Impact of analytic conventions on outcome measures in two longitudinal studies of dental caries

Caries increment is the primary outcome measure used in most longitudinal s tudies of caries. Its computation requires the adoption of specific analyti c conventions for handling examiner misclassification ("reversals"), teeth lost due to caries, and multiple events such as caries initiation and progr ession. However, little is known about the impact of these analytic convent ions on cumulative incidence and incidence density - two additional outcome measures used in longitudinal studies. Objective: We evaluated the impact of analytic conventions on caries outcome measures in two longitudinal cari es studies. Methods: A secondary analysis was undertaken using data from tw o studies: a 2-year observational cohort study of 683 older adults and a 3- year clinical trial of 1754 children. Crude, net and adjusted increments we re compared, effects of inclusion and exclusion of teeth lost due to caries were contrasted and measures of caries initiation and progression were ass essed among subgroups in each study. Results: Cumulative incidence indicate d a high risk of caries (50% or more) in both studies but was not sensitive to subgroup differences. Increment and incidence density measures consiste ntly revealed between-subgroup differences when subgroups had comparable nu mbers of teeth. For all three outcome measures, crude DFS values were large r than net DFS values. The adjusted DFS was similar to crude DFS in older a dults, but similar to net DFS in children. Measurement of caries initiation and progression increased disease estimates for all three outcome measures . Conclusions: In the populations studied, analytic conventions for handlin g examiner misclassification, teeth lost due to caries and multiple events had a greater impact on inferences than the choice of outcome measure.