Chair and Scribe: Floyd Eisenberg

Meeting Access

NOTE: Scroll to the bottom of this page for attendance 

Minutes/notes from:

TopicAgenda item description - NOTE change to Wednesday mornings 10-11 AM ET Discussion
Connectathon Planning and Working Session

2021-05 Clinical Reasoning

(Discussion led by Bryn Rhodes and Gayathri Jayawardena)

Connectathon kickoff provided by Bryn Rhodes.
Cumulative Medication Duration Example

Current methodology for CMD:

Plans based on March 24 call: Add MedicationDispense.whenPrepared since MedicationDispense.whenHandedOver may not be populated but MedicationDispense.whenPrepared will be populated (FHIR-31609) to add MUST SUPPORT for .whenPrepared in QI-Core

Measure examples for discussion:

For review at this April 21, 2021 Call with CQI and Pharmacy representatives - CQL examples for each of the three scenarios identified at the March 31 call:

  1. Start time for cumulative medication duration using MedicationDispense (recap):  The group re-capped the discussion from March 24 indicating that the MedicationDispense.whenHandedOver time may not be populated but the MedicationDispense.whenPrepared would be populated. A question arose during the CQI WG call last Friday (March 26) in that the whenPrepared time does not assure a patient has received the medication, hence, the preference for whenHandedOver. However, if the latter item has no time value (i.e., is null) the start time for the covered days will not be available. The group discussed the fact that MedicationDispense.status options include: preparation | in-progress | cancelled | on-hold | completed | entered-in-error | stopped | declined | unknown. A status = completed will always be present if the patient has received the medication. Prior to the patient picking up the medication status = in-progress, and medications never picked up should show a status = cancelled, stopped, declined, etc.  Therefore, a CQL expression that requests MedicationDispense.whenHandedOver and secondarily MedicationDispense.whenPrepared if whenHandedOver is null will provide a start time for the cumulative medication duration calculation, but the status should always be completed to assure the patient received the medication. The CQI WG did approve FHIR-31609 to require MUST SUPPORT for MedicationDispense.whenPrepared during the March 26, 2021 call.  This discussion today should resolve the concern raised by measure developers about how to use whenPrepared yet assure the medication has been given to the patient.
  2. Handling MedicationDispense events that overlap - The group discussed the scenario in which a second dispensing event occurs before the end of the period during which the previous dispensing events supply duration has been exhausted. Based on discussion, it is common for prescriptions to be dispensed and for patient to pick up their medications before the supply from a previous dispensing is fully exhausted. This scenario is quite common especially with mail-order 90-day prescription refills to avoid patients running out of supply and missing days of medications. Therefore, all cumulative medication duration calculations should begin with the timing of the first dispensing event and assume that each subsequent dispensing event active use begins at the end of the supply duration of the previous dispensing event (i.e., supply duration is added to the end of the previous supply duration):
    1. DO use the date for the first dispensing (when it starts) then add the days (daysSupplied) from each dispense rather than using the date of the dispensing. Count from the end of the period covered by the previous dispense.   Note - if the subsequent dispense .whenHandedOver (or .whenPrepared) occurs after the end of the .daysSupplied from the previous dispensing event, the result is a gap in days covered.
    2. DO NOT shorten the coverage period for the previous dispensing event and start again from the start of the second dispensing event (overlap).
  3. Combined oral and parenteral therapy to determine CMDThe team further discussed the scenario presented by the schizophrenia active treatment measure. The measure example has not been converted to an eCQM and it is based on claims data, specifically, dispensing events and J codes to indicate that IM medications have been administered in a clinical practice setting. The expectation is to covert the measure to one using clinical data.
    1. Data elements (summarized):
      1. Denominator - patients with schizophrenia or schizoaffective disorder diagnosis
      2. Numerator - patients who received active medication treatment for schizophrenia or schizoaffective disorder. Active treatment may either be:
        1. Oral medication in one of 4 classes.
          1. While clinical validation is required to be sure, the team assumed active treatment may include more than 1 of these classes concurrently. Thus, a dispensing overlap should not start the count of days covered from the end of the other dispensing event unless the dispensing events are for the same medication. For different medications dispensed, coalesce the days covered.
        2. Parenteral (IM) medication for which a single dose may be effective for one of three time periods - 14 days, 28 days, or 30 days.
          1. While clinical validation is required to be sure, the team assumed active treatment may include concurrent therapy with an oral medication that starts before the effective period for the parenteral medication ends. Thus the logic expression should collapse the IM administration coverage period and the oral medication dispensing event.
      3. Given the conclusions listed above, the cumulative medication duration should be relatively straightforward assuming that each class of oral agents is identified as a unique data element and each parenteral agent grouping (by effective period) is identified as a unique data element.

See reference information for considerations about CQL expressions based on the information discussed on March 31 (cell to the left in this table): Cumulative Medication Duration Considerations 21April2021.

Bryn presented the changes thus far in the CQL:

Cleaned up timing events. Timing can be an event, a repeat or a code.  

  • Melva suggested that "code" might be represented on a paper prescription but not necessarily on an electronic submission. John Hate suggested that codes such as bid are used to reference different times on each nursing unit in a hospital. The clinician is not familiar with the specific timing requirements for each unit. On community orders, there is no specific timing for codes such as BID.  In the hospital, BID means the hospital pharmacy sends the medication twice and the nursing unit determines the timing. Rob McClure questioned what NCPDP messages would include - e.g., is a code used. The code may also be present only in text and not as a code. Thus, the CQL expression needs to account for both a code for timing schedule. If a code is provided it is understood to be a complete statement and either the code or the structured data can be used to determine timing as repeat.bounds to indicate the frequency per day.
  • Added condition to use .whenPrepared when .whenHandedOver is not available for MedicationDispense with status = completed.
  • Administration - previously used Administration.effective which only specifies the administration time. Now it uses a function but could use a specific value sets that provide specific therapeutic days. OR could provide a code system supplement that provides therapeutic duration and can be accessed for the calculation.  MedicationKnowledge does not seems to have the right element to address duration (has only kinetics.halfLife) - Bryn will create a tracker on the MedicationKnowledge resource to consider addressing therapeutic duration.
  • The discussion was very productive and will continue next week. 
Adjourned at 10:59 AM ET


Supporting Documents

Outline Reference

Supporting Document

Minute Approval

  • No labels