Time for evidence-based screening?

The National Screening Committee (NSC) has published its first annual reportl. The Committee identified almost three hundred screening programmes in the UK, many of them at the research stage but nearly a hundred currently in practice. Only four programmes (breast and cervical screening, and neonatal bloodspot screening for phenylketonuria and hypothyroidism) met the Committee's stringent criteria for both quality and evidence of effectiveness. Furthermore the quality of the breast and cervical screening undertaken in some areas of the UK has recently been found to be inadequate, so the NSC may have been optimistic even in naming four programmes as meeting their criteria. Why has screening for so long been exempt from the critical scrutiny we require for other health care interventions, and what has stimulated interest in the evidence base of screening programmes? For many years, screening was thought of as harmless. The aim of prevention was attractive, compared with the traditional medical model of reacting to disease, and prevention of diseases seemed to offer a way of saving on expensive medical care. By theoretically reaching the whole population, screening also offered a way of reversing the inverse-care law (those most in need least provided for). The perceived popularity of disease prevention with the electorate influenced health policy decisions2. Lately several factors have heightened scepticism about the benefits of screening. First, there is a growing body of evidence that screening can harm individuals, particularly via the adverse effects of false-positive and false-negative results3. In a 1 0-year review of breast screening in the USA, one-third of women had abnormal test results requiring additional evaluation even though no breast cancer was present4. This raises ethical issues for clinicians: when a health intervention is initiated by the clinician rather than the patient, the clinician is under a greater obligation to ensure that the benefits outweigh the harm5. Secondly, far from reversing the inverse-care law, uptake of screening is often lowest among those most in need or with the poorest prognosis see, for example, the UK cervical screening programme6. Similarly, those who do attend may be in better health overall; in the New York mammography trial, women who attended for mammography had a much lower all-cause mortality than women who did not attend7. Thirdly, there has been a sharpening of the Government's mind over the issue of cost-effectiveness of screening in health care8. In addition, increasingly well-informed patients are a further pressure to ensure that policies are evidence-based. In a study of screening for prostatic carcinoma with prostate specific antigen, patients who were informed about the lack of research evidence regarding effectiveness of the screening test were far less interested in undergoing screening9. As well as undertaking an inventory of screening programmes, the NSC has established criteria for appraising the viability, effectiveness and appropriateness of screening programmes. These are essentially an updated version of Wilson's original criteria10 with the additional criterion that, before screening for a condition is initiated, there should be evidence from high-quality randomized controlled trials that the screening programme is effective in reducing mortality or morbidity. There are several reasons why we should welcome this declaration. Observational studies of screening programmes can lead to false conclusions about effectiveness. As already mentioned, patients who attend for screening are healthier overall than patients who do not attend. In addition, milder cases of the target disease will usually be slower to progress and will be in a presymptomatic stage for longer. Patients with mild disease are therefore more likely to be detected by screening, a phenomenon known as 'length-time bias'll. Both these factors mean that screening programmes will selectively recruit patients with a better prognosis. Comparison with a control group, and the inclusion of non-attenders in the analysis of the intervention group, is therefore needed before the results of the screening programme can be assessed. A control group also allows an assessment of underlying change in the prevalence of the target disease over time, which can otherwise be wrongly ascribed to a screening programme. Another reason why randomized trials of preventive programmes are desirable is that they allow a far more complete cost-effectiveness analysis than can usually be achieved with observational studies. Comparison with the control group allows accurate quantification of the additional benefits and costs produced by the programme. The inventory of screening programmes and the drawing up of effectiveness criteria were vital first steps for the NSC. Other work so far undertaken by the Committee includes: the development of a population 0