Realizing early recognition of arthritis in times of increased telemedicine: The value of patient-reported swollen joints

ABSTRACT

Background: Early diagnosis and management of patients with inflammatory arthritis(IA) are critical to improve long-term patient-outcomes. Assessment of joint swelling at joint examination is the reference of IA-identification;early access clinics are constructed to promote this early recognition of IA. However, due to the COVID-19 pandemic the face-to-face capacity of such services is severely reduced. The accuracy of patient-reported swelling in comparison to joint examination has been extensively evaluated in established RA (ρ 0.31-0.67), but not in patients suspected for IA.[1] Objectives: To promote evidence based care in the era of telemedicine, we determined the accuracy of patient-reported joint swelling for actual presence of IA in persons suspected of IA by general practitioners(GP). Methods: Data from two Dutch Early Arthritis Recognition Clinics were studied. These are screening clinics (1.5-lines-setting) where GPs send patients in case of doubt on IA. At this clinic patients were asked to mark the presence of swollen joints on a mannequin with 52 joints. For this study the DIP joints and the metatarsal joints were excluded and, therefore, a total of 42 joints were assessed for self-reported joint swelling. Clinically apparent IA of ≥1 joint determined by the physician was the reference to calculate sensitivity, specificity, positive and negative likelihood ratios (LR+,LR-), and positive and negative predictive values (PPV, NPV) on patient-level. Pearson correlation coefficients(ρ) were determined. Predictive values depend on the prevalence of a disease in a population. Because the prevalence of IA in a 1.5-lines-setting will differ from a primary care setting, post-test probabilities of IA were estimated for two lower prior-test probabilities as example, namely 20% (estimated probability in patients GPs belief IA is likely) and 2% (prior-test probability with less preselection by GPs), using likelihood ratios and nomograms. Results: A total of 1637 consecutive patients were studied. Median symptom duration was 13 weeks. 76% of patients marked ≥1swollen joint at the mannequin. 41% of patients had ≥1swollen joint at examination by rheumatologists. ρ was 0.20(patient-level)-0.26(joint-level). The sensitivity of patients-reported joint swelling was high, 87%, indicating that the majority of patients with IA had marked swelling on the mannequin. However the specificity was 31%, indicating that 69% of persons without IA had also done so. The LR+ was 1.25;the LR-0.43. The PPV was 46%, the NPV 77%. Thus the PPV increased hardly (from 41% to 46%) and the NPV somewhat (from 59% to 77%). Also in settings with prior-test probabilities of 20% and 2%, estimated PPVs (from respectively 20% and 2% to 24% and 2%) and NPVs (from respectively 80% and 98% to 90% and 99%) hardly increased. Conclusion: Patient-reported joint swelling had little value in distinguishing patients with/without IA for different prior-test probabilities, and is less valuable in comparison to self-reported flare detection in established RA.