Understanding the Relationship Between Depression Symptom Severity and Health Care Costs for Patients With Treatment-Resistant Depression

ABSTRACT

Objective: To examine whether measures of depression symptom severity could improve understanding of health care costs for patients with major depressive disorder (MDD) or treatment-resistant depression (TRD) from the health plan perspective.

Methods: In this retrospective cohort study within an integrated health system, cohorts consisted of 2 mutually exclusive groups: (1) adults with TRD based on a standard treatment algorithm and (2) adults with MDD, but no TRD, identified through ICD-9/10-CM codes. Depression severity was measured using the Patient Health Questionnaire-9 (PHQ-9). Patterns of health care resource utilization (HRU) and costs were compared between the TRD and MDD groups overall and within the groups at different symptom levels. A general linear model with a γ distribution and log link for cost outcomes, logistic regression for binary outcomes, and negative binomial regression for count outcomes were used.

Results: Patients with TRD (n = 24,534) had greater comorbidity than those in the MDD group (n = 17,628). Mean age in the TRD group was 52.8 years versus 48.2 for MDD (P < .001). Both groups were predominantly female (TRD: 72.8% vs MDD: 66.9%; P < .001). Overall, the TRD group had greater costs than the MDD group, with 1.23 times (95% CI, 1.21–1.26; P < .001) greater total cost on average over 1 year following index date. Within both groups, those with severe symptoms had greater total mean (SD) costs (TRD: moderate: $12,429 [$23,900] vs severe: $13,344 [$22,895], P < .001; low: $12,220 [$31,864] vs severe: $13,344 [$22,895], P < .001; MDD: moderate: $8,899 [$20,755] vs severe: $10,098 [$22,853]; P < .001; low: $8,752 [$25,800] vs severe: $10,098 [$22,853], P < .001).

Conclusions: MDD and TRD impose high costs for health systems, with increasing costs as PHQ-9 symptom severity rises. Better understanding of subgroups with different symptom levels could improve clinical care by helping target interventions.