Performance of Self-Reported Measures for Periodontitis in Rheumatoid Arthritis and Osteoarthritis

Background: This study evaluates the performance of self-report against the reference standard of clinically defined periodontitis in patients with rheumatoid arthritis (RA) and osteoarthritis (OA) after accounting for factors associated with periodontitis. Methods: Six self-report periodontitis questions were evaluated in patients with RA and OA. Questions were validated against a reference standard of severe and moderate-to-severe periodontitis based on full-mouth examination. Multivariable logistic regression was used to evaluate the performance of: 1) self-report alone; 2) age, sex, education, and smoking status; and 3) a combination of the above. Model performance was assessed using the c-statistic. Convergent validity of self-reported "bone loss/deep pockets'' and "loose teeth'' was assessed; associations of self-report with RA disease characteristics were explored. Results: Self-report performed similarly in RA and OA, with individual question specificity for periodontitis >= 68% and sensitivity from 9.8% to 45%. Question-only models yielded c-statistics of 0.66 to 0.72, whereas risk factor-only models yielded c-statistics of 0.74 to 0.79. The highest-performing models incorporated both self-report questions and periodontitis risk factors, with c-statistics >= 0.79. Greater radiographic alveolar bone loss was observed among participants reporting "bone loss/deep pockets'' (P < 0.001) and "loose teeth'' (P < 0.001). Among patients with RA, "loose teeth,'' but not other self-report items, was associated with rheumatoid factor positivity (P = 0.047) and higher disease activity (P < 0.001). Conclusions: Patient self-report, when combined with other risk factors, performs well in identifying periodontitis among patients with RA and OA. Self-report questions related to alveolar bone loss exhibit excellent convergent validity in these patient subsets.