Effectiveness of water fluoridation in caries prevention

To review the effectiveness of adjusted fluoridation of public water supplies in the prevention of dental caries, with emphasis on results of studies published worldwide since 1990 and to discuss aspects of the design and reporting of these studies compared with those published before 1990.

Studies published worldwide, in any language, reporting the effect of water fluoridation in terms of the dmf/DMF caries index between 1990 and 2010 were examined. The literature search was by professional Internet search, back-tracking from references given in publications, hand-searching all issues of four journals and by contacting colleagues in relevant countries. For the dmf index, age 5 year was preferred, and for the DMF index, age 12 year or older was preferred. The results were compared with results obtained from worldwide literature search prior to 1990 by the same author.

Fifty-nine studies of adjusted water fluoridation were identified, yielding 83 evaluations (30 recording dmft/s and 53 recording DMFT/S) from 10 countries. These numbers are lower than pre-1990 results of 113 studies (66 for primary and 86 for permanent teeth) from 23 countries. For the USA, for example, four studies were identified since 1990 compared with 61 studies before 1990. The most number of recent reports came from Brazil and Australia. There were fewer reports of per cent caries reductions (% CR) above 50% in the recent studies. 86% of the post-1990 investigations were concurrent control cross-sectional studies and, of these, 52% used multivariate statistical analysis to adjust for confounding factors. In the eight studies that provided dmf/DMF data before and after adjustment for confounders, the % CR were little affected by these adjustments.

Fewer studies have been published recently. More of these have investigated effect at the multi-community, state or even national level. The dmf/DMF index remains the most widely used measure of effect. % CR were lower in recent studies, and the 'halo' effect was discussed frequently. Nevertheless, reductions were still substantial. Statistical control for confounding factors is now routine, although the effect on per cent reductions tended to be small. Further thought is needed about the purpose of evaluation and whether measures of effect and study design are appropriate for that purpose.