ADTTE

Time to Event Analysis Dataset

BASIC DATA STRUCTURE · one record per subject per parameter · keys STUDYID, USUBJID, PARAMCD

Variables

Program adtte.sas

Version 1, written from Clinwright's patterns, 2026-10-17, checks passed.

/*==== 00-header ====*/
/*----------------------------------------------------------------------------
  Data set : ADTTE (Time to Event Analysis Dataset)
  Reporting effort : CSR
  ADaM     : ADRS, ADSL
  The header and the finalize step follow from the specification.
----------------------------------------------------------------------------*/
libname sdtm "/work/sdtm" access=readonly;
libname adam "/work/adam";
options validvarname=upcase nofmterr msglevel=i;

/*==== 10-inputs ====*/
/* The subjects of ADSL, with ADSL's variables. */
data work._ds;
  set adam.adsl;
run;

/*==== 20-parameters ====*/
/* ADTTE for oncology: overall survival (OS), progression-free survival (PFS) and duration of response (RSD), as
   admiralonco derives them (see SKILL.md).
   Inputs: ADAM.ADSL (RANDDT, LSTALVDT), ADAM.ADRS (PD, DEATH, LSTA, RSP by the evaluator analysed).
   Output: WORK.ADTTE, one record per subject per parameter. */

/* Each parameter's events and censorings: (parameter, E or C, its order among them, where the date comes from). */
data _rules;
  length PARAMCD $ 8 PARAM $ 40 _kind $ 1 _src $ 10 EVNTDESC CNSDTDSC $ 40;
  input PARAMCD $ _kind $ _order _src $ @;
  input EVNTDESC & $40. CNSDTDSC & $40.;
  select (PARAMCD);
    when ('OS') PARAM = 'Overall Survival';
    when ('PFS') PARAM = 'Progression Free Survival';
    when ('RSD') PARAM = 'Duration of Response';
    otherwise;
  end;
  datalines;
OS E 1 death Death  .
OS C 1 lastalive Alive  Alive During Study
OS C 2 rand Randomization  Randomization
PFS E 1 pd Disease Progression  .
PFS E 2 death Death  .
PFS C 1 lasta Last Tumor Assessment  Last Tumor Assessment
PFS C 2 rand Randomization  Randomization
RSD E 1 pd Disease Progression  .
RSD E 2 death Death  .
RSD C 1 lasta Last Tumor Assessment  Last Tumor Assessment
;
run;

/* The dates each source gives a subject. */
data _dates(keep=USUBJID _src ADT SRCDOM SRCVAR SRCSEQ);
  length _src $ 10 SRCDOM $ 8 SRCVAR $ 8;
  set adam.adrs(keep=USUBJID PARAMCD AVALC ANL01FL ADT ASEQ in=_r) adam.adsl(keep=USUBJID RANDDT LSTALVDT in=_s);
  if _r then do;
    SRCDOM = 'ADRS';
    SRCVAR = 'ADT';
    SRCSEQ = ASEQ;
    if PARAMCD = 'DEATH' and AVALC = 'Y' and ANL01FL = 'Y' then do; _src = 'death'; output; end;
    if PARAMCD = 'PD' and AVALC = 'Y' and ANL01FL = 'Y' then do; _src = 'pd'; output; end;
    if PARAMCD = 'LSTA' and ANL01FL = 'Y' then do; _src = 'lasta'; output; end;
  end;
  if _s then do;
    SRCDOM = 'ADSL';
    SRCSEQ = .;
    _src = 'lastalive'; SRCVAR = 'LSTALVDT'; ADT = LSTALVDT; output;
    _src = 'rand'; SRCVAR = 'RANDDT'; ADT = RANDDT; output;
  end;
run;

/* Each parameter's start: randomization; for the duration of response, the first response. */
data _start(keep=USUBJID PARAMCD STARTDT);
  length PARAMCD $ 8;
  set adam.adsl(keep=USUBJID RANDDT in=_s) adam.adrs(keep=USUBJID PARAMCD AVALC ANL01FL ADT in=_r rename=(PARAMCD=_p));
  if _s then do;
    STARTDT = RANDDT;
    PARAMCD = 'OS'; output;
    PARAMCD = 'PFS'; output;
  end;
  if _r and _p = 'RSP' and AVALC = 'Y' and ANL01FL = 'Y' then do;
    STARTDT = ADT;
    PARAMCD = 'RSD';
    output;
  end;
  format STARTDT yymmdd10.;
run;

/* Candidates: every dated event and censoring of a subject with a start. */
proc sql;
  create table _cand as
  select s.USUBJID, s.PARAMCD, s.STARTDT, r.PARAM, r._kind, r._order, r.EVNTDESC, r.CNSDTDSC,
         d.ADT, d.SRCDOM, d.SRCVAR, d.SRCSEQ
  from _start as s
       inner join _rules as r on r.PARAMCD = s.PARAMCD
       inner join _dates as d on d.USUBJID = s.USUBJID and d._src = r._src
  where s.STARTDT is not missing and d.ADT is not missing;
quit;

/* An event when there is one: the earliest (on one date, the event listed first); else a censoring: the latest
   (on one date, the one listed last). */
data _cand;
  set _cand;
  _event = (_kind = 'E');
  /* Sort key: events after censorings; among events the earliest, among censorings the latest, last. */
  if _event then _key = -ADT;
  else _key = ADT;
  _tie = ifn(_event, -_order, _order);
run;
proc sort data=_cand;
  by USUBJID PARAMCD _event _key _tie;
run;
data adtte;
  set _cand;
  by USUBJID PARAMCD;
  if last.PARAMCD;
  CNSR = 1 - _event;
  if _event then CNSDTDSC = '';
  AVAL = ADT - STARTDT + 1;
  format ADT yymmdd10.;
  drop _kind _order _event _key _tie;
run;

/* Sequence and ADSL's variables. */
proc sort data=adtte;
  by USUBJID PARAMCD;
run;
data work._ds;
  merge adtte(in=_in) adam.adsl;
  by USUBJID;
  if _in;
  if first.USUBJID then ASEQ = 0;
  ASEQ + 1;
run;

/*==== 89-final ====*/
/* The data set as derived; the finalize step keeps the specification's variables. */
data work._final;
  set work._ds;
run;

/*==== 90-finalize ====*/
/* 90-finalize: from the specification; the variables, their order and attributes are not changed here. */
/* The text variables the specification gives no length: as long as their longest value (at least 1). */
%let _len1 = 1; %let _len2 = 1; %let _len3 = 1; %let _len4 = 1; %let _len5 = 1; %let _len6 = 1; %let _len7 = 1; %let _len8 = 1; %let _len9 = 1; %let _len10 = 1; %let _len11 = 1; %let _len12 = 1; %let _len13 = 1; %let _len14 = 1;
data _null_;
  length _ren _asg _siz $32767;
  _dsid = open('work._final');
  if varnum(_dsid, 'STUDYID') > 0 then do; _ren = catx(' ', _ren, 'STUDYID=_rz1'); _asg = catx(' ', _asg, 'STUDYID = _rz1;');
    _siz = catx(' ', _siz, '_len1 = max(_len1, lengthn(STUDYID));'); end;
  if varnum(_dsid, 'USUBJID') > 0 then do; _ren = catx(' ', _ren, 'USUBJID=_rz2'); _asg = catx(' ', _asg, 'USUBJID = _rz2;');
    _siz = catx(' ', _siz, '_len2 = max(_len2, lengthn(USUBJID));'); end;
  if varnum(_dsid, 'SITEID') > 0 then do; _ren = catx(' ', _ren, 'SITEID=_rz3'); _asg = catx(' ', _asg, 'SITEID = _rz3;');
    _siz = catx(' ', _siz, '_len3 = max(_len3, lengthn(SITEID));'); end;
  if varnum(_dsid, 'SEX') > 0 then do; _ren = catx(' ', _ren, 'SEX=_rz4'); _asg = catx(' ', _asg, 'SEX = _rz4;');
    _siz = catx(' ', _siz, '_len4 = max(_len4, lengthn(SEX));'); end;
  if varnum(_dsid, 'TRT01P') > 0 then do; _ren = catx(' ', _ren, 'TRT01P=_rz5'); _asg = catx(' ', _asg, 'TRT01P = _rz5;');
    _siz = catx(' ', _siz, '_len5 = max(_len5, lengthn(TRT01P));'); end;
  if varnum(_dsid, 'TRT01A') > 0 then do; _ren = catx(' ', _ren, 'TRT01A=_rz6'); _asg = catx(' ', _asg, 'TRT01A = _rz6;');
    _siz = catx(' ', _siz, '_len6 = max(_len6, lengthn(TRT01A));'); end;
  if varnum(_dsid, 'SAFFL') > 0 then do; _ren = catx(' ', _ren, 'SAFFL=_rz7'); _asg = catx(' ', _asg, 'SAFFL = _rz7;');
    _siz = catx(' ', _siz, '_len7 = max(_len7, lengthn(SAFFL));'); end;
  if varnum(_dsid, 'ITTFL') > 0 then do; _ren = catx(' ', _ren, 'ITTFL=_rz8'); _asg = catx(' ', _asg, 'ITTFL = _rz8;');
    _siz = catx(' ', _siz, '_len8 = max(_len8, lengthn(ITTFL));'); end;
  if varnum(_dsid, 'PARAMCD') > 0 then do; _ren = catx(' ', _ren, 'PARAMCD=_rz9'); _asg = catx(' ', _asg, 'PARAMCD = _rz9;');
    _siz = catx(' ', _siz, '_len9 = max(_len9, lengthn(PARAMCD));'); end;
  if varnum(_dsid, 'PARAM') > 0 then do; _ren = catx(' ', _ren, 'PARAM=_rz10'); _asg = catx(' ', _asg, 'PARAM = _rz10;');
    _siz = catx(' ', _siz, '_len10 = max(_len10, lengthn(PARAM));'); end;
  if varnum(_dsid, 'EVNTDESC') > 0 then do; _ren = catx(' ', _ren, 'EVNTDESC=_rz11'); _asg = catx(' ', _asg, 'EVNTDESC = _rz11;');
    _siz = catx(' ', _siz, '_len11 = max(_len11, lengthn(EVNTDESC));'); end;
  if varnum(_dsid, 'CNSDTDSC') > 0 then do; _ren = catx(' ', _ren, 'CNSDTDSC=_rz12'); _asg = catx(' ', _asg, 'CNSDTDSC = _rz12;');
    _siz = catx(' ', _siz, '_len12 = max(_len12, lengthn(CNSDTDSC));'); end;
  if varnum(_dsid, 'SRCDOM') > 0 then do; _ren = catx(' ', _ren, 'SRCDOM=_rz13'); _asg = catx(' ', _asg, 'SRCDOM = _rz13;');
    _siz = catx(' ', _siz, '_len13 = max(_len13, lengthn(SRCDOM));'); end;
  if varnum(_dsid, 'SRCVAR') > 0 then do; _ren = catx(' ', _ren, 'SRCVAR=_rz14'); _asg = catx(' ', _asg, 'SRCVAR = _rz14;');
    _siz = catx(' ', _siz, '_len14 = max(_len14, lengthn(SRCVAR));'); end;
  _rc = close(_dsid);
  if _ren ne '' then _ren = cats('(rename=(', _ren, '))');
  call symputx('_ren', _ren);
  call symputx('_asg', _asg);
  call symputx('_siz', _siz);
run;
data _null_;
  set work._final end=_end;
  retain _len1 1 _len2 1 _len3 1 _len4 1 _len5 1 _len6 1 _len7 1 _len8 1 _len9 1 _len10 1 _len11 1 _len12 1 _len13 1 _len14 1;
  &_siz
  if _end then do;
    call symputx('_len1', _len1);
    call symputx('_len2', _len2);
    call symputx('_len3', _len3);
    call symputx('_len4', _len4);
    call symputx('_len5', _len5);
    call symputx('_len6', _len6);
    call symputx('_len7', _len7);
    call symputx('_len8', _len8);
    call symputx('_len9', _len9);
    call symputx('_len10', _len10);
    call symputx('_len11', _len11);
    call symputx('_len12', _len12);
    call symputx('_len13', _len13);
    call symputx('_len14', _len14);
  end;
run;
data adam.adtte(label="Time to Event Analysis Dataset");
  attrib
    STUDYID  length=$&_len1 label="Study Identifier"
    USUBJID  length=$&_len2 label="Unique Subject Identifier"
    SITEID   length=$&_len3 label="Study Site Identifier"
    AGE      length=8 label="Age"
    SEX      length=$&_len4 label="Sex"
    TRT01P   length=$&_len5 label="Planned Treatment for Period 01"
    TRT01PN  length=8 label="Planned Treatment for Period 01 (N)"
    TRT01A   length=$&_len6 label="Actual Treatment for Period 01"
    SAFFL    length=$&_len7 label="Safety Population Flag"
    ITTFL    length=$&_len8 label="Intent-To-Treat Population Flag"
    ASEQ     length=8 label="Analysis Sequence Number"
    PARAMCD  length=$&_len9 label="Parameter Code"
    PARAM    length=$&_len10 label="Parameter"
    STARTDT  length=8 label="Time to Event Origin Date for Subject" format=DATE9.
    ADT      length=8 label="Analysis Date" format=DATE9.
    AVAL     length=8 label="Analysis Value"
    CNSR     length=8 label="Censor"
    EVNTDESC length=$&_len11 label="Event or Censoring Description"
    CNSDTDSC length=$&_len12 label="Censor Date Description"
    SRCDOM   length=$&_len13 label="Source Data"
    SRCVAR   length=$&_len14 label="Source Variable"
    SRCSEQ   length=8 label="Source Sequence Number"
  ;
  set work._final&_ren;
  &_asg
  keep STUDYID USUBJID SITEID AGE SEX TRT01P TRT01PN TRT01A SAFFL ITTFL ASEQ PARAMCD PARAM STARTDT ADT AVAL CNSR EVNTDESC CNSDTDSC SRCDOM SRCVAR SRCSEQ;
run;
proc sort data=adam.adtte; by STUDYID USUBJID PARAMCD; run;

Data (first 20 of 690 records, simulated)