The effects of current body weight on fecundity are well-recognized among women attempting pregnancy spontaneously and through infertility treatment.1 This has led to weight loss being strongly promoted as an effective means of increasing fertility in overweight or obese women.2 However, most of the evidence for weight loss as a treatment for subfecundity comes from small intervention studies in obese women before fertility treatment and observational studies among women undergoing bariatric surgery.3 Only two studies have evaluated the effect of weight change not as a result of surgery or clinical interventions on fecundity. Among overweight or obese women in the Danish National Birth Cohort, every 1-kg decrement in weight between pregnancies was associated with 5.50 (95% confidence interval [CI] 1.35–9.65) days shorter time to pregnancy.4 Among all women in the Danish National Birth Cohort, each 1-kg increment in weight was associated with 2.84 (95% CI 1.33–4.35) days longer time to pregnancy. A study of Danish women planning pregnancy (n=1,651) showed that women who lost more than 5 kg since age 17 years had slightly higher fecundability (fecundability ratio 1.05, 95% CI 0.73–1.52) and woman who gained 15 kg or more since age 17 years had lower fecundability (fecundability ratio 0.72, 95% CI 0.59–0.88).5
Two studies found an inverse association between adolescent body weight and number of children6,7 and another found a positive association between adolescent body weight and risk of ovulatory infertility.8 Taken together, these studies suggest that body weight in adolescence might also be an important risk factor for adult fertility.
The objective of this study was to examine the relationship between weight change since age 18 years, current body mass index (BMI), and BMI at age 18 years and fecundity among a large cohort of U.S. women.
MATERIALS AND METHODS
The Nurses' Health Study 3 is an ongoing Internet-based cohort study of female nurses in the United States and Canada. To be eligible for the study, women had to be a registered nurse, licensed practical or vocational nurse, or nursing student and born on or after January 1, 1965. As of September 2014, 38,016 women had joined the study and 26,798 women had completed at least one follow-up questionnaire, forming the base population for our analysis. Every 6 months, questionnaires are sent to participants to update lifestyle and medical characteristics. The response rate for the second questionnaire is currently 74%; for women who have completed at least two questionnaires, subsequent response rates exceed 80%. An overview of the study flow is illustrated in Figure 1. Briefly, women were eligible for this current analysis if they reported their height and weight on the baseline questionnaire and reported on any of the subsequent questionnaires that they were trying to get pregnant (n=1,970). We excluded women who reported that they were postmenopausal (n=16) or for whom we were missing information on duration of ongoing pregnancy attempt (n=4). After these exclusions, 1,950 women were available for analysis. The study was approved by the institutional review board of the Brigham and Women's Hospital (Boston, Massachusetts). Completion of the web-based questionnaires implied informed consent.
Current height and weight as well as weight at age 18 years were reported on the baseline questionnaire. Current BMI was calculated as weight (kg) divided by height squared (m2), and BMI at age 18 years was calculated as weight at age 18 years (kg) divided by height squared (m2). Weight change since age 18 years was calculated as the difference between current weight and weight at age 18 years. A previous validation study done in a similar cohort showed that self-reported height and weight were highly correlated with measured height (r=0.94) and weight (r=0.97).9 Self-reported weight at age 18 years was also highly correlated with weight recorded in medical records (r=0.84).10
Women who report that they are actively trying to get pregnant are asked to report the current duration of their ongoing pregnancy attempt. Specifically, they are asked: “For how many months have you been actively trying to get pregnant?” Categories for response include: 1 month or less, 2 months, 3 months, 4 months, 5 months, 6 months, 7 months, 8 months, 9 months, 10 months, 11 months, 12 months, 1–2 years, and 3 years or longer. We took a woman's first report of ongoing pregnancy attempt after the baseline questionnaire as her outcome. Validity of self-report of duration of pregnancy attempt has not been assessed in this population. However, there is extensive literature documenting the reproducibility and validity of this outcome.11
Information on potential confounding variables is assessed on the baseline questionnaire including age, race–ethnicity, lifetime pregnancy history, smoking history, marital status, and history of endometriosis and bariatric surgery. Menstrual cycle characteristics including current regularity and length of a usual menstrual cycle are reported on the first follow-up questionnaire. Information on physician-diagnosed polycystic ovary syndrome (PCOS), body hair, use of permanent and temporary hair removal solutions, and acne is also assessed on the second questionnaire. We assessed hirsutism with the simplified version of the Ferriman-Gallwey score, which uses a scale of 0–3 to assess the amount of hair on a woman's chin, upper abdomen, and lower abdomen. Women with a summed score 3 or higher were considered has having hirsutism because a previous study determined this cutoff be highly predictive.12
Women were divided into groups according to the World Health Organization (WHO) BMI classifications for current and age 18 years BMI: less than 18.5 (underweight), 18.5–24.9 (normal weight), 25–29.9 (overweight), and 30 or higher (obese). Weight change since age 18 years was classified into categories based on its distribution in this cohort: lost 4 kg or more, stable weight (±4 kg), gained 4–9.9 kg, gained 10–19.9 kg, and gained 20 kg or more. These variables were also analyzed as continuous variables. Differences in baseline characteristics by weight change since age 18 years categories were compared using a χ2 test for categorical variables and Kruskal-Wallis nonparametric tests for continuous variables.
For our analysis, we applied a current duration approach that uses information collected in a cross-sectional fashion on current duration of ongoing pregnancy attempt to make inferences about actually realized waiting times to pregnancy.13 Numerous studies have utilized this approach to estimate the national prevalence of infertility14 and the association of environmental factors on fecundity.15 Because couples who have long durations of attempting pregnancy are overrepresented in the current duration approach, appropriate statistical models are used to account for this length-biased sampling. The current duration approach and more generally backward recurrence time survival methods allow us to infer the relationship of characteristics to the (unobserved) total duration of pregnancy attempt by using the (observed) current duration of attempt through accelerated failure time models.13 Based on previous research,15 we chose an accelerated failure time model with log normal distribution to estimate the time ratios and 95% CIs. Other outcome distributions such as generalized gamma and Pareto were also explored. Because the log-normal distribution had the lowest −2 log likelihood and Akaike information criterion value, this was chosen as the outcome distribution. The log-normal accelerated failure time model is equivalent to linear regression on the log-transformed current durations. PROC GENMOD in SAS was used to model the effect of various exposures on current duration of pregnancy attempt. The time ratios correspond to exp([beta]) from the log-normal accelerated failure time models. These time ratios can be interpreted as the ratios of the median values of the duration of pregnancy attempts between the compared groups. A time ratio above 1 implies that a given exposure is associated with lower fecundity (eg, prolongs the time to pregnancy), whereas a time ratio below 1 indicates a higher fecundity (eg, shorter time to pregnancy). Tests for linear trend across categories were conducted by using the median values in each category as a continuous variable.
Confounding was evaluated using prior knowledge about suspected confounders and descriptive statistics from our cohort. Variables retained in the final multivariable models were age, race, smoking status, and marital status. Missing covariate data were rare: two women were missing data on smoking status and this was the only variable with missing data. To accommodate these missing responses, a categorical indicator was used. Weight change since age 18 years models were further adjusted for BMI at age 18 years. Menstrual cycle characteristics, diagnosis of PCOS, indicators of hyperandrogenism (such as acne and hirsutism), and current BMI (for BMI at age 18 years models) were evaluated as potential mediators. Additional analyses were also performed to evaluate effect modification by age and gravidity. Separate analyses were carried out for weight change since age 18 years in women with a low BMI (less than 25) or high BMI (25 or higher) at age 18 years. Women were also cross-classified into categories based on WHO categories of BMI at age 18 years and current BMI to address the effect of weight on fecundity depending on where women started in adolescence and ended up in adulthood. P values for heterogeneity were derived from the cross-product interaction term coefficient added to the main-effects multivariable model. SAS 9.3 was used for all analyses. A significance level of P<.05 was used for all analyses.
To assess the robustness of our results, we conducted a more typical time to pregnancy analysis in women in the Nurses' Health Study 3 who enrolled in the Maternal Health Study. Details of this analysis can be found in the supplemental Methods (Appendix 1, available online at http://links.lww.com/AOG/A683).
On average, participants who gained the most weight since age 18 years were older and heavier, more likely to be current smokers, and less likely to be white and nulligravid (Table 1). Women in the highest category of weight gain were also more likely to have been diagnosed with PCOS, have irregular menstrual cycles, higher amounts of body hair, and greater use of temporary hair removals. No women in our cohort reported a history of bariatric surgery. The correlation between BMI at age 18 years and current BMI was 0.68. The estimated proportions of women not pregnant after 12 and 24 months were 16% and 5%, respectively.
Every 5-kg increase in body weight from age 18 years was associated with a 5% (95% CI 3–7%) increase in median current duration of pregnancy attempt. Compared with women who maintained their weight since age 18 years, the multivariable-adjusted time ratios (95% CIs) were 0.85 (0.68–1.06) for women who lost 4 kg or more, 1.08 (0.95–1.22) for women who gained 4–9.9 kg, 1.09 (0.94–1.25) for women who gained 10–19.9 kg, and 1.40 (1.19–1.63) for women who gained 20 kg or more (P trend ≤.001) (Fig. 2). Translated onto an absolute level, compared with women who maintained weight, the adjusted median current duration was 0.5 months shorter in those who lost weight (P difference .14), 0.3 months longer for those who gained 4–9.9 kg (P difference .26), and 10–19.9 kg (P difference .25), and 1.4 months longer for those who gained 20 kg or more (P difference ≤.001). Current BMI was also significantly associated with fecundity (P trend ≤.001). The adjusted effects of being overweight and obese corresponded to 13% (95% CI 0–27%) and 28% (95% CI 13–45%) increases in the median duration of pregnancy attempt compared with women of normal BMI. The effect of overweight and obesity translated into an average of 0.5 and 1.0 month longer current duration of pregnancy attempt compared with women of normal BMI. When analyzed on a continuous level, the adjusted time ratio for a 5-kg/m2 increase in current BMI was 1.08 (95% CI 1.04–1.12). Body mass index at age 18 years appeared to have a U-shaped relationship with fecundity in multivariable adjustment because both underweight (time ratio 1.17, 95% CI 1.00–1.37) and obese women (time ratio 1.22, 95% CI 1.00–1.49) at age 18 years had a longer current duration of pregnancy attempts compared with normal-weight women. We further adjusted these models for current BMI to evaluate whether the relationship between BMI at age 18 years and later fecundity was solely the result of the fact that a woman's BMI tracked into adulthood. After further adjustment for current BMI, only underweight women at age 18 years had a longer current duration of pregnancy attempt compared with normal-weight women at age 18 years (time ratio 1.25, 95% CI 1.07–1.47).
The association between weight change and fecundity was similar in women whose BMIs were less than 25 and higher than 25 at age 18 years (P interaction .33), although this analysis was limited by the low number women who were overweight or obese at age 18 years (n=370). When we cross-classified women into categories based on WHO categories of BMI at age 18 years and current BMI, it became apparent that weight gain was particularly detrimental in underweight women at age 18 years (Fig. 3). Moreover, weight gain was detrimental even if women did not end up being classified as overweight or obese in adulthood (time ratio for being underweight at age 18 years and normal weight in adulthood 1.20, 95% CI 1.00–1.43). There was no evidence of effect modification of the weight change, current BMI, or BMI at age 18 years and fecundity association by age (younger than 37, 37 years or older), smoking status (ever compared with never smoker), or gravidity (nulligravid compared with gravid) (Appendix 2, available online at http://links.lww.com/AOG/A683). Associations between weight change and current BMI and fecundity were slightly attenuated, albeit still significant after adjustment for regular menstrual cycles, diagnosis of PCOS, and hirsutism.
The results of the sensitivity analysis in pregnant women who retrospectively reported their time to pregnancy were consistent with the main analysis with the exception of BMI at age 18 years, which was not associated with fecundability in this cohort (Table 2). Every 5-kg increase in body weight since age 18 years was associated with a 3% (95% CI 1–6%) decrease in fecundability. A significant linear trend was also observed across categories of weight change since age 18 years and current BMI. Overall, compared with the current duration results, effect estimates were slightly weaker for the weight change and current BMI analyses in this cohort, although both were still statistically significant. Of note, women in the retrospective pregnancy cohort were significantly younger, thinner, more likely to be never smokers, married, white, and less likely to be nulligravid, diagnosed with PCOS, and have regular menstrual cycles (P<.05) compared with women in the current duration cohort (Appendix 3, available online at http://links.lww.com/AOG/A683).
In this ongoing cohort of women planning pregnancy, weight gain of 20 kg or more since age 18 years, current BMI of 25 or higher, and BMI less than 18.5 at age 18 years were associated with lower fecundity, although all associations were modest in effect size. For instance, compared with women who had maintained their weight since age 18 years, women who gained 20 kg or more experienced, on average, a 1.4-month longer current duration of pregnancy attempt. The association between weight gain and fecundity was similar in women whose BMIs were less than 25 and 25 or higher at age 18 years and appeared to be independent of current BMI.
Our results for adult weight change and fecundity agree with previous studies that have also shown slightly shorter waiting time to pregnancy in women who lost weight since age 17 years5 or between consecutive pregnancies4 and longer waiting time to pregnancy in women who gained weight in adulthood5 or between pregnancies.4 They are also in agreement with the smaller clinical studies of weight loss and improved ovulatory function among women with PCOS and women undergoing surgery for weight loss.3 We did not find any difference in effect among younger or older women, suggesting that weight loss is beneficial and weight gain is detrimental to fecundity, regardless of timing. One unexpected finding, however, was that gaining weight was particularly detrimental in women who were underweight in adolescence. This suggested to us a critical window of exposure where being underweight at age 18 years significantly affects fecundity to an extent that was exacerbated by later weight gain.
Our results were similar after adjusting for current menstrual cycle characteristics, ever diagnosis of PCOS, and other markers of hyperandrogenism such as body hair patterns, suggesting that mechanisms other than clinical or subclinical PCOS or anovulation are driving this association. Weight loss before in vitro fertilization has been linked to better quality oocytes16 and many studies have shown that weight loss improves insulin sensitivity.17 Because both of these pathways have been linked to higher fecundity, one or both of these mechanisms could be driving this association.18,19 In regard to our findings on weight gain, because our results remained significant after adjusting for current BMI, this suggests that effect of weight gain on fecundity is not completely mediated through current body fatness. An animal study showed that adult female mice reared on chronic caloric restriction who were allowed to eat ad libitum for 1 month (resulting in weight gain) produced fewer good quality oocytes and had a lower number of embryos reaching the blastocyst stage than female mice that were kept on caloric restriction.20 Thus, impaired oocyte quality could be mediating this relationship as well.
The finding of reduced fecundity in overweight and obese women has been observed in many other studies.4,5,21–23 Our study expanded on previous research by directly quantifying the magnitude of this effect on an interpretable scale and suggests that effects of overweight and obesity on fertility are small (eg, on the order of 0.5–1 month delays). Similar to the results on weight change, the effects of current BMI on fecundity did not appear to be completely mediated through menstrual cycle disturbances, PCOS, or markers of hyperandrogenism, which has been shown previously.5 Instead, a growing body of literature suggests that obesity may also exert its negative consequences on oocyte quality24 and thus lengthen time to pregnancy. For example, a recent meta-analysis found that among egg donor recipients, the chance of pregnancy after in vitro fertilization is no different in obese recipients compared with normal-weight recipients.25
The weaker association with BMI that we observed in the retrospective time-to-pregnancy cohort was expected given that women who are involuntary infertile and women experiencing pregnancy losses before 20 weeks of gestation were excluded by design. Because both of these conditions have been linked to obesity,26 this would support the stronger effects we observed in the current duration cohort. Moreover, as a result of the design of his cohort, women had to retrospectively report their time to pregnancy. Although a retrospective report of time to pregnancy is considered reasonably valid, it has been documented that women who experienced longer observed time to pregnancy are more likely to underreport their time to pregnancy later.27 The predicted effect of this underestimation is also bias toward the null.28
Limitations of this study include the possibility of planning bias, residual confounding, and the use of only one exposure assessment, whereas the strengths include the ability to include women with both high fertility and those who are involuntarily infertile, to address menstrual cycle characteristics and subclinical markers of PCOS as possible mechanisms of effect, and to compare our results with a typical retrospective time to pregnancy analysis. These strengths and limitations have been described in more detail in the supplemental Discussion section (Appendix 4, available online at http://links.lww.com/AOG/A683).
The main clinical implications of this study concern the absolute magnitudes of effect for the associations among current BMI, weight change since age 18 years, and underweight at age 18 years. Although previous studies have consistently documented lower fecundity among women who are overweight and obese, most were unable to quantify the exact magnitude of effect on an interpretable scale (eg, as a result of nonparametric modeling) and in a representative population (eg, not just women who become pregnant). As we have shown here, the effect of current overweight and obesity on duration of pregnancy attempt is on the order of 0.5 and 1.0 months, respectively. The effects of weight gain, even at extreme levels of 20 kg or more since adolescence, on current duration of pregnancy attempt are also modest, 1.4 months. This information can provide valuable guidance to women planning pregnancy who wish to understand the absolute magnitude of body weight's effect on fecundity.
1. Jungheim ES, Travieso JL, Hopeman MM. Weighing the impact of obesity on female reproductive function and fertility. Nutr Rev 2013;71(suppl 1):S3–8.
2. Nelson SM, Fleming R. Obesity and reproduction: impact and interventions. Curr Opin Obstet Gynecol 2007;19:384–9.
3. Sim KA, Partridge SR, Sainsbury A. Does weight loss in overweight or obese women improve fertility treatment outcomes? A systematic review. Obes Rev 2014;15:839–50.
4. Ramlau-Hansen CH, Thulstrup AM, Nohr EA, Bonde JP, Sorensen TI, Olsen J. Subfecundity in overweight and obese couples. Hum Reprod 2007;22:1634–7.
5. Wise LA, Rothman KJ, Mikkelsen EM, Sørensen HT, Riis A, Hatch EE. An internet-based prospective study of body size and time-to-pregnancy. Hum Reprod 2010;25:253–64.
6. Jokela M, Elovainio M, Kivimäki M. Lower fertility associated with obesity and underweight: the US National Longitudinal Survey of Youth. Am J Clin Nutr 2008;88:886–93.
7. Jokela M, Kivimäki M, Elovainio M, Viikari J, Raitakari OT, Keltikangas-Järvinen L. Body mass index in adolescence and number of children in adulthood. Epidemiology 2007;18:599–606.
8. Rich-Edwards JW, Goldman MB, Willett WC, Hunter DJ, Stampfer MJ, Colditz GA, et al.. Adolescent body mass index and infertility caused by ovulatory disorder. Am J Obstet Gynecol 1994;171:171–7.
9. Rimm EB, Stampfer MJ, Colditz GA, Chute CG, Litin LB, Willett WC. Validity of self-reported waist and hip circumferences in men and women. Epidemiology 1990;1:466–73.
10. Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC. The validity of recalled weight among younger women. Int J Obes Relat Metab Disord 1995;19:570–2.
11. Joffe M, Villard L, Li Z, Plowman R, Vessey M. Long-term recall of time-to-pregnancy. Fertil Steril 1993;60:99–104.
12. Cook H, Brennan K, Azziz R. Reanalyzing the modified Ferriman-Gallwey score: is there a simpler method for assessing the extent of hirsutism? Fertil Steril 2011;96:1266–70.e1.
13. Keiding N, Kvist K, Hartvig H, Tvede M, Juul S. Estimating time to pregnancy from current durations in a cross-sectional sample. Biostatistics 2002;3:565–78.
14. Thoma ME, McLain AC, Louis JF, King RB, Trumble AC, Sundaram R, et al.. Prevalence of infertility in the United States as estimated by the current duration approach and a traditional constructed approach. Fertil Steril 2013;99:1324–1331.e1.
15. Slama R, Ducot B, Carstensen L, Lorente C, de La Rochebrochard E, Leridon H, et al.. Feasibility of the current-duration approach to studying human fecundity. Epidemiology 2006;17:440–9.
16. Chavarro JE, Ehrlich S, Colaci DS, Wright DL, Toth TL, Petrozza JC, et al.. Body mass index and short-term weight change in relation to treatment outcomes in women undergoing assisted reproduction. Fertil Steril 2012;98:109–16.
17. Holte J, Bergh T, Berne C, Wide L, Lithell H. Restored insulin sensitivity but persistently increased early insulin secretion after weight loss in obese women with polycystic ovary syndrome. J Clin Endocrinol Metab 1995;80:2586–93.
18. Terriou P, Sapin C, Giorgetti C, Hans E, Spach JL, Roulier R. Embryo score is a better predictor of pregnancy than the number of transferred embryos or female age. Fertil Steril 2001;75:525–31.
19. Wang TH, Chang CL, Wu HM, Chiu YM, Chen CK, Wang HS. Insulin-like growth factor-II (IGF-II), IGF-binding protein-3 (IGFBP-3), and IGFBP-4 in follicular fluid are associated with oocyte maturation and embryo development. Fertil Steril 2006;86:1392–401.
20. Selesniemi K, Lee HJ, Muhlhauser A, Tilly JL. Prevention of maternal aging-associated oocyte aneuploidy and meiotic spindle defects in mice by dietary and genetic strategies. Proc Natl Acad Sci U S A 2011;108:12319–24.
21. Nohr EA, Vaeth M, Rasmussen S, Ramlau-Hansen CH, Olsen J. Waiting time to pregnancy according to maternal birthweight and prepregnancy BMI. Hum Reprod 2009;24:226–32.
22. Gesink Law DC, Maclehose RF, Longnecker MP. Obesity and time to pregnancy. Hum Reprod 2007;22:414–20.
23. Bolumar F, Olsen J, Rebagliato M, Saez-Lloret I, Bisanti L. Body mass index and delayed conception: a European Multicenter Study on Infertility and Subfecundity. Am J Epidemiol 2000;151:1072–9.
24. Purcell SH, Moley KH. The impact of obesity on egg quality. J Assist Reprod Genet 2011;28:517–24.
25. Jungheim ES, Schon SB, Schulte MB, DeUgarte DA, Fowler SA, Tuuli MG. IVF outcomes in obese donor oocyte recipients: a systematic review and meta-analysis. Hum Reprod 2013;28:2720–7.
26. 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.
27. Cooney MA, Buck Louis GM, Sundaram R, McGuiness BM, Lynch CD. Validity of self-reported time to pregnancy. Epidemiology 2009;20:56–9.
28. Baird DD, Weinberg CR, Rowland AS. Reporting errors in time-to-pregnancy data collected with a short questionnaire. Impact on power and estimation of fecundability ratios. Am J Epidemiol 1991;133:1282–90.
29. Brunner Huber LR, Stanley WA, Broadhurst L, Dmochowski J, Vick TM, Scholes D. No association between body size and frequency of sexual intercourse among oral contraceptive users. Ann Epidemiol 2014;24:655–9.
30. Abrevaya J, Tang H. Body mass index in families: spousal correlation, endogeneity, and intergenerational transmission. Empir Econ 2010;41:841–64.
31. Sermondade N, Faure C, Fezeu L, Shayeb AG, Bonde JP, Jensen TK, et al.. BMI in relation to sperm count: an updated systematic review and collaborative meta-analysis. Hum Reprod Update 2013;19:221–31.