Objective: To investigate the course of self-reported disability in rheumatoid arthritis (RA) using the major self-report measure of functional impairment, the Health Assessment Questionnaire (HAQ).
Methods: The course of HAQ disability was assessed in 32,525 observations (1,843 patients) in which the HAQ was administered. In addition, a subset of 2,189 visits from 50 patients, followed on average for 17 years, was studied to model the effect of disease duration on the course of HAQ disability in individual patients. Linear and fractional polynomial regression as well as smoothing algorithms were applied to the group of RA patients and then to the individual course of patients.
Results: Group linear and nonlinear models of the effect of disease duration on HAQ disability were found to have 3 characteristics: 1) HAQ disability scores are high at disease onset rather than gradually increasing; 2) HAQ disability increases very slowly over time (0.03 units per year); and 3) all such models fit very poorly, explaining only 5% of the variance in HAQ disability scores. However, application of nonlinear models to individual patient courses (as opposed to groups of patients) explains 37% of the HAQ disability score variation. In some patients, the course of HAQ disability was either 1) chaotic (scores change without any pattern) or 2) determinable, but unrelated to time. When covariates were added to the group model, however, the course of HAQ disability became clearer, and 51% of the variance in the disability score could be explained, statistically, by pain, depression, erythrocyte sedimentation rate, and disease duration.
Conclusion: Individual RA patients have differing, characterizable courses: 1) nonlinear, 2) chaotic, or 3) non-time determined. The model that self-reported physical disability, as measured by the HAQ, occurs as a function of disease acting over time does not fit the data well and is an inadequate model. This discrepancy may also be the result of the patient's upward reappraisal of functional ability with increasing time. The predominant determinants of HAQ disability in RA are disease activity, pain, and psychosocial factors rather than structural abnormality. Although the HAQ is a useful clinical tool and a central measure of disease outcome, it measures both process and outcome, and usually more process than outcome. Individual patient models that include all that is known about the patient--the paradigm of clinical care-may be a more effective way to examine the course of RA than are conventional group-based models.