High concordance between self-reported medication and official prescription database information

Objective We set out to compare the agreement between self-reported psychotropic medication use and information obtained from the administrative prescription database of the Social Insurance Institution of Finland. We compared the point prevalence of psychotropic medication use, and self-reported vs. register-based information on antipsychotic medication dosage. Methods The study population consisted of 905 participants from a population-based genetic study of schizophrenia, of whom 366 had schizophrenia spectrum disorder, 56 had bipolar spectrum disorder, and 483 were unaffected family members. Current medication use was obtained by interview and from the prescription reimbursement database. Agreement between data sources was compared using Cohen's kappa statistic and correlation coefficients. Results The agreement between the two sources was generally good. Kappa values were best for lithium use (0.96; p < 0.0001), followed by antipsychotics (0.87; p<0.0001) and antidepressants (0.77; p<0.0001). Agreement was lowest for benzodiazepines (0.42; p<0.0001). Correlation between antipsychotic medication dose estimates was 0.79 (95% CI 0.76-0.81). Conclusions The concordance between self-reported psychotropic medication use and information obtained from an official prescription database was good for most psychotropic drugs. More studies are needed to replicate results with other forms of medication and patient groups.