Sex differences in trajectories of depressive symptoms among older Taiwanese: the contribution of selected stressors and social factors

Abstract

Objectives: We assessed female–male differences in depressive symptoms among older Taiwanese and quantified the contribution of sex differences in exposure and response to selected covariates in explaining the gap.

Methods: Using data from six survey waves over 18 years for a nationally representative cohort of 4049 Taiwanese aged 60+, we employed growth curve analysis to model individual-level trajectories of depressive symptoms across age.

Results: Among older Taiwanese, women's disadvantage with respect to social position and employment accounted for about 40% of the sex difference in depressive symptoms. Sex differences in decision control and exposure to widowhood and financial decline played surprisingly little role. Although we found no evidence that the effects of marriage, recent widowhood or recent child death varied by sex, living apart from one's children appeared to be more detrimental for women than for men in this society. Moreover, the effect of living with children depended on the arrangement: living with an unmarried son was more strongly associated with depressive symptoms than living with a married son and daughter-in-law.

Conclusion: Sex differentials in social position and employment are major contributors to the sex difference in depressive symptoms among older Taiwanese, yet sex differences in exposure or response to selected stressors appear to play little role. Differential vulnerabilities to particular living arrangements may also contribute to women's excess psychological distress, although more research is needed to elucidate the mechanisms by which living arrangements influence depressive symptoms in Taiwan.

Keywords:

• depression
• sex differences
• socioeconomic status
• Taiwan
• living arrangements

Acknowledgements

This work was supported by the Demography and Epidemiology Unit of the Behavioral and Social Research Program of the National Institute on Aging (grant numbers R01AG16790, R01AG16661); and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant number R24HD047879). Funding for the TLSA came from the Taiwan Department of Health, the Taiwan National Health Research Institute [grant number DD01-86IX-GR601S] and the Taiwan Provincial Government.

We gratefully acknowledge the hard work and dedication of the staff at the Center for Population and Health Survey Research, Bureau of Health Promotion, Taiwan Department of Health. We thank Germán Rodríquez and Scott Lynch for their comments, suggestions, and statistical advice. We are also grateful to I-Fen Lin for her insights regarding intergenerational co-residence and family relations in Taiwan.

Notes

Note: Based on an 8-item subset of the Center for Epidemiologic Studies Depression scale (CES-D), coded according to standard practice (potential range 0 to 24; higher scores indicated more frequent symptoms; α = 0.82 to 0.86 across the six survey waves). Four of the items reflected somatic symptoms (poor appetite, everything an effort, poor sleep, no energy), three items represented depressive affect (bad mood, lonely, sad), and one item reflected interpersonal relations (people unfriendly).

^a1989: during the past three years; 1993–2007: during the three or four years since the previous survey wave.

^b1989: financial situation compared to 10 years ago; 1993–2007: financial situation compared with prior wave. Ranges from 0 (much better) to 4 (much worse).

^aBased on a socioeconomic index developed for Taiwan (Tsai & Chiu, 1991). Occupations scoring above 60 (which included professional, clerical, sales, supervisory, and other skilled positions) were defined as high status, while the reference group comprised semi-skilled workers, service workers, laborers, farmers, and fishermen as well as those with no major lifetime occupation.

^bThe social connections index was constructed by standardizing each item to have a mean of 0 and SD of 1, calculating the mean across the seven items (α = 0.73), and then rescaling so that zero represents the lowest observed score.

^cThe perceived support index was computed as the mean across relevant items (range 0–4, α = 0.89); each item was coded on a 5-point ordinal scale. All 12 items were included for a married respondent with children/grandchildren, whereas the four items pertaining to a spouse were not relevant for unmarried respondents and the four items pertaining to children/grandchildren were not relevant for those with no surviving descendents.

^dAs with perceived support, this index was calculated by taking the mean across relevant items (range 0–4, α = 0.80), each coded on a 5-point ordinal scale.

^aTo ensure that the intercept is meaningful, we specify age in terms of years after age 60, and we rescale continuous variables so that zero represents the lowest observed value. Thus, the intercept represents the CES-D score at age 60 (in 1989/93) for an individual in the reference group for categorical variables and with the lowest observed value for continuous variables.

Note: ***p < 0.001, ** p < 0.01, * p < 0.05.

Note: Based on the coefficients from Model 3 (Table 4), we identify contributions resulting from sex differences in: (a) the means of covariates (‘exposure’), (b) the effects of those covariates (‘response’), and (c) the residual. The residual represents the estimated sex gap at age x for an individual in the reference group for categorical variables and with the lowest observed value for continuous variables, and it is the only quantity that depends on age. In the table above, we show the residual at age 60 (0.82); at age 90, the residual would be smaller (0.53) because the sex × age interaction is negative (albeit not significant). Thus, at age 90, our model would account for 61% (1–0.53/(0.82 + 0.53)) of the sex difference.