Monday, October 9, 2017

Running and Your Heart, Part IV: Running and Mortality


At long last, here is the final post in the "Running and Your Heart" series.  When we started I thought I'd finish this series in about 4-6 weeks.  Now it looks like we're pushing past 6 months.  Hopefully it will have been worth the wait.

So over the past several months we've (rather infrequently) investigated the relationship between long-term endurance exercise and the heart.  We've discussed normal heart function, cardiac adaptations to exercise, abnormalities that can arise from these adaptations, and the impact of marathon and ultramarathon running on coronary artery disease.  (As a brief aside, one of the theories that I discussed in the last post--that the increased coronary calcification seen in marathon runners is quite possibly hard, stable plaque that is less likely to rupture and cause actual problems--has since been supported by some recent studies.  Alex Hutchinson, whose work is generally spot-on, has a good summary of these articles in Runners' World.)  What I've tried to stress is that we should not be alarmist about these issues, but that we should not be naive either in thinking that our running makes us immune from heart disease.  Rather, we need to be aware of the potential problems that can arise from long-term training and be able to address these possibilities with our physicians in a responsible way.

OK, that's all well and good.  But when we start seeing articles with titles like "Fast Running is as Deadly as Sitting on the Couch, Scientists Find" and "Excessive Running Could Kill You", it's natural to feel a bit concerned.  We know running, on the whole, is good for us.  Since Jim Fixx gave us his seminal work The Complete Book of Running in 1977, detailing how running saved him from what seemed to be his genetically destined early cardiac death, we've taken it on faith that diligent training leads to longer lifespans.  But with all the studies in the past decade that have been hinting at correlations between long-term marathon running and paradoxical heart disease, is it possible that we're taking things too far?  Have we reached the point where all this running is actually shortening our lifespans?  (And wait a second, didn't Jim Fixx die pretty young after all?)

Much of the recent concern over the possibility that too much running might actually be bad for you centers on a couple of ideas.  One is the relationship between long-term marathon running and coronary artery disease, which we discussed in detail last time.  Just to sum up where I stand on this: I think it's pretty unambiguous that people who train vigorously for marathon-and-longer distance events for many years do have a higher incidence of coronary artery calcification than those who do not, and that the cardio-protective benefits of regular aerobic exercise require much less mileage and less intensity than many of us (including myself) are actually doing.  However, as I pointed out in the links above, not all coronary calcification is created equal; while you'd certainly prefer less calcification than more, and no calcification to any, the calcific plaques demonstrated in asymptomatic long-term runners are not the same in terms of composition (and possibly long-term risk) as those we'd associate with smoking, uncontrolled hypertension, diabetes, or other native disease states.  Even if we assumed that we may be at higher risk of suffering a serious cardiac-related event as a result of strenuous running (by no means a reliable assumption), I don't think the data support the conclusion that this risk outweighs the mortality benefit of the decreased incidence of high blood pressure, diabetes, and cancer that runners consistently demonstrate over their more sedentary peers.

The other idea that has received a lot of publicity in recent years is the so-called "U-shaped mortality distribution."  This concept is based largely on the work of James O'Keefe and other researchers involved in the Copenhagen City Heart Study, as well as Duck-chul Lee, who in 2012 presented a rather controversial abstract at the American College of Sports Medicine conference.  According to their research, when plotting mortality (the dependent variable) on the y-axis against mileage (the independent variable) on the x-axis, the data shows the highest mortality values at the lowest and highest ends of the exercise spectrum (i.e., a "U-shaped" distribution):

The "U-shaped" mortality curve.
figure: American College of Cardiology 

In other words, those runners logging much less mileage--as little as a mile a day in some cases, and certainly less than 20 miles a week--saw the greatest benefit in mortality, while those running more mileage saw little to no mortality benefit at all!

These are the studies that have prompted most of the terrifying headlines you've read in recent years, and these are the studies I want to talk about today.  Not to argue with their data, but to try to help us understand why some of the authors'--and the media's--conclusions are not necessarily as dire as we've been led to believe.

Lee et. al. demonstrated a 20% reduced risk of mortality for runners vs. non-runners over a 15-year follow-up period--great news!  However, the study appeared to show that runners averaging more than 20 miles per week had not only higher rates of mortality than those running less, but that their mortality rates approached those of sedentary, non-running peers.  However, these findings were based on the researchers' adjustments for various conditions, including body mass index (BMI), smoking, diabetes, and hypertension.  What does this mean?  When comparing different groups of people, researchers can run into a problem with what are called confounding variables.  These are differences between groups that might affect what you're trying to measure.  In this case, the researchers were trying to determine the relationship between running mileage and mortality.  Of course, there are many different variables that contribute to one's mortality risk, and without accounting for these variables, it's difficult to ascertain whether differences between groups of data is due to what you're actually trying to measure (running mileage) or to something else that you're not measuring.  So the researchers performed what's called statistical correction to take these variables into account.  Simply put, they eliminated the effect of as many of these confounding variables as possible, trying to answer the question: "If all other things are equal--if all of these groups are made to be the same in terms of their rates of obesity, diabetes, blood pressure, etc.--then what effect does mileage run have on their mortality risk?"

Let's be clear: there's nothing underhanded about this.  Statistical correction is a perfectly legitimate (and in most cases necessary) part of scientific research; it's how we attempt to discern cause and effect in situations with many different variables in play.  In this particular case, however, we run into a problem.  The reason that running might have a benefit on mortality is that it makes you healthier overall.  That is to say, vigorous runners are less likely to be obese, to have high blood pressure, or to suffer from diabetes.  If you eliminate these benefits as "confounding variables," it only stands to reason that the mortality benefits of running disappear from the data as well.  The problem with this study wasn't that the authors tried to correct for confounders; it was their classification of the benefits of exercise as confounding variables in the first place.  Alex Hutchinson was all over this pretty much right away, and in 2013, cardiologist Thomas Weber pointed out the problem in the journal Heart:

"One possible explanation for the U-shaped that the authors adjust for body mass index, hypertension and hypercholesterolaemia. Running has been shown to lower those risk factors in a dose-dependent fashion with no sign of negative returns until at least 50 miles/week. Arguably, adjusting for all these factors is akin to adjusting for low-density lipoprotein (LDL) values in a study analysing the survival benefit of taking statins to treat hypercholesterolaemia. Put simply, this editorial represents a selective interpretation of the available data, at the best."

What Weber is saying is, if you were studying the impact of a drug for cholesterol on mortality, and you had two groups (one which took the drug and one which didn't), it wouldn't make any sense not to look at the differences in the cholesterol levels between these two groups--how else would you expect the drug to improve mortality if not by impacting cholesterol levels?  Similarly, if we grant that running makes us healthier because it protects against hypertension, diabetes, and obesity, then those are very likely the reasons it would have a mortality benefit; removing those effects from the analysis doesn't make sense.

Indeed, when the final paper of this study was published in 2014, the researchers eliminated the statistical correction--and the U-shaped mortality curve seemed to vanish!  Instead, the authors, now concluded,

"[R]unners across all 5 quintiles of weekly running time, even the lowest quintile of <51 minutes per week had lower risks of all-cause and CVD (ed: cardiovascular disease) mortality compared with non-runners. However, these mortality benefits were similar between lower and higher doses of weekly running time. In fact, among runners (after excluding non-runners in the analyses), there were no significant differences in hazard ratios of all-cause and CVD mortality across quintiles of weekly running time (all p-values >0.10)."

That is to say, running even a little bit lowered mortality risk, and this lower risk appeared constant regardless of the time or distance run per week. Perhaps not surprisingly, this received significantly less media attention than the earlier version of the results.

Similarly, the researchers in the Copenhagen City Heart Study reported findings that seemed to support the U-shaped mortality curve, concluding:

"We found a U-shaped association between jogging and mortality. The lowest mortality was among light joggers in relation to pace, quantity, and frequency of jogging. Moderate joggers had a significantly higher mortality rate compared with light joggers, but it was still lower than that of sedentary nonjoggers, whereas strenuous joggers had a mortality rate that was not statistically different from that of sedentary non joggers." 

and cited Lee's paper in their discussion of the results.  Again, however, these conclusions don't tell the whole story.  While this study followed nearly 1100 runners over a 12-year period, only 40 of these runners qualified as "strenuous joggers" according to the rubric of the study (running at a pace of 9:00/mile or less for at least 2.5 hours/week), and there were only two deaths among this group during the course of the follow-up period--not nearly enough of a rate to draw any meaningful conclusions.  As researcher Steve Farrell pointed out,

"Say that 2 blindfolded men ran across a busy highway and were not struck by a car. Would anyone conclude based on those two events, that it is perfectly safe for everyone to run blindfolded across a busy highway?"

So what to make of all this?  It seems pretty clear that the substantial mortality benefits of aerobic exercise are conferred even after relatively small amounts of running--which is great news for the sedentary population and light exercisers in general--and I'd agree that at some point we reach a rate of diminishing returns, where further increases in mileage or intensity don't offer any additional mortality benefit.  But where that point lies has not been clearly defined, and I think that based on what we currently know, fears of increased mortality as a result of exceeding that threshold appear unfounded.  And generally, most of us who are interested in exploring our physical limits are doing so for reasons that go beyond "living longer."  As Amby Burfoot points out,

"Many aspects of exercise and running also follow a U-curve. This is why many people believe the moderate approach is the smartest path to follow. Of course, you’ll never qualify for the Boston Marathon that way. We all have to make our choices."

Certainly we don't need to run ultramarathons experience all the health benefits of regular exercise.  But it doesn't seem like we need to fear them either.