In part one of this series I looked at why a lack of exercise causes premature deaths; the aim of this second part is to take a look at exactly how many people may be dying from a lack of exercise. My analysis shows that this number is very high, but it also seems that the studies that I found were underestimating the true death toll due to systematically conservative assumptions in approximations used.
The HIPI model, based on Lee et al (http://www.sciencedirect.com/science/article/pii/S0140673612610319), predicts a death toll of 36,815 people a year in England, for ages 40-79, but only from coronary heart disease, colorectal cancer, breast cancer and diabetes. This has a few conservative assumptions from my point of view: I think people start to die of inactivity from childhood (though admittedly in far lower numbers); it ignores many illnesses thought to be caused by inactivity; it uses guideline minimum amounts of exercise (150 minutes) as its baseline for active, and it doesn’t take in the whole of the UK (to allow us to compare to other equivalent statistics).
To correct the last point, we can make an approximation by taking into account the population size of England as compared to the UK, assume that the populations show similar behaviour, and arrive at an estimate of about 44,000 deaths per year in the UK. Of course, this still leaves the other issues – so let’s go back to look at the underlying study – Lee et al.
Lee et al quote an all-cause mortality burden of 16.9% of all deaths in the UK – with a total death toll of about 569,000 per year from the ONS, this yields 96,000 deaths per year. Again, they are assuming that active people are doing guideline amounts of exercise (i.e. over 150 minutes); they also assumed that only people over the age of 40 suffered increased mortality rates. These are two flaws which lead me to expect that this is an underestimate.
What’s the issue with current thinking?
Most biological systems deliver a dose response curve that looks a bit like this:
Basically, initially as we increase dose, you get maximal response up until a point (corresponding to about 20M/L of the drug in the graph above) where the effect of each additional dose decreases. Of course, in the case of exercise, the units of our dose will be measured in calories, minutes of activity or some measure that accounts for volume and intensity of exercise.
The above studies use models for calculating the death toll from inactivity assume that somewhere in the region of 150 minutes of exercise a week (the UK guidelines) deliver almost all of the benefit that exercise can bring. This is based on the dea that 150 minutes is at the ‘plateau’ stage of the above curve – where an increase in dose will be a small, or no, further improvement in health outcomes. I believe that this is not true, and in fact the Lee et al study in the Lancet does acknowledge this as a limitation in their estimates. A large meta-analysis (a review of available evidence) (http://ije.oxfordjournals.org/content/early/2011/09/05/ije.dyr112 ) gives an estimate of a 14% reduction of deaths at 150 minutes of exercise, and 26% – nearly double – at 300 minutes.
In fact we repeatedly find (http://www.indiana.edu/~k562/articles/role/Lee%20PA%20morbidity.pdf, http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(11)60749-6/fulltext ) that in fact, over the range of durations studied, there appears to be an linear relationship between volume of exercise and decreased mortality rates – i.e. a doubling of exercise volume doubles the protective of effect of exercise.
It feels intuitively as if there must come a point where increasing exercise volumes do not keep on reducing death rates per year – but it seems from the studies reviewed that we have not reached that point at 300 minutes, let alone 150!
We can draw a dose-response curve for exercise using currently available data, in fact someone has already done this for me using data from one study:
It seems moderate exercise for various activities yields about 250 calories per 30 minutes; so it looks like we’re reaching the point where the benefits of exercise start slowing at about 2,500 calories exercise per week – or 10 x 30 minutes of exercise a week – 300 minutes (warning – fast and dirty approximation!). Note that it appears that the benefits are still accruing, albeit more slowly, at 3000 Kcal per week.
Interestingly, there seems to be no lower threshold for exercise effectiveness (so there’s no flat spot at the start of the graph) – basically any amount will start improving your health.
So: how many?
The evidence above seem to show that best available protection from deaths by exercise comes at more than 300 minutes of exercise a week. The studies investigated above are based on 150 minutes – which seem to show about half the level of benefit of 300 minutes.
This means that the benefit to those individuals doing less than 150 minutes can be expected to double – but this isn’t the only effect. There will be a fair chunk of the population that were considered ‘active’ under previous studies but who are doing less than 300 minutes of activity a week, and these will also benefit from doing more exercise. This leads me to suggest that the studies above are underestimating the possible protective effect of exercise by at least a factor of 2.
Applying a factor of 2 to Lee et al gives a death toll from inactivity, UK-wide, of 192,000 deaths per year. Note, this is a very fast and dirty estimate and needs further investigation, but can be expected to be a low-end estimate as I have ignored the benefit to those individuals doing between 150 and 300 minutes of exercise a week for simplicity. I have ignored the number from the HIPI model, as it is based on the same underlying data, but is clearly ignoring a vast swathe of deaths caused by inactivity from the discrepency with the number generated by Lee et al.
This puts inactivity as a cause of death ahead of smoking (100,000 deaths per year in the UK, from here: http://ash.org.uk/files/documents/ASH_107.pdf); vastly exceeds deaths directly caused by air pollution from combustion emissions (19,000 from here: http://pubs.acs.org/doi/abs/10.1021/es2040416); and absolutely dwarfs deaths from road traffic collisions (1,754 deaths in 2012).
Notes of caution:
These calculations are based on correlation studies – i.e. they look at how much less likely someone who exercises at different amounts is to die than someone who’s inactive; however, as discussed in other blogs correlation does not always imply causation and there may be confounding factors around more affluent people being more likely to exercise. As affluent people generally live longer, it may be that this is causing the apparent protective effect of exercise. However, there are also other confounders – the types of exercise participated in may expose people to greater risk of dying from other factors – e.g. air pollution from exercising outside. These confounders may cause this study to underestimate the true protective effect of exercise if, say, you were exercising in the country where this air pollution was not present in such high quantities.
- It seems highly likely that any amount of exercise will improve your chance of living a long life – and the more exercise you do, the greater the benefit, until at least 300 minutes of moderate-to-intense activity per week
- Guideline amounts of activity are below the optimum amounts for reducing avoidable deaths according to best available data
- Available studies seem to be underestimating death rates from inactivity by a large amount, possibly a factor of greater than 2
- The reason for these underestimates is an assumption that guideline amounts of activity deliver all possible health benefits from exercise
- Applying a correction factor to Lee et al predicts a death toll from inactivity of 192,000 deaths per year – this is likely to still be a conservative estimate, and most likely puts inactivity as the leading cause of avoidable death in the UK by some considerable margin
- More research is needed in the field to fully quantify the likely death toll, the economic cost to the state and therefore how much needs to be invested in preventing inactivity through clever urban design and sports and physical activity interventions