Using Medical Records to Supplement a Claims-Based Comparative Effectiveness Analysis of Antidepressant Treatment Thomas W. Croghan, Dominick Esposito, Gregory Daniel, Peter Wahl, Michael A. Stoto, and M. Haim Erder June 10, 2008 AcademyHealth 25 th Annual Research Meeting Washington, DC
Research Objectives Compare a branded SSRI (escitalopram) with generically available SSRIs (citalopram, fluoxetine, paroxetine) Phase II: Medical chart review to supplement the claims data analysis –30% of patients discontinued initial antidepressant at 2 months –Patients about 10% more likely to continue escitalopram at 2 months compared with other SSRIs –Total health care costs no different if patient initiates on escitalopram or generically available SSRIs Phase I: Claims data analysis to examine use patterns and costs
Chart Review Description Exploratory analysis of 457 charts for patients with known use patterns based on claims analysis –Four weeks prior to initiating treatment –7 to 21 days after starting treatment Variables –Baseline symptoms of depression and anxiety –Comorbid conditions –Diagnostic tools/education –Symptoms at follow-up –Side effects Escitalopram Continuer Escitalopram Discontinuer Other SSRIs Continuer Other SSRIs Discontinuer
Baseline Symptoms from Chart Review % Present % Absent % Not Recorded Depressed Mood Sleep Crying spells Fatigue/energy Agitation Anhedonia Concentration Weight change Any anxiety disorder Worthlessness Suicidal Total Charts = 457
Baseline Symptom Count Associated with Use Pattern Symptom CountTotalDiscontinueContinue* 4 or more 30.2%23.7%34.0% 2 or %37.3%38.9% Only %26.6%21.5% None8.1%12.4%5.6% Sample Size * p<0.0001, chi-square test of continuers versus discontinuers. Missing recorded as zero.
Difference in Symptom Count Associated with Use Pattern Symptom Count Difference 1 AllDiscontinueContinue* - 3 or more 25.4%20.1%27.8% - 1 or %43.2%44.4% No change 23.5%28.4%18.4% + 1 or more 8.5%8.3%8.7% Sample Size * p < , chi-square test of continuers versus discontinuers 1 Difference in symptom count between first and second visit after initial antidepressant fill.
Physician’s Assessment Associated with Use Pattern Physician Assessment 1 AllDiscontinueContinue* Remitted3.9%5.9%2.8% Improved51.6%38.5%59.4% No change/worse 20.1%24.3%17.8% Not Recorded 24.3%31.4%20.1% Sample Size *p < , chi-square test of continuers versus discontinuers 1 Physician assessment of patient at the second visit after antidepressant initiation.
Among Continuers, Escitalopram Users More Likely to be Judged Improved Continuers Physician Assessment 1 AllEscitalopramOther* Remitted Improved No change/worse Not Recorded Sample Size *p < 0.05, chi-square test of escitalopram versus other 1 Physician assessment of patient at the second visit after antidepressant initiation.
Primary Limitation: Missing Data More than 80% had no side-effects recorded Less than 20% recorded any patient education –Only 25 charts indicated education about antidepressants 71.6% of discontinuers, based on claims data, had no mention in chart (within the 7 to 21 day follow-up period)
Conclusions Most patients had symptoms of depression but few had evidence of major depression Continuation associated with –More symptoms at baseline –Improvement (number of symptoms or physician assessment) Claims data analysis might have underestimated clinical effectiveness –Few differences among drugs but those who continued on escitalopram were more likely to improve