calculating composite score

[Kyra]

Hi,

I have a dataset with outcome Intubation (Yes/No).

I have some variables available in the dataset- age, ethnicity, symtoms at presentation like fever, headache, chest pain etc and laboratory values like AST, ALT etc.

I would like to calculate a composite score using those variable to predict Intubation.

Please let me know if there is a good resource available to learn how to do this.

[PaigeMiller]

Well, that's a very broad question, and there are probably hundreds of ways to compute composite scores. Can you provide any more guidance on what you want?

I'll mention two methods: Partial Least Squares regression (PROC PLS); and logistic regression (PROC LOGISTIC).

[Kyra]

I want to create a score using those variables and want to say that patients with composite score of maybe say 14 are much more likely to be intubated in comparison to patients with composite score of less than 14.

[PaigeMiller]

That's what Logistic regression does. And similarly Partial Least Squares regression can answer that question as well. And as long as I'm thinking about it, Discriminant Analysis can do that (PROC DISCRIM).

[Kyra]

Thanks! Once i have my logistic regression output i will get back to you if i still need help! Thank you very much!