Defining treatment-resistant depression in a real-world setting: impact of a staging model on clinical features and treatment trajectories

Operationalizing treatment-resistant depression (TRD) is essential to guide the rational use of pharmacological and somatic interventions. This study applied a staged TRD model in a real-world inpatient setting to disentangle patterns associated with different degrees of resistance. We prospectively included 538 depressed inpatients and classified them as nonresistant, TRD1 (≥2 antidepressant failures at minimally licensed dose for ≥4 weeks), or TRD2 (≥2 failures of antidepressants from different classes at maximum tolerated dose for ≥4 weeks). Baseline features, comorbidities, and treatments were compared across groups. Depressive symptoms were assessed at admission and at 2, 4, and 8 weeks. TRD1 patients (24%) had an earlier onset and showed the highest prevalence of personality disorders, anxiety disorders, tobacco smoking, AUD, and substance use disorder. In contrast, TRD2 patients (29%) presented with more severe and longer depressive episodes and required more intensive pharmacological and somatic strategies. Only the TRD2 group displayed a significantly attenuated symptomatic improvement over time (time × group interaction; P < 0.001) compared with nonresistant. In a real-world cohort, a simple two-stage TRD model differentiated a “complex but responsive” subgroup (TRD1) from a more biologically refractory profile (TRD2). Graded definitions may reduce misclassification, supporting more tailored management of comorbid conditions.