Births to women aged 40 years and older have increased over recent decades with the birth rate for women 40–44 years approximately doubling between 1990 and 2014.1 Over that same period, births to women 45–49 years of age approximately quadrupled.2 Although many women aged 40 years and older will have uncomplicated pregnancies, pregnancy for women in their 40s is associated with increased risk for adverse outcomes including preterm birth, preeclampsia, gestational diabetes, cesarean delivery, and maternal mortality and severe morbidity.3–9
Maternal race is another major factor in risk for maternal mortality and severe morbidity and other adverse outcomes.3,9,10 Data from the Centers for Disease Control and Prevention’s (CDC) Pregnancy Mortality Surveillance System demonstrated increasing mortality risk for all women over recent decades with risk highest among non-Hispanic black women across maternal age categories. From 2011 to 2013, the mortality risk for non-Hispanic black women was 3.4 times that of non-Hispanic white women, and for non-Hispanic black women aged 40 years or older, the maternal mortality rate was 191.6 per 100,000 livebirths, more than 10 times the national average.9 Black women are also at increased risk for severe morbidity based on CDC criteria.3
Although the role of maternal racial disparities in risk for severe morbidity and mortality is well characterized, how race is related specifically to severe morbidity and other adverse outcomes besides mortality for women 40 years of age or older is unclear. The objectives of this study were to determine how race is associated with adverse maternal outcomes for women aged 40 years or older by characterizing 1) temporal trends in severe maternal morbidity and overall comorbid risk by race in women aged 40 years or older and 2) characterize trends in common pregnancy complications including preeclampsia, gestational diabetes, and cesarean delivery.
MATERIALS AND METHODS
We conducted a retrospective cohort study. The Nationwide Inpatient Sample from the Agency for Healthcare Research and Quality for the years 1998–2014 was used for this analysis. The Nationwide Inpatient Sample is the largest publically available, all-payer inpatient database in the United States and contains a sample of approximately 20% of all hospitalizations nationally. These hospitalizations are selected to generate a sample that is representative for the entire United States and includes academic, community, nonfederal, general, and specialty-specific centers. In 2010, approximately 8 million hospital stays from a total of 45 states were included in the Nationwide Inpatient Sample.11,12 For this analysis, index delivery hospitalizations were captured with International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnosis codes 650 and V27.x. These criteria ascertain greater than 95% of delivery hospitalizations.13 Because the data are deidentified, the Columbia University institutional review board deemed this analysis exempt.
Women aged 40–54 years were included. The primary exposure evaluated was self-reported maternal race and ethnicity as categorized by the Nationwide Inpatient Sample: non-Hispanic white, non-Hispanic black, Hispanic, Asian or Pacific Islander, Native American, or other. The primary outcome of this study was severe maternal morbidity as defined by the CDC. The CDC definition of severe maternal morbidity includes 21 diagnoses including shock, stroke, heart failure, transfusion, and other conditions all identified using ICD-9-CM codes (Appendix 1, available online at http://links.lww.com/AOG/B123).14 Women were considered to have had severe maternal morbidity if they had any codes for any one of these conditions. As an additional sensitivity analysis, because the most common diagnosis in the severe morbidity composite is transfusion (ICD-9-CM code 99.0x), the severe morbidity analysis excluding transfusion and was restricted to the remaining 20 conditions representative of nontransfusion severe morbidity.
We evaluated five sets of secondary outcomes. First, we evaluated temporal trends by race for diagnoses of preeclampsia, cesarean delivery, and gestational diabetes. Second, we evaluated temporal trends by race of specific diagnoses within the CDC composite severe morbidity outcome including 1) acute renal failure, 2) disseminated intravascular coagulation, 3) acute respiratory distress syndrome, 4) stroke, 5) transfusion, and 6) hysterectomy. These diagnoses were chosen because they are relatively common on a population basis and the data use agreement for the Nationwide Inpatient Sample precludes display of small cell sizes. Third, we evaluated risk for maternal death. Fourth, we evaluated temporal trends in comorbid risk by race as measured by an obstetric comorbidity index.15 This comorbidity index provides weighted scores for comorbidity for individual patients based on the presence of specific diagnosis codes and demographic factors present in administrative data. Higher scores are associated with increased risk for severe morbidity. Factors in the comorbidity index include hypertensive diseases of pregnancy, multiple gestation, cardiac conditions, abnormal placentation, substance abuse, pre-existing diabetes, and previous cesarean delivery. In the initial study validating the comorbidity index, patients with the lowest score of 0 had a 0.68% risk of severe morbidity; a score greater than 10 was associated with a risk of severe morbidity of 10.9%.15 This comorbidity index was subsequently validated in an external population.16 Because the comorbidity index includes maternal age, we modified this scoring system excluding maternal age for the present analysis. Fifth, we determined whether risk for severe morbidity differed by race for women with the same comorbidity index score.
Demographic and hospital characteristics were evaluated by Nationwide Inpatient Sample racial categories. Hospital characteristics included bed size (small, medium, or large), location (urban vs rural), region (Northeast, Midwest, South, or West), and teaching status (teaching vs nonteaching). Demographic categories included year of delivery, insurance status (Medicaid, private, Medicare, other, uninsured), and zip code income quartile. Demographic comparisons and temporal trends were evaluated using the χ2 test. Adjusted risk ratios (RRs) for severe morbidity with 95% CIs as measures of effect accounting for demographic and hospital factors were derived from fitting a log-linear regression model. This model was subsequently repeated including the comorbidity index to account for medical and obstetric risk factors. Population weights within the Nationwide Inpatient Sample can be applied to create estimates for the entire U.S. population. These weights were used for this analysis. All analyses were performed with SAS 9.4.
A total of 1,724,694 estimated births to women aged 40–54 years were included in this analysis. The number of births to women in this age group increased over the study period from 73,946 in 1998 to 110,710 women in 2014 (Table 1). Ascertainment of race improved in the Nationwide Inpatient Sample from 2007 on with the proportion of unknown race decreasing from 26.7% in 2007 to 5.2% in 2014. The majority of women in all Nationwide Inpatient Sample racial categories delivered at urban and high-bed number hospitals. White and Asian or Pacific Islander women were more likely to reside in a top quartile income zip code than women of other races.
For women aged 40 years old or older, black women had the highest overall risk of severe maternal mortality with unadjusted risk 104% higher than white women (RR 2.04, 95% CI 1.98–2.04). Hispanic women (RR 1.28 95% 1.24–1.32), Asian or Pacific Islander women (RR 1.29, 95% CI 1.24–1.34), and Native American women (RR 1.20, 95% CI 1.03–1.39) were also at increased risk compared with white women (Table 2). Risk of severe morbidity increased over the study period from 1.6% in 1998–2000 (95% CI 1.5–1.7%) to 3.0% in 2012–2014 (95% CI 2.9–3.1%; P<.01) with risk among black women highest at both the beginning (2.4% in 1998–2000, 95% CI 2.2–2.6%) and the end (4.9% in 2013–2014, 95% CI 4.7–5.1%) of the study period (Fig. 1A). When the analysis was restricted to severe morbidity excluding transfusion, results were similar (Fig. 1B); Black women had the highest risk of severe morbidity with and without transfusion for every 3-year period from 1998 to 2014.
Figure 2 demonstrates proportionate changes in comorbid risk for white, black, Hispanic, and Asian or Pacific Islander women. For all four groups, the proportion of women with the lowest score of 0 decreased, and the proportions of women with a score of 1–2 and greater than 2 increased. Women with scores greater than 2 increased the most proportionately: 78% for white women, 109% for black women, 111% for Hispanic women, and 45% for Asian or Pacific Islander women. However, women with the highest comorbidity scores (greater than 2) represented a relatively small percentage of the overall population (Appendix 2, available online at http://links.lww.com/AOG/B123). Overall, black women were least likely to have a comorbidity score of 0 and most likely to have comorbidity scores of 1–2 and greater than 2.
Risk for severe morbidity increased with comorbidity index score. Risk for severe morbidity was lowest among women with a score of 0 (1.3%, 95% CI 1.3–1.3%), increased for women with a score of 1 or 2 (3.4%, 95% 3.4–3.5%), and was highest for women with scores less than 2 (12.1%, 95% CI 11.7–12.5%). For each of the three comorbidity index score categories (0, 1–2, greater than 2), black women had a significantly higher risk for severe morbidity than white women. Risk for severe morbidity with a score of 0 was 2.4% for black women, 1.5% for Asian or Pacific Islander women, 1.4% for Hispanic women, and 1.1% for white women (P<.01). For scores of 1 or 2, risk for severe morbidity was 5.4% for black women, 3.7% for Hispanic women, 3.9% for Asian or Pacific Islander women, and 3.0% for white women (P<.01). For scores of greater than 2, risk for severe morbidity was 14.2% compared with 13.8% for Hispanic women, 12.2% for Asian or Pacific Islander women, and 11.8% for white women (P<.01) (Fig. 3).
Appendix 3, available online at http://links.lww.com/AOG/B123, demonstrates temporal trends in preeclampsia, gestational diabetes, and cesarean delivery for women aged 40–54 years by race. Incidence of all three outcomes increased significantly for non-Hispanic white, non-Hispanic black, and Hispanic women. Black women were more likely than nonblack women to develop preeclampsia (9.1% vs 5.3%, P<.01) with the diagnosis of preeclampsia rising significantly from 8.0% of pregnancies among black women in 1998 (n=509/6,394, 95% CI 7.3–8.7%) to 11.9% of pregnancies in 2014 (n=1,570/13,140, 95% CI 11.4–12.5%) (P<.01) (Appendix 3A, http://links.lww.com/AOG/B123). Black women were also the more likely than nonblack women to undergo cesarean delivery (50.2% vs 44.8%, P<.01) with the cesarean delivery rate rising 53% for black women from 36.4% in 1998 (n=2,330/6,394, 95% CI 35.3–37.6%) to 55.6% in 2014 (n=7,310/13,140, 95% CI 0.55–0.56%) (Appendix 3C, http://links.lww.com/AOG/B123). Although gestational diabetes increased across all groups (Appendix 3B, http://links.lww.com/AOG/B123), Asian or Pacific Islander and Hispanic women were most likely to receive a diagnosis of gestational diabetes and saw the largest absolute increases in risk over the study period (absolute increases of 13.7–22.4% for Asian or Pacific Islander women and from 11.2% to 19.3% for Hispanic women, respectively, between 1998 and 2014).
Trends in select individual severe maternal morbidity diagnoses are demonstrated in Figure 4. For the entire study period, black women were significantly more likely than white women to experience acute renal failure (P<.01), acute respiratory distress syndrome (P<.01), stroke (P=.03), transfusion (P<.01), hysterectomy (P<.01), and disseminated intravascular coagulation (P<.01). Black women also experienced the absolute largest increase from 1998–2003 to 2010–2014 in risk for acute renal failure, disseminated intravascular coagulation, transfusion, and hysterectomy. For disseminated intravascular coagulation, by the end of the study period, risk was similar between black women and white women (55.8 vs 55.0 per 10,000 hospitalizations, respectively, from 2010 to 2014, P=.97). Stroke among black women did not change substantially over the study period with incidence of 10.1 per 10,000 deliveries in 1998–2003, 7.9 from 2003 to 2009, and 8.8 from 2010 to 2014 (P=.57).
In the adjusted model for severe morbidity, later year of delivery was associated with increased risk for severe morbidity (Table 2). With 1998 as a reference, delivering in 2012, 2013, and 2014 was associated with adjusted RRs of 1.90 (95% CI 1.78–2.03), 1.75 (95% 1.64–1.87), and 1.89 (95% CI 1.77–2.02), respectively. Black race was associated with an 81% increased risk of morbidity in the adjusted model for the entire study period (adjusted RR 1.81 95% CI 1.76–1.87) with white race as the reference; deliveries to Hispanic and Asian or Pacific Islander women were associated with increased risks of 16% (adjusted RR 1.16 95% CI 1.12–1.19) and 22% (adjusted RR 1.23 95% CI 1.18–1.28), respectively. Medicare insurance was associated with increased risk for severe morbidity (adjusted RR 1.85, 95% CI 1.70–2.01) compared with private insurance as was Medicaid insurance (adjusted RR 1.11, 95% CI 1.08–1.14). When the adjusted model was repeated including comorbidity, black race retained the significant association with increased risk for severe morbidity (adjusted RR 1.67 95% CI 1.62–1.72). With a score of 0 as a reference, comorbidity scores of 1–2 and greater than 2 were associated with adjusted RRs of 2.51 (95% CI 2.45–2.56) and 8.38 (95% CI 8.03–8.75), respectively (Appendix 4, available online at http://links.lww.com/AOG/B123).
Overall, there were 231 maternal deaths (13.4/100,000, 95% CI 11.8–15.2) including 71 deaths occurring among white women (8.6/100,000, 95% CI 6.9–10.9), 64 among black women (40.5/100,000, 95% CI 31.7–51.7), 48 among Hispanic women (18.5/100,000, 95% CI 13.9–24.5), and 11 among Asian or Pacific Islander women (11.3/100,000, 95% CI 5.4–17.4). Risk of death was 371% higher for black compared with white women (RR 4.71, 95% CI 3.36–6.61), whereas Hispanic women had more than twice the risk of death (RR 2.13, 95% CI 1.48–3.07) as white women.
In this analysis of delivery hospitalizations for women aged 40 years old or older, risk of severe morbidity increased across all racial groups with the largest absolute increase occurring among black women. A counterintuitive finding relative to mortality risk, which was nearly five times higher for black women compared with white women in our analysis, was that unadjusted risk for severe morbidity was only approximately doubled. Data from the CDC that comprehensively evaluates maternal death in United States found that for black women aged 40 years or older, mortality was three and half times that for white women the same age and more than 10 times that of the general population.9 The cause of the differential for black women aged 40 years or older between morbidity and mortality is unclear. Evidence from other specialties suggests that risk of death is not solely dependent on whether a patient experiences a complication, but also on how and where a patient with a complication is managed and cared for, a concept known as failure to rescue.17 Understanding specific factors related to failure to rescue responsible for maternal mortality among black women may be important in reducing disparities.
Underlying comorbidity may represent another important cause of risk differentials for mortality. In our analysis, comorbidity index scores were higher among black women. In addition to measured comorbidity, there are also likely differences in unmeasured comorbidity given underascertainment with diagnosis codes and that these codes capture the presence or absence of a condition but not severity. Underlying comorbidity may be of particular significance for women aged 40 years and older given assisted reproductive technology (ART), a factor we were not able to reliably measure. Assisted reproductive technology may be used differentially by women of different races and women who become pregnant by ART may be a particularly healthy self-selecting group with lesser comorbidity. Of note, we found adjusted and unadjusted risk for severe morbidity to be highest after 2009. The cause of this differential is unclear, but we speculate that this differential may in part be secondary to a smaller proportion of relatively lower risk women electing to undergo ART because of consequences of the preceding economic downturn. Minimizing comorbid risk through preconceptual consultation, optimization of medical conditions before pregnancy, and interconceptual contraception for at-risk women may be important in reducing overall maternal mortality disparities.
Findings from this analysis demonstrate the complexity and scope of the challenge in reducing risk for women aged 40 years or older in general and black women aged 40 years or older in particular. Severe morbidity rose continuously over the study period both across a number of diagnoses and as a composite. Conditions such as preeclampsia, cesarean delivery, and overall comorbid risk increased, all factors likely associated with the trends noted in severe morbidity. Findings from this analysis support that improving outcomes will likely require a broad range of interventions from improved interconceptual reproductive planning, to improved inpatient safety measures, to multidisciplinary optimization of medical management for at-risk patients. Black women in particular appear to be at disproportionately increased risk relative to measured comorbidity.
In interpreting this analysis, there are several limitations that are important to consider. Most importantly, data on race were missing for a large portion of the cohort with greater than 20% having unknown race from 1998 through 2007. If outcomes based on race differed significantly for patients with unknown compared with known race, reported trends may be biased. The Healthcare Cost and Utilization Project provides a number of recommendations for addressing missing data including imputation.18 We elected not to perform imputation because it was unlikely that missing data on race were random and nondifferential across the years of the Nationwide Inpatient Sample included in this analysis. To minimize missing data, prior analyses of racial disparities in maternal morbidity have used the Healthcare Cost and Utilization Project’s State Inpatient Database including states with high proportions of complete data on race.3 We elected to use the Nationwide Inpatient Sample because of the large, nationally representative sample, availability of a 17-year period for analysis, and, over the last 3 years of the study, low rates of missing data. A second limitation of this analysis is that this administrative data set does not provide information on hospital resources, infrastructure, and staffing, all of which contribute to maternal outcomes and risk. A third limitation is that we are not able to evaluate outpatient management including prenatal care, interdisciplinary care, and access to consultants. A fourth limitation is that for both comorbid exposures and severe morbidity outcomes, validity of the outcomes and ascertainment may vary with administrative data by diagnosis code. A fifth limitation is that valid coding regarding ART interventions before the pregnancy is not available. A sixth limitation is that mortality comparisons by race must be interpreted within the context of a large proportion of missing data and a relatively small number of maternal deaths; however, the risk differentials in this data set align with estimates from the CDC’s Pregnancy Mortality Surveillance System, which uses a more comprehensive system to evaluate maternal deaths. A seventh limitation is that in adjusting for comorbidity, diagnosis codes generally do not specify condition severity. For example, severity of underlying diabetes, renal failure, and chronic hypertension may represent important risk factors that we are not able to account for in our model. Finally, we are not able to determine the degree to which, or whether, prenatal care was attended.
In summary, this analysis found maternal comorbid risk and morbidity increasing for women aged 40 years or older with risk highest among black women. Additionally, among black women, risk was highest for preeclampsia, a number of specific morbidity diagnoses, and mortality, and black women experienced the largest increase in risk for many adverse outcomes. Given that morbidity risk was double for black women and that mortality risk was severalfold higher, improved understanding of failure to rescue may be important in reducing disparities.
1. Martin JA, Hamilton BE, Osterman MJ, Driscoll AK, Mathews TJ. Births: final data for 2015. Natl Vital Stat Rep 2017;66:1.
2. Hamilton BE, Martin JA, Osterman MJ, Curtin SC, Matthews TJ. Births: final data for 2014. Natl Vital Stat Rep 2015;64:1–64.
3. Creanga AA, Bateman BT, Kuklina EV, Callaghan WM. Racial and ethnic disparities in severe maternal morbidity: a multistate analysis, 2008-2010. Am J Obstet Gynecol 2014;210:435.e1–8.
4. Ogawa K, Urayama KY, Tanigaki S, Sago H, Sato S, Saito S, et al. Association between very advanced maternal age and adverse pregnancy outcomes: a cross sectional Japanese study. BMC Pregnancy Childbirth 2017;17:349.
5. Timofeev J, Reddy UM, Huang CC, Driggers RW, Landy HJ, Laughon SK. Obstetric complications, neonatal morbidity, and indications for cesarean delivery by maternal age. Obstet Gynecol 2013;122:1184–95.
6. Lisonkova S, Potts J, Muraca GM, Razaz N, Sabr Y, Chan WS, et al. Maternal age and severe maternal morbidity: a population-based retrospective cohort study. PLoS Med 2017;14:e1002307.
7. Joseph KS, Allen AC, Dodds L, Turner LA, Scott H, Liston R. The perinatal effects of delayed childbearing. Obstet Gynecol 2005;105:1410–8.
8. Jacobsson B, Ladfors L, Milsom I. Advanced maternal age and adverse perinatal outcome. Obstet Gynecol 2004;104:727–33.
9. Creanga AA, Syverson C, Seed K, Callaghan WM. Pregnancy-related mortality in the United States, 2011-2013. Obstet Gynecol 2017;130:366–73.
10. Howell EA, Egorova N, Balbierz A, Zeitlin J, Hebert PL. Black-white differences in severe maternal morbidity and site of care. Am J Obstet Gynecol 2016;214:122.e1–7.
11. Noskin GA, Rubin RJ, Schentag JJ, Kluytmans J, Hedblom EC, Smulders M, et al. The burden of Staphylococcus aureus infections on hospitals in the United States: an analysis of the 2000 and 2001 Nationwide Inpatient Sample Database. Arch Intern Med 2005;165:1756–61.
12. Wright JD, Herzog TJ, Tsui J, Ananth CV, Lewin SN, Lu YS, et al. Nationwide trends in the performance of inpatient hysterectomy in the United States. Obstet Gynecol 2013;122:233–41.
13. Kuklina E, Whiteman M, Hillis S, Jameieson D, Meikle S, Posner S. An enhanced method for identifying obstetric deliveries: implications for estimating maternal morbidity. Matern Child Health J 2008;12:469–77.
14. Centers for Disease Control and Prevention. Severe maternal morbidity in the United States. Available at: https://www.cdc.gov/reproductivehealth/maternalinfanthealth/severematernalmorbidity.html
. Retrieved January 20, 2018.
15. Bateman BT, Mhyre JM, Hernandez-Diaz S, Huybrechts KF, Fischer MA, Creanga AA, et al. Development of a comorbidity index for use in obstetric patients. Obstet Gynecol 2013;122:957–65.
16. Metcalfe A, Lix LM, Johnson JA, Currie G, Lyon AW, Bernier F, et al. Validation of an obstetric comorbidity index in an external population. BJOG 2015;122:1748–55.
17. Ghaferi AA, Birkmeyer JD, Dimick JB. Variation in hospital mortality associated with inpatient surgery. N Engl J Med 2009;361:1368–75.
18. Houchens R. Missing data methods for the NIS and the SID. 2015. HCUP Methods Series report #2015-01. Available at: https://www.hcup-us.ahrq.gov/reports/methods/2015_01.pdf
. Retrieved January 2, 2018.