Download presentation
Presentation is loading. Please wait.
Published byBrice Chapman Modified over 8 years ago
1
Tracing Data Elements through a Standard Data Flow Alistair Dootson Dootsonic Ltd, Manchester, UK 16 February 2016Copyright Dootsonic Ltd 2011
2
16 February 2016Copyright Dootsonic Ltd 2011 StandardsMetadataETLTFLs
3
16 February 2016Copyright Dootsonic Ltd 2011 Trial Design Code Data Collection Systems Code Observed Datasets Code Analysis Datasets Code Tables, Figures & Listings Data Trial Metadata Map Source Metadata Map Observed Data Standards Map Analysis Data Standards Map TFL Definition Metadata Trial Design Standards Data Element Standards Observed Data Standards Analysis Dataset Standards TFL Standards Libraries PRMCDASHSDTMADaM No Standards Industry Standards “Floverview”
4
Standards 16 February 2016Copyright Dootsonic Ltd 2011 Industry Standards PRM CDASH SDTM ADaM Internal Standards CDISC-SHARE ®
5
Metadata What is Metadata? How do we “get” metadata? Proc Contents? 16 February 2016Copyright Dootsonic Ltd 2011
6
Metadata Proc SCAPROC proc scaproc; record "C:\phuse\record.txt" attr; run;... code... proc scaproc; write; run; 16 February 2016Copyright Dootsonic Ltd 2011
7
Metadata /* JOBSPLIT: DATASET INPUT SEQ WORK.DEMO.DATA */ /* JOBSPLIT: LIBNAME WORK V9 'C:\DOCUME~1\LOCALS~1\Temp\SAS Temporary Files\_TD5868' */ /* JOBSPLIT: DATASET OUTPUT SEQ WORK.DEMO1.DATA */ /* JOBSPLIT: LIBNAME WORK V9 'C:\DOCUME~1\LOCALS~1\Temp\SAS Temporary Files\_TD5868' */ /* JOBSPLIT: ATTR WORK.DEMO.DATA INPUT VARIABLE:subjid TYPE:NUMERIC LENGTH:8 LABEL:Subject ID FORMAT:BEST12. INFORMAT:F12. */ /* JOBSPLIT: ATTR WORK.DEMO.DATA INPUT VARIABLE:siteid TYPE:CHARACTER LENGTH:2 LABEL:Center or Site ID FORMAT:$F2. INFORMAT:$F2. */ 16 February 2016Copyright Dootsonic Ltd 2011
8
Metadata 16 February 2016Copyright Dootsonic Ltd 2011
9
Metadata 16 February 2016Copyright Dootsonic Ltd 2011
10
ETL 16 February 2016Copyright Dootsonic Ltd 2011 Clinical DI Studio ® Use Metadata to Produce Jobs Trace that Metadata
11
ETL 16 February 2016Copyright Dootsonic Ltd 2011 A Basic Job
12
ETL 16 February 2016Copyright Dootsonic Ltd 2011 Use Analyze Features DatasetColumn
13
ETL 16 February 2016Copyright Dootsonic Ltd 2011 Full Analysis of ADSL Column
14
TFLs 16 February 2016Copyright Dootsonic Ltd 2011 Subject DemographicsPage x of y hh:mm ddmmmyyyy ITT Populationstatus STUDY # Arm A (pem+cis) Arm B (placebo+cis)Total (N=xxx) Parametern(%)n n Gender Femalexx(xx.x)xx(xx.x)xx(xx.x) Malexx(xx.x)xx(xx.x)xx(xx.x) Race Caucasianxx(xx.x)xx(xx.x)xx(xx.x) Hispanicxx(xx.x)xx(xx.x)xx(xx.x) Asianxx(xx.x)xx(xx.x)xx(xx.x) Age (years) Number of Subjectsxx Meanxx.x SDxx.x Medianxx.x Minimumxx.x Maximumxx.x
15
TFLs 16 February 2016Copyright Dootsonic Ltd 2011 Raw Data ( floor(( intck( 'month',DEMO.BIRTHDT,datepart( DEMO.TRTSTDT )) - ( day( datepart( DEMO.TRTSTDT )) < day( DEMO.BIRTHDT ))) / 12 )); DM.AGE proc MEANS data=DM noprint; var AGE; output out=MEANS median=MedianAge max=MaxAge min=MinAge std=StdAge mean=MeanAge; run; Raw SDTM SDTM ADaM ADaM Display
16
16 February 2016Copyright Dootsonic Ltd 2011 How is it calculated for display in the table? What are the inputs for that calculation in the ADaM data? What are the observed data elements (SDTM) that feed into the ADaM derivation? What are the original data elements in CDASH? Traceability
17
16 February 2016Copyright Dootsonic Ltd 2011 Traceability
18
16 February 2016Copyright Dootsonic Ltd 2011 Traceability
19
16 February 2016Copyright Dootsonic Ltd 2011 Traceability Note: Some columns excluded
20
16 February 2016Copyright Dootsonic Ltd 2011 Some ETL tools provide the facility to trace data elements through their system. The point of traceability is to ensure the quality and integrity of the final variable values. When tracing back variables from the tables, would it shed light on how many variables are collected and not used? What drives the data flow of a study. Are the specifications and database built first? Conclusions
21
16 February 2016Copyright Dootsonic Ltd 2011 However, consider defining the tables first and working backwards. Would that make the data capture and flow more efficient? There seems to be great benefits available for reusability from ETL tools, but not necessarily for full traceability. There’s no one system out there that does this traceability completely, but there are ways of manually putting this together to add value Conclusions
22
16 February 2016Copyright Dootsonic Ltd 2011 Standards Be Proactive Keep up to date with initiatives TFL Development No tools can currently trace from TFL back to Raw Data Transparency of data is not yet required, but it is becoming more visible Conclusions
23
16 February 2016Copyright Dootsonic Ltd 2011 alistair@dootsonic.com Questions?
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.