Quality of Surgical Care Within the Criminal Justice Health Care System

This cohort study reviews perioperative outcomes for patients incarcerated within the Texas Department of Criminal Justice who underwent a general or vascular surgery procedure at the University of Texas Medical Branch between 2012 and 2021.

Key Points

Question

What are the perioperative outcomes of surgical patients managed within the criminal justice health care system?

Findings

This cohort study assessed data from a sample of 6675 patients who were incarcerated and underwent a general or vascular surgery procedure between 2012 and 2021 and found that 30-day morbidity, mortality, and readmission rates were comparable with those in a nonincarcerated population treated at the same institution.

Meaning

Findings of this study suggest that patients who are incarcerated and undergo surgery have overall acceptable perioperative outcomes as demonstrated by equivalent rates of mortality and readmission compared with the general population.

Abstract

Importance

Individuals who are incarcerated represent a vulnerable group due to concerns about their ability to provide voluntary and informed consent, and there are considerable legal protections regarding their participation in medical research. Little is known about the quality of surgical care received by this population.

Objective

To evaluate perioperative surgical care provided to patients who are incarcerated within the Texas Department of Criminal Justice (TDCJ) and compare their outcomes with that of the general nonincarcerated population.

Design, Setting, and Participants

This cohort study analyzed data from patients who were incarcerated within the TDCJ and underwent general or vascular surgery at the University of Texas Medical Branch (UTMB) from 2012 to 2021. Case-specific outcomes for a subset of these patients and for patients in the general academic medical center population were obtained from the American College of Surgeons National Quality Improvement Program (ACS-NSQIP) and compared. Additional quality metrics (mortality index, length of stay index, and excess hospital days) from the Vizient Clinical Data Base were analyzed for patients in the incarcerated and nonincarcerated groups who underwent surgery at UTMB in 2020 and 2021 to provide additional recent data. Patient-specific demographics, including age, sex, and comorbidities were not available for analysis within this data set.

Main Outcome and Measures

Perioperative outcomes (30-day morbidity, mortality, and readmission rates) were compared between the incarcerated and nonincarcerated groups using the Fisher exact test.

Results

The sample included data from 6675 patients who were incarcerated and underwent general or vascular surgery at UTMB from 2012 to 2021. The ACS-NSQIP included data (2012-2021) for 2304 patients who were incarcerated and 602 patients who were not and showed that outcomes were comparable between the TDCJ population and that of the general population treated at the academic medical center with regard to 30-day readmission (6.60% vs 5.65%) and mortality (0.91% vs 1.16%). However, 30-day morbidity was significantly higher in the TDCJ population (8.25% vs 5.48%, P = .01). The 2020 and 2021 data from the Vizient Clinical Data Base included 629 patients who were incarcerated and 2614 who were not and showed that the incarcerated and nonincarcerated populations did not differ with regard to 30-day readmission (12.52% vs 11.30%) or morbidity (1.91% vs 2.60%). Although the unadjusted mortality rate was significantly lower in the TDCJ population (1.27% vs 2.68%, P = .04), mortality indexes, which account for case mix index, were similar between the 2 populations (1.17 vs 1.12).

Conclusions and Relevance

Findings of this cohort study suggest that patients who are incarcerated have equivalent rates of mortality and readmission compared with a general academic medical center population. Future studies that focus on elucidating the potential factors associated with perioperative morbidity and exploring long-term surgical outcomes in the incarcerated population are warranted.

Introduction

People who are incarcerated have substantial protections in place surrounding their participation in medical research. This is largely due to the historical mistreatment sustained by this population, including the toxic pharmaceutical testing within Pennsylvania’s Holmesburg Prison and the Oregon and Washington radiation experiments.^1,2 While these strict limitations are intended to stop potential exploitation, they also present a barrier to understanding the quality and type of medical care provided in these institutions. As laid forth in the 1976 Supreme Court decision Estelle v Gamble, patients who are incarcerated have a right to health care since denying them such would be in violation of the US Constitution’s Eighth Amendment.³ Although not all physicians or surgeons will care for the incarcerated population, approximately 600 000 individuals are released from US prisons annually and an additional 9 million circulate through local jails.⁴ Patients who could not access or did not receive adequate health care while incarcerated will eventually seek care within the community, which has implications for health care professionals outside of the correctional system.

There is limited available information regarding medical care within the prison system, and even less is known about patients who undergo surgery while incarcerated. Much of the literature surrounding the correctional health care system is written from the public health perspective and discusses treatment of communicable diseases (eg, HIV or hepatitis C) and access to mental health services, with a notable lack of data on surgical treatments and outcomes.^5,6 To our knowledge, the largest studies of patients who are incarcerated and treated by a surgical team have focused on trauma and acute care surgery, with studies conducted by the Eastern Association for the Surgery of Trauma and using large trauma databases.^7,8,9 While these studies provided important insight, a considerable proportion of surgical patients, such as those who present for less urgent or elective procedures, remains excluded. This study sought to evaluate quality outcomes from a broader range of surgical patients and subspecialties by using data from one of the largest correctional hospitals in the US, while also comparing these metrics to institutional and national benchmarks.

Methods

Patient Population and Measures

This retrospective cohort study included patients aged 18 years or older within the Texas Department of Criminal Justice (TDCJ) system who underwent a general or vascular surgery procedure at the University of Texas Medical Branch (UTMB) from September 1, 2012, to July 1, 2021. Although the inpatient rooms and holding areas for patients within the TDCJ system are located in Hospital Galveston, a building separate from the hospital used by the nonincarcerated population, all surgical procedures are performed in UTMB operating rooms with the same academic surgical staff as the nonincarcerated population. This study was approved by the UTMB Institutional Review Board and the TDCJ with a waiver of written informed consent because only deidentified data were used. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Data were extracted from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database. The ACS-NSQIP samples from the overall surgical population by analyzing the first 40 cases at a hospital during 8-day periods. Information on 30-day morbidity, mortality, and readmissions was available for analysis. Morbidity was further stratified into specific postoperative complications (eg, pneumonia, acute renal failure). These values from the TDCJ population were compared with the 2021 metrics of the entire UTMB academic medical center (AMC), which was largely composed of patients who were not incarcerated. Due to the summative nature of reporting ACS-NSQIP outcomes at our institution, the TDCJ population could not be entirely excluded from that of the AMC but comprised no more than 20% of the total cases. Current Procedural Terminology codes were also analyzed to determine which procedures had been performed.

We also obtained 2020 to 2021 data on quality outcomes from the Vizient Clinical Data Base, including 30-day mortality, morbidity, length of stay, and readmission. Vizient data included information on all principal general or vascular surgery procedures (ie, operations associated with the primary diagnosis) and were more inclusive of surgical cases compared with the ACS-NSQIP’s sampling method. Mortality index (observed to expected mortality) and length of stay (observed to expected length of stay) were analyzed using the Vizient risk model. Quality metrics using inpatient performance data were compared between all patients in the TDCJ system and nonincarcerated patients who underwent a general or vascular surgery procedure. We were able to directly compare the TDCJ and nonincarcerated populations using data from the Vizient Clinical Data Base, in contrast to the ACS-NSQIP data, which included the entire AMC population.

Statistical Analysis

The ACS-NSQIP database used representative sampling across the overall cohort, while the Vizient Clinical Data Base captured all principal procedures performed by general and vascular surgery faculty. The Fisher exact test was used to compare perioperative outcomes between groups; 2-sided P < .05 was considered to be significant. Because the outcomes of interest were categorical (yes/no), no measures of variability were calculated. The statistical analysis was performed with Prism, version 9.5.1 (GraphPad by Dotmatics).

Results

The sample included data from 6675 patients who were incarcerated and underwent a general or vascular surgery procedure at UTMB from 2012 to 2021. The procedures performed in patients who were incarcerated represented approximately 20% of UTMB’s total surgical volume over that period. Data from the ACS-NSQIP were available for 2304 of these patients who were incarcerated and for 602 patients in the general AMC population. Patients who were incarcerated did not differ significantly from the general AMC population with regard to 30-day mortality (0.91% vs 1.16%) and readmission rates (6.60% vs 5.65%) (Table 1). However, the 30-day morbidity rate was significantly higher for patients who were incarcerated compared with the AMC population (8.25% vs 5.48%, P = .01). The most common postoperative complications in the incarcerated population were superficial surgical site infection (SSI; n = 91), ventilation for more than 48 hours (n = 22), unplanned intubation (n = 21), and pneumonia (n = 21). Although not statistically significant, the rates of these complications were higher when compared with those in the AMC population (SSI, 3.95% vs 2.48%; ventilation for >48 hours, 0.95% vs 0.62%; pneumonia, 0.91% vs 0.16%; and unplanned intubation, 0.91% vs 0.31%).

Table 1. Comparison of American College of Surgeons National Surgical Quality Improvement Program Quality Metrics.

Quality metric	Patients, No. (%)		P value
	In the TDCJ system (n = 2304)	In the AMC (n = 602)
30-d Readmission	152 (6.60)	34 (5.65)	.63
30-d Morbidity	190 (8.25)	33 (5.48)	.01
30-d Mortality	21 (0.91)	7 (1.16)	.45

Abbreviations: AMC, academic medical center; TDCJ, Texas Department of Criminal Justice.

The ACS-NSQIP data for the TDCJ population was further analyzed according to the most frequently performed procedures as reported by Current Procedural Terminology code. The most common operations among patients who were incarcerated, in descending order, were open inguinal herniorrhaphy, open ventral herniorrhaphy, open cholecystectomy, lower extremity endovascular revascularization (which included angioplasty, arthrectomy, and stent placement), and laparoscopic cholecystectomy (Table 2). These procedures encompassed 1397 cases. There were no deaths within 30 days after open inguinal herniorrhaphy, open cholecystectomy, and laparoscopic cholecystectomy; there were 2 deaths (0.53%) after open ventral hernia repair and 4 (3.08%) after lower extremity endovascular revascularization. Laparoscopic cholecystectomy had the lowest 30-day morbidity rate of 0.78%, followed by open inguinal herniorrhaphy (1.46%), open ventral herniorrhaphy (6.93%), open cholecystectomy (10.27%), and lower extremity endovascular revascularization (24.62%). Thirty-day readmission rates ranged from 2.27% for open inguinal herniorrhaphy and laparoscopic cholecystectomy to 15.38% for lower extremity revascularization.

Table 2. Perioperative Outcomes From the American College of Surgeons National Surgical Quality Improvement Program for the Most Common Procedures Among Patients Who Were Incarcerated.

Procedure	Total No. of procedures	No. (%)
		30-d Readmission	30-d Morbidity	30-d Mortality
Open inguinal herniorrhaphy	617	14 (2.27)	9 (1.46)	0
Open ventral herniorrhaphy	375	21 (5.60)	26 (6.93)	2 (0.53)
Open cholecystectomy	146	5 (3.42)	15 (10.27)	0
Lower extremity endovascular revascularization	130	20 (15.38)	32 (24.62)	4 (3.08)
Laparoscopic cholecystectomy	129	3 (2.33)	1 (0.78)	0

The 2020 and 2021 Vizient data included 629 patients who were incarcerated and 2614 patients who were not incarcerated. There were no significant differences between the incarcerated and nonincarcerated populations in readmission rates (12.52% vs 11.30%) and morbidity (1.91% vs 2.60%) (Table 3). The unadjusted mortality rate was significantly lower for patients who were incarcerated (1.27% vs 2.68%, P = .04), but after accounting for the lower case mix index within the TDCJ population, mortality indices were similar between the TDCJ and AMC populations (1.17% vs 1.12%). The length of stay index was higher in the TDCJ population than in nonincarcerated population (1.46 vs 0.93), as was excess days per 100 admissions (44 vs 3).

Table 3. Comparison of Vizient Clinical Data Base Quality Metrics.

Quality metric	Patients, No. (%)		P value
	In the TDCJ system (n = 629)	In the AMC (n = 2614)
30-d Readmission	73 (12.52)	259 (11.30)	.42
30-d Morbidity	12 (1.91)	68 (2.60)	.39
30-d Mortality	8 (1.27)	70 (2.68)	.04
Mortality index (observed/expected)	1.17 (1.27/1.09)	1.12 (2.68/2.39)	NA
Length of stay index (observed/expected)	1.46 (7.89/5.39)	0.93 (7.16/7.68)	NA
Excess days per 100 admissions	44	3	NA

Abbreviations: AMC, academic medical center; NA, not available; TDCJ, Texas Department of Criminal Justice.

Discussion

Despite the lack of reporting within the correctional system and anticipated health care disparities among the incarcerated population, results of this study suggest that perioperative outcomes are not worse overall in patients who are incarcerated than in the nonincarcerated population. According to data from both the ACS-NSQIP and Vizient databases, the incarcerated and nonincarcerated groups had comparable rates of 30-day mortality and readmission. Analysis of data from the Vizient Clinical Data Base also showed no difference in morbidity, although analysis of the ACS-NSQIP data did show a higher 30-day morbidity rate for the incarcerated population. Increased length of stay and excess days in the prison hospital is likely a consequence of logistical constraints involving the safe transportation of patients who are incarcerated from the hospital to other facilities at the time of discharge. To our knowledge, this report represents the largest study of the quality of surgical care provided in the criminal justice health care system and thus provides insight into an area not previously explored.

The lower unadjusted mortality rate for patients who were incarcerated found in the analysis of the Vizient data may initially seem unexpected because the correctional system is often associated with limited resources and logistical challenges. However, patients who are incarcerated are on average younger than the general population,¹⁰ and the prevalence of chronic diseases (eg, hypertension, diabetes, ischemic heart disease) in the prison population may be lower.¹¹ This is consistent with our findings that the TDCJ population had a lower case mix index than the nonincarcerated population treated at UTMB. Age likely contributes to this phenomenon, but the ability to manage and treat chronic health problems in prisons may also be a factor.¹²

The greater morbidity rate for patients who were incarcerated found in the analysis of ACS-NSQIP data appears to be attributable to the higher frequency of SSI and respiratory complications. Skin and soft-tissue infections are common among the incarcerated population secondary to shared living conditions and limitations to maintaining personal hygiene^13,14; these factors are likely also associated with an increased risk for SSIs. Similarly, the higher rates of unplanned intubation, prolonged ventilation, and pneumonia may be attributable to infection as well as to baseline comorbidities. The higher rates of latent tuberculosis, smoking history, and chronic obstructive pulmonary disease among the incarcerated population have been well established and may contribute to these numbers.^15,16,17

With regard to the most commonly performed surgical procedures, rates of perioperative morbidity and mortality within the TDCJ population were largely comparable with published data. The overall complication rate for an open inguinal hernia repair is estimated to be 2% to 8%, with mortality reported as 0.1% to 0.2%.^18,19,20,21 For open ventral herniorrhaphy, previous ACS-NSQIP and Veterans Affairs Surgical Quality Improvement Program studies indicate an overall morbidity rate of approximately 6% and a mortality rate of 0.3% to 0.5%.^22,23 Perioperative morbidity and mortality rates for an open cholecystectomy of 18.1% and 2.5%, respectively, have been reported, and a laparoscopic cholecystectomy is estimated to have a morbidity rate of 3.3% and a mortality rate of 0.3%.²⁴ Thirty-day mortality after endovascular revascularization of a lower extremity is estimated to be 1.6% to 2.9%.^25,26 Morbidity rates have considerably more variability, with rates ranging from 4.6% to 25.4%, likely related to patient comorbidities and severity of disease (ie, intermittent claudication vs critical limb ischemia).^25,27,28 A majority of corresponding values for these procedures in the TDCJ population fell within or below those published ranges. The exceptions were 30-day morbidity of open cholecystectomy (6.93%) and 30-day mortality of lower extremity endovascular revascularization (3.08%). Thirty-day readmissions were infrequently cited in the literature and were therefore not included for comparison. Within the TDCJ population, a high number of open operations was noted. This was primarily due to surgeon preference rather than disease severity. With regard to open inguinal and ventral herniorrhaphies in particular, a limited number of surgeons were available to perform these using a minimally invasive technique.

Taken together, these findings suggest that outcomes of surgical care provided to individuals in the TDCJ system were comparable with those observed in the general population of patients treated at the same medical institution. While it is reassuring that generally equitable quality of perioperative care is being delivered to the incarcerated population, areas of improvement were also identified, namely the greater risk of morbidity seen in the analysis of the ACS-NSQIP data.

Limitations

A limitation of this study includes lack of generalizability to other settings, such as federal prisons and local jails. When providing care to individuals incarcerated in a state prison, Hospital Galveston and UTMB abide by Texas laws and regulations, which may not apply to other locations across the country. The relationship between UTMB and TDCJ is also unique since other correctional systems may not involve academic physicians or may outsource medical care to private companies. However, the system established by UTMB offers benefits, such as improved compliance with managed care programs and fewer health care worker vacancies, and can offer a structure to be replicated elsewhere.²⁹

Another limitation was that both the Vizient and ACS-NSQIP databases used a degree of sampling and thus did not include an exhaustive list of all procedures performed at the AMC. For example, the ACS-NSQIP data included only the first 40 consecutive cases during an 8-day cycle, and no subsequent operations are recorded. The Vizient database was more comprehensive by including all principal procedures performed by general and vascular surgeons, but secondary procedures (ie, operations not associated with the primary diagnosis), if any, were excluded. Consequently, the total TDCJ surgical population comprised 6675 patients, but they were not all included in the quality analyses. An additional confounder within the ACS-NSQIP data was the inability to completely exclude patients who were incarcerated from the AMC population and potentially counting patients within the TDCJ system twice. This situation was secondary to our institution’s manner of reporting AMC outcomes as aggregate data. However, further analysis confirmed that the proportion of patients within the TDCJ system who were counted within the AMC population was at most 20%; thus, the large majority of outcomes in the AMC population reflected the nonincarcerated population.

Finally, our data set was lacking in patient-specific demographics (such as age, sex, and comorbidities) that would allow for a more direct and matched comparison. This paucity of data was primarily a consequence of extracting information from the criminal justice system. However, the available information does serve as a litmus test, and more granular data on a case-by-case basis will be assessed in future studies for the purposes of system-level quality improvement.

Conclusion

To our knowledge, this cohort study represents the largest study of perioperative surgical care within the prison system, with quality metrics demonstrating that patients who are incarcerated have comparable rates of mortality and readmission compared with the nonincarcerated general population. Perioperative morbidity was higher among patients who were incarcerated, and future studies will aim to better understand the potential drivers to serve as a basis for improvement, as well as investigate long-term outcomes.

Supplement.

Data Sharing Statement