Analysis of an administrative database of half a million restorations over 11 years

Aim This paper describes the analytical approach used to extract empirical distributions of the interval from the date of placement of a direct restoration to the date when the next intervention took place on the same tooth, that is, an estimate of restoration survival. Methods Data, based on the complete attendance and treatment history, over the eleven years from 1991 to 2001, of a statistically representative sample of 82,537 General Dental Servicesʹ patients in England and Wales, all of whom received at least one directly placed restoration during the observation period, have been analysed. The patients on the database received a total of 719,009 courses of treatment, and there were 503,965 occasions when a tooth was directly restored. The method of analysis involved, first, the estimation of the probability that the patient will eventually return, given an interval without attending, by analysing the observed patterns of re-attendance. This estimated probability of re-attendance was then used to modify the standard Kaplan-Meier procedure to produce realistic estimates of the hazard of re-intervention. Results The results demonstrate that the newly developed methodology has produced robust estimates of the distribution of survival intervals to next intervention on the same tooth. Regarding attendance patterns, re-attendance probability varies with length of interval since last attendance. There is also a strong association with patient age. The older the patient, the more likely it is that a gap in attendance is indicative that the patient will never return. The detailed cumulative survival curve over eleven years has been plotted and forty-seven per cent of restorations, overall, survived without re-intervention for at least ten years. The entire analysis was then successfully replicated on a second, independently selected, sample of attendance and treatment records, confirming the ten-year survival estimate. Conclusions This newly developed methodology has produced robust estimates of the distribution of survival intervals to next intervention on the same tooth. Patient re-attendance probability varies with length of interval since last attendance and with patient age. The ten-year overall survival rate to next intervention on the same tooth for direct restorations placed within the GDS in England and Wales is circa 47%.