Question

嗨SAS人，

我正在使用医院访问数据开展一些研究项目。我必须确定转换治疗的受试者。

治疗按局部，非生物系统和生物学分组。切换事件被定义为在最后一次治疗取出时未取出的药物的撤回。有一个层次结构，生物制剂位于顶部，然后是非生物系统，然后是局部。

如果首先观察患者撤回局部，则在此之后将任何治疗撤回到生物，非生物系统或新的局部是开关。
如果首先观察到患者撤回非生物系统性疾病，那么之后将任何治疗方法撤回生物或新的非生物系统是一种转变。
任何局部的撤回都不是开关。
如果首先观察到患者撤回生物，则只能转换为另一种生物（与第一种不同）。
任何局部或非生物系统的撤出都不是转换。

根据上述逻辑，一旦患者转为更高级别的治疗，较低等级被认为是伴随用药，并且同时用药的撤销并不表示治疗开关。

简而言之，如果患者从较低级别的治疗转为上级或同一级别（不同的ATC代码），那么将考虑转换。

治疗层级

生物制剂＆gt; Non_Bio_Sys＆gt;主题

我正在为参考添加样本数据集，其中包含可变信息：

LPNR: unique subjID
EDATUM: prescrition date ( treatment recieved date)
ATC_CODE: codes for each of treament type
          'D05AX52',
          'D05AX02' = PSorasis(Topical) ;

          'L01XC02',
          'L04AA24',
          'L04AB01',
          'L04AB02',
          'L04AB04',
          'L04AB05',
          'L04AB06',
          'L04AC03',
          'L04AC07' = Non_Bio_systemic ;

          'A11CC03',           
          'D05BB02',                
          'L01BA01',               
          'L01BB03',               
          'L01XX05',               
          'L04AA21',               
          'L04AD01',               
          'L04AX03' = Biologics ;
TYPE: Treatment type

样本数据：

data have;
  length TYPE $20.;
  input pnr ATC_CODE $ type $ EDATUM $ 10. ;
  date=input(edatum,mmddyy10.);
  format date mmddyy10.;
  cards;
478 L04AX03 Nonbiologic_systemic 2/10/2010 
478 D05AX52 Psoriasis 2/23/2010 
478 L04AX03 Nonbiologic_systemic 3/5/2010 
478 L04AX03 Nonbiologic_systemic 5/18/2010 
478 L04AX03 Nonbiologic_systemic 8/2/2010 
478 L04AX03 Nonbiologic_systemic 11/3/2010
478 L04AX03 Nonbiologic_systemic 2/7/2011 
478 L04AX03 Nonbiologic_systemic 8/16/2011
478 L04AX03 Nonbiologic_systemic 11/22/2011
603 D05AX02 Psoriasis 8/24/2005 
603 D05AX02 Psoriasis 10/13/2005 
603 D05AX02 Psoriasis 6/2/2006
603 D05AX02 Psoriasis 7/19/2006 
603 D05AX02 Psoriasis 4/3/2007
603 D05AX02 Psoriasis 12/7/2007 
603 D05AX02 Psoriasis 1/4/2008
603 D05AX02 Psoriasis 1/4/2008 
603 D05AX02 Psoriasis 10/2/2008
603 D05AX52 Psoriasis 3/16/2009
603 D05AX52 Psoriasis 3/16/2009
603 D05AX52 Psoriasis 5/7/2009
603 D05AX52 Psoriasis 8/21/2009
603 D05AX52 Psoriasis 11/9/2009 
603 D05AX52 Psoriasis 3/19/2010 
603 D05AX52 Psoriasis 7/30/2010 
603 D05AX52 Psoriasis 7/30/2010
603 D05AX52 Psoriasis 1/13/2011 
603 D05AX52 Psoriasis 5/3/2011 
603 D05AX52 Psoriasis 12/12/2011
1103 L04AX03 Nonbiologic_systemic 1/17/2006
1103 L04AX03 Nonbiologic_systemic 8/23/2006
1103 L04AX03 Nonbiologic_systemic 11/9/2006
1103 D05AX52 Psoriasis 4/19/2007 
1103 L04AX03 Nonbiologic_systemic 4/19/2007 
1103 D05AX52 Psoriasis 10/26/2007 
1103 D05AX52 Psoriasis 4/1/2008
1103 L04AX03 Nonbiologic_systemic 10/20/2008 
1103 D05AX52 Psoriasis 10/20/2008
1103 L04AX03 Nonbiologic_systemic 4/7/2009 
1103 L04AX03 Nonbiologic_systemic 11/20/2009 
1103 L04AX03 Nonbiologic_systemic 11/12/2010 
1103 L04AX03 Nonbiologic_systemic 8/12/2011
1103 D05AX52 Psoriasis 10/22/2011 
1103 D05AX52 Psoriasis 12/29/2011
;

我试过的解决方案：

proc sort data=have;
  by pnr date;
run;

/* Identifying the number of subjects who had treatment switch */
data need;
  set have;
  length change $ 10.;
  by pnr;

  /* Create new variable change and keep constant if it is topical for first observation */
  /* Compare topical values with next values if it changes to other non bio or bio then switched */
  if first.pnr and atc_code in ( 'D05AX52','D05AX02') then change="";
  else if not first.pnr and atc_code not in('D05AX52','D05AX02','L01XC02','L04AA24','L04AB01',
                                            'L04AB02','L04AB04','L04AB05','L04AB06','L04AC03',
                                            'L04AC07','A11CC03','D05BB02','L01BA01','L01BB03',
                                            'L01XX05' ,'L04AA21' ,'L04AD01' ,'L04AX03') then change='switched';
  /* Compare non bio values with next values if it changes to other non bio or bio then switched*/
  if first.pnr and atc_code in ('L01XC02','L04AA24','L04AB01','L04AB02','L04AB04',
                                'L04AB05','L04AB06','L04AC03','L04AC07') then change="";
  else if not first.pnr and atc_code not in( 'D05AX52','D05AX02') and 
                            atc_code in('L01XC02','L04AA24','L04AB01','L04AB02','L04AB04','L04AB05',
                                        'L04AB06','L04AC03','L04AC07','A11CC03','D05BB02','L01BA01',
                                        'L01BB03' ,'L01XX05' ,'L04AA21' ,'L04AD01' ,'L04AX03') then change='switched';

  /* Compare bio with next values if it changes to other bio then switched */
  if first.pnr and atc_code in('A11CC03','D05BB02','L01BA01','L01BB03' ,'L01XX05' ,
                               'L04AA21' ,'L04AD01' ,'L04AX03') then change='';
  else if not first.pnr and atc_code not in('D05AX52','D05AX02') and 
                            atc_code in('A11CC03','D05BB02','L01BA01','L01BB03' ,'L01XX05',
                                        'L04AA21' ,'L04AD01' ,'L04AX03') then change='switched';

  if atc_code=lag(atc_code) then change="";                        
run;

结果我希望

pnr   ATC_CODE  TYPE                  EDATUM      Flag_switch
478   L04AX03   Nonbiologic_systemic  2/10/2010
478   D05AX52   Psoriasis             2/23/2010
478   L04AX03   Nonbiologic_systemic  5/18/2010   switch
478   L04AX03   Nonbiologic_systemic  8/2/2010
478   L04AX03   Nonbiologic_systemic  11/3/2010
478   L04AX03   Nonbiologic_systemic  2/7/2011
478   L04AX03   Nonbiologic_systemic  8/16/2011
478   L04AX03   Nonbiologic_systemic  11/22/2011
603   D05AX02   Psoriasis             8/24/2005
603   D05AX02   Psoriasis             10/13/2005
603   D05AX02   Psoriasis             6/2/2006
603   D05AX02   Psoriasis             7/19/2006
603   D05AX02   Psoriasis             4/3/2007
603   D05AX02   Psoriasis             12/7/2007
603   D05AX02   Psoriasis             1/4/2008
603   D05AX02   Psoriasis             10/2/2008
603   D05AX52   Psoriasis             3/16/2009   switch
603   D05AX52   Psoriasis             5/7/2009
603   D05AX52   Psoriasis             8/21/2009
603   D05AX52   Psoriasis             11/9/2009
603   D05AX52   Psoriasis             3/19/2010
603   D05AX52   Psoriasis             7/30/2010
603   D05AX52   Psoriasis             1/13/2011
603   D05AX52   Psoriasis             5/3/2011
603   D05AX52   Psoriasis             12/12/2011
1103  L04AX03   Nonbiologic_systemic  1/17/2006
1103  L04AX03   Nonbiologic_systemic  8/23/2006
1103  L04AX03   Nonbiologic_systemic  11/9/2006
1103  L04AX03   Nonbiologic_systemic  4/19/2007
1103  D05AX52   Psoriasis             10/26/2007
1103  D05AX52   Psoriasis             4/1/2008
1103  L04AX03   Nonbiologic_systemic  10/20/2008  switch
1103  D05AX52   Psoriasis             10/20/2008
1103  L04AX03   Nonbiologic_systemic  4/7/2009    switch
1103  L04AX03   Nonbiologic_systemic  11/20/2009
1103  L04AX03   Nonbiologic_systemic  11/12/2010
1103  L04AX03   Nonbiologic_systemic  8/12/2011
1103  D05AX52   Psoriasis             10/22/2011
1103  D05AX52   Psoriasis             12/29/2011

Answer 1

好。所以这是几种可能的解决方案中的一种。您仍然需要计算出治疗类型3（生物制剂）的开关，并找出如何处理同一日期有多个条目的地方。但这应该让你去！

准备数据

proc format;
  value treatType 1 = "Topical"
                  2 = "Non-biologic systemics"
                  3 = "Biologics";
run;

data have;
  format pnr 8. tType treatType. tCode $10. date YYMMDDs8.;
  informat tType 1. date yymmdd10.;
  input pnr tCode tType date;
  datalines;
478 L04AX03 2 2010-02-10
478 D05AX52 1 2010-02-23
478 L04AX03 2 2010-03-05
478 L04AX03 2 2010-05-18
478 L04AX03 2 2010-08-02
478 L04AX03 2 2010-11-03
478 L04AX03 2 2011-02-07
478 L04AX03 2 2011-08-16
478 L04AX03 2 2011-11-22
603 D05AX02 1 2005-08-24
603 D05AX02 1 2005-10-13
603 D05AX02 1 2006-06-02
603 D05AX02 1 2006-07-19
603 D05AX02 1 2007-04-03
603 D05AX02 1 2007-12-07
603 D05AX02 1 2008-01-04
603 D05AX02 1 2008-01-04
603 D05AX02 1 2008-10-02
603 D05AX52 1 2009-03-16
603 D05AX52 1 2009-03-16
603 D05AX52 1 2009-05-07
603 D05AX52 1 2009-08-21
603 D05AX52 1 2009-11-09
603 D05AX52 1 2010-03-19
603 D05AX52 1 2010-07-30
603 D05AX52 1 2010-07-30
603 D05AX52 1 2011-01-13
603 D05AX52 1 2011-05-03
603 D05AX52 1 2011-12-12
1103 L04AX03 2 2006-01-17
1103 L04AX03 2 2006-08-23
1103 L04AX03 2 2006-11-09
1103 D05AX52 1 2007-04-19
1103 L04AX03 2 2007-04-19
1103 D05AX52 1 2007-10-26
1103 D05AX52 1 2008-04-01
1103 L04AX03 2 2008-10-20
1103 D05AX52 1 2008-10-20
1103 L04AX03 2 2009-04-07
1103 L04AX03 2 2009-11-20
1103 L04AX03 2 2010-11-12
1103 L04AX03 2 2011-08-12
1103 D05AX52 1 2011-10-22
1103 D05AX52 1 2011-12-29
run;

计算开关变量

proc sort data=have;
  by pnr date;
run;

data want;
  set have;
  by pnr;
  format switch 1. tCode_prev $10. tType_prev 3.;
  retain tCode_prev '' tType_prev .;
  format switch 1.;

  if first.pnr then do;
    switch = 0;
  end;

  else if tType_prev = 1 then do;
    if tType > 1 or tCode ne tCode_prev then switch = 1;
    else switch = 0;
  end;

  else if tType_prev = 2 then do;
    if tType > 2 or (tType = 2 and tCode ne tCode_prev) then switch = 1;
    else switch = 0;
  end;

  else if tType_prev = 3 then do;
    /* Here's your homework */
  end;

  output;    
  tCode_prev = tCode;
  tType_prev = tType;

  drop tCode_prev tType_prev;
run;

确定接受治疗的受试者人数