15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ORIGINAL ARTICLE

Association of Appendicular Lean Mass, and

Subcutaneous and Visceral Adipose Tissue With
Mortality in Older Brazilians: The São Paulo Ageing
& Health Study
Felipe M de Santana,1 Diogo S Domiciano,1 Michel A Gonçalves,1 Luana G Machado,1 Camille P Figueiredo,1
Jaqueline B Lopes,1 Valéria F Caparbo,1 Lilliam Takayama,1 Paulo R Menezes,2 and Rosa MR Pereira1*
1
Bone Metabolism Laboratory, Rheumatology Division, Faculdade de Medicina FMUSP da Universidade de Sao Paulo, Sao Paulo, Brazil
2
Department of Preventive Medicine, Faculdade de Medicina FMUSP da Universidade de Sao Paulo, Sao Paulo, Brazil

ABSTRACT
Body composition changes as a result of ageing may impact the survival of older adults. However, its influence on mortality risk is
uncertain. Currently, the best method for body composition analysis in clinical practice is DXA. Nonetheless, the few studies on
body composition by DXA and mortality risk in the elderly have some limitations. We investigated the association between body
composition by DXA and mortality in a cohort of elderly subjects. Eight hundred thirty‐nine community‐dwelling subjects (516
women, 323 men) ≥ 65 years of age were assessed by a questionnaire, clinical data, laboratory exams, and body composition by
DXA at baseline. Total fat and its components (eg, visceral adipose tissue [VAT]) were estimated. Appendicular lean mass (ALM)
adjusted for fat and ALM divided by height² was used to ascertain the presence of low muscle mass (LMM). Mortality was recorded
during follow‐up. Multivariate logistic regression was used to compute ORs for all‐cause and cardiovascular mortality. Over a mean
follow‐up of 4.06 ± 1.07 years, there were 132 (15.7%) deaths. In men, after adjustment for relevant variables, the presence of LMM
(OR, 11.36, 95% CI, 2.21 to 58.37, P = 0.004) and VAT (OR, 1.99, 95% CI, 1.38 to 2.87, P < 0.001, for each 100‐g increase) significantly
increased all‐cause mortality risk, whereas total fat, measured by the fat mass index, was associated with decreased mortality risk
(OR, 0.48, 95% CI, 0.33 to 0.71, P < 0.001). Similar results were observed for cardiovascular mortality. In women, only LMM was a
predictor of all‐cause (OR, 62.88, 95% CI, 22.59 to 175.0, P < 0.001) and cardiovascular death (OR, 74.54, 95% CI, 9.72 to 571.46,
P < 0.001). LMM ascertained by ALM adjusted for fat and fat mass by itself are associated with all‐cause and cardiovascular
mortality risk in the elderly. Visceral and subcutaneous fat have opposite roles on mortality risk in elderly men. Thus, DXA is a
promising tool to estimate risk of mortality in the elderly. © 2019 American Society for Bone and Mineral Research.

KEY WORDS: AGEING; DXA; MORTALITY; SARCOPENIA; VISCERAL ADIPOSE TISSUE

Introduction in clinical practice using simple anthropometric measures, such

as total weight and BMI.(1–4)

A geing is a complex process associated with multiple

changes—those related to body composition are remark-
able. The effects of these changes and how they influence long‐
Thus, many anthropometric methods to estimate body composi-
tion are inadequate for application in clinical practice. For instance,
midarm circumference and other anthropometric correlates for
term survival in older adults are still unclear. estimating muscle mass are no longer recommended.(5) Similarly,
In a three‐compartmental model, body composition can be the skinfold test, previously used as a measure of total adiposity, is
roughly divided into bone mass, fat mass, and lean mass. now widely disregarded.(6) There is still a role for some
Moreover, appendicular lean mass (ALM) corresponds approxi- anthropometric measures in clinical practice, such as waist
mately to the appendicular skeletal muscle mass (ASM); circumference, which has proven to be a good predictor of central
although these terms are conceptually distinct, they have adiposity and cardiovascular risk.(7) In many circumstances,
been used interchangeably in research and clinical practice. however, where a more accurate prediction of body composition
As we age, there is a progressive increase in body fat with a is needed, more sophisticated methods have emerged.
concomitant reduction of muscle mass and water content. The most readily available of these in both clinical and
Many of these body composition changes are difficult to detect research practice are bioelectrical impedance analysis (BIA),

Received in original form October 16, 2018; revised form February 5, 2019; accepted February 23, 2019. Accepted manuscript online May 29, 2019.
Address correspondence to: Rosa MR Pereira, Av Dr Arnaldo, 455 – Reumatologia, 3º andar, sala 3193, Cerqueira Cesar, Sao Paulo, Brazil 01246‐903. E‐mail:
[email protected]
Journal of Bone and Mineral Research, Vol. 34, No. 7, July 2019, pp. 1264–1274
DOI: 10.1002/jbmr.3710
© 2019 American Society for Bone and Mineral Research

◼ 1264
 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
QCT, and DXA. Although BIA is the cheapest and easiest test to from their homes to the research center via a shuttle service
perform, it is also the least accurate. QCT, in turn, is a high‐ specifically hired for the study. Of the 1025 participants in the
precision method seldom used for this purpose in clinical first study, 707 (69% of the original cohort) agreed to take part
practice because of its costs and radiation burden. An in the prospective assessment. Because of varying dates of
assessment of whole‐body composition by DXA has been completion of the baseline survey, the period between baseline
increasingly used both in clinical and research practice because and the second appraisal varied among the participants (range,
it provides an optimal combination of accuracy, cost, and 0.2 to 6.5 years, average 4.06 years). The coverage period for
safety. It is acknowledged by many practitioners as the most mortality data collection was from the date of the baseline
accurate method for estimating body composition in clinical examination through December 31, 2012. At the end of that
practice.(6,8,9) period, 132 (12.9%) individuals had died, and 168 (16.4%) could
Body composition and its association with mortality in older not be contacted because they were living outside of São Paulo
adults have been extensively studied recently. Less‐accurate (lost to follow‐up). Those who could not be contacted were not
methods have mostly been investigated, such as anthropo- known to be alive at the time that analysis was censored; thus,
metry and BIA.(10–23) DXA as a method for estimating mortality they were excluded from the study. Of the 725 subjects who
in older adults has seldom been studied.(13–16) Moreover, the agreed to participate in the second evaluation, 18 (1.8%) were
few studies that assess fat mass and its association with excluded because of a lack of some baseline laboratory or DXA
mortality have not analyzed its subcomponents, such as data. Figure 1 depicts a flow chart of the study.
visceral adipose tissue (VAT). This study was approved by the Local Ethics in Research
Furthermore, a precise definition of low muscle mass (LMM), Committee of the São Paulo University School of Medicine
as well as a consensual approach for diagnosing sarcopenia, are (#0825/09). All participants gave written informed consent.
still lacking. The attempts to associate muscle mass with
mortality have had controversial results.(12–16) Among the many
different definitions,(24) that of Newman and colleagues has not Body composition assessment
yet been prospectively studied.(25) They posit that ALM should
be adjusted for fat mass before ascertaining the presence of At baseline, body composition was determined by DXA using
LMM. This definition has a higher sensitivity in obese Hologic QDR 4500 A densitometry equipment (Discovery
individuals and has already been validated in the Brazilian model; Hologic Inc., Bedford, MA, USA).
elderly population.(26,27) All DXA measurements were performed by the same
The present study was designed to verify the association experienced technologist. The precision error was established
between body composition by DXA and mortality in a cohort of according to the International Society for Clinical Densitome-
Brazilian community‐dwelling older adults. The assessment of try's standards.(31) The least significant change (LSC) was
body composition included total fat mass, VAT, ALM, and 0.033 g/cm² for lumbar spine, 0.047 g/cm² for femoral neck,
different definitions of low muscle mass. and 0.039 g/cm² for total hip.
The ALM was measured by DXA software as the sum of the
lean mass of the upper and lower limbs, assuming that all
nonfat and nonbone tissue is skeletal muscle. Three parameters
Materials and Methods of muscle mass were assessed in the population studied: the
appendicular lean mass index (ALMI) as a continuous variable,
Subjects the presence of LMM by the methods of residuals as described
From June 2005 to July 2007, 1025 older subjects were by Newman et al [LMM (Newman)],(24,25) and the presence of
enrolled in the São Paulo Ageing & Health (SPAH) Study, a LMM according to recommended cut‐offs of ALMI as recently
prevalence survey of osteoporotic fractures. Details about published by the European Working Group on Sarcopenia in
the methodology have been described elsewhere.(28) Briefly,
the SPAH recruited residents ≥ 65 years old from the Butantã
district, an urban district in the city of São Paulo (located in
the south of Brazil) comprised of 377,576 inhabitants at the
time of the study. To minimize selection bias, 66 census
sectors (the smallest administrative areas) from this district
were randomly selected. All older adults living in these
sectors were identified through census records and were
visited by members of our study team. Only those who were
well‐functioning were invited for enrollment. Well‐func-
tioning was defined as being able to independently visit
the research center and to perform basic activities of daily
living. All enrolled subjects were submitted to an inter-
viewer‐administered lifestyle questionnaire, laboratory tests,
whole‐body composition DXA test, and spinal radiographs
performed on the same day. Our study sample was similar in
age, gender, and social class characteristics to the entire Fig. 1. SPAH study flow chart. This flowchart briefly summarizes he
Brazilian elderly population.(29) enrolled individuals. Those lost to follow‐up, those excluded from
Between 2010 and 2012, all subjects were invited to analysis, and those who were included for analysis are highlighted.
participate in a second evaluation.(30) Individuals were recruited SPAH = São Paulo Aging and Health
by telephone calls and door‐to‐door visits. They were brought

Journal of Bone and Mineral Research BODY COMPOSITION & MORTALITY IN OLDER BRAZILIANS 1265 ◼
 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Older People 2 (EWGSOP2) [LMM (EWGSOP2)].(5) ALMI is pressure measurement at examination (systolic ≥ 140 mm Hg
calculated as ALM (kg) divided by height squared (m²).(32) The or diastolic ≥ 90 mm Hg), or a history of recent blood pressure
method described by Newman and colleagues adjusts the medication use. One‐to‐three blood pressure measurements
individual ALM for fat mass.(24,25) According to this method, were performed during data collection. Diabetes mellitus was
expected ALM (expALM) must be calculated from both height defined as self‐reported physician‐diagnosed diabetes mellitus
(m²) and total fat mass (kg) in a linear regression model. This (except gestational diabetes only) or a fasting (12 hours) blood
way we can ascertain whether one's actual ALM, estimated by glucose of 126 mg/dL or greater. Hyperlipidemia was defined
DXA, is higher or lower than the expALM. The difference as self‐reported physician‐diagnosed high cholesterol, non‐
between the actual ALM (actALM) and the expALM is known as high‐density lipoprotein (HDL) cholesterol of 160 mg/dL or
the residual. Thus, the higher the residual, the higher the lean greater, triglycerides of 200 mg/dL or greater, or current use of
mass, and the lower the residual, the lower the lean mass. This lipid‐lowering medication.
definition and its linear regression model were previously The height, weight, and BMI of each participant were measured
validated in two Brazilian populations of older adults.(26,27) The using standard protocols. BMI was calculated by dividing the
equations extracted from this model are as follows(26,27): participant's weight (kg) by their height squared (m2).

Men: expALM (kg) = − 14.51 + 17.27 × height (m) Baseline laboratory assessment
+ 0.20 × fat mass (kg) Blood samples were collected under fasting conditions
Women: expALM (kg) = − 20.67 (between 8:00 AM and 10:00 AM) and stored at − 70 °C until
analysis.
+ 22.48 × height (m)
The concentrations of serum calcium (adjusted for the
+ 0.177 × fat mass (kg) albumin concentration), phosphorus, alkaline phosphatase,
creatinine, total cholesterol, HDL, triglycerides, albumin, and
As mean values of lean mass and fat mass from younger glucose were determined using standard automated laboratory
individuals were not known as a reference for normality by the methods.
time Newman et al(25) hypothesized this definition, the 20th The estimated glomerular filtration rate (eGFR) was calcu-
percentile of the residual distribution was used as the cut‐off lated using the Cockroft‐Gault equation and expressed
point of LMM. This corresponded to − 1.45 in women and categorically according to the U.S. National Kidney Founda-
− 2.06 in men in our population.(26,27) Thus, if these residuals tion's modified classification for chronic kidney disease (CKD):
are lower than − 1.45 and − 2.06, women and men may be stages 1 to 2, eGFR > 60 ml/min/1.73 m2 (reference group);
considered to have LMM, respectively. stage 3a, eGFR = 45 to 59 ml/min/1.73 m2; stage 3b, eGFR = 30
The fat mass was measured by DXA as total fat mass (kg), fat to 44 ml/min/ 1.73 m2; stage 4, eGFR = 15 to 29 ml/min/
percentage (%), VAT, and fat mass index (FMI). The latter 1.73 m2; and stage 5, eGFR < 15 ml/min/1.73 m2.
consisted of total fat mass (kg) divided by height squared (m²). The serum concentration of 25‐hydroxyvitamin D [25(OH)D]
VAT was assessed by whole‐body DXA scan using the APEX was measured using the radioimmunoassay technique (Dia-
software (Version 4.0; APEX Software, San Antonio, TX, USA). Sorin, Stillwater, MN, USA) with a lower detection limit of 5 ng/
ROIs are automatically established by the software and VAT is mL. iPTH serum concentrations were measured using immu-
estimated from the android region. The inferior line of this noradiometric assay (ELSAPTH; CIS Bio International, Gif‐sur‐
region is drawn just at the superior edge of the iliac crest, Yvette, France) with a reference interval of 11 to 65 pg/mL
whereas the superior line is at 20% of the distance between (reference range of the kit tests).
the iliac crest and the inferior edge of the chin. The lateral
android subcutaneous fat is automatically set by the DXA
software; it is used to estimate the anterior and posterior Baseline radiographic assessment
android subcutaneous fat. Finally, VAT results from the At baseline, lateral radiographs of the thoracic and lumbar
subtraction of the subcutaneous android fat from the total spine were obtained using a 40‐inch tube‐to‐film distance
android fat; it was measured as mass (g), area (cm²), and centered at T7 and L2. To identify vertebral fractures, all X‐ray
volume (cm³ ).(33,34) images were analyzed independently by two rheumatologists
who were experienced in vertebral fracture assessment. A
consensus was reached between the readers for any diver-
Baseline clinical assessment
gences in interpretations. The readers evaluated each T4 to L4
At baseline, all of the participants underwent a standardized vertebrae image to determine whether it contained a fracture,
interviewer‐administered questionnaire that ascertained life- using Genant's semiquantitative approach.(35) The reliability
style and health behaviors, including age, dairy product between readers was analyzed using a random subsample of
consumption, smoking status, alcohol consumption, physical 60 X‐ray images, as described. Interobserver agreement was
activity, previous nonvertebral fractures, falls during the 96%, and the kappa coefficient was 0.83.
previous year, comorbidities (including any cardiovascular
event: myocardial infarction, unstable angina, or stroke), and
Outcomes
current medication use. The individuals were asked to bring all
medications to the baseline clinic visit. Details about the At the end of the field data collection, mortality data of the
ascertainment of these baseline characteristics have been original SPAH cohort were collected from the death certificates
provided.(28) available from the publicly available databases of the Programa
Hypertension was determined based on self‐reported de Aprimoramento das Informações de Mortalidade no
physician‐diagnosed hypertension, elevated average blood Município de São Paulo (PRO‐AIM, Improvement Program of

◼ 1266 DE SANTANA ET AL. Journal of Bone and Mineral Research

 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Information on Mortality in São Paulo) Foundation, the organ The univariable analysis for all‐cause mortality in the male
responsible for vital statistics, operating under the auspices of population showed that the deceased participants were older,
the São Paulo State Secretary of Economics and Planning. The less physically active, and more often diabetic. They also had
underlying cause of death was coded using the 10th revision of more recurrent falls; previous cardiovascular events; a higher
the International Classification of Diseases (ICD‐10).(36) Cardio- alcohol intake; higher levels of serum phosphorus and lower
vascular (CVD) mortality was defined as ICD‐10 codes I00 to I99 levels of serum calcium, albumin, and 25(OH)D. The DXA BMD
and non‐CVD mortality as all other causes of death. analysis yielded significantly lower values at the proximal
The outcome for the cohort was determined by examining femur, but no differences in the lumbar spine. Regarding the
the relationship between body composition parameters, body composition analysis, deceased men had a higher
including muscle mass and fat mass, with incident all‐cause prevalence of LMM(25), lower values of total fat mass (FMI),
and cardiovascular mortality during the follow‐up period. and higher values of VAT. We found no significant difference in
ALMI between male groups. We also found no significant
difference in LMM (EWGSOP2).
Statistical analysis
The univariable analysis for cardiovascular mortality in men
The prevalence of any potential predictors of mortality resulted in similar statistical associations.
(clinical, laboratory, and body composition variables from In the multivariable regression model including only male
the baseline assessment: 2005 to 2007) was reported participants, as shown in Table 3, after adjusting for age,
according to a descriptive analysis for dead and living physical activity, falls, alcohol intake, diabetes mellitus,
individuals, separated by sex. To find the potentially cardiovascular event, 25(OH)D, calcium, phosphorus, albumin,
significant differences in the baseline characteristics between and BMD, both LMM(25) and VAT independently increased,
dead and living individuals at the end of the study, the whereas FMI independently decreased all‐cause and cardiovas-
following tests were used for the univariable analysis: chi‐ cular mortality. The presence of LMM(25) increased approxi-
square, Fisher, likelihood ratio, Student t test, and Mann‐ mately 11 times (OR, 11.36, 95% CI, 2.21 to 58.37, P = 0.004) and
Whitney. Given the possible small magnitude of association 14 times (OR, 14.84, 95% CI, 5.15 to 47.72, P < 0.001) all‐cause
for some potential risk factors, which would not reach and cardiovascular mortality, respectively. VAT increased about
statistical significance with a P value < 0.05 because of its two times (OR, 1.99, 95% CI, 1.38 to 2.87, P < 0.001, for each
low prevalence in the study population, the level of 100‐g increase) and 1.5 times (OR, 1.66, 95% CI, 1.31 to 2.10,
significance chosen for the univariable analysis was a P < 0.001) all‐cause and cardiovascular mortality, respectively,
P < 0.10 (two‐tailed). All the risk factors significantly asso- and this association persisted whether we analyzed VAT by
ciated with mortality in the univariable analysis were selected mass, area, or volume. Finally, FMI decreased by half both all‐
for the multivariable regression. Backward stepwise logistic cause (OR, 0.48, 95% CI, 0.33 to 0.71, P < 0.001) and
regression models were built to find which risk factors were cardiovascular (OR, 0.57, 95% CI, 0.43 to 0.76, P < 0.001)
independently associated with mortality, for men and women mortality.
separately. Where variables estimated the same parameter Because of the strong positive association found between
and a high degree of collinearity was to be expected, such as mortality and the presence of LMM, we plotted a ROC curve to
VAT estimated by mass, by volume and area, or muscle mass find the overall performance of the residuals for predicting all‐
estimated by different definitions (ALMI, LMM by Newman cause mortality. This way, we could also ascertain whether
et al(25) and LMM by the EWGSOP2), we decided to run diverse there was a better cut‐off, based on its sensitivity and
logistic regression models, including each of these variables at specificity, than the arbitrarily set 20th percentile. The analysis
a time. Multicollinearity of covariates included in each model of the attained ROC curve resulted in an area under the curve
was tested by measuring the variation inflation factor. The (AUC) of 0.827. Using Youden's index (Sensibility + Specificity –
measure of effect size for each variable was presented as ORs 1), we found − 1.61 to be the best cut‐off residual for predicting
along with its respective 95% CI. all‐cause mortality, considering the balance between sensitivity
A receiver‐operating characteristic (ROC) curve analysis was and specificity. This value differs from the one corresponding to
also performed to test for the overall ability of ALM residuals to the 20th percentile (− 2.06). Figure 2 shows the ROC curve for
predict all‐cause mortality, separated by sex, and Youden's MEN and its analysis results.
index calculation was used to find the residual whose balance The results of the univariable analysis for all‐cause mortality in
between sensitivity and specificity was highest. women were quite similar to that of the male population. When
All analyses were performed using SPSS software version compared with their surviving peers, the deceased women were
20.0 for Windows (IBM Corp, Armonk, NY, USA). older, less physically active, and more often diabetic and
hypertensive. They had a higher prevalence of previous
cardiovascular events; higher levels of both serum phosphorus
Results and iPTH; and lower BMI, eGFR, proximal femur BMD, and 25(OH)
D. Statistically significant differences in body composition were
During a mean follow‐up of 4.06 ± 1.07 years, there were 132 also observed. The muscle mass analysis was similar to that of the
(15.7%) deaths among the 839 participants, of which 57 (43.2%) male population; therefore, the deceased women had a higher
were related to CVD. The remaining 75 noncardiovascular prevalence of LMM.(25) Testing differences in ALMI, however,
deaths could not be further stratified according to other causes yielded positive results; we found a significant difference between
(eg, cancer, infection) because of the small number of dead and living older women. We also found that LMM
observations for each cause. (EWGSOP2) was significantly more common among deceased
The baseline differences between the individuals who women than it was among their living peers. Concerning the fat
survived the follow‐up and those who died are shown in mass analysis, there was no significant difference in FMI; however,
Tables 1,2, according to sex. VAT was lower in women who did not survive the follow‐up.

Journal of Bone and Mineral Research BODY COMPOSITION & MORTALITY IN OLDER BRAZILIANS 1267 ◼
 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1. Baseline Characteristics (2005–2007) of the Male Participants According to Survivala

Baseline Characteristics Death during follow‐up (N = 65) Alive at the end of study (N = 258) P
Clinical Variables
Age (years) 75.66 (6.43) 72.31 (4.65) <0.001
Caucasian, n (%) 48 (73.8) 166 (64.3) 0.147
BMI (kg/m2) 26.59 (4.31) 26.73 (3.78) 0.797
Low physical activity score, n (%) 18 (27.69) 24 (9.30) <0.001
Recurrent falls, n (%) 14 (21.53) 18 (6.97) 0.024
Previous non‐vertebral fractures, n (%) 6 (9.23) 10 (3.87) 0.103
Prevalent vertebral fracture, n (%) 24 (36.92) 81 (31.39) 0.491
Current smoking, n (%) 12 (18.46) 32 (12.40) 0.270
Alcohol intake ≥ 3U/d, n (%) 28 (43.07) 83 (32.17) 0.098
Diabetes mellitus, n (%) 20 (30.76) 39 (15.11) 0.003
Hypertension, n (%) 39 (60) 136 (52.71) 0.210
Hyperlipidemia, n (%) 10 (15.4) 29 (6.5) 0.359
Previous cardiovascular event, n (%) 20 (30.76) 38 (14.72) 0.002
Dairy product intake, mg/day 403.19 (225.09) 413.96 (304.09) 0.794
Laboratory Variables
eGFR, mL/min per 1,73 m2 55.00 (27.51) 59.98 (17.01) 0.202
Serum phosphorus, mg/dl 3.26 (0.66) 3.08 (0.49) 0.042
Serum calcium, mg/dl 9.21 (0.50) 9.36 (0.44) 0.033
Albumin, g/dl 4.26 (0.37) 4.41 (0.26) 0.005
iPTH, pg/ml 55.98 (101.29) 36.59 (16.83) 0.132
25‐OH Vitamin D, ng/ml 19.29 (10.26) 22.16 (9.34) 0.034
Bone Mass
Lumbar spine BMD, g/cm2 0.960 (0.204) 1.000 (0.185) 0.131
Femoral neck BMD, g/cm2 0.710 (0.138) 0.769 (0.139) 0.003
Total hip BMD, g/cm2 0.866 (0.165) 0.943 (0.135) <0.001
Lean Mass
ALMI, kg/m2 8.09 (1.31) 7.83 (1.24) 0.401
LMM (Newman), n (%) 38 (58.46) 48 (18.60) <0.001
LMM (EWGSOP2), n (%) 14 (23.70) 72 (29.30) 0.493
Fat Mass
Total Fat Mass, kg 47.58 (12.09) 22.14 (8.67) <0.001
Total Fat Mass, % 24.96 (5.30) 31.97 (8.67) <0.001
FMI, kg/m2 6.74 (2.22) 9.36 (3.99) <0.001
Visceral Fat Mass
VAT mass, g 723.8 (316.32) 644.77 (274.89) 0.054
VAT volume, cm3 782.37 (341.98) 696.26 (298.09) 0.052
VAT area, cm2 150.05 (65.57) 137.57 (57.18) 0.053
a
All causes of death included (all‐cause mortality).
BMI = Body Mass Index; eGFR = estimated Glomerular Filtration Rate; iPTH = intact Parathyroid hormone; BMD = Bone Mineral Density; ALMI = Appendicular
Lean Mass Index; LMM = Low muscle mass according to Newman et al(25); LMM (EWGSOP2) = Low muscle mass according to EWSGOP2; FMI = Fat Mass Index;
VAT = Visceral Adipose tissue.

Similar results were observed in the univariable analysis for sensitivity of 92.2% and a specificity of 86.2%. Figure 3 shows
cardiovascular mortality. the ROC analysis for women.
The multivariable regression model for the female popula- As previously described in the statistical analysis section, we
tion including LMM(25) is presented in Table 4. Although many decided to run sequential logistic regression models replacing
differences in body composition were shown between dead LMM(25) by ALMI and then by LMM (EWGSOP2). The multi-
and living older women at the univariable analysis, after variable analysis including ALMI revealed it to be indepen-
adjusting for the relevant variables, the only body composition dently associated with all‐cause mortality (OR, 0.50, 95% CI,
parameter in this model that persisted significantly associated 0.37 to 0.67, for each 1 kg/m2 increase, P < 0.001) and
with all‐cause and cardiovascular mortality was the presence of cardiovascular mortality (OR, 0.54, 95% CI, 0.35 to 0.84, for
LMM (Newman et al(25); OR, 62.88, 95% CI, 22.59 to 175.0, P each 1 kg/m2 increase, P = 0.007). The last multivariable model
< 0.001 and OR, 74.54, 95% CI, 9.72 to 571.46, P < 0.001, including LMM (EWGSOP2) instead of LMM(25) also retained it
respectively). Thus, we decided to also carry out a ROC analysis as an independent predictor of all‐cause mortality (OR, 4.21,
to better characterize the overall performance of the residuals 95% CI, 1.80 to 9.87, P< 0.001), but not as a predictor of
for predicting mortality. The AUC calculated was 0.945, even cardiovascular mortality (OR, 2.25, 95% CI, 0.69 to 7.30,
higher than that found for the male population. The best cut‐ P = 0.17). The models including ALMI and LMM (EWGSOP2)
off residual using Youden's index was − 1.32, which yielded a are not shown in the tables.

◼ 1268 DE SANTANA ET AL. Journal of Bone and Mineral Research

 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 2. Baseline characteristics (2005–2007) of the Female Participants According to Survivala

Baseline characteristics Death during follow‐up (N = 67) Alive at the end of study (N = 449) P
Clinical variables
Age (years) 76.64 (7.37) 72.88 (4.78) <0.001
Caucasian, n (%) 46 (68.65) 283 (63.02) 0.371
BMI (kg/m2) 26.45 (5.55) 29.06 (5.29) <0.001
Low physical activity score, n (%) 21 (31.34) 21 (4.67) <0.001
Recurrent falls, n (%) 13 (19.40) 73 (16.25) 0.747
Previous non‐vertebral fractures, n (%) 12 (17.91) 65 (14.47) 0.462
Prevalent vertebral fracture, n (%) 24 (35.82) 124 (27.61) 0.222
Current smoking, n (%) 5 (7.46) 48 (10.69) 0.719
Alcohol intake ≥ 3U/d, n (%) 5 (7.46) 24 (5.34) 0.566
Diabetes mellitus, n (%) 25 (37.31) 90 (20.04) 0.002
Hypertension, n (%) 53 (79.10) 305 (67.92) 0.068
Hyperlipidemia, n (%) 8 (11.9) 68 (15.1) 0.490
Previous cardiovascular event, n (%) 18 (26.86) 47 (10.46) <0.001
Dairy product intake, mg/day 485.52 (317.02) 474.69 (298.19) 0.785
Laboratory variables
eGFR, mL/min per 1,73 m2 50.96 (21.60) 59.47 (18.99) 0.002
Serum phosphorus, mg/dl 3.58 (0.48) 3.50 (0.47) 0.219
Serum calcium, mg/dl 9.55 (0.56) 9.44 (0.48) 0.089
Albumin, g/dl 4.37 (0.26) 4.39 (0.26) 0.608
iPTH, pg/ml 55 (36.43) 40.49 (17.28) 0.002
25‐OH Vitamin D, ng/ml 16.62 (10.17) 18.87 (9.14) 0.064
Bone mass
Lumbar spine BMD, g/cm2 0.811 (0.178) 0.827 (0.170) 0.500
Femoral neck BMD, g/cm2 0.608 (0.140) 0.672 (0.132) <0.001
Total hip BMD, g/cm2 0.717 (0.142) 0.794 (0.132) <0.001
Lean mass
ALMI, kg/m2 6.76 (1.18) 7.71 (1.18) <0.001
LMM (Newman), n (%) 58 (86.56) 53 (11.80) <0.001
LMM (EWGSOP2), n (%) 20 (31.2) 31 (7.1) <0.001
Fat mass
Total Fat Mass, kg 36.21 (8.05) 21.64 (7.74) <0.001
Total Fat Mass, % 33.85 (7.54) 32.00 (8.21) 0.084
FMI, kg/m2 9.30 (3.67) 9.27 (3.89) 0.953
Visceral fat mass
VAT mass, g 529.87 (330.61) 663.62 (270.34) 0.003
VAT volume, cm3 572.82 (357.43) 713.86 (293.69) 0.004
VAT area, cm2 109.88 (68.55) 137.37 (56.00) 0.003
a
All causes of death included (all‐cause mortality).
BMI = Body Mass Index; eGFR = estimated Glomerular Filtration Rate; iPTH = intact Parathyroid hormone; BMD = Bone Mineral Density; ALMI = Appendicular
Lean Mass Index; LMM (Newman) = Low muscle mass according to Newman et al(25); LMM (EWGSOP2) = Low muscle mass according to EWSGOP2; FMI = Fat
Mass Index; VAT = Visceral Adipose tissue.

Therefore, for the male population, two fat mass estimates mortality in older men and to find that Newman and
(VAT and FMI) and one muscle mass estimate (LMM according colleagues’ definition on LMM accurately predicts mortality in
to Newman and colleagues’ definition) were independently older men and women. We also demonstrated that total body
associated with all‐cause and cardiovascular mortality. This was fat seems to play a protective role in the male population; this
distinct from the results of the female population where only is consistent with the few previous studies on the matter. As for
muscle mass estimates were independently associated with women, LMM, particularly when adjusted for fat, seems to be
mortality. These estimates in women included LMM,(25) LMM the prevailing body composition change associated with
(EWGSOP2), and ALMI. higher mortality in the long‐term. Thus, our study showed
that the risks associated with changes in body composition in
Discussion the elderly are different according to sex, and also that the role
Our study was the first to demonstrate that body composition of total fat in older adults may be a protective one, contrary to
parameters assessed by DXA are independently associated with what has been established in the literature for younger
all‐cause and cardiovascular mortality in Brazilian community‐ populations. These results persisted even after adjusting for
dwelling healthy older adults. Furthermore, it was also the first known cardiovascular risk factors, including mineral metabo-
to find VAT, assessed by DXA, as an important predictor of lism parameters such as BMD and PTH, which have been

Journal of Bone and Mineral Research BODY COMPOSITION & MORTALITY IN OLDER BRAZILIANS 1269 ◼
 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 3. Independent Predictors of Mortality for the Male The previously published studies on body composition and
Population After Multivariable Analysisa mortality risk have some limitations. There is considerable
heterogeneity in both their methodologies and their popula-
All‐cause mortality tions. In general terms, there seems to be an association
Baseline characteristics OR CI, 95% P between body composition and mortality, at least for some
Model 1 populations. There are definite published data demonstrating
LMM (Newman) 14.83 5.15 – 42.72 <0.001 strong association of fat mass and muscle mass with mortality
FMI (for each ⇑ 1 kg/m2) 0.56 0.42 – 0.75 <0.001 in patients with specific diseases such as cirrhosis, chronic
VAT mass (for each ⇑ 100 g) 1.65 1.31 – 2.09 <0.001 kidney disease, and chronic heart failure.(20–23) The literature on
Model 2 an older healthy population, however, is far from definite; this is
LMM (Newman) 11.26 2.17 – 58.48 0.004 in part based on the limited number of studies. Moreover, the
FMI (for each ⇑ 1 kg/m2) 0.46 0.30 – 0.70 <0.001 heterogeneous populations of some studies, including young
VAT area (for each ⇑ 100 cm2) 1.40 1.18 – 1.68 <0.001 and elderly adults in the same analysis, hamper the general-
Model 3 ization of the results.(16,18,19) It is widely known that the body
LMM (Newman) 12.07 2.08 – 70.11 0.006 composition correlates differently according to age. For
FMI (for each ⇑ 1 kg/m2) 0.46 0.30 – 0.70 <0.001 instance, BMI is strongly associated with mortality in young
VAT volume (for each ⇑ 1.98 1.39 – 2.84 <0.001 people, but not in older people. In the latter, studies have come
100 cm3) to diverse conclusions, including positive, negative, and even
Low level of physical activity 10.6 1.03 – 109.04 0.047 U‐shaped associations between BMI and mortality.(4) There is
also no standard method used to estimate body composition
Cardiovascular mortality as different studies have used different methods. Most studies
Baseline characteristics OR CI, 95% P measured fat mass and lean mass by anthropometry and
BIA.(10,11) Although easy to perform and inexpensive, these
Model 1
methods are imprecise and incapable of estimating subcom-
LMM (Newman) 14.83 5.15 – 42.72 <0.001
ponents (eg, VAT). On the other hand, QCT and DXA are highly
FMI (for each ⇑ 1 kg/m2) 0.56 0.42 – 0.75 <0.001
accurate tools for the evaluation of body composition.(6)
VAT mass (for each ⇑ 100 g) 1.65 1.31 – 2.09 <0.001
Two prospective studies, designed with QCT, have shown
Diabetes mellitus 3.75 1.33 – 10.55 0.012
that VAT is a risk factor for all‐cause mortality.(18,19) However,
Model 2
both studies included younger and older adults in the same
LMM (Newman) 14.72 5.12 – 42.34 <0.001
analysis. Many studies estimating body composition by DXA
FMI (for each ⇑ 1 kg/m2) 0.56 0.42 – 0.75 <0.001
have been published in the last two decades. Among those
VAT area (for each ⇑ 100 cm2) 1.40 1.18 – 1.68 <0.001
that included older healthy adults, were prospectively designed
Diabetes mellitus 3.81 1.35 – 10.75 0.011 and had mortality as the primary outcome, two of them found
Model 3 an inverse association between total fat mass and mor-
LMM (Newman) 14.73 5.12 – 42.37 <0.001 tality.(14,15) As mentioned above, these findings are consistent
FMI (for each ⇑ 1 kg/m2) 0.56 0.42 – 0.75 <0.001 with our results. A third similar study, however, did not find any
VAT volume (for each ⇑ 1.60 1.29 – 1.98 <0.001 association between total fat mass and mortality.(13)
100 cm3) These previous conflicting results could be explained by
Diabetes mellitus 3.80 1.35 – 10.74 0.011 differences in the methodology used for the fat mass
a
Body composition variables and covariates that remained significant in estimation. For instance, a percentage of fat mass (%) was
the final model. used in the positive studies, whereas absolute fat mass (g) was
Covariates included in the model 1, 2 and 3 for all‐cause mortality are as used in the negative study. Moreover, adipose tissue can be
follows: age, low level of physical activity, recurrent falls, high alcohol intake, divided into different subtypes according to its location, such
diabetes mellitus, previous cardiovascular event, serum phosphorus, as VAT and subcutaneous adipose tissue. Each of them differs
calcium, albumin, 25‐OH Vitamin D and total hip BMD T‐score. not only in spatial distribution, but also in their metabolic
Covariates included in the model 1, 2 and 3 for cardiovascular mortality profiles. Whereas VAT appears to play a deleterious role, being
are as follows: age, low level of physical activity, recurrent falls, diabetes associated with metabolic syndrome and the production of
mellitus, hypertension, previous cardiovascular event, serum phosphorus, pro‐inflammatory cytokines, the subcutaneous adipose tissue
calcium, albumin and total hip BMD T‐score. behaves in a diametrically opposite way.(39) To the best of our
OR = odds ratio; CI = confidence interval; LMM (Newman) = Low muscle mass knowledge, our study is the first to demonstrate the association
according to Newman et al(25); FMI = Fat Mass Index; VAT = Visceral Adipose
of VAT, assessed by DXA, with mortality in older adults.
tissue.
The effect of muscle mass on mortality in the elderly
population is even less conclusive. Rolland and colleagues
previously demonstrated to be predictors of death in this same followed 4575 elderly women for 17 years and did not find an
elderly population.(37) independent association between muscle mass and mor-
SPAH is a cohort study that consisted of a sample of tality.(14) The muscle mass was estimated then by total lean
community‐dwelling older adults. The demographic character- mass (total lean mass / height²), and not by appendicular lean
istics of this population were similar to those of the whole mass, which is a much better predictor of muscle mass. Another
Brazilian population over 65 years of age, according to the important study was published by Bunout et al(13) They
census of the Brazilian Institute of Geography and Statistics followed 1413 elderly men and women and found muscle
(IBGE).(29) Subject loss to follow‐up was reasonably low mass to be a protective factor for mortality (HR 0.85, 95% CI,
(18%).(38) Data on potential risk factors for mortality included 0.74 to 0.98). The higher the appendicular lean mass, the lower
clinical, laboratory, and DXA variables.(37) the long‐term mortality.

◼ 1270 DE SANTANA ET AL. Journal of Bone and Mineral Research

 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 2. ROC curve for calculated residuals from SPAH study male participants, according to the definition of LMM adjusted for fat mass by Newman and
colleagues.(25) In male participants, the residual that yielded the best performance of sensitivity and specificity was − 1.61, according to Youden's index
formula. ROC = receiver operating characteristic; SPAH = São Paulo Ageing and Health; LMM = low muscle mass; expALM = expected appendicular lean
mass; actALM = actual appendicular lean mass; AUC = area under the curve [Color figure can be viewed at wileyonlinelibrary.com].

The dynamic changes in muscle mass also seem to have an the most widely used definitions of LMM still date back to
influence on mortality. Graf and colleagues demonstrated that when sarcopenia was synonymous with LMM. Three of these
a decrease in lean mass index increases mortality in adults aged definitions are most important: The first adjusts ALM for
65 years or older in the long‐term.(40) Two other studies that height(32); the second adjusts ALM for body fat(24,25); and the
evaluated dynamic changes in body composition reached third adjusts ALM for BMI.(43) The definition by Baumgartner
similar conclusions.(41,42) and colleagues is based on the presence of an ALMI (ALM /
Another major issue in body composition studies is how to height2) 2 SDs below the mean for the young adult population
accurately distinguish normal from reduced muscle mass and has been demonstrated to be a poor predictor of
(and its estimates). That said, there is no consensus on which disability.(45) However, this same method based on different
methods and which cut‐offs to rely on. For instance, population data is recommended by the recently published
Baumgartner et al(32) and EWGSOP2(5) suggest ASMI cut‐ European consensus (EGWSOP2).(5) The definition by the FNIH
offs based on standard deviations from a population mean. takes into account well‐defined cut‐offs for low ALM adjusted
The Foundation for the National Institutes of Health (FNIH) for BMI, which have been validated mainly for the American
relies on ASM/BMI cut‐offs that are associated with dis- population.(43) Finally, the definition by Newman et al(25) (ALM
ability.(43) None of these three classification cut‐offs have adjusted for fat) was one of the definitions of LMM chosen in
been proven to predict mortality. A fourth classification, as our study because it proved to be the one with the best
presented by Srikanthan and colleagues, classifies individuals performance in the SPAH population, as previously pub-
based on both muscle and fat mass according to population lished.(26,27) Compared with the definition by Baumgartner
medians. This classification proved to be useful for predicting and colleagues, it achieved a greater sensitivity for the
cardiovascular mortality.(44) It might not accurately discrimi- diagnosis of LMM in populations with a high prevalence of
nate, however, the sarcopenic obese. A given individual overweight and obesity, such as that of the SPAH. Also, ALMI in
might have both fat and muscle mass above the population our study was able to independently predict all‐cause and
median, but his or her muscle mass could be significantly cardiovascular mortality in older women. No association of
reduced compared with his or her own fat mass. This ALMI with mortality was seen in older men. Using ALMI to
individual would be otherwise classified by Newman's classify patients as having LMM or not according to cut‐offs
definition as having LMM. recently recommended by the EWGSOP(5) also proved to be a
Two decades ago sarcopenia was the term used to refer to useful predictor of mortality only for older women, and even
LMM; they were frequently used interchangeably. Currently, then to predict only all‐cause mortality and not cardiovascular
sarcopenia has a much broader definition and refers to a mortality. This is contrary to what was seen when we analyzed
concomitant loss of muscle mass and performance. However, lean mass adjusted for fat mass by the residuals method.

Journal of Bone and Mineral Research BODY COMPOSITION & MORTALITY IN OLDER BRAZILIANS 1271 ◼
 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 4. Independent Predictors of Mortality for the Female ALMI, as is currently recommended by most experts, instead of
Population After Multivariable Analysisa adjusting ALM for fat mass, would not be the best approach in
clinical practice. This is the first study to find an association
All‐cause mortality between LMM by this definition and mortality in older adults,
Baseline characteristics OR CI, 95% P and also to determine the most accurate residual cut‐off to
LMM (Newman) 62.88 22.59 – 175 <0.001 predict mortality in this same population.
Age (for each year) 1.07 1.01 – 1.15 0.035 Overall, our results are not only consistent with many
PTH (for each pg/ml) 1.02 1.01 – 1.04 0.045 previous studies on body composition and mortality, but they
Low level of physical activity 4.86 1.35 – 17.5 0.015 also have plausible grounds based on physiopathology.
Cardiovascular mortality Individuals with LMM, defined by ALMI cut‐offs, have higher
LMM (Newman) 74.54 9.72 – 571.46 <0.001 levels of hemoglobin A1c and higher brachial‐ankle pulse
Age (for each year) 1.11 1.02 – 1.21 0.012 velocity, which is a surrogate of arterial stiffness.(46) These two
biomarkers are, in turn, known established cardiovascular risks.
a
Body composition variables and covariates that remained significant in Decreased fat‐free mass, which is another surrogate of LMM, is
the final model. also associated with a higher prevalence of insulin resistance,
Covariates included in the model for all‐cause mortality are as follows: age, type 2 diabetes mellitus, dyslipidemia, and hypertension, all of
BMI, low level of physical activity, diabetes mellitus, hypertension. previous which are markers of higher mortality.(47) VAT, shown to be a
cardiovascular event, glomerular filtration rate (GFR), serum calcium, PTH,
predictor of mortality in older men in our study, has also been
25‐OHVitamin D, total hip BMD T‐score and VAT.
previously associated with cardiovascular risk factors. Zamboni
Covariates included in the model for cardiovascular mortality are as
and colleagues found that the age‐associated increase in VAT
follows: age, BMI, low level of physical activity, previous non‐vertebral
explains, at least in part, the concomitant increase in many
fracture, current smoking, previous cardiovascular event, glomerular
deleterious metabolic markers, such as serum glucose and
filtration rate (GFR), serum phosphorus, PTH, total hip BMD T‐score and VAT.
OR: odds ratio; CI: confidence interval; LMM (Newman): Low muscle mass
lipids.(48) Similar results were reached when assessing VAT and
according to Newman et al(25)
metabolic and cardiovascular risk in obese individuals.(49)
The limitations of the study include the limited sample size
and the exclusion of individuals from the original cohort who
Classifying older adults as having LMM according to ALM could not be identified as dead or alive at the end of the study
adjusted for fat mass was able to strongly predict both all‐ (loss to follow‐up). These individuals were treated as missing
cause and cardiovascular mortality in women and men. This data and were excluded from the analysis. As such, we treated
might be an indication that classifying LMM solely according to these individuals as a missing completely at random data and

Fig. 3. ROC curve for calculated residuals from SPAH study female participants, according to the definition of LMM adjusted for fat mass by Newman
and colleagues.(25) In female participants, the residual that yielded the best performance of sensitivity and specificity was − 1.325, according to
Youden's index formula. ROC = receiver operating characteristic; SPAH = São Paulo Ageing and Health; LMM = low muscle mass; expALM =
expected appendicular lean mass; actALM = actual appendicular lean mass; AUC = area under the curve [Color figure can be viewed at
wileyonlinelibrary.com].

◼ 1272 DE SANTANA ET AL. Journal of Bone and Mineral Research

 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
we performed our analysis as a complete case analysis. Lost to FMS, DSD, RMRP; Drafting Manuscript: FMS, DSD, RMRP;
follow up individuals were mainly due to change of address Revising Manuscript: FMS, RMRP; Approving Final Version of
and we might not have completely accounted for the Manuscript: FMS, DSD, MAG, LGM, CPF, JBL, VC, LT, PRM, RMRP.
possibility that individuals with LMM might be more prone to
become institutionalized or shift from living independently to
living with their families. In addition, the study did not include References
assessment of muscle strength and physical performance,
1. Batsis JA, Mackenzie TA, Bartels SJ, Sahakyan KR, Somers VK, Lopez‐
which are both known to be important for the current Jimenez F. Diagnostic accuracy of body mass index to identify
definition of sarcopenia and has been shown to be associated obesity in older adults: NHANES 1999–2004. Int J Obes.
with long‐term morbidity and mortality.(5,45,50) We also 2016;40(5):761–767.
acknowledge that the current study enrolled older individuals 2. Bea JW, Thomson CA, Wertheim BC, et al. Risk of mortality
from a single city. Although many social characteristics of our according to body mass index and body composition among
sample were similar to that of the whole Brazilian elderly postmenopausal women. Am J Epidemiol. 2015;182(7):585–96.
population, our results might not hold true for the entire 3. Santos RRd, Bicalho MAC, Mota P, Oliveira D.Rd, Moraes E.Nd.
Obesidade em idosos [Obesity in the elderly]. Revista Medica de
country. Hence, we caution against immediately extrapolating Minas Gerais. 2013;23(1):64–73.
our results to older adults residing abroad and, most
4. Zamboni M, Mazzali G, Zoico E, et al. Health consequences of
importantly, to younger individuals. Finally, we also recognize obesity in the elderly: a review of four unresolved questions. Int J
that many clinical variables were collected through a self‐ Obes. 2005;29(9):1011–29.
reported questionnaire, and as such, are susceptible to 5. Cruz‐Jentoft AJ, Bahat G, Bauer J, et al. Sarcopenia: revised
interviewer and recall bias. For some variables, this was European consensus on definition and diagnosis. Age Ageing.
minimized by complementing data with objective examination 2019;48(1):16–31. Jan
(eg, history of hypertension was complemented by blood 6. Thibault R, Pichard C. The evaluation of body composition: a useful
pressure measurement). tool for clinical practice. Ann Nutr Metab. 2012;60(1):6–16.
In conclusion, the present study demonstrates the role of 7. Jensen MD, Ryan DH, Apovian CM, et al. 2013 AHA/ACC/TOS
body composition by DXA in assessing the risk of death in guideline for the management of overweight and obesity in adults:
a report of the American College of Cardiology/American Heart
community‐dwelling older adults. Although we did not per- Association Task Force on Practice Guidelines and The Obesity
form an assessment for muscle strength and performance, the Society. J Amer Coll Cardiol. 2014;63(25 Part B):2985–3023.
results suggest that the presence of low muscle mass is a risk 8. Rubbieri G, Mossello E, Di Bari M. Techniques for the diagnosis of
factor by itself for increased mortality in elderly men and sarcopenia. Clin Cases Miner Bone Metab. 2014;11(3):181–84.
women. The fat mass seems to play opposite roles according to 9. Marinangeli CP, Kassis AN. Use of dual X‐ray absorptiometry to
its distribution in older men. Visceral fat increases the risk of measure body mass during short‐to medium‐term trials of nutrition
death, whereas total body fat has a protective role, quite and exercise interventions. Nutr Rev. 2013;71(6):332–42.
different to what is established in the literature for young 10. Enoki H, Kuzuya M, Masuda Y, et al. Anthropometric measurements
adults. These findings highlight the importance of body of mid‐upper arm as a mortality predictor for community‐dwelling
Japanese elderly: the Nagoya Longitudinal Study of Frail Elderly
composition by DXA as a promising tool for the evaluation of (NLS‐FE). Clin Nutr. 2007;26(5):597–604.
sarcopenia, body fat, and mortality risk in older adults.
11. Chuang S‐Y. Chang H‐Y, Lee M‐S, Chen RC‐Y, Pan W‐H. Skeletal
muscle mass and risk of death in an elderly population. Nutr, Metab
Cardiovasc Dis. 2014;24(7):784–91.
Disclosures 12. Cesari M, Pahor M, Lauretani F, et al. Skeletal muscle and mortality
results from the InCHIANTI Study. J Gerontol A Biol Sci Med Sci.
2009;64(3):377–84.
All authors state that they have no conflicts of interest. 13. Bunout D, de La Maza MP, Barrera G, Leiva L, Hirsch S. Association
between sarcopenia and mortality in healthy older people.
Australas J Ageing. 2011;30(2):89–92.
Acknowledgments 14. 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. Am J Clin Nutr.
This study was not sponsored by any pharmaceutical company. 2014;100(5):1352–60.
The SPAH project was supported by grants from the Fundacao 15. Lee JSW, Auyeung TW, Kwok T, Li M, Leung J, Woo J. Survival benefit of
de Amparo e Pesquisa do Estado de Sao Paulo (FAPESP #03/ abdominal adiposity: a 6‐year follow‐up study with dual X‐ray
09313‐0, #04/12694‐8, and #09/15346‐4; FAPESP #11/00411‐5 absorptiometry in 3,978 older adults. Age. 2012;34(3):597–608.
to DSD); Conselho Nacional de Ciencia e Tecnologia (CNPQ 16. Toss F, Wiklund P, Nordström P, Nordström A. Body composition
#300559/2009‐7 and #301805/2013‐0 to RMRP), Federico and mortality risk in later life. Age Ageing. 2012;41(5):677–81.
Foundation (to RMRP), and Coordenacao de Aperfeicoamento 17. Navaneethan SD, Kirwan JP, Arrigain S, Schold JD. Adiposity
measures, lean body mass, physical activity and mortality: NHANES
de Pessoal de Nıvel Superior (CAPES to DSD, JBL, and CPF). We 1999–2004. BMC Nephrol. 2014;15(1):108.
further cordially thank Rogerio Ruscitto do Prado, PhD, for
18. Katzmarzyk P, Mire E, Bouchard C. Abdominal obesity and
statistical analysis assistance; Jackeline Couto Alvarenga for mortality: the Pennington Center longitudinal study. Nutr Diabetes.
collecting the blood samples from the subjects; Vera Lucia 2012;2(8):e42–e42.
Barbosa and Maria de Lourdes Floriano for the recruitment of 19. Kuk JL, Katzmarzyk PT, Nichaman MZ, Church TS, Blair SN, Ross R.
study subjects, including door‐to‐door visits; and Maria Luıza Visceral fat is an independent predictor of all‐cause mortality in
Santilli for assistance in data entry. men. Obesity. 2006;14(2):336–41.
Authors’ roles: Study Design: FMS, DSD, PRM, RMRP; Study 20. Honda H, Qureshi AR, Axelsson J, et al. Obese sarcopenia in
Conduct: FMS, DSD, MAG, RMRP; Data Collection: DSD, LGM, patients with end‐stage renal disease is associated with inflamma-
tion and increased mortality. Am J Clin Nutr. 2007;86(3):633–8.
CPF, JBL, VFC, LT; Data Analysis: FMS, DSD; Data Interpretation:

Journal of Bone and Mineral Research BODY COMPOSITION & MORTALITY IN OLDER BRAZILIANS 1273 ◼
 15234681, 2019, 7, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.3710 by Cochrane Mexico, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
21. Kato A, Odamaki M, Yamamoto T, et al. Influence of body 36. World Health Organization. ICD‐10: international statistical classi-
composition on 5 year mortality in patients on regular haemodia- fication of diseases and related health problems/World Health
lysis. Nephrol Dialy Transpl. 2003;18(2):333–40. Organization. Geneva: World Health Organization. 2004. https://
22. Narumi T, Watanabe T, Kadowaki S, et al. Sarcopenia evaluated by www.cms.gov/Medicare/Coding/ICD10/2018‐ICD‐10‐CM‐and‐
fat‐free mass index is an important prognostic factor in patients GEMs.html. Accessed June 6, 2018.
with chronic heart failure. Eur J Intern Med. 2015;26(2):118–22. 37. Domiciano DS, Machado LG, Lopes JB, et al. Bone mineral density
23. Hanai T, Shiraki M, Nishimura K, et al. Sarcopenia impairs prognosis and parathyroid hormone as independent risk factors for mortality
of patients with liver cirrhosis. Nutr. 2015;31(1):193–9. in community‐dwelling older adults: a population‐based prospec-
tive cohort study in Brazil. the São Paulo Ageing & Health (SPAH)
24. Delmonico MJ, Harris TB, Lee JS, et al. Alternative definitions of study. J Bone Miner Res. 2016;31(6):1146–57.
sarcopenia, lower extremity performance, and functional impair-
ment with aging in older men and women. J Am Geriatr Soc. 38. Altman DG. Statistics in medical journals: some recent trends. Statis
2007;55(5):769–74. Med. 2000;19(23):3275–89.
25. Newman AB, Kupelian V, Visser M, et al. Sarcopenia: alternative 39. Kershaw EE, Flier JS. Adipose tissue as an endocrine organ. J Clin
definitions and associations with lower extremity function. J Am Endocrinol Metab. 2004;89(6):2548–56.
Geriatr Soc. 2003;51(11):1602–9. 40. Graf CE, Herrmann FR, Spoerri A, et al. Impact of body composition
26. Figueiredo C, Domiciano D, Lopes J, et al. Prevalence of sarcopenia changes on risk of all‐cause mortality in older adults. Clin Nutr.
and associated risk factors by two diagnostic criteria in community‐ 2016;35(6):1499–505.
dwelling older men: the São Paulo Ageing & Health Study (SPAH). 41. Lee CG, Boyko EJ, Nielson CM, et al. Mortality risk in older men
Osteopor Int. 2014;25(2):589–96. associated with changes in weight, lean mass, and fat mass. J Am
27. Domiciano D, Figueiredo C, Lopes J, et al. Discriminating Geriatr Soc. 2011;59(2):233–40.
sarcopenia in community‐dwelling older women with high 42. Szulc P, Munoz F, Marchand F, Chapurlat R, Delmas PD. Rapid loss
frequency of overweight/obesity: the Sao Paulo Ageing & Health of appendicular skeletal muscle mass is associated with higher all‐
Study (SPAH). Osteopor Int. 2013;24(2):595–603. cause mortality in older men: the prospective MINOS study. Am J
28. Lopes JB, Figueiredo CP, Caparbo VF, et al. Osteoporotic fractures Clinical Nutr. 2010;91(5):1227–36.
in the Brazilian community‐dwelling elderly: prevalence and risk 43. Studenski SA, Peters KW, Alley DE, et al. The FNIH sarcopenia
factors. J Clin Densitom. 2011;14(3):359–66. project: rationale, study description, conference recommendations,
29. IBGE. Instituto Brasileiro de Geografia e Estatística. Available from: and final estimates. J Gerontol A Biol Sci Med Sci.
http://www.ibge.gov.br/home/estatistica/populacao/perfilidoso 2014;69(5):547–58.
2000. 44. Srikanthan P, Horwich TB, Tseng CH. Relation of muscle mass and
30. Domiciano DS, Machado LG, Lopes JB, et al. Incidence and risk fat mass to cardiovascular disease mortality. Am J Cardiol.
factors for osteoporotic vertebral fracture in low‐income commu- 2016;117(8):1355–60.
nity‐dwelling elderly: a population‐based prospective cohort study 45. Newman AB, Kupelian V, Visser M, et al. Strength, but not muscle
in Brazil. The São Paulo Ageing & Health (SPAH) Study. Osteopor mass, is associated with mortality in the health, aging and body
Int. 2014;25(12):2805–15. composition study cohort. J Gerontol A Biol Sci Med Sci.
31. Shepherd JA, Lu Y, Wilson K, et al. Cross‐calibration and minimum 2006;61(1):72–7.
precision standards for dual‐energy X‐ray absorptiometry: the 2005 46. Sanada K, Miyachi M, Tanimoto M, et al. A cross‐sectional study of
ISCD Official Positions. J Clin Densitom. 2006;9(1):31–6. sarcopenia in Japanese men and women: reference values and
32. Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of association with cardiovascular risk factors. Eur J Appl Physiol.
sarcopenia among the elderly in New Mexico. Am J Epidemiol. 2010;110(1):57–65.
1998;147(8):755–63. 47. Karakelides H, Nair KS. Sarcopenia of aging and its metabolic
33. Mohammad A, Rolfe EDL, Sleigh A, et al. Validity of visceral impact. Curr Topics Devel Biol. 2005;68:123–48.
adiposity estimates from DXA against MRI in Kuwaiti men and 48. Zamboni M, Armellini F, Harris T, et al. Effects of age on body fat
women. Nutr Diabetes. 2017;7(1):e238–e238. distribution and cardiovascular risk factors in women. Am J Clin
34. Rothney MP, Xia Y, Wacker WK, et al. Precision of a new tool to Nutr. 1997;66(1):111–5.
measure visceral adipose tissue (VAT) using dual‐energy X‐Ray 49. Neeland IJ, Ayers CR, Rohatgi AK, et al. Associations of visceral and
absorptiometry (DXA). Obesity. 2013;21(1):E134–E6. abdominal subcutaneous adipose tissue with markers of cardiac and
35. Genant HK, Wu CY, Van Kuijk C, Nevitt MC. Vertebral fracture metabolic risk in obese adults. Obesity. 2013;21(9):E439–E47.
assessment using a semiquantitative technique. J Bone Miner Res. 50. Santilli V, Bernetti A, Mangone M, Paoloni M. Clinical definition of
1993;8(9):1137–48. sarcopenia. Clin Cases Miner Bone Metab. 2014;11(3):177.

◼ 1274 DE SANTANA ET AL. Journal of Bone and Mineral Research