Kexin Yu

Background Loneliness precedes the onset of cognitive impairment (CI) in older adults. Although the mechanisms through which loneliness “gets under the skin” to influence the risk of developing CI have been conceptually proposed, they are rarely empirically examined. The Evolutionary Theory of Loneliness posits that loneliness as a stressor could cause dysregulations in multiple physiological systems. The current study investigated whether inflammatory, cardiovascular, and kidney biomarkers mediate the longitudinal association between loneliness and CI. Methods Cross-lagged panel models were used to examine the hypothesized relationships, using 2006, 2010, and 2014 waves of data from the Health and Retirement Study (N = 7,037). Loneliness was measured with the 3-item UCLA loneliness scale. CI was assessed with the modified telephone interview for cognitive status. Biomarkers included HbA1C, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, C-reactive protein, and Cystatin C. Using a stepwise model-building approach, first, the model included only loneliness, CI, and biomarker variables; then, sociodemographic covariates were added; lastly, health status were controlled for. Results In unadjusted and partially adjusted models, loneliness was associated with higher odds of worse cognitive status in an 8-year follow-up. Only HbA1C mediated the longitudinal association between loneliness and CI. However, after further controlling for health status, all associations became nonsignificant. Conclusions Examining a large number of participants and linking a limited number of biological markers with cognition and loneliness longitudinally, our empirical data did not support theoretical propositions, highlighting the critical importance of controlling for confounders in future studies examining longitudinal mediational relationships underlying loneliness and CI.