See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/259875351

Development of a simple screening test for sarcopenia in older adults

Article  in  Geriatrics & Gerontology International · February 2014

DOI: 10.1111/ggi.12197 · Source: PubMed

CITATIONS READS

71 703

12 authors, including:

Shinya Ishii Takashi Higashiguchi

The University of Tokyo Fujita Health University
48 PUBLICATIONS   725 CITATIONS    121 PUBLICATIONS   1,396 CITATIONS   

SEE PROFILE SEE PROFILE

Obuchi Shuichi Hirano Hirohiko

Tokyo Metropolitan Institute of Gerontology Tokyo Metropolitan Institute of Gerontology
163 PUBLICATIONS   1,105 CITATIONS    93 PUBLICATIONS   785 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Do decrease of oral function leads to frailty or sarcopenia View project

falls and fall-related fractures View project

All content following this page was uploaded by Obuchi Shuichi on 24 April 2019.

The user has requested enhancement of the downloaded file.

bs_bs_banner

Geriatr Gerontol Int 2014; 14 (Suppl. 1): 93–101

ORIGINAL ARTICLE

Development of a simple screening test for sarcopenia in

older adults
Shinya Ishii,1 Tomoki Tanaka,2 Koji Shibasaki,1 Yasuyoshi Ouchi,3 Takeshi Kikutani,4
Takashi Higashiguchi,5 Shuichi P Obuchi,6 Kazuko Ishikawa-Takata,7 Hirohiko Hirano,6
Hisashi Kawai,6 Tetsuo Tsuji2 and Katsuya Iijima2
1
Department of Geriatric Medicine, Graduate School of Medicine, 2Institute of Gerontology, The University of Tokyo. 3Federation of
National Public Service Personnel Mutual Aid Associations Toranomon Hospital, 4Division of Clinical Oral Rehabilitation, The Nippon
Dental University Graduate School of Life Dentistry at Tokyo, 6Tokyo Metropolitan Institute of Gerontology, 7Division of Health Promotion
and Exercise, National Institute of Health and Nutrition, Tokyo, and 5Department of Surgery & Palliative Medicine, Fujita Health
University School of Medicine, Toyoake City, Japan

Aim: To develop a simple screening test to identify older adults at high risk for sarcopenia.
Methods: We studied 1971 functionally independent, community-dwelling adults aged 65 years or older randomly
selected from the resident register of Kashiwa city, Chiba, Japan. Data collection was carried out between September
and November 2012. Sarcopenia was defined based on low muscle mass measured by bioimpedance analysis and
either low muscle strength characterized by handgrip or low physical performance characterized by slow gait speed.
Results: The prevalence of sarcopenia was 14.2% in men and 22.1% in women. After the variable selection
procedure, the final model to estimate the probability of sarcopenia included three variables: age, grip strength and
calf circumference. The area under the receiver operating characteristic curve, a measure of discrimination, of the final
model was 0.939 with 95% confidence interval (CI) of 0.918–0.958 for men, and 0.909 with 95% CI of 0.887–0.931
for women. We created a score chart for each sex based on the final model. When the sum of sensitivity and specificity
was maximized, sensitivity, specificity, and positive and negative predictive values for sarcopenia were 84.9%, 88.2%,
54.4%, and 97.2% for men, 75.5%, 92.0%, 72.8%, and 93.0% for women, respectively.
Conclusions: The presence of sarcopenia could be detected using three easily obtainable variables with high
accuracy. The screening test we developed could help identify functionally independent older adults with sarcopenia
who are good candidates for intervention. Geriatr Gerontol Int 2014; 14 (Suppl. 1): 93–101.

Keywords: disability, rehabilitation, sarcopenia, screening, sensitivity and specificity.

Introduction have been vigorously sought and some interventions,

such as resistance training in combination with nutri-
Sarcopenia is a syndrome characterized by progressive tional supplements, appear promising.2–4 It is also
and generalized loss of skeletal mass and strength with becoming apparent that interventions might be more
aging.1 A recent realization that sarcopenia is associated effective early rather than late in the course when
with a risk of adverse events, such as physical disability, patients develop physical disability or functional depen-
poor quality of life and death, has provided significant dence.4,5 The early stage in the course of sarcopenia (i.e.
impetus to sarcopenia research.1 Effective interventions without loss of physical or functional independence)
might therefore represent a valuable opportunity to
carry out interventions to decelerate the progress of
sarcopenia and prevent physical disability.
Accepted for publication 17 October 2013. However, patients with sarcopenia are generally
unaware of their sarcopenic state until the gradual
Correspondence: Dr Katsuya Iijima MD, Institute of
Gerontology, The University of Tokyo, 7-3-1 Hongo, decline in muscle function becomes severe enough to
Bunkyo-ku, Tokyo 113-8656, Japan. Email: be pathological, resulting in physical and functional
[email protected] dependence.4,6 As patients are unlikely to seek medical

© 2014 Japan Geriatrics Society doi: 10.1111/ggi.12197 | 93

 S Ishii et al.

attention for their sarcopenic state, population screen- The study was approved by the ethics committee of
ing to detect sarcopenia before the occurrence of physi- the Graduate School of Medicine, The University of
cal disability could improve the chance of intervention. Tokyo. All participants provided written informed
Currently, the recommended criteria for the diagnosis consent.
of sarcopenia require the documentation of low muscle
mass and either low muscle strength or low physical
Sarcopenia classification and measurement of each
performance.1 Muscle mass is commonly assessed by
component of sarcopenia
dual energy X-ray absorptiometry (DXA) or bioim-
pedance analysis (BIA), muscle strength with handgrip We followed the recommendation of the European
strength, and physical performance with Short Physical Working Group on Sarcopenia in Older People
Performance Battery or usual gait speed.1,7 Unfortu- (EWGSOP) for the definition of sarcopenia.1 The pro-
nately, the feasibility of applying the recommended posed diagnostic criteria required the presence of low
diagnostic algorithm in the setting of population screen- muscle mass plus the presence of either low muscle
ing is limited by the need for special equipment strength or low physical performance.
and training. Hence, a screening test for sarcopenia
simple enough to be carried out on a large scale is
Muscle mass measurement
required.
Using baseline data from the Kashiwa study on func- Muscle mass was measured by BIA using an Inbody 430
tionally independent, community-dwelling older adults, machine (Biospace, Seoul, Korea).8 Appendicular skel-
we designed an analysis to develop a simple screening etal muscle mass (ASM) was derived as the sum of the
test for sarcopenia and examine its ability to estimate the muscle mass of the four limbs. ASM was then normal-
probability of sarcopenia. ized by height in meters squared to yield skeletal muscle
mass index (SMI) (kg/m2).1 SMI values lower than two
standard deviations below the mean values of young
male and female reference groups were classified as low
Methods
muscle mass (SMI <7.0 kg/m2 in men, <5.8 kg/m2 in
women).9
Participants
The Kashiwa study is a prospective cohort study
Muscle strength measurement
designed to characterize the biological, psychosocial
and functional changes associated with aging in Muscle strength was assessed by handgrip strength,
community-dwelling older adults. In 2012, a total of which was measured using a digital grip strength dyna-
12 000 community-dwelling, functionally independent mometer (Takei Scientific Instruments, Niigata, Japan).
(i.e. not requiring nursing care provided by long-term The measurement was carried out twice using their
care insurance) adults aged 65 years or older were ran- dominant hand, and the higher of two trials (in kilo-
domly drawn from the resident register of Kashiwa city, grams) was used for the present analysis. Handgrip
a commuter town for Tokyo in Chiba prefecture, Japan, strength values in the lowest quintile were classified as
and asked by mail to participate in the study. A total of low muscle strength (cut-off values: 30 kg for men,
2044 older adults (1013 men, 1031 women) agreed to 20 kg for women).
participate in the study and comprised the inception
cohort. The sample reflected the distribution of age in
Physical performance measurement
Kashiwa city for each sex.
Baseline examinations were carried out between Sep- Physical performance was assessed by usual gait speed.
tember and November 2012 at welfare centers and Participants were instructed to walk over an 11-m
community centers close to the participants’ residential straight course at their usual speed. Usual gait speed
area, to obviate their need to drive. A team consisting was derived from 5 m divided by the time in seconds
of physicians, nurses, physical therapists, dentists and spent in the middle 5 m (from the 3-m line to the 8-m
nutritionists carried out data collection. To standardize line). Good reproducibility of this measurement was
data collection protocol, they were given the data col- reported previously.10 Usual gait speed values in the
lection manual, attended two sessions for training in lowest quintile were classified as low physical perfor-
the data collection methods and carried out a rehearsal mance (cut-off values: 1.26 m/s for each sex).
of data collection. A total of 73 participants who
did not undergo BIA, usual gait speed or handgrip
Other measurements
strength measurements were excluded, leaving an
analytic sample of 1971 older adults (977 men, 994 Demographic information and medical history of
women). doctor-diagnosed chronic conditions were obtained

94 | © 2014 Japan Geriatrics Society

 Sarcopenia screening

using a standardized questionnaire. Physical activity was ability) was assessed by the area under the receiver
assessed using Global Physical Activity Questionnaire operator characteristic (ROC) curve.16,17 The model fit
and Metabolic Equivalent minutes per week was com- was verified using the Hosmer–Lemeshow goodness-of-
puted.11 Serum albumin was measured at the time of the fit test.18
visit. Anthropometric measurements were obtained There were no missing values of any variable in the
with the participants wearing light clothing and no entire analytic sample.
shoes. Height and weight were measured with a fixed All analyses were carried out using SAS version
stadiometer, and a digital scale and used to compute 9.3 (SAS Institute, Cary, NC, USA) and R stati-
body mass index (BMI). Upper arm, thigh and calf cir- stical software version 2.15.2 (R Foundation, Vienna,
cumferences were measured to the nearest 0.1 cm Austria). Two-sided P < 0.05 was considered statistically
directly over the skin using a measuring tape with the significant.
participant sitting. Upper arm circumference was mea-
sured at the mid-point between the olecranon process Results
and the acromion of the non-dominant arm with the
participant’s arm bent 90° at the elbow. Calf circumfer- There were 32.2% of men and 48.9% of women clas-
ence measurement was made at the maximum circum- sified as having low muscle mass, and 14.2% of men
ference of the lower non-dominant leg with the and 22.1% of women were classified as having
participant’s leg bent 90° degrees at the knee. Thigh sarcopenia. The participant characteristics by the
circumference was measured 15 cm above the upper sarcopenia status in each sex are shown in Table 1.
margin of the patella of the dominant leg. Those with sarcopenia were older and had smaller body
size compared with those without sarcopenia in each
sex (all P < 0.001). Those with sarcopenia were physi-
Statistical analysis
cally less active in each sex. Chronic medical conditions
All analyses were stratified by sex. Differences in par- were in general more prevalent in those with sarcopenia,
ticipant characteristics between those with and without and a statistically significant difference was observed for
sarcopenia were examined using Student’s t-test or hypertension in women, stroke in men and osteoporosis
Wilcoxon rank–sum test. To develop a statistical model in both sexes. Serum albumin was significantly lower in
to estimate the probability of sarcopenia, candidate vari- those with sarcopenia in each sex.
ables were selected by experts based on cost, ease of Table 2 shows the correlation between each compo-
measurement and availability of equipment to measure nent of sarcopenia and the candidate variables. SMI was
them. The candidate variables included age, sex, BMI, correlated with all the variables, with the highest corre-
grip strength, and thigh, calf and upper arm circumfer- lation coefficient observed with calf circumference in
ences. Pearson’s correlation between each component each sex. Usual gait speed was most highly correlated
of sarcopenia and the candidate variables was first com- with age, followed by grip strength and calf circumfer-
puted. We then examined the functional form of the ence in the order of the magnitude of correlation, and
relationships between the variables, and the logit of this finding was consistent in both sexes.
sarcopenia probability using restricted cubic spline plots Visual inspection of the restricted cubic spline plots
and the Wald test for linearity.12 We considered and the Wald test for linearity suggested that the
dichotomization, square and logarithmic transforma- variables were linearly associated with the logit of
tions if the Wald test for linearity was statistically sig- sarcopenia probability, except for grip strength in both
nificant, rejecting the assumption of linearity.12 A sexes and upper arm circumference in women (data not
multivariate logistic regression model including all the shown). However, neither dichotomization nor trans-
candidate variables (“full model”) was constructed. formation improved the model fit, and we decided to
Variable selection with Bayesian Information Criteria use linear terms of these variables in the development of
was carried out to make the model parsimonious, and statistical models.
a multivariate logistic regression model including the Table 3 shows the unadjusted and adjusted associa-
variables selected (“restricted model”) was made.13 A tions between sarcopenia and the variables. In bivariate
bootstrapping procedure was used to obtain estimates analysis, all the variables were significantly associated
of internal validity of the model14 and to derive the final with sarcopenia. In multiple logistic regression with all
models by correcting the regression coefficients for the variables (full model), age was positively, and grip
overoptimism.15 The final model was presented as a strength and calf circumference were inversely associ-
score chart to facilitate clinical application.15 The score ated with sarcopenia, whereas BMI, thigh circumference
chart was created based on rounded values of the and upper arm circumference were not significantly
shrunken regression coefficients. associated. Variable selection resulted in the selection of
The ability of each model to correctly rank order age, grip strength and calf circumference, and the three
participants by sarcopenia probability (discrimination selected variables were significantly associated with

© 2014 Japan Geriatrics Society | 95

96
|
Table 1 Characteristics of study participants

Men Women
Sarcopenia No sarcopenia P Sarcopenia No sarcopenia P
(n = 139) (n = 838) (n = 220) (n = 774)
Age (years) 78.4 ± 5.5 72.2 ± 5.0 <0.001 76.2 ± 5.8 71.8 ± 4.9 <0.001
Height (cm) 160.0 ± 5.6 164.9 ± 5.5 <0.001 148.2 ± 5.6 152.3 ± 5.1 <0.001
Weight (kg) 54.1 ± 7.2 64.3 ± 8.0 <0.001 46.4 ± 5.7 52.9 ± 7.6 <0.001
BMI (kg/m2) 21.1 ± 2.5 23.6 ± 2.6 <0.001 21.1 ± 2.6 22.8 ± 3.2 <0.001
Grip strength (kg) 27.5 ± 4.3 36.0 ± 5.3 <0.001 18.4 ± 3.2 23.6 ± 3.3 <0.001
Thigh circumference (cm) 38.8 ± 3.5 42.4 ± 3.3 <0.001 38.9 ± 3.4 41.7 ± 4.0 <0.001
Calf circumference (cm) 32.8 ± 2.3 36.3 ± 2.5 <0.001 32.1 ± 2.1 34.5 ± 2.7 <0.001
Upper arm circumference (cm) 25.7 ± 2.5 28.4 ± 2.4 <0.001 25.7 ± 2.3 27.3 ± 2.9 <0.001
SMI (kg/m2) 6.34 ± 0.48 7.44 ± 0.58 <0.001 5.25 ± 0.41 6.02 ± 0.60 <0.001
Usual gait speed (m/s) 1.28 ± 0.24 1.51 ± 0.24 <0.001 1.26 ± 0.26 1.51 ± 0.23 <0.001
S Ishii et al.

Physical activity (MET-minutes/week) 1813 (720, 3504) 2540 (1200, 4746) 0.008 1341 (33, 3209) 2587 (1092, 4824) <0.001
Chronic conditions (%)
Hypertension 51.1 46.5 0.32 45.9 38.1 0.04
Diabetes mellitus 18.0 14.9 0.36 8.2 8.9 0.73
Stroke 12.2 6.4 0.01 5.9 4.4 0.35
Osteoporosis 4.3 1.4 0.02 32.7 16.6 <0.001
Use of medications (%)
Statins 18.7 17.4 0.71 29.1 30.6 0.66
Antihypertensives 53.2 45.1 0.08 42.7 36.2 0.08
Albumin (g/dL) 4.37 ± 0.26 4.43 ± 0.23 0.005 4.39 0.23 4.43 ± 0.22 0.04
Values are shown as mean ± standard deviation except for physical activity which was not normally distributed and therefore the mean value and inter-quartile range were
shown. BMI, body mass index; MET, Metabolic Equivalent; SMI, skeletal muscle mass index.

© 2014 Japan Geriatrics Society

 Sarcopenia screening

Table 2 Pearson correlations between components of sarcopenia and six candidate variables

Age BMI Grip Thigh Calf Upper arm

strength circumference circumference circumference
Men
SMI −0.33*** 0.70*** 0.49*** 0.70*** 0.78*** 0.69***
Grip strength −0.46*** 0.21*** 1 0.27*** 0.35*** 0.35***
Usual gait speed −0.35*** 0.007 0.29*** 0.06 0.13*** 0.10**
Women
SMI −0.24*** 0.69*** 0.50*** 0.67*** 0.75*** 0.65***
Grip strength −0.36*** 0.16*** 1 0.22*** 0.33*** 0.21***
Usual gait speed −0.42*** −0.08** 0.36*** 0.01 0.12*** −0.02
*, **, ***Significance at 0.1%, 1%, 5% level, respectively. BMI, body mass index; SMI, skeletal muscle mass index.

sarcopenia in multiple logistic regression (restricted gait speed to capture participants with more severely
model). These findings were consistent in both sexes. impaired muscle function (i.e. strength or performance),
The area under the ROC curve of the full model was and defined them as having sarcopenia, with the same
0.940 (95% confidence interval [CI] 0.920–0.959) for cut-off values for muscle mass as in the main analysis.
men and 0.910 (95% CI 0.888–0.932) for women, We then examined the model performance with all six
showing excellent discriminative ability. The area under variables and with the same set of three variables as
the ROC curve of the restricted model (0.939 with 95% selected in the main analysis (age, grip strength and calf
CI 0.918–0.958 for men and 0.909 with 95% CI 0.887– circumference). The cut-off value of grip strength was
0.931 for women) was not significantly different from 27 kg for men and 17 kg for women, and that of usual
that of the full model in both sexes (P = 0.71 for men, gait speed was 1.16 m/s for men and 1.13 m/s for
0.43 for women). Assessment of internal validity women. The prevalence of sarcopenia was 9.6% in men
showed that discriminative ability of the restricted and 12.7% in women. Both models performed well
model is expected to be good in similar populations (area of the full model: 0.932 for men, 0.919 for women;
(area 0.937 for men, 0.907 for women). area for the restricted model; 0.931 for men, 0.918 for
The final model was presented as a score chart in each women; Figure S2).
sex (Table 4). The use of the score chart with two hypo-
thetical patients is shown in Table S1. The discrimina- Discussion
tive ability of the score chart was comparable with those
of the full and restricted models in each sex (area 0.935 To estimate the probability of sarcopenia in functionally
for men, 0.908 for women; Fig. S1). independent, community-dwelling Japanese older
Figure 1 shows the estimated probabilities corre- adults, we created multivariate models based on the
sponding to the sum scores as calculated with the score three selected variables (age, grip strength and calf cir-
chart in Table 4, and the sensitivity and specificity using cumferences), and found excellent discrimination ability
the sum scores as cut-off values. The sum score that of the models: the area under the curve was 0.939 for
maximized the sum of sensitivity and specificity was 105 men and 0.909 for women. We constructed a score
for men and 120 for women. The corresponding sensi- chart in each sex so that the approximate probability of
tivity, specificity, positive and negative predictive values, sarcopenia could be easily obtained from the values of
and positive and negative likelihood ratios were 84.9%, the three variables, and confirmed that the score charts
88.2%, 54.4% and 97.2%, and 7.19 and 0.17 for men, also had excellent discrimination.
and 75.5%, 92.0%, 72.8% and 93.0%, and 9.44 and Although our multivariate models had excellent dis-
0.27 for women, respectively. crimination capacity, the model’s sensitivity and speci-
ficity at candidate diagnostic thresholds must be
assessed to judge the model’s clinical usefulness.18
Sensitivity analysis
Higher sensitivity can be achieved at the expense of
Because there are no established reference cut-off lower specificity and vice versa. For example, if higher
values for grip strength and usual gait speed in Japanese sensitivity was desired; for example, 90%, then the cut-
older adults, we used the lowest quintiles of the off score would be 101 for men and 104 for women, and
observed distributions to classify low muscle strength the specificity would be lower at 82.2% for men and
and low physical performance. As sensitivity analysis, 70.4% for women. Higher specificity, 90%, could be
we used the lowest deciles of grip strength and usual achieved with the higher cut-off score of 107 for men

© 2014 Japan Geriatrics Society | 97

 S Ishii et al.

and 118 for women, resulting in lower sensitivity of

<0.001

<0.001

<0.001
77.7% for men and 76.8% for women (Fig. 1). The

P
(restricted model)
trade-off between sensitivity and specificity depends
OR (95% CI) on the cost of incorrect classification of those with
Multivariate

(1.04, 1.13)

(0.55, 0.65)

(0.65, 0.78)
sarcopenia relative to the cost of incorrect classification
1.09 of those without sarcopenia. The cost of incorrect

0.59

0.71
answers would vary according to the clinical or research
scenario and personal preferences.16,17
Several observations suggested that the selection of
<0.001

<0.001

<0.001
0.05

0.24

0.10
three variables (age, grip strength and calf circumfer-
P

ence) was not based on chance. First, sarcopenia was

classified based on muscle mass, muscle strength and
OR (95% CI)
Multivariate
(full model)

(1.05, 1.14)

(0.74, 1.00)

(0.53, 0.64)

(0.85, 1.04)

(0.69, 0.91)

(0.98, 1.35)
physical performance, all of which were significantly
correlated with the three variables. Calf circumference
1.10

0.86

0.58

0.94

0.80

1.15 was used to represent muscle mass, considering the

highest correlation between SMI and calf circumference
among the variables considered. A strong correlation
<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

between calf circumference and muscle mass was pre-

P

viously shown in Caucasian older women who were on

average more obese than women in the present.19 Grip
OR (95% CI)

(1.13, 1.20)

(0.78, 0.87)

(0.53, 0.62)

(0.78, 0.86)

(0.64, 0.74)

(0.75, 0.85)

strength was used as an indicator of muscle strength.

Bivariate
Women

Usual gait speed, a measure of physical performance,

1.16

0.82

0.57

0.82

0.68

0.80

was significantly correlated with each of the three vari-

ables. Second, sarcopenia was associated with each of
the three variables in both bivariate and multivariate
0.008

<.001

<.001
Table 3 Unadjusted and adjusted associations between sarcopenia and the variables

analyses in each sex, and P-values for these findings

P
(restricted model)

were comfortably below 0.01. Third, the models with

OR (95% CI)

the three variables had excellent discrimination for

Multivariate

(1.02, 1.12)

(0.68, 0.79)

(0.56, 0.69)

sarcopenia based on more stringent cut-off levels for

grip strength and usual gait speed.
1.07

0.73

0.62

There have been several prior attempts at estimating

the quantity of muscle mass using a variety of variables
0.008

<0.001

<0.001

with varying degrees of accuracy.20–23 Although these

0.69

0.53

0.71

studies were inspired by the desire to facilitate the diag-

P

nosis of sarcopenia, recently developed definitions of

OR (95% CI)

sarcopenia entail the presence of low muscle function,

Multivariate
(full model)

(1.02, 1.12)

(0.78, 1.18)

(0.68, 0.78)

(0.91, 1.21)

(0.53, 0.73)

(0.82, 1.15)

as well as muscle mass.1,24 The present study developed

statistical models with high accuracy for sarcopenia,
1.07

0.96

0.73

1.05

0.62

0.97

which was defined based on muscle mass and muscle

function.
BMI, body mass index; CI, confidence interval; OR, odds ratio.
<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

This study had several limitations. First, the measure-

ment method of usual gait speed was different from
P

those used by the majority of previous studies.25 The

OR (95% CI)

measurement method used in the present study

(0.67, 0.75)

(0.69, 0.78)

(0.52, 0.63)

(0.57, 0.68)
(1.17–1.26)

(0.63–0.74)

required the participant to walk 3 m before the mea-

Bivariate

surement started. An attribute of this method is that

Men

1.21

0.68

0.71

0.73

0.57

0.63

it is less affected by the gait initiation phase where

age-related changes independent of gait speed occur.26,27
Upper arm circumference

This method has been widely used in Japan,9,28 and has

been shown to be reliable,10 but because it starts mea-
Thigh circumference

Calf circumference

suring after the gait initiation phase, it tends to yield

higher values than those obtained with other measure-
Grip strength

ment methods, such as usual gait speed over a 4- or 6-m

Variables

course,25 making direct comparison difficult. Second,

the current analysis was carried out on data from Japa-
BMI
Age

nese older adults, and our findings therefore might not

98 | © 2014 Japan Geriatrics Society

 Sarcopenia screening

Table 4 Score charts for estimated probability of sarcopenia

Variables Value

Men
Age <66 66 68 70 72 74 76 78 80 82 84 86≦
Score 0 +1 +2 +3 +4 +5 +6 +7 +8 +9 +10 +11
Grip strength <20 20 23 26 29 32 35 38 41 44 47 50≦
Score +99 +90 +81 +72 +63 +54 +45 +36 +27 +18 +9 0
Calf circumference <26 26 28 30 32 34 36 38 40 42≦
Score +81 +72 +63 +54 +45 +36 +27 +18 +9 0
Estimated individual
probability of
sarcopenia
Sum score 70 80 90 95 100 105 110 115 120 125 130 135 140 145
Probability (%) 1 2 5 8 13 19 28 39 51 64 74 83 89 93
Women
Age <66 66 68 70 72 74 76 78 80 82 84 86≦
Score 0 +2 +4 +6 +8 +10 +12 +14 +16 +18 +20 +22
Grip strength <14 14 16 18 20 22 24 26 28 30 32 34≦
Score +110 +100 +90 +80 +70 +60 +50 +40 +30 +20 +10 0
Calf leg circumference <26 26 28 30 32 34 36 38 40 42≦
Score +63 +56 +49 +42 +35 +28 +21 +14 +7 0
Estimated individual
probability of
sarcopenia
Sum score 80 90 95 100 105 110 115 120 125 130 135 140 145 150
Probability (%) 1 3 5 8 12 19 28 39 51 63 74 82 88 93
Values for each variable are given with such intervals that the scores show small steps, and scores for intermediate values can be estimated by
linear interpolation. The exact formula to calculate the scores are as follows: score in men, 0.62 × (age – 64) – 3.09 × (grip strength – 50) –
4.64 × (calf circumference – 42); score in women, 0.80 × (age – 64) – 5.09 × (grip strength – 34) – 3.28 × (calf circumference – 42). The
corresponding probabilities of sarcopenia are calculated with the following formulae: probability in men, 1 / [1 + e−(sum score / 10–11.9)]; probability in
women, 1 / [1 + e−(sum score / 10–12.5)].

Figure 1 Estimated probabilities, sensitivity and specificity corresponding to sum scores. The sum scores and corresponding
estimated probabilities are read from Table 3.

© 2014 Japan Geriatrics Society | 99

 S Ishii et al.

be applicable to populations of other race/ethnicity or in Disclosure statement

other countries. Similarly, caution should be exercised
in projecting beyond the range of our data. For example, The authors declare no conflict of interest.
the obese were underrepresented in our data, and the
performance of our models was not assessed for the
obese. However, the present findings suggest that three References
variables, namely age, grip strength and calf circumfer-
ence, should be considered for inclusion in the devel- 1 Cruz-Jentoft AJ, Baeyens JP, Bauer JM et al. Sarcopenia:
European consensus on definition and diagnosis: report of
opment of sarcopenia screening in other populations. the European Working Group on Sarcopenia in Older
Third, although the internal validity was good (i.e. the People. Age Ageing 2010; 39: 412–423.
models would perform well in a similar population), 2 Yamada M, Arai H, Yoshimura K et al. Nutritional supple-
assessment of external validity is still warranted to deter- mentation during resistance training improved skeletal
mine whether the results can be extended to other Japa- muscle mass in community-dwelling frail older adalts.
J Frailty Aging 2012; 1: 64–70.
nese populations. Finally, we could not exclude the 3 Waters DL, Baumgartner RN, Garry PJ, Vellas B. Advan-
possibility of the healthy volunteer effect (i.e. volunteers tages of dietary, exercise-related, and therapeutic interven-
for clinical studies tend to be healthier than the general tions to prevent and treat sarcopenia in adult patients: an
population). Although participants were randomly update. Clin Interv Aging 2010; 5: 259–270.
selected from the resident register, participation was 4 Visvanathan R, Chapman I. Preventing sarcopaenia in
older people. Maturitas 2010; 66: 383–388.
voluntary and the response rate was approximately 5 Peterson MD, Sen A, Gordon PM. Influence of resistance
17%. However, the sensitivity analysis showed that the exercise on lean body mass in aging adults: a meta-analysis.
models’ ability to estimate the probability of sarcopenia Med Sci Sports Exerc 2011; 43: 249–258.
remained excellent when participants with more 6 Rosenberg IH. Sarcopenia: origins and clinical relevance.
severely impaired muscle function were categorized as J Nutr 1997; 127: 990S–991S.
7 Mijnarends DM, Meijers JM, Halfens RJ et al. Validity and
having sarcopenia. reliability of tools to measure muscle mass, strength,
In conclusion, we showed that the presence of and physical performance in community-dwelling older
sarcopenia in older adults could be detected with high people: a systematic review. J Am Med Dir Assoc 2013; 14:
accuracy using three easily obtainable variables. Impor- 170–178.
tantly, we derived the models from a functionally inde- 8 Shafer KJ, Siders WA, Johnson LK, Lukaski HC. Validity
of segmental multiple-frequency bioelectrical impedance
pendent, community-dwelling population. Functionally analysis to estimate body composition of adults across a
independent older adults with sarcopenia are good can- range of body mass indexes. Nutrition 2009; 25: 25–32.
didates for interventions to prevent further physical 9 Tanimoto Y, Watanabe M, Sun W et al. Association
limitations, given their potential for regaining muscle between muscle mass and disability in performing instru-
mass and restoration of muscle function. The score mental activities of daily living (IADL) in community-
dwelling elderly in Japan. Arch Gerontol Geriatr 2012; 54:
charts we developed can be used as an effective screen- e230–e233.
ing tool and help identify functionally independent 10 Nagasaki H, Itoh H, Hashizume K, Furuna T, Maruyama
older adults with sarcopenia. H, Kinugasa T. Walking patterns and finger rhythm of
older adults. Percept Mot Skills 1996; 82: 435–447.
11 Ainsworth BE, Bassett DR, Jr, Strath SJ et al. Comparison
of three methods for measuring the time spent in physical
activity. Med Sci Sports Exerc 2000; 32: S457–S464.
Acknowledgments 12 Frank EH, Jr. Regression Modeling Strategies : With Applica-
tions to Linear Models, Logistic Regression, and Survival Analysis,
This work was supported by a Health and Labor 1st edn. New York, NY: Springer, 2001.
Sciences Research Grant (H24-Choju-Ippan-002) 13 Hastie T, Tibshirani R, Friedman J. The Elements of Statistical
Learning: Data Mining, Inference, and Prediction, 2nd edn.
from the Ministry of Health, Labor, and Welfare of New York, NY: Springer, 2009.
Japan. The authors thank the staff members and par- 14 Steyerberg EW, Harrell FE, Borsboom GJJM, Eijkemans
ticipants of the Kashiwa study and the following indi- MJC, Vergouwe Y, Habbema JDF. Internal validation of
viduals for helping with the acquisition of data: Dr predictive models: efficiency of some procedures for logis-
Yoshiya Oishi PhD DDS, Oishi Dental Clinic. Yuki tic regression analysis. J Clin Epidemiol 2001; 54: 774–
781.
Ohara, Tokyo Metropolitan Geriatric Institute of 15 Steyerberg EW. Clinical Prediction Models, 1st edn. New
Gerontology; Dr Noriaki Takahashi and Dr Hiroyasu York, NY: Springer, 2009.
Furuya, The Nippon Dental University; Seigo 16 Hanley JA, McNeil BJ. The meaning and use of the area
Mitsutake, Tokyo Metropolitan Institute of Gerontol- under a receiver operating characteristic (ROC) curve.
ogy; Mr Masashi Suzuki, Institute of Gerontology, The Radiology 1982; 143: 29–36.
17 Faraggi D, Reiser B. Estimation of the area under the ROC
University of Tokyo; and staff members of The Institute curve. Stat Med 2002; 21: 3093–3106.
of Healthcare Innovation Project, The University of 18 Homer D, Lemeshow S. Applied Logistic Regression. New
Tokyo. York: John Wiley & Sons, 2000.

100 | © 2014 Japan Geriatrics Society

 Sarcopenia screening

19 Rolland Y, Lauwers-Cances V, Cournot M et al. 26 Henriksson M, Hirschfeld H. Physically active older adults
Sarcopenia, calf circumference, and physical function of display alterations in gait initiation. Gait Posture 2005; 21:
elderly women: a cross-sectional study. J Am Geriatr Soc 289–296.
2003; 51: 1120–1124. 27 Polcyn AF, Lipsitz LA, Kerrigan DC, Collins JJ. Age-related
20 Chen BB, Shih TT, Hsu CY et al. Thigh muscle volume changes in the initiation of gait: degradation of central
predicted by anthropometric measurements and correlated mechanisms for momentum generation. Arch Phys Med
with physical function in the older adults. J Nutr Health Rehabil 1998; 79: 1582–1589.
Aging 2011; 15: 433–438. 28 Tanimoto Y, Watanabe M, Sun W et al. Association of
21 Iannuzzi-Sucich M, Prestwood KM, Kenny AM. Preva- sarcopenia with functional decline in community-dwelling
lence of sarcopenia and predictors of skeletal muscle mass elderly subjects in Japan. Geriatr Gerontol Int 2013; 13: 958–
in healthy, older men and women. J Gerontol A Biol Sci Med 63.
Sci 2002; 57: M772–M777.
22 McIntosh EI, Smale KB, Vallis LA. Predicting fat-free mass
index and sarcopenia: a pilot study in community-dwelling
older adults. Age (Dordrecht, Netherlands) 2013; 35: 2423– Supporting information
2434.
23 Kenny AM, Dawson L, Kleppinger A, Iannuzzi-Sucich M, Additional Supporting Information may be found in the
Judge JO. Prevalence of sarcopenia and predictors of skel- online version of this article at the publisher’s web-site:
etal muscle mass in nonobese women who are long-term
users of estrogen-replacement therapy. J Gerontol A Biol Sci Figure S1 Receiver operating characteristic curves of
Med Sci 2003; 58: M436–M440. models estimating the probability of sarcopenia.
24 Muscaritoli M, Anker SD, Argiles J et al. Consensus defi- Figure S2 Receiver operating characteristic curves of
nition of sarcopenia, cachexia and pre-cachexia: joint
document elaborated by Special Interest Groups (SIG) models estimating the probability of sarcopenia based
“cachexia-anorexia in chronic wasting diseases” and on different cut-off values for grip strength and usual
“nutrition in geriatrics”. Clin Nutr 2010; 29: 154–159. gait speed.
25 Abellan van Kan G, Rolland Y, Andrieu S et al. Gait speed Table S1 Application of Score Chart in two hypotheti-
at usual pace as a predictor of adverse outcomes in cal patients.
community-dwelling older people an International
Academy on Nutrition and Aging (IANA) Task Force.
J Nutr Health Aging 2009; 13: 881–889.

© 2014 Japan Geriatrics Society | 101

View publication stats