Important Risk Factors to Proxies of Treatment Failures in Patients With Major Depressive Disorder: Insights From a Claims Database Study
Author(s)
Ling Zhang, MPH, MS, MD1, Carissa S. White Dukes, MPH1, Suzanne St Rose, PhD2, Franco De Crescenzo, PhD2, Anne Kilburg, MSc2, Sigurd Suessmuth, PhD2, Rashmi Patel, PhD, MD3.
1Boehringer Ingelheim, Ridgefield, CT, USA, 2Boehringer Ingelheim, Ingelheim am Rhein, Germany, 3University of Cambridge, Cambridge, United Kingdom.
1Boehringer Ingelheim, Ridgefield, CT, USA, 2Boehringer Ingelheim, Ingelheim am Rhein, Germany, 3University of Cambridge, Cambridge, United Kingdom.
OBJECTIVES: This study aimed to establish a spectrum of proxy treatment failure definitions (from specific to general criteria) in major depressive disorder (MDD) and identify consistent risk factors across definitions.
METHODS: Using the Optum Clinformatics claims database, adults aged 18-65 with ≥2 MDD encounters from January 1, 2012, to March 31, 2022, were included. The index date was the first MDD diagnosis, excluding prior specific mental disorders. MDD episodes (MEPs) were constructed by linking MDD diagnoses or antidepressant use, allowing for a 120-day gap. The baseline was the 4-month period before the index or episode started. Baseline disease characteristics variables were assessed.Lines of treatment (LOTs) were formed by linking antidepressants with a 30-day gap allowance. Treatment failure was defined in three ways: - Fail1: Strict failure: add-on/switching therapy from 1 to~8 months of the current LOT. - Fail2: Broad failure: add-on/switch at any time during LOT. - Fail3: Change of generic names of antidepressants. Risk factors to treatment failure were assessed using LASSO regression.
RESULTS: Among 304,802 MEPs in 203,313 MDD patients, by Fail1, 22.2% episodes having >=1 failure event; 30.8% by Fail2 and 40.6% by Fail3. Episodes with a higher number of failure events (NOF) exhibited increased baseline mental and somatic comorbidities, and greater medication usage. Strong correlations were observed between LOT count and NOF across failure definitions (ρ=0.65-0.70).Baseline use of specific atypical antipsychotics: quetiapine and mood stabilizers: gabapentin was among key predictors of treatment failure. Area under curve values were 0.698 for Fail1, 0.708 for Fail2, and 0.702 for Fail3.
CONCLUSIONS: Frequent treatment changes, whether guideline-based or non-specific medication switches, are associated higher disease severity. These patterns may serve as useful real-world proxies for poor treatment response in MDD. Additionally, certain medications used before diagnosis or during remission may indicate increased risk of treatment failure.
METHODS: Using the Optum Clinformatics claims database, adults aged 18-65 with ≥2 MDD encounters from January 1, 2012, to March 31, 2022, were included. The index date was the first MDD diagnosis, excluding prior specific mental disorders. MDD episodes (MEPs) were constructed by linking MDD diagnoses or antidepressant use, allowing for a 120-day gap. The baseline was the 4-month period before the index or episode started. Baseline disease characteristics variables were assessed.Lines of treatment (LOTs) were formed by linking antidepressants with a 30-day gap allowance. Treatment failure was defined in three ways: - Fail1: Strict failure: add-on/switching therapy from 1 to~8 months of the current LOT. - Fail2: Broad failure: add-on/switch at any time during LOT. - Fail3: Change of generic names of antidepressants. Risk factors to treatment failure were assessed using LASSO regression.
RESULTS: Among 304,802 MEPs in 203,313 MDD patients, by Fail1, 22.2% episodes having >=1 failure event; 30.8% by Fail2 and 40.6% by Fail3. Episodes with a higher number of failure events (NOF) exhibited increased baseline mental and somatic comorbidities, and greater medication usage. Strong correlations were observed between LOT count and NOF across failure definitions (ρ=0.65-0.70).Baseline use of specific atypical antipsychotics: quetiapine and mood stabilizers: gabapentin was among key predictors of treatment failure. Area under curve values were 0.698 for Fail1, 0.708 for Fail2, and 0.702 for Fail3.
CONCLUSIONS: Frequent treatment changes, whether guideline-based or non-specific medication switches, are associated higher disease severity. These patterns may serve as useful real-world proxies for poor treatment response in MDD. Additionally, certain medications used before diagnosis or during remission may indicate increased risk of treatment failure.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EPH138
Topic
Epidemiology & Public Health, Methodological & Statistical Research, Real World Data & Information Systems
Disease
Mental Health (including addition), No Additional Disease & Conditions/Specialized Treatment Areas