Implementing the CDISC standards into an existing CDMS Presented by Ueng-Cheng Yang (yang@ym.edu.tw) Institute of Biomedical Informatics National Yang-Ming University 1
TCT is a resource center to support clinical trials for the National Research Program for Biopharmaceuticals (NRPB) 2
Clinical Data Management System Integrating data by using a global unique identifier Specimen Tracking & Management System Patient Clinical information Protocol Tracking and Management Protocol, System informed consent, etc. Survival % + + + - - + - - Analysis year Specimen High throughput data Protocol application information was collected by a protocol tracking and management system http://www.cims.tw/ 3
Submit protocol to one center s system Application form Attached files Simplify the application of a multi-centered trial by using the same information system in 14 medical centers. Submit xml and zip files to another center s system Relevant parts of the submited protocol The trial summary (TS) data can be downloaded as xml files The downloaded xml file was not CDISC-compliant, so this file has to be converted into a CDSC-compliant xpt file. 4
Converting trial summary (TS) data according to SDTM standard http://www.cdisc.org/system/files/all/cdisc_ar_2013c-pages-web(1).pdf 5
More than 80% of the 13 clinical trial consortia are using the same CDMS supported by TCT http://tc2.ntu.edu.tw/ 6
System setup Set up access rights Randomization Study Design and generate forms Manage patients Collect data Study setup Screen patients Export & analyze data Resource center helps users to make the electronic clinical report forms (ecrfs) 7
ecrf used before adopting the CDISC standards The title of each question, the units, the range of normal values are simple text. The variable names are assigned based on user s codebook. 8
Making annotated CRF in an existing CDMS Using a finished phase I trial to test the workflow approach 9
Making annotated CRF by following the standard variable names and data types 10
Provide users a codebook for data query or analysis Even though a user is not familiar with the CDASH and SDTM standards. He can stilll make query or analyze data based on this codebook. 11
Using a finished phase I trial as an example to test the workflow approach 12
Click this button for detailed information Collected data will be validated based on rules after completing an ecrf This validation only focuses on a single ecrf. The consistency among ecrfs will not be checked at this point. Validated ecrf will be locked. 13
A single mistake may violate many rules The age of patient SCAI-04 was missing, which lead to these 5 violations of rules. 14
Correct the error and re-check this ecrf When research nurse corrected the error and froze the ecrf again, this ecrf was checked again. Since the status was all passed, this ecrf was locked. 15
Using a finished phase I trial as an example to test the workflow approach Data tabulation steps are semi-automatic. The most time consuming part is manual transversing rows and columns. 16
All domains were exported to a single Excel worksheet Exported data were manually separated based on SDTM domains into different worksheets. 17
Import Excel files into SAS Export tabulation results in xpt format Converting xlsx files to xpt files by using SAS The data were converted to xpt format directly without statistical analysis in this proof-of-concept study. In other words, the ADaM standard is not discussed here. 18
Using a finished phase I trial as an example to test the workflow approach 19
Validating data sets by using the OpenCDISC resource The define.xml file was generated by using OpenCDISC resource. This file and the xpt files were validated by the validator tool. http://www.opencdisc.org/projects/using-opencdisc-community 20
Validation report showed that some required data were not collected in this trial The users designed the CRFs themselves based on the needs in science. As a result, a few compulsory information were not collected. Previous validation step on CRF was based on the variables in the form. The CDISC required variables were not checked. 21
Summary The impact on users The users of CDMS did not feel any change except that they had to use a standardized codebook to perform analysis. Users need to work with staffs in resource center to prevent missing the required fields in the CRF design. The impact on resource center This workflow-based approach put most of the workload (i.e. making electronic annotated CRF) on staffs in the resource center. Staff s working experiences are precious for improving the CDMS, such as establishing a question library, etc. The advantage of adopting standards from start Questions are re-usable Quality control can be done in early stage 22