Study participants
Data were from the CLHLS, which is an ongoing nationally representative prospective cohort study investigating factors that influence the health, longevity, and mortality of Chinese older people. More details were available in other studies[21, 22]. The study was initiated in 23 of 34 provinces/municipalities/autonomous regions. Seven waves of data collection were collected every two or three years since 1998, and the latest round was finished in 2018. People who had undergone rigorous training and examinations were selected as interviewers. The current study was based on the 2008, 2011, 2014, and 2018 waves of CLHLS. We excluded 24,951 older adults who were followed up repeatedly and 127 participants who were less than 60 years of the 49,785 interviewees. Respondents who failed to complete SWB or ADL assessment and who were diagnosed with dementia were also excluded. Finally, a total of 13,010 older persons were selected for the study.
Measurement Of Subjective Well-being
Consistent with previous studies, SWB was assessed by eight items[10], which included a person's overall life satisfaction and his or her affective states. Life satisfaction was measured by the question: “How do you rate your life at present?” Positive emotions and negative emotions were two components of affective states. Positive emotions were measured by four questions as follows: 1) “Do you always look on the bright side of things?” 2) “Do you like to keep your belongings neat and clean?” 3) “Can you make your own decisions concerning your personal affairs” 4) “Are you as happy as when you were younger”. Usefulness, anxiety, and loneness, three aspects to reflect negative emotions, could be specifically described as follows: 1) “Do you often feel lonely and isolated?” 2) “Do you feel the older you get, the more useless you are?” 3) “Do you often feel fearful or anxious?”. The response score ranged from 5(very good/always) to 1(very bad/never). For negative emotions, we reversed scoring, that is, “5” means never, while “1” means always. Therefore, the eight questions above ranged from 8 to 40, with higher scores indicating better SWB. We converted SWB to a dichotomous variable. If the score was greater than 30, the participants were considered to have a high level of SWB, if not, they were deemed to have a low level of SWB[9, 10]. The reliability alpha (Cronbach’s alpha) was 0.704, indicating that the internal consistency coefficient for the indices was satisfactory[23].
Measurement Of Adl
Modified Katz index, covering bathing, feeding, continence, indoor transferring, dressing, and toileting, was used to evaluate BADL[24]. Adapted from Lawton scale[25], eight questions as follows were used to assess IADL: 1) “Can you visit your neighbors by yourself?” 2) “Can you go shopping by yourself?” 3) “Can you cook a meal by yourself whenever necessary?” 4) “Can you wash clothing by yourself whenever necessary?” 5) “Can you walk continuously for 1 kilometer at a time by yourself?” 6) “Can you lift a weight of 5kg, such as a heavy bag of groceries?” 7) “Can you continuously crouch and stand up three times?” 8) “Can you take public transportation by yourself?” The coding scale of them was a 3-point scale, this is, 3 represents “no, can’t”, 2 represents “yes, but need some help”, and 1 represents “yes, independently”. The total score of the modified Katz index and adapted Lawton scale varies from 6 to 18 and 8 to 24, with a higher score indicating worse BADL or IADL. The statistical results showed that the internal consistency coefficients for BADL and IADL were alpha = 0.858 and 0.949, respectively. Therefore, both of the alpha values were considered acceptable[23]. A total score of 7 or higher was used as the cutoff for dependence in BADL and a score of 9 or more indicated IADL limitation[26, 27]. The ADL score was constructed from the sum of the score of BADL and IADL.
Covariates
At baseline, a structured questionnaire was used to collect information on sociodemographic characteristics and health-related factors. The sociodemographic characteristics included age, sex, living residence, region, ethnicity, primary occupation before retirement, living arrangement, marital status, income status, and educational level. Age was categorized into three levels:60–69 years old, 70–79 years old, and ≥ 80 years old. Primary occupation before retirement was divided into two categories, namely peasants and others. Living arrangement was composed of three categories, namely with household member(s), alone, and in an institution. According to the National Bureau of Statistics[28], the region was divided into east, central, west, and northeast. Marital status was grouped into five categories, namely married, separate, divorced, widowed, and never married. Income status was evaluated with the following question: how do you rate your economic status compared with others in your local area? (very rich, rich, fair, poor, and very poor). Educational level was categorized as illiterate and educated. Health-related factors included self-reported health, current smoking status (yes or no), and current drinking status (yes or no). Self-reported health was measured by asking the participants how they rate their health at present, with options of very good, good, fair, bad, and very bad. To avoid a small percentage of some categories, data on income status and self-reported health were combined into three groups (income status: rich, fair, and poor; self-reported health: good, fair, and bad). 1.3% of the total data items of covariates were missing, and we used multiple imputation to handle them[29].
Statistical analysis
In this study, SWB is the outcome variable. Various methods were applied to deal with SWB, some researchers regarded it as a continuous variable[30], while others treated it as an ordinal categorical variable[10]. Normality test confirmed that scores of SWB, BADL and IADL were not normally distributed(P < 0.001). Frequency with percentage was to express categorical variables. Baseline characteristics are summarized according to different levels of SWB and compared between participants using the χ2 test. Mann-Whitney U was used to compare the differences in SWB among the elderly with different ADL scores, which were described as medians and interquartile ranges. Spearman correlation coefficients were used to measure the correlation between ADL and SWB. Binary logistic regression was performed to determine the impact of ADL on SWB. Before conducting binary logistic regression analysis, collinearity diagnosis analysis was conducted to assess whether there was a strong linear relationship between covariables. Tolerance less than 0.1 or variance inflation factor (VIF) 10 was used to identify multicollinearity[31].
All statistical analyses were performed using SPSS statistical software version 26.0 (IBM SPSS Inc., New York, NY, USA).