Increasing number of teeth present in a quasi-longitudinal study in adults: a methodological note

This paper discusses the impossible result of an increase in the number of teeth present in a quasi-longitudinal study of an adult population. In 1983 and 1994, random samples were drawn from birth cohorts 1939-48 and 1930-38. The paradoxical result that adults acquired new teeth in the quasi-longitudinal study gave rise to questions about the comparability of the two samples. Four main sources of failure in comparability are described and discussed: instrumental stability, criteria consistency, response bias, and compositional change in aging cohorts. The lower response rate among persons with two edentulous jaws, and possibly among persons with one dentate jaw, was concluded to be the main reason for the increase in the overall group median and mean number of teeth present. In general, in studies with a similar study design, problems of comparability between the samples demand one's attention and should therefore be carefully evaluated. Compositional change of the population as a source of failure in sample comparability cannot be avoided, while the other three sources (instrumental stability, criteria consistency, response bias) can largely be avoided by careful planning and efficient use of resources.