*************************************************************************************************;
* 
*           CLIENT:  ModernaTX, Inc.
*         PROTOCOL:  mRNA-1273-P201
*
*          PURPOSE: Create analysis dataset adsl
*      
*      INPUT FILES: SDTM domains
*     OUTPUT FILES: ADAE.sas7bdat
*
*      USAGE NOTES: 
*      
*************************************************************************************************;
*  (c) 2020 PPD
*  All Rights Reserved. 
*************************************************************************************************;

**Assign global macro variable DSETNAME to reflect the name of the final ADaM dataset**;
%include "madam.sas";

%global DSETNAME;
%let dsetname = adae;
%let adam_spec_loc = &g_projectpath.&g_toplevel.\Documents\Specs;
%let adam_spec = Moderna mRNA1273P201 ADaM spec.xlsm;
 
proc format;
	value $ atoxgr (multilabel)
		'GRADE 1/MILD','1'     = 'Grade 1' 
		'GRADE 2/MODERATE','2' = 'Grade 2' 
		'GRADE 3/SEVERE','3'   = 'Grade 3' 
		'GRADE 4' ,'4'         = 'Grade 4';
		
	invalue atoxgrn
		'GRADE 1/MILD','1'     = 1
		'GRADE 2/MODERATE','2' = 2
		'GRADE 3/SEVERE','3'   = 3
		'GRADE 4' ,'4'         = 4;

	value $ asev 
		'GRADE 1/MILD','1'     = 'Grade 1/Mild' 
		'GRADE 2/MODERATE','2' = 'Grade 2/Moderate' 
		'GRADE 3/SEVERE','3'   = 'Grade 3/Severe' 
	 	'GRADE 4' ,'4'         = 'Grade 4';
		
	invalue asevn
		'GRADE 1/MILD','1'     = 1
		'GRADE 2/MODERATE','2' = 2
		'GRADE 3/SEVERE','3'   = 3
		'GRADE 4' ,'4'         = 4;

	invalue  aereln 
		"NOT RELATED"      = 1
		"RELATED"          = 2
		"NOT APPLICABLE"   = 3;	
	
	picture three
		low-high= "999"; 
run;

proc sort data = adb.adsl out =adsl (keep = tr01sdt tr01sdtm TR01edt TR01edtm  DOSE2DT cutoffdt  DOS2DTM  EOSDT usubjid);
	by usubjid;
run;
 
%revsupp(
libin=TRANS, 
libout=WORK, 
ds=AE, 
supp=SUPPAE,
maploc=&G_PROJECTPATH.\&G_toplevel.\Databases\Transformed,
mapspec=&G_NICKNAME._mapping_spec.xlsx
);

%revsupp(
libin=TRANS, 
libout=WORK, 
ds=HO, 
supp=SUPPHO,
maploc=&G_PROJECTPATH.\&G_toplevel.\Databases\Transformed,
mapspec=&G_NICKNAME._mapping_spec.xlsx
);
 
data ae;
	set ae;
	aespid1 = strip(tranwrd(aespid, 'AE-',''));
run;
data HO;
	set HO;
	hospid1 = strip(tranwrd(hospid, 'HO-',''));
run;

proc sql noprint;
	create table  ae1 as select distinct
		a.*,
		b.HOSTDTC as AEHSTDTC,
		b.HOENDTC as AEHENDTC,
	    strip(c.HOOCCUR) as ICUAD length = 20,
	    c.HODUR as ICUDAY
	    from ae a 
	    	left join HO (where = (upcase(HOTERM)  in ('HOSPITAL'))) b 
	    		on a.usubjid = b.usubjid and a.aespid1 =  b.hospid1
		    left join HO (where = (upcase(HOTERM)   in ('ADMITTED TO ICU'))) c 
	    		on a.usubjid = c.usubjid and a.aespid1 =  c.hospid1 ;
	   quit;
	    
data adae1;
length TOXGGR1 AREL AEHSTDTC AEHENDTC $20  AESOFL $1.  AEMAFL ASEV $20 lnkpid $200;
	merge ae1 (in=a rename=(AESOFL = _AESOFL  AEHSTDTC = _AEHSTDTC AEHENDTC = _AEHENDTC AEMAFL=_AEMAFL) where  = (~missing(AETERM) and REMOVEFL ne 'Y'))
          adsl (in=b);
	by usubjid;
	if a and b;
	
	*supp vars; 
	aesofl      = _aesofl; 
    AEHSTDTC    = _AEHSTDTC;
    AEHENDTC    = _AEHENDTC;
	AEMAFL      = _AEMAFL;
	
	if length(AEENDTC) >= 10  then AENDT  = input(AEENDTC, yymmdd10.);;
	if AEENDTC ne ''          then AENDT  = input(AEENDTC, yymmdd10.); 	
	if length(AEENDTC)=16     then AENDTM = input( AEENDTC ,E8601DT16.);
	if AENDTM ne .            then AENTM  = timepart(AENDTM);

/*Imputation*/
/*ASTDT*/
	if index(aestdtc,'-----') eq 0 then count_AESTDTC = countc(aestdtc, '1234567890');
	If index(upcase(AESTDTC), 'UNKUNK') > 0 or  index(upcase(AESTDTC), 'UNK') or count_AESTDTC < 8 then do;
		If count_AESTDTC = 4 then do;
			ASTDTF = 'M';
			temp_AESTDTC = compress(AESTDTC, ,'kd');
			temp_AESTDTN =  input(temp_AESTDTC, best.);
			yrt = year(tr01sdt);
			Dayb = '-01-01'; 	
			if( aendt ne . and aendt > tr01sdt) or aendt = . then do; 
				if yrt  = yra   then astdt = tr01sdt;
				else astdt = input( compress(AESTDTC||'-'||strip(DAyb)), yymmdd10.);
			end;
		end;

		If count_AESTDTC = 6 then do;
			ASTDTF = 'D';
			yra     = input(substr(strip(AESTDTC),1,4), best.);
			mna     = input(scan(strip(AESTDTC),2,'-'), best.);
			yrt     = year(tr01sdt);
			mnt     = month(tr01sdt);
			Dayb = '01'; 	
			
			if( aendt ne . and aendt > tr01sdt) or aendt = . then do; 
				if yrt  = yra and mnt = mna then do;
					astdt = tr01sdt;
					if tr01sdtm ne . then do;
						asttm = timepart(tr01sdtm);	
						astdtm = tr01sdtm;
					end;
				end;
				else astdt = input( compress(AESTDTC||'-'||strip(DAyb)), yymmdd10.);
			end;
		end;
	end;
	else do;
		ASTDT = input(AESTDTC, yymmdd10.);
	end;
	AESTDTC_  =AESTDTC;
 	*AE start date and time;
 	if ASTDTF ne '' and astdt = tr01sdt and  TR01sdtm ne .  then do;
 		AESTDTC_ = strip(put( TR01sdtm ,E8601DT16.));
 		ASTTMF = 'H';
 	end;
	if length(AESTDTC_)=16 then ASTDTM = input( AESTDTC_ ,E8601DT16.);
	if ASTDTM ne . then ASTTM = timepart(ASTDTM);
	
	 	*MDOSREF;
	if index(upcase(AEREFID), 'DOSE 1')>0 then MDOSREF = 'Vaccination 1';
	if index(upcase(AEREFID), 'DOSE 2')>0 then MDOSREF = 'Vaccination 2';
 
	/*	ASTDY*/
	if nmiss(ASTDT, tr01sdt) eq 0 then do;
		if ASTDT < tr01sdt then ASTDY = ASTDT - tr01sdt;
		else ASTDY = ASTDT - tr01sdt + 1;
	end;

	/*	AENDY*/
	if nmiss(AENDT, tr01sdt) eq 0 then do;
		if AENDT < tr01sdt then AENDY = AENDT - tr01sdt;
		else AENDY = AENDT - tr01sdt + 1;
	end;
	
*ASTAGE;
	if TR01edt ne . then enddate = TR01edt + 27;
	else if tr01sdt ne . then enddate = tr01sdt + 27;
	if (nmiss(astdtm, tr01sdtm) = 0 and astdtm>= tr01sdtm) or (nmiss(astdt, tr01sdt) = 0 and astdt>= tr01sdt and astdtm = .)  then do;
		   			
		if astdt <=enddate then do;
			ASTAGE  = 'Vaccination stage';
			ASTAGEN = 1;
		end; 
	end;
	if nmiss(enddate, astdt)  = 0 and astdt >enddate then do;
			ASTAGE  = 'Follow up stage';
			ASTAGEN = 2;
	end;
	
	*TRTEMFL;
	if nmiss(TR01SDTM, ASTDTM) = 0 and TR01SDTM <= ASTDTM then TRTEMFL= 'Y';
	else if nmiss(TR01SDT, ASTDT) = 0 and TR01SDT <= ASTDT and ASTDTM= . then TRTEMFL= 'Y';
	
	if nmiss(TR01SDTM, AENDTM) = 0 and ASTDTM = . and  TR01SDTM <= AENDTM and ASTDT = . then TRTEMFL= 'Y'; 
	else if nmiss(TR01SDT, AENDT) = 0 and ASTDT= . and  TR01SDT <= AENDT then TRTEMFL= 'Y'; 
	
	if TR01SDTM ne . and nmiss(ASTDTM,AENDTM) = 2 and nmiss(ASTDT,AENDT) = 2 then TRTEMFL= 'Y';
	else if TR01SDT ne . and nmiss(ASTDT,AENDT) = 2 then TRTEMFL= 'Y';
	
	*AREL & ARELN;
	AREL        = AEREL;
	if AREL = 'NOT RELATED'    then  ARELN =1; 
	if AREL = 'RELATED'        then  ARELN =2; 
	if AREL = 'NOT APPLICABLE' then  ARELN =3; 
	
	*Toxicity grades;
	if AETOXGR ne '' then do;
		ATOXGR  = put(AETOXGR, atoxgr.);
		ATOXGRN = input(AETOXGR, atoxgrn.);
		
		if ATOXGRN < 3 then do;
			TOXGGR1  = '<3';
			TOXGGR1N = 1;
		end;
		else do;
			TOXGGR1  = '>=3';
			TOXGGR1N = 2;	
		end;
	end;
	
    *Severity; 
		ASEV  = put(strip(upcase(AETOXGR)), asev.);
		ASEVN = input(strip(upcase(AETOXGR)), asevn.); 
	    
	  *MRDSTDY;	
	 if MDOSREF = 'Vaccination 1' then do;
	 	startdtm = tr01sdtm;
	 	start    = tr01sdt;
	 end;
	  if MDOSREF = 'Vaccination 2' then do; 
	  	startdtm = DOS2DTM ;
	 	start    = dose2dt;
	 end;
 
		if nmiss(startdtm, ASTDTM ) = 0 then do;
			day = (ASTDTM  - startdtm)/(60*60*24) ;
			if mod(day,1) > 0 then day_overshoot = 1;
			else day_overshoot = 0;
			
			MRDSTDY = int(day) + day_overshoot;
		    if day <= 1 then MRDSTDY = 1;
		end;
		else if nmiss(astdt, start)  =0 then do;
			if astdt >= start then  MRDSTDY = astdt - start  + 1;
			else MRDSTDY = astdt - start;
		end;
		
	  *MRDENDY; 
		if nmiss(startdtm, aendtm ) = 0 then do;
			day = (aendtm  - startdtm)/(60*60*24) ;
			if mod(day,1) > 0 then day_overshoot = 1;
			else day_overshoot = 0;
			
			MRDENDY = int(day) + day_overshoot;
		    if day <= 1 then MRDENDY = 1;
		end;
		else if nmiss(aendt, start)  =0 then do;
			if aendt >= start then  MRDENDY = aendt - start  + 1;
			else MRDENDY = aendt - start;
		end;
		
	*D28FL;
	if . < MRDSTDY <=28 then D28FL = 'Y';
	 
	
	*alert for missing ae start date;
	if aestdtc = '' then do;
		put 'ALERT_C: missing AE start date value for subject ' USUBJID  ' ' AETERM;
	end; 
	*alert for missing ae toxicity grade;
	if atoxgr = '' then do;
		put 'ALERT_C: missing toxicity grade for ' USUBJID ' ' AETERM;
	end; 
	
	*ADURN;
	if strip(AECAT) = 'REACTOGENICITY' and MRDSTDY <= 7 then do;
		if aendt ne . then enddate = AENDT;
		else do;
			if nmiss(cutoffdt, EOSDT) = 0 then Enddate = min(cutoffdt, EOSDT);
			else if EOSDT ne .     then Enddate = EOSDT;
			else if cutoffdt  ne . then Enddate = cutoffdt;		
		end;
		
	
		if MDOSREF = 'Vaccination 1' and nmiss(TR01sdt,enddate) ne 2 and ((enddate - TR01sdt + 1) > 7) then  ADURN  = enddate - TR01sdt-6;
		if MDOSREF = 'Vaccination 2' and nmiss(DOSE2DT,enddate) ne 2 and ((enddate - DOSE2DT + 1) > 7) then  ADURN  = enddate - DOSE2DT-6;
	end;

	lnkpid  =  strip(tranwrd(AESPID, 'AE-',''));
	format _all_;
	informat _all_;
	drop tr01sdt tr01sdtm TR01edt TR01edtm  DOSE2DT cutoffdt  EOSDT _: DOS2DTM ;
run;


proc sort data = trans.DS out = DS  ;
	by usubjid;
	where index(upcase(dsdecod), 'ADVERSE EVENT') > 0 and DSSCAT = 'END OF STUDY' and DSCAT = 'DISPOSITION EVENT';
run;

data dsae;
length lnkpid $200.; 
	set ds;

	lnkpid  =  strip(tranwrd(DSSPID, 'DS-',''));
	keep lnkpid usubjid;
run; 

proc sort data = dsae  ;
	by usubjid lnkpid;
run;
proc sort data = adae1  ;
	by usubjid lnkpid;
run;

data &DSETNAME.;
	merge adae1 dsae (in=a);
	by usubjid lnkpid;
	
	if a then AEDISFL = 'Y';
run;

%mergeadsl(&DSETNAME.,keep=usubjid subjid );
%trta(dsname=&DSETNAME);

proc sort data = &DSETNAME  ;
	by USUBJID  AEBODSYS  AEDECOD AESTDTC   AEENDTC AESEQ;
run;

**Generate final dataset by updating certain attributes. Optionally merge common variables & create sequence variable as needed**;
%adam_dataset_update(
	ds=&DSETNAME,
	libin=work,
	libout=output,
	adsllib=adb,
	addcomvar=Y,
	addseq=,
	dropinfmt=Y,
	mapspecfile=&ADAM_SPEC,
	maploc=&ADAM_SPEC_LOC,
	debug=N
	);
	
	
%macro dod;
**Generate Value Level Metadata values for the ad<xx> dataset**;
% dod_vlm(        
	type=ADAM,               
	selmems=&DSETNAME,
	excmems=,
	specloc=&G_PROJECTPATH.&G_TOPLEVEL.\Documents\Specs\,
	specname=&G_NICKNAME._ADaM_Spec.xlsm,
	debug=N
	);

**Generate Enhanced Controlled Terminology for the ad<xx> dataset**;
% dod_enhcd_ct(            
	type=ADAM,               
	selmems=&DSETNAME,
	excmems=,
	specloc=&G_PROJECTPATH.&G_TOPLEVEL.\Documents\Specs\,
	specname=&G_NICKNAME._ADaM_Spec.xlsm,
	map_ct_nm=,
	map_ct_loc=,
	debug=N
	);
%mend;