IJE Advance Access originally published online on November 12, 2005
International Journal of Epidemiology 2006 35(1):169-178; doi:10.1093/ije/dyi212
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Article |
Cause-specific mortality in old age in relation to body mass index in middle age and in old age: follow-up of the Whitehall cohort of male civil servants
1 Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK
2 Department of Epidemiology and Public Health, University College London, London, UK
3 Clinical Trial Service Unit, University of Oxford, Oxford, UK
* Corresponding author. Department of Epidemiology and Public, University College, 1-19 Torrington Place, London WC1E 6BT, UK. E-mail: e.breeze{at}ucl.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionData sources and methodsResultsDiscussionReferences |
|---|
Background The relevance of body mass index (BMI) to cause-specific mortality in old age is uncertain.
Objectives To examine cause-specific 5 year mortality in old age by BMI in old age and middle age (40–69 years).
Methods Cox proportional hazards for mortality rates among 4862 former male civil servants in relation to quartiles of BMI measured when screened in 1968–70 and when resurveyed in 1997–98 (median age 76 years).
Results The association between all-cause mortality after resurvey and BMI in old age was U-shaped with hazard ratios (HRs) of 1.3 (95% CI 1.1–1.5) for the lightest and heaviest categories relative to the middle two. Among ‘healthy’ men the lightest (<22.7 kg/m2) had greatest all-cause mortality. The heaviest men (>26.6 kg/m2) had increased risk of cardiovascular disease (CVD) mortality in the first two years or for the whole period if never-smokers. Respiratory mortality was inversely associated with BMI in old age [adjusted HR for trend per BMI category increase 0.6 (0.5–0.7)] but cancer mortality lacked a clear pattern. Net gain or loss of 10 kg or more between middle and old age was a strong predictor of all-cause and CVD mortality.
Conclusions The shape of the association between BMI in old age and mortality differs by cause of death. Major weight change over time is a warning signal for higher CVD mortality. Having BMI <22.7 kg/m2 in old age is associated with above-average mortality rates even if apparently healthy.
Keywords Body mass index, aged, aged 80 and over, mortality, body weight change, cardiovascular diseases, respiratory tract diseases
Accepted 26 September 2005
|
Introduction |
|---|
|
TopAbstractIntroductionData sources and methodsResultsDiscussionReferences |
|---|
The shape and strength of mortality associations with body mass index (BMI) in old age are uncertain.1–4 Obesity may be a weaker adverse factor for all-cause mortality5 in old age compared with middle age. U-shaped or inverse J-shaped associations between BMI and all-cause or cardiovascular mortality have been reported among men aged 60 years and over.6–9 Explanations put forward for these shapes include: inclusion of people with prior disease resulting in both low BMI and early death10,11; confounding due to smoking10; an increasing ratio of harmful fat mass to beneficial fat-free mass in old age12; and regression dilution from only having a single measurement.13 Weight loss is associated with higher mortality in old age but the reasons are unclear.11,14 Weight change measured over a prolonged follow-up period may be less biased by health selection than that measured over shorter periods.15 Few studies have examined whether BMI in middle age when added to BMI in old age improves the prediction of cause-specific mortality.
The aims of the present analyses were: (i) to assess associations between BMI category in old age and 5 year mortality for all causes combined, for cardiovascular disease (CVD), for malignant cancer, and for respiratory disease; (ii) to assess how net change in BMI over 30 years affects mortality; and (iii) whether BMI–mortality patterns in old age depend on BMI in middle age.
|
Data sources and methods |
|---|
|
TopAbstractIntroductionData sources and methodsResultsDiscussionReferences |
|---|
Between 1967 and 1970 (baseline), 19 019 male civil servants working in London and aged 40–69 years were screened for cardiorespiratory diseases (described in detail elsewhere).16 A questionnaire included self-reported health and health-related factors, and there was also a clinical examination.
During 1997–98 surviving participants were resurveyed by post and asked to visit their family doctor for measurements of blood pressure, height, and weight, and for a blood sample to be taken. The questionnaire contained questions on medical history, smoking, alcohol consumption, socioeconomic and demographic circumstances, social support, activities of daily living (ADL), current medication use, and self-assessed current health. Reminders were sent to non-responders 3–4 months after the initial mailing; abbreviated questionnaires, used for people who would otherwise not provide information, contained insufficient information for use in this paper. The resurvey design and distributions of cardiovascular risk factors have been described previously.17,18 At baseline the original cohort participants were registered (flagged) with the central National Health Service (NHS) Registries for Great Britain, which provided date and cause of death. This article analyses deaths that took place between resurvey and the end of 2002 and had been notified by May 2003. Over 99% of those taking part in the resurvey could be monitored for mortality. Ethics committees of the participating institutions approved the resurvey protocol and procedures.
The present analyses were confined to those who had BMI measured at baseline and resurvey, were flagged with the NHS Registries, and had sufficient information on explanatory variables. Response rate at the resurvey was 83% (7033/8448) but there were further losses because of incomplete flagging (24 men), missing or implausible BMI measure at resurvey (1818 men), or insufficient information on covariates (329 men), leaving 4862 included in the analyses. Figure 1 shows the reasons for exclusion.
|
BMI at resurvey was calculated from height at baseline rather than at resurvey because the prime interest was in weight change. Quartiles of baseline BMI among those who had measures at both times were created and the same cut-points used for resurvey categories. Standard cut-points were not used since some cells would have been too small when baseline and current BMI were combined. Weight change was classified into five groups to distinguish minimal, moderate, and major change.
Potential confounders from the resurvey were marital status and employment grade at time of leaving the civil service, smoking, and alcohol consumption. Potential mediators were poor physical functioning (derived from the SF-36),18 poor ADL functioning, chest pain or shortness of breath or pain in calves when walking on level ground at the same speed as people of the same age [respectively potential angina, chronic obstructive airways disease (COPD) and claudication], diabetes mellitus, blood pressure, and cholesterol concentrations. As substantial numbers did not answer the symptom and functioning questions separate categories for missing answers were retained.
Underlying cause of death was coded at the Office for National Statistics using ICD-9 classification for notifications before October 2002 and ICD-10 thereafter. New rules on assigning underlying cause of death mean that fewer cases are assigned to respiratory disease mortality under ICD-10 than under ICD-9,19 but only 8% of all deaths in the present analysis were coded to ICD-10.
Analysis strategy and methods
Cox proportional hazards models using linear and quadratic BMI terms were fitted to check the general shape of associations between BMI and mortality. For ease of understanding the relative risks of different levels of BMI, we mainly present results from proportional hazards models with BMI divided into categories. We fitted models adjusted for age (by setting the origin for measurement of time as date of birth and start of follow-up as date of resurvey) then added potential explanatory factors. To assess whether overall results were affected by inclusion of people at high mortality risk for other reasons, additional models were run: (i) for a ‘healthy’ group, and (ii) separately for never-smokers and ever-smokers (only 6% were current-smokers). For the respiratory disease analyses, the ‘healthy group’ criteria were no report of cancer or of shortness of breath when walking on level ground. Otherwise the criteria were: satisfying the criteria of not taking medication for CVD, reporting no diagnosis of heart attack, stroke, angina, diabetes, or cancer, and no symptoms of angina, COPD, or claudication. Change in mortality rates by BMI over time were examined to assess evidence of conditions that might be linked both to BMI and an early death. These analyses were carried out for all-cause mortality and separately for deaths from CVD (ICD9 390–459 or ICD10 I1–I99), cancer (ICD9 140–209, ICD10 C1–C99), and respiratory disease (ICD9 416–519, ICD10 J1–J99). Subsidiary analyses were done for ischaemic heart disease (IHD) (ICD9 4100–4149, ICD10 I20–I25).
To assess whether associations between BMI in old age and mortality were dependent on earlier BMI history, we analysed mortality in relation to BMI at baseline and subsequent weight change and then mortality according to combined BMI in middle age and in old age.
Cox proportional hazards models were used for comparing mortality rates between categories of BMI. Tests for proportionality were carried out using scaled Schoenfeld residuals according to Grambsch and Therneau theory.20 Log-likelihood ratio tests were used to test for statistical significance of factors added to models. Log-linear trends are only shown where departure from linearity was not statistically significant and the trend parameter had a P-value <0.05. Stata version 821 was used throughout.
|
Results |
|---|
|
TopAbstractIntroductionData sources and methodsResultsDiscussionReferences |
|---|
The median interval between the original screening and postal resurvey was 28.8 years and the median post-resurvey follow-up period for men included in the present analyses was 5.4 years (maximum 5.7 years). At the time of the resurvey the median age of men in this subset was 76.3 years, (range 67.5–96.4 years). The mean BMI at baseline was 24.5 kg/m2 (SD 2.7) and at resurvey 24.2 kg/m2 (SD 3.1). The median (interquartile) weight change was a loss of 1 kg (a loss of 6 kg to a gain of 4 kg).
In 1997–98 the men who had been measured for BMI at both time points were predominantly married, ex-smokers (64%), and moderate drinkers (56% having 1–14 drinks a week). Exertional symptoms, difficulties with ADL, and poor physical performance were reported by <15% of men (Table 1). The lightest group were of older age, and had greater prevalence of total cholesterol <4.8 mmol/l and of current smoking. Men with the highest BMI had greater prevalence of diastolic blood pressure of 95 mm Hg or more, of pain in the calves, and of poor physical functioning. Those in the second BMI category (22.7–24.1 kg/m2) were least likely to report poor physical performance, inability to do an ADL, or breathlessness. There was a negative association between age or prevalence of low cholesterol levels and weight gain and a positive one for prevalence of high diastolic blood pressure. Men with minimal net weight change had the best overall cardiovascular risk profile.
|
The 2171 men who were excluded from the analyses were older (median age 77.6 years) than those who were included. Comparing baseline characteristics, they were more likely to have been in clerical or manual grades (19% compared with 9% of those included), and more likely to have been smokers (35% compared with 29%). However, they were no more likely to have had signs of heart disease, and the mean BMI, 24.6 kg/m2 (SD 2.9), was similar to those of the included men.
There were 1075 deaths among the men included in the analyses, a mortality rate of 45 per 1000 person years (95% CI 42–48). Of these deaths 44% were for CVD, 28% for cancers and 15% for respiratory disease, and two men had unknown cause of death. The resurvey participants who were excluded from the analyses had twice the mortality rate of those who were included (RR 2.0; 95% CI 1.8–2.1).
Associations between BMI in old age and mortality
The middle BMI groups had the lowest all-cause mortality, the overall shape being a U in the model only adjusted for age, and reverse J after further adjustment for the biological and behavioural covariates (Table 2). Analyses using continuous BMI (data not shown) confirmed the quadratic shape. Of the sample, 26% were in the ‘healthy’ subgroup; their age-standardized mortality rates were about half that of the total group at 24 per 1000 person years (95% CI 21–29). In this group the mortality advantage of the second lightest group is emphasized. Further adjustment for total cholesterol, diastolic blood pressure, and diabetes did not alter the parameters. There is a J-shape for CVD mortality but the association is not statistically significant. After adjustment the association between BMI and respiratory mortality could be represented by a negative log-linear trend per category of BMI. In all analyses the association between BMI and cancer mortality was irregular and not statistically significant (data not shown)—for the adjusted model the hazard ratios (HRs) with respect to the second BMI group were 0.9 (95% CI 0.6–1.2), 0.7 (95% CI 0.5–1.0), and 1.0 (95% CI 0.7–1.4).
|
There were no statistically significant interactions between smoking status and BMI with respect to mortality. Although there was some indication among never-smokers that being in the heaviest groups carried excess all-cause and CVD mortality risk. Never-smokers did not have an advantage over ever-smokers among the heaviest group [crude CVD death rates (95% CI) 22 (16–31) and 21 (17–25)] whereas they did among other BMI groups—for example, the crude CVD death rates in the lightest group were 19 (14–25) per 1000 person years and 24 (20–28), respectively.
The inverse association between BMI category and mortality from respiratory disease persisted through all subgroups.
The mortality rates during the first 2 years after resurvey were very low and increased from the third year onwards except for CVD mortality among the heaviest group (Table 3). There was a statistically significant interaction between BMI and time of follow-up for CVD mortality, with a flattening of the CVD mortality curve for the heaviest group in the later period. Analysis of Schoenfield residuals (data not shown) rejected the hypothesis of proportional hazards for circulatory disease in the healthy and ever-smokers groups. Numbers of deaths were too small for separate analysis by time periods within subgroups. In the lower half of Table 3 there is a suggestion that the increase in cardiovascular mortality rates was greatest for those who had lost most weight but the interaction between time and weight loss with respect to mortality was not statistically significant.
|
Subsidiary analyses for 270 IHD deaths showed higher death rates for the heaviest group relative to the second lightest in the whole sample, among ever-smokers and never-smokers (HR of 1.4, 1.5, and 1.3, respectively, after adjustment for other factors) but none of these associations were statistically significant. Only 29 IHD deaths occurred among the ‘healthy’ group. Although interaction between BMI and time of follow-up was not statistically significant there appeared to be excess IHD mortality among the heaviest group during the first 2 years but not thereafter (data not shown).
Mortality relative to combinations of BMI in middle age and subsequent weight change
There were no statistically significant interactions between BMI in middle age and weight change in their associations with mortality (Table 4). Change in BMI dominated prediction of CVD mortality such that BMI at baseline ceased to be statistically significant once change in BMI was added to the age-adjusted model. On the other hand, there was a positive association with baseline BMI and the subset of IHD deaths (data not shown) and no association with weight change. Adjustment for potential baseline explanatory factors modestly attenuated the HRs for substantial loss and gain in weight with respect to all-cause and CVD mortality. Further adjustment for disability and exertional symptoms reduced the hazards ratio for the group with weight gain 
10 kg with respect to CVD mortality to 1.5 (95% CI 1.1–2.2). For respiratory mortality, both being light in middle age and losing weight accompany greatly increased mortality rates.
|
Table 5 shows (across the rows) that the shapes of the associations between BMI in old age and both all-cause and CVD mortality depend on earlier BMI. The reference group is those in the lightest category at both surveys, as this group had the largest number of deaths. The net effect of weight change and BMI in middle age is that men with low BMI at one time and high BMI three decades earlier or later had the highest all-cause and CVD mortality rates, even after adjustment for covariates. This persisted after 2 years (table available on request). Men in the middle categories at both time points had the lowest mortality rates. For all-cause and CVD mortality there was a 4-fold variation in mortality rate between those experiencing the lowest-risk and those experiencing the highest-risk combination of BMIs in middle age and old age. The differing shapes of the associations between old-age BMI and all-cause mortality or CVD mortality according to BMI in middle age are illustrated in Figure 2a and b. These are derived from age-adjusted models using linear and quadratic terms of continuous old-age BMI allowing for interaction with category of BMI in middle age. Although weight loss had a high HR for respiratory mortality, BMI in old age dominated and knowing BMI in middle age did not add to the predictive power (P = 0.10).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionData sources and methodsResultsDiscussionReferences |
|---|
The differences in patterns by broad cause of death demonstrate that the shape of the associations of BMI in old age with all-cause mortality depend on the distribution of deaths by cause.1 There were no positive trends between BMI in old age and any of the examined causes of mortality. Lower than average weight is not altogether beneficial in old age because those with BMI < 22.7 kg/m2 were most at risk of respiratory disease mortality and among the ‘healthy’ subgroup had the highest risk of all-cause mortality, their risk of death being 65% greater than that for men with BMI between 22.7 and 24.4 kg/m2. Men in the heaviest category in old age (>26.1 kg/m2) had double the risk of CVD mortality compared with the same reference group during the first 2 years and nearly 50% increase averaged over the whole period if they were never-smokers. A substantial net increase in BMI during a 30 year period was associated with increased risk of all-cause and CVD mortality and a substantial net decrease with increased risk of all-cause, CVD, and respiratory mortality. As a result the lowest post-resurvey mortality rates occurred among men who had BMI of 22.7–24.1 kg/m2 in middle age and of 24.4–26.1 kg/m2 in old age. IHD deaths showed a different pattern to total CVD such that high BMI in middle age seemed to be the most important weight-related risk factor for mortality in old age.
The most consistent finding is of relatively high mortality from respiratory disease amongst those who were in the lightest group in middle age or old age. The lack of a clear picture for cancer mortality may arise from the heterogeneity of this group of diseases. As deaths accumulate, separate analyses by type of cancer, as has been done for the 30 year period since baseline,22 may reveal different patterns.
Comparisons between Models 1 and 2 in Table 2 and of deaths in two time periods are consistent with the heavier group developing CVD by the time of resurvey. Moreover, those who were in the heaviest group by old age were less likely to be never-smokers (26% compared to 32% of lighter men, P < 0.001) and were less likely to be in the ‘healthy’ subgroup (23% compared with 29% of lighter men, P = 0.001). IHD deaths in the immediate follow-up to the resurvey also appear to be foreshadowed by pre-existing symptoms and disease, as evidenced by small numbers of deaths in the ‘healthy’ group and a death rate in years 3–5 among the heaviest group that was no greater than in the first 2 years (data not shown). This does not rule out a role for BMI in the development or progression of these diseases. Poor functioning and symptoms experienced while walking appeared to play a small part in the pathway between gain in BMI and CVD mortality.
Separate results for ever-smokers and never-smokers tentatively suggest that for ever-smokers the risks due to smoking per se were more important than BMI, but for never-smokers BMI was important in CVD mortality risk. However, the interaction between smoking status and BMI was not statistically significant.
Limitations of the study
The low mortality rates during the first 2 years of follow-up and subsequent steep increase suggest some health selection. An average mortality rate of 50.7 per 1000 person years among those aged 75–84 at resurvey is substantially <73.6 per 1000 in England and Wales as a whole.23
The sequence of events for weight change and disease-onset was not directly measured. However, reverse causation seems less likely given that BMI change continued to predict mortality in the third–fifth years of follow-up. The men excluded from the analysis had double the mortality of those included, which may mean that the analysis group excluded many of those whose BMI had resulted from their health state. We could not differentiate impact according to how fast or at what stage of life weight was lost or gained.24,25
We could not assess the disadvantages of obesity as only 4% had a BMI over 30 kg/m2 compared with 16% in the 1998 Health Survey for England26 or 14.0% in the MRC Trial of the Assessment and Management of Older People in the Community (personal communication). In a Finnish study a persistently high risk was found among men with BMI of 31 kg/m2 or more1 and others have noted that mild to moderate overweight is not a risk factor for CVD and all-cause mortality among older people.4,27
BMI was the only anthropometric measure available and may be a poor indicator of important factors that account for differences in mortality risk. A constant BMI can mask an increasing fat mass28 and concentration in the abdomen.5 Those who are thin, especially those who have lost weight, could have lost lean muscle.29 This may be part of the explanation of increased CVD mortality risk both with weight loss and with weight gain. Knowing the participants' physical activity could have enhanced our understanding. On the other hand, conventional ‘overweight’ BMI may not be so high risk in old age because of reduced lipolysis in omental fat in old age.29 We used height in middle age to capture weight change more precisely but height shrinkage in old age further dilutes the usefulness of BMI.
Consistency with other research
U-shaped6 or reverse J-shaped11 relationships between BMI and all-cause mortality have been reported even after 15 years among cohorts aged 70 years at baseline.30 Risks for lighter men are not entirely due to recent weight loss, as shown by 5 year mortality among older male health professionals who had not experienced weight loss >10 pounds in 5 years.31 The few papers reporting BMI in relation to cancer in old age are not clear-cut.32,33 Two studies have reported on respiratory mortality in old age and noted increased risk amongst the lightest.31,33
The present study provides unique comparisons of mortality among survivors into old age in relation to BMI nearly 30 years previously or in relation to net BMI change over that period. Consistent with our findings, one long-term study showed a U-shape for never-smokers and ex-smokers2 but another had a linear association between BMI and all-cause mortality among never-smokers.34 Weight loss was associated with increased all-cause or CVD mortality over periods of 10–20 years even after excluding the early years.35–37
Further work needs to be done on the timing of, and reasons for, weight change using surveys with more frequent measurements and then meta-analyses carried out. Having experienced substantial weight loss or weight gain by old age appears to increase the risk of CVD mortality and possibilities of preventative action should be explored by looking at pathways through nutrition, physical activity, blood pressure, and functioning. BMI should be supplemented by other measures that capture the relevant physiological changes more accurately. In particular below-average weight is not by itself wholly beneficial and muscle mass and respiratory fitness also need to be considered.
A.E.F. and E.B. thought up the analyses in this paper. E.B. prepared the resurvey and mortality data, undertook the analyses and wrote the draft. R.C. took the lead in preparing the fieldwork for the Resurvey. M.J.S. prepared the baseline data. All authors commented on drafts and agreed on the manuscript. There are no conflicts of interest in this report.
KEY MESSAGES
|
|
Acknowledgments |
|---|
The British Heart Foundation supported the Whitehall Resurvey, including the salary of E.B. M.G.M. is funded by a Medical Research Council research professorship; the British Heart Foundation funded M.J.S. We are indebted to Rory Collins, Pamela Linksted, Paul Sherliker, and Savita Shah for their help in carrying out the survey at CTSU and to David Leon for his encouragement.
|
References |
|---|
|
TopAbstractIntroductionData sources and methodsResultsDiscussionReferences |
|---|
1 Rissanen A, Heliovaara M, Knekt P, Aromaa A, Reunanen A, Maatela J. Weight and mortality in Finnish men. J Clin Epidemiol 1989;42:781–89.[CrossRef][Web of Science][Medline]
2 Visscher TL, Seidell JC, Menotti A et al. Underweight and overweight in relation to mortality among men aged 40–59 and 50–69 years: the Seven Countries Study. Am J Epidemiol 2000;151:660–66.
3 Strawbridge WJ, Wallhagen MI, Shema SJ. New NHLBI clinical guidelines for obesity and overweight: will they promote health? Am J Public Health 2000;90:340–43.
4 Heiat A, Vaccarino V, Krumholz HM. An evidence-based assessment of federal guidelines for overweight and obesity as the apply to elderly persons. Arch Intern Med 2001;161:1194–203.
5 Elia M. Obesity in the elderly. Obes Res 2001;9 (Suppl 4):244S–8S.[Medline]
6 Allison DB, Gallagher D, Heo M, Pi-Sunyer FX, Heymsfield SB. Body mass index and all-cause mortality among people age 70 and over: the Longitudinal Study of Aging. Int J Obes Relat Metab Disord 1997;21:424–31.[CrossRef][Web of Science][Medline]
7 Cornoni-Huntley JC, Harris TB, Everett DF et al. An overview of body weight of older persons, including the impact on mortality. The National Health and Nutrition Examination Survey I—Epidemiologic Follow-up Study. J Clin Epidemiol 1991;44:743–53.[CrossRef][Web of Science][Medline]
8 Heitmann BL, Erikson H, Ellsinger BM, Mikkelsen KL, Larsson B. Mortality associated with body fat, fat-free mass and body mass index among 60-year-old swedish men-a 22-year follow-up. The study of men born in 1913. Int J Obes Relat Metab Disord 2000;24:33–37.[CrossRef][Web of Science][Medline]
9 Tayback M, Kumanyika S, Chee E. Body weight as a risk factor in the elderly. Arch Intern Med 1990;150:1065–72.
10 Manson JE, Stampfer MJ, Hennekens CH, Willett WC. Body weight and longevity. A reassessment. J Am Med Assoc 1987;257:353–58.
11 Losonczy KG, Harris TB, Cornoni-Huntley J et al. Does weight loss from middle age to old age explain the inverse weight mortality relation in old age? Am J Epidemiol 1995;141:312–21.
12 Allison DB, Faith MS, Heo M, Kotler DP. Hypothesis concerning the U-shaped relation between body mass index and mortality. Am J Epidemiol 1997;146:339–49.
13 Greenberg JA. Biases in the mortality risk versus body mass index relationship in the NHANES-1 Epidemiologic Follow-Up Study. Int J Obes Relat Metab Disord 2001;25:1071–78.[CrossRef][Web of Science][Medline]
14 Reynolds MW, Fredman L, Langenberg P, Magaziner J. Weight, weight change, mortality in a random sample of older community-dwelling women. J Am Geriatr Soc 1999;47:1409–14.[Web of Science][Medline]
15 Dyer AR, Stamler J, Greenland P. Associations of weight change and weight variability with cardiovascular and all-cause mortality in the Chicago Western Electric Company Study. Am J Epidemiol 2000;152:324–33.
16 Reid DD, Brett GZ, Hamilton PJ, Jarrett RJ, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle-aged male Civil Servants. A study of screening and intervention. Lancet 1974;1:469–73.[Web of Science][Medline]
17 Clarke R, Breeze E, Sherliker P et al. Design, objectives, and lessons from a pilot 25 year follow up re-survey of survivors in the Whitehall study of London Civil Servants. J Epidemiol Community Health 1998;52:364–69.[Abstract]
18 Breeze E, Fletcher AE, Leon DA, Marmot MG, Clarke RJ, Shipley MJ. Do socioeconomic disadvantages persist into old age? Self-reported morbidity in a 29-year follow-up of the Whitehall Study. Am J Public Health 2001;91:277–83.[Abstract]
19 Office for National Statistics. Changes in the underlying causes of death and modification rules. 2002. Available at: http://www.statistics.gov.uk/about/classifications/icd10/changes_underlying_cause.asp (Accessed September 8, 2005).
20 Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994;81:515–26.
21 Statacorp. Stata Statistical Software: Release 8.0. College Station, TX, Stata Corporation, 2003.
22 Batty GD, Shipley MJ, Jarrett RJ, Breeze E, Marmot MG, Smith GD. Obesity and overweight in relation to organ-specific cancer mortality in London (UK): findings from the original Whitehall study. Int J Obes Relat Metab Disord 2005;29:1267–74.
23 Office for National Statistics. Table 6.1. Deaths: age and sex. England and Wales. Health Stat Q 2003;20:56.
24 Andres R, Muller DC, Sorkin JD. Long-term effects of change in body weight on all-cause mortality. A review. Ann Intern Med 1993;119:737–43.
25 Pamuk ER, Williamson DF, Madans J, Serdula MK, Kleinman JC, Byers T. Weight loss and mortality in a national cohort of adults, 1971–1987. Am J Epidemiol 1992;136:686–97.
26 Boreham R, Erens B, Falaschetti E, Hirani V, Primatesta P. Risk factors for cardiovascular disease. Ch 3 in Erens B, Primatesta P (eds): Health Survey for England 1998. Cardiovascular Disease. Vol 1 Findings. Norwich: TSO, 1999.
27 Heiat A. Impact of age on definition of standards for ideal weight. Prev Cardiol 2003;6:104–07.[Medline]
28 Prentice AM, Jebb SA. Beyond body mass index. Obes Rev 2001;2:141–47.[CrossRef][Medline]
29 Seidell JC, Visscher TL. Body weight and weight change and their health implications for the elderly. Eur J Clin Nutr 2000;54 (Suppl 3):S33–39.
30 Dey DK, Rothenberg E, Sundh V, Bosaeus I, Steen B. Body mass index, weight change and mortality in the elderly. A 15 y longitudinal population study of 70 y olds. Eur J Clin Nutr 2001;55:482–92.[CrossRef][Web of Science][Medline]
31 Baik I, Ascherio A, Rimm EB et al. Adiposity and mortality in men. Am J Epidemiol 2000;152:264–71.
32 Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003;348:1625–38.
33 Takala JK, Mattila KJ, Ryynanen OP. Overweight, underweight and mortality among the aged. Scand J Prim Health Care 1994;12:244–48.[Medline]
34 Singh PN, Lindsted KD, Fraser GE. Body weight and mortality among adults who never smoked. Am J Epidemiol 1999;150:1152–64.
35 Pamuk ER, Williamson DF, Serdula MK, Madans J, Byers TE. Weight loss and subsequent death in a cohort of U.S. adults. Ann Intern Med 1993;119:744–48.
36 Rhoads GG, Kagan A. The relation of coronary disease, stroke, and mortality to weight in youth and in middle age. Lancet 1983;1:492–95.[CrossRef][Web of Science][Medline]
37 Higgins M, D'Agostino R, Kannel W, Cobb J, Pinsky J. Benefits and adverse effects of weight loss. Observations from the Framingham Study. Ann Intern Med 1993;119:758–63.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. W. Auyeung, J. S. W. Lee, J. Leung, T. Kwok, P. C. Leung, and J. Woo Survival in Older Men May Benefit From Being Slightly Overweight and Centrally Obese--A 5-Year Follow-up Study in 4,000 Older Adults Using DXA J Gerontol A Biol Sci Med Sci, January 1, 2010; 65A(1): 99 - 104. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. E. Yun, H. S. Park, Y.-M. Song, and S.-I. Cho Increases in body mass index over a 7-year period and risk of cause-specific mortality in Korean men Int. J. Epidemiol., September 17, 2009; (2009) dyp282v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Umberson, H. Liu, and D. Powers Marital Status, Marital Transitions, and Body Weight Journal of Health and Social Behavior, September 1, 2009; 50(3): 327 - 343. [Abstract] [PDF]
|
|
|
|
 |
|
 C Giannini, T de Giorgis, A Scarinci, I Cataldo, M L Marcovecchio, F Chiarelli, and A Mohn Increased carotid intima-media thickness in pre-pubertal children with constitutional leanness and severe obesity: the speculative role of insulin sensitivity, oxidant status, and chronic inflammation Eur. J. Endocrinol., July 1, 2009; 161(1): 73 - 80. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. Locher, D. L. Roth, C. S. Ritchie, K. Cox, P. Sawyer, E. V. Bodner, and R. M. Allman Body Mass Index, Weight Loss, and Mortality in Community-Dwelling Older Adults J Gerontol A Biol Sci Med Sci, December 1, 2007; 62(12): 1389 - 1392. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. M. Flegal, B. I. Graubard, D. F. Williamson, and M. H. Gail Cause-Specific Excess Deaths Associated With Underweight, Overweight, and Obesity JAMA, November 7, 2007; 298(17): 2028 - 2037. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S G. Wannamethee, A G. Shaper, L. Lennon, and P. H Whincup Decreased muscle mass and increased central adiposity are independently related to mortality in older men Am. J. Clinical Nutrition, November 1, 2007; 86(5): 1339 - 1346. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. H.N. Nguyen, A. J. Wilcox, R. Skjaerven, and D. D. Baird Men's body mass index and infertility Hum. Reprod., September 1, 2007; 22(9): 2488 - 2493. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. D. Batty, M. Kivimaki, G. D. Smith, M. G. Marmot, and M. J. Shipley Obesity and Overweight in Relation to Mortality in Men With and Without Type 2 Diabetes/Impaired Glucose Tolerance: The original Whitehall Study Diabetes Care, September 1, 2007; 30(9): 2388 - 2391. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||