| Cross-sectional studies are widely used to identify potential risk factors for disease, and to provide estimates of exposure-disease (E-D) associations for use in policy making. Given this reliance on cross-sectional studies, this study aimed to determine whether systematic differences occurred when E-D associations were estimated cross-sectionally versus longitudinally. Data from the Piedmont 65+ Dental Study, a longitudinal study of dental disease among North Carolina elders, were used. Persons examined at 0, 18 and 36 months were included (n = 430). Twenty potential risk factors each were identified for tooth loss, periodontal disease and coronal caries, and E-D associations were estimated. Bivariate prevalence odds ratios were estimated using prevalent disease measured at 18 months, while bivariate incidence odds ratios were estimated using incident disease that occurred from 0 to 18 months, and from 18 to 36 months. Pairs of prevalence odds ratios (OR{dollar}sb{lcub}rm C{rcub}{dollar}) and incidence odds ratios (OR{dollar}sb{lcub}rm L{rcub}{dollar}) were compared. The outcome of interest was the ratio of each prevalence odds ratio to its corresponding incidence odds ratio (OR{dollar}sb{lcub}rm C{rcub}{dollar}/OR{dollar}sb{lcub}rm L{rcub}{dollar}). Overall, prevalence odds ratios were almost twice as large as incidence odds ratios (Mean Ratio (MR) = 1.9, 95% Confidence Interval (CI) = 1.5-2.4). Specifically, prevalence odds ratios were 39% larger than incidence odds ratios for tooth loss (MR = 1.4, 95% CI = 0.97-2.0), 53% larger for periodontal disease (MR = 1.5, 95% CI = 1.1-2.2), and 238% larger for caries (MR = 3.4, 95% CI = 2.2-5.1). Putative causal exposures were somewhat less likely to be overestimated by prevalence data (MR = 1.8, 95% CI = 1.2-2.7), compared to presumed non-causal exposures (MR = 2.0, 95% CI = 1.5-2.7). The over-estimation in prevalence odds ratios compared to incidence odds ratios indicates that the assessment of risk factors using prevalence data can be quite tenuous for the oral conditions studied. |
