/* file to set up libnames, fmt/mac locations */ 

%include "../../../fem_env.sas";

libname bio "&dua_rand_hrs";

%let seed=1230543890;

/*options obs=20000; */  

*** Step 1 - Assign the value for hdl_onewave = 1, from the donor pool (hdl_twowaves = 1)
*** Step 2 - Assign the value for hdl_nowave = 1, from the donor pool (hdl_twowaves = 1)
*** Things need to do before this - 1. Re-create the appropriate grouping variables and add it in the sql step
***                                 3. Reshape the data to wide form (hhidpn wave pct_* hearte hibpe diabe rxchol)


**%macro imp (bio, pctbio, pctbiohat, pctbiosd, imppctbio, impflag);  

proc import out=bio.bio_imp20tiles_wide datafile="&dua_rand_hrs./bio_imp20tiles_wide.dta";
run;

proc import out=bio.bio_hrs_recoded_prepimp_v1 datafile="&dua_rand_hrs./bio_hrs_recoded_prepimp_v1.dta";
run;

proc print data=bio.bio_imp20tiles_wide (where=(hhidpn in (10537011 11802011)));
run;

/*************************************************;
** HDL;
*************************************************;*/


/*   select stddev(pct_a1c_hat) into :a1c_sd from bio.fname;  */
%macro sqlyr(yr,wv); 
	
	proc sql; 
   create table miss_donorhdl&yr as 
      select a.hhidpn,&wv as hdl_nm_wave,a.pct_hdl_hat&yr as pct_hdl_hat,a.pct_hdl&yr as pct_hdl_nm,b.hhidpn as donor_id,
             b.pct_hdl_hat&yr as donorpct_hdl_hat,
             b.pct_hdl06 as donorpct_hdl06,
             b.pct_hdl08 as donorpct_hdl08,
             b.pct_hdl10 as donorpct_hdl10,
             b.pct_hdl12 as donorpct_hdl12,
             b.pct_hdl14 as donorpct_hdl14,
             abs(a.pct_hdl_hat&yr - b.pct_hdl_hat&yr) as distance
             from bio.bio_imp20tiles_wide (where=((hdl_onewave=1 or hdl_twowaves_sas=1) & pct_hdl&yr ne . and pct_hdl_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((hdl_threewaves=1 or hdl_twowaves_nosas=1) & pct_hdl&yr ne . and pct_hdl_hat&yr ne . )) b
             on a.pct_hdl&yr = b.pct_hdl&yr & (a.hearte06 = b.hearte06 or missing(a.hearte06) or missing(b.hearte06) ) & (a.hearte08 = b.hearte08 or missing(a.hearte08) or missing(b.hearte08) )
                & (a.hearte10 = b.hearte10 or missing(a.hearte10) or missing(b.hearte10) ) & (a.hearte12 = b.hearte12 or missing(a.hearte12) or missing(b.hearte12) )
                & (a.hearte14 = b.hearte14 or missing(a.hearte14) or missing(b.hearte14) )
             /*& abs(a.pct_hdl_hat-b.pct_hdl_hat)<=(1.5*a.pct_hdl_sd) */
             order a.hhidpn,calculated distance;
   
   create table no_donorhdl&yr as 
	select b.* from miss_donorhdl&yr (where=(donor_id = .)) a
	left join bio.bio_imp20tiles_wide b
	on a.hhidpn = b.hhidpn ;

   create table miss_donorhdl_nd&yr as 
      select a.hhidpn,&wv as hdl_nm_wave,a.pct_hdl_hat&yr as pct_hdl_hat,a.pct_hdl&yr as pct_hdl_nm,b.hhidpn as donor_id,
             b.pct_hdl_hat&yr as donorpct_hdl_hat,
             b.pct_hdl06 as donorpct_hdl06,
             b.pct_hdl08 as donorpct_hdl08,
             b.pct_hdl10 as donorpct_hdl10,
             b.pct_hdl12 as donorpct_hdl12,
             b.pct_hdl14 as donorpct_hdl14,
             abs(a.pct_hdl_hat&yr - b.pct_hdl_hat&yr) as distance
             from no_donorhdl&yr (where=((hdl_onewave=1 or hdl_twowaves_sas=1) & pct_hdl&yr ne . and pct_hdl_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((hdl_threewaves=1 or hdl_twowaves_nosas=1) & pct_hdl&yr ne . and pct_hdl_hat&yr ne . )) b
             on a.pct_hdl&yr = b.pct_hdl&yr
            /* & abs(a.pct_hdl_hat&yr-b.pct_hdl_hat&yr)<=(1.5*a.pct_hdl_sd&yr) */
             order a.hhidpn,calculated distance;

data miss_donorhdl&yr ; 
	merge miss_donorhdl&yr
        miss_donorhdl_nd&yr ;
        by hhidpn ;

  run;         
   
%mend;

%sqlyr(06,8)
%sqlyr(08,9)
%sqlyr(10,10)
%sqlyr(12,11)
%sqlyr(14,12)

proc sql;
   create table miss_donorhdl as 
      select a.hhidpn,0 as hdl_nm_wave, a.pct_hdl_hat,b.pct_hdl_hat as donorpct_hdl_hat,b.hhidpn as donor_id,
             b.pct_hdl06 as donorpct_hdl06,
             b.pct_hdl08 as donorpct_hdl08,
             b.pct_hdl10 as donorpct_hdl10,
             b.pct_hdl12 as donorpct_hdl12,
             b.pct_hdl14 as donorpct_hdl14,
             abs(a.pct_hdl_hat-b.pct_hdl_hat) as distance
             from bio.bio_imp20tiles_wide (where=(hdl_nowave=1 and pct_hdl_hat ne . )) a
             left join bio.bio_imp20tiles_wide (where=((hdl_threewaves=1 or hdl_twowaves_nosas=1) and pct_hdl_hat ne . )) b
             on (a.hearte06 = b.hearte06 or missing(a.hearte06) or missing(b.hearte06) ) & (a.hearte08 = b.hearte08 or missing(a.hearte08) or missing(b.hearte08) )
                & (a.hearte10 = b.hearte10 or missing(a.hearte10) or missing(b.hearte10) ) & (a.hearte12 = b.hearte12 or missing(a.hearte12) or missing(b.hearte12) )
                & (a.hearte14 = b.hearte14 or missing(a.hearte14) or missing(b.hearte14) )
               /* & abs(a.pct_hdl_hat-b.pct_hdl_hat)<=(1.5*a.pct_hdl_sd) */
             order a.hhidpn,calculated distance;




proc print data=miss_donorhdl06 (obs=200);
run;
proc print data=miss_donorhdl10 (where=(donor_id=.) obs=100);
	title2 donor id missing ; 
run;
title2 selected cases;

proc print data=miss_donorhdl (obs=200);
run;


/* only keep 10 in the donor pool */
data setup checknhdl;
   set miss_donorhdl
       miss_donorhdl06
       miss_donorhdl08
       miss_donorhdl10
       miss_donorhdl12
       miss_donorhdl14;
   by hdl_nm_wave hhidpn ;
   retain n_donor;
   if first.hhidpn then n_donor=0;
   n_donor=n_donor+1;
   if last.hhidpn then do;
      if n_donor>=10 then n_donor_status=3;
      else if n_donor>=5 then n_donor_status=2;
      else if n_donor>0 then n_donor_status=1;
      else n_donor_status=0;
      output checknhdl;
   end;
   /* if one of the candidate donors then assign random number and output */
   if n_donor<=10 then do;
      ran_num=ranuni(&seed);
      output setup;
   end;
run;

proc print data=checknhdl (obs=200) ;
run;

 
proc means data=checknhdl;
  class n_donor_status;
  var n_donor;
  run;
  
/*proc print data=checkn&bio (where=(n_donor<5) obs=20);
title2 hhidpn-wave with fewer than 5 donors;
run;

title2;
*/

proc sort data=setup;
   by hdl_nm_wave hhidpn ran_num;

data imputedhdl;
   set setup;
   by hdl_nm_wave hhidpn ;
   if first.hhidpn;
*  imp_pct_hdl=donorpct_hdl08;
   run;
proc sort data=imputedhdl out=bio.imputedhdl;
	   by hhidpn;
	   
proc print data=bio.imputedhdl (obs=100) ;
run;

	   
data bio.bio_imp_finalhdl;
   merge bio.imputedhdl (in=_inimp keep=hhidpn donorpct_hdl: hdl_nm_wave)
         bio.bio_hrs_recoded_prepimp_v1 (in=_inpct)
         ;
   by hhidpn ;
   
   impflag2_hdl=_inimp;
   
   if missing(pct_hdl) and wave=8 then pct_hdl=donorpct_hdl06;
   if missing(pct_hdl) and wave=9 then pct_hdl=donorpct_hdl08;
   if missing(pct_hdl) and wave=10 then pct_hdl=donorpct_hdl10;
   if missing(pct_hdl) and wave=11 then pct_hdl=donorpct_hdl12;
   if missing(pct_hdl) and wave=12 then pct_hdl=donorpct_hdl14;
   
   if impflag2_hdl=1 and missing(pct_hdl)=1 then impflag2_hdl=9;
   
run;

proc contents data=bio.bio_imp_finalhdl;
run;

proc means data=bio.bio_imp_finalhdl;
   class impflag2_hdl;
   types () impflag2_hdl;
   var pct_hdl ;
   run;


/*************************************************;
** TCHOL;
*************************************************;*/



%macro sqlyr(yr,wv);  
proc sql;
   create table miss_donortchol&yr as 
      select a.hhidpn,&wv as tchol_nm_wave,a.pct_tchol_hat&yr as pct_tchol_hat,a.pct_tchol&yr as pct_tchol_nm,b.hhidpn as donor_id,
             b.pct_tchol_hat&yr as donorpct_tchol_hat,
             b.pct_tchol06 as donorpct_tchol06,
             b.pct_tchol08 as donorpct_tchol08,
             b.pct_tchol10 as donorpct_tchol10,
             b.pct_tchol12 as donorpct_tchol12,
             b.pct_tchol14 as donorpct_tchol14,
             abs(a.pct_tchol_hat&yr - b.pct_tchol_hat&yr) as distance
             from bio.bio_imp20tiles_wide (where=((tchol_onewave=1 or tchol_twowaves_sas=1) & pct_tchol&yr ne . and pct_tchol_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((tchol_threewaves=1 or tchol_twowaves_nosas=1) & pct_tchol&yr ne . and pct_tchol_hat&yr ne . )) b
             on a.pct_tchol&yr = b.pct_tchol&yr & (a.rxchol06 = b.rxchol06 or missing(a.rxchol06) or missing(b.rxchol06) ) & (a.rxchol08 = b.rxchol08 or missing(a.rxchol08) or missing(b.rxchol08) )
                & (a.rxchol10 = b.rxchol10 or missing(a.rxchol10) or missing(b.rxchol10) ) & (a.rxchol12 = b.rxchol12 or missing(a.rxchol12) or missing(b.rxchol12) )
                & (a.rxchol14 = b.rxchol14 or missing(a.rxchol14) or missing(b.rxchol14) )
            /* & abs(a.pct_tchol_hat&yr-b.pct_tchol_hat&yr)<=(1.5*a.pct_tchol_sd&yr) */
             order a.hhidpn,calculated distance;

create table no_donortchol&yr as 
	select b.* from miss_donortchol&yr (where=(donor_id = .)) a
	left join bio.bio_imp20tiles_wide b
	on a.hhidpn = b.hhidpn ;

   create table miss_donortchol_nd&yr as 
      select a.hhidpn,&wv as tchol_nm_wave,a.pct_tchol_hat&yr as pct_tchol_hat,a.pct_tchol&yr as pct_tchol_nm,b.hhidpn as donor_id,
             b.pct_tchol_hat&yr as donorpct_tchol_hat,
             b.pct_tchol06 as donorpct_tchol06,
             b.pct_tchol08 as donorpct_tchol08,
             b.pct_tchol10 as donorpct_tchol10,
             b.pct_tchol12 as donorpct_tchol12,
             b.pct_tchol14 as donorpct_tchol14,
             abs(a.pct_tchol_hat&yr - b.pct_tchol_hat&yr) as distance
             from no_donortchol&yr (where=((tchol_onewave=1 or tchol_twowaves_sas=1) & pct_tchol&yr ne . and pct_tchol_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((tchol_threewaves=1 or tchol_twowaves_nosas=1) & pct_tchol&yr ne . and pct_tchol_hat&yr ne . )) b
             on a.pct_tchol&yr = b.pct_tchol&yr
            /* & abs(a.pct_tchol_hat&yr-b.pct_tchol_hat&yr)<=(1.5*a.pct_tchol_sd&yr) */
             order a.hhidpn,calculated distance;

data miss_donortchol&yr ; 
	merge miss_donortchol&yr
        miss_donortchol_nd&yr ;
        by hhidpn ;
run;

%mend;

%sqlyr(06,8)
%sqlyr(08,9)
%sqlyr(10,10)
%sqlyr(12,11)
%sqlyr(14,12)

proc sql;

   create table miss_donortchol as 
      select a.hhidpn,0 as tchol_nm_wave, a.pct_tchol_hat,b.pct_tchol_hat as donorpct_tchol_hat,b.hhidpn as donor_id,
             b.pct_tchol06 as donorpct_tchol06,
             b.pct_tchol08 as donorpct_tchol08,
             b.pct_tchol10 as donorpct_tchol10,
             b.pct_tchol12 as donorpct_tchol12,
             b.pct_tchol14 as donorpct_tchol14,
             abs(a.pct_tchol_hat-b.pct_tchol_hat) as distance
             from bio.bio_imp20tiles_wide (where=(tchol_nowave=1 and pct_tchol_hat ne . )) a
             left join bio.bio_imp20tiles_wide (where=((tchol_threewaves=1 or tchol_twowaves_nosas=1) and pct_tchol_hat ne . )) b
             on (a.rxchol06 = b.rxchol06 or missing(a.rxchol06) or missing(b.rxchol06) ) & (a.rxchol08 = b.rxchol08 or missing(a.rxchol08) or missing(b.rxchol08) )
                & (a.rxchol10 = b.rxchol10 or missing(a.rxchol10) or missing(b.rxchol10) ) & (a.rxchol12 = b.rxchol12 or missing(a.rxchol12) or missing(b.rxchol12) )
                & (a.rxchol14 = b.rxchol14 or missing(a.rxchol14) or missing(b.rxchol14) )
                /* & abs(a.pct_tchol_hat-b.pct_tchol_hat)<=(1.5*a.pct_tchol_sd) */
             order a.hhidpn,calculated distance;


proc print data=miss_donortchol06 (obs=200);
run;
proc print data=miss_donortchol10 (where=(donor_id=.) obs=100);
	title2 donor id missing ; 
run;
title2 selected cases;

proc print data=miss_donortchol (obs=200);
run;




/* only keep 10 in the donor pool */
data setup checkntchol;
   set miss_donortchol
       miss_donortchol06
       miss_donortchol08
       miss_donortchol10
       miss_donortchol12
       miss_donortchol14;
   by tchol_nm_wave hhidpn ;
   retain n_donor;
   if first.hhidpn then n_donor=0;
   n_donor=n_donor+1;
   if last.hhidpn then do;
      if n_donor>=10 then n_donor_status=3;
      else if n_donor>=5 then n_donor_status=2;
      else if n_donor>0 then n_donor_status=1;
      else n_donor_status=0;
      output checkntchol;
   end;
   /* if one of the candidate donors then assign random number and output */
   if n_donor<=10 then do;
      ran_num=ranuni(&seed);
      output setup;
   end;
run;

proc print data=checkntchol (obs=200) ;
run;

 
proc means data=checkntchol;
  class n_donor_status;
  var n_donor;
  run;
  
/*proc print data=checkn&bio (where=(n_donor<5) obs=20);
title2 hhidpn-wave with fewer than 5 donors;
run;

title2;
*/

proc sort data=setup;
   by tchol_nm_wave hhidpn ran_num;

data imputedtchol;
   set setup;
   by tchol_nm_wave hhidpn ;
   if first.hhidpn;
*  imp_pct_tchol=donorpct_tchol08;
   run;
proc sort data=imputedtchol out=bio.imputedtchol;
	   by hhidpn;
	   
proc print data=bio.imputedtchol (obs=100) ;
run;

	   
data bio.bio_imp_finaltchol;
   merge bio.imputedtchol (in=_inimp keep=hhidpn donorpct_tchol: tchol_nm_wave)
         bio.bio_hrs_recoded_prepimp_v1 (in=_inpct)
         ;
   by hhidpn ;
   
   impflag2_tchol=_inimp;
   
   if missing(pct_tchol) and wave=8 then pct_tchol=donorpct_tchol06;
   if missing(pct_tchol) and wave=9 then pct_tchol=donorpct_tchol08;
   if missing(pct_tchol) and wave=10 then pct_tchol=donorpct_tchol10;
   if missing(pct_tchol) and wave=11 then pct_tchol=donorpct_tchol12;
   if missing(pct_tchol) and wave=12 then pct_tchol=donorpct_tchol14;
   
   if impflag2_tchol=1 and missing(pct_tchol)=1 then impflag2_tchol=9;
   
run;

proc contents data=bio.bio_imp_finaltchol;
run;

proc means data=bio.bio_imp_finaltchol;
   class impflag2_tchol;
   types () impflag2_tchol;
   var pct_tchol ;
   run;



/*************************************************;
** HBA1c;
*************************************************;*/


/*   select stddev(pct_a1c_hat) into :a1c_sd from bio.fname;  */
%macro sqlyr(yr,wv);  
proc sql;
   create table miss_donora1c&yr as 
      select a.hhidpn,&wv as a1c_nm_wave,a.pct_a1c_hat&yr as pct_a1c_hat,a.pct_a1c&yr as pct_a1c_nm,b.hhidpn as donor_id,
             b.pct_a1c_hat&yr as donorpct_a1c_hat,
             b.pct_a1c06 as donorpct_a1c06,
             b.pct_a1c08 as donorpct_a1c08,
             b.pct_a1c10 as donorpct_a1c10,
             b.pct_a1c12 as donorpct_a1c12,
             b.pct_a1c14 as donorpct_a1c14,
             abs(a.pct_a1c_hat&yr - b.pct_a1c_hat&yr) as distance
             from bio.bio_imp20tiles_wide (where=((a1c_onewave=1 or a1c_twowaves_sas=1) & pct_a1c&yr ne . and pct_a1c_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((a1c_threewaves=1 or a1c_twowaves_nosas=1) & pct_a1c&yr ne . and pct_a1c_hat&yr ne . )) b
             on a.pct_a1c&yr = b.pct_a1c&yr & (a.diabe06 = b.diabe06 or missing(a.diabe06) or missing(b.diabe06) ) & (a.diabe08 = b.diabe08 or missing(a.diabe08) or missing(b.diabe08) )
                & (a.diabe10 = b.diabe10 or missing(a.diabe10) or missing(b.diabe10) ) & (a.diabe12 = b.diabe12 or missing(a.diabe12) or missing(b.diabe12) )
                & (a.diabe14 = b.diabe14 or missing(a.diabe14) or missing(b.diabe14) )
             /* & abs(a.pct_a1c_hat-b.pct_a1c_hat)<=(1.5*a.pct_a1c_sd) */
             order a.hhidpn,calculated distance;
             
create table no_donora1c&yr as 
	select b.* from miss_donora1c&yr (where=(donor_id = .)) a
	left join bio.bio_imp20tiles_wide b
	on a.hhidpn = b.hhidpn ;

   create table miss_donora1c_nd&yr as 
      select a.hhidpn,&wv as a1c_nm_wave,a.pct_a1c_hat&yr as pct_a1c_hat,a.pct_a1c&yr as pct_a1c_nm,b.hhidpn as donor_id,
             b.pct_a1c_hat&yr as donorpct_a1c_hat,
             b.pct_a1c06 as donorpct_a1c06,
             b.pct_a1c08 as donorpct_a1c08,
             b.pct_a1c10 as donorpct_a1c10,
             b.pct_a1c12 as donorpct_a1c12,
             b.pct_a1c12 as donorpct_a1c14,
             abs(a.pct_a1c_hat&yr - b.pct_a1c_hat&yr) as distance
             from no_donora1c&yr (where=((a1c_onewave=1 or a1c_twowaves_sas=1) & pct_a1c&yr ne . and pct_a1c_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((a1c_threewaves=1 or a1c_twowaves_nosas=1) & pct_a1c&yr ne . and pct_a1c_hat&yr ne . )) b
             on a.pct_a1c&yr = b.pct_a1c&yr
            /* & abs(a.pct_a1c_hat&yr-b.pct_a1c_hat&yr)<=(1.5*a.pct_a1c_sd&yr) */
             order a.hhidpn,calculated distance;

data miss_donora1c&yr ; 
	merge miss_donora1c&yr
        miss_donora1c_nd&yr ;
        by hhidpn ;

  run;         
                 
%mend;

%sqlyr(06,8)
%sqlyr(08,9)
%sqlyr(10,10)
%sqlyr(12,11)
%sqlyr(14,12)

proc sql;
   create table miss_donora1c as 
      select a.hhidpn,0 as a1c_nm_wave, a.pct_a1c_hat,b.pct_a1c_hat as donorpct_a1c_hat,b.hhidpn as donor_id,
             b.pct_a1c06 as donorpct_a1c06,
             b.pct_a1c08 as donorpct_a1c08,
             b.pct_a1c10 as donorpct_a1c10,
             b.pct_a1c12 as donorpct_a1c12,
             b.pct_a1c14 as donorpct_a1c14,
             abs(a.pct_a1c_hat-b.pct_a1c_hat) as distance
             from bio.bio_imp20tiles_wide (where=(a1c_nowave=1 and pct_a1c_hat ne . )) a
             left join bio.bio_imp20tiles_wide (where=((a1c_threewaves=1 or a1c_twowaves_nosas=1) and pct_a1c_hat ne . )) b
             on (a.diabe06 = b.diabe06 or missing(a.diabe06) or missing(b.diabe06) ) & (a.diabe08 = b.diabe08 or missing(a.diabe08) or missing(b.diabe08) )
                & (a.diabe10 = b.diabe10 or missing(a.diabe10) or missing(b.diabe10) ) & (a.diabe12 = b.diabe12 or missing(a.diabe12) or missing(b.diabe12) )
                & (a.diabe14 = b.diabe14 or missing(a.diabe14) or missing(b.diabe14) )
               /* & abs(a.pct_a1c_hat-b.pct_a1c_hat)<=(1.5*a.pct_a1c_sd) */
             order a.hhidpn,calculated distance;

proc print data=miss_donora1c06 (obs=200);
run;
proc print data=miss_donora1c10 (where=(donor_id=.) obs=100);
	title2 donor id missing ; 
run;
title2 selected cases;

proc print data=miss_donora1c (obs=200);
run;


/* only keep 10 in the donor pool */
data setup checkna1c;
   set miss_donora1c
       miss_donora1c06
       miss_donora1c08
       miss_donora1c10
       miss_donora1c12
       miss_donora1c14;
   by a1c_nm_wave hhidpn ;
   retain n_donor;
   if first.hhidpn then n_donor=0;
   n_donor=n_donor+1;
   if last.hhidpn then do;
      if n_donor>=10 then n_donor_status=3;
      else if n_donor>=5 then n_donor_status=2;
      else if n_donor>0 then n_donor_status=1;
      else n_donor_status=0;
      output checkna1c;
   end;
   /* if one of the candidate donors then assign random number and output */
   if n_donor<=10 then do;
      ran_num=ranuni(&seed);
      output setup;
   end;
run;

proc print data=checkna1c (obs=200) ;
run;

 
proc means data=checkna1c;
  class n_donor_status;
  var n_donor;
  run;
  
/*proc print data=checkn&bio (where=(n_donor<5) obs=20);
title2 hhidpn-wave with fewer than 5 donors;
run;

title2;
*/

proc sort data=setup;
   by a1c_nm_wave hhidpn ran_num;

data imputeda1c;
   set setup;
   by a1c_nm_wave hhidpn ;
   if first.hhidpn;
*  imp_pct_a1c=donorpct_a1c08;
   run;
proc sort data=imputeda1c out=bio.imputeda1c;
	   by hhidpn;
	   
proc print data=bio.imputeda1c (obs=100) ;
run;

	   
data bio.bio_imp_finala1c;
   merge bio.imputeda1c (in=_inimp keep=hhidpn donorpct_a1c: a1c_nm_wave)
         bio.bio_hrs_recoded_prepimp_v1 (in=_inpct)
         ;
   by hhidpn ;
   
   impflag2_a1c=_inimp;
   
   if missing(pct_a1c) and wave=8 then pct_a1c=donorpct_a1c06;
   if missing(pct_a1c) and wave=9 then pct_a1c=donorpct_a1c08;
   if missing(pct_a1c) and wave=10 then pct_a1c=donorpct_a1c10;
   if missing(pct_a1c) and wave=11 then pct_a1c=donorpct_a1c12;
   if missing(pct_a1c) and wave=12 then pct_a1c=donorpct_a1c14;
   
   if impflag2_a1c=1 and missing(pct_a1c)=1 then impflag2_a1c=9;
   
run;

proc contents data=bio.bio_imp_finala1c;
run;

proc means data=bio.bio_imp_finala1c;
   class impflag2_a1c;
   types () impflag2_a1c;
   var pct_a1c ;
   run;
   
   
   
/*************************************************;
** SYSBP;
*************************************************;*/


/*   select stddev(pct_a1c_hat) into :a1c_sd from bio.fname;  */
%macro sqlyr(yr,wv);  
proc sql;
   create table miss_donorsysbp&yr as 
      select a.hhidpn,&wv as sysbp_nm_wave,a.pct_sysbp_hat&yr as pct_sysbp_hat,a.pct_sysbp&yr as pct_sysbp_nm,b.hhidpn as donor_id,
             b.pct_sysbp_hat&yr as donorpct_sysbp_hat,
             b.pct_sysbp06 as donorpct_sysbp06,
             b.pct_sysbp08 as donorpct_sysbp08,
             b.pct_sysbp10 as donorpct_sysbp10,
             b.pct_sysbp12 as donorpct_sysbp12,
             b.pct_sysbp14 as donorpct_sysbp14,
             abs(a.pct_sysbp_hat&yr - b.pct_sysbp_hat&yr) as distance
             from bio.bio_imp20tiles_wide (where=((sysbp_onewave=1 or sysbp_twowaves_sas=1) & pct_sysbp&yr ne . and pct_sysbp_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((sysbp_threewaves=1 or sysbp_twowaves_nosas=1) & pct_sysbp&yr ne . and pct_sysbp_hat&yr ne . )) b
             on a.pct_sysbp&yr = b.pct_sysbp&yr & (a.hibpe06 = b.hibpe06 or missing(a.hibpe06) or missing(b.hibpe06) ) & (a.hibpe08 = b.hibpe08 or missing(a.hibpe08) or missing(b.hibpe08) )
                & (a.hibpe10 = b.hibpe10 or missing(a.hibpe10) or missing(b.hibpe10) ) & (a.hibpe12 = b.hibpe12 or missing(a.hibpe12) or missing(b.hibpe12) )
                & (a.hibpe14 = b.hibpe14 or missing(a.hibpe14) or missing(b.hibpe14) )
             /* & abs(a.pct_sysbp_hat-b.pct_sysbp_hat)<=(1.5*a.pct_sysbp_sd) */
             order a.hhidpn,calculated distance;
             
create table no_donorsysbp&yr as 
	select b.* from miss_donorsysbp&yr (where=(donor_id = .)) a
	left join bio.bio_imp20tiles_wide b
	on a.hhidpn = b.hhidpn ;

   create table miss_donorsysbp_nd&yr as 
      select a.hhidpn,&wv as sysbp_nm_wave,a.pct_sysbp_hat&yr as pct_sysbp_hat,a.pct_sysbp&yr as pct_sysbp_nm,b.hhidpn as donor_id,
             b.pct_sysbp_hat&yr as donorpct_sysbp_hat,
             b.pct_sysbp06 as donorpct_sysbp06,
             b.pct_sysbp08 as donorpct_sysbp08,
             b.pct_sysbp10 as donorpct_sysbp10,
             b.pct_sysbp12 as donorpct_sysbp12,
             b.pct_sysbp14 as donorpct_sysbp14,
             abs(a.pct_sysbp_hat&yr - b.pct_sysbp_hat&yr) as distance
             from no_donorsysbp&yr (where=((sysbp_onewave=1 or sysbp_twowaves_sas=1) & pct_sysbp&yr ne . and pct_sysbp_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((sysbp_threewaves=1 or sysbp_twowaves_nosas=1) & pct_sysbp&yr ne . and pct_sysbp_hat&yr ne . )) b
             on a.pct_sysbp&yr = b.pct_sysbp&yr
            /* & abs(a.pct_sysbp_hat&yr-b.pct_sysbp_hat&yr)<=(1.5*a.pct_sysbp_sd&yr) */
             order a.hhidpn,calculated distance;

data miss_donorsysbp&yr ; 
	merge miss_donorsysbp&yr
        miss_donorsysbp_nd&yr ;
        by hhidpn ;
             
     run;        
             
%mend;

%sqlyr(06,8)
%sqlyr(08,9)
%sqlyr(10,10)
%sqlyr(12,11)
%sqlyr(14,12)

proc sql;
   create table miss_donorsysbp as 
      select a.hhidpn,0 as sysbp_nm_wave, a.pct_sysbp_hat,b.pct_sysbp_hat as donorpct_sysbp_hat,b.hhidpn as donor_id,
             b.pct_sysbp06 as donorpct_sysbp06,
             b.pct_sysbp08 as donorpct_sysbp08,
             b.pct_sysbp10 as donorpct_sysbp10,
             b.pct_sysbp12 as donorpct_sysbp12,
             b.pct_sysbp14 as donorpct_sysbp14,
             abs(a.pct_sysbp_hat-b.pct_sysbp_hat) as distance
             from bio.bio_imp20tiles_wide (where=(sysbp_nowave=1 and pct_sysbp_hat ne . )) a
             left join bio.bio_imp20tiles_wide (where=((sysbp_threewaves=1 or sysbp_twowaves_nosas=1) and pct_sysbp_hat ne . )) b
             on (a.hibpe06 = b.hibpe06 or missing(a.hibpe06) or missing(b.hibpe06) ) & (a.hibpe08 = b.hibpe08 or missing(a.hibpe08) or missing(b.hibpe08) )
                & (a.hibpe10 = b.hibpe10 or missing(a.hibpe10) or missing(b.hibpe10) ) & (a.hibpe12 = b.hibpe12 or missing(a.hibpe12) or missing(b.hibpe12) )
                & (a.hibpe14 = b.hibpe14 or missing(a.hibpe14) or missing(b.hibpe14) )
             /* & abs(a.pct_sysbp_hat-b.pct_sysbp_hat)<=(1.5*a.pct_sysbp_sd) */
             order a.hhidpn,calculated distance;

proc print data=miss_donorsysbp06 (obs=200);
run;
proc print data=miss_donorsysbp10 (where=(donor_id=.) obs=100);
	title2 donor id missing ; 
run;
title2 selected cases;

proc print data=miss_donorsysbp (obs=200);
run;


/* only keep 10 in the donor pool */
data setup checknsysbp;
   set miss_donorsysbp
       miss_donorsysbp06
       miss_donorsysbp08
       miss_donorsysbp10
       miss_donorsysbp12
       miss_donorsysbp14;
   by sysbp_nm_wave hhidpn ;
   retain n_donor;
   if first.hhidpn then n_donor=0;
   n_donor=n_donor+1;
   if last.hhidpn then do;
      if n_donor>=10 then n_donor_status=3;
      else if n_donor>=5 then n_donor_status=2;
      else if n_donor>0 then n_donor_status=1;
      else n_donor_status=0;
      output checknsysbp;
   end;
   /* if one of the candidate donors then assign random number and output */
   if n_donor<=10 then do;
      ran_num=ranuni(&seed);
      output setup;
   end;
run;

proc print data=checknsysbp (obs=200) ;
run;

 
proc means data=checknsysbp;
  class n_donor_status;
  var n_donor;
  run;
  
/*proc print data=checkn&bio (where=(n_donor<5) obs=20);
title2 hhidpn-wave with fewer than 5 donors;
run;

title2;
*/

proc sort data=setup;
   by sysbp_nm_wave hhidpn ran_num;

data imputedsysbp;
   set setup;
   by sysbp_nm_wave hhidpn ;
   if first.hhidpn;
*  imp_pct_sysbp=donorpct_sysbp08;
   run;
proc sort data=imputedsysbp out=bio.imputedsysbp;
	   by hhidpn;
	   
proc print data=bio.imputedsysbp (obs=100) ;
run;

	   
data bio.bio_imp_finalsysbp;
   merge bio.imputedsysbp (in=_inimp keep=hhidpn donorpct_sysbp: sysbp_nm_wave)
         bio.bio_hrs_recoded_prepimp_v1 (in=_inpct)
         ;
   by hhidpn ;
   
   impflag2_sysbp=_inimp;
   
   if missing(pct_sysbp) and wave=8 then pct_sysbp=donorpct_sysbp06;
   if missing(pct_sysbp) and wave=9 then pct_sysbp=donorpct_sysbp08;
   if missing(pct_sysbp) and wave=10 then pct_sysbp=donorpct_sysbp10;
   if missing(pct_sysbp) and wave=11 then pct_sysbp=donorpct_sysbp12;
   if missing(pct_sysbp) and wave=12 then pct_sysbp=donorpct_sysbp14;
   
   if impflag2_sysbp=1 and missing(pct_sysbp)=1 then impflag2_sysbp=9;
   
run;

proc contents data=bio.bio_imp_finalsysbp;
run;

proc means data=bio.bio_imp_finalsysbp;
   class impflag2_sysbp;
   types () impflag2_sysbp;
   var pct_sysbp ;
   run;


/*************************************************;
** crp;
*************************************************;*/

%macro sqlyr(yr,wv); 
	
	proc sql; 
   create table miss_donorcrp&yr as 
      select a.hhidpn,&wv as crp_nm_wave,a.pct_crp_hat&yr as pct_crp_hat,a.pct_crp&yr as pct_crp_nm,b.hhidpn as donor_id,
             b.pct_crp_hat&yr as donorpct_crp_hat,
             b.pct_crp06 as donorpct_crp06,
             b.pct_crp08 as donorpct_crp08,
             b.pct_crp10 as donorpct_crp10,
             b.pct_crp12 as donorpct_crp12,
             b.pct_crp14 as donorpct_crp14,
             abs(a.pct_crp_hat&yr - b.pct_crp_hat&yr) as distance
             from bio.bio_imp20tiles_wide (where=((crp_onewave=1 or crp_twowaves_sas=1) & pct_crp&yr ne . and pct_crp_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((crp_threewaves=1 or crp_twowaves_nosas=1) & pct_crp&yr ne . and pct_crp_hat&yr ne . )) b
             on a.pct_crp&yr = b.pct_crp&yr & (a.cancre06 = b.cancre06 or missing(a.cancre06) or missing(b.cancre06) ) & (a.cancre08 = b.cancre08 or missing(a.cancre08) or missing(b.cancre08) )
                & (a.cancre10 = b.cancre10 or missing(a.cancre10) or missing(b.cancre10) ) & (a.cancre12 = b.cancre12 or missing(a.cancre12) or missing(b.cancre12) )
            /* & abs(a.pct_crp_hat-b.pct_crp_hat)<=(1.5*a.pct_crp_sd) */
             order a.hhidpn,calculated distance;
   
   create table no_donorcrp&yr as 
	select b.* from miss_donorcrp&yr (where=(donor_id = .)) a
	left join bio.bio_imp20tiles_wide b
	on a.hhidpn = b.hhidpn ;

   create table miss_donorcrp_nd&yr as 
      select a.hhidpn,&wv as crp_nm_wave,a.pct_crp_hat&yr as pct_crp_hat,a.pct_crp&yr as pct_crp_nm,b.hhidpn as donor_id,
             b.pct_crp_hat&yr as donorpct_crp_hat,
             b.pct_crp06 as donorpct_crp06,
             b.pct_crp08 as donorpct_crp08,
             b.pct_crp10 as donorpct_crp10,
             b.pct_crp12 as donorpct_crp12,
             b.pct_crp14 as donorpct_crp14,
             abs(a.pct_crp_hat&yr - b.pct_crp_hat&yr) as distance
             from no_donorcrp&yr (where=((crp_onewave=1 or crp_twowaves_sas=1) & pct_crp&yr ne . and pct_crp_hat&yr ne . )) a
             left join bio.bio_imp20tiles_wide (where=((crp_threewaves=1 or crp_twowaves_nosas=1) & pct_crp&yr ne . and pct_crp_hat&yr ne . )) b
             on a.pct_crp&yr = b.pct_crp&yr
            /* & abs(a.pct_crp_hat&yr-b.pct_crp_hat&yr)<=(1.5*a.pct_crp_sd&yr) */
             order a.hhidpn,calculated distance;

data miss_donorcrp&yr ; 
	merge miss_donorcrp&yr
        miss_donorcrp_nd&yr ;
        by hhidpn ;

  run;         
   
%mend;

%sqlyr(06,8)
%sqlyr(08,9)
%sqlyr(10,10)
%sqlyr(12,11)
%sqlyr(14,12)

proc sql;
   create table miss_donorcrp as 
      select a.hhidpn,0 as crp_nm_wave, a.pct_crp_hat,b.pct_crp_hat as donorpct_crp_hat,b.hhidpn as donor_id,
             b.pct_crp06 as donorpct_crp06,
             b.pct_crp08 as donorpct_crp08,
             b.pct_crp10 as donorpct_crp10,
             b.pct_crp12 as donorpct_crp12,
             b.pct_crp14 as donorpct_crp14,
             abs(a.pct_crp_hat-b.pct_crp_hat) as distance
             from bio.bio_imp20tiles_wide (where=(crp_nowave=1 and pct_crp_hat ne . )) a
             left join bio.bio_imp20tiles_wide (where=((crp_threewaves=1 or crp_twowaves_nosas=1) and pct_crp_hat ne . )) b
             on (a.cancre06 = b.cancre06 or missing(a.cancre06) or missing(b.cancre06) ) & (a.cancre08 = b.cancre08 or missing(a.cancre08) or missing(b.cancre08) )
                & (a.cancre10 = b.cancre10 or missing(a.cancre10) or missing(b.cancre10) ) & (a.cancre12 = b.cancre12 or missing(a.cancre12) or missing(b.cancre12) )
                & (a.cancre14 = b.cancre14 or missing(a.cancre14) or missing(b.cancre14) )
             /* & abs(a.pct_crp_hat-b.pct_crp_hat)<=(1.5*a.pct_crp_sd) */
             order a.hhidpn,calculated distance;




proc print data=miss_donorcrp06 (obs=200);
run;
proc print data=miss_donorcrp10 (where=(donor_id=.) obs=100);
	title2 donor id missing ; 
run;
title2 selected cases;

proc print data=miss_donorcrp (obs=200);
run;


/* only keep 10 in the donor pool */
data setup checkncrp;
   set miss_donorcrp
       miss_donorcrp06
       miss_donorcrp08
       miss_donorcrp10
       miss_donorcrp12
       miss_donorcrp14;
   by crp_nm_wave hhidpn ;
   retain n_donor;
   if first.hhidpn then n_donor=0;
   n_donor=n_donor+1;
   if last.hhidpn then do;
      if n_donor>=10 then n_donor_status=3;
      else if n_donor>=5 then n_donor_status=2;
      else if n_donor>0 then n_donor_status=1;
      else n_donor_status=0;
      output checkncrp;
   end;
   /* if one of the candidate donors then assign random number and output */
   if n_donor<=10 then do;
      ran_num=ranuni(&seed);
      output setup;
   end;
run;

proc print data=checkncrp (obs=200) ;
run;

 
proc means data=checkncrp;
  class n_donor_status;
  var n_donor;
  run;
  
/*proc print data=checkn&bio (where=(n_donor<5) obs=20);
title2 hhidpn-wave with fewer than 5 donors;
run;

title2;
*/

proc sort data=setup;
   by crp_nm_wave hhidpn ran_num;

data imputedcrp;
   set setup;
   by crp_nm_wave hhidpn ;
   if first.hhidpn;
*  imp_pct_crp=donorpct_crp08;
   run;
proc sort data=imputedcrp out=bio.imputedcrp;
	   by hhidpn;
	   
proc print data=bio.imputedcrp (obs=100) ;
run;

	   
data bio.bio_imp_finalcrp;
   merge bio.imputedcrp (in=_inimp keep=hhidpn donorpct_crp: crp_nm_wave)
         bio.bio_hrs_recoded_prepimp_v1 (in=_inpct)
         ;
   by hhidpn ;
   
   impflag2_crp=_inimp;
   
   if missing(pct_crp) and wave=8 then pct_crp=donorpct_crp06;
   if missing(pct_crp) and wave=9 then pct_crp=donorpct_crp08;
   if missing(pct_crp) and wave=10 then pct_crp=donorpct_crp10;
   if missing(pct_crp) and wave=11 then pct_crp=donorpct_crp12;
   if missing(pct_crp) and wave=12 then pct_crp=donorpct_crp14;
   
   if impflag2_crp=1 and missing(pct_crp)=1 then impflag2_crp=9;
   
run;

proc contents data=bio.bio_imp_finalcrp;
run;

proc means data=bio.bio_imp_finalcrp;
   class impflag2_crp;
   types () impflag2_crp;
   var pct_crp ;
   run;




  
data bio.bio_imp_all ;
	merge bio.bio_imp_finala1c 
	      bio.bio_imp_finalhdl (keep=hhidpn wave donorpct_hdl06 donorpct_hdl08 donorpct_hdl10 donorpct_hdl12 donorpct_hdl14 pct_hdl impflag2_hdl pct_hdl_hat)
	      bio.bio_imp_finaltchol (keep=hhidpn wave donorpct_tchol06 donorpct_tchol08 donorpct_tchol10 donorpct_tchol12 donorpct_tchol14 pct_tchol impflag2_tchol pct_tchol_hat)
	      bio.bio_imp_finalsysbp (keep=hhidpn wave donorpct_sysbp06 donorpct_sysbp08 donorpct_sysbp10 donorpct_sysbp12 donorpct_sysbp14 pct_sysbp impflag2_sysbp pct_sysbp_hat)
	      bio.bio_imp_finalcrp (keep=hhidpn wave donorpct_crp06 donorpct_crp08 donorpct_crp10 donorpct_crp12 donorpct_crp14 pct_crp impflag2_crp pct_crp_hat)
	      ;
	by hhidpn wave ;
run;