Loneliness and cognitive decline among U.S. adults: A stratified analysis of the BRFSS

Abstract

Background Loneliness is an emerging public health concern linked to adverse mental and physical outcomes. It may play a key role in cognitive aging, yet its population-level association with subjective cognitive decline (SCD) across demographic groups is not well characterized. We evaluated how the frequency of loneliness relates to SCD in U.S. adults and whether associations differ by sex, age and race/ethnicity. Methods We performed a cross‑sectional analysis of adults aged ≥18 years using nationally representative 2016–2023 Behavioral Risk Factor Surveillance System data (BFRSS). Loneliness was categorized as never, rarely, sometimes, usually or always. The primary outcome was self‑reported SCD in the past year. Survey‑weighted logistic regression models adjusted for sociodemographic factors, health insurance, metropolitan status and survey year were used to estimate adjusted marginal probabilities of SCD across loneliness categories. Interaction terms and stratified margins evaluated effect modification by sex, age group (18–44, 45–64 and ≥65 years) and race/ethnicity (non‑Hispanic White, non‑Hispanic Black and Hispanic). Results Among 85,969 adults who reported loneliness, 13,879 (16.2%) experienced subjective cognitive decline (SCD), with a mean age of 65.7 ± 10.6 years. Loneliness showed a strong dose–response relationship with SCD. Predicted probabilities of SCD increased from 9.9% (95% CI, 9.3–10.5%) among respondents who never felt lonely to 15.0% (14.1–15.9%) for rarely, 24.9% (23.6–26.1%) for sometimes, 38.4% (34.4–42.5%) for usually and 45.7% (41.0–50.4%) for always lonely adults (p < 0.001). Women who were always lonely had an adjusted probability of SCD that was 10.7 percentage points higher than men; sex differences were negligible at lower loneliness levels. Age differences were minimal across most loneliness categories; however, among adults who were always lonely, those aged >64 years had significantly lower predicted cognitive function compared with adults aged 18–64 years (p < 0.001). Racial and ethnic differences were modest; the only significant contrast was a 1.7 percentage‑point lower probability of SCD for non‑Hispanic Black adults compared with Whites among those who never felt lonely. Other subgroup differences were not statistically significant. Conclusions Loneliness is independently and strongly associated with higher likelihood of subjective cognitive decline among U.S. adults, and this relationship is most pronounced for chronic loneliness. While sex and age modified the effect of loneliness, racial/ethnic disparities were minimal. These findings identify loneliness as a modifiable social determinant of cognitive health, supporting the need for broad social connection initiatives and targeted efforts for women and mid-life adults with chronic loneliness.