Editor's Choice | This Week in BMJ | Press releases

BMJ No 7105 Volume 315 Editorial Saturday 16 August 1997 Developing high quality clinical databases The key to a new research paradigm Although it is almost 20 years since Arnold Relman heralded the dawning of the "era of assessment and accountability,"(1) we still cling largely to a traditional research paradigm based on ad hoc studies. We do this despite well recognised limitations: such studies are expensive, and so only a limited range of interventions have been, and can ever be, investigated; the rapidity with which health technologies evolve means study results may be of no practical value by the time they are obtained; the results are often of uncertain generalisability as they tend to be carried out in atypical settings; the participating clinicians often have little sense of ownership; and small samples restrict the scope for subgroup analyses and thus the practical value of the results. High quality clinical databases offer an alternative approach, with the potential to bring research closer to practice and audit. The advantages include wide ownership and high generalisability through the participation of many clinicians; relatively low cost for each study, as the expense of data collection is spread over a range of research, audit, and administrative uses; the ability to generate large samples rapidly; the opportunity to study rare conditions or interventions; and the provision of accurate information for clinical practice, audit, and administration. Databases can be used either for non-randomised analyses or to generate hypotheses and provide ready access to clinicians prepared to participate in randomised trials.(2) The benefits of developing high quality clinical databases have long been recognised. Almost 100 years ago, surgeons on both sides of the Atlantic called for routine documentation of clinical practice. In 1907 Ernest Groves, a Bristol surgeon, suggested standardising clinical terminology and measurement,(3) while in Boston Ernest Codman had the temerity to establish an End Results Hospital in which outcome one year after surgery was routinely assessed and recorded.(4) However, he failed to persuade his colleagues to adopt the same approach and, until recently, little progress had been made with this strategy. While lack of progress has partly been a consequence of a lack of interest on the part of clinicians, managers, and researchers, it has also reflected the demanding requirements for creating a high quality database. If the benefits are going to be realised, such databases must include individual data on all consecutive cases, use standard definitions of conditions and outcomes, ensure data are complete and accurate, and include data on all known patient characteristics that affect outcome. Anyone familiar with routine hospital information systems, designed for administrative purposes, will appreciate that they fail to meet these requirements. With developments in information technology, the call for new, carefully designed and managed clinical systems is no longer fanciful. Indeed, in Britain some groups of clinicians have already made considerable progress. Surgeons throughout Lothian, Scotland, established a system in the 1970s(5); obstetricians in North West Thames region did so in the 1980s(6); and haematologists, first in North West and Northern regions(7) and later in other regions, have established a leukaemia register, pioneered by Stephen Proctor in Newcastle, (page 388).(8) However, the most ambitious development has occurred in intensive care. Encouraged by its work validating a risk adjustment model (APACHE II), the British Intensive Care Society is establishing a national database for audit and research.(9) While other groups can learn much from the experiences of the pioneers in this area, it is likely that no single model for organising and managing high quality clinical databases will be appropriate in all circumstances. Databases for long term treatment will differ from those for one off treatments, and the requirements for common, low cost treatments will differ from those for rare, expensive interventions. Regardless of which approach is adopted, early experiences suggest clinical and scientific credibility are more likely if clinicians and health services researchers collaborate in the design, management, and use of the database. Apart from extending this approach to other clinical areas,(10) what other challenges lie ahead for the users of these databases? First, as with any non-randomised comparisons, an accurate method for risk adjustment is needed if meaningful results are to be obtained. While considerable progress has been made in some areas (such as intensive care), in other areas work has barely commenced. Second, the research and audit potential of clinical databases could be enhanced by linking to other databases. For example, the geographical and socioeconomic equity of services could be evaluated by linking to census data.(2) While such developments will enhance their value, the biggest challenge remains getting the databases established, a task that requires clinical knowledge, health services research skills, political nous, and much ingenuity. Nick Black Professor of health services research Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London WC1E 7HT References 1 Relman A S. Assessment and accountability. The third revolution in medical care. N Engl J Med 1988;319:1220-2. 2 Black N A. A regional computerised surgical audit project. Qual Assur Health Care 1990;2:263-70. 3 Groves E W H. Surgical statistics: a plea for a uniform registration of operation results. BMJ 1908;ii:1008-9. 4 Donabedian A. The end result hospital: Ernest Codman's contribution to quality assessment and beyond. Milbank Q 1989;67:233-56. 5 Gruer R, Gordon D S, Gunn A A, Ruckley C V. Audit of surgical audit. Lancet 1986;i:23-7. 6 Patterson C M, Chapple J C, Beard R W, Joffe M, Steer P J, Wright C S W. Evaluating the quality of maternity services-a discussion paper. Br J Obstet Gynaecol 1991;98:1073-8. 7 Charlton B G, Taylor P R A, Proctor S J. The PACE (population-adjusted clinical epidemiology) strategy: a new approach to multi-centred clinical research. Q J Med 1997;90:147-51 8 Interview with Stephen Proctor. BMJ 1997;315:388. 9 Rowan K M. Intensive Care Society has set up a centre for national audit [letter]. BMJ 1996;313:1007-8. 10 Sochart D H, Long A J, Porter M L. Joint responsibility: the need for a national arthroplasty register [editorial]. BMJ 1996;313:66-7. Home | Current issue | Past issues | Classified ads | Career Focus | Feedback Collections | About this site | About the BMJ | BMA | Medline