Secondary Logo

Journal Logo

Contents: Original Research

Racial and Ethnic Disparities in Hospital Readmissions After Delivery

Aseltine, Robert H. Jr PhD; Yan, Jun PhD; Fleischman, Steven MD; Katz, Matthew MS; DeFrancesco, Mark MD

Author Information
doi: 10.1097/AOG.0000000000001090
  • Free

Hospital readmissions have emerged as a focal point in efforts to reduce the cost and improve the quality of health care.1,21,2 This attention has been accompanied by a growing body of evidence suggesting racial and ethnic disparities in readmission rates. Studies of diverse patient populations have found elevated rates of 30-day readmission among black relative to white patients for diabetes,3,43,4 acute myocardial infarction,5,65,6 heart failure,5–75–75–7 pneumonia,5,65,6 and cirrhosis8 and after ambulatory surgery.9 Higher rates of readmissions among Hispanic relative to white patients have been reported for diabetes,4 heart failure, and acute myocardial infarction.10 Rates of 30-day readmission are generally 10–20% higher among racial and ethnic minorities across these conditions.

In contrast, race and ethnic disparities in readmission associated with women's reproductive health have received scant attention. Cesarean delivery typically results in 1.5 to two times the rate of 30-day readmission as vaginal delivery.11–1511–1511–1511–1511–15 Although there is evidence that black women are more likely than white women to deliver by cesarean,12,1612,16 these studies have not examined race and ethnic differences in rates of hospital readmission after this procedure.

This study examines racial and ethnic disparities in rates of hospital readmission in Connecticut after childbirth, which accounts for 8.5% of hospital admissions in the state annually. In addition, this study is distinctive for being among the first to investigate disparities in hospital readmission among young, relatively healthy patients. Previous research examining racial and ethnic disparities in readmission have focused almost exclusively on conditions prevalent among the aged and infirm.3–83–83–83–83–83–8


We analyzed 8 years of data from Connecticut's Acute Care Hospital Inpatient Discharge Database maintained by the Connecticut Department of Public Health.17 The Hospital Inpatient Discharge Database contains patient-level demographic, clinical, and billing data for inpatient and emergency admissions at all 30 acute care hospitals.18 We conducted a retrospective analysis comparing women readmitted and not readmitted to the hospital within 30 days of delivery. Female patients 14 years of age or older with an index admission for vaginal delivery without complicating diagnoses (diagnosis-related group [DRG] 775) and cesarean delivery without complication or comorbidity (DRG 766) during the 2005–2012 Centers for Medicare and Medicaid Services fiscal years were included in the analysis. Five patients who died while hospitalized after their index admission were excluded from the study. Although cause of death was not included in the discharge data, these five patients had comorbidities that included asthma, hypertension, anemia, strep infection, and opioid dependence.

The dependent variable was a binary indicator of readmission (for all causes) to any Connecticut hospital within 30 days of discharge. Same hospital readmissions were identified based on a unique patient identifier reported to the Connecticut Department of Public Health by the Connecticut Hospital Association. Readmissions to a hospital other than the site of the delivery were identified by an identifier constructed from patient's date of birth, race–ethnicity, payer or insurance coverage, and county of residence. Women admitted to different hospitals for postpartum care and examination immediately after delivery (V240) or readmitted for codes reflecting antepartum conditions were not considered 30-day readmissions in this analysis.

Patient's race (white, black, Asian, other) and Hispanic ethnicity and other demographic characteristics (ie, age, source of payment) were obtained at the time of each admission.18 Race and ethnicity were obtained through either self- or observer report. Patient sickness was controlled using Elixhauser's comorbidity measure capturing 29 discrete comorbid or complicating medical conditions during the index admission.19 Up to 10 primary and secondary diagnoses were recorded for each hospitalization. Elixhauser items that included International Classification of Diseases, 9th Revision (ICD-9) codes capturing pregnancy related conditions—ie, hypertension, other neurologic disorders, diabetes without chronic complications, coagulation deficiency, obesity, blood loss anemia, drug abuse—were included as separate predictor variables in the model. In addition, an indicator variable for infectious conditions related to pregnancy (ICD-9 codes 647.00–60, 670.00–04, 659.30–33, 658.40–43, and 646.60–64) was also included in the model. Because of its association with patients' racial status, a separate variable for hereditary hemolytic anemias was also included, coded 1 if ICD-9 code 282.X was listed as a comorbid condition. Because of the large number of null values among the other comorbidity indicators resulting from the young age and relative health of patients giving birth, a summary index was used to control for all remaining comorbidity domains. A dummy variable capturing hospitalization for antepartum conditions before the delivery was also included, coded 1 if the patient was hospitalized for treatment or observation for pregnancy-related complications (ICD-9 640.X-649.X) and zero otherwise. Payer was a categorical measure contrasting those with private insurance and those covered by Medicaid with other sources of payment (which included self-pay or uninsured patients). A proxy measure for patient socioeconomic status was constructed using median 2010 income (in 10,000s) of the zip code for patients' primary residence.20 Patients with zip codes for which no income information was available (eg, post office boxes; n=604 for vaginal delivery, n=237 for cesarean delivery) were assigned the median income for the sample. With the exception of 113 individuals with unknown race (coded as other in the analysis), there were no missing data on any of the other variables included in the analysis. The interval scales measuring facility volume, age, year, and length of stay were centered around their respective means to facilitate interpretation of the coefficient estimates.

We analyzed the data with logistic models for clustered data using generalized estimating equations in R.21,2221,22 Generalized estimating equation is a general method for analyzing such data where observations within a cluster (ie, patients with multiple admissions) may be correlated and observations in separate clusters are independent. In our analysis, we used the inverse of the cluster size as a weight to adjust for patients who had multiple admissions. We used an exchangeable working correlation structure for possible gains in efficiency, which even if misspecified would not affect the consistency of the estimation of the marginal regression parameters.

Three nested models are presented subsequently. For each condition, model 1 presents the regression of a binary variable for 30-day readmission on patient age, race and ethnicity, year of admission, and the number of patients treated for each DRG at the admitting facility; model 2 additionally controls for length of stay, patient comorbidities, patient socioeconomic status, and patient's insurance (Medicaid, other payer compared with private payer). For both age and length of stay, quadratic terms (ie, age and length of stay squared) were included in the model to capture nonlinearities in their association with readmission. Finally, we modeled the statistical interaction between race–ethnicity and payer and race–ethnicity and time in predicting the odds of readmission.

The study was approved by the Human Investigations Committee at the Connecticut Department of Public Health and by the institutional review board of the University of Connecticut Health Center.


There were 167,857 total admissions among 139,792 unique patients for vaginal delivery without complications from 2005 to 2012 in Connecticut hospitals and 75,552 (66,410 unique patients) for cesarean delivery without complications (Table 1). Thirty-day readmission rates (for all causes) were 8.8 (confidence interval [CI] 8.4–9.3) and 15.7 (CI 14.8–16.6) per 1,000 patients, respectively. The average age of women admitted for childbirth was roughly 30 with the majority of admissions involving white women and covered by private health insurance (Table 1). Data presented in Tables 2 and 3 indicate that there were substantial race and ethnic differences in age at delivery, socioeconomic status, payer, and patient comorbidities with black and Hispanic patients significantly younger, of lower socioeconomic status, and with greater levels of comorbidities than white patients. Hospitalization for pregnancy-related complications before vaginal delivery was significantly more common among black and Hispanic patients. In addition, figures capturing the timing of readmissions after vaginal and cesarean delivery by patients' race–ethnic status are included (Figs. 1 and 2).

Table 1
Table 1:
Patient Demographics: Hospital Inpatient Discharge Data for Admissions to Connecticut Hospitals for Selected Diagnosis-Related Groups, 2005–2012
Table 2
Table 2:
Race and Ethnic Differences in Predictor Variables—Vaginal Delivery
Table 3
Table 3:
Race and Ethnic Differences in Predictor Variables—Cesarean Delivery
Fig. 1
Fig. 1:
Cumulative rates of readmission by race and ethnicity after vaginal delivery (diagnosis-related group 775).Aseltine. Disparities in Readmissions After Childbirth. Obstet Gynecol 2015.
Fig. 2
Fig. 2:
Cumulative rates of readmission by race and ethnicity after cesarean delivery (diagnosis-related group 766).Aseltine. Disparities in Readmissions After Childbirth. Obstet Gynecol 2015.

International Classification of Diseases, 9th Revision codes for readmissions presented in Appendix 1 (available online at indicate that the primary reasons for readmission varied by type of delivery. The most common reasons for readmission after vaginal delivery were hypertension, infection, and postpartum hemorrhage. Reasons for readmission after cesarean delivery were generally similar, although the most common reason, ICD-9 code group 674, largely consisted of surgical wound complications (ICD-9 code 674.34; n=189).

Results from nested generalized estimating equations predicting the likelihood of 30-day readmission are presented in Table 4. Black and to a lesser extent Hispanic women were significantly more likely to be readmitted within 30 days of their initial discharge than were white women for both vaginal and cesarean delivery (model 1). The magnitude of these effects was substantial, with black patients approximately twice as likely and Hispanic patients roughly 1.5 times as likely to be readmitted relative to white patients. These odds ratios (ORs) translate to rates of readmission after cesarean delivery of 28.9 per 1,000 for black women (CI 25.5–32.7), 21.4 for Hispanic women (CI 18.9–24.2), and 12.9 for white women (CI 11.9–14.0) and readmission rates after vaginal delivery of 14.6 per 1,000 for black women (CI 13.0–16.5), 10.7 for Hispanic women (CI 9.6–12.0), and 7.5 for white women (CI 7.0–8.1).

Table 4
Table 4:
Results From Generalized Estimating Equations Predicting 30-Day Readmission After Vaginal Delivery (Diagnosis-Related Group 775) and Cesarean Delivery (Diagnosis-Related Group 766), Connecticut Hospitals 2005–2012

Model 2 presented in Table 4 introduces controls for patient sickness, socioeconomic status, and payer. Two of the comorbidity indicator variables—hypertension and obesity—were strongly associated with the odds of readmission after both vaginal and cesarean delivery with ORs ranging between 1.8 and 2.2. Diabetes without complication, infection, and the count of non-pregnancy-related patient comorbidities were also significant predictors of 30-day readmission after vaginal delivery. In contrast, neurologic disorders and infection were also significant predictors of 30-day readmission after cesarean delivery. Length of stay was significantly associated with increased odds of readmission after cesarean delivery, whereas patient socioeconomic status was negatively associated with the odds of readmission after cesarean delivery, indicating that patients residing in higher socioeconomic status zip codes were less likely to be rehospitalized. Patients covered by Medicaid were approximately 1.25–1.3 times more likely to be readmitted within 30 days after discharge relative to those covered by private insurers.

Despite the significant effect of socioeconomic status, patient comorbidities, and payer in predicting readmissions, controlling for these measures had a very modest effect on the race and ethnic differences in readmission observed in model 1. The OR reflecting black–white differences in readmission for cesarean delivery from model 1 was reduced by approximately 21% when these variables were controlled; black–white differences in the odds of readmission after vaginal delivery were reduced by only 13% when payer, patient sickness, and socioeconomic status were controlled. The Hispanic–white differences in readmission after vaginal and cesarean delivery were reduced by only 10–15% when socioeconomic status, patient comorbidities, and payer were controlled (model 2).

We also estimated a series of models to test for differences in race and ethnic readmission patterns 1) over time and 2) among different categories of payers. Separate models including product terms for the interaction of race–ethnicity and 1) date of initial admission; and 2) payer were not statistically significant for either of the DRGs examined in this analysis (Appendices 2 and 3, available online at


This study presents evidence of substantial disparities in hospital readmission after childbirth. Although the base rates for readmission after these procedures are low relative to other conditions,11–1511–1511–1511–1511–15 the large volumes of these procedures result in a problem of significant magnitude because these are two of the most common medical and surgical procedures performed in Connecticut and nationally. Were the rates observed among black women applied to the total Connecticut population, the number of readmissions after vaginal delivery would have exceeded 2,500 as opposed to 1,469 over the period of study, making it one of the most common reasons for rehospitalization across all medical conditions (Appendix 3,

Although a great deal of emphasis is being placed on reducing readmissions in an effort to improve quality,1,21,2 rehospitalization may be less a function of hospital performance than of a lack of community resources, particularly for those on public insurance.5,235,23 Recent research suggests that community-based factors such as availability of general practitioners in the area may be as or more important than hospital factors in determining readmission rates.24 Barriers to accessing ambulatory care among Medicaid beneficiaries are well documented, often leaving such patients with few options other than hospital care for both urgent and nonurgent conditions.25–2725–2725–27 Rates of rehospitalization among Connecticut women covered under public health insurance plans were substantially higher than rates among the privately insured, a trend also observed in a study of diabetic patients in California.4 However, although our data indicate that black and Hispanic women hospitalized for these conditions in Connecticut during this period were three to four times more likely to be covered by Medicaid than were white women, controlling for payer did not explain race and ethnic differences in readmissions after childbirth. These data suggest that barriers to accessing community-based resources may be equally problematic for racial and ethnic minorities who receive health care from private insurers, not just those on Medicaid.

The number of readmissions in our study related to infectious etiologies suggests that genetic and biological factors may also play an important role in disparities. Increased risk of inflammation associated with premature birth among black women is well documented.28 Studies have shown that inflammation pathway-related genetic polymorphisms have a greater prevalence among black mothers and fetuses.29,3029,30 This increased risk of preterm birth predisposes the mother to developing postpartum endometritis, a leading cause of 30-day readmission. Furthermore, preexisting hypertension and obesity are more frequently seen in minority populations and are associated with a number of negative birth outcomes.31–3431–3431–3431–34 However, controlling for the presence of hypertension, obesity, diabetes, infection, and other comorbid conditions at the index admission did not substantially reduce racial disparities in readmission. Further research is needed to specify the extent to which other patient characteristics as well as hospital performance and access to community-based resources are implicated in racial and ethnic disparities in women's reproductive health.

It is important to acknowledge the limitations of our study. Administrative data on patient race and ethnicity were obtained from self-reports or observer reports and may be of limited accuracy and reliability. Recent studies have reported variable levels of agreement between administrative data on race and ethnicity and patient self-reports.35,3635,36 To gauge the reliability of the race and ethnic categorizations in our data, we calculated the level of consistency in these codes among patients with more than one hospital admission from 2005 to 2012. For the DRGs examined in this article, the agreement in race–ethnicity coding across multiple admissions exceeded 96%, suggesting a high level of consistency across admissions. Moreover, the racial and ethnic distributions observed for 2011 hospitalizations in Connecticut were consistent with population distributions derived from the 2010 and 2011 Current Population Surveys (Appendix 4, available online at

Despite these limitations, this study adds to a growing body of evidence on disparities in hospital readmissions in several important respects. First, we have extended the range of medical conditions for which significant racial and ethnic disparities have been documented to include the most common reasons for hospitalization among women. Second, the racial and ethnic disparities observed in this study were considerably more pronounced than disparities generally reported for other diseases and conditions. Third, by utilizing a large statewide database capturing all payers, we were able to control for the effect of insurance status on race and ethnic differences in hospital readmissions after childbirth. These findings should serve as a catalyst for further investigation of the interplay between patient characteristics, hospital practices, and community-based resources and insurance coverage in fostering racial and ethnic disparities in health outcomes.


1. Vaduganathan M, Bonow RO, Gheorghiade M. Thirty-day readmissions: the clock is ticking. JAMA 2013;309:345–6.
2. U.S. Department of Health and Human Services. Report to Congress: national strategy for quality improvement in health care. Washington, DC: HHS; 2011.
3. Jiang HJ, Andrews R, Stryer D, Friedman B. Racial/ethnic disparities in potentially preventable readmissions: the case of diabetes. Am J Public Health 2005;95:1561–7.
4. Kim H, Ross JS, Melkus GD, Zhao Z, Boockvar K. Scheduled and unscheduled hospital readmissions among patients with diabetes. Am J Manag Care 2010;16:760–7.
5. Joynt KE, Orav EJ, Jha AK. Thirty-day readmission rates for Medicare beneficiaries by race and site of care. JAMA 2011;305:675–81.
6. McHugh MD, Carthon JM, Kang XL. Medicare readmissions policies and racial and ethnic health disparities: a cautionary tale. Policy Polit Nurs Pract 2010;11:309–16.
7. Alexander M, Grumbach K, Remy L, Rowell R, Massie BM. Congestive heart failure hospitalizations and survival in California: patterns according to race/ethnicity. Am Heart J 1999;137:919–27.
8. Ganesh S, Rogal SS, Yadav D, Humar A, Behari J. Risk factors for frequent readmissions and barriers to transplantation in patients with cirrhosis. PLoS One 2013;8:e55140.
9. Menachemi N, Chukmaitov A, Brown LS, Saunders C, Brooks RG. Quality of care differs by patient characteristics: outcome disparities after ambulatory surgical procedures. Am J Med Qual 2007;22:395–401.
10. Rodriguez F, Joynt KE, López L, Saldaña F, Jha AK. Readmission rates for Hispanic Medicare beneficiaries with heart failure and acute myocardial infarction. Am Heart J 2011;162:254–261.e3.
11. Burrows LJ, Meyn LA, Weber AM. Maternal morbidity associated with vaginal versus cesarean delivery. Obstet Gynecol 2004;103:907–12.
12. Declercq E, Barger M, Cabral HJ, Evans SR, Kotelchuck M, Simon C, et al.. Maternal outcomes associated with planned primary cesarean births compared with planned vaginal births. Obstet Gynecol 2007;109:669–77.
13. Koroukian SM. Relative risk of postpartum complications in the Ohio Medicaid population: vaginal versus cesarean delivery. Med Care Res Rev 2004;61:203–24.
14. Liu S, Heaman M, Joseph KS, Liston RM, Huang L, Sauve R, et al.. Risk of maternal postpartum readmission associated with mode of delivery. Obstet Gynecol 2005;105:836–42.
15. Lydon-Rochelle M, Holt VL, Martin DP, Easterling TR. Association between method of delivery and maternal rehospitalization. JAMA 2000;283:2411–6.
16. Getahun D, Strickland D, Lawrence JM, Fassett MJ, Koebnick C, Jacobsen SJ. Racial and ethnic disparities in the trends in primary cesarean delivery based on indications. Am J Obstet Gynecol 2009;201:422.e1–7.
17. The State of Connecticut Office of Health Care Access. Inpatient discharge data and Connecticut hospitals 2013. Available at: Retrieved May 21, 2013.
18. Connecticut Department of Public Health. Connecticut health database compendium; a profile of selected databases maintained by the Connecticut Department of Public Health. Hartford (CT): Connecticut Department of Public Health; 2012.
19. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27.
20. U.S. Census Bureau. American Available at: Retrieved December 4, 2013.
21. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13–22.
22. Halekoh U, Hojsgaard S, Yan J. The R package geepack for generalized estimating equations. J Stat Softw 2006;15:1–11.
23. van Walraven C, Bennett C, Jennings A, Austin PC, Forster AJ. Proportion of hospital readmissions deemed avoidable: a systematic review. Can Med Assoc J 2011;183:E391–402.
24. Herrin J, St Andre J, Kenward K, Joshi MS, Audet AJ, Hines SC. Community factors and hospital readmission rates. Health Serv Res 2015;50:20–39.
25. Politzer RM, Yoon J, Shi L, Hughes RG, Regan J, Gaston MH. Inequality in America: the contribution of health centers in reducing and eliminating disparities in access to care. Med Care Res Rev 2001;58:234–48.
26. Manthous CA, Grewal Y, Sofair AN. Medicaid patients' access to subspecialty ambulatory care in Connecticut. J Grad Med Educ 2011;3:582–3.
27. Andrews E, Roccapriore B, Tartajana S. Fixing Medicaid: healing Connecticut's largest health care program. Connecticut health policy project issue brief, 2011. Retrieved June 11, 2013.
28. Hitti J, Nugent R, Boutain D, Gardella C, Hillier SL, Eschenbach DA. Racial disparity in risk of preterm birth associated with lower genital tract infection. Paediatr Perinat Epidemiol 2007;21:330–7.
29. Anum EA, Springel EH, Shriver MD, Strauss JF 3rd. Genetic contributions to disparities in preterm birth. Pediatr Res 2009;65:1–9.
30. Menon R, Pearce B, Velez DR, Merialdi M, Williams SM, Fortunato SJ, et al.. Racial disparity in pathophysiologic pathways of preterm birth based on genetic variants. Reprod Biol Endocrinol 2009;7:62.
31. Fryar CD, Hirsch R, Eberhardt MS, Yoon SS, Wright JD. Hypertension, high serum total cholesterol, and diabetes: racial and ethnic prevalence differences in U.S. adults, 1999–2006. NCHS Data Brief 2010:1–8.
32. Nitert M, Foxcroft K, Lust K, Fagermo N, Lawlor DA, O'Callaghan M, et al.. Overweight and obesity knowledge prior to pregnancy: a survey study. BMC Pregnancy Childbirth 2011;11:96.
33. Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA 2006;295:1549–55.
34. Flegal KM, Carroll MD, Ogden CL, Curtin LR. Prevalence and trends in obesity among US adults, 1999–2008. JAMA 2010;303:235–41.
35. McAlpine DD, Beebe TJ, Davern M, Call KT. Agreement between self-reported and administrative race and ethnicity data among Medicaid enrollees in Minnesota. Health Serv Res 2007;42:2373–88.
36. Kressin NR, Chang BH, Hendricks A, Kazis LE. Agreement between administrative data and patients' self-reports of race/ethnicity. Am J Public Health 2003;93:1734–9.

Supplemental Digital Content

© 2015 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.