Surgical vs general practitioner assessment: diagnostic accuracy in 2-week-wait colorectal cancer referrals.

Aim  It has been recommended that patients with suspected colorectal cancer should proceed straight to an endoscopic test to increase speed of diagnosis, using only the information in the general practitioner’s referral letter. This study aims to establish whether the diagnostic accuracy of the first surgical outpatient assessment is significantly greater than the general practitioner’s assessment and if so by what means. Method  Demographic variables, symptoms and signs were collected from the first surgical outpatient assessment letters and the general practitioners’ referral letters in 2-week-wait colorectal cancer referrals made between 2002 and 2005. Multiple logistic regression models derived from both the surgeons’ and the general practitioners’ letters were compared with receiver operator characteristic curves. Results  Variables were collected from 978 2-week-wait colorectal cancer referrals. The median age was 69 years (range 19–98) and the male to female ratio was 1:2. Seventy-eight referrals were diagnosed with colorectal cancer. Surgeons’ models demonstrated significantly greater diagnostic accuracy than general practitioners’ models (area under the curve, 0.84 vs 0.73; P < 0.003). General practitioners’ letters contained significantly less information than surgeons’ letters (P < 0.001), but correcting for this did not account for the difference in diagnostic accuracy. The single variable that accounted for the difference in diagnostic accuracy was examination of the rectum by rigid sigmoidoscopy. Conclusion  Rigid sigmoidoscopy significantly improves the diagnostic accuracy of clinical assessment in patients with suspected colorectal cancer. If rigid sigmoidoscopy were omitted in a straight-to-test pathway, some patients would be denied the opportunity for immediate diagnosis.