Abstract
Objective
To assess weight change and attempted weight loss during the 12–18 months before spontaneous conception in relation to the risk of pregnancy loss.
Design
Prospective cohort study.
Setting
United States, 2007–2011.
Methods
Women (n=629) who were attempting pregnancy reported at baseline any weight loss attempts over the past 12 months, and their minimum and maximum weights during that time. Follow-up lasted 1–6 menstrual cycles and throughout pregnancy. Using bodyweight measured at 4 weeks’ gestation, participants were categorised as having weight loss ≥5%, weight gain ≥5%, both, or neither, over the previous 12–18 months. Log-binomial models adjusted for potential confounders.
Main Outcome Measures
Risk ratio (RR) and 95% confidence interval (CI) of pregnancy loss.
Results
Attempted weight loss was reported by 44% of women and actual weight loss by 11%, but neither was consistently associated with pregnancy loss. The RR for recent weight gain ≥5% was 1.65 [CI 1.09, 2.49].
Conclusions
Weight gain over the period spanning 12–18 months preconception to 4 weeks’ gestation may increase the risk of pregnancy loss among fertile women with prior pregnancy losses. Attempted and actual weight loss were not associated with pregnancy loss, however, replication is needed from larger studies with data on particular weight-loss methods.
Funding
Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland (Contract Nos. HHSN267200603423, HHSN267200603424, HHSN267200603426).
Keywords: Abortion, Spontaneous; Diet, Reducing; Embryo Loss; Weight Gain; Weight Loss
INTRODUCTION
Women who are obese or overweight may have an increased risk of pregnancy loss in natural and assisted conceptions,1–4 plausibly due to decrements in oocyte quality5 or endometrial receptivity.6 However, among women who were obese and undergoing infertility treatment, a recent trial found that short-term, medically-supervised weight loss made no difference in clinical pregnancy rates and suggested an increased risk of clinical pregnancy loss,7 contradicting earlier studies that found such interventions improved outcomes among women with obesity and infertility.8–10 Moreover, among United States (US) nurses, weight loss since adolescence was associated with a reduced risk of fetal loss (spontaneous abortion or stillbirth) and weight gain was associated with an increased risk,3 although this investigation lacked data on weight-loss attempts (i.e. weight-loss intention).
Indeed, the effect on pregnancy loss from self-initiated weight-loss attempts and from weight change in the period closely preceding conception has received little study,11, 12 although self-initiated weight-loss attempts are common among US women.13 Importantly, prior findings from clinical weight-loss studies may not be generalisable to women’s self-initiated weight-loss attempts due to differences in the study populations, professional support, and weight-loss practices applied.14 Adding to the uncertainty surrounding benefits from self-initiated weight-loss attempts, clinically significant weight loss and maintenance in this setting are relatively rare,14 and some women use unhealthy weight-loss methods.15
Therefore, we prospectively examined women’s self-report of their recent, self-initiated weight-loss attempts, minimum weight, and maximum weight during the 12–18 months prior to achieving natural conception, in relation to the risk of pregnancy loss.
METHODS
The present study used data from the Effects of Aspirin in Gestation and Reproduction (EAGeR) trial, which enrolled 1,228 women with a history of 1–2 pregnancy losses, from 2007–2011 at 4 U.S. study centers.16 Participants were 18–40 years old, attempting pregnancy, had up to 2 live births, regular menstrual cycles lasting 21–42 days, and no major medical problems or infertility. The trial’s primary aim was to assess the effect of preconceptionally initiated, daily 81 mg aspirin on live birth.16 Women were followed for up to six menstrual cycles while trying to become pregnant, and, if they became pregnant, throughout pregnancy. They continued taking the study pill up to gestational week 36, if applicable. The institutional review board at each study center approved the trial protocol, and participants provided written informed consent.
Ascertainment of pregnancy and pregnancy loss
Pregnancies were identified through systematic use of urinary human chorionic gonadotropin (hCG) pregnancy tests (Quickvue®, Quidel, San Diego, California) at home and at study visits following the end of each menstrual cycle, and through beta-hCG assays of stored first-morning urine samples which participants collected daily during the first 2 menstrual cycles of the study (initial test: BioVendor, Asheville, NC; confirmatory test: Diagnostic Automation Inc., Calabasas, CA).17 Participants had clinical confirmation of pregnancy by ultrasound detection of a gestational sac at 6–7 weeks’ gestation; or by any documentation of fetal heart tones or of pregnancy at a later stage. A pregnancy loss could be either an early pregnancy loss, defined as an hCG-detected pregnancy that did not last until clinical confirmation, or a loss detected following clinical confirmation.
Actual weight change
At baseline (i.e., enrollment) participants reported their minimum and maximum weights in the past 12 months via a questionnaire (representing 12–18 months preconception, as women conceived within 6 cycles after enrolling). The exposure of interest was weight change in the time period preceding and proximal to conception. Thus, ‘current weight’ was the weight measured at the time of a positive pregnancy test (approximately 4 weeks’ gestation), which occurred approximately 2 weeks post-conception and <1 week post-implantation. If this measure was unavailable, we used the measured (7% of participants) or self-reported (1% of participants) weight at baseline. All weight change measures were anchored to the current weight because the self-reported minimum and maximum weights clearly preceded current weight. Weight loss was calculated as the maximum weight in the 12 months before enrollment minus the current weight. Percent weight loss was calculated as weight loss divided by the maximum weight. Weight gain and percent weight gain were calculated in the same manner but using the minimum (instead of maximum) weight in the 12 months before enrollment.
Attempted weight change
Attempted weight loss and gain in the 12 months before enrollment was assessed at baseline via a questionnaire, as well as the specified methods for ‘weight control’ (choice of 11 commercial plans, ‘personal plan to control food intake,’ ‘extreme exercising,’ and ‘other’), and the duration of the most recent method used. Participants were categorised as having attempted weight loss if they responded ‘yes’ to attempting weight loss and reported a duration ≥1 month. Participants who reported a weight loss attempt lasting <1 month were analysed in a separate category because we hypothesised any effect of attempted weight loss would require sustained behavior.
Exclusions
An initial screening questionnaire was completed by 5,485 women, and 2,323 (42.4%) were potentially eligible (Figure S1). Women (n=1,577, 67.9%) attended a baseline clinic visit to confirm eligibility, and 1,228 (77.9%) were confirmed eligible and enrolled in EAGeR.18 We excluded women who had no hCG-detected pregnancy when they completed the study (n=303) or withdrew (n=140) and those who withdrew while pregnant (n=14). To avoid confounding by medical conditions, we further excluded women who reported that in the last 12 months they had a live birth (n=60), a pregnancy loss at gestational age >14 weeks (n=32), anorexia and/or bulimia (n=11), or weight loss due to another medical condition (n=13). We also excluded those who reported to have attempted weight gain (n=28) because this behavior and its associated health outcomes were outside the scope of the present research study. Finally, we excluded 12 women with maximum Body mass index (BMI) <18.5 because this small group of underweight women is expected to have different reproductive effects from attempted weight loss and actual weight change. Compared with the 629 women who remained eligible for the analysis, women who were excluded from the analysis due to withdrawing from the study were less likely to have attempted weight loss for ≥1 month compared to women who were included in the analysis (34% vs. 44%, P=0.03), but had a similar probability of weight loss (11% vs. 14%, P=0.45) and gain (42% vs. 51%, P=0.12).
Statistical analyses
Weight change was categorised as weight gain (≥5%), weight loss (≥5%), high weight variability (both weight gain and weight loss ≥5%), and constant weight (no weight gain or loss ≥5%).7, 8, 19 A secondary analysis categorised participants according to absolute weight change ≥4 kg, for consistency with a prior study of weight change and fetal loss.3
Distributions of participants’ baseline characteristics and behaviors were examined overall and by category of weight change. We used log-binomial models to estimate the risk ratio (RR) and 95% confidence interval (CI) of pregnancy loss across categories of weight change (reference: constant weight), weight-loss attempt (reference: no attempt to lose weight), and their combination (reference: constant weight and no attempt to control weight). Stabilised inverse probability of pregnancy weighting accounted for potential selection bias if the exposures affected the chance of pregnancy.20 Models were adjusted for age, parity, number of prior losses, and pregnancy loss in the 12 months before enrollment. The RR for attempted weight loss was also adjusted for maximum BMI in the 12 months before enrollment because we assumed that the maximum weight prompted the attempted weight loss for most women. Because of uncertainty about the timing of the attempted weight loss relative to weight change in the 12 months before enrollment, weight change could be either a cause or effect of attempted weight loss. Thus, we did not adjust either factor for the other.21 The results were essentially unchanged when we further adjusted for vigorous physical activity at baseline as reported in the International Physical Activity Questionnaire – Short Form22 (some versus none) and treatment assignment,18 and when we further adjusted weight change for attempted weight loss. We also stratified analyses by maximum BMI in the 12 months before enrollment (<25 versus ≥25) due to evidence that long-term weight loss was more strongly associated with a reduced risk of pregnancy loss among women who had been overweight.3 Exploratory analyses of secondary outcomes, early pregnancy loss and clinical pregnancy loss, also used these models.
Missing exposure and covariate data were imputed using PROC MI in SAS/STAT® version 9.4 (SAS Institute, Cary, NC). The frequency of missing data by variable were: current weight <1%, attempted weight loss 4%, attempted weight gain 4%, actual weight loss 1%, actual weight gain 5%, maximum BMI in past 12 months <1%, minimum BMI in past 12 months 5%, education <1%, household income <1%, and smoking 1%.
RESULTS
Among the 629 pregnant women analysed, 28% maintained a constant weight in the 12 months before enrollment, 49% gained weight, 14% lost weight, 6% both gained and lost weight, and 3% had their weight change imputed due to missing data (Table 1). Using baseline weight to calculate weight change, instead of weight at 4 weeks’ gestation, resulted in fewer women in the gained weight category (44%) and more women in the constant weight category (35%). Compared to women who maintained a constant weight, both women who gained weight and women who lost weight were more likely to have attempted weight loss, less likely to engage in moderate exercise, and had a higher parity. Women who lost weight were also less likely to have had a recent pregnancy loss; and had more previous pregnancy losses and higher maximum BMI in the 12 months before enrollment.
Table 1.
Baseline characteristics overall and by recent weight change in the past 12–18 months among 629 women who became pregnant during follow-up. The EAGeR Trial, USA, 2007–2011.a
| Overall | Weight change over 12 months preceding enrollment: | ||||
|---|---|---|---|---|---|
| Constant | Gained ≥5% | Lost ≥5% | Gained ≥5% and lost ≥5% |
||
| Number of women, n | 629 | 175 | 311 | 87 | 38 |
| Age ≥35 years, n (%): | 75 (12) | 23 (13) | 32 (10) | 12 (14) | 8 (21) |
| Attempted weight loss ≥1 month, n (%) | 272 (44) | 57 (33) | 157 (50) | 34 (39) | 24 (63) |
| Attempted weight loss <1 month, n(%) | 172 (28) | 43 (24) | 87 (28) | 32 (37) | 10 (26) |
| Current weight (kg), mean (SD) | 72.3 (18.2) | 66.0 (14.1) | 77.1 (19.7) | 66.2 (13.0) | 73.0 (17.8) |
| Weight loss (kg) median (IQR) | 1.0 (0.0–3.1) | 1.4 (0.6–2.3) | 0.0 (0.0–1.1) | 5.9 (4.0–9.2) | 7.4 (4.9–11.1) |
| Weight gain (kg) median (IQR) | 3.7 (1.7–6.9) | 1.5 (0.8–2.3) | 6.5 (4.4–9.3) | 0.8 (0.0–2.2) | 4.9 (4.1–7.4) |
| Maximum BMI, kg/m2, n (%) | |||||
| 18.5–24.9 | 301 (50) | 123 (71) | 126 (41) | 41 (48) | 11 (29) |
| 25–29.9 | 156 (26) | 32 (18) | 90 (29) | 20 (24) | 14 (37) |
| ≥30 | 147 (24) | 19 (11) | 91 (30) | 24 (28) | 13 (34) |
| Vigorous physical activity (hrs/week), median (IQR) | 0.5 (0.0–2.0) | 0.67 (0.0–2.2) | 0.0 (0.0–2.2) | 0.25 (0.0–2.0) | 0.25 (0.0–1.5) |
| Moderate physical activity (hrs/week), median (IQR) | 1.5 (0.0–4.0) | 1.2 (0.0–4.0) | 1.5 (0.0–5.0) | 1.5 (0.0–5.0) | 1.4 (0.0–3.0) |
| Parity, n (%) | |||||
| Nulliparous | 270 (44) | 87 (50) | 135 (43) | 31 (36) | 17 (45) |
| 1 live birth | 223 (36) | 60 (34) | 114 (37) | 36 (41) | 13 (34) |
| 2 live births | 118 (19) | 28 (16) | 62 (20) | 20 (23) | 8 (21) |
| Pregnancy loss in past yearb, n (%) | 512 (84) | 151 (86) | 269 (86) | 67 (77) | 25 (66) |
| Two prior losses, n (%) | 221 (36) | 57 (33) | 105 (34) | 41 (47) | 18 (47) |
| Annual household income ≥$75,000, n (%) | 349 (57) | 103 (59) | 168 (54) | 47 (54) | 31 (82) |
| White, n (%) | 593 (97) | 172 (98) | 303 (97) | 82 (94) | 36 (95) |
| High school diploma, n (%) | 546 (89) | 161 (92) | 272 (87) | 80 (92) | 33 (87) |
| Current regular smoker, n (%) | 24 (4) | 6 (3) | 12 (4) | 5 (6) | 1 (3) |
EAGeR, Effects of Aspirin on Gestation and Reproduction; SD, standard deviation; IQR, interquartile range; BMI, body mass index.
Statistics were calculated from complete data. Certain variables were missing data for: weight change (n=18), maximum BMI (n=8), smoking (n=3).
All pregnancy losses occurred at gestational ages≤14 weeks.
Among the 42% of participants who tried to lose weight, most reported using a “personal plan to control food intake” (79%). Participants could select >1 method and other common choices were “Weight Watchers™” (19%), “Other” (19%, almost half of which were further specified: exercise), “Extreme exercising” (11%), and “Shake diet (e.g., Slimfast™)” (9%).
There were 152 pregnancy losses: 41 early pregnancy losses and 111 losses after clinical confirmation. Relative to maintaining a constant weight, weight gain was associated with an increased risk of pregnancy loss (weight gain ≥5% RR=1.65, 95% CI 1.09, 2.49; weight gain ≥4 kg RR=1.41, 95% CI 0.98, 2.02; Table 2). Stratified by 12 month maximum BMI, among women with maximum BMI 18.5–24.9 kg/m2, weight gain ≥5% RR=1.57 [CI 0.98, 2.50] and weight gain ≥4 kg RR=1.06 [CI 0.71, 1.59], and among women with maximum BMI ≥25 kg/m2, weight gain ≥5% RR=1.49 [CI 0.78, 2.84] and weight gain ≥4 kg RR=2.10 [CI 0.93, 4.71]. Actual weight loss and weight variability showed little association with risk of pregnancy loss overall. Among women with maximum BMI 18.5–24.9 kg/m2, the RR for weight loss ≥5% was 1.77 [CI 0.99, 3.16], while the RR for weight loss ≥4 kg was 1.00 [CI 0.45, 2.25]. Attempted weight loss was not associated with risk of pregnancy loss, relative to no attempt to control weight (RR 1.00, 95% CI 0.67–1.48).
Table 2.
Recent attempted and actual weight loss, in relation to pregnancy loss: 629 pregnant women, the EAGeR Trial, USA, 2007–2011.
| BMI in past 12 monthsc | ||||||
|---|---|---|---|---|---|---|
| Losses, n |
Women, n |
RR (95% CI)a | RR (95% CI) | BMI 18.5–24.9 kg/m2 RR (95% CI) |
BMI ≥25 kg/m2 RR (95% CI) |
|
| Recent weight change (percent) b | ||||||
| Constant: weight change <5% | 35 | 183 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
| Gained ≥5%e | 90 | 321 | 1.55 (1.03, 2.33) | 1.65 (1.09, 2.49) | 1.57 (0.98, 2.50) | 1.49 (0.78, 2.84) |
| Lost ≥5% | 20 | 87 | 1.25 (0.73, 2.14) | 1.14 (0.64, 2.03) | 1.77 (0.99,3.16) | 0.76 (0.28, 2.06) |
| Gained ≥5% and lost ≥5% | 7 | 38 | 1.02 (0.48, 2.19) | 0.92 (0.44, 1.92) | --d | --d |
| Recent weight change (absolute) b | ||||||
| Constant: weight change <4kg | 56 | 255 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
| Gained ≥4kge | 73 | 267 | 1.38 (0.97, 1.98) | 1.41 (0.98, 2.02) | 1.06 (0.71, 1.59) | 2.10 (0.93, 4.71) |
| Lost ≥4kgf | 16 | 73 | 1.13 (0.67, 1.91) | 0.99 (0.56, 1.77) | 1.00 (0.45, 2.25) | 1.27 (0.51, 3.14) |
| Gained ≥4kg and lost ≥4kg | 7 | 34 | 0.98 (0.47, 2.03) | 0.88 (0.44, 1.79) | --d | --d |
| Weight loss attempt in past yearg | ||||||
| No attempted weight loss | 39 | 178 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
| Attempted weight loss ≥1 month | 65 | 274 | 1.16 (0.78, 1.70) | 1.00 (0.67, 1.48) | 0.78 (0.42, 1.45) | 1.60 (0.70, 3.65) |
| Attempted weight loss <1 month | 48 | 177 | 1.23 (0.83, 1.84) | 1.13 (0.78, 1.65) | 1.43 (0.89, 2.30) | 1.54 (0.65, 3.64) |
BMI, body mass index; CI, confidence interval; RR, risk ratio.
Model used stabilized inverse probability of pregnancy weights to adjust for the selection of women who became pregnant.
Model used stabilized inverse probability of pregnancy weights and further adjusted for age≥35 years, number of previous live births (0,1,2), pregnancy loss in the 12 months before enrollment, and number of prior losses (1 vs. 2). Stratified analyses of certain exposure categories used a more parsimonious model, as detailed in the footnotes below.
Analyses of weight gain vs. constant weight were stratified by minimum BMI in the 12 months before enrollment, and analyses of weight loss vs. constant weight were stratified by maximum BMI in the 12 months before enrollment.
Stratified analyses were not done due to the small number of pregnancy losses.
Analyses of gained ≥4kg and gained ≥5% in the stratum of women with BMI ≥25 kg/m2 used a more parsimonious model that adjusted for selection of women who became pregnant, age≥35 years, and pregnancy loss in the 12 months before enrollment.
Analyses of lost ≥4kg stratified by BMI were adjusted for selection of women who became pregnant, age≥35 years, and pregnancy loss in the 12 months before enrollment.
Model used stabilized inverse probability of pregnancy weights and further adjusted for age≥35 years, maximum BMI in the 12 months before enrollment, number of previous live births (0,1,2), pregnancy loss in the 12 months before enrollment, and number of prior losses (1 vs. 2).
Among the 274 women who attempted weight loss, weight loss ≥5% was achieved by 34 (12%) (Table 3). The three pregnancy losses in this group corresponded to a reduction in the adjusted RR (RR=0.44, 95% CI 0.12, 1.29). In contrast, there was a suggestive increased risk of pregnancy loss among women whose attempt at weight loss was accompanied by weight gain (158 women, RR=1.69, 95% CI 0.96, 2.98), but not among women whose attempt was accompanied by constant weight (58 women, RR=1.09, 95% CI 0.47, 2.56). The respective RRs in categories defined by weight change ≥4kg were similar (Table 3). The other categories of combined exposure also produced weak and imprecise associations.
Table 3.
Combined categories of recent attempted and actual weight change in relation to pregnancy loss: 424 pregnant women, the EAGeR Trial, USA, 2007–2011.a
| Losses, n |
Women, n |
Model 1b RR (95% CI) |
Model 2c RR (95% CI) |
Maximum BMI 18.5–24.9 kg/m2 RR (95% CI)c |
Maximum BMI ≥25 kg/m2 RR (95% CI)c |
|
|---|---|---|---|---|---|---|
| Relative weight change | ||||||
| No attempt, constant weight | 15 | 79 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
| Attempted to lose, gained ≥5% | 47 | 158 | 1.63 (0.89, 2.98) | 1.69 (0.96, 2.98) | 0.98 (0.46, 2.08) | 1.73 (0.46, 6.57) |
| No attempt, gained ≥5% | 18 | 74 | 1.21 (0.61, 2.42) | 1.23 (0.64, 2.36) | 0.98 (0.45, 2.13) | 0.51 (0.08, 3.09) |
| Attempted to lose, lost ≥5%d | 3 | 34 | 0.47 (0.12, 1.35) | 0.44 (0.12,1.29) | ---e | ---e |
| No attempt, lost ≥5%d | 5 | 21 | 1.28 (0.51, 3.19) | 1.25 (0.52, 2.99) | ---e | ---e |
| Attempted to lose, constant weightd | 13 | 58 | 1.09 (0.47, 2.52) | 1.09 (0.47, 2.56) | 0.82 (0.35, 1.95) | 1.35 (0.29, 6.24) |
| Absolute weight changef | ||||||
| No attempt, constant weight | 22 | 109 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
| Attempted to lose, gained ≥4kg | 41 | 141 | 1.49 (0.90, 2.46) | 1.48 (0.91, 2.40) | 1.06 (0.39, 2.90) | 2.50 (0.59, 10.61) |
| No attempt, gained ≥4kg | 14 | 54 | 1.10 (0.56, 2.15) | 1.10 (0.58, 2.06) | 1.58 (0.80, 3.13) | 0.86 (0.14, 5.25) |
| Attempted to lose, lost ≥4kg d | 4 | 35 | 0.56 (0.18, 1.32) | 0.64 (0.20, 1.46) | ---e | ---e |
| No attempt, lost ≥4kg d | 3 | 12 | 1.25 (0.36, 2.15) | 1.51 (0.43, 2.32) | ---e | ---e |
| Attempted to lose, constant weight | 17 | 76 | 0.87 (0.44, 1.73) | 0.82 (0.39, 1.73) | 0.87 (0.44, 1.72) | 1.20 (0.21, 6.97) |
CI, confidence interval; RR, risk ratio.
Analysis included 424 women who reported either no attempt to lose weight or attempting to lose weight≥1 month, and who either gained ≥5%, lost ≥5%, or maintained a constant weight <5%. Women in the other categories were excluded due to small numbers. The analysis of categories defined by absolute weight change ≥4kg included 427 women.
Model 1 used stabilized inverse probability of pregnancy weights to adjust for the selection of women who became pregnant.
Model 2 adjusted for the selection of women who became pregnant, age≥35 years, loss in the past year, and number of prior losses (1 vs. 2).
Due to the small numbers of pregnancy losses in this exposure category, results from Models 1 and 2 were calculated using exact logistic regression with no adjustment for selection of women who became pregnant.
This analysis was not performed due to the small numbers of pregnancy losses in this exposure category.
In the stratum BMI 18.5–24.9 kg/m2, the model used stabilized inverse probability of pregnancy weights to adjust for the selection of women who became pregnant.
The secondary analyses of early and clinically-confirmed pregnancy losses as separate outcomes showed only marginal associations with weight gain (Table S1).
DISCUSSION
Main Findings
In a preconception cohort study, weight gain ≥5% in the period spanning the 12 months before enrollment (12–18 months preconception) up to 4 weeks’ gestation was associated with a higher risk of pregnancy loss compared to maintaining a constant weight. Weight loss ≥5% and high weight variability (both weight gain ≥5% and weight loss ≥5%) were not consistently associated with risk of pregnancy loss; given the limited numbers in these categories, small associations cannot be ruled out by these data. Self-initiated attempt to lose weight was commonly reported, and this was not appreciably associated with weight loss. This is the first study, to our knowledge, to report the effects of weight change in combination with self-initiated weight-loss attempts in the recent pre- and periconceptional period on pregnancy loss among women trying to conceive spontaneously.
Strengths and Limitations
Our study has several strengths. First, we were able to analyse both self-initiated, recent weight-loss efforts and actual weight change in a population of women who were not seeking fertility treatment. This extends previous research that has primarily focused on either the attempt15 or long-term, actual weight change3 separately, or was conducted in infertility-clinic populations.7–10 Second, there was near-complete ascertainment of pregnancy loss, including early losses, for participants included in the analysis. Third, the exclusion criteria minimised confounding by major medical conditions, recent gestational weight gain, and recent postpartum weight loss. Fourth, our results from a population of women attempting pregnancy after 1–2 prior pregnancy losses are likely to be of clinical interest, as women in this situation may be interested in lifestyle and behavior modifications in order to support a healthy pregnancy. In support of the generalisability of the results, the study population was similar to US nationally representative data in terms of the observed incidence of pregnancy loss,23 the prevalence of recent weight-control behaviors,13 and the prevalence of overweight/obesity.24
Several limitations also deserve mention. Data on minimum and maximum weights and weight-loss effort in the previous 12 months were self-reported and subject to inaccuracy.25 While we lacked objective measures of recent minimum and maximum weights, several arguments support the validity of our found associations with recent weight change. First, self-report of recent, maximum weight had high validity in the National Weight Control Registry.26 Second, the weights were reported before the pregnancy outcome was objectively ascertained, and therefore the error should be independent and non-differential with respect to the outcome. Thus, any resulting exposure misclassification should bias the association towards the null, and could not explain our significant associations. Women were not instructed to exclude weight during pregnancy when they self-reported their maximum weight, and so we excluded from the analysis women who in the past year had a pregnancy lasting >14 weeks. We lacked data on the temporal order of the weight-loss attempt relative to the minimum and maximum weights over the 12 months before enrollment, which limits conclusions regarding the effects of intentional and unintentional weight change on pregnancy loss. Finally, this was a secondary analysis of data collected for the EAGeR trial, and so this research question did not determine the sample size. The limited sample size resulted in wide confidence intervals around the risk ratios in many categories of exposure and thus limited their interpretation. Notably, the risk ratios were close to the null for attempted weight loss among all women, but the wide confidence intervals show that modest increases in risk cannot be ruled out by our data.
Interpretation
Obesity is associated with an increased risk of pregnancy loss among pregnancies that are conceived spontaneously.2–4 Mechanisms by which weight gain and excess adiposity are thought to harm reproduction include oxidative stress and hormonal and inflammatory disturbances,27 which may be ameliorated by weight loss,28 improved dietary quality,29, 30 and/or physical activity.28 Prolonged, increased insulin concentrations secondary to the insulin resistance of excess adiposity and weight gain may inhibit insulin-like growth factor binding protein-1 synthesis31 and stimulate androgen synthesis.31 These endocrine changes have been shown to impair oocyte and embryonic development5 and endometrial receptivity32 in some studies, though not in others.6, 33 Also, excess intake of refined sugars and saturated fats induces inflammation and oxidative stress,34 contributing to a pro-inflammatory milieu that may impede embryonic development,35 implantation,36 and placental function.37 Although we did not have hormonal and metabolic data through the weight change period, our results support a potential negative impact of pre- and peri-conception weight gain, which involves well-known metabolic changes, on pregnancy loss.
We found an increased risk of pregnancy loss associated with weight gain in the past 12–18 months, suggesting that pregnancy loss could be prevented by avoiding weight gain prior to conception. On the other hand, substantial weight gain in a relatively short time may indicate undetected medical conditions or intense psychological stress, which are risk factors for pregnancy loss.38 We did not find an appreciable association between high weight variability and pregnancy loss, but few participants met the definition of high weight variability (weight gain ≥5% and weight loss ≥5%) over this short time. Indeed, data are scarce on the effect of recent weight cycling on pregnancy outcomes,39 and further study in relevant populations is needed.
The present null association between self-initiated attempted weight loss and pregnancy loss adds information to the gap in research on the real-world experience of self-initiated attempted weight loss on pregnancy outcomes.15 The variety of weight-loss methods reported indicates that results from trials of lifestyle modification for weight-loss may not apply to self-initiated attempts due to differences in practices and professional support. Indeed, of women who had attempted weight loss for ≥1 month, 86% weighed within 4kg of their 12 month maximum weight, and their risk of pregnancy loss was similar to that of women who made no attempt and maintained a constant weight. Given that the self-initiated attempt to lose weight was only rarely accompanied by weight loss, it may not be a feasible way to prevent pregnancy loss.
Conclusion
Among fecund women with 1–2 prior pregnancy losses, recent weight gain was prospectively associated with a higher risk of pregnancy loss. The null association between recent, self-initiated weight-loss attempt and pregnancy loss provides some evidence to reassure women who are attempting to lower their BMI in order to reduce the risk of adverse pregnancy outcomes. Patients who have experienced pregnancy loss desire evidence-based strategies to decrease their risk in subsequent pregnancies. These novel results address a significant data gap concerning the effects of recent weight change and self-initiated weight-loss attempts on pregnancy loss outside of an infertility clinic setting. Further studies are needed to confirm these findings as they suggest a potentially modifiable risk factor for pregnancy loss.
Supplementary Material
Acknowledgments
The authors gratefully acknowledge the outstanding commitments of the EAGeR trial participants and staff.
FUNDING
This study and the original EAGeR study were funded by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland (Contract Nos. HHSN267200603423, HHSN267200603424, HHSN267200603426). This funding was awarded through an external review process.
Footnotes
DISCLOSURE OF INTERESTS
The authors declared no conflict of interest exists. This work was prepared by employees of the U.S. government as part of their official duties and is not subject to copyright.
These findings were previously presented as an abstract at the 29th Annual Meeting of the Society for Pediatric and Perinatal Epidemiologic Research, Miami, FL, June 20–21, 2016; and at the 2016 Epidemiology Congress of the Americas, Miami, FL, June 21–24, 2016.
CONTRIBUTION TO AUTHORSHIP
Dr. Radin analysed and interpreted the data and drafted the manuscript. Dr. Schisterman, Dr. Sjaarda, Dr. Perkins, and Dr. Mumford designed the study, interpreted the data, reviewed the article critically, and revised it for important intellectual content. Dr. Silver, Ms. Lesher, Dr. Lynch, Dr. Wilcox, and Dr. Wactawski-Wende made substantial contributions to the acquisition of the data, reviewed the article critically, and revised it for important intellectual content. Dr. Plowden, Dr. Hinkle, and Dr. Kim interpreted the data, reviewed the article critically, and revised it for important intellectual content.
DETAILS OF ETHICS APPROVAL
- Institutional Review Board, The University of Utah: IRB #1002521 5/24/2007
- The Wright Center for Graduate Medical Education Institutional Review Board (Scranton, Pennsylvania): IRB #HHSN275200403394 May 2007
- University at Buffalo Health Sciences Institutional Review Board: IRB #SPM0900107A 12/20/2007
- Colorado Multiple Institutional Review Board (University of Colorado, Anschutz Medical Campus): IRB #08-0982 April 2009
References
- 1.Wang JX, Davies MJ, Norman RJ. Obesity increases the risk of spontaneous abortion during infertility treatment. Obes Res. 2002;10:551–4. doi: 10.1038/oby.2002.74. [DOI] [PubMed] [Google Scholar]
- 2.Metwally M, Ong KJ, Ledger WL, Li TC. Does high body mass index increase the risk of miscarriage after spontaneous and assisted conception? A meta-analysis of the evidence. Fertil Steril. 2008;90:714–26. doi: 10.1016/j.fertnstert.2007.07.1290. [DOI] [PubMed] [Google Scholar]
- 3.Gaskins AJ, Rich-Edwards JW, Colaci DS, Afeiche MC, Toth TL, Gillman MW, et al. Prepregnancy and early adulthood body mass index and adult weight change in relation to fetal loss. Obstet Gynecol. 2014;124:662–9. doi: 10.1097/AOG.0000000000000478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lashen H, Fear K, Sturdee DW. Obesity is associated with increased risk of first trimester and recurrent miscarriage: matched case–control study. Hum Reprod. 2004;19:1644–6. doi: 10.1093/humrep/deh277. [DOI] [PubMed] [Google Scholar]
- 5.Dokras A, Baredziak L, Blaine J, Syrop C, VanVoorhis BJ, Sparks A. Obstetric outcomes after in vitro fertilization in obese and morbidly obese women. Obstet Gynecol. 2006;108:61–9. doi: 10.1097/01.AOG.0000219768.08249.b6. [DOI] [PubMed] [Google Scholar]
- 6.Bellver J, Ayllón Y, Ferrando M, Melo M, Goyri E, Pellicer A, et al. Female obesity impairs in vitro fertilization outcome without affecting embryo quality. Fertil Steril. 2010;93:447–54. doi: 10.1016/j.fertnstert.2008.12.032. [DOI] [PubMed] [Google Scholar]
- 7.Mutsaerts MA, van Oers AM, Groen H, Burggraaff JM, Kuchenbecker WK, Perquin DA, et al. Randomized trial of a lifestyle program in obese infertile women. N Engl J Med. 2016;374:1942–53. doi: 10.1056/NEJMoa1505297. [DOI] [PubMed] [Google Scholar]
- 8.Clark AM, Thornley B, Tomlinson L, Galletley C, Norman RJ. Weight loss in obese infertile women results in improvement in reproductive outcome for all forms of fertility treatment. Hum Reprod. 1998;13:1502–5. doi: 10.1093/humrep/13.6.1502. [DOI] [PubMed] [Google Scholar]
- 9.Moran L, Tsagareli V, Norman R, Noakes M. Diet and IVF pilot study: short-term weight loss improves pregnancy rates in overweight/obese women undertaking IVF. Aust N Z J Obstet Gynaecol. 2011;51:455–9. doi: 10.1111/j.1479-828X.2011.01343.x. [DOI] [PubMed] [Google Scholar]
- 10.Sim KA, Dezarnaulds GM, Denyer GS, Skilton MR, Caterson ID. Weight loss improves reproductive outcomes in obese women undergoing fertility treatment: a randomized controlled trial. Clin Obes. 2014;4:61–8. doi: 10.1111/cob.12048. [DOI] [PubMed] [Google Scholar]
- 11.Tsagareli V, Noakes M, Norman RJ. Effect of a very-low-calorie diet on in vitro fertilization outcomes. Fertil Steril. 2006;86:227–9. doi: 10.1016/j.fertnstert.2005.12.041. [DOI] [PubMed] [Google Scholar]
- 12.Villamor E, Cnattingius S. Interpregnancy weight change and risk of adverse pregnancy outcomes: a population-based study. Lancet. 2006;368:1164–70. doi: 10.1016/S0140-6736(06)69473-7. [DOI] [PubMed] [Google Scholar]
- 13.Gallup. To lose weight, Americans rely more on dieting than exercise [Internet] Washington, DC: Gallup; 2011. [cited 2016 February 9]. Available from: http://www.gallup.com/poll/150986/lose-weight-americans-rely-dieting-exercise.aspx. [Google Scholar]
- 14.Bartlett SJ, Faith MS, Fontaine KR, Cheskin LJ, Allison DB. Is the prevalence of successful weight loss and maintenance higher in the general community than the research clinic? Obes Res. 1999;7:407–13. doi: 10.1002/j.1550-8528.1999.tb00425.x. [DOI] [PubMed] [Google Scholar]
- 15.Berenson AB, Pohlmeier AM, Laz TH, Rahman M, McGrath CJ. Nutritional and weight management behaviors in low-income women trying to conceive. Obstet Gynecol. 2014;124:579–84. doi: 10.1097/AOG.0000000000000416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Schisterman EF, Silver RM, Perkins NJ, Mumford SL, Whitcomb BW, Stanford JB, et al. A randomised trial to evaluate the effects of low-dose aspirin in gestation and reproduction: design and baseline characteristics. Paediatr Perinat Epidemiol. 2013;27:598–609. doi: 10.1111/ppe.12088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mumford SL, Silver RM, Sjaarda LA, Wactawski-Wende J, Townsend JM, Lynch AM, et al. Expanded findings from a randomized controlled trial of preconception low-dose aspirin and pregnancy loss. Hum Reprod. 2016;31:657–65. doi: 10.1093/humrep/dev329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Schisterman EF, Silver RM, Lesher LL, Faraggi D, Wactawski-Wende J, Townsend JM, et al. Preconception low-dose aspirin and pregnancy outcomes: results from the EAGeR randomised trial. Lancet. 2014;384:29–36. doi: 10.1016/S0140-6736(14)60157-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kiddy DS, Hamilton-Fairley D, Bush A, Short F, Anyaoku V, Reed MJ, et al. Improvement in endocrine and ovarian function during dietary treatment of obese women with polycystic ovary syndrome. Clin Endocrinol. 1992;36:105–11. doi: 10.1111/j.1365-2265.1992.tb02909.x. [DOI] [PubMed] [Google Scholar]
- 20.Hernan MA, Hernandez-Diaz S, Robins JM. A structural approach to selection bias. Epidemiology. 2004;15:615–25. doi: 10.1097/01.ede.0000135174.63482.43. [DOI] [PubMed] [Google Scholar]
- 21.Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology. 2009;20:488–95. doi: 10.1097/EDE.0b013e3181a819a1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ)—Short and Long Forms [Internet] Stockholm, Sweden: Karolinska Institute; 2005. [cited 2017 July 21]. Available from: https://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnx0aGVpcGFxfGd4OjE0NDgxMDk3NDU1YWRlZTM. [Google Scholar]
- 23.Ventura SJ, Curtin SC, Abma JC, Henshaw SK. Estimated pregnancy rates and rates of pregnancy outcomes for the United States, 1990–2008. Natl Vital Stat Rep. 2012;60:1–21. [PubMed] [Google Scholar]
- 24.Branum AM, Kirmeyer SE, Gregory EC. Prepregnancy body mass index by maternal characteristics and state: Data from the birth certificate, 2014. Natl Vital Stat Rep. 2016;65:1–11. [PubMed] [Google Scholar]
- 25.Jain RB. Regression models to predict corrected weight, height and obesity prevalence from self-reported data: data from BRFSS 1999–2007. Int J Obes (Lond) 2010;34:1655–64. doi: 10.1038/ijo.2010.80. [DOI] [PubMed] [Google Scholar]
- 26.McGuire MT, Wing RR, Klem ML, Lang W, Hill JO. What predicts weight regain in a group of successful weight losers? J Consult Clin Psychol. 1999;67:177–85. doi: 10.1037//0022-006x.67.2.177. [DOI] [PubMed] [Google Scholar]
- 27.Brewer CJ, Balen AH. The adverse effects of obesity on conception and implantation. Reproduction. 2010;140:347–64. doi: 10.1530/REP-09-0568. [DOI] [PubMed] [Google Scholar]
- 28.Rector RS, Warner SO, Liu Y, Hinton PS, Sun GY, Cox RH, et al. Exercise and diet induced weight loss improves measures of oxidative stress and insulin sensitivity in adults with characteristics of the metabolic syndrome. Am J Physiol - Endocrinol Metab. 2007;293:E500–E6. doi: 10.1152/ajpendo.00116.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Dai J, Jones DP, Goldberg J, Ziegler TR, Bostick RM, Wilson PW, et al. Association between adherence to the Mediterranean diet and oxidative stress. Am J Clin Nutr. 2008;88:1364–70. doi: 10.3945/ajcn.2008.26528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Willcox DC, Willcox BJ, Todoriki H, Suzuki M. The Okinawan diet: health implications of a low-calorie, nutrient-dense, antioxidant-rich dietary pattern low in glycemic load. J Am Coll Nutr. 2009;28(Suppl):500S–16S. doi: 10.1080/07315724.2009.10718117. [DOI] [PubMed] [Google Scholar]
- 31.Poretsky L, Cataldo NA, Rosenwaks Z, Giudice LC. The insulin-related ovarian regulatory system in health and disease. Endocrine Rev. 1999;20:535–82. doi: 10.1210/edrv.20.4.0374. [DOI] [PubMed] [Google Scholar]
- 32.Carrington B, Sacks G, Regan L. Recurrent miscarriage: pathophysiology and outcome. Curr Opin Obstet Gynecol. 2005;17:591–7. doi: 10.1097/01.gco.0000194112.86051.26. [DOI] [PubMed] [Google Scholar]
- 33.Styne-Gross A, Elkind-Hirsch K, Scott RT. Obesity does not impact implantation rates or pregnancy outcome in women attempting conception through oocyte donation. Fertil Steril. 2005;83:1629–34. doi: 10.1016/j.fertnstert.2005.01.099. [DOI] [PubMed] [Google Scholar]
- 34.Roberts CK, Barnard RJ, Sindhu RK, Jurczak M, Ehdaie A, Vaziri ND. Oxidative stress and dysregulation of NAD(P)H oxidase and antioxidant enzymes in diet-induced metabolic syndrome. Metabolism. 2006;55:928–34. doi: 10.1016/j.metabol.2006.02.022. [DOI] [PubMed] [Google Scholar]
- 35.Gosman GG, Katcher HI, Legro RS. Obesity and the role of gut and adipose hormones in female reproduction. Hum Reprod Update. 2006;12:585–601. doi: 10.1093/humupd/dml024. [DOI] [PubMed] [Google Scholar]
- 36.Comba C, Bastu E, Dural O, Yasa C, Keskin G, Ozsurmeli M, et al. Role of inflammatory mediators in patients with recurrent pregnancy loss. Fertil Steril. 2015;104:1467–74. e1. doi: 10.1016/j.fertnstert.2015.08.011. [DOI] [PubMed] [Google Scholar]
- 37.Jauniaux E, Watson AL, Hempstock J, Bao Y-P, Skepper JN, Burton GJ. Onset of maternal arterial blood flow and placental oxidative stress: A possible factor in human early pregnancy failure. Am J Pathol. 2000;157:2111–22. doi: 10.1016/S0002-9440(10)64849-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Nepomnaschy PA, Welch KB, McConnell DS, Low BS, Strassmann BI, England BG. Cortisol levels and very early pregnancy loss in humans. Proc Nat Acad Sci USA. 2006;103:3938–42. doi: 10.1073/pnas.0511183103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Frederick IO, Rudra CB, Miller RS, Foster JC, Williams MA. Adult weight change, weight cycling, and prepregnancy obesity in relation to risk of preeclampsia. Epidemiology. 2006;17:428–34. doi: 10.1097/01.ede.0000221028.33245.0b. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.