topic Re: Array logic in SAS Procedures

Array logic

robertrao — Sat, 27 Apr 2013 16:34:52 GMT

Hi,

I was learning about different things in Arrays.

I have a couple of questions in the below code:

1)what is the use of vint=.; vext=. for the variables being created;

2)why did we write this

if nmiss(vint, vext)< 2 then output vent_array;

3)what happens if we dont write the above step?????/

Thanks

data vent_array;

format vint vext datetime.;
set have(keep=visit_id mrn1 admission_date discharged

itb1-itb12 ext1-ext12);
by mrn1;
array intubate{12} itb1-itb12;
array extubate{12} ext1-ext12;
do i=1 to 12;

vint=.; vext=.;
if admission_date <= intubate{i} <= discharged then vint=intubate(i);
if admission_date <= extubate{i} <= discharged then vext=extubate(i);
if nmiss(vint, vext)< 2 then output vent_array;

Re: Array logic

PGStats — Sat, 27 Apr 2013 17:18:56 GMT

Note: The code is missing at least end; run; at the end.

1) vint=.; vext=.; resets the default value for these variables as missing instead of the value they had on the previous iteration of the do loop.. If the tests that follow are not conclusive, the variables vint or vext will be missing.

2) if nmiss(vint, vext)< 2 The function nmiss returns the number of missing values among its arguments. Here the test condition it will return TRUE if either vint or vext is missing.

3)This datastep creates a new dataset named vent_array that lists the cases where the intb or ext dates are not valid, i.e. they do not occur between the admission and discharge dates.

Re: Array logic

robertrao — Sat, 27 Apr 2013 17:52:51 GMT

Thanks PG,

Then I think I messed up with the code...

i was thinking in the below code vent_array dataset contains the cases where itb or extb occurs between admission and discharge dates..(i see that in the dataset there are some observation which have missing either itb or extb.But majority of them have both the values and are with in the range of admission and discharged....please explain

and holding dataset contains the patients whose itb or extb are past the discharged date

data vent_array(keep=visit_id mrn1 admission_date discharged vint vext)
intext.HOLDING(keep= mrn1 vint vext admission_date discharged);
/*purposely removed visit_id from Holding to make it compatible for appending to fresh data eaCH MONTH*/
format vint vext datetime.;
set admitsum_vent(keep=visit_id mrn1 admission_date discharged

itb1-itb12 ext1-ext12);
by mrn1;
array intubate{12} itb1-itb12;
array extubate{12} ext1-ext12;
do i=1 to 12;
vint=.; vext=.;
if datepart(admission_date) <= datepart(intubate{i}) <= datepart(discharged) then vint=intubate{i};
if datepart(admission_date) <= datepart(extubate{i}) <= datepart(discharged) then vext=extubate{i};
if nmiss(vint, vext)< 2 then output vent_array;
if last.mrn1 then do;
do i=1 to 12;
if datepart(intubate{i}) >= datepart(discharged) or datepart(extubate{i}) >= datepart(discharged) then do;
/*guys with int/ext past the discharge date are exported to holding dataset*/

vint=intubate{i};
vext=extubate{i};
output intext.HOLDING;
end;
end;
end;
end;
run;

Re: Array logic

robertrao — Sat, 27 Apr 2013 18:04:28 GMT

Cleaarly I see that after I used the code.

the vent_array dataset has observations for which there is non missing vint and vext as well

like shown and also either of them missing

vint vext

value value

missing value

value missing

And these are all within the admissin discharge date ranges...

EVEN THE ONES WITH :

missing value

THE VALUE IS WITHIN THE DATE RANGE

So How can I get to kNOW THE OBSERVATIONS WHOSE VINT AND VEXT ARE NOT WITHIN THE VALID RANGE OF INTUBATE AND EXTUBATES SEPERATELY????????/

aLSO WHEN WE SAID

IF nmiss(vint, vext)< 2 then output vent_array;

WHY IS IT STILL OUTPUTTING OBSERVATIONS WHERE BOTH VALUES ARE NON-MISSING

VALUE VALUE AS SHOWN ABOVE

Re: Array logic

PGStats — Sat, 27 Apr 2013 18:37:06 GMT

Why not simply use :

data

vent_array (keep=visit_id mrn1 admission_date discharged vint vext)

intext.HOLDING (keep= mrn1 admission_date discharged vint vext);

format vint vext datetime.;

set admitsum_vent;

array intubate{12} itb1-itb12;

array extubate{12} ext1-ext12;

do i=1 to 12;

vint=intubate{i};

vext=extubate{i};

if admission_date <= vint <= discharged and

admission_date <= vext <= discharged then output vent_array;

if vint > discharged or vext > discharged then output intext.HOLDING;

end;

run;

Re: Array logic

robertrao — Sat, 27 Apr 2013 19:49:39 GMT

HI,

thanks for your help. The last. Concept is necessary because there are repeats for the mrn1 and it is only for the record encounter we want to check if the patient intubate or extubate is greater than discharge date because by doing that we will know that the patient is not yet discharged completely and we want to hold that records for future analysis agagain

thanks

Re: Array logic

robertrao — Sat, 27 Apr 2013 20:54:06 GMT

Hi,

I am stuck in this logic.Could someone help me figure this out?

As PG explained

if nmiss(vint, vext)< 2 then output vent_array;means

if either of them is missing then output to vent_array

But in my case I am getting observations in the vent_array where both of them are non missing as well as either missing

How am I able get to this results...

i wanted those observations which satify those conditions to be in that vent_array

But that highlighted statement usage should not give what I wanted .....But to my surprise

Thanks

Re: Array logic

PGStats — Sat, 27 Apr 2013 22:13:43 GMT

This then?

data

vent_array (keep = visit_id mrn1 admission_date discharged vint vext)

intext.HOLDING (keep = mrn1 admission_date discharged vint vext);

format vint vext datetime.;

set admitsum_vent; by mrn1;

array intubate{12} itb1-itb12;

array extubate{12} ext1-ext12;

do i = 1 to 12;

vint=intubate{i};

vext=extubate{i};

if (admission_date <= vint <= discharged) and

(admission_date <= vext <= discharged) then output vent_array;

if last.mrn1 and (vint > discharged or vext > discharged) then output intext.HOLDING;

end;

run;

Re: Array logic

Tom — Sat, 27 Apr 2013 23:18:00 GMT

I am confused. Your first restatement of the meaning is wrong, but the second is right.

Perhaps the issue you are seeing is because of the loop over 12 elements of the array?

Because of that you will see values of <visit_id mrn1 admission_date discharged> in VENT_ARRAY when ANY of the 12 pairs have one or zero missing values. The only source records that will not send any records to VENT_ARRAY will be those where ALL of the 24 original variables are missing. Adding the index variable I to the keep list for VENT_ARRAY should help you see this.