Statistical Modeling of Dental-Unit-Water Bacterial Test Kit Performance

Objectives: While it is important to monitor dental-water quality, it is unclear whether in-office test kits provide bacterial counts comparable to the gold standard method (R2A). Studies were conducted on specimens with known bacterial concentrations, and from dental units, to evaluate test kit accuracy across a range of bacterial types and loads. Methods: Colony forming units (CFU) were counted for each of 12 replicate samples from each specimen source, using 9215C/R2A and two types of test kits, and conformity to Poisson distribution expectations was evaluated. Poisson regression was used to test for the effects of source, measurement device, and their interaction, and to estimate rate ratios for test kits relative to R2A. Results: For all devices, distributions were Poisson for low CFU/mL, when only beige-pigmented bacteria were considered. For higher counts, R2A remained Poisson but kits exhibited over-dispersion. Both kits undercounted relative to R2A, but the degree of undercounting was reasonably stable. Kits did not grow pink-pigmented bacteria from dental-unit water identified as Methylobacterium rhodesianum. Conclusions: Only one of the test kits provided results with adequate reliability at higher bacterial concentrations. Undercount bias could be estimated for this device and used to adjust test kit results. Multiplying a Millipore test kit result by a factor of 1.5 yields an estimate that is suitable for testing compliance with respect to the 500 CFU/mL count limit, as it would be determined by R2A for non-pink-pigmented bacteria. As kits do not support the growth of Methylobacteria spp., which are potential pathogens, there is a serious limitation in this surveillance methodology.