To the Editor:

Joint damage is the most prominent feature of disease outcome in rheumatoid arthritis (RA) and it is generally assessed by plain radiography of the hands. Several radiologic damage scores have been developed, each having specific characteristics for reproducibility and sensitivity to change (1). It has been shown that radiographic damage in large joints is significantly related to damage in the hands and feet, the physical disability index, and cumulative disease activity (2). The cost of measuring radiographic damage makes these current methods less suitable for use in developing countries. Cranney et al (3) reported on the use of the deformity index, Orces et al (4) reported the use of the number of deformed joints (NDJ) index, and Johnson et al (5) reported the mechanical joint score as a measure of damage in RA. Zijlstra and colleagues (6) developed and validated the Rheumatoid Arthritis Articular Damage Score (RAAD score). We investigated the RAAD score in Tunisian patients with RA.

Data on disease duration, disease activity score (DAS28) (7), Larsen score, Rheumatoid Arthritis Quality of Life, Arthritis Impact Measurement Scale-2, Health Assessment Questionnaire (HAQ), and RAAD score were collected for 122 patients with RA. In the RAAD score method, 35 joints or joint groups were scored on a 3-point scale (where 0 = no irreversible damage; 1 = partial damage; and 2 = severe damage, ankylosis, or prosthesis). In addition to the original score (RAAD-1), we also studied the 2 alternative RAAD scores (RAAD-2 and RAAD-3) (6). Intraobserver agreement of the RAAD scores, assessed by the same author (IB), on a 1-week test–retest was studied using the intraclass correlation coefficient (ICC) among 61 patients. Kappa statistics were used to assess the intraobserver agreement of the RAAD score for individual joints. Spearman's rank correlation was used to assess the correlation between RAAD scores (first measure), DAS28, disease duration, and Larsen score. The mean disease duration was 14 years (range 0.1–29 years). The mean ± SD HAQ and Larsen scores were 0.81 ± 0.68 and 34.54 ± 29.41, respectively. The RAAD scores were easily applicable and took ∼2 minutes for each patient to complete. The RAAD-1, RAAD-2, and RAAD-3 scores were 11.75, 10.16, and 7.38, respectively. Good intraobserver agreement was found for total scores (n = 61; ICC = 0.94) and individual joints (weighted κ 0.50–0.89). The box plot graph analysis of the RAAD-1 score and the disease duration showed a wide range of RAAD-1 scores for patients with the same disease duration (Figure 1). The correlations of the RAAD-1 score (first time) with the Larsen score and the HAQ were good (rs = 0.76, P = 0.000 and rs = 0.59, P = 0.001, respectively). The RAAD-1 score also correlated with disease duration (rs = 0.64, P = 0.001). Different ways of computing the RAAD score (RAAD-2 and RAAD-3) showed similar degrees of correlation. However, we did not find any correlation between RAAD-1 score and DAS28.

thumbnail image

Figure 1. Box plot showing results of the Rheumatoid Arthritis Articular Damage Score (RAAD) versus disease duration for 122 patients.

Download figure to PowerPoint

The RAAD score is a feasible way to study long-term damage in large groups of patients and compare the outcome with long-term (>5 years) treatment strategies (6). This method is inexpensive and is greatly advantageous in assessing RA damage in developing countries such as Tunisia. Intraobserver variability in our sample was very low. Although Zijlstra et al (6) studied the interobserver variability in 47 patients and found it to be very low, they did not study intraobserver variability. Therefore, our results complete the assessment of the quality of the RAAD score. As a substitute, the Larsen score correlates well with the RAAD score. The use of the RAAD-2 or the RAAD-3 scores did not change the properties of the score significantly. Intraobserver reliability of the RAAD-1 was excellent, similar to that reported for the NDJ (n = 28; ICC 0.94) (4), which also has a good external validity. The mechanical joint score (5) also demonstrated good metrologic properties, and we are planning to compare the RAAD score with these 2 latter scores to choose the method that best meets the needs of Tunisian patients with RA.

The RAAD score is easy to perform and demonstrated good intraobserver reliability. It correlates well with the Larsen score and disease duration and can be used to determine the severity of joint damage within a population.


  1. Top of page

Ismail Bejia MD*, Kamel Ben Salem MD*, Mongi Touzi MD*, Naceur Bergaoui MD*, * EPS Monastir Monastir, Tunisia