Variations in the evaluation of colorectal cancer risk

To test the variability in estimating cancer risk and demonstrate the consequences that subjectivity has on patient care.

Forty-three clinicians were each asked to assess 40 symptomatic colorectal referrals. Each clinician was provided with a comprehensive history on the 40 patients. The clinicians graded the referral according to a malignancy risk score, decided on the required first line investigation and the priority of that investigation. The main outcome measures used was accuracy in cancer detection and appropriateness of investigations selected.

There was a wide degree of variation among all clinicians grading both benign and malignant disease with the overall correct classification of 54% (P-value of <0.001). On average, the clinicians correctly diagnosed 71.3% of the cancer patients as compared to 44% of the benign patients. Of the cancer patients, 47% were correctly classified as an urgent referral whilst 52% of the benign patients were over classified and graded as an urgent referral. The mean number chosen by clinicians to have a flexible sigmoidoscopy as the appropriate first investigation was 13 (of 40 patients); this was despite the diagnosis being possible in all cases with a flexible sigmoidoscopy. The choice to use full colonic investigation was seen throughout all disciplines. Junior doctors demonstrated the highest tendency choosing full colonic investigation in 92.3%. Consultants and senior grades showed the least tendency to choose full colonic imaging although even here colonoscopy or barium enema represented 48.5%.

Subjective assessment of cancer referrals is a significant problem that needs to be confronted. Improvements are needed to resolve the inherent problems of subjectivity and operator bias if uniform quality of patient care and best use of resources is to be achieved.