Coronary Artery Bypass Graft Surgery (Heart Bypass) for Preventing Death over Ten Years

25 for prevented death

Benefits in NNT

1 in 25 were helped (prevented death)
1 in 10-14 were helped (prevented non-fatal heart attack)
96% saw no benefit
4% were helped by preventing death
7-10% were helped by preventing a non-fatal heart attack

Harms in NNT

1 in 83 were harmed (death)
1 in 100 were harmed (stroke)
1 in 43 were harmed (kidney failure)
1 in 28 were harmed (re-operation)
1 in 14 were harmed (extended life support)
1 in 3-5 were harmed (cognitive decline)
>30% were harmed by the operation
1.2% were harmed by dying
1% were harmed by having a stroke
2.3% were harmed by kidney failure
3.6% were harmed by requiring a re-operation
7% were harmed by requiring more than 1 day of life support
>20% were harmed by cognitive decline
View As:


Efficacy Endpoints


Harm Endpoints

Death, Permanent Stroke, Kidney failure, Re-operation, Extended life support, Cognitive decline


Coronary artery bypass surgery is a procedure in which a person’s redundant blood vessels are ‘harvested’ and used to bypass occluded coronary vessels, providing blood flow to heart muscle that, according to symptoms and stress tests, may be needy. In the United States the operation is performed hundreds of thousands of times each year. This review summarizes data from trials comparing surgery to medical (nonsurgical) treatment for patients whose coronary disease is ‘stable’, i.e. surgery is not an emergency.

The data summarized here are from one meta-analysis and review including three medium-sized trials and four smaller trials (n=2649) performed in the 1970s, and two trials performed more recently.

The authors of the meta-analysis report a 4% survival benefit to surgery after ten years. This number is much more complicated than it seems. First of all, this factors in the deaths of people that occur in the peri-operative period, suggesting that at ten years after the start of study 4% more people were alive in the surgical group than the medicines group. However, those who die peri-operatively (up to 5% in the early studies and still over 1% for bypass surgery today) have suffered a much greater loss of life-years than the typical gains seen for those who have surgery.

Neither of the two more recent trials demonstrated a mortality benefit over medical therapy. It is possible, however, that a benefit similar to the above numbers existed but that the size of the trials (n=406, n=763) was too small to detect that difference.

Importantly, the potential harms of surgery are considerable. Perioperative mortality is high with bypass surgery. Recent (2008-2012) national registry data suggest that mortality caused by the operation was 1.2% (NNH 83) for ‘stable’ patients, permanent stroke was 1% (NNH 100), kidney failure was 2.3% (NNH 43), and the need for re-operation was 3.6% (NNH 28). Extended life support (being on a ventilator for more than one day) was 7% (NNH 14).1 Cognitive decline is also common after major surgery2, and perhaps more common after cardiac bypass3, though numbers range and, when compared to healthy controls who do not have surgery, this has been small in some reports.4 A reasonable and credible estimate would be that in the days and months after surgery 20-30% of patients experience some cognitive decline, and a smaller, undefined proportion does not fully recover.


These data are surprisingly limited. Given how common this major and perilous surgery is the number of subjects and the lack of diversity in available studies makes valid inference troublesome.

Subgroup analyses
Some subgroup analyses from these trials have been widely accepted.5 The best example is that patients with ‘left main’ coronary artery disease seem to have fared best with surgery, seeing absolute benefits of over 20% at five years (NNT 4-5) and of 12% at ten years (NNT 8). Those with ‘triple-vessel’ disease also saw a 7% five-year benefit (NNT 15) and a 4% ten-year benefit (NNT 25). Single- and double-vessel disease patients did not see a benefit.

Unfortunately, despite broad use of the ‘left main’ distinction for surgery today, this group comprised 150 subjects across trials, making conclusions shaky. The triple-vessel group was 1341 patients—even this is too small to make definitive statements. Nonetheless, with these data in hand a reasonable best guess is that patients with left main disease can experience a mortality benefit, and there may be a small (NNT 25) benefit for those with triple-vessel disease.

Medical therapy changes
Medical therapy, including antiplatelet agents, aggressive blood pressure control, Mediterranean Diet, cholesterol-lowering drugs, and lifestyle changes, is much more effective today and few if any of these options were used in the control groups of the early bypass surgery trials.

Nonfatal MI and symptom reduction
There was no detectable difference in heart attacks between surgery and medical therapy in the meta-analysis, though the data were inconsistently recorded. In the MASS-II trial, a recent study of >600 patients, there was a roughly 10% reduction in heart attacks at ten years (NNT 10).6The recent BARI2D trial compared intensive modern medical therapy to bypass surgery in 763 diabetics, showing a 7% reduction (NNT 14) at five years.7 We have included these estimates in our overall NNT, with recognition that seven earlier trials were not able to find benefits.

Symptom reduction, however, is not a reliable outcome from these trials for a number of reasons. The most important is that sham surgery has frequently been as effective as real surgery for relieving coronary symptoms.8 9 No sham procedures were performed in these trials.

Deciding about bypass surgery
Overall, the decision about whether or not to undergo bypass surgery is complicated, and it is worth noting that it seemed to be younger patients who gained the most mortality benefit in early studies, where mean age of enrollment was 50. The two later studies found no mortality benefit, with mean ages of 60 and 63.

Moreover, for those considering evaluation for coronary disease it is critical to remember that angiography is the only current path to determine indications for surgery. Angiography carries a roughly 1-2% risk of major harms, and roughly 3% of patients typically will be found to have ‘left main’ coronary disease on angiography. This means that 3% of angiography patients have the potential for a 20% (best-case) benefit at five years. This amounts to a 0.6% chance of mortality benefit (NNT 167), but a five times greater risk of major harm including stroke, kidney failure, major bleeding, cognitive decline, etc., (NNH 33) from the angiography and surgery combined. For most, therefore, and particularly those not suspected of ‘left main’ disease, this pathway means a tiny chance of potential benefit and a much greater chance of harm.

Perhaps in young patients already determined to have ‘left main’ disease the benefits, with mortality NNTs of 5-10, meet or exceed the harms. In those with triple vessel disease 96% (best case) will either die or live and the surgery will not make the difference (NNT 25).

As for nonfatal heart attacks and the 1 in 14 possibility of avoiding these through bypass surgery, patients will have to compare this to the risks of kidney failure, strokes, cognitive decline, and death (which, in aggregate, far exceed the reduction in nonfatal heart attacks).

Because of the complexity of the decision, and the small numbers of enrolled subjects, we have chosen to label this intervention ‘Yellow’, as we would like to see more, more modern, and better trials addressing the issue. The exception, as a best estimate based on very limited data, is for patients with established left main disease, particularly those that are younger, in whom the intervention may be considered ‘Green’ (benefits>harms).


David Newman, MD


July 20, 2014