) were all associated with lower odds of receiving treatment (Table 1). Among life events registered in the NVDRS, intimate partner problems were the most prevalent type of problem before suicide and affected 15,168 (30.3%) of all 50,024 decedents with known circumstances (Table 1).

Compared with persons who died from hanging, those who died by drug poisoning involving a substance that commonly requires prescription (AOR=3.04), by sharp instruments (AOR=1.30), or by falling or jumping (AOR=1.44) had higher odds of recent mental health treatment. Suicides by firearms were associated with lower odds of receiving treatment (AOR=.88) (Table 1). Among 3,758 persons who received treatment and died by poisoning involving commonly prescribed substances, 3,060 (81.4%) were tested for use of antidepressants at the time of death, with 2,278 of them (74.4%) testing positive.

Among the decedents, having recent mental health treatment was positively associated with having depressed mood at time of death (AOR=1.69), a history of suicide attempt (AOR=2.77), and substance use problems other than alcoholism (AOR=1.13) (Table 1). Receiving treatment was inversely associated with having intimate partner conflicts (AOR=.75), perpetrating interpersonal violence (AOR=.64), financial problems (AOR=.87), criminal legal problems (AOR=.60), other legal problems (AOR=.82), and homelessness (AOR=.66).

Nearly a third of suicide decedents received help from some type of mental health care provider before taking his or her own life. Earlier studies have found a similar proportion of service utilization, which was even higher when general health care services—particularly, primary care provider visits—were also taken into account (3). The demographic distribution among suicide decedents known to have received mental health care in the two months prior to death generally reflected patterns of mental health seeking in the general population, in that smaller proportions of males, persons of minority race-ethnicity, individuals ≤30 years, and older adults (≥70 years) were known to access mental health services before suicide. Gender-specific help-seeking behavior, stigma, and socioeconomic factors often play a large role in these treatment disparities (8). However, when controlling for age, race-ethnicity, sex, and history of suicide attempt, we still found that some health- and life-stress–related circumstances were more common among decedents who had sought treatment, which indicates an area for improvement in the delivery of mental health services.

Depressed mood and substance misuse were associated with receiving mental health treatment. Although the effect size for substance misuse was relatively small, this association is consistent with research showing that patients were treated more often for depression when comorbidities were present (9). However, we also found that many suicide decedents who killed themselves by drug poisoning had received mental health treatment before their suicide, and commonly prescribed substances were often involved in these deaths. There is common agreement that drugs should be prescribed only in small package sizes to at-risk individuals to prevent suicide (10). In Britain, reduced pack size of analgesics have been shown to be effective in reducing suicides with paracetamol (acetaminophen in the U.S.) (11). More research is needed to investigate substances used and their responsibility for fatal outcomes among decedents, mechanisms involved, and best prevention practices. In contrast to suicide decedents who used poison, those who used firearms were less likely to have received treatment before death. Firearms are considered one of the most violent and lethal methods of suicide, and the use of violent methods has been described as reflecting a further step in the suicidal process. Individuals choosing firearms as their method may be less inclined to seek or accept treatment (12).

The higher odds of receiving mental health treatment observed among persons with a history of suicide attempts underscore that mental health treatment can provide an opportunity to address the needs of some previous suicide attempters. More follow-up treatment, therapies tailored to specifically reduce self-directed violence (including cognitive or other therapies and strategies intended to improve coping skills to better handle risk factors associated with suicide [13]), and monitoring of prescription medications might reduce the risk of subsequent attempts.

Connecting mental health providers to other services relevant to the circumstances frequently seen among decedents may also help prevent some suicides. Some life events, particularly intimate partner problems, were prevalent for more than 15,000 of all suicide decedents, reflecting sociological concerns with intimacy, including marriage and the association of divorce with high suicide rates (14). However, decedents experiencing partner problems had lower odds of receiving treatment before suicide. Clinicians and other public health professionals may be able to collaborate with successful programs and strategies that involve friends or family of the at-risk individual in order to reach out to individuals affected by family problems (15).

This study had several limitations. NVDRS data do not indicate which type of mental health service was received. Different quantities and treatment types were subsumed as mental health treatment, and findings for specific treatments may be different. The data are not nationally representative but representative of only the 18 states participating in the NVDRS. The information provided on the circumstances of deaths was from proxies and was subject to recall bias. Further, we could not assess whether substances that commonly require prescription were actually prescribed, because drugs were assigned to the prescription drug category only on the basis of the substance name or the name of the metabolite identified. Some of these substances might have been acquired on the street. Even in cases in which the drug was actually prescribed, we cannot rule out that the prescription may have been prescribed for a person other than the decedent.

The findings suggest that the substances used in suicides by poisoning and efforts to reduce access to lethal doses of prescription medications warrant further research. Further, better collaboration between mental health service providers and providers of other services, including outreach to individuals with intimate partner problems, may help reduce suicide deaths.

The authors thank David E. Sugerman, M.D., M.P.H, Jeneita Bell, M.D., M.P.H., and Nimesh Patel, M.S., for their comments on the manuscript. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Diseases Control and Prevention.

The authors report no competing interests.