multicomorbidity analysi 0210

*Calculation of outcome measures
*Load files on baseline and follow-up;
LIBNAME workingd "E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\";
DATA id_list;
	SET workingD.id_list0210;
RUN;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\id_member_LCA.csv"
out=id_member_LCA  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;	

*Merge membership;
PROC sql;
	CREATE table id_list as
	SELECT a.*,
			b.LCA_model_predclass
	FROM id_list a
	LEFT JOIN id_member_LCA b
	ON a.idcard=b.idcard;
QUIT;
DATA id_list1;
	SET id_list;
	IF LCA_model_predclass=. THEN LCA_model_predclass=0;
RUN;

PROC FREQ DATA=id_list1;
	TABLE LCA_model_predclass;
RUN;

*Load ipt and opt data;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\OPT.csv"
out=OPT  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\IPT.csv"
out=IPT  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\OPT_pharm.csv"
out=OPT_pharm  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

DATA IPT;
	SET IPT;
	newadm_time = input(adm_time,yymmdd10.);
	RENAME newadm_time=adm_time;
	newfee1 = input(fee1, 9.);
   drop fee1;
   rename newfee1=fee1;
   	newselffee = input(selffee, 9.);
   drop selffee;
   rename newselffee=selffee;
RUN;
DATA IPT;
	SET IPT;
	year=YEAR(newadm_time);
IF fee1=. THEN fee1=0;
IF selffee=. THEN selffee=0;
IF year=2014 THEN fee1adj=fee1;
IF year=2015 THEN fee1adj=fee1*0.9858;
IF year=2016 THEN fee1adj=fee1*0.9665;
IF year=2017 THEN fee1adj=fee1*0.9516;
IF year=2018 THEN fee1adj=fee1*0.9320;
IF year=2019 THEN fee1adj=fee1*0.9057;
IF year=2020 THEN fee1adj=fee1*0.8800;
IF year=2014 THEN selffeeadj=selffee;
IF year=2015 THEN selffeeadj=selffee*0.9858;
IF year=2016 THEN selffeeadj=selffee*0.9665;
IF year=2017 THEN selffeeadj=selffee*0.9516;
IF year=2018 THEN selffeeadj=selffee*0.9320;
IF year=2019 THEN selffeeadj=selffee*0.9057;
IF year=2020 THEN selffeeadj=selffee*0.8800;
KEEP idcard newadm_time dt_name fee1adj selffeeadj;
RUN;
DATA IPT;
	SET IPT;
RENAME newadm_time=adm_time;
RUN;


PROC sql;
	CREATE table IPT_baseline as
	SELECT a.*
	FROM IPT a
	INNER JOIN id_list1 b
	ON a.idcard=b.idcard AND a.adm_time<b.Dx_date AND a.adm_time>=b.Dx_date-365;
QUIT;
PROC sql;
	CREATE table IPT_folup as
	SELECT a.*
	FROM IPT a
	INNER JOIN id_list1 b
	ON a.idcard=b.idcard AND a.adm_time>=b.Dx_date AND a.adm_time<=b.Rx_end_date_x;
QUIT;
*number of admission;
PROC sql;
	CREATE table IPT_baseline as
	SELECT min(idcard) as idcard,
			count(adm_time) as adm_num_baseline,
			sum(fee1adj) as fee1ipt_baseline,
			sum(selffeeadj) as selffeeipt_baseline
	FROM IPT_baseline
	GROUP BY idcard;
QUIT;
PROC sql;
	CREATE table IPT_folup as
	SELECT min(idcard) as idcard,
			count(adm_time) as adm_num_folup,
			sum(fee1adj) as fee1ipt_folup,
			sum(selffeeadj) as selffeeipt_folup
	FROM IPT_folup
	GROUP BY idcard;
QUIT;

proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\OPT.csv"
out=OPT  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;


proc import datafile="E:\Ningbo_Projects\TB\TB project\Workingd\CSV files\OPT_pharm.csv"
out=OPT_pharm  dbms=csv replace;
getnames=yes;
guessingrows=10000;
RUN;

DATA OPT;
	SET OPT;
	IF clinic_time ^=.;
	year=YEAR(clinic_time);
	newfee1 = input(fee1, 9.);
   drop fee1;
   rename newfee1=fee1;
   	newselffee = input(selffee, 9.);
   drop selffee;
   rename newselffee=selffee;
RUN;
DATA OPT;
	SET OPT;
IF fee1=. THEN fee1=0;
IF selffee=. THEN selffee=0;
IF fee1>100000 THEN DELETE;
IF year=2014 THEN fee1adj=fee1;
IF year=2015 THEN fee1adj=fee1*0.9858;
IF year=2016 THEN fee1adj=fee1*0.9665;
IF year=2017 THEN fee1adj=fee1*0.9516;
IF year=2018 THEN fee1adj=fee1*0.9320;
IF year=2019 THEN fee1adj=fee1*0.9057;
IF year=2020 THEN fee1adj=fee1*0.8800;
IF year=2014 THEN selffeeadj=selffee;
IF year=2015 THEN selffeeadj=selffee*0.9858;
IF year=2016 THEN selffeeadj=selffee*0.9665;
IF year=2017 THEN selffeeadj=selffee*0.9516;
IF year=2018 THEN selffeeadj=selffee*0.9320;
IF year=2019 THEN selffeeadj=selffee*0.9057;
IF year=2020 THEN selffeeadj=selffee*0.8800;
	KEEP idcard clinic_time dt_name fee1adj selffeeadj year;
RUN;

DATA OPT_pharm;
	SET OPT_pharm;
	IF clinic_time ^=.;
	year=YEAR(clinic_time);
	newfee1 = input(fee1, 9.);
   drop fee1;
   rename newfee1=fee1;
   	newselffee = input(selffee, 9.);
   drop selffee;
   rename newselffee=selffee;
RUN;
DATA OPT_pharm;
	SET OPT_pharm;
IF fee1=. THEN fee1=0;
IF selffee=. THEN selffee=0;
IF fee1>100000 THEN DELETE;
IF year=2014 THEN fee1adj=fee1;
IF year=2015 THEN fee1adj=fee1*0.9858;
IF year=2016 THEN fee1adj=fee1*0.9665;
IF year=2017 THEN fee1adj=fee1*0.9516;
IF year=2018 THEN fee1adj=fee1*0.9320;
IF year=2019 THEN fee1adj=fee1*0.9057;
IF year=2020 THEN fee1adj=fee1*0.8800;
IF year=2014 THEN selffeeadj=selffee;
IF year=2015 THEN selffeeadj=selffee*0.9858;
IF year=2016 THEN selffeeadj=selffee*0.9665;
IF year=2017 THEN selffeeadj=selffee*0.9516;
IF year=2018 THEN selffeeadj=selffee*0.9320;
IF year=2019 THEN selffeeadj=selffee*0.9057;
IF year=2020 THEN selffeeadj=selffee*0.8800;
	KEEP idcard clinic_time dt_name fee1adj selffeeadj year;
RUN;

PROC SORT DATA=OPT NODUP; 
BY idcard;
RUN;
PROC SORT DATA=OPT_pharm NODUP; 
BY idcard;
RUN;

*list of insurance;
DATA insurance_list;
	SET OPT OPT_pharm IPT;
	IF dt_name="NA" THEN DELETE;
RUN;
PROC sql;
	CREATE table insurance_list as
	SELECT min(idcard) as idcard,
			min(dt_name) as dt_name
	FROM insurance_list
	GROUP BY idcard;
QUIT;

PROC sql;
	CREATE table id_list2 as
	SELECT a.*,
			b.*
	FROM id_list1 a
	LEFT JOIN insurance_list b
	ON a.idcard=b.idcard;
QUIT;

PROC sql;
	CREATE table OPT_baseline as
	SELECT a.*
	FROM OPT a
	INNER JOIN id_list1 b
	ON a.idcard=b.idcard AND a.clinic_time<b.Dx_date AND a.clinic_time>=b.Dx_date-365;
QUIT;
PROC sql;
	CREATE table OPT_folup as
	SELECT a.*
	FROM OPT a
	INNER JOIN id_list1 b
	ON a.idcard=b.idcard AND a.clinic_time>=b.Dx_date AND a.clinic_time<=b.Rx_end_date_x;
QUIT;
*number of admission;
PROC sql;
	CREATE table OPT_baseline as
	SELECT min(idcard) as idcard,
			sum(fee1adj) as fee1OPT_baseline,
			sum(selffeeadj) as selffeeOPT_baseline
	FROM OPT_baseline
	GROUP BY idcard;
QUIT;
PROC sql;
	CREATE table OPT_folup as
	SELECT min(idcard) as idcard,
			sum(fee1adj) as fee1OPT_folup,
			sum(selffeeadj) as selffeeOPT_folup
	FROM OPT_folup
	GROUP BY idcard;
QUIT;

PROC sql;
	CREATE table OPT_pharm_baseline as
	SELECT a.*
	FROM OPT_pharm a
	INNER JOIN id_list1 b
	ON a.idcard=b.idcard AND a.clinic_time<b.Dx_date AND a.clinic_time>=b.Dx_date-365;
QUIT;
PROC sql;
	CREATE table OPT_pharm_folup as
	SELECT a.*
	FROM OPT_pharm a
	INNER JOIN id_list1 b
	ON a.idcard=b.idcard AND a.clinic_time>=b.Dx_date AND a.clinic_time<=b.Rx_end_date_x;
QUIT;
*number of admission;
PROC sql;
	CREATE table OPT_pharm_baseline as
	SELECT min(idcard) as idcard,
			sum(fee1adj) as fee1OPT_pharm_baseline,
			sum(selffeeadj) as selffeeOPT_pharm_baseline
	FROM OPT_pharm_baseline
	GROUP BY idcard;
QUIT;
PROC sql;
	CREATE table OPT_pharm_folup as
	SELECT min(idcard) as idcard,
			sum(fee1adj) as fee1OPT_pharm_folup,
			sum(selffeeadj) as selffeeOPT_pharm_folup
	FROM OPT_pharm_folup
	GROUP BY idcard;
QUIT;

*Merge the ipt opt and optpharm list to the original cohort list;
PROC sql;
	CREATE table id_list3 as
	SELECT a.*,
			b.*
	FROM id_list2 a
	LEFT JOIN IPT_baseline b
	ON a.idcard=b.idcard;
QUIT;

PROC sql;
	CREATE table id_list3 as
	SELECT a.*,
			b.*
	FROM id_list3 a
	LEFT JOIN IPT_folup b
	ON a.idcard=b.idcard;
QUIT;
PROC sql;
	CREATE table id_list3 as
	SELECT a.*,
			b.*
	FROM id_list3 a
	LEFT JOIN OPT_baseline b
	ON a.idcard=b.idcard;
QUIT;
PROC sql;
	CREATE table id_list3 as
	SELECT a.*,
			b.*
	FROM id_list3 a
	LEFT JOIN OPT_folup b
	ON a.idcard=b.idcard;
QUIT;
PROC sql;
	CREATE table id_list3 as
	SELECT a.*,
			b.*
	FROM id_list3 a
	LEFT JOIN OPT_pharm_baseline b
	ON a.idcard=b.idcard;
QUIT;
PROC sql;
	CREATE table id_list3 as
	SELECT a.*,
			b.*
	FROM id_list3 a
	LEFT JOIN OPT_pharm_folup b
	ON a.idcard=b.idcard;
QUIT;

DATA id_list4;
	SET id_list3;
	IF fee1ipt_folup=. THEN fee1ipt_folup=0;
	IF fee1OPT_folup=. THEN fee1OPT_folup=0;
	IF fee1OPT_pharm_folup=. THEN fee1OPT_pharm_folup=0;
	IF selffeeipt_folup=. THEN selffeeipt_folup=0;
	IF selffeeOPT_folup=. THEN selffeeOPT_folup=0;
	IF selffeeOPT_pharm_folup=. THEN selffeeOPT_pharm_folup=0;
	feeall=fee1ipt_folup+fee1OPT_folup+fee1OPT_pharm_folup;
	selffeeall=selffeeipt_folup+selffeeOPT_folup+selffeeOPT_pharm_folup;
	IF feeall=. OR feeall<0 THEN feeall=0;
	IF selffeeall=. OR selffeeall<0 THEN selffeeall=0;
RUN;

PROC MEANS DATA=id_list4;
	CLASS LCA_model_predclass;
	VAR feeall selffeeall;
RUN;

*Prevalence of mental health conditions;
LIBNAME workingD "E:\Ningbo_Projects\TB\TB project\Workingd\CSV files";
DATA condition_all;
	SET workingD.cci_base_summary;
	IF CCI="B" OR CCI="C";
RUN;
PROC sql;
	CREATE table condition_list as
	SELECT min(label) as label,
			count(idcard) as num_pt
	FROM condition_all
	GROUP BY label;
QUIT;

proc print data=condition_list;RUN;

DATA MH_condition_all;
	SET condition_all;
	IF label in ("Anxiety disorder, unspecified","Bipolar disorder, current episode manic severe with psychotic features",
"Bipolar disorder, current episode manic without psychotic features, unspecified",
"Dissociative and conversion disorder, unspecified","Major depressive disorder, single episode, unspecified",
"Mental disorder, not otherwise specified","Manic episode, unspecified","Schizophrenia, unspecified");
RUN;

PROC sql;
	CREATE table MH_list as
	SELECT min(idcard) as idcard,
			1 as MH_baseline_status
	FROM MH_condition_all
	GROUP BY idcard;
QUIT;

PROC sql;
	CREATE table workingD.id_list_multicormbd_0210 as
	SELECT a.*,
			b.*
	FROM id_list4 a
	LEFT JOIN MH_list b
	ON a.idcard=b.idcard;
QUIT;

PROC sql;
	CREATE table condition_count_list as
	SELECT min(idcard) as idcard,
			count(label) as condition_count
	FROM condition_all
	GROUP BY idcard;
QUIT;

PROC sql;
	CREATE table workingD.id_list_multicormbd_0210 as
	SELECT a.*,
			b.*
	FROM workingD.id_list_multicormbd_0210 a
	LEFT JOIN condition_count_list b
	ON a.idcard=b.idcard;
QUIT;


proc genmod data= temp; 
	class LCA_model_predclass (ref="0") ; 
	model pos_feeall = LCA_model_predclass / dist=binomial link=log type3; 
	output out=dataname pred=var1 resdev=var2 leverage=var3 cooksd=var4;
run;

proc genmod data= temp; 
	class LCA_model_predclass (ref="0") ; 
	model feeall = LCA_model_predclass / dist=gamma link=log type3; 
	output out=dataname pred=var1 resdev=var2 leverage=var3 cooksd=var4;
run;

PROC MEANS DATA=dataname;
	CLASS LCA_model_predclass;
	VAR var1;
RUN;

PROC FREQ DATA=workingD.id_list_multicormbd_0210;
	TABLE dt_type;
RUN;
DATA workingD.id_list_multicormbd_0210;
	SET workingD.id_list_multicormbd_0210;
IF dt_name in ("其他,社会基本医疗保险", "其他社会保险","其他社会保险,社会基本医疗保险","城镇（城乡）居民基本医疗保险","社会基本医疗保险") THEN dt_type="URRMI";
IF dt_name in ("城镇职工基本医疗保险", "城镇职工基本医疗保险,贫困救助,社会基本医疗保险","其他社会保险,社会基本医疗保险","城镇（城乡）居民基本医疗保险") THEN dt_type="UEBMI";
IF dt_type="" THEN dt_type="other";
RUN;

PROC FREQ DATA=workingD.id_list_multicormbd_0210;
	TABLE dt_type;
RUN;

PROC MEANS DATA=workingD.id_list_multicormbd_0210 skew;
	CLASS LCA_model_predclass;
	VAR feeall selffeeall;
RUN;

DATA temp;
	SET workingD.id_list_multicormbd_0210;
	*IF feeall>0;
	*IF LCA_model_predclass ^=0;
	success=0;
	*IF Rx_outcome="Rx_completed" OR Rx_outcome="Cured" THEN success=1;
	*IF adm_num_folup=. THEN success=0;
	IF selffeeall ^=0 THEN success=1;
	pos_feeall=1;
	IF feeall=0 THEN pos_feeall=0;
	IF condition_count=. THEN condition_count=0;
	length_mo=(min(Rx_end_date_x,Death_date,22096)-Dx_date)/30;
	IF length_mo<=0 THEN length_mo=1/30;
	fee1ipt_folup_mo=fee1ipt_folup/length_mo;
	fee1opt_folup_mo=fee1opt_folup/length_mo;
	fee1OPT_pharm_folup_mo=fee1OPT_pharm_folup/length_mo;
	feeall_mo=feeall/length_mo;
	selffeeall_mo=selffeeall/length_mo;
RUN;

proc genmod data= temp; 
	class Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass (ref="0");
	model feeall = age_cal Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass/ dist=gamma link=log type3; 
	output out=dataname pred=var1 resdev=var2 leverage=var3 cooksd=var4;
run;

proc genmod data= temp; 
	class Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass (ref="0");
	model feeall = LCA_model_predclass/ dist=gamma link=log type3; 
	output out=dataname pred=var1 resdev=var2 leverage=var3 cooksd=var4;
run;

proc genmod data= temp; 
	class Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass (ref="0");
	model selffeeall = age_cal Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass/ dist=gamma link=log type3; 
	output out=dataname pred=var1 resdev=var2 leverage=var3 cooksd=var4;
run;

proc genmod data= temp; 
	class Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass (ref="0");
	model selffeeall = LCA_model_predclass/ dist=gamma link=log type3; 
	output out=dataname pred=var1 resdev=var2 leverage=var3 cooksd=var4;
run;

PROC MEANS DATA=dataname n MEAN;
	CLASS LCA_model_predclass;
	VAR VAR1;
RUN;


PROC FREQ DATA=workingD.id_list_multicormbd_0210;
	TABLE age_cat Sex_y dt_type Dx_type___11 Dx_type___18 CCI_cat condition_count/nopercent norow nocol;
RUN;

PROC MEANS  DATA=temp MEAN STD MEDIAN;
	VAR  age_cal condition_count;
RUN;

PROC FREQ DATA=temp;
	TABLE age_cat*LCA_model_predclass 
Sex_y*LCA_model_predclass
dt_type*LCA_model_predclass
Dx_type___11*LCA_model_predclass
Dx_type___18*LCA_model_predclass
CCI_cat*LCA_model_predclass
condition_count*LCA_model_predclass/nopercent norow nocol chisq;
RUN;

PROC MEANS  DATA=temp MEAN STD MEDIAN P25 P75 skew;
	*class LCA_model_predclass;
	VAR  fee1ipt_folup fee1opt_folup fee1OPT_pharm_folup feeall selffeeall;
RUN;

PROC MEANS  DATA=temp n MEAN STD MEDIAN P25 P75 skew;
	*class LCA_model_predclass;
	VAR  fee1ipt_folup_mo fee1opt_folup_mo fee1OPT_pharm_folup_mo feeall_mo selffeeall_mo;
RUN;

PROC ANOVA DATA=temp;
	class LCA_model_predclass;
	MODEL fee1ipt_folup_mo=LCA_model_predclass;
RUN;

PROC FREQ DATA=temp;
	TABLE Rx_outcome*LCA_model_predclass/nopercent norow nocol chisq;
RUN;

PROC GLMSELECT DATA=temp ;
	class LCA_model_predclass (ref="0");
	MODEL success = LCA_model_predclass /selection=lasso(stop=none choose=cvex);
	output out=dataname pred=var1 ;
RUN;


 proc logistic data=temp;
      class Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass (ref="0");
      model success  (ref= "0")= age_cal Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass/ expb;
   run;
 proc logistic data=temp;
      class Sex_y dt_type Dx_type___11 Dx_type___18 LCA_model_predclass (ref="0");
      model success  (ref= "0")= LCA_model_predclass/ expb;
   run;