The longitudinal bidirectional association between sarcopenia and cognitive function in community-dwelling older adults: Findings from the China Health and Retirement Longitudinal Study

Background Although an association between sarcopenia and cognitive function has been demonstrated, the directional association remains unclear. The present study aimed to evaluate the longitudinal reciprocal relationship and identify the possible temporal sequence between sarcopenia and cognitive function in older Chinese adults. Methods Data were collected from the China Health and Retirement Longitudinal Study (CHARLS) baseline survey in 2011 and the follow-up survey in 2015. Cognitive function was measured by episodic memory and executive function. Sarcopenia status (non-sarcopenia, possible sarcopenia and sarcopenia) was defined based on the Asian Working Group for Sarcopenia 2019 criteria. Linear regression analysis and ordinal logistic regression analysis were employed to investigate the relationship between baseline sarcopenia status and follow-up cognition, as well as the association of baseline cognition with follow-up sarcopenia status, respectively. A cross-lagged panel analysis was performed to simultaneously evaluate the bidirectional association and the strength of the temporal relationship. Results Of 2689 participants, the median age was 65.0 years and 1249 (46.5%) were female. After adjusting for potential confounders and baseline measurements, baseline sarcopenia status was dose-dependently associated with subsequent cognitive scores (β = -0.45; P for trend = 0.001), and baseline cognitive scores (in tertiles) were also dose-dependently associated with subsequent sarcopenia status (odds ratio (OR) = 0.86; P for trend = 0.017). The cross-lagged panel analysis indicated that the standardised effect size of sarcopenia status on cognitive function (β = -0.09; P < 0.001) is larger relative to the effect of cognitive function on sarcopenia status (β = -0.05; P = 0.019). Conclusions There is a longitudinal, bidirectional relationship between sarcopenia status and cognitive function in older Chinese adults. Sarcopenia is likely the driving force in these dynamic associations. These ﬁndings imply that interventions in either sarcopenia or cognitive decline may have the ability to generate reciprocal benefits over time. More research is warranted to conﬁrm these ﬁndings and to further elucidate underlying causal pathways.


The longitudinal bidirectional association between sarcopenia and cognitive function in community-dwelling older adults: findings from the China Health and Retirement
Longitudinal Study Table S1.Baseline characteristics of the study participants without baseline mild cognitive impairment (n=2452) Table S2.Baseline characteristics of the study participants without baseline sarcopenia (n=2168) Table S3.Model fit and selection criteria for LCA and LPA at baseline Table S4.Results of LCA and LPA with distal outcomes Table S5.Association between baseline sarcopenia and follow-up cognitive function after additional adjustment for average household income a Table S6.Association between baseline cognitive function and follow-up sarcopenia after additional adjustment for average household income a  a Missing data: 592 for average household income, 47 for social isolation, 14 for nighttime sleep duration, 6 for post-lunch napping duration, 1 for malnutrition, 18 for restriction on ADL, 6 for hypertension, 18 for diabetes, 11 for cancer, 6 for lung disease, 11 for heart diseases, 3 for stroke, 3 for arthritis, 37 for dyslipidemia, 14 for kidney disease, 8 for asthma, 8 for digestive disease, 9 for emotional and mental disorders, 13 for liver disease, 1 for visual impairment, 1 for hearing impairment.The model was performed with excluding missing data on covariates.c β and P values for trend were calculated by treating baseline sarcopenia as ordinal variables (0 = non-sarcopenia; 1 = possible sarcopenia; 2 = sarcopenia), and β represents the change in cognitive score for each one-rank increase in baseline sarcopenia status.Model 5 was adjusted for age, sex, marital status, residence, educational level, drinking status, social isolation, nighttime sleep duration, post-lunch napping duration, BMI category, malnutrition, depressive symptoms, dyslipidemia, baseline cognitive scores, and average household income.Ordinal logistic regression was used to model this association, and the OR is interpreted as the odds of being in a more severe sarcopenia status compared with the reference category.b The model was performed with excluding missing data on covariates.c OR and P values for trend were calculated by treating baseline cognitive score as ordinal variables (0 = lowest tertile; 1 = middle tertile; 2 = highest tertile), and OR indicates the odds ratio of being in a more severe sarcopenia status for each one-rank increase in baseline cognitive scores (tertiles).Model 5 was adjusted for age, sex, marital status, residence, educational level, drinking status, social isolation, nighttime sleep duration, BMI category, depressive symptoms, restriction on ADL, hypertension, cancer, stroke, arthritis, dyslipidemia, baseline sarcopenia status, and average household income.

Figure S1 .
Figure S1.Cross-lagged panel analysis of sarcopenia and cognitive function after additional adjustment for average household income

Figure S1 .
Figure S1.Cross-lagged panel model of sarcopenia and cognitive function after additional adjustment for average household income.Sarcopenia was treated as ordinal variables (0 = non-sarcopenia; 1 = possible sarcopenia; 2 = sarcopenia).The cross-lagged panel model was estimated with weighted least square mean and variance (WLSMV) that was specifically developed for ordinal variables in Mplus.Standardized coefficients were reported.Single-headed arrows represented regression paths.Double-headed arrows represented correlations.

Table S1 . Baseline characteristics of the study participants without baseline mild cognitive impairment (n=2452)
Notes: Statistical significance was assessed by unpaired t-test, Wilcoxon test, or a chi-squared test.CNY, Chinese Yuan; BMI, body mass index; CESD-10, Center for Epidemiologic Studies Depression; ADL, activities of daily living.Values of variables may not sum to 100% due to rounding.P-values in bold indicate <0.05.

Table S2 . Baseline characteristics of the study participants without baseline sarcopenia (n=2168)
Statistical significance was assessed by one-way ANOVA, Kruskal-Wallis test, or a chi-squared test.CNY, Chinese Yuan; BMI, body mass index; CESD-10, Center for Epidemiologic Studies Depression; ADL, activities of daily living.Values of variables may not sum to 100% due to rounding.P-values in bold indicate <0.05.aMissing data: 549 for average household income, 40 for social isolation, 16 for nighttime sleep duration, 4 for post-lunch napping duration, 2 for malnutrition, 15 for restriction on ADL, 7 for hypertension, 19 for diabetes, 11 for cancer, 6 for lung disease, 11 for heart diseases, 3 for stroke, 3 for arthritis, 36 for dyslipidemia, 12 for kidney disease, 10 for asthma, 6 for digestive disease, 8 for emotional and mental disorders, 14 for liver disease, 2 for visual impairment, 1 for hearing impairment.

Table S3 . Model fit and selection criteria for LCA and LPA at baseline
Better model fit is indicated by lower AIC, BIC, and SSABIC values and higher entropy values, and a non-significant value (P>0.05)suggests that the model with one fewer class provides a more parsimonious fit.Selected models were in bold type.AIC, Akaike Information Criteria; BIC, Bayesian Information Criteria; SSABIC, Sample-Size Adjusted Bayesian Information Criteria; LRT, P-value for the Lo-Mendell-Rubin Likelihood Ratio Test.

Table S5 . Association between baseline sarcopenia and follow-up cognitive function after additional adjustment for average household income a
Linear regression analysis was used to model this association.b