Rectal bleeding in general and hospital practice; ‘the tip of the iceberg’

Abstract

To determine the incidence of rectal bleeding and its predictive value for cancer in the general population, patient consultation rates to GPs, referral rates to hospital and the nature of rectal bleeding from benign causes. The study was based on a postal questionnaire in the setting of three General Practices in Portsmouth and South-east Hampshire (two in Portsmouth, one in Petersfield). There were 6000 subjects, over the age of 16 years, randomly selected from the three General Practices. The main outcome measures were incidence of rectal bleeding with respect to age and sex, the number who sought medical advice from their GPs and the numbers referred on to hospital. The frequency of dark red bleeding, of an associated change in bowel habit and the presence or absence of anal symptoms were also determined. The prevalence of colorectal cancer was determined in subjects reporting rectal bleeding in the 5-year period after the postal survey. Six thousand questionnaires were sent and 4006 (67%) were returned. Of the respondents 18% (709/4006) reported rectal bleeding in the previous year. Four percent (175/4006) had had bleeding for the first time and 38% (1522/4006) had bled at some time in the past. Of the 709 individuals, 512 (72%) had not consulted their GPs. Of the 197 (28%) who did, 50 (1:4, 7%) were referred to hospital. Eighteen percent (125/709) described the blood as dark red, and 88% (621/709) had anal symptoms. In 56% (394/709) the rectal bleeding was associated with a change in bowel habit mostly to hard stools (n=208), fewer to loose stools (n=98). Rectal cancer was diagnosed in only one patient 4 years after the first episode of rectal bleeding, giving a predictive value of rectal bleeding for colorectal cancer in the community of 1:709. Many people in the community have painless rectal bleeding, an associated change in bowel habit and dark red bleeding, factors previously thought to indicate a high risk of colorectal cancer. Rectal bleeding has however a low (1:709) predictive value for cancer. As the result of a considerable selection process, only a small number are investigated in hospital. These patients therefore represent a small proportion of all patients with this symptom. The value of the factors used to decide which patients need investigation should be re-assessed.