The National Audit Office report on Regulating Medicines in the UK (January 2003) suggested that the MHRA (then the MCA) "needs to do more to communicate directly with health professionals and the public and improve the effectiveness of its safety messages". The Agency naturally responded as if the underlying problem was to do with image rather than substance.

The upshot was appointment of Simon Gregor, who joined the MHRA in February, as head of its new communications division. In his first public interview, Gregor was asked if he thought the Agency itself, and "not just its PR, needs an overhaul?"

"No, he says, arguing 'the agency needs to communicate more proactively' to engage people and promote understanding: 'Genuinely enthused, he suggests that the way to achieve this is to generate 'dialogue' between the public and scientists on drugs' benefits and risks."

One way and another, the MHRA has been banging on a lot recently about the need to see drug safety issues in content, and always to balance benefit and risk. But the evidence to date clearly shows that they haven’t thought it through. The CSM/MHRA announcement (3 February 2005) about the risk of liver disorders with atomoxetine (Strattera) nicely illustrates the point:

This warning says: "The risk is estimated at below 1 in 50,000 patients treated". This figure seems to have been produced by dividing the total number of patient exposures worldwide (2.3m) by the total number of reports hepatic disorders found worldwide (41 cases). Regulatory estimates made on this basis have previously produced misery and chaos for countless tranquilliser and antidepressant users. This method of calculation flies in the face of common sense and recommended practice (e.g. EMEA/CPMP/2775/99): the presumption that all cases have been reported is misleading and absurd.

The second problem, much harder to deal with, comes to the CSM/MHRA advice that atomoxetine is "an important drug" and "an effective treatment". Such advice underlines the drug-centred, rather than user-centred, perspective that blights so much regulatory communication.

The question Simon Gregor now needs to put to his colleagues is this: what proportion of drug users would actually find atomoxetine to be an effective treatment? I suspect the answer would be of the order of one or two patients in every ten. If the MHRA is intent on effective communication, it needs to avoid making sweeping generalisations, and to explain things in ways that so far as possible predict outcomes for individual drug users. The drawing board would seem a fair place to start.