Association between lifetime criminal legal involvement and acute healthcare utilization in middle-aged and older US adults, 2015–2019

Abstract

Background

Five decades into the era of mass incarceration, a growing number of older adults have experienced criminal legal involvement (CLI) in their lifetime. Studies have shown that prior incarceration is associated with substantial disease burden, but little is known about the distinct needs and utilization patterns of middle-aged and older adults with lifetime CLI compared to those without. Using a nationally representative data set, we tested the association between lifetime CLI exposure and use of acute care services among middle-aged and older adults.

Methods

Our sample included 44,007 US adults (25,074 middle-aged-50-64 years; 18,709 older- ≥65 years) who participated in the National Survey of Drug Use and Health (2015–2019). The data is publicly available. Our independent variable was lifetime CLI. Using separate negative binomial regression models for middle-aged and older adults, we tested the association between lifetime CLI and acute healthcare utilization (ED visits and nights spent inpatient) controlling for relevant sociodemographic covariates.

Results

For middle-aged respondents, 19.1% reported lifetime CLI; for older adults, the rate of exposure was 9.6%. In multivariate models, CLI was associated with increased ED visits in middle-aged adults (IRR 1.18, 95% CI 1.06–1.31) but not older adults (IRR 0.99, 95% CI 0.85–1.16). CLI was associated with increased nights hospitalized in both groups (middle-aged: IRR 1.33, 95% CI 1.08–1.62; older adults: IRR 1.26, 95% CI 1.01, 1.57).

Conclusion

Middle-aged and older adults with lifetime CLI experience higher rates of acute care utilization than their peers with no lifetime CLI, even after adjustment for confounders. As the cohort of adults from the era of mass incarceration ages, understanding the mechanisms by which lifetime CLI impacts health outcomes is crucial in designing interventions to improve outcomes and reduce unnecessary acute healthcare utilization.

Introduction

After five decades of mass incarceration, the size of the United States correctional system persists at over 5.4 million people [1]. Over 70 million Americans have a criminal record [2], and the average age of individuals with criminal legal involvement is rising [3]. Among those incarcerated, the proportion 55 years or older has risen 400% over the preceding three decades [3]. The aging of the system-involved population has significant implications on community health systems because most in this population will return home, and such exposure is associated with increased risk of chronic and communicable diseases [4], worse control over such diseases [5], and worse health outcomes including mortality [6, 7].

The concentration of chronic disease among those with CLI combined with the substantial social risks associated with this exposure results in high rates of emergency department visits and hospitalizations, or acute care use [8, 9]. Those in the community with recent legal system involvement have higher rates of emergency department visits and hospitalizations, including hospitalizations for some ambulatory-sensitive conditions [9, 10]. Further, while people with recent CLI comprise 4% of the U.S. adults, they account for 7% of hospital and 9% of ED expenditures [8].

Most literature examining the health of older legal system-involved adults includes those who are currently incarcerated [11], though several recent studies have examined those in the community with prior incarceration [12]. One study using nationally representative data found that 1 in 15 adults aged 50 or older had experienced incarceration in their lifetime, and that this exposure was associated with increased risk of lung disease, mental health conditions, and alcohol use [12]. Those with prior incarceration were substantially more likely to experience physical and cognitive impairment compared to their never incarcerated peers, even after adjustment for sociodemographic confounders [12]. Another study examining the health of adults aged 50 and greater with recent criminal legal involvement found that they were at increased risk of mental health conditions, substance use disorders and a combination of both these conditions plus medical multimorbidity [13]. However, few analyses consider ever exposure to criminal legal system and its association with health outcomes.

The phenomenon of rapid aging and high burden of physical and behavioral illness among those with criminal legal involvement may have substantial implications on population health, including healthcare system use. However, it is unknown if middle-aged and older adults with lifetime exposure to the criminal legal system use acute care services (emergency department visits and inpatient hospitalization) more than their peers without CLI exposure. Therefore, we used nationally representative data examining the health status and acute healthcare utilization patterns of middle-aged and older adults by lifetime exposure to the criminal legal system. We hypothesized that those with lifetime CLI in both age cohorts would experience substantially higher acute care visits than those without, even after controlling for sociodemographic confounders.

Methods

Study design

The study sample was obtained from the National Survey of Drug Use and Health (NSDUH), an annual nationally representative, cross-sectional survey of the noninstitutionalized U.S. population, age 12 and older. The NSDUH is sponsored by the Substance Abuse and Mental Health Services Administration and provides data on substance use and mental health in the U.S. with about 70,000 individuals interviewed each year [14]. In addition, it provides data on some forms of criminal legal involvement and healthcare utilization. We used data from years 2015 to 2019 for the current study, which we analyzed cross-sectionally and weighted per survey guidelines.

Study population

Our study population included those who were ≥ 50 years old and we analyzed middle-aged (50–64 years) and older (65 + years) adults separately.

Independent variable

Our independent variable was lifetime exposure to the criminal legal system, defined by a self-reported arrest history. Respondents with lifetime CLI were identified as those who responded positively (“yes”) to the question: “Not counting minor traffic violations, have you ever been arrested and booked for breaking the law? Being ‘booked’ means that you were taken into custody and processed by the police or by someone connected with the courts, even if you were then released.”

Outcomes

We considered two groups of study outcomes: clinical conditions and acute care utilization. Clinical conditions included physical and mental comorbidities. Physical comorbidities were self-reported and included a heart condition, hypertension, diabetes, disease, asthma, COPD, cirrhosis, Hepatitis B/C, HIV/AIDS, cancer. Additionally, substance use disorder and mental illness were included, and determined diagnostically by survey responses using DSM IV criteria [15] (SUD) and a previously validated model (mental illness) [16].

Past year acute healthcare utilization was recorded as visits to the emergency department (ED) and nights spent hospitalized in the prior 12 months, which were reported as continuous variables. The number of ED visits was measured by the responses provided to the question: “During the past 12 months, how many different times have you been treated in an emergency room for any reason?” The number of inpatient nights was measured by the responses provided to the question: “During the past 12 months, how many nights were you an inpatient in a hospital?”

Covariates

We included socio-demographic covariates in our models to account for confounders consistent with the Andersen Behavioral Model of Health Services Use [17]. Demographic covariates included: gender (female and male), self-reported race/ethnicity (Non-Hispanic White, non-Hispanic Black, Hispanic and other, including non-Hispanic Native American/Alaskan native, non-Hispanic native Hawaiian or Pacific Islander, non-Hispanic Asian, or more than one race); educational status (less than high school, high school degree or equivalent, some college, and college graduate); marital status (married, widowed/divorced/separated, and never married); employment status (full time, part time, unemployed, and other, including disabled, keeping house full-time, in school/training, retired, some other reason); poverty level (< 100% federal poverty level as determined by the Census Bureau, 100–200% federal poverty level, and > 200% federal poverty level); metropolitan status (large metropolitan, small metropolitan, and non-metropolitan), and insurance status (Medicare, Medicaid, VA, private, and no insurance).

Statistical analyses

First, we described the overall sociodemographic characteristics of our study respondents by age category and compared them by lifetime CLI status using Pearson’s chi-square statistics. Second, we tabulated the percentage of respondents in each age category who reported each clinical condition by lifetime CLI status. Finally, we calculated unadjusted and adjusted incident rate ratios (IRRs) using negative binomial regression models with lifetime CLI status as the independent variable and frequency of healthcare utilization as the dependent variable. Each of the utilization outcomes (ED visits, and inpatient nights) were run as a separate model for each age category. The multivariate negative binomial regression controlled for age, sex, race/ethnicity, education, marital status, employment, poverty level, metropolitan status, and type of insurance coverage. Given the NSDUH’s complex survey design, weights were used per SAMHSA’s guidance to ensure the estimates were representative of the US population. 2-sided p-values of less than 0.05 were considered to be statistically significant. Multiple years of survey data were collapsed into one cross-sectional analysis per SAMHSA’s guidance. All analyses were conducted using Stata version 16 (Stata Corp LP, College Station, TX).

Results

Sociodemographic Variables

Our sample population included 43,783 US adults. Of middle aged adults (N = 25,074), 19.1% reported lifetime CLI; among older adults (N = 18,709), 9.1% reported lifetime CLI. Table 1 shows the demographics of two-age groups stratified by exposure to lifetime CLI.

Table 1.

Demographics of Middle-aged and older adults US adults by Lifetime Criminal Legal involvement status, 2015–2019

	Middle Aged 50–64 years (N = 25,074)		p-value	Older Adults 65 + yrs (N = 18,709)		p-value
	No Lifetime CLI N = 20,142	Lifetime CLI N = 4,932		No Lifetime CLI N = 16,812	Lifetime CLI N = 1,897
Sex			< 0.001			< 0.001
Male	42.3	73.2		41.1	79.3
Female	57.7	26.8		58.9	20.7
Race/Ethnicity			< 0.001			0.01
Non-Hispanic White	68.6	70.4		77.0	75.4
Non-Hispanic Black	10.9	14.5		8.7	12.0
Hispanic	13.0	10.5		8.4	7.6
Other	7.5	4.6		5.9	5.1
Education			< 0.001			< 0.001
< High school	11.0	17.7		14.8	15.7
High school graduate	24.0	31.9		27.6	27.8
Some college	29.1	31.6		25.5	30.9
College graduate	35.9	18.8		32.1	25.6
Marital Status			< 0.001			< 0.001
Married	67.0	46.8		60.1	50.6
Widowed/Divorced	23.6	36.7		35.8	41.0
Never married	9.4	16.5		4.1	8.4
Employment			< 0.001			< 0.001
Full-time	56.8	51.2		11.2	14.2
Part-time	11.9	8.7		11.2	11.6
Unemployed	3.1	4.5		0.8	1.8
Other	28.2	35.5		76.9	72.4
Poverty level			< 0.001			0.002
<100% FPL	10.0	18.7		7.6	9.9
100–200% FPL	14.6	20.9		22.4	25.0
>200% FPL	75.3	60.5		70.1	65.1
Metropolitan Status			< 0.001			0.50
Large metropolitan	55.9	50.4		50.5	48.7
Small metropolitan	29.1	32.3		32.3	33.5
Non-metropolitan	14.9	17.4		17.2	17.9
Insurance Coverage
Medicare	8.4	15.1	< 0.001	93.7	93.3	0.65
Medicaid	10.6	21.4	< 0.001	8.5	13.6	< 0.001
VA Insurance	4.3	7.0	< 0.001	10.5	19.1	< 0.001
Private insurance	75.4	56.8	< 0.001	66.6	55.4	< 0.001
Uninsured	7.2	11.2	< 0.001	0.6	0.7	0.53

For both age-groups, more than 70% of individuals with lifetime CLI were male, compared to just over 40% males for those without CLI. Respondents with lifetime CLI self-identified as Non-Hispanic Black individuals for frequently than those without in both age-groups. Similarly, fewer individuals with CLI were college graduates or married compared to those without CLI. Furthermore, individuals with CLI were more likely to have income below 100% FPL across both age groups. In both age groups, those with CLI were less likely to have private insurance and more like to have Medicaid insurance coverage. There was no difference in Medicare coverage between those with and without CLI for the older adult cohort.

Medical conditions

Table 2 reports the differences in prevalences of comorbidities by lifetime exposure to CLI for middle-aged and older adults. Individuals with lifetime CLI exhibited a higher prevalence of heart disease, diabetes, COPD, cirrhosis, Hepatitis B/C, substance use disorder and mental health conditions across both age-groups. Of note, CLI was associated with a higher prevalence of HIV/AIDS only in the middle-aged group (0.8% vs. 0.3%).

Table 2.

Prevalence of medical and behavioral Health conditions among Middle-aged and older adults by US adults by Lifetime Criminal Legal involvement status, 2015–2019

	Middle Aged			Older Adults
Medical Condition	No Lifetime CLI	Lifetime CLI	p-value	No Lifetime CLI	Lifetime CLI	p-value
Heart disease	10.7	15.5	< 0.001	27.2	32.9	< 0.001
Hypertension	27.1	27.8	0.43	40.7	38.4	0.08
Diabetes	14.6	15.9	0.04	21.7	25.1	0.01
Kidney disease	2.0	2.4	0.19	5.1	5.2	0.83
Asthma	8.1	8.5	0.50	7.9	6.9	0.33
COPD	4.8	10.4	< 0.001	9.0	12.9	< 0.001
Cirrhosis	0.3	1.2	< 0.001	0.6	1.5	< 0.001
Hepatitis B/C	1.3	6.0	< 0.001	1.6	4.7	< 0.001
HIV/AIDS	0.3	0.8	< 0.001	0.1	0.2	0.26
Cancer	7.8	7.0	0.14	17.0	16.5	0.61
Substance use disorder	3.6	15.3	< 0.001	1.5	7.0	< 0.001
Mental health condition	14.4	24.1	< 0.001	10.8	18.9	< 0.001

Acute Healthcare utilization

Table 3 displays the results of the unadjusted and adjusted multivariate negative binomial regression models for the association between lifetime CLI and healthcare utilization (ED visits and inpatient stays) for middle-aged and older adults. In the middle-aged group, CLI was associated with higher rates of ED visits in both unadjusted (IRR 1.53, 95% CI 1.40–1.68; p < 0.0001) and adjusted models (IRR 1.18, 95% CI 1.06–1.31; p = 0.004). Similarly, CLI was associated to higher rates of inpatient stays in both unadjusted (IRR 1.98, 95% CI 1.70–2.31; p < 0.001) and adjusted models (IRR 1.33, 95% CI 1.08–1.62; p = 0.007). In older adults, CLI was not associated to ED visits, as shown by both unadjusted (IRR 1.14, 95% CI 0.98–1.32; p = 0.09) and adjusted models (IRR 0.99, 95% CI 0.85–1.16; p = 0.94). However, CLI was associated with increased rates of inpatient stays in both unadjusted (IRR 1.40, 95% CI 1.14–1.72; p = 0.002) and adjusted models (IRR 1.26, 95% CI 1.01–1.57; p = 0.04).

Table 3.

Unadjusted and adjusted Multivariate Negative Binomial Regression for Association between Lifetime CLI and Emergency Department Visits and Inpatient stays among Middle-aged and older US adults, 2015–2019 (incident risk ratio with 95% confidence interval)

	Middle-Aged Adults		Older Adults
	Unadjusted IRR	Adjusted1 IRR	Unadjusted IRR	Adjusted1 IRR
ED Visits	1.53 (1.40–1.68)	1.18 (1.06–1.31)	1.14 (0.98–1.32)	0.99 (0.85–1.16)
Nights Hospitalized	1.98 (1.70–2.31)	1.33 (1.08–1.62)	1.40 (1.14–1.72)	1.26 (1.01– 1.57)

¹ Adjusted for sex, self-identified race/ethnicity, education, marital status, employment, poverty level, size of county of residence, insurance coverage

Notebold indicates significant at p < 0.05 level

DISCUSSION

In our nationally representative sample of middle-aged and older US adults, lifetime exposure to the criminal legal system was common. Those with lifetime CLI exposure in both age groups reported higher burden of chronic disease, mental health and substance use disorders. Finally, lifetime CLI was largely associated with increased acute care utilization, except for emergency department visits in older adults. The findings persisted after controlling for sociodemographic factors. Our findings suggest that criminal legal exposure across the life course is a significant social risk factor associated with chronic disease and acute care utilization later in life.

Our findings are largely consistent with our hypothesis and existing literature on the concentration of chronic disease among aging populations with CLI [12, 18] and risk for high rates of acute care utilization [8, 9]. In crude numbers, middle-aged adults with CLI had nearly twice as many nights hospitalized as those without; older adults with CLI reported 40% more nights hospitalized. In addition to high rates of mental health conditions, substance use disorders, and communicable diseases, middle-aged and older adults with CLI exposure also reported higher rates of heart disease and diabetes. Older adults in particular reported a substantial burden of cardiovascular disease, with a third reporting a heart condition, one in four diabetes, and a full 40% high blood pressure (though rates of hypertension were similar in both groups). These findings support the mounting literature supports the link between CLI and cardiovascular risk factors [19] and suggest it may be of primary clinical importance among middle-aged and older adults with lifetime CLI.

Prior work has shown that recent CLI is a high risk for acute care utilization [8, 9], but this is the first study to our knowledge to report the same association with lifetime CLI exposure. While the reasons for why CLI exposure may confer medical risk across the life course are not fully elucidated, it supports calls to include incarceration, and CLI more broadly, as a social determinant of health [20]. While our adjusted models controlled for sociodemographic factors which may be adversely influenced by CLI (e.g., employment, insurance, income), it is plausible that in this population with a high degree of social risk, important mediators on the path between CLI and high burden of disease and acute care use were not captured in our models. Such variables could include factors like residential segregation, housing instability, access to preventive healthcare, transportation difficulties, and stigma, which may undermine access to quality preventive care and lead to greater ED use [21–23].

Other work has established that prior CLI is common among older adults. Indeed, a recent study concluded that for older adults, the likelihood of a having criminal legal history is greater than that of having colorectal cancer [12]. Other studies have found that older adults with criminal legal histories have increased health needs and greater functional disability compared to those without [13]. Despite this, there is a paucity of research into interventions for community-dwelling older adults with CLI. A narrative review authored by Onyeali et al. [24] highlighted three key areas for supporting older adults’ reintegration into society: improving healthcare linkages through better discharge planning (e.g., smooth transfer of medical records to the community health-care providers on release, supporting primary care clinics and reentry programs, establishing a national network of nursing homes who need long-term care post-incarceration); ensuring safe, affordable and stable housing in reentry; and facilitating the automatic resumption of health-care coverage, including Medicaid and Medicare, after release from prison. Similarly, a systematic review highlighted a few interventions for older adults in jails and prisons, including counseling, art and music therapy, recreational therapy, group and individual therapies, and self-help groups [25]. Interventions for older adults with exposure to CLI should address the complex social, physical, and mental health problems prevalent in this population. However, there is a great need for further research in this field.

Despite use of a large nationally representative sample, our paper has a few limitations worth noting. First, given the cross-sectional deign of the study, casual inferences cannot be established. Although our models included multiple confounders known to affect healthcare utilization conforming to Andersen’s model [17], all possible confounders could not be accounted for, owing to the variables available in the database. Secondly, our variable for lifetime CLI is a broad measure with variable temporality in terms of time since exposure. The measure included captures a significant social risk and future studies can examine how various forms of CLI and time since exposure may differentially influence health outcomes. a considerable proportion of those with CLI are not included as this survey, as like nearly all nationally representative surveys, NSDUH does not sample those currently incarcerated. However, this sampling design should account for community-dwelling adults with lifetime CLI exposure. Fourth, CLI, healthcare utilization and other clinical covariates were documented based on self-report. A study on a large cohort of adults recently released from prison found that there is adequate agreement between self-report and administrative health records, however, data may be underreported [26].

Conclusion

This study of a nationally representative sample of middle-aged and older adults found that a lifetime exposure to the criminal legal system confers risk of chronic medical conditions and is independently associated with acute healthcare utilization after controlling for sociodemographic confounders. Middle-aged and older adults reported lower socioeconomic status and high prevalence of medical conditions. Additional research should emphasize understanding the drivers of this relationship, particularly focusing on social risks. Identifying and addressing various barriers to care will be critical in optimizing health and reducing unnecessary acute healthcare utilization and costs as the US population with lifetime CLI ages.

Abbreviations

CLI

Criminal Legal Involvement

ED

Emergency Department

NSDUH

National Survey of Drug Use and Health

SUD

Substance Abuse and Mental Health Services Administration

IRR

Incidence rate ratios

CI

Confidence interval

Author contributions

All authors conceived this idea. LEE analyzed the data. SB, LH, RJW, and LEE interpreted the data. SB and LH drafted the manuscript. All authors critically revised the manuscript for intellectual content. All authors reviewed and approved the final manuscript.

Funding

Effort for this study was partially suppored by the National Institute of Diabetes and Digestive and Kidney Diseases (R01DK118038, R01DK120861, PI: Egede; K23DK132505, PI: Hawks), and the National Institute on Minority Health and Health Disparities (R01MD013826, PI: Egede/Walker, R01MD018012, R01MD017574, PI:Egede/Linde). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The dataset supporting the conclusions of this article are publicly available from the National Survey of Drug Use and Health website: https://www.datafiles.samhsa.gov/study-series/national-survey-drug-use-and-health-nsduh-nid13517.

Declarations

Ethics approval and consent to participate

The Medical College of Wisconsin’s Institutional Review Board deemed this study exempt from review because the data are de-identified and publicly available.

Consent for publication

not applicable.

Competing interests

The authors declare no competing interests.