# Covid/Coronavirus and diabetes - the numbers

#### Jamie H

##### Well-Known Member
Ohh yea sorry... Actually glad that's explained as I'm no expert so thanks!! ...

Read that paper too. I see the hazard ratio for well controlled is 1.5... So would that be 1.5 against your relavant age? So under 40 would be 1.5 x 0.07 for example?

#### copilost

##### Well-Known Member
Ohh yea sorry.
No my bad, I should have quoted the reference, sorry again.

#### Jamie H

##### Well-Known Member
No my bad, I should have quoted the reference, sorry again.
No 100 percent I should have read more thoroughly. Just on my point about HR so if you say it's against someone of your same age BMI etc it's 1.5 (Well controlled) x 0.07 (I'm 31 and under 40 risk is 0.07)... Something like that anyway??

#### copilost

##### Well-Known Member
Just on my point about HR so if you say it's against someone of your same age BMI etc it's 1.5 (Well controlled) x 0.07 (I'm 31 and under 40 risk is 0.07)... Something like that anyway??
Ah no! For someone of your age the additional risk of having controlled diabetes is 1.5 as this estimate has been adjusted for age. For someone of 70 the additional risk is also 1.5. Does that mean we think the risk from diabetes is the same for both groups? Short answer we don't know. To find out we would need to run additional statistical models that look for an interaction. i.e does the risk from diabetes change with age. The 1.5 is an average for all age groups, it might be the same whatever your age, it might be more for younger people or for older people, we can't say from this data.

#### Jamie H

##### Well-Known Member
Got you.. I think. But what confuses me is 1.5 times what? Someone who's the same age, sex, BMI etc? Then it's 1.5 times someone of a similar age

If that's the case then would it not infer it changes with age? Just trying to get my head around it all.. Without trying to sound argumentative... Not my area of expertise haha!!

#### Lupf

##### Well-Known Member
Thanks for your posts @Jamie H @copilost @urbanracer @DavidGrahamJones. I agree with the explanation on how to read the risk factors by
@copilost "paper reports adjusted hazard. That means if you compare two people who are same age, sex, BMI, etc etc for everything on the list in the table then the hazard reported for diabetes is the additional hazard regardless of all the other factors."
Two more points: i) When you multiply two risks from this study, correlations are neglected, so it could be an overestimate
ii) while usually the baseline is clear, e.g. no diabetes, for the age related risks, the comparison is made with respect to the 50 to 59 age bracket, so @Jamie H risk is 7% compare to an average 50 to 59 year old.

I use this comment to discuss how science works: We are still learning about Covid. While scientists in the field broadly agree on many issues (spreading and transmission), many questions are still not clear (why does it affect the elderly and diabetics more?, what are the causal mechanism?). When scienties carry out a study, they write a paper with the results and submit it, e.g. to Medrxiv and to a journal such as bmj. At this time the paper will undergo peer review. This means that other specialists in the field will study the paper and try to establish if the methods and results are correct and compare it with other findings. Basically they'll do their best to prove the study wrong. This is looks like scientists not being able to agree on anything, but it is actually not only important but also very healthy. This is how science progresses. Only peer-reviewed results will be accepted by the journals.

We have to be careful when we look at results. Before peer review results could actually be wrong. If different groups arrive at the same result confidence in a result will grow. I try to read papers like the one I quoted at the beginning of the thread. These are made by expert groups and their methods are explained. While I am not a specialist, a paper gives me the chance to read how a result was reached.

Unfortunately many media like to pick up raw numbers such the fact that a quarter of hospital fatalities also had diabetes. Taking such a number without context can easily lead to the wrong conclusion, be used to fear monger in this case, but also to trivialise - R is smaller than one - therefore the lockdown was not required - depending on the intention.

#### copilost

##### Well-Known Member
Then it's 1.5 times someone of a similar age

The figure of 1.5 is compared to a 'reference case' (for diabetes it's someone with no diabetes). The 0.07 for age is compared to someone who is 50-69. If you look at the table there is (ref) in each category, that's what the risk is referencing. For sex the reference is female, so whatever the risk for a female the male risk is 1.99. If the risk to a woman was 1% then the risk for a man is 1% x 1.99=2%. (nb I should be saying hazard really).

it's not easy

#### Jamie H

##### Well-Known Member
I think what Partha Kar is doing is excellent work. My only worry is there may not be enough data around extremely young and to a lesser extent younger type 1 diabetics to analyse. There may be a risk unidentified simply on the basis both groups are a small % of the UK population. Best to not assume though until the data comes out

#### Jamie H

##### Well-Known Member
The figure of 1.5 is compared to a 'reference case' (for diabetes it's someone with no diabetes). The 0.07 for age is compared to someone who is 50-69. If you look at the table there is (ref) in each category, that's what the risk is referencing. For sex the reference is female, so whatever the risk for a female the male risk is 1.99. If the risk to a woman was 1% then the risk for a man is 1% x 1.99=2%. (nb I should be saying hazard really).

it's not easy
Starting to understand it don't worry haha. I get the limitation in the study now that you referred to a few posts back. It's 1.5 and the reference is someone without diabetes... But we simply don't know how the other factors play in.... And you can't simply multiply them either as it could be an overestimate.

I just know my HR is 1.5 against someone of same sex age and weight ... I think!

#### Jamie H

##### Well-Known Member
Thanks for your posts @Jamie H @copilost @urbanracer @DavidGrahamJones. I agree with the explanation on how to read the risk factors by
@copilost "paper reports adjusted hazard. That means if you compare two people who are same age, sex, BMI, etc etc for everything on the list in the table then the hazard reported for diabetes is the additional hazard regardless of all the other factors."
Two more points: i) When you multiply two risks from this study, correlations are neglected, so it could be an overestimate
ii) while usually the baseline is clear, e.g. no diabetes, for the age related risks, the comparison is made with respect to the 50 to 59 age bracket, so @Jamie H risk is 7% compare to an average 50 to 59 year old.

I use this comment to discuss how science works: We are still learning about Covid. While scientists in the field broadly agree on many issues (spreading and transmission), many questions are still not clear (why does it affect the elderly and diabetics more?, what are the causal mechanism?). When scienties carry out a study, they write a paper with the results and submit it, e.g. to Medrxiv and to a journal such as bmj. At this time the paper will undergo peer review. This means that other specialists in the field will study the paper and try to establish if the methods and results are correct and compare it with other findings. Basically they'll do their best to prove the study wrong. This is looks like scientists not being able to agree on anything, but it is actually not only important but also very healthy. This is how science progresses. Only peer-reviewed results will be accepted by the journals.

We have to be careful when we look at results. Before peer review results could actually be wrong. If different groups arrive at the same result confidence in a result will grow. I try to read papers like the one I quoted at the beginning of the thread. These are made by expert groups and their methods are explained. While I am not a specialist, a paper gives me the chance to read how a result was reached.

Unfortunately many media like to pick up raw numbers such the fact that a quarter of hospital fatalities also had diabetes. Taking such a number without context can easily lead to the wrong conclusion, be used to fear monger in this case, but also to trivialise - R is smaller than one - therefore the lockdown was not required - depending on the intention.
Thanks for the feedback. Could you explain how you arrived at 7% if the HR for well controlled diabetes is 1.5? Apologies for my ignorance on this.

#### Lupf

##### Well-Known Member
Thanks for the feedback. Could you explain how you arrived at 7% if the HR for well controlled diabetes is 1.5? Apologies for my ignorance on this.

Apologies if this wasn't clear. What the study says is that your risk is 7% compared to a 50 to 59 year old if all other factors are equal. so this number does not take into account diabetes.
If you want to calculate your overall risk then you can multiply 0.07 (age) times 1.5 (controlled diabetes) = 0.105. This then means: Compared to your 55 year old "twin" with exactly the same lifestyle, weight, genes ... , but no diabetes (could be your dad?) your risk is only 10%. However, when doing this you neglect correlations between age and diabetes, so could be an overestimate (technically even an underestimate, but this is very unlikely).

#### Jamie H

##### Well-Known Member
Apologies if this wasn't clear. What the study says is that your risk is 7% compared to a 50 to 59 year old if all other factors are equal. so this number does not take into account diabetes.
If you want to calculate your overall risk then you can multiply 0.07 (age) times 1.5 (controlled diabetes) = 0.105. This then means: Compared to your 55 year old "twin" with exactly the same lifestyle, weight, genes ... , but no diabetes (could be your dad?) your risk is only 10%. However, when doing this you neglect correlations between age and diabetes, so could be an overestimate (technically even an underestimate, but this is very unlikely).
Got it. Its what I was doing originally but @copilost seemed to think differently?? However 0.07 x 1.5 is 0.1%... Not 10%??

#### Mr_Pot

##### Well-Known Member
While someone at very high risk is likely to act differently to someone at very low risk I can't see that having some precise figure for risk is much help to individuals. You can shield, self isolate, social distance or not bother. There is no point in tailoring your response to exactly match your risk factor.

#### Jamie H

##### Well-Known Member
I think it's a good guide in terms of whether to shield or not... Don't think anyone's going to be acting like they were before covid even if they fit into the lower risk category.

#### Lupf

##### Well-Known Member

There are many self styled experts who make analyses, this one is by far not the worst. However a quick look on the author's twitter profile tells me the person has a political agenda, This study is not science. It starts with the conclusion - lockdown was a mistake - and cherry picks the data to suit their belief, ignoring basic facts - infections were spreading exponentially in March which would have resulted in over a million deaths in Europe alone by now, if no drastic action would have been taken.
I've decided some time ago that there is no point in discussing with them and that this would be a waste of my time.

Whenever I see a claim, result, I ask - where is the paper? If - like in this case - there is none - I move on. I hope you come to the same concluson.

#### Lupf

##### Well-Known Member
Got it. Its what I was doing originally but @copilost seemed to think differently?? However 0.07 x 1.5 is 0.1%... Not 10%??
0.07 is 7%, 1.5 is 150%, 0.07 x 1.5 = 0.105 which is 10%

#### Jamie H

##### Well-Known Member
OK where I've got confused then is I read 0.07 as 0.07% which would tie in with other studies that puts risk of dying of someone under 40 at 0. 1% (rounded up). If its to be taken as 7%... What does the 7% represent then? Surely not 7% risk of dying with covid for someone aged 18-40 .
0.07 is 7%, 1.5 is 150%, 0.07 x 1.5 = 0.105 which is 10%