JMIR Publications

ABSTRACT

Background:

Remote care is on the rise in rheumatology, but the accuracy of visual inspection alone for assessing joint effusions of the hands is unknown.

Objective:

To assess the accuracy of detecting joint effusions of the hands by visual evaluation alone.

Methods:

We conducted a prospective cohort study of rheumatology patients in Edmonton, Canada. Participants were assessed clinically by rheumatologists for the presence of effusions of the MCP and PIP joints between June 13, 2024 to October 29, 2024. Participants had photos and videos of their hands taken on the same day. Photos and videos were assessed separately by four rheumatologists not involved in the initial clinic assessments for the presence of MCP and PIP joint effusions, and compared against the clinical assessments.

Results:

The study cohort included 156 patients (mean age 53, 67% women, 33% had rheumatoid arthritis), for a total of 3120 MCP and PIP joints. 20/156 (12.8%) patients, and 69/3120 (2.2%) joints had effusions per clinical assessment. The average person-level visual assessment sensitivity and specificity of photos were 0.44 and 0.82, respectively. The average visual assessment joint-level sensitivity and specificity of photos were 0.14 and 0.97, respectively. Performance was slightly better for videos. Assessor agreement was poor (Kappa 0.12-0.17).

Conclusions:

Visual inspection of photos and videos can be used to detect joint effusions of the MCP and PIP joints with good specificity and poor sensitivity. Patients and rheumatologists should be aware of these limitations when conducting remote telehealth assessments.