Skip to content

Latest commit

 

History

History
62 lines (44 loc) · 6.68 KB

TODO.md

File metadata and controls

62 lines (44 loc) · 6.68 KB
Raw

Standard Analyses Working Group Task Lists

  1. Standard Analyses (White Papers) to-do list
  2. Standard Scripts (Repository Contents) to-do list
  3. GitHub Infrastructure to-do list
  4. General WG5 Principles and Conventions to-do list

White Papers (Project 08)

Ongoing development & review - White papers that welcome comments and suggestions:
  1. Hepatotoxicity: Clinical Trials and Integrated Summaries see this paper's phusewiki page
  2. Adverse Events: Clinical Trials and Integrated Summaries see this paper's phusewiki page
  3. QT Studies see this paper's phusewiki page
Initial planning - White papers that need further contribution and authoring:
  1. Questionnaire Data see this paper's phusewiki page
  2. Events of Special Interest see this paper's phusewiki page
Final (published) white papers (see PhUSE CS Final Deliverables Catalog):
  1. 2013-10 Measures of Central Tendency: Vital Signs, ECG, Labs (see also this white paper's phusewiki page)
  2. 2014-03 Non-Compartmental Pharmacokinetics (see also this white paper's phusewiki page)
  3. 2014-10 Demographics, Disposition and Medications: Clinical Trials and Integrated Summaries (see also this white paper's phusewiki page)
  4. 2015-09 Analyses of Outliers and Shifts: Vital Signs, ECG, Labs (see also this white paper's phusewiki page)

Repository Contents (Project 02)

White Paper on Measures of Central Tendency (WPCT)

Currently, our main objective is to deliver robust, easy to use and understand SAS and R scripts for WPCT analyses and displays. There are several steps in this process, and therefore several ways that people can participate, learn and contribute.

For details, see our WPCT Task List. See also the topics below for P08/P02 discussion.

Repository Infrastructure (Project 03)

Finalizing and implementing an overall project GitHub structure is the main activity for Project 03.

WG5 Principles and Conventions

There are several general topics to resolve, especially between the Standard Analyses (P08) and Standard Scripts (P03) teams. These arose while implementing WPCT analyses, so we use these analyses to illustrate the general topics.

  • We have drafted:

    • General Output & Formatting Requirements, which should apply to Standard Analyses across white papers, and
    • WPCT General Requirements, which should apply across WPCT displays.
    • Do the White Paper team endorse this approach, and want to finalize these?
    • What about general guidance of SDTM vs ADaM as the source data for standard analyses? The implementation team chose to limit ourselves to ADaM.
    • More generally, should we establish basic ADaM requirements that the Standard Analyses fully support, such as variable names for flags, common derivations, etc?

  • Otherwise, what is the best way to specify and document analysis-specific details such as the following from WPCT?

    • Does the team accept the revised footnotes in the scripts and outputs (accessible via our WPCT task list). How/where do we document these?
    • WPCT Scripts currently restrict data to the safety population (SAFFL = 'Y'). Is this correct; should we specify/document this for each analysis?
    • Most WPCT scripts currently restrict data to "analysis" measurements (e.g., ANL01FL = 'Y'). Is this correct; should we specify/docuemnt this for each analysis?
    • Do we need more detailed specification of the ANCOVA models for pvalues?
    • Section 8.2 mentions that Figures 7.1 & 7.2 could contain additional statistics. Should we implements any specific, optional, stats?
    • Figs. 7.7 & 7.8 should include all subjects with a baseline and post-baseline (see discussion in Section 8.1). What about Fig. 7.6? Does this need clarification? Should the left-hand plot in Fig. 7.8 (which includes right-hand plot of change) repeat & match Fig. 7.6 (which otherwise does not mention change)?
    • Fig. 7.7 in the white paper includes least square means, otherwise not discussed in the paper.
    • Fig. 7.7 footnote describes the ANCOVA model as "Change = Baseline + Treatment + Study", but the table presents a p-value for each study. Is this right? Does running the model by study conflict with "study" as an independent effect? Or is the by-study model different from the "Pooled" model (which includes "study" effect)?
    • Fig. 7.8 in the white paper includes additional results otherwise not discussed in the paper: "TE High n(%)" and "TE Low n(%)". What are these? Should these be included? By default? Only optional?
    • Section 8.3 Discussion implies that typically increases or decreases are of interest for a particular parameter. Should Fig. 7.8 therefore have a parameter-specific definition of "worst", rather than displaying both Min & Max (only one of which is typically meaningful)?