Which test should I use to find the rose among the thorns in SAS studio?

[Konradc]

I have 200 patients who have prescriptions from a pool of 20 drugs, and 100 normal control volunteers who do not take any of the drugs in the pool of 20. Each patient has 0 to 6 prescriptions from the pool of 20 in random combinations. Some of these patients have elevated plasma levels of cholesterol, and I would like to know which of the drugs in the pool elevates the cholesterol levels.

To sum it up:
n=200 patients + 100 controls, each on a combination of up to 6 drugs from a pool of 20; some of the drugs appear only in combination with others
dependent variable: n=1 (cholesterol)

Question: Which drug(s) are responsible for the elevation in cholesterol. I already know that more than one of these drugs are responsible for the elevation, and I also know that some drugs have no effect. I would like to use SAS studio for the analysis.

Thank you

 

[hlsmith]

Well what if some drugs interact or some lower cholesterol? There are probably too many combinations to find associations with confidence. You also don't know their cholesterol level before the drug. Also dosages will very. This is why clinical trials are necessary along with randomization.

You can run linear regression but it will be very limited given your design options.

Proc glm