Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Feb 28.
Published in final edited form as: Popul Res Policy Rev. 2018 Oct 16;38(1):73–93. doi: 10.1007/s11113-018-9496-y

Education and Health Conditions among the Currently Incarcerated and the Non-Incarcerated Populations

Connor M Sheehan 1,a
PMCID: PMC9974178  NIHMSID: NIHMS1830230  PMID: 36860891

Abstract

Previous research has found a strong link between educational attainment and health, where the highly educated live longer and healthier lives than those with lower levels of education. Because such research has relied on samples of the non-institutionalized population, previous research has not explored the association between education and specific chronic and infectious health conditions among the currently incarcerated. Analyzing the relationship between education and health conditions among the incarcerated, whom tend to be less healthy and for whom many of the intermediate mechanisms between education and health are held relatively constant in prison, may yield new insights. Using the 2002–2004 National Health Interview Study (N=74,881), the 2004 Survey of Inmates in State and Federal Correctional Facilities (N=17,553), and interaction terms from logistic regression models, I compared the strength of the association between educational attainment and the presence of chronic and infectious health conditions among the incarcerated and non-incarcerated populations. These models indicated generally stronger negative associations between educational attainment and chronic conditions among the non-incarcerated, while the negative relationship between education and hepatitis was stronger for the incarcerated. These results suggest that while education may play a lesser role for chronic conditions for the incarcerated, it can still important for avoiding risky health behaviors.

Keywords: Incarceration, Education, Health, Hepatitis

Introduction

The currently incarcerated (hereafter incarcerated) population of the United States has grown considerably in the past century. In 1925, the incarceration rate was 79 per 100,000; by 2003, it had grown more than six-fold to almost 500 per 100,000 (Bureau of Justice Statistics 2003; Carson 2014). While recent estimates suggest the growth is subsiding, still in 2013, more than one out of every 100 U.S. adults were incarcerated (Carson 2014), and about 3 percent were under some form of correctional supervision (Pettit 2012; Wakefield and Wildeman 2011). Those who are incarcerated are not a random subset of the adult population, instead they overwhelmingly come from disadvantaged socioeconomic backgrounds (Dolan and Carr 2015), are racial or ethnic minorities (Pettit 2012), have low levels of formal education (Arum and Beattie 1999), and higher levels of chronic and infectious health conditions (Binswanger et al. 2009). As such, mass incarceration reproduces racial inequality, shapes political outcomes, the labor market, and population-health (Kim 2015; Manza and Uggen 2008; Pattillo et al. 2004; Pettit 2012; Roberts 2004).

Mass incarceration also complicates estimates of social phenomenon (Nowotny et al. 2017; Pettit 2012). Specifically, the exclusion of the incarcerated population from many health surveys likely influences population-level estimations of the well-documented relationship between education and a number of health outcomes. Previous research has found that those with higher levels of educational attainment report favorable health (Brown et al. 2014; Liu and Hummer 2008), fewer adverse health conditions (Garrison et al. 1993; Hayward et al. 2000; Mezuk et al. 2008; Siegel et al. 2012), and ultimately live longer (Everett et al. 2013) and proportionally healthier lives (Montez and Hayward 2014) compared to those with lower levels of education (for overviews see: Baker et al. 2011; Hummer and Lariscy 2011). Such findings, however, typically, although not exclusively (e.g. Miech et al. 2011), are based on surveys, such as the National Health Interview Survey or the National Health and Nutrition Examination Survey. These surveys use samples that are representative of the civilian non-institutionalized U.S. population. By design, such surveys exclude Americans who live in an institution (e.g., anyone who is currently in jail or prison). Hence, results from previous research on the relationship between education and health may not apply to the incarcerated population.

The lack of generalizability is not only important given the size of the incarcerated population but also because the incarcerated are not a missing completely at random subset of the population (Alison 2001). As mentioned, the incarcerated have systematically lower levels of educational attainment (Harlow 2003) and higher levels of adverse health conditions (Binswanger et al. 2009) than the rest of the population. This means that excluding the incarcerated from population surveys of health likely provides a sampling frame that is healthier and more highly educated than that in the total population (incarcerated or not) (Binswanger et al. 2009; Maruschak et al. 2015; Nowotny et al. 2016, 2017; Pettit 2012). Indeed, previous research has shown that excluding those who live in group quarters leads to significant underestimates in the prevalence of disability among working-age males and that males with less than a high school education also have disproportionate levels of residence in group quarters—likely prisons (Stapleton et al. 2012). Given the lower levels of incarceration among women, the influence of excluding those who live in group quarters was smaller for women than men (Stapleton et al. 2012). Of course, the relationship between education and health may also differ in highly regulated and formally administered institutions such as prisons that also draw from select (and unhealthy) groups compared to the non-incarcerated population (Nowotny et al. 2017).

Using data from the 2003–2004 Survey of Inmates in State and Correctional Facilities and the 2002–2004 National Health Interview Survey, I descriptively test the relationship between education and chronic and infectious health conditions among the incarcerated population as well as how this relationship may differ relative to that for the non-incarcerated. I descriptively document the prevalence five chronic conditions (hypertension, diabetes, asthma, arthritis, and cancer) and one infectious disease (hepatitis). These conditions were selected due to their overall importance for population health in the non-incarcerated population (Murphy et al. 2016), the incarcerated population (Binswanger et al. 2009; Center of Disease Control 2013; Massoglia and Pridemore 2015; Noonan and Ginder 2013), and due to consistency between datasets. To analyze the relationship between education and the prevalence of these conditions, I first document major demographic differences between the incarcerated and non-incarcerated samples. I then calculate age-standardized prevalence rates for the select health conditions by gender, level of education, and incarceration status. I next fit multivariate models that predict these conditions as a function of educational attainment and important confounders. I also conduct analyses documenting the sensitivity of the results when accounting for factors such as the (lack of) availability of health examinations in prison, the duration of incarceration, the functional form of educational attainment, Inverse Probability Weighted Regression Adjustment (IPWRA to adjust for observable confounders related to selection into prison; Thoemmes and Ong 2016), and document how including the incarcerated influences national level estimates of educational inequality in health.

Descriptively documenting the relationship between education and health conditions among the incarcerated and how this relationship may differ for the non-incarcerated is important for several reasons. First, it can help identify specific segments of the population who face an elevated risk of particular health conditions. This is especially important as the incarcerated population ages and faces greater levels of chronic diseases (Porter et al. 2016; Rikard and Rosenberg 2007). Second, this analysis tests whether the well-documented relationship between education and health in the non-incarcerated population exists among the incarcerated, who are largely excluded from health surveys. Third, this research has important theoretical implications for understanding the relationship between education and health. In prison, many of the “intermediate mechanisms” (Link and Phelan 1995), such as diet, residential location, income, and occupational exposure, through which education may influence health conditions are held relatively constant for persons with different levels of education. Of course, Americans are not randomly assigned to be incarcerated (and those who are incarcerated are likely less healthy than those who are not), which limits the ability to claim that incarceration causally alters the relationship between education and health. Nevertheless, this is a large and non-random segment of the population for which the relationship between education and specific health conditions has heretofore not been analyzed and as such, this research could offer new insights on how (and for whom) education may or may not be associated with specific health conditions.

Background and Hypotheses

The equalization of living conditions and health risks among the incarcerated could lead to smaller educational gradients in health for chronic conditions, which are largely due to the accumulation of negative exposures and risk factors (Bauer et al. 2014; Omran 1971). Prison is one of the most fitting examples of a strict Total Institution (Goffman 1961). Goffman originally developed the concept of a Total Institution to describe settings such as prisons or the military and the highly rigid and supervised environment among those who dwell within them. In Total Institutions aspects of individuality such as choice, freedom, status, and presumably level of education, are minimized in favor of conformance and obedience (Sheehan et al. 2015).

Prisoners have extremely regimented behaviors, are served similar diets, all share the same medical care and doctors, live in the same circumstances and areas, and have certain risky health behaviors formally restricted. Thus, prisons exert direct social control (Umberson 1987, 1992) over behavior generally and health behaviors specifically. In contrast, in the general population, behavior, diet, medical care, and area of residence all serve to help the highly educated gradually garner health advantages relative to those with lower levels of education (Mirowsky and Ross 2005, 2015). Additionally, among the non-incarcerated, those with higher levels of education are able to spend their income to benefit their health (Marmot 2002), whereas in prison the opportunities to spend, especially to benefit health, are almost entirely limited. The lack of variation in ability to differentiate health through gradual decisions is especially important for chronic or man-made health conditions that largely result from long-term life-long health decisions, exposures, and accumulation of risks (Olshansky et al. 2012; Omran 1971). As adverse health behaviors in prison are restricted, intervening mechanisms and circumstances in prison could lead to smaller educational gradients in chronic health outcomes than in the general public where the highly educated are able to reap health benefits while the less educated face substantially worse health risks.

There are other reasons to anticipate equalized risk of chronic conditions by level of education among prisoners. Being incarcerated is a stressful experience adversely affecting overall health (Bottoms 1999; Massoglia and Pridemore 2015; Spaulding et al. 2011). Nevertheless, the relative difference in living conditions caused by incarceration may be less substantial for those with lower levels of education. For example, those who are incarcerated are provided constitutionally mandated healthcare, something that the non-incarcerated with low levels of education may not have access (Schiller et al. 2005). This is not to say that incarceration benefits overall health, as other work has consistently shown that the incarcerated report more chronic conditions (Binswanger et al. 2009) and have higher levels of infectious diseases (Baussano et al. 2010). Rather, the outside environment for those with low levels of education may be so pernicious that prison could protect them from specific health risks and causes of death (Rosen et al. 2008; Schnittker and John 2007; Spaulding et al. 2011). Indeed, a longitudinal analysis of a cohort of prisoners in Georgia found the prisoners faced a lower risk of death during incarceration than after release (Spaulding et al. 2011), when they were more likely to die from drug overdose, HIV/AIDs and homicide, causes of death that were curtailed in prison. Similarly, research has also found that young black men in prison have lower levels of death than those who are not incarcerated (Rosen et al. 2008). Analogous processes could conceivably operate for the emergence of chronic diseases, whereby educational inequality in chronic conditions is minimized. The highly educated who are incarcerated likely also have lower levels of self-control (potentially leading to incarceration), which could also undermine their health advantage. Prisons also draw or select from systematically unhealthy populations—populations so unhealthy that educational gradients in health may not be apparent. Thus, all of these factors culminate in the expectation of smaller educational health disparities in chronic conditions among the incarcerated than the non-incarcerated, leading to Hypothesis #1.

Hypothesis #1: There is smaller educational inequality in chronic health conditions among the incarcerated than among the non-incarcerated.

Although the above hypothesis suggests a shallower educational gradient in health among the incarcerated for chronic health conditions, educational disparities may still be present for other reasons. Persons who achieve high levels of educational attainment are generally from socioeconomically advantaged backgrounds (Dubow et al. 2009). These backgrounds may limit exposures to harmful conditions in childhood, which protect against future risk of chronic conditions (Friedman et al. 2015; Montez and Hayward 2014). In addition, the benefits of education are also long-lasting; for example, research has indicated that cognitively enriching activities in early schooling can benefit neurological functioning, which may lead to greater risk-assessment and decision-making skills throughout adulthood (Baker et al. 2011). These skills may even persist in highly-regulated prison environments to ultimately prevent chronic conditions. Those with greater levels of educational attainment may also practice beneficial health behaviors that carry over to prison and these practices may still slowly differentiate highly educated prisoners from their less educated peers in terms of risk of chronic conditions. It is also worth noting that not all chronic conditions are negatively associated with education. Certain forms of cancer such as prostate cancer for men (Pudrovska and Anishkin 2015) and breast cancer for women (Heck and Pamuk 1997) are positively associated with educational attainment. Of course, the positive relationship between education and these forms of cancer could also be altered among the incarcerated.

Recent research provides even further reasons to anticipate educational gradients in chronic conditions among the incarcerated population. Nowotny et al. (2016), used the Survey of Inmates in State Correctional Facilities, and found that education prior to incarceration was associated with a lower number of adverse health conditions. They also found that gaining education (such as a GED) in prison was associated with lower overall levels of adverse health conditions. I build on Nowotony et al. (2016) in a few ways. First, rather than predicting a sum of the total health conditions (e.g., presence of diabetes + presence of cancer…+ presence of hepatitis), I look at specific health conditions (e.g. diabetes). While summing conditions is a good way to measure overall health, especially at the population level, summing conditions can mask different etiologies of different conditions. Etiologies of specific conditions are important and can reflect unique differences in physiological health; indeed, this is one reason why Nowotny et al. (2016), analyzed one specific health condition: hypertension. Second, in addition to the state inmates whom Nowotny et al. (2016) analyzed, I include the incarcerated from federal prisons. Third, I compare the educational gradients of these conditions among both the incarcerated and the non-incarcerated. Finally, I followed the procedures of Nowotny et al. (2017) and updated national-level estimates for health conditions to include the incarcerated and reassess population-level educational gradients in health conditions. Overall the reasons specified above and the evidence documented by Nowotny et al. (2016) suggests that there could be significant differences in the prevalence of chronic conditions by level of educational attainment, which leads to Hypothesis #2.

Hypothesis #2: There is observable educational inequality in health conditions among the incarcerated, with the highly educated having lower risk of chronic health conditions.

For infectious conditions such as hepatitis, it is plausible that the educational gradient might be significantly greater among the incarcerated than among the non-incarcerated population. Hepatitis rates are considerably higher among the incarcerated population (Center of Disease Control 2013) potentially due to higher levels of stress (Massoglia 2008). In circumstances of elevated risks, the highly educated, with better decision-making skills, may be better able these risks. Better decision-making skills may be especially relevant for avoiding infectious diseases such as hepatitis-C, which is spread through blood transfers and risky behaviors such as unprotected sex, intravenous-drug use, and unsanitized tattoos. Previous research has shown a strong negative association between education and risky behaviors which may adversely affect health (Link 2008). In other words, the high prevalence and risk of hepatitis-C among the incarcerated could heighten the importance of education and accompanying decision making skills, in turn increasing the educational gradient for this disease. This leads to Hypothesis #3.

Hypothesis #3: There is larger educational inequality in some health conditions, particularly Hepatitis, among the incarcerated than the non-incarcerated.

Data and Methods

Data

The data for this investigation came from two nationally representative sources: one of the incarcerated population and one of the non-institutionalized population. For the incarcerated, I used data from the Survey of Inmates in State and Federal Correctional Facilities (SISFCF), which includes 14,499 prisoners from 287 state prisons and 2,984 prisoners from 39 federal prisons. The SISFCF was conducted from 2003–2004. Unfortunately, there was not a more recently released dataset which is generalizable to the entire incarcerated population that also closely aligns to other health surveys.

For the non-incarcerated population, I used data from the 2002–2004 National Health Interview Survey (NHIS). The NHIS is a large nationally representative survey of non-institutionalized Americans conducted annually. I used the publicly available Integrated Health Interview Series version of the NHIS (Minnesota Population Center and State Health Access Data Assistance Center 2016). Previous research regarding the association between education and health also used the NHIS (Masters 2012; Montez and Zajacova 2013; Sheehan et al. 2018). I chose these datasets for consistency with previous research (Binswanger et al. 2009), for similarity in how they ask questions, and for their large samples and representative sampling frames. I also used the 2002–2004 NHIS surveys to be concurrent with SISFCF sampling (the SISFCF was mainly collected in 2003) timeframe and to be consistent with previous research (Binswanger et al. 2009). I excluded data from the Survey of Inmates in Local Jails (SILJ) because its key questions did not align as well with the other data sources and because jail inmates are typically incarcerated for shorter periods of time than those in prison. Nevertheless, in ancillary analysis I found that including SILJ data did not substantially alter the substantive results on educational gradients in chronic conditions.

Consistent with other work research, I analyzed data for all adults 18 to 65 years of age with valid responses for all questions on health conditions (Nowotny et al. 2017). This age range was selected to be consistent with previous research (Binswanger et al. 2009) and due to smaller cell sizes among the older (65+) incarcerated. Only 2 percent of respondents were dropped because of missing values for any of the health condition questions. While the incarcerated were more likely to be dropped because of a missing response, ancillary analyses showed no significant differences in non-response for the health conditions among the incarcerated by level of educational attainment. Other sources of missing data were trivial; for example, only 0.83% of the combined sample had missing values for educational attainment. I handled the small amount of remaining missing data with Stata’s multiple imputation suite. Results were similar when other techniques, such as listwise deletion, were used to handle the missing data. Altogether, the analytic sample had 92,434 respondents, 17,553 incarcerated (SISFCF) and 74,881 non-incarcerated (NHIS).

Measures

Variables in both data sets were coded identically and the datasets were combined (or appended). In both datasets, I examined prevalence of hypertension, diabetes, asthma, arthritis, cancer, and hepatitis. I coded these conditions “1” if the respondent reported having a doctor diagnosis for the condition and “0” if otherwise. Extensive information regarding the consistency of question wording between surveys has been provided elsewhere (Binswanger et al. 2009).

I coded incarceration status as non-incarcerated (0, NHIS) or incarcerated (1, SISFCF). My key predictor variable was educational attainment. Due to the relatively small number of incarcerated individuals with more than a high school education, I used only three education categories: less than high school education (0; reference), high school education or equivalent (1), and greater than high school (2). To examine the sensitivity of the coding of educational attainment I further investigated three additional specifications of educational attainment, such as other categorical specifications and a continuous specification. I elaborate further in the sensitivity analysis section below, and provide more detailed information in the Online Supplement.

For the multivariate analysis, I included control variables associated with health conditions that may vary significantly between the incarcerated and non-incarcerated populations. Critically, wording for these control variables were extremely similar between surveys, allowing me to code them identically for the incarcerated and non-incarcerated samples (Binswanger et al. 2009). I accounted for race/ethnic status by considering respondents as non-Hispanic white (0, reference), non-Hispanic black (1), Hispanic (2), and non-Hispanic other (3). To evaluate if the association between education and health varied significantly by race among the incarcerated, I also interacted education by race in predicting each of the health conditions. I found no systematic differences in the association between education and the health conditions by race, suggesting the association between education and the health conditions did not vary by race/ethnicity among the incarcerated population. Given these results, race/ethnicity was included as a control. I measured age continuously in years at the time of survey, with values ranging from 18 to 65. I also included a categorical measure of marital status with married (0, reference), widowed (1), divorced or separated (2), and never married (3). Finally, I accounted for smoking status, differentiating never smokers (0, reference), former smokers (1), and current smokers (2).

Methods

To compare the extent of educational disparities in health conditions among the incarcerated to non-incarcerated, I employed the following protocol. First, I documented differences between the incarcerated and the non-incarcerated in the prevalence of health conditions and distributions of the control variables. Given that the incarcerated significantly were younger, I next calculated age-standardized prevalence levels of health conditions by gender, incarceration status, and educational attainment. I then estimated a series of logistic regression models (Hoffmann 2004) that predicted the presence of each condition, stratified by incarceration status and gender with controls for age, race, marital status, and smoking history. I also tested for interaction effects (i.e. incarcerated X educational attainment) to gauge whether the association between educational attainment and each health conditions varied by incarceration status.

Results

Table 1 displays the differences in the health conditions and demographic characteristics between the incarcerated (SISFCF) and non-incarcerated (NHIS) populations. The results were consistent with previous research that has shown higher levels of chronic and infectious among the incarcerated than among the non-incarcerated (Binswanger et al. 2009; Maruschak et al. 2015; Nowotny et al. 2016; Pettit 2012). The differences between the incarcerated and the non-incarcerated were statistically significant for women in all six conditions examined and for men in all conditions examined except hypertension and arthritis. These differences are especially striking given the significantly younger average age of the incarcerated.

Table 1:

Basic Health and Demographic Information by Incarceration Status and Gender, Americans Aged 18–65, 2002–2004.

General Population Currently Incarcerated
NHIS SISFCF
Males Females Males Females
Health Conditions
% with Hypertension 20.4% 20.0% 20.3% 24.0%
% with Diabetes 6.1% 6.1% 4.5% 7.4%
% with Asthma 8.7% 12.1% 13.5% 22.5%
% with Arthritis 13.6% 19.3% 14.3% 24.5%
% with Cancer 3.2% 5.3% 1.6% 10.1%
% with Hepatitis 3.5% 3.3% 8.9% 14.2%
Demographic Characteristics
Age (average) 40.5 40.6 35.2 35.7
% Male 45.1% 93.2%
Education
% Less than High School 17.2% 16.5% 65.5% 59.5%
% High School 28.3% 27.5% 19.9% 20.3%
% Greater than High School 54.5% 56.0% 14.6% 20.2%
Race
% Non-Hispanic White 64.4% 61.2% 34.1% 44.3%
% Non-Hispanic Black 12.1% 15.9% 41.9% 33.5%
% Hispanic 19.2% 19.1% 18.4% 14.1%
% Other 4.2% 3.8% 3.0% 2.7%
Marital Status
% Married 50.7% 48.2% 16.6% 17.7%
% Widowed 1.2% 4.4% 1.3% 3.9%
% Divorced/Separated 16.6% 21.2% 23.3% 30.4%
% Never Married 31.5% 26.2% 56.3% 42.5%
Smoking Status
% Never Smoker 51.6% 62.0% 26.0% 21.3%
% Former Smoker 20.8% 15.9% 35.7% 28.7%
% Current Smoker 27.6% 22.0% 38.3% 45.2%
N 33,808 41,073 13,802 3,751
Open in a new tab

Sources: National Health Interview Survey (NHIS), and Survey of Inmates in State and Federal Correctional Facilities (SISFCF).

Bold indicates significant difference (p< .05) as measured by a t-test, between incarcerated and non-incarcerated within Gender.

Italics indicates Significant difference (p< .05) as measured by a Chi-Squared, between incarcerated and non-incarcerated within Gender.

There were also statistically significant differences in educational attainment between the incarcerated compared to the non-incarcerated. Most of the incarcerated population, both male and female, did not graduate high school, while most of the non-incarcerated population, both male and female, had at least some education beyond high school. Specifically, 65.5% of incarcerated men and 59.5% of incarcerated women had less than a high school education. In contrast, only 17.2% of males and 16.5% of females in the non-incarcerated population had less than a high school education. At the upper end of the educational distribution, only 14.6% of incarcerated males and 20.2% of incarcerated females had greater than a high school education, while 54.5% of males and 56.0% of women in the non-incarcerated population had greater than a high school education. Chi-squared tests indicated the differences in the distribution of educational attainment between the incarcerated and non-incarcerated populations were statistically significant.

Given the significant differences in age distribution for the incarcerated and non-incarcerated, in Table 2 I present the age-standardized prevalence rates (standardized to the non-incarcerated age distribution) and relative risk ratios of the health conditions by educational attainment for the incarcerated and non-incarcerated by gender. I also conducted analyses stratified by age group, with results in presented Supplementary Table 1, the results were similar to the overall analysis. For the non-incarcerated population, hypertension and diabetes generally showed the greatest educational gradients, with prevalence of the conditions decreasing among the non-incarcerated population as education increased. I did not find such relationships for asthma, arthritis, and cancer. This was not surprising, given, as mentioned, highly educated Americans have higher rates for some forms of cancer, such as breast cancer for women (Heck and Pamuk 1997) and prostate cancer for men (Pudrovska and Anishkin 2015). The highly educated also have higher survival rates for cancer (Albano et al. 2007); through selective mortality, this could lead to higher rates of reports of “ever” being diagnosed with cancer. The higher rates of asthma for the highly educated are likely due to the greater access to diagnosis.

Table 2:

Age Standardized Prevalence Rates and Risk Ratios of Reported Health Condition by Gender, Level of Education, and Incarceration Status, Americans Aged 18–65, 2002–2004.

Hypertension Diabetes Astdma Arthritis Cancer Hepatitis
GP CI GP CI GP CI GP CI GP CI GP CI
Women
Less than High School 21.8% 29.5% 7.4% 7.4% 12.3% 22.4% 15.7% 29.0% 3.3% 10.3% 3.6% 18.9%
High School 17.1% 24.6% 4.8% 6.4% 11.2% 18.6% 15.4% 24.1% 4.0% 7.9% 2.5% 10.8%
Greater than High School 13.8% 27.5% 3.6% 9.0% 12.5% 22.4% 14.3% 27.6% 4.5% 11.4% 3.1% 9.3%
Men
Less than High School 17.1% 23.1% 5.2% 4.7% 7.7% 12.7% 9.7% 15.7% 1.2% 1.3% 3.3% 10.6%
High School 17.6% 22.9% 4.1% 4.1% 7.9% 11.5% 11.1% 15.5% 1.5% 1.4% 2.9% 8.6%
Greater than High School 15.4% 23.6% 3.4% 4.8% 9.4% 14.4% 9.5% 17.7% 1.8% 2.4% 3.2% 7.8%
Relative Risk Ratios, Relative to Less than High School
Hypertension Diabetes Asthma Arthritis Cancer Hepatitis
GP CI GP CI GP CI GP CI GP CI GP CI
Women
Less than High School 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
High School 0.78 0.83 0.64 0.87 0.91 0.83 0.98 0.83 1.21 0.77 0.69 0.57
Greater than High School 0.63 0.93 0.48 1.22 1.02 1.00 0.91 0.95 1.36 1.11 0.86 0.49
Men
Less than High School 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
High School 1.03 0.99 0.80 0.88 1.03 0.91 1.14 0.99 1.25 1.07 0.88 0.81
Greater than High School 0.90 1.02 0.65 1.02 1.22 1.13 0.98 1.13 1.51 1.90 0.97 0.73
Open in a new tab

GP = General Population, Not-currently incarcerated (Data source: National Health Interview Survey). N = 74,881.

CI = Currently Incarcerated (Data Source: Survey of Inmates in State and Federal Correctional Facilities). N =17,553.

Note: Standardized to Non-Incarcerated Age Distribution.

Among the incarcerated population, there was little evidence of any negative educational gradients for the chronic conditions examined. This is consistent with Hypothesis #1, that educational inequality in chronic conditions is less for the incarcerated than the non-incarcerated. In fact, in most cases for the incarcerated, the highly educated reported higher levels of chronic conditions. There was, however, a discernable negative educational gradient in hepatitis among the incarcerated for both males and females, with little evidence of such a gradient among the non-incarcerated. This is consistent with Hypothesis #3, that educational disparities in hepatitis are larger among the incarcerated population.

Table 3 presents results that illustrate how the association between educational attainment and the health conditions differs by incarceration status in a logistic regression multivariable framework. The models controlled for age, race/ethnicity, marital status, and smoking. I also fit interaction models to test if the effect of educational attainment significantly differed by incarceration status. Coefficients with significant interaction tests are highlighted in bold. Thus, Table 3 demonstrates the significance of the relationship between education and the health conditions net important confounders and whether the educational gradient in the health outcomes is stronger in the non-incarcerated population than incarcerated population.

Table 3:

Odds Ratios from Logistic Models Predicting Health Outcomes by Gender and Incarceration Status, Americans Aged 18–65, 2001–2004.

Hypertension Diabetes Asthma
GP CI GP CI GP CI
Women OR p OR p OR p OR p OR p OR p
Less than High School Diploma
High School 0.75 *** 0.79 * 0.71 *** 0.86 0.83 *** 0.81 †
Greater than High School 0.60 *** 0.93 0.54 *** 1.27 0.96 1.05
Men
Less than High School Diploma
High School 0.95 0.25 0.96 0.53 0.88 0.06 0.89 0.31 0.87 0.04 0.89 0.09
Greater than High School 0.82 0.00 1.02 0.77 0.74 0.00 1.05 0.63 1.07 0.25 1.17 0.03
Arthritis Cancer Hepatitis
GP CI GP CI GP CI
OR p OR p OR p OR p OR p OR p
Women
Less than High School Diploma
High School 0.86 *** 0.79 * 0.97 0.71 * 0.74 *** 0.52 ***
Greater than High School 0.79 *** 0.97 1.09 1.07 0.99 0.46 ***
Men
Less than High School Diploma
High School 0.97 0.94 0.97 1.00 0.99 0.80 **
Greater than High School 0.85 *** 1.09 1.13 1.72 *** 1.19 * 0.72 ***
Open in a new tab
†

p < 0.1

*

p <0.05

**

p <0.01

***

p < 0.001

GP = General Population, Not-currently incarcerated (Data source: National Health Interview Survey). N = 74,881.

CI = Currently Incarcerated (Data Source: Survey of Inmates in State and Federal Correctional Facilities). N =17,553.

Notes: Models additionally control for age, race/ethnicity, marital status, and smoking status.

Bold indicates significant (p < .05) difference as measured by an interaction term between the effect of education and incarceration status.

Table 3 indicated that educated non-incarcerated females, especially those with more than a high school education, had significantly lower odds of hypertension (OR: .60, p < .001), diabetes (OR: .54, p < .001), and arthritis (OR: .79, p < .001) than non-incarcerated females with less than a high school education. For these outcomes and for non-incarcerated females, there was a “step-wise” association between education and the health conditions. Non-incarcerated men showed a similar relationship between education and odds of hypertension, diabetes, and arthritis. There were no significant negative “step-wise” associations between education and asthma, cancer, or hepatitis in the non-incarcerated population. The positive (albeit non-significant) relationships of education with cancer and asthma are consistent with the patterns in Table 2 that are discussed above.

Table 3 indicated fewer statistically significant relationships between education and chronic health conditions for the incarcerated population. Incarcerated women with greater than high school education had no statistically significant negative difference with women with less than high school education on any of the chronic health outcomes. There were some differences between women with a high school education only and those without one on hypertension, arthritis, and cancer. Among incarcerated men, those with more than a high school education had significantly higher odds of asthma (OR: 1.17, p> .05) and cancer (OR: 1.72, p > .05) with no other statistically significant differences evident by level of education. In sum, among the incarcerated population, the relationship between education and chronic conditions was either non-significant or positive for the chronic conditions examined.

A major objective was also to compare educational gradients in health conditions by educational attainment between the incarcerated and non-incarcerated. To accomplish this, I fit interaction models for educational attainment by incarceration status. Significant interaction terms are highlighted in bold in Table 3. The interaction results generally support Hypothesis #1, that the educational gradient in chronic conditions were significantly stronger in the non-incarcerated population. Among women, having more than a high school education was significantly more protective in the non-incarcerated population for hypertension, diabetes, and arthritis. For men, having more than a high school education was significantly more protective in the non-incarcerated population for hypertension, diabetes, arthritis, and cancer.

Note that for the infectious condition, hepatitis, there was an educational gradient in the incarcerated population—persons with high levels of education are significantly less likely to report hepatitis—but this is not evident among the more educated in the non-incarcerated population. Additionally, the association between education and hepatitis was significantly stronger and more protective among the incarcerated than the non-incarcerated. As mentioned, hepatitis is often spread through risky health behaviors such as intravenous drug use, tattoos, and sexual contact, which the more educated may be more likely to avoid, potentially especially so in prisons where there are higher levels of these behaviors. The evidence that education appears to protect incarcerated men and women more than the non-incarcerated supports Hypothesis #3, that there is larger educational inequality in Hepatitis among the incarcerated.

To further explore more proximate pathways between education and hepatitis, in an additional analysis focused exclusively on the incarcerated I additionally controlled for whether incarcerated respondent had ever used a needle to inject drugs. Notably, the question did not indicate when the drug use occurred (in prison or before). Nevertheless, I found that this control explained the association between education and hepatitis for women and weakened the association for men. This indicates that the lower levels of hepatitis for educated incarcerated men and women were at least partially due to their lower propensity to engage in intravenous drug use.

Sensitivity Analyses

I tested the sensitivity of these findings in six important ways. First, I stratified my sample into two age groups to better account for the differences in the age distribution between the incarcerated and non-incarcerated samples (See Supplementary Table 1). Second, while prisoners are provided constitutionally mandated medical care, this care is not without problems (Wilper et al. 2009). Notably, the incarcerated still might lack the same access to and quality of medical care that the non-incarcerated public has. This could lead to underreporting of health conditions. In a sensitivity analysis I therefore limited the incarcerated sample to only those who received a medical examination upon arrival in prison, that is, to those who have seen a doctor relatively recently (Supplementary Table 2). Third, I tested if exposure to prison (measured in years as duration) moderated the effect of educational attainment on health outcomes (it did not, results not shown). Fourth, I tested the robustness of my findings by testing three additional specifications of educational attainment, (Supplementary Table 3). Fifth, I implemented Inverse Probability Weighted Regression Adjustment models (Supplemental Table 4). Finally, I estimated national prevalence estimates of educational inequality for each health condition to include the incarcerated (Supplementary Tables 5 and 6). I briefly discuss these results briefly here and in more detail in the Online Supplement.

The age stratified results were largely consistent with the results presented in the text (See Supplementary Table 1). Results for those who were incarcerated within the past five years and indicated having a medical examination did little to alter the overall conclusions (see Supplementary Table 2). There was not a significant interaction between educational attainment and duration incarcerated for any of the health conditions. I also tested two additional categorical specifications and one linear specification of educational attainment. Again, the substantive results were largely similar to the main analysis. Notably, the results were slightly different for men with hypertension and for women with cancer, but only slightly and for one specification. Supplementary Table 3 presents the full results of the analyses regarding functional form.

Another important concern is the non-random selection that occurs into prison. That is, those who are incarcerated are select on observable and unobservable characteristics. Indeed, to partially account for this methodological challenge, previous research has implemented Propensity Score Matching to analyze the heath behaviors of formerly incarcerated matched to those who were convicted of crimes but not incarcerated (Porter 2014). Consistently I implemented Inverse Probability Weighted Regression Adjustment Models (IPWRA) (Robins et al. 2000; Thoemmes and Ong 2016). IPWRA use probability weighting to create pseudo-populations that emulate random assignment experiments where the covariates and treatments are independent. Compared to multivariate techniques, this minimizes the bias resulting from observed confounders (Thoemmes and Ong 2016). Rendering treatments independent from observed confounders is especially relevant in this research because many variables are likely associated with both incarceration and the health outcomes (including education).

In detail, I first estimated a logistic model predicting incarceration on the whole sample based on age, race, gender, and educational attainment. Each of these predictors was statistically significant and the pseudo R2 (Hoffmann 2004) was relatively high ~0.30. I next used the model results to calculate the probability of being incarcerated and assigned the respondents the inverse probability of incarceration as probability weights. These weights created a “pseudo-population” where the treatment (incarceration) is independent of the measured confounders (including the influence of educational attainment on the likelihood of being incarcerated). I then refit the fully adjusted interaction models (Education X Incarceration) weighted by the inverse probability of being incarcerated. The results, except for hepatitis for men, were substantively similar and the full results of this analysis are presented in Supplementary Table 4.

Finally, I followed the protocols developed by Nowotny et al. (2017), to update national prevalence estimates of educational inequality by including incarcerated individuals. To do this, I took the weighted population estimate from the NHIS and added the weighted population estimate from the SISFCF while simultaneously adjusting the denominator to include the incarcerated. The full results of the updated proportions are presented in Supplementary Tables 5 (women) and 6 (men). Overall, updating the prevalence estimates to include the incarcerated made little difference for women, who have much lower incarceration rates than men. The proportion of men with less than a high school education who reported hypertension decreased slightly when the incarcerated were included in national estimates. However, the national proportion of men with less than a high school education who reported having Asthma and Hepatitis increased when the incarcerated were included. Overall, these findings largely validated results from previous research that found an association between education and health conditions.

Discussion

Previous research has documented a strong relationship between education and health in the U.S. population (Baker et al. 2011; Hummer and Lariscy 2011). However, this research is largely, although not exclusively, based on national surveys which sample the non-institutionalized population. In other words, these surveys exclude the currently incarcerated from the sampling frame during the period that the surveys are being administered. Because those who are incarcerated are a non-trivially sized population and not missing completely at random (Allison 2001), their exclusion could challenge the validity of findings regarding the relationship between education and health. In this study, I analyzed the extent of educational inequality in the prevalence of health conditions among the incarcerated and non-incarcerated. I built on previous research that analyzed the relationship between education and overall health among the incarcerated (Nowotny et al. 2016), by predicting multiple specific health conditions, comparing the association between the incarcerated and non-incarcerated, and updating national level prevalence estimates of health conditions to include the incarcerated.

Consistent with Hypothesis #1 and in opposition to Hypothesis #2, I found little evidence of a relationship between educational attainment and chronic health conditions among the incarcerated but generally significant negative associations between educational attainment and chronic conditions among the non-incarcerated. These results were consistent even when doubly robust IPWRA models were implemented. This accords with previous research showing that in non-restrictive circumstances the highly educated can better accrue gradual and other health advantages than those with low levels of education (Hayward et al. 2015; Hummer and Hernandez 2013; Mirowsky and Ross 2015). These gradual advantages can accumulate and ultimately delay or even prevent chronic “man-made” diseases (Omran 1971). In contrast, in prison most intervening mechanisms, such as diet, residential location, health behaviors, exercise and duration of free time, through which the highly educated gain health advantages are held relatively constant (Link and Phelan 1995; Mirowsky and Ross 2015).

Overall, this analysis found little relationship between educational attainment and chronic health conditions among the incarcerated. While I could not explicitly test the influence of differentials in access to health care (presumably all incarcerated have health care), these results may indicate that further access to health care would minimize socioeconomic based health inequality in chronic conditions. A noteworthy caveat to this finding is the inability to distinguish whether incarceration leads to smaller disparities or if the incarcerated would have also had smaller disparities in the general population had they not been incarcerated. In other words, given the detailed lack of information regarding health before incarceration, it remains unclear whether incarceration is minimizing educational health inequality or if the minimization is based on pre-incarceration health. Thus, it is possible that the lack of the observed educational gradient among the incarcerated could be due to prisons drawing from less healthy populations with smaller educational gradients in health. This question is ripe for future longitudinal research. Regardless the lack of the educational gradient among this population is notable as this is a segment of the population traditionally excluded in nationally representative surveys, and thus a segment of the population that is traditionally not analyzed when examining the relationship between education and health.

Consistent with Hypothesis #3, I found a stronger relationship between education and hepatitis among the incarcerated than the non-incarcerated. In prison, men and women with higher levels of education had significantly lower rates of hepatitis than men and women with low levels of education. The risk of contracting hepatitis was much greater in prison than in the general population (Binswanger et al. 2009), and the highly educated in prison are able to avoid behaviors associated with the risk. Still the descriptive results show that men and women with low levels of educational attainment would be best targeted by policies aimed at preventing hepatitis (such as vaccines) in prison. While some intermediate mechanisms for the relationship between education and health are similar for all in prison, new and elevated health risks are present for infectious conditions. For hepatitis, this analysis documents that the ability to avoid new health risks, especially intravenous drug use, even extends to even the most extremely regulated institutions. Given the structure of the data, however, it is impossible to test the causal nature of education for hepatitis in prison and in the general public, or if those who report having hepatitis contracted it before or during incarceration. Indeed, the IPWRA indicated that the association between education and hepatitis was not stronger for incarcerated males than non-incarcerated males. Taken together, these results suggest that while education may play a lesser role for chronic conditions in prison, education is still critically important for avoiding risky health behaviors.

It is also worth noting that the non-incarcerated population likely includes those who had been incarcerated before, likely leading to conservative findings for our chronic conditions results and potentially exaggerated findings for hepatitis. Unfortunately, the NHIS (like many other population health-based surveys), during those survey years does not have detailed questions regarding incarceration history. Given the extent to which incarceration shapes health and subsequent life-course outcomes, incarceration history is an important area for population health scholars to continue to investigate and survey makers to incorporate in surveys. Understanding how incarceration may influence the education/health relationship remains especially important for racial/ethnic minorities, who are disproportionately incarcerated (Pettit 2012) and often have a weaker education/health relationship than non-Hispanic whites (Hummer and Hernandez 2013). The education/health relationship among the incarcerated may have changed more recently than the period analyzed here (2002–2004). Indeed, recently, the opiate epidemic has disproportionately impacted the longevity of low educated Non-Hispanic white Americans (Case and Deaton 2015). Thus, consistent with incarceration shielding young non-Hispanic blacks from homicide (Rosen et al. 2008), it is possible that incarceration has become protective for the longevity of currently incarcerated low-educated whites, at least those who would be at risk for opiate use.

The fundamental cause perspective promulgates that socioeconomic status can prevent the emergence of health conditions through multiple pathways and resources that can be used flexibly (Phelan et al. 2004). In the highly-regulated environment of prisons, the pathways and the extent to which resources can be used flexibly are substantially constrained. Therefore, it is not surprising that there is a smaller association between education and chronic health outcomes among prisoners than non-prisoners. Prison, however, is also an institutional context where the risk of hepatitis is significantly higher than in the general population. This is because in prison new risks emerge, that may not be present for the non-incarcerated population. In this institutional circumstance, the highly educated can prevent an infectious disease more than the less educated can. However, the relationship between education and health conditions while incarcerated likely changes after release from prison (Schnittker and John 2007), thus future research should also consider how education may alter reintegration after release from incarceration, a critically important time period, with specific attention paid to health behaviors and health (Porter 2014). One of the greatest strengths of population research is the external validity or generalizability of large-scale social processes. I echo previous research (Nowotny et al. 2017; Pettit 2012) that emphasize the importance of population scholars to continue to analyzing well documented processes in specific segments of the population that are systematically excluded from large scale surveys. These segments may even inform processes in the broader population.

Supplementary Material

Supplemental Materials

Acknowledgements:

I thank the National Institute of Aging training grant (T32 AG000037) for support and the University of Minnesota for making the NHIS data available to the public. I thank Becky Petitt, Mark Hayward, Robert Hummer, Tetyana Pudrovska, Daniel Powers, four anonymous reviewers, and Debra Umberson for their helpful comments. The contents of this manuscript are solely the responsibility of the author and do not represent the views of the National Institute of Aging or the University of Southern California.

REFERENCES

  1. Albano JD, Ward E, Jemal A, Anderson R, Cokkinides VE, Murray T, et al. (2007). Cancer mortality in the United States by education level and race. Journal of the National Cancer Institute, 99(18), 1384–1394. [DOI] [PubMed] [Google Scholar]
  2. Allison PD (2001). Missing data (Vol. 136). Sage publications. [Google Scholar]
  3. Arum R, & Beattie IR (1999). High School Experience and the Risk of Adult Incarceration. Criminology, 37(3), 515–540. [Google Scholar]
  4. Baker DP, Leon J, Smith Greenaway EG, Collins J, & Movit M (2011). The education effect on population health: a reassessment. Population and development review, 37(2), 307–332. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Baussano I, Williams BG, Nunn P, Beggiato M, Fedeli U, & Scano F (2010). Tuberculosis incidence in prisons: a systematic review. PLoS Med, 7(12), e1000381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Binswanger IA, Krueger PM, & Steiner JF (2009). Prevalence of chronic medical conditions among jail and prison inmates in the United States compared with the general population. Journal of epidemiology and community health, jech–2009. [DOI] [PubMed] [Google Scholar]
  7. Bottoms AE (1999). Interpersonal violence and social order in prisons. Crime and justice, 26, 205–281. [Google Scholar]
  8. Brown DC, Hummer RA, & Hayward MD (2014). The importance of spousal education for the self-rated health of married adults in the United States. Population research and policy review, 33(1), 127–151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bureau of Justice Statistics, US Dep’t of Justice. (2003). Sourcebook of Criminal Justice Statistics. US Department of Justice, Bureau of Justice Statistics; Washington, DC. [Google Scholar]
  10. Carson E (2014). Prisoners in 2013 (NCJ 247282). Washington, DC: US Department of Justice. [Google Scholar]
  11. Case A, & Deaton A (2015). Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century. PNAS Web site. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Center of Disease Control. (2013). Hepatitis C and Incarceration. https://www.cdc.gov/hepatitis/hcv/pdfs/hepcincarcerationfactsheet.pdf
  13. Dolan K, & Carr JL (2015). The Poor get Prison: the Alarming Spread of the Criminalization of Poverty. Report from the Institute for Policy Studies, DC. [Google Scholar]
  14. Dubow EF, Boxer P, & Huesmann LR (2009). Long-term effects of parents’ education on children’s educational and occupational success: Mediation by family interactions, child aggression, and teenage aspirations. Merrill-Palmer Quarterly (Wayne State University. Press), 55(3), 224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Everett BG, Rehkopf DH, & Rogers RG (2013). The nonlinear relationship between education and mortality: an examination of cohort, race/ethnic, and gender differences. Population research and policy review, 32(6), 893–917. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Friedman E, Montez J, Sheehan C, Guenewald T, & Seeman T (2015). Childhood Adversities and Adult Cardiometabolic Health: Does the Quantity, Timing, and Type of Adversity Matter? Journal of aging and health. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Garrison RJ, Gold RS, Wilson P, & Kannel WB (1993). Educational attainment and coronary heart disease risk: the Framingham Offspring Study. Preventive medicine, 22(1), 54–64. [DOI] [PubMed] [Google Scholar]
  18. Goffman E (1961). Asylums: essays on the soocial situatioon of mental patients and other inmates. [Google Scholar]
  19. Harlow CW (2003). Education and Correctional Populations. Bureau of Justice Statistics Special Report. [Google Scholar]
  20. Hayward MD, Hummer RA, & Sasson I (2015). Trends and group differences in the association between educational attainment and US adult mortality: Implications for understanding education’s causal influence. Social Science & Medicine, 127, 8–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hayward MD, Miles TP, Crimmins EM, & Yang Y (2000). The significance of socioeconomic status in explaining the racial gap in chronic health conditions. American sociological review, 910–930. [Google Scholar]
  22. Heck KE, & Pamuk ER (1997). Explaining the relation between education and postmenopausal breast cancer. American Journal of Epidemiology, 145(4), 366–372. [DOI] [PubMed] [Google Scholar]
  23. Hoffmann JP (2004). Generalized linear models: An applied approach. Addison-Wesley Longman. [Google Scholar]
  24. Hummer RA, & Hernandez EM (2013). The Effect of Educational Attainment on Adult Mortality in the United States*. Population bulletin, 68(1), 1. [PMC free article] [PubMed] [Google Scholar]
  25. Hummer RA, & Lariscy JT (2011). Educational attainment and adult mortality. In International handbook of adult mortality (pp. 241–261). Springer. [Google Scholar]
  26. Kim Y (2015). The effect of incarceration on midlife health: A life-course approach. Population Research and Policy Review, 34(6), 827–849. [Google Scholar]
  27. Link BG (2008). Epidemiological Sociology and the Social Shaping of Population Health∗. Journal of Health and Social Behavior, 49(4), 367–384. [DOI] [PubMed] [Google Scholar]
  28. Link BG, & Phelan J (1995). Social conditions as fundamental causes of disease. Journal of health and social behavior, 80–94. [PubMed] [Google Scholar]
  29. Liu H, & Hummer RA (2008). Are educational differences in US self-rated health increasing?: An examination by gender and race. Social Science & Medicine, 67(11), 1898–1906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Manza J, & Uggen C (2008). Locked out: Felon disenfranchisement and American democracy. Oxford University Press. [Google Scholar]
  31. Marmot M (2002). The influence of income on health: views of an epidemiologist. Health affairs, 21(2), 31–46. [DOI] [PubMed] [Google Scholar]
  32. Maruschak LM, Berzofsky M, & Unangst J (2015). Medical problems of state and federal prisoners and jail inmates, 2011–12. US Department of Justice, Officer of Justice Programs, Bureau of Justice Statistics. Accessed, 6, 2015. [Google Scholar]
  33. Massoglia M (2008). Incarceration as exposure: the prison, infectious disease, and other stress-related illnesses. Journal of health and social behavior, 49(1), 56–71. [DOI] [PubMed] [Google Scholar]
  34. Massoglia M, & Pridemore WA (2015). Incarceration and health. Annual review of sociology, 41, 291–310. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Masters RK (2012). Uncrossing the US black-white mortality crossover: The role of cohort forces in life course mortality risk. Demography, 49(3), 773–796. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Mezuk B, Eaton WW, Golden SH, & Ding Y (2008). The influence of educational attainment on depression and risk of type 2 diabetes. American journal of public health, 98(8), 1480–1485. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Miech R, Pampel F, Kim J, & Rogers RG (2011). The enduring association between education and mortality the role of widening and narrowing disparities. American Sociological Review, 76(6), 913–934. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Mirowsky J, & Ross CE (2005). Education, cumulative advantage, and health. Ageing International, 30(1), 27–62. [Google Scholar]
  39. Mirowsky J, & Ross CE (2015). Education, Health, and the Default American Lifestyle. Journal of health and social behavior, 56(3), 297–306. [DOI] [PubMed] [Google Scholar]
  40. Montez JK, & Hayward MD (2014). Cumulative childhood adversity, educational attainment, and active life expectancy among US adults. Demography, 51(2), 413–435. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Montez JK, & Zajacova A (2013). Trends in mortality risk by education level and cause of death among US White women from 1986 to 2006. American journal of public health, 103(3), 473–479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Nowotny KM, Boardman JD, & Rogers RG (2017). Racial Disparities in Health Conditions among Prisoners Compared with the General Population. SSM-Population Health, 3, 487. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Nowotny KM, Masters RK, & Boardman JD (2016). The relationship between education and health among incarcerated men and women in the United States. BMC Public Health, 16(1), 916. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Olshansky SJ, Antonucci T, Berkman L, Binstock RH, Boersch-Supan A, Cacioppo JT, et al. (2012). Differences in life expectancy due to race and educational differences are widening, and many may not catch up. Health Affairs, 31(8), 1803–1813. [DOI] [PubMed] [Google Scholar]
  45. Omran AR (1971). The epidemiologic transition: a theory of the epidemiology of population change. The Milbank Memorial Fund Quarterly, 49(4), 509–538. [PubMed] [Google Scholar]
  46. Pattillo M, Western B, & Weiman D (2004). Imprisoning America: The social effects of mass incarceration. Russell Sage Foundation. [Google Scholar]
  47. Pettit B (2012). Invisible Men: Mass Incarceration and the Myth of Black Progress: Mass Incarceration and the Myth of Black Progress. Russell Sage Foundation. [Google Scholar]
  48. Phelan JC, Link BG, Diez-Roux A, Kawachi I, & Levin B (2004). “Fundamental causes” of social inequalities in mortality: a test of the theory. Journal of health and social behavior, 45(3), 265–285. [DOI] [PubMed] [Google Scholar]
  49. Porter LC (2014). Incarceration and post-release health behavior. Journal of Health and Social Behavior, 55(2), 234–249. [DOI] [PubMed] [Google Scholar]
  50. Porter LC, Bushway SD, TSAO H-S, & Smith HL (2016). How the US prison boom has changed the age distribution of the prison population. Criminology. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Pudrovska T, & Anishkin A (2015). Clarifying the positive association between education and prostate cancer: a Monte Carlo simulation approach. Journal of Applied Gerontology, 34(3), 293–316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Rikard R, & Rosenberg E (2007). Aging inmates: A convergence of trends in the American criminal justice system. Journal of Correctional Health Care, 13(3), 150–162. [Google Scholar]
  53. Roberts DE (2004). The social and moral cost of mass incarceration in African American communities. Stanford Law Review, 1271–1305. [Google Scholar]
  54. Robins JM, Hernan MA, & Brumback B (2000). Marginal structural models and causal inference in epidemiology. Epidemiology, 11(5), 550–560. [DOI] [PubMed] [Google Scholar]
  55. Rosen DL, Schoenbach VJ, & Wohl DA (2008). All-cause and cause-specific mortality among men released from state prison, 1980–2005. American journal of public health, 98(12), 2278–2284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Schiller JS, Adams PF, & Nelson ZC (2005). Summary health statistics for the US population: National Health Interview Survey, 2003. Vital and health statistics. Series 10, Data from the National Health Survey, (224), 1–104. [PubMed] [Google Scholar]
  57. Schnittker J, & John A (2007). Enduring stigma: the long-term effects of incarceration on health. Journal of health and social behavior, 48(2), 115–130. [DOI] [PubMed] [Google Scholar]
  58. Sheehan CM, Hummer RA, Moore BL, Huyser KR, & Butler JS (2015). Duty, Honor, Country, Disparity: Race/Ethnic Differences in Health and Disability Among Male Veterans. Population Research and Policy Review, 1–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Siegel R, Naishadham D, & Jemal A (2012). Cancer statistics, 2012. CA: a cancer journal for clinicians, 62(1), 10–29. [DOI] [PubMed] [Google Scholar]
  60. Spaulding AC, Seals RM, McCallum VA, Perez SD, Brzozowski AK, & Steenland NK (2011). Prisoner survival inside and outside of the institution: implications for health-care planning. American Journal of Epidemiology, kwq422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Stapleton D, Honeycutt T, & Schechter B (2012). Out of sight, out of mind: Including group quarters residents with household residents can change what we know about working-age people with disabilities. Demography, 49(1), 267–289. [DOI] [PubMed] [Google Scholar]
  62. Thoemmes F, & Ong AD (2016). A Primer on Inverse Probability of Treatment Weighting and Marginal Structural Models. Emerging Adulthood, 4(1), 40–59. [Google Scholar]
  63. Umberson D (1987). Family status and health behaviors: Social control as a dimension of social integration. Journal of health and social behavior, 306–319. [PubMed] [Google Scholar]
  64. Umberson D (1992). Gender, marital status and the social control of health behavior. Social science & medicine, 34(8), 907–917. [DOI] [PubMed] [Google Scholar]
  65. Wakefield S, & Wildeman C (2011). Mass imprisonment and racial disparities in childhood behavioral problems. Criminology & Public Policy, 10(3), 793–817. [Google Scholar]
  66. Wilper AP, Woolhandler S, Boyd JW, Lasser KE, McCormick D, Bor DH, & Himmelstein DU (2009). The health and health care of US prisoners: results of a nationwide survey. American journal of public health, 99(4), 666–672. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Materials
NIHMS1830230-supplement-Supplemental_Materials.docx (51.3KB, docx)

RESOURCES