Intended for healthcare professionals

Education And Debate

Use of personal medical records for research purposes

BMJ 1994; 309 doi: (Published 26 November 1994) Cite this as: BMJ 1994;309:1422
  1. Nicholas Wald, professora,
  2. Malcolm Law, readera,
  3. Tom Meade, professora,
  4. George Miller, senior clinical scientista,
  5. Eva Alberman, professor emeritusa,
  6. John Dickinson, professor emeritusa
  1. a Wolfson Institute of Preventive Medicine, Medical College of St Bartholomew's Hospital, London EC1M 6BQ
  1. Correspondence to: Professor Wald.
  • Accepted 15 November 1994

The established practice of doctors using medical records for research purposes is threatened by the recent proposed guidelines from the Department of Health, the BMA, and the European Commission. The European Commission has proposed that explicit consent should be obtained from each patient before his or her medical records can be used; the proposals from the Department of Health and the BMA would require all research that needs access to personal medical records to be submitted to an ethics committee. We believe that these proposals would seriously impair an entire category of research and suggest therefore that another set of guidelines, proposed by a Royal College of Physicians' working group, should be used to modify the proposals. The guidelines of the working group encourage the use of medical records for research and ensure that such use can be made in a confidential manner without causing harm.

The use of medical records for research purposes is a longstanding practice by doctors of virtually all specialties. The value of such a practice has been immense. It is now under threat from proposed guidelines issued by the Department of Health,1 from a draft bill governing use and disclosure of personal health information produced by the BMA's Multidisciplinary Professional Working Group,2 and a draft data protection directive (SYN 287) from the European Commission. Safeguards to protect the confidentiality of financial and other personal details from unauthorised access by public authorities are being uncritically extended to cover medical records held by doctors.1 2 An entire category of research, which can identify, for example, the side effects of drugs, the effects of toxic substances in the environment, or the safety of medical procedures, would be seriously impaired.


The use of medical records for research purposes has usually taken two forms: systematic record review and what might be termed “record linkage.” Systematic record review is commonplace, carried out, for example, on a consecutive series of patients with the same diagnosis to identify the prevalence of characteristic clinical features, response to treatment, or factors influencing prognosis. Record linkage entails collating medical information from separate sources on individual patients identified by name and date of birth to seek, for example, an association between a drug and a disease. Probably every hazard that we know of has required access to medical records, often with the details of the exposure kept in one place and the details of outcome in another, requiring a formal link between the two. The associations between taking oestrogens in pregnancy and cancer,3 ionising radiation and leukaemia,*RF 4, 4a* and use of oral contraceptives and venous thrombosis5 were all documented in this way. Recently, the risk of limb reduction defects associated with chorionic villus sampling in early pregnancy was identified through linking records.6 Such work can also provide reassuring information on the safety of medical procedures—for example, confirmation that amniocentesis does not cause clubfoot.7 The comparison of the results of blood tests from pregnant women with and without specified congenitally malformed fetuses has led to the development of antenatal screening tests.8 9 For these purposes the personal identification of records is essential for collation, and the systematic review of medical records is impractical without sight of the name of the person to whom they relate.

The importance of personal medical records for research cannot be overestimated. Their use forms a regular part of medical practice, often leading to important discoveries. Such research often arises because an alert doctor with a hunch contacts colleagues. Examining different sets of records that relate to the same patients may clarify the position. Often no discovery results from the enquiry, but occasionally an important finding arises. Such activity is regularly carried out by clinicians, pathologists, epidemiologists, and anyone directly involved in preventing or treating disease who seeks to improve health or the quality of medical practice and is undertaken in confidence on the understanding that the patient will not be contacted. There is, therefore, no possibility of harm, and much opportunity for good.

Proposed guidelines on use of personal records

The European Commission has proposed in its draft directive that explicit patient consent should be obtained before each record can be used, a rule so stringent that this kind of research would probably stop altogether. Even the longitudinal survey carried out by the Office of Population Censuses and Surveys could be under threat from such legislation. Often tens of thousands of records have to be examined to look for clues, and it would be impractical and sometimes impossible to approach everyone. Not uncommonly, the individuals are unavailable or dead. Even if it were practical to approach each individual before a possible association were investigated, this may cause unnecessary anxiety as in most cases the research does not confirm a link between an exposure and disease.

Other guidelines, notably those recently proposed by the Department of Health1 and the BMA,2 are less draconian but are none the less unnecessarily restrictive. They require all applications for support for research using systematic record review or record linkage to be approved by an ethics committee. As in most cases such research may amount to little more than two or three doctors communicating about their patients, prior submission to an ethics committee would introduce an unreasonable barrier to what would otherwise be normal professional interchange. In many cases such exchanges would not even be contemplated if the doctors had to submit a formal application for permission to proceed, particularly when the hunch on which the proposed work is based is tenuous. This would be disastrous, because many important findings have arisen from initially fragile ideas. Indeed, we go further in saying that provided no one is harmed a positive professional duty exists to encourage such activity because doctors have an obligation to extend knowledge and improve medical practice by examination of their collective experience.

Threat to research

Two factors explain the threat to record based research. Firstly, the examination of medical information for research purposes by doctors has unwittingly been included in an all embracing legislation regulating the transfer of personal, financial, social, and criminal details. Secondly, the term “confidential” has been interpreted in an extreme and rigid manner. In practice, degrees of confidentiality exist. At the most restrictive level information given by a patient to a doctor can be divulged to no one else. This is so restrictive as to render the communication useless in all but the most exceptional cases. A doctor can rarely be fully effective acting as a sole agent; usually the information must be shared with colleagues for the patient to benefit— for example, when requesting special tests or arranging treatment. The exchange of information among colleagues can also lead to knowledge that may benefit other patients. In practice, the duty of confidentiality is interpreted as applying not only to the doctor directly involved in the care of the patient but also to those with whom he or she judges the information may be shared; there is, in effect, a professional duty of collective confidentiality. Provided that collective confidentiality is practised no harm will come to the patient and no reasonable objection can be made as it is only through a breach of confidentiality that any harm could arise.

Once the sense of sharing personal medical information among doctors— subject to a strict duty of collective confidentiality—is accepted, it becomes irrelevant whether the activity is for research, teaching, or care. The only relevant issue is to ensure that the collective confidentiality is secure and that no breach occurs that could in any way adversely affect the individuals concerned.

A working group of the Royal College of Physicians has published a report10 that sets out guidelines that encourage the use of medical records for research and ensure that this be done in a confidential manner without causing harm. The guidelines state:

Research involving access to medical records, registers, or existing biological samples only, without direct patient involvement, is not considered to require individual patient consent or independent ethical approval provided that:

  1. explicit consent to access a person's records is obtained either from the official custodian of the records or from the patient's clinician: the decision to access personal medical information should not be left to the sole discretion of the investigator.

  2. the recipient of the information is a senior professional person (eg a consultant medical practitioner or a principal in general practice) who may be disciplined by his or her professional body over any breach of confidentiality.

  3. confidentiality is assured through exercising professional ethical codes of conduct.

  4. anonymity is assured in any report or publication.

The proposals from the BMA and the Department of Health are open for discussion. We believe that they should be modified in line with the guidelines proposed by the Royal College of Physicians' report. This would permit a safe and well established type of medical research and would safeguard confidentiality. It would well serve the interests of patients and the public.