Formating output after proc univariate

[ChuksManuel]

Hello programmers

I want to do a Proc Univariate by tc tn so that i can get N Mean Std for BMI and the dichotomized heart variables (heart1-hear6).

I want to present continuous variables like BMI as N Mean Std.

For the heart variables , i want to do something like a/N %, where a is the number of yes's, N is the total number of non-missings and %= a/N *100 = the percentage of yes's.

In the end i want something like this

                                            Study Population

                                               a/N (%)

Angina                                  xxx/xxx (xx.x%)

Heartburn                             xxx/xxx (xx.x%)

Sleepiness                            xxx/xxx (xx.x%)

Exercise                               xxx/xxx (xx.x%) 

Palpitation                           xxx/xxx (xx.x%)

Any                                       xxx/xxx (xx.x%)

BMI                                      xx.x+xx.x

I've attached the initial dataset and code for you to understand what i want to implement and please any help, advice  or direction is appreciated. 

data me;
input id sex race agecat health bmi bmigroup heart1 heart2 heart3 heart4 heart5 heart6 death tc Tn $ value;
datalines;
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 1 Angina 0
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 2 Heartburn 1
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 3 Sleepiness 0
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 4 Exercise  0
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 5 Palpitation 1
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 6 Any  1
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 1 Angina 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 2 Heartburn 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 3 Sleepiness 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 4 Exercise 1
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 5 Palpitation 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 6 Any  1
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 1 Angina 0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 2 Heartburn 1
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 3 Sleepiness 0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 4 Exercise  0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 5 Palpitation 0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 6 Any 1
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 1 Angina 0
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 2 Heartburn 0
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 3 Sleepiness 1
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 4 Exercise  1
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 5 Palpitation 0
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 6 Any  1
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 1 Angina 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 2 Heartburn 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 3 Sleepiness 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 4 Exercise  0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 5 Palpitation 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 6 Any  0
; run;
proc print; run;

/*format and stringing out*/
Proc format;
value testcodefm
  1= 'Angina'
  2= 'Heartburn'
  3= 'Sleepiness'
  4= 'Exercise'
  5= 'Palpitation'
  6= 'Any'
  7= 'BMI'
;
run;

data You;
set me;
array him heart1-heart6 BMI;
do over him;
  tc = _I_;
  tn = put(tc,testcodefm.);
  value = him;
  output;
end;
keep; 
run;

proc print data=you;
run;

proc univariate data= you; by tc tn;
var heart1 heart2 heart3 heart4 heart5 heart6 BMI
output out= three n=n mean=mean median=median std=std;
 data four; set three;
 newvar= (mean/N)* 100;
 run;
 proc print data= four; var newvar n;
 run;

[PaigeMiller]

Hi, it's great that you posted your data in a data step, that is the best way to provide data, and we appreciate your effort.

A simple data step can produce the printed report you want. 

data _null_;
     file print;
    set four;
    pct=newvar/n;
     put tc testcodefm. +3 newvar '/' n '(' pct percentn7.1 ')';
run;

[Reeza]

You're missing a PROC SORT and semicolon in PROC UNIVARIATE, in the VAR statement. Fix the errors in the order they appear.

[Reeza]

data me;
input id sex race agecat health bmi bmigroup heart1 heart2 heart3 heart4 heart5 heart6 death tc Tn $ value;
datalines;
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 1 Angina 0
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 2 Heartburn 1
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 3 Sleepiness 0
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 4 Exercise  0
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 5 Palpitation 1
1 0 0 5 4 22.494 2 0 1 0 0 1 1 1 6 Any  1
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 1 Angina 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 2 Heartburn 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 3 Sleepiness 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 4 Exercise 1
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 5 Palpitation 0
2 1 0 5 2 20.1801 1 0 0 0 1 0 1 0 6 Any  1
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 1 Angina 0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 2 Heartburn 1
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 3 Sleepiness 0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 4 Exercise  0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 5 Palpitation 0
3 1 0 4 2 26.3606 4 0 1 0 0 0 1 0 6 Any 1
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 1 Angina 0
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 2 Heartburn 0
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 3 Sleepiness 1
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 4 Exercise  1
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 5 Palpitation 0
4 0 0 4 3 25.373 3 0 0 1 1 0 1 0 6 Any  1
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 1 Angina 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 2 Heartburn 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 3 Sleepiness 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 4 Exercise  0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 5 Palpitation 0
6 0 0 4 1 24.8608 3 0 0 0 0 0 0 0 6 Any  0
; run;
proc print; run;

/*format and stringing out*/
Proc format;
value testcodefm
  1= 'Angina'
  2= 'Heartburn'
  3= 'Sleepiness'
  4= 'Exercise'
  5= 'Palpitation'
  6= 'Any'
  7= 'BMI'
;
run;

data You;
set me;
array him heart1-heart6 BMI;
do over him;
  tc = _I_;
  tn = put(tc,testcodefm.);
  value = him;
  output;
end;
run;

proc sort data=you; by tc tn;
run;

proc means  data= you stackods n mean median std; by tc tn;
var heart1 heart2 heart3 heart4 heart5 heart6 BMI;
ods output summary = three;
run;

 data four; set three;
 newvar= (mean/N)* 100;
 run;
 
proc print data=four;
format _numeric_ 8.3;
run;

Combine this with the answer from @PaigeMiller  to get your formatted results. This works for me.

[ChuksManuel]

It worked for all the dichotomous heart variables except BMI. 

I was hoping that the BMI will would appear as (mean +Std) but it appeared as a percentage instead. 

Is there a way i could rewrite the BMI to be  in the (mean+std) format?

/*format and stringing out*/
Proc format;
value testcodefm
  1= 'Angina'
  2= 'Heartburn'
  3= 'Sleepiness'
  4= 'Exercise'
  5= 'Palpitation'
  6= 'Any'
  7= 'BMI'
;
run;

data You;
set me;
array him heart1-heart6 BMI;
do over him;
  tc = _I_;
  tn = put(tc,testcodefm.);
  value = him;
  output;
end;
run;

proc print data=you;
run;

Proc sort; by tc tn;

Proc univariate ; by tc tn;
Var value; 
 Output out =three n=n mean=p std =std; run;
Proc print;  var tc tn n p  std;
Run; 

data four; set three;
 Count= N*P;
 run;
proc print; run;

data five; set four;
percent= (Count/N)*100;
run;
proc print; run;
data _null_;
     file print;
    set five;
    pct=count/n;
     put tc testcodefm. +3 count '/' n '(' pct percentn7.1 ')';
run;

[Reeza]

Add an IF condition to handle that variable differently with a different put statement.

[PaigeMiller · Accepted solution]

---

@ChuksManuel wrote:

It worked for all the dichotomous heart variables except BMI. 

I was hoping that the BMI will would appear as (mean +Std) but it appeared as a percentage instead. 

Is there a way i could rewrite the BMI to be  in the (mean+std) format?

Sure. I don't have a lot of time at the moment, but it would look something like this (where you can fill in the proper variable names and formats):

data _null_;
     file print;
    set five;
    pct=count/n;
    if tc<=6 then put tc testcodefm. +3 count '/' n '(' pct percentn7.1 ')';
    else if tc=7 then put tc testcodefm. +3 meanvariable '+' stdvariable ;
run; 

[ChuksManuel]

Thanks Paige!

[ChuksManuel]

Hello PaigeMiller,

I have a similar problem i posted in the programming section.

Please can you take a look at it? I am trying to write the code based off this example.

Sincerely,

CHucks