Logistic Regression with Mixed effects

#1
I have a logistic regression setup: Outcome = injury (y/n); predictors that are subject specific are gender, age, smoking, and predictors that are class level are $ funds, instructor experience, etc. I want to be able to predict injury based on the subject AND class level data. My setup is very similar to the example 1 in the link below. I've never run this kind of analysis. Does anyone have any suggestions on how to proceed?

http://www.ats.ucla.edu/stat/stata/dae/melogit.htm