Incarceration and Psychiatric Emergency Department Visits Among Black Americans

Abstract

Policy Points.

• The spillover effects of jail and prison incarceration extend to acute psychiatric emergencies in Black communities.

• State‐level drug reform policies and policies that reduce labor market and housing discrimination for former inmates may have salutary mental health benefits for the broader community.

Context

The justice system incarcerates nearly 2.3 million individuals in the United States. Black Americans comprise 40% of those incarcerated despite representing less than 15% of the population. Theoretical work posits that mass incarceration can erode social capital by straining social and family networks as well as inducing carceral churn and coercive mobility within Black communities. Scholars report that greater incarceration may influence population‐level health, specifically in communities of color. However, previous work does not address whether incarceration, as well as the racial disparity in incarceration, corresponds with psychiatric help seeking in the Black community.

Methods

We examine the relation between incarceration and psychiatric emergency department (ED) visits among Black Americans from 2006 to 2015. As the exposure, we use 1) jail and prison incarceration among Black Americans (per 100,000 population), and 2) the ratio of non‐Hispanic Black to non‐Hispanic White American incarcerated populations as a measure of racial inequity in the criminal legal system. We examine, as the outcome, annual psychiatric ED visits (per 100,000 population) among Black Americans in 404 counties from ten US states between 2006 and 2015 (2,360 county‐years). Linear fixed effects analyses control for time‐invariant county factors as well as percentage below the federal poverty line, percentage unemployed, number of hospitals, and arrests for violent crime.

Findings

One additional individual incarcerated (per 100,000 population) corresponds with a 1.4% increase in psychiatric ED visits in Black communities (p < 0.001). An increase in racial inequity in incarceration (non‐Hispanic Black to non‐Hispanic White American ratio) also coincides with a 2.2% increase in psychiatric emergencies among Black Americans (p < 0.001).

Conclusions

The long arm of incarceration may extend to psychiatric emergencies among Black Americans in the broader community. Researchers and policymakers may consider measures, such as drug reform or other social policies, to reduce the influence of mass incarceration on acute and adverse mental health in the Black community.

In 2020, the united states incarcerated nearly two million individuals. ¹ With a rate of 573 per 100,000 population, incarceration in the United States exceeds that of any other nation. ² Racial disproportionality in incarceration is also exceptional. Mass incarceration refers to the concentration of imprisonment among Black Americans as well as the consistency in which it has become a normal stage within the life course in the Black community. ³ , ⁴ , ⁵ Black Americans have an incarceration rate of 2,306 per 100,000 population (vs. White Americans, at 450 per 100,000 population). ² One in three Black men can expect to go to prison in their lifetime, and one in four Black women have a family member in prison. ² , ⁶

The social costs of this expansive correctional system are considerable. Formerly incarcerated people face a greater burden of disease and risk of premature death than individuals who never served time in prison or jail. ⁷ , ⁸ , ⁹ Moreover, the heavily uneven distribution of exposure to this system makes it a critical driver of population health inequities—in that substantial racial inequities in incarceration are a key mechanism driving contemporary racial inequities in health and well‐being. ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴

Though the repercussions of incarceration for health and health equity are in many ways well described, there is room for expansion—particularly in terms of fully quantifying the collateral health consequences of incarceration. Theorists have developed compelling conceptualizations of how incarceration may impact the physical and mental well‐being of people who themselves have not been imprisoned—by removing a family member from a home who would otherwise provide economic and social resources, by disrupting community ties and social support networks—but the empirical evidence demonstrating these ideas remains limited. ¹⁵ , ¹⁶ Although producing evidence of lasting health burdens faced by those who were themselves incarcerated is inarguably valuable, equal efforts are needed to identify the potential health consequences for individuals with secondhand contact to incarceration in order to capture the full extent of the sequalae produced by this system.

Previous work on the collateral consequences of incarceration reports greater rates of preterm birth, sexually transmitted disease, cardiovascular disease, mortality, adverse mental health, and health service utilization. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁷ , ¹⁸ , ¹⁹ In this study, we add to this literature by examining whether increases in county‐level incarceration coincide with greater psychiatric emergencies among Black Americans. We use county‐level jail and prison incarceration data from the Vera Institute of Justice and emergency department (ED) data from the Healthcare Cost and Utilization Project (HCUP) between 2006 and 2015. ²⁰ , ²¹ Using linear regression analyses and specifications to measure within‐county changes over time, we find that increases in jail and prison incarceration corresponds with a 1.4% increase in psychiatric ED visits (per 100,000 population) among Black Americans. This suggests that the collateral consequences of incarceration extend beyond mental health symptoms to acute forms of psychiatric help seeking among Black populations. The disproportionate concentration of Black Americans in correctional systems may additionally drive racial inequities in psychiatric emergencies.

Mechanisms

Previous work indicates that incarceration degrades social capital, cohesion, and integration in Black communities through several mechanisms such as material hardship, stress, carceral churn and coercive mobility, as well as straining social and familial networks. ⁴ , ¹⁵ , ²² , ²³ , ²⁴ Community‐level mechanisms such as carceral churn and coercive mobility, or the removal of individuals from the criminal legal system and the subsequent return to communities, has shown to contribute to community instability, as well as concentrated disadvantage. ²⁴ As a result, the lack of community integration further exacerbates health outcomes within Black communities. ²⁵ Previous work also indicates that incarceration has contributed to restructuring households as well as infringing on social networks in the Black community. ⁴ , ¹⁵ , ²² , ²³ Incarceration strains family relationships, depletes financial resources, and leads to greater social and emotional isolation for the family of the incarcerated person. ²⁶ , ²⁷ , ²⁸ Although scholars assert that moderating factors such as mental illness, addiction, or intimate partner violence may provide relief for some families, the majority of families, by and large, experience further disadvantage and poorer outcomes for children following incarceration. ²⁸ , ²⁹ , ³⁰

Scholars theorize that social networks create the foundation for social capital—defined as networks with shared norms, values, and understandings that facilitate cooperation within or among groups. ¹⁵ , ³¹ Incarceration removes an individual from a household and therefore strains strong ties within the family network. ¹⁵ , ¹⁶ Incarceration of a family member may also reduce the potential for weak ties formed through employment or community activities for the family members, partners, or children of those incarcerated. ¹⁵ , ¹⁶ , ³² Given the substantial proportion of incarcerated men from low‐income Black communities, scholars find that women in these communities often do not replace strong ties after losing their incarcerated male partners. ¹⁵ , ²² This circumstance may lead to strained strong ties with the incarcerated family member or significant monetary and time investments in maintaining the ties. ¹⁵ Additionally, weak ties create bridges from family units to other social networks—thereby building social capital. ¹⁵ Men in prison do not have the opportunity to create connections to beneficial job‐ or community‐related networks for their families, thereby reducing the overall capacity for social capital in these communities. ¹⁵ Although empirical findings in this area remain limited, one study reports that incarceration varies inversely with community attachment and organizational participation. ³³

Mental Health and Well‐Being

Research consistently reports that social ties have salutary benefits on mental health. ¹⁶ , ³⁴ , ³⁵ Scholars contend that relationships among humans comprise three layers that collectively influence psychological well‐being: 1) intimate ties (familial ties), 2) network relations (close relatives and friends), and 3) community participation (participation in voluntary activities or religious organizations). ³⁵ , ³⁶ Although weak ties such as community participation may not provide greater personal interaction, theorists posit that they nonetheless enhance psychological well‐being. ³⁶ Empirical work on network analysis in this area finds that outer layers support inner layers, such that greater community participation makes it possible to create closer relationships with relatives and friends. ³⁶

Individual‐level studies utilizing longitudinal data and rigorous methodologies report that familial incarceration precedes adverse mental health among children and partners of those incarcerated. ³⁷ , ³⁸ , ³⁹ Scholars theorize that forced separation from a parent, material deprivation from the loss of financial support, and compromised parent relationships can harm children's social and emotional well‐being following parental incarceration. ⁴⁰ Nationally representative studies report that parental incarceration precedes problematic externalizing behaviors, as well as depression, posttraumatic stress, and anxiety among adolescents. ³² , ⁴⁰ , ⁴¹ , ⁴² Parental incarceration among Black populations also coincides with increased racial disparities in childhood behavioral problems. ³⁸ Additionally, Black women report higher levels of psychological distress and drug use following incarceration of their male partners. ³⁹ , ⁴³ However, women who report having fixed roles in the community, such as employment, show fewer distress symptoms. ³⁹

At the population level, studies report a multitude of adverse health outcomes, from birth outcomes to infectious disease and mental health as a result of neighborhood‐, county‐, and state‐level incarceration. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁷ , ¹⁸ , ¹⁹ Of most relevance to this study, changes in incarceration coincide with greater ED utilization, as well as adverse mental health symptoms. ¹¹ , ¹⁷ Although researchers argue that having a greater proportion of former inmates results in an increased number of uninsured individuals who utilize the ED for health care services, they also report spillover effects in the broader population by examining within‐state changes in ED visits. ¹⁷ Moreover, Hatzenbuehler and colleagues report the first ecological test of community‐level incarceration and mental health of those not directly tied to incarceration. ¹¹ They report that individuals living in neighborhoods with greater prison admission rates in Detroit have a higher likelihood of current and lifetime depression and anxiety than those living in neighborhoods with lower prison admission rates, after accounting for individual and neighborhood characteristics. ¹¹

Our Contribution

We extend previous work and its limitations in the following ways. First, we test whether incarceration corresponds with greater psychiatric help seeking, such as seeking emergency care because of acute mental health needs. Although studies have shown increases in adverse mental health and ED utilization following current and former incarceration, respectively, further examination of psychiatric ED visits may indicate greater mental health service needs in communities with heightened incarceration. Second, we assess whether racial/ethnic inequities in incarceration coincide with greater psychiatric help seeking among Black Americans. Previous work finds a relation between racial/ethnic inequities in incarceration and preterm birth, suggesting that justice‐related indicators for racial inequity may also precede changes in adverse mental health. ¹⁰ Third, we examine psychiatric help seeking as a function of incarceration at the county level. The long arm of incarceration may contribute to adverse mental health in populations beyond the neighborhood, as studies have found changes in birth outcomes, rates of sexually transmitted diseases, and mortality rates as a function of incarceration within counties. ⁹ , ¹⁰ , ¹⁸ , ¹⁹ Counties may serve as a critical level of aggregation because of county‐level policies, jail administration, and litigation. Fourth, we extend previous work informing theories linking incarceration and mental health as any discovered “indirect effect” of incarceration, which could have implications for psychiatric help seeking beyond the individuals and families of those incarcerated.

We address these limitations and examine whether incarceration corresponds positively with psychiatric ED visits among Black Americans. We also test whether racial inequities in incarceration (Black to White American incarceration ratio) coincide with greater psychiatric ED visits among Black Americans. We use annual variation in psychiatric ED visits among Black Americans as a function of incarceration and the ratio of Black to White American incarceration in 404 counties in the United States between 2006 and 2015. Results may hold relevance to understanding the ecology of the justice system and its potential to exacerbate acute mental health needs in the Black community beyond expected levels.

Methods

Study Population

We obtained the outcome, psychiatric ED visits among non‐Hispanic (NH) Black Americans (per 100,000 population) from the Statewide Emergency Department Database (SEDD). ²⁰ The Agency for Healthcare Research and Quality (AHRQ) makes the SEDD available for purchase under the HCUP. ²⁰ Participating states contract with the HCUP to make visit‐level data available for purchase from all hospitals with an ED. ²⁰ The SEDD does not include hospital data from prison populations. ⁴⁴ Following cross‐validation with American Hospital Association surveys, the SEDD provides 99% of all ED visits for participating states. ²⁰ Much epidemiologic and health care services research utilizes this high‐quality database. ⁴⁵ , ⁴⁶ , ⁴⁷ This study comprised counties from ten US states that report county identifiers, race/ethnicity, and year of visit to the HCUP. States in the study sample included Arizona, California, Florida, Kentucky, Maryland, Massachusetts, New Jersey, New York, North Carolina, and Rhode Island. These states reflect three US regions and comprise approximately 20% of the US population. ⁴⁸ We examined a ten‐year period from 2006 to 2015 in which psychiatric ED visits have increased substantially given that many populations lack access to routine mental health care. ⁴⁹ Mental health diagnoses comprised 6.6% of all ED visits in 2007, 10.9% in 2015, and rose to the seventh primary diagnosis at ED visits in 2019. Non‐Hispanic Black adults have the highest prevalence of psychiatric ED visits (96.8 per 1,000 adults) when compared with all other race/ethnicities. ⁵⁰ Black youth show similar trends, with a 53% increase in psychiatric ED visits between 2011 and 2015. ⁵¹ The University of California, Irvine, Institutional Review Board approved this study (#20195613).

Study Measures

The aggregate county‐year prevalence of psychiatric ED visits among Black Americans (per 100,000 population) is the outcome variable. We classified an ED visit as psychiatric if any visit‐level diagnosis (from diagnostic code 1 to 25) corresponded with a psychiatric International Classification of Diseases, Tenth Revision (ICD‐10) code according to Clinical Classification Software categories developed by the AHRQ. ⁵² We aggregated a total of 8,740,370 psychiatric ED visits among Black Americans by county‐year and converted these counts to population prevalence estimates (per 100,000 population) using race‐specific county‐level population denominators from the US Census Bureau's Population Estimates database. ⁴⁸ Although these outcome data provide the opportunity for monthly resolution, in order to link estimates to the exposure of interest, we aggregated psychiatric ED visits annually. Additionally, we aggregated the outcome at the county level because policies, health services, and funding that may influence ED processes often operate at the county level. ED data at the county level also allowed for geographic linkage to the exposure of interest.

For the exposure, we retrieved data on prevalence of annual prison and jail incarceration (per 100,000 population) among NH Black and White populations from the Vera Institute of Justice “In Our Own Backyard” (IOB) incarceration trends data set. ²¹ IOB provides county‐level data on the annual count of incarcerated populations by race/ethnicity in prisons and jails. ²¹ Incarceration within counties, rather than states, may hold particular relevance to policies deployed by judges, prosecutors, and police at the county level. ²¹ The Vera Institute of Justice validated estimates by comparing state‐level findings with the Bureau of Justice Services National Prisoner Statistics Program. ⁵³ Previous epidemiologic literature has also utilized this data set to examine health outcomes as a function of incarceration. ⁹ , ¹⁰ , ¹⁸ Using race‐specific county‐year population denominators from the US Census Bureau's Population Estimates database, we converted prison and jail population counts to population prevalence estimates (per 100,000 population) for NH Black and NH White Americans. ⁴⁸ We then averaged prison and jail population prevalence estimates (per 100,000 population) to calculate incarceration prevalence estimates (per 100,000 population). A combined measure captures fluctuations in prison and jail populations and the collective removal of individuals from their homes and communities. The social processes involved in removing individuals from communities to jails or prisons differ. Jails often hold individuals who await trials and sentencing or may serve shorter sentences, whereas prisons comprise individuals who have convictions for more serious crimes and serve longer sentences. ⁵⁴ Given the difference in sentence lengths, jail incarceration may abruptly remove and reintroduce individuals to and from households and communities, whereas prison incarceration may indicate continual removal for a lengthy period of time. Although research in this area generally examines jail or prison populations separately, ⁹ , ¹⁰ , ¹¹ , ¹⁷ , ¹⁸ some work, as with our own, does assess incarceration as either jail or prison. ¹² We posit that incarceration in jails and prisons activate the disruption of familial and social networks through different mechanisms. Prison incarceration may remove individuals for longer periods of time and strain social networks; however, consistent movement in and out of jail incarceration may also interfere with maintaining networks, as some literature reports jail recidivism rates as higher than that for prisons. ⁵⁵

We used the ratio of NH Black to NH White incarceration to capture the unequal, racialized burden of incarceration within a county. The ratio reflects the multiplicative difference in incarceration between Black and White populations, which ranges from five to more than seven times greater for Black populations within a county‐year. However, because previous work has also utilized the difference in incarceration between the two race/ethnicities as a measure of structural racism, we used a difference measure as a sensitivity check for the analyses. ¹⁰

Increased arrests for violent crimes within counties may affect both the exposure and outcome, although the literature does not converge on the direction of these associations. ⁵⁶ , ⁵⁷ In our work, we found a modest, inverse correlation between arrests for violent crimes and incarceration, which parallels figures reported by the Bureau of Justice Statistics. ⁵⁷ , ⁵⁸ As such, we utilized data on arrests for violent crimes from the Uniform Crime Reporting (UCR) database as a control variable in our tests. ⁵⁹ , ⁶⁰ , ⁶¹ The Federal Bureau of Investigation (FBI) provides UCR data with information on arrests for violent crime from 18,000 participating city, county, university, tribal, and federal law enforcement agencies. ⁶² As a publicly available data set, UCR remains one of the most extensively used arrest repositories in the United States. ⁶² We converted annual arrest counts to population prevalence estimates (per 1,000 population) and utilized this estimate as a covariate.

We also controlled for county‐level factors that may confound the relation between incarceration and psychiatric ED visits over the study period. We used county‐year arrests (per 1,000 population), number of hospitals reporting visits, percentage unemployed, and percentage below the federal poverty line as control variables. We also incorporated year indicators to reduce confounding from annual factors (i.e., policies associated with drug reform or the Affordable Care Act) that may precede changes in incarceration or psychiatric emergency care during the study period. ⁶³ , ⁶⁴ , ⁶⁵

Analysis

We linked county aggregates of psychiatric ED visits and incarceration with other county‐level covariates using Federal Information Processing Standards (FIPS) codes. ⁶⁶ This linkage yielded a total analytic sample of 404 counties from ten states between 2006 and 2015 (i.e., 4,040 county‐years). Psychiatric ED visits showed a right‐skewed distribution in that more than half of the county‐years (2,095) exhibited psychiatric ED visits (per 100,000 population) between one and 5,000 (Appendix 1). We, therefore, log‐transformed the outcome to approximate a normal distribution and minimize the influence of outliers (Appendix 2). ⁶⁷ We tested whether psychiatric ED visits (per 100,000 population) increase during years with greater incarceration (per 100,000 population) among Black Americans using ordinary least squares linear regression analysis. Additionally, we examined whether the ratio of Black to White American incarceration (an estimate of structural racism) corresponds with greater psychiatric ED visits.

A fixed effects model specification controls for time‐invariant county‐level characteristics that may influence the exposure or the outcome such as access to health care. ⁶⁸ This specification forces within‐county measurement by removing mean values for each county and examining year‐to‐year changes. ⁶⁹ , ⁷⁰ The within‐county estimation strategy, moreover, precludes confounding by cross‐county differences that fundamentally render two (or more) different counties difficult to compare “holding constant” a limited set of covariates.

Two‐way fixed effects models with continuous exposures may, however, lead to a lack of precision when only using within‐unit variation. ⁷¹ These models also assume strict exogeneity in which past values for covariates cannot modify current outcomes and past outcomes cannot modify current covariates, which may bias estimates. Additionally, not including time‐varying confounders in the model may lead to omitted variable bias in the estimates. Lastly, the parallel trends assumption may be difficult to verify given that controls for any given unit (a county, in our study) is the unit itself at a previous timepoint. Given the limitations, scholars argue that more complex models may prove beneficial when researchers face threats of reverse causality or state dependence. Our study may rely on assumptions such as strict exogeneity and parallel trends, but we do not suspect that psychiatric ED visits may lead to greater incarceration among Black Americans within counties. Additionally, our model does not utilize lags or leads, as we do not hypothesize that incarceration in a certain year will correspond with greater psychiatric ED visits in the next year given our annual resolution. ⁷²

We also conducted several sensitivity tests. First, we reestimated the models and adjusted for outliers in the outcome below the fifth or 95th percentile. This adjustment determines whether extreme values in psychiatric ED visits drove any discovered association with incarceration (per 100,000 population) among Black Americans. Second, we assessed the sensitivity of results by using a generalized estimation equation that accounts for gamma distributions and uses a population offset. This model accounts for differences in population sizes, in the outcome variable, that population prevalence estimates may not address. Third, we examined the sensitivity of our results by utilizing a summation of jail and prison prevalence as the exposure. Fourth, we stratified our analyses by assessing psychiatric ED visits among Black Americans as a function of jail and prison prevalence, separately. Fifth, we examined psychiatric ED visits among Black Americans as a function of an additional measure of racial inequity, the difference in incarceration (per 100,000 population) between Black and White individuals within a county.

As additional checks, we conducted the following analyses: 1) whether incarceration among White Americans corresponds with changes in psychiatric ED visits among White Americans, and 2) whether the ratio of Black to White American incarceration coincides with changes in psychiatric ED visits among White Americans. Because incarceration in the United States concentrates in low‐income Black communities, we reasoned that incarceration would show a relatively weaker (or null) association in psychiatric ED visits among White Americans. Similarly, we reasoned that the racial inequity in incarceration (i.e., lower incarceration among White Americans) would not correspond with changes in acute psychiatric needs among non‐Hispanic White Americans. We conducted all analyses using Stata SE version 16.0.

Findings

The study sample shows an average of 10,350.92 psychiatric ED visits per 100,000 population per county‐year among Black Americans (Table 1). Incarceration among Black Americans averages 24.33 per 100,000 population per county‐year (Table 1). As seen in Table 1, the ratio of Black to White American incarcerated populations has a mean value of 6.16 over the study period. Figure 1 plots average psychiatric ED visits (per 100,000 population) among Black Americans across counties over time. Consistent with previous literature, psychiatric ED visits show a secular increase from 2006 to 2015. ⁴⁵ , ⁵¹ Figure 2A plots average incarceration (per 100,000 population) among Black Americans across counties over the study period. Figure 2B shows the average ratio of Black to White incarceration across counties by year. The decrease in incarceration among Black Americans coheres with current population‐level trends in the United States. However, Black Americans still remain overrepresented in the US carceral system.

Table 1.

County‐Level ED, Incarceration, and Socioeconomic Characteristics of 404 Counties From Ten US States, ^a 2006 to 2015

Variable	Mean (±SD)
Psychiatric ED visits among Black Americans (per 100,000 population), n	10,350.92 (±10,074.39)
Incarcerations among Black Americans (per 100,000 population), n	18.99 (±11.56)
Black to White incarceration ratio	6.16 (±3.96)
Arrests (per 1,000 population), n	6.25 (±9.57)
Unemployed, %	7.85 (±3.01)
Below the federal poverty line, %	15.99 (±6.16)
Hospitals, n	123.90 (±969.44)

ED, emergency department; SD, standard deviation.

^a Counties are from the following states: Arizona, California, Florida, Kentucky, Massachusetts, Maryland, New Jersey, New York, Rhode Island, and North Carolina

Figure 1.

Psychiatric ED Visits (per 100,000 Population) Among Black Americans From 404 Counties in Ten US States From the Statewide ED Database, 2006 to 2015

ED, emergency department.

Figure 2.

Incarceration (per 100,000 Population) Among (A) Black Americans and (B) the Racial Inequity in Incarceration (Black to White American Incarceration Ratio) in 404 Counties From Ten US States in the Vera Institute of Justice In Our Own Backyard Data Set, 2006 to 2015

Table 2 shows results from linear fixed effects regression analysis examining log‐transformed psychiatric ED visits (per 100,000 population) as a function of incarceration (per 100,000 population) among Black Americans. One additional incarceration (per 100,000 population) corresponds with a 1.4% increase in psychiatric ED visits (per 100,000 population) among Black Americans in that county‐year (coefficient [coef] = 0.014, standard error [SE] = 0.005, p < 0.001; Table 2). Table 3 reports the relation between the ratio of Black to White American incarceration (i.e., a gauge of the racial inequity in incarceration) and psychiatric ED visits (per 100,000 population) among Black Americans. The racial inequity in incarceration coincides with a slightly higher increase in psychiatric ED visits (2.2%; coef = 0.022, SE = 0.011, p < 0.05; Table 3). Psychiatric ED visits also vary inversely with arrests for violent crime, a control variable in our analysis (coef = −0.003, SE = 0.001, p < 0.001).

Table 2.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of Incarceration Among Black Americans (per 100,000 Population) in 404 Counties From Ten US States, 2006 to 2015

Covariate	Psychiatric ED Visits, Coefficient (±Standard Error) a
Incarceration among Black Americans (per 100,000 population), n	0.014 *** (±0.005)
Arrests (per 1,000 population), n	−0.003 *** (±0.001)
Unemployed, %	0.005 (±0.015)
Below the federal poverty line, %	0.003 (±0.008)
Hospitals, n	2.657E−4 (±2.035E−4)
Year (reference: 2006)
2007	−0.016 (±0.043)
2008	0.139 ** (±0.057)
2009	0.342 **** (±0.092)
2010	0.467 **** (±0.103)
2011	0.598 **** (±0.099)
2012	0.741 **** (±0.092)
2013	0.824 **** (±0.082)
2014	0.999 **** (±0.074)
2015	1.141 **** (±0.075)
N	2,360

ED, emergency department.

^* p < 0.1, **p < 0.05, ***p < 0.01, ****p < 0.001.

^a Values use robust standard errors.

Table 3.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of the Ratio of Incarceration of Black to White Americans in 404 Counties From Ten US States, 2006 to 2015

Covariate	Psychiatric ED Visits, Coefficient (±Standard Error) a
Ratio of incarceration of Black to White Americans	0.022 ** (±0.011)
Arrests (per 1,000 population), n	−0.003 *** (±0.001)
Unemployed, %	0.004 (±0.015)
Below the federal poverty line, %	0.001 (±0.008)
Hospitals, n	2.687E−4 (±2.085E−5)
Year (reference: 2006)
2007	−0.016 (±0.043)
2008	0.146 ** (±0.056)
2009	0.347 **** (±0.091)
2010	0.470 **** (±0.101)
2011	0.594 **** (±0.097)
2012	0.740 **** (±0.090)
2013	0.822 **** (±0.081)
2014	0.998 **** (±0.073)
2015	1.142 **** (±0.074)
N	2,356

ED, emergency department.

^* p < 0.1, **p < 0.05, ***p < 0.01, ****p < 0.001.

^a Values use robust standard errors.

Our sensitivity tests included the following: 1) outlier adjustment (Appendix 3), 2) use of a generalized estimation equation with a gamma distribution and population offset (Appendix 4), 3) summation of jail and prison prevalence as the exposure (Appendix 5), 4) stratified analyses examining jail and prison prevalence alone (Appendix 6A and B), and 5) use of a separate independent variable (i.e., the difference in incarceration between Black and White populations by county‐year as a measure of racial inequity (Appendix 7). Inference from all sensitivity analyses remain robust to the original findings. Lastly, we find no relation between incarceration and psychiatric ED visits among White Americans (Appendix 8). The ratio of Black to White American incarceration also does not correspond with psychiatric ED visits among White Americans (Appendix 9). These checks indicate that the incarceration and psychiatric ED relation concentrates among Black Americans. Figure 3 shows a plot of the point estimates and their confidence intervals for the tests conducted in this study.

Figure 3.

Plot of Point Estimates for Psychiatric ED Visits Among Black and White Americans (per 100k Pop) as a Function of Incarceration (per 100k Pop) and the Racial Inequity in Incarceration in 404 Counties From Ten US States, 2006 to 2015 [Colour figure can be viewed at wileyonlinelibrary.com]

100k, 100,000; ED, emergency department; GEE, generalized estimating equation; OLS, ordinary least squares; pop, population.

Conclusion

Black Americans comprise 40% of the incarcerated population but only 13% of the US population. ² Threats to social community integration and formation caused by overrepresentation in the carceral system may affect psychiatric morbidity in the broader Black population. Our fixed effects analysis indicates a modest, positive relation at the county‐year level between incarceration and psychiatric ED visits among Black Americans. We also find that annual deviations in the intensity of incarceration among Black Americans relative to White individuals (measure of racial inequity) correspond positively with slightly greater psychiatric help seeking among Black Americans.

Our findings align with the theoretical underpinnings of incarceration's influence on the broader population not directly connected to the incarcerated person. Incarceration removes individuals from family and social networks, eliminating financial or supportive contributions to households, as well as participation in communities. ⁴ , ¹⁵ , ²² , ²³ , ⁷³ As evidenced by previous literature, social ties, both strong and weak, have substantial benefits for mental health. ¹⁶ , ³⁴ , ³⁵ Theorists and empirical scholars have also compared the influence of the carceral system in the United States with acts of forced migration. ⁷⁴ This body of work indicates that social destabilization, resulting in the removal of specific groups, has preceded changes to gender ratios in certain regions as well as strains to social cohesion. ⁷⁴ Such disruption may change social norms established within communities that help prevent disease. ⁷⁴ Loss of groups of individuals from mass incarceration, specifically Black men with lower educational backgrounds, results in the forfeiture of monetary and familial support for numerous networks. ²² Disruption of such networks, through removal of male role models, partners, and children, further exacerbates distress in already disadvantaged communities. ²⁸ , ⁷⁵ , ⁷⁶ Although this study does not directly assess whether social capital and family networks mediate or moderate the relation between incarceration and acute mental health, further work would benefit from such an understanding. Well‐established stigmatizing attitudes toward routine mental health treatment and untreated mental illness may also result in use of the ED for psychiatric care in the Black community. ⁷⁵

We find that racial inequities coincide with greater psychiatric help seeking among Black Americans (2.2%). This finding highlights the prominence of differential access to opportunities between Black and White Americans caused by legislation and actions by the criminal legal system, such as disparities in drug arrests or crack cocaine sentencing. ⁷⁷ , ⁷⁸ , ⁷⁹ In turn, these actions reinforce racial/ethnic biases that contribute to adverse health in Black populations. ⁷⁷ , ⁸⁰ These findings offer credence to the often invisible systemic factors that contribute toward racial biases in the criminal legal system and the “indirect effects” on those not directly involved in the system.

Our stratified analyses show modest increases in psychiatric ED visits among Black Americans as a function of jail and prison incarceration (per 100,000 population) separately. We originally utilized a combined measure to capture the joint removal of individuals from communities and the potential transfers that may occur between jails and prisons. Although our inference remains similar across different measures of incarceration (i.e., summation of/average jail and prison prevalence, jail prevalence alone, and prison prevalence alone), stratified findings show greater uncertainty. Although we do not have the granularity in our criminal legal data to evaluate aspects of the jail and prison systems that may contribute to this uncertainty, we encourage future qualitative or mixed‐methods research to further enhance this area of work. Stratified findings also show a modestly greater impact on psychiatric emergencies as a function of prison prevalence. When using the average jail and prison prevalence, as opposed to a summation, our effect size nearly doubles. We speculate that counties that have greater prison populations may not have as many jails and averaging the two would remove its influence on the outcome and therefore produce a greater effect size overall.

We report no difference in psychiatric emergencies as a function of incarceration among White Americans. Reports show that almost half of all Americans have had a family member incarcerated. However, this disproportionately affects Black Americans, with a representative sample showing that Black Americans have a 50% greater likelihood of having a family member incarcerated when compared with White Americans. Black Americans are also three times more likely to have had a family member incarcerated for at least 10 years when compared with White people. ⁸¹ White Americans may therefore not experience the same carceral churn or strain in familial and social networks, which degrades social capital and results in greater acute and adverse mental health experienced by Black communities.

We also do not find a relation between the racial disparity in incarceration and changes in psychiatric ED visits among White Americans. Previous work examining comprehensive measures of structural racism that use economic, education, political, criminal legal, and segregation domains report adverse health outcomes among Black but not White Americans. Within the criminal legal domain, researchers use a measure of state‐level Black to White American incarceration. After creating a latent measure for structural racism by including all domains, they find that Black Americans show adverse mental health, whereas White Americans show enhanced mental health as a function of structural racism. Our findings align with the “no effects/countervailing forces” hypothesis in which White American health is not affected by the racial disparity in incarceration. This disparity may benefit White Americans by providing a greater share of economic resources but also harming them through diminished social cohesion and investment in public goods, resulting in no difference in health. ⁸⁰ Additionally, incarceration further exacerbates existing racial disparities between White and Black Americans by imposing economic and social burdens on Black individuals, families, and communities through fines, fees, and collateral consequences such as difficulty securing housing and employment. ⁸²

Our analyses consistently report a decrease in psychiatric ED visits as a function of greater arrests for violent crime. This finding coheres with previous studies conducted on the concept of “transinstitutionalization” of mental illness from psychiatric hospitals to the justice system. Reductions in public psychiatric hospitals and beds following deinstitutionalization in the United States have led to increases in arrests and incarceration, even after accounting for unemployment and poverty. ⁸² , ⁸³ Further work would benefit from better understanding whether racial bias in policing may lead to transinstitutionalization of Black communities wherein individuals who require psychiatric care for mental health issues are instead incarcerated or killed by police.

Strengths of the analysis include the use of comprehensive, statewide data sets for psychiatric ED visits. We also utilized clinically diagnosed ICD‐10 codes for psychiatric ED visits. The fixed effects specification accounted for time‐invariant characteristics at the county level and “forces” a within‐county analysis (over time) that does not rely on across‐county comparisons that assume comparability across regions. Year indicators also controlled for exogenous shocks that may have altered psychiatric ED utilization or incarceration, such as the Affordable Care Act or drug reform policies. We also accounted for potential confounders, such as heightened arrests for violent crimes and fluctuations in county‐level socioeconomic factors that may correspond with the exposure or the outcome. These findings contribute to the literature on the racial inequities in the criminal legal system and health, operationalized as the ratio of incarceration among Black and White individuals within a county‐year. Extending previous work on neighborhood incarceration and mental health, our findings also suggest that the long arm of incarceration influences psychiatric help seeking within counties by race/ethnicity.

This study has limitations. Our findings do not comprise the prevalence of psychiatric conditions in the population, as not all individuals seek psychiatric care in the ED. Therefore, the results may represent the lower bound of psychiatric conditions in the population. Data limitations prevent examination of the exposure and the outcome in all US counties. However, the study sample includes states from three of the four regions in the United States and comprises 20% of the country's population. Although we did not have the data to do so, we encourage future research to examine the relation between incarceration and psychiatric emergencies after 2015 given changes in state policies that may reduce the racial disparities in incarceration. Additionally, although our study did not examine specific mechanisms between incarceration and psychiatric emergencies, future research should assess mechanisms such as strains in familial and social networks, carceral churn/coercive mobility, degraded social capital, and material hardship and stress.

Additionally, the IOB data set (created from the National Corrections Reporting Program) provides incarceration data on the county where courts impose a sentence, which may differ from the county of residence. ⁵³ This may impede use of theories on family networks and social capital, as incarcerated populations within counties may consist of individuals who reside in other counties within the same state. The Federal Bureau of Prisons considers several factors when determining an inmate's designation following conviction: level of security, programming needs (i.e., substance use treatment), bed availability, and primary residence. Specifically, the Federal Bureau of Prisons prioritizes placement in institutions that remain reasonably close to anticipated home release locations, subject to other factors. ⁷⁷ Jails and prisons may also function differently in this regard in that cities, local districts, and counties operate jails, whereas states and federal governments manage prisons. Measurement of within‐county changes over time in this study, however, may control for time‐invariant characteristics such as types of facilities located within counties and their subsequent programming and security for inmates. This may not account for confounding due to the introduction of new prisons within localities; however, sources report that states have slowed building new prisons since the industry expanded significantly between 1990 and 2005 (prior to our study period of 2006–2015). One study finds that approximately 35% of individuals live within 100 miles of their designated prison; however, limited research exists on whether patterns differ for jails given their localized jurisdiction. ⁷⁸ Use of both prison and jail incarceration in this study, however, may provide the most inclusive evaluation of within‐county changes. Availability of more fine‐grained data with information on county of residence would allow for a more rigorous analysis of the pathways by which incarceration strains family networks and social capital.

Our study findings reflect changes in psychiatric help seeking as a function of the prevalence of jail and prison incarceration among Black Americans. Data limitations prevent more thorough investigations of whether incidence or increases in new entrants to jail and prison may also exacerbate population‐level mental health. Further research disentangling whether prevalence or size of the population incarcerated as opposed to new admissions would provide a better understanding of the community impact of incarceration.

Although our model specifications control for time‐invariant unobserved characteristics, indicators for annual changes, as well as arrests for violent crimes, the proportion below the federal poverty line, unemployment, and the number of hospitals, our model may not rule out other unobserved time‐varying factors. We acknowledge that there are several causal pathways that may underlie the relation between incarceration and psychiatric help seeking in Black communities. Criminal victimization in these communities as well as other socioeconomic factors may likely influence incarceration in addition to psychiatric needs. To best address the potential confounding introduced by omitted time‐varying factors, future analyses could test an exogenous policy, such as state drug reform, that would affect the prevalence of incarceration. A valid instrumental variable, such as funding for corrections, may also remove endogeneity introduced by omitted variable bias. We encourage future research to build on the rigor of our analyses to address this potential methodological concern.

Suicides have increased substantially in the past two decades, becoming the second leading cause of death among Black youth. Suicide rates among certain Black age groups have also surpassed those of White Americans. Previous work reports that familial incarceration precedes greater adverse mental health among children. ³⁷ , ⁴² One study found that racial bias in parental imprisonment corresponds with increased mental health disparities in adolescents. ³⁸ Given the significant racial disparity in parental incarceration, with one in ten Black youth having an imprisoned parent, an evaluation of incarceration and suicide among Black youth may uncover underlying ecological risk factors contributing to the current rise and disparity in suicides.

Since 2010, California, New York, and other states have implemented policies reclassifying drug possession from felonies to misdemeanors. Felony convictions for drug possession have particularly imposed burdens on communities of color. ⁸⁰ Black Americans remain three to four times more likely to get arrested for illegal drugs as opposed to White Americans after controlling for criminal disposition. ⁷⁹ Although incarceration rates for drug offenses have declined over the past decade, with a reduction in racial disparities in drug‐related incarceration in state prisons, racial disparities persist for drug offenses in federal prison. ⁸⁴ California's Proposition 47 (The Safe Neighborhoods and Schools Act in 2014) reclassified low‐level drug and property crimes from felonies to misdemeanors and reinvested funds into community mental health and substance use programs, substantially influencing populations of color. ⁸⁵ Scholars have found that Proposition 47 has not only preceded a decrease in prison populations but has also shown a decline in racial/ethnic disparities in drug arrests. ⁸⁵ , ⁸⁶ Further research would benefit from evaluating whether drug reform policy and the subsequent reduction in incarceration may have salutary benefits for broader community mental health among minoritized race/ethnicities.

In 2022, Black Americans comprised the greatest proportion of individuals in prison for violent crimes (68.4%) as opposed to White Americans (54.8%). ⁸⁴ Investment in interventions to promote public safety and address crime may depart from solely relying on incarceration and prove beneficial to reducing this disparity. Policies that reduce labor market and housing discrimination for former inmates may also provide the stability necessary for better integration, as well as reduction in recidivism. ⁸⁷

Racial disparities in incarceration capture one facet of the criminal legal system within the broader racial discrimination system. ⁸⁸ Theorists posit that residential location, education, health, housing, credit markets, labor, and justice subsystems comprise a reciprocating, integrated system. ⁸⁸ Although evaluating racial inequities in incarceration allows for directed policy change, such as state‐level drug reform or other social policies, populations released from prison or jail may continue to face barriers within housing and labor markets. These barriers may subsequently lead populations back to the criminal legal system, also known as the “revolving door of the criminal legal system.” The reinforcing nature of these subsystems has prompted researchers to create multidimensional measures for structural racism, incorporating racial disparities within each facet. ⁸¹ , ⁸⁹ , ⁹⁰ Findings report that comprehensive measures correspond with adverse health outcomes in the Black community. ⁸¹ , ⁹⁰ Research on psychiatric help seeking as a function of inclusive structural racism measures may also provide the context necessary to better understand racialized inequities in health.

Funding/Support

This work was supported by the National Institute of Mental Health, National Institutes of Health (grant 1R21‐MH‐110815‐01A1).

Conflict of Interest Disclosures

All authors declare no conflicts of interest.

Appendix 1.

Frequency Distribution of Psychiatric ED Visits Among Black Americans (per 100,000 Population) Across 404 Counties (10 States: Arizona, California, Florida, Kentucky, Massachusetts, Maryland, New Jersey, New York, Rhode Island, and North Carolina), Over 10 Years, 2006‐2015.

Appendix 2.

Frequency Distribution of Psychiatric ED Visits Among Black Americans Following Log‐Transformation (per 100,000 Population) Across 404 Counties (10 States: Arizona, California, Florida, Kentucky, Massachusetts, Maryland, New Jersey, New York, Rhode Island, and North Carolina), Over 10 Years, 2006‐2015.

Appendix 3.

Outlier‐Adjusted Linear Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of Incarceration Among Black Americans (per 100,000 Population) in 404 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Incarceration among Black Americans (per 100,000 population)	0.009**	0.004
Arrests (per 1,000 population)	−0.003***	9.554E‐4
Percent unemployed (%)	−0.003	0.014
Percent below the federal poverty line (%)	−3.952E‐4	0.008
Number of hospitals	1.662E‐4	1.386E‐4
Year (reference: 2006)
2007	0.013	0.027
2008	0.180****	0.041
2009	0.360****	0.082
2010	0.467****	0.089
2011	0.586****	0.085
2012	0.689****	0.079
2013	0.759****	0.070
2014	0.917****	0.061
2015	1.054****	0.066
N	2,108

*p<0.1; ^**p<0.05,***p<0.01****p<0.001

^aRobust standard errors

Appendix 4.

Generalized Estimating Equation Fixed Effects Results Predicting the Count of Psychiatric ED Visits Among Black Americans (With a Population Offset and Gamma Distribution) as a Function of Incarceration Among Black Americans (per 100,000 Population) in 404 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Incarceration among Black Americans (per 100,000 population)	0.013***	0.003
Arrests (per 1,000 population)	0.001	0.001
Percent unemployed (%)	0.007	0.011
Percent below the federal poverty line (%)	−0.001	0.005
Number of hospitals	1.1E‐5	7.61E‐6
Year (reference: 2006)
2007	−0.008	0.032
2008	0.145***	0.036
2009	0.327***	0.068
2010	0.449***	0.072
2011	0.581***	0.068
2012	0.732***	0.062
2013	0.813***	0.057
2014	0.981***	0.050
2015	1.027***	0.063
N	2,360

*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors

Appendix 5.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of Black Americans in Prison and Jail (Summation) (per 100,000 Population) in 366 Counties From Ten US States, 2006‐2015.

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Incarceration among Black Americans (per 100,000 population)	0.007***	0.002
Arrests (per 1,000 population)	−0.003***	0.001
Percent unemployed (%)	0.005	0.015
Percent below the federal poverty line (%)	0.003	0.008
Number of hospitals	2.657E‐4	2.035E‐4
N	2,360

*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors

Appendix 6A.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of Black Americans in Prison (per 100,000 Population) in 372 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Incarceration among Black Americans (per 100,000 population)	0.006*	0.003
Arrests (per 1,000 population)	−0.003***	0.001
Percent unemployed (%)	0.006	0.015
Percent below the federal poverty line (%)	0.002	0.008
Number of hospitals	2.726E‐4	2.059E‐4
N	2,398

*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors

Appendix 6B.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of Black Americans in Jail (per 100,000 Population) in 393 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Incarceration among Black Americans (per 100,000 population)	0.004*	0.002
Arrests (per 1,000 population)	−0.003***	0.001
Percent unemployed (%)	0.008	0.015
Percent below the federal poverty line (%)	0.002	0.008
Number of hospitals	2.664E‐4	2.054E‐4
N	2,622

^*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors

Appendix 7.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among Black Americans as a Function of the Difference in Incarceration Between Black and White Americans in 404 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Difference in incarceration between Black Americans and whites (population per 100,000)	0.015***	0.005
Arrests (per 1,000 population)	−0.003***	0.001
Percent unemployed (%)	0.005	0.015
Percent below the federal poverty line (%)	0.003	0.008
Number of hospitals	2.659E‐4	2.033E‐4
Year (reference: 2006)
2007	−0.016	0.0427
2008	0.141**	0.057
2009	0.345****	0.092
2010	0.473****	0.103
2011	0.605****	0.099
2012	0.753****	0.092
2013	0.834****	0.082
2014	1.011****	0.074
2015	1.152****	0.076
N	2,357

^*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors

Appendix 8.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among White Americans as a Function of Incarceration Among White Americans (per 100,000 Population) in 404 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Incarceration among whites (population per 100,000)	−0.020	0.024
Arrests (per 1,000 population)	−0.001	0.001
Percent unemployed (%)	0.001	0.015
Percent below the federal poverty line (%)	0.002	0.010
Number of hospitals	4.082E‐4	2.515E‐4
Year (reference: 2006)
2007	−0.022	0.054
2008	0.193****	0.054
2009	0.341****	0.095
2010	0.436****	0.107
2011	0.552****	0.103
2012	0.665****	0.094
2013	0.718****	0.087
2014	0.885****	0.072
2015	0.965****	0.078
N	2,519

*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors

Appendix 9.

Ordinary Least Squares Fixed Effects Regression Results Predicting Log‐Transformed Psychiatric ED Visits (per 100,000 Population) Among White Americans as a Function of the Ratio of Incarceration Among Black to White Americans in 404 Counties From Ten US States, 2006‐2015

	Psychiatric ED visits
Covariate	Coefficient	Standard Errora
Ratio of incarceration of Black Americans to whites	0.010	0.014
Arrests (per 1,000 population)	−0.002**	8.819E‐4
Percent unemployed (%)	−0.004	0.015
Percent below the federal poverty line (%)	0.006	0.010
Number of hospitals	3.788E‐4	2.399E‐4
Year (reference: 2006)
2007	0.010	0.051
2008	0.197****	0.055
2009	0.359****	0.093
2010	0.457****	0.105
2011	0.572****	0.102
2012	0.672****	0.097
2013	0.714****	0.091
2014	0.870****	0.079
2015	0.971****	0.080
N	2,369

*p<0.1, **p<0.05, ***p<0.01, ****p<0.001.

^aRobust standard errors