Uploaded image for project: 'eCQM Issue Tracker'
  1. eCQM Issue Tracker
  2. CQM-2172

Find a way to include ICU Admissions/Transfers in the "inpatient" populations

XMLWordPrintable

    • Icon: EH/CAH eCQMs - Eligible Hospitals/Critical Access Hospitals EH/CAH eCQMs - Eligible Hospitals/Critical Access Hospitals
    • Resolution: Done
    • Icon: Minor Minor
    • Measure, ValueSet
    • CMS71v6/NQF0436, CMS72v5/NQF0438, CMS91v6/NQF0437, CMS102v5/NQF0441, CMS104v5/NQF0435, CMS105v5/NQF0439, CMS107v5/NQFna
    • "ICU Admission/Transfer" encounters are not qualified under "$EncounterInpatient" nor "$EncoutnerInpatientNonElective" and therefore are excluded from many measure populations (most notably Stroke measures).
    • Hide
      ​Currently VTE-1 and VTE-2 use the “Encounter, Performed” datatype with a value set representing ICU in order to identify accurate start and stop times for the ICU stay. We understand this method of modelling ICU may result in the exclusion of patients with an ICU ‘encounter’ from measures using the $EncounterInpatient or $EncounterInpatientNonElective variables, as the value values sets contained within these variables to not include codes representing the ICU encounter.

      We appreciate your submission of proposed solutions. Unfortunately, neither solution proposed will allow for collection of data that satisfies the measure intent. We would like to take this opportunity to further explain the goals of the logic, the issue, and to propose new solutions for CRP consideration.

      Adding ICU encounters to $EncounterInpatient and $EncounterInpatientNonElective (Not Recommended)

      This suggested solution would ensure patients who have an ICU stay are included in the measure populations for measures using these variables. But, this would not meet the developer’s needs for VTE-1 and VTE-2 to capture ICU length of stay separately from the Inpatient Encounter length of stay.


      Our other option would be to look to different datatypes to represent ICU. The chart-abstracted form of VTE-1 and VTE-2 defines ICU admissions with the data element “ICU Admission or Transfer.” The Notes for Abstraction for this data element specify that in order to select “Yes,” or state that the patient had an ICU Admission, there must be a physician/APN/PA order for admission or transfer to the ICU, or that the order may come from a protocol. Based on this guidance, it may be reasonable to represent the ICU admission with an ‘order’ datatype rather than as an Encounter. The QDM includes the following options for ‘order’ datatypes which we believe would align with the measure intent. We are requesting your feedback on three items:

      1. Our suggestion to represent ICU as an order rather than as an encounter.

      2. The appropriate QDM datatype to use to represent ICU as an order

      3. What perceived implications to measure rates exist, should we represent ICU as an order rather than as an encounter.

      Our options for representing ICU as an order include the following datatypes. For your consideration, we are providing the datatypes, along with their QDM description.

      - Intervention, Order

      o Intervention represents a course of action intended to achieve a result in the care of persons with health problems that does not involve direct physical contact with a patient. Examples include patient education and therapeutic communication.

      o Data elements that meet criteria using this datatype should document a request to perform the intervention indicated by the QDM category and its corresponding value set.

      o NOTE: The start and stop datetime of Intervention, Order reflects the “author time” of the record in QRDA. This corresponds to when the order was signed.

       

      - Procedure, Order

      o Procedure is derived directly from HL7 and Canada Health Infoway: “An Act whose immediate and primary outcome (post-condition) is the alteration of the physical condition of the subject. … Procedure is but one among several types of clinical activities such as observation, substance-administrations, and communicative interactions … Procedure does not comprise all acts of [sic] whose intent is intervention or treatment.”16 A procedure may be a surgery or other type of physical manipulation of a person’s body in whole or in part for purposes of making observations and diagnoses or providing treatment. 17

      o Data elements that meet criteria using this datatype should document a request for the procedure indicated by the QDM category and its corresponding value set.

      o NOTE: The start and stop datetime of Procedure, Order reflects the “author time” of the record in QRDA. This corresponds to when the order was signed.

       

      - Encounter, Order

      o Note, we believe use of “Encounter, Order” would perpetuate the current issues introduced by representing ICU as an encounter, and do not recommend this option as a solution.

       Pro for selecting an option above:

      · We will be able to capture patients that are transferred to the ICU during the timeframe the measure is looking for via an order.

      Con for selecting an option above:

      · We may not be using the datatype as outlined in the Quality Data Model, this may cause confusion as to why we are utilizing the specific datatype. ​

      Logic below:

      CMS108/VTE1

      •Initial Population =
      ◦AND: Age>= 18 year(s) at: Occurrence A of $EncounterInpatient
      ◦AND NOT: Union of:
      ◾Intersection of:
      ◾Occurrence A of $EncounterInpatient
      ◾"Encounter, Performed: Encounter Inpatient" satisfies any:
      ◾(diagnosis: Obstetrics)
      ◾(diagnosis: Venous Thromboembolism)
      ◾(diagnosis: Obstetrics VTE)
      ◾Union of:
      ◾"Diagnosis: Obstetrics"
      ◾$DiagnosisVTE
      ◾starts during Occurrence A of $EncounterInpatient
      ◾Union of:
      ◾"Diagnosis: Obstetrics"
      ◾$DiagnosisVTE
      ◾starts during ("Encounter, Performed: Emergency Department Visit" <= 1 hour(s) ends before or concurrent with start of Occurrence A of $EncounterInpatient )
      •Denominator =
      ◦AND: Initial Population
      •Denominator Exclusions =
      ◦OR: Intersection of:
      ◾Occurrence A of $EncounterInpatient
      ◾"Encounter, Performed: Encounter Inpatient (length of stay < 2 day(s))"
      ◦OR: "Encounter, Performed: ICU Admission or Transfer (length of stay >= 1 day(s))" <= 1 day(s) starts after or concurrent with start of Occurrence A of $EncounterInpatient
       

      CMS190/VTE 2

      •Initial Population =
      ◦AND: Age>= 18 year(s) at: Occurrence A of $EncounterInpatient
      ◦AND NOT: Union of:
      ◾Intersection of:
      ◾Occurrence A of $EncounterInpatient
      ◾"Encounter, Performed: Encounter Inpatient" satisfies any:
      ◾(diagnosis: Obstetrics)
      ◾(diagnosis: Venous Thromboembolism)
      ◾(diagnosis: Obstetrics VTE)
      ◾Union of:
      ◾"Diagnosis: Obstetrics"
      ◾$DiagnosisVTE
      ◾starts during Occurrence A of $EncounterInpatient
      ◾Union of:
      ◾"Diagnosis: Obstetrics"
      ◾$DiagnosisVTE
      ◾starts during ("Encounter, Performed: Emergency Department Visit" <= 1 hour(s) ends before or concurrent with start of Occurrence A of $EncounterInpatient )
      •Denominator =
      ◦AND: Initial Population
      ◦AND: First: "Occurrence A of Encounter, Performed: ICU Admission or Transfer" during Occurrence A of $EncounterInpatient​

      Modeling ICU as a Facility Location: Limitation in Timing Accuracy (Not Recommended)

      We initially considered changing the measure logic so that ICU admissions and transfers are defined as an attribute of “Facility, Location.” However, attributes do not have their own effective times. For VTE-1 and VTE-2, we are specifically looking for ICU length of stay.

      · If we modeled ICU as a facility location, like so: “Encounter, Performed: Encounter Inpatient (facility location: ICU)”

      o And then evaluated (length of stay < 2 day(s))

      · The measure would look to the start and end of the inpatient encounter, NOT the start and end time of the ICU location.

      Therefore, we do not believe this is a solution to the issue and do not suggest the CRP recommend this approach.

      Why change is needed:

      The current logic is impacting the measure rate by not including patients that should be in the measure due to how the logic is written. In addition, as outlined in the initial concern of this ticket, organizations are having to do a workaround to capture patients that should be included in the measure. We found the logic was an issue this year due to the data we had available where previously we did not have this data available to know the “encounter, performed” was not capturing the intended patients
      Show
      ​Currently VTE-1 and VTE-2 use the “Encounter, Performed” datatype with a value set representing ICU in order to identify accurate start and stop times for the ICU stay. We understand this method of modelling ICU may result in the exclusion of patients with an ICU ‘encounter’ from measures using the $EncounterInpatient or $EncounterInpatientNonElective variables, as the value values sets contained within these variables to not include codes representing the ICU encounter. We appreciate your submission of proposed solutions. Unfortunately, neither solution proposed will allow for collection of data that satisfies the measure intent. We would like to take this opportunity to further explain the goals of the logic, the issue, and to propose new solutions for CRP consideration. Adding ICU encounters to $EncounterInpatient and $EncounterInpatientNonElective (Not Recommended) This suggested solution would ensure patients who have an ICU stay are included in the measure populations for measures using these variables. But, this would not meet the developer’s needs for VTE-1 and VTE-2 to capture ICU length of stay separately from the Inpatient Encounter length of stay. Our other option would be to look to different datatypes to represent ICU. The chart-abstracted form of VTE-1 and VTE-2 defines ICU admissions with the data element “ICU Admission or Transfer.” The Notes for Abstraction for this data element specify that in order to select “Yes,” or state that the patient had an ICU Admission, there must be a physician/APN/PA order for admission or transfer to the ICU, or that the order may come from a protocol. Based on this guidance, it may be reasonable to represent the ICU admission with an ‘order’ datatype rather than as an Encounter. The QDM includes the following options for ‘order’ datatypes which we believe would align with the measure intent. We are requesting your feedback on three items: 1. Our suggestion to represent ICU as an order rather than as an encounter. 2. The appropriate QDM datatype to use to represent ICU as an order 3. What perceived implications to measure rates exist, should we represent ICU as an order rather than as an encounter. Our options for representing ICU as an order include the following datatypes. For your consideration, we are providing the datatypes, along with their QDM description. - Intervention, Order o Intervention represents a course of action intended to achieve a result in the care of persons with health problems that does not involve direct physical contact with a patient. Examples include patient education and therapeutic communication. o Data elements that meet criteria using this datatype should document a request to perform the intervention indicated by the QDM category and its corresponding value set. o NOTE: The start and stop datetime of Intervention, Order reflects the “author time” of the record in QRDA. This corresponds to when the order was signed.   - Procedure, Order o Procedure is derived directly from HL7 and Canada Health Infoway: “An Act whose immediate and primary outcome (post-condition) is the alteration of the physical condition of the subject. … Procedure is but one among several types of clinical activities such as observation, substance-administrations, and communicative interactions … Procedure does not comprise all acts of [sic] whose intent is intervention or treatment.”16 A procedure may be a surgery or other type of physical manipulation of a person’s body in whole or in part for purposes of making observations and diagnoses or providing treatment. 17 o Data elements that meet criteria using this datatype should document a request for the procedure indicated by the QDM category and its corresponding value set. o NOTE: The start and stop datetime of Procedure, Order reflects the “author time” of the record in QRDA. This corresponds to when the order was signed.   - Encounter, Order o Note, we believe use of “Encounter, Order” would perpetuate the current issues introduced by representing ICU as an encounter, and do not recommend this option as a solution.  Pro for selecting an option above: · We will be able to capture patients that are transferred to the ICU during the timeframe the measure is looking for via an order. Con for selecting an option above: · We may not be using the datatype as outlined in the Quality Data Model, this may cause confusion as to why we are utilizing the specific datatype. ​ Logic below: CMS108/VTE1 •Initial Population = ◦AND: Age>= 18 year(s) at: Occurrence A of $EncounterInpatient ◦AND NOT: Union of: ◾Intersection of: ◾Occurrence A of $EncounterInpatient ◾"Encounter, Performed: Encounter Inpatient" satisfies any: ◾(diagnosis: Obstetrics) ◾(diagnosis: Venous Thromboembolism) ◾(diagnosis: Obstetrics VTE) ◾Union of: ◾"Diagnosis: Obstetrics" ◾$DiagnosisVTE ◾starts during Occurrence A of $EncounterInpatient ◾Union of: ◾"Diagnosis: Obstetrics" ◾$DiagnosisVTE ◾starts during ("Encounter, Performed: Emergency Department Visit" <= 1 hour(s) ends before or concurrent with start of Occurrence A of $EncounterInpatient ) •Denominator = ◦AND: Initial Population •Denominator Exclusions = ◦OR: Intersection of: ◾Occurrence A of $EncounterInpatient ◾"Encounter, Performed: Encounter Inpatient (length of stay < 2 day(s))" ◦OR: "Encounter, Performed: ICU Admission or Transfer (length of stay >= 1 day(s))" <= 1 day(s) starts after or concurrent with start of Occurrence A of $EncounterInpatient   CMS190/VTE 2 •Initial Population = ◦AND: Age>= 18 year(s) at: Occurrence A of $EncounterInpatient ◦AND NOT: Union of: ◾Intersection of: ◾Occurrence A of $EncounterInpatient ◾"Encounter, Performed: Encounter Inpatient" satisfies any: ◾(diagnosis: Obstetrics) ◾(diagnosis: Venous Thromboembolism) ◾(diagnosis: Obstetrics VTE) ◾Union of: ◾"Diagnosis: Obstetrics" ◾$DiagnosisVTE ◾starts during Occurrence A of $EncounterInpatient ◾Union of: ◾"Diagnosis: Obstetrics" ◾$DiagnosisVTE ◾starts during ("Encounter, Performed: Emergency Department Visit" <= 1 hour(s) ends before or concurrent with start of Occurrence A of $EncounterInpatient ) •Denominator = ◦AND: Initial Population ◦AND: First: "Occurrence A of Encounter, Performed: ICU Admission or Transfer" during Occurrence A of $EncounterInpatient​ Modeling ICU as a Facility Location: Limitation in Timing Accuracy (Not Recommended) We initially considered changing the measure logic so that ICU admissions and transfers are defined as an attribute of “Facility, Location.” However, attributes do not have their own effective times. For VTE-1 and VTE-2, we are specifically looking for ICU length of stay. · If we modeled ICU as a facility location, like so: “Encounter, Performed: Encounter Inpatient (facility location: ICU)” o And then evaluated (length of stay < 2 day(s)) · The measure would look to the start and end of the inpatient encounter, NOT the start and end time of the ICU location. Therefore, we do not believe this is a solution to the issue and do not suggest the CRP recommend this approach. Why change is needed: The current logic is impacting the measure rate by not including patients that should be in the measure due to how the logic is written. In addition, as outlined in the initial concern of this ticket, organizations are having to do a workaround to capture patients that should be included in the measure. We found the logic was an issue this year due to the data we had available where previously we did not have this data available to know the “encounter, performed” was not capturing the intended patients
    • Hide
      ​Providing the feedback received, TJC recommends no changes be made at this time.
      Show
      ​Providing the feedback received, TJC recommends no changes be made at this time.

      1) ICU Admissions/Transfers are defined in the CQM specifications under the "Encounter, Performed" data category.
      1a) This is the same data category where "$EncounterInpatient" and "$EncounterInpatientNonElective" are defined

      2) The Value Sets for "$EncounterInpatient" and "$EncounterInpatientNonElective" do not include any of the codes from the Value set for ICU Admission/Transfer

      3) Since an encounter can only occupy one "Encounter, Performed" status at any given time, when an encounter is qualifying for "Encounter, Performed: ICU Admission or Transfer", that encounter cannot be qualifying at the same time for "$EncounterInpatient" and "$EncounterInpatientNonElective"

      As a result, ICU Admission/Transfer encounters are being excluded from many measures, since they are being recognized as "inpatients". I don't believe this was the intent of the measures.

      Proposed Solutions:
      A) Add ICU Admission/Transfer SNOMED Codes to the "$EncounterInpatient" and "$EncounterInpatientNonElective" Value sets.
      OR
      B) Change measure logic so that ICU Admissions/Transfers are defined by "Facility, Location", rather than "Encounter, Performed"

            JLeflore Joelencia Leflore
            jeffrey.wiles Jeffrey Wiles (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            17 Start watching this issue

              Created:
              Updated:
              Resolved: