| 2067 Impact of Examiner Error on Prediction of Incident Caries Using Past Caries Experience | ||
|
S. PETTI1, S. IANNAZZO1, G. TARSITANI1, and G.D. SLADE2, 1 University of Rome "La Sapienza", Roma, Italy, 2 University of Adelaide, Australia, Australia Implicit in any “high risk” approach for population-control of caries is a capacity to identify, through epidemiological studies, easily measurable and highly predictive patient characteristics. Traditionally, past caries experience (PCE) has been the single most efficient risk predictor, although its predictive properties may vary between populations and be affected by adjustments for examiner error that are necessary when calculating caries incidence. Objectives: To determine the ability of PCE to predict incident caries in two contrasting populations in which incident caries is determined using three methods of adjustment for examiner reversals. Methods: Data were obtained from two observational cohort studies: in Italy, dft was measured at baseline and after 1-2 yrs among 754, 3-to-5-year-old children; in Australia, DFS was measured at baseline and after 2 yrs among 693 adults aged 60+ years. Crude, net and adjusted caries increments were calculated following methods described by Slade and Caplan (CDOE 1999;27:236-48). Increment values greater than zero were used to define incident cases for each method (CCInc, NCInc and AdjCInc, respectively). Relative risks, using a reference group with no PCE, were calculated for three categories of PCE within each population. Results: 75.2% of children had no PCE, and incidence rates in the cohort were low: CCInc=19.2%, NCInc=12.7%, AdjCInc=15.1%. RRs were higher for CCInc than NCInc or AdjCInc, while NCInc did not increase significantly or monotonically with increasing PCE. Among older adults, PCE was extensive (only 7.8% had no PCE) and incidence rates were high (CCInc=51.5%, NCInc=39.7% and AdjCInc=48.3%). Compared with children, RRs for elderly were smaller (range=1.12-2.17) and did not differ for different methods of calculating increment. Conclusions: In populations with low caries frequency, failure to account for examiner error in calculating caries increment (CCInc method) can over-estimate the predictive capacity of PCE. However, NCInc leads to under-estimate of risk associated with PCE. | ||
| Seq #214 - Caries Prediction, Clinical Trials, Diagnosis 3:45 PM-5:00 PM, Friday, 27 June 2003 Svenska Massan Exhibition Hall B | ||
|
Back to the Cariology Research Program
| ||