Resumo

MARTINS, Andréa Maria Eleutério de Barros Lima et al. Challenges of a calibration process for an epidemiological study involving quantitative and ordinal categorical variables: one example. Arq. Odontol. [online]. 2011, vol.47, n.4, pp. 196-207. ISSN 1516-0939.

Aim: To report the challenges faced in a calibration process of examiners for an epidemiological study on malocclusion involving quantitative and ordinal categorical variables, showing the results of inter and intra-examiner diagnostic consistency. Materials and Methods: Volunteers were selected to anticipate the conditions the examiners would find in field. Malocclusion was assessed by the DAI, whose criteria were exhibited in the theoretical training and strengthened in practical training. The examiners were divided into groups and the data collected by examination. Inter and intra-examiner agreement was estimated by means of the Intra-class Correlation Coefficient (ICC) for the general DAI and and the ICC or weighted kappa (Kw) for the DAI components, using the PASW and Excel programs. The examiners with an agreement of less than 0.60 were considered inept and should be recommended for another calibration process. Results: In three calibration exercises, which proved to be lengthy and costly, 33 examiners divided into 7 groups, examined and re-examined 315 students. The inter-examiner agreement for the general DAI, among pairs of examiners, ranged from -0.09 to 0.99 (poor to excellent). The agreement of all examiners together, by group, ranged from 0.51 to 0.96 (reasonable to excellent). In the analyses of components from the seven groups, an agreement of > 0.60 among the pairs of examiners was found for dentitions, crowding and spacing. Disagreements among examiner pairs regarding some components were observed even in groups whose agreement among the same pair of examiners was satisfactory for the general DAI. In the intra-examiner agreement for the general DAI, from 0.37 to 1.00 (poor to excellent), two examiners were considered inept; in the evaluation for DAI components, 11 presented ICC and Kw<0.60. Conclusion: The calibration process of examiners configured itself as a challenge to researchers, demanding time and effort. The agreement for the components of DAI identified a higher number of inept examiners, ensuring greater accuracy regarding the results of the epidemiological survey.

Palavras-chave : Evaluation.; Epidemiology.; Reproducibility of results..