With the increasing availability of biological therapies for patients with rheumatoid arthritis (RA), the need to include reliable and reproducible clinical measures of disease activity in routine practice has been emphasised.1 Disease Activity Score-28 (DAS28) is validated for use in clinical trials;2^–5 however, it is not known how it performs in a clinical setting.

The aim of this study was to evaluate consistency in the measurement of DAS28, and its individual components, by clinic staff in one centre with different levels of experience. Twelve patients with RA (six per exercise) were assessed for the inter-observer exercises. Five patients were assessed for the intra-observer exercise. Six members of the rheumatology team (two staff rheumatologists, two trainees in rheumatology, one advanced nurse practitioner and one rheumatology nurse specialist) participated in the study.

Before each DAS28 reproducibility exercise, the EULAR guidelines for the assessment of swollen and tender joints were reviewed. Patients scored a general health assessment (GHA) once only and the erythrocyte sedimentation rate (ESR) was taken within 48 hours. For intra-observer assessment, one trainee (CAEW) assessed five patients at the same time on three consecutive days. Patients scored a GHA daily. ESR was taken at one time-point only. Single measure (sICC) and average measure (aICC) intraclass correlation coefficients were used to calculate inter-observer reliability. Pearson correlations were used to calculate the intra-observer agreement between joint scores and DAS28 over 3 days.

Swollen joint counts (SJCs) obtained by the assessors in exercise 1 produced a sICC of 0.29 (95% confidence interval (CI), 0.02 to 0.08) and aICC 0.71 (95% CI, 0.10 to 0.95) improving to 0.46 (95% CI, 0.14 to 0.86) and 0.84 (95% CI, 0.50 to 0.97) in exercise 2. Inter-individual correlations also improved (table 1a). Tender joint counts (TJC) were more consistent with sICC of 0.75 (95% CI, 0.47 to 0.96) and aICC 0.95 (95% CI, 0.84 to 0.99) in exercise 1, improving to 0.86 (95% CI, 0.64 to 0.98) and 0.97 (95% CI, 0.92 to 1.00) in exercise 2.

The composite DAS28 values demonstrated excellent correlations between observers in both exercises with sICC of 0.92 (aICC 0.99) in both. Good correlations were seen between all individual observers (table 1b). Intra-observer correlations were performed over 3 successive days. Excellent correlations were observed with r values of 0.99 (TJCs) and 0.95 (SJCs) on days 2 and 3, compared with day 1. The intra-observer correlations for DAS28 were 0.92 and 0.89 on days 2 and 3. The reduced r-value for the DAS28, was mostly due to inconsistencies in the patients’ reported GHA. Daily fluctuations in GHA values were seen in all patients (range 7–32 mm compared with the previous day).

The principal observation in this study was that high ICC values for DAS28 scores were obtained from clinical evaluations performed by six rheumatology practitioners with a range of clinical experience. The low ICC values for the observed SJCs and TJCs are reflected in the statistically determined weighting given to these measures, relative to the more objective ESR, in the DAS28 formula. Moreover, the patient-derived GHA, which is susceptible to daily fluctuations, is allocated the least weighting when deriving DAS28.

Several countries require a minimum baseline level of disease activity for reimbursement when prescribing biological therapies. Thus, important therapeutic decisions in RA demand robust and reproducible instruments for measuring disease activity in the “real-world” setting where different members of the clinical team assess patients at successive visits. This study confirms that DAS28 is a reliable instrument for measuring disease activity and therapeutic response in patients with RA, not only in the structured environment of randomised controlled trials but also in routine clinical practice.