topic Re: Trying to Count & Produce Multiple Names from Variables in SAS Programming

Trying to Count & Produce Multiple Names from Variables

dellerrrr — Thu, 26 Sep 2019 15:16:54 GMT

Hello everyone,

Currently I am struggling with a problem I cannot seem to resolve. I am trying to see the frequency that different antidepressants are appearing in the dataset along with if there are observations with more than one antidepressant. Every observation will have SubstanceName1-SubstanceName30 and Class1-Class30. Below is some sample data of how it appears in the set.

Sample Data

ID# SubstanceName1 SubstanceClass1 SubstanceName2 SubstanceClass2

1 Trazodone Antidepressant Fluoxetine Antidepressant

2 Heroin Opiate Trazodone Antidepressant

Currently the array code I am using below will only return the name of an antidepressant that appears last if there are multiple. Is there any way to count if there are observations with multiple antidepressants like the ID#1, and if so to return both Substance Names? Thank you for any guidance.

Data depress;
set set_1;

Array AntiD $ SubstanceClass1-SubstanceClass30;
Array Name $ SubstanceName1-SubstanceName30;
do i= 1 to 30;
if AntiD(i) = 'Antidepressant' then Depression = Name(i);
end;
run;

Re: Trying to Count & Produce Multiple Names from Variables

PaigeMiller — Thu, 26 Sep 2019 15:18:05 GMT

It seems like this will work:

Data depress;
    set set_1;
    Array AntiD $ SubstanceClass1-SubstanceClass30;
    Array Name $ SubstanceName1-SubstanceName30;
    count=0;
    do i= 1 to 30;
        if AntiD(i) = 'Antidepressant' then Depression = Name(i);
        count=count+1;
        output;
    end;
run;

Re: Trying to Count & Produce Multiple Names from Variables

dellerrrr — Thu, 26 Sep 2019 15:23:38 GMT

PaigeMiller,

Unfortunately this has Depression return as blank,
& count has simply counted from 1 to 30 down the data and then resets at the 31st observation

Re: Trying to Count & Produce Multiple Names from Variables

PaigeMiller — Thu, 26 Sep 2019 15:27:06 GMT

My mistake, how about this

Data depress;
    set set_1;
    Array AntiD $ SubstanceClass1-SubstanceClass30;
    Array Name $ SubstanceName1-SubstanceName30;
    count=0;
    do i= 1 to 30;
        if AntiD(i) = 'Antidepressant' then do;
            Depression = Name(i);
            count=count+1;
            output;
        end;
    end;
run;

Re: Trying to Count & Produce Multiple Names from Variables

dellerrrr — Thu, 26 Sep 2019 15:31:12 GMT

Paige Miller,

Awesome! Thank you. Do you have any thoughts on how to return those antidepressant names where there are multiple? I am trying to get a true measure of the distribution of them.

Thank you

Re: Trying to Count & Produce Multiple Names from Variables

Reeza — Thu, 26 Sep 2019 16:39:16 GMT

Can you restructure your data? If it's in a long form, this is much easier to answer these types of questions.

Re: Trying to Count & Produce Multiple Names from Variables

PaigeMiller — Thu, 26 Sep 2019 16:42:05 GMT

@dellerrrr wrote:
Paige Miller,

Awesome! Thank you. Do you have any thoughts on how to return those antidepressant names where there are multiple? I am trying to get a true measure of the distribution of them.

Thank you

I don't know what this means. Show me the exact output you want.

Re: Trying to Count & Produce Multiple Names from Variables

dellerrrr — Thu, 26 Sep 2019 17:59:17 GMT

I was looking to return the substance names total. So if there is only one match just one name, and if more than return multiple. I tranposed the data like Reeza suggested and have figured it out.

Thank you!

Re: Trying to Count & Produce Multiple Names from Variables

Reeza — Thu, 26 Sep 2019 18:44:17 GMT

If you're in a long format, SQL will do 'aggregations' on character variables and/or count distinct which can make life a lot easier to figure out the number of distinct drugs a client is using.