Yesterday the Medicines and Healthcare Products Regulatory Agency announced it had decided not to prosecute GlaxoSmithKline for offences under medicines legislation, although it says GSK failed to inform it promptly of data it had from clinical trials suggesting that the anti-depressant drug Seroxat (GSK’s UK brand name for the active substance paroxetine – marketed in the US as Paxil) is associated with increased suicidal behaviour in some depressed children. Here is the MHRA webpage with its press release and related documents.

At the time the MHRA found out about this data about seven thousand under-18s were being prescribed the drug – so GSK’s late reporting is bound at the very least to have caused some British teenagers to have suicidal thoughts. Who knows whether some young person died because of this. In fairness I should say that GSK deny they reported the data late. The MHRA claims that GSK knew as early as 1998 that Seroxat was ineffective in treating children’s depression. But it didn’t tell them. Amazingly, it planned in 2003 to apply for a licence specifically for children!

In my view, this whole affair represents a serious failure of the medicines regulatory system, and a number of hard questions need to be asked of the MHRA Chief Executive Kent Woods, and of health ministers. In particular:

Were the MHRA and its lawyers right to think that GSK was not, or may not have been, required to report adverse reactions from Seroxat used in trials outside the terms of its UK marketing authorisation?

Were they right to think Seroxat was not authorised for use in children?

Why were MHRA officials unable to question GSK staff?

Does the MHRA have the powers it needs to properly regulate pharmaceuticals industry?

This monster post is my attempt to explain why I’m not satisfied with the MHRA’s explanation for its decision, and why I think these questions need to be answered.

The key document explaining the decision not to prosecute is this MHRA report into its investigation and legal consideration of prosecution. In the remainder of this post I’ll refer to various paragraphs in that report.

Were the MHRA and its lawyers right to think GSK may not have been required to report adverse reactions in trials outside the terms of Seroxat’s license?

Para. 33 sets out the potential offences that were considered in this case. From 28 February 2002, Schedule 3, paragraph 10(d) of the Medicines for Human Use (Marketing Authorisations Etc.) Regulations 1994 it was an offence for a “qualified person” (a specialist that every drug firm must employ specifically to provide information to the regulator) to fail to provide to the MHRA any information relevant to the risk/benefit analysis of a medicine as required by the European Medicines Code, Directive 2001/83.

I’ve not posted a link to the Regulations because the version on the Statute Law Database is out of date; the MHRA document correctly explains the position as from 2002.

So what did the EU Code require? As para. 42 of the MHRA report says, article 103d of the Code says the qualified person is responsible for providing to national regulators any information relevant to evaluating the risks and benefits of a medicine.

I think that’s all very clear. The EU Medicines Code requires the QP to give the regulator any risk/benefit data; and the UK Regulations make it an offence for the QP to fail to fulfil that responsibility. So why did the MHRA think there was some doubt here about whether late reporting would amount to an offence?

Paragraph 49 of the report explains this. It explains that the requirement in article 103d of the Code to report information on post-authorisation safety studies only covers studies of the drug within its licensed use.

That is I presume because post-authorisation safety studies are defined in article 1.15 of the Code as studies carried out in accordance with the marketing authorisation of a drug. But in addition this view sees article 102 of the Code, which requires national regulators to maintain a pharmacovigilance system to ensure good regulatory decisions having regard to adverse reactions under normal conditions of use, as colouring what information a qualified person must provide. According to this argument, the normal conditions of use are those approved in the marketing authorisation.

I have to say, I have real problems with this legal view. Firstly, article 103d of the Code does not limit the qualified person’s responsibility to report simply to data gathered in “post-authorisation safety studies”. The qualified person must in any event report any other information that relates to risks and benefits. So the definition of “post-authorisation safety studies” is simply irrelevant.

Secondly, I don’t read article 102 as cutting down or limiting what the qualified person must report; it simply explains in very general terms what the purpose of pharmacovigilance is as far as regulators are concerned.

In any event, thirdly I think it is too much to read the phrase normal conditions of use in the Code as meaning use within the terms of a marketing authorisation. It helps when thinking about this to consider article 116, which gives the regulator power to revoke marketing authorisation, and article 117 which enables it to recall products and stop further supplies. Under article 116, revocation can happen where a drug is harmful under the normal conditions of use, is ineffective or where the risk/benefit balance is not positive under the normal conditions of use. Do the MHRA’s lawyers seriously think that the fact that a serious risk arises from a drug only when used outside the strict terms of its authorisation – say, if you dropped dead if you took those two paracetamols just a little less than fours hours after the previous two – means the MHRA could not withdraw authorisation? That’d be silly. If people did start dropping dead for getting their timings wrong, you can be sure the MHRA would revoke authorisation for paracetamol on the basis that the normal conditions of use, in the real world, are not necessarily in all respects strictly limited to the authorised use.

I must be fair, and refer you to article 117. That as I’ve said is about product recall and impounding products at the factory gates, something the regulator can do if the product is harmful under the normal conditions of use, is ineffective or where the risk/benefit balance is not positive under the authorised conditions of use. Now, admittedly, this drafting causes confusion since it seems to suggest that the phrase authorised conditions of use can be used interchangeably with the phrase normal conditions of use. There’s no easy answer to this, European legislation being drafted pretty slackly on the whole. It’d be misconceived to attempt to draw some fine distinction between articles 116 and 117 or even between the different grounds for withdrawal under article 117. I can’t rationalise away this drafting problem.

But equally, it doesn’t work to rationalise it away by concluding that revocation and recall can only happen on the basis of harmful authorised use. That would actually run counter to the purpose of the Code, as revealed by recital (2), which says its essential aim must be to safeguard public health – and EU legislation must of course be interpreted purposively, in line with its aim.

Let me put it more starkly. The MHRA report says use in children is outside the authorised use. Would the MHRA claim that, if those children who were prescribed the drug all killed themselves en masse, it as the regulator would have no power to revoke the authorisation? Of course it wouldn’t. The product would be harmful in the normal conditions of use even if that use was unauthorised.

That ends up being a long way of explaining why I think it’s clear that GSK’s qualified person was required by law to inform the MHRA of the results of its trials, both as regards efficacy and safety, and why any failure to do so was an offence. The view of the MHRA lawyers seems to me an excessively cautious one, implying a very feeble view indeed of the duties of drug firms and the powers of the regulator.

Were the MHRA right to think Seroxat was not authorised for use in children?

Let me begin by saying it seems to be widely believed among drug firms, regulators and health care professionals that a medicine is not licensed for use in children unless the terms of its license explicitly say so. And ministers have said in Parliament that Seroxat was never licensed for use in children. My view of this will seem, therefore, contrarian. But it seems to me that Seroxat probably was licensed for use in children anyway.

How do you know what a drug is, and is not, licensed for? Well, if we could see a document called a marketing authorisation making it clear Seroxat is only licensed for use in adults, that would settle things. But I’m not sure such a document actully exists, or is available online. What does exist is the MHRA’s published “UKPAR”, or UK Public Assessment Report, a document which summarises the scientific basis on which marketing authorisation has been granted. Here’s an UKPAR for generic paroxetine. In that document there is no mention of children except in the text of the SPC or summary of product characteristics, which is a document aimed primarily at doctors explaining the drug’s use and which must be approved by the MHRA. Here’s Seroxat’s SPC. You can see that sections 4.2 on posology, and 4.4 on special warnings, both say Seroxat should not be used in children. It seems to me that it’s on the basis of these statements and these statements alone that the MHRA believe Seroxat is not licensed for children.

Well, fair enough: those statements are pretty clear, and I’ll buy that argument. But what did the SPC say before the MHRA was told about the increased risk of suicidal thoughts? Paragraph 2 of the MHRA report tells us that. The SPC said (presumably in section 4.2 on posology):

The use of Seroxat in children is not recommended as safety and efficacy have not been established in this population.

Is that it? I have to say that that standard wording, to me, seems very different from a clear statement that use in children is not authorised. What’s more, paragraph 14 of the MHRA report tells us that, once the MHRA knew about the increased suicide risk, it asked GSK to submit a variation to the marketing authorisation contraindicating use in the under 18s. So aren’t I right to suggest the authorisation was not clearly limited in that way before?

And another thing: article 116 of the EU Code say that a marketing authorisation can be varied if the risk/benefit balance is not positive in the normal conditions of use. If Seroxat wasn’t licensed for use in children, as the MHRA are now saying, and if the normal conditions of use means simply licensed use, as the MHRA’s lawyers seem to have concluded – how come the MHRA had power to vary the authorisation? None of this adds up.

Why were MHRA enforcement investigators unable to question GSK staff?

Paragraph 23 of the MHRA report tells us that GSK staff, presumably including the qualified person,

declined invitations to attend interviews under caution

Amazing. I think I’d decline an invitation like that, too. But shouldn’t the MHRA have powers to require someone to attend for questioning (which would not be the same as making it compulsory to answer)? Shouldn’t they be able to apply for a warrant of some kind? Why didn’t they get the police involved? After January 2006 they had power to arrest GSK staff for questioning in relation to these offences. Here’s Alison Hannah’s critical explanation of the new powers.

Does the MHRA have the powers it needs to properly regulate pharmaceuticals industry?

This whole affair seems to me a serious failure of the medicines regulatory regime, a regime riddled with legal confusions, doubtful assumptions, excessive caution and misunderstanding. Does the MHRA have power even to vary drug licences in these circumstances? They seem not to think so. How can it be that the regulator thinks the law doesn’t require this safety data to be passed to it? Why does the MHRA seem so toothless in the face of the drugs industry? I’m glad legal changes are being made now, but are they really necessary, or is my analysis correct? Is the regulator being unduly cautious about its powers? Why have these “gaps” in the law, if indeed they are gaps, only been spotted now?

I’m not sure this regulatory system is fit for purpose.