Pre-Programming Processes: a) Can we define all these parameters in ADLB? –Yes!

1. Creating BDS DERIVED Parameters for a Subject-levelFrequency Summary Table?Then this macro can be useful

2. Pre-Programming Processes: • a) Can we define all these parameters in ADLB? –Yes! • b) From this mockup, Baseline flag needs to be defined prior to these parameters – Yes, we need to know baseline flag status. • c) All the subjects in Safety Population must have records of ALT/AST/TBL in the original input dataset – Yes! (for accurate percentage calculation) • d) Can we define all of these parameters in simple macro? – Yes! • Use of Macro makes it possible to automate this type of parameter creation. The Macro runs a SQL statement inside a loop to create each derived parameter. • 1) Use of PROC SQL • 2) SUM function using GROUP BY to summarize records per each qualifying subject. • 3) Conditions that go into SUM function along with PARAMCD/PRAMN/PARAM are appended in each macro key parameter by user-defined delimiter. • Table 1 : This is a mockup table for “Liver Function Test Table” so that SAS programmers get ideas of the table requirement and ADLB Dataset requirement.

3. [ALT or AST ≥ 3xULN and TBL ≥ 2xULN] What conditions do we need to input? • At any post baseline visit (i.e., they don’t need to be the same visits), did a subject experience an AST (Aspartate Aminotransferase) and/or ALT (Alanine Aminotransferase) chemistry panel larger than or equal to three times its upper limit AND was the TBL (Total Bilirubin) chemistry panel larger than or equal to two times its upper limit? • Notice there are two conditions: Condition 1AND Condition 2 • CASE • WHEN SUM (Condition 1)>0 AND SUM(Condition 2)>0 then ‘Y’ • ELSE ‘N’ • END AS AVALC • ~ • GROUP BY USUBJID [SUM functions will summarize the conditions per USUBJID] • Let’s input conditions along with other parameters like PARAMCD/PARA M/PARAMN into the macro

4. † The macro runs each PROC SQL statement in an implicit %do loop defined by 1 to the number obtained from counting function, countw in the sysfunc macro function which turns into the macro variable. For each individual iteration a unique set of values are called in from the macro parameters. The values from iterative parameters are distinguished by a delimiter (&delim), which is carefully selected by users so as to avoid conflict with any macro text strings. An example Call using the two highlights (Yellow & Green) in the mockup shell: % [NAME OF THE MACRO] ( Indta = [NAME OF YOUR INPUT DATA] ,outdta = [NAME OF YOUR OUTPUT DATA] ,condlist= %str (sum (paramcd I n (‘AST’ ‘ALT’) and ablfl=’Y’ and . <aval>=3*input (anrhi, best.)) > 0 &delim. sum (paramcd in (‘AST’ ‘ALT’) and trtelbfl=’Y’ and . <aval>=3*input (anrhi, best.)) > 0 and sum (paramcd=’BILI’ and trtelbfl=’Y’ and . <aval>2*input (anrhi, best.)) > 0 ) ,cdlist = ALSGE3BL&delim.AT3BL2PB ,cdnlist= 111&delim.112 ,nlist = %str (Elevated Total Bilirubin at baseline &delim.Elevated ALT or AST and BILI at pbl records) )

5. Conclusion This paper outlines the use of a reusable macro to streamline programming and improve efficiency The macro is used to create new needed parameters derived for subject-level event summary table within BDS ADaMdataset As more analysis needed, the macro input can be adjusted by concatenating a set of information repeatedly This approach is not limited to only the SUM function, other aggregate functions within the GROUP BY clause such as MEAN, COUNT, MIN and MAX can easily be used. The macro is not applicable in the cases where multiple events of interests need to be compared within the same subject by visits or dates.

6. Contact Information • Name: Masaki Mihaila • Company: Pfizer, Inc. • City/State: SF/ CA • Email: masaki.mihaila@pfizer.com