org355: Doggie diabetes

From The Guardian and BMJ Christmas issue.

"...they discovered that owning a dog with diabetes was associated with a 38% increased risk of having type 2 diabetes compared with owning a healthy hound."

1. What is the RR figure for 38%? Explain what is in the numerator and the denominator to calculate the RR.

2. The word "having" is incorrect. What is the correct word? Why does this matter?

3. After looking at the BMJ article, I see that the Guardian journalist should not have reported the 38% figure. Why?

4. What are the ethical aspects involved with data collection for this study?

The incidence of diabetes in the pets was 1.3 cases per 1000 dog years at risk and 2.2 cases per 1000 cat years at risk.

5. How would you explain the above figures to a non-epidemiologist pet owner?

6. What do you think are the important confounders to consider, and why?

7. Related to #6, is a causal relationship possible?

8. Might this association be extended to hypertension? Educational video here

Re: Doggie diabetes

by | FATHIMA MINISHA - Saturday, 12 December 2020, 9:22 PM

Hello everybody...

1) 38% increase would mean a RR of 1.38. The numerator would be risk of diabetes in owners of dogs with diabetes. The denominator would be risk of diabetes in owners of healthy dogs without diabetes

2) The word 'having' could also mean that they had diabetes even before owning the dog. A better word might be "developing": they discovered that owning a dog with diabetes was associated with a 38% increased risk of developing type 2 diabetes compared with owning a healthy hound.

3) So the 38% increase reported in the article is actually the crude HR. The adjusted ratio is 1.32 (after adjusting for possible confounders. They should have reported a 32% increase.

4) Actually when I first read the article, the ethical aspect was what I though about first. The investigators were able to access health information just using peoples ID number from pet insurance companies. That feels like a violation of the privacy of these participants, because it feels like the pet insurance companies should not ideally be allowed to share client information without their prior consent- especially if their health information is linked to those personal information. Of course, I am not sure what is legally allowed in Sweden- maybe this is not considered a violation.

5) These figures talk about incidence rates. So I would say: There are 13 dogs that get diabetes every 10,000 years at risk, years at risk meaning the total number of years all the dogs are at a risk of getting diabetes.

6) Important confounders to consider here would be age of the owner and dog, sex of the owner, living circumstances, obesity in the owner (this I day because dogs reflect their owners moods and behavior more closely that any other pet), prevalence of diabetes in the area, genetic susceptibility of the owner maybe.

7) It doesn't seem like a actually causal relationship is possible. The CI of the estimate is very close to 1, so I really doubt that based on this study its highly unlikely. If an RCT is done, and healthy individuals randomly given dogs with and without diabetes to own, it might give an idea. But that could be and endless study- till how long would you ideally follow these ppl to see if they develop diabetes.

8) I wouldn't be surprised if a similar ratio is obtained with hypertension, because I think probably life style is one of significant underlying factors in both diseases and the probable reason for this association as well (even though the authors say they have adjusted for a lot of factors).

Fathima

Re: Doggie diabetes

by | Katherine Carr - Tuesday, 15 December 2020, 6:48 AM

1. for the risk ratio:

I thought the RR would have been calculated by looking at: ( number of dog owners with Type 2 Diabetes/ number of dogs with Type 2 diabetes) / (number of dog owners with Type 2 diabetes/number of healthy dogs). (though this is not clear from the Guardian article)

2. The study was looking at incidence of diabetes, not prevalence. The correct word should have been 'developing' not 'having' diabetes.

3. I looked at the BMJ and see that the Guardian's figure of 38% more likely to have diabetes if your dog has diabetes is calculated as follows: "crude hazard ratio for type 2 diabetes in owners of a dog with diabetes compared with owners of a dog without diabetes was 1.38 (95% confidence interval 1.10 to 1.74), with a multivariable adjusted hazard ratio of 1.32 (1.04 to 1.68)".

This means the Guardian has reported the crude rate, not the adjusted rate.

4. Ethical aspects: it seems that all of this information for the study was taken from private pet insurance records (Agria insurance) and public records (population register, patient register, cause of death register and a Database for Health Insurance and Labour Market Studies. It is interesting that people's personal health information is linked to their pet's health data - and that this is accessible (though anonymised) - are pet owners aware of this when they pay for pet insurance??

5. The incidence of diabetes in the pets was 1.3 cases per 1000 dog years at risk and 2.2 cases per 1000 cat years at risk. - I would explain this as:

this study was looking at incidence of diabetes in pets (and owners)- meaning during the study, how many cats and dogs developed diabetes? Since it would be expensive and difficult to track dogs and cats for their whole lives, researchers take a huge number of pets and monitor them for a shorter period giving you data for thousands of dog/cat/person years.

In this case they were looking at over 200,000 dog/owner pairs. When you see a rate of something per 1000 years, it is usually to help us make sense of a small number. In this case - there were 1.3 cases of dog diabetes/1000 dog years. If this had been reported per 100 dog years the number would have been 0.13 cases which isn't easily interpretable.

6. The study incorporates demographic data such as SES, country of birth, level of education but as it does not interview or survey pet owners, we don't know about their diet and exercise which presumably would have a big effect on their own likelihood of developing Type 2 diabetes, as well as an influence on their pets' health.

7. Causal relationship?? Given the influence that diet and exercise exert over likelihood of developing Type 2 diabetes, I don't think there could be a causal relationship between human incidence of diabetes and dog's diabetes status.

8. :))) chihuahua ownership may not lower blood pressure...

P.S Fathima - many kudos to you for being the most dedicated responder to Epi in the News! I've enjoyed reading all your answers!

Re: Doggie diabetes

by | RANMINI SUMUDITA KULARATNE - Friday, 18 December 2020, 8:58 PM

1. Risk ratio
RR figure for 38% is 1.38
Numerator = number of cases of diabetes in per 1000 PYAR in dog owners of dogs wih diabetes
Denominator = number of cases of diabetes per 1000 PYAR in dog owners of healthy dogs

2. Incidence (developing) is the correct word
“Having” equates to prevalence. Matters because no denominator data associated with prevalence to estimate person-time at risk and therefore cannot calculate RR

3. Adjusted hazard ratio should have been used - so 32% instead of 38%

4. Ethical issues
No protection of personal information
No consent given by persons for access of their data
Issues around data confidentiality

5. I like Fathima's explanation

6. Association attributed to shared dietary and physical activity patterns
Confounders: genetics/ family history; environmental factors, other conditions - autoimmune disease that are associated with diabetes

7. Causal relationship not proven and unlikely

Re: Doggie diabetes

by | OLGA VIACHESLAVOVNA KOZHAEVA - Tuesday, 26 January 2021, 10:24 PM

Hi all

adding to the points where I had something extra to add to colleagues above:
---
The incidence of diabetes in the pets was 1.3 cases per 1000 dog years at risk and 2.2 cases per 1000 cat years at risk.

5. How would you explain the above figures to a non-epidemiologist pet owner?

- Over one year, there would be an estimated 1300 new cases of diabetes in dogs and 2200 in cats?

would the above be correct?

7. Related to #6, is a causal relationship possible?

In addition to small OR for dogs, and keeping in mind the Bradford Hill criteria, one might note that there was no evidence of association after adjustment for cats. -By analogy, we may expect the results for dogs and cats to be similar, and the result in cats puts further question to whether the result in dogs are actually due to residual confounding, or perhaps some type of bias?

best wishes
Olga