How to calculate relative risk, and 95%CI

[Bal23]

Thank you. I have spent some time to read it. I understand the theory, but it lacks sample code. It is a little hard for me to put everything together.

Would it be possible  for you to put my code and your suggested line to add together as a sample code for me to work on?

[Reeza]

There's a fully worked example, with fake data I made and code that generates resulsts as indicated. I'm not sure what more sample code could be provided.

[Bal23]

*Create sample data;
data have;
input mygroup $ subgroup $ b d;
cards;
se A 371130 11243
se N 214144 2214
se M 181841 5820
se F 168065 3288
ex M 779199 17987
ex F 155981 4578
ge s20 603224 14766
ge S25 263337 5918
ge o25 68607 1881
ac W 689531 17304
ac B 159304 3714
ac Ot 86345 1547
ed BH 1766 76
ed HS 806844 19491
ed Co 74217 2014
ed B 52171 979
q 99 66736 1373
q 92 376550 8726
q 64 259887 6356
q 49 221302 6009
q 29 2635 83
s fq 811106 19056
s tq 37385 1178
s pq 86689 2331
;

run;

data want;
set have;
by mygroup;
*tell sas to keep the denom across the rows;
retain denom;

*create the denominator value;
if _n_=1 then do;
rr=1;
denom=d/b;
end;

*calculate relative risk;
else rr=(d/b)/denom;

*calculate percent and 95%CI;
pct=d/b;
J=SQRT((((B-D)/D)/B)+(((B-D)/D)/B));
Lcl=EXP(log(pct)-(1.96*(J)));
Ucl=EXP(log(pct)-(1.96*(J)));
/*proc sort; by mygroup;*/

*format variables for appearance;
format pct percent8.1 rr 8.2;

run;

*print results;
proc print data=want;
run;

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

Error message:

ERROR: BY variables are not properly sorted on data set WORK.HAVE.

I did add the by statment for the second time, please see the line / / , but the error message still exists

[Reeza]

Add the NOTSORTED option to your BY statement. However, if your data is not actually sorted correctly, this will provide incorrect answers.

by mygroup NOTSORTED;

Also, look back at my sample code and the first IF, does it match yours?

[Bal23]

Thank you. That is my problem i did not use your code if first.mygroup

after I fixed it, notsorted option seems not necessary.

I need one more option, for each first one, with RR=1, at any subgroup, 95%CI needed to be calculated, that is to leave it as blank,

would you please let me know how to let sas do it?

Thank you again

[Reeza · Accepted solution]

If first.mygroup identifies that record.  Try adding the condition in an IF statement, make sure it goes in the appropriate location (ie after the calculation). 

Give it a try and if you have difficulties post the error with your code. 

Otherwise I think your original question is now answered and please mark this question answered.

[Bal23]

sorry  I do not understand your first sentence; therefore, I do not know how to do

[Bal23]

I meant, the first one, with RR=1, no need to calculate 95%CI. Would you please let me know how to let SAS do this,, put  a marker as "-" for the first one of each subgroup with RR=1?