How to ensure independence of QC in statistical programming 

Posted by Cytel

Nov 11, 2016 10:05:14 AM

magnifyingglass.jpg

A solid and robust QC process is one vital component of ensuring quality programming delivery. Angelo Tinazzi and Cedric Marchand presented at the PhUSE conference recently on the topic of ‘What Auditors want’. As part of this well received presentation, the duo discussed the question of independence of QC and how to make sure this is fully observed.

In this blog we’ll walk through their recommendations to ensure independence of QC in programming.

Folder structures

Programming teams should always use a standard folder structure. As an example, this could be structured as follows:  projects organized by compound/indication, study and then task.  When drilling down to the task level, production and QC should have separate directories with the same folder structure for all types of statistical outputs (i.e. SDTM datasets, ADAM datasets, Tables, Figures, Listings, etc.).  Clearly, it’s vital that access to each project is regulated through robust and documented IT access control.

Closing the Loop

 QC programmers must investigate the reasons for any discrepancies, and discuss with the primary programmer, as well as provide possible solutions.  Tinazzi and Marchand noted that auditors will look for evidence that biostatisticians are kept in the loop with this process, and are available to clarify any uncertainties resulting from the QC activity. Each item which is delivered and resolved should of course have formal documentation of review and approval.

Training

 Team managers should develop training material for statistical programmers and biostatisticians on the key aspects for assuring independency in the QC process of programming outputs.  This should encompass the entire programming development life-cycle, from initial assessment of analysis requirements right through to final production, validation and revision including:

  • Reviewing key points of all applicable SOPs
  • Clarifying differences between QC and Validation of Programs
  • Listing all possible input specifications to be used
  • Output specifications and their review
  • When to use and when not to use standard tools (i.e. most standard tools can be used by the primary programmer only)
  • Recommended methods of QC for each type of output
  • How to conduct a fully independent QC/Review
  • How to handle discrepancies and their resolution.

 

In the associated PhUSE paper, Tinazzi and Marchand also provide their key do’s and don’ts for assuring independence of QC. Click the link to view our handy infographic on this topic. 

Check out Tinazzi and Marchand's PhUSE paper What Auditors Want and their presentation slides here.

 

 Further reading:

 

Lost in Traceability - from SDTM to ADaM

Mind the Gap! How to prepare for SDTM migrations

Infographic: 6 Hot Topics from PhUSE 2016 

 Liked this blog?  Join our audience of biopharma innovators and sign up for biostatistics and biometrics insights, news, best practices and industry trends direct to your inbox .  Click the button below to subscribe to the Cytel blog. 

Subscribe

 

Topics: Statistical Programming, Statistical Analysis, CDM, Trial Quality

The Cytel blog keeps you up to speed with the latest developments in biostatistics and clinical biometrics.  Sign up for updates direct to your inbox. You can unsubscribe at any time.

 

Posts by Topic

see all

Recent Posts