We investigate the prevalence of DM2 in patients with autoimmune thyroiditis and compare with the prevalence of DM2 in common population. The group consisted of 100 patients with AIT without until now documented glucose metabolism disorder. The prevalence of DM2 in our population (4 million adult citizens) is well documented. Can we use in this clinical-epidemiological study not the same range of 100 subjects but a higher number of 500 or 1000 as a control group? We want to create a control group not as a population sample but as a population. Thanks for your opinions.
Hello dr Rudolf
As dr Logan said : Answer is yes
In comparative epidemiological study , ratio of case group : control group can be : 1 case group : 1 control group, 1 : 2 , 1 : 3 , or 1 : 4 ( to increase power of the study) ...
Yes, the control group can be higher, but you should do a propensity score match to see how different the results would be adjusting for the variables that are not of interest. Also, even if you used 500 or 1,000 N as the control is still would be a sample as the population is 4,000,000 you would need to have that as your control sample (which would be your control population). Sorry for the speedy response. Ask for clarification if needed.
Sincerely,
Logan Netzer
Hello dr Rudolf
As dr Logan said : Answer is yes
In comparative epidemiological study , ratio of case group : control group can be : 1 case group : 1 control group, 1 : 2 , 1 : 3 , or 1 : 4 ( to increase power of the study) ...
Hi Rudolf
The answer is YES. In most of the case control studies, the ratio of cases to controls is usually is 1:1. However, this ratio may be increased to 1:2, 1:3, 1:4 or even higher. When we increase this ratio, the power of the study is increased which can detect a significant difference.
Yes, you can use 100 cases of AIT and 500 controls (samples) from the community.
For further details on the effect of case: control ratio on power of study, please see the link below:
https://genominfo.org/journal/view.php?number=292
For case control study, best result is obtained when case: control = 1:4. You can keep it like this. Beyond 4 is not desirable, as it will not give you any more data and error will be increased along with cost.
The answer is yes, the control group should be higher, which increase power of the study
Commonly the number of control is equal or more (2-3 times) than cases. Recruiting control is commonly easier than cases. Valid estimates of the odds ratio may be obtained from any ratio of cases and controls, as long as you follow appropriate sample selection processes. Logistic regression and confounding adjustment would be applied in the usual manner.
When using an OR then a bigger control group would give you more power, and you will be able to statistcally detect smaller differences.
however with a RR it will give a missleading answer.
- Badges
- Science topic
- Similar topics
- Medicine
- Disease
- Metabolic Diseases
- Glucose Metabolism Disorders