Body mass index in early and mid-adulthood, and subsequent mortality: a historical cohort study

Abstract

BACKGROUND: The aim of this study was to investigate the associations between body mass index (BMI) in early and mid-adulthood, and BMI change between these ages, and mortality.

METHODS: Historical cohort study of 629 men, who had height and weight measured at the Student Health Service of the University of Glasgow in 1948–1949 (median age 22?y) and who reported their weight in a postal questionnaire in 1963–1966 (median age 38?y). The participants were followed up until April 2002 (mean follow-up: 35?y). During this time, 124 men died, 68 of cardiovascular disease (CVD) and 33 of cancer.

FINDINGS: Mean BMI increased from 21.4?kg/m2 (standard deviation (s.d.): 2.2?kg/m2) in early adulthood to 24.2?kg/m2 (s.d.: 3.0?kg/m2) in mid-adulthood. All-cause mortality was associated with being overweight (BMI≥25?kg/m2) at age 22 but not at age 38, adjusted hazard ratio (HR): 1.85 (95% confidence interval (CI) 1.09–3.13) and 1.05 (95% CI: 0.73–1.52), respectively. BMI at age 22?y was more strongly associated with CVD mortality than was BMI at age 38?y, adjusted HR22?y: 2.41 (95% CI: 1.26–4.60) and HR38?y: 1.33 (95% CI: 0.82–2.16). There was no clear relationship between cancer mortality and BMI at either age: HR22?y: 0.68 (95% CI: 0.16–2.91), HR38?y: 0.90 (95% CI: 0.44–1.84), although relatively few men died of cancer in the follow-up period. Similar patterns were seen for obesity (BMI≥30?kg/m2) as for being overweight. Analyses of weight patterns indicated particularly detrimental effects of overweight persisting from early to mid-adulthood.

CONCLUSIONS: BMI in early adulthood is positively related to CVD mortality in later life in men. The risk associated with early adulthood adiposity appeared to be greater than that in mid-adulthood. We did not demonstrate an association between weight gain and later mortality. These results reinforce the need to stem the obesity epidemic in children and young adults.

Introduction

Numerous studies have documented the greater mortality rates associated with higher body mass index (BMI) in adulthood. There is also accumulating evidence pointing to the detrimental effects of high BMI in childhood and early adulthood on adult disease.1,2,3 As yet, it is unknown whether there is a critical age at which having a high BMI is particularly harmful.

There are methodological problems with studying BMI and BMI change in relation to later outcomes. Firstly, weight tends to track over the life course, so children and young adults who are overweight are more likely to remain overweight throughout their life. For example, in a Norwegian study, correlation coefficients between BMI at age 20–24?y and BMI approximately 16?y later were 0.80?in men and 0.73 in women.4 These results suggest that inferences regarding the presence of a critical age at which being overweight is particularly deleterious are impossible from studies that use a single measure of BMI.

Secondly, it is important to assess both baseline BMI and BMI change in relation to later outcomes. Each of these may be independently related to later disease. Weight gain during adult life is related to an increased risk of coronary heart disease (CHD)5 and stroke6 as well as breast7 and prostate cancer.8 Furthermore, since weight loss may be indicative of premorbid disease, it is important to either use measures of BMI from many years prior to the development of outcomes, or to exclude individuals from analyses who develop the outcome within the first few years of follow-up.9

In this paper, we report the relative magnitude of the association between BMI in early and mid-adulthood, and BMI change, and mortality risk. With two measures of BMI and BMI change, as well as a long follow-up period, this study overcomes the problem of reverse causality mentioned above and may assist in pointing towards the presence of a critical period in relation to the association between BMI and mortality.

Methods

The study is based on a subset of the Glasgow Alumni Cohort, which has been described in detail elsewhere.10 Briefly, students who were registered at the University of Glasgow at some time between 1948 and 1968 were invited to an annual medical examination at the Student Health Service. Approximately half the invited students attended.10 We have traced approximately 82% of participants through the National Health Service Central Register (NHSCR). The copies of death certificates for these individuals are sent to us on a regular basis.

The current analysis is based on male students who attended the Student Health Service between 1948 and 1949 inclusive, and who responded to a postal survey sent in 1963–1966. At the Student Health Service, heights and weights were measured by a physician in Imperial units. These were converted to metric measures at the time of analysis. Few details of the postal survey remain, but it is thought participants were asked to measure their weight in full clothing and to report what they were wearing at the time of measurement. These measures, recorded on index cards, have been preserved. From the data recorded on these cards, it appears that, at the time of the survey, standard weights were assigned to each item of clothing and subtracted from the participant's reported weight. All measurements were recorded in Imperial units and later converted to metric units using the following conversions: 1?lb=0.455?kg and 1?in=2.54?cm.

We calculated early adult BMI in kg/m2 using the weight and height measured at the Student Health Service. BMI in mid-adulthood was calculated using the reported weight from the postal survey and the student height measure. Cause of death was ascertained from the death certificate using the underlying cause of death. Primary outcomes studied were all-cause mortality and cardiovascular (CVD) mortality (ICD-9 390 to 459). Results for cancer mortality (ICD-9 140 to 209) are also presented, and we further stratified cancer mortality according to whether the aetiology of the cancer is thought to be related to smoking or not.11,12 The Cox proportional hazard models were used to estimate mortality risks. Time at risk was considered to begin for all participants at the date of the return of the postal survey. Date of censoring was the earliest of the following: date of emigration, date of death or the end of April 2002, since it was deemed that the mortality data included in the study were complete up until this date. Age in years was used to define the time line underlying the Cox model.

Three basic models were used, using independent variables as BMI at two different ages and the change in BMI over the intervening time period. For the former two, hazard ratios (HRs) were calculated comparing those participants whose BMI was equal to or above 25?kg/m2 (overweight) to those below 25?kg/m2 (normal weight). Although limited by small numbers, we also investigated the HRs comparing those who were obese (BMI equal to or above 30?kg/m2) to normal weight participants. For models using BMI change, that variable was used as a categorical variable, split into four categories each of 1.5?kg/m2 width. For analyses of BMI change, being overweight (yes/no) at baseline was also included as a potential confounding variable. Other potential confounding factors that were considered were smoking (yes/no), father's social class (I/II vs III–V), height (linear) and diastolic BP (linear—not included in cancer mortality models).

To characterise the mortality risk associated with patterns of overweight from early to mid-adulthood, three categories of weight change were identified. These were (i) normal weight at both ages, (ii) normal weight in early adulthood but overweight in mid-adulthood and (iii) overweight at both the ages. Mortality risk was estimated in each category, using the normal weight at both ages category as the baseline.

Results

From 1948 to 1949, 2953 male students attended the Student Health Service, of whom 695 responded to the postal questionnaire in the 1960s. This fraction is not a true response rate, since no information on the number of people who were contacted is available. The response rate is thought to have been approximately 75%.13 There were no differences in mean age or mean BMI at university between the 2258 men who did not and the 695 men who did respond to the postal questionnaire (mean age: 22.83 vs 22.76?y, P=0.71; mean BMI 21.53 vs 21.55?kg/m2, P=0.79, respectively). We excluded participants with incomplete data on anthropometric measures or potential confounding variables (n=64) and those who had an NHSCR notification of emigration prior to the date of the postal survey (n=2). The results presented are based on the remaining 629 men.

The median age of these men when they attended the Student Health Service for the first time was 22.0?y (interquartile range (IQR): 19.8–24.8?y). At the time of the postal survey, the median age was 38.0?y (36.0–41.1?y). Further characteristics of the participants are given in Table 1. The mean BMI increased from 21.4?kg/m2 (standard deviation (s.d.): 2.2?kg/m2) in early adulthood to 24.2?kg/m2 (s.d.: 3.0?kg/m2) in mid-adulthood. The two BMI measures were correlated (r=0.58, P<0.001). The mean weight gain was 8.6?kg. There were 46 men who were overweight in young adulthood, of whom three were obese and 246 who were overweight in mid-adulthood, of whom 23 were obese.

Table 1 Characteristics of the study population

The participants were followed for a mean of 35.2?y (range 7–41?y) from the postal survey. During this time, 124 men (20%) died, of whom 68 died of CVD and 33 died of cancer. The mean age at death was 68.4?y. The mortality rate for cancer was 1.49 (95% confidence interval (CI) 1.06–2.10) and for CVD was 3.07 (95% CI: 2.42–3.90) per 1000 person-years. Crude HRs comparing overweight to normal weight individuals for all-cause mortality and CVD mortality are shown in Table 2. The results show a strong positive association between being overweight at age 22 and all-cause mortality; the crude HR comparing obese to normal weight men was 5.80 (95% CI: 1.43–23.59) at this age. There was no association between being overweight in mid-adulthood and all-cause mortality risk (see Table 2), but there was a positive relationship between being obese in mid-adulthood and all-cause mortality, 2.39 (95% CI: 1.26–4.57). Comparing the crude with the adjusted HRs, there was no evidence of confounding by height, social class in childhood, diastolic blood pressure or smoking status in young adulthood.

Table 2 Association between BMI in early and mid-adulthood and subsequent mortality

A strong positive association was observed between BMI in young adulthood and CVD mortality. The association between BMI in mid-adulthood and CVD mortality was much weaker. The same pattern was evident for relationships between obesity and CVD mortality, with higher, although less precise, risks associated with obesity in early adulthood (HR: 11.99, 95% CI: 2.91–49.45) than in later adulthood, HR: 3.72 (95% CI: 1.70–8.15).

There was a suggestion of an inverse relationship between BMI at both ages and cancer mortality, although the CIs are wide and are compatible with either an increased or a decreased risk in overweight individuals. Adjusted HR (95% CI) for the association between overweight in early adulthood and cancer mortality was 0.68 (0.16–2.91) and in mid-adulthood was 0.90 (0.44–1.84). These figures were based on 33 men who died of cancer, of whom two were overweight in young adulthood and 12 were overweight in mid-adulthood. Of the three men who were obese in young adulthood, none died of cancer. There was no relationship between obesity in later adulthood and cancer mortality (HR: 0.81, 95% CI: 0.11–6.01).

Seven men died of cancers related to smoking (respiratory tract, urinary tract and pancreas). The site of the cancer was not specified on the death certificate of four further men. The remaining 22 participants died of a cancer not thought to be related to smoking. None of those who died of a smoking-related cancer were overweight in young adulthood. Three of the seven individuals who died of a smoking-related cancer were overweight in later adulthood, HR 1.18 (95% CI: 0.26–5.28). For cancers not thought to be related to smoking, the HRs comparing overweight to normal weight men were 0.49 (95% CI: 0.06–3.69) in early adulthood and 0.87 (0.37–2.09) in mid-adulthood.

The median change in BMI over the 16-y period was 2.7?kg/m2 (IQR: 1.1–4.2?kg/m2). Associations between BMI change and mortality are shown in Table 3. These results indicate that there was no association between change in BMI from early to mid-adulthood and all-cause or CVD mortality. Adjustment for social class in childhood, smoking status in young adulthood, adult height and diastolic blood pressure in young adulthood did not affect these results. Similarly, there was no relationship between BMI change and cancer mortality (adjusted HR (95% CI) per 1?kg/m2 change was 1.09 (0.94–1.27)). In examining the joint effects of BMI in young adulthood and BMI change from young to mid-adulthood on mortality, no effect of BMI change was found for any of the outcomes studied.

Table 3 Association between BMI change from early to mid-adulthood and subsequent mortality

The risks associated with each of the categories of patterns of overweight are shown in Table 4. As only two men who were overweight at a young age subsequently became normal weight, these results are not presented. There was an indication of an adverse effect of being overweight in both young and mid-adulthood with respect to all-cause mortality. Men who were overweight at both ages had two-fold risks of CVD mortality compared to men with stable normal weight. Adjusted results for cancer mortality showed no relationship between being normal weight than overweight compared to men who were of stable normal weight 0.95 (0.44–2.03), nor any relationship between men who were overweight at both ages compared to those who were normal weight at both ages 0.69 (0.16–3.01).

Table 4 Association between weight patterns in early and mid-adulthood and subsequent mortality

To investigate the possibility that morbidity-induced weight change preceding death could have explained some of our results, analyses were repeated excluding the six deaths that occurred in the first 10?y of follow-up, all of which occurred in years 8–10 following the postal survey. There was no difference in CVD mortality risk in men who were normal weight then became overweight compared to those who remained normal weight (HR: 1.12, CI: 0.64–1.98). There remained over two-fold risks in men who were overweight at both ages compared to men with stable normal weight (HR: 2.39, CI: 1.17–4.89).

Discussion

The results presented here indicate a positive association between BMI in early adulthood and all-cause mortality. Cause-specific analyses indicated that this relationship was primarily due to the increased risk of CVD mortality. Increases in BMI from early to mid-adulthood were unrelated to mortality in this cohort. Weight patterns indicated that men who were persistently overweight were at two-fold risk of CVD mortality compared to men who retained normal weight at both ages. There were no clear relationships between BMI at different ages and cancer mortality.

For the early adult measures, we were able to use premorbid height and weight measured by a physician. Although we were unable to validate these, such measures are likely to be more accurate than self-reported or recalled weights. For example, in a study in the US in which girls were followed from menarche for approximately 33?y, there was substantial systematic bias evident in the recall of weight at the time of menarche.14 This ranged from a 10% overestimation of weight by girls who were thinnest in adolescence to a 30% underestimation by those who were fattest. This inverse association between the degree of error and the original BMI percentile was approximately linear. Using contemporaneous weight measures eliminates this recall bias.

For the calculation of BMI in mid-adulthood, we had to rely on self-reported weight at this age and the value of height measured 16?y previously. Despite this limitation, the contemporary self-reported weight measure is likely to be more accurate than a recalled measure, although the precision of self-reported weight is lower than when weight is measured. Error in the self-reported weight measure may account for the observation that the correlation coefficient between the two measures was lower than has been reported from a previous study.4

Measures of potential confounders were taken at baseline and may have lacked precision. For example, because of the small numbers of smokers included in the study, the crude classification of smoking (yes vs no) may have resulted in residual confounding, although using smoking as a three-level variable (nonsmoker, light smoker and moderate/heavy smoker) did not make any appreciable difference to the results. Furthermore, since data on smoking were collected only once (at baseline), some residual confounding may exist in the results for BMI in mid-adulthood, as patterns of smoking are likely to have changed over the course of the study and smoking cessation is commonly associated with weight gain. However, since statistical adjustment did not alter the results for early adulthood BMI, we believe that our results are unlikely to be strongly confounded by the accuracy or timing of measurement of these variables. There does, however, remain the problem of residual confounding by unmeasured variables. For example, exercise patterns in early and mid-adulthood are likely to have substantial effect on BMI at these time points and may also affect the outcomes studied.

A strength of the study is that there were no deaths in the 8?y following the second weight measure, providing reassurance that the associations that we see are not affected by morbidity-induced weight change preceding death. This issue of intentionality of weight change confounding the association between weight change and mortality is one that has impaired the interpretation of results from studies of weight change in older populations, where deaths have occurred soon after second or subsequent weight measures were made.

The main limitation of the study is one of statistical power. The small numbers of cohort members for whom we have self-reported weight in mid-adulthood may have impaired the ability to detect an association with the outcomes that we studied. This was a consequence of basing the results on the subset of the cohort with self-reported weight in mid-adulthood. Longer term follow-up of this subset will allow further cause-specific mortality to be studied. That overweight men were at higher risk of CVD mortality in this cohort was unsurprising. However, the differences in the strength of the associations that we observed between BMI at two time periods and CVD mortality in particular were striking. Comparing overweight to normal weight men, the risk of CVD mortality was over two-fold in young adulthood, and increased by only 20% in later adulthood. There are two possible explanations as to why this could have been observed. Firstly, the self-reported measure of weight in mid-adulthood is likely to have been less accurate than the value measured by a physician at the student health service. This random misclassification could have diluted the true association, indicating that the association could be similar at both ages. Secondly, it is possible that there is a critical period at which being overweight confers an increased risk of disease in later life. If this critical period was early adulthood, this could explain the observed results. There are few cohorts with measured height and weight and sufficient length of follow-up to investigate the joint role of BMI at different ages and later disease outcomes.

BMI change was unrelated to mortality in this cohort. Evidence to support the presence of an association between weight gain and CVD from previous epidemiological research is strong, which suggests that the second measure of weight in this cohort may be imprecise. This evidence comes from cohort studies which have, to varying extents, attempted to control for factors such as smoking, occult disease, comorbidity, and intent of weight loss. Cohort studies that have demonstrated this include a study from Sweden in which the lowest CHD mortality and myocardial infraction (MI) rates were seen in those men with stable weight from early to later adulthood.5 In the Health Professionals' Study, weight gain since the age of 21?y was also positively associated with risk of CHD among men younger than 65?y.15

The evidence that relates weight gain to cancer risk is less consistent than for CVD, and depends on the site of the cancer studied. Studies of prostate cancer are inconsistent8,16,17,18 and there is no evidence to suggest that weight gain is related to colorectal cancer.3,16,19 Our results are consistent with those previously reported null findings, but need to be interpreted in light of the small number of men who died from cancer included in this study.

The timing of the weight gain ought to be stated explicitly for useful comparisons to be made between studies. For example, Stoll suggested that weight gain at the time of menarche may be the most pertinent in terms of subsequent breast cancer risk.20 Prospective cohort studies with repeated measures of weight throughout the life course, coupled with sophisticated analytical methods, will allow further exploration of the possible mortality risks associated with weight gain.

Surprisingly, overweight men who lost weight in this study were at higher risk of CVD mortality in later life. This was not due to the presence of occult disease, since the association persisted having excluded those people who died in the first 10 years of follow-up. One possibility is that the weight loss in these men could partly be due to smoking, which would increase their risk of CVD in later life. However, we do not have data on smoking patterns in this data set to investigate this possibility.

There have been numerous suggestions regarding potential mechanisms that could link overweight and weight gain to disease risk. Firstly, the most usual cause of change in body weight is a result of fat accumulation. It has been proposed that excessive weight gain is therefore a better marker of adiposity in adult life, either in terms of quantity or distribution, than weight per se.20 Fat deposition associated with age-related weight gain tends to occur centrally.21 Abdominal obesity, as measured by waist circumference or waist–hip ratio, is a recognised risk factor for CVD and some cancers.22,23,24

Secondly, BMI change is related to changes in hormonal profiles and metabolic changes in the body. Obesity is also related to lower levels of sex hormone binding globulin,25 resulting in higher levels of bioavailable sex hormones. Furthermore, overweight is associated with hyperinsulinaemia, which in turn is related to increased bioactivity of insulin-like growth factor-I (IGF-I), as well as to its bioavailability through the association with IGF-binding proteins.20,26 Evidence is accumulating to suggest that growth patterns are related to cancer risk, possibly mediated through IGF-I levels.27

In conclusion, we have shown that BMI in early-adulthood is more strongly associated with increased risk of all-cause and CVD mortality than BMI in mid-adulthood. We did not demonstrate an association between weight change from early to mid-adulthood and mortality. Men who were persistently overweight were at two-fold risk of CVD mortality compared to men who retained normal weight at both ages. The findings add to the evidence supporting the continued need to reduce BMI in early adulthood in order to reduce the burden of future chronic disease.

References

  1. 1

    Must A, Strauss R . Risks and consequences of childhood and adolescent obesity. Int J Obes Relat Metab Disord 1999; 23 (Suppl 2): S2–S11.

  2. 2

    Gunnell D, Frankel S, Nanchahal K, Peters T, Davey Smith G . Childhood obesity and adult cardiovascular mortality: a 57-y follow-up study based on the Boyd Orr cohort. Am J Clin Nutr 1998; 67: 1111–1118.

  3. 3

    Whittemore A, Paffenbarger R, Anderson K, Lee J . Early precursors of site-specific cancers in college men and women. J Natl Cancer Inst 1985; 74: 43–51.

  4. 4

    Wilsgaard T, Jacobsen BK, Schirmer H, Thune I, Lochen ML, Njolstad I, Amesen E . Tracking of cardiovascular risk factors: the Tromso study, 1979–1995. Am J Epidemiol 2001; 154: 418–426.

  5. 5

    Rosengren A, Wedel H, Wilhelmsen L . Body weight and weight gain during adult life in men in relation to coronary heart disease and mortality. Eur Heart J 1999; 20: 269–277.

  6. 6

    Rexrode KM, Hennekens CH, Willett WC, Colditz GA, Stampfer MJ, Rich-Edwards JW, Speizer FE, Manson JE . A prospective study of body mass index, weight change, and risk of stroke in women. JAMA 1997; 277: 1539–1545.

  7. 7

    Ziegler RG, Hoover RN, Nomura AM, West DW, Wu AH, Pike MC, Lake AJ, Horn-Ross PL, Kolonel LN, Siiteri PK, Fraumeni JF Jr . Relative weight, weight change, height, and breast cancer risk in Asian-American women. J Natl Cancer Inst 1996; 88: 650–660.

  8. 8

    Cerhan JR, Torner JC, Lynch CF, Rubenstein LM, Lemke JH, Cohen MB, Lubaroff DM, Wallace RB . Association of smoking, body mass, and physical activity with risk of prostate cancer in the Iowa 65+ Rural Health Study (United States). Cancer Causes Control 1997; 8: 229–238.

  9. 9

    Gunnell DJ, Davey Smith G . Weight, weight gain and coronary heart disease mortality. Eur Heart J 1999; 20: 246–248.

  10. 10

    McCarron P, Davey Smith G, Okasha M, McEwen J . Life course exposure and later disease: a follow-up study based on medical examinations carried out in Glasgow University (1948–1968). Public Health 1999; 113: 265–271.

  11. 11

    Davey Smith G, Shipley M, Leon D . Height and mortality from cancer among men: prospective observational study. BMJ 1998; 317: 1351–1352.

  12. 12

    Gunnell DJ, Davey Smith G, Frankel S, Nanchahal K, Braddon FE, Pemberton J, Peters TJ . Childhood leg length and adult mortality: follow up of the Carnegie (Boyd Orr) survey of diet and health in pre-war Britain. J Epidemiol Community Health 1998; 52: 142–152.

  13. 13

    Okasha M, McCarron P, McEwen J, Durnin J, Davey Smith G . Childhood social class and adulthood obesity: findings from the Glasgow Alumni Cohort. J Epidemiol Community Health 2003; (in press).

  14. 14

    Must A, Phillips SM, Naumova EN, Blum M, Harris S, Dawson-Hughes B, Rand WM . Recall of early menstrual history and menarcheal body size: after 30 years, how well do women remember? Am J Epidemiol 2002; 155: 672–679.

  15. 15

    Rimm EB, Stampfer MJ, Giovannucci E, Ascherio A, Spiegelman D, Colditz GA, Willett WC . Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men. Am J Epidemiol 1995; 141: 1117–1127.

  16. 16

    Nomura A, Heilbrun L, Stemmermann G . Body mass index as a predictor of cancer in men. J Natl Cancer Inst 1985; 74: 319–323.

  17. 17

    Schuurman A, Goldbohm R, Dorant E, Van den Brandt P . Anthropometry in relation to prostate cancer risk in the Netherlands Cohort Study. Am J Epidemiol 2000; 151: 541–549.

  18. 18

    Putnam SD, Cerhan JR, Parker AS, Bianchi GD, Wallace RB, Cantor KP, Lynch CF . Lifestyle and anthropometric risk factors for prostate cancer in a cohort of Iowa men. Ann Epidemiol 2000; 10: 361–369.

  19. 19

    Le Marchand L, Wilkens L, Mi M . Obesity in youth and middle age and risk of colorectal cancer in men. Cancer Causes Control 1992; 3: 349–354.

  20. 20

    Stoll B . Timing of weight gain in relation to breast cancer risk. Ann Oncol 1995; 6: 245–248.

  21. 21

    Ballard-Barbash R . Anthropometry and breast cancer. Body size — a moving target. Cancer Suppl 1994; 74: 1090–1100.

  22. 22

    Despres JP, Lemieux I, Prud'homme D . Treatment of obesity: need to focus on high risk abdominally obese patients. BMJ 2001; 322: 716–720.

  23. 23

    Caan B, Coates A, Slattery M, Potter J, Quesenberry C, Edwards S . Body size and risk of colon cancer in a large case–control study. Int J Obes 1998; 22: 178–184.

  24. 24

    Kaaks R, Van Noord P, Den Tonkelaar I, Peeters P, Riboli E, Grobbee D . Breast cancer incidence in relation to height, weight and body-fat distribution in the Dutch “DOM” cohort. Int J Cancer 1998; 76: 647–651.

  25. 25

    Ukkola O, Gagnon J, Rankinen T, Thompson PA, Hong Y, Leon AS, Rao DC, Skinner JS, Wilmore JH, Bouchard C . Age, body mass index, race and other determinants of steroid hormone variability: the HERITAGE Family Study. Eur J Endocrinol 2001; 145: 1–9.

  26. 26

    Leifke E, Gorenoi V, Wichers C, Von Zur MA, Von Buren E, Brabant G . Age-related changes of serum sex hormones, insulin-like growth factor-1 and sex-hormone binding globulin levels in men: cross-sectional data from a healthy male cohort. Clin Endocrinol (Oxford) 2000; 53: 689–695.

  27. 27

    Okasha M, Gunnell D, Holly J, Davey Smith G . Childhood growth and adult cancer. Best Pract Res Clin Endocrinol Metab 2002; 16: 225–241.

Download references

Acknowledgements

This work was undertaken while Mona Jeffreys (nee Okasha) was employed at the University of Bristol. We are grateful for the financial support provided by the Stroke Association, Chest Heart and Stroke (Scotland), NHS R&D CVD Programme and World Cancer Research Fund International.

Author information

Correspondence to M Jeffreys.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Jeffreys, M., McCarron, P., Gunnell, D. et al. Body mass index in early and mid-adulthood, and subsequent mortality: a historical cohort study. Int J Obes 27, 1391–1397 (2003). https://doi.org/10.1038/sj.ijo.0802414

Download citation

Keywords

  • overweight
  • body mass index
  • cardiovascular disease
  • cancer
  • life course
  • epidemiology

Further reading