26.5.08

Understanding the %pageXof Y macro

Compiled by Rupesh R

The following macro needs to be applied for the purpose of presenting page numbers in Page X of Y format within the body of the document.

When calling the macro, the proc report code should be quoted by %nrstr as explained in step 1.a below. To avoid clutter of code, here the proc report code is created as a macro. This proc report code should contain a ‘compute before_page_’ block as described in the step 5 below.

%macro pageXofY (report= /* proc report code, quoted by %nrstr */ /*1.a*/
, dummy=dummy /* name of the dummy output file */
);
%global page pages len; /* 1.b*/


/-- first run --*/
%let page = 0;
%let len = 8; /* 1.c*/
filename _dummy &dummy.; /* 1.d*/
proc printto print = _dummy; run; /*1.e*/
%unquote (&report.) ;/* 1.f*/
proc printto; run;/*1.g*/
filename _dummy clear; /*1.h*/

%*-- second run --*;
%let pages = &page.; /*1.i*/
%let page = 0;/*1.j*/
%let len = %eval(%length(&pages.) * 2 + 4); /* 1.k*/
%unquote(&report.);/*1.l*/
%mend pageXofY;

Explanation of Macro
Step 1 (1.a): Using %nrstr function we mask the special characters and mnemonics .Here we mask & and % symbols in the proc report code.

Step 2 (1.b): Initialize three global macro variables
§ The variable ‘page’ returns the current page number
§ The variable ‘pages’ returns the total number of pages
§ The variable ‘len’ returns the expected length of the string ‘_XofY’ (2.c) .

Step 3 (1.c): The initial values of macro variables ‘page’ and ‘len’ are assigned as 0 and 8 respectively.

Step 4 (1.d): A dummy file path is defined for the output of proc printto

Step 5 (1.e): PRINTTO procedure is used for printing the output in the specified dummy file (_dummy).

Step 6 (1.f): The proc report code created for generating the output is executed using %unquote function in this step.

%UNQUOTE is to restore normal tokenization of a value whose tokenization was altered by a previous macro quoting function. %UNQUOTE takes effect during macro execution. If the value is not unmasked before it reaches the SAS compiler, the DATA step does not compile correctly and it produces error messages.

1) Execute the entire report procedure and generate an output to the specified file path “&dummy.”

2) In report procedure we should include the following set of statements.

compute before/after _page_;
call execute('%let page = %eval(&page. + 1);'); /* 2.a*/
length _XofY $&len.; /* 2.b. */
_XofY = symget('page') ' of ' symget('pages');/*2.c*/
line 'page ' _XofY $&len..; /* 2.d. */
endcomp;
3) Before/After each page of reporting the variable ‘page’ is calculated and so when we exit from this procedure the macro variable ‘page’ has the value of total number of pages. (i.e. The initial value of ‘page’ is ‘0’ when it comes to the first page of file then it becomes ‘1’ and so on, see step (2.a))
4) We get the total pages of report in variable ‘page’.Here the variable “pages” is not initialized. So symget (‘pages’) will return nothing
Step 7 (1.g): Exit the printto procedure and SAS reset the file path to the output window (by default).

Step 8 (1.h): Clear the dummy file.

Step 9 (1.i): Assign the current value of variable ‘page’ to the macro variable ‘pages’. i.e. we assign the total number of pages to the variable ‘pages’

Step 10 (1.j): Reset the variable ‘page’ to ‘0’.

Step 11 (1.k): Calculate the approximate length of the variable ‘_XofY’ based on current value of ‘pages’.

Step 12 (1.l): Re-execute the report procedure for getting the output as per the requirement.
1) Execute the entire report procedure and generate output in the specified path given in the ods rtf file statement.
2) This time the compute block call the ‘page’ value as the current page and ‘pages’ as the total number of pages. So we will get the page number as Page X of Y form.
The following example illustrates the above macro.Here we get page X of Y on the top of each page.

data one;
do var = 1 to 100;
output;
end;
run;

%Macro Report;
proc report data=one nowd;
column var;
define var / display;
compute before _page_;/*2.a*/
call execute('%let page = %eval(&page. + 1);'); /* 2.b*/
length _XofY $&len.; /* 2.c. */
_XofY = symget('page') ' of ' symget('pages');/*2.d*/
line 'page ' _XofY $&len..; /* 2.e. */
endcomp;
run;
%Mend Report;

/* example usage */

options linesize=64 nonumber nodate;
%pageXofY(report=%nrstr(%Report));

16.4.08

Different ODS Footers

If suppose we have ten observations and we are required to present the output such that the observation are distributed on different pages depending on the value of the variable “order”. Also the different pages need to have different footers , footers being presented in the body of the document.As a first step, a variable "page" is created which represents the page number corresponding to the observation in the dataset as described below.

data catval1;
set catval end=eof;
if order le 4 then page=1;
else if 5 le order le 7 then page=2;
else if order ge 8 then page=3;
run;

The footers can be presented in the body of the document using compute before /after statements in proc report as below.

proc report data= catval1 ;
column order var1 var2 ;
………………………………
………………………………
Other SAS statements;
…………………………………
………………………………..
compute before page;
If page=1 then footer= "\li75 @ Subjects who select more than one race” ;
If page=2 then footer=" ";
If page=3 then footer="\li75 + BMI = Weight (kg) / [Height (m)]^2";
endcomp;

/*here _page_ is the SAS generated variable */

compute after _page_ / style=[just=l protectspecialchars=off];
footer11= footer;
footer12="\li75 # Overall p-value for continuous variables “
footer13="\li for Categorical variables from CMH general association test";
line @1 footer11 $300.;
line @1 footer12 $300.;
line @1 footer13 $300.;
endcomp;

/* this will give different footers in each page*/
by page;
run;

The first page will contain the following footers

@ Subjects who select more than one race
# Overall p-value for continuous variables
for Categorical variables from CMH general association test

The second page will contain the following footers

# Overall p-value for continuous variables
for Categorical variables from CMH general association test

The third page will contain the following footers
+ BMI = Weight (kg) / [Height (m)]^2
# Overall p-value for continuous variables
for Categorical variables from CMH general association test

NOTE: Groups are not created because the usage of parameter is DISPLAY.

A SAS dataset named subject is created. The SAS code for the creation of the dataset subject is as follows:
data subject;
input patid name$ sex$ visit $ parameter$ result $;
cards;
0101 lisha f Screening weight 65
0101 lisha f Screening height 150
0101 lisha f Visit1 weight 66
0101 lisha f Visit1 height 150
0102 manu m Screening weight 80
0102 manu m Screening height 165
0102 manu m Visit1 weight 80
0102 manu m Visit2 height 165
;
run;

The aim is to summarize the values across a single observation. The template is as follows:

When using the following code, the output is generated but a Note is also generated in the log as mentioned below:

proc report data=subject nowindows;
column patid name sex visit parameter result;
define patid/"Patient Number" group;
define name/"Subject Number" group;
define sex/"Sex" group;
define visit/"Visit" group;
define parameter/"Test" display;
define result/"Values" display;
run;

The SAS log will look like as follows::
1
2
3 data subject;
4 input patid name$ sex$ visit $ parameter$ result $;
5 cards;
NOTE: The data set WORK.SUBJECT has 8 observations and 6 variables.
NOTE: DATA statement used:
real time 0.01 seconds
cpu time 0.00 seconds
14 ;
15 run;
16
17 proc report data=subject nowindows;
18 column patid name sex visit parameter result;
19 define patid/"Patient Number" group;
20 define name/"Subject Number" group;
21 define sex/"Sex" group;
22 define visit/"Visit" group;
23 define parameter/"Test" display;
24 define result/"Values" display;
25 run;
NOTE: Groups are not created because the usage of parameter is DISPLAY.
NOTE: There were 8 observations read from the data set WORK.SUBJECT.
NOTE: PROCEDURE REPORT used:
real time 0.01 seconds
cpu time 0.01 seconds

A note is produced in the SAS log “NOTE: Groups are not created because the usage of parameter is DISPLAY.”
To remove this note from the log the following code should be used.

proc report data=subject nowindows;
column patid name sex visit parameter result;
define patid/"Patient Number" order;
define name/"Subject Number" order;
define sex/"Sex" order;
define visit/"Visit" order;
define parameter/"Test" display;
define result/"Values" display;
run;

The log is generated as
1
2
3 data subject;
4 input patid name$ sex$ visit $ parameter$ result $;
5 cards;
NOTE: The data set WORK.SUBJECT has 8 observations and 6 variables.
NOTE: DATA statement used:
real time 0.00 seconds
cpu time 0.00 seconds
14 ;
15 run;
16
17 proc report data=subject nowindows;
18 column patid name sex visit parameter result;
19 define patid/"Patient Number" order;
20 define name/"Subject Number" order;
21 define sex/"Sex" order;
22 define visit/"Visit" order;
23 define parameter/"Test" display;
24 define result/"Values" display;
25 run;
NOTE: There were 8 observations read from the data set WORK.SUBJECT.
NOTE: PROCEDURE REPORT used:
real time 0.00 seconds
cpu time 0.00 seconds


The SAS log is now free of the notes, errors and warnings. Instead of using the group variable, use order with display in the define statement.
Applying Cochran-Mantel–Haenzel (CMH) general association test

Compiled by Rupesh R

Here the objective is to determine the general association between categorical variables subject and ranking for each emotion.

The CMH general association test is applied for this purpose. Proc freq is as described below to carry out the test. Proc freq generates p-values corresponding to each emotion and the results are outputted to the dataset hypnosis1.


The dataset is created as follows


data hypnosis;
input subject emotion $ ranking @@;
cards;
1 fear 4 1 joy 3 1 sadness 1 1 calmness 2
2 fear 4 2 joy 2 2 sadness 3 2 calmness 1
3 fear 3 3 joy 2 3 sadness 4 3 calmness 1
4 fear 4 4 joy 1 4 sadness 2 4 calmness 3
5 fear 1 5 joy 4 5 sadness 3 5 calmness 2
6 fear 4 6 joy 3 6 sadness 2 6 calmness 1
7 joy 4 7 joy 1 7 sadness 2 7 calmness 3
8 joy 3 8 joy 4 8 sadness 2 8 calmness 1
;
run;


The following code is used to carry out the analysis. The option ‘cmh’ in the ‘tables’ statement carries out the test. The option ‘cmhga’ presents the P-value for CMH General Association without a warning in the log.


proc sort;
by emotion;
run;



proc freq;
tables subject*ranking / cmh;
by emotion;
output out=hypnosis1 cmhga;
run;


The dataset hypnosis1 is obtained as follows





























emotion_CMHGA_DF_CMHGAP_CMHGA
calmness14140.449711056
fear10100.440493285
joy20.25210.505479124
sadness21210.458944209


In the dataset p_cmhga will be labeled as ’ P-value for CMH General Association’; df_cmhga as “DF for CMH General Association” and _cmhga_ as “CMH General Association” .

15.4.08

Using just=dec for alignment of decimals

Compiled by Prajitha Nair

We have a column in proc report with either 1 or 2 digits right of the decimal and an uncertain number of digits to the left. For enhanced output, the values have to be presented such that they are aligned with respect to the decimal point.

When the display value contains a decimal point, we can use just=dec to align the numbers directly. The JUST=DEC option can be used to align the decimal points in the values in a column

define colname / display "Column/header" style=[just=dec];

where colname is the name of the column name as specified in the SAS.

If we are using ods rtf tags then we can use the pretext option for the purpose

define colname /display “Column Header”

style(column)={pretext="\qj\tqdec\tx1200" protectspecialchars=off just=center};

Extracting Engine and Path name of the code
Compiled by Prajitha Nair




The input parameter in the macro will be the name of the SAS file whose path is to be determined.

%macro PathbyName(progName);

%global fullPath fullPath1 engine;
%if %index(%upcase(&progName),.SAS) eq 0 %then
%let progName=&progName..sas;

proc sql noprint;
select xpath into :fullPath
from dictionary.extfiles where
index(upcase(xpath),"%upcase(&progName) " ) gt 0 ;
select setting into :engine from sashelp.voption
where optname="ENGINE";
quit;

%let engine = %trim(&engine);
%put engine = &engine;
%let fullpath = %trim(&fullPath);

%put fullpath = &fullPath;

%mend PathbyName;

Accessing the current working directory where the file containing the SAS code is stored


Compiled by Prajitha Nair


Let the SAS editor containing the code be KR-PH-XXX-SAS-Init.sas and it is stored in a folder named SAS Programs_Final.


The following %let statements assigns the folder and editor names to macro variables pgmfld and pgm respectively.


%let pgmfld = SAS Programs_Final; /*Folder in which SAS code is stored*/
%let pgm= KR-PH-XXX-SAS-Init.sas; /*Name of the program editor*/


The following macro is then used to extract the path of the editor from sashelp.vextfl and determine the path of the folder as “dir1” and of the editor as “dir2”.


%macro filePath;
%global fpath maxRef;

proc sql noprint;
select xpath into :fPath
from sashelp.vextfl where xpath ? "&pgm";
quit;


%let fpath = %trim(&fpath);
%put &fpath;
%global dir1 dir2 pgm1 pgmfld1;
%let pgm1 =%trim(&pgm);
%let pgmfld1= %trim(&pgmfld);


data _null_;
x=length("&fpath")-length("&pgm1");
y=length("&fpath")-length("&pgm1")-length("&pgmfld1")-1;
call symput("dir2" ,trim(substr("&fpath",1,x)));
call symput("dir1" ,trim(substr("&fpath",1,y)));
run;


%mend filePath;


This macro helps in determining the path of the code and enables the code to be executed in any computer provided the SAS code is saved within a folder and the naming conventions are followed as above.

14.3.08

15.2.08

New Abode

Finally after three years of exemplary hard work of all the present & past employees of KREARA, our dream of a beautiful new office came to frution on January 27th 2008. Thanks to all our clients and employees who have helped us make this dream come true



23.11.07

Randomisation - Data sets

Compiled by Prajitha Nair
Here is a piece of code that will generate mock randomisation list that the programmers could use to generate the outputs before unblinding the data


/*Creating the dummy dataset for randomisation*/

Proc format;
value treat 1='Test'
2='Reference'
3='Placebo';
run;

/*Randomly assigning medical kit number to treatments*/
proc plan seed=111605;
factors medkit_no=220 random block=1 random/noprint;
output out=first;
treatments treat=3 cyclic (1 2 3 );
run;

proc sort data=first out=rand_trt;
by medkit_no;
run;

data rand_trt(keep=medkitno treat);
set rand_trt;
medkitno=put(medkit_no, z3.);
format treat treat.;
run;


/*merge to get the randomised patient included in the study*/
proc sort data=rand_trt;
by medkitno;
run;

proc sort data=medkit_alloc;/*sas dataset imported from mysql*/
by medkitno;
run;

data formo.randl;
merge rand_trt(in=a) medkit_alloc(in=b);
by medkitno;
run;

11.10.07

Read only access to SAS Data sets
Compiled by Soumya Gopinath

Lock the entire SAS data library by using the option Access=Readonly.

Libname libref path access= Readonly;

Silly Proc Report !
Compiled by Soumya Gopinath

An annoying period (.) appeared in the p-value column when the break statement was applied in proc REPORT. How can we eliminate this period (.) from the report (rtf file)?

Solution:
Firstly, the data type of p-value variable is numeric then the break option will replace the missing values with a period. To avoid this situation change the data type of p-value variable to character. Then the missing values will be replaced by blank space only.
Eg:
/* Create dataset with 5 variables and 8 observations*/
data test;
input usubjid $ parameter $ visit $ trtgrp result;
datalines;
001 FEV1 VISIT1 1 2.34
002 FEV1 VISIT1 3 0.98
001 FEV1 VISIT4 1 2.04
002 FEV1 VISIT4 3 1.98
001 FVC VISIT1 1 2.34
002 FVC VISIT1 3 2.98
001 FVC VISIT4 1 2.44
002 FVC VISIT4 3 1.99
;
run;
/* Carrying out ANOVA*/
ods output ModelAnova= test1;
proc glm data=test;
by parameter visit;
class trtgrp;
model result=trtgrp/ss3;
lsmeans trtgrp/adjust=t pdiff;
run;

data test2;
set test1;
test=' '; /* dummy variable*/
keep parameter visit probf Test;
run;

/* Specify the output location*/

ods rtf file="D:\soumya\test.rtf" style=styles.listingstyle;

proc report data = test2 nowd spacing = 2 headline headskip split = '*' missing;
column parameter visit test,(probf);
define parameter / group order=data left 'Parameter' ;
define visit / display left 'Timepoint' ;
define Test / across center "Test" ;
define probf/ display left "p-value";
break after parameter/summarize suppress;
run;

ods _all_ close; /* Closing all ODS outputs statements*/

In the above example probf is a numeric variable within the across variable ‘test’. So in the output a period occurred when the break statement active. Eliminate this we add one more statement in the dataset Test2.
probf1=put (probf, pvalue6.); /* for converting numeric type to character*/

Generally we can say that the numeric variables within the across variable in REPORT procedure should make a period in the blank space and to avoid this convert numeric variables to character by using PUT function.



SAS with mySQL
Compiled by Soumya Gopinath


How to transfer the MySQL Databases named dbformo and dbformo_add to SAS datasets library names formoSQL and AddSQL respectively?

Solution:
libname libname ODBC datasrc= source user= user name password= password schema= database;

By using the procedure COPY we can copy the library formoSQL and AddSQL to our permanent SAS libraries FormoDM and FormoAdd respectively. Before executing the connection string add an ODBC Data source name in the current system, which should be the same as given in datasrc= option. The process,

1) Settings -> Control Panel ->Administrative Tools ->Data Source (ODBC)
2) Click on the Add button.
3) Select the driver for which you want to set up a data source (MySQL ODBC 3.51 Driver) and click Finish button.
4) Add required information in Connector/ODBC window. Click OK button. Then the process is completed.
VB and mySQL
Compiled by Soumya Gopinath

How will you connect a data entry application written in VB to mySQL ?

DRIVER= {MySQL ODBC 3.51 Driver};SERVER=server id; DATABASE=database name; UID=user name; PWD=password";

Wilcoxon Signed Rank test

Compiled by Sreedevi Menon

As part of the efficacy analysis of a recent Phase III study , it was planned to determine the significant difference in efficacy of treatments over the weeks. Each treatment group was analyzed separately. The paired t-test was employed to carry out the analysis on the parametric data.The Wilcoxon Signed Rank test was used for analysis of efficacy of treatments with respect to non-parametric data. The null hypothesis was that in the underlying population of differences among pairs the median difference is equal to 0. The alternate hypothesis may be one sided or two sided.

In SAS, this test is performed using the proc univariate procedure. Before applying the procedure, the difference between the value of parameter at the two time points that are to be compared (usually pre and post dose values) has to be computed. This difference is then defined in the “var” statement of the procedure. The syntax is as followed

ods output BasicMeasures= _A TestsforLocation= _B
proc univariate data= eff_data;
var diff;
run;
ods output close;

The output will contain p-value corresponding to Student’s t, Sign and Signed Rank tests. The p-value corresponding to the signed rank test will be considered. Further, For a 1-sided p-value, we would divide the 2-sided p-value by 2.The ‘ods output statement’ is used to extract descriptive statistics and the p value to two different datasets by using the as shown in the above example

Transpose macro
Compiled by Ajish K Mani

%macro trans(dsn=, outdsn=, vartran=, idtran=, bytran=, copytran=);
proc transpose data=&dsn out=&outdsn let;
var &vartran;
by &bytran;

%if &copytran ne %then %do;
copy ©tran;
%end;

%if &idtran ne %then %do;
id &idtran;
%end;
run;
%mend trans;

data input
input patno 1 sex $ 3 visit 5 weight 7-8 height 10-12;
datalines;
1 m 1 80 150
2 f 1 65 160
3 f 1 70 165
4 f 1 60 170
5 m 1 68 168
;
run;

proc sort data=input out=output;
by patno;
run;


%trans (dsn=output, outdsn=transpose(rename=(col1=Result)), vartran=weight height, bytran=patno);

data transpose;
set transpose;
label _name_ ="Parameter" ;
rename _name_=Parameter;
run;

25.7.07

Asthma Studies

For the analysis of any clinical trial on asthma inhalers, your efficacy parameters are most probably going to be related to the spirometry measures and will almost always include the following
FEV1 - Forced expiratory volume in 1 second
The FEV1 is the volume exhaled during the first second of a forced expiratory maneuver started from the level of total lung capacity.

PEF - Peak expiratory flow
Expiratory peak flow (PEF) is the maximum flow generated during expiration performed with maximal force and started after a full inspiration .

FVC - Forced expiratory vital capacity
The volume change of the lung between a full inspiration to total lung capacity and a maximal expiration to residual volume. Here are more details on Spirometry. Also check out GINA for more Asthma information

Blind Review

First of all the ICH definition - Blind Review is the checking and assessment of data during the period of time between trial completion (the last observation on the last subject) and the breaking of the blind, for the purpose of finalizing the planned analysis.


So how do we do it ? We look at the Clinical Protocol and create a Blind Review Criteria document which will list out all the conditions that needs to be checked against the data to make sure that there is no violations. For eg. if Visit 1 were to happen with in 15-20 days of the screening visit, then we compare the dates of screening and visit1 and if the difference falls below 15 and above 20, then boom we have a violation. Then we write a SAS program to check all those conditions and run it to create a protocol violation document.

Easily said. But to do it and to make sure that all the violations are caught is one hell of a task Johnny.
Hodges-Lehmann ∆

A parametric test, such as the t-test, compares the means of the two samples. A nonparametric method, such as the Wilcoxon Rank Sum Test compares the entire distributions of the two independent samples. The null hypothesis of the Wilcoxon Rank Sum test says the two samples can be viewed as a single sample from one population. The alternative hypothesis is that the first treatment group has a different different distribution (or location) than the second treatment group.

The treatment effect, denoted as ∆, is the difference between treatment groups. If parametric methods were used, means could be calculated for each treatment group, and a subtraction of the means can be used to estimate ∆. However, when the data are not normally distributed and the median value of the response variable of interest is calculated for each treatment group, the estimate of the difference in treatment groups is not as straightforward as subtracting one median from the other.

The difference in medians is estimated using the methodology of Hodges-Lehmann. It is a very simple approach. The following steps can be used to estimate ∆:

• form all possible differences between the first treatment group and the second treatment group, in the response variable of interest. For example, if there are 100 patients in each group then 10,000 (100*100) differences would be calculated.
• the estimator ∆ is the median of those 10,000 differences.

All this taken from this SUPER PAPER.
LOCF

In clinical trials, data are often collected over a period of time from participating patients. In many situations, however, analyses are only based on data from the last time point (the end of the study) or change from the baseline to the last time point. It is often the case that patients drop out before the completion of the study.

So the question arises on how to perform analysis of the last observations, which are defined as observations from the last time point for patients who completed the study and the last observations prior to the dropout for patients who did not complete the study.

An analysis based only on data from patients who completed the study is called a completers analysis. Although a completers analysis is sufficient in some situations, it is often more desirable to perform an all randomized subjects analysis.

The analysis based on all randomized subjects is usually referred to as an intention-to-treat (ITT) analysis. Regulatory agencies generally consider the ITT analysis as the primary analysis for evaluation of efficacy and safety in clinical trials with informative dropout.

When the dropout is informative, the target populations of a completers analysis and an ITT analysis are different. Suppose that the population of patients under certain treatment is stratified according to the time of the last observations; then the target population of a completers analysis is only the subpopulation of patients who completed the study under different treatments while the target populations of an ITT analysis include all the subpopulations under different treatments.

For the last observation carry-forward (LOCF) analysis based on ITT population, the last observations are carried forward to the last time point for patients who dropped out. The LOCF analysis treats the carried-forward data as observed data at the last time point. Therefore when the dropout is informative, the LOCF analysis may introduce biases to the statistical inference, which has been speculated upon by the Food and Drug Administration (FDA) it is still unknown whether or not the LOCF test is asymptotically correct

Here is a macro that might help you with LOCF