Comparing the accuracy of Dutch dentists and dental students in the radiographic diagnosis of dentinal caries

Objectives: To compare the diagnostic accuracy of dental students using bitewing radiographs for the diagnosis of dentinal caries with that of general dental practitioners and to establish a bench mark for testing for improvement in diagnostic performance. Methods: Fourth year dental students (n = 259) viewed bitewing radiographs of 105 proximal tooth surfaces. According to a micro-radiography 'gold standard' 45 surfaces had dentine caries. The students were asked to diagnose presence and absence of dentine caries from which, under supervision, they calculated their sensitivity (TPR) and false positive rate (FPR). The students binary diagnostic data was compared with the historical data from a random sample of Dutch dentists (n = 273). The Summary Receiver Operating Characteristic (SROC) curve method was used to analyse the data. The two groups were tested for differences in (a) diagnostic odds ratio and (b) the area under the SROC curve. Results: The students mean TPR for dentine caries diagnosis was 67.2% (s.d. 11.8%) and the FPR was 8.3% (s.d. 8.4%). The corresponding values for dentists were: TPR 54.0% (s.d. 14.1%) and FPR 3.1% (s.d. 4.5%). Significantly more dentists (85%) than students (72%, P < 0.001) had a diagnostic odds ratio greater than or equal to 21. The area under the SROC curve (Az) was 0.8194 for students and 0.8140 for dentists (P = 0.70). Conclusions: In our study (i) the students had a higher TPR but also a higher FPR than the dental practitioners, (ii) their overall performance (Az) was similar but (iii) for a clinically relevant population the dental practitioners would out-perform the students in diagnostic performance.