Predictive Validity of a Caries Risk Assessment Model at a Dental School

Caries risk assessment (CRA) is an essential component informing clinical decision making for personalized caries management. The aim of this study was to assess the predictive validity of the CRA model used at the University of Michigan School of Dentistry. Data from patients treated in the school's adult integrated clinics between 2011 and 2014 who had a CRA documented were accessed from the electronic health record. Data extracted included caries risk category, date of reassessments, presence of caries risk indicators/factors, and treatments completed. Out of 2,449 patients who had a CRA completed in 2011, 447 had one or more reassessments at least 180 days after the initial CRA. Caries risk status was significantly associated (p<0.0001) with the number of new caries-related treatments during follow-up, with significant increases (p<0.01) in these needs for each caries risk level. For the 11,152 patients with a completed CRA in 2011-14, the prediction model (area under the curve=0.82) included the following risk factors significantly (p<0.001) associated with increased caries experience/needs over time: stagnant plaque (OR 2.6, 95% CI 2.4, 2.9), presence of salivary risk factors (OR 2.6, 95% CI 2.2, 2.9), presence of dietary risk factors (OR 3.2, 95% CI 2.9, 3.6), lack of adequate protective factors (OR 2.1, 95% CI 1.8, 2.4), presence of conditions that affect compliance (OR 2.4, 95% CI 1.9, 3.1), and past/current caries experience (OR 23.7, 95% CI 21.1, 26.7). High risk patients were significantly more likely (p<0.0001) to have restorative, extraction, or endodontic treatments completed due to caries and less likely (p<0.0001) to have diagnostic procedures than low or moderate risk patients. These results suggest that the CRA model used at the dental school can help predict adult patients' future caries experience/needs.