In rheumatoid arthritis (RA), nurses are now increasingly involved in joint count assessment but training is not standardized. The aim was to evaluate and describe the learning curve of nurses for the assessment of swollen and tender joints in RA.

Twenty nurses from university rheumatology centres inexperienced with joint counts were allocated to a rheumatologist from their centre (teacher). Acquisition of skills consisted of Phase 1: (training), a centralized 4hour training session, with (a) lecture and demonstration, and (b) practical sessions on patients with their teachers, followed by Phase 2: (practice) involving further practice on 20 patients in their own hospitals. Primary outcome was achievement of adequate swollen joint agreement between nurse and their teacher (“gold standard”) at the “joint” level defined by prevalence adjusted biased adjusted kappa (PABAK)>0.60. Agreement at the “patient” level of swollen joint count (SJC), tender joint count (TJC) as well as DAS28 between nurse and their teacher were assessed with intra-class correlation coefficients (ICC).

During the training phase, 75% of nurses achieved a swollen joint PABAK>0.60 when compared with their teachers, which further improved to 89% after the 20 practice patients (Phase 2). Median swollen joint PABAK improved from 0.64 (Q1:Q3 0.55,0.86) to 0.83 (Q1:Q3 0.77,1) by the end of Phase 2. At the “patient” level, SJC agreement remained globally stable (ICC, 0.52 to 0.66), while TJC and DAS28 agreement remained excellent throughout.

Nurses inexperienced in joint counts were able to achieve excellent agreement with their teachers in assessment of tender and swollen joints through a short training session; practice further enhanced this agreement. Larger longitudinal studies are required to assess skills retention.