A. Paroxetine is prescribed for depression, anxiety, panic attacks, post-traumatic stress disorder (PSTD), obsessive compulsive disorder (OCD), and most recently, for symptoms of menopause. It is currently marketed under more than 100 different names, including: Aropax, Brisdelle, Casbol, Denerval, Eutimil, Frosinor, Isoxatine, Motivan, Pexeva, Seroxat, Tagonis, and Zwapar.
Q. What is RIAT?
A. RIAT stands for Restoring Invisible and Abandoned Trials. It was the brainchild of Peter Doshi and Tom Jefferson. Where there is access to data, RIAT teams propose to re-author studies that have not been published or have been misleadingly published. Currently, if a company runs 50 trials and 48 show negative results but two show slightly positive results, the company can use only the two “good” ones to get its drug approved and can publish these two, giving the impression that the findings of the two are representative. The RIAT proposal aims at publishing some of the data from the other 48 trials to give a truer, more balanced picture.
GSK have talked about data transparency since 2004, but there has been little progress in ensuring that interested parties have access to all the information necessary to fully understand and evaluate research conclusions. RIAT puts pressure on drug companies to make good on their stated commitment to data access. Study 329 is the first misleadingly reported study to undergo RIAT analysis.
Q. The original Study 329 was published 14 years ago. Why did the new research team decide to re-interpret the data?
A. The original Study 329 had many problems with data classification and interpretation. Its conclusions were so incongruous with the data that in 2004 the Attorney General of New York filed a consumer fraud action against GSK; part of the settlement was that GSK would post trial results on the company website. The company settled another lawsuit brought by the Department of Justice in 2012 for $3 billion. The discrepancy between the data and the conclusion made it a good inaugural case for RIAT.
Q. What are the differences between the findings of Study 329 and the new version, Restoring Study 329?
A. The original Study 329 report concluded that paroxetine (Paxil) was effective and safe for adolescents. It reported that six young people became “emotionally labile” on Paxil, without explaining what that meant, and did not report any problems of dependence.
Restoring Study 329 is based on an assessment of as much of the original data as the researchers could get through in nine months. They were provided with access to scanned copies of redacted patient data through a remote access portal which they could not print or download. The information was not organized in a way that they could clearly see how the original conclusions had been reached, and more than personal identifiers had been redacted. Still, the researchers were able to determine from a detailed review of 77,000 pages of patient records that suicide attempts were significantly higher than what the original study had reported, and there were many other serious adverse events in the Paxil group. They noted that the drug was no more effective than placebo for alleviating depression in adolescents.
Restoring Study 329 concludes that Paxil is neither effective nor safe. The term “emotional lability” turns out to mean suicidality and there were a number of examples of suicidal behavior in the Paxil group that had not been counted at all. Also, many adolescents became dependent on Paxil after just 8 weeks of exposure, as shown by serious adverse events on tapering.
Q. If Paxil is not safe and effective for kids, is it safe and effective for adults?
A: The FDA broadened the Black Box warning about suicide to include young adults, and then all adults, because the problems are not confined to young people.
Q: Why is paroxetine so dangerous?
A: All SSRI antidepressants increase the risk of disturbed thinking for a significant percent of users. This can include manic reactions, psychosis, hostility, violent impulses, social withdrawal, paranoia and suicidal preoccupation. Dependence is a common problem, leading to serious difficulties upon withdrawal. There is evidence that long-term use of SSRIs worsens the prognosis for recovery from depression.
Q. Are selective serotonin-reuptake inhibitors useful medications or not?
A. SSRIs can be useful for a small group of people, especially when used for a limited time. Careful assessment and monitoring are critical to ensure that only people who can truly benefit are given these drugs. Because most people who get prescriptions for them cannot benefit. Given that harms often outweigh the benefits, the current extent of prescribing SSRIs will some day be viewed as a major disaster.
Q. Don’t randomized controlled trials (RCTs) prove drug safety and efficacy?
A. Randomized controlled trials (RCTs) have long been over-rated as a way of judging the real safety and efficacy of medications. When a drug can produce the same outcome as the condition for which it is prescribed – a situation that exists for about half of all medications – RCTs tend to lead to erroneous conclusions, as with Study 329. Where the true situation involves complexity, RCTs cannot answer the relevant questions.
RCTs are useful for one thing: showing when drugs don’t work. This is exactly what the data from Study 329 and other paroxetine trials revealed: that paroxetine did not work. Despite this, the Study 329 trial was presented as a “landmark” trial demonstrating its efficacy and safety! This illustrates just how cynical pharmaceutical company executives can be. GSK executives knew all along that Paxil was neither effective nor safe, so obviously the lives and wellbeing of young people were not the most important considerations for them.
Q. Was Study 329 atypical or do other drug studies have the same serious flaws?
A. Sadly, Study 329 is unlikely to be atypical. Drug studies have been seriously flawed for a long time. We now know that the drug trials that led to Prozac (fluoxetine) had the same problems as Study 329. We know this because of evidence produced in lawsuits. Since other SSRIs and SNRIs are similar, there is no reason to believe that the evidence supporting their suitability is any better. Nor is there any reason to believe that the situation is any better for cardiac or respiratory or migraine or other drugs in any age group. What Study 329 shows seems to standard industry practice.
The basic problem is that the process for safety and effectiveness testing has been delegated to the manufacturers of the drugs, who have an obvious vested interest in seeing positive results. This delegation has not been accompanied by a related requirement that every aspect of data, except for personal identifiers, be made available to all interested parties. Without complete transparency, the temptation for drug companies to manipulate data to make their product look better is irresistible. Now that trials are increasingly run in India and South Africa, where regulation is even less effective, research participants will be at greater risk, and research contractors have even been known to fabricate patients.
Q. Why is access to ALL data necessary?
A. Clinical trials involve a significant amount of interpretation at various stages. Most transparency policies and proposals provide access to MOST data but not All data.
Access to most data is not good enough. We need access to everything to review the interpretation at each stage. This includes the case report forms (CRFs) that are the snapshots by the people in direct contact with the subjects – the “facts” before interpretation.
For example, if a subject in a clinical trial reports feeling a certain way, this is recorded in the CRF. The researcher makes a decision about how to interpret and code these facts. It is important for others to see what the person reported and be able to judge for themselves if they agree or not with the way the report was interpreted, classified, and counted by the researcher.
Restoring Study 329 would not have been possible without access to the CRFs. Our team examined the individual CRFs and interpreted, classified, and counted the events recorded. It is at this level that major differences arose between the original 329 and the current one.
Think of it this way: If the New York Times ran stories without fact checking, it would be the end of its credibility. It is concerning that some of the most prestigious medical journals seem to be unconcerned about fact checking. Unless there is access to all data, the article in which it is reported should not be regarded as a scientific article.
Q. But surely scientific articles are authoritative in a way that stories in the New York Times are not?
A. When there is a large amount of data collected on an issue there are always different ways to interpret it – and it’s important to have the full range of possibilities explored. That can only occur when there is unrestricted access to ALL of the data.
Without ALL the data, scientific articles can make statements that are unverifiable and can be wrong. This is where marketing trumps science, and leaves patients vulnerable to prescriptions they do not really need, that do them more harm than good. Access to data is the only way to ensure that marketing messages get balanced against the facts.
Q. What have we learned about scientific authorship?
A. We now know that for on-patent drugs, a high percent of supposedly scientific articles were “ghost written”. Ghostwriting means that the write-up is done by the drug company or someone hired by it, with the primary goal of delivering a marketing message, and attributed to a well-known academic in the field, who is paid or otherwise rewarded by the drug company to attach their name.
Q. Will ghostwriting of drug studies finally end?
A. Ghost-writing is a problem, and ideally it would not be allowed. However, ghostwriting would be far less of a problem if researchers had full access to ALL data. There is no guarantee that Martin Keller or his other co-authors in the case of Study 329 would have been any more honest than ghostwriter Sally Laden, unless he had known that his data and interpretation would be open to scrutiny. As long as any part of data is held secret, manipulation of results cannot be discovered and the biggest incentive for researchers to be honest is lost. Trust in drug testing has been badly eroded, because we have learned that having the name of a respected academic does not guarantee research quality.
Q. How much money have flawed drug studies cost healthcare systems?
A. The global prescription drug market was estimated at US$955 billion in 2011 by IMS Health. If only 10% of this was wasted on ineffective drugs, this would represent around$100 billion per year, excluding the costs of treating adverse effects.
Q. How many deaths have flawed drug studies caused?
A. Prescription drugs, taken as prescribed, might be the leading cause of death in the OECD. We know that the rate is very high but unfortunately death information is not properly tracked. While we carefully track and seek to minimize deaths from cancer, cardiovascular disease and road traffic accidents, coroners in most countries do not even have a way to record deaths caused by prescription drugs.
Q. Isn’t what you are suggesting as a solution the same solution that the All Trials group has been working toward?
A. No. AllTrials are asking for companies to register all trials and provide company written reports about each trial. We are asking for the underlying patient level data.
Q. Can you give us a brief explanation on the difference between data-based medicine or AllData and evidence-based medicine or AllTrials ?
A. The difference between data-based medicine (DBM) and evidence-based medicine (EBM) is that DBM wants the data so that you can make up your own mind what it shows. EBM wants to offer you their spin on what the data shows – and whether they are pharmaceutical companies or the government or some self-appointed experts – having taken the risks of testing out new drugs, we believe you are entitled to the data – spin is not enough.
Q. But aren’t the data summaries available to researchers sufficient?
A. Medicines can’t be made safer without full access to ALL de-identified, individual patient-level data. Many people have claimed that the provision of data summaries, extractions, and interpretations constitutes sufficient evidence of safety and efficacy (Evidence Based Medicine). We believe that important truths are only revealed in the details of the data. Data must be available so that all interested parties can do their own analysis and interpretation. This is why we need Data Based Medicine.
Q. How can I access a study 329 timeline and supporting documents?
A. Our team has set up a website with background summaries and documents. Go to Study329.org.