Utilizing Clinical SAS Report Templates with ODS Sunil Kumar Gupta, Gupta Programming, Simi Valley, CA

Transcription

1 Utilizing Clinical SAS Report Templates with ODS Sunil Kumar Gupta, Gupta Programming, Simi Valley, CA ABSTRACT SAS progrannners often have the responsibility of supporting the reporting needs of the Clinical Affairs Department of Phannaceutical Companies. This involves generating a variety of reports to fulfill the regulatory requirements of the clinical investigation of the medical device. By establishing clinical reporting templates using ODS for each of the functional aspects of Clinical Data Management, the SAS progranuner can dramatically improve the efficiency of developing and generating reports. With ODS, RTF and PDF files can easily and quickly be generated directly from SAS programs. Examples of these functional reports include patient listings, patient x-ray logs, monitor site visits, monthly status reports, & reports necessary for pre market approval application. Each type of report requires different layouts due to the different objectives of the organization. By establishing a reporting template using ODS for each type, the SAS programmer can make modifications to an already verified and complete program. Only minor adjustments may be required for user specific selections. A software tool used to facilitate this process is complementsoft ASAP"' software application with Clinical Trial Reporting Templates and Data Flow Diagrams. Using SAS report templates with ODS will facilitate the rapid development and execution of clinical reports. Establishing reporting standards will increase efficiency. This reduces time to develop, test and validate programs. INTRODUCTION Supporting the reporting needs of the Clinical Affairs Department can be a challenging task. This is because of the pressure on the programmer to quickly generate a variety of reports. Time is often critical as millions of dollars in sales can be lost due to delays in submission. By establishing clinical reporting templates for each of the functional aspects, the SAS programmer becomes organized and prepared to receive reporting requests. The department improves their efficiency in the development and generation of clinical reports. System standardization can be realized in the following areas: Program Header File Definitions Merged File Definitions Field Properties Standard & Custom Conditions Custom Formats Random Sample Verification of Dataset Contents Output Layout Best Practices: Clinical Trial Reporting Templates ODS R1F and PDF Destinations Standad Study Format Sample Datasets Macro Variables If possible, it is recommended to use the report specification similar to one defined in the SAS Service Request Form before initiating the coding. This paper will review sample templates for patient listings and summary tables. SYSTEM STANDARDIZATION Program Header By completing the program header, proper documentation is maintained for all programs. This is essential for program support. File Definitions All files containing the clinical information are defined with any access & view descriptors. This is a central source of file documentation. Merged File Definitions All links are defmed and established to create new files for data consolidation & reporting. Field Properties For each file, fields are defmed with label, format, & in format statements. These properties are utilized in all data analysis and presentation. Standard & Custom Conditions on the file Any required conditions can be applied to assure data validity. Any user specific conditions can also be applied. Custom Formats Standards in study formats can be recognized. Random Sample With a small random sample, a sufficient amount of data is available for testing. This allows for rapid testing without utilizing all the data in the dataset. 264

2 Verification of Dataset Contents This is important to assure proper data extraction and to verily programming assumptions. Output Layout The end user knows what to expect on the report. The header contains the title, date of execution, and selection conditions. Best Practices: Clinical Trial Reporting Templates One of complementsoft's key feature, Template Manager, enables all programs and macros to be stored in a controlled and shared environment. Each template represents a functional report such as patient listings, summary tables and summary graphs. The following procedures are presented with the macro language: DATA Step, PROC SORT, FORMAT, REPORT, TABULATE & PRINT. Standard Study Format A central standard study format used by all clinical report templates assures consistent display of the results. All updates and additions to the formats should be made in one program. PROC FORMAT; VALUEINV I ~ 'Investigator I' 2 ~ 'Investigator 2' 3 ~ 'Investigator 3' 4 ~ 'Investigator 4' 5 ~ 'Investigator 5'; VALUE DRUG I~ 'Active' 2 ~ 'Placebo'; VALUE SEX I= 'Male' 2 = 'Female'; VALUE RACE I ~ 'Whilte' 2 ='Black' 3 ='Asian'; Another key feature is the Data Flow Diagrams. This enables you to see a visual flow of the SAS procedures and the source data set for each clinical reporting template. QUIT; PICTURE PCTPCT LOW-HIGH= '99"/o' (PREFIX='(~; VALUE VISIT I= 'Preop' 2 = 'Operative' 3 ='Week l' 4='Week2' s~ week3'; Sample Datasets The DEMO data set represents patient demographics and the PVTSIT data set represents patient follow-up visit information. ODS RTF And PDF Destinations Taking advantage ofods's RTF and PDF destinations offers the ability to include in-line fonnats such as superscripts and subscripts in titles and footnotes and style options to control fonts. SAS Procedures including PROC REPORT, PROC TABULATE and PROC PRINT will be displayed. Program Header; *Name - Create_Datasets; Function - Create Datasets for Clinical Reports; Programmer - Sunil K. Gupta; * Date - 7 I ; Verification Date- 7/2/2002 By-; * Revision Log; *Date Programmer * Patient Demographics File; Change; DATA DEMO; INPUT INV 1-2 PATNO 4-5 PTNAME $5. DRUG 12 SEX 14 AGE RACE 20; WHERE INV TN (I,2) /* Custom Conditions *I AND DRUG TN (l) AND SEX TN (I) AND AGE< 45 AND RACE TN(!); 265

3 *IF RANUNI(-1) LE 0.1; /*Random Sample*/ LABEL /* Field Properties */ INV ='Investigator' PATNO = 'Patient*#' PTNAME = 'Patient*Name' DRUG ='Drug' SEX ='Sex' AGE ='Age' RACE ='Race'; FORMAT INV INV. PATNO 2. PTNAME $5. DRUG DRUG. SEX SEX. AGE 3. RACE RACE.; CARDS; 1 JOHN JAN TIM JOE SUE MARY I * Merged File - Demographics & Followup File; PROC SORT DATA=DEMO; BY INV PATNO; PROC SORT DATA=PVISIT; BYINVPATNO; DATA PDVISIT; MERGE DEMO (IN=A) PVISIT (IN=B); BYINVPATNO; *IF RANUNI(-1) LE 0.1; IF A; Verification ofdataset Contents; PROC FREQ DATA=PDVISIT; TABLES!NV PATNO DRUG SEX RACE VISIT; Patient Followup Visits File; DATAPVISIT; INPUTINV 1-2 PATN04-5 VISIT 7-8 EFFI EFF SAFI 16-17; Custom Conditions; *WHERE!NV IN (1,2) AND VISIT IN (1,2,3); * IF RANUNI( -I) LE 0.1; I* Random Sample */ Field Properties; LABEL INV PATNO VISIT EFFI EFF2 SAFI FORMATINV PATNO VISIT EFFI EFF2 SAFI CARDS; I I I I I I I I I = 'Investigator' = 'Patient*#' ='Visit' = 'Efficacy*Variable*l' = 'Efficacy*Variable*2' = 'Safety*Variable*l'; INV. 2. VISIT ; Ma~ro V arlables Level I Controls Level I control macro variables enable the programmer to specify the data set name, factor & analysis variable, & statistics. The factor variable may be more than one variable. The analysis variable is usually one of the efficacy or safety variables. The statistics option inclndes the following: mean, sum, min, max, & std. %LETDATAV =PDVISIT; 'YoLET CLASSY = INV PATNO VISIT; %LETVARV =EFFI; %LET STATVI =MEAN; Leyel 2 Controls Level 2 control macro variables enable the programmer to further subset the data for additional analysis. For example, the programmer can specify the investigator number, patient number, or visit period. %LETINVV %LETPATNOV %LETVISITV =(1,2); =(I, 2, 3, 4, 5); = (1, 2, 3); PATIENT LISTING For a simple patient listing, the REPORT procedure can be utilized. 266

4 Once the file is sorted, then PROC REPORT can be applied. This generates a sorted listing of the investigator name, patient number, visit period, and score value. See output I for the PDF file generated. Output 1: PDF File CUNJCAL REPORT: Pdlknl UD#ng ~ l6./un02, 08.:23 %MACRO OUTPUT!; PROC SORT DATA=&DATAV; BYINVPATNO; ODS PDF FILE = 'C:\outputl.pdf; TITLE! "CLINICAL REPORT: Patient Listing- &SYSDATE, &SYSTIME"; PROC REPORT DATA= &DATAV NO WINDOWS CENTER HEADLINE HEADSKIP MISSING LIST SPLIT='*'; Level2 controls; WHERE INV IN &INVV AND PATNO IN &PATNOV AND VISIT IN &VISITV; Apply custom fonnats; FORMAT INV INV. VISIT VISIT.; List column variables; COLUMN &CLASSY &V ARV; Define each column field; DEFINE!NV f GROUP CENTER 'Investigator*Name'; DEFINE PATNO f GROUP CENTER 'Patient*#'; DEFINE VISIT f GROUP CENTER 'Visit'; DEFINE &V ARV f &STATV! CENTER 'Efficacy*V ariable*l' FORMAT=COMMA6.; * Generate a line after each Investigator; COMPUTE AFTER!NV; LINE 45*'-'; LINE 'Efficacy Variable I Scores for' INV 2. LINE''; ENDCOMP; ODS PDF CLOSE; %MEND OUTPUT!; %OUTPUT!; +2' Investigator '; SUMMARY TABLES a'lkaryvariabii:iscqmfor I ln'!'mlpror For a summary table across follow-up visits, the REPORT procedure is applied again. This procedure generates a sunnnary table by investigator name and patient number. The visit scores reported include: mean, minimum, maximum, and standard deviation. See output 2 and 3 for the reports generated. %LET CLASS=INV PATNO; %MACRO OUTPUT2; ODS PDF FILE = 'C:\output2.pdf; TITLE! "CLINICAL REPORT: Patient Listing- &SYSDATE, &SYSTIME"; PROCREPORTDATA=&DATAV NO WINDOWS NOWD HEADLINE HEADSKIP MISSING LIST SPLIT='*'; Level 2 controls; WHERE INV IN &INVV AND PATNO IN &PATNOV AND VISIT IN &VISITV; * List column fields; COLUMN &CLASSY &VARY &VARV=SMIN &V ARV=SMAX &V ARV=SSTD; Apply custom fonnats; FORMAT INV INV.; Define each column fields; DEFINE INV f GROUP CENTER 'Investigator*Name'; DEFINE PATNO f GROUP CENTER 'Patient*#'; DEFINE &V ARV I &STATVI CENTER 'Efficacy*Variable*l' FORMAT=COMMA6.; DEFINE SMIN f MIN CENTER 'MIN*SCORE' FORMAT=COMMA6.; DEFINE SMAX f MAX CENTER 'MAX*SCORE' FORMAT=COMMA6.; 267

5 DEFINE SSTD I STD CENTER 'STD*SCORE' FORMAT~COMMA6.; COMPUTE AFTER INV; LINE 45*'-'; LINE 'Efficacy Variable 1 Scores for '!NV 2. LINE''; ENDCOMP; +2 ' Investigator '; table (sex race), drug=''* (n='n'*f.=7. pctn<sex race>=' %' *f.=pctpct. ) I box=[labei='baseline ' I* Style for box label *I style=[font_face="arial" font_weight=bold)) rts=43; run; ODS RTF CLOSE; %MEND OUTPUT3; %0UTPUT3; ODS PDF CLOSE; %MEND OUTPUT2; %0UTPUT2; Output 3: RTF File Output 2: PDF File CI.INJCAI. REPORT: Pathmt Usti"g. 16JlJN02, 08:2.l PROC PRINT LISTING Another useful SlllllllllUY table is the Baseline Characteristics table. This procedure generates a sununary of Sex and Race by Drug. The statistics reported include: N, and%. In-line formats are used to create subscripts and superscripts in the RTF file. In addition, style options are applied in the TABULATE procedure to specify the fonts. %MACRO OUTPUT3; OPTIONS ORIENTATION=PORTRAIT nodate center; ODS ESCAPECHAR = '\'; ODS RTF FILE= 'C:\output3.rtf' STYLE= MINIMAL I* Basic MS Word table, no color*/ BODYTITLE; /* Keep titles and footnotes in body */ Title font=arial bold "Baseline Characteristics: Sex, Race\{super a}"; Footnote) ''\S={font_face=arial) \{super a}race: Non-White consists of Black, Hispanic, and Native Arnerican\S={) "; Footnote2 ''\S= {font_ face=arial font_ style=italic) Program: /stat/druga/prograrnl.sas \S={)" ; Footnote3 "\S= { font_face=arial font_ style=italic) Output: Output3.rtf(Date Generated: &sysdate) \S={)"; PROC TABULATE data=demo missing formchar=' ' style=[font_face="arial'1; /*Style for all data*/ I* Style for column and row header */ class sex race drug/style=[font_face="arial" font_weight=bold); In addition to using the REPORT and the TABULATE procedures for clinical report templates, you can take advantage of these similar features in the PRINT procedure. Style options can be applied to several sections within the PRINT procedure to specify different fonts in the listing. %MACRO OUTPUT4; OPTIONS ORIENTA TION=PORTRAIT nodate center; ODS ESCAPECHAR = '\'; ODS RTF FILE = 'C:\output4.rtf STYLE= MINIMAL /*Basic MS Word table, no color*/ BODYTITLE; /* Keep titles and footnotes in body *I Title font=arial bold "Sample Proc Print"; Footnote I "\S= {font_ face=arial font_ style=italic} Program: lstat/druga/prograrnl.sas \S={)"; Footnote2 "\S= { font_face=arial font_ style=italic} Output: Output4.rtf(Date Generated: &sysdate) \S={)"; PROC PRINT data=derno (obs=5) noobs style(header) = [font_face="arial" font_weight=bold] I* Style for headers *I style( data)= [font_face="arial"); I* Style for all data *I var inv patno drug sex race age; run; ODS RTF CLOSE; %MEND OUTPUT4; %0UTPUT4; 268

6 REFERENCES SAS Procedures Guide, ver. 6, third edition; SAS Guide to the Report Procedure - Usage and Reference, ver. 6, first edition; TRADEMARK INFORMATION SAS is a registered trademark of the SAS Institute Inc., Cary, NC,USA. ABOUT THE AUTHOR The author welcomes your connnents & suggestions. SUMMARY By effectively utilizing SAS's flexibility and power for report generation with ODS, it is possible to expedite the process of report development to produce great looking PDF and RTF files. Clinical SAS report templates serve as a set of tools to facilitate the rapid development and execution of clinical reports. A software tool used to facilitate this process is complementsoft ASAP'M software application with Clinical Trial Reporting Templates and Data Flow Diagrams. With the clinical report templates, efficiency is realized because the programmer starts working with code from a similar program. Often, there are similar studies being conducted and the method of clinical review is similar across the efficacy variables. In addition, a great deal of time is saved because the modified code does not need to go through a complete and comprehensive test as that required of the original code. Sunil Kumar Gupta Gupta Programming SAS Certified Professional V 6TM 213 Goldenwood Circle Simi Valley, CA (805) Sunii@GuptaProgramming.com Sunil is a principal consultant at Gupta Programming. He has been using SAS software for over 10 years and is a SAS Certified Professional V 6TM. He has participated in over 6 successful FDA submissions. His consulting projects with pharmaceutical companies include the development of a Macro Based Application for Report Generation and Customized Plots and Charts. He is also the author of a Books By User book on the Output Delivery System and was a SAS Institute Quality PartoerTM for over 5 years. ACKNOWLEDGMENTS The author would like to thank Dr. Fred Hoehler of Data Management Center, Santa Ana, CA and Kirk Paul Lafler of Software Intelligence Corporation, Spring Valley, CA for their assistance in the preparation of this paper. 269