Objective: To determine the correlation between patient self-report joint counts and standard physician joint counts, and to compare pictorial (Mannequin) and text (Rapid Assessment of Disease Activity in Rheumatology, RADAR) formats for obtaining patient self-reports.
Methods: Baseline patient self-report joint counts were mailed and completed by 60 patients with early rheumatoid arthritis (RA) one day before and one day after being examined by a physician. Twenty-seven were randomized to the Mannequin tender and Mannequin swollen joint counts; 33 were randomized to the RADAR tender and swollen joint counts. Agreement between patient and physician self-report joint counts, diagnostic characteristics, and test-retest reliability of patient self-report joint counts was computed. Stepwise regression analyses were performed to identify predictors of patient-physician differences in total joint count.
Results: Means and standard deviations of paired patient and physician total joint counts were not different for Mannequin or RADAR forms. Spearman correlations were moderate (0.58 to 0.69 for Mannequin, 0.37 to 0.58 for RADAR). Agreement (intraclass correlations) was 0.65 for the Mannequin and 0.56 for the RADAR forms. Patient test-retest reproducibility was moderate for RADAR tenderness (0.58) and high (r>0.90) for RADAR swollen and both Mannequin forms. Level of patient education predicted patient-physician differences on the RADAR swollen joint counts (p = 0.003), but was not significant in Mannequin forms, suggesting that education was not a factor in accurate completion of Mannequin forms.
Conclusion: Both pictorial and text format patient self-report joint counts are significantly correlated with physician joint counts. In addition to moderately high patient test-retest reproducibility, this suggests that patient self-reports in both formats may yield accurate measures of improvement in disease activity.