Ratio of number of cases to control

#1
Is taking 20 control and 90 patients in a 2 groups study statistically correct in comparing various parameters? I mean that the patient group is higher in number than control group.
> Is that statistically correct?
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
One needs to think why do they have a case and control group? So you are selected based on outcome status? Are you examining a treatment of interest and you desire to match on background covariates? If so, balancing these covariates may be more important than shear numbers.

The more observations the better is usually the mantra, but if the outcome groups get too imbalanced, than power isn't really being gained. For example, some algorithm may find it easier to classify the outcome as the majority group to achieve accuracy.

You should read on the approach before moving to far forward. Ideally you want to only access the records of patients needed to answer the question in order to respect privacy and minimize hypothetical risk to confidentiality related to data breaches.
 
#3
One needs to think why do they have a case and control group? So you are selected based on outcome status? Are you examining a treatment of interest and you desire to match on background covariates? If so, balancing these covariates may be more important than shear numbers.

The more observations the better is usually the mantra, but if the outcome groups get too imbalanced, than power isn't really being gained. For example, some algorithm may find it easier to classify the outcome as the majority group to achieve accuracy.

You should read on the approach before moving to far forward. Ideally you want to only access the records of patients needed to answer the question in order to respect privacy and minimize hypothetical risk to confidentiality related to data breaches.
Dear hlsmith
Thank you for your concern. I am examining a group of patients against a group of normal people and measuring some parameters to compare them between the 2 groups.
I hope that I have clarified the study design, Looking forward to your kind response.
 
#7
It should be okay, as you have lesser healthier subjects, better limit your tests to 1 or 2. for doing more tests you should take more healthy subject. read more about sample size in the DOE context.
 

hlsmith

Less is more. Stay pure. Stay poor.
#8
I was trying to get at, if this is actually a case-control study or two cohorts.

Of note, historically comparing healthy and a patient population can have paradoxical effect risks - see Berkson Paradox