The use of symptoms to predict colorectal cancer site. Can we reduce the pressure on our endoscopy services?

OBJECTIVE

Our objective was to determine the value of gastrointestinal symptoms and signs in predicting the site of colorectal cancer (CRC). These symptoms can subsequently be used in determining first-line investigation with either sigmoidoscopy or colonoscopy.

METHOD: We interrogated the endoscopic and CRC databases ('Infoflex'), for patients diagnosed with CRC between April 2005 and March 2006 inclusive. These patients were cross-referenced with the pathology database and patient records. Information gathered from these databases include: age, gender, symptoms, site of cancer, histology, Duke's grading, blood parameters, diagnostic tool and treatment.

RESULTS

One hundred fifty-three patients were diagnosed with CRC between April 2005 and March 2006. One hundred twenty-six were initially seen in the out-patient department, of whom 38 (29%) were right-sided (proximal to the splenic flexure), and 88 (70%) were left-sided (splenic flexure and beyond). Change in bowel habit (diarrhoea and constipation) and rectal bleeding were significantly associated with left-sided cancers (P < 0.0024 and P < 0.0001, respectively). Haemoglobin (P < 0.0001) and mean corpuscular volume (P < 0.0001) were significantly lower in right-sided cancers. Weight loss, pain and obstruction were not associated with cancer site. C-reactive protein, albumin and carcinoembryonic antigen are not predictive of cancer site, Duke's stage or influenced by patient age or gender.

DISCUSSION: Symptoms can accurately predict site of cancer, allowing investigations to be tailored accordingly. We would recommend that patients with altered bowel habit and/or rectal bleeding, and no other symptoms, risk factors or anaemia, can be investigated with a flexible sigmoidoscopy to confirm or refute a diagnosis of colorectal cancer.