Abstract
Objective:
To investigate the association between early-pregnancy weight gain and risk of preeclampsia to inform pregnancy weight gain recommendations.
Study Design:
We performed a case-cohort study using a hospital database including 80,812 singleton deliveries from Magee-Womens Hospital, Pittsburgh, Pennsylvania (1998-2011). In each of 6 prepregnancy body mass index (BMI) groups, we abstracted serial antenatal weight measurements from the records of up to 339 preeclampsia cases and 1254 randomly selected pregnancies. Early gestational weight gain (16-19 weeks’ gestation) was standardized for gestational duration using BMI-specific z-score charts. Multivariable log-binomial regression was used to assess the association between weight gain z-score and risk of preeclampsia. We determined the impact of preeclampsia misclassification using probabilistic bias analysis.
Main Outcome Measure:
Risk of preeclampsia
Results:
For normal weight women, there was a steady increase in preeclampsia risk with increasing early gestational weight gain z-score. For example, compared with a weight gain of 1.2 kg (z-score = −1 SD), a 7.2-kg weight gain (z-score = +1 SD) at 16 weeks was associated with 1.3 (0.50, 2.2) excess preeclampsia cases per 100 deliveries. Weight loss at 16–19 weeks among grade 2 or 3 obese women was associated with a reduced risk of preeclampsia. Associations were null among overweight and grade 1 obese women. The bias analysis supported the validity of the conventional analysis.
Conclusions:
Early-pregnancy weight gain may be associated with preeclampsia in some BMI groups. Future revisions of pregnancy weight gain recommendations should account for preeclampsia risks from this and additional studies.
Keywords: Preeclampsia, Gestational weight gain, Body mass index, Gestational hypertension, Obesity, Pregnancy hypertension
Introduction
In 2009, the United States National Academy of Sciences/Institute of Medicine (IOM) published revised guidelines for weight gain during pregnancy to balance risks of low weight gain (e.g., fetal growth restriction, preterm birth, and infant death) and high weight gain (e.g., cesarean delivery, postpartum weight retention, and child obesity) (1). However, the IOM Committee recognized that the validity of its recommendations was threatened by major gaps in our understanding of how gestational weight gain impacts other maternal and child outcomes (1). One of the key outcomes lacking rigorous data is preeclampsia, a common, multi-system complication of pregnancy that leads to substantial infant and maternal morbidity and mortality (2).
Although some research relating gestational weight gain to preeclampsia was available for the 2009 IOM committee to review (3-5), these studies are difficult to interpret because they were based on total pregnancy weight gain at the time of delivery. Preeclampsia can lead to increased vascular permeability and decreased plasma oncotic pressure, resulting in significant oedema and rapid weight gain (6). To evaluate the role of weight gain not attributable to preeclampsia, it is important to measure weight gain before the clinical manifestations of preeclampsia, which cannot be diagnosed before 20 weeks. Few large, contemporary cohorts have the antenatal weight measurements necessary for such an analysis. To our knowledge, only one published study has evaluated early-pregnancy weight gain in relation to preeclampsia (7). These investigators found a positive linear relationship between gestational weight gain before 18 weeks of pregnancy and the risk of preeclampsia. However, this study had too few cases to evaluate these associations separately by prepregnancy body mass index (BMI) category, which is important for informing BMI-specific weight gain recommendations. Our objective was to determine the association between early-pregnancy weight gain and the risk of preeclampsia by prepregnancy BMI group.
Materials and Methods
Astrid is a retrospective pregnancy cohort study aimed at filling research gaps to inform future evidence-based gestational weight gain recommendations. The study is based on medical record data of singleton pregnancies delivered at Magee-Womens Hospital in Pittsburgh, Pennsylvania, from 1998 to 2011. We identified eligible pregnancies using a perinatal database of all deliveries at the hospital, which we supplemented with abstracted antenatal visit data, including serial weight measurements. We sampled from the eligible cohort using a case-cohort design because (i) it provides nearly the same statistical efficiency as a cohort study while reducing medical record abstraction costs; (ii) the subcohort can be used to estimate the prevalence of exposures in the eligible cohort; and (iii) the subcohort can serve as the comparison group for multiple health outcomes of interest (8). The University of Pittsburgh Institutional Review Board approved this study.
There were 114,736 singleton deliveries at Magee-Womens Hospital from 1998 to 2011. Women were eligible for Astrid if they delivered at 16–42 weeks and had complete data on prepregnancy weight (n=80,812). To allow for adequately powered BMI-stratified analyses, we randomly sampled from the eligible cohort 1411 pregnancies in each BMI category. These pregnancies served as representative subcohorts for comparisons, providing more than twice the number of cases in each BMI group. We then augmented each BMI-specific subcohort with up to 500 cases of preeclampsia. Women whose medical records could not be located (5.3%), who had preexisting hypertension (6.5%) or outlying values of weight gain z-score or BMI (0.3%, defined below), who lacked data on covariates of interest (0.4%), or who did not have measured weight at 16 weeks 0 days to 19 weeks 6 days (19%) were excluded. Finally, we excluded underweight women from this analysis because there were only 41 available preeclampsia cases. The final analytic sample included 6314 total observations (from 6035 unique pregnancies), including 1197 total preeclampsia cases, 279 of which served in both the subcohort and case groups.
We used International Classification of Diseases (ICD)-9 codes in the perinatal database to define preeclampsia. The case definition included mild preeclampsia (642.4), severe preeclampsia or HELLP syndrome (642.5), or eclampsia (642.6). We tested the validity of this definition using 467 women with preeclampsia who participated in a previous cohort study in our population (9). In this study, a jury of clinical experts reviewed all medical records to determine whether each subject met the criteria for preeclampsia, which was defined as hypertension and proteinuria, and return of all abnormalities to normal by 12 weeks postpartum (10). Gestational hypertension was defined as systolic blood pressure persistently ≥140 mmHg and/or diastolic blood pressure persistently ≥90 mmHg for the first time after 20 weeks of gestation. Proteinuria was defined as the excretion of >300 mg of protein in 24 hours, a random sample of 2+ protein, a catheterized sample of 1+ protein, or a protein-creatinine ratio >0.3. Using this definition as the gold standard, the coding-based definition of preeclampsia in the database had a sensitivity and specificity of 85% and 96% among normal-weight women, 76% and 95% among overweight women, and 69% and 95% among obese women, respectively.
Data collectors with uniform training performed standardized medical chart abstractions to gather data on serial antenatal weight measurements, antenatal visit dates, ultrasound results, and medical history. Abstractors directly entered data into a piloted computer-assisted data entry system and used a codebook to ensure consistent coding decisions. An inter-rater reliability study demonstrated excellent agreement for key study variables including maternal weight and gestational age (12).
The exposure window of interest was 16 weeks 0 days to 19 weeks 6 days of gestation because it precedes the clinical onset and diagnosis of preeclampsia, as defined above. We calculated gestational weight gain from 16–19 weeks’ gestation as the last measured weight at 16–19 weeks minus prepregnancy weight. If more than one weight measurement was available from this time window, we used the later one. Gestational weight gain was standardized for gestational duration using BMI-specific z-score charts that were developed in the Astrid population (13, 14). For ease of interpretation, we provide absolute weight gains (kg) at 16 weeks for a range of z-scores. Weight gain z-scores <−4 SD or >+4 SD were considered outliers and excluded. We also reviewed weight trajectories throughout pregnancy for implausible patterns through the use of conditional weight percentiles (15). Using this method, observations 4 SD above or below the expected based on the previous weight measurement (or self-reported prepregnancy weight) were flagged and excluded according to their clinical plausibility.
Prepregnancy BMI (weight (kg)/height (m)2) was based on recalled prepregnancy weight and height at the first prenatal visit (median [interquartile range]: 9.1 [7.7–11] weeks’ gestation) and was classified as normal weight (18.5–24.9) overweight (25–29.9), grade 1 obese (30–34.9), grade 2 obese (35–39.9) or grade 3 obese (≥40) (16).
Gestational age at delivery was determined using the best obstetric estimate based on a comparison of menstrual dating and ultrasound dating (17) ascertained from the medical record. Covariates of interest were obtained from the perinatal database or, if missing, were filled in by medical record review where possible. Women were classified based on self-reported race/ethnicity (non-Hispanic White, non-Hispanic Black, other); education (less than high school, high school or equivalent, some college, college graduate); marital status (married, unmarried); smoking during pregnancy (yes, no); and insurance at delivery (private, Medicaid/other). Parity was categorized as 0, 1, or 2 or more.
Statistical analysis
We decided a priori to stratify analyses by prepregnancy BMI category because weight gain recommendations are BMI-specific. Unadjusted measures of risk were calculated after weighting the subcohort by the inverse of the sampling fraction. To assess the association between early gestational weight gain z-score and risk of preeclampsia, we fit multivariable log-binomial regression models. The models were weighted by the inverse of the sampling weight and robust variances were calculated to account for the case-cohort design (8, 18). We estimated risks, risk differences, and their 95% confidence intervals from the regression models via marginal standardization (19). We modeled gestational weight gain z-score as a 3-level categorical variable (<−1 standard deviation [SD], −1 to +1 SD, >+1 SD), where <−1 SD was the referent category because we hypothesized it would be associated with the lowest risk of preeclampsia. We also used restricted cubic spline terms with 3 knots equally spaced across the z-score distribution to allow for flexible non-linear relations (20). The placement and number of knots were selected based on Akaike information criterion (20). We compared the risks at 1-SD increments of z-scores with that at our reference value of −1 SD. Potential confounders (maternal race/ethnicity, education, parity, smoking, age, marital status, and insurance at delivery), were identified using theory-based causal diagrams (21), were included in all models.
To evaluate the impact of ICD coding-based errors in the classification of preeclampsia, we performed a probabilistic bias analysis using a modification of the SAS macro ‘sensmac’ (22, 23). Based on our validation study results described above, we assumed that the misclassification of preeclampsia was differential by BMI category. We used the BMI-specific sensitivities and specificities in a Monte Carlo simulation of the 10,000 datasets, each one yielding estimates of association that were then summarized by the median, 2.5th percentile, and 97.5th percentile. These estimates are analogous to the point estimate and limits of the 95% confidence interval of the conventional analysis, which treats all variables as if they were perfectly measured. The sensmac macro requires binary exposure and outcome variables to be used in a multivariable logistic regression model. Therefore, we approximated case-control data by excluding the cases that also served in the subcohort. We also dichotomized gestational weight gain z-score as ≤+1 standard deviation (SD) or >+1 SD.
Results
There were no meaningful differences in the incidence of preeclampsia or the distributions of key maternal variables comparing the eligible cohort with all deliveries (Table S1). Table S2 displays sample size and sampling probabilities of the eligible cohorts, subcohorts, and case groups. The cohort of eligible singleton pregnancies was 2.5% underweight, 48% normal weight, 28% overweight, 12% grade 1 obese, 5.4% grade 2 obese, and 3.2% grade 3 obese. When we compared women in the eligible cohort with women the sampled subcohort or women the subcohort used in the final analysis, there were no major differences (Tables S3-S5).
In each BMI group, subcohort members in the final analytic sample were predominantly non-Hispanic White, college educated, married, non-smokers and had one previous pregnancy (Table 1). The proportions of women who were non-Hispanic Black, married or with less education tended to increase as the BMI category increased. In the subcohorts, the mean absolute gestational weight gain at 16–19 weeks decreased and the variability increased with increasing prepregnancy BMI category from 5.2 (SD, 3.4) kg among normal weight women to 1.5 (SD, 6.5) kg among grade 3 obese women.
Table 1.
Characteristics of pregnancies in the subcohort and pregnancies with preeclampsia by prepregnancy body mass index category, Magee-Womens Hospital, Pittsburgh, PA, 1998–2011.*,**
| Normal weight | Overweight | Grade 1 obese | Grade 2 obese | Grade 3 obese | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Subcohort n=1254 |
Cases n=339 |
Subcohort n=1061 |
Cases n=290 |
Subcohort n=1004 |
Cases n=283 |
Subcohort n=947 |
Cases n=169 |
Subcohort n=851 |
Cases n=116 |
|
| Maternal race/ethnicity | ||||||||||
| Non-Hispanic White | 79 | 71 | 81 | 74 | 78 | 73 | 74 | 75 | 68 | 74 |
| Non-Hispanic Black | 13 | 22 | 16 | 21 | 20 | 25 | 24 | 22 | 31 | 24 |
| Other | 8 | 7 | 3 | 5 | 2 | 2 | 2 | 3 | 1 | 2 |
| Education, years | ||||||||||
| Less than high school | 8 | 12 | 5 | 7 | 7 | 6 | 8 | 8 | 8 | 7 |
| High school | 20 | 24 | 21 | 25 | 24 | 31 | 28 | 27 | 28 | 36 |
| Some college | 21 | 22 | 25 | 24 | 29 | 27 | 28 | 25 | 34 | 21 |
| College graduate | 51 | 42 | 49 | 44 | 40 | 36 | 36 | 40 | 30 | 35 |
| Parity | ||||||||||
| 0 | 48 | 75 | 40 | 71 | 41 | 63 | 41 | 61 | 40 | 62 |
| 1 | 46 | 23 | 54 | 26 | 50 | 32 | 50 | 35 | 49 | 32 |
| 2 or more | 6 | 2 | 6 | 3 | 9 | 5 | 9 | 4 | 11 | 6 |
| Maternal age, years | 29(6) | 27(6) | 30(6) | 29(6) | 29(6) | 28(7) | 29(6) | 29(6) | 29(6) | 29(5) |
| Married | 68 | 58 | 68 | 63 | 65 | 60 | 59 | 58 | 56 | 53 |
| Smoked during pregnancy | 11 | 13 | 11 | 10 | 13 | 9 | 16 | 13 | 13 | 16 |
| Gestational weight gain at 16–19 wk, kg | 5.2(3.4) | 5.2(3.4) | 4.8(4.6) | 5.0(4.4) | 3.8(5.4) | 4.0(5.1) | 2.7(5.8) | 3.5 (4.6) | 1.5(6.5) | 2.5(5.6) |
| Gestational weight gain z-score at 16–19 wk | −0.13(1.0) | 0.04(1.1) | 0.05(0.9) | 0.09(0.89) | 0.08(0.88) | 0.09(0.92) | 0.11(0.91) | 0.25(0.82) | 0.13(0.79) | 0.25(0.80) |
| Gestational weight gain z-score at 16–19 wk | ||||||||||
| <−1 SD | 19 | 16 | 9 | 8 | 10 | 10 | 10 | 6 | 8 | 6 |
| −1 to +1 SD | 70 | 68 | 77 | 77 | 76 | 75 | 76 | 79 | 82 | 81 |
| >+1 SD | 11 | 16 | 17 | 15 | 12 | 15 | 14 | 15 | 10 | 13 |
| Preterm birth <34 weeks | 1.9 | 9.4 | 2.5 | 13 | 3.4 | 13 | 4.7 | 16 | 2.7 | 12 |
Normal weight, body mass index (BMI) 18.5 to 24.9 kg/m2; overweight, BMI 25–29.9 kg/m2; grade 1 obese, BMI 30–34.9 kg/m2; grade 2 obese, BMI 35–39.9 kg/m2; grade 3 obese, BMI ≥40 kg/m2.
Data in the table are % or mean(standard deviation)
The weighted incidence of preeclampsia was 3.5%, 4.7%, 6.2%, 5.6%, and 7.0% for the normal weight, overweight, grade 1 obese, grade 2 obese, and grade 3 obese subcohorts, respectively. Approximately 9%–16% of preeclamptic pregnancies delivered before 34 weeks. In a majority of the BMI groups, there were greater proportions of cases than subcohort members who were non-Hispanic Black, unmarried, nulliparous, or less educated. No consistent differences were observed for age or smoking between case and subcohort groups.
High gestational weight gain at 16–19 weeks was associated with meaningful increases in risk of preeclampsia among women with normal prepregnancy BMI and grade 2 or grade 3 obesity compared to the reference group of low early gain (< −1SD) (Table 2). For instance, among normal weight women, early weight gain >+1 SD was associated with 1.9 (95% CI 0.2, 3.5) excess cases per 100 deliveries. These gestational weight gain z-score categories correspond to absolute weight gains of <1.2 kg and >7.2 kg at 16 weeks, respectively. Differences in risk between these z-score categories were similar in magnitude for grade 2 and grade 3 obese women, but were less precise due to small samples. When women with grade 2 or grade 3 obesity were combined, >+1 SD weight gain at 16–19 weeks was associated with 3.0 (95% CI 0.01, 5.8) excess cases per 100 deliveries compared with the referent. There was no association among overweight women or women with grade 1 obesity.
Table 2.
Association between gestational weight gain at 16–19 weeks and preeclampsia by prepregnancy body mass index category, Magee-Womens Hospital, Pittsburgh, PA, 1998–2011 *,**
| Prepregnancy body mass index category |
Gestational weight gain z- score at 16–19 weeks |
Equivalent gestational weight gain at 16 weeks (kg) |
Number of preeclampsia cases |
Unadjusted risk of preeclampsia per 100 deliveries (95% CI) |
Adjusted*** risk of preeclampsia per 100 deliveries (95% CI) |
Adjusted*** risk difference per 100 deliveries (95% CI) |
|---|---|---|---|---|---|---|
| Normal weight | < −1 SD | <1.2 | 53 | 2.9 (2.1, 3.8) | 2.8 (1.9, 3.6) | referent |
| −1 to +1 SD | 1.2 to 7.2 | 232 | 3.4 (2.9, 3.7) | 3.5 (3.0, 4.0) | 0.7 (−0.3, 1.7) | |
| > +1 SD | >7.2 | 54 | 4.8 (3.4, 6.2) | 4.6 (3.2, 6.0) | 1.9 (0.2, 3.5) | |
| Overweight | < −1 SD | <−0.65 | 24 | 4.1 (2.4, 5.9) | 4.3 (2.6, 6.1) | referent |
| −1 to +1 SD | −0.65 to 8.5 | 224 | 4.7 (4.1, 5.3) | 4.7 (4.1, 5.4) | 0.4 (−1.5, 2.3) | |
| > +1 SD | >8.5 | 42 | 5.1 (3.5, 6.7) | 4.8 (3.2, 6.5) | 0.5 (−1.8, 2.9) | |
| Grade 1 obese | < −1 SD | <−2.2 | 29 | 6.3 (3.9, 8.7) | 6.7 (4.2, 9.3) | referent |
| −1 to +1 SD | −2.2 to 8.3 | 212 | 6.0 (5.1, 6.8) | 5.9 (5.1, 6.7) | −0.9 (−0.4, 1.8) | |
| > +1 SD | >8.3 | 42 | 7.2 (5.0, 9.5) | 7.5 (5.2, 9.9) | 0.8 (−2.6, 4.3) | |
| Grade 2 obese | < −1 SD | <−3.7 | 10 | 3.5 (1.4, 5.5) | 3.6 (1.5, 5.8) | referent |
| −1 to +1 SD | −3.7 to 7.3 | 133 | 5.8 (4.8, 6.7) | 5.7 (4.8, 6.6) | 2.1 (−0.3, 4.4) | |
| > +1 SD | >7.3 | 26 | 6.3 (4.0, 8.5) | 6.4 (4.0, 8.8) | 2.8 (−0.5, 6.1) | |
| Grade 3 obese | < −1 SD | <−6.0 | 7 | 5.6 (1.8, 9.4) | 5.5 (1.9, 9.1) | referent |
| −1 to +1 SD | −6.0 to 7.8 | 94 | 6.9 (5.6, 8.2) | 6.9 (5.7, 8.2) | 1.5 (−2.3, 5.3) | |
| > +1 SD | >7.8 | 15 | 8.7 (4.7, 12.6) | 8.3 (4.6, 12.1) | 2.9 (−2.3, 8.1) | |
| Grade 2 or 3 obese | < −1 SD | See above | 17 | 4.1 (2.3, 6.0) | 4.2 (2.3, 6.1) | referent |
| −1 to +1 SD | 227 | 6.2 (5.4, 7.0) | 6.2 (5.4, 6.9) | 2.0 (0, 4.0) | ||
| > +1 SD | 41 | 7.0 (5.0, 9.0) | 7.2 (5.1, 9.3) | 3.0 (0.01, 5.8) |
CI, confidence interval; SD, standard deviation
Risks are based on the weighted cohort. The number of women in each subcohort is 1254 normal weight, 1061 overweight, 1004 obese grade 1, 947 obese grade 2, and 851 obese grade 3.
Body mass index (BMI) kg/m2: Normal weight, BMI 18.5 to 24.9; overweight, BMI 25–29.; grade 1 obese, BMI 30–34.9; grade 2 obese, BMI 35–39.9; grade 3 obese, BMI ≥40.
Adjusted for maternal race/ethnicity, maternal age, maternal education, parity, smoking during pregnancy, marital status and insurance.
Figure 1 shows the adjusted curvilinear associations between gestational weight gain z-score at 16–19 weeks and preeclampsia by BMI category, and Table 3 presents the accompanying risk and risk difference estimates.
Figure 1.
Adjusted predicted risk of preeclampsia (based on the weighted cohort) according to gestational weight gain z-score at 16–19 weeks among normal weight (n=339 cases, panel A), overweight (n=290 cases, panel B), grade 1 obese (n=283 cases, panel C), grade 2 obese (n=169 cases, panel D), grade 3 obese women (n=116 cases, panel E), and grade 2 or 3 obese women (panel F), Magee-Womens Hospital, Pittsburgh, PA 1998-2011.
The solid lines represent the point estimate and dashed lines represent its 95% confidence bands. Risks were set at the population average for maternal race/ethnicity, maternal age, maternal education, parity, smoking during pregnancy, marital status and insurance. Gestational weight gain was modeled using a 3-knot restricted cubic spline.
Table 3.
Adjusted risks of preeclampsia according to gestational weight gain z-score at 16–19 weeks and prepregnancy body mass index category.
| Prepregnancy body mass Index* |
Gestational weight gain z-score at 16–19 weeks, kg |
Equivalent weight gain at 16 weeks, kg |
Adjusted** risk per 100 deliveries (95% confidence interval) |
Risk difference per 100 deliveries (95% confidence interval) |
|---|---|---|---|---|
| Normal weight | −2 | −0.87 | 2.5 (1.6, 3.3) | −0.44 (−0.98, 0.09) |
| −1 | 1.2 | 2.9 (2.4, 3.4) | Ref | |
| 0 | 3.9 | 3.5 (3.0, 4.1) | 0.64 (0.10, 1.2) | |
| +1 | 7.2 | 4.2 (3.5, 4.9) | 1.3 (0.50, 2.2) | |
| +2 | 11 | 5.1 (3.4, 6.8) | 2.2 (0.39, 4.0) | |
| Overweight | −2 | −3.8 | 4.6 (3.0, 6.1) | −0.06 (−1.0. 0.92) |
| −1 | −0.65 | 4.6 (3.8, 5.5) | Ref | |
| 0 | 3.4 | 4.7 (4.0, 5.4) | 0.07 (−0.75, 0.89) | |
| +1 | 8.5 | 4.8 (3.9, 5.7) | 0.19 (−1.0, 1.4) | |
| +2 | 15 | 4.9 (2.9, 7.0) | 0.33 (−2.1, 2.7) | |
| Grade 1 obesity | −2 | −5.8 | 7.2 (4.4, 9.9) | 0.85 (−0.95, 2.7) |
| −1 | −2.2 | 6.3 (5.1, 7.5) | Ref | |
| 0 | 2.4 | 5.8 (4.9, 6.7) | −0.52 (−1.7, 0.66) | |
| +1 | 8.3 | 6.4 (5.2, 7.6) | 0.88 (−1.6, 1.8) | |
| +2 | 15 | 7.5 (4.3, 10.8) | 1.2 (−2.4. 4.9) | |
| Grade 2 obesity | −2 | −7.3 | 3.3 (0.95, 5.6) | −1.2 (−2.3, −0.03) |
| −1 | −3.7 | 4.5 (3.0, 5.9) | Ref | |
| 0 | 1.1 | 5.8 (4.7, 6.9) | 1.3 (−0.31, 3.0) | |
| +1 | 7.3 | 6.2 (4.9, 7.4) | 1.7 (−0.29, 3.7) | |
| +2 | 15 | 6.1 (3.1, 9.2) | 1.7 (−1.7, 5.0) | |
| Grade 3 obesity | −2 | −9.9 | 5.5 (1.1, 10.0) | −0.41 (−2.9, 2.0) |
| −1 | −6.0 | 5.9 (3.7, 8.2) | Ref | |
| 0 | −0.30 | 6.5 (5.1, 7.9) | 0.58 (−1.8, 3.0) | |
| +1 | 7.8 | 8.3 (6.3, 10.3) | 2.4 (−0.69, 5.5) | |
| +2 | 19 | 11.2 (4.1, 18.2) | 5.2 (−2.0, 12.5) | |
| Grade 2 or 3 obesity | −2 | See above | 3.9 (1.8, 5.9) | −1.0 (−2.1, −0.01) |
| −1 | 4.9 (3.7, 6.1) | Ref | ||
| 0 | 6.1 (5.2, 7.0) | 1.2 (−0.17, 2.6) | ||
| +1 | 6.9 (5.8, 8.0) | 2.0 (0.35, 3.7) | ||
| +2 | 7.6 (4.6, 10.5) | 2.7 (−0.44, 5.8) |
Normal weight, body mass index (BMI) 18.5–24.9 kg/m2; overweight, BMI 25–29.9 kg/m2; grade 1 obese, BMI 30–34.9 kg/m2; grade 2 obese, BMI 35–39.9 kg/m2; grade 3 obese, BMI ≥ 40 kg/m2.
Adjusted for maternal race/ethnicity, maternal age, maternal education, parity, smoking during pregnancy, marital status and insurance.
For normal weight women, there was a steady increase in preeclampsia risk with increasing z-score at 16–19 weeks (Figure 1, Panel A). Compared with weight gain of 1.2 kg at 16 weeks’ gestation (z-score = −1 SD), weight gains of 3.9 kg (z-score = 0 SD), 7.2 kg (z-score = +1 SD) or 11.4 kg (z-score = +2 SD) at 16 weeks were associated with 0.64 (0.10, 1.2), 1.3 (0.50, 2.2), and 2.25 (0.39, 4.0) excess preeclampsia cases per 100 deliveries, respectively. For overweight and grade 1 obese women, risk curves were nearly flat (Figure 1, Panels B and C) and associations were null.
For both grade 2 and grade 3 obese women, risk curves suggested positive associations (Figure 1, Panels D and E), but the limited number of cases with very low and very high weight gains led to unstable estimates (Table 3). When these two groups of severely obese women were combined for improved precision, an association similar in direction and magnitude to normal weight women was observed (Figure 1, Panel F).
These results were not meaningfully different when we excluded preeclampsia cases delivered before 34 weeks gestation or when we reclassified prepregnancy BMI category using the first measured weight at <10 weeks (data not shown).
Table 4 compares the associations between high gestational weight gain and preeclampsia generated by the conventional analysis with those from the probabilistic bias analysis accounting for the misclassification of preeclampsia. As expected, the bias analysis results were less precise, reflecting the uncertainty due to random and systematic error. Overall, the point estimates suggest a modest amount of bias away from the null in the conventional results, but did not alter the conclusions.
Table 4.
Summary of estimates yielded from the conventional analysis of gestational weight gain z-score at 16–19 weeks in relation to preeclampsia and the probabilistic bias analysis accounting for random and systematic error due to misclassification of preeclampsia.*
| Prepregnancy body mass index category |
Gestational weight gain z- score at 16–19 weeks |
Conventional analysis: Adjusted** odds ratio (95% confidence interval) |
Probabilistic bias analysis accounting for misclassfication of preeclampsia: Adjusted** point estimate (95% bootstrapped sensitivity analysis interval) |
|---|---|---|---|
| Normal weight | ≤ +1 SD | 1.0 (ref) | 1.0 (ref) |
| > +1 SD | 1.5 (0.99, 2.1) | 1.6 (1.0, 2.6) | |
| Overweight | ≤ +1 SD | 1.0 (ref) | 1.0 (ref) |
| > +1 SD | 1.1 (0.72, 1.7) | 1.2 (0.73, 2.0) | |
| Grade 1 obese | ≤ +1 SD | 1.0 (ref) | 1.0 (ref) |
| > +1 SD | 1.4 (0.89, 2.1) | 1.5 (0.87, 2.5) | |
| Grade 2 or 3 obese | ≤ +1 SD | 1.0 (ref) | 1.0 (ref) |
| > +1 SD | 1.2 (0.75, 2.1) | 1.6 (0.77, 4.7) |
SD, standard deviation.
Normal weight, body mass index (BMI) 18.5–24.9 kg/m2; overweight, BMI 25–29.9 kg/m2; grade 1 obese, BMI 30–34.9 kg/m2; grade 2 or 3 obese, BMI ≥ 35 kg/m2.
Adjusted for maternal race/ethnicity, maternal age, maternal education, parity, smoking during pregnancy, marital status and insurance.
Discussion
We observed a positive dose-response relationship between gestational weight gain z-score at 16–19 weeks and the risk of preeclampsia among normal weight and severely obese women, but associations were null in the remaining BMI groups. These results were not meaningfully altered after removing the cases delivered at <34 weeks or when accounting for misclassification of the coding-based definition of preeclampsia.
Our study was observational and cannot determine whether the association between gestational weight gain in early pregnancy and preeclampsia is causal. A meta-analysis of randomized trials to reduce gestational weight gain found that diet and/or exercise interventions reduced the likelihood of excessive pregnancy weight gain at delivery, but had no impact on preeclampsia incidence (24). Nevertheless, it is not known whether the interventions studied reduced weight gain during early pregnancy, which is more likely the critical window of exposure for preeclampsia.
We were not able to separate preeclampsia cases based on gestational age of onset (because we lacked this information) or based on mild or severe symptoms (because of poor specificity of this coding in our dataset). However, the classification of overall preeclampsia that we used in our analysis had a high sensitivity and specificity, and our probabilistic bias analysis accounting for the outcome misclassification was consistent with the conventional analysis. We sampled our study cohort using a population at our hospital that had available data on prepregnancy weight and other key variables. While we cannot rule out the possibility of selection bias, characteristics of women in our final analysis were not meaningfully different from the cohort of eligible women. Prepregnancy BMI was calculated using self-reported prepregnancy weight, which is prone to error (12). However, our results were robust to calculating prepregnancy BMI using a measured weight at <10 weeks.
Strengths of our study include the formal bias analysis for misclassified case status, the use of measured gestational weight gain before the clinical presentation of preeclampsia, and accounting for women’s increased opportunity to gain weight with advancing gestational duration through the use of gestational age-specific z-score charts that were developed in the Astrid population (13, 14). Although some of our estimates for grade 2 or grade 3 obese women were imprecise, quantifying the association between gestational weight gain and preeclampsia in the heaviest women is an important contribution.
Our findings in normal weight women are consistent with the only other published study that we are aware of that evaluated early pregnancy weight gain in relation to preeclampsia. In a large cohort of women in the United Kingdom, MacDonald-Wallis et al. observed a linear relationship between rate of pregnancy weight gain up to 18 weeks’ gestation and risk of preeclampsia (7). Among lean women, women who gained 10.8 kg at 18 weeks had a higher risk than women who gained 3.6 kg at 18 weeks (risks = 4.4 vs. 1.5 per 100, respectively). These investigators observed no variation in the association by prepregnancy BMI, but their ability to detect differences by BMI may have been hampered by the relatively small number of preeclampsia cases (n=201 overall) and their predominantly lean sample (approximately 73% normal or underweight). Many other studies have evaluated the risk of preeclampsia relative to the total amount of weight gained up to 28 weeks’ gestation (25) or delivery (26-34), but edema and other clinical symptoms of preeclampsia, which occur after 20 weeks of pregnancy, cause weight gain. The 2009 IOM committee concluded that reverse causality may explain these findings, and interpretability is therefore limited (2009).
The variation of our findings across prepregnancy BMI categories may be explained by differences in the exposure, outcome, or confounders that we did not measure. Gestational weight gain is a crude measure of nutritional status. Early-pregnancy weight gain may differ in its composition depending on BMI (1), and one compartment of weight gain (fat mass, lean mass, or water) may be a stronger determinant of preeclampsia than the sum of all components (35, 36). Gestational weight gain is also a reflection of dietary intake and physical activity, which vary according to maternal weight (37-39). Unfortunately, we lacked data on these more specific nutritional measures that would help to clarify this finding. It is also possible that the preeclampsia phenotype differs according to prepregnancy BMI (40). While removing the cases delivered at <34 weeks did not change our results, we did not have information on placental pathology or accurate measures of intrauterine growth restriction to otherwise distinguish features of preeclampsia that may suggest underlying etiology (41).
The mechanisms potentially linking early-pregnancy weight gain to preeclampsia have not been tested directly, but research on prepregnancy BMI and preeclampsia may shed light on this association. There is a consistent and strong positive dose-response association between prepregnancy BMI and preeclampsia (42) that is likely due to a complex set of interrelated downstream pathways, including hypertension, insulin resistance, dyslipidemia, vascular dysfunction, and inflammation (43). We speculate that these metabolic and cardiovascular aberrations may also occur in women with excessive early pregnancy weight gain.
Conclusions
Additional research studies with serial antenatal weight measurements, rigorous diagnosis of preeclampsia, and large samples of women in each BMI category are needed to determine whether future pregnancy weight gain recommendations should account for preeclampsia risk when deriving optimal weight gain ranges. If our results are confirmed by others, the inclusion of preeclampsia risk into evidence-based guidelines may favor lower weight gain for some women.
Supplementary Material
Highlights.
Preeclampsia and early weight gain (16-19 weeks gestation) were positively associated
Preeclampsia risk rose steadily alongside early weight gain in normal weight women
Early weight loss was associated with less preeclampsia risk in class 2/3 obese women
Associations were null among overweight and class 1 obese women
Acknowledgements
We would like to thank Melissa J. Papic for excellent data management.
Funding
This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development [R01 HD072008 to LMB and JAH]. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. JAH holds New Investigator awards from the Canadian Institutes of Health Research and the Michael Smith Foundation for Health Research.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declarations of interest: none
References
- 1.Institute of Medicine. Weight Gain During Pregnancy: Reexamining the Guidelines. Washington, DC: National Academies Press; 2009. [PubMed] [Google Scholar]
- 2.Duley L. The global impact of pre-eclampsia and eclampsia. Seminars in Perinatology. 2009. June;33(3):130–7. PubMed PMID: 19464502. Epub 2009/05/26. eng. [DOI] [PubMed] [Google Scholar]
- 3.Cedergren MI. Optimal gestational weight gain for body mass index categories. Obstet Gynecol. 2007. October;110(4):759–64. PubMed PMID: 17906006. Epub 2007/10/02. eng. [DOI] [PubMed] [Google Scholar]
- 4.Thorsdottir I, Torfadottir JE, Birgisdottir BE, Geirsson RT. Weight gain in women of normal weight before pregnancy: complications in pregnancy or delivery and birth outcome. Obstet Gynecol. 2002. May;99(5 Pt 1):799–806. PubMed PMID: 11978290. Epub 2002/04/30. eng. [DOI] [PubMed] [Google Scholar]
- 5.Mbah AK, Kornosky JL, Kristensen S, August EM, Alio AP, Marty PJ, et al. Super-obesity and risk for early and late pre-eclampsia. BJOG. 2010. July;117(8):997–1004. PubMed PMID: 20482533. Epub 2010/05/21. eng. [DOI] [PubMed] [Google Scholar]
- 6.Markham KB, Funai EF. Pregnancy-related hypertension In: Creasy RK, Resnik R, Iams JD, Lockwood CJ, Moore TR, Greene MF, editors. Creasy and Resnik's Maternal-Fetal Medicine, Seventh Edition. Philadelphia, PA: Elsevier; 2014. p. 756–84. [Google Scholar]
- 7.Macdonald-Wallis C, Tilling K, Fraser A, Nelson SM, Lawlor DA. Gestational weight gain as a risk factor for hypertensive disorders of pregnancy. Am J Obstet Gynecol. 2013. May 24. PubMed PMID: 23711667. Epub 2013/05/29. Eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cologne J, Preston DL, Imai K, Misumi M, Yoshida K, Hayashi T, et al. Conventional case-cohort design and analysis for studies of interaction. Int J Epidemiol. 2012. August;41(4):1174–86. PubMed PMID: 22815332. Epub 2012/07/21. eng. [DOI] [PubMed] [Google Scholar]
- 9.Ness RB, Markovic N, Bass D, Harger G, Roberts JM. Family history of hypertension, heart disease, and stroke among women who develop hypertension in pregnancy. Obstet Gynecol. 2003. December;102(6):1366–71. PubMed PMID: 14662228. [DOI] [PubMed] [Google Scholar]
- 10.American College of Obstetricians and Gynecologists Committee on Obstetric Practice. ACOG Practice Bulletin, No. 33. Diagnosis and management of preeclampsia and eclampsia. January 2002. International Journal of Gynaecology & Obstetrics. 2002;77(1):67–75. [PubMed] [Google Scholar]
- 11.Myatt L, Redman CW, Staff AC, Hansson S, Wilson ML, Laivuori H, et al. Strategy for standardization of preeclampsia research study design. Hypertension. 2014. June;63(6):1293–301. PubMed PMID: 24688121. Epub 2014/04/02. eng. [DOI] [PubMed] [Google Scholar]
- 12.Bodnar LM, Abrams B, Bertolet M, Gernand AD, Parisi SM, Himes KP, et al. Validity of birth certificate-derived maternal weight data. Paediatr Perinat Epidemiol. 2014. May;28(3):203–12 PubMed PMID: 24673550. Epub 2014/03/29. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hutcheon JA, Platt RW, Abrams B, Himes KP, Simhan HN, Bodnar LM. A weight-gain-for-gestational-age z score chart for the assessment of maternal weight gain in pregnancy. Am J Clin Nutr. 2013. May;97(5):1062–7. PubMed PMID: 23466397. Pubmed Central PMCID: 3625243. Epub 2013/03/08. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hutcheon JA, Platt RW, Abrams B, Himes KP, Simhan HN, Bodnar LM. Pregnancy weight gain charts for obese and overweight women. Obesity (Silver Spring, Md). 2015. March;23(3):532–5. PubMed PMID: 25707378. Pubmed Central PMCID: Pmc4340088. Epub 2015/02/25. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Yang S, Hutcheon JA. Identifying outliers and implausible values in growth trajectory data. Annals of epidemiology. 2016. January;26(1):77–80.e1-2. PubMed PMID: 26590476. Pubmed Central PMCID: PMC4732581. Epub 2015/11/23. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.WHO Consultation on Obesity. Obesity: Preventing and Managing the Global Epidemic. Geneva, Switzerland: World Health Organization, 2000. WHO Technical Report Series 894. [PubMed] [Google Scholar]
- 17.Abuhamad AZ. ACOG Practice Bulletin, clinical management guidelines for obstetrician-gynecologists number 98, October 2008 (replaces Practice Bulletin number 58, December 2004). Ultrasonography in pregnancy. Obstetrics & Gynecology. 2008. October;112(4):951–61. PubMed PMID: 18827142. Epub 2008/10/02. eng. [DOI] [PubMed] [Google Scholar]
- 18.Barlow WE, Ichikawa L, Rosner D, Izumi S. Analysis of case-cohort designs. J Clin Epidemiol. 1999. December;52(12):1165–72. PubMed PMID: 10580779. Epub 1999/12/02. eng. [DOI] [PubMed] [Google Scholar]
- 19.Localio AR, Margolis DJ, Berlin JA. Relative risks and confidence intervals were easily computed indirectly from multivariable logistic regression. J Clin Epidemiol. 2007. September;60(9):874–82. PubMed PMID: 17689803. Epub 2007/08/11. eng. [DOI] [PubMed] [Google Scholar]
- 20.Harrell FE. Regression modeling strategies with applications to linear models, logistic regression and survival analysis. New York, USA: Springer-Verlag; 2001. [Google Scholar]
- 21.Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol. 2008;8:70. PubMed PMID: 18973665. Pubmed Central PMCID: 2601045. Epub 2008/11/01. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lash TL, Fink AK. Semi-automated sensitivity analysis to assess systematic errors in observational data. Epidemiology. 2003. July;14(4):451–8. PubMed PMID: 12843771. [DOI] [PubMed] [Google Scholar]
- 23.Fox MP, Lash TL, Greenland S. A method to automate probabilistic sensitivity analyses of misclassified binary variables. Int J Epidemiol. 2005. December;34(6):1370–6. PubMed PMID: 16172102. Epub 2005/09/21. eng. [DOI] [PubMed] [Google Scholar]
- 24.Muktabhant B, Lawrie TA, Lumbiganon P, Laopaiboon M. Diet or exercise, or both, for preventing excessive weight gain in pregnancy. Cochrane Database Syst Rev. 2015;6:CD007145. PubMed PMID: 26068707. Epub 2015/06/13. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ruhstaller KE, Bastek JA, Thomas A, McElrath TF, Parry SI, Durnwald CP. The Effect of Early Excessive Weight Gain on the Development of Hypertension in Pregnancy. Amer J Perinatol. 2016 // 28September2016;33(12):1205–10. Epub 04.08.2016. En. [DOI] [PubMed] [Google Scholar]
- 26.Viswanathan M, Siega-Riz AM, Moos MK, Deierlein A, Mumford S, Knaack J, et al. Outcomes of Maternal Weight Gain, Evidence Report/Technology Assessment No. 168. AHRQ Publication No. 08-E-09. Rockville, MD: Agency for Healthcare Research and Quality; 2008. [PMC free article] [PubMed] [Google Scholar]
- 27.DeVader SR, Neeley HL, Myles TD, Leet TL. Evaluation of gestational weight gain guidelines for women with normal prepregnancy body mass index. Obstet Gynecol. 2007. October;110(4):745–51. PubMed PMID: 17906004. Epub 2007/10/02. eng. [DOI] [PubMed] [Google Scholar]
- 28.Kiel DW, Dodson EA, Artal R, Boehmer TK, Leet TL. Gestational weight gain and pregnancy outcomes in obese women: how much is enough? Obstet Gynecol. 2007. October;110(4):752–8. PubMed PMID: 17906005. Epub 2007/10/02. eng. [DOI] [PubMed] [Google Scholar]
- 29.Zhou A, Xiong C, Hu R, Zhang Y, Bassig BA, Triche E, et al. Pre-Pregnancy BMI, Gestational Weight Gain, and the Risk of Hypertensive Disorders of Pregnancy: A Cohort Study in Wuhan, China. PloS one. 2015;10(8):e0136291. PubMed PMID: 26305565. Pubmed Central PMCID: 4548954. Epub 2015/08/26. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Truong YN, Yee LM, Caughey AB, Cheng YW. Weight gain in pregnancy: does the Institute of Medicine have it right? Am J Obstet Gynecol. 2015. March;212(3):362 e1–8. PubMed PMID: 25725659. Epub 2015/03/03. eng. [DOI] [PubMed] [Google Scholar]
- 31.Haugen M, Brantsæter AL, Winkvist A, Lissner L, Alexander J, Oftedal B, et al. Associations of pre-pregnancy body mass index and gestational weight gain with pregnancy outcome and postpartum weight retention: a prospective observational cohort study. BMC Pregnancy and Childbirth. 2014. June 11;14(1):201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Swank ML, Marshall NE, Caughey AB, Main EK, Gilbert WM, Melsop KA, et al. Pregnancy Outcomes in the Super Obese, Stratified by Weight Gain Above and Below Institute of Medicine Guidelines. Obstetrics & Gynecology. 2014;124(6):1105–10. PubMed PMID: 00006250-201412000-00007. [DOI] [PubMed] [Google Scholar]
- 33.Chasan-Taber L, Silveira M, Waring ME, Pekow P, Braun B, Manson JE, et al. Gestational Weight Gain, Body Mass Index, and Risk of Hypertensive Disorders of Pregnancy in a Predominantly Puerto Rican Population. Maternal and child health journal. 2016;20(9):1804–13. PubMed PMID: PMC5007203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Gavard JA. Gestational Weight Gain and Maternal and Neonatal Outcomes in Underweight Pregnant Women: A Population-Based Historical Cohort Study. Maternal and Child Health Journal. 2017 2017/May/01;21(5):1203–10. [DOI] [PubMed] [Google Scholar]
- 35.Wang Y, Qiu J, Zhou M, Du Y. Increasing maternal percentage body fat in early second trimester: a risk factor for preeclampsia. J Matern Fetal Neonatal Med. 2015. February;28(3):293–6. PubMed PMID: 24916361. Epub 2014/06/12. eng. [DOI] [PubMed] [Google Scholar]
- 36.Sween LK, Althouse AD, Roberts JM. Early-pregnancy percent body fat in relation to preeclampsia risk in obese women. Am J Obstet Gynecol. 2015. January;212(1):84 e1–7. PubMed PMID: 25088867. Epub 2014/08/05. eng. [DOI] [PubMed] [Google Scholar]
- 37.Laraia BA, Bodnar LM, Siega-Riz AM. Pregravid body mass index is negatively associated with diet quality during pregnancy. Public Health Nutr. 2007. September;10(9):920–6. PubMed PMID: 17381955. [DOI] [PubMed] [Google Scholar]
- 38.Stuebe AM, Oken E, Gillman MW. Associations of diet and physical activity during pregnancy with risk for excessive gestational weight gain. Am J Obstet Gynecol. 2009. July;201(1):58 e1–8. PubMed PMID: 19467640. Pubmed Central PMCID: 2706304. Epub 2009/05/27. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Deierlein AL, Siega-Riz AM, Evenson KR. Physical activity during pregnancy and risk of hyperglycemia. Journal of women's health. 2012. July;21(7):769–75. PubMed PMID: 22537020. Pubmed Central PMCID: 3387759. Epub 2012/04/28. eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Roberts JM, Bodnar LM, Patrick TE, Powers RW. The Role of Obesity in Preeclampsia. Pregnancy Hypertens. 2011. January 1;1(1):6–16. PubMed PMID: 21532964. Pubmed Central PMCID: 3082136. Epub 2011/05/03. Eng. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.von Dadelszen P, Magee LA, Roberts JM. Subclassification of preeclampsia. Hypertens Pregnancy. 2003;22(2):143–8. PubMed PMID: 12908998. Epub 2003/08/12. eng. [DOI] [PubMed] [Google Scholar]
- 42.Wang Z, Wang P, Liu H, He X, Zhang J, Yan H, et al. Maternal adiposity as an independent risk factor for pre-eclampsia: a meta-analysis of prospective cohort studies. Obes Rev. 2013. June;14(6):508–21. PubMed PMID: 23530552. Epub 2013/03/28. eng. [DOI] [PubMed] [Google Scholar]
- 43.Sattar N, Freeman DJ. Potential mechanisms contributing to gestational diabetes and pre-eclampsia in the obese woman In: Gillman MW, Poston L, editors. Maternal Obesity. Cambridge: Cambridge University Press; 2012. p. 45–55. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.