Estimation of the optimal cut off point in a new immunological faecal occult blood test in a corporate colorectal cancer screening programme.

Abstract

To estimate the optimal cut off point in a new immunological method (OC-Hemodia) for faecal occult blood testing (FOBT). A corporate colorectal cancer screening programme in Japan. METHOD - The screening programme targeted colorectal cancer and adenomatous polyps > or = 10 mm, and was conducted on 27 860 participants (age > or = 40 ) during 1991-92. The follow up consisted of diagnostic management by total colonoscopy on positive screened subjects exceeding the manufacturer recommended cut off level of 50 ng/ml faecal haemoglobin, and the identification of false negative cases by health insurance claims. The optimal cut off point was estimated by the positive predictive value, receiver operating characteristic (ROC) curve, and a cost effectiveness analysis. In this study evaluation was carried out only for cancer as the target disease. - At the current cut off level of 50 ng/ml the sensitivity and specificity were 86.5% and 94.9%. When the optimal cut off point was estimated the highest positive predictive value was obtained at 250-350 ng/ml. The ROC curve showed that the sum of sensitivity and specificity is maximised at 50 ng/ml, but evaluation of the ratio, change in sensitivity/change in false positive rate, pointed to higher optimal cut off points, showing marked changes occurring at about 200 ng/ml. The average cost per case was lowest at 250-300 ng/ml. Overall, the optimal cut off point was estimated to be about 200 ng/ml, at which the sensitivity and specificity of the test would be 77.5% and 98.9%, respectively. The optimal cut off point of the new immunological method of FOBT was estimated to be about 200 ng/ml, a value which, more than the current cut off value, favours specificity over sensitivity.