test difference between physians

[david-ck]

Hi , there are 20 patients. they are assigned to two physicians(P1 and P2) to evaluate 3 disease (a,b,c) 1=disease,0=no disease

to test the difference for for physicians performance for disease p1-a and p2a, I can use mcnemar test. how can I test overall difference between p1 and p2 by combining the 3 diseases? thank you!

id	p1-a	p1-b	p1-c	P2-a	p2-b	p2-c
1	0	0	0	0	0	0
2	0	0	0	0	0	0
3	0	0	0	1	1	0
4	0	0	0	0	0	0
5	0	0	0	0	0	0
6	0	0	1	1	1	0
7	0	0	0	1	1	1
8	0	0	0	0	0	0
9	0	0	0	1	0	0
10	0	0	0	0	0	0
11	0	0	0	0	0	0
12	0	0	0	1	1	0
13	0	0	0	0	0	0
14	0	0	0	0	0	0
15	0	0	0	1	1	0
16	0	0	0	0	0	0
17	0	0	0	0	0	0
18	1	0	0	0	0	0
19	1	0	0	1	1	0
20	0	0	0	1	0	0

[PaigeMiller]

You'd need to re-arrange your data in such a way that it can be more easily analyzed by SAS PROCs

Something like this:

id       physician        disease         outcome
1            1               A                0
1            1               B                0

Then PROC LOGISTIC and/or PROC FREQ can do the statistical tests easily. In PROC LOGISTIC, your model would contain Physician and disease and the interaction between the two.

[david-ck]

for test difference between p1a and p2a,

i use following codes

proc freq data=p1 order=data;
tables p1a*p2a / agree ;
test kappa;
run;

however, how to test overall difference between (p1a p1b p2c)  and (p2a p2b p2c), thank you!

[Watts]

Yes, like PaigeMiller said, you can rearrange your input data. Then a TABLES request like this

proc freq;
tables Disease * P1_Rating * P2_Rating / agree;

produces an overall kappa coefficient (summing over the 3 diseases) together with agreement statistics for each disease.

[david-ck]

what is the sas code to rearrange the data  above? thank you!

[andreas_lds]

A data step can be used to transform your data.

data transposed;
    set have;
    length 
        physician disease $ 1 outcome 8
        i 8 name $ 32
    ;
    /* variable to be excluded from the dataset transposed */
    drop i name p1_a p1_b p1_c p2_a p2_b p2_c;
    
    /* array allows easier processing of the p-variables */
    array old[6] p1_a p1_b p1_c p2_a p2_b p2_c;
    
    do i = 1 to dim(old);
        /* get the name of the variable, returns p1_b if i = 2 */
        name = vname(old[i]);
        /* extract physician and disease from the variable name */
        physician = substr(name, 2, 1);
        disease =  substr(name, 4, 1);
        /* copy the value to the variabel outcome */
        outcome = old[i];
        output;
    end;
run;

EDIT:

It makes suggesting code much easier if data is presented in usable form. Those tables look nice, but not nice to move to sas-code. This form is usable:

data have;
    input id p1_a p1_b p1_c p2_a p2_b p2_c;
    datalines;
1 0 0 0 0 0 0
2 0 0 0 0 0 0
3 0 0 0 1 1 0
4 0 0 0 0 0 0
5 0 0 0 0 0 0
6 0 0 1 1 1 0
7 0 0 0 1 1 1
;
run;

Please note that i was to lazy to write down all rows you have in the table.