PREV-10 - Tobacco Screening - 248 minimum sample logic

[Scott Fitzpatrick]

We ran tests on PREV-10 via the API, trying to decipher when the minimum sample of 248 is met for the various populations.

Test #1

Submission: We submitted a synthetic/test sample of 320 benes from the PREV-10 ranked sample - we set 20 benes as tobacco-users, 15 with interventions

Results: In both the API "score" endpoint and the CMS Front-End UI, it stated the 248 minimum was met with a denominator of 20 and numerator of 15, 75% score.

Test #2 

Submission: We submitted another synthetic/test sample of all 616 benes from the PREV-10 ranked sample -  we set 20 benes as tobacco-users, 15 with interventions

Results: Again, in both the API "score" endpoint and the CMS Front-End UI, it stated the 248 minimum was met with a denominator of 20 and numerator of 15, 75% score.

For both tests, the API "score" endpoint indicated that for population 2, ALL submitted benes were part of the eligible population -  320 in the first test, and 616 in the second.   

For population 2, the API "score" endpoint indicated all non-smokers were being considered exceptions, 300 in the first test, and 596 in the second.

This exception logic seems like it might be wrong.   Based on the measure specs, it seems non-smokers should not be considered part of the eligible population at all.   

The spec doc states "[Denominator] Population 2: Equals Initial Population who were screened for tobacco use and identified as a tobacco user." 

The YouTube video from CMS at the 20 to 30 second mark seems to back up this interpretation.  That is, even though these are all lumped into one measure spec, the populations should all be evaluated separately to determine eligibility.  Does it state somewhere in the specs that non-tobacco users are exceptions for population 2?  

The "GUIDANCE" section of the specs comes close to saying this:  "If tobacco use status of a patient is unknown, the patient does not meet the screening component required to be counted in the numerator and should be considered a measure failure."

It almost seems like the measure specs you are trying to get to are:

     Denominator:  Equals Initial Population who were NOT screened for tobacco use OR screened for tobacco use and identified as a tobacco user.

     Numerator: Patients who received tobacco cessation intervention

     Exceptions:  the usual medical reasons (deceased, hospice, etc)

Perhaps on the next CMS Web Interface webinar we could get a slide of some scenarios showing some numbers of when the 248 minimum is considered met.

Documentation we reviewed for our tests:

    - 2018_WI_PREV10.pdf

    - 2018_measure_226_registry.pdf

    - PREV-10 YouTube video 

    - https://ecqi.healthit.gov/system/files/ecqm/measures/CMS138v6_0.html

Thanks,

Scott

[laura....@semanticbits.com]

Hi Scott,

For population 2, patients who are screened for tobacco use and not identified as tobacco users are not going to be skipped and replaced (denominator exclusions), they will be counted like a denominator exception - marked as complete and removed from the denominator for population 2.  See page 18 on the measure specification for population 2 work flow.  

Thanks,

Laura

[Scott Fitzpatrick]

Laura,

Thanks for the clarification - non-tobacco users are exceptions for population 2. 

Thanks,

Scott Fitzpatrick

P.S.

But just to be clear, this is not stated explicitly anywhere in the document specs.    I believe the screenshot you provided is normally used for Denominator Exclusions, not exceptions. And the screenshot doesn't say anything about these being exceptions.

Exceptions are usually shown in the "Measure Flow;" not the "Measure Confirmation Flow."    Additionally, the spec docs clearly state "[the measure diagrams] should not be used as a substitution for the measure specification."  And the measure spec does not say anything about non-tobacco users being exceptions.

And the NQF 0028 spec definitely does not consider non-tobacco users as denominator exceptions.