Varela Aneiros, IvánFernández Feijoo, JavierGarcía Mato, ElianeDiniz Freitas, MárcioMartínez, IsabelRoca Pardiñas, JavierDiz Dios, PedroLimeres Posse, Jacobo2024-02-122024-02-122022Disability and Health Journal, Volume 15, Issue 2, April 2022, 101229. https://doi.org/10.1016/j.dhjo.2021.1012291936-6574http://hdl.handle.net/10347/32727Background: The dental treatment of individuals with intellectual disability can represent a considerable professional challenge. Objective: To develop a model for predicting the behavior of patients with intellectual disability in the dental office. Methods: The study group comprised 250 patients with Down syndrome (DS), autism spectrum disorder (ASD), cerebral palsy (CP), idiopathic cognitive impairment or rare disorders. We collected their demographic, medical, social and behavioral information and identified potential predictors (chi-squared test). We developed stratified models (Akaike information criterion) to anticipate the patients'behavior during intraoral examinations and to discern whether the dental treatment should be performed under general anesthesia. These models were validated in a new study group consisting of 80 patients. Goodness of fit was quantified with sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and area under the receiver operating characteristic curve (AUC). We developed a mathematical algorithm for executing the models and developed software for its practical implementation (PREdictors of BEhavior in Dentistry, “PREBED”). Results: For patients with DS, ASD and CP, the model predicting the need for physical restraint during examination achieved a PPV of 0.90, 0.85 and 1.00, respectively, and an NPV of 0.66, 0.76 and 1.00, respectively. The model predicting the need for performing treatment under general anesthesia achieved a PPV of 0.63, 1.00 and 1.00, respectively, and an NPV of 1.00, 1.00 and 0.73, respectively. However, when validating the stratified models, the percentage of poorly classified individuals (false negatives + false positives) ranged from 24% to 46.6%. Conclusions: The results of the PREBED tool open the door to establishing new models implementing other potentially predictive variables.eng© 2021 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).Atribución 4.0 Internacionalhttp://creativecommons.org/licenses/by/4.0/DisabilityBehaviorDentistryDental managementGeneral anesthesiaPredictive modelsjournal article10.1016/j.dhjo.2021.1012291876-7583open access