Picking values from an array of variables

[UcheOkoro]

Hello,

Please, I need help writing the codes to pick out values from an array of variables. So, the variables thirty_pct1-thirty_pct15 are variables showing if the patient had a 30% decrease in lab values from baseline, so 1 if 'yes' and 0 if 'no'; lab_date1-lab_date5 are the corresponding dates when the lab tests were done and pct_chg1-pct_chg5 is the percent change from baseline. I want to create columns showing the first increase and next decrease with the corresponding lab_dates and pct_chg. The patient with ID 223 has two decreases and increases so there will be two columns for date of decrease and two for dates of increase.

Thank you!

data WORK.CLASS(label='Student Data');
 input ID$3. thirty_pct1 thirty_pct2 thirty_pct3 thirty_pct4 thirty_pct5 lab_date1 lab_date2 lab_date3 lab_date4 lab_date5 pct_chg1 pct_chg2 pct_chg3 pct_chg4 pct_chg5;
 datalines;
222 0 1 1 0 1 2JAN2022 1FEB2022 4MAR2022 5APR2022 6MAY2022 -24 -31 -35 -19 -32
223 . 1 0 1 0 . 1FEB2022 4MAR2022 5APR2022 6MAY2022 . -31 -28 -33 -25
224 0 . 1 0 0 2JAN2022 . 4MAR2022 5APR2022 6MAY2022 -24 . -35 -19 -32
 ;run;

[ballardw]

Now show what you expect the output to look like. I am not sure how many variables you actually expect to add. Is it one per increase/decrease (means maximum of 4 variables for increase and 4 for possible decreases) or 

[UcheOkoro]

Thank you for your response. I expect the output to look like this:

ID lab_date_of_decrease1  lab_date_of_decrease2  lab_date _of_increase1 labdate _of_increase2 percent_chg1  percent_chg2. The percentage would be for the decrease if there was a decrease or the last increase if there was no decrease.

[ballardw]

What if there are more than 2 increases or decreases? Since what I understand of your data is that the values are against a not-shown value at a some other date it seems like you might possibly have more than 2 possible

[UcheOkoro]

There could be more than 2 or 3 increases and respective decreases the data. The real data has observation with more than 3 increases with respective decreases.

[Tom]

If you want the FIRST how can there be more than one?

Why is your data structured like this?  Why not just make a more normal dataset?

data have ;
  input ID :$3. thirty_pct1-thirty_pct5 (lab_date1-lab_date5) (:date.) pct_chg1-pct_chg5;
  format lab_date1-lab_date5 date9.;
datalines;
222 0 1 1 0 1 2JAN2022 1FEB2022 4MAR2022 5APR2022 6MAY2022 -24 -31 -35 -19 -32
223 . 1 0 1 0 . 1FEB2022 4MAR2022 5APR2022 6MAY2022 . -31 -28 -33 -25
224 0 . 1 0 0 2JAN2022 . 4MAR2022 5APR2022 6MAY2022 -24 . -35 -19 -32
;

data tall ;
  set have;
  array c thirty_pct1-thirty_pct5;
  array d lab_date1-lab_date5;
  array p pct_chg1-pct_chg5;
  do index=1 to dim(d);
    p30 = c[index];
    lab_date = d[index];
    pct_chg = p[index];
    output;
  end;
  format lab_date date9.;
  drop thirty_pct1-thirty_pct5 lab_date1-lab_date5 pct_chg1-pct_chg5;
run;

So you have data like this instead:

Now which of those individual records are the ones you want to keep for your new dataset?

Is it something like:

data want;
  set tall;
  where abs(pct_chg) > 30 ;
run;

[UcheOkoro]

Thank you for your response, Tom. The record I would like to keep are highlighted in yellow. The pattern should be 101010 or 101 or 10 or 1010. It must begin with 1.