• Who will get rich if I don’t get cancer?

    Yesterday’s Times includes an intriguing piece by Roni Caryn Rabin, “Studies Link Daily Doses of Aspirin to Reduced Risk of Cancer.” It summarizes a recent study in The Lancet study by Rothwell and colleagues: “Short-term effects of daily aspirin on cancer incidence, mortality, and non-vascular death: analysis of the time course of risks and benefits in 51 randomised controlled trials.” This article and others suggests that daily aspirin may reduce both the risk of cancer, particularly colorectal cancer, and cancer mortality.

    I’m not convinced that you should run out and buy baby aspirin just yet. Offsetting risks of stroke and GI bleeds should be considered.  An accompanying Lancet commentary by Andrew Chan and Nancy Cook notes that neither the Women’s Health Study nor the Physicians’ Health Study—two huge well-conducted studies–found protective effects. Rothwell and colleagues excluded these papers from their analysis because they involved use of aspirin every other day, rather than daily use. This still seems quite pertinent. Other questions have been raised, too.

    Whatever the specific merits of daily aspirin, the Times’ concluding paragraph provides what might be the sobering comment in the entire piece:

    Some cancer doctors commended the new research, saying said that despite the limitations of the analyses, no other long-term clinical trials of aspirin and cancer are likely to be done because of the enormous expense involved and the fact that aspirin is a cheap generic drug.

    This matter-of-fact indictment of our health care system is especially damning because it is so casually delivered, within a story that has nothing ostensibly to do with the usual health policy debate… Here we have a preventive intervention that might significantly reduce cancer mortality. Unfortunately, the medication in-question is dirt-cheap, and already widely-available. So it’s just not profitable for firms to spend the considerable sums required to nail down its effectiveness in preventing cancer.

    As Ezra Klein observed in the related context of diabetes, this problem is familiar to advocates of public health. If a firm develops some novel therapeutic to more-effectively treat colon cancer, it will make tons of money. If that same firm discovers a simple way to prevent me from getting cancer in the first place?Well, it will be much harder to capture the same financial returns. So reputable firms don’t pursue many simple innovations that might bring huge social benefits. (Too often, the gap in actual evidence is filled in your local nutrition supplement store, whose shelves are stocked with unevaluated products hawked on late-night TV. That’s another matter.)

    This isn’t the usual kind of scandal we ponder when we consider the shortcomings of big pharmaceutical companies. No data are being withheld. No sleazy marketing is undertaken. Nobody at big Pharma is behaving unethically. It really isn’t these companies’ fault. They don’t exist to maximize health outcomes or to minimize the number of life-years lost due to cancer. They exist to earn to a decent return for their shareholders, and to do so in a socially and ethically responsible manner. Their core commercial mission often, but not always, overlaps with broader social goals.

    The real problem resides with us, members of the policy community and the interested public. We rely too much on the private commercial calculations of private firms to implicitly set public health policy. And if pharmaceutical companies face systematically biased incentives to pursue treatment advances involving lucrative patentable drugs, the political economy of health care, the prestige structures of medical research, and the American cultural marketplaces include similar incentives that similarly bias our attention and rewards towards acute care and medical treatment rather than prevention or (God forbid) cost-reducing and simplifying innovations.

    Cancer patients–allied with the people and the ecosystem that cares for them—comprise a powerful, concentrated, and rightly appealing constituency. Meanwhile, beneficiaries of primary cancer prevention are a diffuse, disorganized constituency. Prevention Interventions from smoking cessation on down are worthy, generally boring non-urgent items on everyone’s list. Senators won’t lose their jobs or encounter shaming images of protesters in wheelchairs when they cut money for state smoking quit lines, flu vaccines, or HIV prevention. You won’t win the Nobel Prize in physiology or medicine for a meta-analysis of baby aspirin studies, for brilliant shoe leather epidemiology studies leading to the back-to-sleep recommendations for SIDS, or for health services research showing that we don’t need to use some costly medicine for something when much cheaper medications will do.

    I hope the new PCORI (patient-centered outcomes research institute) does better. The early political signs are mixed. In the knife fight over money to implement health reform, the prevention and public health fund was trimmed with little public opposition or debate. The fund includes some of the most cost-effective spending in the entire new law. Other features of the Affordable Care Act, ranging from the new accountable care organizations to provisions requiring free coverage of evidence-based clinical preventive services, may also provide a better balance of incentives for primary and secondary preventive care.  It will remain an uphill fight because the issue goes much deeper than that, to the basic rewards of our health care financing and research systems.

    Yesterday’s Times happen to include another story, “A Cheap Drug Is Found to Save Bleeding Victims” that gives one novel reason for hope. The piece, by Donald McNeil, Jr., discussed rather similar market failures in the domain of emergency medicine. The article concerns tranexamic acid, which is apparently quite useful to help trauma victims.

    The article begins with a familar lament:

    For months, a simple generic drug has been saving lives on America’s battlefields by slowing the bleeding of even gravely wounded soldiers.

    Even better, it is cheap. But its very inexpensiveness has slowed its entry into American emergency rooms, where it might save the lives of bleeding victims of car crashes, shootings and stabbings — up to 4,000 Americans a year, according to a recent study.

    Because there is so little profit in it, the companies that make it do not champion it.

    In this case, we appear headed to a more sensible ending, in part because American combat surgeons have noticed the drug’s success in British military field hospitals. Dr. David E. Lounsbury, co-author of a 2008 Army textbook, “War Surgery in Afghanistan and Iraq: A Series of Cases, 2003-2007,” tells the Times:

    “An old generic doesn’t have any hair-on-your-chest bravado, so we didn’t even take it to the battlefield…”

    That the British pioneered it “makes complete sense to me,” he added. “I worked in their hospitals, and they did pretty much everything we did — but much more cheaply.”

    We have much to learn from other nations—not just Britain, but many others, too—who operate less-richly-resourced, but hence more disciplined and sometimes more creative medical and public health systems. We don’t much like to learn from others. We prefer to boast: “We have the best health care system in the world.” We actually don’t. Even if we did, our arrogance would still lead us to miss opportunities to improve it.

    • Excellent post. Two further elements that dovetail with what you’ve said.

      First is yet another instance of the phenomenon you describe: pharmaceutical neglecting drugs that could do lots of good but which promise little profit. I learned of this one from my colleague Dan Hausman (http://philosophy.wisc.edu/hausman/papers/Hausman-why-not.pdf):

      “An extreme example of this is the drug eflornithine, which is a highly effective ‘miracle’ cure for sleeping sickness. Until 1999 the drug was produced by a U.S. subsidiary of Aventis, but when eflornithine proved ineffective against cancer (its intended target), Aventis stopped making the drug and gave the production license to the WHO. Only in early 2001, when stocks of the drug were almost exhausted, was the WHO able to find drug companies to manufacture it — and then only because the companies hope to make profits from marketing eflornithine as a cream that removes facial hair. Because the victims of sleeping sickness are so poor, the amount they are willing to pay for eflornithine understates its social value.”

      Second, Thomas Pogge and others have been developing and pushing the idea of adding a second sort of patent to the mix that would pay drug companies in proportion to the amount of good their drugs do in the world. The idea is the drug company would forgo exclusivity, allowing the drugs to be produced by any company and hence driving down costs, and then the company holding the patent would be paid out of a rich-nation-financed fund on the basis of the number of lives saved/QALYs gained/etc. Of course, there are measurement and many other issues here (how do you separate a drug’s impact on QALYs gained from the impact made by improved public health?). But the effort is worth highlighting in this context, I think. More on the proposal at http://www.healthimpactfund.org

    • In the article you mention something described as a market failure, i.e. a pharma company producing a second-best drug instead of a disease curing drug in order to maximize profit. Though it is your job as a TIE blogger I would ask:is this limited to pharma? Isn’t this failure inherent in many services that we receive, such as auto repair or dental work? Physicians may make decisions based on whether or not they get paid. This doesn’t have to imply that they under-perform in order to get patients coming back (not impossible, though probably rare), but perhaps that they over-perform, or request unnecessary treatments and visits.

      I view these as a part of the same problem. Would you agree? Thus, we can point to Nasty Big Corporations, who no doubt are guilty of this, but I would not limit profit maximizing to them alone.

      Also “Because there is so little profit in it, the companies that make it do not champion it.” ?

      Really? I’m not buying this argument … How much profit is there in bananas, and yet somehow I can almost always find bananas for < $1 / lb. Razor-thin margin products exist all over the place, and we all enjoy them daily.

      Farther down in the times article, a more likely cause rears its head:

      "Many companies in India and China make tranexamic acid. Pfizer, which makes an injectable form for hemophiliacs (and donated thousands of doses to the Crash-2 trial), declined to give sales figures or even discuss administering it to trauma patients because the Food and Drug Administration has not approved that use."

    • I followed the link provided to the Lancet article, and although the full article is gated, the main results are reported: out of a total sample size of 69,224 subjects, 562 of the patients on aspirin died from cancer, while 664 of the patients not on aspirin died from cancer. Just doing a crude analysis, this implies that the absolute reduction in cancer mortality was 0.3 percentage points. So I’m not so sure that I agree with you when you characterize this as “a preventive intervention that might significantly reduce cancer mortality.” Yes, the study achieved STATISTICAL significance (p = 0.008) because it had a gigantic sample size (69,224), but I’m not so sure that you can point to the empirically estimated effect size and claim that this is CLINICALLY significant.

      Here’s another way to think about this. To “nail down” the real effectiveness of this intervention as you suggest would basically require a randomized controlled trial with a huge sample, and to do that would require gazillions of dollars, with the final verdict unavailable for many years (if you want to study 20 year mortality, then you’re going to have to follow your subjects for 20 years). Let’s say that trial cost $50 million dollars to run (that would be a low estimate). Is that really the best way to spend our research dollars, tying up $50M to get a precise estimate of an intervention that might improve mortality by less than half a percentage point?

      • You are correct. The absolute mortality risks are low because these are short-term studies in relatively low-risk populations.

      • You’re correct that there is often a difference between clinical and statistical significance, but there is no question that the numbers presented in this paper are clinically significant.

        “Allocation to aspirin reduced cancer deaths (562 vs 664 deaths; odds ratio [OR] 0·85, 95% CI 0·76—0·96, p=0·008; 34 trials, 69 224 participants), particularly from 5 years onwards (92 vs 145; OR 0·63, 95% CI 0·49—0·82, p=0·0005)”

        This suggest that the odds of cancer death are reduced by 15% in the aspirin group overall, while cancer death is reduced by 37% in the long term group. How can something that could prevent a third of cancer deaths not be clinical significant!

        I’d also like to point out that this is a meta-analysis of clinical trails, which are the gold standard in the field. Yes, it is interesting that we don’t see the same results in some observational studies..but these studies are more prone to bias and not as trusted by statisticians.

        Honestly I believe a long term clinical trial would be warranted based on these numbers and it wouldn’t surprise me if the gov’t eventually does fund it despite its large pricetag. At the very least it’ll be closely studied in long term observational studies from now on. Personally I’m optimistic – there is something about cox-2 inhibitors…we can learn more about the biology and other drug companies could hopefully develop other similar drugs that could hit related pathways and hopefully see similar or improved results.

        • Thinking about this a little more – I should emphasize that odds ratios are not a measure of absolute risk. So yes, if we were to implement aspirin as a chemopreventative agent in the general population there would be many people taking aspirin who would never develop or die from cancer. This is similar to the use of statins for CVD. But again if aspirin truly reduced the odds of cancer mortality by a third then it’d be widely recommended. But if not maybe the intervention should only be recommended for those at high risk of cancer (those with BRCA1/2, family history, and other risk factors?)

          Also the editorial in Lancet also mentions this:
          “Nonetheless, as we await data from additional trials (NCT01038583 and NCT00501059) and longer-term follow-up of the WHS and PHS, Rothwell and colleagues’ impressive collection of data moves us another step closer to broadening recommendations for aspirin use.”

          So there are more clinical trials ongoing (5 year follow-up) along with more data from the WHS and PHS which are longer term randomized trials (sorry I initially thought they were observational studies)…just with a different timing of the doses. So regardless the gov’t and other organizations are funding long term clinical trials to look at aspirin for cancer and CVD. Hopefully we’ll have more concrete data soon!

          • Dear danbeachler,

            Thank you for your thoughtful remarks.

            It’s often unclear exactly how to quantify the intuitive notion of “risk” — should we count relative risk or absolute risk? The problem comes when we deal with a small baseline risk, because then tiny changes in absolute risk can result in substantial swings in relative risk. And indeed that’s exactly the situation that we have with this study. My personal preference is that, since this is essentially a preventative measure that must be applied across the population, the absolute risk difference is the relevant statistic. But that’s just my subjective opinion, although I think I could make a strong argument in favor of it (for instance, overall mortality rates will not be changed substantially if the absolute difference is small).

            Yes, clinical trials are indeed the gold standard for medical research. But a meta-analysis is much more questionable. Of course it’s a useful tool, and sometimes (like this case) it’s the only tool available. But unfortunately it isn’t as reliable as we might like. Conversely, observational cohort studies aren’t perfect either, but again they are useful tools and sometimes the only one available. I realize that the results from the Women’s and Physicians’ Health Studies are discouraging, but that’s the way research works, and it’s not fair to discount them just because they don’t tell us what we want to hear. In any case, all the studies agree that the absolute effect size is small.

            The real issue for debate here isn’t whether or not aspirin can prevent cancer. Instead, Harold’s point in this post is that because aspirin is now a generic medication, no private company is willing to invest the enormous sums of money that would be needed to “nail down” (Harold’s words) the effectiveness of such a study. When it comes to “nailing down” this effect, the absolute effect sizes really are important, and such a study would be very, very expensive to conduct. So this was my main point — does it really it make sense to spend a huge amount of money on something that everyone agrees would have a small absolute effect?

            • Thanks for the reply. I agree that the absolute risk is what we should care about, but I don’t believe the absolute effect size is small in this case. The effect size you calculated is low but that was more because of the short follow-up time in these trials.

              There are 500K cancer deaths in the US per year. If the odds of cancers is reduced by 1/3, it would mean that we’re preventing over 150K cancer deaths a year. Now the question is how many people would be harmed if everyone in the general population took aspirin (say 100 million people). Would it be larger than 150K per year? Even if 1% had major negative side effects that would be 1 million people. So maybe…but it’s worth further study.

              Think about what we currently recommend for the general population – mammograms, colonoscopies, etc. They wouldn’t come close to reducing the overall cancer mortality by 1/3. Honestly, if Aspirin really reduces the risk of the two biggest causes of mortality – heart disease and cancer…we have to consider it. I think you’d be right if we were talking about a reduction in one particular cancer, but these results are among ALL cancers. If taking aspirin would reduce cancer mortality by 1/3, it would be almost as good as tobacco cessation.

              Regardless there are clinical trials going on…which are paid for by the gov’t and bayer..so hopefully we’ll know soon enough.

            • Dear danbleacher —

              I agree that if the observed odds ratio were to hold for the long term, that the resulting absolute effect size would be considerable. Of course, that’s a very big if. But the question here isn’t really about aspirin and cancer. It’s about Harold’s original post, in which he bemoaned the lack of interest from the private sector in funding a follow-up study, In Harold’s words, “So it’s just not profitable for firms to spend the considerable sums required to nail down its effectiveness in preventing cancer.” The problem here is that the resources that would be required to “nail down” this issue would be ENORMOUS: it wouldn’t be a little clinical trial with a few hundred people over a short period of time (even that can run into the millions of dollars). This is especially the case if you want to study the long-term effects. I think this is perhaps where you and I really disagree: what the required resources would be in order to actually improve on (“nail down”) the results of the meta-analysis.

              Here’s another way to think about this. We have a meta-analysis with nearly 70,000 people which returned some highly statistically significant results. If that’s not good enough for us, what would we accept as reasonable evidence? Then, when we determine what that would be, we have to ask: we’re spending a lot of money here, just to get a more precise estimate of the effect size. Would it make more sense for us to just accept the result from the meta-analysis and then take that money and spend it on other investigations?

            • Dear danbleacher —

              Incidentally, this issue is hardly restricted to aspirin and cancer. There will be a dramatic drop in the research on Lipitor now that it’s gone generic, and this is true of many other treatments as well. I remember having a conversation a few years ago with a drug company salesman, and he said it was good that I was working on research with psychiatrists. I naively thought that this was because there were all sorts of exciting new results in psychiatry, but instead the real reason was that many of the big psych drugs were still on patent and so companies were still funding research on them, while for instance there had been a big drop-off in interest in heart disease because many of the treatments had by then gone off patent. So Harold’s issue in his post has been around for a while, and there are lots and lots of examples that he could choose.

            • You make a reasonable point. I think the big unknown here is the effect of long term aspirin on survival. So the evidence base would be improved if we had a trial that lasted 15 or 20 years. Maybe that would be too expensive, and if so it would probably need to be further investigated with observational studies. I’d still probably argue that money would be best suited to this long expensive trial than to other potential projects, but there are reasonable arguments on both sides.

    • You know, maybe the government could grant prizes to researchers who uncover low-tech solutions like this.

    • I have to add this great quote from TR, given your last paragraph:

      “It is a mistake for any nation to merely copy another; but it is even a greater mistake, it is a proof of weakness in any nation, not to be anxious to learn from one another and willing and able to adapt that learning to the new national conditions and make it fruitful and productive therein.”

    • Study has been underway since 1971 when the president from the USA declared war on cancer and allocated funds to locate a cure. Nevertheless 40 years have now pasted and also the best they have come up with is a selection of drugs that are marginally more efficient but nonetheless do not cure but may give somebody a little more time. These drugs are very expensive costing about 4000 US dollars a month and 1 of them, herceptin used for a type of breast cancer was said to become 0.six percent better than some of the earlier drugs.