Abstract
Research reveals inmate misconduct results from various factors including age, gang membership, program participation, and mental illness. However, no research has examined the influence of physical illness on misconduct. Per general strain theory, we argue that poor physical health is a significant strain that may negatively affect behavior. Using data from the Bureau of Justice Statistics’ 2004 Survey of Inmates in State Correctional Facilities (SISCF), we investigate how acute illnesses, chronic conditions, and physical disabilities influence misconduct. Results suggest acute conditions increase the likelihood of general, serious, and nonserious misconduct in prison. Conversely, chronic ailments decrease the likelihood of all types of misconduct. We find moderate effects for physical disability. Experiencing acute health conditions while incarcerated significantly increases the likelihood of misconduct, suggesting that by appropriately addressing inmates’ acute ailments, it may be possible to concurrently improve inmate health and decrease misconduct to enhance the lives of those in prison.
Individuals with poor health are overrepresented in the U.S. criminal justice system. Close to 43% of state inmates suffer from a chronic health condition 1 and a quarter struggle with a mental health disorder (Wilper et al., 2009). Research suggests that inmate misconduct, or deviations from the formal rules of the prison, results from numerous factors including mental illness (see Steiner, Butler, & Ellison, 2014, for a systematic overview). Studies have consistently found that inmates suffering from a mental illness are more likely to engage in misconduct (see Adams, 1986; Carr, Eggenberger, Crawford, & Rotter, 2013; Felson, Silver, & Remster, 2012; Houser, Belenko, & Brennan, 2012; McCorkle, 1995; Steiner & Wooldredge, 2009; Wood, 2012; Wood & Buttaro, 2013). However, no studies have examined the role physical health plays in misconduct, despite the high rate of physical health ailments behind bars.
Using Agnew’s general strain theory (GST), we argue that physical health conditions while incarcerated may also be associated with inmate misconduct. In prison, inmates face a myriad of strains from overcrowded conditions to a lack of autonomy and security. Inmates may view poor physical health as an added strain in their already strained lives. In addition, they may lack the appropriate coping strategies to handle the negative emotions that accompany physical illness (see Agnew, 2006b, for a systematic overview of GST). Using data from the Bureau of Justice Statistics’ 2004 Survey of Inmates in State Correctional Facilities (SISCF; U.S. Department of Justice, Bureau of Justice Statistics, 2004), the aim of this article, then, is to examine the significantly understudied connection between physical health and inmate misconduct with the goal of improving correctional health care and reducing prison delinquency.
Theoretical Understandings of Inmate Misconduct
Researchers typically rely on three frameworks to understand inmate misconduct—deprivation theory, importation theory, and situational or management perspectives (Morris, Carriaga, Diamond, Piquero, & Piquero, 2012; Steiner et al., 2014). Deprivation theory, stemming from the work of Clemmer (1940) and Sykes (1958), argues that inmate misconduct is a response to the “pains of imprisonment.” This closed-system model suggests that inmate behavior is an adaptation to the strains, or deprivations, of prison (Blevins, Listwan, Cullen, & Johnson, 2010). Morris et al. (2012) found that deprivations associated with prison, as measured by an environmental strain index, increased the likelihood of violent inmate misconduct. In other words, living in an overcrowded, high-security facility with previously violent, gang-affiliated inmates increased the risk of inmate violence (Morris et al., 2012). Similarly, Rocheleau (2013) observed that prisoners who found it difficult to deal with prison hardships such as boredom and concerns for safety were more likely to engage in misconduct and violence.
Recently, Lindsey, Mears, Cochran, Bales, and Stults (2017) investigated how place of incarceration influences misconduct. As a significant deprivation associated with prison, they found that the farther an inmate lives from home—up to 350 miles—the greater the likelihood of institutional misconduct. Research also found that exposure to solitary confinement—the height of deprivation—did not deter misconduct (Medrano, Ozkan, & Morris, 2017). In fact, Medrano et al. (2017) showed that solitary confinement increased the likelihood of future misconduct.
The importation model, however, argues that preprison experiences shape inmate behavior (Irwin & Cressey, 1962). Inmate behavior is not a reaction to the deprivations of imprisonment, but is a reflection of experiences, values, and beliefs held prior to incarceration. Stemming from this model, inmates who espoused criminal values or beliefs prior to prison should be more likely to engage in misconduct while incarcerated than those who did not. Mears, Stewart, Siennick, and Simons (2013) found that an adherence to a criminal belief system is associated with a greater likelihood of inmate violence. In other words, preincarceration values that promote violence increased the risk of violent behavior in prison (Mears et al., 2013).
Situational perspectives suggest that inmate misconduct is a result of the dynamic interaction between inmates and the prison environment such as management strategies or practices, prison architecture, and staff characteristics (Camp, Gaes, Langan, & Saylor, 2003; DiIulio, 1987; Jiang & Fisher-Giorlando, 2002; Morris & Worrall, 2010; Useem & Kimball, 1989). DiIulio (1987) found that prisons with weak or inadequate management strategies are more likely to have higher rates of misconduct. In addition, Useem and Kimball (1989) observed that when administration breaks down, inmates view their confinement as unjust and violence thrives. Similarly, Morris and Worrall (2010) suggested that although prison architecture has little effect on violence, it was associated with nonviolent misconduct.
Health, Crime, and Misconduct
Mental illness is one factor related to inmate misconduct that aligns with multiple theoretical frameworks described above. In particular, mental illness can be imported into the prison or it can manifest in an inmate due to the deprivations associated with incarceration. Studies have found that mental health is a consistent, significant predictor of inmate misconduct (Adams, 1986; Carr et al., 2013; Felson et al., 2012; Houser et al., 2012; McCorkle, 1995; Steiner & Wooldredge, 2009; Wood, 2012; Wood & Buttaro, 2013). Felson et al. (2012) observed that the type of mental illness differentially affects offending behaviors in prison. In particular, psychosis and major depression significantly increased the likelihood of misconduct involving aggression and nonaggression compared with anxiety disorders (Felson et al., 2012). Similarly, co-occurring mental illness and substance use disorders strongly predicted institutional misconduct and victimization for both state and federal inmates (Houser et al., 2012; Wood, 2012; Wood & Buttaro, 2013).
Despite the growing body of research on mental illness and inmate misconduct, no study has empirically assessed the relationship between physical health and prison misconduct. Research has, however, assessed the relationship between poor physical health and offending and/or substance abuse outside of prison (Ford, 2014; Kort-Butler, 2017; Piquero, Daigle, Gibson, Piquero, & Tibbetts, 2007; Piquero, Shepherd, Shepherd, & Farrington, 2011; Schroeder, Hill, Haynes, & Bradley, 2011; Stogner & Gibson, 2010, 2011). Schroeder et al. (2011) found that decreases in physical health directly affected both the onset of offending and crime escalation. Stogner and Gibson (2010) observed that acute health problems (e.g., headaches, stomachaches, sore throat, joint pain) significantly increased the likelihood of nonviolent delinquency. Stogner and Gibson (2011) also revealed that greater amounts of minor health issues increased the likelihood of substance use initiation. Along similar lines, Ford (2014) found that respondents who reported health strain had a greater risk of substance use. However, respondents who reported fewer chronic health conditions (e.g., asthma, cirrhosis, diabetes, heart disease, tuberculosis) were more likely to drink heavily (Ford, 2014). Ford (2014) argued that, although this is counterintuitive to GST, drinking is often a social and peer-involved activity. Individuals with more chronic health ailments may be less likely to be in social situations that are conducive to heavy drinking (see also Umberson & Montez, 2010).
GST, Prison Life, and Health
According to Agnew (1992, 2006b), crime results from the lack of legal coping mechanisms to effectively deal with the negative emotions that may result from three types of strain—the presence of negatively valued stimuli, the loss of positively valued stimuli, and the inability to achieve positively valued goals. There is consistent evidence that strain increases the likelihood of criminal or delinquent coping for many groups, including inmates (e.g., Agnew, 2006a, 2013; Agnew, Brezina, Wright, & Cullen, 2002; Baron, 2004, 2006; Blevins et al., 2010; Brezina, 1999; Broidy, 2001; Eitle & Eitle, 2016; Moons, Morash, McCluskey, & Hwang, 2009; Morris et al., 2012; Snyder et al., 2016). Many of the studies that examine the relationship between poor health and criminal/delinquent behavior, such that adverse physical health influences offending, have framed poor health as a significant strain per GST (see Ford, 2014; Kort-Butler, 2017; Schroeder et al., 2011; Stogner & Gibson, 2010, 2011).
Poor health is itself a negative stimulus. It may also lead to a loss of positively valued stimuli (e.g., peer relationships, participation in favorable activities). Poor health may also affect one’s ability to achieve positively valued goals (e.g., marriage, money). Experiencing poor health in prison only adds to the already strain-filled situation. Prison tends to lack positively valued goals (e.g., programming, visitation), it removes positively valued stimuli (e.g., personal possessions, autonomy), and it contains a surplus of noxious stimuli (e.g., high noise, crowded conditions, victimization). When combined with poor physical health, inmates suffering from any type of physical ailment are at an increased risk of misconduct.
Agnew (2001) argues that strains are most likely to lead to criminal coping when they are viewed as unjust and high in magnitude. Poor health in prison may be perceived as particularly unjust due to the quality of correctional health care and the possibility that they may not have gotten sick, or as sick, had they been in a different environment or been able to access health care quicker. Health issues in prison are similarly likely to be high in magnitude, which makes criminal coping more likely. In addition, the options for noncriminal coping with the deleterious emotional effects of physical health issues in prison may be particularly limited, resulting in a greater propensity to engage in misconduct. As Agnew (1992, 2001, 2006b) argues, strain is likely to lead to negative emotions such as anger and depression, which negatively affects one’s ability to cope via legitimate avenues. It follows, then, that poor health in prison, may lead to anger, irritability, and/or depression, which creates a pressure for misconduct as a way to alleviate the strain and associated negative emotions.
Although illness, regardless of its duration or impact, might be viewed as a strain, acute and chronic health conditions as well as physical disabilities (e.g., hearing and/or vision problems, problems with functional mobility and daily activities) may have differential effects on behavior (see Ford, 2014; Stogner & Gibson, 2010). Acute and chronic illnesses as well as physical disabilities are often accompanied by discomfort, exhaustion, impulsivity, anger, and heightened emotional responses. Acute symptoms, however, tend to be short lived and not often suffered. However, chronic conditions such as arthritis, heart disease, diabetes, and cirrhosis and functional disabilities are more salient, persistent, and long lasting. Although Agnew (2013) would argue that chronic health conditions or physical disabilities are high in magnitude, due to the significant duration of said conditions, individuals may actually be accustomed to being sick or disabled and have the coping mechanisms in place to handle the associated stress.
As explored by Van Gelder (2013), chronic conditions, including chronic physical disabilities, may result in a “cold” affect, which is associated with more self-control and a better ability to weigh the costs and benefits associated with decision making. This particular type of affect, then, may decrease the likelihood of criminal involvement. In contrast, those who experience acute symptoms might not be used to dealing with their illness and may be more prone to the associated negative emotions such as anger, irritability, and depression, thus increasing the likelihood of institutional misconduct. In addition, for inmates who acquire an acute illness while in prison, the symptoms may not only act as a significant strain but also prevent the inmate from coping with other significant strains experienced while incarcerated. Unlike the “cold” affect that might accompany chronic illnesses or disabilities, acute symptoms might lead to “hot” affect, which is argued to lead to less self-regulation, more impulsive behavior, and seemingly irrational decisions (Van Gelder, 2013). Therefore, it is likely that individuals suffering from an acute condition rather than a chronic illness or physical disabilities may be more likely to engage in crime and misconduct.
Current Study
Inmates experience significant daily strains while incarcerated—they face overcrowded conditions, are deprived of personal possessions, lack an individual identity, and are acutely aware of potential violence. Physical illness while in prison should be a particularly salient stressor as these individuals are already experiencing a substantial level of strain. In particular, we predict that experiencing acute health conditions, which may be viewed as more immediate than chronic health ailments or physical disabilities, will increase the likelihood that an inmate will engage in misconduct. Conversely, we predict that suffering from a chronic health condition or having a physical disability, due to their prolonged and sustained impacts on one’s body, will decrease the likelihood of inmate misconduct.
Method
Study Design and Sample
The data for this study come from the 2004 SISCF, which include a nationally representative sample of inmates held in state prisons. In-person interviews and computer-assisted personal interviewing (CAPI) were conducted from October 2003 through May 2004. Interviews covered a range of topics including their current offense, criminal history, family background, demographic characteristics, prior drug and alcohol use, and prison activities, programs, and services. The sampling procedure employed a stratified, two-stage selection process. Correctional facilities were separated into two sampling frames: one for male prisons and one for female prisons. The prison sample was first selected from a universe of 1,585 state prisons across the two gendered sampling frames and 301 prisons were selected at random for inclusion in the study. A total of 287 prisons participated in the study. In the second stage of the sampling process, inmates were randomly selected for participation. Participation was voluntary and 14,499 inmates participated in the study (89.1% response rate; U.S. Department of Justice, Bureau of Justice Statistics, 2004).
Measures
Inmate Misconduct
We used three dependent variables to measure types of inmate misconduct variety across models. Variety measures of deviance, crime, and misconduct have been shown to have greater internal consistency, stability over time than comparable frequency measures, and stronger associations with conceptually related variables over the Raw Frequency Scale (Bendixen, Endresen, & Olweus, 2003; Sweeten, 2012). Researchers also suggest that variety measures may be better suited for assessing overall prevalence within a particular type of misconduct (general, serious, nonserious), rather than the incidence of misconduct within each of the categories using frequency measures (Steiner et al., 2014; Wolff, Blitz, Shi, Bachman, & Siegel, 2006). Although variety and frequency scales of inmate misconduct have both been used as valid measures in criminological research (Steiner et al., 2014), we used variety measures here to measure prevalence within each type of misconduct due to greater internal reliability when compared with frequency measures within the data. We compared the internal consistency of all three variety measures against frequency scales for the same forms of misconduct. Lower alpha coefficients were found for the frequency measures of general misconduct (.57), serious misconduct (.48), and nonserious misconduct (.57).
The first dependent variable is a constructed general variety scale of all misconduct that combines whether or not an inmate has been written up or found guilty of any of the following 10 offenses: drug violation, alcohol violation, weapon possession, possession of stolen property, found in possession of an unauthorized object, verbal assault of a staff member, verbal assault of another inmate, escape or attempted escape, being out of place, or disobeying orders. We coded each form of misconduct “1” if the inmate was ever written up or found guilty of that misconduct and “0” if not. We then added the 10 offense types together to create a scale of general misconduct variety (α = .63).
The second and third dependent variables measure serious and nonserious misconduct to examine how physical health conditions relate differently to specific forms of misconduct in prison. The Serious Misconduct Scale includes drug violation, alcohol violation, weapon possession, possession of stolen property, and escape or attempted escape. We measured each of these as binary items and then combined the five items to create a scale of serious misconduct, in the same manner as the General Misconduct Scale (α = .61). The third dependent variable, Nonserious Misconduct, includes possession of an unauthorized object, verbal assault of a staff member, verbal assault of another inmate, being out of place, and disobeying orders. We added these five binary items to create the variety measure of nonserious misconduct (α = .60). The physical assault of an inmate or a staff member as an act of misconduct was intentionally excluded from all outcome measures. Because inmates may incur injuries or other physical health conditions because of an altercation with another inmate or guard, we leave these measures of misconduct out of all scales due to concerns of reverse causality.
The alpha coefficients for the misconduct measures here are slightly lower than the typically accepted cutoff point of .7 (Schmitt, 1996). After assessing coefficients for all the combinations of misconduct items to develop the total, serious, and nonserious scales, the scales used here achieved the greatest inter-item reliability while maintaining theoretical sense for what constitutes a serious or nonserious type of misconduct. Because most inmates report being written up for zero incidents or only one type of misconduct incident, generally low correlations between misconduct incidents also influence alpha coefficients here (correlation range = .08-.33). However, the measures here provide a good approximation of the variety of misconduct types while enabling separate analyses for different forms of misconduct based on seriousness.
In a supplemental analysis (available upon request), we created dichotomous measures of each of these misconduct scales to assess the influence of physical health factors on whether or not inmates engage in any misconduct for general, serious, and nonserious misconduct categories. Inmates who did not engage in any misconduct were coded “0” and those who had engaged in any form of misconduct within that type of misconduct were coded “1.” Results using these measures largely confirmed the results found using the misconduct variety scales.
Physical Health
The main predictor variables of interest measure various aspects of an inmate’s physical health. The first group of health predictors, acute health conditions, contains measures “since admission” and includes having had surgery, accidental injury (not related to an altercation with another inmate), dental problems, 2 and illness (including a cold, virus, or the flu). We measured each of these using a binary outcome coded “1” if the inmate indicated “yes” for each. All acute issues are considered non–life threatening and nonchronic. Prior research on the relationship between acute conditions and crime outside of prison mainly focuses on symptoms of acute illness such as headache, stomachache, feeling dizzy or weak, or having chest pains (see Kort-Butler, 2017; Stogner & Gibson, 2010, 2011); Although the measures of acute conditions here do not directly measure these previously used symptoms of illness, they represent health problems that are likely accompanied by many similar indications.
The second category of physical health, chronic health conditions, measures whether the respondent currently suffers from a chronic physical health issue. Respondents were asked whether they are currently experiencing any of the following 12 chronic physical health conditions (0 = no, 1 = yes): arthritis or rheumatism, asthma, any type of cancer, cirrhosis, diabetes or high blood sugar, heart problems (including angina, arrhythmia, arteriosclerosis, heart attack, heart disease, valve damage, or tachycardia), hepatitis, hypertension, kidney problems, any type of paralysis, a sexually transmitted disease, or experience of a stroke or brain injury. We created a binary measure to indicate whether the respondent suffers from any of the conditions (1) or none of them (0). Prior studies have used similar chronic conditions such as heart disease, diabetes, hepatitis, and high blood pressure to assess the relationship between health and crime outside of prison (Ford, 2014; Umberson & Montez, 2010).
The final category of physical health assessed is current physical disability. Respondents were asked three questions regarding current physical disabilities related to visual impairment, hearing difficulties, and the use of aids (e.g., cane, wheelchair, walker) for daily functional activity. We combined these three measures and then recoded to reflect whether the respondent suffers from any of these physical disabilities (1) or is without a current physical disability (0). The relationship between physical disability and misconduct is clearly understudied and may be an important factor in assessing the influence of health on crime overall.
The categories of physical health conditions are not mutually exclusive, but rather indicate whether the respondent is suffering from any of the conditions within each category. As a result, it is possible that an inmate may suffer from a chronic condition as well as particular acute issues. However, the data do not indicate whether acute issues are a direct result of a chronic problem and, thus, both are included in the analysis. Correlations among all illness variables included in the analyses suggest no issues of multicollinearity that would arise from including highly correlated variables in the same regression model. The highest correlation among illness variables is between illness since admission and dental conditions (.22) and the relationship between chronic conditions and functional disability (.22). Although it is likely that many ill respondents suffer from comorbid health conditions, the present analysis is designed to examine the differential effects of these illness types.
Mental Health, Anger, and Drug Dependence
In addition to the physical health measures used as predictors of interest, three variables were included to account for the inmate’s mental health and drug dependency. We used a dummy measure to indicate whether the respondent has ever been diagnosed with a mental or emotional condition (0 = no, 1 = yes). We included a scale measure comprised of three items to control for recent anger. The items making up the Anger Scale include being angry more frequently than usual, losing one’s temper more easily than usual, and breaking things or hurting someone as a result of anger (α = .950). Although anger may be a negative emotion that mediates the relationship between health strain and misconduct according to GST (Agnew, 1992), the cross-sectional data used here do not allow for mediation testing. We include the measure of anger in all regression models as a control for heterogeneity related to misconduct among inmates given that alternative measures such as self-control or IQ are unavailable (noted below). We also include a binary measure of any current drug dependence, including alcohol (0 = no, 1 = yes).
Criminal History and Current Sentence
We included a number of variables to control for different incarceration experiences across the sample population. We measured criminal history using the survey’s definition to indicate whether the inmate is a first-time offender (reference category) or a recidivist with a past violent or nonviolent offense (coded “1”). We measured the inmate’s current offense using a categorical variable with the following categories: violent offense (reference), property offense, drug offense, and public order offense. We measured the number of prior incarcerations continuously (observed range = 0-161). We also used a continuous measure for the number of months served of the current sentence, ranging from less than 1 month to 523 months (i.e., 43.6 years). We measured sentence length (in months) using a continuous variable. A binary measure indicates whether the inmate currently has a work assignment on prison grounds (1 = yes). A second binary measure, coded 1 for “yes,” indicates whether the inmate has been a part of a vocational or job-training program since admission.
Demographic Controls
We included demographic indicators to control for variation across the state prison sample. We measured age as a continuous variable (observed range = 16-84 years). We measured gender as a binary variable where male is coded “1.” We measured race using the following discrete categories: White non-Hispanic (reference), Black non-Hispanic, Hispanic, and Other/multiple race (non-Hispanic). Finally, we measured respondent education continuously based on the highest grade level attended prior to incarceration from less than kindergarten through graduate education at equal yearly intervals.
Analytic Strategy
Approximately 15% of the data were missing after accounting for all analytic variables. We carried out multiple imputation using chained equations (MICE) to impute missing values for all variables to decrease the risk of bias from excluded cases and to ensure proper statistical power across all analyses (Little, Jorgensen, Lang, & Moore, 2014; Schlomer, Bauman, & Card, 2010). MICE fills in missing values for multiple variables using a flexible sequence of univariate imputation methods, preserving the distribution of all variables in the analysis via fully conditional specifications (StataCorp, 2015; van Buuren, 2007; van Buuren, Boshuizen, & Knook, 1999). Recent studies assessing competing methods of handling missing data suggest that multiple imputation is appropriate when missing data result in a reduction of more than 10% of the data and bias may result from insufficient statistical power (Cheema, 2014; Langkamp, Lehman, & Lemeshow, 2010). We used chained equations to impute the missing values with 20 imputations for all variables via the “mi impute chained” command in Stata. This results in a full analytic sample of 14,499 for all analyses.
Following imputation, we regressed each of the variety measures of inmate misconduct on all variables described above including physical health, mental health, anger, drug dependence, criminal history, and demographic data. We used negative binomial regression to account for overdispersion 3 in the dependent measures and the high number of “zero” responses indicating no misconduct (60% of the sample). Negative binomial regression also provides greater flexibility than the Poisson distribution because it does not force equality of its mean and variance (Land, McCall, & Nagin, 1996). We expect that responses indicating no misconduct are due to the fact that the inmate simply did not engage in any illicit behavior, rather than due to a secondary process that excludes particular respondents. We, therefore, used a standard negative binomial regression technique, rather than a zero-inflated negative binomial approach. 4 Results of the negative binomial analysis are reported using robust standard errors and incidence rate ratios (IRRs), interpreted as a relative difference measure similar to odds ratios. We used the sample weights provided in the publicly available data set to account for the complex sampling design of the study for all models. We conducted all analyses using Stata 14.2.
Results
Descriptive Results
Table 1 reports the descriptive results for all variables used in the multivariate analyses. Of 10 possible misconduct offenses, most inmates were written up or found guilty of either no offense or a single offense (M = 0.80). Approximately 40% of all inmates were written up or found guilty of at least one offense. Being found guilty of serious misconduct is relatively rare (M = 0.13) whereas participation in nonserious misconduct is much higher (M = 0.67). Regarding physical health issues, about 12% of the sample has had surgery since admission and 22% of the respondents have been hurt in an accident, which can include a broken bone, knocking out or chipping teeth, internal injuries, losing consciousness, or receiving some sort of abrasion. Roughly half of the respondents experienced some sort of dental problem while incarcerated. About 60% experienced some sort of illness including a cold, the flu, or a virus, while in prison. About 17% of respondents reported some kind of physical disability, whereas roughly 40% of inmates are dealing with a chronic health issue. Almost 40% of the sample has a drug dependence and about 30% of inmates have been diagnosed with a mental health disorder.
Unweighted Descriptive Statistics for All Variables (Preimputation; n = 12,323)
The sample is predominantly male (80%) and the average respondent is about 35 years old. Most inmates are White, non-Hispanic (36%) or Black, non-Hispanic (40%) with a mean education level achieved of about 11 years, equivalent to completing one’s junior year in high school. Forty-six percent of offenders are incarcerated for a violent offense and the average number of prior incarcerations is just below two. The average sentence length served is about 55 months or roughly 4.6 years and the average inmate has a sentence length of about 11 years. Most inmates currently have a work assignment (61%), and just more than a quarter of the sample (28%) has participated in job or vocational training. On a scale of 0 to 3, the sample has a mean recent anger “score” of 0.70 with a standard deviation of about 0.98.
Multivariate Results
Table 2 presents the findings for all three negative binomial models after imputation. In the first model, we regressed a general variety measure of inmate misconduct on all independent variables. The results indicated that all four acute physical health conditions were statistically significant predictors of general misconduct, with IRRs ranging from 1.15 (surgery) to 1.40 (accidental injury). This indicated that inmates who have had surgery were 15% more likely to engage in misconduct and those who were accidentally injured were 40% more likely to engage in misconduct (p < .001). These results support the hypothesis that acute conditions increase the likelihood of inmate misconduct. However, both physical disability (IRR = 0.93, p = .045) and having a current chronic illness (IRR = 0.90, p < .001) were associated with a decrease in misconduct (a 7% and 10% decrease, respectively), indicating that disability and serious illness may render the individual too sick or physically unable to engage in misconduct.
Weighted Negative Binomial Incidence Rate Ratios and Robust Standard Errors, Misconduct Variety (Imputed;
Note. IRR = incidence rate ratio.
p < .05. **p < .01. ***p < .001.
We found similar relationships between serious and nonserious misconduct and physical health conditions. All four acute physical health conditions were significantly associated with an increase in serious misconduct. The results indicated that inmates who have had surgery were 20% more likely to engage in serious misconduct (p = .008). Those who experienced dental problems were 29% more likely to receive disciplinary action for serious misconduct. Inmates who suffered an accidental injury were 36% more likely to be written up by officials for serious misconduct. Similarly, inmates who experienced an acute illness were 54% more likely to engage in serious misconduct. In line with the general misconduct results, findings suggested that both chronic illness and physical disability significantly decreased the likelihood of serious misconduct. Inmates with a physical disability were about 17% less likely to engage in serious misconduct than those without a physical disability (p = .017). Similarly, those suffering from a chronic illness were about 12% less likely to engage in serious misconduct than those without a chronic illness (p = .034).
Regarding nonserious misconduct, inmates who suffered from any of the four acute physical health conditions were significantly more likely than those without an acute condition to engage in nonserious misconduct. However, these associations were smaller than that of the relationships with serious misconduct (IRR range = 1.13-1.40). Accidental injury had the largest association with nonserious misconduct (IRR = 1.40, p < .001), indicating that inmates who were accidentally injured were 40% more likely to engage in nonserious misconduct. Having a physical disability was not significantly associated with nonserious misconduct. However, having a chronic illness decreased the likelihood of nonserious misconduct by about 9% (p < .001).
In supplementary logistic regression models, almost all acute conditions were positively associated with misconduct. The one exception was having had surgery, which was not associated with engaging in serious misconduct. Having a physical disability decreased the risk of serious misconduct by about 19%, whereas having a chronic condition decreased the risk of general and nonserious misconduct by about 14% each. The full supplementary logistic analyses are available upon request. The following section discusses these results as well as their implication for improving inmate health and decreasing the risk of misconduct in prisons.
Discussion and Conclusion
An inmate’s likelihood of engaging in misconduct while incarcerated is influenced by a variety of factors including age, drug use, prior criminal justice involvement, gang membership, prison program involvement, and mental illness. To date, no research has examined the relationship between physical health conditions and prison misconduct despite the well-documented association between mental health illness and prison behavior. The purpose of this study was to assess the relationship between adverse physical health ailments and inmate misconduct. We found that acute health conditions significantly increase one’s risk for general, serious, and nonserious misconduct. In addition, physical disability is associated with a decrease in the likelihood of general and serious misconduct, but not nonserious misconduct. Finally, chronic illness decreases the likelihood of all forms of misconduct.
Surgery, accidental injury, dental problems, and illness all significantly increase inmate misconduct. These results indicate that acute health conditions may be viewed as an unjust and particularly salient strain that inmates must endure in an already strained environment. Following a GST approach, inmates who experience an acute physical health issue may not be accustomed to being sick and, as a result, may lack the appropriate coping mechanisms to handle these health ailments or the negative emotions that accompany these conditions. This is supported by the results indicating that experiencing an accidental injury or having an illness has the most robust influence on all forms of misconduct. More so than surgery or having dental problems, the experience of an accidental injury or illness may be particularly unexpected, leaving the inmate to cope with pain and symptoms that they may be unaccustomed to. Illness is especially consequential for serious misconduct (IRR = 1.54, p < .001), which includes alcohol and drug offenses. Inmates may use illicit substances that result in a misconduct citation, especially if they are dealing with pain or illness that makes the experience of being a prisoner more difficult than it already is. Ultimately, acute physical symptoms appear to increase the likelihood that an inmate will engage in various types of misconduct with particular concerns for injury and illness.
Having a physical disability decreases the likelihood of both general and serious misconduct. However, no relationship was found for nonserious misconduct. Although the level of significance was lower for the influence of physical disability, the results suggest that having a physical impairment may prevent inmates from engaging in misconduct. For serious misconduct, the impairment may be debilitating enough that they are unable to do things such as attempt an escape or take the necessary steps to possess an unauthorized substance or weapon. The lack of a relationship found for nonserious misconduct suggests that having a physical disability may act somewhat as a strain toward certain forms of misconduct in prison, but that the individual may be more adept at coping with this strain. Thus, the disability is not incapacitating enough to decrease misconduct involvement or enough of a strain to lead to coping through misconduct. To explore this further, future research should focus not only on whether an individual experiences a physical disability but also the duration and magnitude of that disability to indicate how debilitating it might be and how much of a strain it poses on the inmate.
As predicted, suffering from a chronic illness was significantly associated with a decreased likelihood of all forms of inmate misconduct. This finding suggests that when a condition becomes prolonged and sustained, misconduct is not a viable response to the strain experienced and the risk of misconduct decreases. This leads to three separate, yet related, conclusions. First, there might be a threshold to the relationship between strain and misconduct. When the strain associated with chronic health conditions or a physical disability becomes too physically debilitating, misconduct becomes unlikely. Future longitudinal research should more fully examine this potential threshold. In particular, research should determine where that threshold lies with regard to illness and/or disability as a strain. When does a chronic illness or physical disability become too much that misconduct becomes less likely? Second, those with chronic health conditions and functional disabilities may simply lack the physical ability to engage in misconduct. Although certain disabilities such as hearing and visual impairment may still enable an inmate to engage in misconduct, those using an aid to walk around or complete basic daily tasks, or activities of daily living (ADLs), may not have the capacity to do so. Third, chronic illness and functional disabilities tend to be associated with age. Research consistently finds that older inmates experience higher levels of chronic illness and functional decline in comparison with younger inmates and the general population (Aday & Krabill, 2012; Binswanger, Krueger, & Steiner, 2009; Harzke et al., 2010; Williams et al., 2010). As a result, many inmates with chronic illnesses or physical disabilities are simply older and their lack of misconduct might simply reflect their “aging out” of crime and misbehavior. Although age is controlled across all models, those inmates who are elderly and experiencing multiple forms of illness may be too sick or physically limited to engage in misconduct or have less interest in possessing illicit substances or getting into altercations with others.
Despite this possibility, research on the aging prison population suggests that when physical health status worsens, older prisoners experience higher distress levels (Baidawi & Trotter, 2016; Burling, 1999; Murdoch, Morris, & Holmes, 2008). In particular, this distress may result from difficulties they encounter with prison-specific ADLs, such as climbing on and off top bunks and hearing guards’ orders. Although chronic health ailments and physical disabilities may present significant challenges for older inmates, which lead to distress, the data in this study suggest that both physical disability and chronic conditions decreases the risk of misconduct. It is possible that inmates with a chronic health condition or physical disability are in fact engaging in misconduct such as disobeying orders—potentially because they cannot hear the orders given to them. However, correctional officers may be sympathetic and simply less likely to write a disciplinary report for these inmates because they may view their behavior as unintentional. Future research should examine the discretion used by correctional officers in writing disciplinary reports, especially for those inmates suffering from chronic illnesses and physical disabilities.
Additional research suggests that inmates suffering from co-occurring disorders are more likely to engage in misconduct as compared with those with either no diagnosis or a singular diagnosis (Houser et al., 2012; Wood, 2012; Wood & Buttaro, 2013). However, co-occurring disorders as they relate to issues of misconduct are usually defined as having a mental health issue in combination with substance use disorders. Future research should investigate the co-occurring relationship between different forms of physical health conditions, as well as the influence of physical health in conjunction with mental health issues and/or substance use disorder on prison misconduct.
Limitations and Policy Implications
There are several limitations associated with this study. First, the data used are from 2004. Although dated, this is the most current iteration of the Survey of Inmates in State and Federal Correctional Facilities data set and offers the most detailed data available regarding inmate health and misconduct. Once a more recent data set is released, it is imperative to reassess the relationship between inmate physical health and misconduct, especially as the number of aging inmates continues to rise, given that this group is more likely to experience chronic health ailments and physical functional disabilities than their younger counterparts are.
Second, the data used for this study are cross sectional and we cannot definitively determine the causal ordering of the variables of interest. Although acute physical symptoms appear to increase the likelihood that an inmate will engage in various types of misconduct, it is important to recognize potential causality issues. In particular, acute physical health issues might result from involvement in misconduct. In other words, it is possible that engagement in misconduct leads to acute health issues such as illness or accidental injury. Engagement in particular forms of misconduct such as drug use while incarcerated may also lead to chronic health conditions. As a result, we cannot determine whether the strain of physical illness leads to misconduct or whether misconduct contributes to physical health issues. Research by Piquero and colleagues support this limitation. Using longitudinal data from the Cambridge Study in Delinquent Behavior, Piquero et al. (2011) found that, by the age of 48 years, high-rate chronic offenders had the highest risk of poor health outcomes, particularly disabilities and hospitalizations. Similarly, Piquero et al. (2007) also demonstrated that, by the ages 27 to 33 years, life course–persistent offenders were more likely to experience adverse physical health conditions than adolescence-limited offenders and nonoffenders. This suggests, then, that adverse health conditions—including disabilities and acute ailments—might result from misconduct. It is possible that causality runs in both directions and, therefore, future studies should use longitudinal data to more clearly address this issue of causal order across different forms of misconduct and physical health conditions.
Finally, our variables were limited in certain ways. First, the inter-item reliability coefficients for the dependent misconduct measures were slightly lower than the commonly accepted cutoff point for alpha coefficients (Schmitt, 1996). This was likely due to generally low correlations among misconduct incidents given that most inmates surveyed reported either zero incidents or only a single misconduct incident. Future research on inmate misconduct should strive to develop internally reliable scales that account for as many types of misconduct as possible to improve reliability among individual items. Second, we did not have access to the full array of GST variables used in other studies. More specifically, GST argues that strain leads to negative emotions such as anger and/or depression. Although we did have measures of anger, the cross-sectional data did not provide us with any other measures of additional negative emotions such as depression, irritability, or frustration. Thus, we include anger as a control variable to address heterogeneity across the sample but we do not include it in a mediation analysis test given the cross-sectional nature of the data. As a result, we are not able to properly determine whether physical health illness affects a range of negative emotions and whether those negative emotions increase the likelihood of inmate misconduct. Future research should more fully account for the role of negative affect in the relationship between physical health illness and inmate misconduct using these variables within available longitudinal data sets. Third, despite prior research that indicates institutional-level factors (e.g., security classification, prison population density) influence inmate misconduct, this data set does not include institutional-level variables. Future studies should include these variables in their analyses to determine whether similar findings result even after controlling for institutional-level factors.
The final limitation is the lack of measures of population heterogeneity such as self-control or IQ. Without these measures, it could be the case that inmates low in self-control are simply more likely to have both adverse physical health problems and engage in misconduct. The relationship we found might be spurious and caused by some underlying trait that encourages both poor health and misconduct. Research by Pratt, Barnes, Cullen, and Turanovic (2016) highlights the idea that some people, including inmates, fail at everything. They find that being arrested (i.e., failing at crime) significantly predicts other life failures such as unemployment, divorce, and dropping out of high school. In other words, “people who fail at crime also fail at a lot of other things” (Pratt et al., 2016, p. 847). In line with this study, then, inmates who are written up for misconduct in prison (i.e., fail at crime while incarcerated) may simply also fail at maintaining good health. It is imperative that future research includes important measures of population heterogeneity to determine whether a true relationship exists between poor health and misconduct.
Despite these limitations, there are significant resultant policy implications, particularly with regard to correctional health care. As the findings indicate, inmates with fewer acute health conditions are significantly less likely to engage in misconduct. As a result, improving health care within prisons may decrease misconduct. Beginning in the early 1970s, the National Commission on Correctional Health Care (NCCHC; 2014) was tasked with detailing accreditation standards for improving the quality of health care provided in jails and prisons (Steinwald, Alevizos, & Aherne, 1973). Despite the 1976 Supreme Court decision in Estelle v. Gamble that ruled prisoners are entitled to appropriate health care, adhering to the NCCHC’s accreditation standards remains voluntary for correctional facilities. Recent research argues for the need for higher standards in correctional health care (Rich, Allen, & Williams, 2015). By increasing health care standards in prison and potentially making accreditation mandatory for correctional facilities, we will not only improve inmate health but also the quality of life for incarcerated individuals.
In addition, in at least 35 states, inmates are required to pay a copayment to receive medical services (Brennan Center for Justice, New York University School of Law, n.d.). Not all illnesses, however, require a copayment. In some states, for example, copays are required for nonemergency sick calls, whereas chronic disease treatment and long-term care are exempt (Brennan Center for Justice, New York University School of Law, n.d.). Requiring inmates to pay a minimal fee to see the doctor is designed to reduce the number of unnecessary visits to an already strained medical department (Brennan Center for Justice, New York University School of Law, n.d.; Vogt, 2002). Although, in theory, the idea of copayments seems beneficial to the efficiency of correctional facilities, it is possible that inmates who experience acute illnesses will avoid going to the doctor because either they do not want to pay the copay or they simply lack the funds to pay the copayment. When an inmate makes a nominal amount of money for a prison job assignment (if he or she is working at all), he or she may be extremely reluctant to use that money for a doctor’s visit for a seemingly minor illness. However, the research presented here highlights the potential consequences of avoiding medical services for acute illness—an increase in the likelihood of misconduct. If inmates are able to receive quality care for acute conditions in a timely manner for little to no cost, then they may view their illness as less unjust, less salient, and, may therefore, be less likely to misbehave while in prison. We argue that improving correctional health care is necessary, not only because inmates get sick but also because improved health care, in terms of availability of treatment, time to receive services, and cost of visits, may enhance the overall prison environment by reducing the amount of violence and misconduct behind bars. We caution scholars, however, from interpreting the findings here in such a way that results in more restrictive confinement conditions for individuals experiencing acute physical health ailments. As previously mentioned, due to the data being cross sectional, we cannot definitively claim that acute health conditions lead to an increased likelihood of misconduct. It is possible that misconduct leads to acute physical health ailments such as surgery, accidental injury, or dental problems.
Conclusion
Although research has consistently found that mental health affects inmate behavior, no research to date has examined the influence of physical health on inmate conduct, despite the fact that physical illness is a significant stressor experienced by already strained inmates. As our study indicates, acute health conditions significantly increase the likelihood of inmate misconduct, whereas both physical disability and chronic illness decrease the risk of misconduct. By addressing inmates’ acute illnesses in a more timely and cost-efficient manner, it may be possible to concurrently improve inmate health and decrease misconduct to enhance the lives of those in prison.