76 zyxwvutsrqpon Journal of Public Health Dentistry Validation of a Simple Approach to Caries Risk Assessment z James D. Bader, DDS, MPH; Nancy A. Perrin, PhD; Gerard0 Maupome, BDS, MS, PhD; Brad Rindal, DDS; William A. Rush, PhD Abstract zyxwvutsrqp Objective: zyxwvutsrqp This study examined the predictive validity of a simple subjective method promoted to dentists for assessing their patients’ caries risk. Methods: Data from two large group practices that have used guideline-assisted caries risk assessment (CRA) for several years were analyzed retrospectively to determine the receipt of caries-related treatment following a CRA. Patient age and receipt of caries preventive treatment subsequent to the CRA were control variables in logis- tic regressions to determine the likelihood of caries-related treatment for low, mod- erate, and high risk groups. Results: Among 45,693 individuals in the two plans, those categorized as being at high caries risk were approximately four times as likely to receive any caries-related treatment as those categorized as being at low caries risk. Those categorized as at moderate risk were approximately twice as likely to receive any treatment. In addition, for those at elevated risk who required any treatment, the number of teeth requiring treatment was larger. Conclusion: The results of this study provide the first large-scale, generalizable evidence for the validity of dentists’ subjective assessment of caries risk. Key Words: dental caries; caries risk assessment; dental caries incidence; dentist practice patterns Introduction Methods to assess the likelihood that an individualwould developone or more zyxwvuts L\ caries lesions at some time zyxwvuts 111 zyxwvutsrqponmlkjihg ‘ > : lure have been studied for morL __,. rhree decades (1).A re- cent systematic review that consid- ered only multivariate models of car- ies experience progression included 43 studies designed to identify risk predictors for future caries while ex- cluding an additional 126 (2). How- ever, it was not until the publication of a supplement to the Journal of the American Dental Association in 1995 that the practical application of the growingbody of information concern- ing diverse risk indicators for dental caries was clearly explicated for prac- ticing dentists (3). The supplement urged that caries be regarded as a ”chronic, infectious multifactorial disease process,” and that individu- als be assessed routinely for the pres- ence or absence of risk indicators for this process. The supplement listed a variety of risk indicators to consider when as- sessing an individual’s risk of dental caries, and suggested combinations of these indicators that might be sug- gestive of low, moderate, and high risk categories, but it did not present any computational method to combine and/or differentially weight the risk indicators. At about the same time that the supplement appeared, a small number of caries risk assessment sys- tems designed to be applied clinically were described that incorporated com- putational approaches (4-8).Some of these systemswere based directly on the results of multivariate modeling of caries incidence data (7, 8), while others used many of the same risk in- dicators, but they were weighted ac- cording to expert opinion (4-6). All but one of these risk classification systems called for an extensive variety of in- puts, typically requiring salivaryflow rates, zyxw mutans and/or lactobacilli counts, and formal diet histories in addition to results of clinical exami- nations of the teeth. All required arith- metic operationsto produce a catego- rization score. Until recently, none of the caries risk assessment systems described in the literature had been validated in clinical practice. In this context, vali- dation means that the accuracy of the caries risk categorizations made at baseline are compared to actual car- ies experience determined through one or more follow-up examinations in a population other than that used to develop the predictive formula. Within the past few years, validation studies have been reported for two computational systems, the Cariogram system (9, 101, and the Dentprog system (11, 12). Both sys- tems were effective in categorizing children, and in the case of the Cariogram, elderly adults, by the ex- tent of future caries they would expe- rience. However, these validity stud- ies involved relatively small numbers of socially homogenous subjects and were performed using data collected during epidemiological studies and secondary analyses of clinical records. The practicality of using the systems in and the generalizabilityof these results to dental practice remain in question. Indirect evidence from dental claims data suggest that practicing dentists may not employ caries risk assessment strategiesin their practices (13,141. The lack of early response to Send correspondence to Dr. James D. Bader, Sheps Center, Ste. 200, 725 Airport Rd., Chapel Hill, NC, 27514. Phone: (919) 966-5727; Fax: (919) 966-3811. E-mail: jim bader8unc.edu. Reprints will not be available. Dr. Bader is affiliated with the School of Dentistry and Sheps Center for Health Services Research, University of North Carolina. Drs. Perrin and Maupome are affiliated with the Kaiser Permanente Center for Health Research, Portland, OR. Drs. Rindal and Rush are affiliated with Healthpartners Research Foundation, Minneapolis, MN. Acknowledgement: Supported by grant R01 HS 013339 from the Agency for Healthcare Research and Quality. Manuscript received 6/7/04; returned to authors for revision 7/27/04; final version accepted for publication 8/27/04.