Abstract
About one in every five Americans suffers from inadequate sleep, a key indicator of physical and mental well-being. Using data from the National Longitudinal Study of Adolescent Health, we investigate the association between incarceration and two related sleep problems: short sleep duration and symptoms of insomnia. We find that relative to those who have not been incarcerated, those with a history of incarceration have an elevated risk of short sleep duration and insomnia symptoms. These associations persist even after controlling for a host of potential confounders, including prior sleep problems. Additionally, the associations do not differ significantly across race.
Introduction
Over 600,000 individuals are released from state and federal prisons each year (Carson 2015), and millions more are released from local jails (Minton and Zeng 2015). For these individuals, incarceration will likely compound preexisting hardships, increasing labor market, familial, social, and housing difficulties (Travis, Western, and Redburn 2014; Wakefield and Uggen 2010; Western et al. 2015). Incarceration is also associated with an increased risk of several mental and physical health conditions, including depression, anxiety, heart disease, functional limitations, and infectious diseases (for a review, see Massoglia and Pridemore 2015). Although prisoners tend to be in poor health at intake (Conklin, Lincoln, and Tuthill 2000), research suggests that incarceration also worsens health by disproportionately exposing individuals to stressors and infectious diseases (Massoglia 2008).
An important component of the incarceration-health nexus that has been largely overlooked is sleep behavior. Balancing stress levels is crucial to maintaining a healthy sleeping pattern (Van Reeth et al. 2000), and indeed current inmates exhibit high rates of sleep problems compared to the general public. For example, approximately 40 percent of state prisoners and 50 percent of jail inmates exhibit symptoms of insomnia or hypersomnia (James and Glaze 2006) compared to 30 percent of the general population (Roth 2007). The incidence of sleep problems among inmates is concerning given the close connection between sleep and overall well-being (Colten and Altevogt 2006). In fact, insufficient sleep is currently considered a pervasive public health problem facing the U.S. population in general (Centers for Disease Control and Prevention 2015; Colten and Altevogt 2006).
Sleep is especially tied to psychological well-being (Banks and Dinges 2007). For instance, M. P. Walker (2009:185) notes: “Nearly all psychiatric and neurological mood disorders express co-occurring abnormalities of sleep.” Sleep difficulties tend to co-occur with mental health problems such as depression, anxiety, and psychological distress (Benca et al. 1992; Benca and Peterson 2008; Breslau et al. 1996; Ford and Kamerow 1989; Jansson and Linton 2005; Reid et al. 2006). Inadequate sleep has also been tied to increases in suicide and suicidal ideations (Bernert et al. 2015; Ribeiro et al. 2012). Although it is difficult to disentangle the role of sleep since poor sleep can cause, co-occur with, or be caused by an individual’s mental health (Anderson and Bradley 2013; Buysse et al. 2008; M. P. Walker 2009), it remains clear that mental health and sleep are inexorably linked.
To be sure, a number of studies establish an association between incarceration and mental health directly (Massoglia 2008; Porter and Novisky forthcoming; Schnittker, Massoglia, and Uggen 2012; Turney, Wildeman, and Schnittker 2012). However, sleep is an important variable to consider in its own right—rather than simply as a proxy or risk marker. Sleep is a modifiable behavior and therefore represents a promising avenue for health interventions among the formerly incarcerated population (Grandner et al. 2016). Second, it is connected to a wide variety of outcomes related to cognitive functioning and mood disorders, which also have implications for everyday behavior. For example, lack of sleep can lead individuals to be less amicable in their relationships, less effective as employees, and less productive in general (Dinges et al. 1997; Kahn-Greene et al. 2007; Scott, McNaughton, and Polman 2006; Swanson et al. 2011; M. P. Walker 2009)—all of which can exacerbate the reintegration challenges already confronted by former inmates (Harding et al. 2014; Visher and Travis 2003; Western et al. 2015). Third, sleep problems are unevenly distributed across the population, along with many other health concerns. An estimated 50 to 70 million Americans suffer from inadequate sleep (Colten and Altevogt 2006); however, blacks are more likely than whites to report inadequate sleep duration (Hale and Do 2007) and unintentionally falling asleep during the day in the past month. Blacks are also twice as likely to have nodded off while driving (Centers for Disease Control and Prevention 2011).
Those who spend time in jails or prisons are at high risk of developing sleep problems given their disproportionate exposure to stressors. However, relatively little is known about how incarceration affects sleep patterns. While there is some descriptive information available for current inmates, formerly incarcerated persons represent a segment of the population so large, they are now considered a “felon class” (Uggen, Manza, and Thompson 2006). This group faces a complex and powerful set of challenges that stand to threaten health and well-being. Sleep problems, another penalty of incarceration that may be incurred by this group, could further exacerbate the ability to cope with other reintegration hurdles such as maintaining steady employment and forging prosocial relations (Ailshire and Burgard 2012; Burgard and Ailshire 2009; Knudsen, Ducharme, and Roman 2007). Thus, in the current study, we investigate whether a history of incarceration is associated with poor sleep.
Using the National Longitudinal Study of Adolescent Health (Add Health), we examine the relationship between incarceration and two related sleep problems: insomnia symptoms and short sleep duration. We examine these outcomes at Wave IV, when respondents were between the ages of 24 and 34. We also control for other risk factors known to affect sleep—most importantly, prior sleep problems during adolescence. Finally, we test whether the association between incarceration and sleep is contingent on race given prior research suggesting racial differences in the effects of incarceration as well as racial differences in sleep patterns.
Background
The Consequences of Incarceration for Stress-related Health
Stressful life events have negative implications for health generally (Thoits 1995, 2010) and sleep behavior specifically (Åkerstedt, Kecklund, and Axelsson 2007; Burgard and Ailshire 2009; Van Reeth et al. 2000). Indeed, psychological stress is associated with the development of sleep problems and poor sleep quality (Morin, Rodrigue, and Ivers 2003; Pillai et al. 2014). When stressed, the body releases neurotransmitters and hormones such as adrenaline, noradrenaline, and cortisol, which are related to difficulty falling asleep and sleep disturbances throughout the night (Burgard and Ailshire 2009; Kumari et al. 2009; Meerlo, Sgofio, and Suchecki 2008). For instance, prior research finds stress leads 43 percent of adults to lie awake at night, more than 50 percent of adults to report feeling sluggish or lazy after a poor night sleep, and 42 percent of adults to report either poor or fair sleep quality while feeling stressed (American Psychological Association 2014). Those who experience ongoing stress are also at heightened risk for insomnia and lower quality sleep (Pillai et al. 2014).
Prior research suggests that incarceration constitutes a particularly stressful life event. Specifically, the experience of incarceration can be likened to “stress proliferation” as outlined by Pearlin (1989), which distinguishes between primary and secondary stressors (Pearlin, Aneshensel, and LeBlanc 1997). In terms of primary stressors, incarceration results in an immediate deprivation of liberty and autonomy (Massoglia 2008; Sykes 1958), coupled with a series of daily stressors such as overcrowding, a lack of privacy, as well as exposure to violence (Massoglia and Pridemore 2015; Walker 2016). While each of these can be conceptualized as stressors, the living conditions of prison may disrupt sleep in a more direct way. Inmates usually do not have their own private space to sleep and in fact may often find themselves sleeping in a room with multiple people. In addition, silence and darkness are rare commodities inside a prison. Even at night, prisons are dimly lit for security reasons (Dewa etal. 2017; Rocheleau 2013; Zamble and Porporino 1988). Following release from prison, former inmates are then exposed to a host of challenges serving as potential secondary stressors, including barriers in the labor market, fractured social ties, housing dilemmas, and the social stigma of being an “ex-con” (Harding etal. 2014; Pager 2007; Wakefield and Uggen 2010; Western etal. 2015) Indeed, the diminished status of former prisoners may be a particularly disruptive force in terms of their mental health since status affects “the degree of autonomy and control individuals have and their opportunities for full social engagement” (Marmot 2004:46). Additionally, Maruna (2001) finds that former inmates tend to adopt a “doomed to deviance” mindset and feel that their futures are largely out of their control.
Consistent with these contentions, a growing body of research links incarceration to stress-related health outcomes. Massoglia (2008) finds that individuals with a history of incarceration are more likely to be diagnosed with hypertension, anxiety, chronic headaches, and sleep problems. Porter (2014) also finds that incarceration is linked to stress-related health behaviors, including cigarette smoking and fast food consumption. Other research has focused more explicitly on mental health itself, finding that incarceration is associated with an increased risk of depression and dysthymia (Porter and Novisky forthcoming; Schnittker etal. 2012; Turney etal. 2012). A handful of these studies also explored mechanisms, finding that declines in mental health among former inmates were operating primarily through variables relating to material hardship, family functioning, and socioeconomic status (Porter and Novisky forthcoming; Turney etal. 2012).
We also explore the possibility that any effect of incarceration on sleep may be contingent on race. On average, blacks exhibit more sleep problems than whites (Hale etal. 2009; Mezick etal. 2008). Such differences may be due to differential exposure to social stressors and chronic psychosocial stress (Hale and Do 2007; Hicken etal. 2013; Mezick etal. 2008). In particular, chronic psychosocial stress related to institutional and interpersonal racial discrimination is also more prevalent among blacks and is related to racial differences in sleep (Grandner etal. 2012; Hicken etal. 2013). Blacks may be disproportionately vulnerable to certain stressors as well. For example, even when exposure to stressful life events is similar across blacks and whites, these events tend to account for more variation in depressive symptoms among blacks (Brown, Meadows, and Elder 2007). Similarly, loss-related events are more common among black respondents, and these events explain more variation in depression among blacks than for whites (George and Lynch 2003). It could be that blacks have fewer resources, on average, than whites for combatting stressors. On the one hand, a differential exposure perspective suggests that blacks may accumulate more stressors than whites prior to their incarceration experiences, dampening any impact of incarceration at that point. However, black inmates may be more vulnerable to the stressors of incarceration and have fewer resources—financially, socially, and personally—meaning the impact of incarceration could be more powerful on their sleep patterns.
Finally, Porter and Novisky (forthcoming) show no race differences in the association between incarceration and depressive symptoms, suggesting that some of the harmful effects of incarceration may not discriminate based on race. Taken together, this research tells a conflicting narrative about the race-specific effects of incarceration. As such, we hypothesize that incarceration should exhibit race-specific associations with sleep problems, but the expected direction is uncertain.
Current Study
In the current study, we investigate the relationship between incarceration and sleep problems. We analyze two key outcomes—sleep duration and insomnia symptoms. Additionally, incarceration disproportionality impacts racial minorities (Travis etal. 2014), and sleep problems tend to be unequally distributed across the population (Chen etal. 2015; Hale and Do 2007; Lauderdale etal. 2006; Ruiter etal. 2010, 2011). Thus, we investigate racial differences in the link between incarceration and post-release sleep patterns. We examine these relationships using Add Health, a national representative sample of young adults.
Data and Methods
Sample
Add Health is a nationally representative survey of adolescents enrolled in Grades 7 through 12 during the 1993-1994 academic year. At Wave I, an in-school survey was distributed to more than 90,000 students in 132 schools. Following the initial survey, approximately 20,000 participants were selected for in-home interviews. These interviews were between one and two hours and covered an array of topics, including respondents’ health, decision-making processes, educational aspirations, family dynamics, delinquency, and sexual behaviors. Three subsequent follow-up interviews have been conducted since the initial survey, the most recent of which was administered in 2007 and 2008 (Wave IV). At this wave, respondents were between the ages of 24 and 34. The Wave IV interview included items about contact with the criminal justice system as well as detailed information on sleep patterns. This study draws on data from 10,174 respondents and uses data from Waves I and IV.
Variables
Incarceration
At Wave IV, respondents were asked the following survey question: “Have you ever spent time in a jail, prison, juvenile detention center or other correctional facility?” By the time of the Wave IV interview, approximately 15 percent of the sample reported being incarcerated at some point in their lives. Thus, exposure to incarceration is measured as a dichotomy signifying whether a respondent has spent any versus no time behind bars. 1
Sleep Patterns
We measure sleep patterns at Wave IV using two variables: insomnia symptoms and short sleep duration. Insomnia symptoms are measured using a scale calculated by combining responses to the following questions: “How often do you have trouble falling asleep?” and “How often do you have trouble staying asleep?” Each response was scaled 0 to 4, where higher numbers indicated greater frequency of insomnia. Responses to these items were combined to generate a scale ranging from 0 to 8. Short sleep duration was measured as a dichotomy indicating whether respondents sleep less than seven hours on “free” nights (i.e., nights when they do not need to get up for work or related activities the next day). Short sleep duration is defined as less than seven hours given the guidelines provided by the National Heart, Lung, and Blood Institute and the Centers for Disease Control, which recommend seven to eight hours of sleep for adults. Respondents reported the time at which they usually go to bed and the time at which they usually wake up on these days. These times were used to calculate a total number of hours spent sleeping, and the variable was then transformed into a dichotomy capturing those who slept less than seven hours versus seven hours or more. About 2 percent of respondents were coded as missing on short sleep duration due to providing times that seemed either implausible or impossible. For example, 254 respondents had negative values on this variable. In some cases, it appears that respondents accidentally switched the reporting of their sleep and wake-up times. These respondents did not vary significantly on incarceration history, prior sleep patterns, or health. 2
Controls
A number of controls are included that may confound the relationship between incarceration and sleep patterns. We control for demographic characteristics, including the age, race, and sex of respondents. We also control for parental education by Wave I, which captures the highest degree earned across parents, as well as whether the respondent graduated from high school. We control for prior sleep problems at Wave I (trouble falling or staying asleep), including the average number of hours respondents slept per night and whether respondents felt they “usually” get enough sleep. We also control for Wave I measures of prior delinquency, hard drug use (drug use other than marijuana), alcohol use, and tobacco use. Delinquency is measured as a weighted scale based on involvement in vandalism, shoplifting, other theft, burglary, fighting, selling drugs, and robbery (α = .79). Drug use indicates whether a respondent had ever used an illicit drug other than marijuana during adolescence. Alcohol and cigarette smoking were each measured as the product of the frequency of use and the number of items “typically” consumed each time. Child abuse experiences were gathered at Wave IV, when respondents were asked if they were ever hit with a fist, kicked, or thrown on the floor, into a wall, or down the stairs by a parent or adult caregiver before the age of 18. In addition, respondents were asked if they were ever touched in a sexual way or forced to touch this person in a sexual way or forced to have sexual relations. Individuals who experience these traumatic events are more likely to be incarcerated, and childhood victimization experiences are also associated with sleep disturbance (Noll etal. 2006). Finally, we control for prior depressive symptoms. Respondents were asked how often during the past seven days they experienced nine symptoms, including feeling depressed, too tired to do things, sad, and having trouble focusing. Responses were summed to create a single scale ranging from 0 to 27, with higher scores indicating a higher prevalence of depressive symptoms.
Analytic Strategy
Logistic and negative binomial regressions are used to analyze the relationship between incarceration and sleep patterns. We use logistic regression analyses to model the relationship between incarceration and short sleep duration since it is a binary outcome. Negative binomial regressions are used to model the relationship with insomnia symptoms given that this variable is positively skewed and has a large portion of zero values. Because we have sleep data at Waves I and IV, we are able to assess within-individual changes in sleep across these waves. About 20 percent of the sample dropped out between Waves I and IV. The bias introduced by attrition between Waves I and IV has been found “small in magnitude” after applying sample weights (Harris 2013). Accordingly, all estimates in this study are survey adjusted to account for the multistage cluster design of the Add Health study and to adjust for sample attrition between Waves I and IV.
Results
Descriptive statistics are shown for the Add Health sample in Table 1. Survey-adjusted means and standard errors are displayed as well as the range of values for each variable. Approximately 11 percent of respondents sleep fewer than seven hours on free nights. On average, respondents report relatively low levels of insomnia (2.38 on a scale of 0 to 8). Former inmates constitute about 14 percent of the sample. Interestingly, whites report more insomnia than blacks (2.45 vs. 2.15), somewhat inconsistent with prior research (not shown). However, blacks report sleeping less than seven hours per night at almost twice the rate of whites (18 percent vs. 10 percent).
Descriptive Statistics for Add Health Sample (N = 12,441).
Table 2 displays results of short sleep duration regressed on incarceration and other variables. In Model 1, we regress short sleep duration on incarceration and controls, including prior sleep issues. Incarceration exhibits a significant association with short sleep duration, indicating that those who have been incarcerated have 1.2 the odds of being sleep deprived. Expressed as predicted probabilities, about 11.5 percent of respondents with a history of incarceration are predicted to sleep less than seven hours per night compared to 9.5 percent of those who have not been incarcerated. Not surprisingly, the results also suggest continuity in sleep patterns, with trouble sleeping and shorter sleep durations at Wave I exhibiting strong correlations with sleep durations in early adulthood. Race and a history of child abuse also strongly predict the likelihood of short sleep duration. In Models 2 and 3, we replicate this model for whites and blacks separately. These results suggest that there is no significant association between incarceration and short sleep duration for whites but a strong and statistically significant association for blacks. The effect size is nearly eight times larger for blacks, although based on a test of the equality of coefficients, we cannot claim that the magnitude of these effects is significantly different (Paternoster etal. 1998). In particular, blacks who have been incarcerated are about 8 percentage points more likely to be sleep deprived than blacks who have not (22 percent vs. 14.7 percent), suggesting that more than one out of five blacks with a history of incarceration suffer from inadequate sleep. The lack of a significant finding is not due to the differences in the coefficient sizes but rather due to the fact that the standard error around the estimate for whites is so large. In addition, some of the controls appear to operate somewhat differently across groups, with hours of sleep at Wave I predicting sleep duration at Wave IV for whites but not for blacks. Again, these differences do not appear to be statistically significant.
Logistic Regression of Short Sleep Duration on Incarceration and Other Covariates.
p < .05. **p < .01.
In Table 3, we show results for insomnia symptoms. Consistent with short sleep duration findings, those who spend time in jail or prison experience trouble staying or falling asleep at 1.2 the rate of those who have not. The expected “count” on the scale for insomnia symptoms is 2.8 out of 8 among formerly incarcerated individuals compared to 2.3 for never-incarcerated individuals. The scores on this scale correspond to how often a respondent has trouble falling or staying asleep, meaning that those who have been incarcerated have trouble more often than those who have not. As before, prior sleep problems predict sleep patterns in early adulthood. In addition, being a victim of child abuse and prior depressive symptoms are also strongly associated with insomnia in early adulthood. We next test whether there is any difference in this association between blacks and whites. In Models 2 and 3, it can be observed even from a visual standpoint that the effect sizes are quite similar. However, unlike sleep duration models, we find that the effect is statistically significant for whites but not for blacks. For whites, former inmates have an expected insomnia symptom count of 2.9 compared to 2.4, very similar to the results from the full model.
Negative Binomial Regression of Insomnia Symptoms on Incarceration and Other Covariates.
p < .05, **p < .01.
Discussion
Poor sleep is categorized as a serious public health problem that affects millions of Americans and results in billions of dollars annually in direct and indirect medical costs, hospital services, prescriptions, over-the-counter medications, lost productivity, and sleep-related accidents and injuries (Colten and Altevogt 2006; Lamberg 2004). Estimates suggest the direct health care costs of treating insomnia alone are $13.9 billion annually (Walsh and Englehardt 1999), but the indirect costs of sleep problems due to accidents and lost productivity raise that figure much higher (Kessler etal. 2011; Leger 1994). 3
The current study examined whether incarceration contributes to sleep problems among former inmates. We find that incarceration adversely affects sleep duration and symptoms of insomnia. These results persist after accounting for a host of potential confounders, including prior sleep behavior, delinquent activity, drug use, and prior health. Contrary to our hypotheses, we did not find statistically significant differences in the effects across race. However, the lack of differences is also interesting. Overall, this may suggest that blacks and whites are similarly vulnerable to the pains of imprisonment. However, blacks have higher rates of sleep problems, and some studies suggest that blacks are more vulnerable to stressors. Thus, this may speak to the potency of incarceration as a stressor (see also Porter and Novisky forthcoming). As sleep deprivation and insomnia are related to a variety of chronic illnesses, one implication of our findings is that poor sleep among former inmates will likely translate to higher rates of major depression, anxiety, disease and premature mortality (Foley etal. 2004; Kripke etal. 2002). 4
A few limitations of this study must be noted. First, there may be unobserved factors that confound the relationship between incarceration and post-release sleep behavior. Although we control for a number of key variables, it is possible that there are unobserved factors that are related to both incarceration and sleep—such as undiagnosed mental illness. Accordingly, future research should investigate alternative sources of data and methodological approaches to identify the influence of incarceration on sleep behavior. Second, as with prior research using Add Health data (Porter 2014; Siennick, Stewart, and Staff 2014), the dichotomous measure of incarceration does not specify whether an individual was incarcerated in jail, prison, or some other correctional facility. It is possible that unique facility types could differentially influence sleep behavior of former inmates. While prior research on incarceration and health has not been able to assess these contextual factors (Massoglia and Pridemore 2015), it would be worthwhile for future research to investigate the heterogeneity in incarceration experiences.
We end on a note about the implications of our work for intervention. The findings suggest that if we are to improve physical and mental health functioning of former inmates, criminal justice practitioners may be wise to devote greater attention and resources to sleep behavior modifications. Indeed, emerging evidence suggests that achieving optional levels of sleep can reduce negative emotional functioning and improve mental health (Goldstein and Walker 2014; Wassing etal. 2016; Yoo etal. 2007). For instance, one avenue to treat and modify this behavior apart from having to change structural or social risk factors is through behavioral interventions that educate individuals about sleep and provide behavioral guidelines to help combat insomnia, including reducing the total time in bed, getting up at the same time every day, not going to bed unless tired, and not staying in bed unless asleep (Buysse etal. 2011). Given the significant personal and societal costs related to inadequate sleep and the strong connection to health, functioning, and overall quality of life, targeting interventions at sleep behavior may have profound impacts on the health of former inmates—and on health inequalities more generally.
