Measuring psychotropic drug exposures in register-based studies--validity of a dosage assumption of one unit per day in older Finns.

Pharmacoepidemiological studies provide valuable information on the relationships between psychotropic drug use and adverse outcomes in older people. To minimize the influence of misclassification bias in pharmacoepidemiological studies, more emphasis should be given to methodological aspects of exposure assessment. This study evaluated the validity of a dosage assumption of one unit per day for measuring legend duration of psychotropic drug exposures among older people. Using data from the Finnish Prescription Register, the study analysed 62,320 psychotropic drug prescriptions dispensed for people aged ≥ 75 years (n = 52,729) in September 2009. The proportions of prescriptions in which the prescribed dose deviated from one unit per day were assessed for categories and subcategories of psychotropic drugs. The prescription was considered misclassified (a) if the prescribed drug was intended for "as needed" use, (b) if the prescription included a dose range, or (c) if the prescribed dose was below or above one unit per day. Among antidepressants, less than every fourth (23.7%) prescription was misclassified. The proportions of misclassification varied substantially across subcategories, being 13.1% for selective serotonin reuptake inhibitors (SSRIs), 25.3% for other antidepressants and 53.8% for tricyclic antidepressants. Of the benzodiazepine and antipsychotic prescriptions, 79.9% and 57.6%, respectively, were misclassified. In conclusion, the dosage assumption of one unit per day is valid for measuring the legend duration of SSRI and other antidepressant exposures among older people. Among other psychotropic drugs, the dosage assumption is likely to lead to severe exposure misclassification.