Association between active commuting and incident cardiovascular disease, cancer, and mortality: prospective cohort study, by Carlos A Celis-Morales et al, BMJ 2017;357:j1456

Researchers from the University of Glasgow grabbed a lot of international media attention a few weeks ago with a paper published in the British Medical Journal showing a very strong correlation between good health and cycling to work (Association between active commuting and incident cardiovascular disease, cancer, and mortality: prospective cohort study):

We found that cycling to work was associated with a 41% lower risk of dying overall compared to commuting by car or public transport. Cycle commuters had a 52% lower risk of dying from heart disease and a 40% lower risk of dying from cancer. They also had 46% lower risk of developing heart disease and a 45% lower risk of developing cancer at all.

A 40% lower risk of dying from cancer! Those who walked to work got some benefit too; a 27% lower risk of heart disease. But their risk of dying from all causes or from cancer was the same as those who drove or took public transport. The authors suggest their findings show what actions should be taken:

Initiatives to encourage and support active commuting could reduce risk of death and the burden of important chronic conditions.

These are impressive findings and were eagerly lapped up by both social media and mainstream media, who saw a direct causal link between cycling and good health. The New Daily headlined its report, Cycling to work cuts risks of cancers: study. BBC News was equally direct, proclaiming Cycling to work can cut cancer and heart disease, says study; The Australian asserted Cycling to work cuts risk of cancer; and Australia’s Bicycle Network reported the news under the headline, Bikes steer you clear of early grave.

The accompanying editorial in the same issue of the British Medical Journal forthrightly claimed Active commuting is beneficial for health. The researchers also published a widely-read summary article in The Conversation on 20 April, under the heading, Cycling to work: major new study suggests health benefits are staggering.

It all sounds fantastic so it’s no wonder the significance of the findings was taken at face value. But the research doesn’t establish causality, as pointed out by several writers to the British Medical Journal. Professor David Colquhoun from University College London wrote:

It worries me that, although the paper has the usual token nod to the problem of causality, it’s written entirely as though it demonstrates that biking to work has all sorts of benefits. The related editorial is even more cavalier: the title “Active commuting is beneficial for health” is a direct statement of causality. In particular, the paper doesn’t discuss at all the possibility of reverse causality… Of course I’m 100 percent in favour of people walking or biking to work. It would make the streets a lot more pleasant for everyone. But it isn’t good when evangelism replaces science. Causality really matters.

Causality does indeed matter. While I don’t doubt regular cycling to work is good for your health at the individual level if you don’t otherwise get adequate exercise, I think there might be a simpler explanation for much of the difference identified by the researchers. They looked at potential health and socio-demographic confounders, but there will always be unobserved differences between the two groups.

An alternative explanation is that the 3% of workers in England and Wales who currently cycle for the journey to work are different in a number of ways – including in terms of their health prospects – from the 67% who commute to work by private vehicle and the 18% who take public transport. The researchers note that 90% of cyclists in the sample met current physical activity guidelines compared to only 51% of motorists. None of this should be surprising; it’s hardly news that the current tiny cohort of on-road cyclists in Australia is quite different from the great majority who use other modes e.g. they’re less risk averse.

The implication is that if cyclists made up a much larger proportion of the journey to work population, the differences with respect to other commuters in terms of health outcomes would be smaller e.g. the next 3% of (prospective) cycling commuters might not cycle as intensively as the first. We already know the average commuter cyclist in high-cycling countries is closer in profile to the average citizen than their counterparts in the UK or Australia. It’s interesting to note that notwithstanding the authors’ claim that cycling commuters in their sample have a 40% lower risk of dying from cancer, Denmark and The Netherlands have notably higher cancer rates than the UK.

The researchers hypothesise that the lower health benefits from walking to work compared to riding reflect the higher exercise intensity of cycling. That seems plausible, but it’s not the only possible explanation. It might be that it’s mainly a function of the larger population of walkers i.e. 11% of workers in England and Wales commute to work on foot compared to 3% by bicycle.

If cycling to work is your main form of exercise, it’s likely it’s good for you. If you’re among the very small proportion of Australians who cycle to work, it’s likely your health prospects are much better than those who drive; but it’s due to a number of your attributes and would probably hold in large part even if you didn’t commute by bike.