According to the excellent Yellow Card scheme. Although, as acknowledged, reporting is patchy, and the real figure is probably much higher than the estimate of 3,000 fatalities (plus a further 13,000 adverse reactions).
Drug reactions "kill thousands"
Anyway, I was particularly amused by the MHRA's response, from our good friend, Dr June "like the mountains I'm blue" Raine:
"Our role is to ensure that the benefits of medication outweigh the risks. It is important to note that a report of an adverse drug reaction does not necessarily prove that it was caused by the drug.
"Other factors such as underlying disease or other medicines may contribute to suspected adverse reactions."
Which conveniently misses the point. Drugs have acknowledged side effects, and to then suggest (without further investigation), that the patient may be at fault looks piss poor to me. Besides which, this approach obliges the patient to prove that the drug caused the side effect, which was acknowledged as being a risk of taking the drug.
Stop protecting the Worshipful Company, June: it's not big, and it's not clever. Another possibility is that the true incidence of a side effect has been masked, or its existence suppressed, altogether. Why don't we discuss that? By the way, June, you forgot to say something about "robustness," and "fact-basedness," which is important, I think, if you're going to quote from the Agency's* tagline:
We enhance and safeguard the health of the public by ensuring that medicines and medical devices work and are acceptably safe. No product is risk-free. Underpinning all our work lie robust and fact-based judgements to ensure that the benefits to patients and the public justify the risks.
Finally, June, you'd best make sure that the drugs you license have a benefit, against which to assess the risks, eh?
* The MHRA is four years old.