Skip to main content

Cardiovascular Disease

500,000 Excess Deaths From Vioxx? Where?

Death rate chart
Cardiovascular chart
Several people have sent this along, or similar writeups. The claim is that Vioxx actually was responsible for over 500,000 deaths, which would be about ten times the absolute highest (and much disputed) estimate that’s been seen to date. Ron Unz, the publisher of the American Conservative who makes these allegations, says that they’re obvious in death-rate figures, particularly in the older population:

“We find the largest rise in American mortality rates occurred in 1999, the year Vioxx was introduced, while the largest drop occurred in 2004, the year it was withdrawn,” says Unz. “Vioxx was almost entirely marketed to the elderly, and these substantial changes in the national death-rate were completely concentrated within the 65-plus population.”


I found this claim very hard to believe. (For one thing, how could all those patients and lawyers suing Merck have let this get past them?) Looking at the statistics themselves, I can see no evidence for Unz’s claim. Here, for example, is the death rate in the US, crude and age-adjusted, over this time span:
And to get more specific, here are the numbers for cardiovascular deaths for people 65 and over. (They’re in a chart comparing them to cancer death rates as well):
I do not see the effects that Unz is talking about. Not at all. A single-cause change in the death rate of the magnitude that he’s proposing should most certainly show up in these figures (particularly the latter chart), but it isn’t there. I see no reason to take this claim seriously.
For those wanting more, there are more arguments against this theory in the comments here).

24 comments on “500,000 Excess Deaths From Vioxx? Where?”

  1. Hap says:

    500K/6 years = 83K deaths per year
    It seems like 83K deaths a year for overall mortality, much less in a single age group, would be hard to miss. Wouldn’t that be something like 2% of the overall deaths in the US during that period?
    Also, was Vioxx marketed exclusively to older people? Older people are more likely to have arthritis, so they might be a larger part of the set, but I thought Vioxx was marketed to a far larger set of the population (which was why any effects were seen). Am I confused?
    I can hardly wait to see his economics suggestions.

  2. Electrochemist says:

    The data Derek shows seem pretty clear in refuting Unz’ argument. In reading the links, it appears that Unz is basing his conclusions on total death rate (mortality) trends, not stratified by cause of death. Those data, however, also cannot be used to support his claims.
    Here is a link to total mortality trends for US males, 1960-2007:
    and here are the data for US females:
    There are no “large rises” in total mortality rates for either gender in 1999, nor are there large drops in 2004.
    Just another example of how little attention is paid to verifying claims like Unz’ prior to publication in the press.

  3. exGlaxoid says:

    I would rather see a chart showing the number of older people who die early due to wasting way from not being able to keep active or exorcising due to pain. I have seen this in several relatives, one of who took Vioxx, and it was nearly a miracle drug for them, as they had too many issues with Aspirin, Aleve, or Advil to use them any more, and Celebrex has never worked well for them. Once it was off the market, they have been unable to keep as active, and it is slowly killing them. But that type of benefit is rarely noticed by the press.
    As well, many other withdrawn drugs worked very well for certain people, and I wish the nuts who want to ban most prescription drugs as “dangerous” would just go away and leave me alone to make the risk-benefit analysis with my doctor. I can think of at least 2 other drugs withdrawn which were also safe and very effective when used properly.
    The one area that made that tough was the FDA allowing direct to consumer ads, which have then overwhelmed the Docs so that now they can’t do that as easily. And companies over-marketing the drugs and pushing unapproved uses is just as bad.

  4. cirby says:

    During the clinical trials for Vioxx, I made a fair amount of money running slides for the presenters. I also spent a lot of time converting 35mm film slides to Power Point versions.
    One of the things I remember about staring at those slides over the course of a couple of years was the low to nonexistent death rates in their test populations. Basically, there weren’t any deaths, if you discount the one or two from non-drug incidents (I believe they lost one participant to a car wreck).
    I’d be willing to bet that the 500K “extra deaths” is a statistical artifact.

  5. Aspirin says:

    How many people still think that pulling Vioxx from the market was a bad idea? Unfortunately, not enough.

  6. MIMD says:

    While I agree the VIOXX numbers here are likely erroneous, the point of the cheapening of the value of American life is depressingly accurate.
    For instance, look how readily companies lay people off, ruining them, and perhaps forcing them out of the workforce forever.
    Also, currently being pushed by HHS is a medical device for rapid national implementation known to cause injury and death. The government is partially financing it to the tune of tens of billions of dollars, probably with Chinese money no less.
    There are financial penalties for medical refuseniks (non-adopters).
    However, FDA, the Institute of Medicine and others readily admit in publication thay have no idea of the magnitude of the harm because of lack of data collection, impediments to information diffusion and even legal censorship of the harms data. In effect, we don’t even know if the benefits exceed the harms, and FDA and IOM admit it. FDA in fact refers to the known injuries and deaths from this device as “likely the tip of the iceberg.”
    Perhaps to some it’s no longer a big deal if people are injured and/or die in data gathering for this medical enterprise.
    E.g., see “FDA Internal Memo on H-IT Risks“, and the Inst. of Medicine report in the same issues here.
    It’s all for the greater social good, they might say.

  7. tgibbs says:

    It is worth noting that GI toxicity of nonselective COX inhibitors, such as aspirin, causes substantial mortality. The actual numbers are uncertain, but I’ve seen estimates as high as over 10,000 per year in the US, and there is evidence that GI risk was substantially lower with rofecoxib (Vioxx). So a proper analysis would need to also consider lives saved by Vioxx. Also, it now seems likely that there is some degree of cardiovascular toxicity for older NSAIDS, with the likely exception of aspirin, so it is unclear to what extent the availability of Vioxx increased actual mortality in this patient population. It would not be surprising to me if, when all factors were considered, Vioxx turned out to be pretty much a wash in terms of overall safety. And I’ve had a number of people with joint pain tell me that it was the most effective drug they’d ever tried.

  8. NoDrugsNoJobs says:

    Many countries did not pull Vioxx but strongly affirmed its staying on the market (e.g. Canada) I believe. Many people forget that even the fda advisory board did not vote to pull vioxx but Merck voluntarily pulled it – Merck, I think, tried to do the right thing but probably would have been smarter to leave it on the market with restricted access – IMO

  9. Okay, normally I try not to feed the tinfoil hats, but exactly which government is it that is financing the dastardly undisclosed death-causing device to the tune of tens of billions of dollars with Chinese money?
    I would have thought the Chinese government was the only one that could logically do that, but they don’t count their expenses in dollars, do they?

  10. Watson says:

    @6 You keep using that word – “device” – I do not think it means what you think it means

  11. MIMD says:

    ‘medical device’ is the term chosen by FDA and SMPA (EU).
    But that’s a distraction from the points I raise in the linked post about the experiment.

  12. Watson says:

    The linked article is discussing the poor state of “medical device records” because of a lack of uniform specifications with respect to Health Information Technology, i.e. how these technologies code data and the challenges of making the data obtained uniform across a wide variety of implementations and vendors.
    It seems that the concern, far from being that Health Information Technology is “killing” people, is that the Medical Device Records may contain duplicate reports for adverse health events because of health care providers encoding the data more than once for each event. This problem with replication exists because there are different health record systems where this data needs to be input, and perhaps the same patient uses different physicians who have different systems, but all of which are required to report adverse events.
    In other words, “Health Information Technology” is not some monolithic “device”, and your conflation of “HIT” which is more properly an abstract term with the “devices” which are used to generate some forms of patient data is in my view the real distraction.
    Yes, some of the “devices” (a blood pressure monitor for example) may have underlying issues, which the FDA regulations for “medical device records” are designed to identify. The FDA, as a governmental entity has no constitutional power to mandate certain devices or implementations are to be used. The power that the FDA does have is to inspect that the manufacturer of a device keeps appropriate medical device records (e.g. a lot of syringes, or a batch of formulated drug) and addresses any complaints about the device to the satisfaction of the FDA.

  13. anonymous says:

    Sorry, but as non-PC as this might appear, the TRUE crime was pulling Vioxx off the market – see comments #3 and (particularly) #7.

  14. milkshaken says:

    “66.67% of all statistic studies are just made up numbers”

  15. Fred says:

    “Sorry, but as non-PC as this might appear, the TRUE crime was pulling Vioxx off the market – see comments #3 and (particularly) #7. ”
    For sure. And you know what? Some day, some non-American generic house (TEVA, maybe) will make a killing on profits that rightly belonged to Merck for the Vioxx IP– or off Celebrex (Searle,Pharmacia,Borg), which I believe, goes off-patent this year.
    The abject failure of the FDA to weigh benefit against (minimal) risks is huge factor in driving a stake through the pharmaceutical sector. That plus biologicals, out-sourcing, and the dearth of VC investment tells the whole story. There are plenty of great drug targets out there– sitting around gathering dust on company and university file servers.

  16. Kevin says:

    Well, Fred, speaking as someone with bad joint pain from a very misspent youth, I personally can’t wait. Vioxx was awesome – I could move without pain. Because of reactions to sulfa drugs, I can’t use Celebrex. When they pulled Vioxx, I gained 20 pounds due to decreased activity (fixed that now but I am still less active) and pretty much live with constant pain. I’d take the risk to get movement back.

  17. MIMD says:

    It seems that the concern, far from being that Health Information Technology is “killing” people, is that the Medical Device Records may contain duplicate reports for adverse health events because of health care providers encoding the data more than once for each event
    Yes, fix just that little problem and then the problems with clinical IT are solved! (Actually,I’m not even sure what you’re referring to, but the evidence is that fixing it as you suggest is the cure.)
    The FDA, as a governmental entity has no constitutional power to mandate certain devices or implementations are to be used.
    You are also right about FDA. They were completely toothless even in this situation.

  18. MIMD says:

    In other words, “Health Information Technology” is not some monolithic “device”, and your conflation of “HIT” which is more properly an abstract term with the “devices” which are used to generate some forms of patient data is in my view the real distraction.
    Those who conducted the Tuskegee experiments probably felt the same way.
    It’s all about definitions, not ethics, and not data – which FDA as well as IOM or the National Academies, our highest scientific body, among others, admits, as in the linked posts in #6, is quantitatively and structurally lacking on risks and harms.
    I don’t really mean to laugh at you, not knowing how little you really know about the Medical Informatics domain, but you bring to mind this Scott Adams adage on logical fallacy:
    Example: My house is on fire! Quick, call the post office and tell them to hold my mail!
    And with that, I move on, letting others enjoy the risible comments from surely to follow! 🙂

  19. Watson says:

    I read the articles you originally linked to, and my comments were based upon trying to interpret your meaning from those selections. I worked in the industry and had to deal with GMP, and had to make sure to follow all of the guidelines with respect to medical device manufacturing and electronic records. I understand the terminology very well. Luckily, I never had to deal with “health IT”, but I did have to pore over enough pages of Federal Register legalese to know that what is sufficient is not necessarily what is best.
    Is that risible enough for you?
    The link that you provided in @17 was a much more concrete example, and if you had referenced it in your original post, would have cleared up much of the confusion that I (and I assume @9) faced in understanding what it was you were trying to convey. It would have been useful if you had explained which device or devices you were talking about. If you had more than conjecture to back up the Chinese money trail, and if you had provided an example of a company that has been damaged by being a refusenik, those would have supported your argument as well.
    Straw man and ad hominem fallacies are pretty transparent around here, and I wish you the best with both.
    I assure you that I recognize what’s important, that I have ethics, and that I care about people having reliable healthcare.

  20. MIMD says:

    The FDA taxonomy of HIT safety issues in the leaked Feb. 2010 “for internal use only” document “H-IT Safety Issues” available at the link in my post #6 is quite clear:
    – errors of commission
    – errors of omission or transmission
    – errors in data analysis
    – incompatibility between multi-vendor software applications or systems
    This is further broken down in Appendices B and C, with actual examples.
    Both this FDA internal report and the public IOM report of 2011 (as well as Joint Commission Sentinel Events Alert on health IT of a few years ago, and others) make it abundantly clear there is a dearth of data on the harms, due to multiple cultural, structural and legal impediments to information diffusion.
    Yes, it’s in the linked IOM report at #6 entitled “Health IT and Patient Safety: Building Safer Systems for Better Care”. See for instance the summary pg. S-2 where IOM states about limited[ed] transparency on H-IT risk that “these barriers to generating evidence pose unacceptable risks to safety.”
    Argue with them, not me.
    Back to my original point: national rollout of this medical device (whatever you call it is irrelevant to my point, but see Jeff Shuren’s statement to that effect here) under admitted conditions of informational scarcity regarding risks and harms represents a cheapening of the value of patient’s lives. Cybernetics Over All.
    As to your other misdirection, spare me the lecture. It’s not ad hominem to call statements like “The FDA, as a governmental entity has no constitutional power to mandate certain devices or implementations are to be used” for what they are – laughable (and I am being generous).
    FDA’s authority is statutory, not written in the Constitution. Same with their parent, HHS. To get quite specific, on human subjects experimentation, which the H-IT national experiment is, the statutory authority for HHS research subject protections regulations derives from 5 U.S.C. 301; 42 U.S.C. 300v-1(b); and 42 U.S.C. 28.
    A real scientist would have known things like this before posting, or have made it their business to know.
    Tell me: are you in sales? Not to point fingers, but with your dubious evasion of the ethical issue that was the sole purpose of my post, and your other postings using misdirection and logical fallacy to distract, you fit that mindset.
    You certainly don’t sound like a scientist. Any med chemist worth their salt (pun intendend) would have absorbed the linked reports and ethical issues accurately, the first time.

  21. MIMD says:

    Note: since this topic is about health IT, not pharma per se, I am taking this comment thread to the Healthcare Renewal blog, at the post “Nimnod, or Industry Shill? You Be The Judge.

  22. ginny says:

    I am just a lay person with no expertise when it comes to molecular science and have, unfortunately, become versed in how the pharmaceutical companies operate and how much control they have over the judicial system, doctors and health care organizations. I can say, based on the information I have obtained, Vioxx was rushed to market, despite Merck’s knowledge of the higher risks of heart attack and stroke.It is blatantly apparent Merck did not consider the many lives that were affected and was only concerned with profit. They flooded the market with samples, “neutralized” doctors that voiced their concerns of patient health, but, and foremost, they knew they were killing people! Don’t tell my doctor there is no evidence of an increased risk of heart attack and stroke. The sales reps were given scripts on how to handle or divert the attention away from the risks when selling the drug. It is proven that there was no real study when this drug hit the street. Merck didn’t care that Vioxx was killing people; they just wanted to find out how best to market the drug. They practiced “seeding” and put it out there. If a doctor or other health care professional objected to their fake findings that Vioxx was a safe drug, they were “hit” and “neutralized” to conform. A case in point is Dr. Topol. What happened to him???? Look it up. I could go on and on………..However, and Ms. Reicin said it very directly, that if you put the aspirin back in the drug, which prevents the heart attacks, strokes and thrombotic events, you don’t have the benefits of treating patients without the side effects. I am sorry, but if I would have had the choice, I would take a stomach ache over a dead heart any day. Did the drug work? Yes, but I could continue to live, and I say that in the most literal sense, with a golf elbow; I can’t live without my heart. Nor can I function with the many strokes I have daily. But, hell, I am just a case number, waiting to be victimized again by Merck, their lawyers and the judicial system.It is a sad day when you realize just how much power the pharmaceutical companies have in this country. All power, profit and control…….and it is all under the guise of “saving lives”….Give me a freaking break.

  23. MIMD says:

    In an interview with the Clinical Risk Management & Safety Surveillance group in Merck Research Labs ca. early 2004, I was told by drug safety scientists that their access to safety data was impaired by sluggish and backwards IT support. They were definitely irritated.
    This was in the context of my showing them, on a laptop, novel human-computer interaction designs I’d developed for use by busy biomedical researchers for a large clinical study in the Middle East that were well-accepted, as well as a comprehensive invasive cardiology information system of my design to detect drug, device and treatment modality risks developed for a regional center performing more than 6,000 procedures/year. The latter system made database querying and risk evaluation of new invasive cardiology drugs, devices and techniques much simpler.
    The drug safety scientists and Exec. Director I spoke with were eager to have such a flexible data retrieval and analysis tool, because they stated they would get partial excerpts of drug safety data from IT on paper and in other formats that were difficult to use or analyze. I was recommended for hiring by the Executive Director of the group and ran out to Rahway to sepak to the Exec. Director’s boss.
    Paradoxically (or not), he Exec. Director’s boss, a former FDA official, dismissed me after just a few minutes, as I documented here.
    It’s possible the company is being honest about not really knowing the safety profile and thus acting in “good faith.” (It’s not bad faith per se if you can’t find a fever because you don’t take a proper temperature.)
    Cutting scientist access to SciFinder in mid 2003 was another move that could be interpreted in a negative light vis-a-vis risk awareness.

  24. Peter Schaeffer says:

    From 1998 to 1999 Vioxx prescriptions rose from zero to 4.845 million and death rates rose. From 1999 to 2000 they rose by 15.785 million and death rates fell. Given that the larger change in Vioxx prescriptions was associated with a decline in death rates, this would tend to indicate the Vioxx reduced mortality.
    From 2003 to 2004 Vioxx prescriptions fell by 5.965 million and death rates fell. From 2004 to 2004 Vioxx prescriptions fell by 13.994 million (to zero) and death rates rose. Given that the larger change in Vioxx prescriptions was associated with an increase in death rates, this would tend to indicate the Vioxx reduced mortality.
    Of course, it is unlikely (but not inconceivable) that Vioxx reduced mortality while it was on the market and the withdrawal of Vioxx increased mortality. A more likely explanation for the observed data is that other factors (much) more than offset the impact of Vioxx either way.
    However, it turns out that more sensitive analysis of the data is possible. Vioxx was introduced in late May of 1999 and withdrawn on September 30th of 2004. The CDC publishes monthly mortality statistics that can be used to compare the actual 12 months before and after Vioxx was introduced and the 12 months before and after Vioxx was recalled.
    It turns out that the average monthly death rate was almost identical (70.575 versus 70.599) for the 12 months before and after Vioxx was introduced. This is a better test than using calendar years because Vioxx was only sold for 7 months in calendar 1999. By comparing 1998/06 – 1999/05 to 1996/06 – 2000/05 we are testing 12 months of no Vioxx (some Vioxx in late May 1999 is possible) versus 12 months of full marketing.
    The same approach can be used to analyze the Vioxx withdrawal. For the period from 2003/10 – 2004/09 the average monthly death rate was 69.444. For the 12 months after Vioxx as recalled, the average monthly death rate was 68.708. Net, the average monthly death rate declined by 0.736. By contrast, the average monthly death rate for calendar 2003 was 70.175 versus 68.107 for calendar 2004 (a net decline of 2.068). Using actual 12 month periods is a better test because Vioxx was prescribed for 9 months in 2004 making comparisons with 2003 difficult. Stated differently the death rate decline, using actual 12 month periods was 35.58% of the decline using calendar year periods.
    Was the 0.736 decline in average monthly mortality a consequence of the Vioxx recall? Perhaps, perhaps not. Most of 0.736 decline appears to be related to a large P&I (Pneumonia and Influenza) epidemic in late 2003 / early 2004. Of course, the data is predictably more complex. The monthly death rate in March of 2005 was higher than the death rate in March of 2004. Once again, Vioxx appears to have been a secondary (or lesser) influence on mortality.
    Thank you
    Peter Schaeffer

Comments are closed.