SHIT Happens – 2

October, 31, 2013 | 23 Comments


  1. From Peter Gotzsche:

    David, see p 127 in my book about reporting serious ADEs immediately:

    In 2010, the FDA warned Pfi zer in a 12- page letter for failing to quickly report serious and unexpected potential side effects from its drugs after having conducted a 6- week inspection of Pfi zer’s headquarters.108 Pfizer had misclassifi ed or downgraded reports to non- serious without reasonable justifi cation and had failed to submit reports on blindness caused by Viagra (sildenafil) and similar medications within the agency’s 15- day deadline.

    Pfizer was also warned in 2009, but the FDA noted that the company’s delays in telling the agency about harms had only grown. Pfizer was told that failure to fi x the problems could result in legal action without notice and delays in approving the company’s pending drugs.

    In 2012, Roche was reprimanded by the EMA for failing to report up to 80,000 possible adverse reactions from its drugs, including 15 161 deaths in the United States.109 Regulators identified additional defi ciencies related to the evaluation
    and reporting to national drug agencies of suspected adverse reactions in 23,000 other patients and 600 participants in clinical trials.

    See Gotzsche P. Deadly medicines and organised crime: How big pharma has corrupted health careLondon: Radcliffe Publishing; 2013

    • Thanks for this David and Peter.

      I freely admit that while being aware of the incidents that you describe, my knowledge of them is not detailed.

      The lens through which I view these events may be simple minded. It is simply based on my personal relationships with many folks involved in regulatory affairs within the industry. I can’t say that any of them would participate knowingly in any sort of vast coverup of AEs.

      The underreporting of AEs is unacceptable in any circumstance, whether through incompetence or criminal intent. But the questions I would put to you would be:

      1) If the intent was to deliberately hide these events, why were the records not destroyed rather than left in files subject to FDA inspection and revelation?

      2) What was the impact of the non-reporting of these events? What drugs were pulled from the market when they were revealed? Which received black box warnings? Did the non-reporting of these adverse events materially affect the perceived risk/benefit ratio of any drug? Did their revelation, when it finally occurred, negatively impact the financial interests of either of these companies in a material way?

      To be honest, I don’t know the answer to these questions. I think the answers are important in interpreting intent.


      • David and Peter:

        Just an additional thought that occurred to me after posting:

        In the case of Roche, I think the 80,000 unreported AEs (including over 15,000 deaths) is clearly unacceptable. But I think it also needs to be put in the context that Roche’s primary business is in oncology. The AE rates in this indication tend to be extremely high, as the disease itself has a fatal outcome when it proceeds unchecked. Indeed, in the case of certain types of leukemia, the best overall survival rates are achieved when the number of deaths from the treatment approaches that from the direct effects of the disease.

        Roche sells about $5.5 billion dollars worth of Herceptin every year; about $6 billion worth of Rituxan, and $6 billion worth of rituxan. If one assumes that the average patient worldwide receives $50K worth of drug, its about 350,000 patients per year. According to the Rituxan FDA prescribing information, 99% of patients in clinical trials experienced AEs, and 57% experienced Grade 3/4 AEs. The AEs recorded described in the Avastin and Herceptin package inserts are similar in magnitude and severity.

        So the expected rate of reportable AEs for just these three drugs is at least 350,000/year (some, probably most patients will have more than one), with over half of these being severe or life threatening.

        While 80,000 unreported AEs is clearly unacceptable, in the absence of other information that you might provide, I’m not sure that I believe that the intent or outcome here was to materially change perceptions of the risk/benefit ratio of these drugs.

        Any additional data you can provide to shed light on this will be appreciated.

        • While intent is important, culture is also. We have a culture today which suggests that if an event hasn’t been demonstrated to happen at a statistically significant rate in controlled trials, then it isn’t in fact happening. Against this background it may well be that regulatory and company personnel are less bothered by an accumulation of adverse events. They are essentially just filed away. They rarely lead to action.


          • From John (Defeated by Captcha)

            I’d have to agree with you that “mistakes” which on some level serve the interests of the person making the mistake must always be viewed with suspicion.

            But the title of the book being cited is “Deadly medicines and organised crime: How big pharma has corrupted health care”. It seems to me that these events are being presented as an example of deliberate criminal acts of monumental proportions.

            I’m only pointing out that while this behavior is unacceptable, the revelation of the data did not lead to any drug withdrawals, any new black box warnings, any loss of revenue, or any substantive re-evaluation of the risk-benefit ratio for any drug, at least to the best of my knowledge. In terms of consequences to patients, it seems to have been a non-event. Thus it strikes me as over-reaching to use this incident to support the implied claims in the title of the book. There is nothing in these incidents that I can see that can be objectively associated with the words “deadly”, “crime”, or “corrupted”. Nor for that matter, with “organized” :>).

            But I suppose it is all in the eye of the beholder….

          • Well I’m speaking for Peter here and as someone who hasn’t used the words deadly, organized or crime but the system we have at present does swallow up tens of thousands of deaths without blinking an eye.

            And this links to the point made in the following post which has GSK sitting on a statistically significant increase in suicidality in children in Study 329 and yet heavily promoting this study as a means it seems to get children put on Paxil.

            Study 329 was also one where the consent form says the children will not be exposed to risks that differ from normal clinical care but they were it would appear put on double the dose of imipramine that would normally have been used and all it seems to obtain a marketing advantage.

            This is deeply disturbing. No recalls of the drug, no drop in revenue, GSK’s CEO knighted. Its good we have some Peter Goetzsches.


  2. These days the press is full of babble about “E-patients,” smartphone apps and social networking as tools to help Pharma give us better drugs. And there’s no doubt Pharma is listening. They’ve even gone so far as to found social networks of their own, so they can listen in on the scuttlebutt. (Here’s a “community” set up by the contract-research giant Quintiles: ) And every angry-patient website has had multiple “hits” from drug-company servers.

    The question is, what are they doing with that information? I’d bet they are analyzing the hell out of it but NOT reporting it externally. Instead of using it to correct or warn of problems with the drug, they’re using it to craft counter-arguments … and even to plan counter-research. (The best theme is “it’s the disease, not the drug.” Thus it’s schizophrenia, not antipsychotics, that put you at risk for diabetes, etc. etc.) The other thing they can do, of course, is monitor complaints about the competition and look for ways to make their own product appear safer – whether it is or not.

    Data-mining of this kind gives them what the CIA would call “plausible deniability”. Even if they admit to reading these complaints, they’re anonymous, and they lack key details. Who’s to say what shape these people were in BEFORE they took our drug? Or what other drugs they were on, or what they were drinking and smoking, or what kind of stress and strain they were living through? In reality they may consider this data very solid – but they don’t have to admit that.

    Here’s where we figure out why they are not burning up the phone lines calling What RxISK offers them is Too Much Information (and too high quality). Plus which they can’t just drop in and read RxISK reports without identifying themselves. No plausible deniability – therefore, too hot to handle. Better to say this is just slanted stuff collected by a guy who has (OMG!) worked with (yecch) lawyers! That’s what THEIR lawyers will tell them to say, anyway …

  3. “An industry where Ian Hudson, then of GSK, now head of Britain’s FDA (MHRA), can get hundreds of reports of a problem like violence on Paxil-Seroxat and still say he has seen no evidence that Paxil-Seroxat causes any problems”

    I don’t see any contradiction there at all David. If you put 100,000 patients on Lipitor for life, at the end of a year 10 of them will have died in motor vehicle accidents. That’s hardly evidence that statins cause auto accidents.

    We’ve seen again and again the reports from the Institute of Safe Medication practices and others suggesting that antidepressants cause violence. Most commonly the reasoning goes like this: XX% of spontaneously reported AEs for SSRIs involve violent actions, but only YY% of spontaneously reported AEs for antibiotics involve violence. Therefore SSRIs are causing violence.

    This reasoning fails because of attribution error (most people are more likely to attribute violent behavior to a CNS drug than to an antibiotic, and most doctors know most antibiotics do not enter the CNS) and because of good old fashioned Baysean statistics. Individuals taking psychiatric drugs are more likely to have psychiatric problems than those taking antibiotics, just as those who enter the ER with an active antibiotic prescription are more likely to die in the hospital of an infection. This reasoning conflated the effects of treatment with the effects of the patient’s underlying condition.

    In study after study, geographical and time series correlations of violent crime with antidepressant use show either negative or non-significant correlations. Antidepressant use in the US has doubled since the mid-1990s, but violent crime has dropped by a third. A CDC study looking at suicide rates by US county found an inverse relationship to SSRI use.

    At worst the evidence on this issue is a wash, at best it is suggestive of a reduction in violence. Suggesting that the high incidence of violence among those taking psychotropic medications is due to the medication itself is like blaming antibiotics for pnuemonia deaths.

    But Time magazine never runs a story with the headline “Shooter not taking psychotropic drugs kills 5” and lead their story by asking if the failure to treat this patient with psychotropics contributed to the incident.

    • John
      Here’s the problem. The best evidence that a drug causes a problem like violence or wetting the bed or whatever is patient reports for instance constant unexpected thoughts of violence soon after starting – these stop when the drug stops and for a clincher they restart when the drug is restarted.

      If this happens to you, then the drug has caused it to you. If several reasonable people report it then this reaction is more than idiosyncratic – the drug causes it but we don’t know the frequency. There is no easy way to know the likely frequency with which it happens. This may depend on having a genetic marker.

      Determining the frequency is easier in the case for instance of clozapine and wetting the bed when the drug but not the illness causes the problem.

      Now Ian Hudson at the time of the Tobin case in which he was deposed had a number of compelling reports just like this. All of the companies had them – reports where when they applied the Naranjo or related algorithms the companies had determined internally that their drug had caused the problem.

      But Hudson and others can still say under oath that I have seen no evidence that our drug can cause a problem – in fact other senior GSK employees come very close to saying there is no evidence Paxil can cause any problem. The basis of this claim is that until there is clinical trial evidence of a doubling of violent acts on Paxil compared to placebo company personnel are not going to concede there is any evidence their drug causes a problem even though they have internally written all over adverse event reports – until their lawyers stopped them doing it – that their drug caused this or that episode of violence.

      Now around the time he was deposed, the unpublished clinical trial evidence showed Paxil approximately doubled the rates of acts of violence and did so to a statistically significant extent.

      This should not be taken as evidence that Paxil causes violence – but rather evidence that it leads to more violent acts than it mitigates – it is linked to an excess of violent acts. Clearly to be linked to an excess of violent acts it has to cause violence but the causal demonstration lies in convincing case reports.

      What would it take to have confidence in Ian Hudson as a national regulator against this background? A clear public statement saying that adverse event reports may offer convincing evidence that a drug causes a problem and that when both a drug and an illness cause a problem that clinical trials are often close to useless in determining cause and effect.

  4. “The best evidence that a drug causes a problem like violence or wetting the bed or whatever is patient reports for instance constant unexpected thoughts of violence soon after starting – these stop when the drug stops and for a clincher they restart when the drug is restarted.If this happens to you, then the drug has caused it to you.”

    I would say yes and no. First there is the potential for attribution error even in a challenge/dechallenge experiment. Placebo effects seen in clincal trials, not just for psychotripic drugs but even in indications with objectively measureable endpoints, are an excellent example of this. Its hard to design experiments that eliminate confounding factors well enough that once can accurately characterize the effect of a drug on an average patient, and to me it seems well nigh impossible to try to break this out on a patient by patient basis.

    That being said, I don’t think your decription of clinical trial AE rates as being the net result of increased and decreased outcomes in individual subjects is completely unreasonable. I’m just not sure it has utility unless you can identify a biomarker that allows you to indentify and segregate the two subgroups. And of course retrospective detection of these sorts of effects and responding to them is part of why we require many of these drugs to be taken under the supervision of a physician.

    So why do I object to the approach you are suggesting? Ultimately because I don’t think the average medical practiioner, let alone the public, is as intelligent as you are and as comfortable with such a complex view of causality. They just absorb the message that all these drugs do is harm people. We never see the headline “Psychiatric patient not treated with psychotropic drugs shoots up shopping mall”.

    I’m a dyed in the wool pharma guy. I know all kinds of bad things have been done in the pursuit of profits, but I look around and I see people with HIV standing upright instead of dead in the ground. The average life expectancy in the US is about 85 years, while that in countries that do not have access to the industry’s products is about half that. (This is of course not completely attributable to pharmaceuticals, but a large proportion of the excess death rates in these countries can be traced to infectious disease and other treatable causes.) But when I go to Wikipedia, the average description of a drug there contains 300 lines of text on adverse effects and dedicates only 4 lines on what the drug is for. It makes no sense to me. These folks should live for a year in a world without pharmaceuticals and learn what the industry has really accomplished.

    Like everything else in life, pharma is a mixed bag. Always has been, always will be. Not as good as it was perceived to be 25 years ago, but a lot better than how it is perceived today.

    • John

      The primacy of a visible effect whether grossly on a patient or on a biomarker is not foolproof but is something we need to recover. This applies to both the benefits and adverse effects of a drug. Have a look at the video on RxISK – The Man who Changed Psychiatry – you can go to this through the post – A Black Box Warning for Clinical Trials.

      One of the problems with trials is that beating placebo as a criterion rather than visible effects means that we now in general have weaker drugs than we had in the 1950s and 1960s.

      Another problem is that because trials say these drugs work, we have people on 10-15 drugs or more – this rarely if ever happened in the 1960s and this is maintained even though the visible appearance of the patient makes it clear that they are doing poorly. Life expectancy far from rising is falling and trials of de-prescribing now show that reducing medication burden by 50% of so reduces mortality, hospitalization and of course drug cost.

      But aside from questions of cause and effect, there seems to be a basic dishonesty with Pharma looking to their profit balance rather than the wellbeing of any people taking their drugs. They could likely regain a lot of public trust by speaking straight – Pharma Man speak with forked tongue?


  5. Well, David, I am not familiar with the data supporting some of those statements, but I am sure you have good reason to believe they are true.

    As I stated previously, I do not deny that there has been all kinds of bad behavior. Frankly, that is the history of the industry, and I think the change of the last 25 years is more one of declining tolerance for such things than an uptick in the underlying behavior. The 1950’s to 1970’s brought us amphetamines pushed for depression, “low energy” and weight loss, barbituates for insomnia, and birth control pills which went on the market with nothing close to adequate testing (a dodged bullet to be sure). They also brought the first safe antibiotics for gram-(-) infections, the first successful chemotherapies (I was told at age 12 that my cousin with Hodgkins would not see his 18th B-day, he turned 50 last month), and early drugs for rheumatoid arthritis.

    So yes, we have Vioxx, withdrawn diabetes drugs, and the whole 9 yards of off-label promotion and harm to patients. We also have great drugs from rheumatoid arthritis, hepatitis C, HIV, multiple sclerosis, breast cancer, and some really neat drugs for melanoma, hepatitis B, and lung cancer in the near term pipeline.

    From my POV, it is what it is. It could be better, but it could be a hell of a lot worse.

    • John

      There is no question we could point to areas of medicine that are as bad or worse – the horrors of prostate screening and treatment just to name one.

      My problem is far from seeing industry get better, I see things getting worse. Efforts to rewrite Study 329 – see the post tomorrow on GSK’s Transparency and Access Journey – will let us perhaps work out whether GSK have changed spots as their public pronouncements are busily trying to persuade people. From my point of view it looks like they have fooled AllTrials and they far from being more transparent – as I’ve been writing for a year – GSK have come up with a way to be even less transparent.

      To get to where you would like to be GSK, AbbVie and others would have to have a primary commitment to patient welfare rather than profit and that is simply not the case at present, nor likely to be in the near future.

      Re the new drugs, the marketing of the MABs appears more egregious than the marketing of prior older drugs. The sooner we have Biosimilars the better.


  6. “We’ve seen again and again the reports from the Institute of Safe Medication practices and others suggesting that antidepressants cause violence. Most commonly the reasoning goes like this: XX% of spontaneously reported AEs for SSRIs involve violent actions, but only YY% of spontaneously reported AEs for antibiotics involve violence. Therefore SSRIs are causing violence. ”

    How about we turn things around? Does the current symptom checker for depression include holding up garages, impersonating police officers, robbing banks, running around wielding an axe, stealing shopping trolleys and driving a stolen golf cart whilst firing random bullets on a golf course?

    Just 6 examples.

    All 6 subjects were diagnosed with depression, all were treated with SSRi’s. Could we say they were misdiagnosed or could we say they had some sort of drug induced psychosis?

    Would they have carried out these acts if their depression went untreated?

  7. Bob, some of those individuals were likely taking Lipitor, vitamin supplements, or even eating a lot of Twinkies.

    I don’t mean to be sarcastic, but how does the data you are citing support causation?

  8. If the evidence for SSRI’s causing suicide and violence – at least in some – were really as weak as a simple observation that “suicides are more common in people on antidepressants than people on antibiotics” then we would not have much to talk about. But even in controlled trials, it’s clear that when the figures are re-analyzed (to weed out the confounding of patients in cold-turkey withdrawal and patients who never took the drug, for example) there are more suicidal acts on the drug than on placebo. David Healy is far from alone in recognizing this – that’s why calls for the retraction of GSK’s infamous Study 329 on Paxil in depressed kids have come from so many respected figures in mainstream medicine.

    Clearly this effect doesn’t happen to everyone – if it did, with 11% of Americans aged 12 and up (and 23% of middle-aged women) on antidepressants at any given time, the morgues would be full and the country would be in collapse. To me it’s worrisome enough that the suicide rate among women 45-60 is rising roughly three times as fast as the overall US suicide rate in the past ten years, according to the CDC. Even if they won’t admit the drugs are causing some of the suicides, I hope someday our medical establishment can at least begin to admit that they aren’t doing such a hot job of preventing them.

    But a drug effect doesn’t need to hit every taker to be real. Nor do we need to find a “biomarker” before we take it seriously. To this day I don’t think we have a “biomarker” for allergy to penicillin, and it only affects a small portion of the population. Yet that doesn’t stop us from monitoring people for the allergy, promptly withdrawing the drug when it pops up and even urging those affected to wear Medic Alert bracelets for the rest of their lives, lest someone give them penicillin by accident. (Now if only SSRI’s did as much undisputed good as penicillin!)

    To me, the most compelling evidence has come with the sharp rise in antidepressant prescriptions to people who are NOT depressed. To take just one example, Cymbalta (duloxetine), Eli Lilly’s multi-billion dollar blockbuster, is now prescribed for low back pain, fibromyalgia, diabetic neuropathy, urinary urgency and arthritic knees. People taking the drug for these non-psych indications are reporting identical side effects and withdrawal problems as those taking it for depression. Peter Gotzsche has analyzed duloxetine studies in Europe, where it’s widely prescribed for women’s bladder problems, and found the suicide rates increased as much among the bladder patients as the depressed.

    I hope the medical establishment does not simply yawn and speculate that developing a leaky bladder could cause you to commit suicide all by itself! But I can predict that they will obtain a “meta-analysis” with just that conclusion, and it will be published in a very good peer-reviewed journal … sad to say.

    • Johanna’s comment has an excellent point. The trend of repurposing so-called antidepressants may be their undoing. We can only hope.

  9. Well, ok. But generalizations blur over a lot of important details. Suicide and violence against others do have commonalities, but they are not identical behaviors and evidence supporting a relationship with one does not necessarily prove a relationship with the other.

    The studies I’ve seen point to an increase in suicidiality in children but not in adults. As for violence against others, I’ve yet to see a compelling study showing a relationship. Violent crime has fallen precipitously in the US as the number of people taking antidepressants has skyrocketed. Other studies show that when data for multiple countries is broken out by geography rather than over time, the same negative correlation is obtained.

    I agree these drugs have side effects, and the withdrawal can be pretty severe. But I think its important to be careful not to overgeneralize, and to consider details. At most I’d consider the evidence on this issue to be mixed.

  10. But Johanna, just to be specific about “suicides are more common in people on antidepressants than in people on antibiotics”, I raise that example because it has actually been presented and published in a peer reviewed journal, though on review, it was in reference to varenicline not an antidepressant.

    What is shocking to me is that this hopelessly confounded analysis was published with a standing member of an FDA advisory committee as a co-author.

  11. John

    I hope you’re not feeling ganged up on, but some details.

    Re suicide, the suicidal act rates are as high in 45-65 year olds in the 2006 FDA database as in 16-25 year olds.

    But these trials in children or adults don’t prove ADs cause suicide – they point to an excess of suicides and suicidal acts on ADs. Causality comes from challenge-dechallenge and rechallenge and clinicians have been observing this since 1959.

    Re violence – see the series of posts on Rxisk stories under Shooters on Prescription
    Drugs. ( There is compelling CDR evidence for violence, an excess of violent acts in the RCT data we have, and an easy explanation for the falling violence rates in some populations – there is a falling number of young men on the loose.

    There is also a very active marketing campaign to roll back warnings re suicidality even for children and efforts to deny a link to violence. This is the kind of detail marketing departments – part of the commercial side of pharma you mentioned in another comment – handle.

    I think its a mistake to ignore Bob’s point. If women with bladder problems suddenly start committing suicide, industry are quite capable of spinning a line that bladder problems cause depression and depression causes suicide.

    What is shocking to me is that doctors from – not just the average doctor but media doctors, PhD doctors and everyone else – lets them get away with it.


  12. If I didn’t want to be ganged up on, why would I come to David Healy’s blog and spout conventional wisdom :>).

    I’ll look into this over the weekend and reply.

    Your Captcha is way too hard for me. Usually takes me 6 tries!

Leave a Reply