Transcript Document
Analyses and Displays for Vital Signs, ECG, and Laboratory Analyte Measurements Mary Nilsson, Eli Lilly and Company; Wei Wang, Eli Lilly and Company; Qi Jiang, Amgen OBJECTIVE White Paper 1 – Finalized October 2013 The objective of this poster is to receive feedback on recommended displays for vital signs, electrocardiogram, and laboratory analyte measurements as outlined in PhUSE Computational Science Symposium (CSS) white papers as part of a standardization and code-sharing effort. The Vision – Development of Standard Scripts for Analysis and Reporting Working Group Recommended Tables and Figures Traditional Tables and Figures • Only has mean and SD - Lacks additional useful summary statistics • Outliers not displayed • Can be misleading when data are non-normal • Allows for visual inspection of changes over time • Can visually assess the potential impact of outliers on the central tendency summary statistic • Out of range values in red • Easy to see treatment differences • Summary table compliments box plot so numbers are available for textual summaries White Paper 2 – Under st 1 • Lack of information on patient level information o Lack of magnitude of shift • Difficult to assess overall treatment difference o Very popular, however, users tend to count and create grouped percentages manually for those shifting to high from low/normal (or low from normal/high). Round Review! FDA Industry Academic Where to Find Things Other White Papers Standard Targets and Validated Scripts White Papers to Fill the Gaps of Standardization of Tables, Figures, and Listings (Creating Standard Targets) Clinical Data Flow Industry Standards Alignment Trial Design PRM Data Collection Systems CDASH Observed Datasets SDTM Analysis Datasets ADaM Tables, Figures and Listings No TFL Stds Exist Examples of Discussed Topics Whether to report p-values and confidence intervals Handling of measures collected in reflex manner, repeated measures, unplanned measures, measurements post drug exposure Defining baseline Central versus local laboratories Choices of lab reference limits Units and transformations ECG correction factors Two pages per measurement o One for Max. baseline vs. Max. post-baseline (shown on the left) o Another identical one for Min. baseline vs. Min. post-baseline • Scatter plot o Shows patient level information o Allows quick browsing through a large amount of information • Shift table o Shows useful summary level statistics • Treatment emergent high/low table o Shows concise summary level statistics o Allows for easy assessment of treatment difference PhUSE CSS Final Deliverables: www.phuse.eu, click Publications PhUSE CSS Work in Progress: www.phusewiki.org, click CSS Working Groups, then Development of Standard Scripts Script Repository (reusable code library): https://code.google.com/p/phuse-script In Development: Adverse Events Demographics, Disposition, Medications Hepatotoxicity Non-compartmental PK Planned: QT Studies Conclusion: Industry standards have evolved over time for data collection (CDASH), observed data (SDTM), and analysis datasets (ADaM). Development of standard tables and figures with associated analyses is the next step! Having an organized process for shared learning of improved methodologies can lead to earlier safety signal detection and better characterization of the safety profile of our products. We welcome new members! Contact information on phusewiki.org. Acknowledgements: Special acknowledgement to members of the PhUSE Computational Science Symposium Standard Scripts for Analysis and Programming Working Group, White Papers Project Team who have contributed to the white papers, and to those who have provided review comments.