Reduction in long-term disability in patients with rheumatoid arthritis by disease-modifying antirheumatic drug-based treatment strategies.

OBJECTIVE: Therapeutic strategies for rheumatoid arthritis (RA) have been evolving from the traditional "pyramid" approach toward one based upon early and sustained use of disease-modifying antirheumatic drugs (DMARDs), in the hope of improving long-term health outcomes. However, few data to have been presented to document the effects of this approach. We sought to directly assess associations between consistent DMARD use and long-term functional outcomes.

METHODS: We studied 2,888 RA patients who were followed up prospectively for up to 20 years (average 9 years) at 8 databank centers. The independent variable was the proportion of patient encounters that resulted in treatment with > or = 1 DMARD (hydroxychloroquine, sulfasalazine, auranofin, intramuscular gold, D-penicillamine, methotrexate, and/or azathioprine). The dependent variable was each patient's last recorded Disability Index value from the Health Assessment Questionnaire (HAQ).

RESULTS: Increased DMARD use was strongly associated with better long-term Disability Index values (P < 0.0001). The association was strengthened when restricted to more seriously affected (rheumatoid factor (RF)-positive) patients. The magnitude of the effect, unadjusted, was a difference of 0.53 HAQ Disability units (scale 0-3) between 100% DMARD use and 0%. Correlation coefficients ranged up to 0.26. Effects were similar for all disease duration periods (0-4, 5-9, 10-14, 15-19, and 20+ years). "Control" correlations, with variables computed to represent the proportion of time in which patients were taking either nonsteroidal anti-inflammatory drugs or prednisone, failed to show positive associations. A multiple linear regression model, which controlled for age, disease duration, sex, RF positivity, proportion of visits under a prednisone regimen, and initial disability level, included the proportion of time in which patients were taking DMARDs (P < 0.0001), with a model R2 of 0.54. These results were obtained despite an adverse selection bias in which more severely affected individuals were given DMARDS more frequently, and despite absence of data on drug use early in the disease course of many patients. Thus, these results, which suggest up to a 30 percent reduction in long term disability with consistent DMARD use, are most likely conservative.

CONCLUSION: An association between consistent DMARD use and improvement in long-term functional outcomes in RA is supported by these data.