Accuracy and precision in the radiographic diagnosis of clinical instability in Brånemark dental implants.

The aim was to evaluate accuracy and precision in the radiographic diagnosis of clinical instability in Brånemark dental implants. In two clinics specialized in the Brånemark osseointegration technique we identified all patients in whom lack of clinical stability of one or more fixtures had been observed during a 5-year period. Radiographs of these fixtures (n=62) were mixed with radiographs of clinically stable fixtures (n=158). Eight observers were asked to determine whether or not a perifixtural radiolucency was seen by means of a 5-point rating scale expressing the confidence with which the presence or absence of a radiolucency was determined. Receiver operator characteristic analysis was used to calculate the A(z) values and the variability in A(z) values between and within observers was determined. In addition, we evaluated the agreement in rating codes within observers between repeat examinations. The A(z) values (mean=0.844 at the first reading and 0.856 at the second) indicated that the accuracy in radiographic diagnosis of clinical fixture instability was at least as good as that associated with other radiographic tasks such as approximal caries diagnosis and diagnosis of small periodontal bone lesions. Interobserver variability was larger than intraobserver variability. Intraobserver agreement in rating codes was relatively high and less than an average of 6% of all ratings differed with two rating codes or more. Despite the relatively good diagnostic accuracy, the probability of predicting clinical fixture instability from a radiographic examination can be low in populations with a low prevalence of fixtures showing clinical instability.