Dear editor
I read with interest the cross-sectional study by Hu et al, which examined intergenerational support, social networks, attitudes toward aging, and depressive symptoms among 1157 older adults in China. The integration of familial and social resources is thus valuable. However, several inconsistencies in the published descriptive and mediation results warrant clarification.1
First, the published total means do not equal the sums of their corresponding component means. The intergenerational support subscale means (10.57, 6.97, and 7.82) sum to 25.36, versus the reported total mean of 25.61. Family- and friend-network means (5.86 and 6.01) sum to 11.87, versus a social-network total of 13.61.
Aging-attitude component means (26.16, 26.06, and 24.62) sum to 76.84, versus a total of 80.58. Rounding alone cannot explain these differences. This comparison assumes the same respondents and compatible scoring conventions; importantly, the psychosocial-loss component is reverse-scored.
Clarification of total-score derivation, item recoding, and respondent denominators would help reconcile these summaries.
Second, the interpretation of the two-way emotional-support mediation pathway is difficult to reconcile with the reported effect signs. The bootstrap analysis yields a negative direct association with depressive symptoms (−0.24; 95% CI −0.307 to −0.171) and a positive indirect association through attitudes toward aging (0.017; 95% CI 0.005 to 0.039). Because higher Geriatric Depression Scale scores indicate more depressive symptoms, the reported positive indirect association would offset, rather than reinforce, the negative direct association.
In contrast, the family-network indirect estimate is negative (−0.031; 95% CI −0.069 to −0.011). Opposing direct and indirect signs can arise in inconsistent mediation or suppression, but their interpretation differs from a uniformly protective mediation pathway. The published estimates should be reconciled with the conclusion that reciprocal emotional support indirectly mitigates depression.2
Third, the structural equation modeling results describe the association from bidirectional emotional support to aging attitudes as negative, while displaying β = 0.113; similarly, the association from aging attitudes to depression is described as negative but displayed as β = 0.163. Both numerical coefficients are positive. Figure 1 depicts the proposed paths but does not display signed estimates.
Clarification of the coefficient signs and scale orientation, ideally supported by the signed model estimates, is important for interpreting the positive indirect association.
Finally, consistent with the authors’ stated limitations, a single cross-sectional assessment cannot establish the temporal sequence of intergenerational support, aging attitudes, and depressive symptoms.3 Reconciling the scale summaries and pathway directions would strengthen confidence in the reported associations without presuming errors in the underlying dataset or analysis.