An algorithm to identify antidepressant users with a diagnosis of depression from prescription data

Authors

  • Helga Gardarsdottir PharmD,

    1. Division of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands
    Search for more papers by this author
  • Antoine C. G. Egberts PhD,

    1. Division of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands
    2. Department of Clinical Pharmacy, University Medical Centre Utrecht, Utrecht, The Netherlands
    Search for more papers by this author
  • Liset van Dijk PhD,

    1. Netherlands Institute for Health Services Research (NIVEL), Utrecht, The Netherlands
    Search for more papers by this author
  • Miriam C. J. M. Sturkenboom PhD,

    1. Department of Medical Informatics, Erasmus University Medical Centre, Rotterdam, The Netherlands
    Search for more papers by this author
  • Eibert R. Heerdink PhD

    Corresponding author
    1. Division of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, Utrecht, The Netherlands
    • Division of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, PO Box 80082, 3508 TB Utrecht, The Netherlands.
    Search for more papers by this author

Abstract

Purpose

Antidepressants are used for many indications besides depression. This makes investigating depression treatment outcomes in prescription databases problematic when the indication is unknown. The aim of our study is to develop an algorithm to identify antidepressant drug users from prescription data that suffer from depression.

Methods

Data for deriving the algorithm were obtained from the Second Dutch National Survey of General Practice, carried out in 2001 by The Netherlands Institute for Health Services Research (NIVEL), and for validation the Integrated Primary Care Information (IPCI) database was used. Both sets included adults receiving their first antidepressant drug in 2001 (n = 1855 and 3321, respectively). The outcome was a registered diagnosis of depression. Covariates investigated for developing the algorithm were patient and prescribing characteristics, and co-medication.

Results

The predictive algorithm included age, SSRI prescribed on the index date, prescribed dose, general practitioner as prescriber and the number of antidepressant prescriptions prescribed plus medication for treating acid related disorders, laxatives, cardiac therapy or hypnotics/sedatives prescribed in the 6 months prior to index date. The probability that the algorithm correctly identified an antidepressant drug user as having a depression diagnosis was 79% with a sensitivity of 79.6% and a specificity of 66.9%.

Conclusion

In conclusion, we developed and validated an algorithm that can be used to compose cohorts of patients treated with antidepressants for depression from prescription databases. Copyright © 2008 John Wiley & Sons, Ltd.

Ancillary