A methodology study on assessment of cancer screening test / 中国医学科学院学报

ABSTRACT

<p><b>OBJECTIVE</b>To analyze the effect of gold standard, blind comparison and different cut-points choosing on screening techniques assessment, and to promote the application of evidence-based medicine theory in screening study.</p><p><b>METHODS</b>A screening study for cervical cancer in rural China in 1999, where 1997 women had been tested for pathology as gold standard and simulating situations without gold standard, blind comparison and under different cut-points. Indices such as detectable rate, sensitivity and specificity were calculated for each technique. Receiver operating characteristics (ROC) curves were drawn and areas under ROC curves between screening techniques were tested.</p><p><b>RESULTS</b>Without gold standard, diagnostic techniques could not be evaluated correctly, and without the blind comparison, the sensitivity and specificity of the tests would be subjectively increased. Furthermore, use of different cut-points led to different sensitivities and specificities of test.</p><p><b>CONCLUSIONS</b>Gold standard, blind comparison and perfect cut-points can improve the quality of screening test and drawing ROC curves is an effective way to confirm cut-points and evaluate diagnostic techniques. It is necessary to enforce the application of evidence-based medicine theory in scientific research.</p>