INTRODUCTION

The number of incarcerated older adults is increasing in the United States (US), as the population of adults age > 55 in prisons increased 280% from 1999 to 2016.1 Incarcerated adults have high rates of HIV/AIDS, cirrhosis, cardiovascular disease,2,3,4 mental illness,5 and substance use disorders (SUDs).6 The population of community-dwelling older adults who are involved in the criminal justice system (“justice-involved adults”—those recently arrested, on probation, or parole) is also growing. Middle-aged and older justice-involved adults are at high risk for these same chronic conditions,7 and are more likely to experience a higher burden of medical multimorbidity (> 2 chronic medical diseases) and geriatric conditions.8 Yet the amount of intersection of chronic disease alongside mental illness and SUDs in older justice-involved adults is unknown.

Most studies about co-occurring conditions in the US prison population focuses on mental illness and SUDs among younger adults.6, 9, 10 However, medical multimorbidity, mental illness, and SUDs are all frequently interrelated and place individuals at risk for adverse outcomes. In general, older adults with multiple conditions have high rates of healthcare utilization and receive poorly coordinated care.11, 12 Therefore, older justice-involved adults, many of whom have a history of poor access to healthcare and social services, are even less likely to receive coordinated care that addresses co-occurring conditions. This is particularly important for adults who are returning to the community from correctional institutions,13 a period that represents high susceptibility to health-related mortality.14, 15

Few studies have estimated the prevalence of medical multimorbidity, mental illness, and SUDs among older justice-involved adults,7 and none have examined their co-occurrence. Therefore, this study aims to use nationally representative data to describe the health of middle-aged and older justice-involved adults through a multimorbidity framework to inform the development of multidisciplinary interventions.

METHODS

Data Source and Study Population

We performed data analysis on the 2015 through 2018 administrations of the National Survey on Drug Use and Health (NSDUH). NSDUH is a nationally representative annual cross-sectional survey of non-institutionalized, community-living individuals in the USA.16 Surveys are administered via computer-assisted interviewing conducted by an interviewer and audio computer-assisted self-interviewing. Sample weights are provided by NSDUH to account for the complex survey design, non-response, selection probability, and population distribution. We limited analysis to individuals age ≥ 50 (n = 34,898).

Justice-Involved Status

Participants were asked three questions regarding whether in the past year they were (1) arrested and booked for an offense, (2) on probation, or (3) on parole (or supervised release) in the past year. Similar to other studies,17, 18 we coded these variables into a binary variable indicating any criminal justice involvement in the past year.

Covariates

We considered the following demographic characteristics: age (50–64 years of age and 65 years of age and older), gender, race/ethnicity, education, annual family income, relationship status, and by whether or not the participant reported having health insurance.

Outcomes

Mental Illness

NSDUH includes an indicator for past-year mental illness developed and validated by the Substance Abuse and Mental Health Services Administration and National Institute of Mental Health, which is based on responses to specific questions.19 The items include level of depression, emotional distress, functional impairment, and suicidal thoughts. The responses from each individual are coded into a variable indicating mild, moderate, and severe mental illness. Our definition for mental illness was limited to moderate or severe mental illness, equivalent to a Global Assessment of Functioning (GAF) score of < 60.19

Medical Multimorbidity

Participants were asked if they have been told by a doctor that they have asthma, chronic obstructive pulmonary disease (COPD), cancer, cirrhosis, diabetes, a heart condition, hepatitis B or C, high blood pressure, HIV/AIDS, or kidney disease. We also coded a variable indicating whether participants reported being diagnosed with ≥ 2 conditions to indicate medical multimorbidity.11

Substance Use Disorder

Participants reporting past-year use of psychoactive substances were asked to answer DSM-IV questions about potential “abuse” and “dependence.” Similar to other studies,20, 21 while NSDUH is not a diagnostic interview, we considered the proxy diagnoses of either as past-year SUD. The specific substances queried included use of alcohol, cannabis, cocaine, heroin, methamphetamine, and prescription opioids, stimulants, and tranquilizers or sedatives, with the latter two categories including benzodiazepines, muscle relaxants, zolpidem, eszopiclone, zaleplon products, and barbiturates.

Analysis

We estimated the prevalence of criminal justice involvement and compared whether there were differences in criminal justice involvement based on demographics, mental illness, medical multimorbidity, and SUDs. Bivariable comparisons were conducted using Rao-Scott chi-square22 to determine differences for each of these conditions with respect to past-year justice involvement. Next, using separate multivariable models which used forced entry of covariates, we determined whether criminal justice involvement was associated with each outcome, and with co-occurrence of outcomes. This resulted in 16 separate multivariable models which controlled for demographic characteristics (age, gender, race/ethnicity, marital status, education, income, and health insurance). To account for potential family-wise error resulting from multiple models, we used a Bonferroni statistical correction (α = 0.003; 0.05/16 separate outcomes) to determine significance. We also ensured that multicollinearity was not present. Weights were included in all analyses and Stata 13 SE was used to analyze all data and we used Taylor series estimation to provide accurate standard errors.23 Secondary analysis of this publicly available data was exempt for review by New York University’s Institutional Review Board.

RESULTS

An estimated 1.2% (95% confidence interval [CI] = 1.1–1.3) of adults in the US age > 50 experienced criminal justice involvement. Table 1 presents demographic characteristics according to criminal justice involvement. Those who were younger, male, Black, not married, not having health insurance, those reporting lower educational attainment, and lower family annual income were more likely to report criminal justice involvement (all ps < 0.001).

Table 1 Demographic Characteristics of Middle-Aged and Older Adults According to Justice-Involved Status in the USA, 2015–2018
Full size table

As is shown in Table 2, justice-involved adults age ≥ 50 reported a higher percentage of mental illness (20.8% vs. 6.2%; p < 0.001), COPD (12.1% vs. 7.4%, p < 0.001), cirrhosis (2.1% vs. 0.5%, p < 0.001), and hepatitis B or C (9.0% vs. 2.0%, p < 0.001), but were less likely to report high blood pressure (23.1% vs 33.7%, p < 0.002) compared with adults not justice-involved. There was no statistically significant difference between the two groups regarding medical multimorbidity. In terms of SUDs, justice-involved adults reported a higher percentage across all substances examined, and for having any SUD (34.7% vs. 3.7%, p < 0.001). In addition, justice-involved adults age ≥ 50 were more likely to report all combinations of co-occurring outcomes (any SUD, mental illness, and medical multimorbidity) including having all three simultaneously (3.3% vs. 0.3%, p < 0.001).

Table 2 Chronic Disease, Mental Illness, and Substance Use Disorder Prevalence among Middle-Aged and Older Adults According to Justice-Involved Status in the USA, 2015–2018
Full size table

In our adjusted models shown in Table 3, with all else being equal, among adults age ≥ 50, justice-involved adults were at increased odds for mental illness (adjusted odds ratio [aOR] = 3.04, 95% CI = 2.09–4.41), past-year SUD for each substance examined, and for any SUD (aOR = 8.10, 95% CI = 6.12–10.73). In terms of co-occurring conditions, justice-involved adults were at increased odds of all combinations of the outcomes, including all three simultaneously (aOR = 8.56, 95% CI = 4.10–17.86). In sensitivity analysis (Supplemental Table 1), we found that adults arrested or booked had higher odds of mental illness, any SUD, and all combinations of the outcomes compared with adults who were on probation or parole. We also found that being on probation was not a significant correlate of mental illness when utilizing the Bonferroni correction (p = 0.01) and that those on probation or parole did not have a significantly increased odds of co-occurring medical multimorbidity, mental illness, or for all three outcomes concurrently while considering the Bonferroni correction (p = 0.03 for parole and p = 0.004 for probation). Thus, recent arrest appeared to somewhat drive significant associations with respect to mental illness. Patterns of medical multimorbidity and SUD for alcohol, cocaine, heroin, prescription opioids, and all opioids remained similar across the 3 subgroups.

Table 3 Multivariable Models Examining Justice Involvement as a Risk Factor for Chronic Disease, Mental Illness, and Substance Use Disorder among Middle-Aged and Older Adults in the USA, 2015–2018
Full size table

DISCUSSION

This study of adults age ≥ 50 living in the community with involvement in the criminal justice system found a high prevalence of SUDs and mental illness and is consistent with past research.7, 13 While most SUD interventions for criminal justice populations focus on opioids and alcohol,24, 25 our study adds a more granular examination of SUD patterns and reveals a high prevalence of alcohol, opioid, cocaine, and methamphetamine use disorders. As older adults who use stimulants may be at particularly high risk for death,26 these findings should be considered in harm reduction interventions targeting this population. While middle-aged and older adults with justice involvement have high rates of opioid use disorder, only a small percentage of justice-involved adults receive evidence-based medications (e.g., methadone or buprenorphine).27 Increasing the availability of medications for opioid use disorder for justice-involved adults, such as low-threshold buprenorphine programs,28 especially for older adults with chronic medical disease is critical to reduce the risk of health-related harms and overdose.29

While we found a higher prevalence of cirrhosis, hepatitis, and COPD, consistent with prior studies of incarcerated adults,2, 4 we did not identify a higher prevalence of medical multimorbidity. This may be due to differences in the definition of medical multimorbidity from other studies.8 In addition, we found a lower percentage of hypertension among justice-involved adults and no significant difference in heart disease, which contrasts with the literature of higher rates of cardiovascular disease and hypertension among recently incarcerated adults.14, 30 This could be because this population may have more limited access to healthcare and therefore have fewer opportunities to receive a medical diagnosis. Furthermore, we used national data of justice-involved adults who are living in the community, which may differ from other studies that have a heterogeneity of justice-involved populations.31 Despite similar rates of medical multimorbidity in justice- and non-justice-involved middle-aged and older adults, we found justice-involved adults age ≥ 50 still had higher odds of having two or more co-occurring chronic conditions (including medical multimorbidity with mental illness and medical multimorbidity with SUD) compared with those not justice-involved. In particular, our sensitivity analysis showed that those who were arrested or booked had higher odds of having mental illness and experiencing co-occurring conditions relative to those on probation or parole. Thus, recent arrestees may represent a particularly high-risk group within middle-aged and older justice-involved adults that has implications for targeted interventions.

Mental illness and SUDs themselves are chronic diseases and therefore coupled with chronic medical diseases can represent a compound multimorbidity. The intersection of substance use, mental illness, and medical multimorbidity is complex and such high levels of multimorbidity among justice-involved middle-aged and older adults place this population at risk for poor outcomes. Interventions for this population must consider the interplay between each condition and how it contributes to overall multimorbidity to reduce harms and improve the management of chronic diseases. Several interventions have been developed or proposed to address either mental illness, SUD, or chronic medical diseases in this population, especially for those transitioning back to the community.24, 25, 32,33,34 Our results highlight the need to address these problems simultaneously. Existing multimorbidity models of geriatric-based care are best suited to accomplish this but must incorporate treatment for SUD and mental illness with traditional models that address medical multimorbidity, functional health, and geriatric conditions. The ability to integrate geriatric-based care with addiction medicine29 will be critical given the high prevalence of SUDs in this population.

Our study has limitations in that information is self-reported and therefore susceptible to limited recall and social desirability bias. Multiple testing could be a limitation, but test results (ps < 0.001) remained significant in light of a Bonferroni correction (alpha = 0.003). In addition, our study is limited to community-living, justice-involved adults age 50 and older and not representative of adults currently incarcerated or sub-populations of justice-involved adults such as homeless individuals who do not use shelters or adults living in institutional group quarters. Also, this paper focused on any criminal justice involvement. Sensitivity tests found that many associations appear driven by recent arrestees, but we must keep in mind that only a small percentage (< 1%) of participants were on probation or parole and this could have limited our tests. Finally, NSDUH does not include a diagnostic interview. Responses to survey questions related to mental illness and SUD are used to determine proxy diagnosis and not a direct diagnosis. However, these proxy diagnoses in NSDUH are used extensively in the literature.7, 16,17,18,19,20,21

Despite these limitations, this study using nationally representative data finds that middle-aged and older justice-involved adults living in the community are at increased risk for co-occurring morbidity from medical multimorbidity, mental illness, and SUDs. As most research focused on justice-involved populations examine SUDs and psychiatric illness among younger adults, our study adds to the literature by focusing on middle-aged and older adults who not only also experience high rates of SUD and mental illness but also are at higher risk for chronic medical diseases. Finally, this study indicates the need to consider how co-occurring conditions contribute to poor health outcomes to deliver patient-centered care to an often-overlooked medically vulnerable population.