Hi Folks:   I'm trying to assign ID to identify unique individual patients to the best of the information in hand. So far, data rows matching on the personal variables such as: birth year, birth month, sex, race, ethnicity and 5-digit ZIP take same ID. However, I noticed that some records had hospital admissions after 'death' registered. My question is how to account for 'death' in the process of assigning ID to the rows? Right now, i.e., it's not logical that ID=1 patient has hospital admission (DIS_Y) in 2004 after death (DISP=20) registered in 1997. After accounting for the death-logic, only first two rows would take the same ID. Same for ID=2 as his/her death happened (DISP=20) in 1997 this record takes an independent ID from the following two rows as shown in NEW_ID.  Thanks for your time in advance.   Please see below is the short mock data. More complete data can be found as an attachment to the post.  DATA BE_LOGICAL; input ID count birth_y birth_m SEX $ ZIP RACE ETHNIC DISP dis_y NEW_ID; cards; 1 3 1909 04 F 10002 02 1 06 1996 1 1 3 1909 04 F 10002 02 1 20 1997 1 1 3 1909 04 F 10002 02 1 06 2004 2 2 3 1916 11 F 10458 02 2 20 1997 3 2 3 1916 11 F 10458 02 2 01 2005 4 2 3 1916 11 F 10458 02 2 01 2006 4 3 4 1917 04 F 14450 01 2 20 1996 5 3 4 1917 04 F 14450 01 2 03 1999 6 3 4 1917 04 F 14450 01 2 01 2007 6 3 4 1917 04 F 14450 01 2 03 2008 6 4 3 1917 12 F 10025 02 2 05 1999 7 4 3 1917 12 F 10025 02 2 06 2002 7 4 3 1917 12 F 10025 02 2 20 2003 7 5 3 1919 09 F 11561 01 2 06 1998 8 5 3 1919 09 F 11561 01 2 06 1999 8 5 3 1919 09 F 11561 01 2 20 2006 8 6 4 1920 02 F 10023 01 2 01 1999 9 6 4 1920 02 F 10023 01 2 06 2000 9 6 4 1920 02 F 10023 01 2 20 2003 9 6 4 1920 02 F 10023 01 2 06 2006 10 ; proc print; run; /*DISP=20 indicates death event, DIS_Y=discharge year*/   ... View more

Hi Folks: I my data is stored by calendar years and I concatenated as shown in the code below. I'm curious if there was more efficient way to concatenate the datasets? I have more than 11 variables to read in actuality and I have to add variables onto all the lists every time I need I decide to add more variables into the analyses.  data c.concatenated(COMPRESS=YES); set c.data95(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data96(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data97(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data98(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data99(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data00(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data01(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data02(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data03(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data04(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data05(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data06(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data07(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data08(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data09(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data10(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data11(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data12(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11) c.data13(keep=var1 var2 var3 var4 var5 var6 var7 var8 var9 var10 var11); run;   ... View more

Hi Folks:       I have an unidentified data (no data on how many times each patient was hospitalized). Therefore, I'd like to understand the extent of patients/data rows that share the same birth_y, birth_m, sex, zip, discharge_y and discharge_m. Given my research question is concerned with a rapidly fatal (max survival time ~ 6month) rare disease (~8 per 100,000 people), one could relatively safely assume that it's unlikely that the two individuals diagnosed with this rare medical condition is to occur to have the same birth_y, birth_m, sex, zip, discharge_y and discharge_m. This gives me a hope that I could create a synthetic unique individual identifier based on these variables. I'm aware of proc sort nodupkey by listing these variables to de-duplicate the data. But I have to assess the reliability of this assumption before I get to the point of de-duplication.      Do you know how to create unique identifier based on the multiple variables? A patient hospitalized twice a year could take different discharge_y and discharge_m. But this could be solved later based on this initial screening.   Thanks for your time in advance. See mock data below, if that helps.   data have; input birth_y birth_m sex zip discharge_year discharge_month; cards; 1980 2 1 12202 1991 3 1982 2 1 12202 1991 3 1970 6 2 12307 1971 8 1965 7 2 12907 1968 9 ;   ... View more

HI Folks: I'm trying to identify cases based on the array which consists of 25 variables initialized with DX in my data. The variable IDENTIFIED is the correct answer in the HAVE data where the IDENTIFIED variable correctly captured value of '157' in any variable of the array. In other words, all DX01-DX03 are diagnoses and any of this series of variables can indicate the diagnosis (157 by ICD-9) that I'm looking up to identify.    Thanks for your time in advance.   data HAVE; input DX01 $ DX02 $ DX03 $ IDENTIFIED; cards; 157 157 200 1 157 200 100 1 100 157 100 1 ; data WANT; set HAVE; array DXIN [3] DX01-DX03; do I = 1 to dim(DXIN); IDENTIFY=DXIN[I]='157'; end; drop I; run;   ... View more

Hi Folks:   I'd like to substring the first 3 digits of multiple variables all initialized with dx. In real dataset, I have dx01-25. My code posted below doesn't work.  Could you please help correct the existing code or suggest alternative approach to substring all dx01-dx25 without substr each variable individually?  Thanks for your time in advance.  data have; input dx01 $ dx02 $ dx03 $ case; cards; 15701 1576 2007 1 15701 2006 1007 1 10001 1576 1007 1 ; data want; set have; array diag[*] dx01-dx03; do i = 1 to dim(diag); dx=diag[i]; DX_substr=substr(left(dx),1,3); end; drop i; run;   ... View more