Implications of less-than-perfect reliability of clinical parameters for the misclassification of periodontitis.

OBJECTIVE

This study aimed to evaluate and compare the misclassification of periodontitis using two operational protocols currently employed in research, as a result of intra-examiner less-than-perfect reliability of their formative clinical parameters.

METHODS

Full-mouth duplicate recordings (six sites per tooth) of probing depth (PD) and clinical attachment level (CAL) were obtained by a single examiner, 1 week apart, from 148 adults with self-perceived need for periodontal treatment. They were used to classify periodontitis subjects according to the protocols proposed by the 5th European Workshop in Periodontology (EWP5) and the collaboration between the US Centre for Diseases Control and Prevention and the American Academy of Periodontology (CDC-AAP).

RESULTS

Hundred percent and 91.2% of the subjects were classified as periodontitis cases according to the EWP5 and CDC-AAP protocols, respectively. Test-retest agreement was observed in 95% and 85% of the classified subjects according to the EWP5 and CDC-AAP protocols, respectively. The difference was statistically significant (P < 0.01). Alignment of protocols, in terms of number of classification categories, resulted in a similar test-retest agreement of 95% using the EWP5 protocol for extensive periodontitis and 90% using the CDC-AAP protocol for severe periodontitis (P = 0.09). Misclassification was nondifferential when subjects were classified according either to the EWP5 or the aligned CDC-AAP protocols, while the reliability using the original CDC-AAP protocol depended on the extent of bleeding on probing, CAL ≥3 mm and PD ≥4 mm.

CONCLUSION

This study suggests the EWP5 protocol is less error-prone than the CDC-AAP protocol in adult subjects with perceived need for periodontal treatment, as a result of intra-examiner less-than-perfect reliability of periodontal clinical parameters. Extensive or severe cases presented similar misclassification. Further studies using survey-like conditions are recommended to confirm these findings.