A method for adjusting caries increments for reversals due to examiner misclassification

Abstract

Two types of reversals occur in longitudinal caries studies, remineralization (true reversals) and reversals due to examiner misclassification (D3 to S; F to S). Since 1966, the standard practice has been to subtract examiner reversals from the crude increment (CCI), resulting in a net caries increment (NCI). The use of the NCI has been based on the assumption that examiners make an equal number of false positive and false negative errors both at baseline and follow-up examination. Consequently, the difference between the two caries prevalence scores would provide an unbiased estimate of caries incidence between examinations. The NCI considers all reversals to be true reversals which is an extreme strategy, particulary when the level of diagnosis is set at lower thresholds. In this study we compromised between the NCI and CCI by creating a simple formula to calculate the caries increment using a prevalence-based adjustment for reversals. The formula is ADJCI = y2(1-(y3/(y3 + y4))), where y2 = S to D or F; y3 = D or F to S; y4 = D to D/F or F to F. The impact of this adjustment is illustrated using data from a random sample of 452 older black and white adults followed over a 3-yr period. The ADJCI was more likely to result in higher caries increments and more observed intergroup differences than the NCI, a finding that has implications for clinical trials. For example, the crude 3-yr coronal DS increment for whites was 0.62, the NCI was 0.26 and the ADJCI was 0.56. There were significant black-white differences for the CCI and ADJCI, but not the NCI. We conclude that the NCI was too severe of an adjustment for reversals for this study population.