Docb
Moderator
- Relationship to Diabetes
- Type 2
I have restarted this thread having now read and I think sort of understand the two papers on which the original announcements were made. I'll try and summarise where I am at which may or may not be helpful to others.
There are two documents, one by Barren et al and the second by Holmen et al. Quite a few people appear on both lists of authors so it can be assumed that the papers are complementary and produced with collaboration rather than rivalry.
Both papers look at the mortality data for COVID up to the 1st of May. Both use the data collected, and reported, by GPs to determine presence of diabetes in the population at large and in people who have died with the COVID virus present. Both paper use statistical methods to try and separate out the effect of the presence of diabetes from other factors that influence the severity of the effect of COVID. This needs to be done because the presence of diabetes can be associated with age and ethnicity and these confound ( that is obscure) the effect if not treated properly.
The Barren paper is the simplest. It starts with the simple observation that a lot of people who have died with COVID had a diabetes diagnosis, far more than would be expected if you look at the prevalence of diabetes in the general population. Some numbers. There are 63 and a bit million people in the study of whom 2.8 million have type 2 diabetes, a ratio of 0.04:1 At the time of the report 28000 people had died with COVID of which 7500 had type 2 diabetes a ratio of 0.26:1 Note, I have rounded the numbers a bit. At first sight this might suggest that you are six times more likely to die of COVID if you have diabetes than if you don't but like all things numeric, a little more scrutiny is needed, which is what the authors did. What they did was to separate out the effect of diabetes from the effect of age, ethnicity, and a couple of other things that might be important. This was done statistically using methods I only vaguely understand - its years since I did that sort of stuff - but it is reasonable to assume that knew what they were doing.
What they came up with is an odds ratio. You take a characteristic, set a reference and then get the odds of finding people with other values of the characteristic in the group you are looking at. Take the age data after eliminating the effects of all the other things. Their reference age is the range 60-69, which has the odds ratio of 1. If you look at the age range 80+, this ratio is just over 9. Quite how you headline that I don't know but it illustrates the fact that age is a big factor when it comes to effect of COVID. For the diabetes data (removing the effects of age and all the other stuff) they have used no diabetes as the reference. The odds ratio for T1 is 3.5 and T2 it is 2. All the results are significant.
So, what they have shown is that when all other things have been taken into account, there are twice as many T2's in the death stats than you might have predicted suggesting that the presence of diabetes in itself has some influence of the severity of the outcome of a COVID infection. It is small compared with age. A simple look suggests that having T2 is something like putting 10 years on your age once you are over 50.
The Holmen paper digs a bit deeper. It uses the same data sources but as far as I can see a different subset to the Barren paper. The interesting thing is that they introduce HbA1c and BMI as variables although the annoying thing is that they calculate a different statistic called the hazard ratio. Quite how that relates to the "odds ratio" I don't know. Eddy Edson might, he's more au fait with the statistical methods.
What this paper shows is that age and sex (quality not quantity) is the big factor, just like the Barren paper. It also shows that HbA1c has an effect. They take range, 49-53mmol/mol as their reference and the hazard ratio increases as HbA1c increases. Nowhere near the effect of age, but it is there. It also shows an effect of duration of diagnosis, particularly for T1. Same with BMI. There is an effect of BMI which is more pronounced in T1 than T2.
So, what do I make of it? The studies do show that patients with diabetes are more prevalent in mortality statistics than those without diabetes. Out of control diabetes makes things a bit worse as does being overweight. Age alone dominates everything. To my mind the findings don't warrant scary headlines but they should be brought to the attention of people with diabetes so that they can work out what is best for them.
An interesting question which is not looked at is whether the same pattern is seen as you go down the severity scale, for example, do more people with diabetes finish up in hospital than you might expect and do more people with diabetes become infected than you might expect.
Anyway, stop the seriousness and to go into newspaper mode. DocB's summary is:-
If you are a young, well-controlled, svelte diabetic then carry on. If you get it then you are in the same boat as the rest of the population.
If you are old, fat, male and have a screaming HbA1c then it's a bad idea to inhabit crowded sweaty places, unless you don't give a monkeys and are determined to have a good time before you go.
Everybody else is somewhere in the middle, but if you are a lady, then stay that way, and in any case don't grow old.
There are two documents, one by Barren et al and the second by Holmen et al. Quite a few people appear on both lists of authors so it can be assumed that the papers are complementary and produced with collaboration rather than rivalry.
Both papers look at the mortality data for COVID up to the 1st of May. Both use the data collected, and reported, by GPs to determine presence of diabetes in the population at large and in people who have died with the COVID virus present. Both paper use statistical methods to try and separate out the effect of the presence of diabetes from other factors that influence the severity of the effect of COVID. This needs to be done because the presence of diabetes can be associated with age and ethnicity and these confound ( that is obscure) the effect if not treated properly.
The Barren paper is the simplest. It starts with the simple observation that a lot of people who have died with COVID had a diabetes diagnosis, far more than would be expected if you look at the prevalence of diabetes in the general population. Some numbers. There are 63 and a bit million people in the study of whom 2.8 million have type 2 diabetes, a ratio of 0.04:1 At the time of the report 28000 people had died with COVID of which 7500 had type 2 diabetes a ratio of 0.26:1 Note, I have rounded the numbers a bit. At first sight this might suggest that you are six times more likely to die of COVID if you have diabetes than if you don't but like all things numeric, a little more scrutiny is needed, which is what the authors did. What they did was to separate out the effect of diabetes from the effect of age, ethnicity, and a couple of other things that might be important. This was done statistically using methods I only vaguely understand - its years since I did that sort of stuff - but it is reasonable to assume that knew what they were doing.
What they came up with is an odds ratio. You take a characteristic, set a reference and then get the odds of finding people with other values of the characteristic in the group you are looking at. Take the age data after eliminating the effects of all the other things. Their reference age is the range 60-69, which has the odds ratio of 1. If you look at the age range 80+, this ratio is just over 9. Quite how you headline that I don't know but it illustrates the fact that age is a big factor when it comes to effect of COVID. For the diabetes data (removing the effects of age and all the other stuff) they have used no diabetes as the reference. The odds ratio for T1 is 3.5 and T2 it is 2. All the results are significant.
So, what they have shown is that when all other things have been taken into account, there are twice as many T2's in the death stats than you might have predicted suggesting that the presence of diabetes in itself has some influence of the severity of the outcome of a COVID infection. It is small compared with age. A simple look suggests that having T2 is something like putting 10 years on your age once you are over 50.
The Holmen paper digs a bit deeper. It uses the same data sources but as far as I can see a different subset to the Barren paper. The interesting thing is that they introduce HbA1c and BMI as variables although the annoying thing is that they calculate a different statistic called the hazard ratio. Quite how that relates to the "odds ratio" I don't know. Eddy Edson might, he's more au fait with the statistical methods.
What this paper shows is that age and sex (quality not quantity) is the big factor, just like the Barren paper. It also shows that HbA1c has an effect. They take range, 49-53mmol/mol as their reference and the hazard ratio increases as HbA1c increases. Nowhere near the effect of age, but it is there. It also shows an effect of duration of diagnosis, particularly for T1. Same with BMI. There is an effect of BMI which is more pronounced in T1 than T2.
So, what do I make of it? The studies do show that patients with diabetes are more prevalent in mortality statistics than those without diabetes. Out of control diabetes makes things a bit worse as does being overweight. Age alone dominates everything. To my mind the findings don't warrant scary headlines but they should be brought to the attention of people with diabetes so that they can work out what is best for them.
An interesting question which is not looked at is whether the same pattern is seen as you go down the severity scale, for example, do more people with diabetes finish up in hospital than you might expect and do more people with diabetes become infected than you might expect.
Anyway, stop the seriousness and to go into newspaper mode. DocB's summary is:-
If you are a young, well-controlled, svelte diabetic then carry on. If you get it then you are in the same boat as the rest of the population.
If you are old, fat, male and have a screaming HbA1c then it's a bad idea to inhabit crowded sweaty places, unless you don't give a monkeys and are determined to have a good time before you go.
Everybody else is somewhere in the middle, but if you are a lady, then stay that way, and in any case don't grow old.
