Study population
CLHLS is an extensive and ongoing longitudinal study on the determinants of healthy longevity in China. The CLHLS utilized a multistage stratified cluster sampling approach, implemented across 22 provinces selected from China’s 31 provincial administrative divisions. A total of 631 municipal and county units were randomly chosen through this framework, collectively encompassing approximately 85% of the national population [15].
All participants provided paper-based informed consent before data collection. During the data collection procedure, only the participant and the interviewer were present. The study was approved by the Biomedical Ethics Committee of Peking University, Beijing, China (IRB00001052–13074) [15].
This cross-sectional study utilizes data from the 2018 wave of CLHLS, which was conducted between 2017 and 2018 and comprised of a total of 15,874 respondents, with 12,411 of them being newly interviewed in 2018. The dataset was freely downloaded from Peking University Open Research Data (http://opendata.pku.edu.cn/). After excluding participants without complete dietary information to calculate the healthy eating index and those with missing data on key variables for assessing sarcopenia, a total of 14,257 participants aged 60 years and older were included. The detailed flowchart is shown in Fig. 1.
Flowchart of participants included in the CLHLS 2018
Calculation of the healthy eating index HEI
The HEI score was calculated based on previously established methods with modified according to the CLHLS dietary information [10, 16, 17]. The current consumption frequency of 13 food groups, used to construct the HEI, was collected through face-to-face interviews by trained interviewers using structured food frequency questionnaire. These food groups included vegetables, fruits, meats, fish, eggs, beans and their products, tea, garlic, nuts, mushrooms or algae, dairy products, salt-preserved vegetables, and sugar.
Scores were assigned to each food group’s consumption frequency as follows: For vegetables and fruits, scores ranged from 0 to 3, with “rarely or never” scoring 0, “occasionally” scoring 1, “except in winter” scoring 2, and “almost every day” scoring 3. For the other 11 food groups, scores were assigned based on the frequency of consumption as follows: “rarely or never” scored 0 points, “not every month, but occasionally” scored 1 point, “not every week, but at least once a month” scored 2 points, “not every day, but at least once a week” scored 3 points, and “almost every day” scored 4 points, while sugar and salt-preserved vegetables were reverse scored [16].
Finally, the score of each food group was summed together to obtain the HEI score of participants. The HEI score ranged 0 ~ 50 with the higher HEI score representing a healthier diet.
Assessment of sarcopenia
As described in previous studies in CLHLS [18], the SARF-C questionnaire, a widely used screening tool for sarcopenia with good internal consistency reliability (Cronbach’s α varied from 0.76 to 0.81) in three cohorts and low sensitivity (13.7–37.9%) but high specificity (94.8–98.1%) in Chinese community elders [19,20,21], was employed in this study. The questionnaire consists of five questions assessing strength, assistance walking, rising from a chair, climbing stairs, and falls, as described previously with slight modifications [22].
In brief, in the CLHLS, the strength was measured by the question, “Are you able to carry a 5 kg weight?” Assistance walking was assessed by the question, “Are you able to walk one kilometer?” Climbing stairs was not directly measured in the CLHLS but was substituted with the question, “Are you able to crouch and stand three times?” to assess lower limb performance. For both the strength and walking questions, scores were assigned as follows: 2 points for “yes,” 1 point for “a little difficult,” and 0 points for “unable to do so.”
Rising from a chair was assessed by the question, “Are you able to stand up from sitting in a chair?” with 2 points for “yes, without using hands,” 1 point for “yes, using hands,” and 0 points for “no.” Falls were measured by the number of falls in the past year, with 2 points assigned for four or more falls, 1 point for 1 to 3 falls, and 0 points for no falls.
The total SARC-F score ranges from 0 to 10, with respondents considered to have sarcopenia with the SARC-F score ≥ 4 in this study [22, 23].
Potential covariables
Potential confounding factors, which include socio-demographic characteristics (sex, age, residence, co-residence type, economic status, marital status, education level) and health-related factors (smoking, drinking, physical activity, health status, body mass index [BMI], and chronic disease status), were selected based on prior evidence of their associations with sarcopenia and diet, and were adjusted to improve the accuracy of the results [2, 7]. The adjusted social-demographic factors and some health-related factors, including smoking, alcohol consumption, physical activity and history of diseases were collected by trained interviewer through face-to-face interview.
Age were categorized into two groups:<75 years and ≥ 75 years. Co-residence type was categorized into two groups: “alone” for participants who lived alone, and “not alone” for those who lived in an institution or with household members. Education level was categorized into four groups based on the years of education: Illiterate (0 years), Primary school (1 ~ 6 years), Middle school (7 ~ 9 years) and High school or above (≥ 10 years). Marital status was stratified into two groups: “currently married and living with spouse/cohabiting” and “separated/divorced/widowed/never married.” Economic status was categorized as “difficulty,” “average,” or “wealthy” according to participants’ responses to the question, “How do you rate your economic status compared with other local people?” Smoking, drinking, and exercise habits were classified into three groups: “never,” “former,” and “current.” Health status was assessed and stratified into three distinct categories by trained interviewers: (1) “surprisingly healthy” for participants reporting no chronic conditions and maintaining full functional independence; (2) “relatively healthy” for those with only minor ailments but preserving basic daily functioning; and (3) “ill” for individuals with moderate to major degrees of major ailments or illnesses or with significant functional impairments. BMI was calculated as weight (kg)/[height (meter)]2 and divided into four groups(underweight [< 18.5 kg/m2], normal [18.5 kg/m2 ≤ BMI < 24 kg/m2], overweight [24.0 kg/m2 ≤ BMI < 28.0 kg/m2] and obesity[BMI ≥ 28 kg/m2]) according to China Working Group criteria [24]. A history of hypertension is asserted for participants with SBP ≥ 140 mmHg and (or) DBP ≥ 90 mmHg, or who have been diagnosed by a doctor or are currently taking medication for hypertension. A history of diabetes, heart disease, stroke, and cancer is asserted if participants have been diagnosed by a doctor or are currently taking medication for these conditions.
Statistic methods
Demographic characteristics of the study subjects were summarized using standard descriptive methods. HEI scores were analyzed both as a continuous variable and as quartiles (Q1-Q4). Variance analysis or Kruskal-Wallis test was used for continuous variables and the chi-square test was used for categorical variables to compared the difference between quartiles of HEI.
Three logistic regression models were built to explore the association between HEI and sarcopenia. Specifically, Model 1 was the crude model, Model 2 adjusted for sociodemographic factors, including age, sex, residence, co-residence type, economic status, marital status, and education level. Model 3 additionally included smoking, drinking, physical activity, health status, BMI and history of diseases. Model diagnostics including Hosmer-Lemeshow goodness-of-fit test and Nagelkerke’s R² were performed to confirm model appropriateness [25].
In addition to analyzing HEI scores as a continuous variable, the association was also assessed using HEI quartiles (Q1-Q4) as a categorical variable. A trend test (P-trend) was performed with the medium of the HEI within each category to assess whether the association showed a significant decreasing trend across increasing HEI quartiles [26]. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to quantify the associations. Finally, subgroup analysis was performed with multiple logistic regression models to evaluate the consistency of observed results between different predefined subgroups, considering potential effect modifier such as age, gender, marital status, Residential area, co-residence type, economic status and physical activity.
Missing values in covariates were handled using Markov Chain Monte Carlo (MCMC)-based multiple imputation. Five imputed datasets were generated with 20 iterations, and pooled estimates (OR and 95% CI) were derived via Rubin’s rules to minimize bias [27].
Restricted cubic splines (RCS) were utilized to investigate potential non-linear relationships between HEI and prevalence of sarcopenia. In the spline models, the 10th percentile of the ln-transformed HEI distribution was set as the reference value (OR = 1.00), with knots at the 5th, 35th, 65thand 95th percentiles and adjusted for covariables in model 3 [28].All statistical analyses were conducted using SPSS version 26.0 (IBM Corp., Armonk, NY, USA) and R version 4.0.0 (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was defined as a two-tailed P-value < 0.05.