RESUMEN
OBJECTIVE: This study aims to explore the contribution of implicit attitudes and associations towards conventional disease-modifying antirheumatic drugs (cDMARDs), alongside explicit measures, on medication-taking behaviour and clinical outcomes in adult patients with rheumatoid arthritis (RA). METHODS: In this observational study, implicit attitudes (positive-negative) and health-related associations (health-sickness) were measured with Single Category Implicit Association Tests, whereas explicit outcomes were measured with a bipolar evaluative adjective scale and the Beliefs about Medicines Questionnaire Specific. The primary outcome of this study was medication-taking behaviour subjectively measured by self-report (i.e. validated Compliance Questionnaire on Rheumatology) and objectively measured with electronic drug monitors over a 3 month period. Spearman rank correlations were used to describe correlations between implicit and explicit outcomes. Nested linear regression models were used to assess the additional value of implicit measures over explicit measures and patient-, clinical-, and treatment-related characteristics. RESULTS: Of the 1659 initially-invited patients, 254 patients with RA agreed to participate in this study. Implicit attitudes correlated significantly with necessity-concerns differential (NCD) scores (ρ = 0.13, P = 0.05) and disease activity scores (ρ = -0.17, P = 0.04), whereas implicit health-related associations correlated significantly with mean scores for explicitly reported health-related associations (ρ = 0.18, P = 0.004). Significant differences in age, number of DMARDs, biologic DMARD use, NCD-scores, and self-reported correct dosing were found between the four attitudinal profiles. Nested linear regression models revealed no additional value of implicit measures in explaining self-reported medication-taking behaviour and clinical outcomes, over and above all other variables. CONCLUSION: Implicit attitudes and associations had no additional value in explaining medication-taking behaviour and clinical outcomes over and above often used explicitly measured characteristics, attitudes and outcomes in the studied population. Only age and NCD scores contributed significantly when the dependent variable was correct dosing measured with self-report.