Method: A sample of 84 head and neck cancer patients was selected before radiotherapy. The frequency of periodontitis was defined according to the diagnostic criteria of Lopez et al. (2002), Centers for Diseases Control and Prevention (CDC, 2007), Hujoel et al. (2006), Beck et al. (1990), Machtei et al. (1992) and Consensus (2005), based on the clinical parameters obtained by a single calibrated examiner (Kw = 0.71). The kappa test was used to evaluate the agreement between definitions. Sensitivity (ST), specificity (SP) and area under ROC curve (ROC) were calculated using CDC (2007) criterion as the gold standard.
Result: The frequencies of periodontal disease, ST, SP and ROC were, respectively: 53.6%, 0.92, 0.73, 0.82 (Lopez et al.); 81.0%, 0.56, 0.97, 0.76 (Hujoel et al.); 40.5%, 0.88, 0.53, 0.70 (Beck et al); 26.2%, 0.92, 0.34, 0.63 (Machtei et al.) and 13.1%, 1.0, 0.19, 0.60 (Consensus). The agreements between CDC and the other criteria were: 0.59 (Hujoel et al.); 0.56 (Lopes et al); 0.31 (Beck et al); 0.18 (Machtei et al.); 0.12 (Consensus).
Conclusion: The use of different case definitions had a great impact on the frequency of periodontitis. CDC and Hujoel et al. (K=0.59) and CDC and Lopez et al. (K=0.56) had a moderate agreement. The criterion with higher sensitivity (0.92) and specificity (0.73) was Lopez et al. (2002), which showed the highest area under the ROC curve (0.82).