I thank the committee for inviting me here today to engage in this dialogue.
Let me begin by saying that although pharmaceuticals are assessed for safety and efficacy at the pre-market stage, current pre-market evaluation is really recognized as incomplete. While the strength of randomized clinical trials, which is the primary evidence on which drugs are approved to be on the market, is the determination of short-term efficacy, safety is often unresolved within the limited controlled context of these pre-market randomized control trials. Their small size and short duration don't allow them to detect late-onset or less frequent adverse drug reactions. Patients with comorbid diseases are frequently excluded from such trials, and pregnant women and children. So pre-market randomized controlled trials really have limited generalizability. At the same time, these pre-market trials are essential to establish efficacy from an initial assessment perspective for safety before widespread population exposure. So we can think of pre-market randomized trials as necessary but woefully insufficient.
Canada's lack of systematic monitoring of marketed drugs means that adverse drug reactions are often not evident until years after a drug is on the market. As a result, many drugs with unacceptable harm-benefit ratios remain on the market for prolonged periods of time, thereby leaving Canadians exposed to unanticipated risks.
As representatives from the FDA noted in a publication--I believe it was The New England Journal of Medicine--the immense biological subtlety of human pharmaceuticals often cannot practically or adequately be detected in formal clinical studies. In the conclusion of that particular article they advised against taking a new medicine in the first two years in which it's marketed. I think that's a huge statement.
Researchers who have taken a look at the effect of adverse drug reactions have found that they are between the fourth and sixth leading cause of death in the U.S., contributing to more than 100,000 deaths and 1.5 million hospitalizations yearly. Since we have basically the same drugs in the Canadian market as in the U.S., we can extrapolate that in Canada the figure would be about 10,000 deaths annually due to adverse effects, or about 150,000 hospitalizations yearly. That means our current passive system of adverse event reporting--we do have a system in place, and it's after the fact--captures less than 5% of all of these adverse drug reactions. We're really only capturing the tip of the iceberg and not the whole picture.
The medicine digoxin offers a very good example. While the FDA received only 82 reports of adverse drug reactions annually, a post-market study that involved data-mining of hospital records found that it was associated with over 200,000 hospitalizations in seven years. So there were 82 reports from patients through their doctors, but 200,000 by taking a look at hospital records and seeing what hospitalizations were linked to that particular drug.
I find it very counterintuitive that just as a new drug enters the market and its use increases exponentially, its effects and patterns of use are no longer actively monitored. This is an oversight that could be ameliorated through a system of post-market surveillance.
I would like to share some insights with you from a comparative analysis that colleagues and I have done on how the U.S., Britain, France, New Zealand, Australia, Norway, and the European Medicines Agency approach pharmacosurveillance. I'm going to highlight the role that research networks can play, and how the knowledge they produce can be used by drug regulators to inform their regulatory decisions.
In these countries, two main approaches are used to look at the safety and efficacy of drugs after they're in the market. The first involves risk management plans. How these risk management plans work is that when a drug sponsor, a manufacturer, is about to have a product approved for the market, they negotiate the terms of the marketing and how they will assess risk on the market. The second involves developing a national network of research centres that can be commissioned or can submit proposals to conduct independent research. That could include observational studies that incorporate electronic health care databases.
Let me begin with the risk management plans. These currently are used by the European Medicines Agency, EMEA, in which the regulator requests drug sponsors to develop and implement a risk management plan to monitor each new drug's safety once it's on the market.
What we found is that the EMEA's risk management plans really lack transparency. What we can find about the plans is just a very brief summary on the Internet. More seriously, though, they don't involve the systematic approach to developing a study protocol. Instead, these protocols are developed on a case-by-case, ad hoc basis and they often don't entail any rigorous scientific method.
For example, they can involve developing a registry of patients who are taking a particular medication without any control group. The importance of having a control group is so you can monitor patients, but how will you know their level of adverse effects is higher than that of someone who is not taking it if you're just observing them? So you really need a control group, and that would be a far more rigorous approach. Alternatively, they can involve educational initiatives to physicians for patients, but as we all know, when industry is involved in promoting their medications, you don't necessarily get the full story on the risks and the benefits.
If we were to go that route in Canada, risk management plans should be informed by a process of risk assessment that precedes risk management, in which the magnitude of the harm is modelled by incorporating all pre-market data that's been generated and anticipating population exposure levels, so a far more rigorous approach. Unless Health Canada adopts a more systematic, rigorous approach to the risk management plans than the European Medicines Agency, including defined standards for research methods similar to the pre-market phase one, two, and three trials, and sets conditions regarding blinded assessments, alternate methods should be used to assess post-market safety and effectiveness. Such a study is developed and conducted by independent pharmaco-epidemiologic research centres.
In our study, we concluded the EMEA risk management plans are of limited value, even though they lend the impression of systematic surveillance. Just as a point of emphasis, they put something in place so the public is under the impression their drugs are being monitored, when often what is being done is not an effective approach to surveillance at all.
Reliance on drug company studies can also be imprudent, given the inherent conflict of interest, even when safeguards are incorporated. For example, in France, where they take a look at some of these risk management plans, they have an oversight committee to ensure industry adopts a scientific approach, but even those studies have weaknesses, because in the end, industry analyzes the data and interprets what they mean.
Getting industry to conduct these studies through the marketing process is an inadequate process. But at the same time, industry could still fund the research, as in the case of Italy, in which manufacturers contribute the equivalent of about 5% of their promotional budget to the Italian Medicines Agency to fund post-market research by university and clinical researchers. It would also allow studies of an entire drug class. And what I mean by that is often a pharmaceutical company will study a particular drug, and even though there are competing drugs on the market that address the same problem, they refuse to do a head-to-head comparison of one with another. So as a drug benefit plan, you're trying to decide which drugs to fund; as a physician you're trying to decide which drug to prescribe to your patient. If you don't have information on how the benefits of one drug compared to another are in an overall study framework, it's very difficult to know which one to choose.
Pharmaceutical companies have also been shown to report their research selectively, by either publishing studies with only positive results or finding a way to convey a positive outcome for those that have negative results. Recently, Turner et al. reported in 2008 that it would appear that 94% of trials for selective serotonin re-uptake inhibitors, SSRIs, were positive. In contrast, they conducted a separate analysis of the entire range of trial data submitted to the FDA and found that only half of those studies showed positive results. So again, industry can find a way to lend a positive approach even to studies that are negative.
These issues highlight the need for public oversight to ensure post-market studies address key research questions, that they are designed to produce valid results, and that they are reported accurately.
Minimizing study duplication is also likely to foster continued cooperation from doctors. A lot of these studies--these randomized controlled trials--involve working with physicians who prescribe the medication to their patients and then monitor how the patients are doing. If you're going to do separate little studies all over the place.... They found in France, where they're starting to do some of these observational studies, that physicians are starting to say they've done enough and that their patients are tired. The physicians just won't engage.
So if you're going to develop some kind of a strategy, you should use your resources, including physician time and patient time, in the best possible way.
The second approach to post-market surveillance that I would like to propose is an independent research network that creates a framework to allow oversight of study design, ensures validity, facilitates independence from commercial interests, and makes comparisons of drugs in the same class possible. Research would also be publicly available instead of being proprietary. Often we have a problem where the drug benefit plans in the provinces would like more information on the safety and efficacy of a particular product and Health Canada tells them it can't share that information because it's proprietary. So there's a problem with transparency in terms of getting the information to the drug plans and ultimately the physicians and patients who will be using that information. If we have research that's conducted in a public forum, then we will allow that data to be publicly accessible.
Several national regulators commission post-market studies from several centres. The U.S. has a research network called DEcIDE. New Zealand has a national pharmaco-vigilance centre. And in the European Union they are developing an EU-wide approach to commission international pharmaco-surveillance research from over 60 research centres across Europe. This is in the early stages of development.
Canada is well positioned to realize the potential for a national network of research centres. Our provincial health care plans have electronic health care records and pharmacy dispensing records that could be used to conduct observational research that would augment an activity already underway in Canada.
But a commitment and a will to support cooperation among provincial health care systems and Health Canada is essential to develop the infrastructure needed for pharmaco-vigilance and public health impact studies.