In recent weeks my description of the possible scale of the Vioxx Disaster has begun getting a little coverage on the web and in the British press, leading to some strong “push back” by people who say I can’t possibly be right. They may certainly be correct in their opinion, but I think their reasoning is mistaken, so I thought I’d briefly summarize the analysis once more, emphasizing again that the evidence is purely circumstantial.
I realize most readers may be growing increasingly weary of Vioxx mortality disputes—I certainly am—but given the tens or more likely hundreds of thousands of American deaths at issue, adding a few short paragraphs of text seem not totally unwarranted.
(*) In 1999, Vioxx was marketed to the American people by Merck as a particularly effective anti-pain medication with minimal side-effects, a sort of super-aspirin substitute targeted at arthritis sufferers in the over-65 category. Backed by an eventual half-billion dollars of advertising, it soon became one of the most widely popular—and lucrative—drugs in this country and the world, with some twenty-five million total American prescriptions. As a consequence, it also became one of Merck’s most important revenue sources.
(*) In 2004, a detailed published FDA study proved that Vioxx had deadly consequences in its patients, greatly increasing the risk of sudden cardiovascular death, and had probably killed at least 30,000-60,000 Americans since its introduction. Learning of the pending publication of this study, Merck immediately pulled the drug from the market. The media later discovered that Merck had apparently been aware of these huge cardiovascular health risks from the very beginning, but had decided to ignore them, presumably because the drug was so lucrative. Merck eventually paid some $8 billion dollars in total government fines, legal expenses, and damages for Vioxx-related deaths.
(*) As it happened, the 2004 American death rate unexpectedly dropped by 50,000, the greatest such national decline in sixty years, a decline whose cause completely mystified American health authorities, who searched in vain for some possible logical explanation. This decline was almost entirely due to fewer deaths in 65+ age range, mostly due to a large drop in cardiovascular fatalities.
(*) Interestingly enough, an examination of the American mortality data freely available on the government CDC website reveals a corresponding rise in deaths for Americans 65+ which had previously occurred in 1999, the year Vioxx was introduced. This 1999 mortality rise was the largest in the past fifteen years, and—perhaps coincidentally—a sharp shift in the rate of cardiovascular deaths had once again been the leading factor.
It is completely impossible for me to say whether or not the recall of a an extremely popular but deadly drug proven to cause cardiovascular deaths among its 65+ target population had any direct connection to the huge drop in cardiovascular deaths among Americans 65+ during that same year. Similarly, the earlier sharp rise in 65+ cardiovascular deaths the year the drug had been introduced may or may not be purely coincidental. But one would think these intriguing facts might arouse a bit of curiosity within American media and government circles.
Based on these items, I have advanced a speculative hypothesis suggesting a much higher Vioxx death-toll than is currently accepted. One of the main arguments which various critics have made against my hypothesis is that although the American 65+ death rate did undergo rather surprising upward and downward shifts during 1999 and 2004, the years that Vioxx was introduced and then removed, the behavior of the death rate during the intervening years was far less remarkable. A few people have argued that the widespread use of a deadly drug during 2000-2003 would surely have caused large, continuing changes in the mortality figures, and their absence tends to completely eliminate the possibility.
Unfortunately, this reasoning is incorrect, and confuses an impact upon total longevity with an impact upon mortality rates. This can easily be understood if we consider an extreme thought-experiment.
Suppose, for example, that the government required everyone aged 65 and above to immediately take cyanide tablets, and established this as a permanent policy going forward, with mandatory cyanide doses being a fixture of every 65th year birthday party.
Obviously, this would lead to many premature American deaths and a very substantial change in American lifespans. Indeed, since our current life-expectancy is around 78, the vast majority of Americans henceforth would be killed by government cyanide, instead of dying naturally. Over the next hundred years, the overwhelming majority of all deaths would be from cyanide, and the total cyanide death-toll in America might approach the half billion mark. Clearly, cyanide would become a very major negative health factor in American society.
However, the actual impact upon the annual American death-rate would be small or perhaps even favorable during nearly the entire period in question, a totally astonishing result. This seeming paradox follows from the fact that everyone eventually dies of something, and therefore there would automatically be huge drops in cancer, heart attack, strokes, and car accident fatalities which would almost exactly balance out the rise in cyanide deaths.
Consider, for example, the American population one hundred years from now and compare it with a non-cyanide scenario. In the former case, there would be no one aged 65+, with that portion of the population having succumbed to cyanide; but those would be the *only* differences in total net-fatalities compared to the base-case Every other American death would have been the same under the two scenarios, though certainly with different timing. And if we average that small slice of additional deaths over the one hundred years in question, the average annual impact is fairly small.
Obviously, the first year of a mandatory-cyanide scenario would see a huge die-off of all those 65+. But mortality rates after that would generally be pretty ordinary, perhaps even sometimes *lower* than under the normal situation, depending upon the shape of the evolving age-distribution curve. Indeed, it is quite possible that people just looking at the mortality rates for the ninety-nine following years and comparing these with current projections might notice they were somewhat reduced, and wrongly conclude that mandatory cyanide might have significant beneficial properties, since it seemed to cut mortality rates. This rather counter-intuitive result might be termed “The Mortality Rate Paradox.”
However, if at any point, the mandatory-cyanide policy were discontinued, that particular year would see a remarkable *drop* in the annual death rate, followed by smaller changes in subsequent years, until eventually a new age-mortality equilibrium was established. Thus, the only significant signals of a mandatory cyanide policy found in the annual mortality rates would come at the beginning and at the end of the policy.
Obviously, Vioxx did not remotely have the lethality of cyanide, nor was its use universal among the elderly. Moreover, any Vioxx-related mortality shifts were substantially masked by much larger directional mortality trends due to the aging of the population, improvements in life-saving and other medical technology, and all sorts of other factors. Distinguishing signal from noise is not as trivial as examining the slope of a curve.
But it does seem a bit intriguing that the mortality-curve for Americans 65+ followed a very similar trajectory to that of the extreme thought-experiment: a sharp rise in the year of introduction, a few years of relative stability, and then a very sharp drop in the year of recall.
Most of the Vioxx defenders put the total six-year death toll perhaps around 33,000, or roughly 6,000 additional deaths per year. But the actual shifts we find at the crucial starting and stopping points are far higher than this. For example, elderly deaths actually rose 35,000 in the year Vioxx was introduced, a figure several times larger than the average for the preceding few years, and dropped by 67,000 in the year it was withdrawn, which was similarly anomalous and remarkable, many times higher than the recent average change. Both these mortality shifts were heavily driven by the cardiovascular category.
It seems to me that a Vioxx-induced premature American death toll which was well into the hundreds of thousands is the most parsimonious explanation of these surprising mortality statistics.