Jameshow
Veteran
Since you’re all such fans of evidence, let’s have a look at the newest, largest, and most comprehensive study there’s ever been.
Garcia et al published a new meta-study as recently as 3 months ago, and it’s huge: 196 articles, 94 cohorts with >30 million participants, 163 million person-years of data, and 810,000 deaths. That’s 17 times as many person-years, and 7 times as many deaths as the previous biggest study, which has for the first time enabled them to sub-divide by disease, and produce dose-response curves. Any studies with fewer than 10,000 participants were excluded to reduce the risk of them degrading the accuracy. (A meta-study is a collation of all the relevant studies so that the results can be pooled to increase statistical significance, and avoid cherry-picking.)
Taking at the all-cause mortality for example, their results look like this:
View attachment 685106
As you can see, as exercise level increases, the risk of death drops markedly at first, and then begins to level off. The zone shaded in dark is the lowest 75% of person-years of data, and the vertical dotted lines mark the 37.5th and 75th centiles. The paler region beyond has been presumed, and constrained, to be linear at the same slope as the 75th centile.
Exercise dose is measured in marginal MET hours per week. METs are a unit for measuring metabolic rate in kcals per hour per kg of bodyweight, so MET hours per week are effectively kcals per week per kg. Marginal means that they are measuring activity over and above sedentary levels: that is, the increase imposed by activity.
In more familiar terms of cycling rather than marginal METhr/wk, if we take a typical cycling speed of 12mph, the metabolic rate from the ACSM Compendium is about 7.4METs. So as the rate for sedentary behaviour is 1MET, the marginal rate is 6.4METs, therefore in one hour you complete 12 miles for 6.4 mMET hours. That gives a ratio of 12/6.4 = 1.875 miles/mMEThr, or for the sake of round numbers, say about two.
We can then calibrate the same graph above in terms of weekly mileage on the bike assuming 12mph (Scale appended in red):
View attachment 685107
So as you can see, anyone cycling more than about 25 miles a week is off the end of the region for which the data is reliable, and at 60 miles/wk you’re beyond the point where there’s any data at all. I’ve included (in blue) a marker indicating my average exercise level for the 20 year period from 1992-2011. My average for 2009 was twice that again.
Looking at individual conditions separately, you can see that even within the limited range for which there’s reliable data, the risk of some is starting to rise as exercise levels increase:
View attachment 685108
The paper also comes with an interactive calculator website, so you can amuse yourself playing around with individual conditions, but AF in particular is too specific to appear as a separate graph.
It’s easy to see why the reliable data ends where it does when you compare the risk plot against the distribution of exercise dose among the study subjects:
View attachment 685109
The most common level of exercise is the equivalent of just 7miles of cycling a week, and half the population are doing less than 20m/wk. So, yes, exercise is very good for you, but the only evidence for that is at levels far below what a typical cycling enthusiast is likely to be doing. In short, Garcia shows good evidence that moderate exercise is better than none, but not any evidence that a lot of exercise is better still, or even that it isn’t worse.
That distribution doesn’t bode well for the prospect of getting any large scale studies into AF when large volumes of exercise are the very thing you need to be able to test, and especially as there’s a 20-30 year latency period from exposure to developing the arrhythmia. One reason that existing studies are of limited accuracy is that they’re mostly retrospective, with difficulties in retrospectively quantifying exercise dose, and all the confounding variables you need to control for. On the other hand, if you want a prospective study, not only are you likely to have to wait decades for the result, but you have the job of recruiting a huge number of people willing to keep accurate exercise records for that length of time. I think those who demand more statistical rigour have a choice: make a timely decision using the best information available, or wait until it’s too late in the hope of getting what you want. Even recent meta-studies are still using quite old data, because there isn’t much new stuff coming through yet.
There are those on here who concede that cycling increases the risk of AF, but insist that it only applies to athletes, so what’s the definition of an athlete? Well, perhaps unsurprisingly, there isn’t one, or at least not a standardised one.
Pelliccia et al in the 2020 guidelines from the ESC (European Society of Cardiology) for cardio patients participating in sport have the most to say on the matter:
“The ESC defines an athlete as ‘an individual of young or adult age, either amateur or professional, who is engaged in regular exercise training and participates in official sports competition’. Similarly, the American Heart Association (AHA) and others define a competitive athlete as an individual involved in regular (usually intense) training in organized individual or team sports, with an emphasis on competition and performance.”
In essence, a competitive cyclist, on the grounds that they’re the ones most likely to be motivated to train hard, but they then go on to say:
“In a proposed classification of athletes based on the minimum volume of exercise, ‘elite’ athletes…generally exercise ≥10h/week; ‘competitive’ athletes…exercise ≥6h/week; and ‘recreational’ athletes exercise ≥4h/week.”
And then finally, they finish with:
“This distinction is somewhat arbitrary since some recreational athletes, such as long-distance cyclists and runners, engage in exercise at higher volumes than some professional athletes”
No sh!t? I was averaging over 14 hours a week in 2009.
In Recommendations For Participation In Leisure-Time Physical Activity And Competitive Sports (2006) Heidbuchel et al said:
"It needs to be stressed, however, that there is no clear division between recreational and (semi)-competitive sports. Some patients may engage in high intensity exercise during leisure time activities”
In AF In Athletes And The Interplay Between Exercise And Health (2013) La Gerche et al said:
“..what is seen as vigorous exertion in cardiovascular epidemiology studies would not even be considered warm up tempo in sports science parlance”
It’s the same with the definition of ‘vigorous’ or ‘high intensity’. Pelliccia above defines:
Moderate Intensity as 55-74% MHR and
High intensity as 75-90% MHR, (70-85% HRR)
(But beware that Verdicchio says that %HRR is not a valid measure of intensity in AF patients because its correlation as a proxy for %VO2 doesn’t hold true.)
On the other hand, Garcia defines vigorous exercise as 7METs, which is about 12mph cycling, and in the 2013 Guidelines for Exercise Testing and Prescription, the ACSM define vigorous as 6METs, which is only 10mph.
As LaGerche et al says: “The non-standardised gradations of exercise intensity complicate interpretation of exercise studies.”
So you have a catch-22, without a clear definition of the difference between the control and study group it’s difficult to avoid cross-contamination of the two groups, and with cross-contamination the effect of the exercise becomes diluted in the test results. Some on here are quoting data for all cardiac conditions, but that will also mask the data for AF in particular.
With all that in mind, Calvo et al have produced the first ever dose-response curve similar to the Garcia ones for exercise-induced AF. They found that the most powerful predictor was lifetime-accumulated hours of high-intensity endurance training. AF cases had a history of nearly 5 times the training hours of the controls, and those with a history of more than 2000 hours had nearly four times the odds of developing AF. Endurance sports like cycling were much more risky than team sports.
What I found particularly interesting is that after controlling for exercise hours, competitive athletes were less likely to develop AF than non-athletes. So it’s the exercise hours that were the determinant of risk, not whether you’re a competitive athlete! My guess is that this might be because athletes benefit from organised coaching, which would put them in a better position to avoid overtraining.
View attachment 685110
So from this plot, 1000 hours is about optimum for maximising health benefit, and by the time you reach 2000hr all that benefit is lost.
As ever, the intensity is not objectively defined, Calvo uses four subjective levels assessed with questionnaire:
1. Sedentary: no physical effort, minimal walking
2. Light: minimal effort, eg. slow walking, no increase in HR, no perspiration
3. Moderate:, not exhausting, slightly increase HR, some light perspiration
4. High-intensity: vigorous, intense sweating and deep-breathing
For my own part, I can account for over 8,000 hours, with more still that went unaccounted, so if even a quarter of that was high intensity that puts me into the risk zone, especially as I’m 6’5”, and height is another major risk factor which exacerbates the effect of training (see below).
Crump et al is an interesting paper, not only because it’s huge like Garcia, with over 1.5m male participants, but it also uses objectively measured data harvested from military conscription records. They’re measuring aerobic fitness at age 18 though, which is a poor substitute for lifetime exercise hours, so it doesn’t show such a strong effect.
The purpose of it, however, was to explore the interaction of other risk factors like height and weight, and what they found was that not only were height weight and fitness risks for AF, but fitness magnified the effect of height and weight. Tall stature and high aerobic fitness was the highest risk of all, and fitness is actually protective for short men. If other studies aren’t controlling for height and weight, this may explain a lot of the variation in past results.
View attachment 685112
View attachment 685114
It occurs to me that the lower risk for athletes compared to non-athletes, and the lower risk by fitness rather than by exercise hours might be indicating that the root cause of AF risk is overtraining rather than either fitness or exercise hours per se.
So what’s the current advice?
Well, the 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation say:
“Physical activity should be considered to help prevent AF incidence or recurrence, with the exception of excessive endurance exercise, which may promote AF.”
This is categorised as
Class IIa: Weight of evidence/opinion is in favour of usefulness/efficacy. Should be considered
and
Level C: Consensus of opinion of the experts and/or small studies, retrospective studies, registries.
The supplementary data to the above guide lists all the risk factors for incident AF:
View attachment 685116
Nobody would argue that eating as much as you can is healthy on the grounds that starving is unhealthy, I think it’s bizarre to suggest that exercise is any different.
Well I'm going to keep cycling....!
.
