In the first "Anatomy of a Measure" post we discussed the three components of a clinical quality measure (CQM): measure logic, value sets and attribution. Today we're going to jump into measure logic – the "ANDs" and "ORs" that constitute a CQM.
UDS expresses measure logic in narrative form, so the specification is a long .PDF with paragraphs and bullets that explain how measures work. Check out this snippet on the Adult Weight Screening and Follow-up measure:
Most of the time this is a fine way to explain how a measure works, but, as we've noted before, ambiguity can be a killer! Can you spot the ambiguity in the specification?
Here are a few things I see:
- What constitutes a visit? Fortunately, that's explained earlier in the UDS manual.
- What are "normal parameters" for a BMI? That's explained on the next page, though it probably could've been included in the numerator paragraph.
- The denominator refers to the patient having had a "medical visit" during the reporting year, but the numerator refers to the patient's "most recent visit" without clearly explaining that it's medical. While it’s probably safe to assume the visit is a medical one, the wording is ambiguous.
- When should the follow-up plan be documented? The spec only states: "a follow-up plan is documented.” Does that mean that it was documented during the reporting year? Sometime after the out-of range-BMI? The spec is not clear.
That last one is tricky, and has been the topic of many support tickets. We actually look for follow-up plan documentation within six months of the most recent visit (ie. the same time frame that we check for the BMI). This may seem strange because we're looking for a follow-up plan within a six-month period prior to the most recent visit. Why?
Before we that question, let's examine how the Meaningful Use eCQMs express measure logic. Each eCQM is defined via a Health Quality Measures Format (HQMF) document. HQMF is a standard way of defining measures that is maintained by the standards organization HL7. These HQMF documents are ideal for programmers because they thoroughly describe the measure logic and provide explicit references to value sets for every data element in the measure. But if you've ever seen a "human readable" version of an HQMF document, you may argue about how much clarity they actually provide!
So let's get back to adult weight screening and follow-up. Here's the human readable HQMF: (click image to enlarge)
It might take a while to digest, but the spec leaves no ambiguity. Take note of the follow-up timeframe: it’s within six months of the most recent encounter! We decided to follow the same logic for our UDS measure – primarily for consistency's sake. We generally do our best to consolidate measure logic across specifications, and it looks like others are following that trend. HRSA recently announced that UDS would begin to align with the eCQMs.
I'll admit that there's a lot going on here, and that the learning curve can be steep. But if we want standardized clinical quality measurement, we need good, unambiguous specifications. They must be accessible to a large audience (like the UDS spec) and they must provide the level of detail programmers expect (like the eCQMs). There's a lot of grey area in healthcare, but clinical quality measurement logic shouldn't be part of it.
Note:
Still confused about why the measure checks for the follow-up in the six months prior to the visit? So are we! There is a ticket open in the ONC's Jira tracker that seeks clarification on why the measure works this way.
Eric Gunther is an engineer at Azara Healthcare.