Apparently the young are all dying of cancer because they’re all so damn fat. That’s the conclusion to draw from this latest research – the young are showing disturbing incidences of obesity related cancers. Sadly, there’s a logical error here. Or at least a potential logical error. We don’t actually know that the cancers are being caused by obesity. Could be something else. Might, just to invent some reasons, be sugar consumption, lack of exercise or the existence of Simon Cowell. That last has certainly soured a few guts in its time.
Normally this isn’t a problem in epidemiology. We’ve a correlation here and that’s what we’re exploring, obesity rates and cancer rates. But we do have a problem because of the way we decide that a cancer is obesity related. We just assume that some percentage of cancers are, of their type, caused by obesity. Which again is usually fine. But that doesn’t work if we think we’re seeing a structural change in the world. A structural change pretty much being defined by our thinking that we’re seeing a change in the rates of incidence.
[perfectpullquote align=”full” bordertop=”false” cite=”” link=”” color=”” class=”” size=””] Rates of obesity-fuelled cancers are now rising in successively younger age groups, a landmark study shows. Experts said “shocking” levels of disease linked to growing waistlines across the globe threaten to reverse decades of progress in the war on cancer. The Lancet study shows that rates of obesity-related cancers are rising faster in adults aged 25 to 49 than in older generations – despite the fact cancer is seen as a disease of old age. The research, which examined 12 types of cancer linked to obesity, divided patients into five-year age groups from 25-29 to 80-84 years old. In six of the main cancers – including bowel, pancreatic and kidney disease – it was found that the younger the age group, the greater the increase in incidence. [/perfectpullquote]We start by stating that certain cancers are obesity related. As with booze deaths, 25% – or whatever the number – of exploding livers are assigned to booze as a cause. Now we see a rise in exploding livers, we say that booze related exploding livers are rising. By 25% of the total number. But that doesn’t work if there’s some new thing causing livers to explode – say, Hepatitis C, or E,F,G, however far we’ve got in exploring. We can’t and shouldn’t assign 25% of the deaths from that to booze. But we will do, given that we count 25% of all deaths from exploding livers as coming from booze.
We might well be seeing a rise in youth cancers. And they might well be those types of cancer which we associate with obesity. But the next step, the rise in cancer is being caused by the rise in obesity, this isn’t something we can say. Because we’re not actually checking that the cancers are being caused by obesity, we’re assuming it.
Not that that will stop those who insist we should all starve ourselves into being whippets but it’s worth pointing it out.
The language we use matters - it provides clarity to our own thoughts and enables…
It is now generally acknowledged that the structure of the NHS needs to be overhauled…
In the film Apollo 13, a loss of oxygen causes the crew to start inadvertently…
There's an idea out there which seems intuitive but then so many ideas do seem…
When we think about the darkly opaque goals of modern central bankers as they relate…
As the papers recently filled with the distressing images of desperate souls looking to escape…
View Comments
First, as I learned from the estimable John Brignell, find out how big the numbers are. In a low-incidence cancer a couple of cases could show a large percentage change. These epidemiologists will do anything to get published. Tricks of presentation might explain all the claimed effect.
Quite correct. ALL medical assertions should nowadays be taken with a large pinch of salt - especially those which seem to confirm current fashionable thinking.
Statistical studies offer so many opportunities for ensuring that the 'correct' result is obtained that they should be treated with great suspicion.
I looked for the study, but could not find it. However, I found a comment on it here:
https://www.thelancet.com/journals/lanpub/article/PIIS2468-2667(19)30017-9/fulltext
I append a few salient features:
"....However, despite the plausibility, the investigators' hypothesis and report have constraints that are important to consider.
First, the hypothesis assumes that cancer risk factors have a similar influence on cancer development across the lifespan. However, some early onset cancers can have a distinct cause from their later onset forms..................Pooled analyses of multiple prospective studies will be required to characterise associations of risk factors with the early onset of most cancers, and to estimate latency for early life exposures. These considerations .........underscore the speculative nature of the investigators' interpretation.
......................... In particular, the report acknowledged insufficient evidence regarding the association of obesity with some cancers grouped with the 18 additional cancers investigated .................. Such cancers did not show temporal trends of increasing incidence in younger adults in Sung and colleagues' study.
Additionally, ............ they chose to aggregate subtypes for other cancer types............... The interpretation of the data for the aggregated groups is challenging; notably, the small number of subtype-specific findings presented in the appendix suggest moderate differences in age-specific and birth cohort-specific incidence .........
Lastly, Sung and colleagues did not comment on why only some obesity-related cancers, and not all 12, showed temporal trends of markedly rising younger adult incidence, or why some obesity-related cancers appeared to have declining rather than increasing incidence in the older age groups. ...."
I miss Numberwatch :-(
Exactly, already made that point elsewhere. In young people the incidences of these cancers are extremely low, so on or two extra cases can show a "large increase" when it is nothing of the sort - particularly as cancer cases are always whole numbers.
At 10/100,000, on extra case shows a 10% increase, (which is higher than those being claimed!) but obvioulsy the difference between 10 and 11 per 100,000 is just random.
It's more likely that both obesity and the increasing incidence of cancer (and other non-communicable diseases) are caused by a third factor.
Processed food at a guess.
And how would "processed food" cause anything?
Or even that increasing consumption of processed food is also caused by another factor. This is world data as far as I know, and the black, brown and yellow world as a whole is getting richer, and living in cities with sanitation, not dying from communicable diseases, taking a little less exercise, and using motor transport a little more. Processed food is a way of feeding this busier urbanised population with less waste.
Life expectancy still nudging up though.