Sunday, 20 May 2012

Stop attacking my heart

The idea behind "revascularising" coronary arteries is very appealing: “My blood vessels were blocked and the doctors unblocked them”. Like so many things addressed in my blog, this sounds good and seems hard to argue with, unless you look at it scientifically and ask the right questions. ...

The debate between the cardiothoracic surgeons (on the side of coronary bypass grafts: CAGs) and the interventional cardiologists (on the side of angioplasty / stenting) continues. There are dozens of clinical studies comparing these two treatments for heart disease. Everybody wants to know which one is best, but I am more interested in whether either of them is better than NOT doing them.

The best evidence tells us that there is no difference between these two methods when it comes to the chance of dying, and not much difference for anything else, except that you are more likely to need another “revascularisation” with stenting, but that would only be important if revascularisation was important in the first place. The fact that the stents block up a lot more than the CAGs didn’t seem to affect the outcome that much. Even the newer, more expensive drug-eluting stents don’t confer any advantage over the old ones when it comes to keeping you alive or preventing a future heart attack, and yet they are much less likely to block up within the first year. If the blockage is the problem, why doesn’t it matter if your revascularisation blocks up or not?

As a pragmatist, I would like to know if revascularising my blood vessels actually changes my chance of dying. I could spend some time explaining how my heart has compensated for the blocked vessel, as witnessed by my continued existence, or I could ask why the high blockage rates in these stents don’t actually seem to confer any disadvantage, but that is just stacking biological explanations against opposing explanations, a theoretical game that can go on forever. The arguments about what treatment is best and how it works distract us from asking the most important question: “Am I less likely to die if I have this procedure, compared to if I don’t?”

So what is the evidence? For “stable” heart disease (not an acute heart attack), the largest and best known study comparing stenting to NOT doing a stent showed no advantage to stenting in any of the outcomes measured: mortality, heart attack, or hospitalisation. And the most recent review of this topic came to the same conclusion.

Even for “acute coronary syndrome” (like a heart attack), a review of the randomised trials shows that there is no significant advantage in overall survival over 5 years for patients having routine invasive angiography / stenting.

Recently, the American College of Cardiology has put stenting on its Top 5 list of tests and procedures whose necessity should be questioned. So even they agree that you should question your doctor if he or she wants you to have your arteries unblocked - it is not as simple as it sounds.

Addit 5 Nov 2012: This study shows how cardiologists will choose to stent patients, even when they know there is no clinical benefit, for all the wrong reasons (just in case, medicolegal, theory, relieve anxiety, avoid regrets etc).
Addit 13 Nov 2012: Overview of angioplasty debate here.


  1. Hi Dr. S - as a heart attack survivor, I too am concerned about the "stent happy" cardiologist like Mark Midei et al whose overuse of this expensive treatment option has given stents a black eye (and interventionalists big headaches!)

    But a slight addition to your post is in order here: the 2010 J Am Coll Cardiology review you mention in fact concluded:

    "Routine invasive strategy reduces long-term rates of cardiovascular death or MI and the largest absolute effect in seen in higher-risk patients."

    This is an important difference for those of us high-risk patients who survived ACS/MI and were subsequently stented.

    1. Thanks CZ,
      The conclusion you refer to is correct, but my question is: Is that the conclusion you are interested in? Despite the reduction in CARDIOVASCULAR death or MI in high risk patients, there was still no significant difference in the OVERALL deaths.
      This is a common finding in many studies. For example, cancer studies talk about disease-specific mortality often being reduced with treatment X, but overall mortality is often the same.
      I always look at overall mortality in studies of treatments that are meant to reduce mortality. Measures of subgroup mortality can be spurious.

    2. "I always look at overall mortality in studies of treatments that are meant to reduce mortality."

      That's because you've likely never survived a heart attack.

      Those who do, trust me on this, are VERY interested in future CV death and MI risks. True, I could get hit by a bus tomorrow or develop a brain tumour or die of sepsis after a botched knee surgery, but my odds of having a subsequent MI are what actually keep me awake at night. . .

    3. Thanks for providing a patient perspective, and I understand the important role of peace of mind, but I am not talking about being hit by a bus, I am talking about mortality associated with the intervention.
      A good example is breast cancer screening, where large randomised studies of mammography (over 600,000 women) have shown a significant reduction in breast cancer mortality, but because the overall mortality is the same, it means that those who underwent mammography were more likely to die of other causes. There was real harm for the thousands of women who had false positive diagnoses and subsequently underwent surgery, radiotherapy, chemotherapy etc.
      I do not want to ignore the reduction in cardiovascular death, but I still think that what epidemiologists call 'fact of death' trumps 'cause of death' every time.

  2. I think one possible explanation for the failure to reduce overall mortality in so many of these disparate studies (cardiovascular, oncology etc.) is that death is an overdetermined variable. There are many possible causes of death all competing with each other with an overall probability that rises with age, eventually exceding 1.0.

    Lets say a patients additive mortality probability is 1.2 over 10 years of which cardiac causes contribute 0.5. Reducing the later by 40% with some intervention still leaves a probability of death at 1.0 at 10 years ie besides 'benefiting' the patient will still die.
    This principle also explains why, even if all cancers were to be cured overnight, the breakthrough would add only 3-4 years to average life expectancy.

    1. Thanks Secuti, and this problem is particularly noticeable in conditions affecting older people, like interventions to reduce mortality in hip fracture patients, as the average age is 80 and the 12 month mortality is about 25%. Disease specific mortality is, well, more specific, but overall mortality is more important, and can uncover harms from the interventions.
      Your comment highlights another one of my recurring points, that the perception of benefit from interventions is much greater than the reality. Many people think that cancer treatments and screening and stents save lives - that if you are still alive one year later, it is due to the intervention, and that you would not be alive unless you had received the treatment.

    2. There is pretty good evidence, however, that doing a PCI early is better than doing one late:

      (Of course, there's also a host of potential confounders around the circumstances under which one might be done early or late, e.g. general quality of care).

    3. Thanks Anon,
      Thanks for the link. Agree with confounding issues, as that study is observational. Timing issues have been addressed in other areas and can be a bit confusing (in establishing cause and effect), and associations seen in observational studies can disappear on adjusted analyses or later randomised trials.

  3. One of the first randomized studies that compared medical therapy alone with coronary bypass surgery in stable CAD patients was the Coronary Artery Surgery Study (CASS) trial, published in 1983. In this study, 780 CAD patients were randomized to one of the two strategies and followed for 5 years. Interestingly, in this study, the average annual mortality rate for patients assigned to medical therapy was 1.6% and to surgery 1.1% (P = 0.34). Analyzing only the patients with an ejection fraction ≥ 0.50 (75% of the entire population of the trial), those assigned to medical therapy had annual mortality rates of 1.1%, 0.6%, and 1.2%, respectively, for single-, double-, and triple-vessel disease. Patients with an ejection fraction ≥ 0.50 assigned to surgery had similar mortality rates 0.8%, 0.8%, and 1.2%, respectively, for single-, double-, and triple-vessel disease. There were no statistical differences between the two treatment strategies.