Obesity is associated with elevated levels of morbidity1 and mortality,2 and is considered to be one of the most important threats to public health.3 Obesity increases the risk of sexual and reproductive dysfunction in women4–6 and men,7–9 and obese people are less likely to get married than people of normal weight.10 Very low body mass index (BMI) has also been associated with dysfunctions of reproductive physiology in both men and women.4,8
Despite the global increase in the prevalence of obesity,3 very little is known about the potential influence of body weight on fertility (ie, the number of children) in the general population. In a sample of parents of college students, Ellis and Haman11 found that higher BMI was associated with greater number of children, particularly in women. However, this study was not prospective and, therefore, the direction of causality between weight and number of children could not be determined definitely. The number of births (ie, parity) is associated with increased risk of obesity in women,12 and parenthood may also increase the risk of obesity in men.13 Moreover, the sample of Ellis and Haman11 included only individuals with one or more children, and potential confounding factors, such as age or level of education, were not controlled for.
Prospective epidemiologic studies with representative samples are needed to evaluate the influence of body weight on fertility in the general population. In the present study we examined whether adolescent BMI predicted the number of children in adulthood 21 years later in a population-based sample of Finnish women and men participating in a prospective study.14,15 We hypothesized that there is an inverted U-shaped association between BMI and number of children, because both low and high BMI have been associated with risks for reproduction. We also assessed whether the influence of adolescent BMI was mediated by adulthood body weight or the likelihood of attaining a mate. In women, high adolescent BMI has been shown to be related to younger menarche16 which, in turn, may predict earlier initiation of sexual behavior.17 We therefore examined whether age at menarche was involved in the association between BMI and fertility. Skinfold thickness was used as an additional indicator of body fatness.
The participants were 1298 women (n = 715) and men (n = 583) participating in the on-going population-based study of Cardiovascular Risk in Young Finns.14,15 In this study, a randomly selected sample of 3596 Finnish healthy children and adolescents from 6 birth cohorts (age 3, 6, 9, 12, 15, and 18 years at baseline) have been followed since 1980, focusing on the development of cardiovascular risk factors. Complete details of the study are provided elsewhere.14,15 For the present study, we included 3 of the oldest cohorts, who were adolescents (ie, 12, 15 and 18 years of age) at baseline and thus were adults (ie, 33, 36 and 39 years of age) in the most recent follow-up phase in the year 2001. Of 1790 study participants in these 3 cohorts at baseline, 1298 had data on the present study variables (Table 1). The participants gave written informed consent, and the study was approved by local ethics committees. There was no linear or nonlinear association between adolescent BMI and the likelihood of having participated in the follow-up, indicating that adolescent BMI was not related to study attrition.
Measures in Adolescence
Adolescent height and weight were measured at the baseline when the participants were 12, 15, and 18 years of age. Measurements were taken in a medical examination with a Seca weight scale and anthropometer. BMI was calculated as weight in kilograms/(height in meters)2.
Measures of skinfold thickness were obtained at the baseline examination by Harpenden calipers (Holtain and Bull-British Indicators instruments) to 0.2-mm readings. The combined thickness of 3 skinfold measurements (subscapular, triceps, and biceps) was used in the analysis. The partial correlation (controlling for age and sex) between BMI and skinfold thickness was r = 0.69.
The women participants reported the age at menarche at the baseline and in 2 subsequent follow-up phases 3 and 6 years later. For those who had reported the age at menarche in more than one follow-up, data from the earliest possible follow-up phase were used.
Measures in Adulthood
Body mass index was calculated as above based on adulthood height and weight as measured in a medical examination in the follow-up phase in 2001. Waist circumference was measured midway between iliac crest and lowest rib as the average of 2 measurements with an accuracy of 0.1 cm. The partial correlation (controlling for age and gender) between adulthood BMI and waist circumference was r = 0.91.
The number of the participant's children and their years of birth were reported by the participants in the follow-up phase when the participants were 33, 36, and 39 years of age. For partnership history we created a dichotomous variable that indicated whether or not the participant had ever lived together with a partner (ie, had either been married or cohabiting). Data for this variable were taken from all the 6 follow-up phases, in which the participants have reported their current marital status and changes in marital status between the follow-ups.
Education level was measured by the completed years of education in adulthood reported by the participants. Place of residence in adulthood was reported by the participants on a four-point scale (1 = city; 2 = suburban area; 3 = rural area; 4 = remote rural area), and was used as a continuous covariate in the analyses.
The association between BMI and the number of children was assessed with Poisson regression,18 for which the equation with n independent variables is
The Poisson model provided an adequate fit for the present data (Pearson goodness-of-fit χ2=1310.65, df =1291), and there was no significant overdispersion in the dependent variable (mean = 1.6, variance = 1.7). The Poisson regression was fitted with adolescent BMI and its square as the independent variables; adulthood number of children was the dependent variable; and covariates were age, sex, years of education and place of residence. The associations of skinfold thickness and of age at menarche with the number of children were also assessed with Poisson regression, with the same covariates as mentioned above. The parameter estimates were expressed as incidence-risk ratios (IRRs).
In order to determine the temporal patterning of the fertility differences, discrete-time survival analysis19 was used to examine whether and when the participants had their first, second, and third child, and whether adolescent BMI predicted these events. The birth of the fourth or any subsequent child was not considered since only 6% of the participants had a fourth child and only 2% had 5 or more children (Table 1). Survival analysis takes into account the phenomenon of censoring, ie, the fact that not all participants have their children within the study period, and some participants do not have children at all. In survival analysis a participant is censored either when the event of interest occurs (ie, a child is born) or when the study period ends (ie, at the age of 33, 36, or 39, depending on whether the participant belonged to the youngest, middle or oldest birth cohort). The survival analyses thus allowed us to determine the population estimates for the probabilities of having the first, second, and third child by the age 39. Age, sex, education and place of residence were entered as covariates. The age-specific fertility is known to follow a bell-shaped curve, so the effect of time was modeled as a nonlinear function. The parameter estimates were expressed as odds ratios of logit hazard function for one unit increase in the independent variable.
Adolescent BMI Categorization for Illustration
The results were illustrated by categorizing participants into groups of underweight, normal weight, overweight, and obese on the basis of their adolescent BMI. For the overweight and obese groups we used the international cut-off values provided by Cole et al20 Based on these cut-offs, 7% of adolescent females and 9% of males were categorized as overweight and 0.9% of females and 1.4% of males were categorized as obese. In the absence of established cut-off values for adolescent underweight, we categorized BMI in the lowest 5% as underweight. This categorizing was carried out within sex and birth cohort groups, and provided the respective BMI cut-off values of 14.8, 16.6, and 17.6 for females 12, 15, and 18 years of age, and the corresponding values of 15.2, 17.0, and 18.3 for males. Next we determined the median adolescent BMI values within the underweight, normal weight, overweight and obese groups, (16.2, 19.4, 25.1, and 30.4, respectively). The corresponding values were 15.7, 19.5, 25.1, and 29.9 for females, and 16.4, 19.2, 25.1, and 30.5 for males. These values were used as prototypical or average cases of the 4 adolescent BMI groups in illustrations of the model-predicted estimates. In these calculations other covariates were assigned their mean values.
The association between adolescent BMI and partnership history was assessed with logistic regression analysis. Adolescent BMI and its square were the independent variables; adulthood number of children was the dependent variable; and the covariates were age, sex, years of education and place of residence.
BMI was positively skewed, so we examined whether a transformation of BMI affected the results using an inverse transformation (ie, X=1/BMI) which corrected the skewness. The results were similar and, in terms of confidence intervals of the estimates, even slightly more precise when the transformed BMI was used (data not shown). However, the original scale provided more convenient regression coefficients, so the analyses were fitted with the original BMI scale. Outlier diagnostics indicated that extreme observations did not substantially influence the results.
With one exception (see below), the findings were similar for women and men, with no statistical evidence of differences by sex. However, because of the sex-specific nature of fertility, we also present the results separately for women and men. In some analyses the confidence intervals of the relevant parameter estimates included 1.00 in separate analyses for women and men, but not when sexes were combined. In the absence of statistical interaction by sex, we will interpret these results based on the estimates for women and men combined.
There was a nonlinear association between adolescent BMI and the number of children in adulthood (Table 2, Model 1), indicating that underweight and obese individuals had fewer children than those with normal weight (Fig. 1A). The predicted number of children of average members (ie, median values) of adolescent underweight, normal weight, overweight, and obese groups were, respectively, 1.49, 1.78, 1.64, and 1.10 in women and 1.32, 1.46, 1.40, and 0.99 in men. In other words, average underweight, overweight, and obese women had, respectively, 16%, 8%, and 38% fewer children than women with normal weight. The corresponding percentages in men were 10%, 4%, and 32%. Individuals with BMI about 2 units above the mean (ie, around the cut-off point of the 75th percentile) were estimated to have 2–3% more children than individuals with mean BMI. Note that women had more children than men because women begin to have children at a younger age.21,22
The survival analyses indicated that there was a nonlinear association between adolescent BMI and the probability of having the first, second or third child at a given age (Table 3; Fig. 2). There were no linear or nonlinear interaction effects between BMI and time, indicating that the strength of the association between BMI and the probability of having children did not increase or decrease with time. Figure 2 shows the cumulative probability (ie, P = 1 – value of survival function) of having children plotted against age in women and men. Table 4 shows in numerical form the likelihoods of having the first, second, and third child by the age of 39 in women and men.
The partial correlation (controlling for age and sex) between Year-0 adolescent and Year-21 adulthood BMI was r=0.54. We tested whether adulthood BMI mediated the association between adolescent BMI and the number of children by entering both adolescent and adulthood BMI into the same model. The results differed by sex (sex × adulthood BMI interaction effect: linear IRR=1.28, 95% confidence interval [95% CI] = 1.02–1.60; quadratic IRR = 0.996, 95% CI = 0.992–0.999). When both BMIs were in the model, only adolescent BMI was important in women, while only adulthood BMI was important in men (Table 2, Model 3). In other words, the influence of adolescent BMI was mediated by adulthood BMI in men but not in women. This result was the same when waist circumference rather than BMI was used as an indicator of adulthood body fatness (data not shown).
Next we assessed whether adolescent BMI predicted the likelihood of having ever lived with a partner (ie, having been married or cohabiting), and the degree to which this accounted for the association between BMI and the number of children. Adolescent BMI predicted in a nonlinear fashion the likelihood of having ever lived with a partner in women (linear OR = 1.83, 95% CI = 1.03–3.26; quadratic OR = 0.986, 95% CI = 0.973–0.999) and men (linear OR = 1.62, 95% CI = 0.996–2.64; quadratic OR = 0.989, 95% CI = 0.978–0.999; Fig. 1B) with the respective probability estimates of 0.84, 0.90, 0.90, and 0.81 in women and 0.79, 0.84, 0.84, and 0.74 in men for average members of the 4 BMI groups. Adolescent BMI was still associated with number of children even after controlling for partnership history (Table 2, Model 2).
When the Poisson regression models were fit including only participants who had ever lived together with a partner, the evidence of the BMI-fertility association was weaker, all (n = 1130): linear IRR = 1.14, 95% CI = 0.99–1.33; quadratic IRR = 0.997, 95% CI = 0.993–1.000; Women (n = 640): linear IRR = 1.15, 95% CI = 0.94–1.42; quadratic IRR = 0.997, 95% CI = 0.992–1.001; Men (n = 490): linear IRR = 1.13, 95% CI = 0.91–1.40; quadratic IRR = 0.997, 95% CI = 0.992–1.002], although not absent. In this subsample, the predicted number of children in the 4 adolescent BMI groups was 1.77, 1.93, 1.82 and 1.46 in women, and 1.63, 1.69, 1.56, and 1.21 in men. In other words, underweight, overweight and obese women had approximately 8%, 5%, and 24% fewer children, respectively, than women with normal adolescent weight. In men the corresponding percentages were 4%, 8%, and 29%, respectively.
Among women the partial correlation (controlling for age) between BMI and age at menarche was r=−0.27, indicating that heavier women had earlier menarche. There was a nonlinear association between age at menarche and the number of children, such that women with early or late menarche had more children than those in between (linear IRR=0.44; CI=0.25–0.76; quadratic IRR = 1.031; 95% CI = 1.010–1.053; Fig. 1C). Controlling for age at menarche did not change the association between adolescent BMI and fertility (data not shown).
The association between adolescent skinfold thickness and adulthood number of children was in the same direction (all: linear IRR = 1.13, 95% CI = 0.98–1.31; quadratic IRR = 0.988, 95% CI = 0.973–1.004; women: linear IRR = 1.13, 95% CI = 0.92–1.39; quadratic IRR = 0.986, 95% CI = 0.960–1.013; men: linear IRR = 1.16, 95% CI = 0.92–1.48; quadratic IRR = 0.987, 95% CI = 0.963–1.012) as with BMI, although the confidence intervals of the parameter estimates included 1.00.
Our findings suggest that underweight and obesity are associated with reduced reproduction in the general population. The participants of the present study were still in their reproductive years, so their current fertility may not have accurately represented their completed fertility. However, the survival analyses suggested that this may not have been a major limitation, because the reproductive differentials between BMI groups became stronger over time. Moreover, less than 5% of babies in Finland are born to women over 39 years of age,21 the corresponding percentage being somewhat higher for men.22 It is therefore unlikely that the observed fertility differences related to body weight would represent only transitory effects, although this should be confirmed with further follow-up of this cohort.
In women, underweight is known to cause menstrual dysfunction,23 and obesity has been associated with decreased fecundability and increased risk of pregnancy complications.4–6 Obesity is also a risk factor for polycystic ovary syndrome, which causes infertility.24 In men, low (<20 kg/m2) and high (>25 kg/m2) adulthood BMI have been associated with decreased semen quality,8 and obesity has been shown to increase the risk of erectile dysfunction.9 These physiological correlates of body weight may contribute to the lower fertility in underweight and obese people.
The lowered fertility of underweight and obese individuals was partly accounted by their decreased likelihood of attaining a partner. We were not able to examine whether other factors related to mating (eg, the number of partners, the stability of relationships) might also be involved. The role of other psychologic and social variables should also be considered. Taller young women have been found to have lower maternal tendencies and decreased preference for having children.25 Whether a similar phenomenon is related to BMI is unclear26; the possible correlation between childbearing motivation and body weight merits further research. The extremes of body weight may also be related to psychiatric disorders.27,28
Research suggests that there is assortative mating (ie, nonrandom mating that results in similarity between spouses) for BMI in general,29 and obesity in particular.30 This may act to cumulate the influence of BMI on fertility in underweight and obese couples. Assortative mating also implies that the BMI-fertility association observed in one sex might be partly accounted by their spouse's BMI, for which we did not have data. This possibility has received little attention in fertility research. Sallmen et al7 found that men's overweight decreased the likelihood of conception independently of their wives’ BMI. Further data on couples are needed to investigate the combined influence of partners’ body weight on the number of their children.
Adolescent BMI predicted fertility independently of adulthood BMI in women but not in men. This difference may be because in women the association between adulthood BMI and fertility is probably confounded by the influence of pregnancies on adulthood BMI.12 It may also be that women's adolescent body weight has independent long-term implications for reproductive potential. This would parallel the finding that adolescent overweight may increase adulthood mortality independently of adulthood body weight.2 Indeed, adolescent overweight has been implicated as a risk factor for adulthood polycystic ovary syndrome31 and menstrual disturbances.31–33 Women with a history of eating disorders and severe underweight appear to suffer no major long-term consequences of infertility,34–36 although they are more likely to have obstetric complications.27,34–36 More detailed information on possible biologic and social mechanisms would help to evaluate the relative importance of adolescent and adulthood body weight on fertility.
The ongoing increases in obesity3 may represent a concern for future reproductive health.37 In Finnish adolescents the prevalence of obesity increased 2- to 3-fold from 1977 to 1999.38 Hence, the lowered fertility associated with high BMI may affect an increasing proportion of the population in generations younger than the present study participants, who were adolescents in the early 1980s.
Adolescent BMI is a highly heritable trait with an estimated heritability of about 85% in Finnish adolescent twins.39 Heritable traits associated with reproductive differentials have the potential to evolve via the mechanism of natural selection.40 The present results suggest that stabilizing selection may be acting on BMI, since the low and high ends of BMI are selected against in terms of reproductive success. On the other hand, individuals with a BMI about one-half standard deviations above the mean were estimated to have 2–3% more children than those with mean BMI. If this pattern were to hold over generations, genetic factors might be shifting the mean BMI of the population about 2 units upwards in future generations, albeit only at a modest rate. Reproductive differentials are unlikely to account for any substantial portion of the recent obesity epidemic.
The higher fertility of women with early or late menarche was an unexpected finding. Early-maturing girls tend to enter sexual and marital relations earlier,17 which might contribute to their greater number of children in adulthood. On the other hand, a study conducted in China41 found that late rather than early menarche was associated with increased fertility. The pattern of the present finding suggests that timing of menarche may be under disruptive natural selection where individuals in the extreme ends of a trait have a reproductive advantage over the others. It would be of interest to investigate further the developmental pathways linking early and late menarche to increased adulthood fertility.
1. Power C, Lake JK, Cole TJ. Measurement and long-term health risks of child and adolescent fatness. Int J Obes Relat Metab Disord
2. Must A, Jacques PF, Dallal GE, Bajema CJ, Dietz WH. Long-term morbidity and mortality of overweight adolescents. A follow-up of the Harvard growth study of 1922 to 1935. N Engl J Med
3. World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Report of a WHO Consultation.
Geneva, Switzerland: WHO; 2000.
4. Zaadstra BM, Seidell JC, Van Noord PAH, et al. Fat and female fecundity: prospective study of effect of body fat distribution on conception rates. BMJ
5. Norman RJ, Clark AM. Obesity and reproductive disorders: a review. Reprod Fertil Dev
6. Siega-Riz AM, Siega-Riz AM, Laraia B. The implications of maternal overweight and obesity on the course of pregnancy and birth outcomes. Matern Child Health J
7. Sallmen M, Sandler DP, Hoppin JA, et al. Reduced fertility among overweight and obese men. Epidemiology
8. Jensen TK, Andersson AM, Jorgensen N, et al. Body mass index in relation to semen quality and reproductive hormones among 1,558 Danish men. Fertil Steril
9. Bacon CG, Mittleman MA, Kawachi I, et al. A prospective study of risk factors for erectile dysfunction. J Urol
10. Gortmaker SL, Must A, Perrin JM, et al. Social and economic consequences of overweight in adolescence and young adulthood. N Engl J Med
11. Ellis L, Haman D. Population increases in obesity appear to be partly due to genetics. J Biosoc Sci
12. Smith DE, Lewis CE, Caveny JL, et al. Longitudinal changes in adiposity associated with pregnancy. The CARDIA Study. Coronary Artery Risk Development in Young Adults Study. JAMA
13. Weng HH, Bastian LA, Taylor DH, et al. Number of children associated with obesity in middle-aged women and men: results from the Health and Retirement Study. J Womens Health
14. Åkerblom HK, Uhari M, Pesonen E, et al. Cardiovascular Risk in Young Finns. Ann Med
15. Raitakari OT, Juonala M, Kähönen M, et al. Cardiovascular risk factors in childhood and carotid artery intima-media thickness in adulthood: the Cardiovascular Risk in Young Finns Study. JAMA
16. Biro FM, Khoury P, Morrison JA. Influence of obesity on timing of puberty. Int J Androl
17. Udry RJ, Cliquet RL. A cross-cultural examination of the relationship between ages at menarche, marriage, and first birth. Demography
18. Long JS, Freese J. Regression Models for Categorical Dependent Variables Using Stata, Revised Edition.
College Station, TX: Stata Press; 2003.
19. Singer JB, Willett JB. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence.
Oxford, Oxford University Press; 2003.
20. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ
21. Gissler M. Hedelmällisyys Suomessa ennen, nyt ja tulevaisuudessa [Fertility in Finland in the past, now, and in the future] Yhteiskuntapolitiikka
22. Nikander T. Suomalaismiehen perheellistyminen
[The family formation in Finnish men] (Statistics Finland, Helsinki, 1995; in Finnish).
23. Frisch RE. Body fat, menarche, fitness and fertility. Hum Reprod
24. Pasquali R, Gambineri A, Pagotto U. The impact of obesity on reproduction in women with polycystic ovary syndrome. BJOG
25. Deady DK, Smith MJL. Height in women predicts maternal tendencies and career orientation. Pers Ind Diff
26. Ball K, Crawford D, Kenardy J. Longitudinal relationships among overweight, life satisfaction, and aspirations in young women. Obes Res
27. Finfgeld DL. Anorexia nervosa: analysis of long-term outcomes and clinical implications. Arch Psychiatr Nurs
28. Simon GE, Von Korff M, Saunders K, et al. Association between obesity and psychiatric disorders in the US adult population. Arch Gen Psychiatry
29. Silventoinen K, Kaprio J, Lahelma E, Viken RJ, Rose RJ. Assortative mating by body height and BMI: Finnish twins and their spouses. Am J Hum Biol
30. Hebebrand J, Wulftange H, Goerg T, et al. Epidemic obesity: are genetic factors involved via increased rates of assortative mating? Int J Obes Relat Metab Disord
31. Laitinen J, Taponen S, Martikainen H, et al. Body size from birth to adulthood as a predictor of self-reported polycystic ovary syndrome symptoms. Int J Obes Relat Metab Disord
32. Rich-Edwards JW, Goldman MB, Willett WC, et al. Adolescent body mass index and infertility caused by ovulatory disorder. Am J Obstet Gynecol
33. Lake JK, Power C, Cole TJ. Women's reproductive health: the role of body mass index in early and adult life. Int J Obes Relat Metab Disord
34. Crow SJ, Thuras P, Keel PK, Mitchell JE. Long-term menstrual and reproductive function in patients with bulimia nervosa. Am J Psychiatry
35. Bulik CM, Sullivan PF, Fear JL, et al. Fertility and reproduction in women with anorexia nervosa: a controlled study. J Clin Psychiatry
36. Wentz E, Gillberg IC, Gillberg C, Rastam M. Fertility and history of sexual abuse at 10-year follow-up of adolescent-onset anorexia nervosa. Int J Eat Disord
37. Skakkebaek NE, Jorgensen N, Main KM, et al. Is human fecundity declining? Int J Androl
38. Kautiainen S, Rimpelä A, Vikat A, Virtanen SM. Secular trends in overweight and obesity among Finnish adolescents in 1977–1999. Int J Obes Relat Metab Disord
39. Pietiläinen KH, Kaprio J, Rissanen A, et al. Distribution and heritability of BMI in Finnish adolescents aged 16y and 17y: a study of 4884 twins and 2509 singletons. Int J Obes Relat Metab Disord
40. Darwin C. On the Origin of Species.
Cambridge: Harvard University Press; 2005/1859.
© 2007 Lippincott Williams & Wilkins, Inc.
41. McKibben SL, Poston Jr., DL. The influence of age at menarche on the fertility of Chinese women. Soc Biol.