Assessing the Role of Cardiac Disease in Severe Maternal Morbidity: A Population-Based Cohort Study

Authors: Kelly F. Darmawan, Elizabeth B. Sherwin, Stephanie A. Leonard, Danielle M. Panelli, Jonathan A. Mayo, Anna I. Girsen, Suzan L. Carmichael, Abha Khandelwal, Katherine Bianco

Abstract

Individuals with cardiac disease are at significantly increased odds of all types of severe maternal morbidity, especially cardiac and pulmonary events.

Cardiac disease is a major contributor to severe maternal morbidity (SMM) and mortality.^1–3^ The American College of Obstetricians and Gynecologists defines SMM as adverse labor and delivery outcomes that result in significant or long-term sequelae to maternal health.^4^ The Centers for Disease Control and Prevention defines SMM through diagnosis codes.^5^ In the United States, rates of SMM have been increasing over the past 2 decades.^6,7^

The increasing rates of SMM likely reflect the increasing number of high-risk pregnancies. As risk factors for noncongenital cardiac disease such as advanced maternal age, diabetes mellitus, and hypertension have increased, so have pregnancies in people with cardiac disease.^8–11^ An additional contributor is that earlier detection and improvements in the management of congenital cardiac diseases have led to an increasing number of reproductive-aged people with congenital cardiac disease who pursue pregnancy.^11^

Maternal cardiac disease has been associated with increased SMM, including nontransfusion SMM, during pregnancy and postpartum.^10,12–16^ Nontransfusion SMM has been studied in prior studies on SMM including in people with cardiac disease because blood transfusion alone without other SMM occurs often during delivery hospitalization and is nonspecific.^16^ It is important to identify areas of possible intervention to prevent SMM and ultimately to promote safe obstetric care in this population.^17^ Previous studies have focused on establishing risk of SMM and cardiac complications by severity of cardiac disease, which helps to stratify risk of people with cardiac disease, but are limited in determining what types of complications can be prevented.^12–15^ Another study that examined SMM by type in people with cardiac disease did not include the postpartum period.^16^ Assessing risk of SMM through 42 days postpartum is important because of the expected physiologic changes and because readmissions often occur in the 30 days after discharge from delivery hospitalization.^18^ In addition, although newer guidance recommends individualized timing of postpartum visits, postpartum visits have traditionally been at 6 weeks postpartum.^19^

The objective of this population-based study was to evaluate whether people with preexisting cardiac disease were at increased odds of nontransfusion SMM and specific types of SMM during delivery hospitalization through 42 days postpartum compared with those without cardiac disease.

METHODS

We conducted a population-based, retrospective cohort study using data from California deliveries from 2007 to 2020. The study dataset includes linked vital records and maternal and neonatal hospital discharge records obtained from the California Department of Health Care Access and Information.^20^ The California birth and fetal death certificates are filled out by both patients and medical staff and contain information on demographic characteristics, medical diagnoses, delivery complications, and infant outcomes. Race and ethnicity data were self-reported by individuals and collected from the California birth and fetal death certificates. Discharge records were obtained from all licensed acute care facilities in California and include International Classification of Diseases, 9th and 10th Revisions, Clinical Modification (ICD-9-CM and ICD-10-CM). The ICD-9 codes were used from 2007 through September 2015; the ICD-10 codes were used from October 2015 until 2020. Analyses, interpretations, and conclusions reached regarding birth and fetal death data are attributed to the authors of the linked records and not to the California Department of Public Health.^20^ Permission was obtained to use this database. This study was reviewed and approved by the Stanford IRB (43209) and the California State Committee for the Protection of Human Subjects (17-04-2932).

All singleton deliveries (live births and stillbirths) at 20–44 weeks of gestation to people 13–55 years old were included in this study. Multiple, repeated pregnancies to the same individual were allowed to maximize power given the rare exposures and outcomes. This approach is consistent with a prior study demonstrating minimal effects on statistical outcomes compared with restricting the cohort to one pregnancy per individual.^21^ Deliveries in people with diagnosis codes for cardiac disease with a noncardiac cause (eg, renal disease) were excluded to better evaluate the association between primary cardiac disease and SMM. Deliveries were excluded if an implausible height, weight, or body mass index (BMI) was documented for the birthing individual. Complete case analysis was conducted, excluding births with missing information for age, race and ethnicity, insurance type, education level, timing of prenatal care initiation, height, weight, or BMI.

Our cohort consisted of people with diagnoses of cardiac disease preexisting the delivery hospitalization. Cardiac disease was identified with ICD-9-CM and ICD-10-CM codes from delivery hospitalization records (Appendix 1, available online at http://links.lww.com/AOG/E524). Care was taken to use diagnosis codes for chronic cardiac conditions such as arrhythmias, congenital heart disease, valvular heart disease, acquired cardiomyopathy, or chronic heart failure. Diagnosis codes that referenced an acute cardiac problem or overlapped with diagnosis codes defined as SMM were not included to further isolate people with preexisting cardiac disease.

The primary outcome was nontransfusion SMM, which was defined as the presence of SMM indicators, except for blood transfusion, from delivery through 42 days postpartum. The Centers for Disease Control and Prevention and Alliance for Innovation on Maternal Health have defined SMM by ICD-9-CM and ICD-10-CM diagnosis codes (Appendix 2, http://links.lww.com/AOG/E524). These diagnosis codes have also been validated in prior studies to measure SMM.^22^ We chose nontransfusion SMM as the primary outcome because blood transfusion alone without other SMM occurs often during delivery hospitalization and is nonspecific. Transfusion has been excluded in other literature assessing SMM.^15,16,23^ We assessed SMM with transfusion as a secondary outcome.

We further categorized SMM indicators into cardiac SMM (acute myocardial infarction, aneurysm, cardiac arrest or ventricular fibrillation, conversion of cardiac rhythm, heart failure or arrest during surgery, or acute heart failure), pulmonary SMM (adult respiratory distress syndrome, pulmonary edema, temporary tracheostomy, or mechanical ventilation), renal SMM (acute renal failure), hemorrhage SMM (disseminated intravascular coagulation, shock, or hysterectomy), infection SMM (sepsis), obstetric SMM (amniotic fluid embolism, eclampsia, severe anesthesia complications, or air and thrombotic embolism), and other medical SMM (puerperal cerebrovascular disorders or sickle cell disease with crisis). This categorization was completed in similar fashion to prior literature examining SMM as an outcome.^23^

Demographic characteristics were selected a priori according to prior literature on SMM. These included maternal age, race and ethnicity, prepregnancy BMI, method of payment, education level, trimester of prenatal care initiation, parity, and mode of delivery. Race and ethnicity were included because prior literature showed that risk of SMM varied by race and ethnicity in people with cardiac disease.^16^ As an additional covariate, we included a modified expanded obstetric comorbidity score that has been validated to improve comparisons of nontransfusion SMM and SMM rates across patient populations using ICD-10 codes.^24,25^ The modified obstetric comorbidity score included 25 chronic and gestational comorbidities identified by ICD-9-CM and ICD-10-CM codes and excluded cardiac disease because this was our exposure (Appendix 3, http://links.lww.com/AOG/E524). All characteristics were compared in people with and those without cardiac disease with the Pearson χ^2^ test for categorical variables and Wilcoxon rank-sum test for continuous variables.

To assess associations between preexisting cardiac disease and SMM, we used multivariable logistic regression models to estimate odds ratios (ORs) with 95% CIs. A value of P<.05 was considered statistically significant. In the first set of models (model 1), we adjusted for patient characteristics that included age, race and ethnicity, method of payment, education level, trimester of prenatal care initiation, prepregnancy BMI, and parity. In the second set of models (model 2), we additionally adjusted for the modified obstetric comorbidity score and mode of delivery to assess odds of SMM independently of these factors. Sociodemographic factors and medical conditions were sequentially adjusted for in models 1 and 2 to determine how adjusting for sociodemographic factors affected the odds of SMM compared with adjusting for the modified obstetric comorbidity score and mode of delivery.

We then conducted additional analyses to assess the timing of nontransfusion SMM events in people with and people without cardiac disease. Among people with SMM, we classified the first nontransfusion SMM event as happening either during delivery hospitalization or at a separate postpartum hospital encounter after discharge until 42 days postpartum. The distributions of these first SMM events between people with and people without cardiac disease were then compared with the Pearson χ^2^ test. A value of P<.05 was considered significant for results of the Pearson χ^2^ test and Wilcoxon rank-sum test. All analyses were performed with SAS 9.4 and R 4.2.2.^26^

RESULTS

Among 5,848,486 deliveries, 30,359 (0.5%) were affected by maternal cardiac disease (Fig. 1). Of people with cardiac disease, the most common cardiac disease category was arrhythmias (45.2%) (Appendix 4, http://links.lww.com/AOG/E524). The next most common categories of cardiac disease were valvular disease (29.8%), congenital disease (14.6%), cardiomyopathy (6.1%), and presence of cardiac device in situ or transplantation (6.0%). The other categories of cardiac disease made up less than 5.0% of the cohort, including ischemic cardiac disease, myocarditis, pulmonary cardiac disease, and acquired heart failure.

Compared with those without cardiac disease in pregnancy, more individuals with cardiac disease were 35–39 years old, and had cesarean deliveries and elevated obstetric comorbidity scores (Table 1). However, that group also had a significantly higher proportion of people with higher education level (college-level education or higher), who had private insurance, and who initiated prenatal care in the first trimester. All characteristics in Table 1 were significantly different in people with cardiac disease compared with those without cardiac disease (P<.01).

People with cardiac disease in pregnancy had markedly higher rates of nontransfusion SMM (9.1% vs 0.8%) and any SMM (11.5% vs 1.7%) compared with those without cardiac disease (Table 2). When grouped by organ system, rates of SMM were significantly increased in all categories in people with cardiac disease (Fig. 2).

There was a significantly increased odds of nontransfusion SMM, any SMM, and each category of SMM in people with cardiac disease in the crude logistic regression model. After adjustment for demographic and social covariates in model 1, the odds of nontransfusion SMM were increased 12-fold and the odds of cardiac SMM were increased at 88-fold. The odds of other types of SMM and any SMM were either similar or mildly decreased after adjustment for these covariates (model 1, Table 3).

After adjustment for the obstetric comorbidity composite score and mode of delivery in addition to the demographic and social covariates (model 2), the odds of nontransfusion SMM were reduced but remained statistically significant (adjusted OR [AOR] 6.88, 95% CI, 6.57–7.19, Table 3). Similarly, the odds of any SMM (AOR 4.69, 95% CI, 4.50–4.88) and all categories of SMM were attenuated. The odds of cardiac SMM remained especially high (AOR 53.49, 95% CI, 49.82–57.40), as did the odds of pulmonary SMM (AOR 11.79, 95% CI, 10.84–12.81). The odds of other types of SMM (obstetric, infection, medical, renal, and hemorrhage) were also significantly increased. However, the odds of these respective SMM types were lower compared with nontransfusion, cardiac, and pulmonary SMM, ranging from a twofold to fivefold increase. Of these remaining types of SMM, obstetric SMM had the highest odds, and hemorrhage SMM had the lowest odds.

In an examination of the timing of first nontransfusion SMM events in people with cardiac disease, 91.9% occurred during the delivery hospitalization compared with 8.1% that occurred after discharge from delivery hospitalization through 42 days postpartum. Compared with people without cardiac disease, those with cardiac disease were significantly more likely to experience their first nontransfusion SMM event during delivery hospitalization (91.9% vs 75.5%, P<.001).

DISCUSSION

Cardiac disease is a major risk factor for SMM events.^15,16,27^ Our study builds on prior work by showing that people with cardiac disease have elevated odds of SMM across every type, beyond the previously recognized cardiac SMM and overall SMM events in aggregate.^15,16^ This work shows that cardiac disease is associated with maternal morbidity beyond the cardiovascular system. Similar to Denoble et al,^15^ we also adjusted for a validated obstetric comorbidity score and mode of delivery to assess the odds of SMM independently of these factors.^24^ This study found that when the odds of SMM were adjusted for sociodemographic factors (model 1), there was minimal difference in the OR. When additionally adjusted for the validated obstetric comorbidity score and mode of delivery, the odds of SMM continued to be significantly increased but were attenuated, showing that adjustment for the validated obstetric comorbidity score and mode of delivery was necessary to isolate the independent association between cardiac disease and SMM.

Cardiac output increases by almost 50% at the time of delivery and up to 60–80% immediately postpartum. Plasma volume increases postpartum, in part as a result of autotransfusion of up to 500 mL immediately after delivery, but increases again between days 3 and 6 postpartum because of fluid shifts.^28^ This increase in volume stresses the cardiovascular and pulmonary systems in those with existing cardiac disease, which likely explains the significantly increased risk of both cardiac and pulmonary SMM in our findings.

This work also found that there was a significantly increased odds of SMM in people with cardiac disease from delivery through 42 days postpartum, with the majority of first SMM events occurring during delivery hospitalization in people with cardiac disease. However, many people may be discharged while postpartum fluid shifts are still occurring, so close outpatient follow-up is needed. Because of these physiologic changes, it is prudent to include the postpartum period in an assessment of SMM in people with cardiac disease. In addition, people with cardiac disease should be carefully monitored before discharge and should have a plan for close follow-up. The American College of Obstetricians and Gynecologists recommends that people with cardiac disease should have postpartum follow-up with a primary care clinician or cardiologist within 7–14 days of delivery. They also recommend a team-based approach with a multidisciplinary team such as a maternal heart pregnancy team.^4^ Several studies have shown that patients followed up by a multidisciplinary program with cardiology and obstetrics have relatively decreased rates of cardiovascular events.^4,29–31^

Strengths of this study include that it was a large, contemporary study over a long period of time. This enabled us to assess rare but severe complications in pregnancy. Because this was a population-based study, there was no selection bias. Our cohort also used a large, diverse state population, which makes the results more generalizable. Rates and distribution of types of cardiac disease in our study population were also consistent with prior literature.^15,16,30^ Finally, this study excluded cardiac disease that was secondary to other causes such as renal disease because these comorbidities would likely increase the risk of SMM.

Limitations of this study included the use of diagnosis codes, which introduces potential misclassification bias, and inaccuracies resulting from coding errors. In a study on the validity of obstetric diagnosis and procedure codes on hospital discharge abstracts, the sensitivity of reporting cardiovascular disease was poor (12–27%) and of pregnancy-related comorbidities (ie, chorioamnionitis, preeclampsia) was moderate (60–80%).^32^ False-negative errors were generally more frequent than false-positive errors; thus, our findings are more likely to be underrepresentative of SMM events. The ICD diagnoses transitioned from the ICD-9 to the ICD-10 coding system on October 1, 2015. After the transition to ICD-10 codes for SMM, there was a significant decrease in the incidence of SMM that is likely attributed to the increased specificity of ICD-10 codes.^33^ In addition, due to the use of diagnosis codes, this study was unable to determine the severity or management of medical conditions.

In conclusion, people with cardiac disease are at significantly increased odds of all types of SMM indicator events, notably cardiac and pulmonary SMM. The odds of SMM remained remarkably elevated even after adjustment for obstetric comorbidities, showing an independent association between cardiac disease and SMM. These results will help clinicians inform people with preexisting cardiovascular disease who are considering pregnancy about pregnancy-related risks during preconception counseling. These findings also emphasize the need for preventive strategies to reduce the risk of SMM for people with cardiac disease, such as the use of a multidisciplinary team of clinicians to manage these patients during pregnancy and throughout the postpartum period.