Publicatie

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

Gardarsdottir, H., Egberts, T.C.G., Dijk, L. van, Sturkenboom, M., Heerdink, E.R. An algorithm to identify antidepressant users with a diagnosis of depression from prescription data. Pharmacoepidemiology and Drug Safety: 2008, 17(suppl. 1), p. S1. Abstract. 24th International Conference on Pharmacoepidemiology & Therapeutic Risk Management, Kopenhagen, 18 augustus 2008.
Background: Investigating depression treatment outcomes in prescription databases is problematic when information on indication for antidepressant prescriptions is unavailable. Objectives: To develop and validate an algorithm using prescription data to identify antidepressant drug users who suffer from depression. Methods: Data for deriving the algorithm were obtained from the 2nd Dutch National Survey of General Practice (NIVEL). The Integrated Primary Care Information (IPCI) database was used for validation. Both sets included adults receiving their first antidepressant in 2001 (N=1,855 respectively N=3,321). The outcome was defined as depression diagnosed by a general practitioner (GP). Covariates investigated were patient and prescribing characteristics, and comedication. The association between the outcome and the covariates was quantified using univariate logistic regression. The multivariate logistic model included covariates with an association p-value<0.2. The resulting algorithm was reduced by excluding predictors with p-value>0.1. Results: The derivation set was 67% female with a mean age of 50 years. The algorithm included age, an SSRI prescribed on index date, prescribed dose, GP 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 six months prior to index date. The probability that the algorithm correctly identified an antidepressant user as having a depression diagnosis was 79% with a sensitivity of 80% and a specificity of 67%. Conclusions: 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. (aut. ref.)