Risk Factors and Timing of Acute Myocardial Infarction Associated With Pregnancy: Insights From the National Inpatient Sample

The data that support the findings of this study are available from the corresponding author upon reasonable request. We utilized the National Inpatient Sample (NIS), collected by the Healthcare Cost and Utilization Project. It is the largest, publicly available all‐payer inpatient healthcare database in the United States and contains ≈7 million unweighted stays yearly and 35 million weighted inpatient visits. We included discharge records related to pregnancy, labor and delivery, and postpartum periods from 2003 to September 2015 and utilized the International Classification of Diseases, Ninth Revision (ICD-9) codes to identify discharge records. We did not utilize data beyond September 2015 because hospital administrative data began using International Classification of Diseases, Tenth Revision, Clinical Modification, Procedure Coding Classification System (ICD-10-PCS). The data within the NIS are de‐identified and therefore International Review Board approval was not required for this study.

Diagnosis and procedure codes were used to identify records of pregnancy, labor and delivery (cesarean or vaginal), and postpartum (Data S1). Postpartum diagnosis is defined by ICD-9 code and not by time. AMI was further defined by diagnosis codes. Records with age <18 or >55 years and male or unknown sex were excluded. Patient characteristics were obtained based on status of pregnancy, labor and delivery or, postpartum, only female sex, age categories of 18 to 24, 25 to 29, 30 to 34, 35 to 39, 40 to 44, 40 to 55, race/ethnicity of White, Black, Hispanic, Asian or Pacific Islander, or other/unknown, income quartile of 0 to 25th percentile, 26th to 50th percentile (median), 51st to 75th percentile, 76th to 100th percentile or other, and primary expected payer of Private insurance, Medicaid, Medicare, or Other/unknown. AMI was defined by type, location, and cause. Specific comorbidities were considered and identified by diagnosis codes. Elixhauser Comorbidity Index Scores were utilized through the Healthcare Cost and Utilization Project's Elixhauser Comorbidity Software to assign 2 variables, readmission score and in‐hospital mortality, which identify comorbidities.^{10, 11} Complications in labor and delivery were also identified. Records defined as having eclampsia (severe preeclampsia, mild preeclampsia, and preeclampsia or eclampsia superimposed on pre‐existing hypertension and unspecified hypertension complicating pregnancy childbirth or the puerperium were utilized) included those with hypertensive disorders of pregnancy. MACCE (defined as arterial embolism and thrombosis, acute renal failure, arrhythmia, bleeding/transfusion, cardiac arrest, cardiac complications of anesthesia or other sedation in labor and delivery, cardiogenic shock, heart failure, in‐hospital death, obstetrical pulmonary embolism, postpartum hemorrhage, respiratory failure or arrest, stroke, atheroembolism, pulmonary embolism, and infarction) and obstetric event rates were obtained.

χ² tests or χ² tests with exact P values based on Monte Carlo simulation when small cell counts (<5) existed in 2‐way contingency tables were utilized to examine the marginal association between categorical variables and AMI, as well as between categorical variables and MACCE among patients with AMI. Wilcoxon rank sum tests were used to compare unadjusted marginal differences in continuous variables between patients with and without AMI and between patients with AMI with and without MACCE. The status of records and other factors related to each outcome that were significant (P<0.05) based on univariate analysis were further considered in multivariable logistic regression models. The linear trends in the rate of AMI and eclampsia among AMI records over years were examined using the complex survey analysis method with officially provided trend weight for records in years 2003 to 2011 and with discharge weight for records in years 2012 to 2015.¹² The SURVEYFREQ procedure was used to generate yearly frequency of records with AMI and any type of eclampsia among AMI records. Stratum identifier and hospital identifier in NIS data files were used as stratification variable and cluster variable, respectively. In order to correctly account for rate and make accurate inferences about the national populations as suggested by Berglund¹³ and West et al,¹⁴ all NIS records from year 2003 to year 2015 remain in the analysis and interested subgroup analysis was carried out using the domain options. P value for year was created from the SURVEYLOGISTIC procedure to assess the annual trend over time. Statistical analysis was performed using SAS 9.4 (SAS Institute Inc., Cary, NC) and significance level was set at 0.05.

From 2003 to 2015, a total of 11 297 849 records for pregnancy, labor and delivery, and postpartum were extracted and among these records, 913 instances of AMI were recorded (0.008%). There was a significant increasing in trend for AMI in pregnancy from 2003 to 2015 (P=0.0005; Figure 1). Of these 913 records, 111 (12.2%) women experienced AMI during labor and delivery, 338 (37.0%) during pregnancy, and 464 (50.8%) during the postpartum period. Among the women who experienced AMI, 661 (72.4%) were age ≥30 years and this increased with age (Table 1). Most women were White (38.4%), followed by those of Black race (23.4%) as secondarily most frequent. A large proportion of records were described as insured by public insurance (Medicaid or Medicare, 44.9%). Most (55.2%) were described as in <50th median percentile for income quartile. Many patients were described as having coronary artery disease (CAD) as a prior comorbidity (33.6%), with a relative minority noted with prior percutaneous coronary intervention, myocardial infarction, or coronary artery bypass grafting in the AMI group. Prior tobacco use was common (18.9%), as was prior hyperlipidemia (13.0%).

There was no cause of AMI determined for 59.4% of patient records; cause was found in 40.6%. Of these, the cause was coronary atherosclerosis of the native coronary artery (28.3%) or atherosclerosis (0.1%), other specified forms of chronic ischemic heart disease (3.4%), chronic total occlusion of coronary artery (1.3%), dissection of coronary artery (15.0%), primary and secondary thrombophilia hypercoagulable state (2.3%), arterial embolism, and thrombosis (0.7%).

Table 2 describes the location and timing of AMI as well as care patterns for revascularization. Notably, 60% of AMI were subendocardial. In addition, only 51.6% received any recorded form of revascularization.

Table 3 describes the timing and frequency of MACCE events in AMI records. Most major events occurred postpartum, including death, heart failure, arrhythmia, cardiogenic shock, and cardiac arrest, among others. Events during pregnancy were next most frequent.

Obstetric complications were recorded and analyzed among this population of women who experienced AMI and are represented in Table 4. These complications were found to be more common in the AMI population. Considering specifically hypertensive syndromes, 18.3% of this population experienced any type of eclampsia/preeclampsia. Although there was a significantly increasing in trend for AMI in pregnancy, there was no significant increase in trend noted for eclampsia (P=0.1571).

Based on univariate analysis for association between AMI and potential factors (patients' age, race∕ethnicity, primary expect payer, income quartile, records status, length of stay, Elixhauser Comorbidity Index Scores of readmission, Elixhauser Comorbidity Index Scores of in‐hospital mortality, comorbidities, obstetric complications, and complications), significant factors were adjusted in a multivariable logistic regression model (Figure 2 and Table S1). The greatest predictor of AMI during pregnancy was known CAD (odds ratio [OR], 517.4; 95% CI, 420.8–636.2). Other cardiac factors that were strongly associated with AMI were heart failure (OR, 8.2; 95% CI, 1.9, 35.2), prior valve replacement (OR, 6.4; 95% CI, 2.4, 17.0), and atrial fibrillation (OR, 2.7; 95% CI, 1.5–4.7). In terms of known risk factors for traditional atherosclerosis, hyperlipidemia, obesity, and smoking history, these were all significantly associated with AMI. Other substance abuse and thrombophilia were significant for AMI as well. Hypertensive syndromes of pregnancy such as eclampsia (OR, 6.0; 95% CI, 3.3, 10.8) and preeclampsia (OR, 3.2; 95% CI, 2.5, 4.2) were significant risk factors for AMI. Black race (OR, 1.3; 95% CI, 1.0, 1.5) was notably significant. Obstetric risk factors included postpartum hemorrhage (OR, 2.5; 95% CI, 1.9, 3.2), placental abruption (OR, 1.6; 95% CI, 1.0, 2.3), uterine rupture (OR, 4.2; 95% CI, 1.4, 13.0), postpartum infection (OR, 4.0; 95% CI, 2.7, 5.7) and thrombotic event (OR, 3.8; 95% CI, 2.3, 6.2).

Risk factors for MACCE are shown in Figure 3 and Table S2. Factors were significant variables based on univariate analysis between MACCE and potential factors (patients' age, race/ethnicity, primary expected payer, income quartile, records' status, AMI type and location, length of stay, Elixhauser Comorbidity Index Scores of readmission, Elixhauser Comorbidity Index Scores of in‐hospital mortality, comorbidities, obstetric complications, and complications). In terms of timing, women in labor and delivery (OR, 3.4; 95% CI, 2.0, 5.7) or in pregnancy (OR, 3.7; 95% CI, 2.5, 5.3) had higher odds of MACCE than those in the postpartum. In terms of location, inferoposterior myocardial infarction was most likely among other locations to have MACCE (OR, 8.2; 95% CI, 1.2, 54.0). Those with preeclampsia or eclampsia superimposed on pre‐existing hypertension and unspecified hypertension complicating pregnancy childbirth or the puerperium were at increased risk of MACCE (OR, 2.3; 95% CI, 1.3, 3.9). We also found that that having a prior percutaneous coronary intervention (OR, 6.6; 95% CI, 1.4, 31.2), or fluid and electrolyte imbalance (OR, 2.4; 95% CI, 1.3, 4.2) were risk factors of MACCE in patients with AMI.

Our study has several important limitations. The NIS was especially valuable in studying AMI, which has a low incidence in pregnancy, because it allowed for a large, contemporary, representative US sample of records. This data set has been utilized in the past to study hospitalization trends and its predictors in patients with pregnancy and other forms of heart disease, such as preexisting cardiomyopathy, congenital heart disease, and myocardial disorders.^{23, 24} However, the data are restricted to inpatient, labor‐ and delivery‐related hospitalizations, and maternal data on disease verification among others were not available. Deaths, events, or hospitalizations after the index labor and delivery hospitalization would not be included. Late maternal mortality is a real and devastating issue. Although ICD-9 codes have imperfect sensitivity and specificity, prior data do show a high level of accuracy (>90%) in ascertaining cardiovascular disease diagnoses.²⁵ Undercoding or miscoding are possibilities, although this would be unlikely to bias the results, because even a small number of misclassifications would not have a sizable effect on summary estimates of the large number of records included herein. Investigating the role of increased body mass index as a risk factor for AMI in this population was limited because of body mass index of the pregnant population being confounded by gestational weight gain. Our inability to distinguish whether increased body mass index was caused by obesity at the start of pregnancy or caused by the gestational weight gain led to the use of the diagnosis code of obesity rather than numerical body mass index value. Limitations on the cause of AMI were apparent and warrant further study in prospective registries. Moreover, the NIS does not distinguish whether a condition is present on admission; therefore, verification of preexisting diagnoses can be difficult; furthermore, certain aspects of patient history, imaging data, laboratory values, medications, and long‐term follow‐up were not available for analysis. Despite these limitations, it is validating that our data substantiated existing paradigms about AMI, including cardiac and obstetric complications.