Heys, Michelle*; Schooling, C Mary*; Jiang, Chaoqiang†; Cowling, Benjamin J.*; Lao, Xiangqian*; Zhang, Weisen†; Cheng, Kar Keung‡; Adab, Peymane‡; Thomas, G Neil*; Lam, Tai Hing*; Leung, Gabriel M.*
Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
Age of menarche is affected by both genetic and environmental factors. With economic development since the Industrial Revolution in the West, mean age of menarche fell by 4 months per decade from 1830 to 19501 and then fell more slowly, if at all.2 More recently, in East Asia, there has been a rapid fall in the age of menarche of about 8 months per decade.3,4 Parts of highly developed East Asia, such as Hong Kong, now have one of the lowest mean ages of menarche world-wide, at 11.7 years.5 Currently mean age of menarche in urban China is between 12.0 and 13.4 years.6,7
Cardiovascular disease is increasingly rapidly in China, and is now one of the leading causes of death.8 The metabolic syndrome is associated with increased risk of cardiovascular disease morbidity and mortality in both Chinese and those of European descent.9–13
In the West, earlier age of menarche has been associated with individual and clusters of cardiovascular risk factors in adulthood (including obesity, insulin resistance, and hypertension14–17), although the evidence is not very consistent, possibly because of the relatively narrow range of age of menarche and because of confounding by differential behavioral patterns. There are no studies examining this relation in Chinese women. There is also little evidence as to whether earlier age of menarche, with the associated greater lifetime estrogen exposure, continues to be a cardiovascular risk factor after the menopause, when women are more susceptible to cardiovascular disease. We took advantage of a large study in southern China to examine the relation between age of menarche and a composite measure of cardiovascular risk, the metabolic syndrome. This population has a relatively high age at menarche (and wide age range); nutritionally limited childhood conditions; a diet low in animal source foods18; and social circumstances, norms, and attitudes involving women that largely preclude earlier puberty being associated with unhealthy behavior such as use of alcohol, smoking, or obesity.19 To clarify the etiologic role of age of menarche, we tested the hypothesis that cardiovascular risk (proxied by the metabolic syndrome) varies inversely with age of menarche in the Guangzhou Biobank Cohort Study.
The Guangzhou Biobank Cohort Study is an ongoing collaboration between the Guangzhou People’s Number 12 Hospital and the Universities of Hong Kong and Birmingham; it is described in detail elsewhere.20 Subjects are recruited from The Guangzhou Health and Happiness Association for the Respectable Elders (GHHARE), a community social and welfare association aligned with the municipal government. Membership is open to anyone for a monthly fee of 4 Yuan (50 US cents). About 7% of permanent Guangzhou residents age 50 years and over are members of GHHARE. Eleven percent of members were enrolled in this study. They were included if they were ambulatory, capable of consenting, and not receiving treatment modalities which, if interrupted, might result in immediate life-threatening risk (such as chemotherapy or radiotherapy for cancer, or dialysis for renal failure). Of those eligible, 90% of the men and 99% of the women participated. Subjects underwent a detailed medical interview, including questions regarding lifestyle and socioeconomic status, and a physical examination in 2003–2004. The Guangzhou Medical Ethics Committee of the Chinese Medical Association approved the study and all subjects gave written, informed consent prior to participation.
Detailed methods of measurement and investigation have been reported elsewhere.20 In brief, seated blood pressure was taken as the average of the last 2 of 3 measurements, using the Omron 705CP sphygmomanometer (Omron Corporation, Kyoto, Japan). Waist circumference was measured as the smallest horizontal circumference between the ribs and iliac crest, or, for obese participants, the circumference at the level of the navel. Standing height was measured without shoes to the nearest 0.1 cm. Fasting total serum cholesterol, HDL-c, triglycerides, and fasting plasma glucose levels were determined by Shimadzu CL-8000 Automatic Chemical Analyzer (Shimadzu Corporation, Kyoto, Japan) in the hospital laboratory.
The primary outcome was the metabolic syndrome and secondarily its components. The National Cholesterol Education Program-Adult Treatment Panel III21 guidelines classify individuals as having the metabolic syndrome if at least 3 of the following 5 criteria are present: (a) central obesity (defined as waist circumference ≥88 cm in females); (b) raised blood pressure (systolic blood pressure ≥130 or diastolic blood pressure ≥85 mm Hg or treatment of previously diagnosed hypertension); (c) raised fasting plasma glucose (≥6.1 mmol/L) or treatment for previously diagnosed type 2 diabetes; (d) fasting plasma triglyceride (≥1.69 mmol/L); and (e) reduced fasting HDL-c: (≤1.29 mmol/L in females). We adopted Asian-specific cutpoints for waist circumference (≥ 80cm in females11) and the revised cutpoint for fasting plasma glucose (≥5.6 mmol/L22).
Age of Menarche
The exposure, age of menarche, was recorded in years (as per the Gregorian calendar and related interpretation of age), rounded during data collection to the nearest year (eg, 13 years represents the onset of menarche from 12 years 6 months to 13 years 5 months). We categorized age of menarche as <12.5 years, 12.5 to <14.5 years and ≥14.5 years, to reflect current age of menarche in China and to be consistent with other studies,14–16 within the constraints of the data collected.
Multivariable logistic regression was used to assess the relation between age of menarche, considered as a 3-level categorical variable, and the metabolic syndrome and its components. Age of menarche ≥14.5 years was the reference category. We considered the following, potential confounders categorized as in Table 1: age, education, current annual personal income, job type, smoking status, alcohol use, exercise, age of menopause (as a categorical variable), number of reproductive years (age of menopause minus age of menarche; continuous), number of years since menopause (chronological age minus age of menopause, continuous), and number of pregnancies (continuous). Women were asked to recall their age of menopause and whether or not they had ever used hormone replacement therapy (HRT) or undergone a hysterectomy. Menopausal status was categorized as women with a history of a hysterectomy and women without a history of hysterectomy divided into age of menopause (≤48 years; 49–51 years; and ≥52 years), giving a total of 4 categories. Exercise was quantified using the short version of the International Physical Activity Questionnaire.23 Smoking status and alcohol use were considered dichotomously because of their low prevalence in these women.8
Potential confounders, other than age and education, were assessed for inclusion in the final model according to a change-in-estimate criterion (at least 5% change)24 for the metabolic syndrome to create a parsimonious model. We presented crude odds ratios (ORs) and odds ratios adjusted for age, education and number of pregnancies, together with their 95% confidence intervals (CIs). We additionally adjusted for waist circumference to assess whether the effect of menarche on the other components of the metabolic syndrome were due to central obesity. We separately adjusted for height as a marker of childhood conditions.
To identify whether the relation of age of menarche with the metabolic syndrome and its components was linear, log linear, or threshold (age of menarche <12.5 years), we compared model fit, using the Akaike Information Criterion, for choosing among regression models.25 In addition we calculated prevalence of the metabolic syndrome and its components adjusted for age, education, and number of pregnancies by age of menarche as a categorical variable (8 categories: <12.5, 12.5–13.4, 13.5–14.4, 14.5–15.4, 15.5–16.4, 16.5–17.4, ≥18.5).
To investigate the potentially confounding effect of menopause, we examined the relation of 3 menopause-related factors (age of menopause [categorical variable], years since menopause, and number of reproductive years) to the metabolic syndrome and its components. We also examined whether the effects of age of menarche were consistent across educational levels, age groups (<65 years and ≥65 years), and menopausal age groups (above or below the median), based on the heterogeneity of effect across strata (with age as a continuous variable in these tests) and the comparison of model fit with and without interaction terms. Finally, a sensitivity analysis was performed using the original definition21 (fasting blood glucose ≥6.1 mmol/L and waist circumference ≥88 cm) and the International Diabetes Federation definitions for the metabolic syndrome.26
Reliability of the questionnaire was confirmed 6 months into recruitment by recalling 200 randomly selected subjects. The intraclass correlation coefficients for continuous variables were age of menarche (0.92) and age at menopause (0.70). Kappa values for the categorical variables were smoking (0.96 and 0.88 for the 2 questions on smoking status), drinking (0.60), physical activity (0.58), education (0.90), occupation (0.80), and parity (0.93).
Of the 7349 women recruited, 7108 (97%) had complete data for all variables. The women were age 50 to 94 years, with a mean ± SD of 64.0 ± 6.0 years. Mean age of menarche was 15.4 ± 2.1 year (range = 8–25 years). Most of the women had attended primary school or above (83%). The majority of women (99%) had never taken HRT, and a small proportion (7%) had a history of a hysterectomy. Most women had never smoked (95%) and never drunk alcohol (91%). Around two-thirds of women reported regular physical exercise as defined by the International Physical Activity Questionnaire.23
Earlier age of menarche was more common in women with earlier age of menopause. Women with age of menarche less than 12.5 years were otherwise similar to those with age of menarche between 12.5 and 14.5 years. Women with late onset of menarche (≥14.5 years) were more likely to be older, have had more pregnancies, and report lower education, lower personal income, and a manual occupation. (Table 1) There was no clear pattern for age of menarche in relation to smoking, alcohol use, or exercise.
According to the US definition,21 with revision only for Asian waist circumference, 30% of the women had metabolic syndrome, which was similar to the population representative rate (28%) for Chinese women using the same definition.27 Using the guidelines revised for both waist circumference and fasting glucose, 38% of women in our study had the metabolic syndrome, 61% had raised blood pressure, 51% had high fasting blood glucose, 35% had high fasting plasma triglycerides, 10% had reduced fasting HDL-c, and 51% had central obesity.
Table 2 shows the relation between age of menarche in 3 groups (<12.5 years, 12.5–<14.5 years and ≥14.5 years) and the metabolic syndrome and its components, using age of menarche ≥14.5 years as the reference category. Age of menarche before 12.5 years was associated with the presence of the metabolic syndrome, central obesity, high blood pressure, high fasting plasma glucose and high triglycerides.
The highest OR for low HDL-c was found for the intermediate age of menarche group (12.5–<14.5 years), for which there is no plausible biologic explanation. Given the small number of women (n = 45) who had reduced HDL-c and age of menarche age <12.5 years, this is most likely to be a false-positive result. Recruitment of subjects occurred steadily throughout the year; cardiovascular risk factors such as dietary pattern vary with season.28 We therefore checked for any potential effect of seasonality by additionally adjusting for month of recruitment (as a categorical variable), which removed the apparent peak in reduced HDL-c at age of menarche 12.5 to <14.5 years, but did not affect any other results (data not shown).
Adjustment for confounders strengthened the positive association between earlier age of menarche and the metabolic syndrome and all of its components. Additional adjustment for waist circumference attenuated effects, but the associations remained. Adjusted for waist circumference, young age of menarche (<12.5 years) compared with age of menarche ≥14.5 years was associated with a higher risk of raised blood pressure (OR = 1.27; 95% CI = 1.03–1.56), fasting glucose (1.33; 1.08–1.63), and high triglyceride levels (1.25; 1.01–1.54).
Adjustment for height strengthened the associations. Adjusted for height, young age of menarche (<12.5 years) compared with age of menarche ≥14.5 years was associated with a higher risk of the metabolic syndrome (1.57; 1.28–1.92), central obesity (1.47; 1.20–1.80), high blood pressure (1.35; 1.09–1.66), high fasting glucose (1.43; 1.17–1.75), and high triglyceride levels (1.37, 1.12–1.68).
Age of menopause, years since menopause, and number of reproductive years had little association with the metabolic syndrome or its components, with narrow confidence intervals around the null (data not shown). Additional adjustment for measures of menopausal status did not affect the reported relation between age of menarche and the metabolic syndrome and its components (data not shown). There was no evidence of effect modification by age (<65 years compared with ≥65 years), educational level, or age of menopause (<49.5 years compared with ≥49.5 years); (data not shown).
Adjusted prevalence of the metabolic syndrome and its components were usually highest for those with age of menarche <12.5 years (Fig. 1). A model with a threshold effect for age of menarche <12.5 years fit best for the metabolic syndrome, central obesity and high fasting blood glucose. On the other hand, a model with a linear relation with age of menarche fit best for high blood pressure and high triglycerides. (Appendix Table A1, available with the online version of the paper.)
The associations were qualitatively and quantitatively similar with the original National Cholesterol Education Program—Adult Treatment Panel III21 definition (without Asian-specific cut offs) and the International Diabetes Federation26 definition for the metabolic syndrome (Appendix Table A2).
This study has examined age of menarche and the metabolic syndrome in postmenopausal women. Young age of menarche (<12.5 years) was associated with the presence of the metabolic syndrome, central obesity, and high fasting glucose; younger age of menarche was associated with high blood pressure and high triglyceride levels. Adult central obesity may be an important mediating factor but does not fully explain these associations, which remained after adjustment for central obesity. Consistent with other studies,29,30 age of menopause, years since menopause, and number of reproductive years (as a reflection of lifetime estrogen exposure) were not clearly related to the metabolic syndrome, and did not confound or modify the relation between age of menarche and the metabolic syndrome and its components. Adjustment for education strengthened the positive association between earlier age of menarche and the metabolic syndrome and its components.
Despite differences in setting, our study shares some features with studies in the West. The Bogalusa Heart Study16—the only other study to have examined a clustering of cardiovascular risk factors—found menarche before the age of 12, associated with development of these risk factors in early adulthood, with an odds ratio of 1.54 (1.14–2.07), very similar to our odds ratio of 1.51 (1.23–1.85) for the metabolic syndrome. Young age of menarche is consistently associated with obesity,31 and after adjustment for adiposity may also be associated with fasting blood glucose and postchallenge blood glucose.14 Findings on the relation between age of menarche and blood pressure are less consistent.14–16,32 Little association has been found between lipid levels and age of menarche.16
To the best of our knowledge, no previous study has examined the form of the relation between age of menarche and cardiovascular risk. We found a threshold effect of early age of menarche (<12.5 years) for the metabolic syndrome, central obesity, and elevated fasting blood glucose, which is suggestive of a critical period of young age of menarche. However, we also found a linear dose-response relationship for elevated blood pressure and high triglycerides, suggesting that, as the age of menarche falls, there is a steady increase in the risk of hypertension and high triglycerides. This suggests that there may be different pathophysiological pathways by which early menarche affects hypertension and central obesity. These findings are similar to results from another recent study of Filipino women, which concluded that the early-life determinants of diabetes and coronary heart disease are probably different.33
It has been suggested that earlier puberty may be associated with unhealthy lifestyles, thereby contributing to higher cardiovascular risk in adulthood. However, we found negligible differences in adult health-related behavior (smoking, alcohol use, and exercise) across ages of menarche, and adjustment for lifestyle factors had little effect. Additionally, our findings could not be attributed to differences in socioeconomic status, adiposity, or height.
Alternatively, our findings could be a cohort effect resulting from the lifetime circumstances of this cohort and the different ages within it. Living standards in southern China prior to 1949 were broadly similar to preindustrial Europe, with gradual improvement punctuated by periods of great hardship until 1978, after which there was rapid economic development in Guangdong.34 All the women would have been exposed to relatively limited living conditions in childhood, and such conditions would have continued later into adulthood for the older women. Nevertheless, the associations between age of menarche and the metabolic syndrome were similar in the younger and older women. Although the lifetime experiences and lifestyles of these older women may be different from the experiences of young women growing up today in the more developed parts of China, these changes in lifestyle are unlikely to affect the relation between age of menarche and the metabolic syndrome unless they modify the effect of age of menarche—for which there is no evidence. Moreover, the recent and ongoing epidemiologic transition experienced by the women in this cohort fortuitously provides a range of childhood environments that is different from those experienced by the populations usually studied. In this specific context and cohort, the environmental impact on age of menarche and their corresponding metabolic consequences may be more visible than in universally well-fed populations who achieve menarche close to the earliest time allowed by their genetic potential. This does not necessarily mean that earlier puberty mediated by economic development and better living conditions will not affect cardiovascular risk, but that such an effect might not always be visible.
Whether early menarche is an independent risk factor or a marker of other environmentally determined mediators of long-term cardiovascular risk (such as childhood obesity,35 smaller size at birth,36 or an increased tempo of growth37) cannot be determined in our setting, although high levels of childhood obesity are implausible.38 Nevertheless, the childhood environmental conditions that result in earlier age of menarche also increase cardiovascular risk in older adult life. Thus, the current secular trend of decreasing age of menarche in China may already have contributed to increased risk of cardiovascular disease. Clear understanding of these childhood antecedents of disease is vital to formulating appropriate public health policies in such societies in transition.
There are a number of potential limitations of the study. We used cross-sectional data and the woman’s recalled age of menarche. Menarche is a discrete event and a key milestone in a girl’s physiological development that historically has strong cultural significance,39 so is likely to be memorable even in later adulthood. Recall of age of menarche is reliable over many years.40 We demonstrated good reliability on repeat questioning of age of menarche. Furthermore there is no reason to suspect recall bias, as subjects were unaware of this hypothesis at the time of interview. Random misclassification would make our findings conservative.
Although our cohort may not be a representative sample of the population, prevalence of relevant morbidities (eg, hypertension and diabetes) were similar to those in a recent representative sample of urban Chinese.20 Our findings would be biased if either women with healthy metabolic profiles and early age of menarche, or women with unhealthy metabolic profiles and late age of menarche were disproportionately excluded. We have little reason to believe either situation is likely.
In conclusion, menarche before the age of 12.5 years was associated with the metabolic syndrome and its components in a large Chinese cohort. Further research should establish whether these associations are due to the tempo of maturity or are a marker of other risk factors. By 1991 the average age of menarche in urban populations in China had fallen to 12.5 years. Experience in more economically developed but ethnically and culturally similar Chinese settings, such as Hong Kong, suggests that age at menarche will continue to fall throughout China. Earlier age of menarche in younger Chinese women today could already be contributing to an increase in the metabolic syndrome and thus cardiovascular disease—a trend that would become more evident over the next few decades as these women age. Such an increase, rooted in economic conditions in childhood and adolescence, would most likely be beyond the sphere of personal and public health control. This would emphasize the importance of modifiable risk factors for metabolic syndrome (such as obesity and teenage smoking), particularly among women who undergo early sexual maturation.
We thank Sir R Peto and ZM Chen of the Clinical Trial Service Unit, The University of Oxford for their support. The Guangzhou Cohort Study investigators include: Guangzhou No. 12 Hospital: XQ Lao, WS Zhang, M Cao, T Zhu, B Liu, CQ Jiang (Co-PI); The University of Hong Kong: GN Thomas, CM Schooling, SM McGhee, GM Leung, RF Fielding, TH Lam (Co-PI); The University of Birmingham: P Adab, Y Peng, KK Cheng (Co-PI). Our thanks also go to Steven Riley for his help with graphics and finally to 2 anonymous reviewers for their contribution to the final manuscript.
1. Marshall W, Tanner J. Puberty. In: Falkener F, Tanner J, eds. Human Growth: A Comprehensive Treatise
. 2nd ed. New York: Plenum Press; 1986.
2. Demerath EW, Li J, Sun SS, et al. Fifty-year trends in serial body mass index during adolescence in girls: the Fels Longitudinal Study. Am J Clin Nutr
3. Graham MJ, Larsen U, Xu X. Secular trend in age at menarche in China: a case study of two rural counties in Anhui Province. J Biosoc Sci
4. Hwang JY, Shin C, Frongillo EA, et al. Secular trend in age at menarche for South Korean women born between 1920 and 1986: the Ansan Study. Ann Hum Biol
5. Tang CS, Yeung DY, Lee AM. Psychosocial correlates of emotional responses to menarche among Chinese adolescent girls. J Adolesc Health
6. Zhang YP, et al. Investigation on menarche age and psychosexual behavior of the girl students in the railway middle school of seven cities (Chinese). Med Soc
7. Chen SK, Guo XM, Ma ZJ, et al. The relationship between body measurements and pubertal growth among 3–16 year olds in Nanning. (Chinese). Zhongguo Er Tong Bao Jian Za Zhi
8. He J, Gu D, Wu X, et al. Major causes of death among men and women in China. N Engl J Med
9. Isomaa B, Almgren P, Tuomi T, et al. Cardiovascular morbidity and mortality associated with the metabolic syndrome. Diabetes Care
10. Dou XF, Zhang HY, Sun K, et al. [Metabolic syndrome strongly linked to stroke in Chinese]. Zhonghua Yi Xue Za Zhi
11. Heng D, Ma S, Lee JJ, et al. Modification of the NCEP ATP III definitions of the metabolic syndrome for use in Asians identifies individuals at risk of ischemic heart disease. Atherosclerosis
12. Thomas GN, Schooling CM, McGhee SM, et al. Metabolic syndrome increases all-cause and vascular mortality: the Hong Kong Cardiovascular Risk Factor Study. Clin Endocrinol
13. Lao XQ, Thomas GN, Jiang CQ, et al. Association of the metabolic syndrome with vascular disease in an older Chinese population: Guangzhou Biobank Cohort Study. J Endocrinol Invest
14. Saquib N, Kritz-Silverstein D, Barrett-Connor E. Age at menarche, abnormal glucose tolerance and type 2 diabetes mellitus: the Rancho Bernardo Study. Climacteric
15. Remsberg KE, Demerath EW, Schubert CM, et al. Early menarche and the development of cardiovascular disease risk factors in adolescent girls: the Fels Longitudinal Study. J Clin Endocrinol Metab
16. Frontini MG, Srinivasan SR, Berenson GS. Longitudinal changes in risk variables underlying metabolic Syndrome X from childhood to young adulthood in female subjects with a history of early menarche: the Bogalusa Heart Study. Int J Obes Relat Metab Disord
17. Cooper GS, Ephross SA, Weinberg CR, et al. Menstrual and reproductive risk factors for ischemic heart disease. Epidemiology
18. Du S, Lu B, Zhai F, et al. A new stage of the nutrition transition in China. Public Health Nutr
19. Yu MY, Sarri R. Women’s health status and gender inequality in China. Soc Sci Med
20. Jiang C, Thomas GN, Lam TH, et al. Cohort Profile: the Guangzhou Biobank Cohort Study, a Guangzhou-Hong Kong-Birmingham collaboration. Int J Epidemiol
21. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). JAMA
22. Bloomgarden ZT. Second International Symposium on triglycerides and HDL: metabolic syndrome. Diabetes Care
23. International Physical Activity Questionnaire. Available at: www.ipaq.ki.se
. Accessed October 29, 2005.
24. Rothman JK, Greenland S. Modern Epidemiology
. 2nd ed. Philadelphia: Lippincott-Raven Publishers; 1998.
25. Harrell FEJ. Regression Modeling Strategies With Applications to Linear Models, Logistic Regression, and Survival Analysis
. New York: Springer-Verlag; 2001.
26. Alberti KG, Zimmet P, Shaw J. The metabolic syndrome—a new worldwide definition. Lancet
27. Gu D, Reynolds K, Wu X, et al. Prevalence of the metabolic syndrome and overweight among adults in China. Lancet
28. Zou XN, Taylor PR, Mark SD, et al. Seasonal variation of food consumption and selected nutrient intake in Linxian, a high risk area for esophageal cancer in China. Int J Vitam Nutr Res
29. Gohlke-Barwolf C. Coronary artery disease—is menopause a risk factor? Basic Res Cardiol
. 2000;95(suppl 1):I77–183.
30. Schneider JG, Tompkins C, Blumenthal RS, et al. The metabolic syndrome in women. Cardiol Rev
31. Pierce MB, Leon DA. Age at menarche and adult BMI in the Aberdeen children of the 1950s cohort study. Am J Clin Nutr
32. Hardy R, Kuh D, Whincup PH, et al. Age at puberty and adult blood pressure and body size in a British birth cohort study. J Hypertens
33. Langenberg C, Araneta MR, Bergstrom J, et al. Diabetes and coronary heart disease in Filipino-American women: role of growth and life-course socioeconomic factors. Diabetes Care
34. Maddison A. Chinese Economic Performance in the Long Run
. Paris: OECD; 1998.
35. Must A, Naumova EN, Phillips SM, et al. Childhood overweight and maturational timing in the development of adult overweight and fatness: the Newton Girls Study and its follow-up. Pediatrics
36. Adair LS. Size at birth predicts age at menarche. Pediatrics
37. Barker DJ, Osmond C, Forsen TJ, et al. Trajectories of growth among children who have coronary events as adults. N Engl J Med
38. Piazza AL. Food Consumption and Nutritional Status in the PRC
. Boulder: Westview Press; 1986.
39. Goldstein J. Scissors, surveys and psycho-prophylactics: prenatal health care campaigns and state building in China, 1949-1954. J Hist Sociol
40. Must A, Phillips SM, Naumova EN, et al. Recall of early menstrual history and menarcheal body size: after 30 years, how well do women remember? Am J Epidemiol
© 2007 Lippincott Williams & Wilkins, Inc.