Intended for healthcare professionals


Using clinical databases in practice

BMJ 2003; 326 doi: (Published 04 January 2003) Cite this as: BMJ 2003;326:2

Individualised prediction of survival for patients with cancer may be possible

  1. Nick Black, professor of health services research (Nick.Black{at}
  1. London School of Hygiene and Tropical Medicine, London WC1E 7HT

    Information in practice p 29

    In the past decade clinical databases have become increasingly widely used in all industrialised countries. This has been accompanied by enhancements in their quality as a result of greater understanding of the requirements for scientific rigour and the availability of technology that can automate processes such as validity checking. Meanwhile recognition has been growing of the uses to which high quality clinical databases can be put—evaluative research, clinical audit, and managing services.1 A further but less widely recognised application is that of helping patients, together with their practitioners, to make informed decisions about their clinical management.

    An example of such an application is the use of a breast cancer database in Finland (p 29).2 The Finprog study uses data on about 2000 women followed up for 10 years to enable an individualised prediction of survival for a new patient by matching her disease profile to that of many previous patients with breast cancer whose outcome is known. The patient and her practitioner can obtain a survival curve for the entire available follow up period, not simply an estimate for a single point in time. Such a system could be applied to any clinical database that includes accurate information on those characteristics of patients that affect clinical outcome.

    Such a development could make a major contribution to the promotion of patient centred care and help make meaningful shared decision making a reality.3 The need for such decision support was recognised by the inquiry into paediatric cardiac surgery in Bristol, which noted the failure of staff to provide parents with accurate prognostic information.4 This was not because the information was withheld but because it wasn't available.

    The Finprog study illustrates the potential value of such an approach, but it also highlights three challenges that lie ahead. Patients and practitioners are going to require information that is up to date and reflects local clinical services. At present, users of the Finprog study obtain information on the outcomes for a cohort of women diagnosed and treated 10 years ago. But clinical care has moved on. With ongoing recruitment, databases would be able to provide more up to date information (at least for short term outcomes) reflecting current treatment outcomes. The second enhancement needed is the ability to provide data on the outcomes achieved by the healthcare providers a patient is attending, although inevitably the relatively small volume of patients treated in any one setting will limit the statistical confidence of any estimation of prognosis. The third challenge will be to show that this approach not only promotes patients' participation in making decisions but also leads to health benefits. 5 6

    The potential scope for using high quality clinical databases in this way is rapidly expanding with the growth in the availability of such databases. To encourage their use and enhance their quality, a web based directory of clinical databases ( has recently been developed.7 This directory is restricted to the United Kingdom, but similar websites could be created in other countries. When complete the directory will provide a description of all multicentre clinical databases that exist in the country and an independent assessment of the extent and quality of the data collected. The growing availability of software such as that developed in Finland is an exciting step forward in promoting the use of databases to inform and support clinical decisions that practitioners and patients face every day.


    • Competing interests NB leads the Directory of Clinical Databases (DoCDat) project


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    View Abstract