Use of SAS ® for Clinical Trial Management and Risk-Based Monitoring of Multicenter Clinical Trial...
27
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, PhD NESUG 2013 Presentation
-
Upload
bertha-henderson -
Category
Documents
-
view
223 -
download
0
Transcript of Use of SAS ® for Clinical Trial Management and Risk-Based Monitoring of Multicenter Clinical Trial...
Use of SAS® for Clinical Trial Management and Risk-Based Monitoring of Multicenter Clinical Trial Data from Electronic Data Capture Tools
Bob Hall1, MS, Rebecca V. Fink2 MPH, David Gagnon1, MD, MPH, PhD
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
Safety (e.g. Adverse Events, Pharmacy Data)Efficacy (e.g. Outcomes Data for Trial Objectives)Study 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 risk-based 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:
Initiate re-training efforts, improve communication with sites, increase on-site monitoring activities.
Probation efforts if needed.
Electronic Data Capture:
Electronic Data Capture (eDC):• 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
Ability to establish professional level reports in many formats.
• Physical Report DocumentFound sites were more responsive to a physical report.
• Reports in Near Real TimeDaily reports based on schedule cron jobs.Timing was considered acceptable for task.
• SAS Knowledge BaseResearch 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:Listing of outstanding CRFsDays outstanding from expectedSummary numbers
• Report for Management Teams:CRF completion rates by siteCompletion 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
Report frequencyAbility to generate ad hoc reports
• Time Limit / Grace Period for CRF CompletionProtocol defined / May vary by CRF type
• Risk-Based MonitoringOverall metrics:
Insure that percentage of CRFs complete w/in time limitNo more than 10% outside 14 days
Graphing completion rates from site launch
Monitoring Data Clarifications:
Flow of Data Clarification (DCF) : • DCFs Types:
Automated – Fire through a system scriptManual – Entered by coordinating center
• Expected Site Responses: Data correction because of true data error.Resolving DCF since data value is correct.
• Good Metric for Assessing DCF Correction is DCF Aging.
Monitoring Data Clarifications:
Business Use Case: • eDC system - Didn’t Provide Best Metric for DCF Aging:
Based on date/time DCF opened by staff.Aging continued after site response.
• Build a Better Report:Use DCF status tableRe-calculate aging based on time DCF fired Define calculation solely on site actionDevelop reports for sites and management teams
SAS Tools for Reporting:
SAS Tools for Reporting:• Output Delivery System (ODS) Functionality
Professional level reports in many formats (HTML, PDF, RTF)
• Access to SQL Relational DatabasesODBC Connections
• Macros to Facilitate ReportingMacro Looping to Generate by Site ReportsCustomized Reporting of Study Benchmarks
Customized DCF Report:
Business Case• Aggregate Report of Site Level Activity
Example - DCF Aging
• Repeatability of Reports• Professional Looking Report – ODS Approach• Modular Reporting Process
Flexibility in Defining/Re-Defining Metrics Flexibility in Ordering of Metrics in Report
DCF Report - Elements
Site 1 Site 2 Site 3 Site 4 Site 5 Site 6
exposurecat exposure N(%) N(%) N(%) N(%) N(%) N(%)
DCF Age DCF Age Mean 43.66 (53.22) 33.02 (50.85) 40.69 (58.54) 19.91 (41.1) 24.7 (37.96) 25.62 (50.37)
DCF Age Median 32 (2,59) 9 (0,48) 15 (0,58) 7 (1,21) 7 (0,30) 7 (0,28)
DCF Age 0-15 346 (43.7) 536 (58.7) 802 (50.3) 845 (71) 670 (58.1) 298 (60.7)
DCF Age 16-30 49 (6.2) 75 (8.2) 115 (7.2) 133 (11.2) 196 (17) 93 (18.9)
DCF Age 30-45 84 (10.6) 64 (7) 153 (9.6) 88 (7.4) 83 (7.2) 27 (5.5)
DCF Age 45-60 117 (14.8) 68 (7.4) 145 (9.1) 41 (3.4) 35 (3) 16 (3.3)
DCF Age >60 195 (24.7) 170 (18.6) 378 (23.7) 83 (7) 170 (14.7) 57 (11.6)
DCF Status DCF Status Open 49 (6.2) 31 (3.4) 26 (1.6) 24 (2) 11 (1) 10 (2)
DCF Status Answered 0 (0) 5 (0.5) 4 (0.3) 6 (0.5) 12 (1) 4 (0.8)
DCF Status Closed 742 (93.8) 877 (96.1) 1563 (98.1) 1160 (97.5) 1131 (98) 477 (97.1)
Site Number Site Number
The SAS System
Site Type
VA Non-VA
DCF Report – Format Table
Site 1 Site 2 Site 3 Site 4 Site 5 Site 6
exposurecat exposure N(%) N(%) N(%) N(%) N(%) N(%)
DCF Age DCF Age Mean 43.66 (53.22) 33.02 (50.85) 40.69 (58.54) 19.91 (41.1) 24.7 (37.96) 25.62 (50.37)
DCF Age Median 32 (2,59) 9 (0,48) 15 (0,58) 7 (1,21) 7 (0,30) 7 (0,28)
DCF Age 0-15 346 (43.7) 536 (58.7) 802 (50.3) 845 (71) 670 (58.1) 298 (60.7)
DCF Age 16-30 49 (6.2) 75 (8.2) 115 (7.2) 133 (11.2) 196 (17) 93 (18.9)
DCF Age 30-45 84 (10.6) 64 (7) 153 (9.6) 88 (7.4) 83 (7.2) 27 (5.5)
DCF Age 45-60 117 (14.8) 68 (7.4) 145 (9.1) 41 (3.4) 35 (3) 16 (3.3)
DCF Age >60 195 (24.7) 170 (18.6) 378 (23.7) 83 (7) 170 (14.7) 57 (11.6)
DCF Status DCF Status Open 49 (6.2) 31 (3.4) 26 (1.6) 24 (2) 11 (1) 10 (2)
DCF Status Answered 0 (0) 5 (0.5) 4 (0.3) 6 (0.5) 12 (1) 4 (0.8)
DCF Status Closed 742 (93.8) 877 (96.1) 1563 (98.1) 1160 (97.5) 1131 (98) 477 (97.1)
Site Number Site Number
The SAS System
Site Type
VA Non-VA
VARNAME: Variable Name TYPE: Continuous / CategoricalFFORMAT / VARLABEL / ORDER
VARCAT: Row Variable Category
DCF Report – Macro Variables
Site 1 Site 2 Site 3 Site 4 Site 5 Site 6
exposurecat exposure N(%) N(%) N(%) N(%) N(%) N(%)
DCF Age DCF Age Mean 43.66 (53.22) 33.02 (50.85) 40.69 (58.54) 19.91 (41.1) 24.7 (37.96) 25.62 (50.37)
DCF Age Median 32 (2,59) 9 (0,48) 15 (0,58) 7 (1,21) 7 (0,30) 7 (0,28)
DCF Age 0-15 346 (43.7) 536 (58.7) 802 (50.3) 845 (71) 670 (58.1) 298 (60.7)
DCF Age 16-30 49 (6.2) 75 (8.2) 115 (7.2) 133 (11.2) 196 (17) 93 (18.9)
DCF Age 30-45 84 (10.6) 64 (7) 153 (9.6) 88 (7.4) 83 (7.2) 27 (5.5)
DCF Age 45-60 117 (14.8) 68 (7.4) 145 (9.1) 41 (3.4) 35 (3) 16 (3.3)
DCF Age >60 195 (24.7) 170 (18.6) 378 (23.7) 83 (7) 170 (14.7) 57 (11.6)
DCF Status DCF Status Open 49 (6.2) 31 (3.4) 26 (1.6) 24 (2) 11 (1) 10 (2)
DCF Status Answered 0 (0) 5 (0.5) 4 (0.3) 6 (0.5) 12 (1) 4 (0.8)
DCF Status Closed 742 (93.8) 877 (96.1) 1563 (98.1) 1160 (97.5) 1131 (98) 477 (97.1)
Site Number Site Number
The SAS System
Site Type
VA Non-VA
DSN: Report DataRDSN: Format TableLIBOUT / FILEOUT
COLVAR: Column Variable- Site
BYVAR: Column Groups- VA vs Non-VA
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 n1• aging_s_x 1 DCF Age Median n2• dcfagecat 2 DCF Age c2 dcfagef.• dcfstatus 3 DCF Status c2 dcfstaf.• ; • run;
- Elements from DCF Table- n1, n2, c1, c2 represent different types of metrics
Customized Reporting:
Macro: %tab1mac_doc - Do-Loop w/ Call Symput
• %macro tab1mac_doc (rdsn, dsn, colvar, byvar, libout, fileout);• ………• %do count = 1 %to &maxcount;• data tmpk1; set &rdsn;• if _n_ eq &count;• call symput('vcat',varcat);• call symput('rf',varname);• call symput('vartype',type);• call symput('varlab',varlabel);• call symput('vorder',order);• call symput('fformat',fformat);• run;• %end;
DSN: Report DataRDSN: Format TableCOLVAR: Column Var.BYVAR: By Var.LIBOUT / FILEOUT
VARNAME: Variable Name TYPE: Continuous / CategoricalFFORMAT / VARLABEL / ORDER
Customized Reporting:
Macro: %tab1mac_doc / Ex. Continuous Variable
• **** for Continuous variables (mean and sd)****;• %if &vartype eq n1 %then %do;• proc means data=&dsn noprint; • by &byvar &colvar;• var &rf;• output out= contout1 mean=mean std=std;• run;• data contout1; set contout1;• length exposure $42;• sorder = &vorder; exposurecat = "&vcat";• exposure = "&varlab";• run;• data results_m; set results_m contout1;• run;• %end;
Data Table: TabFormat /N1 == Calculate Means and Standard Deviation
Formatting StatementsData Table: results_m(holds Mean, SD values)
Customized Reporting:
Macro: %tab1mac_doc / Proc Report Statements
• %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;
Generate a dataset tabres – combination of all results (continuous, categorical).Formatting of table columns for final report
Customized Reporting:
ODS Output Destination – This case MS Excel
• ods listing;• ods tagsets.ExcelXP style=listing• options (Sheet_Interval='proc' embedded_titles='Yes' Index='Yes' • Absolute_Column_Width='20,20,20'• Row_Heights='15,15,15,15,15,15,15')• file="O:\SAS\Users\VA101010\Reports\Tables.xls";• • ods tagsets.ExcelXP options(sheet_name="Table 1a");• proc document name=tabout.Table1a;• replay ;• run;• ods tagsets.ExcelXP close;• run;
Proc Document allows replay of stored reports.
DCF Report
Site 1 Site 2 Site 3 Site 4 Site 5 Site 6
exposurecat exposure N(%) N(%) N(%) N(%) N(%) N(%)
DCF Age DCF Age Mean 43.66 (53.22) 33.02 (50.85) 40.69 (58.54) 19.91 (41.1) 24.7 (37.96) 25.62 (50.37)
DCF Age Median 32 (2,59) 9 (0,48) 15 (0,58) 7 (1,21) 7 (0,30) 7 (0,28)
DCF Age 0-15 346 (43.7) 536 (58.7) 802 (50.3) 845 (71) 670 (58.1) 298 (60.7)
DCF Age 16-30 49 (6.2) 75 (8.2) 115 (7.2) 133 (11.2) 196 (17) 93 (18.9)
DCF Age 30-45 84 (10.6) 64 (7) 153 (9.6) 88 (7.4) 83 (7.2) 27 (5.5)
DCF Age 45-60 117 (14.8) 68 (7.4) 145 (9.1) 41 (3.4) 35 (3) 16 (3.3)
DCF Age >60 195 (24.7) 170 (18.6) 378 (23.7) 83 (7) 170 (14.7) 57 (11.6)
DCF Status DCF Status Open 49 (6.2) 31 (3.4) 26 (1.6) 24 (2) 11 (1) 10 (2)
DCF Status Answered 0 (0) 5 (0.5) 4 (0.3) 6 (0.5) 12 (1) 4 (0.8)
DCF Status Closed 742 (93.8) 877 (96.1) 1563 (98.1) 1160 (97.5) 1131 (98) 477 (97.1)
Site Number Site Number
The SAS System
Site Type
VA Non-VA
Concluding Remarks:
Centralized monitoring activity can improve efficiency of multi-center trials.
• Part of a Risk-Based approach to monitoring. eDC applications contains operational tables that
can assist in these activities.• Beyond defined reports in eDC 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
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: Rebecca Fink / David Gagnon
• Staff at VA CSP Boston Coordinating Center / MAVERIC Erika Holmberg / Allan Lewis
Audience Questions
