Gestational weight gain as a risk factor for hypertensive disorders of pregnancy

1See Journal Club, page 391

2See Journal Club, page 391

3The hypertensive disorders of pregnancy (HDP), including gestational hypertension and preeclampsia, remain a leading cause of maternal and perinatal mortality and morbidity worldwide,,,, and can affect up to 10% of pregnancies. At present, effective treatments are limited, and consequently, there is considerable interest in their prevention. Prepregnancy maternal adiposity is the strongest modifiable risk factor for HDP,,, and thus provides a potential means of prevention; however, it is increasingly recognized that reducing overweight/obesity in all women of reproductive age is extremely difficult. As a result there has been increased interest in the extent to which gestational weight gain (GWG) may influence HDP and hence provide a potential target for interventions aimed at reducing its risk.

4Several studies have shown that HDP are more likely to develop in women who have greater GWG,,,; however, most of these studies have important methodological limitations. The primary concern is that previous studies have related total GWG across pregnancy to the risk of HDP. This is problematic because women who develop HDP are more likely to experience edema during pregnancy than women who remain normotensive, and this in turn may result in greater GWG. Therefore, a positive association of total GWG with HDP could represent increased edema in women with HDP, causing greater weight gain, rather than greater GWG influencing increases in blood pressure. Furthermore, although a recent metaanalysis of interventions designed to limit GWG did show a reduction in HDP, this could be explained by the treatment of gestational diabetes mellitus that occurred in the 2 largest studies contributing to the metaanalysis.,

5Therefore, clarification of whether greater GWG precedes HDP and therefore could be a potential causal risk factor is urgently needed. The primary aim of this study was to examine whether GWG up to 18 weeks' gestation is associated with the subsequent development of HDP. GWG prior to 18 weeks' gestation, the time of the nadir in blood pressure, is unlikely to be affected by edema caused by HDP because it occurs before blood pressure begins to increase, and generally clinical edema does not occur until the second half of pregnancy.

6To further examine whether GWG influences blood pressure in pregnancy, we also investigated whether greater GWG in different periods of pregnancy precedes greater increases in blood pressure. In this second analysis, we excluded women with HDP to exclude the possibility that the relationships are explained by changes in weight caused by HDP.

7The Avon Longitudinal Study of Parents and Children (ALSPAC) is a prospective birth cohort study investigating the health and development of children. The study has been described in full elsewhere and on the website (Available at: www.bris.ac.uk/alspac. Accessed March 20, 2013). Women with expected delivery dates between April 1, 1991, and Dec. 31, 1992, who were living in a defined area of Avon during their pregnancy were eligible for recruitment. Ethical approval was obtained from the ALSPAC Law and Ethics Committee and from the National Health Service local ethics committee.

8A total of 14,541 women were enrolled, 13,678 women had a singleton live birth, and 13,461 of these had data abstracted from obstetric records. We excluded women who reported having a previous diagnosis of hypertension outside pregnancy (n = 445), leaving 13,016 women of whom 12,522 had at least 1 measurement of blood pressure and weight during pregnancy.

9All weight and blood pressure measurements (median [interquartile range] per woman: 12 [10–14] and 14 [11–16], respectively) taken routinely as part of antenatal care by midwives or obstetricians were abstracted from obstetric records by 6 trained research midwives. There was no between-midwife variation in the mean values of the data abstracted, and error rates were consistently less than 1% in repeated data entry checks.

10Gestational age at delivery was calculated as the difference between the delivery date and the mother's reported last menstrual period date or updated if ultrasound information was available, which led to a reassessment of gestation. At the time of recruitment, first-trimester ultrasound gestational age dating was not routine clinical practice, and the information abstracted from clinical records did not indicate which few women had their gestational age adjusted.

11HDP was defined according to International Society for the Study of Hypertension in Pregnancy criteria of systolic blood pressure (SBP) of 140 mm Hg or greater or diastolic blood pressure (DBP) of 90 mmHg or greater, measured on 2 occasions after 20 weeks' gestation for gestational hypertension and the same criteria in conjunction with proteinuria of at least 1+ on dipstick testing for preeclampsia.

12Prepregnancy weight was obtained from the predicted weight at 0 weeks' gestation from multilevel models based on the antenatal weight measurements (see the following text). This correlated highly with self-reported prepregnancy weight from a questionnaire administered during pregnancy (r = 0.94). Absolute weight gain was calculated by subtracting the predicted weight at 8 weeks (because there were few measurements prior to this time) from the predicted weight at 40 weeks' gestation, using a multilevel model for weight change across pregnancy including women with term pregnancies (see the following text) and was then categorized according to Institute of Medicine (IOM) recommendations (Appendix; Supplementary Table 1). Consistent results to those presented here for associations of IOM categories with HDP were found if self-reported prepregnancy weight and the last antenatal weight measurement (in which this was after 37 weeks) were used to define absolute weight gain.

13Maternal age at delivery and offspring sex were abstracted from obstetric records. Maternal height, parity, highest educational qualification, and smoking status were obtained from questionnaires administered during early pregnancy. Prepregnancy body mass index (BMI) was calculated as predicted prepregnancy weight (kilograms)/self-reported height (meters) squared and classified according to the World Health Organization categories (Supplementary Table 1). Smoking status was classed as never for women who did not smoke immediately prior to or during pregnancy, before pregnancy/first trimester for women who smoked only immediately prior to pregnancy or in the first 3 months of gestation, or throughout for women who continued to smoke after the first trimester.

14For women with complete data on IOM categories of GWG and covariates (n = 9596), we used a multinomial regression model to assess the risk of developing gestational hypertension and preeclampsia (with normotensive women as the comparison group) by the IOM category of GWG, adjusting for prepregnancy BMI, maternal age, parity, smoking, education, and sex of offspring.

15Using all weight measurements up to 18 weeks' gestation for women with at least 1 weight measurement prior to this time (n = 11,760), we developed a multilevel model for the change in weight with gestational age, using a linear slope. The model had 2 levels: measurement occasion (occasion level) within woman (individual level); and 2 individual-level random effects representing weight at 0 weeks (the intercept; referred to as prepregnancy weight) and average GWG per week up to 18 weeks. The individual-level random effects were then used as exposure variables in multinomial logistic regression models.

16The outcome variable had 3 categories: normotensive (comparison group), gestational hypertension, and preeclampsia. We adjusted for maternal height, age, parity, smoking, education, and offspring sex, and to examine the exposure of GWG up to 18 weeks, we additionally adjusted for prepregnancy weight. We tested for nonlinearity in these associations using fractional polynomials, and any nonlinear relationships were approximated using linear splines. We also tested for an interaction between prepregnancy BMI category and GWG up to 18 weeks.

17We have previously fitted multilevel linear spline models for changes in SBP, DBP, and weight with gestational age separately. We combined these models to develop 2 bivariate multilevel linear spline models, the first including changes in SBP and weight in parallel and the second including changes in DBP and weight in parallel (Supplementary material).

18This analysis was restricted to normotensive women who had a term pregnancy (≥37 weeks) (n = 9855). The model with SBP and weight as outcomes had 5 random-effect parameters for SBP: SBP at 8 weeks (baseline) and rate of SBP change at 8-18 weeks (early pregnancy), 18-29 weeks (midpregnancy), 29-36 weeks (late pregnancy), and 36 weeks onward (very late pregnancy) and 4 random-effect parameters for weight: weight at 0 weeks (prepregnancy weight) and rate of GWG at 0-18 weeks (early pregnancy), 18-29 weeks (midpregnancy), and 29 weeks onward (late pregnancy). The model with DBP and weight as outcomes had the equivalent random-effects parameters for DBP as for SBP and the same weight parameters. (Estimates of these parameters are shown in Supplementary Table 2.)

19In the second stage, we used the bivariate multilevel models to derive associations of prepregnancy weight and GWG (exposures) with blood pressure change (outcome), adjusting for maternal height, age, parity, smoking, education, and offspring sex in model 1. In model 2 we additionally adjusted for baseline weight and SBP/DBP and GWG and SBP/DBP changes in periods of pregnancy prior to the exposure period. We then conceptualized GWG as the outcome with changes in blood pressure as a potential risk factor. We therefore derived associations of GWG in each period of gestation (outcome) with SBP/DBP at 8 weeks and SBP/DBP change in previous periods of gestation (exposure).

20We adjusted for maternal characteristics in model 1. In model 2 we additionally adjusted for baseline weight and SBP/DBP, GWG, and the SBP/DBP change in all periods prior to the exposure period and GWG in the exposure period.

21We completed sensitivity analyses, restricting to women who contributed at least 8 measurements of blood pressure and weight to the bivariate models (n = 6666), and second, restricting to women who contributed no more than 11 measurements (n = 6135). The findings were not meaningfully changed from those in the main analyses. We also performed a sensitivity analysis using an 8 week baseline for weight.

22Statistical analyses were carried out using MLwiN version 2.25, STATA version 12.1 and runmlwin and reffadjust Stata commands (StataCorp, College Station, TX).

23The characteristics of women in the full dataset, and subsets with complete data for each analysis, by categories of HDP are shown in Table 1, and distributions of prepregnancy BMI and early-pregnancy GWG are shown in Supplementary Figures 1 and 2, respectively. Each subset had similar characteristics to the full dataset.

24Nulliparous women, nonsmokers, and women aged over 35 years were more likely to develop HDP (Supplementary Table 3). Gaining more than the IOM recommended weight was associated with an increased risk of gestational hypertension and preeclampsia compared with gaining within the recommended range (Figure; odds ratio [OR], 1.51; 95% confidence interval [CI], 1.32–1.73 and OR, 2.14; 95% CI, 1.46–3.12, respectively). Also, lower than IOM-recommended GWG was associated with a reduced risk of gestational hypertension (OR, 0.75; 95% CI, 0.62–0.91) and preeclampsia (OR, 0.70; 95% CI, 0.38–1.27) compared with GWG within the recommended range (although the confidence interval was wide for preeclampsia and included the null).

25A higher prepregnancy weight was associated with an increased risk of developing gestational hypertension and preeclampsia (Table 2). This association was nonlinear, with evidence of a stronger association between prepregnancy weight and gestational hypertension for women who weighed 90 kg or less than for those who weighed more than 90 kg before pregnancy, although associations with preeclampsia were similar in each group.

26After adjustment for prepregnancy weight, greater early-pregnancy GWG was independently associated with an increased risk of developing gestational hypertension and preeclampsia (per 200 g/wk increase in GWG up to 18 weeks: OR, 1.26; 95% CI, 1.16–1.38 and OR, 1.31; 95% CI, 1.07–1.62, respectively).

27These findings are further illustrated by Table 3, which shows the predicted probabilities of developing gestational hypertension or preeclampsia for different prepregnancy weights and rates of early-pregnancy GWG. This shows clearly that, even among women with healthy prepregnancy weights (eg, 50 or 60 kg), those who gain more weight have important increases in the risk of gestational hypertension and preeclampsia. We found no evidence that associations of GWG up to 18 weeks with the risk of gestational hypertension and preeclampsia differed by prepregnancy BMI category (P value for interaction =.18).

28Analysis of the mean differences in SBP and DBP change for a 200 g/wk increase in GWG in the same and preceding period of gestation are shown in Table 4. For completeness, associations of all possible combinations of prepregnancy weight and GWG in each gestational period with later SBP and DBP changes are also shown in Supplementary Tables 4 and 5, respectively.

29In each period of pregnancy, GWG was positively associated with the changes in SBP and DBP that occurred during the same gestational period (Table 4). In addition, an increase in GWG in early pregnancy (up to 18 weeks) was associated with a greater increase in SBP and DBP in the subsequent period of gestation (midpregnancy: 18-29 weeks) (Table 4). However, greater GWG in early pregnancy was also associated with a smaller increase in SBP and DBP in late pregnancy (Supplementary Tables 4 and 5). GWG in midpregnancy (18-29 weeks) was not associated with SBP or DBP change in late pregnancy (29-36 weeks) (Table 4).

30To assess whether rises in blood pressure may precede increases in edema and thus weight gain, even in these normotensive women, we also examined associations of changes in SBP and DBP with subsequent GWG (Table 5). All possible associations of baseline SBP and DBP and changes in SBP and DBP with GWG in later periods of gestation are shown in Supplementary Table 6. In fully-adjusted models, there were no associations between SBP or DBP change and GWG in subsequent periods of pregnancy. There was evidence of positive associations between SBP and DBP at baseline and GWG in early pregnancy (Supplementary Table 6). However, when we repeated the bivariate analyses with the baseline set at 8 weeks for weight, neither SBP nor DBP at baseline was associated with GWG in early pregnancy after adjustment for baseline weight (no other findings were changed in this sensitivity analysis).

31In this large cohort, greater prepregnancy weight and GWG in early pregnancy were both independently associated with an increased risk of gestational hypertension and preeclampsia. We also found that, in normotensive women, gaining more weight in early pregnancy was associated with a greater increase in blood pressure in the subsequent period of gestation. In addition, in each period of gestation, GWG was positively associated with the change in blood pressure during the same gestational period. These associations are unlikely to be explained by blood pressure increases leading to increased edema because we found little evidence that blood pressure changes were associated with subsequent GWG in any period of pregnancy in these women.

32The review of outcomes associated with GWG by Viswanathan et al did not find any strong evidence of an association of GWG with risk of HDP. Although the majority of studies reviewed reported a positive association, the review authors noted that the number of studies was limited and most were assessed to be of weak design. Since the review, 2 studies have reported a positive relationship of excess GWG according to 1990 IOM recommendations with a risk of HDP in 5377 Canadian and 1043 Latina women, respectively, and others have reported positive associations between net GWG and HDP risk., However, because IOM criteria combine prepregnancy BMI and GWG over the whole of pregnancy, it is not possible using these criteria to distinguish the impact of prepregnancy BMI from GWG, and none of the studies have addressed whether it is edema associated with HDP that is driving the association.

33Our study was limited by the use of routinely collected clinic blood pressure and weight measurements, which is likely to produce random measurement error but unlikely to introduce bias. It also has the advantage that our results could be translated into clinical practice if they are replicated because the measures represent those obtained in clinical practice. Women who develop preeclampsia have reduced plasma volume expansion in early pregnancy compared with women who remain normotensive.,, However, studies have suggested that this is due to increased capillary permeability and a redistribution of plasma to interstitial fluid., We were unable to account for plasma volume expansion in our analyses, but any bias in our findings resulting from the reduction in plasma volume would be expected to attenuate associations toward the null.

34The study's strengths are its large size, inclusion of women with a range of BMIs, the high number of repeated weight and blood pressure measurements, and longitudinal modeling of GWG and blood pressure changes in different periods of pregnancy. Because of the repeat measurements of blood pressure and proteinuria, we were able to apply a standard international definition of HDP to all women and not rely on clinical diagnoses.

35Outside pregnancy, greater body mass is an established risk factor for high blood pressure,, and therefore, the association of greater GWG in early pregnancy with HDP and of GWG throughout pregnancy with blood pressure increases in women without HDP may reflect the effect of greater maternal pregnancy adiposity acquisition on blood pressure.

36Also, markers of inflammation, interleukin-6 and C-reactive protein, are increased in obese pregnant women, and this inflammation may contribute to endothelial dysfunction. Inflammation, measured by C-reactive protein concentrations, has been found in 1 study to mediate the association between prepregnancy BMI and preeclampsia risk. In our study, we were unable to examine this mechanism. Alternatively, these associations may be partly explained by biological changes related to pregnancy influencing both GWG and blood pressure change or from an underlying predisposition to obesity and cardiovascular risk.

37Although our observational data are not able to prove causality, they are a step change from the previous publications and demonstrate temporality by showing that GWG precedes the development of HDP and also a greater increase in blood pressure.

38We have also adjusted for a wide range of potential confounders in our analyses. If GWG does have a causal influence on blood pressure change, the potential implications are that limiting GWG in pregnancy may be able to both reduce the incidence of HDP and the severity in those who develop HDP. A recent systematic review and metaanalysis of pregnancy interventions aimed at managing weight gain estimated a 33% reduction in the risk of preeclampsia and a 70% reduction in the risk of gestational hypertension had been achieved through interventions that limited GWG. However, the 2 largest included studies, which drove the associations, included only women at high risk for gestational diabetes and an intervention aimed at preventing its development. Hence, it is unclear whether the decreased risk of HDP was achieved through the monitoring and management of gestational diabetes because the conditions are strongly linked.

39The clinical significance of our findings is indicated by the potential risk reduction for preeclampsia that could be achieved by optimizing GWG. Our results show that the risk of gestational hypertension and preeclampsia is markedly lower in women with lower GWG in all strata of prepregnancy weight (Table 3).

40These results combined with evidence from recent trials showing that lifestyle interventions can effectively limit GWG, show the potential to prevent HDP by targeting GWG. Furthermore, the risk reduction that might be achieved by targeting GWG compares favorably with the relative risk reduction of 0.75 (95% CI, 0.66–0.85) observed with antiplatelet agents for the risk of preeclampsia and may be independent of this. However, any benefits of limiting GWG in terms of reducing HDP risk should be balanced with the potential adverse effects because there is evidence of an increased risk of preterm birth and small-for-gestational-age offspring at the lower end of the GWG distribution. We were unable to assess the optimum levels of GWG required to achieve this balance in this study.

41In summary, we found that greater GWG up to 18 weeks, independently of prepregnancy weight, was associated with a greater risk of developing HDP and in normal pregnancy was associated with a greater midpregnancy rise in blood pressure. GWG in all periods of gestation were positively associated with concurrent rises in blood pressure. This suggests that interventions aimed at limiting GWG from early pregnancy onward may have the potential to reduce the incidence of gestational hypertension and preeclampsia. Our results support establishing randomized trials in nonspecific populations to assess the effectiveness of such interventions in the prevention of these disorders.