Background Although running is a popular leisure-time physical activity, little is known about the long-term effects of running on mortality. The dose-response relations between running, as well as the change in running behaviors over time, and mortality remain uncertain.
Objectives We examined the associations of running with all-cause and cardiovascular mortality risks in 55,137 adults, 18 to 100 years of age (mean age 44 years).
Methods Running was assessed on a medical history questionnaire by leisure-time activity.
Results During a mean follow-up of 15 years, 3,413 all-cause and 1,217 cardiovascular deaths occurred. Approximately 24% of adults participated in running in this population. Compared with nonrunners, runners had 30% and 45% lower adjusted risks of all-cause and cardiovascular mortality, respectively, with a 3-year life expectancy benefit. In dose-response analyses, the mortality benefits in runners were similar across quintiles of running time, distance, frequency, amount, and speed, compared with nonrunners. Weekly running even <51 min, <6 miles, 1 to 2 times, <506 metabolic equivalent-minutes, or <6 miles/h was sufficient to reduce risk of mortality, compared with not running. In the analyses of change in running behaviors and mortality, persistent runners had the most significant benefits, with 29% and 50% lower risks of all-cause and cardiovascular mortality, respectively, compared with never-runners.
Conclusions Running, even 5 to 10 min/day and at slow speeds <6 miles/h, is associated with markedly reduced risks of death from all causes and cardiovascular disease. This study may motivate healthy but sedentary individuals to begin and continue running for substantial and attainable mortality benefits.
This study has been heralded as a groundbreaking study, but this is wildly inappropriate misrepresentation that has done little more than confuse those who read the myriad poor reports of this study. To give an example, a major media outlet not only conflated causation with association/correlation, but also misleadingly reported that running as little as 5 minutes per day could reduce your risk of dying from a heart attack or stroke by 45% and your risk of all-cause mortality by 30%. This coverage provided on Medscape is similarly disappointing and misleading.
ReplyDeleteLet's take a quick peek at the study.
The study was large, with 55,137 individuals for the all-cause mortality analysis and 52,941 individuals for the analysis of CVD mortality. The study was long, with the mean (interquartile range, IQR) follow-up was 14.7 years and 14.6 years for the respective analyses. Unfortunately, the study was also an observational study. Herein lies a fatal flaw in reporting that this study shows that running reduces anything. Only in very, very specific instances can cause-and-effect claims be made from observational data (however, we have seen this, e.g., cigarette smoking as an inferred cause for lung cancer). This study doesn't meet these specific criteria. The best this study can say is that the authors found an association, and even with as much as I love exercise, this study has weaknesses that threaten even the ability to claim an association. The authors determined if and how much people ran by self-reported measures (which is always fraught with potential recall bias, not to mention social desirability bias/a pseudo Hawthorne effect). The study also extended the period of exercise recall to 3 months, which is longer than is typically employed (thus potentially furthering the possible effects of recall bias). There are further limitations with internal validity and generalizability, but for the sake of the length of this post, I'll forego getting into any further details on issues with the study’s methods. Suffice it to say: There are problems. Let's get back to that impressive claim that running as little as 5 minutes a day can reduce your risk of all-cause mortality by 30% and your risk of heart attack or stroke by 45%.
The authors did do a pseudo dose-response analysis (more on that below), but interestingly, analysis of the aggregate data resulted in the aforementioned associated reductions in relative risk. So, when looking at all the runners compared to the non-runners, the authors found a hazard ratio (HR) of 0.70 for all-cause mortality (thus an associated 30% decreased relative risk) and an HR of 0.55 for cardiovascular (CV) mortality (thus an associated 45% decreased relative risk). Even though I put the word "relative" in italics and bold, it is worth stating explicitly that the much-touted metrics have been the relative figures, not the absolute figures. I'll get to that in just a bit.
So, what about that pseudo dose-response thing I was talking about earlier?
Below, I've included data from Table 3 from the study (not all the data, and I tried to recreate the table as a table, but formatting issues precluded me from doing so):
ReplyDeleteDeath rates for all-cause and cardiovascular mortality by quintile of weekly running time
***Note: All death rates are reported as death rate per 10,000 person-years***
Non-runners
Death rate for all-cause mortality: 45.9
Death rate for CV mortality: 17.8
Quintile of running time, minutes/week
Quintile 1 (< 51 minutes/week)
Death rate for all-cause mortality: 31.7
Death rate for CV mortality: 8.0
Quintile 2 (51–80 minutes/week)
Death rate for all-cause mortality: 29.7
Death rate for CV mortality: 9.0
Quintile 3 (81–119 minutes/week)
Death rate for all-cause mortality: 29.8
Death rate for CV mortality: 10.3
Quintile 4 (120–175 minutes/week)
Death rate for all-cause mortality: 31.5
Death rate for CV mortality: 9.1
Quintile 5 (≥ 176 minutes/week)
Death rate for all-cause mortality: 33.8
Death rate for CV mortality: 11.6
***Again, please note that all death rates are reported as death rate per 10,000 person-years***
So, comparing folks who run < 51 minutes/week to folks who don’t run at all, the study found all-cause death rates of 31.7 per 10,000 person-years versus 45.9 per 10,000 person-years, which yields an absolute risk difference of 14.2 per 10,000 person-years for all-cause mortality (45.9 – 31.7 = 14.2). However, quick math shows that this risk difference would equate to an HR of approximately 0.69 (31.7/45.9 or 1 – [14.2/4.59]), which would translate into an associated 31% reduction in relative risk (14.2/45.9).
Repeating the same for CV mortality, one finds an absolute risk difference of 9.8 per 10,000 person-years and an HR of approximately 0.45, which would translate into an associated 55% reduction in relative risk.
These are the best-case-scenario numbers for this study (other analyses yield more modest associated reductions). Also, note I spoke of absolute risk difference instead of absolute risk reduction. This is because the data are observational, so one should talk about absolute risk difference by convention. I like to think that’s the convention in order to remind folks we’re talking about observing differences, not claiming something reduces something (a cause-and-effect statement).
That sure puts things in perspective, doesn't it? Note how different the absolute risk differences are compared to their associated HRs. (Also note how incredibly close the HRs for the aggregate data are to the HRs for the quintile that reported running < 51 minute/week.)
ReplyDeleteFor those who look closely at the data I included from Table 3, you’ll also note something interesting that may be nothing more than a “data fluke” or possibly an indicator of the likely-present biases in the data: increasing the amount of running per week isn't consistently associated with a commensurately lower risk of death from all causes or from cardiovascular causes. For instance, compare running 51-80 minutes/week to running 120-175 minutes/week and death rates for all-cause mortality. No statistical tests were conducted on these differences, though. Again, I am most certainly not suggesting this means anything other than the two aforementioned possibilities, but it’s a neat thing to notice, and again, even if statistical tests had been conducted on these differences, I’d still advise against making any strong conclusions from this study.
To make matters worse, one of the study’s authors is on camera saying: “Running one to two times per week, uh, running less than an hour per week, was still producing very substantial reductions in total and cardiovascular mortality.” (See his statement starting 49 seconds into the video at the following URL - http://www.nbcnews.com/nightly-news/running-5-minutes-day-could-reduce-heart-attack-risk-n167046)
That’s highly, highly unfortunate. To me, that’s an unmistakable cause-and-effect statement. Again, though, the study he co-authored cannot support such claims.
Am I saying exercise isn't good? You'd be way off-base if you think the intent of this post is to suggest anything of the sort. The totality of the evidence in the literature does obviously suggest that exercise is beneficial for our health, and you needn't be an Ironman competitor, Olympian-level athlete, or anything like that to glean such benefits, but I also certainly wouldn't be comfortable saying a mere 5 minutes will do the trick, either. The point of this admittedly-longer-than-originally-intended post is to point out serious problems with the study and how folks - including at least one of the study's authors - are mishandling it.
MM Physician assistant