The Government’s National Health Service reforms are designed to increase cost effectiveness, widen consumer choice and improve the quality of care. In making the case for these reforms, the White Paper Working for Patients (CM 555) pointed to the wide variations in performance throughout the health service. Variations in the average cost of treating acute hospital in-patients of up to 50 per cent between health authorities, two-fold variations in the medicine costs per head of population, and twenty fold variations in the rates at which general practitioners (GPs) refer patients to hospitals were all cited as evidence of inefficiency. It is the Government’s aim to reduce these variations and to raise the standards of all hospitals and GP practices to those of the best.

It is by no means clear, however, that the way in which the problem of variations is being approached will achieve this goal. By concentrating on high levels of activity (and cost)—whether these relate to prescribing, referrals or intensity of hospital service provision—there is a real danger that crude cost containment will displace the more relevant aim of cost effectiveness.

To ensure the efficient use of resources in health care it is necessary to establish the appropriate or optimal level of treatment. This can only be determined in relation to the costs of treatment, its clinical outcomes and the patient benefits associated with different outcomes. Unfortunately, there are few areas of clinical activity where knowledge of all these factors is precise or complete enough to form the basis of actual policy. Nonetheless, there is sufficient evidence to demonstrate that present levels of under-treatment (and the denial of potential health benefits to patients) are possibly as important as wasteful over-treatment.

In the light of these concerns, and in the belief that it was in this context that variations in doctors’ behaviour should be examined, a symposium was organised by the Office of Health Economics to consider factors influencing clinical decisions in general practice. This Briefing reports some of the main findings of this symposium. It is divided into three main sections. First, the evidence on variations in GP’s referral rates of patients to hospitals is examined. Second, variations in GP prescriptions for medicines are considered. And third, current policy responses towards these variations are discussed briefly with an assessment of their likely impact upon the cost and quality of care.