Answers are grounded in this paper and cite their sources.
Background Lifetime episode count and illness duration may convey different information about peripheral inflammation, but estimates from small cross-sectional samples require cautious interpretation. Methods This exploratory analysis examined adult inpatients with bipolar or unipolar depression recruited in Shenzhen, China, during 2020–2022. Primary log-linear models jointly included log 2 (episode count + 1) and log 2 (illness duration in years + 1), adjusting for age, sex, body mass index, diagnosis, and 17-item Hamilton Depression Rating Scale (HAMD-17) total. HC3 standard errors and Benjamini–Hochberg adjustment across six burden coefficients were used. Results Primary analyses included 63 patients for IL-6 and TNF-α and 57 for IL-1β. Per doubling of episode count plus one, conditional associations were observed for TNF-α (geometric-mean ratio [GMR], 2.238; 95% confidence interval [CI], 1.424–3.517; q = 0.004) and IL-6 (GMR, 3.591; 95% CI, 1.305–9.882; q = 0.029), whereas the IL-1β estimate remained inconclusive (GMR, 1.767; 95% CI, 0.815–3.831; q = 0.175). Duration was inversely associated with TNF-α at fixed episode count (GMR, 0.622; 95% CI, 0.442–0.877; q = 0.023). The TNF-α episode association was nonsignificant without duration adjustment and did not meet multiplicity-adjusted significance after excluding annotated records (n = 47; q = 0.130). A model with fewer covariates supported the association within the original sample, but its estimate was less precise in all 75 age-eligible patients (GMR, 1.815; 95% CI, 0.950–3.467; q = 0.106). Excluded eligible patients differed in depressive severity and inflammatory-marker distributions. Conclusion The associations did not consistently meet multiple-testing-corrected significance thresholds in sensitivity analyses and should be interpreted strictly as hypothesis-generating, requiring confirmation in larger longitudinal cohorts. Missing treatment and assay-quality information further limits interpretation. The results do not establish causality, neuroprogression, protective effects of longer illness, or predictive utility.
No takes yet. Share an insight, caveat, or question.
Han et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: