Vice President, Research OM1, Inc. Tufts University School of Medicine Boston, United States
Background: The relationships between social determinants of health and major depressive disorder (MDD) are complex. There is interest in understanding the role of financial risk factors in disparities in MDD.
Objectives: To describe the association between credit risk score and measures of MDD burden using observed and estimated PHQ-9 (PHQ-9/ePHQ-9) in a US real-world cohort.
Methods: This retrospective observational cohort study was conducted in the PremiOMTM MDD Dataset (OM1, Boston, MA), a multi-source real-world data network with linked health care claims, social determinants of health, and electronic medical records data on US patients with MDD from all 50 states. For this study, adult patients with MDD were identified during the study period (3/2015 - 2/2021). Patient age, race, sex, insurance type, and PHQ-9 and ePHQ-9 scores (estimated by a machine learning model) were assessed on the date of initial diagnosis (index). In subgroups defined by credit risk score, PHQ-9/ePHQ-9 scores and number and percent of patients with mental health care visits and antidepressant prescriptions were assessed in the 12 months after index.
Results: The study included 3,469,613 patients [68% female, mean age 54 years (SD=18), 89% White]. Median household income for patients with MDD and high risk credit scores was $47K (Q1-Q3=$31-66K) compared to $80K (Q1-Q3=$50-130K) for patients with low risk credit scores, with 15% and 3% having household incomes <$25K, respectively. The proportion of Black patients in the high risk group was 16% compared with 5% in the low risk group.
For patients with available PHQ-9/ePHQ-9 scores at index, the mean score for patients with high credit risk was greater than that of low credit risk patients (13.3 vs 12.4; P < .001). Among high credit risk patients, 45% had a PHQ-9 ≥ 15 (moderately severe to severe MDD) versus 40% of low risk patients. Emergency care (3.3% vs 1.9%; p < .001) and inpatient mental health care use (9.7% vs 6.8%; p < .001) was higher in high risk patients while outpatient mental health visits were lower (30.2% vs. 31.2%; p < .001). Prescription fills for antidepressant therapy in the 12 months post index were significantly lower for low versus high risk patients (1.9 vs 2.2; p <.001). In the 18 months post index, the proportion of patients with moderately severe to severe MDD remained higher among high risk credit score patients versus low risk credit score patients (25% vs 20%; p < 0.001).
Conclusions: In a real-world cohort of MDD patients, higher risk credit scores were associated with increased disease burden as measured by observed and estimated PHQ-9 scores. Improving access to mental health care may increase opportunities for more effective treatment of depressive symptoms to improve outcomes and reduce disparities in MDD.
