Study Design and Participants
This was a longitudinal observational study of Chinese older adults living in the community of Shanghai. A total of 4050 participants aged ≥ 60 years were originally recruited for an investigation of chronic diseases and geriatric syndromes during 2015 (baseline) from Shanghai. After eliminating invalid answers to questionnaires, the valid response proportion was 98.5% (3988/4050). Multi-stage random sampling was used to select subjects. Two of 11 streets were randomly selected, and then four communities were randomly selected from each street. Simple random sampling was used to select family addresses in which older adults aged ≥ 60 years lived in through a household registration information system. Residents aged ≥ 60 years who lived in the community for at least 6 months of the year from the selected families were all approached to participate in the investigation. Written informed consent was obtained from each study participant, and the Research Ethics Committee of the Division for the Prevention and Control of Chronic Non-communicable Diseases, China Center for Disease Control and Prevention, approved the study protocol. Finally, the average age and its standard deviation (SD) of participants was 69.38 (7.06), with 43.5% male, 79.8% married, and more than half never educated (as shown in Table 1). During the next 2 years, there were 127 deaths and 23 participants lost to follow-up.
Frailty Index Construction and Frailty Transitions
According to the methodology of FI construction, a multiple deficits approach was used to construct the Frailty Index. Deficits were defined as “symptoms, signs, disabilities, and diseases,” the prevalence of which must increase with age. Thirty-six eligible items were eventually selected covering the self-reported presence of current diseases (5 items), cognitive and mental symptoms (9 items), ability in the activities of daily living (15 items), as well as physical and neurological signs (7 items) (Details in Appendix 1). All of the items for FI were dichotomized into the presence (1) or absence (0) of a frailty deficit. The FI was calculated as the proportion of the number of deficits for an individual to the maximum total number of deficits. According to previous studies,[23, 35] we categorized the FI score into a three-level variable: robust (FI ≤ 0.10), prefrail (0.10 < FI ≤ 0.21), and frail (FI > 0.21). Frailty transitions were included for the three kinds of frailty status (robust, prefrail, and frail) changing into to each other among survivors after 2 years, which were defined as worsening, stability, or improvement. Outcomes also included death from each of the three kinds of frailty status.
Sociodemographic factors included age (Years), gender (Male/Female), educational level (Illiteracy/Primary school/Junior high school or above), marital status (Married/Unmarried; ‘Unmarried’ included never married, divorced, and widowed), and working status (Yes/No; ‘Yes’ means an individual still works for payment or for free, such as a volunteer, and ‘No’ means an individual has retired or has no work).
Lifestyle factors included living alone (Yes/No), has a shower facility at home (Yes/No), annual physical examination (Yes/No), cigarette smoking (Current smoker, Former smoker, Nonsmoker), alcohol intake (Yes/No; ‘Yes’ means an individual drinks sometimes or more often, and ‘No’ means an individual never drinks), daily tea (Yes/No), reading (Yes/No; ‘Yes’ means an individual reads books or newspapers sometimes or more often, and ‘No’ means an individual almost never reads), plays cards or mahjong (Yes/No; ‘Yes’ means an individual plays cards or mahjong sometimes or more often, and ‘No’ means an individual almost never plays), physical exercise (Almost never/Several times per week/Everyday; ‘Almost never’ means an individual does exercises over 10 minutes for only several times per month or less), meeting with children (Almost never/Several times per week/Everyday; ‘Almost never’ means an individual meets with children for only several times per month or less), neighbor interaction (Almost never/Several times per week/Everyday; ‘Almost never’ means an individual interacts with neighbors over 10 minutes for only several times per month or less), and social participation (Almost never/Several times per month/Several times per week; ‘Almost never’ means an individual takes part in various social activities for several times per year or never). The number and proportion of each lifestyle variable is presented in Table 1.
Descriptive statistics for demographic and lifestyle variables are presented as the frequency and percentage, continuous variables are described as means and SD, and the Chi-square test was used to evaluate the distribution of the three kinds of frailty status at baseline between groups according to demographics and lifestyle. The health outcomes after 2 years and the three kinds of frailty transitions were also described and compared according to gender and age. Multinomial logistic regressions were used to explain whether the 2-year change of frailty was associated with lifestyle factors, with unadjusted and adjusted coefficients both reported. Windows-based SPSS version 22.0 (SPSS Inc., Chicago, IL, USA) was used for all of the statistical analysis, and a P value of less than 0.05 was considered to be statistically significant.