Understanding hospital referral rates: a user's guide

Abstract

Detailed referral information from one practice was used to investigate the effect of calculating referral rates in several different ways. Referral rates for individual general practitioners should be related to the number of consultations carried out and not to the number of registered patients; for whole practices list size may be used as the denominator. Most doctors will not need to control for age and sex of patients when comparing referral rates but may need to control for case mix when comparing referral rates to individual specialties. In addition, a method is described for distinguishing systematic variation between the referral rates of individual doctors from the random variation that may arise from data based on fairly small numbers of referrals. The method indicates whether systematic variation is greater than would be expected by chance, and it can be extended to indicate whether variability in referral rates is greater in one specialty than another. Because of random variation with time a year's data may not be sufficient to allow reliable interpretation of referral rates to individual specialties, except for the largest. At present there is no known relation between high or low referral rates and quality of care. Nevertheless, if doctors are to interpret their own rates of referral they need those rates to be reliable and valid. Use of the 10 guidelines described in this paper will help to prevent unwarranted conclusions being drawn from information on general practitioners' rates of referral to hospital.