Michael Massoglia (University of Wisconsin) began his presentation on the current state of knowledge with reference to estimates on the number of incarcerated and ex-incarcerated individuals in the United States. Crudely put, Massoglia said, the same number of people are graduating college as being released from prison: the number of people released from prison each year has fluctuated between 600,000 and 700,000 over the last decade (the vast majority of whom are males) (Guerino et al., 2012); according to the National Center for Education Statistics, 678,000 males received bachelor’s degrees in 2008. He noted that there has been a significant expansion in criminal justice involvement in the past several decades. This expansion of the penal system has not occurred in a vacuum or because of a single new law or initiative: it has resulted from a pattern of interconnected laws—state, local, and federal—and it has had effects at all societal levels.
Recent research has studied not only the effects of incarceration on incarcerated adults, but also the health effects on the partners and children of incarcerated individuals and the communities to which they belong. The evidence shows that incarceration generally has negative effects on mental and physical health, for both morbidity and mortality, as well as increasing the risk of anxiety and depression. Those effects on health can be both short term and long term. Massoglia noted that the effects do not differ significantly by race: rather, the effects (including by race) are more often a function of differential exposure, with communities
that have higher rates of incarceration being more susceptible to negative health effects.
Massoglia acknowledged that prison conditions can be very harsh (see Chapter 1), but he pointed out that incarcerated people can derive some benefits from the penal system that are comparatively better than what they were exposed to or had access to in their communities, including consistent dwelling spaces, basic sustenance, possible removal or protection from other risks (such as protection from domestic abuse), and access to health care.
Massoglia told the group that identifying a single treatment variable through which to measure criminal justice contact has proven challenging. There is also a question, however, as to whether the selection of one treatment variable over another would have much bearing on understanding the effects of incarceration. He noted the variables that are often considered: whether a person has been convicted of a felony; if that conviction led to incarceration; if so, whether the person spent time in a jail or prison; if applicable, whether federal or state prison; and length of incarceration. Massoglia offered three examples on how greatly the outcome can vary for convicted or incarcerated individuals:
- Convicted felons spend an average of 37.5 months in federal prisons, which involves extended separation from family and community, as well as removal from the labor market.
- About 65 percent of jailed individuals spend one night or less in a jail and may not ultimately even be convicted of a felony.
- It is possible for a person to be convicted of a felony and not spend any time in a prison or jail.
Being arrested has not been a major component of the research on the effects of criminal justice involvement. Since there is no consensus in the current research about which particular measure of contact with the criminal justice system to use, it raises the question of whether having a single, uniform definition or measure can encompass the differential effects that criminal justice contact can have on individuals, families, and communities.1 Massoglia said that it is important to consider each criminal justice treatment and whether it leads to different health outcomes than other treatments.
Massoglia described measures that are important in identifying the causal relationship between incarceration and health. He approached
1 Massoglia drew an analogy with how the word “employed” encompasses a number of dimensions that include full-time or part-time status, type of occupation, dual earnings and that, as such, saying one is employed does not provide much insight into the person’s specific condition or expected outcome.
the issue from both perspectives: from a health standpoint, longitudinal data can provide the degree of detail needed to identify patterns and problems. Information like help-seeking behavior, lifestyle indicators (e.g., exercise, drug use), and even self-reported illness can prove informative in this regard. From a criminal justice perspective, longitudinal data can also prove very useful, as well as data on type and length of contact and conviction history. Self-reported criminal behavior can provide unique insight and allow researchers to differentiate the effects of being in prison from the effects of leading a risky lifestyle out of prison. Many of these data efforts would require collaboration across multiple agencies at several levels, such as the Department of Justice, the National Institutes of Health, and state and local health and corrections agencies.
Limited data on criminal justice involvement can be gleaned from household surveys such as the Fragile Families Study, the National Longitudinal Study of Adolescent to Adult Health, and the National Longitudinal Survey of Youth. These surveys collect meaningful information, but because the primary focus differs slightly for each, and none of the three is focused primarily on criminal justice involvement, they serve as inconsistent measures of treatment variables for health as it pertains to incarceration. (See Chapter 5 for information on how national household-based population health surveys collect information on incarcerated populations.)
Massoglia outlined the key research questions that he is hopeful can be answered by investments in new survey data collection: causal mechanisms, comparison groups, confounding factors, drug use, global effects, first-order and second-order effects, demographics, and breadth of effects.
Specific Causal Mechanisms There is a general consensus in the literature that incarceration matters, and research has provided evidence that suggests that incarcerated individuals are at a health disadvantage. Why, though, does incarceration matter? Is it because an individual has been removed from the community and so lost social support? Is it the exposure to a unique set of stressors and adverse health effects during incarceration? Or is it possibly the stigmatization people face as they attempt to rejoin a community after release?
Comparison Group What is the baseline for measurement? Should individuals be treated as their own control, or should researchers make intergroup comparisons of people with similar levels of criminal justice contact? By taking the latter approach, incarcerated individuals could be compared with: ex-incarcerated individuals; those in jails but not in prisons; those who have been jailed but not convicted; and those who have only been arrested. It is straightforward to get estimates or counts of incarcerated populations from administrative and survey data. But for other popu-
lations, such as people formerly involved in the criminal justice system, getting accurate data would be considerably more challenging.
Confounding Factors How great a role does institutional variation among incarceration facilities play in determining the health effects on individuals? Most health measures only ask if a person was or was not incarcerated. If one views incarceration as a black-or-white effect, the length of the term may not matter much. However, if one considers such possibilities as whether or not the effects are exacerbated by the quality of facilities, the type of confinement (e.g., solitary confinement or in the general prison population), or whether multiple incarcerations yield duplicative effects, the duration of stay would matter.
Drug Use Drug use is an important comorbidity/confounding factor, albeit one that is very difficult to measure in this context. Incarceration may prevent access to some drugs but enable exposure to others. However, it is clear that drug use has a detrimental effect on health, and it is an important factor to consider for this population. As essential as longitudinal data can be, they do not always shed light on the effects of drug use because the health problems associated with drug use may have a long dormant period.
Global Effects The field is still in the early stages of understanding the relationship between incarceration and health. As mass incarceration is a relatively new phenomenon, it is still unknown whether the health effects of incarceration are likely to become stronger when the incarcerated and formerly incarcerated populations age. If the health effects become stronger, it will have implications for government-funded health programs, such as Medicare and parts of the Affordable Care Act.
First-Order and Second-Order Effects As noted above, being incarcerated exposes a person to a regular regime of physical exercise, activity, and food—access to certain daily life activities that may have been lacking for those who come from the lower income strata of society. Are positive effects stronger for certain demographic groups and do those effects become negative over time? Do they ever outweigh the effects of exposure to the negative stressors of incarceration?
Demographics (Gender, Race, and Age) Female offenders are the fastest growing group in the prison system, yet the field lacks sufficient data on the effects of incarceration on their health outcomes. Race is another important factor: Hispanics and non-black minorities are particularly understudied populations. In addition, the age at which an individual experiences
criminal justice involvement can be another mediating factor when it comes to measuring the relationship between incarceration and health.
Breadth of Effects How does the incarceration experience influence the behavior and mental health of incarcerated and ex-incarcerated persons? Massoglia said that some of the above questions can be answered using preliminary and aggregate-level data, but for some questions longitudinal and individual-level data would be necessary. As a starting point for all concerns, there is a great need for population counts. The count of currently incarcerated individuals is readily attainable, but as noted above the numbers of ex-prisoners and ex-felons in the United States are considerably harder to attain. Those numbers are key, however, to considering mortality, migration, reincarceration, and other societal effects.
Massoglia stressed the following next steps to reduce the negative impacts of incarceration on health. First, there is a need to identify best practices and intervention points, both for individuals while they are institutionalized and after they are released from prison. Health-related best practices could include health education, access to drug treatment, and perhaps even adding a health care component to post-release monitoring or parole. Socially, programs for prison inmates to help them maintain family bonds and pre-release counseling may have a positive effect on health. Other programs, such as job training and skill development, can help prisoners find stable employment opportunities once they are released. These social and health components can be mutually reinforcing.
Candace Kruttschnitt (University of Toronto) pointed out that many female prisoners have children while they are in prison, and so their needs are separate from those of male prisoners. She suggested that a longitudinal survey could collect information on prenatal care and follow up on children born in prison. Massoglia agreed that female prisoners do have different medical needs, but the research is lacking on how they are different from male prisoners and how health effects differ for them. In that respect, he agreed that longitudinal data would provide an opportunity to understand those differences and figure out intervention points.
Daniel Nagin (Carnegie Mellon University) commented that he does not believe it sufficient to say that each individual could potentially serve as his or her own control in a longitudinal study. Different “treatments” will have different relative effects, based on an individual’s life course, therefore causing the baseline for comparison to potentially change. In terms of medical care, specifically, Nagin commented that it is important
to remember the context for comparison: Is one comparing the quality of care before prison to the care in prison? Or is one comparing the quality of care in prison to other, better health care offerings? Massoglia acknowledged that it is important to understand the counterfactual—to whom is one comparing the incarcerated population and what method is being used for the comparison?
Given the data limitations in the field, when information is not always available on a control group, Massoglia suggested the analytical approach of using a longitudinal fixed-effects model. John Hagan (Northwestern University) said that the majority of the research work in the field of incarceration and health has been a reapplication or building on datasets of developmental criminology, psychology, and life-course sociology. He referred to the point made by Massoglia about looking more closely at the experiences of subpopulations within the incarcerated community, which would require a different method of sampling and different allocation of research resources. Hagan asked Massoglia if he thought it was time for that kind of new consideration and reallocation. Massoglia answered that he would first paint with broad strokes, looking at basic demographics—race, gender, etc.—and then move to identifying other causal relationships.
Alexis Bakos (Office of Minority Health) asked whether or not the current research on quality of care in prisons and jails has taken into account the possible iatrogenic effects resulting from subpar practices before major reform changes, such as the Affordable Care Act, were implemented. Could one say that in-prison care was better than community care under those conditions? Massoglia responded by saying that because the data on quality of care and the research community’s understanding of it are both limited, it is difficult to ascertain.
Josiah Rich (Brown University) commented that the difference between prison facilities in the United States and those in other developed countries, and between prison and civilian medical facilities stateside, is a lack of oversight. While nonprison medical facilities are subject to the Joint Commission and the Accreditation of Hospitals, where deficiencies can have serious fiscal and other consequences, prison medical facilities do not have anything similar. There is a voluntary accreditation program through the National Commission on Correctional Healthcare, but only about 20 percent of facilities (about 500 nationwide) participate, and lack of accreditation has no significant consequences. Information on quality usually surfaces only when it is truly substandard and there is litigation; however, the Prison Litigation Reform Act has substantially limited the use of litigation as a tool to drive improvements and oversight of health care in correctional facilities.
Emily Wang (Yale School of Medicine) noted that, while causality is important, there are crucial questions that are as of yet unanswered about what the effects of having a large incarcerated population have had on
society and about how incarceration actually leads to negative outcomes for individuals, their families, and communities. In terms of effects, how does serving a continuous 30-year sentence compare with serving multiple shorter sentences for that same length of time? She suggested that bibliometric information be collected on surveys to get reliable information on health status of individuals instead of relying on self-reporting of health conditions and that latent-class analysis be used to tease out heterogeneous effects.
Ross Matsueda (University of Washington) asked Massoglia if there is any research on the heterogeneity of experiences within prisons. Massoglia said “no” and added that information on prison quality and prison experience is still sparse. Chris Wildeman (Cornell University) responded that departments of corrections have become more forthcoming over time in terms of sharing their data, and the hope is to explore questions on treatment heterogeneity.
This workshop session was aimed at considering the causal components of criminal justice involvement on health; identifying critical unanswered research questions and the data needed to explore them; and identifying optimal data collection strategies—that is, whether any of the proposed existing data collection instruments are able to meet the most critical uses given resource constraints.
Matsueda began the discussion by summarizing the viewpoints and information covered in earlier workshop presentations and raising other critical questions. For example: If one views incarceration as a treatment, what, then, is the counterfactual? How can one best deal with treatment heterogeneity? Matsueda asked the session participants to also address the issue of resource constraints on research.
Ingrid Binswanger (Kaiser Permanente Colorado and University of Colorado School of Medicine) focused on the principles that should guide decisions regarding collecting health information on the criminal justice involved population. From her perspective as a physician and health services researcher, Binswanger described the challenges in understanding the health outcomes of this population. She said that she considers the population to be “hidden” in public health surveillance and research: although the Bureau of Justice Statistics (BJS) collects statistics on the health of people in jails and prisons, they are largely excluded from major health surveys, despite being at high risk for many negative public health outcomes. Undercounting this population severely limits knowledge about their access to quality health care, as well as the ability to obtain accurate public health estimates.
Binswanger listed 10 important reasons that the bidirectional linkage between criminal justice and health needs to be more deeply explored:
- surveillance—improving measures of prevalence and incidence of epidemics, illness, and other emerging health issues;
- alignment—providing corresponding data on what is available on external populations in order to make appropriate comparisons and measure quality gaps;
- policy—addressing emerging policy questions, for example, gauging the impact of the Affordable Care Act on health care for criminal justice involved populations;
- equity—measuring variability, mortality, or other disparities by race, ethnicity, gender, geographic location;
- prevention—measuring the availability and delivery of preventive services (such as vaccinations);
- patient centeredness—assessing the quality of care to ensure that it responds to each individual patient’s needs;
- transparency—providing accurate information about treatment, allowing public participation, promoting accountability;
- operations—using relevant data to guide criminal justice health care delivery, organization, and administration (such as human resources and contracting);
- performance and value—focusing on the value of care; and
- human rights and legality—assuring that patient rights are respected and constitutional or legal mandates are met.
Nagin highlighted the challenge of determining and measuring counterfactual outcomes when estimating the causal impact of incarceration on health. When one measures the effects of incarceration, to what condition is it being compared? What is the “alternative” treatment? In clinical trials, the control and treatment groups are well defined. When studying incarceration, Nagin encouraged the group to think about the best comparative treatment. The possibilities include individuals who were convicted but not incarcerated; individuals who were arrested but not convicted; and individuals who were not arrested. For more serious crimes, such as murder, incarceration post-conviction is imminent. However, for misdemeanors, summary offenses, and what Nagin refers to as “marginal felonies,” the outcomes can vary greatly.
Nagin also picked up a point from Massoglia’s presentation (see Chapter 4), noting that when comparing treatment during incarceration to treatment when not incarcerated it is important to consider both the health condition and the quality of treatment in both situations. There are some situations for which an individual may benefit from the access to health care
that is provided in the correctional institution. These types of treatment heterogeneities merit serious exploration.
Evelyn Patterson (Vanderbilt University) talked about the collateral consequences of incarceration and how the effects of incarceration can reverberate throughout a person’s life and the lives of his or her family and friends. She proposed a theoretical framework to understand the life experiences of people who have been incarcerated that highlights three major components: age, duration, and spells.
- Age: As people go through life they assume new social roles, some of which are cumulative—being old enough to work, to vote, becoming parents, etc. These roles are each affected when a person is incarcerated. Whether or not these roles are maintained during incarceration can provide insight into the effects of the situation on the individual and his or her family.
- Duration: Duration is also closely tied to social roles, in that the length of incarceration can be inversely proportionate to an individual’s ability to preserve his or her social roles. How well can the incarcerated individual cope with the incarceration, or weather the storm, so to speak?
- Spell: A person’s proximity to the incarceration experience affects how well he or she is able to understand it. Patterson pointed out that there are communities where the prevalence of incarceration is low and therefore general understanding of the experience and its effects are low.
Patterson noted that people’s experiences and outlooks can be colored by their social, physical, and civic development and involvement. For example, although a person may be physically healthy, he or she may not have a fully formed civic or social life, which would hamper his or her chances of or expectations for upward mobility. Because this theory applies both to people who have and have not been incarcerated, it is possible that the quality of these theoretical lives can be similar across the two groups and not just within them.
Massoglia asked Nagin to elaborate on how predictive measures can help researchers hone in on the most important treatment variables for the criminal justice population. Nagin spoke about the benefits of longitudinal surveys like the National Longitudinal Study of Adolescent to Adult Health and how early waves of the study can inform researchers about the health statuses of candidates as they prepare for data collection in later waves. Binswanger commented that she still has difficulty processing incarceration as a treatment, since she sees it as a much-overused last resort for medical conditions (addiction, mental health) that could best be addressed in
other settings. Nagin responded that, scientifically, the concept of treatment effect comes into play when one considers which treatments are necessary to make an inference that being sent to jail has a deleterious effect on an individual’s health.
Josiah Rich (Brown University) picked up on Binswanger’s mention of incarceration being improperly used as a treatment for addiction, mentioning that the lack of actual treatment for the problem often leads to a person’s regression (and often overdose) when released. Rich suggested that Binswanger’s 10 points be revisited with a focus not just on reducing incarceration, but also on doing so in a way that will be practical from a political and public acceptance standpoint, while also highlighting the beneficial effects and minimizing the detrimental effects of reduced incarceration.
John Laub (University of Maryland) asked Evelyn Patterson to offer an alternative strategy for collecting better measures of incarceration and health. Patterson said that she believes the key is to view incarceration as a social institution, exposure to which can affect the civic and social wellbeing of those currently and formerly incarcerated. She noted that health surveys do have questions that can measure social embeddedness or civic engagement. Including questions on incarceration in national health surveys can help researchers in parsing out the effects of incarceration, especially the compounding effects of incarceration spells.
Rashida Dorsey (U.S. Department of Health and Human Services [HHS]) thanked Patterson for her suggestion and reiterated one of the objectives of the workshop—to formulate criminal justice involvement questions that can be fielded in HHS population health surveys. The limitation of this approach is that these surveys are not given to currently incarcerated individuals, and the percentages of the survey samples that have had any criminal justice contact are very small. Patterson responded with the recommendation that questions be asked not just about a respondent’s experience, but also about the household’s exposure to the criminal justice system—i.e., proxy questions. She added that cognitive testing is required before fielding criminal justice questions in a health survey to understand respondent comprehension and recall.
Matsueda asked workshop participants who have experience conducting surveys and analyzing survey results to comment on what key data points are missing and what criminal justice experience questions should have priority for inclusion. Amanda Geller (New York University) pointed out the utility of linking survey data to administrative records. Speaking from her experience with the Fragile Families Study (see Chapter 7), getting consent from survey respondents to link their data to administrative records will complement the information from surveys. Administrative records are better at documenting duration and spells of incarceration than a survey, which relies on respondents’ memory.
Christopher Uggen (University of Minnesota) referred to Wendy Manning’s presentation (see Chapter 2) that highlighted the importance of disaggregating jail and prison experiences. Jail time may not substantially alter a life trajectory, while incarceration in prison can have a myriad of adverse effects, including loss of employment and increased risk of suicide and, for juveniles, increased risk of victimization. He supported Patterson’s suggestion that duration is a key factor, and questions should be crafted in a manner that will aid people’s recollection. Manning (Bowling Green State University) countered that while the two incarceration types differ vastly in design, prison overcrowding has resulted in longer jail stays for people who are convicted. And while jails might be less restrictive than prisons, they lack the structure and programs that are offered to prison inmates.
Following on this point, Laub suggested that researchers consider incarceration in prison and its recency and duration first, followed by filter questions on jails and probation. He mentioned the need to better understand prevalence and provide estimates of effects of any of these experiences on health. Cantor suggested that respondents may have an easier time recalling the duration of the stay than they will in differentiating their stays in a jail or a prison. Binswanger emphasized the importance of understanding the effects of the transition to different levels of criminal justice involvement when considering health outcomes.
The workshop participants discussed opportunities for combining resources and data linkages across surveys. Wang suggested that, as a start, surveys could ask about arrest history and compare the response rate on that question to the incarceration rate. She also reminded workshop participants that although surveys like the National Health and Nutrition Examination Survey (NHANES) do not specifically ask criminal justice questions, they do ask items that provide information on other social determinants of health, such as enrollment in the food stamp program. By elaborating on current questions (e.g., asking if a family has ever been banned from food stamps),2 one can get information on the collateral consequences of criminal justice involvement.
Ann Carson (U.S. Department of Justice) described her agency’s efforts to collect health data on incarcerated populations through collaboration: with the National Survey of Drug Use and Health, by including modules on risk, family, and social engagement for those who answer yes to probation or parole question; with the National Center for Health Statistics, to change the National Death Certificate Database to report if a death occurred in a
2 According to the Pew Charitable Trust Legal Action Center, 6 states impose a full ban on food stamps for individuals convicted of a drug felony, and 24 other states impose a partial ban. (The District of Columbia and Puerto Rico were not included in this measure.)
correctional facility. She added that BJS has redesigned the inmate surveys to include modules on mental health.
Nagin asked survey practitioners if physiological measures can be collected from prison inmates. Kathryn Porter (National Center for Health Statistics) and Carson described the logistical challenges in collecting biological specimens in criminal justice institutions: NHANES uses mobile clinics; BJS relies on health care facilities in jails and prisons.
Jordyn White (Committee on National Statistics) reminded the workshop participants that space on existing national health surveys is limited, and as such the number of criminal justice questions that can be realistically included may well be less than what Wildeman proposed. She also encouraged participants to think about question priorities: a question asking about currently incarcerated family members may be more pertinent than one for respondents and other current household members who have had criminal justice experience, as an absent family member has a more pervasive effect on the family.
Massoglia pointed out that the desired answers will determine the proposed questions. Given resource limitations, it is important to understand what effects are of most interest to the research and policy communities. The questions need to be phrased according to research and policy priorities. Dorsey and Alexis Bakos responded that their agencies aim for better understanding of individual, family, and community effects, in that order of priority. With the awareness that formulation of question(s) depends on the constitution of the survey and respondent burden, the agencies are open to the possibility of rotating modules and linking across surveys.