Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Confidential: For Review Only
Predicted lean body mass, fat mass, and all-cause and
cause-specific mortality in men: results from a prospective
US cohort study
Journal: BMJ
Manuscript ID BMJ.2018.043414.R1
Article Type: Research
BMJ Journal: BMJ
Date Submitted by the Author: 01-Apr-2018
Complete List of Authors: Lee, Dong Hoon; Harvard University T H Chan School of Public Health, Nutrition Keum, NaNa; Harvard University T H Chan School of Public Health, Nutrition; Dongguk University, Department of food science and Biotechnology Hu, Frank; Harvard University T H Chan School of Public Health, Nutrition and Epidemiology; Brigham and Women’s Hospital and Harvard Medical School, Channing Division of Network Medicine, Department of Medicine Orav, Endel; Harvard University T H Chan School of Public Health, Biostatistics; Brigham and Women’s Hospital, Department of Medicine Rimm, Eric; Harvard University T H Chan School of Public Health, Nutrition and Epidemiology; Brigham and Women’s Hospital and Harvard Medical
School, Channing Division of Network Medicine, Department of Medicine Willett, Walter; Harvard University T H Chan School of Public Health, Nutrition and Epidemiology; Brigham and Women’s Hospital and Harvard Medical School, Channing Division of Network Medicine, Department of Medicine Giovannucci, Edward; Harvard University T H Chan School of Public Health, Nutrition and Epidemiology; Brigham and Women’s Hospital and Harvard Medical School, Channing Division of Network Medicine, Department of Medicine
Keywords: body mass index, body composition, lean body mass, fat mass, mortality, obesity paradox
https://mc.manuscriptcentral.com/bmj
BMJ
Confidential: For Review Only
1
Predicted lean body mass, fat mass, and all-cause and cause-specific mortality in men:
results from a prospective US cohort study
Dong Hoon Lee, NaNa Keum, Frank B. Hu, E. John Orav, Eric B. Rimm, Walter C. Willett,
Edward L. Giovannucci
Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA 02115, USA,
Dong Hoon Lee, post-doctoral research fellow, NaNa Keum, post-doctoral research fellow,
Frank B. Hu, professor, Eric B. Rimm, professor, Walter C. Willett, professor, Edward L.
Giovannucci, professor
Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA 02115,
USA, Frank B. Hu, professor, Eric B. Rimm, professor, Walter C. Willett, professor, Edward L.
Giovannucci, professor
Department of food science and Biotechnology, Dongguk University, Goyang, South Korea,
NaNa Keum, assistant professor, Channing Division of Network Medicine
Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston,
MA 02115, USA, Frank B. Hu, professor, Eric B. Rimm, professor, Walter C. Willett, professor,
Edward L. Giovannucci, professor
Department of Medicine, Brigham and Women’s Hospital, Boston, MA 02115, USA, E. John
Orav, associate professor
Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, MA 02115,
USA, E. John Orav, associate professor.
Corresponding author:
Edward L. Giovannucci, Department of Nutrition, Harvard T.H Chan School of Public Health,
665 Huntington Avenue, Bldg. 2, Room 371, Boston, MA 02115
Phone: 617-432-4648, Fax: 617-432-2435, Email: [email protected]
Word count: 4,229
Number of tables and figures: 4 tables and 1 figure
Page 1 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
2
ABSTRACT
Objective: To investigate the association of predicted lean body mass (LBM), fat mass (FM) and
body mass index (BMI) with all-cause and cause-specific mortality in men.
Design: Prospective cohort study.
Setting: Health professionals in the United States
Participants: 38,021 men (aged 40-75 years) from the Health Professionals Follow-up Study
were followed-up for death (1987-2012).
Main outcome measures: All-cause and cause-specific mortality.
Results: Using validated anthropometric prediction equations developed from the National
Health and Nutrition Examination Survey, LBM and FM were estimated for all participants.
During a mean of 21.4 years of follow-up, we identified 12,356 deaths. We consistently observed
a J-shaped association between BMI and mortality. Multivariable-adjusted Cox models including
both predicted FM and LBM showed a strong positive monotonic association between predicted
FM and mortality. Compared to those in the lowest quintile of predicted FM, men in the highest
quintile had 35% (95% confidence interval (CI): 26 to 46%), 67% (95% CI: 47 to 90%), and 24%
(95% CI: 8 to 42%) increased risk of mortality due to all causes, cardiovascular disease, and
cancer. In contrast, a U-shaped association was found between predicted LBM and mortality due
to all causes, cardiovascular disease, and cancer (P for non-linearity<0.001). However, there was
a strong inverse association between predicted LBM and mortality due to respiratory disease (P
for trend<0.001). Compared to those in the lowest quintile of predicted LBM, men in the highest
quintile had 50% (95% CI: 39 to 65%) decreased risk of death due to respiratory disease.
Conclusions: The shape of the relationship between BMI and mortality was determined by the
relationship between two body components (LBM and FM) and mortality. Our finding suggests
that the ‘obesity paradox’ controversy may be largely explained by low LBM, rather than low
FM, in the lower range of BMI.
Keywords: body mass index, body composition, lean body mass, fat mass, mortality, obesity
paradox
Page 2 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
3
“What this paper adds” box
Section 1: What is already known on this topic
• Numerous epidemiological studies have shown unexpected J-or U-shaped relationship
between body mass index (BMI) and mortality (‘obesity paradox’).
• The controversial issue of ‘obesity paradox’ may have arisen in part due to
underappreciation of different contributions of lean body mass (LBM) and fat mass (FM)
to BMI.
• Direct measure of body composition is difficult in large epidemiological settings, thus the
relationship between body composition and mortality is still unknown.
Section 2: What this study adds
• Using validated anthropometric prediction equations for body composition, this study
represents the first effort to comprehensively examined the association between lean
body mass, fat mass and mortality in a large prospective cohort study.
• Predicted fat mass showed a strong positive monotonic association with mortality, while
predicted lean body mass showed a strong U-shaped association with mortality.
• The ‘obesity paradox’ controversy may be explained largely by low LBM, rather than
low FM, in the lower range of BMI
Page 3 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
4
INTRODUCTION
Obesity is a major public health challenge in the United States and around the world.1 In 2013-
2014, more than two thirds of Americans were classified as overweight (defined as body mass
index (BMI) of 25-29.9 kg/m2) or obese (BMI of ≥30 kg/m
2).
2 BMI is known as a reasonably
good measure of general adiposity3, and many epidemiologic studies have provided evidence
supporting that obesity, assessed by BMI, is a significant risk factor for increased risk of many
chronic diseases as well as mortality.4-6
However, details of the shape of the association between
BMI and mortality has been a topic of considerable discussion as epidemiologic studies have
found various types of J-shaped, U-shaped, and linear relationships between BMI and mortality.7
For instance, in some studies, overweight was associated with increased mortality8, but in others,
the lowest mortality was observed among overweight individuals and mortality tended to
increase with lower BMI, even after accounting for smoking (residual confounding) and
preexisting disease (reverse causation).9 10
This pattern has come to be known as the “obesity
paradox”.11
Given the existing and rising number of overweight and obese adults in the US, these
divergent findings could cause a great deal of confusion among researchers, policy makers, and
the general public.
One important but underexplored methodological limitation in the current obesity
research is that BMI is an imperfect measure of adiposity.12-15
While BMI indicates overweight
relative to height, it does not discriminate between fat mass (FM) and lean body mass (LBM).16-
18 Given the same BMI, body composition is highly variable among individuals. This is
particularly important because FM and LBM may act differently on health outcomes including
mortality. Excess FM has shown to be detrimental for health,19
while growing evidence suggests
Page 4 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
5
that skeletal muscle, which accounts for majority of LBM, may be beneficial for health.20 21
Therefore, understanding different contributions of LBM and FM to BMI may provide new
insights on the ‘obesity paradox’ and deliver important clinical and public health messages
regarding healthy body composition beyond BMI. However, direct measurement of LBM is
particularly difficult in large epidemiological studies because it requires expensive and
sophisticated technologies like dual-energy X-ray absorptiometry (DXA) or imaging
technologies. Therefore, little is known about the influence of body composition, particularly
LBM, on mortality. A limited number of studies have used less accurate surrogate measures (e.g.,
arm circumference,22 23
total body potassium,24
skinfold,25
and bioelectrical impedance26
) or
direct measures27-33
to estimate body composition but these studies had relatively small sample
size, short period of follow-up, restricted study population (e.g., elderly) and/or potential biases
(e.g., confounding and reverse causation). Moreover, the association of LBM and FM with
cause-specific mortality is largely unknown.
Therefore, we used validated anthropometric prediction equations to estimate body
composition and examine the association of predicted LBM, FM and BMI with all-cause and
cause-specific mortality in a large prospective US cohort study of men. Application of validated
equations in a large cohort allowed us to estimate LBM and FM and examine the independent
roles of two different body components in relation to mortality, accounting for potential biases.
Page 5 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
6
METHODS
Study population
The Health Professionals Follow-up Study was initiated in 1986 when 51,529 male health
professionals aged 40–75 were enrolled. Participants were mailed questionnaires at baseline and
every two years thereafter to collect updated demographics, lifestyle, and medical information.
For the analysis, we included participants who had information on age, race, height, weight and
waist circumference, which were required to create predicted LBM and FM (N=40,764). We
excluded participants previously diagnosed with cancer or cardiovascular diseases (N=2,118)
and those with BMI <12.5 or >60 kg/m2
(N=625) at baseline. The final sample size was 38,021
men.
Exposure assessments
Derivation and validation of the predicted LBM and FM has been described in detail
previously.34
Briefly, we used a large US representative sample of 7,531 men who had measured
DXA from the National Health and Nutrition Examination Survey (NHANES). With DXA-
measured LBM and FM each as a dependent variable, a linear regression was performed using
age, race, height, weight, and waist circumference as independent predictors. Then, we validated
the developed equations in an independent validation group of 2,292 men and using obesity-
related biomarkers (i.e., triglycerides, total cholesterol, high-and low-density lipoprotein
cholesterol, glucose, insulin and C-reactive protein). The anthropometric prediction equations
had high predictive ability for LBM (R2=0.91, standard error of estimate (SEE)=2.6 kg) and FM
(R2=0.90, SEE=2.6 kg). Cross-validation in the validation group showed robustly high
Page 6 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
7
agreement between the actual and predicted LBM and FM with no evidence of bias. In an
additional validation, DXA-measured values and predicted values showed consistently high
agreement with similar errors across the range of LBM and FM. Scatter plots of the difference
between DXA and predicted values against DXA values showed no strong non-linear pattern
(Supplementary figure 1 and 2 and supplementary table 2 and 3). Moreover, the developed
equations performed well across different subgroups of the validation group (i.e., age, BMI, race,
smoking status, and disease status), and predicted FM showed similar correlations with obesity-
related biomarkers as DXA-measured FM.34
For a sensitivity analysis, we also used different
prediction equations that include additional polynomial terms of anthropometric measures. These
equations had similar R2 and SEEs but slightly improved fit in the extreme range of LBM and
FM (Supplementary figure 1 and 2 and supplementary table 2 and 3). The anthropometric
prediction equations are shown in the supplement (Supplementary table 1). Using the equations,
predicted LBM and FM were calculated for each cohort member based on their age, race, height,
weight, and waist circumference. Predicted LBM and FM were available in 1987, 1996, and
2008.
We collected information on height at enrollment in 1986, and weight from biennial
questionnaires.35 36
Distinct from the biennial questionnaire, participants were asked to measure
and report their waist circumferences to the nearest one-quarter inch using provided tape
measures and following the same instructions in 1987, 1996, and 2008. Non-responders received
follow-up mailings to increase the response rate. In our validation study, the correlation between
self-reported and technician-measured height, weight, and waist circumference were 0.94, 0.97,
and 0.95, respectively.35
Page 7 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
8
Ascertainment of outcomes
Deaths were identified by reports from the next of kin, postal authorities, or by searching the
National Death Index. More than 98% of deaths were ascertained from the follow up. Cause of
death was determined by physician review of medical records and death certificates. ICD-8
codes (International Classification of Diseases, 8th revision) were used to classify death due to
cardiovascular disease (codes 390-459, 795), cancer (codes 140-239), respiratory disease (codes
460-519), and other causes.
Ascertainment of covariates
Detailed information on age, race, smoking, and physical activity were collected in 1986 and
updated every two years from biannual questionnaires. Family history of cardiovascular disease
and cancer were assessed periodically. Dietary information was collected via validated food
frequency questionnaires every four years. The Alternate Healthy Eating Index (AHEI) was
calculated as an overall measure of diet quality.37
Statistical analyses
A Spearman correlation was calculated between predicted LBM and FM. Person-time of follow-
up was calculated from the age when the baseline predicted LBM and FM were available until
the age at death or the end of study (January 2012), whichever came first. Cox proportional
hazards models were used to estimate hazard ratios and 95% confidence interval (CI)s. We
stratified the analysis by age in months and calendar year of the questionnaire cycle.
Predicted FM and LBM were categorized into quintiles on the basis of the distribution of
exposures. We used predefined cut points for BMI (<18.5, 18.5-20.4, 20.5-22.4, 22.5-24.9, 25-
Page 8 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
9
27.4, 27.5-29.9, 30-34.9, and ≥35 kg/m2). For the main analysis, we used predicted FM, LBM,
and BMI measured at baseline to minimize the impact of underlying diseases on mortality. To
account for variation in body size, which is particularly important for LBM, we adjusted for
height by using residuals from the regression of LBM on height for LBM and by including
height as a continuous variable for FM in the models. In multivariable models, we adjusted for
potential confounders including race, family history of cardiovascular disease, family history of
cancer, smoking status, physical activity, total energy intake, alcohol consumption, and AHEI.
To examine the independent association of predicted LBM and FM in relation to mortality, we
further ran a multivariable model including both predicted LBM and FM. Test for trend was
conducted by treating the categorical predicted scores and BMI as continuous variables in the
model after assigning a median value for each category.
We also used restricted cubic splines with 5 knots at 5th
, 35th
, 50th
, 65th
, and 95th
percentiles to flexibly model the association between LBM and FM and mortality. We tested for
potential non-linearity using a likelihood ratio test comparing the model with only a linear term
to the model with linear and cubic spline terms.38-40
Given our a priori hypothesis that people
with low LBM in the lower BMI range cause the J-or U-shaped relationship between BMI and
mortality, we examined how the shape of BMI-mortality relationship changes after excluding
those with low LBM. For a sensitivity analysis, we additionally examined the shape of BMI-
mortality relationship after excluding those with low FM.
To evaluate the latency between predicted LBM and FM and mortality, we conducted
analyses using different lag times (approximately 0, 4+, 8+, and 12+ years). For each lagged
analysis, the baseline was shifted to 1987, 1990, 1994, and 1998, respectively, and predicted
LBM and FM were updated using three repeated measures accordingly. For example, for no lag
Page 9 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
10
time analysis (simple updated), we used the most updated predicted LBM and FM that were
closest to the time of death. For a lag time of 4+ years, we used predicted measures in 1987 for
follow-up from 1990 to 2000 and predicted measures in 1996 for follow-up from 2000 to 2012.
Similarly, for a lag times of 8+ years, we used predicted measures in 1987 for deaths in 1994-
2004 and predicted measures in 1996 for deaths in 2004-2012. Moreover, we conducted
stratified analyses to explore whether the association of predicted LBM and FM with mortality
varied across smoking status and age.
Several sensitivity analyses were conducted with no adjustment for physical activity,
exclusion of deaths that have occurred in the early follow-up period (2 years) and right-censoring
criteria for age (>85 years), and inclusion of baseline illness. We also conducted analyses using
different categories for predicted LBM, FM and BMI (i.e., quintiles and deciles). Lastly, we
tested the robustness of our findings using other prediction equations with polynomial terms. All
statistical tests were two-sided and P<0.05 was considered to determine statistical significance.
We used SAS 9.4 for all analyses (SAS institute).
Patient involvement
No patients were involved in setting the research question or the outcome measures, nor were
they involved in the design and implementation of the study. There are no plans to involve
patients in dissemination.
Page 10 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
11
RESULTS
Study participants
A total of 38,021 men were included in the analyses. Baseline characteristics of participants
according to BMI categories are presented in Table 1. The mean age was 54.4 years and the
mean BMI was 25.4 kg/m2. Predicted LBM increased with higher BMI. Predicted FM slightly
deceased in the second category of BMI (18.5-20.4 kg/m2) and then increased with higher BMI.
Moreover, men with lower BMI tended to have higher physical activity and AHEI score, peaking
in the third category of BMI (20.5-22.4 kg/m2). Although the number of men with underweight
(BMI<18 kg/m2) was small, they were taller and had higher waist circumference and lower
physical activity and AHEI score. The Spearman correlation between predicted LBM and FM
was 0.66 in men.
Page 11 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
12
Table 1 Age-standardized baseline characteristics according to body mass index in men (Health Professionals Follow-up
Study, 1987-2012) Body Mass Index (kg/m
2)
<18.5 18.5-20.4 20.5-24.9 22.5-24.9 25.0-27.4 27.5-29.9 30.0-34.9 ≥35.0
Person-years 1839 15337 92790 254122 243335 95023 52320 8275
Age (year)a 55.5 (10.4) 54.0 (10.8) 53.8 (10.2) 54.0 (9.9) 54.5 (9.7) 54.9 (9.6) 55.1 (9.4) 55.5 (10.1)
Height (cm) 185.4 (12.7) 179.4 (7.8) 178.7 (6.2) 178.5 (6.4) 178.2 (6.5) 178.7 (6.8) 178.5 (7.1) 176.4 (10.0)
Weight (kg) 60.8 (8.2) 64.1 (5.7) 69.6 (5.1) 75.9 (5.8) 82.9 (6.4) 91.4 (7.3) 101.2 (9.0) 118.2 (13.9)
Waist circumference (cm) 86.6 (12.4) 82.8 (5.3) 86.8 (5.1) 91.2 (5.6) 96.7 (6.1) 102.9 (6.8) 110.6 (7.9) 123.4 (11.4)
BMI (kg/m2) 17.6 (0.8) 19.8 (0.5) 21.7 (0.5) 23.7 (0.7) 26.0 (0.7) 28.5 (0.7) 31.7 (1.3) 37.9 (3.6)
Predicted fat mass (kg) 13.3 (5.0) 13.1 (2.5) 15.9 (2.4) 19.1 (2.6) 22.8 (2.9) 27.1 (3.3) 32.3 (4.1) 41.2 (6.5)
Predicted Lean body mass (kg) 40.4 (5.8) 47.5 (2.2) 50.6 (1.9) 53.9 (2.1) 57.4 (2.3) 61.2 (2.6) 65.9 (3.4) 75.2 (6.0)
Total energy intake (kcal/day) 2132 (610) 2023 (570) 2045 (599) 2002 (595) 1992 (609) 2002 (625) 2036 (639) 2089 (657)
Alcohol consumption (g/day) 14.2 (18.7) 9.7 (14.2) 10.9 (14.3) 11.5 (14.7) 11.8 (15.4) 11.7 (15.5) 10.9 (16.1) 8.9 (15.1)
AHEI (score) 51.4 (13.5) 54.1 (12.7) 54.3 (12.0) 53.8 (11.6) 52.4 (11.1) 51.5 (10.9) 50.7 (11.0) 49.3 (10.8)
Physical activity (MET-h/wk) 21.4 (35.8) 22.6 (27.0) 24.2 (28.6) 22.3 (27.4) 19.4 (23.9) 16.8 (22.0) 14.4 (20.9) 11.7 (14.9)
White (%) 98.4 99.2 99.5 99.3 99.2 98.8 98.7 99.4
Family history of CVD (%) 35.3 32.2 33.0 33.4 33.7 33.8 35.2 35.5
Family history of cancer (%) 17.6 16.8 17.2 16.8 17.5 16.9 16.8 15.4
Smoking status (%)
Never 47.4 56.5 56.0 50.5 45.8 44.1 42.3 41.1
Past 34.0 32.3 35.2 42.2 46.0 47.5 50.0 50.6
Current 18.6 11.2 8.9 7.3 8.2 8.4 7.8 8.3
Abbreviation: BMI, body mass index; AHEI, alternate healthy eating index; CVD, cardiovascular disease
Data are presented as means (SD) for continuous variables and percentages for categorical variables, unless otherwise indicated. a Value is not age adjusted
Page 12 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
13
All-cause mortality
During up to 25 years of follow-up (mean of 20.4 years), we identified 12,356 deaths. The
association of predicted FM and LBM with all-cause mortality in men is presented in Table 2. A
multivariable adjusted model showed a positive association between predicted FM and all-cause
mortality, while predicted LBM showed a U-shaped association with all-cause mortality. In a
mutually adjusted model including both predicted FM and LBM, we consistently observed a
strong positive association between predicted FM and all-cause mortality. Compared to those in
the lowest quintile of predicted FM, men in the highest quintiles had 35% (95% CI: 26 to 46%)
increased hazard of all-cause mortality. Moreover, predicted LBM showed a stronger U-shaped
association with all-cause mortality in the mutually adjusted model. Compared to those in the
lowest quintile of predicted LBM, men in the second to fourth quintiles had 8 to 10% decreased
hazard of all-cause mortality.
In Figure 1, we used restricted cubic splines to flexibly model and visualize the
relationship between predicted FM and LBM with all-cause mortality in men. The risk of all-
cause mortality was relatively flat and increased slightly until around 21 kg of predicted FM, and
then started to increase rapidly afterwards (P for non-linearity<0.001). The average BMI for men
with 21 kg of FM is 25 kg/m2. In respect to the strong U-shaped relationship between predicted
LBM and all-cause mortality, the plot showed a substantial reduction of the risk within the lower
range of predicted LBM, which reached the lowest risk around 55 kg and then increased
thereafter (P for non-linearity<0.001).
Page 13 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
14
Table 2. Hazard ratio (95% CI) of all-cause mortality according to predicted fat mass and lean body
mass in men (Health Professionals Follow-up Study)
Analysis
Hazard Ratio (95% CI)
No of
deaths
IR
/100,000py Model 1 Model 2 Model 3
Fat Massa,b
Quintile 1 1937 1265 1 (reference) 1 (reference) 1 (reference)
Quintile 2 2298 1504 1.09 (1.03 to 1.16) 1.06 (1.00 to 1.12) 1.08 (1.01 to 1.15)
Quintile 3 2297 1504 1.03 (0.97 to 1.09) 0.98 (0.92 to 1.04) 1.01 (0.94 to 1.07)
Quintile 4 2726 1789 1.23 (1.16 to 1.31) 1.13 (1.06 to 1.20) 1.16 (1.09 to 1.24)
Quintile 5 3098 2038 1.51 (1.42 to 1.60) 1.33 (1.25 to 1.41) 1.35 (1.26 to 1.46)
P-trend
<.001 <.001 <.001
Lean Body Massa,b
Quintile 1 2996 1969 1 (reference) 1 (reference) 1 (reference)
Quintile 2 2419 1585 0.93 (0.88 to 0.98) 0.93 (0.88 to 0.98) 0.92 (0.87 to 0.97)
Quintile 3 2324 1521 0.95 (0.90 to 1.01) 0.93 (0.88 to 0.98) 0.90 (0.85 to 0.96)
Quintile 4 2282 1494 1.03 (0.98 to 1.09) 1.00 (0.95 to 1.06) 0.92 (0.87 to 0.98)
Quintile 5 2335 1529 1.26 (1.20 to 1.34) 1.16 (1.10 to 1.23) 0.97 (0.91 to 1.04)
P-trend <.001 <.001 0.49
Model 1: adjusted for age.
Model 2: adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no),
physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol consumption (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total
energy intake (quintiles), and smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), Alternate Healthy Eating Index (quintiles).
Model 3: additionally, mutually adjusted for predicted fat mass and predicted lean body mass. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 14 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
15
When we used BMI alone, we observed a J-shaped relationship between BMI and all-
cause mortality in men (Table 3 and Figure 1). We also examined the influence on BMI when we
excluded men with low predicted LBM. When we excluded those in the lowest 2.5th
percentiles
of predicted LBM, the J-shaped relationship between BMI and mortality disappeared. Upon
excluding more participants with low predicted LBM (5th
and 10th
percentiles), the BMI-
mortality relationship became more linear and slightly stronger. However, the J-shaped
relationship still existed when excluding those with low FM (Supplementary table 4).
We further examined how the association of predicted FM and LBM with all-cause
mortality changes by different lag times (Supplementary table 5). With shorter lag times,
predicted FM showed a less linear positive association with all-cause mortality, while predicted
LBM showed a stronger U-shaped association with all-cause mortality. We also examined the
associations stratified by smoking status and age (Supplementary table 6 and 7). The relationship
between predicted FM and all-cause mortality was stronger and more linear among never-
smokers compared to current-smokers and among younger adults compared to older adults. On
the other hand, we observed a stronger U-shaped association between predicted LBM and all-
cause mortality among current-smokers compared to never-or past-smokers. We observed a
similar U-shaped relationship for predicted LBM across all age groups.
Our findings remained robust in several sensitivity analyses (Supplementary table 8, 9
and 10). The results did not change with no adjustment for physical activity, exclusion of deaths
in the early follow-up period and right-censoring criteria for age, inclusion of baseline illness and
use of quintiles and deciles for exposures. Moreover, using other prediction equations with
polynomial terms showed consistent results (data not shown).
Page 15 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
16
Table 3. Hazard ratio (95% CI) of all-cause mortality according to body mass index in men (Health Professionals Follow-
up Study)
Analysis
Hazard Ratio (95% CI)
No of
deaths
IR
/100,000py Model 1
a Model 2
a Model 3
b Model 4
c Model 5
d
BMI
<18.5 53 2883 1.74 (1.33 to 2.28) 1.65 (1.25 to 2.16) NA NA NA
18.5-20.4 269 1754 1.16 (1.03 to 1.32) 1.09 (0.96 to 1.24) 0.95 (0.76-1.18) 0.95 (0.68 to 1.32) 0.79 (0.38 to 1.66)
20.5-22.4 1358 1464 1.01 (0.95 to 1.08) 1.03 (0.97 to 1.09) 1.01 (0.94-1.08) 0.99 (0.92 to 1.06) 0.93 (0.85 to 1.03)
22.5-24.9 3740 1472 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 3986 1638 1.09 (1.04 to 1.14) 1.04 (0.99 to 1.09) 1.04 (1.00-1.09) 1.05 (1.00 to 1.10) 1.06 (1.01 to 1.11)
27.5-29.9 1753 1845 1.31 (1.24 to 1.39) 1.21 (1.14 to 1.28) 1.22 (1.15-1.29) 1.23 (1.16 to 1.30) 1.24 (1.17 to 1.31)
30.0-34.9 1001 1913 1.48 (1.38 to 1.59) 1.31 (1.22 to 1.41) 1.31 (1.22-1.41) 1.33 (1.24 to 1.43) 1.34 (1.25 to 1.44)
≥35.0 196 2368 2.28 (1.98 to 2.64) 2.01 (1.74 to 2.33) 2.02 (1.75-2.34) 2.04 (1.76 to 2.36) 2.06 (1.78 to 2.38)
P-trend
<.001 <.001 <.001 <.001 <.001
Abbreviation: BMI, body mass index; NA, not available (no cases available after exclusion).
Model 1: adjusted for age.
Model 2: adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical activity (<3, 3-8.9, 9-
17.9, 18-26.9, or >27 MET-hour/week), alcohol consumption (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy intake (quintiles), smoking status (never, ever, 1-14,
15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles).
Model 3: additionally, excluded 2.5%ile of total participants with low lean body masse
Model 4: additionally, excluded 5%ile of total participants with low lean body masse
Model 5: additionally, excluded 10%ile of total participants with low lean body masse a Number of deaths/person-years for each category of BMI: 53/1839, 269/15337, 1358/92790, 3740/254122, 3986/243335, 1753/95023, 1001/52320, and 196/8275. b Number of deaths/person-years for each category of BMI: 0/26, 80/7196, 1147/85194, 3695/252987, 3980/243212, 1751/94960, 1000/52311, and 196/8275. c Number of deaths/person-years for each category of BMI: 0/0, 36/3402, 884/72686, 3603/250459, 3976/243080, 1751/94960, 1000/52311, and 196/8275. d Number of deaths/person-years for each category of BMI: 0/0, 7/781, 509/48989, 3295/239196, 3963/242606, 1751/94928, 1000/52311, and 196/8275. e For exclusion analyses, height-adjusted lean body mass was used after regressing out variation due to height.
Page 16 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
17
Cause-specific mortality
We further examined the association of predicted FM and LBM with cause-specific mortality
(Table 4). Mutually adjusted models showed a linear positive association between predicted FM
and mortality due to cardiovascular disease and cancer. Compared to those in the lowest quintile
of predicted FM, men in the highest quintile had 67% (95% CI: 47 to 90%) and 24% (95% CI: 8
to 42%) increased hazard of death due to cardiovascular disease and cancer, respectively. In
contrast, predicted LBM showed a U-shaped association with mortality due to cardiovascular
disease and cancer in the mutually adjusted models. However, predicted LBM showed a strong
inverse association with mortality due to respiratory disease (P for trend<.001). Compared to
those in the lowest quintile of predicted LBM, men in the highest quintile had 50% (95% CI: 39
to 65%) decreased hazard of death due to respiratory disease. When we examined the association
between BMI and cause-specific mortality, we observed a U-shaped association for
cardiovascular disease death but a positive association for cancer death and an inverse
association for respiratory disease death.
Page 17 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
18
Table 4. Hazard ratio (95% CI) of cause-specific mortality according to predicted fat mass and lean
body mass in men (Health Professionals Follow-up Study)
Analysis Hazard Ratio (95% CI)
CVD death Cancer death Respiratory death Other death
No. of deaths 4296 3723 960 3377
IR/100,000py 558 483 124 438
Fat Massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 1.11 (1.00 to 1.24) 1.15 (1.03 to 1.29) 0.92 (0.74 to 1.14) 1.01 (0.90 to 1.14)
Quintile 3 1.10 (0.99 to 1.23) 1.06 (0.94 to 1.19) 1.06 (0.85 to 1.31) 0.84 (0.74 to 0.94)
Quintile 4 1.30 (1.16 to 1.46) 1.15 (1.02 to 1.30) 1.10 (0.88 to 1.38) 1.02 (0.90 to 1.15)
Quintile 5 1.67 (1.47 to 1.90) 1.24 (1.08 to 1.42) 1.26 (0.97 to 1.64) 1.13 (0.98 to 1.30)
P-trend <.001 0.01 0.03 0.05
Lean Body Massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.96 (0.87 to 1.06) 0.97 (0.88 to 1.08) 0.61 (0.51 to 0.74) 0.94 (0.84 to 1.04)
Quintile 3 0.95 (0.86 to 1.05) 0.94 (0.84 to 1.04) 0.58 (0.47 to 0.71) 0.95 (0.86 to 1.07)
Quintile 4 0.96 (0.87 to 1.07) 0.95 (0.84 to 1.06) 0.57 (0.46 to 0.71) 1.00 (0.89 to 1.12)
Quintile 5 1.11 (0.98 to 1.24) 1.02 (0.90 to 1.16) 0.50 (0.39 to 0.65) 0.98 (0.86 to 1.12)
P-trend 0.10 0.86 <.001 0.99
BMI
<18.5 1.45 (0.87 to 2.41) 0.66 (0.32 to 1.40) 5.33 (3.10 to 9.17) 1.86 (1.15 to 3.01)
18.5-20.4 1.12 (0.90 to 1.38) 0.99 (0.78 to 1.25) 1.93 (1.36 to 2.73) 0.92 (0.72 to 1.19)
20.5-22.4 0.95 (0.85 to 1.06) 0.97 (0.87 to 1.09) 1.30 (1.06 to 1.60) 1.09 (0.97 to 1.22)
22.5-24.9 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 1.16 (1.08 to 1.26) 1.01 (0.93 to 1.09) 0.91 (0.78 to 1.08) 0.98 (0.92 to 1.06)
27.5-29.9 1.40 (1.27 to 1.54) 1.13 (1.02 to 1.25) 1.09 (0.89 to 1.35) 1.10 (0.98 to 1.23)
30.0-34.9 1.75 (1.56 to 1.96) 1.12 (0.98 to 1.28) 0.81 (0.60 to 1.09) 1.18 (1.02 to 1.35)
≥35.0 2.66 (2.11 to 3.36) 1.55 (1.17 to 2.04) 0.90 (0.43 to 1.92) 2.13 (1.63 to 2.77)
P-trend <.001 <.001 <.001 0.002
Abbreviation: BMI, body mass index; CVD, cardiovascular disease; NA, not available (no cases available after exclusion).
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or
no), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy
intake (quintiles), smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles). Fat mass and lean
body mass were mutually adjusted in the model. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 18 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
19
DISCUSSION
Principal findings
In a large prospective cohort study of men, we used validated anthropometric prediction
equations to examine the association of LBM and FM with all-cause and cause-specific mortality.
We found a strong positive association between predicted FM and mortality due to all causes,
cardiovascular disease, and cancer. In contrast, predicted LBM showed a U-shaped association
with mortality due to all causes, cardiovascular disease, and cancer, and an inverse association
with mortality due to respiratory disease.
Comparison with other studies
Numerous epidemiological studies have examined the relationship between BMI and mortality,
but controversy and confusion exist around the unexpected J- or U-shaped association between
BMI and mortality.9 A systematic review and meta-analysis of 141 prospective studies in 2013
reported that, relative to normal weight, both grade 2 and grade 3 obesity (BMI≥30 kg/m2) were
associated with higher all-cause mortality but overweight (BMI 25-29.9 kg/m2) was associated
with lower all-cause mortality. In contrast, an individual participant-data meta-analysis of 239
prospective studies by the Global BMI Mortality Collaboration in 2016 showed evidence that
increased risk of all-cause mortality among overweight was largely due to confounding by
aspects of smoking and reverse causation from underlying disease and frailty at older ages.8 41 42
There are ongoing controversies around the ‘obesity paradox’43-45
with many studies reporting
excess mortality at the lower BMI range. More importantly, these studies acknowledged the
Page 19 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
20
major limitation of BMI as a measure of adiposity but did not directly investigate two distinct
compartments of body composition (LBM and FM) in relation to mortality.8 9
Our findings on BMI were in line with the previous findings, whereby we consistently
observed a J-shaped relationship with mortality even after accounting for age, smoking, and
baseline diseases. Two different shapes in mortality risk for FM and LBM taken together can
explain the observed J-shaped relationship between BMI and mortality in our study. The
increased risk of mortality in the lower BMI range (<25 kg/m2) could be attributed to a
combination of the high risk among men with low predicted LBM, which over-rides the modest
positive association between predicted FM and mortality in this lower range of BMI. The
increase of mortality risk at the BMI range of 25-30 kg/m2 is likely due to the high risk
associated with predicted FM in combination with only a moderate risk associated with predicted
LBM. Lastly, the rapid increase of mortality risk in the higher BMI range (>30kg/m2) could be
due to a very high risk associated with both predicted FM and LBM. Of note, at the high end of
BMI (>30kg/m2), the vast majority of individuals have high predicted FM and LBM. Those with
high predicted LBM almost invariably have high FM; for example, the average predicted FM for
those in the highest decile of predicted LBM was 31 kg (Supplementary table 10).
These observed patterns for FM and LBM were further supported from our additional
analyses of BMI and mortality after excluding those in the lower end of predicted LBM, which
resulted in a strong linear positive relationship between BMI and mortality. This shows that
separating lean and healthy (low BMI and normal LBM) vs. lean and unhealthy (low BMI and
low LBM) individuals could be a key to explain the ‘obesity paradox’ phenomenon. Our data
directly address the controversial hypothesis that accumulating excess fat may be causally
beneficial for survivor, and show that this is not likely to be true.46
Page 20 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
21
To date, only a limited number of studies have examined mortality in relation to directly
measured body composition using DXA or computed tomography.27-33
Most studies have been
done among elderly populations with approximate mean age of 75 years.27-29 31 33
The findings
showed inconsistent and various shapes of the relationship. An Italian study of 934 older people
showed no association of calf muscle and fat mass area with 6-year mortality28
, while another
study of 2292 elderly from the US found that low leg muscle area was associated with increased
risk of 6-year mortality in men only.29
In contrast, other studies found that lower fat percent23 27
or leg fat mass29
was associated with increased mortality27 31 33
but among these studies, only one
study of 477 community-dwelling older people from Netherlands reported that lower
appendicular skeletal muscle mass was associated with increased risk of 12-year mortality.33
Although these studies used a direct measure of body composition, the study samples were
restricted to elderly population which limits the generalizability of the findings. Moreover, they
also had other limitations such as small sample size, short follow-up, exposure measured at one-
time point, lack of information on important confounders (especially smoking) and no
examination on cause-specific mortality. Nonetheless, our finding was consistent with a recent
large-scale Canadian study that measured DXA from participants referred for bone mineral
density testing.30
That study found that high percent fat and low BMI were independently
associated with increased risk of mortality when percent fat and BMI were simultaneously
adjusted in the models. However, the observed associations might have been confounded by
smoking or physical activity due to lack of information on those variables, and the study did not
directly use LBM in the analysis.
The BMI-mortality relationship is prone to reverse causation by preexisting diseases that
can cause weight loss and also increase risk of mortality, and this is more likely to be a concern
Page 21 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
22
with shorter lag times. We found that, with shorter lag time periods, the positive association
between predicted FM and mortality was attenuated, while the U-shaped association between
predicted LBM and mortality tended to be strengthened. Therefore, the stronger U-shaped
relationship between BMI and mortality with shorter lag time periods can be mostly attributed to
the pronounced U-shaped association for predicted LBM, which may be an indicator of health
status capturing any preexisting undiagnosed medical condition, including frailty at older ages.
The influence of smoking is particularly important in investigating the obesity-mortality
relationship. Not only is smoking a strong risk factor for death, but it also affects body weight
and body composition.47-50
Similar to the BMI-mortality association, predicted FM showed
weaker and less linear association with mortality among current-smokers than past-or never-
smokers. Interestingly, we found a stronger U-shaped relationship between predicted LBM and
mortality among current-smokers than past-or never-smokers. Although we cannot completely
rule out the residual confounding by smoking, our findings showed some evidence that the
frequently observed U-shaped relationship between BMI and mortality among smokers may be
affected by the strong U-shaped association between LBM and mortality.
Strengths and limitations of the study
Our study has several strengths. First, the innovative approach of validated anthropometric
prediction equations allowed us to practically estimate LBM and FM in large epidemiological
settings. This is the first and one of the most comprehensive analyses to examine the association
of predicted body composition with all-cause and cause-specific mortality in a large prospective
cohort study. Second, the Health Professionals Follow-up Study is a well-established prospective
cohort study that has a large number of deaths over long-term follow-up period. Third, detailed
Page 22 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
23
and updated information on lifestyle and health-related factors allowed adequate control for
confounding. Fourth, repeated measures on exposures (i.e., predicted scores) allowed prospective
analyses of different lag time periods to examine reverse causality in the obesity-mortality
relationship.
There are several limitations as well. First, predicted LBM and FM are not perfect
measures of actual LBM and FM. Nonetheless, the validation results from the NHANES showed
high predictive ability of the anthropometric equations with no systematic bias. In fact, the very
high R2 between FM and LBM (>0.90) for direct DXA measurements and predicted measures in
an independent dataset indicate that a direct DXA measure would give very similar answers to
ours; this is further supported by the equal predictive ability of the predicted measures and DXA
measures for various obesity-related biomarkers. Moreover, given the prospective study design,
any mismeasurement in the exposures would likely be random with respect to endpoints,
resulting in conservative associations. Second, we cannot entirely rule out the possibility of
unmeasured or unknown confounding factors that may account for the associations observed in
this study. However, the homogeneity of the study population and comprehensive data on the
risk factors minimized potential confounding. Third, the generalizability of the findings may be
limited given that the study participants were restricted to health professionals and
predominantly White men. However, we believe that our main findings will be broadly
applicable.
Clinical and public health implications
The current study provides strong evidence that excess FM increases the risk of mortality.
Increased FM was not protective for mortality which is counter to the premise of the ‘obesity
Page 23 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
24
paradox’. On the other hand, low LBM was associated with increased risk of mortality in the
lower range of BMI. Our study suggests that understanding different contributions of LBM and
FM to BMI may explain the shape of the relationship between BMI and mortality as well as the
controversial issue of ‘obesity paradox’. Our findings support the current recommendations on
body weight for prevention of chronic diseases to keep a normal body weight defined by BMI of
18.5 to 24.9 kg/m2, and further suggest that the lowest mortality risk can be observed at the lower
normal range of BMI if the influence of LBM on mortality is accounted for. The current
recommendations should highlight not only the importance of normal body weight but also
healthy body composition (e.g., healthy lean) to reduce confusion around the optimal weight
(e.g., ‘obesity paradox’) for overall health. Interventions and strategies to promote healthy body
composition via lifestyle modification (e.g., physical activity and diet) may be an important next
step to improve population health.
Conclusions
We found a strong positive association between predicted FM and mortality, and a U-shaped
association between predicted LBM and mortality in men. Low LBM, rather than low FM, may
be driving the increased risk of mortality in the lower BMI range. Understanding the independent
role of LBM and FM has important implications for clarifying the ‘obesity paradox’
phenomenon in the relationship between BMI and mortality.
Page 24 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
25
Acknowledgements: We thank the participants and staff of the HSPF for their valuable
contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO,
CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH,
OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for
analyses and interpretation of these data.
Contributors: DHL and ELG had full access to all of the data in the study and take
responsibility for the integrity of the data and the accuracy of the data analysis. DHL and ELG
conceived and designed the study. NK, FBH, EJO, EBR, WCW, and ELG acquired the data.
DHL and ELG drafted the manuscript. All the authors critically revised the manuscript for
important intellectual content. DHL did the statistical analysis. FBH, EBR, WCW, and ELG
obtained funding. DHL and ELG were responsible for administrative, technical, or material
support. ELG was responsible for study supervision. DHL is the guarantor.
Funding: This work was supported by the National Institutes of Health (UM1 CA167552 and
R01 HL35464). The funders had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; and preparation, review, or approval of the
manuscript; and 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
Page 25 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
26
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: This investigation was approved by the Institutional Review Board of the
Harvard T.H. Chan School of Public Health and Brigham and Women’s Hospital.
Transparency declaration: The lead author (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 as
planned (and, if relevant, registered) have been explained.
Data sharing: No additional data available.
Copyright/License for Publication: The Corresponding Author has the right to grant on behalf
of all authors and does grant on behalf of all authors, a worldwide licence to the Publishers and
its licensees in perpetuity, in all forms, formats and media (whether known now or created in the
future), to i) publish, reproduce, distribute, display and store the Contribution, ii) translate the
Contribution into other languages, create adaptations, reprints, include within collections and
create summaries, extracts and/or, abstracts of the Contribution, iii) create any other derivative
work(s) based on the Contribution, iv) to exploit all subsidiary rights in the Contribution, v) the
inclusion of electronic links from the Contribution to third party material where-ever it may be
located; and, vi) licence any third party to do any or all of the above.
Page 26 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
27
References
1. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World
Health Organization technical report series 2000;894:i-xii, 1-253. [published Online
First: 2001/03/10]
2. Centers for Disease Control and Prevention. National Health and Nutrition Examination
Survey. Available at: http://www.cdc.gov/nchs/about/major/nhanes/datalink.htm.
Accessed March, 2018.
3. Hu F. Obesity epidemiology: Oxford University Press 2008.
4. Adams KF, Schatzkin A, Harris TB, et al. Overweight, obesity, and mortality in a large
prospective cohort of persons 50 to 71 years old. The New England journal of medicine
2006;355(8):763-78. doi: 10.1056/NEJMoa055643
5. Berrington de Gonzalez A, Hartge P, Cerhan JR, et al. Body-mass index and mortality among
1.46 million white adults. The New England journal of medicine 2010;363(23):2211-9.
doi: 10.1056/NEJMoa1000367
6. Chen Z, Yang G, Offer A, et al. Body mass index and mortality in China: a 15-year
prospective study of 220 000 men. International journal of epidemiology
2012;41(2):472-81. doi: 10.1093/ije/dyr208
7. Manson JE, Bassuk SS, Hu FB, et al. Estimating the number of deaths due to obesity: can the
divergent findings be reconciled? Journal of women's health 2007;16(2):168-76. doi:
10.1089/jwh.2006.0080
8. Global BMIMC, Di Angelantonio E, Bhupathiraju Sh N, et al. Body-mass index and all-cause
mortality: individual-participant-data meta-analysis of 239 prospective studies in four
Page 27 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
28
continents. Lancet 2016;388(10046):776-86. doi: 10.1016/s0140-6736(16)30175-1
[published Online First: 2016/07/18]
9. Flegal KM, Kit BK, Orpana H, et al. Association of all-cause mortality with overweight and
obesity using standard body mass index categories: a systematic review and meta-
analysis. Jama 2013;309(1):71-82. doi: 10.1001/jama.2012.113905
10. Veronese N, Cereda E, Solmi M, et al. Inverse relationship between body mass index and
mortality in older nursing home residents: a meta-analysis of 19,538 elderly subjects.
Obesity reviews : an official journal of the International Association for the Study of
Obesity 2015;16(11):1001-15. doi: 10.1111/obr.12309
11. Greenberg JA. The obesity paradox in the US population. The American journal of clinical
nutrition 2013;97(6):1195-200.
12. Visscher TL, Seidell JC, Molarius A, et al. A comparison of body mass index, waist-hip ratio
and waist circumference as predictors of all-cause mortality among the elderly: the
Rotterdam study. International journal of obesity and related metabolic disorders :
journal of the International Association for the Study of Obesity 2001;25(11):1730-5. doi:
10.1038/sj.ijo.0801787
13. Romero-Corral A, Lopez-Jimenez F, Sierra-Johnson J, et al. Differentiating between body fat
and lean mass-how should we measure obesity? Nature clinical practice Endocrinology
& metabolism 2008;4(6):322-3. doi: 10.1038/ncpendmet0809
14. Okorodudu DO, Jumean MF, Montori VM, et al. Diagnostic performance of body mass
index to identify obesity as defined by body adiposity: a systematic review and meta-
analysis. International journal of obesity 2010;34(5):791-9. doi: 10.1038/ijo.2010.5
Page 28 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
29
15. Pischon T. Commentary: Use of the body mass index to assess the risk of health outcomes:
time to say goodbye? International journal of epidemiology 2010:dyp388.
16. Gallagher D, Visser M, Sepulveda D, et al. How useful is body mass index for comparison of
body fatness across age, sex, and ethnic groups? American journal of epidemiology
1996;143(3):228-39.
17. Gallagher D, Ruts E, Visser M, et al. Weight stability masks sarcopenia in elderly men and
women. American journal of physiology Endocrinology and metabolism
2000;279(2):E366-75.
18. Harris TB. Invited commentary: body composition in studies of aging: new opportunities to
better understand health risks associated with weight. American journal of epidemiology
2002;156(2):122-4; discussion 25-6.
19. Haslam DW, James WP. Obesity. (1474-547X (Electronic))
20. Wannamethee SG, Atkins JL. Muscle loss and obesity: the health implications of sarcopenia
and sarcopenic obesity. Proceedings of the Nutrition Society 2015;74(04):405-12.
21. Rolland Y, Czerwinski S, Van Kan GA, et al. Sarcopenia: its assessment, etiology,
pathogenesis, consequences and future perspectives. The Journal of Nutrition Health and
Aging 2008;12(7):433-50.
22. Wannamethee SG, Shaper AG, Lennon L, et al. Decreased muscle mass and increased
central adiposity are independently related to mortality in older men. The American
journal of clinical nutrition 2007;86(5):1339-46.
23. Allison DB, Zhu SK, Plankey M, et al. Differential associations of body mass index and
adiposity with all-cause mortality among men in the first and second National Health and
Nutrition Examination Surveys (NHANES I and NHANES II) follow-up studies.
Page 29 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
30
International journal of obesity and related metabolic disorders : journal of the
International Association for the Study of Obesity 2002;26(3):410-6. doi:
10.1038/sj.ijo.0801925
24. Heitmann B, Erikson H, Ellsinger B, et al. Mortality associated with body fat, fat-free mass
and body mass index among 60-year-old Swedish menFa 22-year follow-up. The study of
men born in 1913. International journal of obesity and related metabolic disorders :
journal of the International Association for the Study of Obesity 2000;24:33-37.
25. Lee CD, Blair SN, Jackson AS. Cardiorespiratory fitness, body composition, and all-cause
and cardiovascular disease mortality in men. The American journal of clinical nutrition
1999;69(3):373-80.
26. Bigaard J, Frederiksen K, Tjonneland A, et al. Body fat and fat-free mass and all-cause
mortality. Obesity research 2004;12(7):1042-9. doi: 10.1038/oby.2004.131
27. Auyeung TW, Lee JS, Leung J, et al. Survival in older men may benefit from being slightly
overweight and centrally obese—a 5-year follow-up study in 4,000 older adults using
DXA. Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences
2009;65(1):99-104.
28. Cesari M, Pahor M, Lauretani F, et al. Skeletal muscle and mortality results from the
InCHIANTI Study. Journals of Gerontology Series A: Biomedical Sciences and Medical
Sciences 2009;64(3):377-84.
29. Newman AB, Kupelian V, Visser M, et al. Strength, but not muscle mass, is associated with
mortality in the health, aging and body composition study cohort. The Journals of
Gerontology Series A: Biological Sciences and Medical Sciences 2006;61(1):72-77.
Page 30 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
31
30. Padwal R, Leslie WD, Lix LM, et al. Relationship Among Body Fat Percentage, Body Mass
Index, and All-Cause Mortality: A Cohort Study. Annals of internal medicine 2016
31. Rolland Y, Gallini A, Cristini C, et al. Body-composition predictors of mortality in women
aged≥ 75 y: data from a large population-based cohort study with a 17-y follow-up–. The
American journal of clinical nutrition 2014;100(5):1352-60.
32. Toss F, Wiklund P, Nordström P, et al. Body composition and mortality risk in later life. Age
and ageing 2012;41(5):677-81.
33. Wijnhoven HA, Snijder MB, Deeg DJ, et al. Region-specific fat mass and muscle mass and
mortality in community-dwelling older men and women. Gerontology 2012;58(1):32-40.
34. Lee DH, Keum N, Hu FB, et al. Development and validation of anthropometric prediction
equations for lean body mass, fat mass and percent fat in adults using the National Health
and Nutrition Examination Survey (NHANES) 1999–2006. British Journal of Nutrition
2017;118(10):858-66.
35. Rimm EB, Stampfer MJ, Colditz GA, et al. Validity of self-reported waist and hip
circumferences in men and women. Epidemiology 1990;1(6):466-73.
36. Troy LM, Hunter DJ, Manson JE, et al. The validity of recalled weight among younger
women. International journal of obesity and related metabolic disorders : journal of the
International Association for the Study of Obesity 1995;19(8):570-2.
37. McCullough ML, Willett WC. Evaluating adherence to recommended diets in adults: the
Alternate Healthy Eating Index. Public health nutrition 2006;9(1a):152-57.
38. Durrleman S, Simon R. Flexible regression models with cubic splines. Statistics in medicine
1989;8(5):551-61.
Page 31 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
32
39. Govindarajulu U, Spiegelman D, Thurston S, et al. Comparing smoothing techniques for
modeling exposure-response curves in Cox models. Stat Med 2007;26(3735):52.
40. Smith PL. Splines as a useful and convenient statistical tool. The American Statistician
1979;33(2):57-62.
41. Yu E, Ley SH, Manson JE, et al. Weight History and All-Cause and Cause-Specific
Mortality in Three Prospective Cohort StudiesWeight History and Mortality in Three
Prospective Cohort Studies. Annals of internal medicine 2017;166(9):613-20.
42. Berrington de Gonzalez A, Hartge P, Cerhan JR, et al. Body-mass index and mortality among
1.46 million white adults. New England Journal of Medicine 2010;363(23):2211-19.
43. Flegal KM, Ioannidis JPA. A meta-analysis but not a systematic review: an evaluation of the
Global BMI Mortality Collaboration. (1878-5921 (Electronic))
44. Bhupathiraju SN, Di Angelantonio E, Danesh J, et al. Commentary on “A meta-analysis but
not a systematic review: an evaluation of the Global BMI Mortality Collaboration”.
Journal of clinical epidemiology 2017;88:30-32.
45. Flegal KM, Ioannidis JP. A meta-analysis of individual participant data constructed to align
with prior expert views: comments on Bhupathiraju et al. Journal of clinical
epidemiology 2017;88:33-36.
46. Tobias DK. Addressing Reverse Causation Bias in the Obesity Paradox Is Not “One Size Fits
All”. Diabetes care 2017;40(8):1000-01.
47. Chiolero A, Faeh D, Paccaud F, et al. Consequences of smoking for body weight, body fat
distribution, and insulin resistance. The American journal of clinical nutrition
2008;87(4):801-09.
Page 32 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
33
48. Molarius A, Seidell JC, Kuulasmaa K, et al. Smoking and relative body weight: an
international perspective from the WHO MONICA Project. Journal of epidemiology and
community health 1997;51(3):252-60.
49. Canoy D, Wareham N, Luben R, et al. Cigarette Smoking and Fat Distribution in 21, 828
British Men and Women: A Population‐based Study. Obesity 2005;13(8):1466-75.
50. Leite M, Nicolosi A. Lifestyle correlates of anthropometric estimates of body adiposity in an
Italian middle-aged and elderly population: a covariance analysis. International journal
of obesity 2006;30(6):926-34.
Page 33 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
34
Figure legends
Figure 1. The association of predicted body composition* and body mass index with all-
cause mortality in men. 1a. Fat mass and all-cause mortality. 2b. Lean body mass and all-
cause mortality. 3c. Body mass index and all-cause mortality.
Hazard ratios are indicated by solid lines and 95 % CIs by dashed lines. The reference point is
the lowest value for each fat mass and lean body mass and 25 kg/m2 for body mass index, with
knots placed at the 5th
, 35th
, 50th
, 65th
and 95th
percentiles of each fat mass and lean body mass
distribution. The models adjusted for the same cofounders in Table 2 plus mutually adjusted for
predicted fat mass and predicted lean body mass.
* Percentiles (0, 2.5, 5, 10, 25, 50, 75, 90, and 100%ile): 7, 13, 14, 15, 18, 21, 25, 29, and 66 kg
for fat mass, 24, 48, 49, 51, 53, 56, 59, 63, and 103 kg for lean body mass and 14.2, 21.5, 21.2,
22.0, 23.4, 25.1, 27.0, 31.0 and 62.0 kg/m2.
Page 34 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Figure 1. The association of predicted body composition* and body mass index with all-cause mortality in men. 1a. Fat mass and all-cause mortality. 2b. Lean body mass and all-cause mortality. 3c. Body mass
index and all-cause mortality.
311x116mm (96 x 96 DPI)
Page 35 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary online contents
Supplementary figure 1. Scatter plots for LBM in an independent validation group from the NHANES 1999-
2006
Supplementary figure 2. Scatter plots for FM in an independent validation group from the NHANES 1999-
2006
Supplementary table 1. Anthropometric prediction equations for lean body mass and fat mass in men
Supplementary table 2. Validation of predicted LBM with DXA-measured LBM in an independent validation
group from the NHANES 1999-2006
Supplementary table 3. Validation of predicted FM with DXA-measured FM in an independent validation
group from the NHANES 1999-2006
Supplementary table 4. Hazard ratio (95% CI) of all-cause mortality according to body mass index in men
(exclusion of low fat mass).
Supplementary table 5. Hazard ratio (95% CI) of all-cause mortality according to predicted fat mass, lean
body mass and body mass index by different lag-time periods
Supplementary table 6. Hazard ratio (95% CI) of all-cause mortality according to predicted fat mass, lean
body mass and body mass index stratified by smoking status.
Supplementary table 7. Hazard ratio (95% CI) of all-cause mortality according to predicted fat mass, lean
body mass and body mass index stratified by age
Supplementary table 8. Sensitivity analysis of predicted fat mass, lean body mass and body mass index in
relation to all-cause mortality in men
Supplementary table 9. Hazard ratio (95% CI) of all-cause and cause-specific mortality according to quintiles
of body mass index in men
Supplementary table 10. Hazard ratio (95% CI) of all-cause mortality according to deciles of predicted fat
mass, lean body mass and body mass index in men
This supplementary material has been provided by the authors to give readers additional information about their
work.
Page 36 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary figure 1. Scatter plots for LBM in an independent validation group from the NHANES 1999-2006
(n=2292). Scatter plots of DXA-measured LBM (y-axis) against predicted LBM (x-axis) (A. Original equations and B.
Equations with polynomial terms). Scatter plots of difference between DXA and predicted LBM (y-axis) against DXA-
measured LBM (x-axis) (C. Original equation and D. Equations with polynomial terms). For A and B, solid lines
represent regression and dotted lines present 95% CI. For C and D, dotted lines represent mean difference of 0. LBM, lean
body mass; DXA, dual-energy X-ray absorptiometry; NHANES, National Health and Nutrition Examination Survey
1999-2006
Page 37 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary figure 2. Scatter plots for FM in an independent validation group from the NHANES 1999-2006
(n=2292). Scatter plots of DXA-measured FM (y-axis) against predicted FM (x-axis) (A. Original equations and B.
Equations with polynomial terms). Scatter plots of difference between DXA and predicted FM (y-axis) against DXA-
measured FM (x-axis) (C. Original equation and D. Equations with polynomial terms). For A and B, solid lines represent
regression and dotted lines present 95% CI. For C and D, dotted lines represent mean difference of 0. Abbreviation: FM,
fat mass; DXA, dual-energy X-ray absorptiometry; NHANES, National Health and Nutrition Examination Survey
Page 38 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 1. Anthropometric prediction equations for lean body mass and
fat mass in menƚ
Original equations
LBM (kg)
= 19.363 + 0.001*age (yr) + 0.064*height (cm) + 0.756*weight (kg) - 0.366*waist (cm) -
0.066*Mexican + 0.231*Hispanic + 0.432*Black - 1.007*Other ethnicity
R2=0.914
SEE=2.548
FM (kg)
= -18.592 - 0.009*age (yr) - 0.080*height (cm) + 0.226*weight (kg) + 0.387*waist (cm) +
0.080*Mexican - 0.188*Hispanic - 0.483*Black + 1.050*Other ethnicity
R2=0.900
SEE=2.603
Equations with polynomial terms
LBM (kg)
= 12.052 – 0.003*age (yr) + 0.049*height (cm) + 1.133*weight (kg) – 0.486*waist (cm) –
0.112*Mexican + 0.135*Hispanic + 0.359*Black - 0.849*Other -0.006*(weight (kg)*waist (cm))
+ 0.001*(weight (kg))2 + 0.003*(waist (cm))
2
R2=0.918
SEE=2.494
FM (kg)
= -14.321 – 0.005*age (yr) – 0.064*height (cm) – 0.236*weight (kg) + 0.639*waist (cm) +
0.112*Mexican – 0.099*Hispanic – 0.363*Black + 0.888*Other + 0.008*(weight (kg)*waist
(cm)) – 0.002*(weight (kg))2 – 0.005*(waist (cm))
2
R2=0.905
SEE=2.543
Abbreviation: LBM, lean body mass; FM, fat mass. Race variables are binary variables (1 if yes, 0 if no), and White is the reference group.
ƚ Equations were developed and validated using the National Health and Nutrition Examination Survey 1999-2006
Page 39 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 2. Validation of predicted LBM with DXA-measured LBM in an independent validation group from the NHANES
1999-2006 (n=2292)
Deciles of DXA-measured LBM
Mean (SD) 1 2 3 4 5 6 7 8 9 10
DXA-measured LBM (kg) 42.70 (2.85) 47.70 (0.98) 50.40 (0.72) 52.78 (0.64) 54.91 (0.58) 57.25 (0.69) 59.69 (0.70) 62.32 (0.82) 65.80 (1.21) 72.60 (4.20)
Original equations
Predicted LBM (kg) 44.37 (2.76) 48.64 (2.38) 50.96 (2.21) 52.95 (2.34) 54.78 (2.11) 57.23 (2.13) 59.31 (2.52) 61.62 (2.75) 65.16 (2.95) 71.96 (5.44)
Diff (DXA-predicted) (kg) -1.67 (1.98) -0.94 (2.14) -0.56 (2.13) -0.18 (2.28) 0.13 (2.02) 0.02 (2.12) 0.39 (2.41) 0.70 (2.53) 0.64 (2.69) 0.64 (3.18)
Equations with polynomial terms
Predicted LBM (kg) 43.83 (3.06) 48.5 (2.37) 50.97 (2.28) 53.06 (2.37) 54.94 (2.16) 57.46 (2.18) 59.57 (2.51) 61.86 (2.74) 65.36 (2.82) 71.85 (5.03)
Diff (DXA-predicted) (kg) -1.13 (1.95) -0.80 (2.09) -0.57 (2.19) -0.28 (2.30) -0.04 (2.06) -0.21 (2.18) 0.13 (2.41) 0.47 (2.52) 0.44 (2.56) 0.75 (2.97)
Abbreviation: LBM, lean body mass; DXA, dual-energy X-ray absorptiometry; NHANES, National Health and Nutrition Examination Survey 1999-2006
Page 40 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 3. Validation of predicted FM with DXA-measured FM in an independent validation group from the NHANES
1999-2006 (n=2292)
Deciles of DXA-measured FM
Mean (SD) 1 2 3 4 5 6 7 8 9 10
DXA-measured FM (kg) 9.80 (1.32) 13.21 (0.87) 15.93 (0.73) 18.34 (0.67) 20.55 (0.62) 22.46 (0.57) 24.52 (0.65) 27.00 (0.82) 30.28 (1.14) 38.55 (5.42)
Original equations
Predicted FM (kg) 9.63 (2.71) 13.79 (2.22) 16.39 (2.38) 18.82 (2.34) 21.1 (2.24) 22.79 (2.31) 24.77 (2.39) 26.74 (2.76) 29.63 (2.70) 36.38 (4.84)
Diff (DXA-predicted) (kg) 0.17 (2.17) -0.59 (2.03) -0.47 (2.30) -0.49 (2.15) -0.54 (2.14) -0.33 (2.27) -0.25 (2.34) 0.26 (2.59) 0.65 (2.57) 2.17 (3.07)
Equations with polynomial terms
Predicted FM (kg) 10.22 (2.24) 13.95 (2.00) 16.40 (2.15) 18.73 (2.15) 20.86 (2.13) 22.45 (2.18) 24.4 (2.36) 26.4 (2.85) 29.33 (2.79) 36.71 (5.56)
Diff (DXA-predicted) (kg) -0.42 (1.76) -0.75 (1.84) -0.48 (2.07) -0.39 (1.97) -0.30 (2.03) 0.01 (2.14) 0.12 (2.32) 0.60 (2.67) 0.95 (2.66) 1.84 (3.17)
Abbreviation: FM, fat mass; DXA, dual-energy X-ray absorptiometry; NHANES, National Health and Nutrition Examination Survey
Page 41 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 4. Hazard ratio (95% CI) of all-cause mortality according to body mass index in men (exclusion of low fat mass*).
Hazard Ratio (95% CI)
No of
deaths
IR
/100,000py Model 1 Model 2 Model 3 Model 4 Model 5
BMI
<18.5 53 2883 1.74 (1.33 to 2.28) 1.64 (1.25 to 2.15) 2.24 (1.16 to 4.32) 1.66 (0.75 to 3.71) 1.75 (0.79 to 3.91)
18.5-22.4 1626 1504 1.04 (0.98 to 1.10) 1.03 (0.98 to 1.10) 1.04 (0.97 to 1.10) 1.03 (0.97 to 1.10) 1.10 (1.03 to 1.19)
22.5-24.9 3740 1472 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 3987 1639 1.09 (1.04 to 1.14) 1.04 (0.99 to 1.09) 1.04 (0.99 to 1.09) 1.04 (0.99 to 1.09) 1.05 (1.00 to 1.09)
27.5-29.9 1753 1845 1.31 (1.24 to 1.39) 1.21 (1.14 to 1.28) 1.21 (1.14 to 1.28) 1.21 (1.15 to 1.29) 1.22 (1.15 to 1.29)
30.0-34.9 1001 1913 1.48 (1.38 to 1.59) 1.31 (1.22 to 1.40) 1.31 (1.22 to 1.41) 1.32 (1.22 to 1.41) 1.32 (1.23 to 1.42)
≥35.0 196 2368 2.28 (1.98 to 2.64) 1.99 (1.72 to 2.30) 2.02 (1.75 to 2.33) 2.02 (1.75 to 2.34) 2.03 (1.76 to 2.35)
P-trend
<.001 <.001 <.001 <.001 <.001 Abbreviation: BMI, body mass index. Model 1: adjusted for age
Model 2: adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total daily energy intake
(quintiles), and smoking (never, ever, 1-14, 15-24, ≥25 cigs/day), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), Alternate Healthy Eating Index (quintiles). Model 3: additionally, excluded 2.5%ile of total participants with low fat mass (height-adjusted)
Model 4: additionally, excluded 5%ile of total participants with low fat mass (height-adjusted)
Model 5: additionally, excluded 10%ile of total participants with low fat mass (height-adjusted) * Derived from validated anthropometric prediction equations.
Page 42 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 5. Hazard ratio (95% CI) of all-cause mortality according to body mass index,
predicted fat mass, lean body mass and body mass by different lag-time periods
Hazard Ratio (95% CI)
0-12y 4-16y 8-18y 12-22y
No. of deaths 12356 10726 8214 5982
IR/100,000py 1419 1580 1694 1856
Fat massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.98 (0.93 to 1.04) 1.03 (0.97 to 1.10) 1.04 (0.96 to 1.12) 1.02 (0.93 to 1.11)
Quintile 3 0.95 (0.89 to 1.01) 0.99 (0.93 to 1.06) 1.02 (0.94 to 1.10) 1.02 (0.93 to 1.12)
Quintile 4 1.04 (0.98 to 1.11) 1.11 (1.04 to 1.19) 1.12 (1.04 to 1.22) 1.14 (1.03 to 1.25)
Quintile 5 1.22 (1.13 to 1.31) 1.29 (1.19 to 1.40) 1.38 (1.26 to 1.51) 1.36 (1.22 to 1.51)
P-trend <.001 <.001 <.001 <.001
Lean body massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.87 (0.83 to 0.92) 0.87 (0.83 to 0.93) 0.90 (0.84 to 0.96) 0.93 (0.86-1.01)
Quintile 3 0.84 (0.80 to 0.89) 0.88 (0.83 to 0.94) 0.88 (0.82 to 0.94) 0.93 (0.85-1.01)
Quintile 4 0.85 (0.80 to 0.91) 0.89 (0.83 to 0.95) 0.90 (0.84 to 0.97) 0.93 (0.85-1.02)
Quintile 5 0.90 (0.84 to 0.97) 0.95 (0.88 to 1.03) 0.94 (0.87 to 1.03) 1.00 (0.91-1.11)
P-trend 0.001 0.21 0.21 0.91
BMI
<18.5 2.32 (2.00 to 2.70) 1.45 (1.11 to 1.90) 1.46 (1.06 to 2.02) 1.55 (1.06 to 2.27)
18.5-22.4 1.20 (1.13 to 1.26) 1.06 (1.00 to 1.13) 1.02 (0.95 to 1.09) 0.99 (0.91 to 1.08)
22.5-24.9 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 0.94 (0.90 to 0.99) 0.98 (0.93 to 1.03) 1.01 (0.95 to 1.07) 1.03 (0.97 to 1.10)
27.5-29.9 1.09 (1.03 to 1.15) 1.16 (1.09 to 1.23) 1.17 (1.09 to 1.26) 1.24 (1.14 to 1.35)
30.0-34.9 1.20 (1.12 to 1.29) 1.28 (1.19 to 1.38) 1.35 (1.24 to 1.46) 1.36 (1.22 to 1.50)
≥35.0 1.64 (1.44 to 1.87) 1.61 (1.39 to 1.86) 1.65 (1.39 to 1.95) 1.92 (1.58 to 2.34)
P-trend 0.27 <.001 <.001 <.001 Abbreviation: BMI, body mass index; NA, not available (no cases available after exclusion).
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical
activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy intake (quintiles), smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles). Fat mass and lean body mass were mutually adjusted in the model. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 43 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 6. Hazard ratio (95% CI) of all-cause mortality according to
predicted fat mass, lean body mass and body mass stratified by smoking status
Hazard Ratio (95% CI)
Never-smokers Past-smoker Current-smoker
No. of deaths 6791 4947 618
IR/100,000py 1457 1377 1366
Fat massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference)
Quintile 2 1.06 (0.97 to 1.15) 1.14 (1.03 to 1.26) 0.95 (0.72 to 1.26)
Quintile 3 1.01 (0.93 to 1.11) 1.02 (0.92 to 1.13) 0.92 (0.68 to 1.24)
Quintile 4 1.17 (1.07 to 1.28) 1.16 (1.04 to 1.29) 1.22 (0.90 to 1.65)
Quintile 5 1.40 (1.26 to 1.54) 1.34 (1.19 to 1.51) 1.17 (0.84 to 1.65)
P-trend <.001 <.001 0.18
Lean body massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.95 (0.88 to 1.02) 0.91 (0.83 to 0.99) 0.66 (0.50 to 0.86)
Quintile 3 0.90 (0.83 to 0.98) 0.94 (0.86 to 1.03) 0.72 (0.54 to 0.94)
Quintile 4 0.92 (0.85 to 1.00) 0.94 (0.85 to 1.03) 0.81 (0.61 to 1.08)
Quintile 5 0.97 (0.89 to 1.07) 0.98 (0.88 to 1.10) 0.86 (0.63 to 1.19)
P-trend 0.46 0.97 0.61
BMI
<18.5 1.57 (1.08 to 2.29) 1.67 (1.07 to 2.60) 1.96 (0.78 to 4.94)
18.5-22.4 1.01 (0.93 to 1.09) 1.05 (0.95 to 1.15) 1.33 (1.04 to 1.70)
22.5-24.9 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 1.05 (0.99 to 1.11) 1.02 (0.95 to 1.10) 1.15 (0.92 to 1.42)
27.5-29.9 1.23 (1.14 to 1.33) 1.20 (1.10 to 1.31) 1.24 (0.95 to 1.63)
30.0-34.9 1.32 (1.19 to 1.45) 1.27 (1.14 to 1.42) 1.56 (1.12 to 2.17)
≥35.0 2.02 (1.65 to 2.48) 2.06 (1.66 to 2.56) 1.89 (0.86 to 4.14)
P-trend <.001 <.001 0.19 Abbreviation: BMI, body mass index; NA, not available (no cases available after exclusion).
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+
g/day), total energy intake (quintiles), smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index
(quintiles). Fat mass and lean body mass were mutually adjusted in the model. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 44 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 7. Hazard ratio (95% CI) of all-cause mortality according to
predicted fat mass, lean body mass and body mass stratified by age
Hazard Ratio (95% CI)
Age <70yrs Age 70-84yrs Age ≥85yrs
No. of deaths 2406 6845 3105
IR/100,000py 402 2789 11983
Fat massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference)
Quintile 2 1.14 (0.99 to 1.31) 1.08 (0.99 to 1.18) 1.02 (0.91 to 1.15)
Quintile 3 0.95 (0.81 to 1.11) 1.01 (0.93 to 1.10) 1.03 (0.91 to 1.16)
Quintile 4 1.22 (1.04 to 1.42) 1.17 (1.07 to 1.28) 1.10 (0.97 to 1.25)
Quintile 5 1.52 (1.28 to 1.80) 1.39 (1.25 to 1.53) 1.15 (0.99 to 1.33)
P-trend <.001 <.001 0.04
Lean body massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.85 (0.73 to 0.97) 0.92 (0.85 to 0.99) 0.97 (0.88 to 1.07)
Quintile 3 0.87 (0.76 to 1.01) 0.90 (0.83 to 0.97) 0.94 (0.84 to 1.05)
Quintile 4 0.94 (0.81 to 1.09) 0.88 (0.81 to 0.96) 1.03 (0.91 to 1.16)
Quintile 5 0.98 (0.84 to 1.15) 0.92 (0.84 to 1.01) 1.05 (0.90 to 1.22)
P-trend 0.55 0.08 0.58
BMI
<18.5 1.80 (0.96 to 3.37) 1.69 (1.19 to 2.39) 1.45 (0.80 to 2.63)
18.5-22.4 1.05 (0.91 to 1.21) 1.04 (0.96 to 1.13) 1.03 (0.92 to 1.14)
22.5-24.9 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 1.12 (1.00 to 1.24) 1.03 (0.97 to 1.09) 1.02 (0.94 to 1.12)
27.5-29.9 1.37 (1.21 to 1.56) 1.18 (1.09 to 1.27) 1.18 (1.04 to 1.33)
30.0-34.9 1.59 (1.38 to 1.84) 1.28 (1.16 to 1.40) 1.12 (0.94 to 1.34)
≥35.0 2.33 (1.80 to 3.03) 2.05 (1.70 to 2.47) 1.29 (0.79 to 2.13)
P-trend <.001 <.001 0.04 Abbreviation: BMI, body mass index; NA, not available (no cases available after exclusion).
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+
g/day), total energy intake (quintiles), smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index
(quintiles). Fat mass and lean body mass were mutually adjusted in the model. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 45 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 8. Sensitivity analysis of predicted fat mass, lean body mass and body
mass in relation to all-cause mortality in men
Hazard Ratio (95% CI)
Model 1* Model 2
† Model 3
‡ Model 4
§
Fat massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 1.09 (1.02 to 1.16) 1.09 (1.02 to 1.16) 1.09 (1.02 to 1.17) 1.08 (1.01 to 1.14)
Quintile 3 1.02 (0.96 to 1.09) 1.03 (0.96 to 1.10) 0.99 (0.92 to 1.07) 1.02 (0.96 to 1.09)
Quintile 4 1.19 (1.11 to 1.27) 1.18 (1.10 to 1.26) 1.18 (1.09 to 1.27) 1.15 (1.08 to 1.23)
Quintile 5 1.41 (1.31 to 1.51) 1.37 (1.27 to 1.47) 1.41 (1.30 to 1.53) 1.36 (1.26 to 1.46)
P-trend <.001 <.001 <.001 <.001
Lean body massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.91 (0.86 to 0.96) 0.92 (0.87 to 0.97) 0.90 (0.84 to 0.95) 0.90 (0.86 to 0.95)
Quintile 3 0.89 (0.84 to 0.95) 0.90 (0.85 to 0.96) 0.89 (0.83 to 0.95) 0.89 (0.84 to 0.94)
Quintile 4 0.91 (0.86 to 0.97) 0.92 (0.86 to 0.98) 0.90 (0.84 to 0.96) 0.91 (0.86 to 0.97)
Quintile 5 0.96 (0.90 to 1.03) 0.98 (0.91 to 1.05) 0.95 (0.88 to 1.02) 0.96 (0.90 to 1.03)
P-trend 0.33 0.62 0.27 0.29
BMI
<18.5 1.68 (1.28 to 2.20) 1.36 (1.00 to 1.84) 1.85 (1.39 to 2.46) 1.48 (1.14 to 1.92)
18.5-22.4 1.04 (0.98 to 1.10) 1.03 (0.97 to 1.09) 1.05 (0.98 to 1.12) 1.04 (0.98 to 1.10)
22.5-24.9 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 1.05 (1.00 to 1.10) 1.03 (0.99 to 1.08) 1.04 (0.99 to 1.10) 1.03 (0.99 to 1.08)
27.5-29.9 1.23 (1.16 to 1.31) 1.21 (1.14 to 1.28) 1.23 (1.15 to 1.31) 1.21 (1.14 to 1.28)
30.0-34.9 1.35 (1.26 to 1.45) 1.31 (1.22 to 1.41) 1.35 (1.25 to 1.46) 1.32 (1.23 to 1.41)
≥35.0 2.10 (1.81 to 2.42) 2.04 (1.76 to 2.36) 2.09 (1.79 to 2.43) 1.86 (1.61 to 2.14)
P-trend <.001 <.001 <.001 <.001 Abbreviation: BMI, body mass index.
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy intake (quintiles),
smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles). Fat mass and lean body mass were mutually
adjusted in the model. *
Model 1: no adjustment for physical activity (number of deaths/person-years: 12356/763041). †
Model 2: exclusion of deaths occurred in the early follow-up period (2 years) (number of deaths/person-years: 11940/760100). ‡
Model 3: exclusion of right censoring criteria for age (>85 years) (number of deaths/person-years: 9764/743476). §
Model 4: inclusion of baseline illness (number of deaths/person-years: 13377/788198).
a Derived from validated anthropometric prediction equations. b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 46 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 9. Hazard ratio (95% CI) of all-cause and cause-specific mortality according to quintiles of
body mass index in men
Hazard Ratio (95% CI)
All cause death CVD death Cancer death Respiratory death Other death
No. of deaths 12356 4296 3723 960 3377
IR/100,000py 1619 558 483 124 438
BMI
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.98 (0.93 to 1.04) 1.01 (0.91 to 1.12) 1.05 (0.95 to 1.17) 0.74 (0.61 to 0.90) 0.97 (0.87 to 1.09)
Quintile 3 0.96 (0.91 to 1.02) 1.08 (0.98 to 1.20) 0.97 (0.87 to 1.07) 0.72 (0.60 to 0.88) 0.93 (0.83 to 1.03)
Quintile 4 1.06 (1.00 to 1.12) 1.27 (1.15 to 1.40) 1.07 (0.97 to 1.19) 0.64 (0.53 to 0.78) 0.96 (0.87 to 1.07)
Quintile 5 1.25 (1.18 to 1.32) 1.60 (1.46 to 1.77) 1.19 (1.08 to 1.32) 0.71 (0.58 to 0.87) 1.11 (1.00 to 1.24)
P-trend <.001 <.001 0.01 0.03 0.05
Abbreviation: BMI, body mass index; CVD, cardiovascular disease; NA, not available (no cases available after exclusion).
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical
activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy intake (quintiles), smoking status
(never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles). Fat mass and lean body mass were mutually adjusted in the model. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 47 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
Supplementary table 10. Hazard ratio (95% CI) of all-cause mortality according to deciles of predicted fat mass,
lean body mass and body mass index in men
Hazard Ratio (95% CI)
FM (kg)
Mean (SD)
LBM (kg)
Mean (SD)
BMI (kg/m2)
Mean (SD)
No of
deaths Model 1 Model 2 Model 3
Fat massa,b
Decile 1 13.7 (1.5) 51.9 (3.3) 21.6 (1.4) 1009 1 (reference) 1 (reference) 1 (reference)
Decile 2 16.5 (0.5) 53.3 (3.2) 22.9 (1.2) 928 0.85 (0.78 to 0.93) 0.87 (0.79 to 0.95) 0.89 (0.81 to 0.98)
Decile 3 18.0 (0.4) 54.1 (3.3) 23.6 (1.3) 1094 0.97 (0.89 to 1.06) 0.95 (0.87-1.04) 0.99 (0.90 to 1.08)
Decile 4 19.3 (0.4) 54.6 (3.3) 24.1 (1.3) 1204 1.04 (0.96 to 1.14) 1.02 (0.94-1.11) 1.07 (0.98 to 1.17)
Decile 5 20.5 (0.4) 55.5 (3.5) 24.8 (1.4) 1147 0.94 (0.86 to 1.02) 0.89 (0.82-0.97) 0.94 (0.86 to 1.03)
Decile 6 21.8 (0.4) 56.0 (3.6) 25.3 (1.4) 1150 0.96 (0.88 to 1.04) 0.93 (0.85-1.01) 0.98 (0.90 to 1.08)
Decile 7 23.2 (0.5) 56.9 (3.7) 26.0 (1.5) 1303 1.08 (0.99 to 1.17) 1.00 (0.92-1.08) 1.06 (0.97 to 1.16)
Decile 8 25.0 (0.6) 57.8 (4.1) 26.7 (1.6) 1423 1.20 (1.10 to 1.30) 1.11 (1.02-1.20) 1.18 (1.07 to 1.29)
Decile 9 27.5 (1.0) 59.4 (4.3) 27.9 (1.8) 1515 1.28 (1.18 to 1.39) 1.17 (1.08-1.27) 1.24 (1.13 to 1.37)
Decile 10 33.8 (4.5) 64.6 (6.3) 31.3 (3.2) 1583 1.52 (1.40 to 1.65) 1.32 (1.21-1.43) 1.36 (1.23 to 1.51)
P-trend <.001 <.001 <.001
Lean body massa,b
Decile 1 17.4 (4.3) 48.5 (2.4) 21.5 (1.4) 1656 1 (reference) 1 (reference) 1 (reference)
Decile 2 18.3 (3.7) 51.6 (0.5) 22.8 (1.1) 1340 0.89 (0.83 to 0.96) 0.90 (0.84 to 0.97) 0.91 (0.84 to 0.97)
Decile 3 19.2 (3.7) 53.0 (0.4) 23.5 (1.1) 1242 0.88 (0.82 to 0.95) 0.89 (0.82 to 0.95) 0.88 (0.82 to 0.95)
Decile 4 20.0 (3.7) 54.2 (0.3) 24.2 (1.1) 1177 0.88 (0.82 to 0.95) 0.88 (0.82 to 0.95) 0.87 (0.81 to 0.94)
Decile 5 20.5 (3.7) 55.3 (0.3) 24.7 (1.1) 1163 0.87 (0.81 to 0.94) 0.87 (0.80 to 0.93) 0.85 (0.78 to 0.92)
Decile 6 21.3 (3.9) 56.4 (0.4) 25.3 (1.2) 1161 0.94 (0.87 to 1.01) 0.92 (0.85 to 0.99) 0.88 (0.81 to 0.95)
Decile 7 22.2 (3.8) 57.6 (0.4) 26.0 (1.2) 1099 0.92 (0.86 to 1.00) 0.91 (0.84 to 0.99) 0.85 (0.79 to 0.93)
Decile 8 23.5 (4.1) 59.1 (0.5) 26.8 (1.2) 1183 1.04 (0.96 to 1.12) 1.00 (0.93 to 1.08) 0.90 (0.83 to 0.98)
Decile 9 25.6 (4.5) 61.3 (0.9) 28.1 (1.4) 1137 1.08 (1.00 to 1.17) 1.03 (0.96 to 1.11) 0.88 (0.81 to 0.96)
Decile 10 31.2 (6.3) 67.0 (4.4) 31.5 (3.1) 1198 1.33 (1.23 to 1.44) 1.20 (1.11 to 1.30) 0.94 (0.85 to 1.03)
P-trend <.001 <.001 0.08
BMI
Decile 1 14.9 (2.6) 49.4 (2.7) 21.0 (0.9) 1221 1 (reference) 1 (reference) NA
Decile 2 17.1 (2.2) 52.0 (1.8) 22.5 (0.3) 1112 0.93 (0.86 to 1.01) 0.95 (0.87 to 1.03) NA
Decile 3 18.6 (2.4) 53.3 (1.9) 23.4 (0.2) 1188 0.96 (0.89 to 1.04) 0.97 (0.89 to 1.05) NA
Decile 4 19.4 (2.4) 54.3 (1.9) 24.0 (0.2) 884 0.94 (0.86 to 1.03) 0.94 (0.86 to 1.02) NA
Decile 5 20.7 (2.5) 55.2 (1.9) 24.6 (0.2) 1175 0.98 (0.90 to 1.06) 0.96 (0.89 to 1.04) NA
Decile 6 21.6 (2.6) 56.2 (2.0) 25.3 (0.2) 1222 0.95 (0.87 to 1.02) 0.92 (0.85 to 0.99) NA
Decile 7 22.9 (2.7) 57.3 (2.0) 26.0 (0.3) 1408 1.07 (0.99 to 1.16) 1.02 (0.94 to 1.10) NA
Decile 8 24.3 (2.9) 58.8 (2.2) 27.0 (0.3) 1306 1.10 (1.02 to 1.19) 1.04 (0.96 to 1.13) NA
Decile 9 26.7 (3.2) 60.8 (2.6) 28.3 (0.5) 1387 1.23 (1.14 to 1.33) 1.14 (1.06 to 1.24) NA
Decile 10 32.6 (5.4) 66.3 (5.0) 31.9 (2.8) 1453 1.46 (1.35 to 1.57) 1.29 (1.19 to 1.40) NA
P-trend <.001 <.001
Abbreviation: BMI, body mass index; FM, fat mass; LBM, lean body mass; NA, not available.
Model 1: adjusted for age.
Model 2: adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol consumption (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy intake (quintiles), and smoking status (never, ever, 1-14, 15-24, ≥25
cigs/day), Alternate Healthy Eating Index (quintiles).
Model 3: additionally, mutually adjusted for predicted fat mass and predicted lean body mass. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 48 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
1
Predicted lean body mass, fat mass, and all-cause and cause-specific mortality in men:
results from a prospective US cohort study
Dong Hoon Lee, NaNa Keum, Frank B. Hu, E. John Orav, Eric B. Rimm, Walter C. Willett,
Edward L. Giovannucci
Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA 02115, USA,
Dong Hoon Lee, post-doctoral research fellow, NaNa Keum, post-doctoral research fellow,
Frank B. Hu, professor, Eric B. Rimm, professor, Walter C. Willett, professor, Edward L.
Giovannucci, professor
Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA 02115,
USA, Frank B. Hu, professor, Eric B. Rimm, professor, Walter C. Willett, professor, Edward L.
Giovannucci, professor
Department of food science and Biotechnology, Dongguk University, Goyang, South Korea,
NaNa Keum, assistant professor, Channing Division of Network Medicine
Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston,
MA 02115, USA, Frank B. Hu, professor, Eric B. Rimm, professor, Walter C. Willett, professor,
Edward L. Giovannucci, professor
Department of Medicine, Brigham and Women’s Hospital, Boston, MA 02115, USA, E. John
Orav, associate professor
Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, MA 02115,
USA, E. John Orav, associate professor.
Corresponding author:
Edward L. Giovannucci, Department of Nutrition, Harvard T.H Chan School of Public Health,
665 Huntington Avenue, Bldg. 2, Room 371, Boston, MA 02115
Phone: 617-432-4648, Fax: 617-432-2435, Email: [email protected]
Word count: 4,229
Number of tables and figures: 4 tables and 1 figure
Page 49 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
2
ABSTRACT
Objective: To investigate the association of predicted lean body mass (LBM), fat mass (FM) and
body mass index (BMI) with all-cause and cause-specific mortality in men.
Design: Prospective cohort study.
Setting: Health professionals in the United States
Participants: 38,021 men (aged 40-75 years) from the Health Professionals Follow-up Study
were followed-up for death (1987-2012).
Main outcome measures: All-cause and cause-specific mortality.
Results: Using validated anthropometric prediction equations previously developed from the
National Health and Nutrition Examination Survey, LBM and FM were estimated for all
participants. During a mean of 21.4 years of follow-up, we identified 12,356 deaths. We
consistently observed a J-shaped association between BMI and mortality. Multivariable-adjusted
Cox models including both predicted FM and LBM showed a strong positive monotonic
association between predicted FM and mortality. Compared to those in the lowest quintile of
predicted FM, men in the highest quintile had 35% (95% confidence interval (CI): 26 to 46%),
67% (95% CI: 47 to 90%), and 24% (95% CI: 8 to 42%) increased risk of mortality due to all
causes, cardiovascular disease, and cancer. In contrast, a U-shaped association was found
between predicted LBM and mortality due to all causes, cardiovascular disease, and cancer (P for
non-linearity<0.001). However, there was a strong inverse association between predicted LBM
and mortality due to respiratory disease (P for trend<0.001). Compared to those in the lowest
quintile of predicted LBM, men in the highest quintile had 50% (95% CI: 39 to 65%) decreased
risk of death due to respiratory disease.
Conclusions: The shape of the relationship between BMI and mortality was determined by the
relationship between two body components (LBM and FM) and mortality. Our finding suggests
that the ‘obesity paradox’ controversy may be largely explained by low LBM, rather than low
FM, in the lower range of BMI.
Keywords: body mass index, body composition, lean body mass, fat mass, mortality, obesity
paradox
Page 50 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
3
“What this paper adds” box
Section 1: What is already known on this topic
• Numerous epidemiological studies have shown unexpected J-or U-shaped relationship
between body mass index (BMI) and mortality (‘obesity paradox’).
• The controversial issue of ‘obesity paradox’ may have arisen in part due to
underappreciation of different contributions of lean body mass (LBM) and fat mass (FM)
to BMI.
• Direct measure of body composition is difficult in large epidemiological settings, thus the
relationship between body composition and mortality is still unknown.
Section 2: What this study adds
• Using validated anthropometric prediction equations for body composition, this study
represents the first effort to comprehensively examined the association between lean
body mass, fat mass and mortality in a large prospective cohort study.
• Predicted fat mass showed a strong positive monotonic association with mortality, while
predicted lean body mass showed a strong U-shaped association with mortality.
• The ‘obesity paradox’ controversy may be explained largely by low LBM, rather than
low FM, in the lower range of BMI
Page 51 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
4
INTRODUCTION
Obesity is a major public health challenge in the United States and around the world.(1) In 2013-
2014, more than two thirds of Americans were classified as overweight (defined as body mass
index (BMI) of 25-29.9 kg/m2) or obese (BMI of ≥30 kg/m
2).(2) BMI is known as a reasonably
good measure of general adiposity(3), and many epidemiologic studies have provided evidence
supporting that obesity, assessed by BMI, is a significant risk factor for increased risk of many
chronic diseases as well as mortality.(4-6) However, details of the shape of the association
between BMI and mortality has been a topic of considerable discussion as epidemiologic studies
have found various types of J-shaped, U-shaped, and linear relationships between BMI and
mortality.(7) For instance, in some studies, overweight was associated with increased
mortality(8), but in others, the lowest mortality was observed among overweight individuals and
mortality tended to increase with lower BMI, even after accounting for smoking (residual
confounding) and preexisting disease (reverse causation).(9, 10) This pattern has come to be
known as the “obesity paradox”.(11) Given the existing and rising number of overweight and
obese adults in the US, these divergent findings could cause a great deal of confusion among
researchers, policy makers, and the general public.
One important but underexplored methodological limitation in the current obesity
research is that BMI is an imperfect measure of adiposity.(12-15) While BMI indicates
overweight relative to height, it does not discriminate between fat mass (FM) and lean body mass
(LBM).(16-18) Given the same BMI, body composition is highly variable among individuals.
This is particularly important because FM and LBM may act differently on health outcomes
including mortality. Excess FM has shown to be detrimental for health,(19) while growing
Page 52 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
5
evidence suggests that skeletal muscle, which accounts for majority of LBM, may be beneficial
for health.(20, 21) Therefore, understanding different contributions of LBM and FM to BMI may
provide new insights on the ‘obesity paradox’ and deliver important clinical and public health
messages regarding healthy body composition beyond BMI. However, direct measurement of
LBM is particularly difficult in large epidemiological studies because it requires expensive and
sophisticated technologies like dual-energy X-ray absorptiometry (DXA) or imaging
technologies. Therefore, little is known about the influence of body composition, particularly
LBM, on mortality. A limited number of studies have used less accurate surrogate measures (e.g.,
arm circumference,(22, 23) total body potassium,(24) skinfold,(25) and bioelectrical
impedance(26)) or direct measures(27-33) to estimate body composition but these studies had
relatively small sample size, short period of follow-up, restricted study population (e.g., elderly)
and/or potential biases (e.g., confounding and reverse causation). Moreover, the association of
LBM and FM with cause-specific mortality is largely unknown.
Therefore, we used validated anthropometric prediction equations to estimate body
composition and examine the association of predicted LBM, FM and BMI with all-cause and
cause-specific mortality in a large prospective US cohort study of men. Application of validated
equations in a large cohort allowed us to estimate LBM and FM and examine the independent
roles of two different body components in relation to mortality, accounting for potential biases.
Page 53 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
6
METHODS
Study population
The Health Professionals Follow-up Study was initiated in 1986 when 51,529 male health
professionals aged 40–75 were enrolled. Participants were mailed questionnaires at baseline and
every two years thereafter to collect updated demographics, lifestyle, and medical information.
For the analysis, we included participants who had information on age, race, height, weight and
waist circumference, which were required to create predicted LBM and FM (N=40,764). We
excluded participants previously diagnosed with cancer or cardiovascular diseases (N=2,118)
and those with BMI <12.5 or >60 kg/m2
(N=625) at baseline. The final sample size was 38,021
men.
Exposure assessments
Derivation and validation of the predicted LBM and FM has been described in detail
previously.(34) Briefly, we used a large US representative sample of 7,531 men who had
measured DXA from the National Health and Nutrition Examination Survey (NHANES). With
DXA-measured LBM and FM each as a dependent variable, a linear regression was performed
using age, race, height, weight, and waist circumference as independent predictors. Then, we
validated the developed equations in an independent validation group of 2,292 men and using
obesity-related biomarkers (i.e., triglycerides, total cholesterol, high-and low-density lipoprotein
cholesterol, glucose, insulin and C-reactive protein). The anthropometric prediction equations
had high predictive ability for LBM (R2=0.91, standard error of estimate (SEE)=2.6 kg) and FM
(R2=0.90, SEE=2.6 kg). Cross-validation in the validation group showed robustly high
Page 54 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
7
agreement between the actual and predicted LBM and FM with no evidence of bias. In an
additional validation, DXA-measured values and predicted values showed consistently high
agreement with similar errors across the range of LBM and FM. Scatter plots of the difference
between DXA and predicted values against DXA values showed no strong non-linear pattern
(Supplementary figure 1 and 2 and supplementary table 2 and 3). Moreover, the developed
equations performed well across different subgroups of the validation group (i.e., age, BMI, race,
smoking status, and disease status), and predicted FM showed similar correlations with obesity-
related biomarkers as DXA-measured FM.(34) For a sensitivity analysis, we also used different
prediction equations that include additional polynomial terms of anthropometric measures. These
equations had similar R2 and SEEs but slightly improved fit in the extreme range of LBM and
FM (Supplementary figure 1 and 2 and supplementary table 2 and 3). The anthropometric
prediction equations are shown in the supplement (Supplementary table 1). Using the equations,
predicted LBM and FM were calculated for each cohort member based on their age, race, height,
weight, and waist circumference. Predicted LBM and FM were available in 1987, 1996, and
2008.
We collected information on height at enrollment in 1986, and weight from biennial
questionnaires.(35, 36) Distinct from the biennial questionnaire, participants were asked to
measure and report their waist circumferences to the nearest one-quarter inch using provided
tape measures and following the same instructions in 1987, 1996, and 2008. Non-responders
received follow-up mailings to increase the response rate. In our validation study, the correlation
between self-reported and technician-measured height, weight, and waist circumference were
0.94, 0.97, and 0.95, respectively.(35)
Page 55 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
8
Ascertainment of outcomes
Deaths were identified by reports from the next of kin, postal authorities, or by searching the
National Death Index. More than 98% of deaths were ascertained from the follow up. Cause of
death was determined by physician review of medical records and death certificates. ICD-8
codes (International Classification of Diseases, 8th revision) were used to classify death due to
cardiovascular disease (codes 390-459, 795), cancer (codes 140-239), respiratory disease (codes
460-519), and other causes.
Ascertainment of covariates
Detailed information on age, race, smoking, and physical activity were collected in 1986 and
updated every two years from biannual questionnaires. Family history of cardiovascular disease
and cancer were assessed periodically. Dietary information was collected via validated food
frequency questionnaires every four years. The Alternate Healthy Eating Index (AHEI) was
calculated as an overall measure of diet quality.(37)
Statistical analyses
A Spearman correlation was calculated between predicted LBM and FM. Person-time of follow-
up was calculated from the age when the baseline predicted LBM and FM were available until
the age at death or the end of study (January 2012), whichever came first. Cox proportional
hazards models were used to estimate hazard ratios and 95% confidence interval (CI)s. We
stratified the analysis by age in months and calendar year of the questionnaire cycle.
Predicted FM and LBM were categorized into quintiles on the basis of the distribution of
exposures. We used predefined cut points for BMI (<18.5, 18.5-20.4, 20.5-22.4, 22.5-24.9, 25-
Page 56 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
9
27.4, 27.5-29.9, 30-34.9, and ≥35 kg/m2). For the main analysis, we used predicted FM, LBM,
and BMI measured at baseline to minimize the impact of underlying diseases on mortality. To
account for variation in body size, which is particularly important for LBM, we adjusted for
height by using residuals from the regression of LBM on height for LBM and by including
height as a continuous variable for FM in the models. In multivariable models, we adjusted for
potential confounders including race, family history of cardiovascular disease, family history of
cancer, smoking status, physical activity, total energy intake, alcohol consumption, and AHEI.
To examine the independent association of predicted LBM and FM in relation to mortality, we
further ran a multivariable model including both predicted LBM and FM. Test for trend was
conducted by treating the categorical predicted scores and BMI as continuous variables in the
model after assigning a median value for each category.
We also used restricted cubic splines with 5 knots at 5th
, 35th
, 50th
, 65th
, and 95th
percentiles to flexibly model the association between LBM and FM and mortality. We tested for
potential non-linearity using a likelihood ratio test comparing the model with only a linear term
to the model with linear and cubic spline terms.(38-40) Given our a priori hypothesis that people
with low LBM in the lower BMI range cause the J-or U-shaped relationship between BMI and
mortality, we examined how the shape of BMI-mortality relationship changes after excluding
those with low LBM. For a sensitivity analysis, we additionally examined the shape of BMI-
mortality relationship after excluding those with low FM.
To evaluate the latency between predicted LBM and FM and mortality, we conducted
analyses using different lag times (approximately 0, 4+, 8+, and 12+ years). For each lagged
analysis, the baseline was shifted to 1987, 1990, 1994, and 1998, respectively, and predicted
LBM and FM were updated using three repeated measures accordingly. For example, for no lag
Page 57 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
10
time analysis (simple updated), we used the most updated predicted LBM and FM that were
closest to the time of death. For a lag time of 4+ years, we used predicted measures in 1987 for
follow-up from 1990 to 2000 and predicted measures in 1996 for follow-up from 2000 to 2012.
Similarly, for a lag times of 8+ years, we used predicted measures in 1987 for deaths in 1994-
2004 and predicted measures in 1996 for deaths in 2004-2012. Moreover, we conducted
stratified analyses to explore whether the association of predicted LBM and FM with mortality
varied across smoking status and age.
Several sensitivity analyses were conducted with no adjustment for physical activity,
exclusion of deaths that have occurred in the early follow-up period (2 years) and right-censoring
criteria for age (>85 years), and inclusion of baseline illness. We also conducted analyses using
different categories for predicted LBM, FM and BMI (i.e., quintiles and deciles). Lastly, we
tested the robustness of our findings using other prediction equations with polynomial terms. All
statistical tests were two-sided and P<0.05 was considered to determine statistical significance.
We used SAS 9.4 for all analyses (SAS institute).
Patient involvement
No patients were involved in setting the research question or the outcome measures, nor were
they involved in the design and implementation of the study. There are no plans to involve
patients in dissemination.
Page 58 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
11
RESULTS
Study participants
A total of 38,021 men were included in the analyses. Baseline characteristics of participants
according to BMI categories are presented in Table 1. The mean age was 54.4 years and the
mean BMI was 25.4 kg/m2. Predicted LBM increased with higher BMI. Predicted FM slightly
deceased in the second category of BMI (18.5-20.4 kg/m2) and then increased with higher BMI.
Moreover, men with lower BMI tended to have higher physical activity and AHEI score, peaking
in the third category of BMI (20.5-22.4 kg/m2). Although the number of men with underweight
(BMI<18 kg/m2) was small, they were taller and had higher waist circumference and lower
physical activity and AHEI score. The Spearman correlation between predicted LBM and FM
was 0.66 in men.
Page 59 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
12
Table 1 Age-standardized baseline characteristics according to body mass index in men (Health Professionals Follow-up
Study, 1987-2012) Body mass index (kg/m
2)
<18.5 18.5-20.4 20.5-24.9 22.5-24.9 25.0-27.4 27.5-29.9 30.0-34.9 ≥35.0
Person-years 1839 15337 92790 254122 243335 95023 52320 8275
Age (year)a 55.5 (10.4) 54.0 (10.8) 53.8 (10.2) 54.0 (9.9) 54.5 (9.7) 54.9 (9.6) 55.1 (9.4) 55.5 (10.1)
Height (cm) 185.4 (12.7) 179.4 (7.8) 178.7 (6.2) 178.5 (6.4) 178.2 (6.5) 178.7 (6.8) 178.5 (7.1) 176.4 (10.0)
Weight (kg) 60.8 (8.2) 64.1 (5.7) 69.6 (5.1) 75.9 (5.8) 82.9 (6.4) 91.4 (7.3) 101.2 (9.0) 118.2 (13.9)
Waist circumference (cm) 86.6 (12.4) 82.8 (5.3) 86.8 (5.1) 91.2 (5.6) 96.7 (6.1) 102.9 (6.8) 110.6 (7.9) 123.4 (11.4)
BMI (kg/m2) 17.6 (0.8) 19.8 (0.5) 21.7 (0.5) 23.7 (0.7) 26.0 (0.7) 28.5 (0.7) 31.7 (1.3) 37.9 (3.6)
Predicted fat mass (kg) 13.3 (5.0) 13.1 (2.5) 15.9 (2.4) 19.1 (2.6) 22.8 (2.9) 27.1 (3.3) 32.3 (4.1) 41.2 (6.5)
Predicted Lean body mass (kg) 40.4 (5.8) 47.5 (2.2) 50.6 (1.9) 53.9 (2.1) 57.4 (2.3) 61.2 (2.6) 65.9 (3.4) 75.2 (6.0)
Total energy intake (kcal/day) 2132 (610) 2023 (570) 2045 (599) 2002 (595) 1992 (609) 2002 (625) 2036 (639) 2089 (657)
Alcohol consumption (g/day) 14.2 (18.7) 9.7 (14.2) 10.9 (14.3) 11.5 (14.7) 11.8 (15.4) 11.7 (15.5) 10.9 (16.1) 8.9 (15.1)
AHEI (score) 51.4 (13.5) 54.1 (12.7) 54.3 (12.0) 53.8 (11.6) 52.4 (11.1) 51.5 (10.9) 50.7 (11.0) 49.3 (10.8)
Physical activity (MET-h/wk) 21.4 (35.8) 22.6 (27.0) 24.2 (28.6) 22.3 (27.4) 19.4 (23.9) 16.8 (22.0) 14.4 (20.9) 11.7 (14.9)
White (%) 98.4 99.2 99.5 99.3 99.2 98.8 98.7 99.4
Family history of CVD (%) 35.3 32.2 33.0 33.4 33.7 33.8 35.2 35.5
Family history of cancer (%) 17.6 16.8 17.2 16.8 17.5 16.9 16.8 15.4
Smoking status (%)
Never 47.4 56.5 56.0 50.5 45.8 44.1 42.3 41.1
Past 34.0 32.3 35.2 42.2 46.0 47.5 50.0 50.6
Current 18.6 11.2 8.9 7.3 8.2 8.4 7.8 8.3
Abbreviation: BMI, body mass index; AHEI, alternate healthy eating index; CVD, cardiovascular disease
Data are presented as means (SD) for continuous variables and percentages for categorical variables, unless otherwise indicated. a Value is not age adjusted
Page 60 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
13
All-cause mortality
During up to 25 years of follow-up (mean of 20.4 years), we identified 12,356 deaths. The
association of predicted FM and LBM with all-cause mortality in men is presented in Table 2. A
multivariable adjusted model showed a positive association between predicted FM and all-cause
mortality, while predicted LBM showed a U-shaped association with all-cause mortality. In a
mutually adjusted model including both predicted FM and LBM, we consistently observed a
strong positive association between predicted FM and all-cause mortality. Compared to those in
the lowest quintile of predicted FM, men in the highest quintiles had 35% (95% CI: 26 to 46%)
increased hazard of all-cause mortality. Moreover, predicted LBM showed a stronger U-shaped
association with all-cause mortality in the mutually adjusted model. Compared to those in the
lowest quintile of predicted LBM, men in the second to fourth quintiles had 8 to 10% decreased
hazard of all-cause mortality.
In Figure 1, we used restricted cubic splines to flexibly model and visualize the
relationship between predicted FM and LBM with all-cause mortality in men. The risk of all-
cause mortality was relatively flat and increased slightly until around 21 kg of predicted FM, and
then started to increase rapidly afterwards (P for non-linearity<0.001). The average BMI for men
with 21 kg of FM is 25 kg/m2. In respect to the strong U-shaped relationship between predicted
LBM and all-cause mortality, the plot showed a substantial reduction of the risk within the lower
range of predicted LBM, which reached the lowest risk around 55 kg and then increased
thereafter (P for non-linearity<0.001).
Page 61 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
14
Table 2. Hazard ratio (95% CI) of all-cause mortality according to predicted fat mass and lean body
mass in men (Health Professionals Follow-up Study)
Analysis
Hazard Ratio (95% CI)
No of
deaths
IR
/100,000py Model 1 Model 2 Model 3
Fat massa,b
Quintile 1 1937 1265 1 (reference) 1 (reference) 1 (reference)
Quintile 2 2298 1504 1.09 (1.03 to 1.16) 1.06 (1.00 to 1.12) 1.08 (1.01 to 1.15)
Quintile 3 2297 1504 1.03 (0.97 to 1.09) 0.98 (0.92 to 1.04) 1.01 (0.94 to 1.07)
Quintile 4 2726 1789 1.23 (1.16 to 1.31) 1.13 (1.06 to 1.20) 1.16 (1.09 to 1.24)
Quintile 5 3098 2038 1.51 (1.42 to 1.60) 1.33 (1.25 to 1.41) 1.35 (1.26 to 1.46)
P-trend
<.001 <.001 <.001
Lean body massa,b
Quintile 1 2996 1969 1 (reference) 1 (reference) 1 (reference)
Quintile 2 2419 1585 0.93 (0.88 to 0.98) 0.93 (0.88 to 0.98) 0.92 (0.87 to 0.97)
Quintile 3 2324 1521 0.95 (0.90 to 1.01) 0.93 (0.88 to 0.98) 0.90 (0.85 to 0.96)
Quintile 4 2282 1494 1.03 (0.98 to 1.09) 1.00 (0.95 to 1.06) 0.92 (0.87 to 0.98)
Quintile 5 2335 1529 1.26 (1.20 to 1.34) 1.16 (1.10 to 1.23) 0.97 (0.91 to 1.04)
P-trend <.001 <.001 0.49
Model 1: adjusted for age.
Model 2: adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no),
physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol consumption (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total
energy intake (quintiles), and smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), Alternate Healthy Eating Index (quintiles).
Model 3: additionally, mutually adjusted for predicted fat mass and predicted lean body mass. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 62 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
15
When we used BMI alone, we observed a J-shaped relationship between BMI and all-
cause mortality in men (Table 3 and Figure 1). We also examined the influence on BMI when we
excluded men with low predicted LBM. When we excluded those in the lowest 2.5th
percentiles
of predicted LBM, the J-shaped relationship between BMI and mortality disappeared. Upon
excluding more participants with low predicted LBM (5th
and 10th
percentiles), the BMI-
mortality relationship became more linear and slightly stronger. However, the J-shaped
relationship still existed when excluding those with low FM (Supplementary table 4).
We further examined how the association of predicted FM and LBM with all-cause
mortality changes by different lag times (Supplementary table 5). With shorter lag times,
predicted FM showed a less linear positive association with all-cause mortality, while predicted
LBM showed a stronger U-shaped association with all-cause mortality. We also examined the
associations stratified by smoking status and age (Supplementary table 6 and 7). The relationship
between predicted FM and all-cause mortality was stronger and more linear among never-
smokers compared to current-smokers and among younger adults compared to older adults. On
the other hand, we observed a stronger U-shaped association between predicted LBM and all-
cause mortality among current-smokers compared to never-or past-smokers. We observed a
similar U-shaped relationship for predicted LBM across all age groups.
Our findings remained robust in several sensitivity analyses (Supplementary table 8, 9
and 10). The results did not change with no adjustment for physical activity, exclusion of deaths
in the early follow-up period and right-censoring criteria for age, inclusion of baseline illness and
use of quintiles and deciles for exposures. Moreover, using other prediction equations with
polynomial terms showed consistent results (data not shown).
Page 63 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
16
Table 3. Hazard ratio (95% CI) of all-cause mortality according to body mass index in men (Health Professionals Follow-
up Study)
Analysis
Hazard Ratio (95% CI)
No of
deaths
IR
/100,000py Model 1
a Model 2
a Model 3
b Model 4
c Model 5
d
BMI
<18.5 53 2883 1.74 (1.33 to 2.28) 1.65 (1.25 to 2.16) NA NA NA
18.5-20.4 269 1754 1.16 (1.03 to 1.32) 1.09 (0.96 to 1.24) 0.95 (0.76-1.18) 0.95 (0.68 to 1.32) 0.79 (0.38 to 1.66)
20.5-22.4 1358 1464 1.01 (0.95 to 1.08) 1.03 (0.97 to 1.09) 1.01 (0.94-1.08) 0.99 (0.92 to 1.06) 0.93 (0.85 to 1.03)
22.5-24.9 3740 1472 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 3986 1638 1.09 (1.04 to 1.14) 1.04 (0.99 to 1.09) 1.04 (1.00-1.09) 1.05 (1.00 to 1.10) 1.06 (1.01 to 1.11)
27.5-29.9 1753 1845 1.31 (1.24 to 1.39) 1.21 (1.14 to 1.28) 1.22 (1.15-1.29) 1.23 (1.16 to 1.30) 1.24 (1.17 to 1.31)
30.0-34.9 1001 1913 1.48 (1.38 to 1.59) 1.31 (1.22 to 1.41) 1.31 (1.22-1.41) 1.33 (1.24 to 1.43) 1.34 (1.25 to 1.44)
≥35.0 196 2368 2.28 (1.98 to 2.64) 2.01 (1.74 to 2.33) 2.02 (1.75-2.34) 2.04 (1.76 to 2.36) 2.06 (1.78 to 2.38)
P-trend
<.001 <.001 <.001 <.001 <.001
Abbreviation: BMI, body mass index; NA, not available (no cases available after exclusion).
Model 1: adjusted for age.
Model 2: adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or no), physical activity (<3, 3-8.9, 9-
17.9, 18-26.9, or >27 MET-hour/week), alcohol consumption (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy intake (quintiles), smoking status (never, ever, 1-14,
15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles).
Model 3: additionally, excluded 2.5%ile of total participants with low lean body masse
Model 4: additionally, excluded 5%ile of total participants with low lean body masse
Model 5: additionally, excluded 10%ile of total participants with low lean body masse a Number of deaths/person-years for each category of BMI: 53/1839, 269/15337, 1358/92790, 3740/254122, 3986/243335, 1753/95023, 1001/52320, and 196/8275. b Number of deaths/person-years for each category of BMI: 0/26, 80/7196, 1147/85194, 3695/252987, 3980/243212, 1751/94960, 1000/52311, and 196/8275. c Number of deaths/person-years for each category of BMI: 0/0, 36/3402, 884/72686, 3603/250459, 3976/243080, 1751/94960, 1000/52311, and 196/8275. d Number of deaths/person-years for each category of BMI: 0/0, 7/781, 509/48989, 3295/239196, 3963/242606, 1751/94928, 1000/52311, and 196/8275. e For exclusion analyses, height-adjusted lean body mass was used after regressing out variation due to height.
Page 64 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
17
Cause-specific mortality
We further examined the association of predicted FM and LBM with cause-specific mortality
(Table 4). Mutually adjusted models showed a linear positive association between predicted FM
and mortality due to cardiovascular disease and cancer. Compared to those in the lowest quintile
of predicted FM, men in the highest quintile had 67% (95% CI: 47 to 90%) and 24% (95% CI: 8
to 42%) increased hazard of death due to cardiovascular disease and cancer, respectively. In
contrast, predicted LBM showed a U-shaped association with mortality due to cardiovascular
disease and cancer in the mutually adjusted models. However, predicted LBM showed a strong
inverse association with mortality due to respiratory disease (P for trend<.001). Compared to
those in the lowest quintile of predicted LBM, men in the highest quintile had 50% (95% CI: 39
to 65%) decreased hazard of death due to respiratory disease. When we examined the association
between BMI and cause-specific mortality, we observed a U-shaped association for
cardiovascular disease death but a positive association for cancer death and an inverse
association for respiratory disease death.
Page 65 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
18
Table 4. Hazard ratio (95% CI) of cause-specific mortality according to predicted fat mass, lean body
mass and body mass index in men (Health Professionals Follow-up Study)
Analysis Hazard Ratio (95% CI)
CVD death Cancer death Respiratory death Other death
No. of deaths 4296 3723 960 3377
IR/100,000py 558 483 124 438
Fat massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 1.11 (1.00 to 1.24) 1.15 (1.03 to 1.29) 0.92 (0.74 to 1.14) 1.01 (0.90 to 1.14)
Quintile 3 1.10 (0.99 to 1.23) 1.06 (0.94 to 1.19) 1.06 (0.85 to 1.31) 0.84 (0.74 to 0.94)
Quintile 4 1.30 (1.16 to 1.46) 1.15 (1.02 to 1.30) 1.10 (0.88 to 1.38) 1.02 (0.90 to 1.15)
Quintile 5 1.67 (1.47 to 1.90) 1.24 (1.08 to 1.42) 1.26 (0.97 to 1.64) 1.13 (0.98 to 1.30)
P-trend <.001 0.01 0.03 0.05
Lean body massa,b
Quintile 1 1 (reference) 1 (reference) 1 (reference) 1 (reference)
Quintile 2 0.96 (0.87 to 1.06) 0.97 (0.88 to 1.08) 0.61 (0.51 to 0.74) 0.94 (0.84 to 1.04)
Quintile 3 0.95 (0.86 to 1.05) 0.94 (0.84 to 1.04) 0.58 (0.47 to 0.71) 0.95 (0.86 to 1.07)
Quintile 4 0.96 (0.87 to 1.07) 0.95 (0.84 to 1.06) 0.57 (0.46 to 0.71) 1.00 (0.89 to 1.12)
Quintile 5 1.11 (0.98 to 1.24) 1.02 (0.90 to 1.16) 0.50 (0.39 to 0.65) 0.98 (0.86 to 1.12)
P-trend 0.10 0.86 <.001 0.99
BMI
<18.5 1.45 (0.87 to 2.41) 0.66 (0.32 to 1.40) 5.33 (3.10 to 9.17) 1.86 (1.15 to 3.01)
18.5-20.4 1.12 (0.90 to 1.38) 0.99 (0.78 to 1.25) 1.93 (1.36 to 2.73) 0.92 (0.72 to 1.19)
20.5-22.4 0.95 (0.85 to 1.06) 0.97 (0.87 to 1.09) 1.30 (1.06 to 1.60) 1.09 (0.97 to 1.22)
22.5-24.9 1 (reference) 1 (reference) 1 (reference) 1 (reference)
25.0-27.4 1.16 (1.08 to 1.26) 1.01 (0.93 to 1.09) 0.91 (0.78 to 1.08) 0.98 (0.92 to 1.06)
27.5-29.9 1.40 (1.27 to 1.54) 1.13 (1.02 to 1.25) 1.09 (0.89 to 1.35) 1.10 (0.98 to 1.23)
30.0-34.9 1.75 (1.56 to 1.96) 1.12 (0.98 to 1.28) 0.81 (0.60 to 1.09) 1.18 (1.02 to 1.35)
≥35.0 2.66 (2.11 to 3.36) 1.55 (1.17 to 2.04) 0.90 (0.43 to 1.92) 2.13 (1.63 to 2.77)
P-trend <.001 <.001 <.001 0.002
Abbreviation: BMI, body mass index; CVD, cardiovascular disease; NA, not available (no cases available after exclusion).
All models were adjusted for age, race (white or non-white), family history of cardiovascular disease (yes or no), family history of cancer (yes or
no), physical activity (<3, 3-8.9, 9-17.9, 18-26.9, or >27 MET-hour/week), alcohol (0, 0.1-4.9, 5-9.9, 10-14.9, or 15.0+ g/day), total energy
intake (quintiles), smoking status (never, ever, 1-14, 15-24, ≥25 cigs/day), and Alternate Healthy Eating Index (quintiles). Fat mass and lean
body mass were mutually adjusted in the model. a Derived from validated anthropometric prediction equations.
b Height was adjusted by including height as a continuous variable for fat mass and by regressing out variation due to height for lean body mass.
Page 66 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
19
DISCUSSION
Principal findings
In a large prospective cohort study of men, we used validated anthropometric prediction
equations to examine the association of LBM and FM with all-cause and cause-specific mortality.
We found a strong positive association between predicted FM and mortality due to all causes,
cardiovascular disease, and cancer. In contrast, predicted LBM showed a U-shaped association
with mortality due to all causes, cardiovascular disease, and cancer, and an inverse association
with mortality due to respiratory disease.
Comparison with other studies
Numerous epidemiological studies have examined the relationship between BMI and mortality,
but controversy and confusion exist around the unexpected J- or U-shaped association between
BMI and mortality.(9) A systematic review and meta-analysis of 141 prospective studies in 2013
reported that, relative to normal weight, both grade 2 and grade 3 obesity (BMI≥30 kg/m2) were
associated with higher all-cause mortality, but overweight (BMI 25-29.9 kg/m2) was associated
with lower all-cause mortality. In contrast, an individual participant-data meta-analysis of 239
prospective studies by the Global BMI Mortality Collaboration in 2016 showed evidence that
increased risk of all-cause mortality among overweight was largely due to confounding by
aspects of smoking and reverse causation from underlying disease and frailty at older ages.(8, 41,
42) There are ongoing controversies around the ‘obesity paradox’(43-45) with many studies
reporting excess mortality at the lower BMI range. More importantly, these studies
acknowledged the major limitation of BMI as a measure of adiposity but did not directly
Page 67 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
20
investigate two distinct compartments of body composition (LBM and FM) in relation to
mortality.(8, 9)
Our findings on BMI were in line with the previous findings, whereby we consistently
observed a J-shaped relationship with mortality even after accounting for age, smoking, and
baseline diseases. Two different shapes in mortality risk for FM and LBM taken together can
explain the observed J-shaped relationship between BMI and mortality in our study. The
increased risk of mortality in the lower BMI range (<25 kg/m2) could be attributed to a
combination of the high risk among men with low predicted LBM, which over-rides the modest
positive association between predicted FM and mortality in this lower range of BMI. The
increase of mortality risk at the BMI range of 25-30 kg/m2 is likely due to the high risk
associated with predicted FM in combination with only a moderate risk associated with predicted
LBM. Lastly, the rapid increase of mortality risk in the higher BMI range (>30kg/m2) could be
due to a very high risk associated with both predicted FM and LBM. Of note, at the high end of
BMI (>30kg/m2), the vast majority of individuals have high predicted FM and LBM. Those with
high predicted LBM almost invariably have high FM; for example, the average predicted FM for
those in the highest decile of predicted LBM was 31 kg (Supplementary table 10).
These observed patterns for FM and LBM were further supported from our additional
analyses of BMI and mortality after excluding those in the lower end of predicted LBM, which
resulted in a strong linear positive relationship between BMI and mortality. This shows that
separating lean and healthy (low BMI and normal LBM) vs. lean and unhealthy (low BMI and
low LBM) individuals could be a key to explain the ‘obesity paradox’ phenomenon. Our data
directly address the controversial hypothesis that accumulating excess fat may be causally
beneficial for survivor, and show that this is not likely to be true.(46)
Page 68 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
21
To date, only a limited number of studies have examined mortality in relation to directly
measured body composition using DXA or computed tomography.(27-33) Most studies have
been done among elderly populations with approximate mean age of 75 years.(27-29, 31, 33)
The findings showed inconsistent and various shapes of the relationship. An Italian study of 934
older people showed no association of calf muscle and fat mass area with 6-year mortality(28),
while another study of 2292 elderly from the US found that low leg muscle area was associated
with increased risk of 6-year mortality in men only.(29) In contrast, other studies found that
lower fat percent23 27
or leg fat mass29
was associated with increased mortality(27, 31, 33) but
among these studies, only one study of 477 community-dwelling older people from Netherlands
reported that lower appendicular skeletal muscle mass was associated with increased risk of 12-
year mortality.(33) Although these studies used a direct measure of body composition, the study
samples were restricted to elderly population which limits the generalizability of the findings.
Moreover, they also had other limitations such as small sample size, short follow-up, exposure
measured at one-time point, lack of information on important confounders (especially smoking)
and no examination on cause-specific mortality. Nonetheless, our finding was consistent with a
recent large-scale Canadian study that measured DXA from participants referred for bone
mineral density testing.(30) That study found that high percent fat and low BMI were
independently associated with increased risk of mortality when percent fat and BMI were
simultaneously adjusted in the models. However, the observed associations might have been
confounded by smoking or physical activity due to lack of information on those variables, and
the study did not directly use LBM in the analysis.
The BMI-mortality relationship is prone to reverse causation by preexisting diseases that
can cause weight loss and also increase risk of mortality, and this is more likely to be a concern
Page 69 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
22
with shorter lag times. We found that, with shorter lag time periods, the positive association
between predicted FM and mortality was attenuated, while the U-shaped association between
predicted LBM and mortality tended to be strengthened. Therefore, the stronger U-shaped
relationship between BMI and mortality with shorter lag time periods can be mostly attributed to
the pronounced U-shaped association for predicted LBM, which may be an indicator of health
status capturing any preexisting undiagnosed medical condition, including frailty at older ages.
The influence of smoking is particularly important in investigating the obesity-mortality
relationship. Not only is smoking a strong risk factor for death, but it also affects body weight
and body composition.(47-50) Similar to the BMI-mortality association, predicted FM showed
weaker and less linear association with mortality among current-smokers than past-or never-
smokers. Interestingly, we found a stronger U-shaped relationship between predicted LBM and
mortality among current-smokers than past-or never-smokers. Although we cannot completely
rule out the residual confounding by smoking, our findings showed some evidence that the
frequently observed U-shaped relationship between BMI and mortality among smokers may be
affected by the strong U-shaped association between LBM and mortality.
Strengths and limitations of the study
Our study has several strengths. First, the innovative approach of validated anthropometric
prediction equations allowed us to practically estimate LBM and FM in large epidemiological
settings. This is the first and one of the most comprehensive analyses to examine the association
of predicted body composition with all-cause and cause-specific mortality in a large prospective
cohort study. Second, the Health Professionals Follow-up Study is a well-established prospective
cohort study that has a large number of deaths over long-term follow-up period. Third, detailed
Page 70 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
23
and updated information on lifestyle and health-related factors allowed adequate control for
confounding. Fourth, repeated measures on exposures (i.e., predicted scores) allowed prospective
analyses of different lag time periods to examine reverse causality in the obesity-mortality
relationship.
There are several limitations as well. First, predicted LBM and FM are not perfect
measures of actual LBM and FM. Nonetheless, the validation results from the NHANES showed
high predictive ability of the anthropometric equations with no systematic bias. In fact, the very
high R2 between FM and LBM (>0.90) for direct DXA measurements and predicted measures in
an independent dataset indicate that a direct DXA measure would give very similar answers to
ours; this is further supported by the equal predictive ability of the predicted measures and DXA
measures for various obesity-related biomarkers. Moreover, given the prospective study design,
any mismeasurement in the exposures would likely be random with respect to endpoints,
resulting in conservative associations. Second, we cannot entirely rule out the possibility of
unmeasured or unknown confounding factors that may account for the associations observed in
this study. However, the homogeneity of the study population and comprehensive data on the
risk factors minimized potential confounding. Third, the generalizability of the findings may be
limited given that the study participants were restricted to health professionals and
predominantly White men. However, we believe that our main findings will be broadly
applicable.
Clinical and public health implications
The current study provides strong evidence that excess FM increases the risk of mortality.
Increased FM was not protective for mortality which is counter to the premise of the ‘obesity
Page 71 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
24
paradox’. On the other hand, low LBM was associated with increased risk of mortality in the
lower range of BMI. Our study suggests that understanding different contributions of LBM and
FM to BMI may explain the shape of the relationship between BMI and mortality as well as the
controversial issue of ‘obesity paradox’. Our findings support the current recommendations on
body weight for prevention of chronic diseases to keep a normal body weight defined by BMI of
18.5 to 24.9 kg/m2, and further suggest that the lowest mortality risk can be observed at the lower
normal range of BMI if the influence of LBM on mortality is accounted for. The current
recommendations should highlight not only the importance of normal body weight but also
healthy body composition (e.g., healthy lean) to reduce confusion around the optimal weight
(e.g., ‘obesity paradox’) for overall health. Interventions and strategies to promote healthy body
composition via lifestyle modification (e.g., physical activity and diet) may be an important next
step to improve population health.
Conclusions
We found a strong positive association between predicted FM and mortality, and a U-shaped
association between predicted LBM and mortality in men. Low LBM, rather than low FM, may
be driving the increased risk of mortality in the lower BMI range. Understanding the independent
role of LBM and FM has important implications for clarifying the ‘obesity paradox’
phenomenon in the relationship between BMI and mortality.
Page 72 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
25
Acknowledgements: We thank the participants and staff of the HSPF for their valuable
contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO,
CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH,
OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for
analyses and interpretation of these data.
Contributors: DHL and ELG had full access to all of the data in the study and take
responsibility for the integrity of the data and the accuracy of the data analysis. DHL and ELG
conceived and designed the study. NK, FBH, EJO, EBR, WCW, and ELG acquired the data.
DHL and ELG drafted the manuscript. All the authors critically revised the manuscript for
important intellectual content. DHL did the statistical analysis. FBH, EBR, WCW, and ELG
obtained funding. DHL and ELG were responsible for administrative, technical, or material
support. ELG was responsible for study supervision. DHL is the guarantor.
Funding: This work was supported by the National Institutes of Health (UM1 CA167552 and
R01 HL35464). The funders had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; and preparation, review, or approval of the
manuscript; and 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
Page 73 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
26
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: This investigation was approved by the Institutional Review Board of the
Harvard T.H. Chan School of Public Health and Brigham and Women’s Hospital.
Transparency declaration: The lead author (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 as
planned (and, if relevant, registered) have been explained.
Data sharing: No additional data available.
Copyright/License for Publication: The Corresponding Author has the right to grant on behalf
of all authors and does grant on behalf of all authors, a worldwide licence to the Publishers and
its licensees in perpetuity, in all forms, formats and media (whether known now or created in the
future), to i) publish, reproduce, distribute, display and store the Contribution, ii) translate the
Contribution into other languages, create adaptations, reprints, include within collections and
create summaries, extracts and/or, abstracts of the Contribution, iii) create any other derivative
work(s) based on the Contribution, iv) to exploit all subsidiary rights in the Contribution, v) the
inclusion of electronic links from the Contribution to third party material where-ever it may be
located; and, vi) licence any third party to do any or all of the above.
Page 74 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
27
This is an Open Access article distributed in accordance with the Creative Commons Attribution
Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt,
build upon this work non-commercially, and license their derivative works on different terms,
provided the original work is properly cited and the use is non-commercial. See:
http://creativecommons.org/licenses/by-nc/4.0/.
Page 75 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
28
References
1. Obesity: preventing and managing the global epidemic. Report of a WHO consultation.
World Health Organization technical report series. 2000;894:i-xii, 1-253.
2. Centers for Disease Control and Prevention. National Health and Nutrition Examination
Survey. Available at: http://www.cdc.gov/nchs/about/major/nhanes/datalink.htm. Accessed
March, 2018.
3. Hu F. Obesity epidemiology: Oxford University Press; 2008.
4. Adams KF, Schatzkin A, Harris TB, Kipnis V, Mouw T, Ballard-Barbash R, et al.
Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old.
The New England journal of medicine. 2006;355(8):763-78.
5. Berrington de Gonzalez A, Hartge P, Cerhan JR, Flint AJ, Hannan L, MacInnis RJ, et al.
Body-mass index and mortality among 1.46 million white adults. The New England journal of
medicine. 2010;363(23):2211-9.
6. Chen Z, Yang G, Offer A, Zhou M, Smith M, Peto R, et al. Body mass index and
mortality in China: a 15-year prospective study of 220 000 men. International journal of
epidemiology. 2012;41(2):472-81.
7. Manson JE, Bassuk SS, Hu FB, Stampfer MJ, Colditz GA, Willett WC. Estimating the
number of deaths due to obesity: can the divergent findings be reconciled? Journal of women's
health. 2007;16(2):168-76.
8. Global BMIMC, Di Angelantonio E, Bhupathiraju Sh N, Wormser D, Gao P, Kaptoge S,
et al. Body-mass index and all-cause mortality: individual-participant-data meta-analysis of 239
prospective studies in four continents. Lancet. 2016;388(10046):776-86.
Page 76 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
29
9. Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality with
overweight and obesity using standard body mass index categories: a systematic review and
meta-analysis. Jama. 2013;309(1):71-82.
10. Veronese N, Cereda E, Solmi M, Fowler SA, Manzato E, Maggi S, et al. Inverse
relationship between body mass index and mortality in older nursing home residents: a meta-
analysis of 19,538 elderly subjects. Obesity reviews : an official journal of the International
Association for the Study of Obesity. 2015;16(11):1001-15.
11. Greenberg JA. The obesity paradox in the US population. The American journal of
clinical nutrition. 2013;97(6):1195-200.
12. Visscher TL, Seidell JC, Molarius A, van der Kuip D, Hofman A, Witteman JC. A
comparison of body mass index, waist-hip ratio and waist circumference as predictors of all-
cause mortality among the elderly: the Rotterdam study. International journal of obesity and
related metabolic disorders : journal of the International Association for the Study of Obesity.
2001;25(11):1730-5.
13. Romero-Corral A, Lopez-Jimenez F, Sierra-Johnson J, Somers VK. Differentiating
between body fat and lean mass-how should we measure obesity? Nature clinical practice
Endocrinology & metabolism. 2008;4(6):322-3.
14. Okorodudu DO, Jumean MF, Montori VM, Romero-Corral A, Somers VK, Erwin PJ, et
al. Diagnostic performance of body mass index to identify obesity as defined by body adiposity:
a systematic review and meta-analysis. International journal of obesity. 2010;34(5):791-9.
15. Pischon T. Commentary: Use of the body mass index to assess the risk of health
outcomes: time to say goodbye? International journal of epidemiology. 2010:dyp388.
Page 77 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
30
16. Gallagher D, Visser M, Sepulveda D, Pierson RN, Harris T, Heymsfield SB. How useful
is body mass index for comparison of body fatness across age, sex, and ethnic groups? American
journal of epidemiology. 1996;143(3):228-39.
17. Gallagher D, Ruts E, Visser M, Heshka S, Baumgartner RN, Wang J, et al. Weight
stability masks sarcopenia in elderly men and women. American journal of physiology
Endocrinology and metabolism. 2000;279(2):E366-75.
18. Harris TB. Invited commentary: body composition in studies of aging: new opportunities
to better understand health risks associated with weight. American journal of epidemiology.
2002;156(2):122-4; discussion 5-6.
19. Haslam DW, James WP. Obesity. (1474-547X (Electronic)).
20. Wannamethee SG, Atkins JL. Muscle loss and obesity: the health implications of
sarcopenia and sarcopenic obesity. Proceedings of the Nutrition Society. 2015;74(04):405-12.
21. Rolland Y, Czerwinski S, Van Kan GA, Morley J, Cesari M, Onder G, et al. Sarcopenia:
its assessment, etiology, pathogenesis, consequences and future perspectives. The Journal of
Nutrition Health and Aging. 2008;12(7):433-50.
22. Wannamethee SG, Shaper AG, Lennon L, Whincup PH. Decreased muscle mass and
increased central adiposity are independently related to mortality in older men. The American
journal of clinical nutrition. 2007;86(5):1339-46.
23. Allison DB, Zhu SK, Plankey M, Faith MS, Heo M. Differential associations of body
mass index and adiposity with all-cause mortality among men in the first and second National
Health and Nutrition Examination Surveys (NHANES I and NHANES II) follow-up studies.
International journal of obesity and related metabolic disorders : journal of the International
Association for the Study of Obesity. 2002;26(3):410-6.
Page 78 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
31
24. Heitmann B, Erikson H, Ellsinger B, Mikkelsen K, Larsson B. Mortality associated with
body fat, fat-free mass and body mass index among 60-year-old Swedish menFa 22-year follow-
up. The study of men born in 1913. International journal of obesity and related metabolic
disorders : journal of the International Association for the Study of Obesity. 2000;24:33-7.
25. Lee CD, Blair SN, Jackson AS. Cardiorespiratory fitness, body composition, and all-
cause and cardiovascular disease mortality in men. The American journal of clinical nutrition.
1999;69(3):373-80.
26. Bigaard J, Frederiksen K, Tjonneland A, Thomsen BL, Overvad K, Heitmann BL, et al.
Body fat and fat-free mass and all-cause mortality. Obesity research. 2004;12(7):1042-9.
27. Auyeung TW, Lee JS, Leung J, Kwok T, Leung PC, Woo J. Survival in older men may
benefit from being slightly overweight and centrally obese—a 5-year follow-up study in 4,000
older adults using DXA. Journals of Gerontology Series A: Biomedical Sciences and Medical
Sciences. 2009;65(1):99-104.
28. Cesari M, Pahor M, Lauretani F, Zamboni V, Bandinelli S, Bernabei R, et al. Skeletal
muscle and mortality results from the InCHIANTI Study. Journals of Gerontology Series A:
Biomedical Sciences and Medical Sciences. 2009;64(3):377-84.
29. Newman AB, Kupelian V, Visser M, Simonsick EM, Goodpaster BH, Kritchevsky SB, et
al. Strength, but not muscle mass, is associated with mortality in the health, aging and body
composition study cohort. The Journals of Gerontology Series A: Biological Sciences and
Medical Sciences. 2006;61(1):72-7.
30. Padwal R, Leslie WD, Lix LM, Majumdar SR. Relationship Among Body Fat Percentage,
Body Mass Index, and All-Cause Mortality: A Cohort Study. Annals of internal medicine. 2016.
Page 79 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
32
31. Rolland Y, Gallini A, Cristini C, Schott A-M, Blain H, Beauchet O, et al. Body-
composition predictors of mortality in women aged≥ 75 y: data from a large population-based
cohort study with a 17-y follow-up–. The American journal of clinical nutrition.
2014;100(5):1352-60.
32. Toss F, Wiklund P, Nordström P, Nordström A. Body composition and mortality risk in
later life. Age and ageing. 2012;41(5):677-81.
33. Wijnhoven HA, Snijder MB, Deeg DJ, Visser M. Region-specific fat mass and muscle
mass and mortality in community-dwelling older men and women. Gerontology. 2012;58(1):32-
40.
34. Lee DH, Keum N, Hu FB, Orav EJ, Rimm EB, Sun Q, et al. Development and validation
of anthropometric prediction equations for lean body mass, fat mass and percent fat in adults
using the National Health and Nutrition Examination Survey (NHANES) 1999–2006. British
Journal of Nutrition. 2017;118(10):858-66.
35. Rimm EB, Stampfer MJ, Colditz GA, Chute CG, Litin LB, Willett WC. Validity of self-
reported waist and hip circumferences in men and women. Epidemiology. 1990;1(6):466-73.
36. Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC. The validity of
recalled weight among younger women. International journal of obesity and related metabolic
disorders : journal of the International Association for the Study of Obesity. 1995;19(8):570-2.
37. McCullough ML, Willett WC. Evaluating adherence to recommended diets in adults: the
Alternate Healthy Eating Index. Public health nutrition. 2006;9(1a):152-7.
38. Durrleman S, Simon R. Flexible regression models with cubic splines. Statistics in
medicine. 1989;8(5):551-61.
Page 80 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
33
39. Govindarajulu U, Spiegelman D, Thurston S, Eisen E. Comparing smoothing techniques
for modeling exposure-response curves in Cox models. Stat Med. 2007;26(3735):52.
40. Smith PL. Splines as a useful and convenient statistical tool. The American Statistician.
1979;33(2):57-62.
41. Yu E, Ley SH, Manson JE, Willett W, Satija A, Hu FB, et al. Weight History and All-
Cause and Cause-Specific Mortality in Three Prospective Cohort StudiesWeight History and
Mortality in Three Prospective Cohort Studies. Annals of internal medicine. 2017;166(9):613-20.
42. Berrington de Gonzalez A, Hartge P, Cerhan JR, Flint AJ, Hannan L, MacInnis RJ, et al.
Body-mass index and mortality among 1.46 million white adults. New England Journal of
Medicine. 2010;363(23):2211-9.
43. Flegal KM, Ioannidis JPA. A meta-analysis but not a systematic review: an evaluation of
the Global BMI Mortality Collaboration. (1878-5921 (Electronic)).
44. Bhupathiraju SN, Di Angelantonio E, Danesh J, Hu FB. Commentary on “A meta-
analysis but not a systematic review: an evaluation of the Global BMI Mortality Collaboration”.
Journal of clinical epidemiology. 2017;88:30-2.
45. Flegal KM, Ioannidis JP. A meta-analysis of individual participant data constructed to
align with prior expert views: comments on Bhupathiraju et al. Journal of clinical epidemiology.
2017;88:33-6.
46. Tobias DK. Addressing Reverse Causation Bias in the Obesity Paradox Is Not “One Size
Fits All”. Diabetes care. 2017;40(8):1000-1.
47. Chiolero A, Faeh D, Paccaud F, Cornuz J. Consequences of smoking for body weight,
body fat distribution, and insulin resistance. The American journal of clinical nutrition.
2008;87(4):801-9.
Page 81 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
34
48. Molarius A, Seidell JC, Kuulasmaa K, Dobson AJ, Sans S. Smoking and relative body
weight: an international perspective from the WHO MONICA Project. Journal of epidemiology
and community health. 1997;51(3):252-60.
49. Canoy D, Wareham N, Luben R, Welch A, Bingham S, Day N, et al. Cigarette Smoking
and Fat Distribution in 21, 828 British Men and Women: A Population‐based Study. Obesity.
2005;13(8):1466-75.
50. Leite M, Nicolosi A. Lifestyle correlates of anthropometric estimates of body adiposity in
an Italian middle-aged and elderly population: a covariance analysis. International journal of
obesity. 2006;30(6):926-34.
Page 82 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
Confidential: For Review Only
35
Figure legends
Figure 1. The association of predicted body composition and body mass index* with all-
cause mortality in men. 1a. Fat mass and all-cause mortality. 2b. Lean body mass and all-
cause mortality. 3c. Body mass index and all-cause mortality.
Hazard ratios are indicated by solid lines and 95 % CIs by dashed lines. The reference point is
the lowest value for each fat mass and lean body mass and 25 kg/m2 for body mass index, with
knots placed at the 5th
, 35th
, 50th
, 65th
and 95th
percentiles of each fat mass and lean body mass
distribution. The models adjusted for the same cofounders in Table 2 plus mutually adjusted for
predicted fat mass and predicted lean body mass.
* Percentiles (0, 2.5, 5, 10, 25, 50, 75, 90, and 100%ile): 7, 13, 14, 15, 18, 21, 25, 29, and 66 kg
for fat mass, 24, 48, 49, 51, 53, 56, 59, 63, and 103 kg for lean body mass and 14.2, 21.5, 21.2,
22.0, 23.4, 25.1, 27.0, 31.0 and 62.0 kg/m2.
Page 83 of 83
https://mc.manuscriptcentral.com/bmj
BMJ
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960