Examiner agreement in estimating changes in periodontal bone from conventional and subtraction radiographs

It has previously been shown that computer-assisted subtraction of serially-obtained radiographs yields a higher degree of accuracy in the detection of small changes in the marginal periodontal bone than conventional radiographs. This study concerns another important aspect of such diagnostic methods, namely the intra- and inter-examiner agreement rates. On 10 patients with moderate to severe periodontal bone loss involved in a clinical trial to evaluate the effectiveness of subgingival antimicrobial irrigation, conventional radiographs were taken at baseline and after 32 and 52 weeks. The changes in terms of gain or loss of bone occurring between the examination at baseline and that after 32 weeks, as well as those occurring between the examination performed after 32 weeks and that made after 52 weeks, were evaluated by means of conventional radiographs and by subtraction images. Both types of image were interpreted by each of 4 observers. The images were read a 2nd time after a period of at least 2 weeks. The inter-observer as well as the intra-observer agreement were assessed by calculating 3 different measures of agreement: the overall agreement, the specific agreement and the kappa value. Regardless of measure, higher inter-observer agreement rates were found for all observers when the subtraction technique was employed. A similar pattern was found with respect to the intra-observer agreement rates.