Agreement amongst examiners assessing dental fluorosis from digital photographs using the TF index

To compare the scoring of dental fluorosis by experienced examiners from digital photographs using the TF index.

120 images were selected from 703 photographs obtained during a clinical trial (Tavener et al., 2004). The selection process was stratified so that the full range of defects seen in the main study was included. The children, aged 8-10 years, were from deprived areas of Manchester, England with fluoride levels in the drinking water of less the 0.1 ppm F. The photographs of the upper and lower anterior sextants were taken after cleaning and drying the teeth. The examiners were identified by searching Medline for individuals who had previously used the TF index or had experience of scoring dental fluorosis. Of the 12 examiners identified, 10 agreed to take part. Each examiner was provided with identical CDs containing a PowerPoint presentation of the images. Twelve images were duplicated and interspersed amongst the 120 images to assess intra examiner agreement. Each examiner was also supplied with a table listing the criteria and illustrations for each of the TF index scores (Fejerskov et al., 1988).

The prevalence of fluorosis (TF > 0) amongst the 10 examiners ranged from 43% to 70% and from 2% to 13% for the more severe scores (TF 3 or 4). Paired agreements amongst subject scores for the 10 examiners, measured using a weighted Kappa score, ranged from 0.40 to 0.71.

It is concluded that although the criteria for the TF index are well defined, it is possible that examiners may interpret the criteria in different ways and conditions in which images are viewed may need to be standardised. This study may explain some of the differences in the prevalence and severity of fluorosis reported in different studies. There is a need to standardise the methods used to score dental fluorosis.