topic Re: how to make combine different observations in relation to a variable to make a new combo variabl in SAS Programming

how to make combine different observations in relation to a variable to make a new combo variable

mcalde33 — Tue, 25 Oct 2022 18:22:20 GMT

I need to make table A into Table B:

Table A

Patient_ID	drug_ID	drug

9900001	1	vinorelbine
9900001	2	carboplatin
9900001	2	paclitaxel
9900003	1	fluorouracil
9900005	1	5-FU
9900005	1	leucovorin
9900005	2	fluorouracil
9900005	3	fluorouracil
9900005	3	leucovorin
9900008	1	fluorouracil
9900008	2	fluorouracil
9900008	2	leucovorin
9900008	3	fluorouracil
9900008	3	irinotecan
9900008	3	leucovorin

Table B

Patient_ID	drug_ID	drug_combo

9900001	1	vinorelbine
9900001	2	carboplatin paclitaxel
9900003	1	fluorouracil
9900005	1	5-FU leucovorin
9900005	2	fluorouracil
9900005	3	fluorouracil leucovorin
9900008	1	fluorouracil
9900008	2	fluorouracil leucovorin
9900008	3	fluorouracil irinotecan leucovorin

Re: how to make combine different observations in relation to a variable to make a new combo variabl

PeterClemmensen — Tue, 25 Oct 2022 18:30:50 GMT

data have;
input Patient_ID $ drug_ID drug :$20.;
datalines;
9900001 1 vinorelbine  
9900001 2 carboplatin  
9900001 2 paclitaxel   
9900003 1 fluorouracil 
9900005 1 5-FU         
9900005 1 leucovorin   
9900005 2 fluorouracil 
9900005 3 fluorouracil 
9900005 3 leucovorin   
9900008 1 fluorouracil 
9900008 2 fluorouracil 
9900008 2 leucovorin   
9900008 3 fluorouracil 
9900008 3 irinotecan   
9900008 3 leucovorin   
;

data want(drop = drug);
   do until (last.drug_ID);
      set have;
      by Patient_ID drug_ID;
      length drug_combo $200;
      drug_combo = catx(' ', drug_combo, drug);
   end;
run;

Result:

Patient_ID  drug_ID  drug_combo
9900001     1        vinorelbine
9900001     2        carboplatin paclitaxel
9900003     1        fluorouracil
9900005     1        5-FU leucovorin
9900005     2        fluorouracil
9900005     3        fluorouracil leucovorin
9900008     1        fluorouracil
9900008     2        fluorouracil leucovorin
9900008     3        fluorouracil irinotecan leucovorin

Re: how to make combine different observations in relation to a variable to make a new combo variabl

ballardw — Tue, 25 Oct 2022 20:01:05 GMT

What exactly is superior about Table B?

Do you expect the result to always be fluorouracil leucovorin, to pick one combination, when the order in the original data might result in leucovorin fluorouracil from different sequences of data?

Do you have any idea what the maximum number of items might need to be combined? This goes to setting the maximum length of the target variable.

Suggestion: If you must combine such place a character between values that will not appear in your values so that it is possible to separate them back to the originals later when it is needed (and likely will be needed). Very easy to do with @PeterClemmensen's code. Place a comma, for example in the CATX function call: Catx(',' ,<other parameters>) for example places a comma between all the values.

Re: how to make combine different observations in relation to a variable to make a new combo variabl

Reeza — Tue, 25 Oct 2022 20:22:01 GMT

With your current approach order matters.

carboplatin paclitaxel will not be the same as paclitaxel carboplatin and because there's no delimiter it'll be harder to separate out drugs that have two names.

If this matters sort your data before you do the combination.

What are you trying to achieve that makes you think this is the correct format?

Re: how to make combine different observations in relation to a variable to make a new combo variabl

mcalde33 — Tue, 25 Oct 2022 21:01:35 GMT

This didnt work for me, and why was 200 used as the length of drug combo?

Re: how to make combine different observations in relation to a variable to make a new combo variabl

mcalde33 — Tue, 25 Oct 2022 21:09:39 GMT

9900001 1 n
2 9900001 2 l
3 9900001 2 i
4 9900003 1 c
5 9900005 3 c
6 9900005 2 c
7 9900005 1 9
8 9900005 3 n
9 9900008 3 n
10 9900008 3 n
11 9900008 3 c
12 9900008 2 n
13 9900008 2 c
14 9900008 1 c

Re: how to make combine different observations in relation to a variable to make a new combo variabl

mcalde33 — Tue, 25 Oct 2022 21:21:14 GMT

The employer wants it that way, im not sure.

Re: how to make combine different observations in relation to a variable to make a new combo variabl

Reeza — Tue, 25 Oct 2022 21:51:26 GMT

The two main method are outlined here:

https://gist.github.com/statgeek/d583cfa992bf56da51d435165b07e96a

The length is set to a large number to ensure it can hold the combined values at the end. If we don't know the data, we usually guess an arbitrarily large number. Feel free to customize it to your requirements.

If a solution doesn't work, you need to explain how in details.

Is there an error in the log? It could be anything from a missing semicolon to an incorrect solution.
Saying it doesn't work doesn't convey any information.

Re: how to make combine different observations in relation to a variable to make a new combo variabl

mcalde33 — Tue, 25 Oct 2022 22:15:05 GMT

it has already been mentioned. I sent the output I was getting, that way I conveyed the output to explain it.

Re: how to make combine different observations in relation to a variable to make a new combo variabl

Reeza — Tue, 25 Oct 2022 22:19:50 GMT

@mcalde33 wrote:
it has already been mentioned. I sent the output I was getting, that way I conveyed the output to explain it.

Do you mean this post?

Did it get cut off? Because I'm not seeing any details or explanations of what went wrong or how just a random series of data that looks nothing like your original posted data and therefore have no idea what it means.

9900001 1 n
2 9900001 2 l
3 9900001 2 i
4 9900003 1 c
5 9900005 3 c
6 9900005 2 c
7 9900005 1 9
8 9900005 3 n
9 9900008 3 n
10 9900008 3 n
11 9900008 3 c
12 9900008 2 n
13 9900008 2 c
14 9900008 1 c

Re: how to make combine different observations in relation to a variable to make a new combo variabl

mcalde33 — Tue, 25 Oct 2022 22:25:06 GMT

This is the code used:
DATA drugs;
Input Patient_ID Drug_ID Drug :$20;

Datalines;

9900001 1 vinorelbine
9900001 2 carboplatin
9900001 2 paclitaxel
9900003 1 fluorouracil
9900005 1 5-FU
9900005 1 leucovorin
9900005 2 fluorouracil
9900005 3 fluorouracil
9900005 3 leucovorin
9900008 1 fluorouracil
9900008 2 fluorouracil
9900008 2 leucovorin
9900008 3 fluorouracil
9900008 3 irinotecan
9900008 3 leucovorin
;

RUN;

Data Drug_combo (DROP=drug);
DO until (last.drug_ID);
SET drugs;
BY Patient_ID Drug_ID;
LENGTH drug_combo $200;
drug_combo= catx('', drug_combo, drug);
end;
RUN;

My log:

1 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
68
69 DATA drugs;
70 Input Patient_ID Drug_ID Drug :$20;
71
72 Datalines;

NOTE: SAS went to a new line when INPUT statement reached past the end of a line.
NOTE: The data set WORK.DRUGS has 14 observations and 3 variables.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
user cpu time 0.00 seconds
system cpu time 0.00 seconds
memory 669.78k
OS Memory 23976.00k
Timestamp 10/25/2022 10:23:33 PM
Step Count 24 Switch Count 2
Page Faults 0
Page Reclaims 167
Page Swaps 0
Voluntary Context Switches 13
Involuntary Context Switches 0
Block Input Operations 0
Block Output Operations 264

89 ;
90
91 RUN;
92
93 Data Drug_combo (DROP=drug);
94 DO until (last.drug_ID);
95 SET drugs;
96 BY Patient_ID Drug_ID;
97 LENGTH drug_combo $200;
98 drug_combo= catx('', drug_combo, drug);
99 end;
100 RUN;

NOTE: There were 14 observations read from the data set WORK.DRUGS.
NOTE: The data set WORK.DRUG_COMBO has 9 observations and 3 variables.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
user cpu time 0.01 seconds
system cpu time 0.00 seconds
memory 1070.75k
OS Memory 24236.00k
Timestamp 10/25/2022 10:23:33 PM
Step Count 25 Switch Count 2
Page Faults 0
Page Reclaims 210
Page Swaps 0
Voluntary Context Switches 13
Involuntary Context Switches 0
Block Input Operations 0
Block Output Operations 264

101
102
103
104
105
106 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
116

My output:

Obs Patient_ID Drug_ID Drug
1 9900008 3 c
2 9900008 2 c
3 9900008 2 n
4 9900008 1 c
5 9900008 3 n
6 9900008 3 n
7 9900005 3 c
8 9900005 2 c
9 9900005 1 9
10 9900005 3 n
11 9900003 1 c
12 9900001 2 l
13 9900001 1 n
14 9900001 2 i

Re: how to make combine different observations in relation to a variable to make a new combo variabl

Reeza — Tue, 25 Oct 2022 22:39:21 GMT

The first step is incorrect. Note the log:

NOTE: SAS went to a new line when INPUT statement reached past the end of a line.
NOTE: The data set WORK.DRUGS has 14 observations and 3 variables.
NOTE: DATA statement used (Total process time):

If you print the input data set, the data is truncated from when it's imported, leading to the problem you're seeing.

The solution is actually fine assuming you'd be applying to actual data not just test data.

DATA drugs;
length drug $ 20;
Input Patient_ID Drug_ID Drug;

Datalines;

9900001 1 vinorelbine
9900001 2 carboplatin
9900001 2 paclitaxel
9900003 1 fluorouracil
9900005 1 5-FU
9900005 1 leucovorin
9900005 2 fluorouracil
9900005 3 fluorouracil
9900005 3 leucovorin
9900008 1 fluorouracil
9900008 2 fluorouracil
9900008 2 leucovorin
9900008 3 fluorouracil
9900008 3 irinotecan
9900008 3 leucovorin
;

RUN;



Data Drug_combo (DROP=drug);
DO until (last.drug_ID);
SET drugs;
BY Patient_ID Drug_ID;
LENGTH drug_combo $200;
drug_combo= catx('', drug_combo, drug);
end;
RUN;