jurnal kohort survival

Lifetime Weight Patterns in Male Physicians:The Effects of Cohort and Selective SurvivalBethany B. Barone,* Jeanne M. Clark,* Nae-Yuh Wang, Lucy A. Meoni, Michael J. Klag,* andFrederick L. Brancati*

AbstractBARONE, BETHANY B., JEANNE M. CLARK, NAE-YUH WANG, LUCY A. MEONI, MICHAEL J. KLAG,AND FREDERICK L. BRANCATI. Lifetime weightpatterns in a cohort of male physicians: an analysis of cohorteffects and selective survival in the Johns HopkinsPrecursors Study. Obesity. 2006;14:902908.Objective: The natural history of lifetime weight change isnot well understood because of conflicting evidence fromcross-sectional and longitudinal studies. Cross-sectionalanalyses find that adult weight is highest at 60 years ofage and lower thereafter. Longitudinal analyses have notfound this pattern. Our objective was to test whether cohorteffects and selective survival may explain the differencesobserved between cross-sectional and longitudinal studies.Research Methods and Procedures: We analyzed data onwhite men from the Johns Hopkins Precursors Study (n 1197). Weight and height were measured at enrollmentduring medical school. The Precursors Study collected sub-sequent weight measurements by self-report and follows allparticipants for mortality.Results: In preliminary analyses that ignored cohort andsurvival effects, average weight increased 0.16 kg/yr to age65 (p 0.001) and declined 0.10 kg/yr thereafter (p 0.002). When controlling for differing rates of weightchange by cohort and survival group, the apparent declineafter 65 years of age was mostly explained.Discussion: These data suggest that, in white men, weightincreases steadily until age 65 and then plateaus. These

findings emphasize the necessity of longitudinal rather thancross-sectional data to describe lifetime weight patterns.

Key words: aging, body weight changes, cohort effect,mortality, prospective studies

IntroductionObesity is one of the most significant public health prob-

lems facing the United States today (13). High BMI isassociated with greater risk of cardiovascular diseases (2),type 2 diabetes (4), cancer (5), and all-cause mortality (6,7).Although reducing obesity was a major goal of HealthyPeople 2000 (information regarding Healthy People 2000is available at http://www.cdc.gov/nchs/about/otheract/hp2000/hp2k.htm), the prevalence of obesity increased inevery state during the 1990s (3). By 2000, the prevalence ofoverweight (BMI 25 kg/m2) in the United States hadreached 64%, and the prevalence of obesity (BMI 30kg/m2) had reached a staggering 30% (8). Understandingthe natural history of obesity is important for developingeffective treatment and prevention programs.

Cross-sectional evidence from the National Health andNutritional Examination Survey (NHANES)1 as well asother studies has suggested that the natural pattern of weightchange is an increase in weight until the seventh decade oflife followed by a decline in weight thereafter (911). Thispattern was also observed when NHANES repeated weightmeasurements 10 years after the first weight measurement(12). However, several longitudinal studies of weight pat-terns have found sustained weight gain past the seventhdecade of life as opposed to a decline (13,14). We hypoth-esized that the inverted U trajectory in body weight ob-served in cross-sectional analyses is not the true lifetimepattern of weight change but rather the result of severalepidemiological effects. First, increases in the prevalence ofobesity over the past few decades may be creating a birth

Received for review February 1, 2005.Accepted in final form March 7, 2006.The costs of publication of this article were defrayed, in part, by the payment of pagecharges. This article must, therefore, be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.*Department of Epidemiology, Department of Medicine, Department of Biostatistics, andDepartment of Health Policy and Management, The Johns Hopkins Bloomberg School ofPublic Health and The Johns Hopkins University School of Medicine, Baltimore, Maryland.Address correspondence to Jeanne M. Clark, 2024 E. Monument Street, Suite 2-600,Baltimore, MD 21287.E-mail: [email protected] 2006 NAASO

1 Nonstandard abbreviation: NHANES, National Health and Nutritional Examination Sur-vey.

902 OBESITY Vol. 14 No. 5 May 2006

cohort effect where individuals in earlier cohorts have loweraverage weight (8). Thus, a cross-sectional analysis wouldshow a lower weight among the elderly group relative toyounger groups. Second, the relationship between obesityand all-cause mortality may be creating a survival effect inwhich those who die earlier are heavier, on average, thanthose who live longer (11). Therefore, the mean weight atsuccessive ages would decrease as a greater proportion ofthe heavier individuals are removed from the samplinggroup. A final possibility is that disease-related weight losscauses a decrease in average weight among the elderly (15).To our knowledge, no other analysis has looked at bothcross-sectional weight patterns and the contribution of lon-gitudinal epidemiological effects in the same population.

Research Methods and ProceduresSetting and Sample Description

Caroline Bedell Thomas, MD, initiated the Johns Hop-kins Precursors Study in 1947 to prospectively study pre-cursors of cardiovascular disease. The study recruited a totalof 1337 Johns Hopkins medical students who graduatedbetween 1948 and 1964. Baseline information thought to berelevant to cardiovascular disease was measured at the timeof enrollment by a thorough medical history and physicalexam. Baseline height was measured using a stadiometer,and weight was measured using a balance beam scale.Subsequent collection of body weight data occurred inter-mittently throughout follow-up by self-report on mailedquestionnaires (every 5 years after graduation from 1948 to1965, annually from 1966 to 1985, and in 1988, 1989, 1993,1998, 2001, 2002, and 2003). The Precursors Study contin-ues today with annual questionnaires and follow-up fordisease outcomes and death, including National Death In-dex searches. Vital status is known for 99% of the studypopulation. The study has excellent retention, with responserates of at least 85% for every 5-year follow-up interval.Further information about the Johns Hopkins PrecursorsStudy can be found in more than 150 publications (16). Allstudy procedures were approved by the Institutional ReviewBoard (Johns Hopkins Medical Institution, Baltimore, MD).

The study population for this analysis includes white,male participants of the Johns Hopkins Precursors StudyCohort (n 1197). Women (n 121) and non-white men(n 19) were not included because these subgroups weretoo small to analyze independently. Of the remaining par-ticipants, those without a baseline height or weight mea-surement (n 23) were excluded. Participants with baselineweight measurements but no subsequent reported weights(n 38) were also excluded because we were specificallyinterested in studying weight trajectories over time. Partic-ipants in the final study population had a variable number offollow-up weight responses ranging from 1 to 25, with anaverage of 15. The distribution of weight responses was the

result of a function of individual response rate, year ofenrollment in the study, and the design of weight datacollection by the Precursors Study. The year for each weightobservation was recorded as the year the individual wasmeasured at baseline or the year a questionnaire was re-turned by the participant to the Precursors Study. The finalstudy population consisted of 1136 (95% of the entirecohort) white men with both baseline anthropometric mea-surements and at least one follow-up weight observation.

Statistical MethodsDescriptive statistics (mean, standard deviation, median),

including birth year, baseline age, baseline BMI, and num-ber of weight observations per individual, were calculatedfor the study population. All-cause mortality was also cal-culated. The full study population was separated by birthyear into three cohorts of approximately equal sample size,and the same statistics were calculated for each cohort.Average length of follow-up was calculated. The cross-sectional, running mean of BMI on age using unweightedkernal smoothing (17) was plotted for the calendar years1970, 1985, and 2002 to establish the cross-sectional patternof BMI with age seen in our study population. These plotsincluded all members of the Johns Hopkins PrecursorsStudy who reported a weight in the given calendar year, anda given individual could contribute to the plot for up to 3calendar years. These calendar years were arbitrarily chosento show cross-sectional plots across different age ranges ofour entire study population.

Generalized estimating equations multiple linear regres-sion models with robust estimation of variance were used totest the association of age, birth cohort, and survival withattained weight adjusted for height squared. We chose at-tained weight as the outcome measure and adjusted forheight squared as an independent variable in all models. Wefelt that changes in weight would be easier to interpret thanchanges in BMI. We fit three models. In Model 1, we usedcontinuous age with a linear spline at age 65, adjusting forheight squared as a predictor of attained weight. This modelrepresents evaluating the data without considering birthcohort effects and survival effects. The purpose of the linearspline was to determine whether weight declines after thisage as suggested by cross-sectional analyses (911). InModel 2, we used continuous age with spline, birth cohort(1, 2, and 3), and survival status (alive/dead) as predictors ofattained weight, adjusting for height squared. Survival sta-tus was categorized by the reported condition (alive/dead)of an individual in December 2002. Our second modeladded the main effects of cohort and survival status to themodel. In Model 3, we added birth cohort-by-age and sur-vival status-by-age interactions. Our third model studiedwhether rates of weight change differed by cohort andsurvival status. Because illness preceding death may alsocause weight loss in the elderly, we repeated all models

Lifetime Weight Patterns in Male Physicians, Barone et al.

OBESITY Vol. 14 No. 5 May 2006 903

excluding any weight data reported in the 3 years before anindividuals death. In all models, age was centered at themedian baseline age (22 years) rather than age zero to allowfor main effects to be compared at a baseline age of 22rather than a baseline age of 0. A random effects model wasused for all analyses to allow for individual variability.Variograms of the model residuals vs. age were generatedfor a random sample of the study population to determinewhether the exchangeable correlation structure was appro-priate.

The STATA 7 Statistical Package (StataCorp LP., Col-lege Station, TX) was used for data analysis.

ResultsSelected characteristics of the study population are re-

ported for all participants and by cohort (Table 1). Atenrollment, the mean age was 23 years, and the averageBMI was 23 kg/m2. By December of 2002, approximatelyone fifth of the cohort was reported to have died. Medianbirth year, age at enrollment, and mortality differed signif-icantly across birth cohorts. There was no statistical differ-ence in number of follow-up observations by cohort. Base-line BMI did not differ significantly by the number offollow-up observations. Average length of follow-up was42 11 (standard deviation) years.

Cross-sectional plots of BMI by age in 3 different calen-dar years showed that average BMI increased and thendecreased as age increased (Figure 1). The average BMI ofthe study population at a given age was generally greater inlater calendar years.

In Model 1 (Table 2), we observed the expected result ofa yearly increase in average weight (0.16 kg/yr, p 0.001)before the spline at age 65 and a yearly decrease in averageweight (0.10 kg/yr, p 0.002) thereafter. This analysis, inwhich cohort and survival effects are not considered, showsa basic longitudinal weight trajectory, the inverted U.

In Model 2 (Table 3), which added birth cohort andsurvival status, the average rate of weight change did notchange from Model 1. At age 22 (where age was centered),average weight did not differ significantly by cohort. How-ever, average weight was 1.78 kg higher (p 0.05) at age22 in men who died by the end of follow-up compared withmen who survived through 2002. This analysis evaluatedsimple overall effects of cohort and survival group but wasnot optimal because it did not allow for the cohort andsurvival effects to vary with age (before and after the splineset at age 65), as these types of effects likely would in atrajectory.

In Model 3 (Table 4), we added age-by-birth cohort andage-by-survival status interactions to assess whether thesurvival and cohort effects were better modeled as rates ofchange with age rather than just a one-time effect. In thismodel, the reference group included men in the first cohortwho were alive at the end of follow-up. Their lifetime

Table 1. Selected characteristics of white men in the full study population and by birth cohort in the JohnsHopkins Precursors Study, Baltimore, MD, 1948 to 2002

Cohort

Overall 1902 to 1927 1928 to 1933 1934 to 1941

N 1136 431 363 342Median birth year* 1930 1925 1930 1936Mean baseline age (years)* 23.1 (2.6) 24.7 (3.3) 22.3 (1.4) 21.9 (1.0)Mean baseline BMI (kg/m2) 23.1 (2.6) 23.2 (2.7) 23.1 (2.5) 23.1 (2.5)Median number of weight observations 16 (6) 16 (7) 16 (6) 15 (6)Cumulative mortality by December 2002 (%)* 19 34 15 6

* Characteristic differed significantly (p 0.05) when compared across cohorts using ANOVA. Values are means (standard deviation).

Figure 1: Cross-sectional plot of running mean of BMI on ageusing unweighted kernel smoothing in 1970, 1985, and 2002 in theJohns Hopkins Precursors Study.



weight pattern was described by the coefficients seen underage. This pattern continued to be a highly significant rate ofweight gain with age up to age 65 (0.14 kg/yr, p 0.001).However, we no longer observed a significant decline inweight beyond age 65. There was no significant differencein weight at age 22 by survival status or cohort. The maineffects of cohort and survival status alone are not statisti-cally significant in this model after adjusting for the inter-action terms. However, we did identify two important in-teraction effects in our study population. First, we found acohort-age effect, in that individuals in the most recent birthcohort had a statistically greater rate of weight gain up toage 65 compared with the oldest birth cohort. Second, we

found a survival-age effect, in that there was a highlysignificant rate of decline in average weight among menover age 65 who died by December 2002 compared withmen alive at the end of follow-up. The main or simpleeffects of cohort and survival status were attenuated in thethird model because the significant effects were rates ofchange with age (the interaction terms). Model 3 betteradjusted for the cohort and survival effects and confirmedthe hypothesis that the weight decline observed in cross-sectional studies after age 65 is significantly attenuated onceproperly modeled survival and cohort effects are removed.

To understand how weight varies with age in the groups ofmen other than the reference group for Model 3, the coefficientof weight change for age must be combined with the interac-tion terms applicable to the group of interest. For example, tounderstand how weight changes up to age 65 for men who aredeceased by the end of follow-up in Cohort 3, the up to age65 coefficient (0.14 kg/yr) would be combined with theage-by-Cohort 3 coefficient (0.05 kg/yr) and the up to age65 deceased coefficient (0.02 kg/yr).

Similar results were observed when all models wererepeated after excluding any weight measurements in the 3years before an individuals death (data not shown).

DiscussionThe results observed in the final and best model (Model

3) showed that, contrary to the impression created by cross-sectional studies such as NHANES III (911), body weight

Table 2. Model 1: rate of weight change by age inthe Johns Hopkins Precursors Study

Covariate(age) Coefficient*

95% confidenceinterval

Up to age 65 0.16 kg/yr 0.15,0.18After age 65 0.10 kg/yr 0.16,0.04

* Age is centered at the median baseline age (22 years), and allcoefficients are adjusted for other covariates in the table andheight2. p 0.001. p 0.01.

Table 3. Model 2: rate of weight change by age and weight differences by cohort and survival status in the JohnsHopkins Precursors Study

Covariate Coefficient*95% confidence

interval

AgeUp to age 65 0.16 kg/yr 0.15, 0.18After age 65 0.10 kg/yr 0.16, 0.04

Birth cohort1 (1902 to 1927) Reference2 (1928 to 1933) 0.05 kg 1.19, 1.293 (1934 to 1941) 0.71 kg 0.65, 2.07

Survival statusAlive on December 31, 2002 ReferenceDeceased before December 31, 2002 1.78 kg 0.24, 3.32

* Age is centered at the median baseline age (22 years), and all coefficients are adjusted for other covariates in the table and height2. p 0.001. p 0.01. p 0.05.



does not generally decrease in individuals after age 65.Rather, the apparent decline is largely explained by twoepidemiological phenomena: 1) a birth cohort effect whereaverage weight increases at a greater rate in the youngestcohort compared with older cohorts and 2) a survival effectby which heavier individuals are removed from the cohortas they die, and, thus, the average weight decreases overtime, especially after age 65. As was expected, the cohorteffect is observed as a significantly greater increase inaverage weight among the youngest cohort up to age 65(age-by-cohort covariate, up to age 65Cohort 3), and asimilar but non-significant trend is seen in men over age 65in more recent cohorts (age-by-cohort covariate, age 65 andafterCohort 2, and age 65 and afterCohort 3). Thesignificantly greater decrease in average weight observed inmen after age 65 who died explains much of the observeddecrease in average weight after age 65 (age-by-survivalstatus covariate, age 65 and afterdeceased). This is likelyan effect of the heaviest individuals being removed from the

group and average weight going down as a result. Disease-related weight loss may also contribute to this effect. Ouranalysis showed the complex nature of the cohort andsurvival effects in that they need to be evaluated longitudi-nally, rather than the cross-sectionally, to truly understandlifetime weight patterns. Furthermore, we must adjust forcohort and survival effects as interaction terms with aging(age-by-cohort, age-by-survival status) and not just simple,main effect covariates (age, cohort, survival status). Onlywhen using these interaction terms, as in Model 3, do we seesignificant effects by both cohort and survival status, as wellas the attenuation of the weight loss effect after age 65.

The greatest strength of our analysis was the prospectivedesign of the Johns Hopkins Precursors Study, which hascollected information on weight and has followed all par-ticipants for mortality for more than 50 years. This rarecollection of longitudinal data is ideal for analyses of bodyweight patterns over time because it allows for the assess-ment of selective mortality and limits variability by the use

Table 4. Model 3: rate of weight change by age, weight differences by cohort and survival status, andage-by-cohort and age-by-survival status interaction terms in the Johns Hopkins Precursors Study

Covariate* Coefficient* 95% confidence interval

AgeUp to age 65 0.14 kg/yr 0.12, 0.17Age 65 and after 0.06 kg/yr 0.14, 0.02

Birth cohort1 (1902 to 1927) Reference2 (1928 to 1933) 0.01 kg 1.17, 1.193 (1934 to 1941) 0.42 kg 1.67, 0.83

Survival statusAlive in 2002 ReferenceDeceased before end of 2002 1.36 kg 0.02, 2.74

Age-by-cohortUp to age 65cohort 1 ReferenceUp to age 65cohort 2 0.00 kg/yr 0.04, 0.03Up to age 65cohort 3 0.05 kg/yr 0.01, 0.09Age 65 and aftercohort 1 ReferenceAge 65 and aftercohort 2 0.11 kg/yr 0.04, 0.23Age 65 and aftercohort 3 0.39 kg/yr 0.16, 0.94

Age-by-survival statusUp to age 65alive ReferenceUp to age 65deceased 0.02 kg/yr 0.03, 0.07Age 65 and afteralive ReferenceAge 65 and afterdeceased 0.36 kg/yr 0.53, 0.18

* Age is centered at the median baseline age (22 years), and all coefficients are adjusted for other covariates in the table and height2. p 0.001. p 0.05.



of repeated measures. Another strength of this analysis isthe fairly large, homogenous sample of white, educatedmen, which reduces variability caused by race, sex, andsocioeconomic status.

Our longitudinal results are generally consistent with thefew previous studies that have been done, although none hashad such a long follow-up. The Tromso Study (13) and theFels Longitudinal Study (14) both observed that BMI didnot decrease with age, even among elderly white men. TheTromso Study also found a faster rate of weight gain amongmore recent cohorts. However, the Tromso Study could notinvestigate the effects of selective survival or weight losspreceding death because participation was limited to indi-viduals attending all follow-up visits (13). Among men whowere still alive at the end of follow-up, we observed anon-significant weight decline of 0.06 kg/yr after age 65after adjusting for cohort and interaction terms. This resultis less than the 0.1-kg weight loss per year observed in acohort of elderly, independent, healthy white men (18).While we were surprised to find a slower rate of weight lossamong elderly men in our analysis, the men who were aliveat the end of follow-up were likely to be a sample of thehealthiest individuals in the Precursors Study cohort.

Despite its strengths, several limitations of this studywarrant discussion. First, most data on body weight wereself-reported. However, these data were prospectively col-lected, and the excellent correlation ( 0.98) of self-reported weight to measured weight has been previouslyshown in this cohort (19). A second potential for bias wasthat individuals excluded because they lacked a baselineweight observation (n 38) had a statistically higher mor-tality rate. However, most deaths in the excluded groupoccurred in young adulthood and, thus, are not likely to berelated to obesity.

Variable follow-up (the number of weight observationsranged from 2 to 26 per person) may also be a source ofbias. One possibility is that individuals with cyclical weightpatterns may be less likely to report a weight when at a peakand more likely to report a weight when in a valley. There-fore, some individuals may be misclassified as having alower average weight.

Additionally, the benefit of a homogenous populationcomes at the cost of generalization. White, educated, malephysicians may be thinner and less likely to gain weightthan the general population. This possibility is supported bythe comparatively slow rate of weight gain we observed(14,20). However, we do not believe that this quantitativedifference would explain the pattern we found.

Despite a range in birth year of 39 years, the differencebetween median birth years by cohort was 6 years. Thedistribution of baseline age and attained age also differed bycohort, mainly in that the oldest cohort has a greater rangethan the two younger cohorts. Both factors may have re-duced the ability to observe the main effects differences by

cohort that have been shown in other analyses (13,2022),although we did observe a significant age-by-cohort inter-action effect.

Increasing longevity may be another source of bias in thisanalysis because we know that later cohorts are heavier butlife expectancy has increased. While the cohort effects weadjusted for may incorporate this effect, a more complexmodel (e.g., including a different spline point for eachcohort) would be needed to take into account lifespan in-creases with time. Furthermore, increasing longevity bycohort may also have prevented us from observing a moredistinct cohort effect.

Finally, information on voluntary vs. involuntary disease-related weight loss, body composition, and body fat distri-bution was not available for this analysis. We did attempt toremove the effects of unintentional weight loss immediatelypreceding death by excluding weight measurements within3 years of death. The results of this analysis were similar,suggesting that our inferences about cohort and survivaleffects were not distorted by weight loss immediately beforedeath.

Our findings have several implications for future re-search. The first is that, even among affluent, well-educated,medically sophisticated men, BMI continues to increaseacross the lifespan at least to age 65 without evidence ofsignificant decline in healthy adults thereafter. Furthermore,BMI increases at a greater rate in successive cohorts, andthe decline of average weight in the older ages seems to bea survival effect. Therefore, we should not be deceived bycross-sectional studies that suggest that weight naturallydeclines after age 65 in individual adults, because thisphenomenon is explained mostly by cohort effects andselective survival. This knowledge offers the opportunity tocontinue efforts and counseling for weight managementacross the healthy adult lifetime. The effects of voluntaryand involuntary disease-related weight loss on expectedlifetime weight trajectories among healthy individuals, therole of body composition and body fat distribution in lon-gitudinal weight patterns with aging, and the potential ben-efit of voluntary weight loss in the elderly await furtherresearch.

AcknowledgmentsThis work was supported by NIH Grants AG01760 and

DK02856. F.B. was supported by National Institute of Di-abetes & Digestive & Kidney Diseases Grant K24DK62222-01. The authors thank Drs. Brad Astor, GretaBunin, and Lawrence Cheskin for helpful suggestions onearlier versions of the manuscript.

References1. Visscher TLS, Seidell JC. The public health impact of obe-

sity. Annu Rev Public Health. 2001;22:35575.



2. Must A, Spadano J, Coakley EH, Field AE, Colditz G,Dietz WH. The disease burden associated with overweightand obesity. JAMA. 1999;282:15239.

3. Nelson DE, Bland S, Powell-Griner E, Klein R, Wells HE,Hogelin G, Marks JS. State trends in health risk factors andreceipt of clinical preventive services among US adults duringthe 1990s. JAMA. 2002;287:265967.

4. Brancati FL, Wang NY, Mead LA, Liang KY, Klag MJ.Body weight patterns from 20 to 49 years of age and subse-quent risk for diabetes mellitus. Arch Intern Med. 1999;159:95763.

5. Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ.Overweight, obesity, and mortality from cancer in a prospec-tively studied cohort of U.S. adults. N Engl J Med. 2003;348:162538.

6. Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CWJr. Body mass index and mortality in a prospective cohort ofU.S. adults. N Engl J Med. 1999;341:1097105.

7. Peeters A, Barendregt JJ, Willekens F, Mackenbach JP, AlMamun A, Bonneux L. Obesity in adulthood and its conse-quences for life expectancy: a life-table analysis. Ann InternMed. 2003;138:2432.

8. Flegal KM, Carroll MD, Ogden CL, Johnson CL. Preva-lence and trends in obesity among US adults, 19992000.JAMA. 2002;288:17237.

9. National Health and Nutritional Examination. PublishedTables. http://www.cdc.gov/nchs/about/major/nhanes/datatblelink.htm (Accessed May 2, 2006).

10. Fogelholm M, Kujala U, Kaprio J, Sarna S. Predictors ofweight change in middle-aged and old men. Obes Res. 2000;5:36773.

11. Seidell JC, Visscher TLS. Body weight and weight changeand their health implications for the elderly. Eur J Clin Nutr.2000;54:S339.

12. Williamson DF. Descriptive epidemiology of body weightand weight change in U.S. adults. Ann Intern Med. 1993;119:6469.

13. Jacobsen BK, Njolstad I, Thune I, Wilsgaard T, LochenML, Schirmer H. Increase in weight in all birth cohorts in ageneral population. Arch Intern Med. 2001;161:46672.

14. Guo SS, Zeller C, Chumlea WC, Siervogel RM. Aging,body composition, and lifestyle: the Fels Longitudinal Study.Am J Clin Nutr. 1999;70:40511.

15. Losonczy KG, Harris TB, Cornoni-Huntley J, SimonsickEM, Wallace RB, Cook NR, Ostfeld AM, Blazer DG. Doesweight loss from middle age to old age explain the inverseweight mortality relation in old age? Am J Epidemiol. 1995;141:31221.

16. Thomas CB. Observations on some possible precursors ofessential hypertension and coronary artery disease. Bull JohnsHopkins Hosp. 1951;89:41941.

17. Stone CJ. Consistent nonparametric regression. Ann Stat.1977;5:595625.

18. Chumlea WC, Garry PJ, Hunt WC, Rhyne RL. Distribu-tions of serial changes in stature and weight in a healthyelderly population. Human Biol. 1988;60:91725.

19. Klag MJ, He J, Mead LA, Ford DE, Pearson TA, LevineDM. Validity of physicians self-reports of cardiovasculardisease risk factors. Ann Epidemiol. 1993;3:4427.

20. Lewis CE, Jacobs DR Jr, McCreath H, et al. Weight gaincontinues in the 1990s: 10-year trends in weight and over-weight from the CARDIA Study. Am J Epidemiol. 2000;151:117281.

21. McTigue KM, Garrett JM, Popkin BM. The natural historyof the development of obesity in a cohort of young U.S. adultsbetween 1981 and 1998. Ann Intern Med. 2002;136:85764.

22. Lahti-Koski M, Jousilahti P, Pietinen P. Secular trends inbody mass index by birth cohort in eastern Finland from 1972to 1997. Int J Obes Relat Metab Disord. 2001;25:72734.



Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

jurnal kohort survival

Documents