Prevalence and Correlates of Suicidal Ideation Among Parolees

Suicide is the tenth leading cause of death in the general population in the United States, and suicidal behavior leads to nearly a half-million emergency room visits annually (1). Suicide is often the second or third leading cause of death among state prison inmates (2,3). Compared with rates in the general population, rates of self-harm are higher among prison inmates before incarceration (7.5 times higher than the general population) (4) and while in custody (six times higher) (5,6). Each additional year of prison stay increases the risk of mortality by almost 16% (7).

Several studies outside the United States found that recently released inmates had higher suicide rates than the general population (6,8–11). In the first year after their release from prison, males were eight times more likely than males in the general population and females were 36 times more likely than females in the general population to die of self-inflicted harm (11).

Prevalence rates of suicidal ideation and suicidal behaviors are higher among prisoners than in the general population (5). Prisoners with suicidal ideation tend to be young, single, and undereducated (5). They have little family support and have a history of adverse life events, including self-harm and physical and sexual abuse (5,12). In the general population, persons who have poor general medical health (13), who live alone (14), or who are substance abusers (5,15) are also more likely to commit suicide. A study by Hayes (16), which examined intake records of all jail inmates, found that during the intake process a third of inmates who later committed suicide in jail had reported prior suicide attempts. Reviews of completed suicides of inmates in California (17) and Missouri (18) found that most had a known history of suicidal behavior or ideation.

Although details of inmates’ mental disorders are often unknown, one study found that 34% of jail inmates who committed suicide had a history of mental illness (16). In a review of case files of inmates who completed suicide, 73% had a history of mental illness (17). The relationship between specific mental disorders and suicide is less clear for several reasons, including psychiatric comorbidity (19), concerns related to valid estimates of certain types of mental illness (20), and misclassification of causes of death (21). Psychosis or affective disorders (22), including depression (16) and neurotic disorders (5), were the mental disorders most commonly found among prisoners who committed suicide. These findings concur with the literature on specific mental disorders as suicide risk factors in the general population. Suicidal behaviors have been found to be higher in the general population among persons with a major depressive disorder, panic disorder, social phobia, posttraumatic stress disorder, conduct disorder, and alcohol abuse (19). Way and colleagues (15) found schizophrenia to be a significant suicide risk factor among prisoners. Many community surveys, however, exclude measures of schizophrenia because of concerns about valid estimates in community settings (20).

The risk of death from suicide among former prison inmates in the first month of their release is three times that of the general population (23,24). The transition from prison to the community is often difficult. Released prisoners must obtain housing, reestablish relationships, find employment, and access health care (25). Former offenders often lack adequate education and employment opportunities and experience addiction and mental illness.

The presence of suicidal ideation is an indicator of poor emotional or physical well-being (26). Therefore, the presence of suicidal ideation among former prison inmates, regardless of whether it leads to suicidal behavior, needs to be addressed as a risk factor for poor adjustment to community life. A limited number of studies have examined the prevalence and correlates of suicidal ideation among parolees. This study compared the prevalence and correlates of suicidal ideation among parolees and among individuals in the general population who were not parolees.

Methods

NSDUH data

National Survey on Drug Use and Health (NSDUH) data from 2009 to 2011 were analyzed to examine the prevalence and correlates of suicidal ideation among respondents who were parolees and those who were not. Approximately 70,000 noninstitutionalized civilians participate in the survey annually (27). The sample is selected by using a multistage area proportional probability method. State is the first level of stratification, followed by state sampling region (SSR), census tract, and dwelling unit. About 50% of SSRs, not respondents, are overlapped for five years to increase the validity of estimates in trend analysis (27). A subset of survey respondents is interviewed to ensure validity of answers. If the answers cannot be verified, the survey respondent is dropped from the data. Respondents are offered a $30 incentive to increase response rates. To ensure confidentiality, the NSDUH is conducted in the privacy of the respondent’s home, and a computer-assisted personal interview is utilized.

Sample weight

The Substance Abuse and Mental Health Services Administration (SAMHSA) utilizes NSDUH data to obtain national estimates of the prevalence of substance use (27). National estimates are possible because each respondent is given a person-level weight for the number of individuals he or she represents on the basis of the multistage sampling method. Sample weights are calculated to reflect each state and the U.S. general population (28).The final sampling weights are adjusted on the basis of seven factors, including nonresponse and extreme weights. SAMHSA recommends applying the final weight variable to obtain unbiased national estimates of prevalence. Because the application of the weight produces underestimated standard errors, the final weight is applied to estimate prevalence but is not applied in the regression analysis (29). Westlake and colleagues (30) provided detailed information regarding the sample weight calculation.

Statistical analysis

Only respondents age 18 and older were asked questions related to mental health. The final nonweighted sample consisted of 1,249 parolees and 114,033 respondents who were not parolees (nonparolees). Presence of suicidal ideation was measured by the following NSDUH question: “At any time in the past 12 months . . . including today, did you seriously think about trying to kill yourself?” The grouping variable was parole status, which was measured by a question about whether the respondent was “on parole, supervised release, or other conditional release from prison at any time during the past 12 months.”

The first step of analysis examined prevalence of suicidal ideation among parolees and nonparolees with the sample weight applied. Multivariate logistic regression analysis was conducted for the two groups, using suicidal ideation as the dependent variable to understand its correlates. The year of data collection was included to control for time-specific confounding impacts.

Results

Prevalence of suicidal ideation

Table 1 presents prevalence rates for suicidal ideation by parole status. The average rate for parolees was more than twice that for nonparolees (8.6% versus 3.7%). The rate for parolees fluctuated, compared with the rate for nonparolees. The 2009 rate of 10.1% for parolees declined to 6.9% in 2010 (6.9%) and increased to 8.3% in 2011. No data were available to determine how much of the change was caused by sampling errors from the relatively small representation of parolees in the three annual subsamples.

Characteristics of parolees and nonparolees

Table 2 summarizes data on characteristics of parolees and nonparolees. Results were consistent with previous research: a large percentage of parolees were from racial-ethnic minority groups. Among nonparolees, 11.5% were black, compared with 25.8% of parolees. The disparity was similar for Hispanics—14.0% of nonparolees and 25.7% of parolees. Parolees were less likely than nonparolees to be married (26.0% versus 53.8%), employed (80.0% versus 93.9%), and a high school graduate (57.7% versus 85.5%). They were also less likely to have attended religious services in the past month (57.2% versus 63.0%).

Almost a quarter of parolees (24.3%) reported using illegal substances in the past month, about three times the rate for nonparolees (8.5%). Despite parolees' higher past-year prevalence of psychological distress (22.2% versus 10.2%) and depressive episodes (12.6% versus 6.6%), about the same percentage of parolees and nonparolees reported taking prescription medication for a mood disorder (7.2% versus 5.1%). Almost half of parolees (45.8%) had no health insurance, as opposed to 16.2% of nonparolees. Not surprisingly, a larger percentage of parolees than nonparolees rated their health as poor (4.1% versus 2.6%). Given that 37.6% of parolees were under the age 30, compared with 21.8% of nonparolees, this is an unexpected finding. In addition, parolees were almost six times more likely than nonparolees to have attacked someone in the past year (7.1% versus 1.3%).

Correlates of suicidal ideation

Table 3 presents multivariate logistic regression results. For parolees, many sociodemographic characteristics were not associated with suicidal ideation. For parolees, being married, being age 30 and older, being employed, and attending religious services were not related to having a lower rate of suicidal ideation; however, among nonparolees, these characteristics were significantly associated with a lower rate. Being a high school graduate was associated with increased suicidal ideation among parolees (odds ratio [OR]=1.62); however, among nonparolees, being a high school graduate was associated with a lower rate of suicidal ideation (OR=.91). For both groups, serious psychological distress, major depressive episode, illicit substance use, and inpatient stay were related to increased suicidal ideation. For parolees, these variables increased the odds of suicidal ideation by a factor of 2 to 5.

For parolees, having health insurance decreased the odds of suicidal ideation by half (OR=.50); however, having insurance was not significantly related to suicidal ideation among nonparolees. Among parolees, those taking prescription medication for a mood disorder did not have a higher likelihood of having suicidal ideation, whereas this variable was related to increased suicidal ideation among nonparolees (OR=1.53).

Discussion

This study examined the prevalence and correlates of suicidal ideation among respondents to a national survey who reported being on parole and among those who did not. Suicidal ideation was more than twice as prevalent among parolees. The results highlight the importance of providing parolees with access to health care. Having health insurance decreased the odds of suicidal ideation by half among parolees. For nonparolees, access to health care was not related to suicidal ideation, even though most of the other demographic and clinical variables examined were significantly related to suicidal ideation among nonparolees. Thus the finding about health insurance may point to an additional vulnerability among parolees. Nonparolees without health care coverage may have family and other social institutional support that lowers the incidence of suicidal ideation.

In addition, despite their relative youth, parolees were more likely than nonparolees to rate their health as poor and to have stayed in a hospital overnight. Although it was not possible to determine whether parolees with health insurance utilized health care or to assess the quality of the care they received, providing health insurance for this vulnerable population seems a promising approach to reducing suicidal ideation.

The results also indicated that parolees who had received a prescription for a mood disorder medication were no more likely than nonparolees to have suicidal ideation. Therefore, policies and practices that ensure parolees’ access to health care and adherence to psychiatric medication regimens may improve parolee management and reduce the prevalence of suicidal ideation.

Parolees were more than five times as likely as nonparolees to have attacked someone in the past year with intent to seriously harm the person. This finding is alarming in two respects. First, almost half of parolees did not have health insurance coverage. If they sustained injuries during physical altercations, they may not have received proper medical care or may have resorted to utilizing emergency care. In addition, attacking someone may lead to arrest, which may mean parole revocation.

Despite high unemployment among parolees, their employment status was not associated with health care coverage or to suicidal ideation in multivariate and bivariate analyses (results not shown), whereas employment status significantly predicted a lower likelihood of suicidal ideation among nonparolees in both analyses. A possible explanation for this finding is that parolees may be engaged in employment that does not provide health insurance or does not generate enough income to purchase private insurance.

As with any self-report survey, NSDUH data may reflect underreporting due to social stigma or social desirability (31). To estimate sampling error in the NSDUH and to assess confidence intervals and validity of the data, SAMHSA calculates standard errors and design effects (27). Nine substance use and treatment variables were used to assess sampling error by comparing observed outcomes with expected outcomes for three age groups and the entire sample. None of the examined variables were judged to be of poor precision to cause a concern about sampling error. Although this assessment lends confidence to the results of this study, it does not directly address whether the observed variations in prevalence of parolees’ suicidal ideation reflect true changes or sampling error. It is not possible to accurately determine the cause of changes because suicidal ideation was not used to assess the validity of the sampling design. Because of the observed changes in prevalence, we included the survey year as a predictor, using the latest year (2011) as the reference group to control for time-specific disturbances. The results from models with and without the survey year (results not shown) were largely the same. Therefore, the correlates of suicidal ideation identified in our model proved to be robust and were unaffected by disturbances associated with temporal factors.

Conclusions

It should be emphasized that although most individuals with suicidal ideation may never attempt suicide, decreasing suicidal ideation is a first step in preventing suicidal behavior and helping parolees transition to life in the community (2,23,24,32,33). Our findings indicate that suicidal ideation among parolees was not entirely determined by events that happened during incarceration or by health conditions and that suicidal ideation may be amenable to postrelease interventions. Specifically, our data indicated that a lack of health insurance was positively correlated with suicidal ideation.

These preliminary findings underscore the importance of implementing a seamless transition of care for parolees with mental health needs. Increasing eligibility for and utilization of public health insurance (for example, Medicaid) among parolees at the time of release is paramount for this continuity of care. Formerly incarcerated persons are overreliant on public health services. Without Medicaid, they must rely on charity care to meet their health care needs. Charity care is not a promising alternative to Medicaid. It is not regulated and must rely on voluntary initiatives of hospitals and doctors. Recent research from the Oregon Health Insurance Experiment found that expanding Medicaid for low-income adults increased health care utilization and lowered the probability of a positive screen for depression by 30% (34).

It is anticipated that prisoners will make up a large percentage of the Medicaid expansion population under the Affordable Care Act. To ensure seamless continuity of care, correctional departments will need to hire and train case managers or discharge planners to identify prisoners who are eligible for Medicaid and other health care benefits, assist them with completing the application forms, and ensure that the parolee, if not automatically enrolled in Medicaid, completes the application process at a social service office within the first few days of release and receives a health insurance card.

In addition to providing help with insurance coverage and medication, parole supervisors can help decrease suicidal ideation and behavior by identifying and referring parolees with specific mental disorders to psychiatric treatment, especially those with depression, substance use disorders, and schizophrenia (35,36). It is important to work with at-risk former offenders (37,38) and follow up with those who have expressed suicidal ideation or made a suicide attempt (39).

National data indicate that about a third of inmates who commit suicide had expressed suicidal ideation during the prison intake process (16). Nevertheless, actions are often not taken to prevent inmate suicides (40). A comprehensive study of more than 16,000 correctional facilities found that most jails and prisons did not have effective suicide prevention programs (16). This neglect applies to recently released inmates as well. Preventive screening can now be done efficiently; several standardized tools for assessment of self-harm risk have been validated among criminal offenders (41). Many of these tools are readily available to correctional authorities, allowing targeted follow-up of suicidal ideation and attempts.

Parole supervision is often focused on monitoring offenders for violations rather than ensuring that they receive needed services and treatments (42–44). Prevalence of suicidal ideation among parolees has not been adequately addressed because suicidal ideation is not understood as a major obstacle to successful social integration. In fact, suicide prevention is often not a priority for community correctional agencies, and parole officers are rarely trained to recognize warning signs. Findings of this study may prompt parole agencies to implement screening for current suicidal ideation and to devise preventive measures. Although some risk factors for suicide, such as substance use, aggression, and mental or general medical conditions may be difficult to modify or change, this study identified a protective factor that is relatively easy to implement: health insurance coverage for parolees. Implementation of the Affordable Care Act may lead to lower suicide rates among parolees and may help improve their general health status.