BACKGROUND: Public health and other population-based studies often depend on participants' self-reported disease status to assess prevalence, incidence, and disease trends. We sought to assess the feasibility of self-reported periodontal disease measures using dental history questions combined with demographic and medical history to predict periodontal disease. METHODS: We evaluated results from two separate population-based studies carried out at the University at Buffalo, Buffalo, New York, i.e., the "Periodontal Infection and Risk for Myocardial Infarction Study," a study of 1,578 adults assessing the association between periodontal disease and myocardial infarction and the "Periodontal Disease Research Center" (the Erie County Study), an epidemiologic risk assessment study of 1,438 adults. In each study, an extensive list of oral health questions was asked, and a comprehensive medical history, blood analysis using chemistry and hematology tests, and demographic data were collected. RESULTS: Using a predefined measure of severity of periodontal disease, we compared patients with severe disease to all others (i.e., those with moderate and no or mild disease). We examined areas under the curve (AUC) of the receiver operating curve to determine the best models, adding one, two, or three dental variables in all possible combinations. The AUC maximized at 0.76, and the combined sensitivity and specificity maximized at 142 and were comparable in both studies. CONCLUSIONS: Self-reported measures of periodontal disease are moderately predictive of clinical attachment loss. The demographic variables of age, race, smoking, gender, and diabetes mellitus accounted for much of the predictive power for self-reported periodontal disease; however, increases in sensitivity and specificity in the C statistics occurred when questions, including "Gum surgery in the past?," "Sore gums in the past?," "Scaling in the past?," "Bleeding gums now?," "Periodontal surgery in the past 2 years?," and "Chewing satisfaction?," were added to the model.
BACKGROUND: Public health and other population-based studies often depend on participants' self-reported disease status to assess prevalence, incidence, and disease trends. We sought to assess the feasibility of self-reported periodontal disease measures using dental history questions combined with demographic and medical history to predict periodontal disease. METHODS: We evaluated results from two separate population-based studies carried out at the University at Buffalo, Buffalo, New York, i.e., the "Periodontal Infection and Risk for Myocardial Infarction Study," a study of 1,578 adults assessing the association between periodontal disease and myocardial infarction and the "Periodontal Disease Research Center" (the Erie County Study), an epidemiologic risk assessment study of 1,438 adults. In each study, an extensive list of oral health questions was asked, and a comprehensive medical history, blood analysis using chemistry and hematology tests, and demographic data were collected. RESULTS: Using a predefined measure of severity of periodontal disease, we compared patients with severe disease to all others (i.e., those with moderate and no or mild disease). We examined areas under the curve (AUC) of the receiver operating curve to determine the best models, adding one, two, or three dental variables in all possible combinations. The AUC maximized at 0.76, and the combined sensitivity and specificity maximized at 142 and were comparable in both studies. CONCLUSIONS: Self-reported measures of periodontal disease are moderately predictive of clinical attachment loss. The demographic variables of age, race, smoking, gender, and diabetes mellitus accounted for much of the predictive power for self-reported periodontal disease; however, increases in sensitivity and specificity in the C statistics occurred when questions, including "Gum surgery in the past?," "Sore gums in the past?," "Scaling in the past?," "Bleeding gums now?," "Periodontal surgery in the past 2 years?," and "Chewing satisfaction?," were added to the model.
Authors: Ted R Mikuls; Jeffrey B Payne; Fang Yu; Geoffrey M Thiele; Richard J Reynolds; Grant W Cannon; Jeffrey Markt; David McGowan; Gail S Kerr; Robert S Redman; Andreas Reimold; Garth Griffiths; Mark Beatty; Shawneen M Gonzalez; Debra A Bergman; Bartlett C Hamilton; Alan R Erickson; Jeremy Sokolove; William H Robinson; Clay Walker; Fatiha Chandad; James R O'Dell Journal: Arthritis Rheumatol Date: 2014-05 Impact factor: 10.995
Authors: Michael J LaMonte; Kathleen M Hovey; Amy E Millen; Robert J Genco; Jean Wactawski-Wende Journal: J Periodontol Date: 2013-12-19 Impact factor: 6.993
Authors: Hon K Yuen; Matthew S Shotwell; Kathryn M Magruder; Elizabeth H Slate; Carlos F Salinas Journal: J Spinal Cord Med Date: 2009 Impact factor: 1.985