Background. Ambient air pollution has been associated with depression in individual-level studies, yet global comparisons are vulnerable to contextual confounding and rarely distinguish cross-national contrasts from within-country change. Objective. To estimate associations of population-weighted fine particulate matter (PM2.5) and tropospheric nitrogen dioxide (NO2) with depressive-disorder years lived with disability (YLD), test persistence within countries, and quantify residual spatial dependence. Methods. We conducted a spatial ecological panel study of 445 country-years from 153 countries (2019–2021). Depressive-disorder YLD per 100,000 were linked to annual population-weighted PM2.5, satellite-derived tropospheric NO2, gross domestic product per capita, urbanization, region, and year. Pollutants were scaled to interquartile ranges (IQR). The primary pooled model used country-clustered standard errors; fixed-effects, between-country, spatial-error, and sensitivity analyses evaluated robustness. Results. In the pooled model, each PM2.5 IQR (17.994 µg/m³) was associated with −41.66 YLD per 100,000 (95% confidence interval [CI], −80.76 to −2.55; p = 0.037); NO2 showed no statistically significant association (−14.24; 95% CI, −40.06 to 11.59; p = 0.280). The PM2.5 estimate did not persist with country and year fixed effects (6.35; 95% CI, −61.97 to 74.67), balanced-panel fixed effects (10.78; 95% CI, −58.20 to 79.76), or first differences (−1.71; 95% CI, −88.83 to 85.41). Residual spatial autocorrelation was present (Moran's I = 0.134; permutation p = 0.008). Conclusion. The inverse pooled PM2.5 association reflected cross-national structure, not within-country change, and should not be read as protective or causal. NO2 estimates were imprecise. Longer panels with individual-level linkage are needed to separate pollution effects from contextual determinants.
Copyrights © 2026