Sex modifies the relationship between depression and risk for dementia: implications for targeted prevention

Alzheimer’s disease and related dementias (ADRD) are projected to affect 13.8 million Americans by 2060, with women comprising nearly two-thirds of those diagnosed1. Depression, a highly prevalent and modifiable risk factor, is nearly twice as common in women and has been independently linked to elevated ADRD risk2-4. Yet, the joint effect of sex and depression on cognitive decline remains under-explored in large-scale longitudinal studies. We analyzed 75,069 visits from 11,091 participants aged 65 and older, using data from 42 Alzheimer’s Disease research centers across the US. The final sample had a mean of 9.71 visits (SD=3.43; range: 2-19) over an average follow-up period of 3,679 days, equivalent to approximately 10 years (range: 5-18 years). We used Cox proportional hazards models to examine time to cognitive impairment, defined as a clinical diagnosis of mild cognitive impairment or dementia, in relation to Geriatric Depression Scale (GDS-15) score, sex, age, and apolipoprotein E ε4 (APOE ε4) genotype. For ease of interpretation, depression was coded as binary (depressed: GDS-15 score ≥5; not depressed: GDS-15 score <5). Controlling for age and APOE ε4 genotype, depression increased the risk of cognitive impairment by 77% in men (hazard ratio, HR=1.77, 95% CI: 1.68-1.87, p<0.001) and by 119% in women (HR=2.19, 95% CI: 2.02-2.38, p<0.001). The interaction between depression and sex was statistically significant (HR=1.24, p<0.001). While women had lower baseline hazard of cognitive impairment (HR=0.62), they experienced a disproportionately greater increase in impairment risk when depressed. We then used a time-varying covariate model, which accounts for changes in depression status across repeated observations, to test whether risk trajectories differed over time. In this model, at baseline, men with depression had a 165% higher risk of cognitive impairment (HR=2.65), while women had only a 19% increase (HR=1.19). The depression-sex interaction was HR=0.45. However, over time, men who became depressed were 23% less likely to develop cognitive impairment, while women who became depressed were 19% more likely to develop impairment at some point. Therefore, the depression-sex interaction at follow-up became HR=1.55 (95% CI: 1.12-2.15, p<0.01). This suggests that depression that emerges or persists later in life may have a more harmful effect on women’s brain health than men’s. Importantly, these effects remained consistent across older age groups, indicating that the depression-sex interaction is stable across the later life course. In additional analyses, we modeled depression continuously and found that each additional point on the GDS-15 increased the risk of cognitive impairment by approximately 14.5% (HR=1.14, 95% CI: 1.139-1.150, p<0.001). We also categorized depression severity into three groups using GDS-15 scores: normal (0-4), mild (5-8), and moderate-to-severe (9-15). Our analyses showed a clear dose-response relationship between depression severity and risk of cognitive impairment. Mild depression was associated with a 69% increased risk (HR=1.69; 95% CI: 1.57-1.82), while moderate-to-severe depression increased risk by 80% (HR=1.80; 95% CI: 1.59-2.04). Importantly, time-varying effects indicated that the impact of moderate-to-severe depression slightly increased over time (p<0.001), suggesting cumulative risk, whereas the impact of mild depression remained stable. These results were adjusted for age, APOE ε4 genotype, and accounted for time effects, underscoring the importance of recognizing even mild depression as a significant and clinically actionable ADRD risk factor. To assess the robustness of our findings, we conducted several sensitivity analyses. First, we examined attrition bias, and found that dropout rates were low and comparable between sexes (men: 1.6%, women: 1.5%, p=0.72). Next, we assessed baseline depression severity, finding that men had slightly higher mean GDS-15 scores than women (4.14 vs. 4.02; Wilcoxon W=110,604,030, p<0.001). We conducted an additional test to assess significant differences in depression severity by sex: for the 6,186 observations in the depressed group, there were none (W=4,475,864, p=0.07). Therefore, any sex differences in cognitive impairment predicted by depression are not driven by differences in depression severity between groups. Third, we accounted for center-level differences, by including ADRD research center-level frailty terms in all models. Finally, we expanded the sample to include participants under age 65, and found that the depression-by-sex interaction remained significant (HR=1.22, p<0.001). Together, these results indicate that sex modifies the relationship between depression and ADRD, and that this effect is not due to measurement or sampling bias, but likely reflects a biologically meaningful difference. Several mechanisms could account for women’s heightened susceptibility to cognitive impairment. Compared to men, women with depression may exhibit greater neuroinflammatory responses5, more rapid hippocampal atrophy6, and heightened hypothalamic-pituitary-adrenal axis dysregulation7, all of which may magnify the long-term neurodegenerative impact of depression. Clinically, our findings support the prioritization of depression screening and treatment as a key strategy for ADRD prevention, especially for women. Despite global efforts8, 9, sex remains underutilized as a stratifying factor in both research and practice. Furthermore, depression is often treated as a static covariate in models of cognitive decline, despite being dynamic and potentially interactive. We offer the following recommendations for future research and clinical care: a) routine screening and early intervention for depression, especially in mid- to late-life women; b) inclusion of sex-stratified and interaction models in ADRD risk prediction and observational research; c) clinical trials of depression treatment that examine cognitive outcomes by sex across the lifespan; d) mechanistic studies exploring sex-specific inflammatory, hormonal, and stress-related pathways linking depression and neurodegeneration. In sum, our longitudinal analysis of over 75,000 visits across US Alzheimer’s Disease research centers revealed that depression significantly increases ADRD risk for both sexes, but the trajectory and magnitude differ by sex. The interaction between sex and depression persists across time, analytic methods, and population subsets. Women with depression are at disproportionately higher risk for cognitive impairment as they age. These findings underscore the urgency of tailoring ADRD prevention strategies to account for sex-specific vulnerabilities.