Bleeding oesophageal varices: referral bias and survival

The survival of subjects with varices and gastrointestinal bleeding transferred to a referral centre from outside hospitals was compared with the survival of those admitted directly. The clinical features of the two groups were very similar and survival was 72 per cent. This suggests that the bias introduced by including such external referrals in series from regional centres studying survival in these patients is small and that they provide information relevant to patient management in smaller units. Of the referrals 79 per cent were from within 30 miles, transfer occurred rapidly and early mortality was low (7 per cent by one week). These factors help to minimise bias.