Use of SAS for Clinical Trial Management and

Use of SAS® for Clinical Trial Management and Risk-Based Monitoring of Multicenter Clinical Trial Data from Electronic Data Capture Tools Bob Hall 1, MS, Rebecca V. Fink 2 MPH, David Gagnon 1, MD, MPH, Ph. D NESUG 2013 Presentation

Objectives of Presentation: § Introduce Risk-Based Monitoring and Electronic Data Capture § Discuss Business Use Cases § Provide a SAS Approach for Generating Modular Reports for Site Monitoring Metrics

Multi-Center Clinical Trials: § Multi-center trials can be complicated: • Multiple visits for long term trials • Multiple CRFs with different purposes ØSafety (e. g. Adverse Events, Pharmacy Data) ØEfficacy (e. g. Outcomes Data for Trial Objectives) ØStudy Activity (e. g. Protocol Deviations; Disposition) § Trials require frequent monitoring of data to insure quality.

Risk-Based Monitoring: § Trial monitoring has focused on on-site activities § FDA Guidance Document that describes a riskbased approach to monitoring: • Recently finalized – August 2013 • Combines on-site monitoring and remote ‘centralized monitoring’. • Targeted monitoring of site activity remotely.

Risk-Based Monitoring: § Improve efficiency and costs of clinical trials • Reducing the need for frequent on-site monitoring. • Focus on sites that don’t meet defined study metrics. • Dictate corrective actions with sites: Ø Initiate re-training efforts, improve communication with sites, increase on-site monitoring activities. Ø Probation efforts if needed.

Electronic Data Capture: § Electronic Data Capture (e. DC): • Real-time data capture and correction. • Data maintained in relational database architecture. • Trial management functions (CRFs, DCFs, etc. ). § Operational Tables for Trial Management: • • Participant Status CRF/Form Status DCF Status Master Data Files (i. e. tables for all data transactions)

SAS for Reporting: § Points to Consider - SAS for Risk-Based Monitoring • Output Delivery System (ODS) Functionality Ø Ability to establish professional level reports in many formats. • Physical Report Document Ø Found sites were more responsive to a physical report. • Reports in Near Real Time Ø Daily reports based on schedule cron jobs. Ø Timing was considered acceptable for task. • SAS Knowledge Base Ø Research infrastructure had more SAS experience.

Business Use Cases: Business Use Cases for Discussion: 1) Case Report Form (CRF) Completion 2) Data Clarification (DCF) Status

Case Report Form (CRF) Completion: § Business Use Case: • Important to have data entered during or soon after participant visits. • Report for Site Staff: Ø Listing of outstanding CRFs Ø Days outstanding from expected Ø Summary numbers • Report for Management Teams: Ø CRF completion rates by site Ø Completion rates by CRF and visit window

CRF Completion: § Form Status Table • Captured information on form status markers. • Record for each expected CRF during course of study. § Determination of Incomplete CRFs Based on: • Date of participant enrollment • Expected date of study visit • Acceptable grace period for completion

CRF Completion:

CRF Completion: § Points to Consider: • Timing of Report Generation Ø Report frequency Ø Ability to generate ad hoc reports • Time Limit / Grace Period for CRF Completion Ø Protocol defined / May vary by CRF type • Risk-Based Monitoring Ø Overall metrics: ²Insure that percentage of CRFs complete w/in time limit v. No more than 10% outside 14 days ²Graphing completion rates from site launch

Monitoring Data Clarifications: § Flow of Data Clarification (DCF) : • DCFs Types: Ø Automated – Fire through a system script Ø Manual – Entered by coordinating center • Expected Site Responses: Ø Data correction because of true data error. Ø Resolving DCF since data value is correct. • Good Metric for Assessing DCF Correction is DCF Aging.

Monitoring Data Clarifications: § Business Use Case: • e. DC system - Didn’t Provide Best Metric for DCF Aging: Ø Based on date/time DCF opened by staff. Ø Aging continued after site response. • Build a Better Report: Ø Use DCF status table Ø Re-calculate aging based on time DCF fired Ø Define calculation solely on site action Ø Develop reports for sites and management teams

SAS Tools for Reporting: § SAS Tools for Reporting: • Output Delivery System (ODS) Functionality Ø Professional level reports in many formats (HTML, PDF, RTF) • Access to SQL Relational Databases Ø ODBC Connections • Macros to Facilitate Reporting Ø Macro Looping to Generate by Site Reports Ø Customized Reporting of Study Benchmarks

Customized DCF Report: § Business Case • Aggregate Report of Site Level Activity Ø Example - DCF Aging • Repeatability of Reports • Professional Looking Report – ODS Approach • Modular Reporting Process Ø Flexibility in Defining/Re-Defining Metrics Ø Flexibility in Ordering of Metrics in Report

DCF Report - Elements

DCF Report – Format Table VARCAT: Row Variable Category VARNAME: Variable Name TYPE: Continuous / Categorical FFORMAT / VARLABEL / ORDER

DCF Report – Macro Variables BYVAR: Column Groups - VA vs Non-VA DSN: Report Data RDSN: Format Table LIBOUT / FILEOUT COLVAR: Column Variable - Site

Customized Reporting: Table to Define Report Structure • • • data tabformat; input varname $14. order @18 varlabel $14. @34 type $2. +2 fformat $8. ; if fformat eq ' ' then fformat = '4. '; if _n_ gt 3 then varcat = 'DCF Status’; else varcat ='DCF Age'; call symput('maxcount', _n_); datalines; aging_s_x 1 DCF Age Mean n 1 - Elements from DCF Table aging_s_x 1 DCF Age Median n 2 - n 1, n 2, c 1, c 2 represent dcfagecat 2 DCF Age c 2 dcfagef. different types of metrics dcfstatus 3 DCF Status c 2 dcfstaf. ; run;

Customized Reporting: Macro: %tab 1 mac_doc - Do-Loop w/ Call Symput • • • • DSN: Report Data RDSN: Format Table COLVAR: Column Var. BYVAR: By Var. LIBOUT / FILEOUT %macro tab 1 mac_doc (rdsn, colvar, byvar, libout, fileout); ……… %do count = 1 %to &maxcount; data tmpk 1; set &rdsn; if _n_ eq &count; call symput('vcat', varcat); call symput('rf', varname); VARNAME: Variable Name call symput('vartype', type); TYPE: Continuous / Categorical call symput('varlab', varlabel); call symput('vorder', order); / VARLABEL / ORDER call symput('fformat', fformat); run; %end;

Customized Reporting: Macro: %tab 1 mac_doc / Ex. Continuous Variable • • • • **** for Continuous variables (mean and sd)****; %if &vartype eq n 1 %then %do; proc means data=&dsn noprint; by &byvar &colvar; var &rf; output out= contout 1 mean=mean std=std; run; data contout 1; set contout 1; length exposure $42; sorder = &vorder; exposurecat = "&vcat"; exposure = "&varlab"; run; data results_m; set results_m contout 1; run; %end; Data Table: Tab. Format / N 1 == Calculate Means and Standard Deviation Formatting Statements Data Table: results_m (holds Mean, SD values)

Customized Reporting: Macro: %tab 1 mac_doc / Proc Report Statements • • • Generate a dataset tabres – combination of all results (continuous, categorical). Formatting of table columns for final report %if (&byvar ne ) and (&vcat ne ) %then %do; ods listing; ods document name=&libout. . &fileout (write); proc report data=tabres nowd headline ps=130; column exposurecat exposure &byvar, &colvar, freqpercent ; define exposurecat/ group order=data width=55; define exposure /group order=data width=45; define &byvar /across order=internal format=byf. ; define &colvar /across order=internal format=colf. ; define freqpercent /group width=20; run; ods document close; run; quit; %end;

Customized Reporting: ODS Output Destination – This case MS Excel • • • • ods listing; ods tagsets. Excel. XP style=listing options (Sheet_Interval='proc' embedded_titles='Yes' Index='Yes' Absolute_Column_Width='20, 20' Row_Heights='15, 15, 15, 15') file="O: SASUsersVA 101010ReportsTables. xls"; ods tagsets. Excel. XP options(sheet_name="Table 1 a"); proc document name=tabout. Table 1 a; Proc Document allows replay ; replay of stored reports. run; ods tagsets. Excel. XP close; run;

DCF Report

Concluding Remarks: § Centralized monitoring activity can improve efficiency of multi-center trials. • Part of a Risk-Based approach to monitoring. § e. DC applications contains operational tables that can assist in these activities. • Beyond defined reports in e. DC application. § SAS / ODS Reporting Features can be used to generate professional reports. • Modular reporting macro can assist in providing center metrics.

Concluding Remarks: § Full Macro Information • Available at (http: //people. bu. edu/gagnon) § Acknowledgements • SAS® Acknowledgements Ø SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. • Co-authors: Ø Rebecca Fink / David Gagnon • Staff at VA CSP Boston Coordinating Center / MAVERIC Ø Erika Holmberg / Allan Lewis Audience Questions