Reopening to move the status to 'Resolved' until test deck can be updated for the next release.

Howard,

While updating the test patient data for the Cypress 2.5 release, two patients were noted to have a time difference of minutes between the time of depression diagnosis and the time of depression screening. The patient that you brought to our attention was one of them.

We believed that we had adjusted the test data in both patients to evaluate the difference of the two data elements to the day. However upon closer inspection, we realized that the patient you mentioned was not updated. That patient continues to have a depression diagnosis one hour starts before the start of the depression screening (i.e. depression diagnosis on 3/1 @ 3pm and depression screening on 3/1 @ 4 pm).

We intend to move the depression screening to 7/1 @ 2pm for the next bundle release (during the encounter that occurs on 7/1). This change will provide a 4 month time span between diagnosis and screening.

We apologize for not addressing the timing concerns in the specific patient that you cited in your JIRA posting. The ATLs will be provided guidance to continue to accept both outcomes for the patient – the patient can be included in the denominator or excluded from the denominator - until the next bundle is released.

Thank you for your feedback,
Sharon

I still question why this statement was added and what the significance is. "Data criteria will continue to be evaluated to the minute unless otherwise specified." I read that to mean that the basic premise of the issue is incorrect, but you are going to yield on this particular example anyway.

Howard,

My apologies for the delay in getting back to you. This issue required a lot of discussion between ONC and the Cypress team. I have posted the guidance above, and will update the issue when a new measure bundle is released. If you have any questions, please let me know.

Can we please have resolution of thsi issue, which is marked appropriately with Blocker priority?

Thanks Rob. I'm not sure I'm following you here, maybe a phone call would help. In this particular case, I do think the scenario isn't realistic clinically, so we have a weaker leg to stand on when arguing a principle. During an encounter, I might administer a test, which convinces me of depression. I might add depression to the problem list before I enter in the test result - that would exclude the patient. In addition, it's rare that anyone would document date of onset of depression to the minute. When capturing an item on a problem list, there is the author time, which is very precise, and there is the clinician's assertion of date of onset, generally only to the day. So, depending on what the "time" in the test case means, it should be more or less precise, but to resolve this particular issue, my suggestion would be to have the diagnosis on a different day altogether from the administration of the test.

Isn't realistic clinically from their system standpoint, that's not to say that all systems act the same way that theirs does. If this is an overall clinically unrealistic scenario for the majority of systems then the record can change.

What I cannot let go by without challenge is the logic that because CDA allows for variability in the precision of time fields that data can be dropped on the floor and ignored. That is a flawed argument to make. CDA (QRDA in the instances we are talking about), defines the means to convey the information about the patient, thats it. One cannot read into the QRDA and transpose meaning back to the CQM's and say well CDA says this so I don't have to do this in the measure, thats just wrong, and if I let the argument go here than I'ld have to let it go everywhere that someone uses variability in CDA as an argument as to how their system can ignore whatever they want. One could say that their system across the board stores dates with only the precision to the day and seeing how CDA times are variable in nature all those CQMs that have data criteria that measure the difference between events in minutes or hours are invalid. Thats an extreme example but its the same logic so I can't let that go.

Hi, I'm chiming in at Howard's request. I'm a bit fuzzy on the concern, so let me try to summarize my understanding: While the scenario isn't realistic clinically, given that the times are provided to the minute, then the test patient would meet the exclusion criteria.

Ben,

I'm familiar with the way time elements work in CDA. You seem to be using the argument that because time elements can be represented to varying precision that you can just use whatever precision you feel like. The test records that you are given have known times and you are dropping the precision down from what it is stated in the records i.e the known precision. The ability of time elements to have varying precision is representative of what time elements are capable of. I think that is quite a bit different than stripping a portion of a time element off to change the precision it was originally in.

Rob, you are correct - that was from R2. My apologies.

In R1 it's the same, but it's harder to find. Basically, HQMF uses the HL7 data type specs.

So, according to that data type spec, timestamps can be at whatever precision:

A calendar expression can be of variable precision, omitting parts from the right.

This agrees with Bob's statement in CQM-648 that "effectiveTime need only be represented to the precision known (e.g. the month, year, or decade)".

So we have that TS can have varying precision. But now we need to figure out how to compare TS of varying precision, which is tricky. This article helped me understand it; one relevant quote:

the TS value of 20100404 implies the day 4-Apr 2010, and the implicit time is from 00:00 to 23:99 on that day (actually, [201004040000;20100405000[)

Ben,

I believe that ActCriteria.temporallyRelatedInformation is in HQMF R2 which is not part of MU2. I cant find any reference to it in HQMF R1.

I also do not see how [CQM 648] backs up your assertion. The only place in the CAT I documents that reference the US Realm Date Time Constraints which is the constraints that you make reference to in [CQM 648] are in the a few sections in the header only, not in the measure data itself. I do not see anything in that thread that states you can ignore any level of date precision

Hey Robert,

I think there is some confusion. While it's true that the measure compares to the minute, the data can be at whatever precision we record. See CQM-648.

The HQMF spec gives some info on how to deal with data having varying levels of precision in the section about ActCriteria.temporallyRelatedInformation. You can see that if the diagnosis is stored with day-level precision, it's equivalent to the interval [3/1/12 12 AM, 3/2/12 12 AM), and hence not before 3/1/12 11 AM. So the patient would not be excluded.

Let me know if you would like more information about why it makes sense to store diagnoses to day-level precision, or if there's anything else I can help with.

Howard,

I do not believe your interpretation of HQMF to be correct. If you look at Appendix C in the CQM logic guidance document (http://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/Downloads/2014_eCQM_LogicGuidance_June2013.pdf) you will see that time interval calculations that do not have a level of granularity specified default to minute based comparrisons, not days as you allude to be the case. As such the patient record itself is valid and the calculation results that are expected are valid as well.

Hi Renee. The problem here is one of time interval precision. The test patient could have been configured with the problem start time set to days before the encounter, weeks before, or months/years before. Any of those would have reflected a realistic scenario. But instead it was configured with the start time 1 hour before the encounter/screening. First, this is not realistic clinically. Second, by doing so, Cypress is basically saying that we must make a time distinction down to the precision of minute/hour, we have to evaluate the diagnosis as before the encounter, when the HQMF allows us to say that the diagnosis and the encounter are concurrent and not therefore not exclude the patient.

Hello Howard, thank you for your comment. I was hoping to gain a little more information from you. How does this time specification differ from any other Cypress patient that has a time asssociated with a concept? Just trying to gain a little more clarification on why this patient is being brought forward as your example and how we should consider it in relation to others. Thank you, Renee