Advertisement

Maternal outcomes by race during postpartum readmissions

Published:February 17, 2019DOI:https://doi.org/10.1016/j.ajog.2019.02.016

      Background

      Maternal race may be an important risk factor for postpartum readmissions and associated adverse outcomes.

      Objective

      To determine the association of race with serious complications during postpartum readmissions.

      Study Design

      This repeated cross-sectional analysis used the National (Nationwide) Inpatient Sample from the Healthcare Cost and Utilization Project from 2012 to 2014. Women ages 15–54 readmitted postpartum after a delivery hospitalization were identified by Centers for Disease Control and Prevention criteria. Race and ethnicity were characterized as non-Hispanic white, non-Hispanic black, Hispanic, Asian or Pacific islander, Native American, other, and unknown. Overall risk for readmission by race was determined. Risk for severe maternal morbidity during readmissions by race was analyzed. Individual outcomes including pulmonary edema/acute heart failure and stroke also were analyzed by race. Log-linear regression models including demographics, hospital factors, and comorbid risk were used to analyze risk for severe maternal morbidity during postpartum readmissions.

      Results

      Of 11.3 million births, 207,730 (1.8%) women admitted postpartum from 2012 to 2014 were analyzed, including 96,670 white, 47,015 black, and 33,410 Hispanic women. Compared with non-Hispanic white women, non-Hispanic black women were at 80% greater risk of postpartum readmission (95% confidence interval, 79%–82%) whereas Hispanic women were at 11% lower risk of readmission (95% confidence interval, 10%–12%). In unadjusted analysis, compared with non-Hispanic white women, non-Hispanic black women admitted postpartum were at 27% greater risk of severe maternal morbidity (95% confidence interval, 24%–30%) whereas Hispanic women were at 10% lower risk (95% confidence interval, 7%–13%). In the adjusted model, non-Hispanic black women were at 16% greater risk for severe maternal morbidity during readmission than non-Hispanic white women (95% confidence interval, 10%–22%), whereas Hispanic women were at 7% lower risk (95% confidence interval, 1%–12%). Differences in severe maternal morbidity risk between other racial groups and non-Hispanic white women were not significant. In addition to overall morbidity, non-Hispanic black women were at significantly greater risk for eclampsia, acute respiratory distress syndrome, and renal failure than other racial groups (P<.05 all). Black women were at 126% greater risk for pulmonary edema/acute heart failure than white women (95% confidence interval, 117%–136%).

      Conclusion

      Black women were more likely (1) to be readmitted postpartum, (2) to suffer severe maternal morbidity during readmission, and (3) to suffer life threatening complications such as pulmonary edema/acute heart failure. At-risk women including black women with cardiovascular risk factors may benefit from short-term postpartum follow-up.

      Key words

      Obstetric readmissions may be of increasing clinical importance. Overall risk for postpartum readmission is increasing, with a recent study finding an increase from 1.7% in 2004 to 2.2% in 2011.
      • Clapp M.A.
      • Little S.E.
      • Zheng J.
      • Robinson J.N.
      A multi-state analysis of postpartum readmissions in the United States.
      The American College of Obstetricians and Gynecologists recently has updated recommendations regarding postpartum care, stating “To optimize the health of women and infants, postpartum care should become an ongoing process, rather than a single encounter, with services and support tailored to each woman’s individual needs.”
      ACOG Committee Opinion No. 736
      Optimizing Postpartum Care.
      Traditional approaches to postpartum care are being reconsidered in hopes of reducing adverse postpartum maternal outcomes and risk for readmission.
      • Murray Horwitz M.E.
      • Molina R.L.
      • Snowden J.M.
      Postpartum care in the United States— new policies for a new paradigm.

      Why was the study conducted?

      • To evaluate whether there are racial disparities in postpartum readmissions.

      Key findings

      • Black women were more likely to be readmitted, more likely to experience severe morbidity, and were at greater risk for life-threatening complications such as pulmonary edema and heart failure.

      What does this add to what is known?

      • Black women are at particularly high risk for specific postpartum readmission complications.
      An important consideration in optimizing postpartum care is appropriately characterizing maternal risk. Previous studies have identified a range of risk factors for readmission, including advanced age, socioeconomic status, chronic comorbidities, multiple gestation, cesarean delivery, and other high-risk pregnancy conditions.
      • Clapp M.A.
      • Little S.E.
      • Zheng J.
      • Robinson J.N.
      A multi-state analysis of postpartum readmissions in the United States.
      • Aseltine Jr., R.H.
      • Yan J.
      • Fleischman S.
      • Katz M.
      • DeFrancesco M.
      Racial and ethnic disparities in hospital readmissions after delivery.
      • Sharvit M.
      • Rubinstein T.
      • Ravid D.
      • Shechter-Maor G.
      • Fishman A.
      • Biron-Shental T.
      Patients with high-risk pregnancies and complicated deliveries have an increased risk of maternal postpartum readmissions.
      Maternal race may represent a particularly important risk factor for postpartum readmissions and associated complications. Maternal race is a major risk factor adverse obstetric outcomes, and a number of previous analyses have addressed disparities in overall risk for severe morbidity and mortality.
      • Creanga A.A.
      • Syverson C.
      • Seed K.
      • Callaghan W.M.
      Pregnancy-related mortality in the United States, 2011-2013.
      • Creanga A.A.
      • Bateman B.T.
      • Kuklina E.V.
      • Callaghan W.M.
      Racial and ethnic disparities in severe maternal morbidity: a multistate analysis, 2008-2010.
      • Howell E.A.
      • Egorova N.
      • Balbierz A.
      • Zeitlin J.
      • Hebert P.L.
      Black-white differences in severe maternal morbidity and site of care.
      Although race has been associated with increased risk for postpartum risk in smaller samples,
      • Aseltine Jr., R.H.
      • Yan J.
      • Fleischman S.
      • Katz M.
      • DeFrancesco M.
      Racial and ethnic disparities in hospital readmissions after delivery.
      national estimates of postpartum readmissions and associated complications have not been well characterized. A better understanding of how race is associated with postpartum risk may be useful in risk stratification and designing improvements in maternal care.
      Given the knowledge gap regarding race and postpartum readmissions, the objectives of this study were to characterize (1) risk for postpartum readmissions by race, and (2) describe risk for severe, life-threatening complications during these hospitalizations.

      Materials and Methods

      The National (Nationwide) Inpatient Sample (NIS) from the Healthcare Cost and Utilization Project for the years 2012–2014 was used for this repeated cross-sectional analysis. The NIS is a large, publicly available, all-payer inpatient contains a sample of approximately 20% of all hospitalizations in the United States. These hospitalizations are selected via a stratified systemic random sample to generate a population representative of the entire United States across medical specialties that includes academic, community, nonfederal, general, and specialty-specific centers. Weights can be applied to create national estimates. Approximately 8 million hospital stays from a total of 45 states were included in the NIS in 2010.
      Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project (HCUP)
      Introduction to the HCIP Nationwide Readmissions Database (NRD) 2010-2015.
      For this analysis, postpartum readmissions after a delivery hospitalization were captured with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes. The Centers for Disease Control and Prevention (CDC) provided the algorithm used to identify postpartum hospitalizations.
      • Callaghan W.M.
      • Creanga A.A.
      • Kuklina E.V.
      Severe maternal morbidity among delivery and postpartum hospitalizations in the United States.
      Centers for Disease Control and Prevention
      Severe Maternal Morbidity in the United States.
      We included women aged 15–54 years. The primary exposure of interest was self-reported maternal race and ethnicity. Maternal race is categorized by the NIS as non-Hispanic white, non-Hispanic black, Hispanic, Asian or Pacific Islander, Native American, other, or unknown. The years 2012–2014 were used for the analysis, given that the proportion of missing data for race in the NIS for these years is small compared with preceding iterations. Women with a code for a delivery hospitalization (ICD-9-CM 650 or V27.x) were excluded. The Columbia University Institutional Review Board deemed this analysis exempt given that the data are deidentified.
      The primary outcome of this study was risk for severe maternal morbidity (SMM) as defined by the CDC during a postpartum readmission. The composite CDC definition of SMM includes 21 diagnoses, including shock, stroke, heart failure, transfusion, and other conditions all identified using ICD-9-CM codes.
      Centers for Disease Control and Prevention
      Severe Maternal Morbidity in the United States.
      A sensitivity analysis was performed excluding transfusion and retaining the remaining 20 diagnoses, given that transfusion is the most common complication in the composite. In addition to the primary outcome of SMM, we evaluated 2 secondary outcomes. First, we evaluated overall risk for postpartum readmissions. Because the NIS is cross-sectional and does not allow linkage of hospitalizations for individual patients, readmission risk was calculated with the number of deliveries each year as denominator for each racial category. Second, we evaluated risk for 11 individual severe morbidity outcomes within the CDC composite: (1) acute renal failure, (2) acute respiratory distress syndrome, (3) disseminated intravascular coagulation, (4) eclampsia, (5) stroke, (6) hysterectomy, (7) shock, (8) embolism, (9) sepsis, (10) pulmonary edema and acute heart failure, and (11) transfusion. These conditions were chosen because they represent a broad range clinical pathology and were anticipated to be sufficiently prevalent to make meaningful comparisons.
      • Callaghan W.M.
      • Creanga A.A.
      • Kuklina E.V.
      Severe maternal morbidity among delivery and postpartum hospitalizations in the United States.
      Demographic factors, hospital characteristics, and comorbidity were evaluated by NIS race categories and compared using the χ2 test or Fisher exact test as appropriate. Hospital characteristics included bed size (small, medium, or large), location and teaching status (urban teaching, urban non-teaching, and rural), and region (Northeast, Midwest, South, or West). Demographic categories included year of delivery, insurance status (Medicaid, private, Medicare, other, uninsured), and ZIP code income quartile.
      Comorbidity was evaluated using an obstetric comorbidity index that measures underlying patient risk.
      • Bateman B.T.
      • Mhyre J.M.
      • Hernandez-Diaz S.
      • et al.
      Development of a comorbidity index for use in obstetric patients.
      An important consideration in evaluating population-based outcomes is appropriate risk adjustment. For evaluating morbidity and death, there are comorbidity indices that have been developed in medical and surgical populations that account for underlying risk. This comorbidity index has been developed specifically for an obstetric population.
      In creating the obstetric comorbidity index, a list of maternal comorbidities possible associated with increased risk for maternal morbidity and mortality was developed. These risk factors were analyzed within an initial obstetric sample population and used to construct an adjusted model using a fully stepwise selection algorithm for both entry and retention in the model. The outcome was either severe maternal morbidity or mortality during the delivery hospitalization or postpartum. Potential risk factors included 24 maternal comorbidities and maternal age categorized as <19, 20–34, 35–39, 40–44, and >44 years of age.
      Risk factors found to be significant were retained in the model and then tested in a validation cohort of the same population. The final model includes 20 maternal conditions and maternal age. This comorbidity index provides weighted scores for comorbidity for obstetric patients based on specific diagnosis codes and demographic factors that can be ascertained in administrative data. Greater scores are associated with increased risk for severe morbidity. In the initial study validating the comorbidity index in a general obstetric population, patients with the lowest score of 0 had a 0.7% risk of severe morbidity whereas scores of >10 were associated with a risk of severe morbidity of 10.9%. This comorbidity index was subsequently validated in an external population.
      • Metcalfe A.
      • Lix L.M.
      • Johnson J.A.
      • et al.
      Validation of an obstetric comorbidity index in an external population.
      We categorized women based on comorbidity index scores: 0 (lowest risk), 1 or 2, and >2 (highest). Because maternal age is presented separately in our analysis to demonstrate the effect of this factor, this variable was omitted in calculating the comorbidity index score. A supplemental analysis includes the full comorbidity score and excludes maternal age as a separate variable. Adding the comorbidity index score to the adjusted analysis allows for the analysis to better characterize the effect of race accounting for other underlying risk factors. Adjusted risk ratios (aRRs) for severe morbidity including transfusion with 95% confidence intervals (CIs) as measures of effect accounting for comorbidity, demographic characteristics, and hospital factors were derived from fitting a log-linear regression model. Population weights can be applied to data in the NIS to create national estimates; these weights were applied in this study. All analyses were performed with SAS 9.4 (SAS Institute, Cary, NC).

      Results

      From 2012 to 2014, an estimated 11.3 million births and 207,730 (1.8%) postpartum readmissions were ascertained from the NIS and included in the analysis. The proportion of non-Hispanic black women readmitted relative to delivery hospitalizations was significantly larger (3.09%; 95% CI, 3.06%–3.12%; n=47,015/1,523,190) than other groups, including non-Hispanic white women (1.71%; 95% 1.70%–1.72%; 96,670/5,650,075), Hispanic women (1.52%; 95% CI, 1.50%–1.54%; 33,410/2,194,500), Asian or Pacific Islander women (1.17%; 95% CI, 1.14%–1.20%; 6930/592,025), Native American women (2.25%; 95% CI, 2.15%–2.35%; 1855/82,315), and women of other and unknown race (1.74%; 95% CI, 1.70%–1.78%; 8895/516,450 and 1.83%; 95% CI, 1.80%–1.86%; 12,865/702,315, respectively) (P<.01). Overall, readmissions for non-Hispanic black women relative to delivery hospitalizations were 80.4% more likely than for non-Hispanic white women (95% CI, 78.5%–82.4%, P<.01).
      Among readmissions, Hispanic women were younger than non-Hispanic black and non-Hispanic white women, whereas Asian or Pacific Islander women were the oldest (Table 1). Non-Hispanic black and Hispanic women were more likely to receive Medicaid insurance than Asian or Pacific Islander or non-Hispanic white women (62.9% vs 62.5% vs 32.9% vs 39.8%, respectively, P<.01) and more likely to be from the lowest ZIP code income quartile (49.1% vs 36.2% vs 14.2% vs 25.1%, respectively, P<.01). Compared with non-Hispanic white, Hispanic, and Asian or Pacific Islander women, non-Hispanic black women had the greatest rates of mild, severe, and superimposed preeclampsia (P<.01). Finally, excluding age and preeclampsia, non-Hispanic black women had greater comorbidity scores; 3.6% of non-Hispanic black women had a score >2 and 21.8% a score of 1 or 2 compared with 1.2% and 16.7% of non-Hispanic white women, 0.9% and 12.7% of Hispanic women, and 2.5% and 9.3% of Asian or Pacific Islander women (P<.01).
      Table 1Demographics of patients readmitted postpartum
      White non-HispanicBlack non-HispanicHispanicAsian/Pacific IslanderNative AmericanOtherUnknown
      All patients (n)96,67047,01533,41069301855898512,865
      %%%%%%%
      Maternal age, y
       15–171.8%2.4%4.0%0.6%2.2%1.9%2.5%
       18–2426.9%31.5%33.1%12.3%36.1%27.7%26.7%
       25–2926.9%25.6%25.6%23.4%27.5%26.7%27.8%
       30–3426.3%22.5%21.4%35.0%19.1%25.4%26.7%
       35–3913.7%13.1%12.3%21.6%11.6%13.8%12.5%
      >404.5%4.9%3.6%7.1%3.5%4.6%3.8%
      Year
       201233.7%32.7%34.5%33.6%37.7%36.0%31.8%
       201332.6%33.4%32.1%31.0%32.3%30.6%35.5%
       201433.7%33.9%33.4%35.4%29.9%33.4%32.7%
      Hospital bed size
       Small12.8%9.9%11.9%10.5%13.5%11.9%13.4%
       Medium26.6%29.2%28.6%29.4%26.1%28.2%27.1%
       Large60.6%60.9%59.5%60.1%60.4%60.0%59.5%
      Insurance status
      Unknown insurance status not demonstrated given cell size <10.
       Medicare1.8%4.1%0.9%0.8%0.8%1.2%3.7%
       Medicaid39.8%62.9%62.5%32.9%59.3%54.1%38.8%
       Private51.3%27.1%25.4%60.4%23.7%35.1%49.9%
       Self-pay3.0%3.2%7.4%3.2%5.9%6.0%3.1%
       Other3.9%2.5%3.5%2.6%10.0%3.4%4.2%
      Hospital location
       Rural11.4%3.9%3.9%3.2%25.6%4.6%12.6%
       Urban non-teaching31.8%24.3%33.6%28.4%26.7%29.2%20.2%
       Urban teaching56.7%71.8%62.5%68.4%47.7%66.2%67.2%
      ZIP income quartile
       1st (lowest)25.1%49.1%36.2%14.2%45.3%30.3%23.7%
       2nd27.0%22.3%25.8%18.0%24.5%23.2%29.8%
       3rd24.9%16.1%22.9%26.8%15.9%23.9%25.8%
       4th (highest)21.8%10.2%12.2%39.4%7.8%19.8%18.9%
       Unknown1.2%2.2%2.9%1.7%6.5%2.8%1.7%
      Hospital region
       Northeast18.5%19.1%15.9%19.8%4.6%30.8%4.4%
       Midwest22.4%19.4%7.4%11.5%14.6%17.0%60.3%
       South41.6%53.3%37.0%20.6%31.8%36.0%16.7%
       West17.6%8.1%39.7%48.0%49.1%16.2%18.7%
      Comorbidity index
       082.2%74.5%86.4%88.2%76.3%85.7%81.5%
       1 or 216.7%21.8%12.7%9.3%23.2%13.0%16.9%
       >21.2%3.6%0.9%2.5%0.6%1.3%1.5%
      Preeclampsia
       Mild6.1%9.2%4.1%5.9%2.7%6.1%4.6%
       Severe4.9%8.0%4.1%5.9%5.4%6.5%6.0%
       Superimposed1.8%4.2%1.5%1.9%1.6%2.3%1.6%
      Comorbidity index excludes maternal age and hypertensive diagnoses.
      Aziz et al. Maternal outcomes by race during postpartum readmissions. Am J Obstet Gynecol 2019.
      a Unknown insurance status not demonstrated given cell size <10.
      During readmissions, non-Hispanic black women were at greater risk for severe maternal morbidity, including transfusion (24.3%), than all other groups including non-Hispanic white (19.1%), Asian or Pacific Islander (20.5%), Native American women (21.0%), and Hispanic women (17.2%) as well as women of unknown (20.2%) or other (19.1%) race (P<.01). In the unadjusted model with non-Hispanic white women as the reference, this difference translated into a 27% increased risk (risk ratio [RR], 1.27; 95% CI, 1.24–1.30) for non-Hispanic black women while Hispanic women were at 10% lower risk (RR, 0.90; 95% CI, 0.87–0.93). Women with Medicare compared with private insurance were at increased risk (RR, 1.37; 95% CI, 1.30–1.45). Women age 40 or older and women aged 35–39 years were at increased risk for severe morbidity compared with women 18–24 years of age (RR, 1.53; 95% CI, 1.46–1.59; RR, 1.24; 95% CI, 1.20–1.28). Greater obstetric comorbidity index scores of 1–2 or >2 also were associated with increased risk for severe morbidity (RR, 1.35; 95% CI, 1.32–1.38; RR, 2.61; 95% CI, 2.54–2.69, respectively). In the adjusted analysis, many of these risk factors retained significance. Non-Hispanic black women were at increased risk for severe morbidity compared with non-Hispanic white women (aRR, 1.16; 95% CI, 1.10–1.22) whereas Hispanic women were at slightly lower risk (aRR, 0.93; 95% CI, 0.88–0.99). Medicare insurance was retained as a risk factor (aRR, 1.18; 95% CI, 1.06–1.30) (Table 2). Women age 40 years or older and women age 35–39 years were similarly at increased risk for severe morbidity compared with women 18–24 years of age (aRR, 1.40; 95% CI, 1.29–1.52; aRR, 1.18, 95% CI, 1.11–1.26). Finally, comorbidity index score of 1 or 2 and >2 were associated with increased risk compared with scores of 0 (aRR, 1.28; 95% CI, 1.23–1.34; aRR, 2.47, 95% CI, 2.34–2.60, respectively). When the analysis was repeated excluding maternal age as a separate variable and incorporating it into the obstetric comorbidity index, the estimate for comorbidity index score >2 compared with 0 was greater (aRR, 3.14; 95% CI, 2.98–3.31) whereas estimates for other factors in the analysis was similar (Supplemental Table 1). When the sensitivity analysis was performed restricted to severe morbidity without transfusion, the risk differential was larger. Non-Hispanic black women were at 41.8% increased risk for severe morbidity excluding transfusion compared with non-Hispanic white women (21.0% vs 14.8%) with risk also lower for Hispanic (13.3%), Asian or Pacific Islander (16.3%), and Native American women (16.7%) (P<.01).
      Table 2Unadjusted and adjusted models for severe morbidity during postpartum readmissions
      RR95% CIaRR95% CI
      Race
       Non-Hispanic white1.00Reference1.00Reference
       Non-Hispanic black1.271.24–1.301.161.10–1.22
       Hispanic0.900.87–0.930.930.88–0.99
       Asian or Pacific Islander1.091.03–1.151.050.95–1.17
       Native American1.100.99–1.211.080.89–1.32
       Other1.000.95–1.051.030.93–1.13
       Unknown1.051.01–1.101.020.94–1.11
      Year
       20121.00Reference1.00Reference
       20131.051.03–1.081.040.99–1.09
       20141.061.03–1.081.030.98–1.08
      Hospital bed size
       Small1.00Reference1.00Reference
       Medium1.051.01–1.081.020.95–1.09
       Large1.111.07–1.141.060.99–1.12
      Insurance status
       Private1.00Reference1.00Reference
       Medicare1.371.30–1.451.181.06–1.30
       Medicaid0.980.96–1.000.980.94–1.02
       Other0.850.81–0.900.870.78–0.97
       Uninsured1.000.95–1.061.030.93–1.14
      Hospital location
       Urban non-teaching1.00Reference1.00Reference
       Rural0.960.93–1.000.990.91–1.07
       Urban teaching1.021.00–1.040.970.93–1.02
      Income quartile
       1st (lowest)1.00Reference1.00Reference
       2nd0.970.94–0.991.000.95–1.05
       3rd0.940.92–0.960.970.92–1.02
       4th (highest)0.920.90–0.950.940.88–1.00
      Hospital region
       Northeast1.00Reference1.00Reference
       Midwest1.191.16–1.241.201.12–1.29
       South1.251.22–1.291.251.18–1.33
       West1.181.15–1.221.271.18–1.36
      Maternal age
       15-170.960.90–1.030.970.84–1.11
       18-241.00Reference1.00Reference
       25-291.051.02–1.081.030.98–1.09
       30-341.141.11–1.171.111.05–1.17
       35-391.241.20–1.281.181.11–1.26
       >391.531.46–1.591.401.29–1.52
      Comorbidity index
       01.00Reference1.00Reference
       1-21.351.32–1.381.281.23–1.34
       >22.612.54–2.692.472.34–2.60
      aRR, adjusted risk ratio; CI, confidence interval; RR, risk ratio.
      Adjusted model included all factors in this table (year, bed size, insurance status, hospital location, income quartile, hospital region, hospital teaching status, maternal age, and race). Comorbidity index excludes maternal age.
      Aziz et al. Maternal outcomes by race during postpartum readmissions. Am J Obstet Gynecol 2019.
      Evaluating individual severe morbidity outcomes, we found that non-Hispanic black women were at significantly increased risk compared with non-Hispanic white women for several outcomes. Non-Hispanic black women were at increased risk for acute renal failure (RR, 1.56; 95% CI, 1.46–1.68), acute respiratory distress syndrome (RR, 1.38; 95% CI, 1.30–1.49), eclampsia (RR, 1.69; 95% CI, 1.55–1.83), and stroke (RR, 1.41; 95% CI, 1.29–1.54) (Figure 1) with non-Hispanic white women as the reference. Non-Hispanic black women were at particularly high risk for pulmonary edema and acute heart failure, with 8.4% readmissions receiving diagnosis compared with 3.7% of non-Hispanic white women (RR, 2.26; 95% CI, 2.17–2.36) (Figure 2). Comparing these groups, we found that non-Hispanic black women were at significantly lower risk for shock (RR, 0.79; 95% CI, 0.70–0.89), transfusion (RR, 0.88; 95% CI, 0.85–0.93), sepsis (RR, 0.85; 95% CI, 0.80–0.90), and hysterectomy (RR, 0.58; 95% CI, 0.50–0.68). Differences for risk between non-Hispanic black and white women were not significant for disseminated intravascular coagulation and embolism.
      Figure thumbnail gr1
      Figure 1Risk for individual severe morbidity diagnoses by maternal race
      Compared with non-Hispanic white women, non-Hispanic black women were at significantly increased risk for acute renal failure, ARDS, eclampsia, and stroke (P<.01 for all). Black women were at significantly decreased risk for hysterectomy and shock (P<.01), whereas differences in DIC and embolism were not significantly different.
      ARDS, acute respiratory distress syndrome; DIC, disseminated intravascular coagulation.
      Aziz et al. Maternal outcomes by race during postpartum readmissions. Am J Obstet Gynecol 2019.
      Figure thumbnail gr2
      Figure 2Risk for individual severe morbidity diagnoses by maternal race
      Compared with non-Hispanic white women, non-Hispanic black women were at significantly increased risk pulmonary edema and acute heart failure but decreased risk for transfusion or sepsis.
      Aziz et al. Maternal outcomes by race during postpartum readmissions. Am J Obstet Gynecol 2019.

      Discussion

      Main findings

      In this analysis of postpartum readmissions, non-Hispanic black women were at significantly greater risk (1) to be readmitted postpartum, (2) to suffer severe maternal morbidity during readmission, and (3) to suffer a range of life-threatening complications. Although non-Hispanic black women were at lower risk for complications such as sepsis, shock, hysterectomy, and transfusion, they were particularly likely to receive a diagnosis of pulmonary edema/acute heart failure, with risk more than twice as high as that of non-Hispanic white women.

      Implications of the findings and future research directions

      The differential risk found in this analysis may have a number of implications for optimizing postpartum care. First, because non-Hispanic black women were at much greater risk for readmissions relative to the number of births in the population, efforts to reduce maternal risk in this demographic may represent an important component of overall strategies to reduce postpartum risk and racial disparities. Although this dataset is not able to link readmissions across hospitalizations, identifying preceding obstetric, medical, and social risk factors among non-Hispanic black women that contribute to differentials will be necessary to identify who may most benefit from more intensive postpartum care and surveillance.
      Second, non-Hispanic black women were at particularly high risk for cardiopulmonary complications. Because this dataset does not include outpatient data and chart reviews are not possible, we are not able to determine to what degree these conditions were preceded by outpatient symptoms or whether changes in management could have prevented readmissions. However, it is reasonable to assume that for at-risk women who develop cardiopulmonary complications, optimal postpartum care may facilitate timely diagnosis, multidisciplinary care, and transfer to an appropriate level of inpatient care.
      Third, the relatively large differentials in risk for adverse outcomes supports the American College of Obstetricians and Gynecologists’ recommendation that postpartum care be tailored. In comparison with intrapartum management, where a small number of conditions such as hemorrhage, hypertensive diseases of pregnancy, and thromboembolism account for the majority of risk among most women, postpartum risk may be more differentially distributed and comparative effectiveness research is needed to determine which interventions may most benefit individual populations. Specifically, further data are needed to determine potential benefits of longer inpatient stays for particularly high-risk women, shorter office follow-up, mobile health interventions, increased coordination with medical specialists, and other approaches to reducing maternal risk. Fourth, although non-Hispanic black women were at lower risk than non-Hispanic white women for hysterectomy, sepsis, shock, and transfusion during readmissions, on a population basis relative to delivery hospitalizations non-Hispanic black women were at similar risk for hysterectomy and increased risk for the latter 3 conditions, given overall likelihood of readmission.

      Strengths and limitations

      There are several limitations that are important to consider in interpreting this study’s findings. First, as noted, these data are cross-sectional, and we are not able to account for factors that occurred during the delivery or after hospital discharge prior to the readmission. We are not able to determine whether a complication occurred on an outpatient basis or during the readmission. As a result, although there were clear differentials in outcomes noted between racial groups, we are not able to make inferences related to preventability or when a complication occurred. Second, this administrative data set does not provide information on hospital resources, infrastructure, protocols and staffing, all of which contribute to maternal outcomes and risk.
      Third, because administrative data are used primarily for billing data both underascertainment and misclassification of secondary diagnoses are concerns. Although administrative data are appropriate in assessing population-level resource use and disease burden, granularity of many important clinical factors is often limited. Fourth, data on race was missing for 6% of patients of this study. If outcomes based on race differed significantly for patients with unknown compared with known race, the disparities reported may be biased.
      The Healthcare Cost and Utilization Project provides a number of recommendations for addressing missing data, including imputation.
      • Houchens R.
      Missing Data Methods for the NIS and the SID. 2015. HCUP Methods Series Report # 2015-01 ONLINE. January 22, 2015. U.S. Agency for Healthcare Research and Quality.
      We decided not to perform imputation because it was unlikely that missing data on race were (1) randomly related to other measurable factors and (2) nondifferential across the years of NIS included in this analysis. In determining whether to perform imputation important considerations are (1) whether specific outcomes in the actual population are related to the greater likelihood that data are missing, and (2) whether factors associated with the outcome of interest are related differentially to missing and non-missing data. HCUP documents support that specific outcomes may in fact have been related to missingness; for these missing data, differential associations may be present between other measured factors.
      • Houchens R.
      Missing Data Methods for the NIS and the SID. 2015. HCUP Methods Series Report # 2015-01 ONLINE. January 22, 2015. U.S. Agency for Healthcare Research and Quality.
      To address these concerns, HCUP in part suggests incorporating data outside the NIS, an approach beyond the scope of this analysis.
      Fifth, because hospitalizations are not linked, we are not able to determine the time interval between delivery hospitalization discharge and readmission; subsequent analyses with data sources that allow evaluation of duration between discharge and readmissions may be important in optimizing care. Sixth, many complications that are primarily managed on an outpatient basis, such as wound complications, cannot be captured using this data. Seventh, we cannot control for the effect of multiple pregnancies occurring to individual women. We elected to use the NIS because it is a large, nationally representative sample, and for the 3 years of the study there were low rates of missing data compared with prior iterations in which data on race were missing for >20% of all hospitalizations. Other strengths of this study include measurement of readmissions, which are an adverse event in and of themselves, and assessment of a broad range of outcomes demonstrating major differentials in risk.
      In conclusion, this analysis demonstrated increased risk among non-Hispanic black women for postpartum readmission and associated complications. Addressing this risk aligns with urgent priorities of optimizing postpartum care for at-risk women and reducing racial disparities in adverse obstetric outcomes. At-risk women, including black women with cardiovascular risk factors, may in particular benefit from closer surveillance, including short-term postpartum follow up.

      Appendix

      Supplemental Table 1Adjusted models for severe morbidity during postpartum readmissions with comorbidity index inclusive of maternal age
      Adjusted risk ratio95% CI
      Race
       Non-Hispanic white1.00Reference
       Non-Hispanic black1.201.14–1.26
       Hispanic0.930.88–0.99
       Asian or Pacific Islander1.060.95–1.18
       Native American1.090.90–1.33
       Other1.050.95–1.16
       Unknown1.040.96–1.13
      Year
       20121.00Reference
       20131.040.99–1.09
       20141.040.99–1.09
      Hospital bed size
       Small1.00Reference
       Medium1.030.96–1.10
       Large1.071.00–1.13
      Insurance status
       Private1.00Reference
       Medicare1.010.91–1.12
       Medicaid0.950.91–0.99
       Other0.990.90–1.10
       Uninsured0.830.74–0.92
      Hospital Location
       Urban non-teaching1.00Reference
       Rural1.000.93–1.09
       Urban teaching1.000.96–1.05
      Income quartile
       1st (lowest)1.00Reference
       2nd1.000.95–1.05
       3rd0.980.93–1.03
       4th (highest)0.950.89–1.01
      Hospital region
       Northeast1.00Reference
       Midwest1.211.13–1.29
       South1.271.19–1.34
       West1.261.17–1.35
      Comorbidity index
       01.00Reference
       1–21.301.24–1.35
       >23.142.98–3.31
      CI, confidence interval.
      Adjusted model included all factors in this table (race, year, bed size, insurance status, hospital location, income quartile, hospital region, and comorbidity index score).
      Aziz et al. Maternal outcomes by race during postpartum readmissions. Am J Obstet Gynecol 2019.

      References

        • Clapp M.A.
        • Little S.E.
        • Zheng J.
        • Robinson J.N.
        A multi-state analysis of postpartum readmissions in the United States.
        Am J Obst Gynecol. 2016; 215: 113.e1-113.e110
        • ACOG Committee Opinion No. 736
        Optimizing Postpartum Care.
        Obstet Gynecol. 2018; 131: e140-e150
        • Murray Horwitz M.E.
        • Molina R.L.
        • Snowden J.M.
        Postpartum care in the United States— new policies for a new paradigm.
        N Engl J Med. 2018; 379: 1691-1693
        • Aseltine Jr., R.H.
        • Yan J.
        • Fleischman S.
        • Katz M.
        • DeFrancesco M.
        Racial and ethnic disparities in hospital readmissions after delivery.
        Obstet Gynecol. 2015; 126: 1040-1047
        • Sharvit M.
        • Rubinstein T.
        • Ravid D.
        • Shechter-Maor G.
        • Fishman A.
        • Biron-Shental T.
        Patients with high-risk pregnancies and complicated deliveries have an increased risk of maternal postpartum readmissions.
        Arch Gynecol Obstet. 2014; 290: 629-633
        • Creanga A.A.
        • Syverson C.
        • Seed K.
        • Callaghan W.M.
        Pregnancy-related mortality in the United States, 2011-2013.
        Obstet Gynecol. 2017; 130: 366-373
        • Creanga A.A.
        • Bateman B.T.
        • Kuklina E.V.
        • Callaghan W.M.
        Racial and ethnic disparities in severe maternal morbidity: a multistate analysis, 2008-2010.
        Am J Obstet Gynecol. 2014; 210: 435 e1-435 e8
        • Howell E.A.
        • Egorova N.
        • Balbierz A.
        • Zeitlin J.
        • Hebert P.L.
        Black-white differences in severe maternal morbidity and site of care.
        Am J Obstet Rynecol. 2016; 214: 122 e1-122 e7
        • Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project (HCUP)
        Introduction to the HCIP Nationwide Readmissions Database (NRD) 2010-2015.
        (Available at:) (Accessed January 10, 2019)
        • Callaghan W.M.
        • Creanga A.A.
        • Kuklina E.V.
        Severe maternal morbidity among delivery and postpartum hospitalizations in the United States.
        Obstet Gynecol. 2012; 120: 1029-1036
        • Centers for Disease Control and Prevention
        Severe Maternal Morbidity in the United States.
        (Available at:)
        • Bateman B.T.
        • Mhyre J.M.
        • Hernandez-Diaz S.
        • et al.
        Development of a comorbidity index for use in obstetric patients.
        Obstet Gynecol. 2013; 122: 957-965
        • Metcalfe A.
        • Lix L.M.
        • Johnson J.A.
        • et al.
        Validation of an obstetric comorbidity index in an external population.
        BJOG. 2015; 122: 1748-1755
        • Houchens R.
        Missing Data Methods for the NIS and the SID. 2015. HCUP Methods Series Report # 2015-01 ONLINE. January 22, 2015. U.S. Agency for Healthcare Research and Quality.
        (Available at:)