Accuracy of Visual Inspection Alone to Assess Joint Effusions of the Hand: Cross-Sectional Study

Physical examination is the cornerstone of diagnosing and monitoring inflammatory arthritis, with the detection of joint effusions being one of the most crucial components of the examination. Rheumatologists largely rely on palpation, supported by other examination techniques, such as evaluating the range of motion and visual inspection, to assess signs of joint swelling. However, remote care is on the rise in rheumatology, and aside from visual inspection, physical examination of joints is limited during telehealth visits. The ability of rheumatologists to accurately detect hand synovitis through photos or videos of the hands, without the benefit of direct tactile examination, is currently unknown.

This study aimed to assess the accuracy of detecting joint effusions of the hands by remote visual evaluation alone.

We conducted a prospective cohort study of patients assessed by a rheumatologist in Edmonton, Alberta. Participants were assessed clinically by rheumatologists for the presence of effusions of the metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints between June 13, 2024, and October 29, 2024. Participants had photos and videos of their hands taken by study staff on the same day. Photos and videos were remotely assessed separately by 4 rheumatologists not involved in the initial clinical assessments for the presence of MCP and PIP joint effusions and compared with the clinical assessments.

The study cohort included 156 patients (mean age 53, SD 14.2 years, n=105, 67.3% were female and n=52, 33.3% had rheumatoid arthritis), for a total of 3120 MCP and PIP joints. Effusions were identified in 12.8% (20/156) of patients and in 2.2% (69/3120) of joints per clinical assessment. The average visual assessment joint-level sensitivity and specificity of photos were 0.14 (95% CI 0.07-0.19) and 0.97 (95% CI 0.96-0.98), respectively. The average visual assessment joint-level sensitivity and specificity of videos were 0.24 (95% CI 0.13-0.33) and 0.98 (95% CI 0.97-0.99), respectively. The average person-level visual assessment sensitivity and specificity of photos were 0.44 (95% CI 0.30-0.60) and 0.82 (95% CI 0.76-0.87), respectively. The average person-level visual assessment sensitivity and specificity of videos were 0.48 (95% CI 0.35-0.60) and 0.84 (95% CI 0.79-0.89), respectively. Assessor agreement was poor (κ=0.12-0.17).

Visual inspection of photos and videos to detect MCP and PIP joint effusions was poor at both the joint and person levels. Patients and rheumatologists should be aware of these limitations when conducting remote telehealth assessments.