Integrating clinical proxies and metabolic data identifies and distinguishes high-risk depression subtypes in a real-world first-hospitalization cohort.

BACKGROUND

Early identification of high-risk depression subtypes, specifically, recurrent (RD) and treatment-resistant (TRD) depression, is critical for improving long-term outcomes, yet practical stratification tools based on routinely available clinical and metabolic data remain limited. This study aimed to characterize these subtypes within a real-world, first-hospitalization cohort by integrating clinical proxy indicators with metabolic biomarkers.

METHODS

In a cross-sectional analysis of 1,436 first-hospitalized patients with first-episode depression (FED) and RD, we compared demographic, clinical, and metabolic characteristics. TRD was operationally defined by electroconvulsive therapy (ECT) exposure. Multivariable logistic regression identified factors associated with RD (vs. FED) and TRD (within RD).

RESULTS

Compared to FED patients, RD patients were older (47.1 vs. 42.4 years, p<0.001), had longer hospital stays, and exhibited a worse metabolic profile, including higher triglycerides (1.53 vs. 1.39 mmol/L, p = 0.014) and greater prevalence of elevated glucose (23.4% vs. 19.0%, p=0.047) and low HDL-C (39.5% vs. 31.4%, p=0.002). A disease duration >12 months was the strongest factor associated with a first-episode hospitalization diagnosis (OR = 10.75, p<0.001). Among RD patients, those with TRD (n=112) were distinguished by a higher rate of completing a 6-week treatment observation period (82.1% vs. 59.1%, p<0.001) and a greater prevalence of documented suicide risk (25.9% vs. 13.1%, p < 0.001). Completion of the observation period was the strongest predictor of TRD status (OR = 3.04, p < 0.001).

CONCLUSION

In first-hospitalized patients, RD is associated with adverse metabolic markers, while TRD is characterized by clinical indicators of failed adequate treatment and high acute risk. A prolonged illness duration in first-episode patients may signal significant treatment delay. An integrated assessment of these accessible clinical and metabolic proxies could facilitate early risk stratification in routine care.