Construction and validation of a risk prediction model for oral frailty among community-dwelling older adults

WU Qian; CHENG Fang; CAI Wenjun; JIANG Yan; SHI Qiang

doi:10.12102/j.issn.1009-6493.2025.18.003

Abstract：Objective:To explore risk factors for developing oral frailty among community-dwelling older adults,and to construct and validate a predictive model.Methods:Older adults residing in 3 communities in Hefei city,Anhui province were selected as the study participants from January 2022 to January 2024.Among them,411 older adults from 2 communities were assigned to the modeling set,while 175 older adults from 1 community were assigned to the validation set.Data were collected using a general information questionnaire,Oral Frailty Index-8(OFI-8),Frailty Phenotype(FP),Oral Health Assessment Tool(OHAT),and Mini-Nutritional Assessment Short Form(MNA-SF).LASSO regression and Logistic regression analyses were employed to identify influencing factors of oral frailty.A predictive model was developed using the rms package in R software.The goodness-of-fit and predictive performance were evaluated using the Hosmer-Lemeshow test and receiver operating characteristic(ROC)curve analysis.Decision curve analysis(DCA)was used to assess the clinical applicability of the model.Results:In the modeling set,230 out of 411 older adults(56.0%)were diagnosed with oral frailty,while in the validation set,88 out of 175 older adults(50.3%)were diagnosed with oral frailty.Logistic regression analysis revealed that passive smoking,living arrangement,types of chronic diseases,physical frailty,denture use,and nutritional status were included in the predictive model(P<0.05).The area under the ROC curve(AUC)for the modeling set was 0.952(95%CI 0.934-0.970),with a Youden index of 0.785,sensitivity of 89.6%,specificity of 89.0%,and an optimal cutoff value of 0.505.The AUC for the validation set was 0.936(95%CI 0.902-0.970),with a Youden index of 0.691,sensitivity of 86.4%,specificity of 82.8%,and an optimal cutoff value of 0.505.The Hosmer-Lemeshow test for the modeling set yielded χ2=13.375(P=0.100).The Hosmer-Lemeshow test for the validation set yielded χ2=19.497(P=0.112).The calibration curves for both the modeling set and the validation set indicated good agreement between the predicted and observed outcomes.Internal validation yielded an AUC of 0.878.DCA results showed positive net benefits within risk thresholds of 5%～96%for the modeling set and 5%～92%for the validation set.Conclusions:The constructed risk prediction model for oral frailty among community-dwelling older adults demonstrated clinical utility.It could serve as a tool for community workers to facilitate early identification,prevention,and effective management of oral frailty.