Association between active commuting and incident cardiovascular disease, cancer, and mortality: prospective cohort study

Dear Sirs,

In response to the comments of Gee Yen Shin and Rohini Manuel:

You express valid concern that the authors do not mention risk of RTA to cyclists. Only 3,339 cyclist KSI's were reported in 2015 (DfT figures, http://bit.ly/2qbhK0F ). It is thus likely that zero cyclists, so few as to be statistically invalid, were involved in serious accidents that the authors did not report this. Taken against total miles cycled, the KSI rate for cyclists is 0.95/million miles cycled. Nevertheless, lack of consideration of RTAs is a valid criticism of this paper. Official statistics fail to report minor injuries to cyclists, which may be a significant cause in work time lost- a metric of obvious interest.

You also express concern about a lack of correction for exposure to air pollution. Can Gee and Manuel present evidence that active commuters are in fact at any greater exposure to this than inactive commuters? Published work does indeed suggest that cyclists are affected by air pollution (https://www.ncbi.nlm.nih.gov/pubmed/281790030), but again fails to show that cycling causes greater damage than any other activity.

In response to the comments of Saripandis:

I feel your concerns about the threat that cycling poses to the automotive industry are somewhat overstated. With pedal cycling accounting for only ~1% of journeys in the UK, doubling it would reduce journeys by car by a little more than 1%. This is very unlikely indeed to translate into catastrophic loss of volume. Indeed, doubling the number of journeys made by pedal cycle is unlikely to have any negative effect on the automotive industry whatsoever. I hypothesise that cycling would have to increase by ~2,000% to cause any significant loss of scale to the automotive industry beyond normal churn.

I agree with Dr Anand re potentially serious health problems of daily long distance bike riding. In many ways it applies to walking long distances to and from work.

I’d like to add to the discussion on the benefits and drawbacks of the daily bike ride exercise to and from work a pearl of wisdom based on my own experience of walking for many years to work and back home 5.2 km each way (10.4 km a day, five days per week), in Sydney (Australia).

It was a great exercise until I started developing calluses, painful skin abrasions and painful thickened skin on the soles of my (overheated by walking) feet, and other unpleasant symptoms of what I called a “repetitive strain injury” I could avoid some streets with a peak hours’ high traffic density, but not all. I could not breathe deeply in those areas.

I realised that doing heavy house work was equally, if not more effective exercise as my intensive walking. Minus “repetitive strain injury” and contaminated air.

Needless to say I stopped walking to and from work; I started riding a bus (I called it affectionately my stretch limo) instead, which was instrumental in striking up a conversation with a lady on a bus stop and, through her, meeting Leif Karlsson, an electronics engineer, who became my (late) third husband and who at my suggestion developed a breathing monitor for babies; first a home monitor alarming when breathing stopped or became shallow (for which I reinvented the term hypopnea, and introduced the term stress-induced breathing pattern, beyond 5% of the volume of normal breathing for 20+ seconds) and soon after also a microprocessor-based research unit, which besides alarming, also retained the recorded babies’ breathing. Babies had alarms (recorded by parents on special forms we provided) after vaccination showing characteristic clocking and clustering along the critical days (all displayed on the microprocessor computer printouts of babies’ breathing), to me indicating a causal link. There were some near-misses thanks to alarms successfully attended to by the parents, but a baby never died on Cotwatch. To my surprise, because then I did not know the controversy surrounding vaccination, the computer records of babies breathing and manual records of alarms by parents had a few paediatricians rising their eyebrows, but many more (including some at the Health Department in Sydney, now sadly deceased) encouraged me to look deeper into the issues. I was also invited to present the results of the data collection of babies breathing to a pro-vaccination conference in Canberra (27th-29th May 1991).

I realised that we were onto something important and I headed straight for the medical libraries and embarked on systematic research of the relevant published medical research papers.

After my early retirement from the “day job” and besides a heavy house work I also took up heavy duty gardening around my house, enjoying the fresh Blue Mountains air followed by innumerable evenings spent in the nearest medical library. The librarians there called me their best client and even asked me to contribute to a birthday present for one of the librarians; I gave them my last five dollars. [They bought her an umbrella with a duck’s head.]

The first results of the data collection of babies’ breathing have been presented by myself to the 2nd Immunisation Conference , held from 27th to 29th May 1991(organised by the Public Health Association) and the abstract is published in the Proceedings of that conference. More details are published in the Journal of the Australasian College of Nutritional and Environmental Medicine: Scheibner 2002; Shaken baby syndrome diagnosis on shaky ground. J ACNEM 20(2): 5-8&15 and Scheibner 2003; Dynamics of critical days as part of the non-specific stress syndrome discovered during monitoring with Cotwatch breathing monitor. J ACNEM; 23(3): 1-5).

The upshot? You never know what’s good for you. A “repetitive strain injury” caused by walking some 10km a day to and from work for many years, inevitably abandoned in favour of a daily bus ride, may change the whole direction of your life.

Any casual observer standing aside an arterial road leading into any major town or city in UK would attest to the incontrovertible fact that commuting cyclists are usually super fit with a healthy BMI and usually recipients of envious glances from the less well endowed. So it is not at all surprising that the cyclists are healthy.[1].

The old and infirm on a bicycle would be like a fish out of water. People with high BMI cannot even buy a decent comfortable bicycle seat, let alone ride one to work.

This study might a good example to illustrate the statistical fallacy that underlying baseline differences could be somehow adjusted by statistical modelling, design and computations. Statistical sleight of hand cannot turn Apples into Pears. [2].

References

1 Celis-Morales CA, Lyall DM, Welsh P, et al. Association between active commuting and incident cardiovascular disease, cancer, and mortality: prospective cohort study. BMJ 2017;357:j1456. doi:10.1136/bmj.j1456

2 Song M, Hu FB, Wu K, et al. Trajectory of body shape in early and middle life and all cause and cause specific mortality: results from two prospective US cohort studies. BMJ 2016;353:i2195. doi:10.1136/bmj.i2195

We would like to commend Celis-Morales and colleagues on a well-written piece that provides important evidence in support of public policies that promote active commuting. Following publication, several authors have noted that Celis-Morales and colleagues failed to discuss the possibility of reverse causation. We wish to add to the discussion by reminding readers that the possibility of reverse causation is further increased in this study by the relatively short follow-up period (i.e., 5.0 years for mortality and 2.1 years for incident CVD and cancer) and the measurement of the exposure at just one point in time. The effects of the exposures under study may take more than two to five years to develop and some of the observed associations may be a result of undetected antecedent disease. Although the large sample size helps increase the precision of the estimates, these estimates may still be spurious. Multiple exposure measurements would have helped minimize errors in the independent variables particularly given that they were self-reported (i.e., minimized regression dilution bias(1)) and contributed to disentangling the directionality of the observed associations. To more fully address the possibility of reverse causation, ideally future analyses would incorporate both longer follow-up and multiple exposure measurements.

We also wish to raise an additional concern regarding the possibility of misclassification bias - particularly in relation to the referent group. The risks of cardiovascular disease (CVD) and cancer incidence, as well as CVD, cancer, and all-cause mortality across different types of commuting behaviours were compared to the risks observed in non-active commuters. Non-active commuters were defined as participants who relied on car/motor vehicles and/or public transport only. While we agree that cycling and walking may be more desirable than reliance on public transport, we believe that public transit users should not be grouped, conceptually or methodologically, with car/motor vehicle users.

Adults who do not have access to a vehicle are more physically active than adults who have access to a vehicle and are also more likely to walk 30 minutes or more per day to or from transit.(2) For those that currently rely on personal vehicles for transport, transitioning to using public transit (where this infrastructure exists), may represent an easier ‘first step’ for incorporating regular physical activity into daily life than cycling. In light of the other comments received in response to Celis-Morales and colleagues’ paper, encouraging the use of public transit may also have wider benefits such as helping to reduce air pollution and traffic accidents in large cities by reducing car traffic. By including public transit users in the referent group, the authors may have biased the associations - particularly for walking - towards the null. Building new public transit stops in neighbourhoods has been linked to increases in physical activity.(3,4) This is likely because at least some active commuting is required to reach a public transit stop – as opposed to reliance on a car/personal vehicle since cars are usually parked in close proximity to an individual’s residence and workplace. Given the potential benefits of public transport on human health,(5) it may be helpful for researchers to disaggregate the role of public transport from car/vehicle use.

Samantha Hajna & Simon Griffin

References:
1. Hutcheon JA, Chiolero A, Hanley JA. Random measurement error and regression dilution bias. BMJ. Jun 23 2010;340:c2289.
2. Besser LM, Dannenberg AL. Walking to public transit: Steps to help meet physical activity recommendations. American Journal of Preventive Medicine. Nov 2005;29(4):273-280.
3. Miller HJ, Tribby CP, Brown BB, et al. Public transit generates new physical activity: Evidence from individual GPS and accelerometer data before and after light rail construction in a neighborhood of Salt Lake City, Utah, USA. Health & Place. Nov 2015;36:8-17.
4. Brown BB, Werner CM. A new rail stop: tracking moderate physical activity bouts and ridership. American Journal of Preventive Medicine. Oct 2007;33(4):306-309.
5. Wasfi RA, Ross NA, El-Geneidy AM. Achieving recommended daily physical activity levels through commuting by public transportation: unpacking individual and contextual influences. Health & Place. Sep 2013;23:18-25.