To explore the relative utility of cross-sectional fluoride intake measures and Area-Under-the-Curve (AUC) measures of fluoride intake based upon longitudinal follow-up as predictors of dental fluorosis in the early permanent dentition, motivated by the lack of universal availability of longitudinal measures.
Method:
Multiple measures of fluoride intake were available for participants in the Iowa Fluoride Study, a cohort study of children recruited perinatally: intakes based upon dietary report at 6, 12, 24, 36 and 48 months of age (cross-sectional) and AUCs reflecting intake from birth until 12, 24, 36 and 48 months (longitudinal). All measures were available for 237 of 544 children participating in an associated study of genetic and environmental determinants of dental fluorosis. Quantitative definitions of dental fluorosis were based upon the number of affected teeth or zones based upon the Fluorosis Risk Index (FRI). Spearman correlational analyses were used to assess which fluoride intake measurements had the strongest correlations with fluorosis phenotypes; performance of competing predictors was assessed using the method of Dunn and Clark (1969).
Result:
Most fluoride intake variables were strongly correlated among themselves. Among cross sectional measures of fluoride intake, the strongest correlations with fluorosis outcomes were for fluoride intake at 36 months (r=0.22-0.27; all p<0.001); among longitudinal measures, it was AUC from birth to 48 months (r=0.24-0.31; all p<0.0002). There was no evidence that these intake measures differed significantly in predictive ability.
Conclusion:
These results suggest that cross-sectional fluoride intake measures may reasonably be used when longitudinal measures are not available. Similarly, if the stringent completeness criteria of AUC measures create significant missing data, substitution of cross-sectional measures results in increased sample size and greater statistical power. We recommend the use of the cross-sectional measurement at 36 months as a substitute for the AUC measures.