propensity scoring with 4 levels of treatment exposure

niugrads1999 — Fri, 18 Mar 2022 14:56:03 GMT

I am unsure how to accomplish propensity score matching when there is a control group (0) and 3 levels of exposure (1,2,3). I have attempted to develop 3 different data sets comparing 0-1, 0-2, 0-3. The first dataset produces a new table with the expected number of observations. However, the second and third do not produce a matched dataset (see attached excel file). Not sure what the issues is?

/*propensity matching CMT 0 vs 1*/
proc psmatch data=SMTgroup01;
class CMT_group Gender;
psmodel CMT_group(Treated="1")=claim_count allowed gender age R_Risk;
match method=optimal(k=1) stat=lps exact=gender caliper=0.25;
output out(obs=match)=propscore01 matchid=_MatchID;
run;

/*propensity matching CMT 0 vs 2*/
proc psmatch data=SMTgroup02;
class CMT_group Gender;
psmodel CMT_group(Treated="2")=claim_count allowed gender age R_Risk;
match method=optimal(k=1) stat=lps exact=gender caliper=0.25;
output out(obs=match)=propscore02 matchid=_MatchID;
run;

/*propensity matching CMT 0 vs 3*/
proc psmatch data=SMTgroup03;
class CMT_group Gender;
psmodel CMT_group(Treated="3")=claim_count allowed gender age R_Risk;
match method=optimal(k=1) stat=lps exact=gender caliper=0.25;
output out(obs=match)=propscore03 matchid=_MatchID;
run;

topic propensity scoring with 4 levels of treatment exposure in Statistical Procedures

propensity scoring with 4 levels of treatment exposure