[Validation of the reproducibility of the DMF Index used in bucco-dental epidemiology and evaluation of its 2 clinical forms].

The DMF index is the one most commonly used in oral epidemiology to quantify the extent of caries. It represents the total of untreated decayed, missing and filled teeth in a person's mouth. A review of the literature shows that while this index is internationally recognized by the scientific community and people engaged in clinical dental practice, information demonstrating its reproducibility is scanty. The initiation and promotion of studies on the validity of the DMF index depend on first evaluating its reproducibility. This evaluation is based on an analysis of the rate of concordance between the results observed when a sample of subjects is examined twice, either by the same examiner or by two examiners at different times. Three types of comparison were carried out as part of this presentation: comparisons of results by one examiner, between two examiners, and between the two clinical forms of the DMF index according to whether it is calculated by tooth (DMFT) or tooth surface (DMFS). Each of the comparisons covered a group of 50 or so adult subjects selected solely for having an adequate number of teeth presenting different conditions. In all 215 persons were examined twice, making up a total of over 21,000 characters recorded. The state of each of the 32 permanent teeth, or of their 148 surfaces for the DMFS index, was codified using a digit selected from the seven possibilities corresponding to the letters D, M, F, and S (tooth that has not developed, untreated decayed, extracted, filled, healthy, replaced by a fixed or removable prosthesis). The statistical analysis covered 5 variables: the number of decayed, missing, filled and healthy teeth and the DMF index, for the whole mouth and tooth by tooth. The main finding emerging from this study is that for each comparison the rates of concordance are very high, higher for the one-examiner comparisons (> 64%) than for the two-examiner comparisons (> 45%). Comparison between the two types of data-gathering shows rates of over 72% for all the variables. The correlation coefficients between the scores attributed in two successive examinations are all over 0.9, with very small variation intervals even when the rates of concordance are lower. In fact, the differences between the results are generally of the order of +/- 1 tooth. Except in two cases, the Student test (t test) is not significant, which indicates that the means of each of the 5 variables did not differ statistically from one observation to another.(ABSTRACT TRUNCATED AT 400 WORDS)