Participants and survey design
This study uses data from the second-, fourth- and sixth-wave surveys of the Study of Health and Living Status of the Middle-Aged and Elderly in Taiwan, a longitudinal, multidisciplinary national survey representing the population of individuals aged 50 and over in Taiwan. The first-wave survey of the study was conducted in 1989 on people aged 60 and over. The subsequent surveys were conducted every three years. To maintain the representation of people aged 60 and above and to extend the elderly population to individuals aged over 50, a second cohort was included in the second survey in 1996, and a third cohort was included in the fifth survey in 2003.
Written informed consent was obtained from the participants before their participation. The research protocol was approved by the Institutional Review Board of the Bureau of Health Promotion, Department of Health, R.O.C. (Taiwan).
In this study, we used information from the second, fourth and sixth waves of the surveys, which were conducted in 1993, 1999 and 2007. A total of 3155 individuals aged 64 and over comprised the sample of the second wave survey in 1993, the sample of the fourth wave survey in 1999 included 4440 persons aged over 53, and the sample in 2007 included 4534 persons aged 54 and over. For comparability, only individuals 60 years old and older were included in our analysis. To evaluate the impact of intergenerational family transfer, including co-residence with a partner, co-residence with children and care for grandchildren, on elders’ loneliness and depression, only those elders who had grandchildren and were capable of completing the Center for Epidemiological Studies Depression Scale (CES-D scale) themselves were included in this study. In addition, if an individual participated in more than one wave of the survey, only the information from the last wave was used in the study. Overall, there were 914 elders in 1993, 1792 elders in 1999 and 2292 elders in 2007 who were included in the final analysis.
Elders’ personal characteristics, including age, gender, education, ethnic group, living place and employment status, were used in the study. In addition, information about their self-reported health status, collected with the question “how do you feel about your current health status?”, was used in the analysis. Responses were provided using a 5-point Likert scale, ranging from “very good” to “very bad”.
To evaluate elders’ intergenerational family transfers, elders were asked whether they were currently co-residing with an adult child, whether they were currently co-residing with a partner and whether they currently cared for their grandchildren to help their adult children. In the 1999 and 2007 questionnaires, the frequency of taking care of grandchildren, rated as “usual” or “sometimes”, was also collected. To compare this with the data from 1993, responses of “usual” and “sometimes” were classified as “yes” in this analysis.
The CES-D scale was used to determine elders’ depression in the study. In 1993, there were only 5 items from this scale that were evaluated in the questionnaire. These items were “I was in a bad mood”, “I felt lonely”, “People were unfriendly”, “I felt sad” and “I had no energy for things”. Responses were provided using a 4-point Likert scale, ranging from “none” to “always”. To compare the data with those from 1993, those 5 items in 1999 and 2007 were used for analysis. The internal consistency reliability of scale scores of all waves of the Study of Health and Living Status of the Middle-Aged and Elderly in Taiwan was confirmed in a previous study [18].
Statistical analysis
Two approaches were used to analyze the data. For summary statistics, the chi-square test was used to compare elders’ personal characteristics, including age, gender, education, ethnic group, living place, employment status and self-reported health status, between the 1993, 1999 and 2007 samples. Elders’ intergenerational family transfers, including co-residence with partners, co-residence with their children and care for their grandchildren, as well as their loneliness and depression, as evaluated by the CES-D scale, were also compared among the 1993, 1999 and 2007 samples by chi-square test.
Each CES-D scale item was classified into two levels, either “depressed” or not; thus, responses other than “none” were classified as “depressed”. All CES-D item scores were also totaled to represent the levels of elders’ depression. The respondents in the upper tertile of CES-D scores were defined as a high-risk depressive group. Then, multiple logistic regressions were performed to determine the associations between elders’ intergenerational family transfers and their depressive symptoms for each year and for the total samples after controlling for potential confounding factors, including age, gender, education, ethnic group, living place, employment status and self-reported health status. Observations with missing data in any variable were excluded from the multivariable model.
The odds ratios (ORs), the 95% confidence intervals (CIs) and the 0.05 significance level were calculated using the software SPSS version 18.0.