Statistical analysis
After the SBRs of all sampling sites were calculated using the method
mentioned in the previous study8, a receiver operating
characteristic (ROC) curve for SBRs of fluorescence colposcopy was drawn
by a custom-written Matlab (MathWorks, the United States) code, and the
cut-off value was set at the maximum of Euclidean’s index. SBR greater
than the cut-off value was considered a positive examination result.
When analyzing on the per-patient basis, we took the sampling site with
the highest SBR value for each patient as the measurement index.
The diagnostic efficiency
according to the patient and the sampling point respectively were
calculate by comparing the results of the colposcopy with the
pathological diagnosis. The indicators of diagnostic efficiency included
accuracy, sensitivity, specificity,
positive predictive value (PPV),
negative predictive value (NPV)
and associated 95% confidence intervals (CI). To compare the diagnostic
efficiency of fluorescence colposcopy and conventional colposcopy,
chi-square test or Fisher’s exact test were performed by IBM SPSS
statistics 22. Differences were considered to be statistically
significant at P<0.05.