Influence of culture, residential segregation and socioeconomic development on rural elderly health-related quality of life in Guangxi, China

This cross-sectional community-based survey was carried out among ethnic groups in the rural areas of Guangxi Province. The province consists of 12 minority autonomous counties and 58 minority autonomous townships based on areas inhabited by ethnic minorities. In each autonomous township, single ethnic minority households should account for over 30 % of the total population [15]. We selected the study sample based on the principles of both the concentration of the ethnic group and maintenance of culture and tradition. To begin with, all of the 58 minority autonomous townships were divided into 7 groups based on the areas inhabited by minorities such as Zhuang, Yao, Miao, Dong, Mulao, Maonan and Jing. In each study township, high-, middle- and low-income population groups were determined based on annual household income at township level. Then, we randomly selected one township per ethnic group from each of these three groups, yielding a total of 24 selected townships. Each township was divided into three blocks based on the size of the ethnic population and adjacent geographical location. In each of the 72 blocks, in order to guarantee a sufficient representative sample size, two minority villages were chosen from the list of villages based on probability proportional to size if the number of the total villages in each block was more than two; otherwise, one minority village was chosen. A total of 116 villages were selected in the final reckoning. Finally, we obtained a list of the individuals aged ≥ 60 years from the village administrative committee in each selected village, and we used simple random sampling to select eligible individuals from each village.

The HRQoL of the elderly was evaluated using the European Quality of Life - 5 Dimensions - 3 Levels questionnaire (EQ-5D-3 L), which is a standardized health-related quality of life questionnaire developed by the EuroQol Group in 1990 [16]. Up to date, the EQ-5D-3 L has been translated into more than 160 official language versions, including the Chinese version, which was applied in this study [15]. The Chinese version of the EQ-5D-3 L instrument has demonstrated acceptable construct validity and fair to moderate levels of test-retest reliability in general populations [17, 18], and an ability to distinguish well for known groups [3, 19].

The instrument classifies respondents’ current health status into five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) employing three response levels (no problems, some or moderate problems, extreme problems), which theoretically results in 243 unique health states. The EQ-5D has also a visual analog scale (VAS) part, allowing respondents to evaluate their current health status on a range from 0 (representing the worst health status) to 100 (representing the best health status). In addition, we calculated the EQ5D index values as an aggregated utility index based on the value set that has become recently available for the Chinese version of the EQ-5D-3 L instrument [20].

The elderly characteristics comprised age (years), gender, ethnicity, marital status, educational attainment and annual average income that was assessed based on the average income for each family member living in household over the previous year.

Household-level variables included empty-nested family that is defined as one containing elderly individuals with no children or whose children lived far away from them, type of housing (brick-, earth- and wooden-structure) and household health insurance meant participation in China’s National Cooperative Health Insurance Scheme.

Residential segregation was assessed using these three indicators: monocultural village status, geographical setting and distance to the nearest county seat. A village comprising more than two minorities, and the proportion of the majority ethnic population was not less than 70 % of the total was defined as monocultural one. A community of people as monoculture that the majority population should have such a predominance so that the prevailing overall culture of that setting is dominated by the culture of the majority population. Geographic setting of village (flatland, hilly and mountainous area) and distance to the nearest county seat that was defined as how far, on average, the village was from the nearest county administrative seat. In general, rural mountainous areas constitute the harshest natural living environment, consisting of poorer infrastructure and more deficient arable soil than other environments. As a rule, the longer the distance, the poorer the village is.

A structured questionnaire was used in individual faceto-face interviews to obtain information on the elderly and their households. All of them were interviewed at their home using their local language or dialect by trained interviewers who were recruited from Guilin Medical University. In order to ensure the accuracy and comparability of the data collection, a workshop was conducted before the commencement of fieldwork to teach them how to use the questionnaire and check the interview. Village information was also collected from the administrative committee of the village in which the participant lived.

During the data collection process, the respondents were given a full explanation of the research purpose before being invited to participate, and, after they signed the informed consent, a face-to-face interview was conducted. As a quality control, the supervisors checked the completeness of the questionnaire at the end of each day. If information was missing, the interviewer went back on the same or the following day to obtain the missing information.

The characteristics of the respondents were summarized in terms of frequency and percentage for categorical variables or mean and standard deviation for continuous variables. The distribution of the respondents by ethnicity was calculated together with their percentages. Additionally, the percentage of problems reported in each EQ-5D dimension by ethnic group was also calculated. Descriptive statistics with 95 % confidence intervals were provided for the EQ-VAS, five dimensions and index values for each ethnic population by socio-demographic status and residential segregation.

The multilevel linear regression model was employed to predicate the EQ-5D index values according to sociodemographic status, ethnic culture and residential segregation. In the multilevel analysis, the individual characteristics were set at the first level, and the family and village information were set at the second and third levels, respectively. The p-value of the likelihood ratio to the chi-square was used as a guide to the model’s goodness of fit. All p-values were two-tailed and statistical significant level was set as less than 0.05. Finally, data analysis were performed in R software (R version 3.2.2) using epicalc, lme4 and sjPlot packages.

An ethical consideration application form for this study was submitted to and approved by the Ethics Committee of the Faculty of Medicine, Prince of Songkla University, Hat Yai, Songkhla Province, Thailand (Reference number: 57-188-18-5), and further endorsed by the Ethics Review Committee of Guilin Medical University, Guilin, Guangxi Zhuang Autonomous Region, P.R. of China before the research was carried out.

In total, 6,998 eligible elderly from 5,541 households in 116 villages agreed to join the survey giving a response rate of 93 %. The demographic characteristics of the elderly by ethnicity were summarized in Table 1. The age range of the respondents participating in the study was 60 ~ 105 years. Nearly half of the entire study subjects belonged to the 60 ~ 69 years age group, except for the Jing minority, whose mean age was 74.7 years. Over four-fifths of respondents had a primary school and below educational level. The highest illiteracy rate was found in the Miao minority, which was three times that of the Zhuang minority. Over 70 % of the respondents who belonged to the Yao, Miao and Dong minorities reported an average annual income less than 3,000 Yuan, but 40 % of the Jing minority the income was higher than 10,000 Yuan. This figure was nearly 40 times higher compared with those reported in both the Miao and Maonan groups. Nearly half of both the Zhuang and Han elderly lived in empty-nested family, whereas this proportion was less than 12.5 % among the Miao communities. Almost all of both the Miao and Dong elderly lived in housing quarters constructed wholly with wooden material, and in villages that were farther away from the nearest county seat than their counterparts. Meanwhile, the Jing and Zhuang minorities mainly lived in brick house and flatland areas.

The respondents’ self-reported problems based on the EQ-5D dimensions by ethnic group were summarized in Table 2. Across the five dimensions, pain/discomfort was the most prominent domain, followed by anxiety/depression, mobility and usual activities. Despite the fact that self-care was the least reported, the proportion of respondents experiencing severe problems with self-care was higher than that for the mobility domain. Overall, the EQ-5D index values were plotted against the mean VAS scores before and after adjustment for other variables (Fig. 1) across the ethnic groups. The correlation coefficients were 0.82 and 0.77, respectively. The Zhuang elderly had the largest values on both scales, whereas a lower level of consistency was observed in the other ethnic groups on these two scales.

Table 3 summarizes the VAS, the index values and problems reported in all EQ-5D dimensions by SES and residential segregation. The oldest age group people was more likely to be associated with lower VAS and index values, and a higher problem frequency in each EQ-5D dimension, except for the anxiety/depression dimension. Married individuals reported fewer problems than both those who were single or divorced. Overall, the elderly with illiteracy tended to report more problems and lower VAS scores. The EQ-5D index values tended to rise gradually with an increase in annual income. Concerning residential segregation, the elderly living in wooden-structure housing or farther away from the county administrative seat had lower VAS scores than those living in brick house of the flatland.

Table 4 shows the findings from the multilevel linear regression model, which was employed to analyze the predictors of HRQoL in these elderly populations. As can be seen, significant differences in EQ-5D index values were noted in age group, gender, ethnicity, educational level and household income. Compared with the Han ethnic group, the Zhuang minority showed higher values, whereas the Yao minority had significantly lower scores. The difference in values between the Zhuang and Yao groups was 0.046. Nevertheless, no significance was observed among the subgroups of marital status, empty-nested family, health insurance, type of housing, monoculture, distance from county seat and geographic setting of village.

Despite its large sample size and high response rate, information was collected by face-to-face interviews. Barriers from the participants’ local language or dialect, language bias should be taken into account. Secondly, detail attributes on the ethnic cultural belief have not been collected to be included in statistical analyses in this study yet. Future studies are needed to obtain some qualitative information on the diversity on lifestyle choices, religion and value in order to understand the way in which ethnic and cultural differences influence health behaviors and HRQOL. Finally, the relationship between HRQoL and its predictors may not be causal due to the cross-sectional nature of the study.