|Department of Pharmacology & Therapeutics|
|DR. M PIRMOHAMED||Ashton Street Medical School|
|Liverpool L69 3GE|
|03/11/98||Telephone: 0151 794|
|Facsimile: 0151 794 5540|
Dear Mr Medawar,
DRUG-INDUCED IATROGENIC ILLNESS
We apologise for not having written to you sooner following your letters and e-mails about drug-induced iatrogenic illness. We have had a tremendous amount of correspondence to cope with following the publication of the BMJ review. The purpose of the review was to focus attention on what we feel is an important cause of disease, and I think we have done that.
Before we go on to answer your specific questions, there are two points we would like to mention:
First, the views expressed in the article and in this letter are entirely our own views and do not necessarily represent the views of any institutions we represent, either individually or collectively.
Second, we are greatly concerned by your letter to Dr Richard Smith, Editor of the BMJ (dated 17 July), in which you suggested that there was "some pretty funny data" in our BMJ article on ADRs. The data presented in that article has all been published before in the medical literature (as outlined below). With regard to the peer review, we would also like to point out that this was one of the most extensively peer-reviewed papers that we have published. You might like to know that we received 3 pages of referees comments, and the whole process from submission of the first draft to acceptance took approximately 12 months (there were numerous drafts in between). The paper as originally submitted was almost 4000 words but had to be cut down to about 2000 words in keeping with BMJ policy on publication of reviews.
It is also important to note that this was a review article, i.e. it summarised what we feel was important in the field of adverse drug reactions at the time the paper was written. We did not include any original data nor did we calculate any incidences from published data. At the time of submission and acceptance, we were unaware of the paper by Lazarou et al which was published in JAMA one week after our review. This is the reason why the article was not cited in our review.
Regarding your questions, our answers are as follows:
1 . The studies cited were by authors in US institutions. However, the paper by Einarson (1993) is in fact a summary of studies published world-wide. It includes 4 studies from the UK, from NI (%ADR admissions 2.9%), various centres (10.5%), Southampton (4.0%) London (6.8-8.6%) and Carlisle (6.4%). The numbers studied varied and criteria for inclusion also varied amongst the different studies. However, the weighted meta-analytic estimate was 5.1% (95% Cl 4.4-5.8%) when all the studies were considered. I think most investigators would agree that 5% is a realistic figure. This has also been borne out by a pilot study conducted by a pharmacist at the Royal Liverpool University Hospital.
2. Please see comments above that our article was a review of published literature and thus no original data was included.
3. and 4. The data that you refer to was presented in a "box" where it is not possible to use citations. The studies where the figures were derived from were as follows:
Siedl et al 1965; Epidemiological studies of adverse drug reactions. Am J Public Health, 65, 1170.
Hurwitz and Wade, 1969; Intensive monitoring of adverse reactions to drugs. Br Med J., i, 356.
Gardner and Watson, 1970; Adverse drug reactions: a pharmacist-based monitoring system. Clin Pharmacol Ther, 2, 802.
These studies are rather old but unfortunately there have been no other recent studies (apart from those already cited in the paper) which have addressed the question of ADRs whilst an in-patient. This largely reflects the difficulty in detecting and diagnosing ADRs while in hospital (as reflected by the poor ADR reporting rates). Once computerised prescribing systems are introduced in the UK, it will hopefully make it easier to collect such data.
With regard to fatal reactions, the figures were taken from the following studies:
Shapiro et al; 1971; Fatal drug reactions among medical in-patients. JAMA, 216, 467.
Porter and Jick. 1977. Drug-related deaths among medical in patients. JAMA, 237, 879.
Armstrong et al, 1976. Fatal drug reactions in patients admitted to surgical services. Am J Surg. 132, 643.
The figures are lower than those reported in the recent meta-analysis by Lazarou et al (1998). I think figures reported from meta-analysis do have to be treated with some caution, largely because of the problem of publication bias. In other words, the figures are derived from studies which have been published, and other studies which may not have been published, perhaps because the data was not as striking, will not have been included in the meta-analysis, thus giving a falsely high death rate.
5. About 25% of ADRs are idiosyncratic, and by definition, these cannot be predicted and therefore cannot be prevented with the current state of knowledge. The other 75% are so-called dose-dependent predictable adverse drug reactions. It is unlikely that all of these latter type of ADRs are preventable, and thus the estimate that half of all ADRs are avoidable seems reasonable.
To finish off, I think one point that has not been made in many of the recent publications is that drugs are a victim of their own success. Clearly we are now in a position to treat many diseases which previously ran a fatal course. This applies particularly to infectious diseases. Thus, while ADRs may be the fourth commonest cause of death, this is partly related to the fact that other diseases are now treatable. Nevertheless, the ideal situation would be not to have iatrogenic disease in the list of causes of death at all, and this was our intention in trying to draw attention to the problem in our article.
We trust this answers your questions satisfactorily.
|M. Pirmohamed||A.M. Breckenridge|
|N.R. Kitteringham||B.K. Park|
CLICK HERE TO READ ON