Great! I'll close this ticket again.
Best,
Lizzie

Ok, it makes sense. These test cases have been corrected. Thank you!

Hi Lynn,

Bonnie appears to be calculating correctly.

The logic for Strat 2 looks like it was modified to rely on "Inpatient Encounter". This appears to have additional criteria in it. For example, that the occurrence is less than or equal to 120 days. This is not the case for test case "IPPFail Enc>120Days". Similarly for "IPPFail EncAfterMP", this does not satisfy the "Inpatient Encounter" logic.

If you don't want this characteristics included in the strat 2 definition, you will need to modify that definition to not rely on "Inpatient Encounter".

Let me know if this answers your question.

Best,
Lizzie

Reopening this ticket as the functionality is not performing in CQL Bonnie as expected based on previous issue discussions. Two test cases IPPFail, Enc>120Days and IPPFail, EncAfterMP are expected to meet strat 2 criteria for no ICU encounter even though they do not meet the IP criteria. But the CQL calculation is not showing them as meeting strat 2. A similar test case, IPPFail, 2Young does calculate as meeting strat 2 even though it does not meet the IP criteria. The only difference being that the 2Young case was able to calculate the IP criteria, whereas in the others it was not. Can you please verify if this is related to a calculation issue based on the IP and if so, how can we resolve this?

Yes, this makes sense. Thank you for the thorough explanation this will aid in explaining to the measure developer. thank you.

Hi Angela,

This is due to the stratification logic within the QDM-Logic version of the measure, as well as how Bonnie handles stratifications. You'll notice that if you add an ICU encounter during that encounter, the Strat 1 is satisfied with a count of 1 even though the IPP has a count of 0. You'll also notice that if you make this test case pass the IPP, Strat 2 would show a count of 1.

There are several things happening here to cause this behavior.

1. The definition of stratification 1 does not return any encounters when it is satisfied.

 AND : 
     AND NOT: Encounter, Performed: Icu Admission Or Transfer during  $EncounterInpatient

When this is satisfied, there is no "Encounter, Performed: Icu Admission Or Transfer", so the count is 0

2. Bonnie tries to account for this, but fails when the IPP is also not satisfied
A similar situation must have occurred previously when the IPP was satisfied but STRAT was returning 0. This appears to have been handled by saying, "the STRAT should be at least the count of the IPP", so the following logic was added to Bonnie:

if stratification is true and stratification count is less than IPP count, then set stratification count to be equal to IPP count.

However, when the IPP is not satisfied, we don't have a way of knowing what this count should be.

3. The stratification definition here looks like it may be incorrect
It appears the intent of this stratification definition is to find cases where an inpatient encounter occurred, but there was no ICU visit during that encounter. As it's written now, this will return true even without an inpatient encounter occurring. If we are reading this correctly, this should be rewritten along the lines of:

 AND : 
     AND: $EncounterInpatient
     AND NOT: Encounter, Performed: Icu Admission Or Transfer during  $EncounterInpatient

I believe if the logic were rewritten in this way, this stratification would return a count of 1, so would show "1" even with the IPP showing "0".

4. The CQL logic reflects the definition described in (3) rather than in (1).
The CQL has:

define "Stratum 2":
    ["Encounter, Performed": "Encounter Inpatient" ] Encounter without ["Encounter, Performed": "ICU Admission or Transfer" ] ICU
        such that ICU.relevantPeriod during Encounter.relevantPeriod

This is saying, "return an encounter inpatient that does not have an ICU admission/transfer that occurs during that encounter". This parallels the suggestion in item (3).

Let us know if you are satisfied with this response, or if you would still like us to take an action.

Thank you,
Lizzie

Hi Angela,

Thank you for reporting this issue. We will look into it.

Best,
Lizzie