Participants
Data were drawn from the Long-term Care Demand Assessment (LCDA), a population-based face-to-face survey of 60 years old or over from rural and urban areas of China. A multistage, cluster sampling scheme was conducted by the Xiangya-Oceanwide Health Management Research Institute, Central South University. Initially, four regions (Hunan, Jiangsu, Tianjin, and Heilongjiang) were randomly selected from the northern, southern, central, western, and eastern regions, respectively, of China. Then, cities with a representative population and economic development were selected (Changsha, Changde, Zhuzhou, Xiangtan, Hengyang, Nantong and Daqing). Finally, communities and nursing homes from these places were selected randomly. A total of 3340 individuals were enrolled in this project. Then, a total of 1324 individuals were excluded due to coma (141), missing depression data (872), disability (224), Alzheimer's disease (63) and other dementia (24), resulting in 2016 participants included in this analysis. This study was approved by the Central South University Ethics Committee (approval number: 2018011). All participants provided written informed consent at the time of enrollment.
Measures
Functional ability The Barthel Index (BI) was used to measure the individuals’ level of dependency in ADL. The BI is a sum score across ten domains of ADL. Each domain is scored on a weighted numerical scale with the lowest score indicating total dependency and the highest score indicating complete independence. The minimum total score is 0 (completely dependent), the maximum total score is 100 (completely independent). Time for completion is approximately 5-10 minutes[7].
Intrinsic capacity
Depression The patient health questionnaire depression scale (PHQ-9) was used to measure the individual’s depressive symptoms. As a severity measure, the PHQ‐9 score ranges from 0 to 27, since each of the 9 items can be scored from 0 (not at all) to 3 (nearly every day)[8].
Nutrition The Mini Nutritional Assessment short-form (MNA-SF) was used to measure the individual’s nutritional status. It consists of an ordinal scale with values ranging from 0-14. Lower scores are indicative of a worse nutritional status. The sum of its scores enables the following classification: from 0 to 7 presents malnutrition; from 8 to 11 presents having the risk of malnutrition; from 12 to 14 presents normal nutritional status[9].
Hearing and visual function Hearing function were evaluated through talking to the individuals and wearing hearing-aid was allowed. Hearing and visual functions were categorized into three groups, namely, severe or moderate impairment, mild impairment and intact.
Cognition The cognitive function was measured by the Cognitive Function Rating Scale (Shanghai Version). This scale is widely used in Shanghai for rating the care-level of the elderly. It composes of four categories including short-term memory, procedural memory, orientation function, and judgment function. The score ranges from 0 to 40 and a higher score indicates worse cognitive function. As a severity measure, the scale is scored from 0-1(intact cognition), 2-11(mild cognition decline), 12-25(moderate cognition decline) and 26 and over (severe cognitive decline). Its interrater reliability (0.896) was assessed using a sample of 30 individuals and its test-retest coefficient of reliability (r = 0.918) was calculated for a time interval of two weeks and value of Cronbach’s α was 0.911.
Sociodemographic and clinical characteristics
Baseline sociodemographic variables were collected including age, gender, education, marital status, economy, and living condition. Age was a continuous variable. Categorized variables were divided into groups, including education (<-5 years, 6-11 years, >-12 years); living condition (nursing homes and communities); economy (no vs yes; no indicating not specially funded by governments and yes indicating specially funded by governments); and marital status (partnered vs not partnered).
Clinical characteristics, namely, history of fall (yes vs no), auxiliary tools (yes vs no), self-rating health (very good, good, fair and poor), social support (sufficient material and emotional support, material support, emotional support, and insufficient material and emotional support), and social participation (active, regular, occasional and never) were collected. Five common diseases were collected from medical records (hypertension, stroke, coronary disease, diabetes, and cataract) as they were common diseases among the elderly in our research data. (Figure S1.)
Statistical analysis
We used categorical patient characteristics such as the sex by ADL using the chi-square test, or, if expected cell counts were less than 5, continuity correction was applied. We used Wilcoxon tests to compare two groups on age, cognition, nutrition and pain scores as all tests of normality did not show normal distribution (p for Kolmogorov-Smirnov test < 0.05). These tests were all 2 sided.
To evaluate the association between IC and ADL, a multivariate analysis was fit with the use of dichotomous variables of ADL scale that was coded as ‘0’ (76 ≤ ADL Scale score ≤100) and ‘1’ (0 ≤ ADL Scale score ≤75) as the dependent variable[10]. The variables included in the multivariate analysis were variables with a p-value of less than 0.20[11, 12].
Missing values of variables (9 participants without marital status, 1 participant without living condition, and 13 without self-rating health) were imputed with the mode as they are category variables. Precisely, missing marriage status, living conditions, self-rating health self-rating health were replaced with not partnered, nursing home and fair, respectively. Statistical analysis was performed with free open-source R statistical software (www.r-project.org).