Preconception Cardiovascular Risk Factor Differences Between Gestational Hypertension and Preeclampsia: Cohort Norway Study.

Preconception predictors of gestational hypertension and preeclampsia may identify opportunities for early detection and improve our understanding of the pathogenesis and life course epidemiology of these conditions. Female participants in community-based Cohort Norway health surveys, 1994 to 2003, were prospectively followed through 2012 via record linkages to Medical Birth Registry of Norway. Analyses included 13 217 singleton pregnancies (average of 1.59 births to 8321 women) without preexisting hypertension. Outcomes were gestational hypertension without proteinuria (n=237) and preeclampsia (n=429). Mean age (SD) at baseline was 27.9 years (4.5), and median follow-up was 4.8 years (interquartile range 2.6-7.8). Gestational hypertension and preeclampsia shared several baseline risk factors: family history of diabetes mellitus, pregravid diabetes mellitus, a high total cholesterol/high-density lipoprotein cholesterol ratio (>5), overweight and obesity, and elevated blood pressure status. For preeclampsia, a family history of myocardial infarction before 60 years of age and elevated triglyceride levels (≥1.7 mmol/L) also predicted risk while physical activity was protective. Preterm preeclampsia was predicted by past-year binge drinking (≥5 drinks on one occasion) with an adjusted odds ratio of 3.7 (95% confidence interval 1.3-10.8) and by past-year physical activity of ≥3 hours per week with an adjusted odds ratio of 0.5 (95% confidence interval 0.3-0.8). The results suggest similarities and important differences between gestational hypertension, preeclampsia, and preterm preeclampsia. Modifiable risk factors could be targeted for improving pregnancy outcomes and the short- and long-term sequelae for mothers and offspring.

Hypertensive disorders of pregnancy are prevalent complications and include preexisting chronic hypertension, gestational hypertension and preeclampsia, and preeclampsia superimposed on chronic hypertension.[1] Among these conditions, preeclampsia associates with the most significant immediate risks to offspring and mother,[1,2] increases women’s long-term risk of end-stage renal disease,[3] and cardiovascular morbidity and mortality.[4-8] Preeclampsia is conceptualized into 2 primary stages: the first being altered placental perfusion as a result of abnormal early trophoblast growth and differentiation, poor placentation, or other pathologies and the second involving maternal responses to placental factors excreted because of a dysfunctional placenta.[9-17] Further, the pathophysiology of preterm and late-onset preeclampsia may differ with poor placentation being more important for preterm than term preeclampsia.[18]

Preexisting cardiovascular risk factors may identify women at risk for de novo hypertensive disorders of pregnancy, where pregnancy acts as a metabolic and vascular stress test for women with underlying acquired or genetic predispositions.[19] Gestational hypertension and preeclampsia and term and preterm preeclampsia may have important etiologic differences, but no study to date has compared the preconception risk factor differences between these outcomes. We, therefore, evaluated the extent of differences and similarities in the pregravid cardiovascular risk factors associated with gestational hypertension, preeclampsia, and preterm preeclampsia using a prospective study design linking regional health surveys with subsequent pregnancies identified in the Norwegian Medical Birth Registry.

We are aware of only 2 earlier studies, of substantially smaller sample sizes, that evaluated preconception risk factors for their ability to predict preeclampsia.[20,21] As these studies had discrepant findings, perhaps owing to the smaller sample sizes, the current analyses provide a more robust evaluation of preconception risk factors. Also, the current study provides novel preconception risk factor information regarding physical activity and binge drinking and is unique in the literature for its evaluation of preconception differences and similarities in risk factors for gestational hypertension and preeclampsia.

Methods

Participants in Cohort Norway (CONOR) health surveys (1994–2003) were linked, using a national identification number, to the Medical Birth Registry of Norway for births subsequent to CONOR participation (through to December 31, 2012). The Materials and Methods in the online-only Data Supplement provides additional details and the study design figure. The reproductive-aged women from CONOR health surveys[22] came from 3 regions (24.2% from Oslo, 49.9% from Nord-Trøndelag, and 21.3% from Troms). An earlier study evaluated pregravid risk factors in 3494 women in Nord-Trøndelag: a subset of the current analyses.[21] The majority of CONOR participants were ethnic Norwegians, but only 61.9% of women of reproductive age in Oslo surveys were born in Norway, given an immigrant survey component, in contrast to 97.5% in Nord-Trøndelag and 95.0% in Troms.

Record linkages identified 17 320 births with a mother that participated in CONOR before delivery (Figure S1 in the online-only Data Supplement). Exclusions included preexisting hypertension, nonviable births, mother pregnant during or delivered <1 year before CONOR participation, and multiple birth pregnancies, resulting in 13 217 singleton births for analyses (representing 8321 women; 1.59 births/woman).

CONOR Assessments

Assessments included height and weight, past-year leisure-time light and vigorous physical activity,[23] alcohol consumption frequency and binge drinking,[24] blood pressure, nonfasting lipids,[22] and a family history of diabetes mellitus, stroke, or myocardial infarction before 60 years of age in first-degree relatives. Analyses used the average of the last 2 of 3 systolic and diastolic blood pressure readings taken by an automatic device (DINAMAP, Criticon, Tampa, FL). Binge drinking was defined as ≥5 drinks/d at least once in the past year. Those who reported drinking alcohol but who did not answer the binge drinking question were assigned to a nonresponse category. Binge drinking frequency was also evaluated (none, 1–5, and ≥6 binges in past year). Binge drinking was not assessed in Nord-Trøndelag; otherwise, missing data were low (<0.1%) for the majority of parameters, with the exception of physical activity (7%) and smoking (5%).

Definition of Outcomes

Gestational hypertension was defined as hypertension diagnosis after 20 weeks of gestation (systolic BP ≥140 mm Hg or a diastolic BP ≥90 mm Hg, or both). Preeclampsia diagnosis required the additional presence of proteinuria (≥0.3 g in 24 hour urine or ≥1 point increase on a urinary dipstick).[25] Term and preterm preeclampsia were defined based on gestational age at delivery (≥37 or <37 weeks or when gestational age was missing (n=287/13 217), having a birth weight ≥2500 g or <2500 g. Gestational age was determined by ultrasound for 80.0%, last menstrual period for 17.8%, and birth weight for remaining 2.2% of pregnancies.

Statistical Methods

Descriptive characteristics are reported as mean (SD) and percent. Multivariable multinomial logistic regression analyses provided odds ratios (OR) and 95% confidence intervals (CI) of characteristics for their prediction of gestational hypertension and preeclampsia and in a separate analyses of term and preterm preeclampsia. Multivariable models included parity (0, 1, or 2+), length of follow-up, baseline age (years), daily smoking (yes versus no), pregravid diabetes mellitus, and a history of gestational hypertension or preeclampsia in a prior pregnancy (obtained via record linkages to pre-CONOR pregnancies), educational level (≤12, 13–16, ≥17 years), marital status (married/common law partner versus other), and region of survey (Oslo versus other). Mother’s pseudo-ID was entered as a cluster variable. When lipids were evaluated, oral contraceptive use was added to the multivariable model.

Each potential additional risk factor was evaluated separately with the above mentioned parameters. Risk factors evaluated included physical activity (≥3 hours per week of light or vigorous activity on average in the past year), body mass index classifications (<25, 25–29.9, and ≥30 kg/m2), oral contraceptive use (yes versus no), alcohol consumption frequency (≥1/week, 1–3 times/month, and <1/month which included abstainers), binge drinking, a high total cholesterol/high-density lipoprotein (HDL) cholesterol ratio (≥5 versus lower), a high triglyceride level (≥1.7 mmol/L versus lower), and blood pressure status (normal: systolic BP <130 mm Hg and diastolic BP <85 mm Hg; elevated: systolic BP 130–139 mm Hg or diastolic BP 85–89 mm Hg; hypertensive: systolic BP ≥140 mm Hg or diastolic BP ≥90 mm Hg). Family history of chronic diseases was evaluated in unadjusted analyses. Selected postestimation tests were conducted to evaluate whether differences in regression coefficients between outcome groups were statistically significant. Stata 12 (StataCorp LP, College Station, TX) was used in analyses; significance was determined by P<0.05.

Sensitivity Analyses

Three sensitivity analyses were conducted: (1) restricted to those with a follow-up <7.2 years (approximately the 70th percentile); (2) restricted to nulliparous women; and (3) for preterm preeclampsia, restricted to preterm births.

Results

Average age (SD) at the time of the baseline CONOR assessment was 27.9 years (4.5). At baseline, 24.1% patients had a higher education, 21.3% were married or had a common law partner, and 55.9% engaged in physical activity of ≥3 hours per week (Table 1). The median follow-up was 4.8 years (interquartile range 2.6–7.8; maximum 17.5 years). A total of 666 (5.0%) pregnancies were affected by either gestational hypertension (without proteinuria) or preeclampsia, of whom 114 were preterm and 315 term preeclampsia. Only 23 of 237 gestational hypertensive pregnancies resulted in preterm deliveries. The percent small-for-gestational age by sex (<10th percentile) was 16.5%, 36.8%, and 14.0% for gestational hypertension, preterm, and term preeclampsia groups, respectively, and the percent very small-for-gestational age (<2.5th percentile) was 5.9%, 14.9%, and 3.8%, for the 3 groups, respectively.

Table 1.

Preconception Baseline Demographic and Cardiovascular Risk Factors: Cohort Norway and Medical Birth Registry of Norway (N=13 217)*

A family history of diabetes mellitus and women’s pregravid diabetes mellitus predicted both gestational hypertension and preeclampsia, whereas a family history of myocardial infarction before 60 years of age predicted preeclampsia, but not gestational hypertension (postestimation test for differences in coefficients, P=0.053; Table 2). A family history of stroke predicted the combined outcome of gestational hypertension or preeclampsia (OR 1.5; 95% CI: 1.02–2.10), with similar but nonsignificant ORs noted for the 2 outcomes separately evaluated (Table 2).

Table 2.

Odds Ratios (OR) and 95% Confidence Intervals (CI) for Gestational Hypertension and Preeclampsia by Family History of Disease and Prepregnancy Diabetes Mellitus: Cohort Norway and Medical Birth Registry of Norway (N=13 217)*

Physical activity was protective for preeclampsia (OR 0.8; 95% CI 0.61–0.97) and particularly for preterm preeclampsia (OR 0.5; 95% CI 0.32–0.76), but not for gestational hypertension (Table 3). Body mass index classifications and baseline hypertensive status predicted both outcomes, albeit with stronger effects noted for gestational hypertension than for preeclampsia (postestimation tests, P<0.05). Further, the Oslo region had a greater risk of gestational hypertension (OR 1.9; 95% CI 1.38–2.67) than the other regions, but no regional differences in preeclampsia were noted. Baseline educational attainment, marital status, smoking, and oral contraceptive use were not significantly related to gestational hypertension or preeclampsia (Table 3). A high total cholesterol/HDL cholesterol ratio predicted both gestational hypertension and preeclampsia. In contrast, an elevated triglyceride level only predicted preeclampsia.

Table 3.

Preconception Risk Factors for Gestational Hypertension and Preeclampsia: Cohort Norway and Medical Birth Registry of Norway (N=13 217)*

Weekly alcohol consumption relative to none or less than once a month was associated with lower risk of preeclampsia (OR 0.7; 95% CI 0.48–0.95) and term preeclampsia (OR 0.7; 95% CI 0.47–0.99; Table 3). In the analyses limited to the subcohort with available binge drinking data, weekly alcohol consumption (adjusted for binge drinking and other covariates) was associated with an OR for preeclampsia of 0.5 (95% CI 0.27–0.78) with a stronger protective effect noted for preterm preeclampsia (Table 4). Binge drinking (adjusted for alcohol consumption frequency and other covariates) was associated with increased risk of preeclampsia (OR 1.8; 95% CI 1.16–2.92) with an especially strong association noted for preterm preeclampsia (OR 3.7; 95% CI 1.25–10.78). However, there was no evidence for a binge frequency dose–response effect given that the OR associated with 1 to 5 binges was the same as that associated with ≥6 binges in the past year (Table 4).

Table 4.

Subcohort Preconception Past-Year Alcohol Consumption Frequency and Binge Drinking in Gestational Hypertension and Preeclampsia: CONOR and Medical Birth Registry of Norway (N=6488)*

The adverse effects of binge drinking for preterm preeclampsia remained significant when analyses were restricted to preterm births (n=703; OR 3.2; 95% CI 1.07–9.57), restricted to pregnancies occurring within 7.2 years of follow-up (n=9419; OR 3.8; 95% CI 1.11–12.81), and when restricted to nulliparous women (n=3975; OR 4.5; 95% CI 1.15–17.30).

In contrast, the protective effect of weekly alcohol consumption was not consistently observed in the sensitivity analyses (ie, a protective association was noted when restricting analyses to within 7.2 years of follow-up, but not when restricting analyses to nulliparous women or to preterm deliveries).

The protective effect of physical activity for preterm preeclampsia, however, persisted in analyses limited to nulliparous pregnancies (OR 0.5; 95% CI 0.28–0.85); to those with a follow-up within 7.2 years (OR 0.5; 95% CI 0.33–0.89); and to preterm deliveries (OR 0.5; 95% CI 0.30–0.81). Further, the majority of all results reported remained unaltered in the sensitivity analyses with the exception that in nulliparous pregnancies, a history of oral contraceptive use predicted gestational hypertension (OR 1.9; 95% CI 1.16–3.01).

Discussion

The results presented provide evidence of similarities and potentially important differences in predisposing factors for gestational hypertension and preeclampsia and preterm preeclampsia. Gestational hypertension and preeclampsia shared several baseline risk factors: a family history of diabetes mellitus, pregravid diabetes mellitus, baseline blood pressure status, obesity, a high total cholesterol/HDL cholesterol ratio, and a nonsignificant tendency to have a family history of stroke. Preeclampsia, however, was also predicted by a family history of myocardial infarction before 60 years of age, physical inactivity, an elevated triglyceride level, and binge drinking.

The notable adverse association of binge drinking with preterm preeclampsia in the current study mirrors the deleterious cardiovascular disease effects of heavy or binge drinking noted in the literature.[24,26-28] Biologically plausible mechanisms for the observed association between binge drinking and increased risk of preterm preeclampsia, but not gestational hypertension or term preeclampsia, likely relate to alcohol’s impairment of placentation and utero-placental growth and function through a variety of mechanisms.[29-32]

There are only a few studies that report on alcohol consumption’s association with preeclampsia. In the Screening for Pregnancy End points study of 5628 participants, early pregnancy alcohol consumption including binge drinking was not associated with preeclampsia or any other adverse outcome.[33] In a large study of over 1 million singleton birth in Missouri, binge drinking was not accessed, but 1 to 2 drinks per week was associated with a lower multivariable adjusted OR for preeclampsia (OR 0.82; 95% CI, 0.74–0.90).[34] Further, in an evaluation of blood pressure in pregnancy, those reporting light alcohol drinking during pregnancy were reported to have significantly lower blood pressure.[35] Although our study found reduced risk of preeclampsia associated with preconception weekly alcohol consumption, the results were not consistent in our sensitivity analyses. In contrast, our findings that preconception past-year binge drinking was associated with increased risk of preterm preeclampsia were consistently observed in all subsequent sensitivity analyses. We did not, however, observe a dose–response effect related to past-year binge drinking frequency. We speculate that there may have been reluctance to report the frequency of binge drinking in the current study or that binge drinking is a marker for other risk factors not measured.

The protective association of physical activity with preterm preeclampsia may reflect several protective underlying mechanisms beyond weight management, such as reduced inflammation and oxidative stress and improved endothelial function and placental growth and vascular development.[36] The preponderance of evidence suggests physical activity is protective of preeclampsia with a few notable exceptions.[37]

Our lipid results indicate that nonfasting lipid levels are useful in predicting preeclampsia and gestational hypertension and note that nonfasting lipid levels have been successfully used in cardiovascular research.[38,39] In the previous prospective study in Norway, which forms a subset of our current study, cholesterol, low-density lipoprotein cholesterol and nonHDL cholesterol were positively related to preeclampsia but trends associated with serum triglyceride quintiles were nonsignificant.[21] In Finland, triglycerides were associated with increased risk of preeclampsia but not cholesterol, low-density lipoprotein cholesterol, or HDL cholesterol.[20] Our lipid results corroborate the preponderance of existing evidence that elevated triglyceride levels is a risk factor for preeclampsia[40] and identified that a high total cholesterol/HDL cholesterol ratio was a shared risk factor for gestational hypertension and preeclampsia. Nonfasting triglyceride levels reflect exposures to atherogenic remnant lipoproteins[39] and elevated triglyceride levels associate with small, dense low-density lipoprotein cholesterol,[41,42] which is readily oxidized. Of relevance to preeclampsia is that oxidized low-density lipoprotein inhibits the fetal trophoblast invasion of the uterus.[43] Thus, further investigation of the role of derangements in lipid metabolism in preeclampsia is warranted.

Obesity was an important risk factor for gestational hypertension and preeclampsia as expected. However, our postestimation test for equality in coefficients identified that obesity was a significantly stronger predictor of gestational hypertension than preeclampsia, highlighting that for preeclampsia, there are other factors that increase risk.

Strengths and Weaknesses

Strengths of the current study include the large population-based cohort with uniformly assessed preconception risk factors and complete linkage to the Medical Birth Registry: strengths which contribute to the generalizability of the study. Further, the national healthcare system in Norway and our multivariate adjustment for educational and marital status and region are strengths of the study in that they minimize the possibility for disparities in the adequacy of prenatal care to influence results. Also, self-reported alcohol consumption data were obtained before pregnancy and would, therefore, not be influenced by under-reporting associated with the stigma of drinking during pregnancy. However, we recognize that under-reporting of alcohol consumption is also possible in nonpregnant study populations. Limitations of the study include the lack of information of fasting glucose or HbA1C, apolipoproteins, urine samples, and family history of hypertension. Other limitations include the largely ethnic Norwegian study population and inability to generalize to more ethnically diverse populations. Further, changes in risk factors over time could not be assessed. In our study, smoking had a nonsignificant association with a lower risk of preeclampsia in multivariable analyses, with the observed OR of 0.8 being greater than the anticipated OR of 0.5.[44] However, because of the sharp decline in smoking in women in Norway over the past decade, the attenuated smoking association may result from misclassification, where smokers identified in CONOR would be less likely to be smokers at the time of their subsequent pregnancy.

Perspectives

The study suggests similarities and some potentially important differences between risk factors for gestational hypertension and preeclampsia and between preterm and term preeclampsia. The results are intriguing given the lifetime increased risk of cardiovascular morbidity and mortality observed in women with a history of preeclampsia.[4-8] Further, although the majority of prospective studies of long-term consequences have focused on preeclampsia, an 18-year follow-up of mothers found that women with a history of gestational hypertension had similar predicted 10-year cardiovascular disease risk based on the Framingham score as women with a history of preeclampsia.[45] Our findings support the hypotheses that pregnancy unmasks predisposing familial and modifiable cardiometabolic risk. However, in the current study, a greater number of risk factors predicted preeclampsia than gestational hypertension. The presence of risk factors in women of reproductive age could help clinicians identify women needing greater clinical monitoring and lifestyle changes. Promoting better lifestyles in women of reproductive age would be advantageous for preventing the short- and long-term outcomes associated with hypertension disorders of pregnancy.

Acknowledgments

We acknowledge the CONOR Steering Committee and research teams which contributed the majority of participants for analyses: The Tromsø Study (IV and V), Troms and Finnmark Health Study (TROFINN), Oslo Health and Immigrant Health Study (HUBRO and I-HUBRO), and HUNT II Study, Nord-Trøndelag. We also thank Hordaland Health Study (HUSK) and Oppland and Hedmark Health Study (OPPHED).

Disclosures

None.