I'm trying to use propensity scores to compare outcomes between 'cases' and 'controls.' I have 24 cases and 200 controls. I used proc psmatch using the greedy neighbor option (version SAS 9.4). I successfully created the propensity scores output. I asked for 3:1 matching, but my output was closer to 2:1.
ods graphics on;
proc psmatch data=ribs.casecontrols region=cs;
class obs gender ethnicity dm cad mi chf htn kidney cva dementia
copd asthma smoker flail side;
psmodel obs(Treated='case')= gender ethnicity dm cad mi chf htn kidney
cva dementia copd asthma smoker flail side age bmi charlson
num_rib chest_ais ISS;
match method=greedy(k=3) stat=lps caliper=0.25;
assess lps var=(gender ethnicity dm cad mi chf htn kidney cva dementia
copd asthma smoker flail side age bmi charlson num_rib
chest_ais ISS) / weight=none plots=(boxplot barchart);
output out(obs=match)=Outgs lps=_Lps matchid=_MatchID;
run;
However, I'm at a loss of what do to next. My outcome variables are both continuous and categorical (all binary except one). How do I analyze the new dataset taking into account the newly matched observations with propensity scores?
Specifically, I need help with the code on how to run 1.) a chi-sq to compare the binary outcomes and both 2.) t-test and 3.) wilcoxon rank sum test for the continuous variables.
Lastly, 4.) I'm not sure how best to analyze the variable with more than 2 values (i.e., variable = 'dispo' and possible values = 'home' 'rehab' 'death' 'acute' 'hospice'). I can use chi-sq, but some of the cells are pretty low in counts.
