PDDI Info Model Task Force - Content Sub-team minutes for 5/19

Previous Topic Next Topic
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

PDDI Info Model Task Force - Content Sub-team minutes for 5/19

Richard Boyce-2
Hi, the minutes and link to meeting recording for the May 19th meeting of the content sub-team of the PDDI Info Model Task Force are pasted below. Kind regards, -R

Minutes for 5/19/2016 (Content subgroup)

In Attendance:  Evan Draper, Brian LeBaron, Richard Boyce, Dan Malone, John Poikonen, Michel Dumontier, Scott Nelson, Jeff Nielsen, Serkan Ayvaz, John Horn, Elizabeth Garcia, Louisa Zhang

Meeting recording: http://goo.gl/lESwy5


  • Introductions

  • Refresher from last meeting

    • Agreeing on interactions to work on

    • Decision trees for certain interactions

  • Decision Trees

    • Beta blocker+Epinephrine, Warfarin-NSAIDs; K-sparing diuretics/KCl just added

    • Goal is to create decision trees for selected interactions that identify clinical consequences, patient factors, specific drugs involved, specific actions to take

    • Qualtrics survey sent to Standards Team; results pending

      • User-centered definition of clinical consequence

    • Beta blocker+Epinephrine and Warfarin+NSAIDs decision trees

  • Discussion of Suggested DDIs

    • 4 additional PDDIs were agreed upon: https://goo.gl/rYpmjt

    • Several categories were discussed in depth

    • Evidence supporting the interaction is weak

      • Citalopram+Amiodarone interaction will be removed

        • In the case of citalopram+ amiodarone, there is a lot of evidence, but there is little risk associated with the combination

      • Warfarin + antibiotics that don’t inhibit CYP2C9 is a much better example of weak evidence

      • Distinction between two separate ideas:

        • 1 – Trying to identify if there is a problem with a drug pair

          • Most interactions don’t have good evidence

        • 2 – People have studied it, but is there an increased risk of harm?

          • Not in the case of Citalopram+Amiodarone

      • What exactly are we talking about in terms of “evidence supporting the interaction”

        • Different answers based on different questions

          • Is there a potential interaction? Yes

          • Is there a clinically important/pharmacodynamic interaction? No

          • Is there a potential kinetic interaction? Yes, but we don’t know if there’s evidence of it

      • Question of scoping – is this model/are these examples focusing on things with clinical effects, or is it broader?

        • What about pharmacokinetic effects?  What about what patients are interested in?

      • Needs further thought

    • Mechanism is not known

      • Warfarin + fibrates:  does not necessarily apply for the class; removed

      • Pravastatin + paroxetine:  issue of potentially spurious relationship based on data mining; removed

    • Frequency of exposure is/is not available

    • Frequency of adverse effects is/is not available

      • Discussion about definitions of “frequency” and “exposure”

        • What is the purpose of including these as information items?

          • If we keep the frequency is/is not available categories in, they will allow us to fill in gaps with our datasets

      • Really depends on your numerator and denominator

        • Defining the populations; depending on source, you can get lots of different information

      • How do you define when exposure matters?

        • Can we define a cohort of patients, and what kind of exposure matters (when, where, under what setting)?

      • Not a straightforward set of categories; need to define further

      • Pharmacy Quality Alliance (PQA) – quality measures at health system level; creating a list of drug interactions of concern based on large claims database

        • Not included in the list of our stakeholders (focused on researchers, drug compendia editors)

          • Should we include this stakeholder group?

        • Needs further discussion

  • Next Steps

    • Will send out several more questionnaires to move things forward

      • Stakeholder descriptions

      • Drug interaction categories

    • Need to define evidence

      • Absence of evidence vs. evidence showing lack of clinical relevance

      • Kinetic interaction vs. interaction requiring clinical intervention

    • Frequency of exposure data and adverse event data

      • Defining for the minimum information model

      • Their purpose as information items

    • Consider health service quality measures vs. clinically-oriented concerns

Richard D Boyce, PhD
Assistant Professor of Biomedical Informatics
Faculty, Center for Pharmaceutical Policy and Prescribing 
Faculty, Geriatric Pharmaceutical Outcomes and Gero-Informatics Research and Training Program
University of Pittsburgh
[hidden email]
Office: 412-648-9219
Twitter: @bhaapgh