Participants and study design
Three cross-sectional studies – Evaluation of healthy city projects were conducted in three cities in China-Shanghai, Zhengzhou, and Baoji from June 2017 to October 2018. Participants were selected by a multi-level stage sampling procedure. In the first stage, 16 districts from Shanghai (eastern China), 28 districts from Zhengzhou (central China), and 5 districts from Baoji (western China) were sampled randomly. The three cities represented the eastern, central and western China in our previous study about drinking and social capital. In the second stage, 5% of the population aged 18 years or older were randomly selected from each district selected in the first stage. In total, 22,290 participants were randomly selected to answer undergo the questionnaires anonymously, but 856 participants rejected participation. 20,716 participants were involved in this study after excluding 481 participants with missing data and 237 participants aged under 18 years old. This study was approved by the Institutional Review Board of Fudan University (Ethic Approval Code: IRB#2018-03-0666).
The Chinese version of the FRAIL scale was used to measure frailty, which consists of five dichotomous (yes/no) items (fatigue, resistance, ambulation, illness, and loss of weight) from both the Frailty Index (FI) and Freid’s Frailty Phenotype (FP), which is widely used in the Asia-Pacific region (3). The FRAIL scores range from 0 to 5 (i.e., 1 point for each component; 0 = best, 5= worst) and represent frail (3–5), pre-frail (1–2), and robust (0) health status.
Alcohol consumption pattern (ACP)
Alcohol consumption pattern was measure by the following questions. Firstly, the participants responded to “Have you ever consumed any alcohol such as beer, wine, spirits during the past 12 months?”. If their responses were yes to the first questions, they continued to respond to four dichotomous (yes=1/no=0) items of the Chinese version of CAGE . A higher score on the scale indicates alcohol dependence or addiction, and a total of 2 or more is considered problematic drinking . Participants whose responses to the first question was no were recognised as nondrinkers; Participants whose responses to the first question was yes, but CAGE scale was 0 or 1 were recognised as a non- problematic drinker; Participants whose responses to the first question was yes and their CAGE scale was greater than or equal to 2 were recognised as a problematic drinker.
Based on a literature review, several sociodemographic factors were selected as covariates: gender, age (10-year categories), educational level (elementary school or lower, junior high school, senior high school, and college or higher), occupational status (recoded into employed and unemployed, including unemployed and retirement), and marital status (recoded into married and unmarried, including unmarried, divorced, and widowed). Other covariates in this study also included self-rated health (categorised as very good, good, fair, or bad), smoking (yes/no), and physical activity (none, irregular, and regular). The methods used to measure the above covariates were described in detail in other literature [22, 23].
As well as descriptive analyses, univariate analyses including χ2 trend tests and Spearman regressions were used to examine the distributions of frailty and ACP through covariates. Second, we used multinomial logistic regressions to examine the association of ACP with pre-frailty and frailty. We estimated the adjusted ORs and their 95% confidence intervals (CIs) of independent variables for frailty. STATA version 13.0 (StataCorp LP., College Station, TX, USA) was used to carry out all analyses.