Skip navigation

Correspondence: Adverse drug reactions

We received our copies of Bandolier 28 recently and thought that a local perspective on adverse drug reactions might be of interest to you.

The prospective cohort study reported by Bates et al that you featured in that issue reiterated the message that adverse reactions to drugs are common, and identified 247 adverse drug "events" in a study population of 4031 admissions (6.1%). This compares with an Oxford study (Br J Clin Pharmacol in press), in which we collected adverse drug reaction data on 20,695 consecutive acute general medical admissions to seven general medical wards at the John Radcliffe Hospital between April 1990 and March 1993.

This was a spontaneous reporting scheme in which nurses, pharmacists, doctors, and medical students were invited to report adverse drug reactions occurring in patients during hospital admission. Deliberate or accidental overdoses were not included; nor were patients who were not admitted overnight to hospital. There were 1420 reports to the Oxford scheme, a rate of 6.9%.

In a parallel study in the John Radcliffe Hospital, 1071 acute medical admissions were systematically reviewed by an experienced adverse drug reaction pharmacist, and this study indicated that the spontaneous reporting scheme was effective in identifying 60% of all adverse drug reactions occurring on the general medical wards. This suggests that the true incidence of adverse drug reactions is nearer 10%.

In 25% of these cases the adverse drug reaction was the main cause of admission to hospital, and the remaining 75% arose after admission; i.e. on average each year 172 medical admissions to the medical wards at the John Radcliffe Hospital are due to adverse drug reactions, and a further 516 patients suffer an adverse reaction during their stay in hospital. In the 3 years covered by the study 2 patients died as a direct result of their adverse drug reactions.

The Boston group chose to use the term adverse drug event, which we believe causes confusion. This term is well established in another context, namely that of pre--marketing clinical trials, in which it is used to mean any adverse event that occurs during the trial; this is important, because the use of the term "event" rather than "reaction" avoids the need to make a judgement on causality at a time when little is known about the unwanted effects of the investigational drug. This use can also be extended to large post-marketing trials. For example, the increases in violent and accidental deaths seen in trials of lipid-lowering drugs were correctly classified as adverse events, and although they may have proved a lucrative field for those writing grants for studies on cholesterol and serotonin, in all likelihood they were not adverse reactions to the drug therapy.

However, the Boston group have extended the use of this term to other types of adverse drug reaction, including errors in prescribing, misuse or malfunction of infusion pumps, and potential adverse reactions (e.g. potential drug interactions). Of course, these events do not come within the definition offered by the WHO, as they and you correctly pointed out, but it is also confusing to label them as adverse events. They are often in fact toxic reactions arising from poor prescribing.

The Bandolier article incorrectly states that ADRs are "idiosyncratic and rare", whereas in fact, as we and many others have shown, they are very common and are usually dose-related. In our view the WHO definition is a poor one. Any event that is attributable to a drug should be called an adverse reaction; dose-related adverse reactions can then be subdivided according to whether they are toxic reactions or side effects.

JK Aronson and DJM Reynolds
Department of Clinical Pharmacology
Radcliffe Infirmary, Oxford OX2 6HE
previous story in this issue