Community non-specialists’ periodontal tool: identifying periodontitis in older adults with diabetes

Periodontitis and diabetes have a bidirectional relationship, and timely referral for periodontal care is important for adults with diabetes. Existing questionnaire-based approaches may be difficult to implement in community settings. We therefore sought to develop a concise, non-specialist-administered tool for older adults with diabetes. We developed an 11-item questionnaire model and a stepwise-derived 3-item simplified model in 219 older adults with diabetes or at high risk of diabetes, and externally evaluated the simplified model in an independent cohort of 479 participants. Reporting followed the TRIPOD and STARD 2015 statements. Logistic regression was the primary modeling method; discrimination, calibration, and decision-curve analyses were assessed against the study’s prespecified clinical reference outcome. The 11-item model showed good discrimination (AUC = 0.813, 95% CI: 0.756–0.870) and calibration (mean absolute error = 0.033). The 3-item model showed acceptable discrimination (AUC = 0.775, 95% CI: 0.716–0.834) and calibration (mean absolute error = 0.038). In the external dataset, the 3-item model had an AUC of 0.755. Decision-curve analysis suggested potential clinical utility across the evaluated threshold range. The questionnaire models may help community non-specialists identify older adults with diabetes who could benefit from periodontal referral. They are screening and referral-support tools, not substitutes for a comprehensive periodontal examination, and require further multicenter validation.