# How do they know how many undiagnosed?

Discussion in 'Type 2 Diabetes' started by danziger, Apr 21, 2022.

1. ### danziger Type 2 · Well-Known Member

Messages:
145
167
Trophy Points:
83
I’m doing some research for a writing project that touches on my experience of T2 but don’t have a science background so don’t really get this.

When researchers say how many people have type 2 but don’t know it, what is that based on/how is it calculated/what is that extrapolated from?

TIA to anyone who can help!

#1
2. ### KennyA Type 2 (in remission!) · Moderator Staff Member

Messages:
739
404
Trophy Points:
123
In my experience this sort of guesstimate is usually based on an extrapolation. For instance: someone looks at a number of people who have (eg) not been diagnosed with T2, and then tests their blood glucose. Suppose 100 are tested and 10 have glucose at "diabetic" levels. So 10% of undiagnosed people "have diabetes" and don't know it. So...10% of the country's undiagnosed population can be claimed to "have diabetes" and be unaware. There's a lot wrong with this sort of calculation - the initial test population is very small, it may not be typical of the wider population, etc. It's a huge jump to extrapolate from a small sample one-off to making statements about the population, but it's done all the time. The attached article is a good read on junk science. https://gizmodo.com/i-fooled-millions-into-thinking-chocolate-helps-weight-1707251800

• Agree x 2
#2
3. ### danziger Type 2 · Well-Known Member

Messages:
145
167
Trophy Points:
83
Ahhh, thank you! I guess I need to track down some of these studies and take a good look at the methodology. It’s wild how these stats are taken as fact and repeated over and over.

#3
4. ### ianf0ster Type 2 · Well-Known Member

Messages:
1,608
728
Trophy Points:
173
Not just those types of studies!
Dodgy methodology or selective subject recruitment (biased towards sponsor's desired outcome) are rife. I think an ex-editor of the BMJ suggested well over 50 % of studies are actually biased garbage.

• Agree x 1
#4
5. ### Ronancastled Type 2 · Well-Known Member

Messages:
1,028
655
Trophy Points:
173
They reckon 1/3 of Americans are pre-diabetic & 90% of these don't even know it.

It's a headline grabber for sure

#5
6. ### jape Type 2 · Well-Known Member

Messages:
149
87
Trophy Points:
88
I don't know how they made the calculation, but here is how I would do it:

Take a random sample of people (say a 1,000) properly stratified among different demographics that may indicate a higher/lower prevalence for diabetes such as gender, age, ethnicity, income levels, regional, lifestyle etc to reflect the entire population. Then test everyone in the sample for diabetes, and determine the percentage of diabetics. Apply this percentage to the entire population, and compare with the actual known number of diagnosed diabetics to determine the number of undiagnosed diabetics. Of course, this is an estimate and there will be a statistical error factor.

This process is broadly similar to polling during election times.

• Agree x 1
#6
7. ### Oldvatr Type 2 · Expert

Messages:
6,952
3,790
Trophy Points:
198
• Agree x 1
#7
8. ### KennyA Type 2 (in remission!) · Moderator Staff Member

Messages:
739
404
Trophy Points:
123
The PHE (not UK gov, only England) paper referred to stresses "estimates" throughout. It is still guesswork based on assumptions - maybe much better assumptions but data are soft. Getting access to actual medical records even for the purposes of research is difficult (in some cases impossible) time consuming and expensive.

Edited to include a link to the relevant organization for England. https://www.hra.nhs.uk/approvals-amendments/what-approvals-do-i-need/confidentiality-advisory-group/

#8
Last edited: Apr 22, 2022
• ### Meet the Community

Find support, connect with others, ask questions and share your experiences with people with diabetes, their carers and family.

Did you know: 7 out of 10 people improve their understanding of diabetes within 6 months of being a Diabetes Forum member. Get the Diabetes Forum App and stay connected on iOS and Android

Grab the app!