Quality and efficacy of a telecolposcopy programme.

We tested the performance of telecolposcopy in the diagnosis of various squamous intraepithelial lesions. There were three groups of physicians: Group 1 consisted of 15 gynaecologists who were not trained in colposcopy; Group 2 consisted of six experienced colposcopists (specialists); Group 3 was a supervising team consisting of three experts (colposcopists with more than 15 years of colposcopy experience). The first group took colpophotographs during gynaecological examinations and used a Web browser to compare them with examples from a database of colposcopy findings. Group 2 made their own diagnoses independently. Then the supervising team examined the findings from all 250 patients and made diagnoses that could be considered the 'gold standard'. The findings were identical for Groups 1 and 2 in 219 cases, i.e. an inter-observer agreement of 88%; Cohen's kappa was 0.81. The findings were identical for Groups 1 and 3 in 208 cases, i.e. an inter-observer agreement of 83%; Cohen's kappa was 0.74. The findings were identical for Groups 2 and 3 in 239 cases, i.e. an inter-observer agreement of 96%; Cohen's kappa was 0.93. There was high inter-observer agreement between all participants. However, there were also significant differences in some cases. In the cases of suspected invasive carcinoma, the results showed a significant difference (chi2=4.8, P< 0.005) because gynaecologists from Group 1 were not sure of their diagnosis and they wanted to obtain a second opinion by referring the decision to a higher level.