Introduction

Worldwide, more than 10.3 million people are in prison at any given time [1], and an estimated 30 million people move through prisons annually [2]. This population experiences a disproportionate burden of morbidity and mortality [3, 4], which is particularly large at the time of release from prison [2, 5–8].

The United Nations Nelson Mandela Rules define States’ minimum obligations to provide equitable health care in prison and to attend to aftercare at the time of prison release [9]. Data on health care utilization may be used to indicate access to health care and population health status for this population. Several studies have determined that in prison, people have high rates of ambulatory care utilization [10–14] and hospitalization [13–16]. Research similarly shows that after release from prison, people have high rates of primary care use [17, 18], emergency department use [19–22], and hospitalization [5, 20, 21, 23], though findings regarding ambulatory care use are inconsistent [20, 21]. Several studies found that compared to use by the general population, ambulatory care use was higher in prison [10], and emergency department use and hospitalization were higher post-release [5, 22, 23]. However, most studies conducted to date have limitations that challenge their internal and external validity, including small or select samples [5, 19–21, 24], the lack of general population comparator groups [11, 12, 14, 15, 17–21, 24, 25], the use of self-reported data [15, 17–20, 25], and the use of data from over 20 years ago [11, 13]. Further, no studies have examined health care utilization both in prison and on release.

In Ontario, Canada, there is a universal, publicly funded health care system, in which the provincial government pays for physician and hospital services [26] in provincial prison as well as in the community. Access to provincial administrative health data provide a unique opportunity to comprehensively assess health care access for the population of persons who experience imprisonment, specifically in a wealthy country with a high performing health care system [26] and without the barriers to health care on prison release that may exist in countries that lack universal health insurance coverage [27].

In this study, we aimed to describe health care utilization rates for people released from provincial prison in Ontario in 2010, and to compare health care utilization with the general population.

Methods

Results

Of 53,955 persons released from provincial prison in Ontario in 2010, we achieved valid linkage for 52,546 (97.4%) (S1 Fig). For persons released for at least one day in 2010, there were significant differences between those linked and not linked to an IKN (see S1 Table), with an older median age for those linked, longer median length of stay for those linked, and different distribution of self-reported race, but no difference in sex. We excluded an additional 3,685 persons who were not released to the community in 2010. We matched each person in the prison release group with four age- and sex-matched controls, for a total of 195,444 general population controls.

Table 1 shows characteristics of persons in the prison release group and general population controls.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of persons released from provincial prison in Ontario in 2010 and general population controls.

https://doi.org/10.1371/journal.pone.0201592.t001

Compared to general population controls, persons in the prison release group had a higher prevalence of all conditions examined, except for hypertension (see Table 2). The relative rate of mental disorders ranged from 5.23 for substance use disorders to 11.0 for schizophrenia for the prison release group compared to the general population. In the prison release group, the HIV prevalence was 3.84 times higher and the COPD prevalence was 2.20 times higher than in the general population controls. Persons in the prison release group also had a higher number of ADGs, as indicated by the median number and proportion of people with 5–9 or 10 or more ADGs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Health status of persons released from provincial prison in Ontario in 2010 and general population controls.

https://doi.org/10.1371/journal.pone.0201592.t002

In the prison release group, ambulatory care utilization and psychiatric hospitalization rates were higher in prison than in the three months after release (Table 3). In contrast, the rates of emergency department use and medical-surgical hospitalization were lower in prison than in the three months after release. Comparing the time in prison to the week after release, ambulatory care utilization decreased and the rate of emergency department utilization, medical-surgical hospitalization, and psychiatric hospitalization all increased substantially. Rates of ambulatory care utilization were lower in all time periods post-release and rates of emergency department utilization and medical-surgical hospitalization were higher in all time periods post-release compared to in prison, while psychiatric hospitalization rates decreased over the 3 months post-release, reaching rates in post-release days 30–89 that were lower than rates in prison.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Health care utilization for persons released from provincial prison in Ontario in 2010 (N = 48,861) and general population controls (N = 195,444), by health care type and period relative to time in prison^*.

https://doi.org/10.1371/journal.pone.0201592.t003

Compared to general population controls, the rate of health care utilization was significantly higher for the prison release group across all time periods and types of health care, as shown in Table 3 and Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Rate ratio* of health care utilization of persons released from provincial correctional facilities in Ontario in 2010 (N = 48,861) and age- and sex-matched general population controls (N = 195,444), by health care type and period relative to time in prison† *Adjusted for neighbourhood income quintile and rurality.

†Or the corresponding dates for general population controls.

https://doi.org/10.1371/journal.pone.0201592.g001

Discussion

In this population-based study, we found that people who experience imprisonment had significantly elevated rates of health care utilization while in prison and after release and for all types of health care examined. Rates of utilization of ambulatory care and emergency departments and of hospitalization for medical-surgical and psychiatric reasons were high compared to age- and sex-matched general population controls. There were substantial changes in rates of health care utilization at the time of release, with a large decrease in ambulatory care use and increase in emergency department use and hospitalization in the week after release.

Our study findings on health care utilization in prison are consistent with prior data on ambulatory care, which identified use rates of 6 to 20 visits per year [10, 11, 13–15], and hospitalization in Canadian and UK studies [13, 40]. Emergency department use rates in prison were several fold higher in prison than those in a 1999 Canadian study [40]. Post-release, relative rates of emergency department use and hospitalization were increased, similar to recent US studies of persons with recent criminal justice system involvement [5, 22, 41]. The increased comorbidity in the prison release group compared to the general population was also consistent with other research [3, 4].

Strengths of this study include our comprehensive assessment of health care utilization: in prison, on release, and across several types of health care. Our sample was population-based and the sample size was large. We achieved a high rate of linkage between correctional data and health administrative data, at 97.4%. Instead of relying on self-reported health care use, we accessed health administrative data, which are comprehensive for Ontario residents in the setting of universal health care. We compared rates of use with several age- and sex-matched general population controls for each person in the prison release group.

This study has several potential limitations. For health care in prison, we were not able to discern if ambulatory care visits were for administrative reasons only, or the degree to which Ministry-required physician assessments were captured in health administrative data. Similarly, on release, some persons may be obligated to participate in treatment for mental illness or substance use disorders, for example, through court-ordered drug treatment, and we were not able to distinguish care for those purposes. The data did not specify whether people were released from prison for compassionate reasons, which may be associated with health care use on release such as medical-surgical hospitalization. However, the population age distribution and the rarity of compassionate release in Ontario provincial prison make it unlikely that this would substantially affect our findings. Indigenous persons are overrepresented in provincial prisons, and health care utilization on First Nations is not captured in provincial health administrative data, which may mean that health care utilization rates post-release underestimate use; this would lead to a conservative bias. Administrative data may misclassify comorbidity status as well as socio-demographic variables. We note the substantial proportion of persons in the prison release group with missing race data (9.2%), which could bias the estimates of the proportions in each race category, and the proportion of persons with missing neighbourhood income quintile (4.7%), which is lower than expected given the frequency of homelessness at the time of prison release [42] and may therefore not accurately indicate geographic location. The prevalence of comorbidities may be inaccurate, including for mental disorders since we used diagnostic codes that have not been validated, and for HIV since provincial prisons do not routinely offer screening for bloodborne infections. We did not examine the independent effect of imprisonment on health care utilization, as this was not our study objective; a different analytic strategy would be required to explore that issue. We did not look at indicators of quality of care [43–45], as this was also beyond the scope of this study, however, this information would be relevant to understanding how well health care services in prison and in the community met general and disease-specific standards of care. The study results may not be generalizable to other jurisdictions, for example those with different health care systems or with different patterns of morbidity and mortality in prison or the general population.

Our results show that even in the context of universal health care and with high rates of ambulatory care use, rates of emergency department use and hospitalization remain very high for this population. The high rates of admission to hospital necessarily reflect the high morbidity of this population, since only people with substantial illness are admitted to hospital. It is unclear to what extent suboptimal access to quality health care may contribute to high morbidity and to emergency department use and hospitalization, for example whether people are receiving the health care they need in prison and on release to meet their needs in prison and to support continuity of care. Further work should examine ambulatory care access and quality for this population, especially in light of prior research showing that for people with chronic illness, ambulatory care may focus on specific conditions at the cost of other health conditions and preventive care [46].

A novel finding of this study is the difference in health care utilization rates in prison and post-release. Compared to use after release, the relatively high rate of ambulatory care use in prison could be appropriate for patients’ preferences and high needs, or could reflect increased need in prison due to distress related to imprisonment [10] or people not attending to their health needs after release because of competing priorities, reluctance to disclose that they were imprisoned to community-based care providers, or a lack of access to care [13]. Administrative issues may also contribute to use in prison, such as institutional requirements for a clinical assessment or to see a physician for care that is available in the community without a physician [10, 13], for example over-the-counter medications. Though ambulatory care utilization decreased substantially on release, the rates in the prison release group remained high compared to the general population, which is notable given that people face many challenges on release that may compete with attending to health needs and accessing health care, including housing access, relapse to substance use, and family issues [47–52]; this suggests people may have urgent unmet health care needs at the time of release.

Increased rates of emergency department use and hospitalization on release may reflect changes in health behaviours such as substance use, which drives high mortality on release [8, 53], worsening health status because of medication non-adherence or lack of housing on release, or a lack of adequate access to primary care [13] for routine care including prescriptions. In prison, health care providers function as gatekeepers to the emergency department, which may prevent emergency department use in prison, whether appropriately or inappropriately. Finally, hospital-based health care providers may make different decisions regarding whether to admit a patient to hospital based on whether a person is in prison or in the community, for example based on perceptions about access to monitoring and care. In particular, the substantial change in psychiatric admission rates may illustrate such differential access to care, but could also reflect acute worsening illness, for example due to medication non-adherence or substance use. Additional research is needed to understand both the reasons for high use and differences in rates of use on release, with a particular focus on continuity of care for persons with specific diseases such as mental illness, substance use disorders, HIV, and other chronic diseases, and, if indicated, evidence-based strategies should be implemented to support persons with comorbidity in achieving continuity of care [18, 21, 54–57].

The large increase in emergency department use and hospitalization on prison release indicates the need to focus on health care and health status at the time of release from prison, consistent with the obligations defined in the Nelson Mandela Rules [9]. Even in the context of a wealthy country such as Canada with universal public health insurance, we need to consider ways to change the health care system to address the needs of this population. We should advance work to improve appropriate access to and acceptability of health care for this population, building on evidence in this field [18, 21, 54–56], such as linkage with appropriate health care and social supports at the time of release from prison. Interventions should focus on health care in prison and in the community at the health system, provider, and patient levels, with explicit attention to supporting people with multi-morbidity and social complexity.

Supporting information

Acknowledgments

This study was supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). We acknowledge the Ontario Ministry of Community Safety and Correctional Services (MCSCS), which provided data for the study, and we appreciate the contributions of Mr. Michael Kirk to the study.

Parts of this material are based on data and/or information compiled and provided by CIHI. However, the analyses, conclusions, opinions and statements expressed in the material are those of the authors, and not necessarily those of CIHI.

The analyses, opinions, results and conclusions reported in this paper are also independent from the other sources that provided data and funding. No endorsement by ICES, the MOHLTC, or the MCSCS is intended or should be inferred.

We thank Dr. Gordon Guyatt for providing comments on an earlier draft of this manuscript.