Consequences for research if use of anonymised patient data breaches confidentiality

Editor—During this century well conducted clinical and epidemiological research has formed the basis of remarkable advances in medical knowledge. The benefits provided to countless patients by these advances have greatly outweighed the possible risks entailed in clinical studies. The success of these endeavours has been well safeguarded by established laws, directives, regulations, and guidelines. Research using patient records has provided important information on factors predisposing to disease and successful outcomes.

A recent legal case—Source Informatics (now owned by IMS Health) v Department of Health, 28 May 1999—now threatens much healthcare research by suggesting that the use of patient data that have been subjected to procedures ensuring the anonymity of individual people may, nevertheless, constitute a breach of confidentiality. The consequences of this judgment are potentially so far reaching that the development of new treatments, particularly new drugs, may be seriously handicapped in the future. Studies that would be affected include individual case reports, retrospective archival research, the use of tissue samples, some case-control studies, observational cohort studies, prescription based investigations, and studies using computerised databases of medical records.

The unique value of this research is that it can provide information that is inaccessible to randomised, controlled clinical trials, which require ethical approval and informed patient consent because they are prospective and experimental. These requirements greatly reduce the inclusion of young children, pregnant women, very old and very sick people, and those unable to give informed consent. However, medical practice includes a high proportion of such patients who are underrepresented or excluded from clinical trials. Furthermore, as has been reported by Rawlins and Jeffreys,¹ considerations of feasibility and cost often limit the numbers of patients exposed to a drug to a maximum of a few thousand for comparatively short times. Computerised databases in primary care can extend times to many years of continuous care and the numbers of patients to millions; this would be impossible to do in any other way.

The judge in the recent legal case recognised that, in certain circumstances, the benefits derived from anonymised patient data might outweigh any problems of confidentiality but also raised the concept that this would not be an acceptable exemption if the purpose(s) for which the data were intended were to be classified as “commercial.” Uncertainties about the precise meaning of this term raise further doubts about using anonymised patient data in research, particularly as some of the studies we envisage might be regarded as market research into the effective use of new drugs.

We strongly support the acceptability of using anonymised patient data for healthcare research without this being regarded as a breach of confidentiality. We agree that ethical scrutiny of specific study plans is generally desirable but that informed consent is both unnecessary and often impractical. The purpose(s) for which a study is conducted should be clearly stated but would not be disallowed on the grounds that certain aspects could be described as commercial or relating to commerce.

We recommend that written guidelines, such as those proposed by a working group of the Royal College of Physicians,² be used to clarify the current uncertain situation. Any measures that would restrict non-interventional, observational research on unidentifiable patients by the setting up of unnecessary and impractical requirements will inevitably have deleterious effects on progress in health care.