Confidential: For Review Only - BMJ · Confidential: For Review Only Weight change across adulthood in relation to all cause and cause specific mortality: a prospective cohort study

Confidential: For Review OnlyWeight change across adulthood in relation to all cause and cause specific mortality: a prospective cohort study of U.S.

adults

Journal: BMJ

Manuscript ID BMJ-2019-050369

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 23-Apr-2019

Complete List of Authors: Chen, Chen; Huazhong University of Science and Technology, School of Public HealthYe, Yi; Huazhong University of Science and Technology, School of Public HealthZhang, Yanbo; Huazhong University of Science and Technology, School of Public HealthPan, Xiong-Fei; Huazhong University of Science and Technology, School of Public HealthPan, An; Huazhong University of Science and Technology, School of Public Health

Keywords: Cohort study, Mortality, Obesity, Weight change

https://mc.manuscriptcentral.com/bmj

BMJ

Confidential: For Review Only

1

Weight change across adulthood in relation to all cause and cause specific

mortality: a prospective cohort study of U.S. adults

Chen Chen MBBS, Yi Ye MBBS, Yanbo Zhang MBBS, Xiong-Fei Pan Postdoc researcher, An Pan

Professor

Affiliations: Department of Epidemiology and Biostatistics, School of Public Health, Tongji Medical

College, Huazhong University of Science and Technology, Wuhan, 430030, China; Key Laboratory of

Environment and Health, Ministry of Education & Ministry of Environmental Protection, School of

Public Health, Tongji Medical College, Huazhong University of Science and Technology, Wuhan,

430030, China

Correspondence: An Pan, Department of Epidemiology and Biostatistics, School of Public Health,

Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430030, China;

telephone and fax numbers: +86-27-83627309; email: [email protected].

Key words: Body weight changes, Young adult, Middle Aged, Aged, Mortality

Word count: 4731

3 main tables, 2 main figures

Supplemental materials: tables S1-S10 and figures S1-S3

Page 1 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


2

What is already known on this topic

High body mass index in adulthood is associated with higher risk of premature deaths.

Scientific evidence about weight change from young to late adulthood or from middle to late

adulthood in relation to risk of mortality was not entirely consistent.

Little is known about associations between weight change across adulthood especially from

young to middle adulthood and all cause and cause specific mortality.

What this study adds

Weight gain from young to middle adulthood was associated with increased risk of mortality from

all cause, cardiovascular disease and cancer in US adults. Weight loss over this period was not

significantly related to mortality. Individuals who remained obese had the highest risk of mortality.

Across adulthood, association between weight gain and mortality attenuated with increasing age

while the association with weight loss became stronger and significant in middle to late adulthood.

Our results highlight the importance of maintaining normal weight across adulthood, especially

preventing weight gain in early adulthood, for preventing premature deaths in later life.

Page 2 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


3

ABSTRACT (word count 399)

OBJECTIVE

To investigate the association between weight changes across adulthood and mortality.

DESIGN

Prospective cohort study.

SETTING

U.S. National Health and Nutrition Examination Survey (NHANES) 1988-1994 and NHANES 1999-

2014.

POPULATION

36 812 participants aged 40 years or over who recalled their weight at young adulthood (at age 25

years) and at middle adulthood (at ten years prior to NHANES survey, mean age of 47 years) and had

measured weight at late adulthood (at baseline NHANES survey, mean age of 57 years).

MAIN OUTCOME MEASURES

All cause and cause specific mortality from baseline until 31 December 2015.

RESULTS

During a mean follow-up of 10 years, 10 936 deaths occurred. Compared with participants remaining

normal body mass index (BMI), those who gained weight from non-obese BMI at young adulthood to

obese BMI at middle adulthood had a 33% (hazard ratio 1.33, 95% confidence interval 1.22 to 1.45),

52% (1.52, 1.28 to 1.81) and 26% (1.26, 1.02 to 1.56) higher risk of all cause mortality, cardiovascular

mortality and cancer mortality respectively; losing from obese to non-obese BMI over this period was

not significantly associated with mortality risk. Similar results were found for the relations of mortality

with weight change from young to late adulthood. Weight loss from middle to late adulthood was

Page 3 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


4

associated with increased risk of all cause mortality (1.44, 1.28 to 1.61) and cardiovascular mortality

(1.66, 1.32 to 2.09). Weight gain with BMI moving from non-obese to obese category over this period

was unrelated to mortality. Maintaining an obese BMI in all three intervals was consistently associated

with increased risk of mortality, while the associations were stronger when stable obesity occurred

from young to middle adulthood (1.93, 1.72 to 2.16) or to late adulthood (1.81, 1.59 to 2.05) compared

to that from middle to late adulthood (1.33, 1.22 to 1.45). Maximum overweight individuals had a very

modest or null association with mortality across adulthood. The association between weight change and

mortality across adulthood was more pronounced among participants who were less than 60 years old

at baseline and never smokers compared to their counterparts.

CONCLUSIONS

Stable obesity across adulthood, weight gain from young to middle adulthood and weight loss from

middle to late adulthood were associated with increased risk of mortality. Our study implies that

maintaining normal weight across adulthood, especially preventing weight gain in early adulthood, is

important for preventing premature deaths in later life.

Page 4 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


5

Introduction

Obesity is a major global public health problem. The global prevalence of adult obesity has increased

from 3% in 1975 to 11% in 2016 among men and from 6% to 15% among women. In the United States,

the prevalence of adult obesity was 36% for men and 38% for women in 2016 compared with 11% for

men and 14% for women in 1975.1 The U-shape or J-shape relationship between body mass index (BMI)

and mortality has been well established.2 3 However, many previous cohort studies only included a single

time measurement of BMI, which ignored the dynamic feature of body weight over time. Thus more

studies are needed to assess the long-term consequence of weight change during certain life periods. A

recent meta-analysis of 26 prospective studies among participants aged 40 to 65 years old at baseline

reported that both weight loss and weight gain from middle adulthood to old age were related to elevated

mortality risk.4 Another meta-analysis of 17 prospective studies among adults aged 60 years or older also

reached similar conclusion.5 However, adults gain weight more rapidly from young adulthood to middle

adulthood and excess adiposity mostly accrues in this period compared with the period from middle to

late adulthood when weight begins to stabilize or decrease.6 In addition, the prevention of weight gain

from young to middle adulthood might be more important than promoting weight loss because achieving

long-term weight loss and maintenance is difficult once an individual becomes obese.7 Thus, it is

important to evaluate the long-term impact of weight change, particularly weight gain from young to

middle adulthood, on future health.

Although most previous cohort studies reported that weight gain from young adulthood to either

midlife or late adulthood was associated with increased mortality risk,8-11 the results were not entirely

consistent with some reporting no significant association.12 13 Some studies found that weight loss was

associated with increased mortality risk12 14 15 while others did not.10 11 In addition, some studies only had

Page 5 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


6

two measurements of weight (from young adulthood around 20 years old to baseline survey),9-11 14 or

used the trajectory analysis approach when there were more than two measures of weight,16 17 or were

conducted in specific groups rather than the general population.10 13 A previous study in the US National

Health and Nutrition Examination Survey (NHANES) 1988-2004 evaluated the relations of all cause

mortality with weight status at age 25, at ten years before baseline and at baseline survey among adults

with a baseline age of 50 to 74 years.18 However, this study focused on the statistical modeling of obesity

history and suggested that a duration model performed well in predicting mortality.

Using updated data from the NHANES 1988-2014, we aimed to examine the relations of weight

changes from young adulthood (age of 25 years) to midlife (mean age of 47 years) and to late adulthood

(mean age of 57 years) with all cause and cause specific mortality and explore how the impact of weight

history varies by age.

Subjects and Methods

Study Population

Details of the NHANES have been described elsewhere.19 Briefly, NHANES used a complex, multistage,

probability sampling method to collect nationally representative health-related data of US population. It

was conducted periodically before 1999 and on a continuous basis thereafter. Data were obtained by in-

person interview, mobile physical examination and laboratory tests.

In the current analysis, we only included participants aged 40 or over at baseline survey, i.e., 40 052

participants from the NHANES III (1988-1994) and continuous NHANES (1999-2014) datasets, given

that death mostly occurred in older adults. We excluded participants who received household interview

but did not participate in physical examination (n=3198). We further excluded individuals without

Page 6 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


7

mortality data (n=42). Therefore, a total of 36 812 participants remained in our analytic cohort

(supplementary figure 1).

Assessments of weight change and covariates

Data on weight at age 25 years and at ten years prior to baseline NHANES survey were recalled in

baseline surveys. Baseline weight, height and waist circumference were measured during mobile physical

examination. BMI at age 25 years (BMI25), at ten years before survey (BMI10prior, mean age of 47 years,

range 30 to 80, interquartile range 36 to56) and at survey (BMIbaseline, mean age of 57 years, range 40 to

90, interquartile range 46 to 66) were calculated as weight (kg) divided by the square of height (m2). We

categorized each of the three BMI variables into standard groups: underweight (<18.5 kg/m2), normal

weight (18.5-24.9 kg/m2), overweight (25.0-29.9 kg/m2) and obese (≥30.0 kg/m2).20 Waist circumference

was divided into quartiles. Information on covariates including age, sex, race/ethnicity, education level,

household income level, marital status, smoking status, drinking status, family history of chronic diseases

(only including family histories of diabetes and heart diseases which have been consistently collected in

various survey cycles) and self-reported baseline diseases of diabetes, hypertension, cardiovascular

disease (CVD), chronic obstructive pulmonary disease (COPD) and cancer was available through

questionnaires at survey.

We created weight change patterns for three time intervals: BMI25 to BMI10prior, BMI25 to BMIbaseline,

and BMI10prior to BMIbaseline. Using BMI at two time points, we defined five weight change patterns for

each of the three time intervals: stable normal (<25.0 kg/m2 at both times), maximum overweight (25.0-

29.9 kg/m2 at either time but not ≥30.0 kg/m2 at the other time), weight loss (≥30.0 kg/m2 at younger age

while <30.0 kg/m2 later), weight gain (<30.0 kg/m2 at younger age while ≥30.0 kg/m2 later), and stable

Page 7 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


8

obesity (≥30.0 kg/m2 at both times). The method has been described in detail in a previous publication

using the NHANES data.21 Based on a recent study conducted by Zheng et al,11 we also classified

absolute weight change in each time interval into five groups as the following: weight loss ≥2.5 kg,

weight loss or weight gain <2.5 kg (reference group), 2.5 kg≤ weight gain <10.0 kg, 10.0 kg≤ weight

gain <20.0 kg, and weight gain ≥20.0 kg.

Ascertainment of deaths

Mortality status was identified by probabilistic matching to the National Death Index through 31

December 2015 using a unique study identifier. Further details of the matching methodology are

available at the National Center for Health Statistics.22

Causes of deaths were classified according to the codes of International Statistical Classification of

Diseases and Related Health Problems, 10th Revision (ICD-10). Primary outcomes for our study were

mortality from all cause, cardiovascular disease (CVD, including heart diseases and stroke, codes I00-

I09, I11, I13, I20-I51, I60-I69), cancer (codes C00-C97) and other causes (such as diabetes, Alzheimer

disease, influenza, and accidents).

Statistical Analysis

We accounted for complex survey design factors for the NHANES including sample weights, clustering

and stratification whenever feasible. We defined baseline as the time when participants had their physical

examinations. Person years were counted from baseline to the date of death, loss to follow-up or 31

December 2015, whichever came first. We used cox proportional hazards models with time in study as

the underlying time metric to calculate the hazard ratios (HRs) and corresponding 95% confidence

Page 8 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


9

intervals (95% CIs) for all cause and cause specific mortality in relation to weight change patterns, using

stable normal pattern as the reference. We examined proportional hazards assumption by creating a time-

dependent covariate defined by the cross product of follow-up time and weight change patterns.

Likelihood ratio tests comparing models with and without the variable were not significant, suggesting

no departure from the proportional hazards assumption.

For the main analyses, we examined associations between weight change patterns and mortality in the

three time intervals: BMI changes from age 25 years to ten years prior to baseline (BMI25 to BMI10prior),

to approximate weight change from young to middle adulthood; BMI changes from ten years before

baseline to baseline (BMI10prior to BMIbaseline), to approximate weight change from midlife to late

adulthood; BMI changes from age 25 years to baseline (BMI25 to BMIbaseline), to approximate weight

change during the whole adulthood period. We also investigated associations between absolute weight

change groups and mortality in the above three time intervals. In addition, we tested the possible

nonlinear relationship between absolute weight change and mortality risk by non-parametrically

restricted cubic splines,23 and tests for nonlinearity were conducted through the likelihood ratio test

comparing the models with and without the cubic spline terms. Participants with missing BMI data in

certain age point were excluded based on specific analysis.

We adjusted for baseline age (years, continuous), sex, and race/ethnicity (non-Hispanic White, non-

Hispanic Black, Mexican American, and others) in model 1. We further adjusted for education level (less

than high school, high school or equivalent, and college or above), household income level, marital status

(married, separated including widowed and divorced groups, and never married groups), smoking (never

smoker, ever smoker and current smoker) and drinking status (never drinker and ever drinker), and family

history of diabetes and heart diseases in model 2. In the absolute weight change analyses, we also

Page 9 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


10

included baseline height and weight at the beginning of the change period as potential confounders in

model 2. Finally, we additionally adjusted for baseline histories of diabetes, hypertension, CVD, cancer

and COPD, as those comorbidities could be mediators for the association between weight change and

mortality. Percentages of missing values of covariates were less than 7%, and we included them as

missing indicators accordingly. We conducted stratified analyses and potential effect modifications by

baseline age (<60 and ≥60 years), sex and smoking status (ever and never smokers). To evaluate the

impact on premature death, we also examined the relations with deaths occurred before 65 years old,

between age 66 and 75 years, between age 76 and 85 years and after 85 years old.

We conducted a series of sensitivity analyses to test the robustness of the results: performing separated

analysis in the NHANES 1988-1994 and 1999-2014 to examine the cohort effect; redefining the weight

change patterns by combinations of standard BMI categories in each two time points which generated a

16-category weight change group instead of using the five weight change patterns; excluding

underweight people to reduce the influence of underweight on the results; and excluding those who died

within the first three years of follow-up to minimize potential reverse causation caused by illness.

All analyses were performed using SAS 9.4 (SAS Institute, Inc., Cary, North Carolina). A two-tailed

P value of <0.05 was determined as statistically significant.

Patient involvement

No participants were involved in setting the research questions or outcome measures, or in the design

and implementation of the study.

Results

Page 10 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


11

Baseline characteristics and weight change pattern

Regarding weight change across adulthood, participants tended to gain more weight from age 25 years

to ten years prior to baseline survey (mean age of 47 years) than in the ten years period before baseline

(from mean age of 47 years to mean age of 57 years), and weight loss was more likely to occur over this

ten years period. Across adulthood, more than 15% participants gained weight with BMI moving from

non-obese to obese range, whereas less than 4.2% participants lost weight from obese to non-obese BMI

(supplementary table 1).

Table 1 presents baseline characteristics of study participants across weight change groups from age

25 years to baseline. Compared with individuals in the stable normal group, those losing from obese to

non-obese BMI were older, more likely to be men, less educated, current drinkers, to have lower

household income and to have comorbidities except for cancer, but were less likely to be non-Hispanic

Whites, married and never smokers. However, those who gained from non-obese BMI to obesity were

more likely to be non-Hispanic Black, never smokers, to have family history of chronic metabolic

diseases and to develop chronic diseases except cancer. The distributions of participants’ characteristics

were generally similar when classifying the participants with weight change in different time periods

(supplementary tables 2 and 3).

Relations of weight and weight change patterns with mortality

Among 36 812 participants, 10 936 deaths occurred during a mean follow-up of 10 years (363 811 person

years), including 3013 CVD deaths and 2373 cancer deaths. When evaluating the weight status at each

time point (supplementary table 4), we found that overweight and obesity at age 25 years were

significantly associated with increased risk of all cause mortality, while underweight showed null

Page 11 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


12

association. The BMI-mortality association changed to U-shape or reversed J-shape with increasing age

that underweight was associated with increased risk of mortality at ten years before baseline and at

baseline while the association with overweight disappeared or even reversed and the association with

obesity was substantially attenuated. Baseline waist circumference showed a U-shape association with

mortality, and significantly increased risks were observed for both low and high waist groups.

Separate analysis of the relations of weight change with mortality in the NHANES 1988-1994 and

1999-2014 showed no significant cohort effect (supplementary table 5), and thus we used the pooled

sample of the two periods in our final analysis. Table 2 shows the association between weight change

patterns in the three adulthood period in NHANES 1988-2014 and mortality risk using stable normal

BMI group as the reference. As expected, stable obese participants had increased risk of all cause

mortality across the adulthood, but the associations were stronger when the exposure was from age 25

years to ten years before baseline (HR 1.93, 95% CI 1.72 to 2.16) and from age 25 years to baseline

(1.81, 1.59 to 2.05) compared to that in the ten years period before baseline (1.33, 1.22 to 1.45). Gaining

weight with BMI moving from non-obese range at age 25 years to obese range in ten years prior to

baseline was associated with a 33% higher risk of all cause mortality (1.33, 1.22 to 1.45), while the

association was modest with weight gain from at age 25 years to baseline (1.11, 1.01 to 1.21) and not

significant in the ten years period before baseline (1.09, 0.95 to 1.25). Maximum overweight group

showed very modest or null association with all cause mortality in all three time intervals. Losing weight

from obesity to non-obese BMI in the ten years period before baseline was associated with a 44% higher

risk of all cause mortality (1.44, 1.28 to 1.61), but the association was not significant for the weight loss

pattern from age 25 years to ten years before baseline or to baseline.

We observed similar patterns of results for CVD mortality but with slightly stronger effect estimates.

Page 12 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


13

For example, the HR (95% CI) was 2.52 (1.97 to 3.23) for stable obesity from age 25 years to ten years

before baseline, 2.24 (1.71 to 2.94) from age 25 years to baseline and 1.45 (1.22 to 1.73) in the ten years

period before baseline. Likewise, those who had been non-obese at age 25 years but became obese at ten

years before baseline had an HR of 1.52 for CVD mortality (1.52, 1.28 to 1.81). We found no significant

associations between BMI change and cancer mortality except for those gained weight from non-obese

BMI at 25 years to obese BMI at ten years before baseline and the stable obesity group in the ten years

period before baseline. As for deaths from other causes, the results were generally comparable to all

cause mortality, except that losing from obesity to non-obese BMI was associated with increased risks

in all three time intervals (table 2). Further adjustment of baseline comorbidities attenuated the

associations, but stable obesity at all three time intervals remained significantly associated with higher

mortality risk while the association with weight gain was only significant for the period from age 25

years to ten years before baseline (supplementary table 6). When classifying the participants into the 16-

catergory weight change groups, the conclusion was generally consistent with the main results

(supplementary figure 2).

In the stratified analyses, we found significant interactions with baseline age and smoking status but

not sex (Figure 1), and the associations were stronger among participants who were less than 60 years

old at baseline and never smokers compared to their counterparts. When stratified by age at death, the

magnitude of associations was larger among participants who died before 65 years old than those who

died later (supplementary table 7). The relations of weight change patterns with all cause mortality were

materially unchanged after excluding underweight participants (supplementary table 8) or when

participants died within three years of follow-up were not included (supplementary table 9).

When evaluating the absolute weight changes, we identified a J-shape or U-shape association for all

Page 13 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


14

cause mortality with weight change across the three time intervals (Figure 2). When classified into

categories (table 3), extreme weight gain (weight gain ≥20 kg) was consistently associated with increased

risk of all cause mortality in all three time intervals, compared with stable weight group (weight change

within 2.5 kg). Moderate to large weight gain (10 kg ≤ weight gain <20 kg) was related to a modestly

increased risk of all cause mortality in the ten years period before baseline (1.12, 1.02 to 1.23), while

small to moderate weight gain (2.5 kg ≤ weight gain <10 kg) was not associated with all cause mortality

in any of the three time intervals. Those losing weight of more than 2.5 kg from age 25 years to baseline

had an HR of 1.21 (1.06 to 1.38) for all cause mortality and that of 1.30 (1.17 to 1.44) in the ten years

period before baseline. Significantly increased risk of CVD mortality was observed in extreme weight

gain from age 25 years to ten years before baseline and in the ten years period before baseline. Cancer

mortality risk was significantly higher in participants losing weight of more than 2.5 kg from age 25

years to baseline (1.48, 1.17 to 1.86) and in the ten year period before baseline (1.29, 1.09 to 1.53). The

relation of weight change with mortality of other causes was comparable to that of all cause mortality

(table 3). With the exception of the linear association of CVD mortality with weight change from age 25

years to ten years before baseline, the J-shape or reversed J-shape association between absolute weight

change across adulthood and cause specific mortality appeared in the restricted cubic spline analyses

(supplementary figure 3). Further adjustment of baseline comorbidities weakened the relations, with only

weight loss and extreme weight gain in the ten years period before baseline remained significant for all

cause mortality (supplementary table 10).

Discussion

In this large prospective study of national representative US adults, the lowest risk of mortality was

Page 14 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


15

observed in stable normal BMI participants, while stable obesity and weight gain across adulthood,

particularly weight gain from young to middle adulthood, were associated with increased risk of

mortality; in addition, weight loss from middle to late adulthood was significantly related to increased

mortality risk. Therefore, the findings underscore the importance of maintaining a normal weight across

adulthood, especially the prevention of weight gain in early adulthood, for reducing mortality risk in later

life.

Comparison with other studies

The association between BMI and mortality has been extensively studied in many prospective cohort

studies and in general a U-shape association was reported.24 However, many studies only measured BMI

once and thus could not capture the dynamic change feature of body weight. A number of studies with

repeated measures of BMI reported that both weight loss and weight gain in middle-aged to elderly adults

were associated with increased mortality risk.4 Similar findings were also observed in our recent analysis

in the Singapore Chinese population.25 In our current study, stable obesity and weight loss with BMI

moving from obese to non-obese range from middle to late adulthood had significantly higher mortality

risk (33% and 44% higher risks, respectively), but weight gain from non-obese BMI to obesity was not

significant (9% higher risk). In the previous meta-analysis, weight loss compared with stable weight was

associated with about 45% higher risk of all cause mortality, while weight gain was related to only 7%

higher risk (although statistically significant).4 Therefore, our results are largely consistent with previous

cohort studies on the relations of weight change from middle to late adulthood and mortality risk.

Adult weight gain mostly occurs from young to middle adulthood while studies on this critical period

are still lacking.26 In addition, we identified a J-shape association for all cause mortality and a linear

Page 15 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


16

association for CVD mortality, while very few studies reported the dose-response relationship between

weight change over this period and mortality. Using data from the National Institutes of Health-AARP

cohort with 109 947 never smokers aged 50 to 71 years at entry, Adams et al10 found that weight gain

was positively related to mortality, with stronger associations for gain between ages 18 and 35 years and

ages 35 and 50 years than between ages 50 and 69 years. The findings were consistent with ours that the

relations of stable obesity and weight gaining from non-obese to obese BMI with mortality were

generally stronger from young to middle or late adulthood than from middle to late adulthood. The

National Institutes of Health-AARP study also found that weight loss in late adulthood, but not in early

adulthood, was significantly associated with increased mortality risk, which again was consistent with

our results. A study in 1657 Finnish men with average age of 47 years at baseline reported that mortality

risk was significantly increased in the highest quartile of weight gain (≥15.0 kg) compared with the

lowest quartile.13 In the Southern Community Cohort study with 56 868 US participants aged 40 to 65 at

baseline, Hirko et al15 reported that maintaining an obese BMI and losing from obese to non-obese BMI

from age 21 years to approximately 52 years were associated with elevated mortality risk, while gaining

weight from non-obese to obese BMI was associated with a 8% lower risk of mortality. Reasons for the

inconsistent results between this study and ours might be that our study used stable normal BMI as the

reference group while this study used stable non-obesity as the reference group (thus people with

overweight at either time were included in the reference); this study also included more than two thirds

of black Americans and they previously reported that obesity in mid-to-late adulthood among blacks was

not associated with the same excess mortality risk seen among whites.27 Taken together, our results

indicate that avoiding obesity at young age and preventing weight gain from young to middle adulthood

could be an important strategy to reduce future mortality risk.

Page 16 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


17

A number of studies have also investigated the relations of weight change from young to late adulthood

and mortality risk, but the results were not entirely consistent. Similar to our results, most studies showed

gaining weight over this period was associated with increased risk of all cause mortality.9-11 18 In contrast,

a few studies reported that weight loss, but not weight gain, was related to increased mortality risk.12 28

The controversy might be due to the differences in sample size and statistical power and definitions of

weight change across studies, which made direct comparisons difficult. In addition, the baseline age

varied in different studies and previous studies showed that the obesity-mortality association tends to

decline with increasing age.29 30 As also shown in our study, the association between weight gain and

mortality attenuated with age while the association with weight loss became stronger and significant in

middle to late adulthood.

In the present study, the strongest association between weight gain and CVD mortality was observed

in participants who changed from non-obese BMI in young adulthood to obese BMI in middle adulthood.

Association between weight loss and CVD mortality was more pronounced in participants who moved

from obese to non-obese BMI from middle to late adulthood. In previous studies, similar results were

found.11 31 While it is logical that weight gain from young to middle adulthood may be a risk factor for

CVD death because obesity is a well-established risk factor for CVD mortality, it is unexpected that

weight loss in middle to late adulthood was also a risk factor for CVD mortality. One possible reason for

the positive association between weight loss and CVD mortality may be that a large decrease of BMI

from middle to late adulthood may indicate sarcopenia or muscle loss that can cause frailty and increase

risk of death.32 Another possibility is confounding by smoking and reverse causation due to underlying

severe chronic diseases. However, in our attempts to minimize such confounding by restricting analysis

in never smokers or excluding participants died within three years of follow-up, our results remained

Page 17 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


18

unchanged.

In our study, maximum overweight pattern was generally not significantly related to mortality risk

except for all cause and CVD mortality over the period of age 25 years to ten years before baseline.

Findings for this pattern were in accordance with most prior studies reporting null or modest association

between mild weight change and mortality.9 25 31 33

Strengths and limitations of our study

This study had several strengths including its large and nationally representative sample, high follow-up

rate and detailed analysis of weight change from young adulthood through midlife to late adulthood. The

importance of categorizing weight change throughout the entire adulthood into two special periods lies

in that natural tendency of weight gain from young to middle adulthood and weight loss from middle

adulthood to late adulthood could be differentiated and accounted. Furthermore, weight gain from young

to middle adulthood mostly reflects accumulation of fat mass while weight loss from middle to late

adulthood is usually accompanied with a decrease of lean mass and an increase in fat mass,7 30 therefore

there might be different underlying mechanisms of the harmful effect of weight change in certain life

period.

Several limitations also existed in the present study. First, our approximation of middle adulthood (age

range of 30 to 80, IQR 36 to 56) and late adulthood (age range 40 to 90, IQR 46 to 66) was overlapped.

Therefore, we conducted a stratified analysis by baseline age, and results in participants less than 60

years old at baseline (which means participants were in their 30s to 50s ten years before baseline, thus

considered as middle adulthood) and more than 60 years old at baseline (which means participants were

in their 60s to 90s ten years before baseline, thus considered as late adulthood) further confirmed that

Page 18 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


19

weight gain in younger ages was associated with much higher mortality risk. Second, we used recalled

and self-reported weight data at age of 25 years and ten years prior to the baseline NHANES survey.

Because of the probable inaccuracy of recalled weight compared with actual weight and the fact that

people tend to underreport their weight, misclassification bias may be introduced. However, several

validation studies showed that self-reported weight is highly correlated with measured weight.34-36 Third,

some possible confounders such as diet and physical activity were not adjusted for because of lack of

consistent data in different cycles of NHANES surveys. Nevertheless, previous studies reported that

adiposity, diet and physical activity may act independently in relation to mortality.37 38 Fourth, we could

not differentiate intentional from unintentional weight change and supplementary information regarding

reasons of weight change in future studies may help better interpret the results and offer appropriate

intervention strategies accordingly. Lastly, we did not evaluate the relations of changes in other

adiposity-related markers such as waist circumference and fat mass with mortality due to lack of data.

Further studies with repeated data on those markers may provide more comprehensive picture of the

changes in obesity status and mortality risk.

Conclusions and public health implications

Our study showed that weight gain across adulthood was associated with increased risks of all cause,

CVD and cancer mortality with weight gain from young to middle adulthood showing the strongest

association, while weight loss in middle to late adulthood conferred elevated risks of all cause and CVD

mortality. Future studies are required to unravel the mechanisms underlying the association of weight

change in whole adulthood and mortality, particularly the relations of changes in body composition on

mortality. In addition, since weight loss is less achievable (less than 4.2% of study participants changed

Page 19 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


20

from obese to non-obese category across adulthood), our results suggest that the prevention of weight

gain might be more important. Taken together, our findings indicate that maintaining a normal weight

throughout the adulthood, especially the prevention of weight gain in early adulthood, should be

encouraged to reduce mortality risk, however, further studies are still needed to consider other health-

related and obesity-related factors, and clinical trials of weight loss on long-term health consequences

are also warranted.

Page 20 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


21

Contributors: CC, XF and AP conceived the study design. CC and YY did the statistical analysis. CC

and AP wrote the manuscript. All authors provided critical revisions on the draft and approved the

submitted draft. AP funded this study. AP is the guarantor and attests that all listed authors meet

authorship criteria and that no others meeting the criteria have been omitted.

Funding: The work was supported by the National Key Research and Development Program of China

to Dr. An Pan (2017YFC0907504). The funders had no role in design and conduct of the study; collection,

management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript;

or the decision to submit the manuscript for publication.

Competing interests: All authors have completed the ICMJE uniform disclosure form at

http://www.icmje.org/coi_disclosure.pdf and declare: no support from any organization for the submitted

work other than those described above; no financial relationships with any organizations that might have

an interest in the submitted work in the previous three years; no other relationships or activities that could

appear to have influenced the submitted work.

Ethical approval: Approval of the study from the National Center of Health and Statistics Research

ethics review board was waived because the research was relied on publicly used, de-identified secondary

data.

Transparency declaration: The manuscript’s guarantor affirms that the manuscript is an honest,

accurate, and transparent account of the study being reported; that no important aspects of the study have

been omitted; and that any discrepancies from the study have been explained.

Data sharing: No additional data available.

Page 21 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


22

References

1. Abarca-Gómez L, Abdeen ZA, Hamid ZA, et al. Worldwide trends in body-mass index,

underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416

population-based measurement studies in 128· 9 million children, adolescents, and

adults. Lancet 2017;390:2627-42. doi: 10.1016/S0140-6736(17)32129-3.

2. Di Angelantonio E, Bhupathiraju SN, Wormser D, et al. Body-mass index and all cause

mortality: individual-participant-data meta-analysis of 239 prospective studies in four

continents. Lancet 2016;388:776-86. doi: 10.1016/S0140-6736(16)30175-1.

3. Bhaskaran K, dos-Santos-Silva I, Leon DA, Douglas IJ, Smeeth L. Association of BMI with

overall and cause specific mortality: a population-based cohort study of 3· 6 million

adults in the UK. Lancet Diabetes Endocrinol 2018;6:944-53. doi: 10.1016/S2213-

8587(18)30288-2.

4. Karahalios A, English DR, Simpson JA. Change in body size and mortality: a systematic

review and meta-analysis. Int J Epidemiol 2017;46:526-46. doi: 10.1093/ije/dyw246.

5. Cheng FW, Gao X, Jensen GL. Weight Change and All cause Mortality in Older Adults: A

Meta-Analysis. J Nutr Gerontol Geriatr 2015;34:343-68. doi:

10.1080/21551197.2015.1090362.

6. Sheehan TJ, DuBrava S, DeChello LM, Fang Z. Rates of weight change for black and white

Americans over a twenty year period. Int J Obes Relat Metab Disord 2003;27:498-504.

doi: 10.1038/sj.ijo.0802263.

7. Fontana L, Hu FB. Optimal body weight for health and longevity: bridging basic, clinical, and

population research. Aging Cell 2014;13:391-400. doi: 10.1111/acel.12207.

Page 22 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


23

8. Strandberg TE, Strandberg A, Salomaa VV, Pitkala K, Miettinen TA. Impact of midlife weight

change on mortality and quality of life in old age. Prospective cohort study. Int J Obes

Relat Metab Disord 2003;27:950-4. doi: 10.1038/sj.ijo.0802313.

9. Shimazu T, Kuriyama S, Ohmori-Matsuda K, Kikuchi N, Nakaya N, Tsuji I. Increase in body

mass index category since age 20 years and all cause mortality: a prospective cohort

study (the Ohsaki Study). Int J Obes (Lond) 2009;33:490-6. doi: 10.1038/ijo.2009.29.

10. Adams KF, Leitzmann MF, Ballard-Barbash R, et al. Body mass and weight change in

adults in relation to mortality risk. Am J Epidemiol 2014;179:135-44. doi:

10.1093/aje/kwt254.

11. Zheng Y, Manson JE, Yuan C, et al. Associations of Weight Gain From Early to Middle

Adulthood With Major Health Outcomes Later in Life. JAMA 2017;318:255-69. doi:

10.1001/jama.2017.7092.

12. Yarnell JW, Patterson CC, Thomas HF, Sweetnam PM. Comparison of weight in middle

age, weight at 18 years, and weight change between, in predicting subsequent 14 year

mortality and coronary events: Caerphilly Prospective Study. J Epidemiol Community

Health 2000;54:344-8.

13. Jeffreys M, McCarron P, Gunnell D, McEwen J, Smith GD. Body mass index in early and

mid-adulthood, and subsequent mortality: a historical cohort study. Int J Obes Relat

Metab Disord 2003;27:1391.

14. Corrada MM, Kawas CH, Mozaffar F, Paganini-Hill A. Association of body mass index and

weight change with all cause mortality in the elderly. Am J Epidemiol 2006;163:938-49.

doi: 10.1093/aje/kwj114.

Page 23 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


24

15. Hirko KA, Kantor ED, Cohen SS, Blot WJ, Stampfer MJ, Signorello LB. Body mass index in

young adulthood, obesity trajectory, and premature mortality. Am J Epidemiol

2015;182:441-50. doi: 10.1093/aje/kwv084.

16. Mehta NK, Stenholm S, Elo IT, Aromaa A, Heliovaara M, Koskinen S. Weight histories and

mortality among finnish adults: the role of duration and peak body mass index.

Epidemiology 2014;25:707-10. doi: 10.1097/ede.0000000000000147.

17. Song M, Hu FB, Wu K, et al. Trajectory of body shape in early and middle life and all cause

and cause specific mortality: results from two prospective US cohort studies. BMJ

2016;353:i2195. doi: 10.1136/bmj.i2195.

18. Preston SH, Mehta NK, Stokes A. Modeling obesity histories in cohort analyses of health

and mortality. Epidemiology 2013;24:158-66. doi: 10.1097/EDE.0b013e3182770217.

19. Centers for Disease Control and Prevention (CDC). National Center for Health Statistics

(NCHS). National Health and Nutrition Examination Survey. Available from

https://www.cdc.gov/nchs/nhanes/about_nhanes.htm. Accessed 1 February 2019.

20. Health NIo. NHLBI obesity education initiative expert panel on the identification, evaluation,

and treatment of overweight and obesity in adults. The Evidence Report NIH 1998.

21. Stokes A, Collins JM, Grant BF, et al. Obesity Progression Between Young Adulthood and

Midlife and Incident Diabetes: A Retrospective Cohort Study of U.S. Adults. Diabetes

Care 2018;41:1025-31. doi: 10.2337/dc17-2336.

22. National Center for Health Statistics. Available from https://www.cdc.gov/nchs/data-

linkage/mortality-public.htm#. Accessed 1 February 2019.

23. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989;8:551-

Page 24 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


25

61.

24. Aune D, Sen A, Prasad M, et al. BMI and all cause mortality: systematic review and non-

linear dose-response meta-analysis of 230 cohort studies with 3.74 million deaths

among 30.3 million participants. BMJ 2016;353:i2156. doi: 10.1136/bmj.i2156.

25. Pan X-F, Yuan J-M, Koh W-P, Pan A. Weight change in relation to mortality in middle-aged

and elderly Chinese: the Singapore Chinese Health Study. Int J Obes (Lond) 2018:1.

doi: 10.1038/s41366-018-0259-y.

26. Dietz WH. Obesity and Excessive Weight Gain in Young Adults: New Targets for Prevention.

JAMA 2017;318:241-42. doi: 10.1001/jama.2017.6119.

27. Cohen SS, Signorello LB, Cope EL, et al. Obesity and all cause mortality among black

adults and white adults. Am J Epidemiol 2012;176:431-42. doi: 10.1093/aje/kws032

28. Droyvold WB, Lund Nilsen TI, Lydersen S, et al. Weight change and mortality: the Nord-

Trondelag Health Study. J Intern Med 2005;257:338-45. doi: 10.1111/j.1365-

2796.2005.01458.x.

29. Elliott AM, Aucott LS, Hannaford PC, Smith WC. Weight change in adult life and health

outcomes. Obes Res 2005;13:1784-92. doi: 10.1038/oby.2005.217.

30. Ferrucci L, Alley D. Obesity, disability, and mortality: a puzzling link. Arch Intern Med

2007;167:750-1. doi: 10.1001/archinte.167.8.750.

31. Park SY, Wilkens LR, Maskarinec G, Haiman CA, Kolonel LN, Marchand LL. Weight change

in older adults and mortality: the Multiethnic Cohort Study. Int J Obes (Lond)

2017;42:205-12. doi: 10.1038/ijo.2017.188.

32. Walston J, Hadley EC, Ferrucci L, et al. Research agenda for frailty in older adults: toward

Page 25 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


26

a better understanding of physiology and etiology: summary from the American

Geriatrics Society/National Institute on Aging Research Conference on Frailty in Older

Adults. J Am Geriatr Soc 2006;54:991-1001. doi: 10.1111/j.1532-5415.2006.00745.x.

33. Mulligan AA, Lentjes MA, Luben RN, Wareham NJ, Khaw K-T. Weight change and 15 year

mortality: results from the European Prospective Investigation into Cancer in Norfolk

(EPIC-Norfolk) cohort study. Eur J Epidemiol 2018;33:37-53. doi: 10.1007/s10654-017-

0343-y.

34. Casey VA, Dwyer JT, Berkey CS, Coleman KA, Gardner J, Valadian I. Long-term memory

of body weight and past weight satisfaction: a longitudinal follow-up study. Am J Clin

Nutr 1991;53:1493-98.

35. Perry GS, Byers TE, Mokdad AH, Serdula MK, Williamson DF. The validity of self-reports

of past body weights by US adults. Epidemiology 1995:61-66.

36. Stokes A, Ni Y. Validating a summary measure of weight history for modeling the health

consequences of obesity. Ann Epidemiol 2016;26:821-26. e2. doi:

10.1016/j.annepidem.2016.10.005.

37. Koster A, Harris TB, Moore SC, et al. Joint associations of adiposity and physical activity

with mortality: the National Institutes of Health-AARP Diet and Health Study. Am J

Epidemiol 2009;169:1344-51. doi: 10.1093/aje/kwp053.

38. Ekelund U, Ward HA, Norat T, et al. Physical activity and all cause mortality across levels

of overall and abdominal adiposity in European men and women: the European

Prospective Investigation into Cancer and Nutrition Study (EPIC). Am J Clin Nutr

2015;101:613-21. doi: 10.3945/ajcn.114.100065.

Page 26 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


27

Page 27 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


28

Figure Legends

Fig 1 Associations between weight change patterns across adulthood and risk of all cause mortality stratified by baseline age, smoking status and sex in NHANES 1988-2014.

Risk estimates were adjusted for baseline age (not adjusted in subgroup analysis by age), sex (not adjusted in subgroup analysis by sex), race/ethnicity, education level, household income level, marital status, drinking and smoking status (not adjusted in subgroup analysis by smoking status) and family history of diabetes and heart attack. P for interaction was <0.001, 0.95, 0.01 by baseline age, sex and smoking status respectively in age 25 years to ten years before baseline. P for interaction was <0.001, 0.49 and 0.03 by baseline age, sex and smoking status respectively in age 25 years to baseline. P for interaction was <0.001, 0.53 and 0.01 by baseline age, sex and smoking status respectively in ten years period before baseline.

Fig 2 Dose response association between weight change across adulthood and risk of all cause mortality.

Associations were examined by multivariable Cox regression based on restricted cubic splines. Dash line represents estimates of hazard ratios and dotted line represents 95% confidence intervals. Risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status and family history of diabetes and heart attack. For weight change from age 25 years to ten years before baseline or to baseline, weight at age 25 and baseline height were also adjusted. For weight change from ten years before baseline to baseline, weight at 10 years ago and baseline height were adjusted. P for overall association <0.001 in all three periods. P for non-linear association <0.001 in all three periods.

Page 28 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


29

Table 1 Baseline characteristics of study participants aged 40 or over in NHANES 1988-2014 according to patterns of weight change from age 25 years to baseline* Weight change patterns from age 25 years to baseline†

Characteristics Stable normal Maximum overweight Loss Gain Stable obese Total

No. of participants, n (%) 8529 (29.1) 12 955 (37.7) 590 (1.4) 9956 (26.9) 1750 (4.9) 33 780Age, mean (95% CI), years 56.8 (56.2 to 57.4) 57.8 (57.4 to 58.3) 59.9 (58.3 to 61.5) 56.9 (56.5 to 57.3) 53.5 (52.8 to 54.2) 57.1 (56.7 to 57.5)Women 4799 (63.3) 5523 (43.6) 201 (31.4) 5736 (56.9) 898 (47.8) 17 157 (52.9)BMI, mean (95% CI), kg/m2

At age 25 years 20.8 (20.7 to 20.8) 23.1 (23.1 to 23.2) 32.9 (32.5 to 33.3) 24.3 (24.2 to 24.3) 34.7 (34.4 to 35.0) 23.4 (23.3 to 23.5)At 10 years prior baseline 22.2 (22.1 to 22.2) 26.0 (25.9 to 26.1) 30.2 (29.6 to 30.9) 30.1 (29.9 to 30.2) 37.7 (37.1 to 38.3) 26.6 (26.5 to 26.7)Maximum 24.6 (24.5 to 24.7) 29.4 (29.2 to 29.4) 35.1 (34.5 to 35.7) 36.4 (36.2 to 36.5) 43.6 (43.0 to 44.2) 30.6 (30.5 to 30.8)At survey 22.2 (22.2 to 22.3) 27.1 (27.0 to 27.1) 26.4 (26.1 to 26.7) 34.5 (34.3 to 34.6) 39.3 (38.8 to 39.9) 28.2 (28.1 to 28.4)

Waist circumference, mean (95% CI), cm 83.6 (83.3 to 83.8) 96.9 (96.7 to 97.1) 97.1 (95.7 to 98.6) 111.9 (111.6 to 112.3) 121.8 (120.8 to 122.8) 98.3 (97.9 to 98.6)Race

Non-Hispanic White 4826 (80.6) 6678 (78.7) 267 (73.7) 4580 (74.7) 813 (76.5) 17 164 (80.0)Non-Hispanic Black 1533 (7.1) 2530 (9.0) 140 (12.1) 2574 (12.6) 538 (14.4) 7315 (9.7)Mexican American 990 (2.5) 2407 (4.7) 134 (5.9) 1918 (5.7) 282 (4.9) 5731 (4.4)Others 1180 (9.9) 1340 (7.5) 49 (8.3) 884 (7.0) 117 (4.3) 3750 (7.9)

EducationLess than high school 2529 (19.4) 4460 (23.0) 298 (36.1) 3393 (24.9) 558 (21.0) 11 238 (22.5)High school or equivalent 2105 (26.6) 3215 (28.3) 121 (29.4) 2538 (29.5) 453 (30.2) 8432 (28.2)College or above 3864 (53.7) 5257 (48.5) 168 (34.4) 4010 (45.5) 734 (48.7) 14 033 (49.0)

Household income levelBelow or at poverty 1272 (8.3) 1958 (8.8) 134 (14.6) 1652 (10.4) 329 (11.4) 5345 (9.3)Above poverty 6528 (85.0) 9883 (84.4) 408 (79.5) 7492 (82.5) 1307 (83.7) 25 618 (83.9)

Marital statusMarried 5275 (67.3) 8421 (70.5) 342 (63.9) 6186 (67.5) 1012 (66.0) 21 236 (68.5)Separated 2614 (26.1) 3666 (23.5) 195 (28.0) 3040 (26.6) 495 (21.5) 10 010 (25.0)Never married 562 (5.7) 745 (5.2) 48 (7.6) 632 (5.2) 228 (11.8) 2215 (5.6)

Alcohol drinkingNever drinker 7492 (89.5) 11221 (87.4) 459 (73.1) 8756 (88.3) 1523 (86.6) 29 451 (88.0)Ever drinker 958 (9.5) 1623 (11.9) 120 (24.1) 1138 (11.0) 217 (12.9) 4056 (11.2)

Smoking statusNever smoker 3947 (45.3) 5992 (45.2) 224 (33.3) 4995 (47.6) 913 (50.5) 16 071 (46.0)Ever smoker 2355 (28.5) 4458 (35.1) 207 (36.1) 3389 (35.3) 521 (32.2) 10 930 (33.1)

Page 29 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


30

Current smoker 2222 (26.2) 2500 (19.7) 158 (30.5) 1567 (17.1) 316 (17.3) 6763 (20.9)16 797 (49.9) Family history of chronic metabolic

diseases3729 (44.7) 6131 (47.6) 303 (53.8) 5545 (56.3) 1089 (62.3)

Self-reported chronic diseasesDiabetes 604 (4.5) 1609 (8.4) 170 (16.9) 2020 (15.9) 548 (26.4) 4951 (10.3)Hypertension 2096 (19.0) 4489 (30.6) 242 (33.6) 4740 (45.3) 922 (53.0) 12 489 (32.3)Cardiovascular diseases 1054 (8.7) 1916 (11.5) 150 (20.6) 1639 (13.8) 353 (16.3) 5112 (11.6)Chronic obstructive pulmonary diseases 797 (9.7) 986 (7.9) 65 (15.6) 998 (11.1) 182 (10.4) 3028 (9.5)Cancer 1211 (14.5) 1727 (13.4) 82 (12.2) 1141 (12.4) 176 (9.8) 4337 (13.3)*Complex survey designs are accounted for all estimates; data are presented as No. (%) unless otherwise indicated. Percentages may not add up to 100% due to missing values.†Stable normal: BMI25 <25 kg/m2 and BMIbaseline <25 kg/m2; maximum overweight: either BMI25 or BMIbaseline was 25.0-29.9 kg/m2 and the other BMI not ≥30.0 kg/m2; loss: BMI25 ≥ 30 kg/m2 and 0 < BMIbaseline < 30 kg/m2; gain: 0 < BMI25 < 30 kg/m2 and BMIbaseline ≥ 30 kg/m2; stable obese: BMI25 ≥ 30 kg/m2 and BMIbaseline ≥ 30 kg/m2.

Page 30 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


31

Table 2 Hazard ratios (95% CI) of all cause and cause specific mortality with weight change patterns across adulthood among 36 812 participants in NHANES 1988-2014*

Weight change patterns†

Cause of death Stable normal Maximum overweight Loss Gain Stable obeseFrom age 25 years to ten years before baseline All cause

No. of deaths/Person-years 3094/133 393 3849/127 257 160/3724 1837/54 404 524/16 220Age-adjusted mortality rate‡ 26.4 (25.5 to 27.3) 27.8 (26.9 to 28.6) 38.3 (32.2 to 44.4) 30.2 (28.8 to 31.6) 39.1 (35.6 to 42.5)Model 1 1.00 1.05 (0.98 to 1.12) 1.42 (1.06 to 1.92) 1.30 (1.18 to 1.42) 1.85 (1.63 to 2.09)Model 2 1.00 1.08 (1.01 to 1.15) 1.28 (0.92 to 1.79) 1.33 (1.22 to 1.45) 1.93 (1.72 to 2.16)

Cardiovascular diseasesNo. of deaths 745 1096 44 515 162Model 1 1.00 1.18 (1.04 to 1.35) 1.24 (0.84 to 1.83) 1.48 (1.24 to 1.76) 2.45 (1.89 to 3.16)Model 2 1.00 1.21 (1.06 to 1.39) 1.10 (0.74 to 1.64) 1.52 (1.28 to 1.81) 2.52 (1.97 to 3.23)

CancerNo. of deaths 789 872 27 380 90Model 1 1.00 0.96 (0.83 to 1.12) 1.10 (0.60 to 2.02) 1.20 (0.97 to 1.49) 1.07 (0.81 to 1.43)Model 2 1.00 1.01 (0.88 to 1.17) 0.98 (0.51 to 1.88) 1.26 (1.02 to 1.56) 1.16 (0.87 to 1.55)

Other causesNo. of deaths 1560 1881 89 942 272Model 1 1.00 1.03 (0.94 to 1.12) 1.69 (1.12 to 2.55) 1.27 (1.13 to 1.43) 2.03 (1.67 to 2.49)Model 2 1.00 1.05 (0.96 to 1.14) 1.53 (0.99 to 2.38) 1.29 (1.15 to 1.44) 2.08 (1.71 to 2.54)

From age 25 years to baselineAll cause

No. of deaths/Person-years 2677/87 892 3915/131 968 264/4923 2302/98 464 423/15 156Age-adjusted mortality rate‡ 29.1 (28.0 to 30.2) 27.8 (26.9 to 28.7) 45.2 (39.5 to 50.8) 26.0 (24.9 to 27.2) 36.3 (32.6 to 40.1)Model 1 1.00 0.91 (0.84 to 0.99) 1.38 (1.10 to 1.73) 1.08 (0.99 to 1.17) 1.68 (1.48 to 1.92)Model 2 1.00 0.94 (0.86 to 1.03) 1.21 (0.93 to 1.58) 1.11 (1.01 to 1.21) 1.81 (1.59 to 2.05)



Other causes

Page 31 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


32

No. of deaths 1386 1925 137 1128 225Model 1 1.00 0.87 (0.80 to 0.96) 1.57 (1.18 to 2.08) 1.05 (0.94 to 1.17) 1.84 (1.48 to 2.28)Model 2 1.00 0.89 (0.82 to 0.98) 1.38 (1.00 to 1.90) 1.07 (0.96 to 1.20) 1.94 (1.56 to 2.40)

From ten years before baseline to baseline All cause

No. of deaths/Person-years 2362/81 771 3924/131 762 852/16 428 1137/58 251 1652/56 972Age-adjusted mortality rate‡ 28.7 (27.5 to 29.9) 27.8 (26.9 to 28.7) 40.5 (37.7 to 43.4) 25.3 (23.7 to 26.9) 29.7 (28.2 to 31.2)Model 1 1.00 0.93 (0.85 to 1.02) 1.46 (1.27 to 1.68) 1.07 (0.95 to 1.20) 1.27 (1.15 to 1.40)Model 2 1.00 0.96 (0.88 to 1.04) 1.44 (1.28 to 1.61) 1.09 (0.95 to 1.25) 1.33 (1.22 to 1.45)



Other causesNo. of deaths 1190 1974 444 546 843Model 1 1.00 0.94 (0.85 to 1.04) 1.51 (1.25 to 1.82) 1.11 (0.95 to 1.29) 1.28 (1.13 to 1.46)Model 2 1.00 0.96 (0.87 to 1.07) 1.46 (1.23 to 1.74) 1.13 (0.96 to 1.33) 1.33 (1.17 to 1.50)

*Complex survey designs are considered for all estimates. A total of 3382, 3032, 2230 participants were excluded in analyses from age 25 years to ten years before baseline, age 25 years to baseline, ten years before baseline to baseline respectively due to missing values of BMI at both times. Model 1 were adjusted for baseline age, sex, race/ethnicity. Model 2 were additionally adjusted for additionally adjusted for education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack.†Weight change patterns were defined by BMI at two time points. Stable normal participants had BMI <25.0 kg/m2 at both times, maximum overweight participants had BMI in 25.0-29.9 kg/m2 at either time but not ≥30.0 kg/m2 at the other time, weight loss participants had BMI ≥30.0 kg/m2 at younger age while <30.0 kg/m2 later, weight gain participants had BMI <30.0 kg/m2 at younger age while ≥30.0 kg/m2 later, and stable obese participants had BMI ≥30.0 kg/m2 at both times. ‡Mortality rates per 1000 person-years, directly standardized to the age distribution of the entire study population

Page 32 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


33

Table 3 Hazard ratios of all cause and cause specific mortality with absolute weight change groups across adulthood in 36 812 participants in NHANES 1988-2014*

Weight change in three life periodsCause of death Weight loss ≥2.5 kg Weight loss <2.5

or gain <2.5 kg 2.5 kg ≤ Weight gain <10.0 kg 10 kg ≤ Weight gain <20 kg Weight gain ≥20 kg

From age 25 years to ten years before baselineAll cause

No. of deaths 762 2152 2661 2191 1698 Age-adjusted mortality rate† 37.2 (34.5 to 39.8) 30.0 (28.7 to 31.3) 26.2 (25.2 to 27.2) 26.2 (25.1 to 27.3) 31.1 (29.6 to 32.7) Model 1 1.30 (1.11 to 1.53) 1.00 0.90 (0.82 to 0.99) 0.95 (0.86 to 1.05) 1.17 (1.05 to 1.30) Model 2 1.07 (0.92 to 1.25) 1.00 0.97 (0.87 to 1.07) 1.06 (0.95 to 1.18) 1.28 (1.15 to 1.42)

Cardiovascular diseases No. of deaths 211 565 703 618 465 Model 1 1.17 (0.96 to 1.43) 1.00 0.82 (0.70 to 0.96) 0.85 (0.72 to 1.00) 1.11 (0.96 to 1.30) Model 2 0.90 (0.75 to 1.10) 1.00 0.88 (0.74 to 1.05) 0.96 (0.80 to 1.15) 1.26 (1.08 to 1.47)Cancer No. of deaths 170 510 662 453 363 Model 1 1.34 (1.05 to 1.72) 1.00 0.94 (0.79 to 1.11) 0.96 (0.77 to 1.11) 1.14 (0.91 to 1.44) Model 2 1.19 (0.92 to 1.53) 1.00 0.99 (0.83 to 1.19) 1.04 (0.83 to 1.31) 1.21 (0.96 to 1.53)Other causes No. of deaths 381 1077 1296 1120 870 Model 1 1.37 (1.10 to 1.72) 1.00 0.95 (0.83 to 1.09) 1.02 (0.91 to 1.15) 1.23 (1.09 to 1.40) Model 2 1.12 (0.89 to 1.42) 1.00 1.02 (0.88 to 1.18) 1.14 (1.01 to 1.29) 1.35 (1.19 to 1.53)From age 25 years to baselineAll cause No. of deaths 1801 1059 2027 2417 2277 Age-adjusted mortality rate† 44.4 (42.2 to 46.6) 29.1 (27.3 to 30.9) 26.5 (25.4 to 27.7) 25.6 (24.6 to 26.6) 25.7 (24.5 to 26.8) Model 1 1.44 (1.26 to 1.64) 1.00 0.97 (0.86 to 1.09) 0.91 (0.82 to 1.02) 1.09 (0.98 to 1.21) Model 2 1.21 (1.06 to 1.38) 1.00 1.01 (0.90 to 1.13) 0.99 (0.89 to 1.10) 1.16 (1.05 to 1.28)Cardiovascular disease No. of deaths 506 296 563 622 605 Model 1 1.18 (0.96 to 1.46) 1.00 0.88 (0.72 to 1.08) 0.84 (0.71 to 0.99) 1.01 (0.83 to 1.22) Model 2 0.95 (0.75 to 1.19) 1.00 0.92 (0.76 to 1.13) 0.92 (0.79 to 1.08) 1.11 (0.92 to 1.34)Cancer

No. of deaths 394 210 475 572 537Model 1 1.67 (1.35 to 2.07) 1.00 1.08 (0.89 to 1.31) 1.03 (0.85 to 1.26) 1.16 (0.98 to 1.37)Model 2 1.48 (1.17 to 1.86) 1.00 1.12 (0.93 to 1.34) 1.10 (0.89 to 1.36) 1.20 (1.01 to 1.42)

Page 33 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


34

Other causes No. of deaths 901 553 989 1223 1135 Model 1 1.50 (1.25 to 1.79) 1.00 0.98 (0.83 to 1.15) 0.92 (0.79 to 1.08) 1.13 (0.98 to 1.30) Model 2 1.26 (1.03 to 1.53) 1.00 1.02 (0.87 to 1.19) 1.00 (0.86 to 1.16) 1.20 (1.06 to 1.37)From ten years before baseline to baselineAll cause No. of deaths 3853 2205 2184 1115 570 Age-adjusted mortality rate† 38.5 (37.2 to 39.7) 25.5 (24.4 to 26.6) 23.6 (22.6 to 24.6) 25.4 (23.7 to 27.1) 28.5 (25.3 to 31.6) Model 1 1.49 (1.36 to 1.64) 1.00 0.99 (0.91 to 1.07) 1.14 (1.04 to 1.25) 1.68 (1.43 to 1.97) Model 2 1.30 (1.17 to 1.44) 1.00 1.01 (0.93 to 1.09) 1.12 (1.02 to 1.23) 1.57 (1.31 to 1.87)Cardiovascular disease No. of deaths 1106 613 567 267 153 Model 1 1.37 (1.12 to 1.67) 1.00 1.03 (0.86 to 1.23) 1.02 (0.80 to 1.31) 1.82 (1.35 to 2.44) Model 2 1.16 (0.94 to 1.44) 1.00 1.04 (0.88 to 1.23) 0.99 (0.78 to 1.25) 1.70 (1.25 to 2.31)Cancer No. of deaths 773 478 562 281 130 Model 1 1.45 (1.24 to 1.71) 1.00 1.02 (0.86 to 1.21) 1.07 (0.90 to 1.29) 1.36 (1.01 to 1.82) Model 2 1.29 (1.09 to 1.53) 1.00 1.04 (0.87 to 1.25) 1.06 (0.88 to 1.29) 1.24 (0.93 to 1.66)Other causes No. of deaths 1974 1114 1055 567 287 Model 1 1.57 (1.41 to 1.75) 1.00 0.95 (0.84 to 1.07) 1.24 (1.09 to 1.42) 1.80 (1.41 to 2.30) Model 2 1.38 (1.22 to 1.54) 1.00 0.97 (0.86 to 1.09) 1.23 (1.07 to 1.41) 1.70 (1.32 to 2.19)

*Complex survey designs were taken into account for all estimates. A total of 3382, 3032, 2230 participants were excluded in analyses from age 25 years to ten years before baseline, age 25 years to baseline, ten years before baseline to baseline respectively due to missing values of BMI at both times. Model 1 were adjusted for baseline age, sex and race/ethnicity. In analyses from age 25 years to ten years before baseline and age 25 years to baseline, model 2 were adjusted for baseline age, sex, race/ethnicity, height, weight at age 25, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack. In analysis from ten years before baseline to baseline, model 2 were adjusted for baseline age, sex, race/ethnicity, height, weight at ten years ago, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack. †Mortality rates per 1000 person-years, directly standardized to the age distribution of the entire study population

Page 34 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Fig 1 Associations between weight change patterns across adulthood and risk of all cause mortality stratified by baseline age, smoking status and sex in NHANES 1988-2014.

Risk estimates were adjusted for baseline age (not adjusted in subgroup analysis by age), sex (not adjusted in subgroup analysis by sex), race/ethnicity, education level, household income level, marital status,

drinking and smoking status (not adjusted in subgroup analysis by smoking status) and family history of diabetes and heart attack. P for interaction was <0.001, 0.95, 0.01 by baseline age, sex and smoking status

respectively in age 25 years to ten years before baseline. P for interaction was <0.001, 0.49 and 0.03 by baseline age, sex and smoking status respectively in age 25 years to baseline. P for interaction was <0.001,

0.53 and 0.01 by baseline age, sex and smoking status respectively in ten years period before baseline.

230x148mm (96 x 96 DPI)

Page 35 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Fig 2 Dose response association between weight change across adulthood and risk of all cause mortality.

Associations were examined by multivariable Cox regression based on restricted cubic splines. Dash line represents estimates of hazard ratios and dotted line represents 95% confidence intervals. Risk estimates

were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status and family history of diabetes and heart attack. For weight change from age 25 years to ten years before baseline or to baseline, weight at age 25 and baseline height were also adjusted. For weight change from ten years before baseline to baseline, weight at 10 years ago and baseline height

were adjusted. P for overall association <0.001 in all three periods. P for non-linear association <0.001 in all three periods.

151x298mm (96 x 96 DPI)

Page 36 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Page 37 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary tables/figures

Table S1 Weight change patterns across adulthood in NHANES 1988-2014

Table S2 Characteristics of 36 812 study participants aged 40 years or over in NHANES1988-2014 according to weight change from age 25 years to ten years before

baseline

Table S3 Characteristics of 36 812 study participants aged 40 years or over in NHANES1988-2014 according to weight change in the ten year period before baseline

Table S4 Hazard ratios of all cause mortality with obesity status at three age points among 36 812 participants in NHANES 1988-2014

Table S5 Hazard ratios of all cause mortality with weight change patterns stratified by study period

Table S6 Hazard ratios of all cause and cause specific mortality with weight change patterns across adulthood in 36 812 participants in NHANES 1988-2014 with further

adjustment of disease status

Table S7 Hazard ratios of all cause mortality with weight change patterns among 36 812 stratified by age at death in NHANES 1988-2014

Table S8 Hazard ratios of all cause mortality with weight change patterns after excluding participants who were underweight in NHANES 1988-2014

Table S9 Hazard ratios of all cause and cause specific mortality with weight change patterns across adulthood among 31 077 participants after excluding deaths within the

first three years of follow-up in NHANES 1988-2014

Table S10 Hazard ratios of all cause and cause specific mortality with absolute weight change groups across adulthood in 36 812 participants in NHANES 1988-2014 with

Page 38 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


further adjustment of disease status

Figure S1 Flow chart of inclusion and exclusion of participants

Figure S2 Associations between fine weight change patterns across adulthood and all cause mortality in NHANES 1988-2014

Figure S3 Dose response associations between weight change across adulthood and risk of mortality in the NHANES 1988-2014

Page 39 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S1 Weight change patterns across adulthood in NHANES 1988-2014From age 25 yearsto ten years before baseline From age 25 years to baseline From ten years before baseline

to baselineAbsolute weight change, median (IQR), kg 6.80 (1.81, 13.6) 11.70 (4.10, 21.07) 3.67 (-1.61, 10.13)Stable Normal

No. of participants 12 201 8529 7785Percent (%) 42.3 29.1 26.9

Maximum overweight No. of participants 12 804 12 955 12 932Percent (%) 36.1 37.7 37.2

Loss No. of participants 403 590 1982Percent (%) 1.2 1.4 4.2

Gain No. of participants 6104 9956 5502Percent (%) 15.3 26.9 15.4

Stable obese No. of participants 1918 1750 6381Percent (%) 5.2 4.9 16.3

Page 40 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S2 Baseline characteristics of study participants aged 40 or over in NHANES1988-2014 according to patterns of weight change from age 25 years to ten years before baseline a

Weight change patterns bCharacteristics Stable normal Maximum overweight Loss Gain Stable obese Total

No. of participants, n (%) 12 201 (42.3) 12 804 (36.1) 403 (1.2) 6104 (15.3) 1918 (5.2) 33 430Age, mean (95% CI), years 55.0 (54.5, 55.5) 58.7 (58.3, 59.2) 57.9 (56.6, 59.2) 59.8 (59.3, 60.2) 54.3 (53.6, 54.9) 57.1 (56.7, 57.4)Women 7350 (64.9) 5292 (40.7) 158 (37.3) 3154 (51.1) 925 (45.6) 16 879 (52.7)BMI, mean (95% CI), kg/m2

At age 25 years. 20.9 (20.8, 20.9) 23.9 (23.8, 24.0) 33.2 (32.6, 33.8) 25.1 (25.0, 25.2) 34.6 (34.3, 34.9) 23.5 (23.4, 23.5) At 10 years prior baseline 22.2 (22.1, 22.2) 27.0 (27.0, 27.1) 26.8 (26.3, 27.2) 33.8 (33.6, 34.0) 38.2 (37.7, 38.6) 26.6 (26.5, 26.7) Maximum 26.0 (25.9, 26.2) 31.0 (30.9, 31.1) 35.7 (34.8, 36.5) 37.6 (37.4, 37.8) 43.1 (42.6, 43.7) 30.6 (30.5, 30.8) At survey 24.4 (24.3, 24.5) 28.8 (28.7, 28.9) 29.3 (28.4, 30.2) 34.0 (33.8, 34.2) 38.1 (37.6, 38.5) 28.2 (28.1, 28.4)

Waist circumference, mean (95% CI), cm 88.5 (88.2, 88.9) 100.9 (100.7, 101.2) 103.0 (101.3, 104.7) 111.6 (111.1, 112.1) 119.6 (118.6, 120.6) 98.2 (97.9, 98.6)

RaceNon-Hispanic White 6489 (79.0) 6569 (78.3) 208 (78.0) 2923 (76.0) 869 (75.4) 17 058 (78.1)Non-Hispanic Black 2465 (8.4) 2575 (9.4) 83 (9.5) 1498 (12.1) 589 (14.8) 7210 (9.7)Mexican American 1664 (3.2) 2367 (4.9) 79 (4.4) 1187 (5.6) 327 (5.1) 5624 (4.3)Others 1583 (9.4) 1293 (7.4) 33 (8.1) 496 (6.3) 133 (4.7) 3538 (7.9)

Education levelLess than high school 3555 (19.6) 4345 (23.2) 174 (28.4) 2265 (26.9) 665 (23.0) 11 004 (22.3)High school or equivalent 3120 (27.7) 3159 (28.3) 94 (29.0) 1535 (29.4) 473 (29.9) 8381 (28.3)College or above 5498 (52.4) 5273 (48.2) 134 (42.6) 2293 (43.4) 772 (47.0) 13 970 (49.2)


Marital statusMarried 7648 (68.4) 8337 (70.5) 245 (67.2) 3735 (66.0) 1100 (65.3) 21 065 (68.6)Separated 3573 (24.9) 3612 (23.6) 125 (25.3) 1990 (28.6) 563 (22.5) 9683 (24.9)Never married 868 (5.8) 720 (5.1) 30 (6.9) 331 (4.7) 238 (11.5) 2187 (5.7)


Smoking statusNever smoker 5685 (45.3) 6031 (45.4) 146 (35.6) 3029 (48.2) 976 (49.2) 15 867 (45.9)

Page 41 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Ever smoker 3514 (30.5) 4450 (35.7) 166 (40.3) 2145 (34.5) 563 (31.7) 10 838 (33.1)Current smoker 2994 (24.2) 2319 (18.9) 91 (24.1) 929 (17.3) 378 (19.1) 6711 (21.0)

Family history of chronic metabolic diseases

5570 (47.0) 6236 (48.6) 220 (60.1) 3451 (57.2) 1163 (60.5) 16 640 (50.0)Self-reported chronic diseases

Diabetes 707 (4.0) 1753 (10.0) 92 (12.5) 1712 (22.3) 628 (27.3) 4892 (10.3)Hypertension 3098 (21.2) 4877 (34.9) 146 (30.5) 3241 (50.1) 1013 (53.1) 12 375 (32.4)Cardiovascular diseases 1250 (7.4) 2061 (13.2) 90 (20.3) 1268 (17.6) 414 (16.7) 5083 (11.7)Chronic obstructive pulmonary diseases

1099 (9.4) 993 (8.1) 48 (17.1) 660 (12.2) 197 (10.3) 2997 (9.5)Cancer 1442 (12.7) 1771 (14.1) 54 (13.9) 836 (14.3) 202 (9.6) 4305 (13.3)

a Complex survey designs are considered for all estimates; percentages may not add up to 100% due to missing values. b From age 25 years to ten years before baseline, stable normal: BMI25 < 25 kg/m2 and BMI10prior < 25 kg/m2; maximum overweight: either BMI25 or BMI10prior was 25.0-29.9 kg/m2 and the other BMI not ≥30.0 kg/m2; loss: BMI25 ≥ 30 kg/m2 and 25 ≤ BMI10prior < 30 kg/m2; gain: 25 ≤ BMI25 < 30 kg/m2 and BMI10prior≥ 30 kg/m2; stable obese: BMI25 ≥ 30 kg/m2 and BMI10prior ≥ 30 kg/m2.

Page 42 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S3 Baseline characteristics of study participants aged 40 or over in NHANES1988-2014 according to weight change patterns in ten years period before baseline a

Weight change patterns bCharacteristics Stable normal Maximum overweight Loss Gain Stable obese Total

No. of participants, n (%) 7785 (26.9) 12 932 (37.2) 1982 (4.2) 5502 (15.4) 6381 (16.3) 34 582Age, mean (95% CI), years 56.3 (55.7, 56.9) 58.1 (57.6, 58.6) 61.6 (60.5, 62.6) 54.9 (54.4, 55.4) 57.7 (57.3, 58.2) 57.2 (56.8, 57.6)Women 4370 (63.4) 5578 (44.3) 867 (42.7) 3273 (58.6) 3457 (52.2) 17 545 (52.8)BMI, mean (95% CI), kg/m2

At age 25 years 20.8 (20.7, 20.9) 23.0 (22.9, 23.1) 25.8 (25.4, 26.1) 23.7 (23.6, 23.9) 27.9 (27.6, 28.2) 23.4 (23.3, 23.5)At 10 years prior baseline 21.6 (21.6, 21.7) 25.5 (25.5, 25.6) 32.9 (32.6, 33.2) 26.8 (26.7, 26.9) 35.5 (35.2, 35.7) 26.6 (26.5, 26.7)Maximum 24.2 (24.1, 24.3) 28.9 (28.9, 29.0) 34.4 (34.1, 34.7) 34.7 (34.5, 34.8) 40.1 (39.9, 40.4) 30.6 (30.5, 30.8)At survey 22.1 (22.0, 22.2) 26.8 (26.7, 26.8) 27.5 (27.3, 27.6) 33.3 (33.2, 33.5) 36.9 (36.7, 37.1) 28.2 (28.1, 28.3)

Waist circumference, mean (95% CI), cm 83.1 (82.9, 83.4) 96.2 (96.0, 96.4) 98.9 (98.3, 99.4) 109.4 (109.0, 109.9) 117.1 (116.7, 117.6) 98.2 (97.9, 98.5)

RaceNon-Hispanic White 4422 (80.6) 6670 (78.5) 865 (73.2) 2419 (73.5) 2999 (75.8) 17 375 (77.6)Non-Hispanic Black 1399 (7.0) 2433 (8.7) 484 (13.7) 1457 (13.0) 1704 (12.9) 7477 (9.8)Mexican American 862 (2.4) 2394 (4.7) 468 (6.8) 1094 (5.9) 1174 (5.4) 5992 (4.5)Others 1102 (10.0) 1435 (8.1) 165 (6.3) 532 (7.7) 504 (5.8) 3738 (8.1)

EducationLess than high school 2273 (19.1) 4506 (23.3) 946 (35.4) 1849 (24.6) 2236 (24.7) 11 810 (23.1)High school or equivalent 1922 (26.4) 3152 (28.1) 443 (27.4) 1396 (29.0) 1619 (29.8) 8532 (28.0)College or above 3560 (54.2) 5243 (48.4) 586 (37.0) 2250 (46.3) 2514 (45.3) 14 153 (48.6)


Marital statusMarried 4812 (67.2) 8387 (70.3) 1128 (62.3) 3423 (68.2) 3845 (65.7) 21 595 (68.1)Separated 2355 (25.8) 3660 (23.5) 725 (31.7) 1622 (25.0) 1993 (26.6) 10 355 (25.2)Never married 548 (6.2) 760 (5.3) 116 (5.5) 398 (6.0) 491 (7.0) 2313 (5.9)


Smoking statusNever smoker 3576 (45.6) 6043 (45.0) 901 (45.5) 2689 (46.1) 3324 (49.8) 16 533 (46.1)

Page 43 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Ever smoker 2123 (28.3) 4406 (35.1) 634 (30.1) 1807 (35.4) 2149 (34.3) 11 119 (32.9)Current smoker 2082 (26.1) 2476 (19.9) 446 (24.4) 1004 (18.5) 907 (15.9) 6915 (20.9)

17 062 (49.6) Family history of chronic metabolic diseases

3288 (43.5) 6012 (47.6) 1058 (52.0) 2994 (54.6) 3710 (59.0)

Self-reported chronic diseasesDiabetes 413 (3.5) 1487 (7.7) 614 (23.5) 791 (11.2) 1847 (23.8) 5152 (10.4)Hypertension 1725 (17.3) 4342 (29.8) 961 (44.5) 2297 (40.3) 3454 (52.2) 12 779 (32.3)Cardiovascular diseases 855 (7.6) 1883 (11.5) 509 (22.1) 780 (12.2) 1248 (16.1) 5257 (11.8)

3089 (9.5)Chronic obstructive pulmonary diseases

715 (9.6) 969 (7.8) 217 (12.4) 516 (10.2) 672 (11.5)

Cancer 1059 (14.0) 1729 (13.6) 286 (14.7) 551 (11.5) 776 (12.5) 4401 (13.2)a Complex survey design factors were taken into account for all estimates; percentages may not add up to 100% due to missing values.b From ten years before baseline to baseline, stable normal: BMI10prior < 25 kg/m2 and BMIbaseline < 25 kg/m2; max overweight: either BMI10prior or BMIbaseline was 25.0-29.9 kg/m2 and the other BMI not ≥30.0 kg/m2; losing: BMI10prior ≥ 30 kg/m2 and 25 ≤ BMIbaseline < 30 kg/m2; gaining: 25 ≤ BMI10prior < 30 kg/m2 and BMIbaseline ≥ 30 kg/m2; stable obese: BMI10prior ≥ 30 kg/m2 and BMIbaseline ≥ 30 kg/m2.

Page 44 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S4 Hazard ratios of all cause mortality with adiposity status at different age points among 36 812 participants in NHANES 1988-2014 aStandard BMI groupsUnderweight Normal weight Overweight Obese

BMI at age 25 yearsMean (95% CI) 17.5 (17.4, 17.6) 21.9 (21.8, 21.9) 26.8 (26.8, 26.9) 34.3 (34.0, 34.6)No. of participants 1902 21 841 7814 2358No. of deaths/person-years 455/21 174 6175/225 062 2323/73 018 696/20 201Model 1 1.00 (0.89, 1.12) 1.00 1.19 (1.11, 1.27) 1.65 (1.45, 1.87)Model 2 0.98 (0.86, 1.12) 1.00 1.17 (1.09, 1.25) 1.43 (1.24, 1.65)

BMI at ten years before baselineMean (95% CI) 17.4 (17.3, 17.6) 22.4 (22.4, 22.5) 27.2 (27.2, 27.3) 34.9 (34.8, 35.1)No. of participants 547 12 651 13 119 8401No. of deaths/person-years 125/6034 3376/137 158 3971/129 407 2524/73 664Model 1 1.53 (1.23, 1.91) 1.00 1.06 (1.00, 1.13) 1.42 (1.32, 1.53)Model 2 1.52 (1.20, 1.92) 1.00 1.02 (0.95, 1.09) 1.22 (1.13, 1.32)

BMI at baselineMean (95% CI) 17.4 (17.3, 17.5) 22.6 (22.5, 22.6) 27.3 (27.3, 27.4) 35.2 (35.1, 35.4)No. of participants 547 9827 13 216 12 461No. of deaths/person-years 275/4282 3347/98 960 3923/135 093 2955/120 644Model 1 1.76 (1.45, 2.13) 1.00 0.92 (0.85, 0.99) 1.16 (1.07, 1.25)Model 2 1.75 (1.42, 2.15) 1.00 0.89 (0.82, 0.96) 1.02 (0.94, 1.11)

Baseline waist quartile Q1 Q2 Q3 Q4Mean (95% CI) 81.3 (81.2, 81.5) 94.1 (94.0, 94.1) 103.0 (102.9, 103.1) 118.6 (118.3, 118.9)No. of participants 8600 8587 8704 8597No. of deaths/person-years 2453/91 738 2446/89 923 2554/86 128 2332/78 984Model 1 1.12 (1.03, 1.22) 1.00 1.11 (1.02, 1.22) 1.39 (1.24, 1.55)Model 2 1.16 (1.06, 1.27) 1.00 1.10 (1.00, 1.20) 1.23 (1.10, 1.37)

a Complex survey designs are considered for all estimates. A total of 2897, 2094, 761, 2324 participants were excluded due to missing values of adiposity indicator at specific age point respectively. Model 1 were adjusted for baseline age, sex, race, education level, marital status, income level, drinking status, smoking status (never, ever, current) and family history of chronic metabolic diseases. Model 2 were additionally adjusted for baseline history of diabetes, hypertension, cardiovascular diseases, cancer and chronic obstructive pulmonary diseases based on model 1.

Page 45 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S5 Hazard ratios of all cause mortality with weight change patterns stratified by study period aWeight change patterns bStratified by study period Stable Normal Maximum overweight Loss Gain Stable obese

From age 25 years to ten years before baselineNHANES 1988-2014

No. of deaths/person years 3094/133 393 3849/127 257 160/3724 1837/54 404 524/16 220HR (95% CI) 1.00 1.08 (1.01, 1.15) 1.28 (0.92, 1.79) 1.33 (1.22, 1.45) 1.93 (1.72, 2.16)

NHANES Ⅲ (1988-1994)No. of deaths/person years 1777/65 057 2023/54 490 81/1555 891/20 638 241/5700HR (95% CI) 1.00 1.09 (1.01, 1.17) 1.21 (0.80, 1.84) 1.40 (1.25, 1.57) 1.98 (1.70, 2.30)

NHAENS 1999-2014No. of deaths/person-years 1317/68 336 1826/72 767 79/2169 946/33 766 283/10 520HR (95% CI) 1.00 1.05 (0.96, 1.15) 1.57 (1.18, 2.08) 1.25 (1.13, 1.38) 1.83 (1.56, 2.16)

From age 25 years to baselineNHANES 1988-2014

No. of deaths/person years 2677/87 892 3915/131 968 264/4923 2302/98 464 423/15 156HR (95% CI) 1.00 0.94 (0.86, 1.03) 1.21 (0.93, 1.58) 1.11 (1.01, 1.21) 1.81 (1.59, 2.05)

NHANES Ⅲ (1988-1994)No. of deaths/person years 1515/43 435 2114/59 845 122/2059 1173/39 153 204/5277HR (95% CI) 1.00 0.94 (0.85, 1.05) 1.10 (0.77, 1.57) 1.18 (1.06, 1.32) 1.89 (1.59, 2.24)

NHAENS 1999-2014No. of deaths/person-years 1162/44 457 1801/72 123 142/2864 1129/59 312 219/9878HR (95% CI) 1.00 0.93 (0.85, 1.03) 1.53 (1.24, 1.89) 0.99 (0.90, 1.08) 1.62 (1.35, 1.95)

From ten years before baseline to baselineNHANES 1988-2014

No. of deaths/person years 2362/81 771 3924/131 762 852/16 428 1137/58 251 1652/56 972HR (95% CI) 1.00 0.96 (0.88, 1.04) 1.44 (1.28, 1.61) 1.09 (0.95, 1.25) 1.33 (1.22, 1.45)

NHANES sⅢ (1988-1994)HR (95% CI) 1371/40 942 2126/60 056 427/6886 612/24 403 810/20 852Multivariable-adjusted HR (95% CI) 1.00 0.95 (0.86, 1.06) 1.45 (1.22, 1.72) 1.16 (0.96, 1.39) 1.41 (1.26, 1.58)

NHAENS 1999-2014No. of deaths/person years 991/40 829 1798/71 706 425/9542 525/33 848 842/36 120HR (95% CI) 1.00 0.97 (0.88, 1.07) 1.45 (1.23, 1.70) 0.98 (0.87, 1.11) 1.22 (1.11, 1.35)

a Complex survey designs are considered for all estimates. A total of 3382, 3032, 2230 participants were excluded in analyses from age 25 years to ten years before baseline, age 25 years to baseline, ten years before baseline to baseline respectively due to missing values of BMI at both times. Risk estimates were adjusted for baseline age, sex,

Page 46 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack.b Weight change patterns were defined by BMI at two time points. Stable normal participants had BMI <25.0 kg/m2 at both times, maximum overweight participants had BMI in 25.0-29.9 kg/m2 at either time but not ≥30.0 kg/m2 at the other time, weight loss participants had BMI ≥30.0 kg/m2 at younger age while <30.0 kg/m2 later, weight gain participants had BMI <30.0 kg/m2 at younger age while ≥30.0 kg/m2 later, and stable obese participants had BMI ≥30.0 kg/m2 at both times.

Page 47 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S6 Hazard ratios of all cause and cause specific mortality with weight change patterns across adulthood in 36 812 participants in NHANES 1988-2014 with further adjustment of disease status a

Weight change patternsCause of death Stable normal Maximum overweight Loss Gain Stable obeseFrom age 25 years to ten years before baselineAll cause

No. of deaths/Person-years 3094/133 393 3849/127 257 160/3724 1837/54 404 524/16 220HR (95% CI) 1.00 1.04 (0.97, 1.11) 1.19 (0.82, 1.73) 1.16 (1.06, 1.27) 1.56 (1.35, 1.79)

Cardiovascular diseasesNo. of deaths 745 1096 44 515 162HR (95% CI) 1.00 1.14 (0.99, 1.31) 1.06 (0.69, 1.63) 1.28 (1.06, 1.54) 1.98 (1.51, 2.59)

CancerNo. of deaths 789 872 27 380 90HR (95% CI) 1.00 1.01 (0.87, 1.17) 0.94 (0.48, 1.82) 1.22 (0.98, 1.53) 1.11 (0.86, 1.45)

Other causesNo. of deaths 1560 1881 89 942 272HR (95% CI) 1.00 1.04 (0.96, 1.14) 1.49 (0.93, 2.37) 1.26 (1.12, 1.42) 1.98 (1.62, 2.43)

From age 25 years to baselineAll cause

No. of deaths/Person-years 2677/87 892 3915/131 968 264/4923 2302/98 464 423/15 156HR (95% CI) 1.00 0.93 (0.85, 1.01) 1.09 (0.81, 1.47) 1.00 (0.92, 1.10) 1.47 (1.27, 1.70)

Cardiovascular diseasesNo. of deaths 687 1071 80 629 125HR (95% CI) 1.00 0.97 (0.82, 1.15) 1.09 (0.75, 1.60) 1.05 (0.91, 1.22) 1.74 (1.31, 2.30)



From ten years before baseline to baselineAll cause

No. of deaths/Person-years 2362/81 771 3924/131 762 852/16 428 1137/58 251 1652/56 972HR (95% CI) 1.00 0.93 (0.84, 1.02) 1.23 (1.10, 1.38) 1.00 (0.87, 1.14) 1.13 (1.03, 1.24)

Cardiovascular diseases

Page 48 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


No. of deaths 605 1058 270 318 455HR (95% CI) 1.00 0.89 (0.76, 1.06) 1.40 (1.11, 1.79) 1.03 (0.84, 1.26) 1.18 (0.98, 1.42)



a For all cause mortality, risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current), family history of diabetes and heart attack and baseline history of diabetes, hypertension, cardiovascular diseases, cancer and chronic obstructive pulmonary diseases. For cardiovascular disease mortality, risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current), family history of diabetes and heart attack and baseline history of diabetes, hypertension, cancer and chronic obstructive pulmonary diseases. For cancer mortality, risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current), family history of diabetes and heart attack and baseline history of diabetes, hypertension, cardiovascular diseases and chronic obstructive pulmonary diseases. For other-causes mortality, risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current), family history of diabetes and heart attack and baseline history of cardiovascular diseases and cancer.

Page 49 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S7 Hazard ratios of all cause mortality with weight change patterns among 36 812 stratified by age at death aWeight change patternsStable Normal Maximum overweight Loss Gain Stable obese P for interaction

From age 25 years to ten years before baseline <.0001Stratified by death age groupDead before 65y

No. of deaths 522 459 23 238 140Multivariable-adjusted HR (95% CI) 1.00 1.10 (0.91, 1.32) 1.96 (1.02, 3.79) 1.54 (1.25, 1.89) 2.10 (1.68, 2.63)

Dead between 66y and 75yNo. of deaths 558 688 29 407 132Multivariable-adjusted HR (95% CI) 1.00 1.12 (0.96, 1.31) 1.40 (0.89, 2.21) 1.06 (0.91, 1.24) 1.27 (0.95, 1.68)

Dead between 76y-85yNo. of deaths 975 1341 58 682 156Multivariable-adjusted HR (95% CI) 1.00 1.11 (0.99, 1.26) 1.02 (0.69, 1.52) 1.24 (1.09, 1.41) 1.38 (1.16, 1.66)

Dead after 85yNo. of deaths 1039 1361 50 510 96Multivariable-adjusted HR (95% CI) 1.05 (0.92, 1.21) 0.87 (0.49, 1.55) 1.29 (1.08, 1.55) 1.84 (1.41, 2.40)

From age 25 years to baselineStratified by death age group <.0001Dead before 65y

No. of deaths 335 704 44 421 120Multivariable-adjusted HR (95% CI) 1.00 0.91 (0.72, 1.16) 1.38 (0.91, 2.10) 1.24 (1.00, 1.53) 2.12 (1.64, 2.74)

Dead between 66y and 75yNo. of deaths 457 1352 55 510 107Multivariable-adjusted HR (95% CI) 1.00 0.95 (0.79, 1.13) 1.18 (0.81, 1.72) 0.86 (0.72, 1.01) 1.10 (0.79, 1.53)


Dead after 85yNo. of deaths 1013 3915 79 551 67Multivariable-adjusted HR (95% CI) 1.00 1.00 (0.87, 1.15) 1.12 (0.65, 1.91) 1.13 (0.92, 1.39) 1.51 (1.13, 2.01)

From ten years before baseline to baselineStratified by death age group <.0001Dead before 65y 307 460 98 255 284

No. of deaths 1.00 0.93 (0.72, 1.19) 1.64 (1.16, 2.32) 1.21 (0.93, 1.59) 1.65 (1.31, 2.08)

Page 50 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Multivariable-adjusted HR (95% CI)

Dead between 66y and 75y 395 685 175 238 396No. of deaths 1.00 0.97 (0.81, 1.17) 1.15 (0.88, 1.51) 0.87 (0.66, 1.15) 0.94 (0.80, 1.10)Multivariable-adjusted HR (95% CI)


Dead after 85yNo. of deaths 891 1441 272 254 386Multivariable-adjusted HR (95% CI) 1.00 0.96 (0.83, 1.09) 1.40 (1.11, 1.77) 1.00 (0.77, 1.32) 1.26 (1.03, 1.53)

a Complex survey designs are considered for all estimates. Risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack.

Page 51 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S8 Hazard ratios of all cause mortality with weight change patterns after excluding participants who were underweight in NHANES 1988-2014 aWeight change patternsStable Normal Maximum overweight Loss Gain Stable obese

From age 25 years to ten years before baselineAll participants

No. of deaths 3094 3849 160 1837 524HR (95% CI) 1.00 1.08 (1.01, 1.15) 1.28 (0.92, 1.79) 1.33 (1.22, 1.45) 1.93 (1.72, 2.16)

Underweight people excludedNo. of deaths 2751 3741 159 1788 524HR (95% CI) 1.00 1.09 (1.02, 1.16) 1.31 (0.94, 1.82) 1.35 (1.23, 1.47) 1.94 (1.73, 2.18)

From age 25 years to baselineAll participants

No. of deaths 2677 3915 264 2302 423HR (95% CI) 1.00 0.94 (0.86, 1.03) 1.21 (0.93, 1.58) 1.11 (1.01, 1.21) 1.81 (1.59, 2.05)


From ten years before baseline to baselineAll participants

No. of deaths 2362 3924 852 1137 1652HR (95% CI) 1.00 0.96 (0.88, 1.04) 1.44 (1.28, 1.61) 1.09 (0.95, 1.25) 1.33 (1.22, 1.45)


a Complex survey designs are considered for all estimates. A total of 3382, 3032, 2230 participants were excluded in analyses from age 25 years to ten years before baseline, age 25 years to baseline, ten years before baseline to baseline respectively due to missing values of BMI at both times. Risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack.

Page 52 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S9 Hazard ratios of all cause and cause specific mortality with weight change patterns across adulthood among 31 077 participants after excluding deaths within three years of follow-up in NHANES 1988-2014 a

Weight change patternsCause of death b Stable normal Maximum overweight Loss Gain Stable obese

All cause

From age 25 years to ten years before baselineNo. of deaths 2597 3118 127 1462 404Model 1 1.00 1.05 (0.97, 1.13) 1.30 (0.89, 1.88) 1.32 (1.20, 1.45) 1.98 (1.74, 2.26)Model 2 1.00 1.02 (0.94, 1.10) 1.24 (0.84, 1.84) 1.17 (1.06, 1.29) 1.64 (1.41, 1.91)From age 25 years to baseline No. of deaths 2172 3187 188 1923 343Model 1 1.00 0.93 (0.84, 1.02) 1.19 (0.86, 1.63) 1.12 (1.02, 1.24) 1.90 (1.65, 2.18)Model 2 1.00 0.92 (0.83, 1.01) 1.10 (0.78, 1.54) 1.03 (0.93, 1.14) 1.58 (1.36, 1.85)From ten years before baseline to baselineNo. of deaths 1940 3172 634 982 1335Model 1 1.00 0.93 (0.84, 1.03) 1.42 (1.23, 1.63) 1.12 (0.96, 1.31) 1.34 (1.22, 1.48)Model 2 1.00 0.91 (0.81, 1.01) 1.23 (1.08, 1.41) 1.03 (0.88, 1.20) 1.16 (1.04, 1.28)

Cardiovascular diseases

From age 25 years to ten years before baselineNo. of deaths 623 879 39 406 117Model 1 1.00 1.16 (1.00, 1.35) 1.20 (0.79, 1.81) 1.50 (1.25, 1.81) 2.56 (1.93, 3.40)Model 2 1.00 1.11 (0.95, 1.30) 1.20 (0.78, 1.84) 1.30 (1.07, 1.59) 2.09 (1.55, 2.83)From age 25 years to baseline No. of deaths 552 859 54 529 99Model 1 1.00 0.98 (0.81, 1.18) 1.11 (0.73, 1.69) 1.21 (1.04, 1.41) 2.37 (1.77, 3.17)Model 2 1.00 0.95 (0.79, 1.15) 1.07 (0.69, 1.66) 1.08 (0.92, 1.26) 1.93 (1.43, 2.60)From ten years before baseline to baselineNo. of deaths 485 858 194 279 367Model 1 1.00 0.91 (0.75, 1.09) 1.57 (1.20, 2.04) 1.14 (0.92, 1.40) 1.45 (1.19, 1.77)Model 2 1.00 0.87 (0.72, 1.05) 1.36 (1.03, 1.80) 1.03 (0.84, 1.27) 1.22 (1.00, 1.50)

Cancer

From age 25 years to ten years before baseline

Page 53 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


No. of deaths 646 684 18 294 70Model 1 1.00 1.02 (0.87, 1.20) 1.04 (0.51, 2.13) 1.26 (1.00, 1.59) 1.22 (0.86, 1.74)Model 2 1.00 1.02 (0.87, 1.20) 1.03 (0.51, 2.09) 1.23 (0.97, 1.57) 1.19 (0.85, 1.65)From age 25 years to baselineNo. of deaths 474 733 29 440 60Model 1 1.00 0.98 (0.81, 1.18) 0.86 (0.46, 1.61) 1.16 (0.97, 1.38) 1.28 (0.85, 1.92)Model 2 1.00 0.99 (0.82, 1.20) 0.86 (0.46, 1.59) 1.16 (0.98, 1.37) 1.26 (0.85, 1.86)From ten years before baseline to baselineNo. of deaths 451 697 99 227 278Model 1 1.00 0.96 (0.77, 1.19) 1.14 (0.79, 1.64) 1.06 (0.80, 1.40) 1.28 (1.04, 1.58)Model 2 1.00 0.97 (0.78, 1.20) 1.11 (0.77, 1.59) 1.07 (0.82, 1.40) 1.27 (1.03, 1.57)

Other causes

From age 25 years to ten years before baselineNo. of deaths 1328 1555 70 762 217Model 1 1.00 1.02 (0.92, 1.12) 1.48 (0.90, 2.43) 1.27 (1.12, 1.45) 2.16 (1.73, 2.69)Model 2 1.00 1.02 (0.92, 1.12) 1.44 (0.86, 2.42) 1.26 (1.10, 1.43) 2.08 (1.66, 2.60)From age 25 years to baselineNo. of deaths 1146 1595 105 954 184Model 1 1.00 0.88 (0.79, 0.98) 1.39 (0.96, 2.03) 1.09 (0.96, 1.23) 2.01 (1.59, 2.55)Model 2 1.00 0.88 (0.79, 0.99) 1.35 (0.90, 2.01) 1.07 (0.94, 1.21) 1.94 (1.54, 2.45)From ten years before baseline to baselineNo. of deaths 1004 1617 341 476 690Model 1 1.00 0.94 (0.83, 1.05) 1.47 (1.20, 1.80) 1.13 (0.96, 1.39) 1.33 (1.16, 1.53)Model 2 1.00 0.93 (0.83, 1.05) 1.44 (1.18, 1.76) 1.12 (0.93, 1.35) 1.30 (1.13, 1.50)

a Complex survey designs are considered for all estimates. Model 1 were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack. b For all cause mortality, model 2 were additionally adjusted for baseline history of diabetes, hypertension, cardiovascular diseases, cancer and chronic obstructive pulmonary diseases. For CVD mortality, model 2 were additionally adjusted for baseline history of diabetes, hypertension, cancer and chronic obstructive pulmonary diseases. For cancer mortality, model 2 were additionally adjusted for baseline history of diabetes, hypertension, cardiovascular diseases and chronic obstructive pulmonary diseases. For other-causes mortality, model 2 were additionally adjusted for baseline history of cardiovascular diseases and cancer.

Page 54 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Table S10 Hazard ratios (95% CI) of all cause and cause specific mortality with weight change in three life periods in 36 812 participants in NHANES 1988-2014 with further adjustment of disease status a

Weight change in three life periodsCause of death b Weight loss ≥2.5 kg Weight loss <2.5 kg

or gain <2.5 kg 2.5 kg ≤ Weight gain < 10 kg 10 kg ≤ Weight gain <20 kg Weight gain ≥20 kg

From age 25 years to ten years before baselineAll cause

No. of deaths 762 2152 2661 2191 1698 Age-adjusted mortality rate c 37.2 (34.5, 39.8) 30.0 (28.7, 31.3) 26.2 (25.2, 27.2) 26.2 (25.1, 27.3) 31.1 (29.6, 32.7)HR (95% CI) 1.12 (0.95, 1.31) 1.00 0.97 (0.87, 1.08) 0.98 (0.87, 1.10) 1.10 (0.98, 1.23)

Cardiovascular diseases No. of deaths 211 565 703 618 465

HR (95% CI) 0.96 (0.79, 1.17) 1.00 0.87 (0.73, 1.04) 0.87 (0.72, 1.05) 1.03 (0.87, 1.21)Cancer No. of deaths 170 510 662 453 363 HR (95% CI) 1.18 (0.92, 1.51) 1.00 1.00 (0.83, 1.19) 1.03 (0.81, 1.30) 1.16 (0.91, 1.49)Other causes No. of deaths 381 1077 1296 1120 870 HR (95% CI) 1.15 (0.90, 1.45) 1.00 1.02 (0.88, 1.19) 1.13 (0.99, 1.29) 1.33 (1.17, 1.51)From age 25 years to baselineAll cause No. of deaths 1801 1059 2027 2417 2277 Age-adjusted mortality rate c 44.4 (42.2, 46.6) 29.1 (27.3, 30.9) 26.5 (25.4, 27.7) 25.6 (24.6, 26.6) 25.7 (24.5, 26.8) HR (95% CI) 1.16 (1.00, 1.35) 1.00 0.97 (0.87, 1.09) 0.94 (0.84, 1.05) 1.00 (0.90, 1.11)Cardiovascular disease No. of deaths 506 296 563 622 605 HR (95% CI) 0.95 (0.75, 1.21) 1.00 0.89 (0.73, 1.09) 0.86 (0.73, 1.01) 0.93 (0.78, 1.12)Cancer

No. of deaths 394 210 475 572 537 HR (95% CI) 1.46 (1.16, 1.84) 1.00 1.12 (0.93, 1.35) 1.10 (0.89, 1.35) 1.17 (0.99, 1.39)Other causes No. of deaths 901 553 989 1223 1135 HR (95% CI) 1.26 (1.04, 1.54) 1.00 1.02 (0.87, 1.19) 1.00 (0.86, 1.16) 1.17 (1.03, 1.34)From ten years before baseline to baselineAll cause

Page 55 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


No. of deaths 3853 2205 2184 1115 570 Age-adjusted mortality rate c 38.5 (37.2, 39.7) 25.5 (24.4, 26.6) 23.6 (22.6, 24.6) 25.4 (23.7, 27.1) 28.5 (25.3, 31.6) HR (95% CI) 1.25 (1.12, 1.38) 1.00 1.00 (0.92, 1.08) 1.09 (0.98, 1.20) 1.37 (1.14, 1.66)Cardiovascular disease No. of deaths 1106 613 567 267 153 HR (95% CI) 1.14 (0.92, 1.41) 1.00 1.04 (0.88, 1.23) 0.97 (0.76, 1.22) 1.49 (1.09, 2.05)Cancer No. of deaths 773 478 562 281 130 HR (95% CI) 1.27 (1.07, 1.50) 1.00 1.04 (0.87, 1.26) 1.07 (0.88, 1.29) 1.22 (0.92, 1.62)Other causes No. of deaths 1974 1114 1055 567 287 HR (95% CI) 1.34 (1.19, 1.51) 1.00 0.96 (0.85, 1.08) 1.20 (1.04, 1.38) 1.59 (1.22, 2.08)a Complex survey designs were taken into account for all estimates. A total of 3382, 3032, 2230 participants were excluded in analyses from age 25 years to ten years before baseline, age 25 years to baseline, ten years before baseline to baseline respectively due to missing values of BMI at both times. In analyses from age 25 years to ten years before baseline and age 25 years to baseline, risk estimates were adjusted for baseline age, sex, race/ethnicity, height, weight at age 25, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack. In analysis from ten years before baseline to baseline, risk estimates were adjusted for baseline age, sex, race/ethnicity, height, weight at ten years ago, education level, household income level, marital status, drinking and smoking status (never, ever, current) and family history of diabetes and heart attack.b For all cause mortality, risk estimates were additionally adjusted for baseline history of diabetes, hypertension, cardiovascular diseases, cancer and chronic obstructive pulmonary diseases. For cardiovascular disease mortality, risk estimates were additionally adjusted for baseline history of diabetes, hypertension, cancer and chronic obstructive pulmonary diseases. For cancer mortality, risk estimates additionally adjusted for baseline history of diabetes, hypertension, cardiovascular diseases and chronic obstructive pulmonary diseases. For other-causes mortality, risk estimates additionally adjusted for baseline history of cardiovascular diseases and cancer based on model 2.

c Mortality rates per 1000 person-years, directly standardized to the age distribution of the entire study population.

Page 56 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Confidential: For Review OnlyFigure S1 Flow chart of inclusion and exclusion of study participants

Page 57 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Figure S2 Association between fine weight change patterns across adulthood and all cause mortality in NHANES 1988-2014. All risk estimates were calculated using multivariable Cox regression adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status and family history of diabetes and heart attack.a Only two participants were in the group so that the effect estimates were not stable and were not shown.b Only three participants were in the group so that the effect estimates were not stable and were not shown.c Only five participants were in the group so that the effect estimates were not stable and were not shown.

Page 58 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Confidential: For Review OnlyFigure S3 Dose response association between weight change across adulthood and risk of cause specific mortality. Associations were examined by multivariable Cox regression based on restricted cubic splines. Dash line represents estimates of hazard ratios and dotted line represents 95% confidence intervals. Risk estimates were adjusted for baseline age, sex, race/ethnicity, education level, household income level, marital status, drinking and smoking status and family history of diabetes and heart attack. For weight change from age 25 years to ten years before baseline or to baseline, weight at age 25 and baseline height were also adjusted. For weight change from ten years before baseline to baseline, weight at 10 years ago and baseline height were adjusted.

Page 59 of 58


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Documents

Confidential: For Review Only - BMJ · Confidential: For Review Only Weight change across adulthood in relation to all cause and cause specific mortality: a prospective cohort study