Derivation and validation of a multivariable model to predict when primary care physicians prescribe antidepressants for indications other than depression
Authors Wong J, Abrahamowicz M, Buckeridge DL, Tamblyn R
Received 1 October 2017
Accepted for publication 4 December 2017
Published 18 April 2018 Volume 2018:10 Pages 457—474
Checked for plagiarism Yes
Review by Single-blind
Peer reviewer comments 2
Editor who approved publication: Professor Henrik Toft Sørensen
Jenna Wong, Michal Abrahamowicz, David L Buckeridge, Robyn Tamblyn
Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, QC, Canada
Objective: Physicians commonly prescribe antidepressants for indications other than depression that are not evidence-based and need further evaluation. However, lack of routinely documented treatment indications for medications in administrative and medical databases creates a major barrier to evaluating antidepressant use for indications besides depression. Thus, the aim of this study was to derive a model to predict when primary care physicians prescribe antidepressants for indications other than depression and to identify important determinants of this prescribing practice.
Methods: Prediction study using antidepressant prescriptions from January 2003–December 2012 in an indication-based electronic prescribing system in Quebec, Canada. Patients were linked to demographic files, medical billings data, and hospital discharge summary data to create over 370 candidate predictors. The final prediction model was derived on a random 75% sample of the data using 3-fold cross-validation integrated within a score-based forward stepwise selection procedure. The performance of the final model was assessed in the remaining 25% of the data.
Results: Among 73,576 antidepressant prescriptions, 32,405 (44.0%) were written for indications other than depression. Among 40 predictors in the final model, the most important covariates included the molecule name, the patient’s education level, the physician’s workload, the prescribed dose, and diagnostic codes for plausible indications recorded in the past year. The final model had good discrimination (concordance (c) statistic 0.815; 95% CI, 0.787–0.847) and good calibration (ratio of observed to expected events 0.986; 95% CI, 0.842–1.136).
Conclusion: In the absence of documented treatment indications, researchers may be able to use health services data to accurately predict when primary care physicians prescribe antidepressants for indications other than depression. Our prediction model represents a valuable tool for enabling researchers to differentiate between antidepressant use for depression versus other indications, thus addressing a major barrier to performing pharmacovigilance research on antidepressants.
Keywords: antidepressant; indications; predictive studies; predictors; primary care; pharmacovigilance
This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution - Non Commercial (unported, v3.0) License. By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms.Download Article [PDF] View Full Text [HTML][Machine readable]