03-16-2012 05:01 PM
I have a 4 level- variable that I want to use as one of the independent variable (renalicd9 ) in proc logiotic model,
The variable is coded as
0, (Normal renal finction)
1- Stage 3 chrnoic renal disease
2. Stage 4 chrnoic renal disease
3- Stage 5 chr rnal dis with dialysis.
I want to obtain the Odds ratio for mortality (variable "died") for each of this category compared against those with normal renal function.(code = 0)
Can I use this multilevel variable as it is in the regression model or should I convert into 3 separate binary variables (coded as 0 and 1) and the run 3 different models?
What would give me correct results?
At present after including the multilevel variable (highlighted) the the SAS code looks like this:
proc logistic data =library.nismicathcabg4 descending;
class female (ref= first) renalicd9 (ref=first) dm dmcx htn_c aids alcohol ANEMDEF arth race1(ref=first) ZIPINC_QRTL(ref=first) hosp_location h_contrl(ref=first) hosp_teach
bldloss chf chrnlung coag depress drug hypothy liver lymph lytes mets neuro obese para perivasc psych pulmcirc renlfail tumor
ulcer valve wghtloss cararrhythmia/param=ref;
model died= age female dm dmcx htn_c aids alcohol ANEMDEF arth race1 zipinc_qrtl hosp_location h_contrl hosp_teach TOTAL_DISC
bldloss chf chrnlung coag depress drug hypothy liver lymph lytes mets neuro obese para perivasc psych pulmcirc renalicd9 tumor ulcer valve wghtloss cararrhythmia;
title 'Logi Reg in-hosp mortality vs renalicd9 in MI patients with race + income';
The output i got after running this is attached here..
03-16-2012 05:15 PM
Recoding as binary is effectively what SAS does.
See the design matrix for the variable recoded on page1/2 of your output and your estimates on page 10.
YOU DO NOT NEED TO RUN THREE DIFF MODELS.
03-16-2012 05:26 PM
Thanks Reeza. Yes, I already had a look at the design variables and estimates and it looks good. But as i never used a multilevel indepedent variable in the model before, just wanted to confirm if it is alright to do and if the results are correct.
So, great, it means i can report the 3 Odds Ratios from page 10 of the output..