Only Members of Parliament can table PQs, but anyone can propose them – and here’s an opportunity to test their effectiveness in measuring the quality of the drug regulatory system, especially in relation to the safety and effectiveness of SSRI antidepressants. Social Audit has set up a Bulletin Board, annexed to this website, as a forum for framing the kinds of questions that legislators might be asking about the adequacy of drug control.

From time to time, we shall add to this thread on this discussion board, but the questions posted so far are about:

Speed of scrutiny is a cornerstone of the Agency’s own development strategy. The guarantee of a quick turnaround puts the MHRA at a competitive advantage with other regulatory bodies - all of them now largely or wholly dependent on industry fees. Drug regulatory agencies in different countries operate in a market of their own: they too need to attract business to survive.

The MHRA’s pride in speedy drug licensing is reflected in many of the agency’s targets and performance indicators, and has reportedly been achieved "without reducing quality of the approval service" (NAO, 2003). However, little evidence has been given to support this contention, and there are reasons to doubt it (Medicines out of Control? pp 148 - 152).

Consider what is now happening with the MHRA/CSM review on SSRI antidepressants. The review was started in May 2003, and apparently the publication date has just been postponed again. In January 2004, the completion date was said to be May/June, but now it’s been put back to September. In other words, the SSRI review will have taken, overall, about 10 times as many working days as it takes to screen a new drug licence application. It takes the MHRA, on average, around 35 working days to assess a new drug application – well within the EU target of 55 – 70 working days.

This is another example of the way in which the antidepressant crisis is pointing to flaws in the whole basis of drug regulation. Two obvious questions arise here, both dependent on the answer to another question: how many working days (as opposed to elapsed time) have been spent so far on pursuing this review? That is a question for Parliament: See: Proto-PQs.

In the absence of that information, one can only speculate. My guess is that the public would be wrong to infer that there isn’t a lot of hard and important work now going on behind the scenes. Very probably, the answers would have come faster, if the Pharmas had been pressing for speed – but a key lesson to be learned from this regulatory fiasco is that speed kills.

This review is taking time because the regulators are at last doing what they should have done a long time ago. What is happening is that the regulators are now re-examining data that they failed to scrutinise properly, years ago. When licensing drugs, the regulators have routinely relied on the applicant company’s helpful summaries of the raw data – but not the data themselves. The point is developed in Medicines out of Control? (e.g. pp.141 - 142).

Probably the main lesson to be learned is not that this review is taking unduly long, but that the license approval process is done at breakneck speed. Under the present fee system, that is inevitable. Companies pay a fee of up to £80,000 for a new drug application, a minuscule amount in relation to blockbuster sales, or the lifetime cost of monitoring a drug’s safety and efficacy. The present fee represents the kind of sum that a successful sales representative might earn in a year.

On Social Audit's new Bulletin Board, we start to develop a 'CoG Taxonomy', a schedule of components, drivers and points of systems failure. The aim is to identify better ways of doing things and to understand more clearly how much could be gained by finding solutions that really worked.

This 'CoG Taxonomy' is constructed around four main actors: government, commerce, professionals and people (users, consumers, patients, taxpayers, 'you' and 'me').

Meanwhile, the official warnings on SSRI use in pregnancy (in the Label/SPC) seem, for the most part, studiedly unhelpful. Thus, the warnings for citalopram, paroxetine and sertraline say only that, "the safety of xxx during human pregnancy has not been established. As with all drugs xxx should only be used in pregnancy if the potential benefits of treatment to the mother outweigh the possible risks to the developing foetus." The venlafaxine warning says much the same, but adds: "If venlafaxine is used until or shortly before birth, discontinuation effects in the newborn should be considered."

The paucity of these warnings again underlines the regulators’ inability to grapple with and communicate uncertainties. Instead they take refuge in something approximating to the NERO defence – ‘no evidence of risk equals evidence of no risk". The advice to weigh benefits and risks is self-evident, yet the official warnings give precious little basis for making such decisions.

Here, once again, the regulators demonstrate an incapacity to see reason – failing to see risk in the absence of firm evidence of harm. The existing warnings in no way reflect the risks of neonatal withdrawal suggested in many case reports – and that seems all the more astonishing because of the evidence of such problems with fluoxetine.

Because of its long half life, neonatal withdrawal problems with fluoxetine are likely to be much less marked than with the other drugs – yet fluoxetine is the only SSRI that carries a credible warning, simply because better quality evidence is available. It would clearly be safer to assume that neonatal withdrawal problems with all SSRIs and related drugs would be at least as serious as this:

The need for proper warnings with all such drugs has been underlined by the recent report on The Reproductive And Developmental Toxicity Of Fluoxetine from an expert panel of the US National Toxicology Program and Center for the Evaluation of Risks to Human Reproduction (NTP-CERHR, 2004). The overall conclusions of this study included the following:

The NTP-CERHR Expert Panel also identified a range of Critical Data Needs:

"… studies in humans were generally limited in statistical power by small sample sizes and were not designed or reported in a manner that would allow a clear distinction between the effects of the underlying disease and the effects of the medication. Data were generally not available to permit a comparison of the pregnancy outcome effects of medication with the effects of nonmedication therapies, e.g., cognitive behavioral or interpersonal therapies, in pregnant women. Further, information was generally lacking on criteria for diagnosis of depression and on severity of disease. In addition, confounding factors such as smoking, alcohol consumption, use of other medications including dietary supplements, age, prior reproductive history, and comorbid illnesses often were not adequately reported or controlled. Future studies should take these factors into consideration, because such a design would permit longitudinal ascertainment of exposure data and other relevant covariates. Additional and better comparisons of fluoxetine effects with effects of other SRIs are needed."