This (long) post is about how Pharma marketing can play a not-unimportant but still subsidiary role in pushing a misleading health message onto both the medical profession and the public. I focus here on an issue that I haven't discussed that much previously, stroke prevention and treatment.
Let me summarize what many of my evidence-based friends give as the recent history of stroke treatment in the US. It's a frankly biased account but at the end I'll provide some specific references that bolster the case. Then I want to assess the relative responsibilities of various parties for putting this message out there.
Several studies were done in the early 1990s on treatment of stroke with clot-buster drugs--a cheaper older drug, streptokinase, and an expensive new drug, tissue plasminogen activator (tPA). As about 3/4 of strokes are caused by clots and not by bleeding into brain tissue, it seemed reasonable to try this approach, especially as similar studies had proven these drugs helpful in many cases of heart attack, similarly caused by clots. (Aggressive neurologists were starting to mount the campaign that we should rename stroke as "brain attack" to emphasize both the similarities with heart attack, and the need for specialized units to treat stroke akin to coronary care units.)
Most of these studies were discouraging. The clot-buster drugs either provided no benefit or else caused too many new brain bleeds. However, in 1995, one major trial, the NINDS study (funded by NIH), came out showing a slight benefit for tPA. The benefit was that patients given tPA within 3 hours of onset of stroke symptoms ended up, at 3 months out, having fewer disabling neurologic symptoms than the placebo group. And the price paid for this benefit was an increased number of serious bleeds in those getting tPA.
Based on this one slightly positive study, a number of interesting things happened that we must, in the end, account for. One might have thought that because of the slim nature of the evidence for clot-busting, the main lesson from NINDS would have been the need to do more extensive studies and to replicate the findings if possible. But, over the next decade, this happened not at all. Very recently, a new study has been published which claims again to be slightly in favor of tPA therapy. But for most of the interval, investigators were content to simply rehash and slice-and-dice the NINDS data. It was almost as if, having finally found one study that seemed to support tPA, both the manufacturer and the neurologist community were afraid to do a new trial that might come out differently.
Next, a new set of policies were developed out of the NINDS results. Rallying around the motto, "time is brain," neurologists browbeat their colleagues to create emergency care systems to assure that any patient with a new onset of stroke could be delivered to a specialized stroke treatment center within the magic three-hour window, so as to allow the safe use of tPA. These policies downplayed another important limitation of NINDS--the fact that so many exclusion criteria were applied, to eliminate all those patients at higher risk of bleeding from the treatment, that in the end, a very small percentage of all stroke patients would be candidates for the treatment. Ideally, maybe 1 in 6 stroke patients might meet all the qualifying criteria; but in real-world settings, it is not unusual for only 1% of patients to qualify. A relative minority of physicians have questioned the cost-effectiveness of creating an entire stroke treatment system around such bizarre and unfavorable numbers.
Two recent publications indicate a sense of how far we have come in treating stroke as a result of all this activity. (Hat tip to Rick Bukata and Jerry Hoffman, for their invaluable Primary Care Medical Abstracts audio series, in identifying and commenting on these studies.) Kleinig et al. from Australia identified 259 consecutive patients admitted to a specialized stroke unit, and did an intensive analysis of them. They calculated that, of the 183 patients who ended up diagnosed with disabling or fatal stroke, 135 had a risk factor that was being suboptimally managed. Based on standard criteria, they concluded that 70 of these strokes were preventable, had relatively simple and cheap measures such as controlliong high blood pressure been employed. By contrast, they took the most favorable existing statistics for tPA treatment of stroke (such as calculating that 15% of patients rather than 1% would end up being tPA candidates), and concluded that if tPA performed as it did under the most favorable circumstances, at most, it could have prevented disability in 4 of the 183 strokes. Kleinig and colleagues put forth this difference (benefiting 4 vs 70 patients) as a hint that maybe we have got it all wrong, and that the primary focus of so-called "stroke units" ought to be on prevention using known and cheap interventions, and not rushing patients to the center to get tPA after the fact. At any rate, it seems to huge amount of resources invested in aggressive stroke treatment might have been better employed.
That in turn raises the question of how tPA actually performs, given that the advantage noted by Kleinig et al., 4 out of 183 patients with some improvement, represents the best-case scenario, and is far from the usual outcome in community practice today. That's the focus of the second article by Jerome R. Hoffman and David L. Schriger of UCLA. With considerable effort, they managed to get hold of the complete raw data from NINDS, and the article (which on line runs 43 pages) represents their attempt to display the outcome data from NINDS graphically in every possible way. Hoffman and Schriger invite the reader to look at the graphs and draw their own conclusions.
The conclusions that the two believe that a fair-minded reader will come to are, first, that all the supposed advantage of tPA was most likely due to the fact that fewer subjects in the tPA group had severe stokes, compared to the placebo group. They point out that if you compare how patients with severe, moderate, and mild strokes did on tPA vs. placebo, within each category the outcomes were nearly identical. But, because there were fewer severe and more moderate and mild strokes in the tPA group, the overall outcomes were better with tPA administration. So the apparent success of tPA in NINDS ultimately reflects a failure of adequate randomization rather than the drug effect.
The second conclusion that Hoffman and Schriger believe we'll arrive at from their graphs is to examine what happens when tPA (or placebo for that matter) is given during each possible time interval after the stroke. According to the "time is brain" theory, those lucky enough to get tPA only 70 minutes post-stroke (as early as anyone managed to give it) would have done better than those who only got it 3 hours later. But the raw data of NINDS reveals no difference at all based on time to medication. In short, "time is brain" received no empirical support from NINDS.
OK, so I'll conclude that these two recent papers, along with a lot of criticism of NINDS already published elsewhere, goes to raise considerable doubts as to the wisdom of focusing on post-stroke clot-busting drugs as a major tool in preventing the worst consequences of stroke--and that this focus may well have diverted us from things that really would work like better stroke prevention. Now, if you have read this blog regularly, you are waiting for me to claim that this must be the result of over-the-top marketing of tPA.
My conclusion is that drug marketing has played some role in all this, probably most specifically, the failure to do adequate replicatory trials after NINDS. But it would be unfair to industry to attribute all the blame in that direction. The other group that must bear responsibility is the neurology community, the folks that my primary care colleagues sometimes refer to as the "brain attack mafia," who have been beating the drums for stroke treatment centers, faster EMT transport of stroke patients, and the "time is brain" theory ever since the ink dried on the first NINDS publication. These folks, so far as I can tell, were to some extent motivated by jealousy of their friends the cardiologists--who get to make really big bucks by sticking thin tubes into people having heart attacks and doing nifty procedures and injecting nifty drugs. But much more, these neurologists have been discouraged by the helpless feeling of watching stroke patients either get better or not, without having a "magic bullet" intervention that you believe will really help. So that when something came along and there was the tiniest smidgeon of evidence that it might be the magic bullet, these folks all leapt onto that bandwagon. The hope of finally being able to help a group of patients that medicine had previously been unable to do much for seemed to be the big motive here.
Pharma marketing can do a certain amount of damage to rational medical care on its own. It can do much more damage when it joins forces with physician over-exuberance, however benignly intended.
Kleinig TJ, Kimber TE, Thompson PD. Stroke prevention and stroke thrombolysis: quantifying the potential benefits of best practice therapies. Medical Journal of Australia 190:678-682, 2009.
Hoffman JR, Schriger DL. A graphic reanalysis of the NINDS trial. Annals of Emergency Medicine 54:329-336, 2009.