Social Audit Ltd
P O Box 111 London NW1 8XG
Telephone/Fax 44 (0)171 586 7771

[email protected]


Professor Alasdair M Breckenridge
Department of Pharmacology & Therapeutics
University of Liverpool, Box 147
LIVERPOOL L69 3BX 28 April 1998

Dear Professor Breckenridge,


I am writing to you as a senior member of the CSM and a specialist on adverse drug reactions (ADRs), though mainly in connection with the paper from your Department, recently published in the BMJ (25 April 1998, 316, 1295-1298). I appreciate that Dr Pirmohamed is identified as principal author and correspondent, and I shall write him a note to explain. I hope he will understand that my concern is not so much about the paper, as about its public policy implications.

This will also partly explain why this letter will shortly be posted on the Internet site where we now publish our work. I hope you can feel reasonably relaxed about this, if only on the grounds that it gives you much more control over how you are represented than any conventional press/media outlet might allow. True, it also intended as a gentle blow for open government, peer review and consumer participation, but I would not expect you to object to that.

The policy implications of your Department's paper seem to me most important, since there is now something of a basis for an official estimate of the extent of the problem of drug injury. Previously there was none. Interestingly, I wrote to the Secretary for Health about this late last year, and again only a couple of weeks ago - though, I have yet to receive a response. I am enclosing copies of this correspondence (which can also be accessed on our website), also a briefing paper which summarises in broad terms what I would think the magnitude of the problem might be. You will see that the main figures I have used do broadly accord with yours - not that I find them very satisfactory, even if they seem about as close to the mark as one can get.

I drafted that paper, when preparing for a debate with your CSM colleague, Professor Rawlins, a few weeks ago. I did this partly because, the last time I discussed this issue with him (on a BBC Radio 4 interview in 1992) he emphasised he could not accept the figures I used; (these were based mainly on the review of 40-odd studies by D'Arcy (1986), rather than on the one study to which Rawlins referred). Though he doubted whether, at that time, anyone had even begun to measure the extent of the problem in the UK, he clearly thought the estimate of 5% of hospital admissions attributable to ADRs was too high.

"No, I think you've slightly misinterpreted the evidence, to be honest. It is true that the one major published survey in the United Kingdom suggests that 5% of patients who are admitted to hospital were admitted as a result of taking drugs. That was a study conducted in 1965, in Belfast, by Natalie Hurwitz and Owen Wade (BMJ, 1 March 1969, 539-540). That study showed that half those patients actually were suffering from self-poisoning - or almost half. The remainder were admitted to four specialised wards: one surgical, one medical, one dermato-logical (which is actually rather important because it was a dermatology ward taking in patients from virtually the whole of Northern Ireland, so you would tend to get a very biased view of the numbers there), and a psychiatric ward. You can't extrapolate that data from 1965, from that relatively small cross-section of hospital patients, to what is going on today"

Against this background, I would be grateful if you could let me know a bit more about the basis on which your own estimates were prepared.

1. The main estimate provided - that ADRs account for about 5% of all hospital admissions) is cited to four US studies. Can I take it from this that you were unable to find any convincing evidence relating to UK experience?

2. As your estimate is cited only to US studies, on what basis was it established that the situation in the US would be directly comparable to that in the UK?

3. Your estimate relates numbers of ADRs to all hospital admissions. I wondered about this, partly in the light of Mike Rawlins' comments, also because it would be helpful to identify which areas of medicine (and categories of patients) were more or less at risk. Could you please clarify this point by indicating the base number of hospital admissions/year you would rely on?

4. The paper estimates that ADRs affect 10-20% of all in-patients (ie regardless of cause of admission), but no evidence is given for it, nor is there any indication of severity. I would be interested to know how this figure was arrived at; it seems much higher than most estimates from the US.

5. The US studies I've looked at seem to agree that perhaps half of all ADRs resulted from avoidable oversights by prescribers. Would you think it reasonable to suppose the situation was much the same in the UK?

6. I wondered why the paper did not mention the large majority of ADRs that do not lead to hospitalisation or happen in hospital. Clearly many patients would be affected, and some of these ADRs are serious, even if most are not. Even if only 1% of all prescriptions resulted in a significant ADR, several million would be affected and the burden on doctors, and cost to the NHS, would be very high.

7. Your paper states that ADRs cause deaths in 0.1% of medical and 0.01% of surgical inpatients - but no evidence is given for these estimates and they are much lower than those from the recently published meta-analysis of US studies of ADRs in hospitalised patients (Lazarou et al., 1998). This study suggested that fatal ADRs amounted to 0.32% of all hospital admissions, ie around 100,00 deaths per year, "making these reactions between the fourth and six leading cause of death." in US hospitals. On what basis did you and your colleagues decide that US and UK experience differed so markedly in this respect, and on what UK data was your estimate based?

8. If one takes as a base just the 6 million/year or so "ordinary admissions" for "general and acute episodes" (ie excluding another 4 million day cases, maternity, psychiatric, learning disability and 'well baby' episodes), your data imply that, each year, at least 300,000 people end up in hospital because of ADRs, and that perhaps one million more experience ADRs while there. Among other things, these data imply:

[a] the numbers of serious and fatal ADRs reported under the Yellow Card scheme would be much closer to 1% that the "less than 10%" figure you mention, and cast further doubt on the CSM's view (1995) that "the proportion of doctors who have ever sent in a yellow card is encouragingly high."

[b] some lack of commitment to get to grips with the problem, given the cost of ADRs to the NHS/taxpayer would be reckoned in terms of billions/year.

9. Under the heading "Strategy for Drug Safety", your paper makes a number of recommendations relating to clinical procedures and the need for vigilance and so on. But given the apparently overwhelming lack of UK data, and the colossal size of the problem, I was very surprised to see no recommendation to do some research to actually find out what's going on. Wouldn't that be the place to start?

I would appreciate any comments you and your colleagues might have on these points, and shall look forward to hearing from you.

Yours sincerely,
Charles Medawar

Reminder sent on 6 July; no acknowledgement/reply was received. SA therefore wrote on 17 July to Dr M Pirmohamed, a member of the CSM's Sub-Committee on Pharmacovigilance (with copies and covering notes both to the Editor of the BMJ and to Prof. Breckenridge).


Contents page
List of MCA/CSM Correspondence