Intra- and inter-examiner agreement when assessing radiographic implant bone levels: Differences related to brightness, accuracy, participant demographics and implant characteristics

ObjectivesEvaluate intra- and inter-examiner agreement of radiographic marginal bone level (MBL) assessment around Branemark single implants; and whether agreement related to radiograph brightness, discrimination level (accuracy), participant demographics or implant characteristics. Materials and MethodsSeventy-four participants assessed MBLs of 100 digital radiographs twice with normal brightness, and twice with increased brightness. Intra-examiner agreement with and without increased brightness to the same thread, and within one thread; and inter-examiner agreement as compared with the group (defined by the mode) for the first assessments with and without increased brightness, to the same thread, and within one thread were calculated with Cohen's Kappa. Relationships between agreement, thread discrimination level (accuracy), brightness, participant and implant characteristics were explored. ResultsWhen assessing 100 Normal radiographs twice, a participant on average assessed 24% differently to themselves (poor intra-examiner agreement, median Kappa 0.58, range 0.21-0.82); and 28% differently to other participants (poor inter-examiner agreement, median Kappa 0.53, range 0.05-0.80). Agreement within examiners improved when radiographs were Bright (median Kappa 0.58 vs. 0.62, p<0.001, accuracy to same thread; median Kappa 0.94 vs. 0.96, p<0.001, accuracy within one thread). Agreement between examiners was neither better nor worse when radiographs were Bright (median Kappa 0.53 vs. 0.55, p=0.64, accuracy to same thread; median Kappa 0.93 vs. 0.93, p=0.23, accuracy within one thread). Intra- and inter-examiner agreements were lower when accuracy to the same thread was required (p<0.001, p<0.001). Neither intra- nor inter-examiner agreement related to age, time since graduation, specialty, viewing device, implant experience, external hex familiarity, periimplantitis treatment experience, implant location or width (p-values 0.05-0.999). Intra-examiner agreement increased across dental assistants (n=11), general dentists (n=16) and specialists (n=47) (Bright assessments, p=0.045, median Kappa's 0.55, 0.60, 0.65 respectively); and for females (n=8, males=58) (Normal assessments, p=0.019, median 0.68 vs. 0.55), but female numbers were low. ConclusionsAgreement within and between examiners when assessing MBLs was poor. Disagreement occurred around 25% of the time, potentially affecting consistent disease assessments. No participant or implant characteristic clearly affected agreement. Brighter radiographs improved intra-examiner agreement. Overall, perceived MBL changes below 1mm are likely due to human, not biological variation.