Leisure-time physical activity modulates associations of sarcopenia and functional disability with mortality in older adults

Summary

Whether physical activity modifies the associations of sarcopenia and basic/instrumental activities of daily living (ADL/IADL) disability with mortality remains unclear. The study included 64,146 participants from three nationally representative cohorts: Health and Retirement Study (HRS), Survey of Health, Aging and Retirement in Europe (SHARE), and China Health and Retirement Longitudinal Study (CHARLS). In HRS, the greater risk of mortality associated with each dysfunction was more pronounced in inactive participants, with multivariable-adjusted hazard ratios (HRs) (95% confidence intervals [CIs]) in Cox models of 3.88 (3.57–4.23) for sarcopenia, 1.82 (1.70–1.95) for ADL disability, and 1.95 (1.81–2.09) for IADL disability, while the corresponding HRs (95% CIs) were 3.32 (2.92–3.77), 1.39 (1.20–1.61), and 1.44 (1.26–1.65) in regularly active participants (all P for multiplicative interaction < 0.010). Consistent interaction patterns were observed in SHARE and CHARLS. This study revealed that regular activity meeting the WHO recommendations significantly mitigated the hazards of mortality associated with sarcopenia and ADL/IADL disability.

Graphical abstract

Highlights

• •Physical activity mitigates the mortality hazards associated with functional disability
• •Sarcopenia or functional disability gains more survival benefits from regular activity
• •This interaction pattern remained for the majority of components of functional disability

Health sciences

Introduction

Compelling evidence has associated physical activity with reduced risk of all-cause mortality.¹ Current guidelines encourage the general population to incorporate physical exercise into daily routines,^2,3,4,5,6 even for older adults with cardiovascular disease or disabled physical function.^2,3,4,5 The World Health Organization (WHO) guidelines recommend that for all adults, physical activity should amount to at least 150 min of moderate-intensity exercise, 75 min of vigorous-intensity exercise, or an equivalent combination each week.⁶ However, nearly one-third of adults worldwide were inactive, and the proportion of inactivity was even higher in older adults with cardiovascular diseases or disabled physical functions, approximately more than 50% or 62%.^5,7,8

Previous evidence has suggested that physical activity confers survival benefits both in primary and secondary cardiovascular disease prevention in middle-aged and elderly adults,^9,10 highlighting that adults with cardiovascular disease may benefit to a greater extent than their counterparts without cardiovascular disease.¹⁰ By contrast, refined evidence on the benefits of physical activity for older adults with reduced physical function is limited, restricting the promotion of physical activity in healthy aging. Older adults are susceptible to prolonged sedentary lifestyles and physical inactivity,^8,11 which can create viciously cycled with declined physical functional capacity including sarcopenia and basic/instrumental activities of daily living (ADL/IADL) disability,^12,13 thereby accelerating subsequent mortality risk.¹⁴ Notably, sarcopenia and ADL/IADL disability represent early and actionable dimensions of age-related functional decline.^15,16 Sarcopenia typically precedes and predicts the onset of frailty, while ADL/IADL disability directly threatens independent living,¹⁷ making both critical targets for clinical and public health intervention.¹⁶ Nevertheless, whether and how physical activity interacts with sarcopenia and ADL/IADL disability on mortality risk in older adults remains unknown. Delineating this interaction pattern could deepen our understanding of the beneficial association of physical activity in these high-risk people and improve evidence-based prevention strategies to foster healthy aging and ameliorate economic burdens.^18,19

To this end, we investigated the interaction of physical activity with sarcopenia and ADL/IADL disability on all-cause mortality risk in older adults, based on three large-scale nationally representative cohorts from 17 high- and middle-income countries across the United States, China, 14 European countries, and Israel. We specifically elucidated whether and in what pattern the associations of sarcopenia, ADL disability, and IADL disability with all-cause mortality risk are modified by physical activity, and whether the survival benefit of physical activity varies among adults with and without these declined physical functional capacities.

Results

Baseline characteristics

A total of 64,146 participants were included in this study. The baseline characteristics of participants in the Health and Retirement Study (HRS), the Survey of Health, Aging and Retirement in Europe (SHARE), and the China Health and Retirement Longitudinal Study (CHARLS) are presented in Table 1. The median ages of the participants in HRS, SHARE, and CHARLS ranged from 58.0 to 66.0 years, and more than half of them were women. The proportion of participants with regular activity was 48.3% in HRS, 47.1% in SHARE, and 56.4% in CHARLS. Across three cohorts, the proportion of confirmed sarcopenia ranged from 8.7% to 16.7%, the proportion of ADL disability ranged from 10.4% to 20.0%, and the proportion of IADL disability ranged from 13.3% to 25.5%. Baseline characteristics of participants in the three cohorts by physical activity levels are shown in Table S1. Proportions of sarcopenia and ADL/IADL components were higher in the inactivity group compared with the regular activity group (Figure S2). The significant but weak pairwise correlations between physical activity, sarcopenia, ADL disability, and IADL disability suggested no evident collinearity across these variables (Table S2).

Table 1.

Baseline characteristics of 64,146 participants in HRS, SHARE, and CHARLS cohorts

Characteristic	HRS cohort n = 20,061	SHARE cohort n = 30,758	CHARLS cohort n = 13,327
Follow-up years of study	2010–2021	2006–2020	2011–2020
Total person-years	154,616	268,437	112,979
Age, median (IQR), y	66.0 (57.0–75.0)	64.0 (57.0–72.0)	58.0 (52.0–65.0)
Sex, no. (%)
Men	8,539 (42.6)	13,933 (45.3)	6,418 (48.2)
Women	11,522 (57.4)	16,825 (54.7)	6,909 (51.8)
Country	USA	14 European countries and Israel	China
Race/ethnicity, no. (%)
Non-Hispanic White	14,615 (72.9)	NA	NA
Non-Hispanic Black	3,880 (19.3)	NA	NA
Hispanic	967 (4.82)	NA	NA
Other race or ethnicity	599 (2.99)	NA	NA
Residence location, no. (%)
Rural	5,697 (28.4)	8,383 (27.3)	8,441 (63.3)
Urban	14,364 (71.6)	22,375 (72.7)	4,886 (36.7)
Education level, no. (%)
Below high school	3,961 (19.7)	14,993 (48.7)	11,832 (88.8)
High school	11,733 (58.5)	9,760 (31.7)	1,286 (9.65)
College or above	4,367 (21.8)	6,005 (19.5)	209 (1.57)
Work status, no. (%)
Never worked or retired	12,204 (60.8)	20,881 (67.9)	5,022 (37.7)
Employed	7,857 (39.2)	9,877 (32.1)	8,305 (62.3)
Annual household PCI, no. (%)
Low	9,957 (49.6)	15,255 (49.6)	6,593 (49.5)
High	10,104 (50.4)	15,503 (50.4)	6,734 (50.5)
Cigarettes smoking, no. (%)
Never	8,661 (43.2)	16,141 (52.5)	8,014 (60.1)
Former	8,366 (41.7)	8,733 (28.4)	1,185 (8.89)
Current	3,034 (15.1)	5,884 (19.1)	4,128 (31.0)
Alcohol drinking, no. (%)
Never	9,022 (45.0)	4,980 (16.2)	7,794 (58.5)
Former	9,112 (45.4)	8,157 (26.5)	2,985 (22.4)
Current	1,927 (9.61)	17,621 (57.3)	2,548 (19.1)
Physical activity, no. (%)
Inactivity	10,373 (51.7)	16,257 (52.9)	5,812 (43.6)
Regular activity	9,688 (48.3)	14,501 (47.1)	7,515 (56.4)
Sarcopenia status, no. (%)
No sarcopenia	14,635 (81.3)	21,803 (76.6)	7,305 (57.6)
Probable sarcopenia	886 (4.92)	4,191 (14.7)	3,268 (25.8)
Confirmed sarcopenia	2,491 (13.8)	2,467 (8.67)	2,114 (16.7)
ADL, no. (%)
Normal	16,049 (80.0)	27,550 (89.6)	11,208 (84.1)
Disabled	4,012 (20.0)	3,208 (10.4)	2,118 (15.9)
IADL, no. (%)
Normal	13,442 (74.5)	26,670 (86.7)	10,591 (79.5)
Disabled	4,609 (25.5)	4,088 (13.3)	2,736 (20.5)
Death, no. (%)
No	14,091 (70.2)	23,752 (77.2)	12,095 (90.8)
Yes	5,970 (29.8)	7,006 (22.8)	1,232 (9.24)
Heart disease, no. (%)
No	15,449 (77.0)	26,071 (84.8)	11,724 (88.0)
Yes	4,612 (23.0)	4,687 (15.2)	1,603 (12.0)
Stroke, no. (%)
No	18,313 (91.3)	29,290 (95.2)	13,008 (97.6)
Yes	1,748 (8.71)	1,468 (4.77)	319 (2.39)
Cancer, no. (%)
No	17,273 (86.1)	28,828 (93.7)	13,212 (99.1)
Yes	2,788 (13.9)	1,930 (6.27)	115 (0.86)
Diabetes, no. (%)
No	15,653 (78.0)	26,960 (87.7)	12,525 (94.0)
Yes	4,408 (22.0)	3,798 (12.3)	802 (6.02)
Hypertension, no. (%)
No	8,431 (42.0)	18,537 (60.3)	9,888 (74.2)
Yes	11,630 (58.0)	12,221 (39.7)	3,439 (25.8)
Lung disease, no. (%)
No	18,178 (90.6)	28,769 (93.5)	12,067 (90.5)
Yes	1,883 (9.39)	1,989 (6.47)	1,260 (9.45)

The number of missing values for sarcopenia was 2,049 in HRS cohort, 2,297 in SHARE cohort, and 640 in CHARLS cohort; the number of missing values for IADL was 2,010 in HRS cohort; and the number of missing values for ADL was 1 in CHARLS cohort. Definitions for each longitudinal cohort study and variable of physical activity, sarcopenia, and ADL/IADL are shown in Methods S1–S6.

CHARLS, China Health and Retirement Longitudinal Study; HRS, Health and Retirement Study; NA, not applicable; PCI, per capita income; SHARE, Survey of Health, Aging and Retirement in Europe.

Associations of sarcopenia and ADL/IADL disability with mortality were modulated by physical activity levels

During a median follow-up of 9.3 (interquartile range [IQR], 5.8–9.8) years in HRS, 10.0 (5.5–12.7) years in SHARE, and 9.0 (8.9–9.0) years in CHARLS, 5,970, 7,006, and 1,232 participants died in the three cohorts, respectively. Across three cohorts, sarcopenia, ADL disability, and IADL disability were individually associated with increased risks of mortality as compared with their reference counterparts, with multivariable-adjusted hazard ratios (HRs) of mortality ranging from 2.82 (95% confidence interval [CI]: 2.65–3.01) to 3.76 (3.52–4.01) associated with confirmed sarcopenia, from 1.69 (1.49–1.91) to 2.23 (2.10–2.37) associated with ADL disability, and from 1.71 (1.52–1.94) to 2.44 (2.31–2.59) associated with IADL disability (Tables S3–S5).

In HRS, SHARE, and CHARLS, the risk for all-cause mortality associated with sarcopenia or ADL/IADL disability was more pronounced in inactive participants than in regularly active participants (P for interaction ≤ 0.031 in three cohorts; Table 2). Specifically, in HRS, compared with no sarcopenia, the multivariable-adjusted HRs (95% CIs) of mortality associated with probable sarcopenia and confirmed sarcopenia were 1.81 (1.61–2.03) and 3.88 (3.57–4.23), respectively, in inactive participants, while the corresponding HRs (95% CIs) were 2.34 (1.89–2.90) and 3.32 (2.92–3.77) in regularly active participants (P for interaction = 0.001). Compared with normal ability, ADL disability was associated with a greater risk of mortality in inactive participants (multivariable-adjusted HR: 1.82; 95% CI: 1.70–1.95) than in regularly active participants (1.39; 1.20–1.61, P for interaction = 0.010). Similarly, the HRs (95% CIs) for mortality associated with IADL disability versus normal ability were 1.95 (1.81–2.09) in inactive participants and 1.44 (1.26–1.65) in regularly active participants (P for interaction = 0.003). Consistent interaction patterns were also observed in SHARE (all P for interaction ≤ 0.031) and CHARLS (all P for interaction ≤ 0.022).

Table 2.

Association of sarcopenia and ADL/IADL with all-cause mortality risk in older participants stratified by physical activity levels in HRS, SHARE, and CHARLS cohorts

Cohort	Inactivity			Regular activity			Pinteraction
	No. of death/total participants	Total person-years	HR (95% CI)	No. of death/total participants	Total person-years	HR (95% CI)
HRS cohort
Sarcopenia status (n = 18,012)
No sarcopenia	1,657/6,674	54,971	1.00 (Ref.)	1,018/7,961	68,909	1.00 (Ref.)	0.001
Probable sarcopenia	340/644	4,501	1.81 (1.61–2.03)	93/242	1,900	2.34 (1.89–2.90)
Confirmed sarcopenia	1,378/1,698	7,942	3.88 (3.57–4.23)	429/793	5,297	3.32 (2.92–3.77)
ADL (n = 20,061)
Normal	2,310/7,183	55,213	1.00 (Ref.)	1,464/8,866	74,200	1.00 (Ref.)	0.010
Disabled	1,929/3,190	18,793	1.82 (1.70–1.95)	267/822	6,410	1.39 (1.20–1.61)
IADL (n = 18,051)
Normal	1,672/5,759	45,039	1.00 (Ref.)	1,149/7,683	64,910	1.00 (Ref.)	0.003
Disabled	1,911/3,369	20,923	1.95 (1.81–2.09)	357/1,240	9,751	1.44 (1.26–1.65)
SHARE cohort
Sarcopenia status (n = 28,461)
No sarcopenia	2,335/10,184	89,712	1.00 (Ref.)	1,402/11,619	109,146	1.00 (Ref.)	0.031
Probable sarcopenia	884/2,673	21,946	1.22 (1.13–1.33)	241/1,518	14,060	1.14 (0.99–1.32)
Confirmed sarcopenia	1,152/1,831	11,313	2.70 (2.50–2.93)	230/636	4,936	2.48 (2.11–2.91)
ADL (n = 30,758)
Normal	3,562/13,570	116,818	1.00 (Ref.)	1,849/13,980	129,575	1.00 (Ref.)	0.002
Disabled	1,462/2,687	17,581	2.12 (1.98–2.27)	133/521	4,463	1.47 (1.19–1.81)
IADL (n = 30,758)
Normal	3,219/13,020	113,058	1.00 (Ref.)	1,762/13,650	126,847	1.00 (Ref.)	0.002
Disabled	1,805/3,237	21,341	2.34 (2.19–2.50)	220/851	7,191	1.88 (1.61–2.18)
CHARLS cohort
Sarcopenia status (n = 12,687)
No sarcopenia	153/2,695	23,148	1.00 (Ref.)	181/4,610	40,159	1.00 (Ref.)	0.022
Probable sarcopenia	200/1,598	13,364	1.71 (1.38–2.12)	111/1,670	14,386	1.51 (1.19–1.93)
Confirmed sarcopenia	321/1,130	8,844	3.39 (2.74–4.19)	153/984	8,207	2.47 (1.93–3.15)
ADL (n = 13,326)
Normal	468/4,637	38,962	1.00 (Ref.)	382/6,571	56,705	1.00 (Ref.)	<0.001
Disabled	305/1,174	9,220	1.92 (1.65–2.24)	76/944	8,090	1.07 (0.83–1.38)
IADL (n = 13,327)
Normal	419/4,321	36,362	1.00 (Ref.)	356/6,270	54,127	1.00 (Ref.)	<0.001
Disabled	355/1,491	11,821	1.88 (1.62–2.19)	102/1,245	10,668	1.19 (0.94–1.50)

The number of missing values for sarcopenia was 2,049 in HRS cohort, 2,297 in SHARE cohort, and 640 in CHARLS cohort; the number of missing values for IADL was 2,010 in HRS cohort; and the number of missing values for ADL was 1 in CHARLS cohort. HRs (95% CIs) were adjusted for age, sex, race (only for HRS cohort), residence location, education level, work status, annual household income per capita, alcohol drinking, cigarette smoking, heart disease, stroke, cancer, diabetes, hypertension, and lung disease. Physical activity was categorized as inactivity (<150 min/week of moderate activity and <75 min/week of vigorous activity, and physical activity index < 9) and regular activity (≥150 min/week of moderate activity or ≥75 min/week of vigorous activity, or physical activity index ≥ 9), based on the frequency and duration of physical activity performed by participants. Definitions for each longitudinal cohort study and variable of physical activity, sarcopenia, and ADL/IADL are shown in Methods S1–S6.

ADL, activities of daily living; CHARLS, China Health and Retirement Longitudinal Study; CI, confidence interval; HR, hazard ratio; HRS, Health and Retirement Study; IADL, instrumental activities of daily living; Ref., reference; SHARE, Survey of Health, Aging and Retirement in Europe.

There were also significant positive additive interactions of confirmed sarcopenia or ADL/IADL disability with physical inactivity on all-cause mortality risk (Table 3). In HRS, for confirmed sarcopenia and physical inactivity, the relative excess risk due to interaction (RERI) was 2.26 (95% CI: 1.83, 2.71), the attributable proportion due to interaction (AP) was 0.39 (0.33, 0.45), and the synergy index (SI) was 1.90 (1.66, 2.17); for ADL disability and physical inactivity, the RERI was 0.82 (95% CI: 0.58, 1.06), the AP was 0.28 (0.20, 0.35), and the SI was 1.72 (1.42, 2.09); for IADL disability and physical inactivity, the RERI was 0.95 (0.71, 1.18), the AP was 0.30 (0.23, 0.37), and the SI was 1.81 (1.51, 2.17). Such significant additive interactions were also consistently observed in SHARE and CHARLS.

Table 3.

Additive interactions of sarcopenia and ADL/IADL with physical activity on mortality in older adults in HRS, SHARE, and CHARLS cohorts

	RERI	AP	SI
HRS cohort
Confirmed sarcopenia and inactivity	2.26 (1.83, 2.71)	0.39 (0.33, 0.45)	1.90 (1.66, 2.17)
ADL disability and inactivity	0.82 (0.58, 1.06)	0.28 (0.20, 0.35)	1.72 (1.42, 2.09)
IADL disability and inactivity	0.95 (0.71, 1.18)	0.30 (0.23, 0.37)	1.81 (1.51, 2.17)
SHARE cohort
Confirmed sarcopenia and inactivity	1.27 (0.89, 1.64)	0.31 (0.23, 0.39)	1.72 (1.43, 2.07)
ADL disability and inactivity	1.22 (0.89, 1.54)	0.37 (0.27, 0.45)	2.11 (1.62, 2.73)
IADL disability and inactivity	1.23 (0.91, 1.54)	0.34 (0.26, 0.41)	1.88 (1.55, 2.29)
CHARLS cohort
Confirmed sarcopenia and inactivity	0.91 (0.37, 1.45)	0.27 (0.12, 0.40)	1.64 (1.19, 2.27)
ADL disability and inactivity	1.22 (0.82, 1.63)	0.48 (0.34, 0.59)	4.81 (1.74, 13.30)
IADL disability and inactivity	1.08 (0.70, 1.47)	0.43 (0.29, 0.54)	3.49 (1.66, 7.34)

Additive interactions were calculated based on the reference group without confirmed sarcopenia or ADL/IADL disability and with physical activity. HRs (95% CIs) were adjusted for age, sex, race (only for HRS cohort), residence location, education level, work status, annual household income per capita, alcohol drinking, cigarette smoking, heart disease, stroke, cancer, diabetes, hypertension, and lung disease.

ADL, activities of daily living; AP, attributable proportion due to interaction; CHARLS, China Health and Retirement Longitudinal Study; CI, confidence interval; HR, hazard ratio; HRS, Health and Retirement Study; IADL, instrumental activities of daily living; Ref., reference; RERI, relative excess risk due to interaction; SHARE, Survey of Health, Aging and Retirement in Europe; SI, synergy index.

In sensitivity analysis, after excluding participants with a history of cancer or cardiovascular disease, the interaction patterns (i.e., the positive associations of sarcopenia, ADL disability, and IADL disability with mortality risk were greater in inactive participants than in active participants) persisted across the three cohorts (Table S6). These results remained consistent when further excluding participants diagnosed with cancer during the first follow-up survey (Table S7). When ADL/IADL disability was categorized by severity into three groups, the interaction patterns remained consistent (Table S8). Additional sensitivity analyses confirmed the robustness of the primary interaction patterns under various conditions: using physical activity data without imputation in CHARLS (Table S9), restricting to participants aged ≥ 50 years in CHARLS (Table S10), adjusting for marital status and psychiatric disorders across three cohorts (Table S11), applying multiple imputation for missing covariates across three cohorts (Table S12), and excluding deaths occurring within the first year of follow-up across three cohorts (Table S13).

Similarly, all components of sarcopenia and ADL/IADL disability were associated with increased risks of mortality (Tables S3–S5), and most of the positive associations of sarcopenia and ADL/IADL components with mortality were attenuated with regular activity in HRS, SHARE, and CHARLS (Figure 1). In particular, sarcopenia components such as low muscle mass and low performance, ADL components such as having difficulties in dressing, bathing, eating, getting out of bed, and toileting, and IADL components such as having difficulties in cooking, shopping, taking medications, managing money, using a map, and doing household chores showed significant interactions with physical activity on mortality in at least one cohort (all P for interaction ≤ 0.045).

Figure 1.

Association of sarcopenia components and ADL/IADL components with all-cause mortality risk in older participants stratified by physical activity levels in HRS, SHARE, and CHARLS cohorts

HRs (95% CIs) were adjusted for age, sex, race (only for HRS cohort), residence location, education level, work status, annual household income per capita, alcohol drinking, cigarette smoking, heart disease, stroke, cancer, diabetes, hypertension, and lung disease. Physical activity was categorized as inactivity (<150 min/week of moderate activity and <75 min/week of vigorous activity, and physical activity index < 9) and regular activity (≥150 min/week of moderate activity or ≥75 min/week of vigorous activity, or physical activity index ≥ 9), based on the frequency and duration of physical activity performed by participants. Definitions for each longitudinal cohort study and variable of physical activity, sarcopenia, and ADL/IADL are shown in Methods S1–S6.

ADL, activities of daily living; CHARLS, China Health and Retirement Longitudinal Study; CI, confidence interval; HR, hazard ratio; HRS, Health and Retirement Study; IADL, instrumental activities of daily living; NA, not applicable; SHARE, Survey of Health, Aging and Retirement in Europe.

The survival benefit of regular activity was more evident in adults with sarcopenia and ADL/IADL disability than in normal adults

Regular physical activity was associated with a decreased risk of mortality compared with inactivity, with multivariable-adjusted HRs (95% CIs) of 0.53 (0.50–0.56) in HRS, 0.56 (0.53–0.59) in SHARE, and 0.64 (0.57–0.73) in CHARLS (Tables S3–S5). The decreased risk for all-cause mortality associated with regular activity was more evident in participants with sarcopenia or ADL/IADL disability than in their reference counterparts (Figure 2). In HRS, the multivariable-adjusted HRs (95% CIs) of mortality associated with regular activity compared with inactivity were 0.66 (0.61–0.72) in participants with no sarcopenia but were 0.50 (0.45–0.56) in participants with confirmed sarcopenia. The association between regular activity and decreased risk of mortality was more pronounced in disabled ADL participants (multivariable-adjusted HR: 0.49; 95% CI: 0.43–0.56) than in normal ability participants (0.62; 0.58–0.67); similar association patterns were also observed by IADL disability status. Consistent interaction patterns were also observed in SHARE and CHARLS.

Figure 2.

Association of physical activity with all-cause mortality risk in older participants with or without sarcopenia and ADL/IADL disability in HRS, SHARE, and CHARLS cohorts

HRs (95% CIs) were adjusted for age, sex, race (only for HRS cohort), residence location, education level, work status, annual household income per capita, alcohol drinking, cigarette smoking, heart disease, stroke, cancer, diabetes, hypertension, and lung disease. Physical activity was categorized as inactivity (<150 min/week of moderate activity and <75 min/week of vigorous activity, and physical activity index < 9) and regular activity (≥150 min/week of moderate activity or ≥75 min/week of vigorous activity, or physical activity index ≥ 9), based on the frequency and duration of physical activity performed by participants. Definitions for each longitudinal cohort study and variable of physical activity, sarcopenia, and ADL/IADL are shown in Methods S1–S6.

ADL, activities of daily living; CHARLS, China Health and Retirement Longitudinal Study; CI, confidence interval; HR, hazard ratio; HRS, Health and Retirement Study; IADL, instrumental activities of daily living; SHARE, Survey of Health, Aging and Retirement in Europe.

When stratified by each component of sarcopenia or ADL/IADL disability, the interaction between each component and physical activity, that is, the more pronounced inverse association between regular activity and mortality in participants with disability than in normal participants, was observed in at least one cohort of HRS, SHARE, and CHARLS (Figure 3).

Figure 3.

Association of physical activity with all-cause mortality risk in older participants with or without components of sarcopenia and ADL/IADL in HRS, SHARE, and CHARLS cohorts

HRs (95% CIs) were adjusted for age, sex, race (only for HRS cohort), residence location, education level, work status, annual household income per capita, alcohol drinking, cigarette smoking, heart disease, stroke, cancer, diabetes, hypertension, and lung disease. Physical activity was categorized as inactivity (<150 min/week of moderate activity and <75 min/week of vigorous activity, and physical activity index < 9) and regular activity (≥150 min/week of moderate activity or ≥75 min/week of vigorous activity, or physical activity index ≥ 9), based on the frequency and duration of physical activity performed by participants. Definitions for each longitudinal cohort study and variable of physical activity, sarcopenia, and ADL/IADL are shown in Methods S1–S6.

ADL, activities of daily living; CHARLS, China Health and Retirement Longitudinal Study; CI, confidence interval; HR, hazard ratio; HRS, Health and Retirement Study; IADL, instrumental activities of daily living; NA, not applicable; SHARE, Survey of Health, Aging and Retirement in Europe.

Discussion

In three large-scale nationally representative cohorts, we found that physical activity defined according to the WHO recommendations significantly interacted with sarcopenia and ADL/IADL disability on all-cause mortality risk in older adults. The increased risk of mortality associated with sarcopenia, ADL disability, or IADL disability was amplified in inactive adults but was attenuated in regularly active adults. On the other hand, the beneficial association of regular activity with survival was more evident in adults with sarcopenia, ADL disability, or IADL disability than in those without these functional disabilities. This interaction pattern was consistent across the three cohorts from different countries and diverse populations, persisted after excluding adults with a history of cancer or cardiovascular disease, and remained for the majority of components of sarcopenia and ADL/IADL disability, strengthening the replicability and robustness of the results. Our findings indicated that the risk of mortality associated with sarcopenia and ADL/IADL disability could be mitigated by regular activity meeting WHO recommendations in the older population, and adults with sarcopenia or ADL/IADL disability appeared to derive greater benefit more from regular activity.

Previous evidence suggests that declined physical functional capacities such as sarcopenia phenotypes and ADL/IADL disability are risk factors for mortality.^20,21 To date, only three studies have explored the potential interaction between specific traits of functional capacity and physical activity on mortality, revealing inconsistent evidence.^22,23,24 One study using 498,135 adults aged 40–69 years from the UK Biobank reported that the higher hazard of mortality associated with lower physical activity was greatest in adults with the lowest grip strength (HR: 1.11; 95% CI: 1.09–1.14) and lowest in those with the highest grip strength (1.04; 1.01–1.08).²² On the contrary, the Healthy Survey of Northern Trøndelag (second wave) study of 2,529 elderly Norwegian women showed no significant interactions of handgrip strength and chair stand test with physical activity on all-cause mortality,²³ as well as one recent nationwide study of 177,360 older adults with a mean age of 71.3 years from the United States, which also reported non-significant differences in hazards of mortality associated with ADL/IADL disability between adults meeting or not meeting physical activity levels based on WHO recommendations.²⁴ These controversial findings might be attributable to differences in sample size, regions, or follow-up period, and reliance on a single functional indicator (e.g., handgrip strength alone) may not fully capture the complex interplay between functional capacity and physical activity in relation to mortality. In our study, taking advantage of precise and comprehensive measurements to assess sarcopenia and related performance, including ADL and IADL, we elucidated consistent interactions that the higher mortality risk attributable to sarcopenia and ADL/IADL disability significantly amplified in inactive adults but diminished in regularly active adults across three representative cohorts from different countries. Our findings extended previous evidence by providing novel and refined insights that older adults who are physically inactive might be more susceptible to mortality risk associated with a broader range of physical capacity indicators.

From another perspective, our study indicated that the lower mortality risk associated with physical activity was most prominent in older adults with sarcopenia or ADL/IADL disability, suggesting that these subgroups of people might gain more survival benefit from interventions to promote physical activity based on WHO recommendations. One recent meta-analysis of 42 randomized controlled trials showed that in older adults with sarcopenia, high or moderate certainty evidence supports that exercise with or without nutrition can improve handgrip strength and physical function measured by usual gait speed, timed up-and-go test, and the five-repetition chair stand test.²⁵ Moreover, in obese older people, aerobic and resistance exercise improves muscle protein synthesis and myocellular quality, thereby maintaining muscle mass during weight-loss therapy.²⁶ Our observations are supported by previous evidence, collectively suggesting that physical activity is a promising individualized intervention to improve survival, especially for older adults with declined physical capacity. The strengths of this study include the standardized definition of physical activity according to the WHO recommendations,⁶ the precise measurements of sarcopenia and ADL/IADL based on comprehensive components following the guidelines of European or Asian Working Groups,^27,28 and the consistent findings in three large-scale, nationally representative cohorts of multi-countries and diversified populations.

In summary, this study demonstrated consistent interactions of physical activity with sarcopenia and ADL/IADL on mortality risk in older adults from three nationally representative cohorts. Regular physical activity meeting the WHO recommendations significantly mitigates the greater risk of mortality associated with sarcopenia and ADL/IADL disability, and older adults with sarcopenia or ADL/IADL disability might gain more survival benefits from regular activity.

Limitations of the study

This study also has several limitations. First, the analysis did not distinguish between specific exercise types, such as aerobic and resistance training. While using total leisure-time physical activity enhances the generalizability of the study findings, this approach limits our ability to offer nuanced recommendations for personalized clinical interventions. The potential heterogeneity of effects across different exercise types warrants further investigation to identify the most suitable forms of activity for individuals with specific clinical profiles. Second, due to the observational nature of the cohort design, our findings cannot support causal inferences. Although extensive statistical adjustments were performed, residual confounding remains a possibility. Future research would benefit from more harmonized data that integrates database information with detailed clinical biomarkers to better control for confounding factors. Third, although most findings were consistent across the three cohorts, heterogeneity in population characteristics may partly explain the variations in effect estimates observed in certain subgroups. Consequently, further research using randomized controlled trials, Mendelian randomization, and gene-environment interaction analyses is needed to establish causality and improve precision prevention. Looking forward, future research should shift toward a life-course epidemiological perspective, focusing on middle-aged populations to detect early warning signs of functional decline. Moreover, developing and validating predictive models to identify individuals at high risk of progressing from possible sarcopenia to overt disability could enable truly preventive interventions.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Tiange Wang (tiange.wang@shsmu.edu.cn).

Materials availability

This study did not generate new reagents.

Data and code availability

• •HRS data are available online at: https://hrsdata.isr.umich.edu/data-products/rand. SHARE data are available online at: https://share-eric.eu/. CHARLS data are available online at: https://charls.pku.edu.cn/en/.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We extend our gratitude to all team members who contributed to the HRS, SHARE, and CHARLS studies. This study used data from these datasets. The HRS study is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. The SHARE study is funded by the European Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, DASISH: GA N°283646), and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, SHARE-COVID19: GA N°101015924) and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, VS 2020/0313, SHARE-EUCOV: GA N°101052589, and EUCOVII: GA N°101102412, and from various national funding sources (see https://share-eric.eu/infrastructure/funding). The CHARLS study is supported by the Behavioral and Social Research Division of the National Institute on Aging of the National Institute of Health (grants 1-R21-AG031372-01, 1-R21-AG033675-01A1, 1-R01-AG037031-01, and 1-R01-AG037031-03S1); the Natural Science Foundation of China (grants 7 70773002, 70910107022, and 71130002), the World Bank (contracts 7145915 and 7159234), and is conducted by Peking University. This work was supported by the grants from the Noncommunicable Chronic Diseases-National Science and Technology Major Project (2023ZD0508603), the National Natural Science Foundation of China (82370820 and 82570955), the “Shu-Guang Scholar Program” from Shanghai Municipal Education Commission, the Ruijin Hospital Youth Development Program-Outstanding Scholar (2025JC001), and the Innovative Research Team of High-level Local Universities in Shanghai.

Author contributions

Z. Zhu, data curation, formal analysis, writing original draft, writing review, and editing; X.Z., writing review and editing; M.C., writing review and editing; C.D., writing review and editing; D.L., writing review and editing; L.K., writing review and editing; C.Y., writing review and editing; M.X., writing review and editing; Y.X., writing review and editing; M.L., writing review and editing; Z. Zhao, writing review and editing; J.Z., writing review and editing; J.L., writing review and editing; Y.C., writing review and editing; W.W., writing review and editing; G.N., conceptualization, supervision, writing review, editing, and funding acquisition; Y.B., conceptualization, supervision, writing review, editing, and funding acquisition; T.W., conceptualization, supervision, writing review, editing, and funding acquisition.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
The data for 20,061 participants from the HRS cohort (2010)	HRS database	https://hrsdata.isr.umich.edu/data-products/rand
The data for 30,758 participants from the SHARE cohort (2006)	SHARE database	https://share-eric.eu/
The data for 13,327 participants from the CHARLS cohort (2011)	CHARLS database	https://charls.pku.edu.cn/en/
Software and algorithms
R language	R software	https://www.r-project.org/

Experimental model and study participant details

Study settings and participants

Study participants were from three nationally representative cohorts: HRS, SHARE, and CHARLS. Detailed information on the three cohorts is described in Methods S1. Briefly, HRS is a nationwide representative longitudinal survey of adults aged ≥50 years in the USA.²⁹ SHARE is a cross-national panel study of adults aged ≥50 years across 14 European countries and Israel.³⁰ CHARLS is a nationally representative study of adults aged ≥45 years in China.³¹ In this study, we selected the earliest visit with large sample sizes and available measurements on sarcopenia and ADL/IADL as the baseline for each cohort, and the follow-up time was the latest visit, utilizing data from: 2010–2021 for HRS, 2006–2020 for SHARE, and 2011–2020 for CHARLS. As shown in the flowchart (Figure S1), from 21,042 participants in HRS, 36,079 participants in SHARE, and 17,106 participants in CHARLS, we excluded participants with missing data on physical activity and covariates, with incomplete data to define at least one functional disability indicator (i.e., sarcopenia, ADL, and IADL), and with missing data on mortality ascertainment. Finally, the final analysis included 20,061 participants in HRS, 30,758 participants in SHARE, and 13,327 participants in CHARLS. The ethical approval was covered by the original surveys for HRS, SHARE, and CHARLS. This analysis was reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline for cohort studies.

The HRS study was approved by the Institutional Review Board at the University of Michigan (approval number: HUM00061128). The SHARE study was reviewed and approved by the Ethics Council of the Max Planck Society. Further details on ethics approvals are available at: https://share-eric.eu/fileadmin/user_upload/Ethics_Documentation/SHARE_ethics_approvals.pdf. The CHARLS study was approved by the Institutional Review Board of Peking University (approval number: IRB00001052-11015). Ethical approvals were obtained from the respective institutional review boards for each cohort, and informed consent was obtained from all participants. No additional ethical approval was required for the use of publicly available de-identified data.

Method details

Assessment of leisure-time physical activity

In HRS, SHARE, and CHARLS, leisure-time physical activity was measured by questionnaires, and the questions that collect physical activity information are shown in Methods S2. In HRS and SHARE, physical activity was assessed by intensity (vigorous, moderate, and light in HRS and vigorous and moderate in SHARE) and frequency (“never”, “1–3 times/month”, “1 time/week”, and “>1 time/week”). We followed a widely used method to produce a physical activity index score by summing vigorous, moderate, and light physical activity levels.³² Moderate activity level was coded as: 0 = “never”, 1 = “1–3 times/month”, 3 = “1 time/week”, and 6 = “>1 time/week”; light activity level was coded as: 0 = “never”, 0.5 = “1–3 times/month”, 1.5 = “1 time/week”, and 3 = “>1 time/week”; vigorous activity level was coded as: 0 = “never”, 2 = “1–3 times/month”, 6 = “1 time/week”, and 12 = “>1 time/week”. The index was calculated by summing the scored points of these three questions, ranging from 0 to 18. An index ≥9 is recommended as regular activity, equivalent to meeting or exceeding the WHO physical activity recommendations.^6,32

In CHARLS, the questionnaire constructed physical activity variables recording the frequency and duration of vigorous and moderate physical activity in a usual week, based on the International Physical Activity Questionnaire (Short Form), which is one of the most widely used instruments for measuring physical activity levels in adults worldwide.³³ The total physical activity time was calculated by multiplying the frequency and duration of physical activity. Based on the WHO recommendations, regular activity was defined as at least 150 min/week of moderate physical activity, or 75 min/week of vigorous physical activity, or a combination of moderate and vigorous physical activity (the volume of vigorous activity is twice that of moderate activity).⁶

Assessment of sarcopenia

Sarcopenia was diagnosed by muscle strength, physical performance, and muscle mass. Muscle strength was measured by grip strength, an adequate marker of overall limb strength, which was alternately measured twice for each hand using a handheld dynamometer in HRS, SHARE, and CHARLS. Muscle strength was calculated as the mean of the maximum value from valid grip strength measurements of the right and left hands, and sex-specific cutoff points were used to define low muscle strength according to European or Asian Working Groups for Sarcopenia (details see Methods S3).^27,28

Physical performance is a multidimensional indicator that includes three standard examinations: gait speed, 5-time chair stand test, and the short physical performance battery (SPPB).^27,28,34 Physical performance was assessed by gait speed in HRS, by gait speed or chair stand test in SHARE, and by gait speed, chair stand test, or SPPB in CHARLS (Details see Methods S4). Gait speed was evaluated by measuring time as the participants walked the 2.5-meter course two times, the faster speed of the two laps was used for statistical analysis. The chair stand test was evaluated by measuring time as the participants performed five repetitions of sit-to-stand at their fastest pace while keeping their arms folded across the chest. SPPB is an established tool developed by the National Institute on Aging to predict disability across diverse populations, summing a composite score based on gait speed, chair stand test, and standing balance.^35,36

In HRS, SHARE, and CHARLS, the muscle mass was estimated by the total-body skeletal mass index (SMI). SMI was obtained by dividing skeletal muscle mass (SMM) by height square to adjust body size (SMI = SMM/height2).³⁷ SMM was calculated using the validated anthropometric equation (Methods S5).³⁸ Using the magnetic resonance imaging (MRI)-measured SMM as the gold standard, agreement between the SMM equation and MRI-measured SMM was strong (R² = 0.85, SEE = 2.6 kg).³⁹ Low muscle mass was categorized based on the sex-specific lowest 20% of SMI in the study population.³⁷

In HRS, SHARE, and CHARLS, sarcopenia was defined as probable, confirmed, or severe, according to the consensus of the European Working Group on Sarcopenia in Older People (EWGSOP2) and the Asian Working Group for Sarcopenia (AWGS2019).^27,28 Probable sarcopenia is suspected whereby low muscle strength or low chair stand test performance is found. Confirmed sarcopenia is defined as the presence of low muscle mass. When both muscle strength and physical performance are impaired, sarcopenia is considered severe. In this study participants, only 321 (1.6%) of HRS, 365 (1.2%) of SHARE, and 654 (4.9%) of CHARLS had severe sarcopenia, so participants with severe sarcopenia were merged within the confirmed sarcopenia group following previous research.⁴⁰

Assessment of ADL/IADL disability

The information on ADL components (dressing, bathing, eating, getting in and out of bed, and using the toilet)⁴¹ and IADL components (cooking, shopping, taking medication, and managing money)⁴² was collected through questionnaires in HRS, SHARE, and CHARLS. Besides, walking across a room belonging to ADL and using a map belonging to IADL were collected in both HRS and SHARE. Maintaining continence belonging to ADL and doing housework belonging to IADL were collected in CHARLS. In HRS, SHARE, and CHARLS, any difficulty in performing any ADL or IADL components was considered ADL or IADL disability. Detailed ADL/IADL components in three cohorts are shown in Methods S6.

Ascertainment of mortality

In HRS, mortality was determined by reconciling HRS trace data, the National Death Index, the National Health Service’s Central Registry, or Medicare claims.⁴³ Exit interviews were conducted with the deceased participant’s relative or other informant.⁴³ In SHARE, mortality was determined from exit interviews with the deceased participant’s relative, household member, neighbor, or any other close person, who provided information on the precise time, location, and cause of death.³⁰ In CHARLS, mortality was determined from the official registration and certification of death submitted by the deceased participant’s proxy respondents, which is typically issued by the hospital or the local community authority.³¹ Such documentation constitutes a thorough record, detailing the death registration number, the precise time and location of the event, underlying death causes, and pertinent preexisting information of the deceased participant.³¹

Covariates measurement

Information on sociodemographic factors (age, sex, residence location, education level, work status, and annual household income per capita), lifestyle factors (cigarette smoking and alcohol drinking), and common diseases (self-reported diagnosis by physician of heart disease, stroke, cancer, diabetes, hypertension, or lung disease) was derived from questionnaires in HRS, SHARE, and CHARLS. In HRS, information on race/ethnicity was also collected, with four categories including non-Hispanic White, non-Hispanic Black, Hispanic, and others.

In HRS, SHARE, and CHARLS, education was categorized as below high school, high school, or college or above. Work status was divided into never worked or retired and employed based on labor force codes. Annual household income per capita, the ratio of annual household income to the number of household members, was classified as low or high by comparing with the median level. Cigarette smoking and alcohol drinking were categorized as never, former, and current status. Six self-reported diseases were classified as yes or no based on a diagnosis by professional physicians.

Quantification and statistical analysis

Baseline characteristics of participants were presented as median (IQR) for continuous variables or number of participants (percentage) for categorical variables. Pairwise correlations between physical activity, sarcopenia, ADL, and IADL were explored by Kendall’s tau-b correlation coefficients. For the missing physical activity data at baseline in CHARLS, we imputed these values using the corresponding measurements from the same participants obtained during their first follow-up visit, applying the “next observation carried backward (NOCB)” method.⁴⁴

The Cox proportional hazards model was performed to estimate HRs and 95% CIs for the risk of all-cause mortality associated with physical activity, sarcopenia, ADL/IADL disability, and their components within each cohort. Time to event for each participant was determined from the enrollment date to the death or follow-up end date, whichever came first. Associations of sarcopenia and ADL/IADL disability with mortality risk stratified by physical activity levels and associations of physical activity with mortality risk stratified by sarcopenia and ADL/IADL disability status were analyzed by Cox proportional hazards models. Multiplicative interaction terms were incorporated in the main regression models using a likelihood ratio test to assess the interaction between physical activity and sarcopenia and ADL/IADL disability on mortality risk. Additive interaction was also examined by calculating the RERI, AP, and SI to examine the additive effect of sarcopenia or ADL/IADL disability with physical inactivity on the all-cause mortality risk. The RERI indicates if the combined effect of both exposures is greater than the sum of their individual effects, and an RERI>0 indicates positive additive interaction. The AP indicates the proportion of risk attributable to the combined exposure, and the SI > 1 indicates a synergistic interaction. For association and interaction analyses, Cox proportional hazards models were adjusted for age, sex, race (only for HRS), residence location, education level, work status, annual household income per capita, alcohol drinking, cigarette smoking, heart disease, stroke, cancer, diabetes, hypertension, and lung disease.

We conducted multiple sensitivity analyses to assess the robustness of our findings. We replicated the interaction analysis of physical activity with sarcopenia and ADL/IADL on all-cause mortality after excluding participants with a history of heart disease, stroke, or cancer at baseline to assess whether the interaction pattern may be influenced by worse health status. We further extended this approach by excluding participants diagnosed with cancer during the first follow-up survey, which occurred after intervals of 2 years for HRS (2010–2012), 4 years for SHARE (2006–2010), and 2 years for CHARLS (2011–2013). To explore interactive associations across severity levels of ADL/IADL disability, we categorized ADL/IADL disability into three groups (0, 1–2, and >2 components of disability) and repeated the interaction analysis across three cohorts. Additional sensitivity analyses included: repeating interaction analyses after excluding participants with missing physical activity data (without imputation) in CHARLS; restricting to participants aged ≥50 years in CHARLS; further adjusting for marital status and psychiatric disorders in HRS, SHARE, and CHARLS; applying multiple imputation for missing covariates across three cohorts; and excluding deaths occurring within the first year of follow-up across three cohorts.

All analyses were conducted using R software version 4.3.3 (Lucent Technologies, Bell Laboratories, NJ, USA) using the coxph function in survival package, the TableSubgroupCox function in Jstable package, the mice function in mice package, and the interactionR function in interactionR package. Statistical significance was set at a 2-sided value of p < 0.05.

Published: April 9, 2026