Author Topic: Statistics and bacon  (Read 151 times)

bluehillside Retd.

  • Hero Member
  • *****
  • Posts: 13608
Re: Statistics and bacon
« Reply #50 on: April 17, 2019, 05:49:51 PM »
Prof,

Quote
Yup completely misleading, particularly as the comparison was between people who eat on average 76g of red and processed meat a day compared to people eating 21g a day.

You also have to be careful that the study fully factors out other linked effects - e.g. obesity. It could be that being overweight is the driving factor rather than red and processed meat eating per se, and that people who tend to eat the most red and processed meat tend also to be the more overweight.

You need to be very careful that association and causation aren't confused.

Yep - and you have to be careful too about unintended consequences. I have no idea whether bacon contains anything that's good for us but if, say, it's a main source of iron (as I believe red meat is) and people stopped eating it the consequence could be that different illnesses would increase. It's a bit like the cycling stats - one-in-X rides will result in a fatality, but if the Y million cyclists there are sat on the sofa instead the death toll from heart disease etc would be much higher. Of course the answer is to make the options non-binary - go to the gym instead, but then perhaps there'd be more road accidents, more exhaust pollution as a result of people driving there. Risk and perception of risk are in other words not the same thing.     
"To understand via the heart is not to understand."

Michel de Montaigne

ProfessorDavey

  • Hero Member
  • *****
  • Posts: 10449
Re: Statistics and bacon
« Reply #51 on: April 17, 2019, 05:57:48 PM »
Prof,

Yep - and you have to be careful too about unintended consequences. I have no idea whether bacon contains anything that's good for us but if, say, it's a main source of iron (as I believe red meat is) and people stopped eating it the consequence could be that different illnesses would increase. It's a bit like the cycling stats - one-in-X rides will result in a fatality, but if the Y million cyclists there are sat on the sofa instead the death toll from heart disease etc would be much higher. Of course the answer is to make the options non-binary - go to the gym instead, but then perhaps there'd be more road accidents, more exhaust pollution as a result of people driving there. Risk and perception of risk are in other words not the same thing.   
True - we need to look at the range of factors that affect health and how they inter-relate.

And perhaps the elephant in the room is poverty - I wouldn't be surprised if there is a relationship between deprivation and eating large amounts of processed meat, which tends to be both cheap and lasts a long time compared to fresh ingredients. The link between deprivation and poor health is scary. Where I work in East London, one of the most deprived areas in the UK, the life expectancy is five years less than in Kensington & Chelsea, which is less than 6 miles away.

Walter

  • Hero Member
  • *****
  • Posts: 2980
Re: Statistics and bacon
« Reply #52 on: April 17, 2019, 06:23:47 PM »
when you say 'deprived areas' are you referring to levels of intelligence ?

Udayana

  • Hero Member
  • *****
  • Posts: 3834
  • βε ηερε νοω
    • The Byrds - My Back Pages
Re: Statistics and bacon
« Reply #53 on: April 17, 2019, 07:33:46 PM »
True - we need to look at the range of factors that affect health and how they inter-relate.

And perhaps the elephant in the room is poverty - I wouldn't be surprised if there is a relationship between deprivation and eating large amounts of processed meat, which tends to be both cheap and lasts a long time compared to fresh ingredients. The link between deprivation and poor health is scary. Where I work in East London, one of the most deprived areas in the UK, the life expectancy is five years less than in Kensington & Chelsea, which is less than 6 miles away.

They seem to have adjusted for deprivation and other health factors in their analysis:

Diet and colorectal cancer in UK Biobank: a prospective study
Ah, but I was so much older then ... I'm younger than that now

ProfessorDavey

  • Hero Member
  • *****
  • Posts: 10449
Re: Statistics and bacon
« Reply #54 on: April 17, 2019, 08:03:45 PM »
when you say 'deprived areas' are you referring to levels of intelligence ?
No - levels of deprivation, broadly speaking poverty.

jeremyp

  • Admin Support
  • Hero Member
  • *****
  • Posts: 23341
Re: Statistics and bacon
« Reply #55 on: April 17, 2019, 08:04:20 PM »
I don't think it is misleading - it just means that the final risk is 20% higher than the initial (6%) risk, in other words 7.2%.

Where are we getting 6% is the initial risk from? If 6% is the overall rate of bowel cancer in the UK, then it already has the people who eat too much red meat baked in. So, if you eat more than the four portions a week, your risk is probably higher than 6% but not as by anything like as much as 20% higher and reducing your intake to less than two portions a week may reduce your risk to lower than 6% but not necessarily to 5%.
This post and all of JeremyP's posts words certified 100% divinely inspired* -- signed God.
*Platinum infallibility package, terms and conditions may apply

Walter

  • Hero Member
  • *****
  • Posts: 2980
Re: Statistics and bacon
« Reply #56 on: April 17, 2019, 08:12:17 PM »
I've just listened to this report on ITV NEWS and although very slickly presented the science content was sadly lacking to the extent it was meaningless .
typical!

Udayana

  • Hero Member
  • *****
  • Posts: 3834
  • βε ηερε νοω
    • The Byrds - My Back Pages
Re: Statistics and bacon
« Reply #57 on: April 17, 2019, 08:35:07 PM »
Where are we getting 6% is the initial risk from? If 6% is the overall rate of bowel cancer in the UK, then it already has the people who eat too much red meat baked in. So, if you eat more than the four portions a week, your risk is probably higher than 6% but not as by anything like as much as 20% higher and reducing your intake to less than two portions a week may reduce your risk to lower than 6% but not necessarily to 5%.
It's from the same databank used for the study, which contains health and diet data for about half a million people over 6 years. The 20% is the difference between those consuming around 29g processed meat vs 5g average per day.   
Ah, but I was so much older then ... I'm younger than that now

Udayana

  • Hero Member
  • *****
  • Posts: 3834
  • βε ηερε νοω
    • The Byrds - My Back Pages
Re: Statistics and bacon
« Reply #58 on: April 17, 2019, 08:44:35 PM »
I've just listened to this report on ITV NEWS and although very slickly presented the science content was sadly lacking to the extent it was meaningless .
typical!

The Independent link provide by ProfD seems to have a reasonable level of detail (apart from the general rubbish all over the website). The BBC page is also OK.   

On broadcast media they assume that no-one has the attention span for anything mathsy (even the maths programmes occasionally hidden away on BBC4). "More or Less" on R4 is probably the best available.
 
Ah, but I was so much older then ... I'm younger than that now

ProfessorDavey

  • Hero Member
  • *****
  • Posts: 10449
Re: Statistics and bacon
« Reply #59 on: April 17, 2019, 09:00:53 PM »
Where are we getting 6% is the initial risk from? If 6% is the overall rate of bowel cancer in the UK, then it already has the people who eat too much red meat baked in. So, if you eat more than the four portions a week, your risk is probably higher than 6% but not as by anything like as much as 20% higher and reducing your intake to less than two portions a week may reduce your risk to lower than 6% but not necessarily to 5%.
Yup - that was my point in a later post I made - the 20% is the increase from eating low levels of red and processed meat to high levels, not the increase from bowel cancer levels in the general population.