BACKGROUND: Researchers need valid methods to assess whether patients are taking their antidepressant medications. Two important sources of data on drug exposure are patients' self-reports and pharmacy claims. OBJECTIVE: To compare self-report and claims data for antidepressant exposure. RESEARCH DESIGN: Cross-sectional analysis. SUBJECTS: This study comprised 422 contemporaneous self-report and claims data points obtained from 164 unique patients in a longitudinal depression study in which patients completed up to five surveys during an 18-month period. MEASURES: For the self-report measure, the following question was asked: Do you now take any prescription medicines for depression? Using claims data, patients were considered to be using an antidepressant if they had filled at least one antidepressant prescription in the 90 days before survey dates. RESULTS: Self-report and claims agreed in 85% (358/422) of cases, with a kappa of 0.69. Eighty-eight percent (56/64) of discrepant cases using other study data sources was resolved. Reasons for discrepancies included the use of medications for conditions other than depression (32/64), recent AD discontinuations (6/64), samples usage (3/64), and low-frequency/PRN use (7/64). CONCLUSIONS: Self-report and claims showed good concordance, but they reflect different truths. Self-report identifies medications intended primarily for the treatment of depressive disorders, whereas claims data identify use of medicines with antidepressant effects. Our assessment of discordant cases showed self-report to be more valid than claims to assess current antidepressant use for depression therapy.
BACKGROUND: Researchers need valid methods to assess whether patients are taking their antidepressant medications. Two important sources of data on drug exposure are patients' self-reports and pharmacy claims. OBJECTIVE: To compare self-report and claims data for antidepressant exposure. RESEARCH DESIGN: Cross-sectional analysis. SUBJECTS: This study comprised 422 contemporaneous self-report and claims data points obtained from 164 unique patients in a longitudinal depression study in which patients completed up to five surveys during an 18-month period. MEASURES: For the self-report measure, the following question was asked: Do you now take any prescription medicines for depression? Using claims data, patients were considered to be using an antidepressant if they had filled at least one antidepressant prescription in the 90 days before survey dates. RESULTS: Self-report and claims agreed in 85% (358/422) of cases, with a kappa of 0.69. Eighty-eight percent (56/64) of discrepant cases using other study data sources was resolved. Reasons for discrepancies included the use of medications for conditions other than depression (32/64), recent AD discontinuations (6/64), samples usage (3/64), and low-frequency/PRN use (7/64). CONCLUSIONS: Self-report and claims showed good concordance, but they reflect different truths. Self-report identifies medications intended primarily for the treatment of depressive disorders, whereas claims data identify use of medicines with antidepressant effects. Our assessment of discordant cases showed self-report to be more valid than claims to assess current antidepressant use for depression therapy.
Authors: Sebastian Schneeweiss; Amanda R Patrick; Daniel H Solomon; Jyotsna Mehta; Colin Dormuth; Matthew Miller; Jennifer C Lee; Philip S Wang Journal: Arch Gen Psychiatry Date: 2010-05
Authors: Barbara Guerard; Vincent Omachonu; Raymond A Harvey; S Robert Hernandez; Bisakha Sen Journal: Health Serv Res Date: 2015-09-15 Impact factor: 3.402
Authors: Danielle M Crookes; Ryan T Demmer; Katherine M Keyes; Karestan C Koenen; Shakira F Suglia Journal: Epidemiology Date: 2018-07 Impact factor: 4.822
Authors: Marianne Haapea; Jouko Miettunen; Sari Lindeman; Matti Joukamaa; Hannu Koponen Journal: Int J Methods Psychiatr Res Date: 2010-06 Impact factor: 4.035
Authors: Cynthia A Beck; Scott B Patten; Jeanne V A Williams; Jian Li Wang; Shawn R Currie; Colleen J Maxwell; Nady El-Guebaly Journal: Soc Psychiatry Psychiatr Epidemiol Date: 2005-10 Impact factor: 4.328