28 JADA 142(1) http://jada.ada.org January 2011 STORY COVER D iabetes is the seventh leading cause of death in the United States. 1 In 2007, 17.9 million people in the United States had diagnosed diabetes, and 5.7 million people had undiagnosed diabetes. 1 Investigators have pro- posed programs to screen for undiag- nosed diabetes, which could be cost- effective if implemented. 2-7 Most of these studies, however, were con- ducted in general medical settings. 5-7 Investigators have proposed pro- grams to screen for undiagnosed dia- betes in the dental office, but few have proposed and tested specific clinical algorithms. 8-14 In April 2007, the investigators associated with The Scottsdale Project, a consensus group that brought together a wide range of medical and dental experts, discussed the quality of existing evi- dence related to associations between diabetes and periodontal disease. 11 They concluded that “there is sufficient evidence of a bidirec- tional relationship between diabetes and periodontal disease to formulate guidelines for screening for undiag- nosed diabetes and the comanage- ment of patients with diabetes in the clinical practice of dentistry and dental hygiene.” However, a specific clinical guideline was not formulated for use with dental patients. The oral manifestations of dia- betes, which might give a dentist the first indication that a patient has the disease, include dry mucous mem- branes (xerostomia or dry mouth), periodontal diseases, oral candidi- asis, burning mouth sensation (glos- Dr. Li is a doctoral candidate, Department of Epidemiology, Harvard School of Public Health, Boston. Dr. Williams is a senior lecturer, Department of Biostatistics, Harvard School of Public Health, Boston. Dr. Douglass is a professor emeritus Harvard School of Dental Medicine, and a professor, Harvard School of Public Health. Address reprint requests to Dr. Douglass at Harvard School of Dental Medicine, 188 Longwood Ave., Boston, Mass. 02115, e-mail “chester_douglass@hsdm.harvard.edu”. Development of a clinical guideline to predict undiagnosed diabetes in dental patients Shanshan Li, MD, MSc; Paige L. Williams, PhD; Chester W. Douglass, DMD, PhD ABSTRACT Background. In 2007, 17.9 million people in the United States had diagnosed diabetes, and 5.7 million had undiagnosed diabetes. The authors developed a clinical guideline to help den- tists identify patients with undiagnosed diabetes. Methods. The authors used classification and regression tree (CART) methods to generate different prediction models using data from the Third National Health and Nutrition Examination Survey (NHANES III) (1988-1994) and data from NHANES 2003-2004 for external validation. They classified participants who answered “No” to the question “Have you ever been told by a physician that you have diabetes?” and who had a fasting plasma glucose level greater than or equal to 126 milligrams per deciliter as having undiagnosed dia- betes. The authors used oral examination data regarding the pres- ence or absence of periodontitis and waist circumference, as well as data on participants’ self-reported oral health status, weight, age, family history and race or ethnicity. The authors chose the best pre- diction model by means of 10-fold cross-validation, as well as internal and external validation methods, which evaluated each prediction model by comparing sensitivity, specificity, area under the receiver operating characteristic curve and ease of use criteria (N = 7,545). Results. The authors’ final clinical guideline for predicting undiag- nosed diabetes in dental patients had a sensitivity of 82.4 percent, a specificity of 52.8 percent and a receiver operating characteristic area under the curve of 0.72. They found that waist circumference, age, self-reported oral health status, self-reported race or ethnicity and self-reported weight information could be used to predict the risk of having undiagnosed diabetes (range, 0.1 to 9.1 percent). Conclusion. Dental care providers should consider using a clin- ical guideline that includes the following predictors: waist circumfer- ence, age, self-reported oral health, self-reported weight and self- reported race or ethnicity, as well as any additional information on periodontal status and family history of diabetes. Clinical Implications. This clinical guideline could help den- tists identify patients with undiagnosed diabetes, resulting in the early identification of dental patients who require treatment for dia- betes and, thus, reduce morbidity and health care costs. Key Words. Undiagnosed diabetes; dental office; clinical guideline. JADA 2011;142(1):28-37. J A D A C O N T I N U I N G E D U C A T I O N ✷ ✷ ® A R T I C L E 1 Copyright © 2011 American Dental Association. All rights reserved.