A smartphone-based joint self-assessment app showed modest agreement with physician joint counts and tended to overestimate active rheumatoid arthritis. In the study, patient-derived joint counts led to composite disease activity scores that were better at confirming low disease activity or remission than identifying active disease.
Patient-reported assessments had a 95% positive predictive value for identifying low disease activity and a 94% specificity, but only a 50% negative predictive value and 54% sensitivity, indicating limited ability to detect active disease. Overall agreement between patient and physician classifications was modest.
Researchers led by Swamy Venuturupalli, MD, of Cedars-Sinai Medical Center in Los Angeles evaluated the app-based, cartoon-guided training module in 100 patients with rheumatoid arthritis across six US community rheumatology sites.
Patients completed app-guided assessments of 28 tender and swollen joints prior to an in-office visit, during which physicians performed corresponding joint counts. The paper states visits were intended within 3 days of the app assessment, although Table 1 and the statistical analysis section report intervals of up to 6 days.
Physicians recorded fewer affected joints than patients. Mean patient-reported counts were 10.8 tender joints and 8.2 swollen joints compared with 5.8 and 1.3, respectively, recorded by physicians.
Agreement between patient and physician assessments was modest and higher for tender joint counts than for swollen joint counts, with correlation coefficients of 0.5 and 0.33, respectively.
In multivariable analyses, physician-assessed joint counts were the primary predictors of patient-reported counts. Fibromyalgia and osteoarthritis were associated with higher patient-reported tender joint counts but were not associated with swollen joint counts. Model fit was greater for tender joints than for swollen joints.
These findings suggest patient-derived joint counts may be useful for confirming low disease activity or remission but are less reliable for identifying active disease. Patients tended to report higher joint counts than physicians, particularly for swollen joints.
The authors said modest agreement may reflect real-world differences in patient environment and timing, difficulty distinguishing pain from tenderness, the influence of comorbid conditions such as fibromyalgia, and limitations of app-based training without in-person or video instruction.
“A 95% PPV and a 50% NPV indicated that using patient-derived counts may have utility for confirming that patients with RA are remaining in a low disease activity state or remission but not for accurately assessing RA disease activity for those with less well-controlled disease,” the researchers wrote.
Author disclosures are available in the published study.
Source: ACR Open Rheumatology
