Participants
Data were collected from the China Health and Retirement Longitudinal Survey (CHARLS), a dynamic cohort study with first-wave data collected in 2011 and three follow-up surveys in 2013, 2015, and 2018, were sponsored by the National Development Research Institute of Peking University, the Chinese Social Sciences Survey Center of Peking University, and the League Committee of Peking University. It is one of the international health and pension survey series and recruit a representative sample of Chinese elders from 150 counties and 450 communities (villages) of 28 provinces in China. The survey areas covered 1.16 billion people, accounting for 85% of the total population in China. Meanwhile, hypertension management is behaviorally demanding and complex [12], and the core component of their care considered to be self-management [14]. Given the prevalence of hypertension and our aging population are increasing, the impact of poor adherence to self-management behaviors on the health of the population are likely to become worse increasingly [4]. Therefore, we choose hypertensive patients as the participants in our study.
The current study used CHARLS’s last three waves of the data in 2013, 2015 and 2018 (recorded as T1, T2 and T3). We selected a total of 2,801 hypertensive patients from 14,277 elders who participated in all three-wave investigation, excluding 429 of them with mental retardation and physical disabilities, and 1864 of them with missing values on analysis variables; the final sample for analyses, therefore, consisted of 508 respondents.
Outcome variables
Self-management behaviors were the interaction of health behaviors and their related processes that patients and families engage in were to care for a chronic condition [39]. On the basis of relevant literatures [7, 10, 11, 40] and CHARLS questionnaires, self-management behaviors in this study encompassed both pharmacological and non-pharmacological management behaviors, including medication use, self-monitoring, physical activity, and tobacco and alcohol avoidance.
Medication use: A single question “Are you now taking any of the following medications to treat or control your hypertension?” was administered to assess participants’ medication use status. Each option that taking Chinese traditional medicine or taking Western modern medicine was scored 1 when the answer was yes and 0 for none of the above, so higher scores indicated better medication use.
Self-monitoring: A question of “During last year (last 12 months), how many times have you had blood pressure examination?” was inquired to inform the patients’ self-monitoring behavior. 0 time of examination was scored by 0, 1 for [1,6), 2 for [6,12), and ≥12 for 3. Higher scores indicated better self-monitoring behavior.
Physical activity: The participation of physical activities (PA) were binary answers to the questions of whether an individual took vigorous physical activities (VPA), participated in moderate physical activities (MPA), or walked for at least 10 minutes continuously every week (WALK). CHARLS defined VPA as activities which made people breathe much harder than normal and might include heavy lifting, digging, plowing, aerobics, fast bicycling, and cycling with a heavy load; MPA as activities which made individuals breathe somewhat harder than normal might include carrying light loads, bicycling at a regular pace, or mopping the floor; WALK as walking those individuals might do solely for recreation, sport, exercise, or leisure. PA in the database was divided into four levels according to exercise intensity, exercise volume and exercise time. The PA standard of level 1 was more than once a week with no less than 30 minutes of VPA each time, more than 3 times a week with no less than 30 minutes of MPA each time, or more than 5 times a week with no less than 30 minutes’ WALK each time. Level 2 PA was no less than 30 minutes of MPA three times a week or no less than 30 minutes of WALK of four or five times a week; Level 3 PA was for no less than 3 times a week with at least 30 minutes’ WALK each time; Level 4 PA was no participating in physical exercise. Scored 0-4 points were from level 4 PA to level 1 PA, and a higher score indicated better physical activity taking.
Tobacco avoidance: To understand patients’ tobacco avoidance status, we used questions of “Have you ever chewed tobacco, smoked a pipe, smoked self-rolled cigarettes, or smoked cigarettes/cigars?” and “Do you still have the habit or have you totally quit?” to investigate the patients’ smoking history. The option of quitting or never smoking was scored for 1, and 0 for still smoking.
Alcohol avoidance: The question of “Did you drink any alcoholic beverages, such as beer, wine, or liquor in the past year? How often?” was used to understand the patients’ alcohol avoidance. The response of having ever drunk was scored for 0, and never drinking was scored for 1.
Independent variable
Self-perceived disease control
Self-perceived disease control of hypertensive patients was reported by a question “Compared to when we interviewed you in R’s LAST IW MONTH, YEAR, is your condition better, about the same as it was then or worse?”. When the patient answered better, he or she was scored for 1, while -1 for worse and 0 for same.
Mediating variable
Usually, individual subjective life expectancy was gathered by subjective probability of survival for a defined age or self-rated life expectancy. To calculate the subjective residua life (SLE), we refer to Spaenjers & Spira (2015)’s study and calculate it as a proxy variable of SLE [41]. The calculation formula is as follows:
Subjective residual life = expected age at death − current age (1)
Expected age at death = average life expectancy + (target age − average life under the same probability) (2)
The “current age” refers to the age of the respondents at the time of each survey.
The “average life expectancy” was determined by the China Life Insurance Mortality Table (2010-2013)[*] (male=79.5 years-old, female=84.6 years-old).
The “average life under the same probability” was searched in the China Life Insurance Mortality Table (2010-2013) by the individuals’ ages and genders, and the same probability refers to an individual’s subjective survival probability (SPS) which was investigated by the CHARLS. The question is: “Suppose there are 5 options, where the lowest option represents the smallest chance and the highest option represents the highest chance, on what option do you think is your chance of reaching the age of [...]?” So, the response options for this item were 1=almost impossible, 2=not very likely, 3=maybe, 4=very likely, 5=almost certain, which correspond to the SPS of 0%, 25 %, 50 %, 75 % and 100 %, respectively.
The “target age” was determined by the current age of the interviewee, as shown in Table 1.
Table 1 Target age (years) in subjective survival probability
Age
|
<65
|
65-69
|
70-74
|
75-79
|
80-84
|
85-89
|
Target age
|
75
|
80
|
85
|
90
|
95
|
100
|
Data Source:CHARLS
Covariates
We included a number of covariates that were known to be associated with self-perceived disease control and self-management behaviors were controlled in our statistical analyses, in order to minimize the disturbing possibility of other variables and to maximize the parsimony of our analytic model [11, 42, 43]. The covariates including gender (0=male, 1=female), age (continuous variable), marital status (1=married, 2=unmarried, 3=others (divorced and widowed)(reference)), Hukou types (0=agricultural Hukou, 1=non-agricultural Hukou), medical insurance (1=urban employee medical insurance, 2=urban and rural resident medical insurance, 3=other medical insurance (reference)), education (1=illiterate(reference), 2=primary school and below, 3= Junior school and above), living arrangement (0=living alone, 1=living with others), comorbidity (Whether the individual have other chronic diseases? (1=Yes, 0=No), these chronic diseases were dyslipidemia (elevation of low density lipoprotein, triglycerides, and total cholesterol, or a low high density lipoprotein level); diabetes or high blood sugar; cancer or malignant tumor (excluding minor skin cancers); chronic lung diseases (such as chronic bronchitis); emphysema (excluding tumors, or cancer); liver disease (except fatty liver, tumors, and cancer); heart attack, coronary heart disease, angina, congestive heart failure, or other heart problems; stroke; kidney disease (except for tumor or cancer); stomach or other digestive disease (except for tumor or cancer); emotional, nervous, or psychiatric problems; memory-related disease; arthritis or rheumatism; asthma). Life satisfaction was assessed by a single question: “Please think about your life-as-a-whole. How satisfied are you with it? (1 =Not satisfied at all; 2=Not very satisfied; 3=Somewhat satisfied; 4=Very satisfied; 5 =Completely satisfied). Social participation was assessed by a single question: “Have you participated in the following social activity in the past month?” for which there were 10 activities (interacted with friend; played mahjong, chess, or cards, or went to community club; provided help to family, friends, or neighbors who did not live with you and did not pay you for the help; went to a sport, social, or other kind of club; took part in a community-related organization; did voluntary or charity work; cared for a sick or disabled adult who did not live with you and who did not pay you for the help; attended an educational or training course; stocked investment; used the Internet.); participating in each activity was scored by 1, otherwise, it was 0; and the range of total score was 0-10 [44]. Self-rated health was evaluated by a question of “What do you feel about your health status?” (1 =Very poor, 2 =Poor, 3 =Very good, 4 =Good, 5 =Fair). Medical intervention status was investigated by the question “Have you ever received any medical intervention?” The intervention items contained blood pressure examination, weight control, physical exercise advice, diet advice, and smoking control. The answer “yes” was scored for 1, ranging from 0 to 5. Depression was determined by using a short form of the Center for Epidemiologic Studies Depression Scale (CES-D10) developed by Andresen et al. (1994). Lei et al. (2014) tested the reliability and validity of CES-D10 by using CHARLS data to confirm the validity of CES-D10 through Chinese population studies, and CESD-10 covered a range of depressive symptomatology with emphasis on current levels of depressive affect [45, 46]. Items were weighted by frequency of symptom occurrence in the last week, using a 4-point Likert-type response format, and each item was rated from 0 (rarely or no time) to 3 (most or all the time). Individuals were divided into three groups based on these ranges identified by Andresen et al. (1994) [45]: 1=depression (score≥10), 0=non-depression (score<10). Activities of daily living (ADL) functional status was an index that indicated individual functional status of middle-aged and elders when they dealt with ADL on their own. According to the international standard ADL index developed by Katz (1963), ADL functional status contained six indices, which were the functional status of eating, dressing, transferring, bathing, using the toilet, and continence [47]. Each item was independently completed with 1 point, otherwise was 0, and the range of total score was 0-6.
Statistical approach
SPSS v26 (IBM Corp 2019) was used to test the demographic differences among self-management behaviors (independent-sample t-test and one-way ANOVA) and the correlation among self-perceived diseases control, SLE, and self-management behaviors (Pearson correlation analysis and multilinear test).
Using two types of separate autoregressive cross-lagged models in Mplus v7.4 (Muthen & Muthen 1998-2015) to estimate the main hypotheses [48]. The first model examined the bidirectional association between self-perceived disease control and self-management behaviors, and the second model added SLE into the first model to test its role in bidirectional mediation. Both two types of full models included stability paths within variables across time (i.e., autoregressive paths), concurrent associations among variables within each assessment wave, and associations among variables across time (i.e., cross-lagged paths). All analyses used the robust maximum likelihood (ML) estimator because the data was in non-normal distribution which was determined by the K-S normality test. Model fit was evaluated by using criterions proposed by Hu and Bentler (1998; 1999), which used multiple fit indices comprising root mean square error of approximation (RMSEA) [49, 50], standardized root mean square residual (SRMR) and the ratio of chi-square to degrees of freedom (χ2/df). Model fit is good when RMSEA<0.06, SRMR<0.08, χ2/df <3.
Furthermore, because the data distribution of variables was skewed, we used the bootstrapping method, an approach for implementing statistical tests and constructing confidence intervals without the use of the traditional statistical assumption of normality, to test the statistical significance of the paths and to compute an estimation of the indirect effect with a 95% CI. The indirect effect was deemed to be significant when the confidence interval did not include zero.
Footnote:
[*] The China Life Insurance Mortality Table (2010-2013) was released by the China Insurance Regulatory Commission in 2016. It collected data of 340 million policies and 1.85 million claims, covering 180 million people, ranking first in the world in terms of sample data. It accurately calculates the mortality probability of people of different ages and genders, and is one of the most authoritative standard for life expectancy measurement in China.