Use of Selective Serotonin Reuptake Inhibitors - Validity of Self-Report versus Plasma Concentrations and Pharmacy Dispensations - A Cross-Sectional Analysis of the Norwegian Women and Cancer Study.

Purpose: To validate self-reported current use of selective serotonin reuptake inhibitors (SSRI) in the Norwegian Women and Cancer study (NOWAC) and to identify factors associated with discordance between data sources. Material and Methods: This is a cross-sectional record-linkage study comparing SSRI-use derived from four data sources: 1) Specific SSRI questions in the main NOWAC questionnaire, 2) Open questions on medication use in the small questionnaire following blood samples, 3) plasma concentration measurements for a subsample, and 4) pharmacy dispensations from the Norwegian prescription database (NorPD) where current use of SSRI was defined by Legend Time Duration (LTD). Among 105 855 women, aged 46 to 64 years and randomly selected from the general population, 70,191 had data on SSRI-use from both NOWAC and NorPD. Plasma concentration was measured for 93 pairs of self-reported SSRI-users and non-users, with dispensation data available for 68 pairs. Validity was assessed by sensitivity and specificity; agreement was assessed by Cohen's kappa. Factors associated with discordance between information sources were analyzed by multiple binary logistic regression. Results: We found high sensitivity (89.5%) and specificity (98.7%) for the specific questions in the main questionnaire compared with pharmacy dispensations. Measured against plasma concentrations, current SSRI-use defined by open questions and pharmacy dispensations both had high sensitivity (100% and 92.5%, respectively) and specificity (98.6% both). Agreements (kappa) were similarly high for all comparisons (≥0.80). The factors associated with discordance between data sources included poor health, comorbidity, being single and not being in full time work. Education was inversely associated with discordance. Conclusion: Self-reported current use of SSRI from the NOWAC questionnaires is highly valid and, according to plasma concentrations, perhaps even more so than pharmacy dispensations. Factors associated with discordance between information sources should be taken into account in the interpretation of future analyses which include SSRI-use in the NOWAC study.