Cumulative and Incremental Value of Sarcopenia Components on Predicting Adverse Outcomes

Abstract

Objectives:

To examine the incremental value of sarcopenia components, following the diagnosis algorithm of the Asian consensus, on predicting adverse outcomes.

Design:

A prospective cohort study.

Setting and Participants:

Four thousand community-dwelling Chinese adults (2000 men) aged 65 years or older in Hong Kong (mean age = 72.5 ± 5.2).

Methods:

SARC-F was used as the initial predictor of 9 adverse outcomes. In step 2, muscle strength (ie, grip strength) and/or functions (ie, chair-stand, walking speed) were added on top of SARC-F. In step 3, height-, weight-, and body mass index–adjusted appendicular skeletal mass (ASM) measured by dual-energy x-ray absorptiometry (DXA) were added separately to all models formulated in step 2. The areas under the receiver operating characteristic curve (AUCs) were calculated for the models formulated in all steps. Each cumulative AUC would be compared with the AUC yielded in the previous step to evaluate the incremental prediction value.

Results:

On top of SARC-F, assessing grip strength, walking speed, or 5-time chair-stand significantly increased the AUC for most adverse outcomes. In particular, assessing both grip strength and gait speed yielded the highest AUC in most prediction models (AUC = 0.539–0.770) and significantly increased the AUC for all outcomes except for recurrent falls. With both muscle strength and function assessed, adding ASM failed to significantly increase the AUC except for 2 conditions. In the 2 conditions, however, a higher height-adjusted ASM was associated with a higher risk of having worsened physical limitations [OR 1.25, 95% confidence interval (CI) 1.12–1.40] and decline in the physical quality of life (OR 1.18, 95% CI 1.06–1.33) in women.

Conclusions and Implications:

Assessing muscle strength and function provides additional power to predict adverse outcomes on top of SARC-F. Further assessment of muscle mass with DXA provides no extra constructive value in bettering the prediction regardless of the adjustment parameters. Alternative technologies to measure muscle mass might be required.

Sarcopenia was introduced to describe the loss of muscle mass and function. Receiving raising attention as a common disease in the older adult population, it was assigned an International Classification of Diseases, Tenth Revision, Clinical Modification code (M62.84) in 2016.¹ Although the definitions of sarcopenia that have been proposed vary widely, muscle strength, muscle function, and muscle mass were common components.^2–7 According to the guidelines, older adults often need to go through a series of testing for the diagnosis of sarcopenia. This usually involves starting off with an optional screening questionnaire (eg, SARC-F), followed by testing on muscle strength and function (eg, grip strength, chair-stand test, gait speed), and confirming the diagnosis with the assessment on muscle mass.^2–5,7,8

Based on these definitions, people who were diagnosed as having sarcopenia were found to possess a higher risk of having adverse outcomes, including falls and fractures,^9,10 physical limitations,¹¹ cardiometabolic diseases,¹² cognitive impairment,¹³ lower quality of life,¹⁴ hospitalization, and even mortality.^11,15–17 However, it remains unclear whether the significant prediction of sarcopenia to these adverse outcomes is predominantly mediated by poor muscle strength, poor muscle function, less muscle mass, or by all of the factors.^9,10,14

Several studies have separately investigated the ability of muscle strength and muscle mass on predicting various adverse outcomes. Although most studies showed that muscle strength is a consistent predictor of adverse outcomes,^15,18,19 the ability of muscle mass to predict adverse outcomes remains controversial.^9,15,20–22 Nevertheless, it has been reported that muscle strength measurements are better than muscle mass in predicting physical limitations^23,24 and mortality.¹⁵ In the recent international clinical practice guidelines for sarcopenia, there was strong evidence that grip strength and gait speed should be conducted in clinical practice for sarcopenia diagnosis whereas the evidence for dual-energy x-ray absorptiometry (DXA)–eassessed lean mass was only conditional.⁶ Whether muscle mass could provide significant value in the prediction of adverse outcomes on top of muscle strength and function is unclear.

As the population is aging, there is an increasing number of older adults who require screening for sarcopenia. Assessment of muscle strength and functions are relatively simple but muscle mass assessment requires hefty resources. It would be important to understand the additional value that could be offered by doing each additional assessment in the diagnosis process. Therefore, this study aims to use a stepwise model to examine the incremental value of sarcopenia components in predicting multiple adverse outcomes following the steps of the latest diagnosis algorithm formulated by the Asian Working Group for Sarcopenia (AWGS).⁷ With the insufficient number of studies comparing the value of different adjustment methods for muscle mass,⁷ this study also aims to compare the value of different adjustment iterations of muscle mass in predicting adverse outcomes.

Methods

Participants

Four thousand community-dwelling older men and women aged 65 years and older recruited for the Mr. OS and Ms. OS Hong Kong cohort study from August 2001 to December 2003 were included in the analysis. The sample was age-stratified so there is an approximately equal number of participants in the age range of 65 to 69, 70 to 74, and 75 or older. The participants were mainly recruited through posting advertisement in housing estates and local community centers. Individuals with the following characteristics were excluded from the study: (1) unable to walk independently, (2) had bilateral hip replacement, (3) had a reduced chance of survival in the 4 years following recruitment because of known medical conditions (eg, cancer, heart diseases, end-stage renal diseases, chronic lung diseases), and (4) inability to provide informed consent. The methodology has been previously described.²⁵ All surviving participants were invited to the second examination between August 2003 and December 2005, and to the third examination between August 2005 and November 2008. The assessment finding in the third examination constituted some of the adverse outcomes included for analysis in the present study. The study was approved by the clinical research ethics committee of the Chinese University of Hong Kong and informed consents were obtained from the participants.

Sarcopenia-Related Assessments

Each participant completed a series of assessments in 1 day in the first visit with a team of trained research assistants. Sarcopenia-assessments were picked according to the AWGS consensus.⁷

Sarcopenia screening

The 5-question SARC-F questionnaire was administered to the participants at baseline.²⁶

Assessments of muscle strength and functions

Grip strength was measured using a dynamometer (JAMAR Hand Dynamometer 5030JO; Sammons Preston, Bolingbrook, IL). Two readings were taken from the left and right side, respectively. The maximum value of the 4 readings was used for analysis.

Gait speed was measured by asking the participants to complete a 6-m walkway at a comfortable speed.

The 5-time chair-stand test has been proposed to be a measure of muscle function related to sarcopenia.^3,7 The time taken to complete 5 chair-stands as quickly as possible was recorded.

Assessment of muscle mass

DXA was used to measure lean mass.²⁷ The assessment was conducted by certified DXA operators, and standardized procedures for scanning were used to ensure measurement reproducibility. Total appendicular skeletal mass (ASM) was calculated as the sum of appendicular lean mass minus bone mineral content of the arms and legs, with the operator adjusting the cut lines of the limbs according to specific anatomical landmarks. Body height was measured by a Holtain Harpenden stadiometer (Holtain Ltd, Crosswell, UK) for the calculation of height-adjusted ASM. Bodyweight was measured to the nearest 0.1 kg with participants wearing a light gown, using the Physician Balance Beam Scale (Healthometer, McCook, IL). Weight-adjusted ASM (ASM/weight) was calculated. Body mass index (BMI)–eadjusted ASM was also calculated based on the height and weight measured.

Outcome Measures Assessed at the Fourth Year

Worsened physical limitations

Physical limitations at baseline and in the fourth visit were assessed using the following 2 questions: (1) Do you have any difficulty in climbing stairs (possible answers: no, a little, a lot)? and (2) Do you have any difficulty in carrying out the following household activities, such as moving chairs or tables (possible answers: no, a little, a lot)? Participants were categorized as having worsened physical limitations if the answer to either question was downgraded (eg, from no to a little, from no to a lot, or from a little to a lot).

Recurrent falls

Prospective fall at the fourth year was collected during a follow-up visit. The number of falls in the previous year was obtained in an interview. Recurrent fallers were those who had 2 or more falls during the 1-year period. A fall was defined as any unintentional rest on the ground resulting from a loss of balance.

Living in nursing

Living status was assessed in the fourth year, as all participants were lving in the community at baseline. The subjects were classified as (1) living alone; (2) living with relatives, friends, or maids; or (3) living in nursing home or old age home. Participants who lived in a nursing home or old age home at the fourth year were classified as having an adverse outcome.

Cumulative days of hospital stay

Length of hospital stay from baseline to the fourth year follow-up was obtained from the Hong Kong Hospital Authority records, which covered more than 93% of the hospitalizations in the Hong Kong population. The cutoff date for determining the length of hospital stay was September 30, 2008. A length of stay equal or more than 10 days was used as an outcome measure in the models.

Quality of life assessment

Health-related quality of life (HR-QOL) was evaluated by the 12-Item Short Form Health Survey questionnaire (SF-12), which derives summary scores from specific items from the 8 domains of the SF-36, with physical component summary score (summary of physical functioning, role-physical, bodily pain, and general health) and mental component summary score (summary of vitality, social functioning, role-emotional, and mental health).²⁸ SF-12 was assessed at baseline and at the fourth year, a decline of 5 units or more was used as an indicator of a decrease in quality of life. The cutoff score of 5 units was determined by using half of the standardized standard deviation according to the original US population, which is also close to the minimal important difference previously established.²⁹

Outcome Measures Assessed at the 10th Year

Fracture incidence

Fracture occurrence was determined by carrying out a search in the Hospital Authority electronic database, which includes all visits to Accident and Emergency Departments and outpatient clinics from baseline to October 2013. The database covers all publicly funded hospitals in Hong Kong. Hip fracture incidence and major osteoporotic fracture (MOF) incidence were recorded. A MOF was defined as a fracture of the hip, clinical spine, wrist, or humerus.

Mortality

Mortality data were ascertained from the Hong Kong Government Death Registry. The last search was undertaken on January 31, 2019. Mortality within 10 years since the baseline recruitment of each subject was used in this analysis.

Statistical Analysis

Statistical analysis was conducted using the statistical package SAS, version 9.4 (SAS Institute, Inc, Cary, NC). Baseline characteristics across genders were compared using the t test for continuous variables and chi-square test for categorical variables. Analyses were conducted separately for men and women considering that there were significant differences in muscle strength, function, and mass across sexes at baseline.

Logistic regression was conducted to evaluate the ability of each individual sarcopenia component in predicting adverse outcomes of interest. Odds ratios (ORs) are presented per standard deviation increase in the predictors’ value. Receiver operating characteristic (ROC) curves were constructed to evaluate the ability of each sarcopenia component to discriminate between people with or without the presence of the adverse outcomes in the follow-up. The area under the ROC curve (AUC) was used to measure the concordance of the predictive values with the actual outcomes.

The subsequent analysis framework follows the consensus recently published by the AWGS.⁷ In step 1, SARC-F was used as the only predictor. In step 2, muscle strength and function were added on top of SARC-F in 5 models: (1) grip strength, (2) gait speed, (3) 5-time chair-stand test, (4) grip strength and gait speed, and (5) grip strength and 5-time chair-stand test. In step 3, ASM measured by DXA, and adjusted by height-squared, weight, and BMI, were separately added to the 5 models formulated in the second step. AUCs were calculated for all models formulated in each step. The ORs of the predictors added in each step with adjustment of other predictors entered in previous steps were also calculated.

To evaluate the incremental value that could be provided by assessing each additional step of assessment, the cumulative AUCs formulated in each step would be compared to the one formulated in the previous step. The AUC comparisons were conducted using the Delong method approach, and the chi-square test of homogeneity was used. All statistical tests were 2-sided. A P value of less than .05 was considered statistically significant. For AUC comparisons, a P value of less than .10 was considered as marginally significant given the confirmative nature of the analysis.

Results

Characteristics of Participants

Two-thousand men and 2000 women were assessed at baseline. After 4 years, 1566 men (78.3%) and 1587 women (79.4%) returned for reassessment. At baseline, men had significantly better muscle strength and function, higher ASM, and fewer previous falls than women. The percentage of subjects suffering from the adverse outcomes was also significantly different across genders except for nursing home placement and hip fracture incidence (Table 1).

Table 1.

Participants’ Characteristics

	Men (n = 2000)	Women (n = 2000)	P Value*
Baseline
Age, mean (SD)	72.39 (5.01)	72.58 (5.36)	.25
SARC-F, mean (SD)	0.38 (0.84)	0.92 (1.35)	<.001
Grip strength, kg, mean (SD)	33.90 (6.73)	22.31 (4.41)	<.001
Gait speed, m/s, mean (SD)	1.07 (0.23)	0.96 (0.21)	<.001
Five-time chair stand, s, mean (SD)	12.65 (3.90)	13.42 (5.04)	<.001
Appendicular skeletal mass, mean (SD)
Height adjusted	7.20 (0.82)	6.06 (0.73)	<.001
Weight adjusted	0.31 (0.02)	0.26 (0.02)	<.001
BMI adjusted	0.82 (0.09)	0.58 (0.07)	<.001
Physical limitation, n, mean (SD)	361 (18.05%)	757 (37.85%)	<.001
Fall history in the past year, n			<.001
0	1693 (84.65%)	1518 (75.9%)
1	234 (11.7%)	320 (16%)
≥2	73 (3.65%)	162 (8.1%)
SF-12 PCS, mean (SD)	50.53 (7.59)	46.60 (8.76)	<.001
SP-12 MCS, mean (SD)	55.83 (6.77)	55.05 (7.76)	.001
No. of diseases, n			.56
0	345 (17.25%)	325 (16.25%)
1–2	1037 (51.85%)	1068 (53.4%)
≥3	618 (30.9%)	607 (30.35%)
No. of medications, n			.89
0	896 (44.8%)	905 (45.25%)
1–2	793 (39.65%)	795 (39.75%)
≥3	311 (15.55%)	300 (15%)
Follow-up	(n = 1566)	(n = 1587)
Physical limitations after 4 y	490 (31.29%)	871 (54.88%)	<.001
Worsening of physical limitations after 4 y	381 (24.33%)	630 (39.7%)	<.001
Recurrent falls in the fourth year	65 (4.15%)	103 (6.49%)	.004
Living in nursing home at the fourth year	35 (1.9%)	47 (2.46%)	.24
Days of hospital stay ≥10 after 4 y	546 (27.3%)	404 (20.21%)	<.001
SF-12 PCS decline ≥5 in 4 y	432 (27.59%)	513 (32.33%)	.004
SF-12 MCS decline ≥5 in 4 y	214 (13.67%)	271 (17.08%)	.008
Major osteoporotic fracture in 10 y	139 (6.95%)	236 (11.8%)	<.001
Hip fracture in 10 y	63 (3.15%)	69 (3.45%)	.60
Mortality in 10 y	490 (24.5%)	281 (14.05%)	<.001

MCS, mental component summary; PCS, physical component summary; SD, standard deviation.

Unless otherwise noted, values are n (%).

^*P value of t test or chi-square test.

Ability of Individual Sarcopenia-Related Assessment in Predicting Adverse Outcomes

SARC-F significantly predicts 5 adverse outcomes in men and 6 in women. The 3 muscle strength and performance related assessments predict 8 outcomes in men and 5 to 8 outcomes in women. Better muscle strength and functions were consistently associated with a lower chance of suffering from adverse outcomes (Table 2).

Table 2.

Ability of Individual Sarcopenia Component on Predicting Adverse Outcomes

	Worsened Physical Limitations After 4 y				Recurrent Falls in the 4 y				Living in Nursing Home at the 4 y
	Men (n = 1566)		Women (n = 1587)		Men (n = 1566)		Women (n = 1587)		Men (n = 1842)		Women (n = 1910)
	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC
SARC-F	1.26 (1.11, 1.44)	0.548	1.16 (1.03, 1.29)	0.540	1.52 (1.25, 1.85)	0.622	1.48 (1.25, 1.77)	0.644	1.28 (0.98, 1.66)	0.558	1.37 (1.08, 1.74)	0.590
Grip strength	0.68 (0.60, 0.77)	0.611	0.85 (0.76, 0.94)	0.544	0.80 (0.62, 1.04)	0.566	0.87 (0.71, 1.07)	0.542	0.43 (0.30, 0.61)	0.733	0.57 (0.42, 0.78)	0.659
Gait speed	0.60 (0.52, 0.68)	0.628	0.68 (0.61, 0.76)	0.607	0.67 (0.51, 0.88)	0.603	0.81 (0.66, 1.002)	0.538	0.46 (0.32, 0.66)	0.726	0.49 (0.37, 0.65)	0.709
Five-time chair-stand	1.36 (1.21, 1.53)	0.580	1.36 (1.21, 1.52)	0.597	1.51 (1.24, 1.84)	0.608	1.22 (1.03, 1.44)	0.517	1.35 (1.05, 1.72)	0.680	1.19 (0.95, 1.49)	0.598
Appendicular skeletal mass
a. Height adjusted	0.97 (0.86, 1.09)	0.511	1.22 (1.10, 1.35)	0.555	0.85 (0.66, 1.11)	0.554	0.98 (0.80, 1.21)	0.505	0.79 (0.55, 1.13)	0.553	0.67 (0.49, 0.92)	0.606
b. Weight adjusted	0.78 (0.69, 0.88)	0.575	0.93 (0.84, 1.03)	0.522	0.73 (0.56, 0.95)	0.583	0.76 (0.61, 0.95)	0.578	0.78 (0.55, 1.11)	0.566	1.25 (0.94, 1.66)	0.542
c. BMI adjusted	0.75 (0.67, 0.85)	0.578	0.86 (0.78, 0.96)	0.542	0.78 (0.60, 1.01)	0.570	0.72 (0.58, 0.90)	0.589	0.69 (0.48, 0.99)	0.607	1.03 (0.77, 1.39)	0.481
	Days of Hospital Stay ≥10 After 4 y				SF-12 PCS Decline ≥5 in 4 y				SF-12 MCS Decline ≥5 in 4 y
	Men (n = 2000)		Women (n = 2000)		Men (n = 1566)		Women (n = 1587)		Men (n = 1566)		Women (n = 1587)
	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	Yes (n = 35)	OR* (95% CI)	AUC
SARC-F	1.44 (1.30, 1.59)	0.566	1.42 (1.28, 1.58)	0.577	1.04 (0.91, 1.19)	0.513	1.01 (0.90, 1.13)	0.504	1.17 (1.001, 1.37)	0.534	1.27 (1.11, 1.44)	0.550
Grip strength	0.67 (0.61, 0.75)	0.613	0.78 (0.70, 0.88)	0.567	0.85 (0.75, 0.95)	0.547	0.91 (0.82, 1.02)	0.526	0.94 (0.81, 1.10)	0.527	0.83 (0.72, 0.95)	0.547
Gait speed	0.69 (0.62, 0.76)	0.606	0.66 (0.58, 0.74)	0.612	0.87 (0.77, 0.97)	0.542	0.88 (0.79, 0.98)	0.531	0.86 (0.74, 1.002)	0.541	0.86 (0.75, 0.98)	0.532
Five-time chair-stand	1.34 (1.21, 1.48)	0.588	1.23 (1.11, 1.35)	0.566	1.19 (1.06, 1.33)	0.550	1.10 (0.99, 1.22)	0.538	1.06 (0.92, 1.23)	0.514	1.18 (1.04, 1.33)	0.567
Appendicular skeletal mass
a. Height adjusted	0.90 (0.81, 0.99)	0.530	0.94 (0.84, 1.05)	0.516	1.01 (0.90, 1.14)	0.502	1.15 (1.03, 1.28)	0.535	0.89 (0.77, 1.03)	0.529	0.86 (0.75, 0.99)	0.536
b. Weight adjusted	0.91 (0.82, 1.004)	0.528	1.06 (0.95, 1.19)	0.508	0.99 (0.89, 1.11)	0.503	0.95 (0.85, 1.06)	0.516	0.93 (0.80, 1.08)	0.523	0.92 (0.80, 1.06)	0.525
c. BMI adjusted	0.85 (0.77, 0.94)	0.547	0.96 (0.86, 1.08)	0.516	0.94 (0.84, 1.05)	0.521	0.91 (0.82, 1.02)	0.521	0.95 (0.82, 1.10)	0.513	0.86 (0.75, 0.99)	0.539
	Major Osteoporotic Fracture in 10 y				Hip Fracture in 10 y				Mortality After 10 y
	Men (n = 2000)		Women (n = 2000)		Men (n = 2000)		Women (n = 2000)		Men (n = 2000)		Women (n = 2000)
	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	AUC	OR* (95% CI)	Yes (n = 35)	OR* (95% CI)	AUC
SARC-F	1.15 (0.99, 1.34)	0.548	1.07 (0.93, 1.23)	0.523	1.20 (0.98, 1.47)	0.561	1.24 (0.996, 1.54)	0.551	1.35 (1.22, 1.49)	0.559	1.12 (0.99, 1.27)	0.517
Grip strength	0.67 (0.56, 0.80)	0.612	0.82 (0.71, 0.94)	0.552	0.58 (0.44, 0.75)	0.651	0.67 (0.52, 0.86)	0.609	0.58 (0.52, 0.65)	0.645	0.71 (0.62, 0.81)	0.597
Gait speed	0.72 (0.60, 0.86)	0.590	0.79 (0.69, 0.91)	0.553	0.52 (0.39, 0.67)	0.681	0.63 (0.49, 0.80)	0.601	0.59 (0.53, 0.66)	0.638	0.72 (0.63, 0.82)	0.588
Five-time chair-stand	1.22 (1.05, 1.42)	0.568	1.12 (0.99, 1.26)	0.528	1.25 (1.02, 1.53)	0.572	1.19 (0.98, 1.44)	0.555	1.48 (1.33, 1.64)	0.612	1.19 (1.07, 1.33)	0.557
Appendicular skeletal mass
a. Height adjusted	0.80 (0.67, 0.95)	0.561	1.05 (0.92, 1.21)	0.519	0.67 (0.52, 0.87)	0.616	0.72 (0.55, 0.93)	0.577	0.76 (0.68, 0.84)	0.576	0.85 (0.74, 0.97)	0.542
b. Weight adjusted	1.02 (0.85, 1.21)	0.497	1.18 (1.03, 1.35)	0.534	1.14 (0.89, 1.47)	0.530	1.29 (1.02, 1.63)	0.530	0.87 (0.79, 0.97)	0.540	1.05 (0.92, 1.19)	0.492
c. BMI adjusted	0.95 (0.79, 1.13)	0.509	1.05 (0.91, 1.20)	0.508	1.01 (0.78, 1.30)	0.514	1.09 (0.86, 1.38)	0.504	0.88 (0.79, 0.98)	0.539	0.91 (0.80, 1.03)	0.539

OR, odd ratio; CI, confidence interval; MCS, mental component summary; PCS, physical component summary.

^*Crude OR (95% CI) presented as per standard deviation increase in the predictor.

Muscle mass–related predictors predict 3 to 4 outcomes in men and 3 to 6 outcomes in women. However, 2 significant predictions from height-adjusted ASM (worsened physical limitations, OR 1.22; SF-12 physical component summary decline, OR 1.15) and weight-adjusted ASM (major osteoporotic fractures, OR 1.18; hip fractures, OR 1.29) showed that higher ASM was associated with higher chances of having the adverse outcomes (Table 2).

Additional Prediction Value of Muscle Strength and Function (Step 2)

Adding grip strength to SARC-F (marginally) significantly increased the AUC of the prediction for all 9 outcomes in men and 4 outcomes in women. Adding gait speed increased the AUC for 6 and 4 outcomes in men and women, respectively. Adding 5-time chair-stand improved the prediction in 5 outcomes in both genders (Table 3). Having better muscle strength or functions was found to be less likely to have adverse outcomes after adjusting for SARC-F (Supplementary Table 1).

Table 3.

Cumulative Area Under the Receiver Operating Curve of SARC-F Plus Muscle Strength and/or Function on Predicting Adverse Outcomes (Step 1 and 2)

	Worsened Physical Limitations After 4 y		Recurrent Falls in the 4 y		Living in Nursing Home at the 4 y
	Men (n = 1566)	Women (n = 1587)	Men (n = 1566)	Women (n = 1587)	Men (n = 1842)	Women (n = 1910)
Step 1: SARC-F only	0.548	0.540	0.622	0.644	0.558	0.590
Step 2:
1. + Grip strength	0.622*	0.557	0.663†	0.640	0.741*	0.675*
2. + Gait speed	0.634*	0.608*	0.653	0.643	0.730*	0.711*
3. + Five-time chair-stand	0.598*	0.601*	0.652	0.639	0.685*	0.629*
4. + Grip strength and Gait speed	0.653*	0.608*	0.661	0.641	0.770*	0.739*
5. + Grip strength and Chair-stand	0.633*	0.601*	0.662	0.637	0.751*	0.679*
	Days of Hospital Stay ≥10 After 4 y		SF-12 PCS Decline ≥5 in 4 y		SF-12 MCS Decline ≥5 in 4 y
	Men (n = 2000)	Women (n = 2000)	Men (n = 1566)	Women (n = 1587)	Men (n = 1566)	Women (n = 1587)
Step 1: SARC-F only	0.566	0.577	0.513	0.504	0.534	0.550
Step 2:
1. + Grip strength	0.637*	0.597*	0.548†	0.527	0.560†	0.569
2. + Gait speed	0.626*	0.620*	0.542	0.536	0.562	0.562
3. + Five-time chair-stand	0.614*	0.591†	0.549†	0.535	0.542	0.578*
4. + Grip strength and Gait speed	0.650*	0.623*	0.554*	0.539†	0.563†	0.573†
5. + Grip strength and Chair-stand	0.647*	0.605*	0.562*	0.541†	0.559†	0.581*
	Major Osteoporotic Fracture in 10 y		Hip Fracture in 10 y		Mortality After 10 y
	Men (n = 2000)	Women (n = 2000)	Men (n = 2000)	Women (n = 2000)	Men (n = 2000)	Women (n = 2000)
Step 1: SARC-F only	0.548	0.523	0.561	0.551	0.559	0.517
Step 2:
1. + Grip strength	0.616*	0.553	0.660*	0.614*	0.656*	0.594*
2. + Gait speed	0.593†	0.554	0.681*	0.600	0.648*	0.588*
3. + Five-time chair-stand	0.572	0.529	0.580	0.559	0.632*	0.559*
4. + Grip strength and Gait speed	0.624*	0.570*	0.707*	0.633*	0.682*	0.617*
5. + Grip strength and Chair-stand	0.623*	0.556	0.665*	0.620*	0.679*	0.610*

MCS, mental component summary; PCS, physical component summary.

^*P < .05 compared with SARC-Feonly AUC.

^†P < .10 compared with SARC-Feonly AUC.

Comparing to SARC-F alone, adding both grip strength and gait speed (model 4) (marginally) significantly improved the prediction for all outcomes in both genders except for recurrent falls. Adding both grip strength and chair-stand (model 5) was also found to provide (marginally) significant additional prediction value on top of SARC-F for all outcomes except for recurrent falls in both genders and major osteoporotic fracture in women (Table 3).

Relative to using only 1 of the muscle strength or function assessments as predictor, combining grip strength with gait speed (model 4) or chair-stand (model 5) was shown to yield higher AUCs in 16 and 15 of the 18 analyses (9 outcomes × 2 genders), respectively. Combining grip strength with gait speed generally yielded a higher AUC than the combination of grip strength and chair-stand (Table 3).

Additional Prediction Value of Muscle Mass (Step 3)

In men, muscle mass was shown to add no significant prediction value on top of SARC-F+ muscle strength and/or function. The results were insignificant in all 15 combinations (5 models formulated in step 2 × 3 adjustment iterations of muscle mass) for all outcomes (Table 4). The associated logistic regression analysis is presented in Supplementary Table 2.

Table 4.

Cumulative Area Under the Receiver Operating Curve With the Addition of Appendicular Skeletal Mass to SARC-F Plus Muscle Mass and/or Physical Function in Men (Step 3)

	Worsened Physical Limitations After 4 y (n = 1566)			Recurrent Falls in the 4 y (n = 1566)			Living in Nursing Home at the 4 y (n = 1842)
	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM
1. SARC-F + Grip strength	0.622	0.631	0.627	0.669	0.672	0.659	0.741	0.742	0.744
2. SARC-F + Gait speed	0.634	0.644	0.644	0.664	0.671	0.665	0.724	0.733	0.741
3. SARC-F + Chair-stand	0.599	0.614	0.619	0.670	0.657	0.657	0.658	0.669	0.685
4. SARC-F + Grip strength + Gait speed	0.653	0.660	0.657	0.668	0.675	0.666	0.770	0.770	0.771
5. SARC-F + Grip strength + Chair-stand	0.633	0.645	0.640	0.674	0.663	0.661	0.751	0.750	0.752
	Days of Hospital Stay ≥10 After 4 y (n = 2000)			SF-12 PCS Decline ≥5 in 4 y (n = 1566)			SF-12 MCS Decline ≥5 in 4 y (n = 1566)
	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM
1. SARC-F + Grip strength	0.639	0.637	0.637	0.550	0.547	0.547	0.545	0.554	0.556
2. SARC-F + Gait speed	0.626	0.627	0.629	0.543	0.542	0.547	0.562	0.562	0.563
3. SARC-F + Chair-stand	0.614	0.613	0.618	0.550	0.549	0.552	0.547	0.548	0.552
4. SARC-F + Grip strength + Gait speed	0.651	0.650	0.650	0.556	0.554	0.554	0.562	0.562	0.563
5. SARC-F + Grip strength + Chair-stand	0.648	0.647	0.647	0.564	0.561	0.562	0.546	0.557	0.558
	Major Osteoporotic Fracture in 10 y (n = 2000)			Hip Fracture in 10 y (n = 2000)			Mortality After 10 y (n = 2000)
	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM
1. SARC-F + Grip strength	0.618	0.622	0.622	0.669	0.675	0.675	0.655	0.657	0.657
2. SARC-F + Gait speed	0.604	0.594	0.593	0.702	0.692	0.687	0.656	0.651	0.648
3. SARC-F + Chair-stand	0.592	0.579	0.570	0.641	0.612	0.586	0.643	0.634	0.635
4. SARC-F + Grip strength + Gait speed	0.624	0.628	0.627	0.710	0.720	0.717	0.683	0.682	0.683
5. SARC-F + Grip strength + Chair-stand	0.626	0.627	0.627	0.675	0.679	0.679	0.680	0.679	0.680

MCS, mental component summary; PCS, physical component summary.

^*P < .05 compared with the model without muscle mass.

^†P < .10 compared with the model without muscle mass.

In women, height-adjusted ASM was shown to provide significant additional prediction value in hip fracture (AUC increased from 0.600 to 0.652) and mortality (AUC increased from 0.588 to 0.609) in the SARC-F+ gait speed combination (model 2). BMI-adjusted ASM added marginally significant prediction ability for SF-12 mental component summary decline (AUC increased from 0.569 to 0.583) on top of the SARC-F+ grip strength combination (model 1) (Table 5).

Table 5.

Cumulative Area Under the Receiver Operating Curve With the Addition of Appendicular Skeletal Mass to SARC-F Plus Muscle Mass and/or Physical Function in Women (Step 3)

	Worsened Physical Limitations After 4 y (n = 1587)			Recurrent Falls in the 4 y (n = 1587)			Living in Nursing Home at the 4 y (n = 1910)
	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM
1. SARC-F + Grip strength	0.590*	0.561	0.566	0.639	0.651	0.658	0.689	0.700	0.689
2. SARC-F + Gait speed	0.622*	0.608	0.611	0.642	0.652	0.657	0.742	0.722	0.709
3. SARC-F + Chair-stand	0.616	0.600	0.605	0.638	0.651	0.656	0.664	0.624	0.614
4. SARC-F + Grip strength + Gait speed	0.625*	0.609	0.611	0.640	0.652	0.659	0.752	0.753	0.745
5. SARC-F + Grip strength + Chair-stand	0.620†	0.602	0.606	0.636	0.651	0.658	0.690	0.702	0.690
	Days of Hospital Stay ≥10 After 4 y (n = 2000)			SF-12 PCS Decline ≥5 in 4 y (n = 1587)			SF-12 MCS Decline ≥5 in 4 y (n = 1587)
	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM
1. SARC-F + Grip strength	0.598	0.599	0.597	0.553	0.532	0.531	0.575	0.575	0.583†
2. SARC-F + Gait speed	0.624	0.622	0.620	0.556†	0.539	0.541	0.569	0.569	0.581
3. SARC-F + Chair-stand Gait speed	0.599	0.590	0.591	0.550	0.536	0.537	0.576	0.583	0.589
4. SARC-F + Grip strength +	0.625	0.626	0.624	0.565*	0.543	0.544	0.577	0.579	0.584
5. SARC-F + Grip strength + Chair-stand	0.606	0.606	0.604	0.561	0.544	0.544	0.584	0.586	0.592
	Major Osteoporotic Fracture in 10 y (n = 2000)			Hip Fracture in 10 y (n = 2000)			Mortality After 10 y (n = 2000)
	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM	+ Height-Adjusted ASM	+ Weight-Adjusted ASM	+ BMI-Adjusted ASM
1. SARC-F + Grip strength	0.561	0.566	0.557	0.629	0.631	0.629	0.597	0.601	0.595
2. SARC-F + Gait speed	0.555	0.573	0.556	0.652*	0.626	0.613	0.609*	0.589	0.590
3. SARC-F + Chair-stand	0.523	0.529	0.519	0.613	0.546	0.552	0.577	0.556	0.564
4. SARC-F + Grip strength + Gait speed	0.571	0.581	0.571	0.662	0.649	0.647	0.622	0.621	0.618
5. SARC-F + Grip strength + Chair-stand	0.562	0.563	0.556	0.636	0.628	0.630	0.610	0.613	0.610

MCS, mental component summary; PCS, physical component summary.

^*P < .05 compared with the model without muscle mass.

^†P < .10 compared with the model without muscle mass.

Although height-adjusted ASM was also shown to provide significant additional prediction value on worsened physical limitation and SF-12 decline (Table 5), results from the logistic regression analyses showed that an increase in height-adjusted ASM significantly increased the risk of having these 2 adverse events (worsened physical limitation: OR 1.19–1.30; SF-12 decline: OR 1.14–1.20) (Supplementary Table 3). Therefore, none of the muscle mass iterations provide a constructive increase in AUC relative to SARC-F+ muscle strength and function (models 4 and 5).

Discussion

SARC-F, muscle strength, and functions could predict most included adverse outcomes in community-dwelling older people. Muscle mass was found to be less predictive, with some of the predictions going in an unfavorable direction (ie, high muscle mass leading to high chances of suffering from adverse outcomes). The results from the present study were in accordance with previous observations, which suggest that muscle strength is better than muscle mass in predicting adverse outcomes.^9,15,20,24

Our stepwise analysis showed that grip strength, walking speed, and 5-time chair-stand could all add significant predictive power on top of SARC-F. Assessment of both muscle strength and functions, particularly grip strength and gait speed, is preferred. It also provided direct evidence that ASM assessed by DXA offers little additional value on top of SARC-F, muscle strength, and function in predicting adverse outcomes. It could be postulated that the significant findings showing that sarcopenia predicts adverse outcomes were largely attributed to the muscle strength and function components.^{9–11,14–17}

Our finding validated the advocate of the European consensus that low muscle strength should be a principal determinant of sarcopenia in an Asian population.³ Moreover, it raised concern about whether height-adjusted ASM, the most commonly used assessment of muscle mass in the diagnosis of sarcopenia,⁸ is an appropriate measurement to be taken during the diagnosis of sarcopenia in view of adverse outcomes prediction. This is alarming when a higher height-adjusted ASM was found to be associated with a higher chance of suffering from worsened physical limitations and decline in the physical aspect of quality of life in older women. The current practice of including low height-adjusted ASM as a component to diagnose sarcopenia might indeed compromise, rather than improve, the prediction of some adverse outcomes in older women.

We have previously reported that height-adjusted ASM follows a U-shaped relationship with physical limitations.³⁰ For women, a large part of the association between height-adjusted ASM and physical limitations follows a positive relationship indeed. A large cohort conducted in Canada has also reported that height-adjusted ASM is inversely associated with physical performance.³¹ A potential explanation is that muscle mass increases with weight, BMI, and fat mass.³² The apparently negative association between height-adjusted ASM and physical limitations could hence be modulated by body size or fat mass.³¹ This is supported by our further analyses in the present study showing that the prediction of ASM on worsened physical limitations in women becomes insignificant when it is adjusted by weight or BMI instead (Supplementary Table 3). Furthermore, higher weight- and BMI-adjusted, but not height-adjusted, ASM were found to be significant protective factors to worsened physical limitation in the logistic regression analysis after adjusting for SARC-F, muscle strength, and muscle function in men (Supplementary Table 2). Similar findings were observed by Janssen et al,³³ which showed that muscle mass assessed by bioelectrical impedance analysis (BIA) was associated with functional impairment and physical disability after adjusting for body weight. Weight-adjusted ASM was also shown to be better than height-adjusted ASM in predicting physical limitations.^33,34 The adjustment method of muscle mass appears to have great impact on the prediction ability, which might even reverse the prediction direction.

Another example that demonstrated the importance of adjustment method for muscle mass was shown in the analysis of hip fracture in women. After adjusting for SARC-F and muscle strength and/or function, height-adjusted ASM was a protective factor for hip fracture whereas weight-adjusted ASM was a risk factor (Supplementary Table 3). This could be, again, due to the positive correlation between ASM and weight,³² whereas lower weight is a known risk factor for hip fractures.³⁵ Given that height-adjusted and weight-adjusted ASM predict some of the adverse outcomes in an unfavorable direction, BMI-adjusted ASM might have an edge to be adopted as a diagnostic criterion instead, as it only yielded significant prediction to adverse outcomes in a favorable direction.

Nevertheless, regardless of the iterations of ASM adopted, they provided limited additional value on top of muscle strength and function assessments in predicting all adverse outcomes. The reason for the finding could lie within the limitations of DXA technology. Although DXA is one of the most recommended measurement tool for assessing muscle mass at present,^2,3,8 it actually measures appendicular lean mass, which also includes fibrotic and connective tissue and water.³⁶ It also could not separate fat infiltration from muscle in the measurement of lean mass.³⁷ A recent study has demonstrated that muscle mass determined by D₃-creatine dilution is associated with physical performance, falls, and mobility limitation in older men.³⁶ Further studies using other technologies to measure muscle mass should be conducted.

Limitation

We have included only individuals who can walk and come to our center for assessment. Our participants represent a healthier sample of the community population in Hong Kong, particularly for outcomes assessed in the fourth year. The findings have to be confirmed in other cohorts with different ethnicities and lifestyles. Some outcomes have few people-reported occurrences (ie, recurrent falls, living in a nursing home, hip fracture), which compromised the power of detecting significance. We do not have data for some of the assessments listed in the AWGS guideline for a fully comprehensive analysis (ie, calf circumference, short physical performance battery).

Conclusions and Implications

With SARC-F adopted as a screening tool, further assessments of muscle strength and function provide additional power to predict adverse outcomes. Assessing muscle mass with DXA on top of SARC-F, muscle strength, and function provides no extra constructive value in bettering the prediction of adverse outcomes regardless of the adjustment parameters. Alternative technologies to measure muscle mass might be needed for the diagnosis of sarcopenia in view of adverse outcome prediction.