Pregnancy Outcomes in US Prisons, 2016–2017

METHODS

From May 2016 to December 2016, state prison systems and Federal Bureau of Prisons (BOP) institutions were recruited through purposive and snowball sampling. Using demographic data published by the BJS, we targeted initial recruitment to the 18 state prison systems housing at least 2000 women. Custody and health care administrators at these state prison systems were contacted via e-mail or telephone and invited to participate. As a means of broadening the sample, networks established through the National Institute of Corrections and the National Resource Center on Justice Involved Women were used to notify prison systems housing fewer than 2000 women. BOP administrators were contacted, and the acting director of the Department of Justice at the time approved BOP institutions’ participation in reporting pregnancy data.

Both state and federal prisons house people sentenced to serve terms typically longer than 1 year, so most imprisoned women will be in custody for the duration of their pregnancies. Otherwise, state and federal prisons differ in their administrative structure and oversight, geographic locations, and types of convictions. Pregnancy care at federal prisons must follow BOP policies, whereas each state prison falls under the jurisdiction of the state’s department of corrections policies.

Twenty-two state prison systems and all 26 BOP prisons housing women enrolled in the study (see the box on this page; Figure A, available as a supplement to the online version of this article at http://www.ajph.org). Study reporting occurred between May 2016 and January 2018, with the exact 1-year time frame for each prison based on when it was enrolled. According to 2016 BJS prison census data, the study prisons included 57% of all women in US prisons (53% of women in state prisons and 86% of women in federal prisons).² In 17 of the 22 state systems, pregnant people were housed at 1 facility. In 4 states more than 1 prison housed pregnant women, and their study reporting reflected all pregnant women in the state. Wisconsin housed pregnant women at 4 locations, but data were reported for only a single prison. BOP reporting excluded outcomes among 25 women who were transferred to privately run federal prisons. Rhode Island and Vermont combine jail and prison populations (with jail populations being predominantly of pretrial status and serving shorter sentences), and midway through the study Maryland began housing pregnant pretrial women in prisons; thus, these states’ data reflect both convicted and pretrial populations.

Designated reporters at the different sites (medical directors, other health care personnel, prison research coordinators, women’s programming coordinators, and, in one instance, a warden) reported data at the end of each month. We provided sites with examples of tracking systems, as the majority of the participating prisons had no existing process for tracking pregnancy data. Data were reviewed monthly by study staff to check for discrepancies; site reporters were contacted directly to resolve inconsistencies. Midway through the study, we approached all of the site reporters for an interview to ensure that we could understand their tracking system and to ensure accuracy of reporting; 43% of site reporters responded and were interviewed, and all were reporting accurately.

The secure, Web-based REDCap (research electronic data capture) application was used to collect and manage study data.²⁶ Site reporters could also complete electronic documents and return them via e-mail, fax, or mail. Data were collected for 13 consecutive months with the exception of Louisiana, which temporarily suspended reporting from August 2016 to February 2017 as a result of flooding and then resumed. Thirteen months of data were collected because we assumed that the first month would be a trial period for pregnancy tracking systems. Analyses were performed on data from months 2 to 13 (a total of 12 months).

Prisons reported the following aggregate, deidentified numbers each month: pregnant women admitted and in custody on the last day of the month; live births and stillbirths, stratified by term (more than 37 weeks), preterm (24 weeks to 36 weeks, 6 days), and very early preterm (20 weeks to 23 weeks, 6 days); preterm and cesarean deliveries; miscarriages; abortions; ectopic pregnancies; maternal deaths during custody (deaths during pregnancy or within 6 weeks of the pregnancy ending); and newborn deaths within the first 3 days of life. This differs from the standard definition of neonatal death within the first 30 days because prison systems could consistently know what happened to the infant only during the time the mother was in the hospital for her postpartum recovery. Information was also collected on pregnancy testing policies and on whether the prison’s health care system was privatized or accredited.

As a result of the nature of aggregate data collection from each prison system, we were not able to gather information on individual women’s specific demographic characteristics (e.g., race and age). We also could not assess whether any of these pregnancies occurred in transgender men or non-binary individuals. Prisons also reported their overall female census and total number of admitted women on December 31, 2016 (consistent with BJS reporting), and on July 31, 2016, so that we could calculate the proportion of imprisoned women who were pregnant and the proportion of admitted women who were pregnant at 2 points in time; July 31 census data were available only for 16 states and admission data for 13.

Data were analyzed for frequencies and other descriptive statistics. Proportions of each pregnancy outcome were calculated for pregnancies in which the outcomes were known; that is, outcomes did not account for women who were released from prison while they were still pregnant or those who were in custody but still pregnant at the end of the study reporting period. We performed analyses for all prisons combined and also broken down by state and federal jurisdiction. Because the data represent complete counts from all state prison systems (with the exception of Wisconsin), we were not able to calculate confidence intervals.

RESULTS

The participating state prisons represented a range of small, medium, and large prisons holding between 132 and 12 711 women, with most sites (54%) housing 500 to 2000 women (Table 1). Sixty-four percent of prisons had some form of voluntary health care accreditation, and 50% contracted health care delivery to a private corporation. In total, 1396 pregnant women were admitted to these state and federal prisons over the 12-month study period (with a range of 0–26 women admitted each month at individual prison systems; Table 2).

At the 13 state prisons for which data on total female admissions were available, 65 of 1646 women (3.9%) in July 2016 and 63 of 1654 (3.8%) in December 2016 were pregnant at the time of their admission (with a range of 0% to nearly 17% of admissions in individual states; Table 2). The overall pregnancy prevalence at these sites on July 31, 2016, was 0.9%. The 2016 year-end pregnancy prevalence at all state and federal prisons was 0.6% (Table 2). Among women who were already in state prisons, 5 new pregnancies were diagnosed during a 6-month reporting period. Three of these women became pregnant during work release; the timing of the other 2 pregnancies was not reported.

Overall, there were 753 live births, representing 92% of all known pregnancy outcomes (Table 3); 685 births occurred among women incarcerated in state prisons and 68 among women incarcerated in BOP sites. Among live births in state prisons, 6% were preterm and 0.3% were very early preterm. The majority of live births were vaginal, with 32% cesarean deliveries. All but 6 births occurred in a hospital; 3 were attributable to precipitous labor with prison nurses or paramedics in attendance, and details were not available for the others. Of the remaining 8% of non-live-birth pregnancy outcomes, 6% were miscarriages (n = 46), 1% were abortions (n = 11), 0.5% were stillbirths (n = 4), and 0.25% were ectopic pregnancies (n = 2). The 4 stillbirths occurred among women who delivered preterm. There were 3 newborn deaths and no maternal deaths. Table 3 presents data on selected outcomes according to state and federal jurisdiction.

More miscarriages were reported at state prisons that administered pregnancy tests within 48 hours of admission (n = 33) than at sites that conducted pregnancy tests only at the woman’s request or at the clinician’s discretion (n = 9 miscarriages), suggesting higher detection of early pregnancy outcomes.

DISCUSSION

To our knowledge, this study is the first systematic investigation to collect prospective data on pregnancy frequencies and outcomes among women in US prisons. Our study demonstrates that it is feasible to prospectively track pregnancy outcome data among incarcerated women. The vast majority of pregnancies (92%) ended in live births. Approximately 4% of women admitted to state prisons were pregnant, a percentage similar to the figure reported in the 2004 BJS survey that included self-reported pregnancy status at intake.¹⁷ This represents a notable proportion of women who will arrive pregnant and will have pregnancy-specific health care needs that must be addressed in a timely fashion. Collection of data on pregnancy frequencies and outcomes among women in prison, who are not accounted for in existing databases, is part of a broader public health strategy of tracking maternal health data to reduce adverse outcomes and promote equity.²⁷

The overall prevalence of pregnancy on December 31, 2016, was 0.7% at all state sites and 0.3% at all federal prisons. This overall pregnancy prevalence differs from the proportion of admitted women who were pregnant because the former encompasses in the denominator all women in prison, including those who have been there for many years and are less likely to become pregnant because of their confinement and age; women newly admitted to prison, in contrast, were more recently in the community, where they may have become pregnant.

It is notable that there was significant variation by state. For instance, Texas and Ohio had some months when there were more than 50 pregnant women in their prisons on the last day of the month, whereas other states had no incarcerated pregnant women; this is likely related to Texas having the largest and Ohio the fourth largest populations of incarcerated women in the United States.² Likewise, the percentages of newly admitted women who were pregnant in July and December varied widely by state, with percentages as high as 17% in Maryland and as low as 0% in Maine. Of note, Maryland’s July to December 2016 increase in the proportion of admitted pregnant women was probably attributable to the previously mentioned policy change according to which some pretrial pregnant women were housed at the state prison. Overall, miscarriages represented 6% of known pregnancy outcomes; these percentages also varied by state, however, with miscarriages representing 20% or more of pregnancy outcomes in some states.

Numerous factors likely contribute to these state-by-state variations in pregnancy frequency and miscarriage, including state sentencing laws, prison health care policies, community reproductive health care access, individual patient attributes, and other variables that our study did not measure.

It is useful to compare our study results with national pregnancy statistics for the general population; however, such a comparison is limited by differences in the ways in which national statistics are collected and our study statistics were gathered. The National Survey of Family Growth (2011–2013) reported that 5% of women 15 to 44 years old were pregnant or postpartum at the time of their interview.²⁸ To estimate a proportion of imprisoned pregnant women from our study with a similar age denominator, we can assume, on the basis of BJS data, that 75% of all women at Pregnancy in Prison Statistics study sites were 18 to 44 years old in 2016²; this yields a pregnancy prevalence of 0.8%. The national percentage was higher than our study’s prison pregnancy prevalence; however, our study did not include postpartum women in the denominator.

The general fertility rate, a measure of the number of live births among women 15 to 44 years of age, was 62 per 1000 women in the United States in 2016.²⁹ For our study data, we can roughly estimate a general fertility rate of 18 per 1000 imprisoned women 18 to 44 years old. Although this estimate is understandably lower than national rates, it is still notable, and it speaks to the need to address the numerous complexities of birth in custody, such as the medically unsafe practices of placing pregnant women in solitary confinement and shackling women in labor, ensuring proper pregnancy and postpartum care, and determining who will care for the infants born to mothers in custody.^11,30

Nationally, nearly 10% of live births in 2016 were preterm,²⁹ as compared with 6% of live births in prisons in our study. This lower preterm birth proportion in these prisons may be partially related to the relative presence of prenatal care, food, and shelter and the limited access to illicit substances, conditions that may be different for some pregnant women not in custody; however, this explanation should be considered cautiously given the variability in access to and quality of prenatal care from prison to prison. In addition, some state prison systems had preterm birth rates that exceeded the national rate, suggesting that the context of the individual prison system and preincarceration conditions may play a role.

Our relative preterm birth rates should further be considered with caution given the problematic implication that prisons—designed for punishment, not health care—may somehow exert a “protective” effect on pregnancy. Such an interpretation fails to consider the broader social and structural determinants of health that differentially affect people before, during, and after incarceration.¹²