/*
   remerge-with-addresses-09jan01.sas

   directory:  C:\Documents and Settings\John Bailer\
                 My Documents\baileraj\Grants\ODHS-oct98\SAS-programs

   modified:   merge-mds-odh-31jan2000.sas
*/

/* data files -- MDS-Plus quarterly surveys 
              -- one record per facility


*** includes 2 separate fixes for invalid H20 entries in the ODH data
    (based on fix1-nomatch-31jan2000.sas and fix2-*.sas)

*/

Libname insas "d:\baileraj\grants\odhs-oct98\Datasets"; **** changed 5jul00;

/*
data mdsmar98;   set insas.summar98;  matcher = provider;   * <-- no observations!!! ;
*/

data mdsdec97;   set insas.sumdec97;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdssep97;   set insas.sumsep97;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdsjun97;   set insas.sumjun97;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdsmar97;   set insas.summar97;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;

data mdsdec96;   set insas.sumdec96;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdssep96;   set insas.sumsep96;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdsjun96;   set insas.sumjun96;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdsmar96;   set insas.summar96;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;

data mdsdec95;   set insas.sumdec95;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdssep95;   set insas.sumsep95;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdsjun95;   set insas.sumjun95;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;
data mdsmar95;   set insas.summar95;  matcher = provider;
  keep matcher provider casebar casemed caseq1 caseq3 
       agebar agemed ageq1 ageq3 rugrp1-rugrp45 quarter numres;

/* data files -- ODH annual surveys  */

libname inodh "d:\baileraj\grants\odhs-oct98\ODHS-survey";

/* set up to include some text for fixing mismatched records  */

filename job95 "d:\baileraj\grants\odhs-oct98\SAS-programs";

data odh95; set inodh.odh95all;

%include job95(fix1-nomatch-31jan2000); * 1st set of fixes - ifix1 defined;
%include job95(fix2-nomatch-31jan2000); * 2nd set of fixes - ifix2 defined;
if ifix1=0 and ifix2=0 then matcher=substr(h20,1,7);

*  matcher = h20;

  if H19B>0 or H19C>0;
*  if H19B + H19C > 0;          * restrict to Medicaid certified;

  facname = a1b;
  facadd = a1c;  faccity = a1d;  facstat = a1e; faczip = a1f; 
  facphone = a1h;
  county = a1g;
  orgtype = b1;
  license = b2a;

  palzday = .;
  if e1b1 > 0 then   palzday = b4c/e1b1;   * proportion alzheimer days;
  adjbed = NF_BEDS;         * adjusted bed capacity;
*   adjbed = h18;         * adjusted bed capacity;
  pmcarebd = .;  pmcaidbd = .;
  if h9 > 0 then do;
     pmcarebd = h19b/h9;   * proportion medicare beds;
     pmcaidbd = h19c/h9;   * proportion medicaid beds;
  end;

  ventil = "";           * ventilator data missing in 95 and 96;

  nurshrs = (c1d1 + c1e1)*36 + (c1d4 + c1e4)*18;  * nurse hours;
  aidehrs = c1f1*36 + c1f4*18;                    * aide hours;
  therhrs = (c1j1 +c1k1 + c1l1 + c1m1 + c1p1)*36 +
            (c1j4 +c1k4 + c1l4 + c1m4 + c1p4)*18; * therapy hours;

  occpncy = .; admperbd = .;  dthperbd=.;
  if h18 > 0 then do;
     occpncy = (e1b1-e1b7)/(NF_BEDS*365.25); * proportion of beds occupied in a year;
     admperbd = (e1c1-e1c7)/NF_BEDS;         * admission per bed available;
     dthperbd = (e1e1-e1e7)/NF_BEDS;         * dead per bed available;
* modified to use NF_BEDS instead of h18 (24 July 00);
* modified to remove rest home beds -- non-MDS stuff (19 July 00);
  end;

  pprivpay=.;  pmcarepy=.; pmcaidpy=.;
  if e7i > 0 then do;
     pprivpay = (e7e1+e7e2+e7e3)/e7i; * proportion private pay;
     pmcarepy = e7a/e7i;              * proportion medicare pay;
     pmcaidpy = e7c/e7i;              * proportion medicaid pay;
  end;

  mcareavg = g1a;                  * avg. $ medicare per day;
  mcaidavg = g1b;                  * avg. $ medicaid per day;
  privavg = g1d3;                 * avg. $ private pay per day;
  rename provider=nprovidr;    * defined as NUMERIC in conflict with MDS CHARACTER;
  keep h20 matcher--privavg ifix1 ifix2;


/*
   intermediate printing of 1995 ODH data <--------------------------<<
*/

/*
proc sort data=odh95; by ifix;
proc print data=odh95; by ifix;
title 1995 ODH data;
  var matcher H20;
run;
*/

data odh96; set inodh.odh96all;

%include job95(fix1-nomatch-31jan2000); * 1st set of fixes - ifix1 defined;
%include job95(fix2-nomatch-31jan2000); * 2nd set of fixes - ifix2 defined;
if ifix1=0 and ifix2=0 then matcher=substr(h20,1,7);

*  matcher = h20;

  if H19B>0 or H19C>0;
*  if H19B + H19C > 0;          * restrict to Medicaid certified;

  facname = a1b;
  facadd = a1c;  faccity = a1d;  facstat = a1e; faczip = a1f; 
  facphone = a1h;
  county = a1g;
  orgtype = b1;
  license = b2a;

  palzday = .;
  if e1b1 > 0 then palzday = b4c/e1b1;   * proportion alzheimer days;

  adjbed = NF_BEDS;         * adjusted bed capacity;
*   adjbed = h18;         * adjusted bed capacity;

  pmcarebd = .;  pmcaidbd = .;
  if h9 > 0 then do;
     pmcarebd = h19b/h9;   * proportion medicare beds;
     pmcaidbd = h19c/h9;   * proportion medicaid beds;
  end;

  ventil = "";           * ventilator data missing in 95 and 96;

  nurshrs = (c1d1 + c1e1)*36 + (c1d4 + c1e4)*18;  * nurse hours;
  aidehrs = c1f1*36 + c1f4*18;                    * aide hours;
  therhrs = (c1j1 +c1k1 + c1l1 + c1m1 + c1p1)*36 +
            (c1j4 +c1k4 + c1l4 + c1m4 + c1p4)*18; * therapy hours;

  occpncy = .; admperbd = .;  dthperbd=.;
  if h18 > 0 then do;
     occpncy = (e1b1-e1b7)/(NF_BEDS*365.25); * proportion of beds occupied in a year;
     admperbd = (e1c1-e1c7)/NF_BEDS;         * admission per bed available;
     dthperbd = (e1e1-e1e7)/NF_BEDS;         * dead per bed available;
* modified to use NF_BEDS instead of h18 (24 July 00);
* modified to remove rest home beds -- non-MDS stuff (19 July 00);
  end;

  pprivpay=.;  pmcarepy=.; pmcaidpy=.;
  if e7i > 0 then do;
     pprivpay = (e7e1+e7e2+e7e3)/e7i; * proportion private pay;
     pmcarepy = e7a/e7i;              * proportion medicare pay;
     pmcaidpy = e7c/e7i;              * proportion medicaid pay;
  end;

  mcareavg = g1a;                  * avg. $ medicare per day;
  mcaidavg = g1b;                  * avg. $ medicaid per day;
  privavg = g1d3;                 * avg. $ private pay per day;

  a1anum = a1a*1.;  * coerce a1a to numeric (currently character in odh96);
  rename provider=nprovidr;    * defined as NUMERIC in conflict with MDS CHARACTER;
  keep a1anum h20 matcher--privavg ifix1 ifix2;

/*
   intermediate printing of 1996 ODH data <-------------------------<<
*/

/*
proc sort data=odh96; by ifix;
proc print data=odh96; by ifix;
title 1996 ODH data;
  var matcher H20;
run;
*/

/*
   1997 does not have H20 defined and organiz. type is also not defined!
*/
data odh97; set inodh.odh97all;

data junk96a; set odh96; keep a1anum h20 orgtype matcher ifix1 ifix2;
data junk96b; set junk96a; a1a = a1anum; drop a1anum;

proc sort data=junk96b out=s96; by a1a;
data s96; set s96; keep a1a h20 matcher ifix1 ifix2;

proc sort data=odh97 out=s97; by a1a;

data newodh97; merge s97 s96; by a1a;
*  matcher = h20;
  if H19B>0 or H19C>0;
*  if H19B + H19C > 0;          * restrict to Medicaid certified;

  facname = a1b;
  facadd = a1c;  faccity = a1d;  facstat = a1e; faczip = a1f; 
  facphone = a1h;
  county = a1g;

  palzday = .;
  if e1b1 > 0 then palzday = alz1c/e1b1;   * proportion alzheimer days;

  adjbed = NF_BEDS;         * adjusted bed capacity;
*   adjbed = h18;         * adjusted bed capacity;

  pmcarebd = .;  pmcaidbd = .;
  if h9 > 0 then do;
     pmcarebd = h19b/h9;   * proportion medicare beds;
     pmcaidbd = h19c/h9;   * proportion medicaid beds;
  end;

  ventil = vent2a;      * ventilator data missing in 95 and 96;

  nurshrs = (c1d1 + c1e1)*36 + (c1d4 + c1e4)*18;  * nurse hours;
  aidehrs = c1f1*36 + c1f4*18;                    * aide hours;
  therhrs = (c1j1 +c1k1 + c1l1 + c1m1 + c1p1)*36 +
            (c1j4 +c1k4 + c1l4 + c1m4 + c1p4)*18; * therapy hours;


  occpncy = .; admperbd = .;  dthperbd=.;
  if h18 > 0 then do;
     occpncy = (e1b1-e1bR)/(NF_BEDS*365.25); * proportion of beds occupied in a year;
     admperbd = (e1c1-e1cR)/NF_BEDS;         * admission per bed available;
     dthperbd = (e1e1-e1eR)/NF_BEDS;         * dead per bed available;
* modified to use NF_BEDS instead of h18 (24 July 00);
* modified to remove rest home beds -- non-MDS stuff (19 July 00);
  end;

  pprivpay=.;  pmcarepy=.; pmcaidpy=.;
  if e7i > 0 then do;
     pprivpay = (e7e1+e7e2+e7e3)/e7i; * proportion private pay;
     pmcarepy = e7a/e7i;              * proportion medicare pay;
     pmcaidpy = e7c/e7i;              * proportion medicaid pay;
  end;

  mcareavg = g1a;                  * avg. $ medicare per day;
  mcaidavg = g1b;                  * avg. $ medicaid per day;
  privavg = g1d3;                  * avg. $ private pay per day;
  keep h20 matcher--privavg ifix1 ifix2;

run;

/* 
  try to merge
*/

/*  SORT the ODH annual data                  */

proc sort data=odh95 out=sodh95; by matcher;
proc sort data=odh96 out=sodh96; by matcher;
proc sort data=newodh97 out=sodh97; by matcher;

/*  SORT all of the MDS quarterly data        */

proc sort data=mdsmar95 out=smar95; by matcher;
proc sort data=mdsmar96 out=smar96; by matcher;
proc sort data=mdsmar97 out=smar97; by matcher;

proc sort data=mdsjun95 out=sjun95; by matcher;
proc sort data=mdsjun96 out=sjun96; by matcher;
proc sort data=mdsjun97 out=sjun97; by matcher;

proc sort data=mdssep95 out=ssep95; by matcher;
proc sort data=mdssep96 out=ssep96; by matcher;
proc sort data=mdssep97 out=ssep97; by matcher;

proc sort data=mdsdec95 out=sdec95; by matcher;
proc sort data=mdsdec96 out=sdec96; by matcher;
proc sort data=mdsdec97 out=sdec97; by matcher;

/*   now merge each quarter's MDS with the annual ODH demographic data */

data mermar95; merge smar95 sodh95; by matcher;
data mermar96; merge smar96 sodh96; by matcher;
data mermar97; merge smar97 sodh97; by matcher;
run;

data merjun95; merge sjun95 sodh95; by matcher;
data merjun96; merge sjun96 sodh96; by matcher;
data merjun97; merge sjun97 sodh97; by matcher;
run;

data mersep95; merge ssep95 sodh95; by matcher;
data mersep96; merge ssep96 sodh96; by matcher;
data mersep97; merge ssep97 sodh97; by matcher;
run;

data merdec95; merge sdec95 sodh95; by matcher;
data merdec96; merge sdec96 sodh96; by matcher;
data merdec97; merge sdec97 sodh97; by matcher;
run;

/*
  now comes the fun ...
  combine all of the data sets ...
*/

data y95; set mermar95 merjun95 mersep95 merdec95; run;
data y96; set mermar96 merjun96 mersep96 merdec96; run;
data y97; set mermar97 merjun97 mersep97 merdec97; run;

data y9596; set y95 y96; run;
data y959697; set y9596 y97; 
  if numres > 0;   * delete facilities with zero residents reported (19jul00);
  run;

data insas.ny959697; set y959697; 
  if county="ASHT" or county="CUYA" or county="GEAU" or county="LAKE" or
     county="LORA" or county="MEDI" or county="PORT" or county="SUMM" then
     geopeer = "NECMSA";

  else if county="BROW" or county="BUTL" or county="CLER" or county="HAMI" or
     county="WARR" then
     geopeer = "SWCMSA";

  else if county="FULT" or county="LUCA" or county="WOOD" or county="ALLE" or
     county="AUGL" or county="MONT" or county="MIAM" or county="CLAR" or
     county="GREE" or county="CRAW" or county="RICH" or county="DELA" or
     county="LICK" or county="FRAN" or county="MADI" or county="PICK" or
     county="FAIR" or county="LAWR" or county="STAR" or county="CARR" or 
     county="TRUM" or county="MAHO" or county="COLU" or county="JEFF" or
     county="BELM" or county="WASH" then
     geopeer = "MSA";

* spelled out OTHER geopeer counties ... 5 july 2000;
  else if county="WILL" or county="DEFI" or county="PAUL" or county="VANW" or
     county="MERC" or county="DARK" or county="PREB" or county="HENR" or
     county="PUTN" or county="SHEL" or county="HANC" or county="HARD" or
     county="LOGA" or county="CHAM" or county="CLIN" or county="HIGH" or
     county="ADAM" or county="FAYE" or county="UNIO" or county="MARI" or 
     county="WYAN" or county="SENE" or county="SAND" or county="OTTA" or 
     county="ERIE" or county="HURO" or county="UNIO" or county="MORR" or 
     county="ROSS" or county="PIKE" or county="SCIO" or county="ASHL" or 
     county="KNOX" or county="HOCK" or county="VINT" or county="JACK" or 
     county="GALL" or county="MEIG" or county="ATHE" or county="PERR" or 
     county="MUSK" or county="COSH" or county="HOLM" or county="WAYN" or 
     county="TUSC" or county="GUER" or county="NOBL" or county="MORG" or 
     county="MONR" 
     then geopeer ="OTHER";

else geopeer="      ";

if geopeer="NECMSA" or geopeer="SWCMSA" or geopeer="MSA" then urban=1;
  else if geopeer="OTHER " then urban=0;
  else urban=.;

* defined per capita hours variables;
  pnurshrs = nurshrs/adjbed;   * by # of beds 25 July 2000;
  paidehrs = aidehrs/adjbed;   * hrs/bed/week;
  ptherhrs = therhrs/adjbed; 

if pnurshrs <=30 AND paidehrs <=30 and ptherhrs <=15;
  * eliminating impossible values;

if orgtype <=3 then corgtype = "for profit";
else if orgtype <=6 then corgtype = "non profit"; * grouping orgtype;
else if orgtype <=9 then corgtype = "    public"; *   25 july 2000  ;

if occpncy >=1 then occpncy=.;   * set occupancy>=1 to missing 25july2000;

run;

data insas.ndec97; set insas.ny959697;
  if quarter="dec97";
run;

 
/*
proc sort; by quarter;
proc means data=insas.ny959697; by quarter;
  var occpncy admperbd dthperbd pnurshrs paidehrs ptherhrs;
run; 
*/

/*
proc freq data=insas.ny959697;
  table county*geopeer geopeer*urban; run;
*/

proc contents data=insas.ny959697; run;

/*
data testjunk; set insas.y959697;
  imatch = (matcher EQ provider);
  inomissn = (matcher NE "       ") and (provider NE "       ");
*  if quarter = "sep97";

  proc freq data=testjunk;
    table ifix1 ifix2 ifix1*ifix2 imatch imatch*ifix1 imatch*ifix2;

run;

proc sort data=testjunk; by imatch inomissn matcher;
proc print data=testjunk;
 var imatch inomissn matcher h20 facname provider 
     quarter ifix1 ifix2 agebar casebar;
run;

data test2; set testjunk;
  if imatch=0 or inomissn=0;
run;

proc sort data=test2; by h20 facname;
proc print data=test2;
title PRINT of not matched or not missing;
  var imatch inomissn matcher h20 facname provider 
      quarter ifix1 ifix2 agebar casebar;
run;

proc sort data=test2; by h20 facname provider;
proc print data=test2;
  var facname h20 county provider casebar;
run;

data bigboy; set mermar95 merjun95 mersep95 merdec95
                 mermar96 merjun96 mersep96 merdec96
                 mermar97 merjun97 mersep97 merdec97;
run;

*/


/*
data mermar97; merge smar97 sodh97; by matcher;
run;

proc print data=mermar95;
 var matcher provider h20 a1b a1g b1 b2a h1 agebar casebar;
run;

proc print data=mermar97;
 var matcher provider id a1b a1g b1 b2a h1 agebar casebar;
run;

proc contents p data=smar95; run;
proc contents p data=sodh95; run;
proc contents p data=smar97; run;
proc contents p data=sodh97; run;

*/