Hypertensive Disorders of Pregnancy and Maternal Cardiovascular Disease Risk Factor Development

During reproductive life, approximately 15% of parous women have at least 1 pregnancy complicated by a hypertensive disorder, such as gestational hypertension or preeclampsia (1). Growing evidence suggests that these women are nearly twice as likely as those who are normotensive during pregnancy to develop cardiovascular disease (CVD) (24). Hypertensive disorders of pregnancy (HDP) may reveal subclinical CVD risk under the physiologic stress test of pregnancy, providing early insight into CVD risk that might be leveraged to identify high-risk women for targeted prevention from an early age (1, 5). Although the 2011 American Heart Association guidelines recommend that clinicians evaluate CVD risk by screening for a history of HDP, few data exist on which risk factors should be screened for as well as the frequency and timing of screening (6). Hypertensive disorders of pregnancy have been consistently linked to future chronic hypertension despite the fact that blood pressure returns to normal during the postpartum period (2, 712). Women with a history of HDP have higher risks for impaired glucose tolerance and insulin resistance and a 3- to 4-fold increased risk for type 2 diabetes mellitus (T2DM) (9, 1322). Those with a history of preeclampsia also have higher levels of total and low-density lipoprotein cholesterol and triglycerides; however, these differences are not consistently statistically significant (16, 1821, 2325). Many previous studies were limited by small sample size; short follow-up; or incomplete adjustment for potential confounders, such as prepregnancy smoking status, body mass index (BMI), and family history. Further, although these associations have been observed over variable lengths of follow-up, little is known about the specific timing of risk factor development, which is critical to inform screening guidelines. We examined associations of gestational hypertension and preeclampsia with development of chronic hypertension, T2DM, and hypercholesterolemia. These associations were evaluated in a large longitudinal cohort study with up to 50 years of follow-up from first birth. Methods Cohort Description and Selection The NHS II (Nurses' Health Study II) is a prospective cohort study of 116429 female U.S. registered nurses who were enrolled at age 25 to 42 years in 1989. Participants are followed biennially via questionnaires, which collect information on lifestyle, health-related behaviors, and incident disease. Study Sample Analyses were restricted to participants who responded to the 2009 questionnaire (n= 76840), which allowed pregnancies to be dated and linked to specific complications. We excluded women who were nulliparous (n= 13253), those who were missing a valid year of first pregnancy (n= 12), those younger than 18 years or older than 45 years at their first birth (n= 846), and those who were missing gestation length or had a value that was incompatible with the pregnancy outcome reported (n= 292). Women also were excluded if they reported chronic hypertension, type 1 or 2 diabetes, hypercholesterolemia, myocardial infarction, or stroke before their first pregnancy (n= 2470) or if they were missing the date of diagnosis or reported diagnosis of these conditions before 1980 (which precluded dating of those events) (n= 1210). Finally, because undetected chronic hypertension before pregnancy may be incorrectly captured as incident chronic hypertension directly after pregnancy, we excluded women who reported chronic hypertension within 1 year after their first birth (n= 86). This yielded an analytic sample of 58671 women. This analysis was approved by the Institutional Review Board at Brigham and Women's Hospital. Hypertensive Disorders of Pregnancy In 2009, women retrospectively reported their complete pregnancy history. Hypertensive disorders of pregnancy were self-reported as pregnancy-related high blood pressure (gestational hypertension) or preeclampsia/toxemia. The primary analysis focused on the first pregnancy because this is when HDP predominantly occurs (24). To assess the validity of self-reported preeclampsia, we reviewed medical records of 598 women who reported preeclampsia on biennial questionnaires from 1991 to 2001 for provider report of preeclampsia or evidence of gestational hypertension (new-onset high blood pressure [systolic blood pressure 140 mm Hg or diastolic blood pressure 90 mm Hg] after 20 weeks' gestation) and proteinuria (protein excretion 300 mg per 24 hours, proteincreatinine ratio 0.3, or dipstick reading 1+) (26). There were 411 cases of preeclampsia confirmed by medical records, for a positive predictive value of 69%. Given the complexity of validating preeclampsia (confirming normotension before 20 weeks and elevated blood pressures and proteinuria after 20 weeks), several components of the medical record are required. We excluded 136 medical records with insufficient information available for validation (for example, those that were missing laboratory data or prenatal and/or labor and delivery records), resulting in a positive predictive value of 89%. Having complete medical record information for all 598 women would likely have resulted in a positive predictive value between 69% and 89%. Recurrent HDP was analyzed in a secondary analysis with follow-up starting at age 40 years. This analysis was restricted to 45815 parous women who had not experienced a CVD event or developed CVD risk factors of interest by age 40 years and had no additional pregnancies at this age or later. CVD Risk Factors Risk factors for CVD (chronic hypertension, T2DM, and hypercholesterolemia) were self-reported on biennial questionnaires beginning in 1989. The 1989 questionnaire retrospectively captured any physician diagnoses of high blood pressure (excluding during pregnancy), diabetes: not during pregnancy, and elevated cholesterol and the year of diagnosis (before 1980, 1980 to 1984, or 1985 to present). Women prospectively reported incident diagnoses of CVD risk factors on biennial questionnaires beginning in 1991. The midpoint of each date range was assigned as the year of diagnosis for chronic hypertension and hypercholesterolemia; for T2DM, the year of diagnosis was obtained via a supplemental questionnaire. Previous validation of self-reported high blood pressure in NHS II indicated good agreement, with sensitivity of 94% and specificity of 85% (27). Women who reported a new diagnosis of diabetes received a supplemental questionnaire to report diagnostic test results, symptoms, and treatment. This information was used to classify cases into categories proposed by the National Diabetes Data Group and the American Diabetes Association (2830). Information on self-reported use of cholesterol-lowering medication has been collected since 1999. We defined hypercholesterolemia as self-report of hypercholesterolemia or cholesterol-lowering medication use. Self-reported hypercholesterolemia was validated in a similar cohort and showed a positive predictive value of 86% and a negative predictive value of 85% (31). Lifestyle Factors and Medical History In 1989, participants reported height, current weight, and weight at age 18 years. Participants updated their weight on all biennial questionnaires. Body mass index was calculated from reported height and weight at age 18 years and was updated every 2 years from 1989 to 2013. Body mass index was derived for ages at which weight was not reported, with incorporation of data on weight at age 18 years; weights reported on each questionnaire; and somatograms at ages 20, 30, and 40 years (see the Appendix). A previous validation study in NHS II found high correlations between physical examination records and both recalled weight at age 18 years (r= 0.87) and self-reported height (r= 0.94) (32). Race/ethnicity, family history of chronic hypertension, and strenuous physical activity at ages 18 to 22 years were reported at baseline. History of smoking, alcohol consumption, and oral contraceptive use were also reported in 1989 and updated during follow-up. Biennial questionnaires after 1989 queried participants about family history of diabetes, parental education, and diet. Food-frequency questionnaires were used to derive a dietary quality score from the 2010 Alternative Healthy Eating Index (33). Self-reported physician diagnoses of myocardial infarction and stroke were verified through medical record review. Prepregnancy information was drawn from the biennial questionnaire immediately before the first pregnancy. Because most first births (85%) occurred before baseline, health-related behavior in high school and within varying age ranges from 13 through 42 years that was retrospectively reported in 1989 was used to assign prepregnancy values for these women. Statistical Analysis Characteristics of the analytic sample were age-standardized and stratified by HDP status in the first pregnancy (Table 1). We used Cox proportional hazards models to estimate associations between HDP in the first pregnancy and chronic hypertension, T2DM, and hypercholesterolemia (28). Women contributed person-time from their first birth until development of the CVD risk factor of interest, occurrence of a CVD event (nonfatal myocardial infarction, fatal coronary heart disease, or nonfatal or fatal stroke), death, the last returned questionnaire, or 2013. They also were censored at antihypertensive medication use for the chronic hypertension analysis and at type 1 diabetes diagnosis for the T2DM analysis. For the 85% of women who delivered their first child before 1989, the analysis included an average of 9.8 years (SD, 5.5) of follow-up before enrollment. Table 1. Age-Standardized Characteristics of NHS II Participants, by Hypertensive Disorders in First Pregnancy* Log-rank tests were used to determine whether the distributions of age at and time to CVD risk factor development differed between HDP groups. We calculated multivariable-adjusted hazard ratios (HRs)