OBJECTIVES: Many studies have identified predictors of the health assement questionnaire (HAQ), but few have evaluated the predictors of HAQ trends among RA patients. To investigate the predictors of worsening disability trends compared to other patterns among RA patients over at least 2 years. METHODS: A total of 646 RA patients from the ongoing biannual NDB-Portugal cohort with at least four consecutive HAQ scores per patient during their follow-up were used.The proportion defined by the number of 6-month positive increments in HAQ scores (worsening function) divided by the total number of differences was computed per patient and used to define a patient' trend.  The outcome was then defined as the presence of a trend of worsening disability (when proportion>0.5). This meant that a patient´s tendency of worsening was higher than their tendency of improving during their own follow-up. Univariate (UV) and multivariate (MV) generalized estimating equations (GEE) were used to study the predictors of a worsening disability trend. Age, education, disease duration, paid work, retirement, number of total major comorbidities, SF-36 mental component, RADAI, the VAS scales of sleep, fatigue and pain, the use of current TNF (with or without concomitant DMARDs) vs. traditional DMARD therapy and steroids, were used as possible predictors. RESULTS: A total of 26% patients had worsening disability trends. The UV analyses showed that all of the following factors were statistically relevant: age, educational level, number of major comorbidities, sleep disturbances and fatigue, RADAI and the use of TNF therapy. The final MV model included pain (OR: 1.003 (95%CI: (1.000; 1.005))), age (OR: 1.02 (1.01; 1.02)) and the use of TNF (OR: 0.94 (0.91; 0.97)). CONCLUSIONS: In our study, we showed that older age and more pain predicted worsening HAQ disability trends. The use of TNF therapy was the only factor that decreased the odds of having a worsening HAQ trajectory.