In this episode of Ask Dr. Mingle, Dr. Dan Mingle provides an in-depth look at the data required for healthcare quality reporting.
Click play to listen to this episode now:
Question One: Data Requirements for Healthcare Quality Measurement
Kyle asks: “We often find confusion and experience a recurring set of questions about data for quality reporting. What data do you need to calculate quality measures?”
Let’s start with a simple definition of healthcare quality measurement.
It doesn’t matter whether we measure for the Merit-Based Incentive Payment System (MIPS), the APM Performance Pathway (APP), HEDIS (the Healthcare Effectiveness Data and Information Set), or any other healthcare measurement set. Every quality measure asks a variation on a single fundamental question:
“Have our patients received the healthcare we intend them to receive?”
We can define this in terms of an outcome, like the level of blood pressure control or Hemoglobin A1C. Or we can define it as an intervention, like ACE inhibitors in our patients with left ventricular systolic dysfunction, annual influenza vaccination, or potassium level for patients on diuretics.
There are other quality measures, but the most common ones we experience involve the relationship between two numbers: the numerator and the denominator.
The denominator defines and counts, for any given outcome or intervention, the eligible instances in our practices for that outcome or intervention. When I say “instance”:
- It could be patients eligible for a specific intervention, like women over 50, eligible for a mammogram.
- Or visits eligible for intervention, like health care visits during which a provider assesses the patient’s pain level.
The numerator defines and counts, relative to each instance in the denominator, those eligible instances that achieved the designated level of control or received the desired intervention.
Typically, we express quality measure as a fraction—the numerator over (divided by) the denominator:
- Numerator: the count of those who have achieved the outcome or received the intervention.
- Denominator: the count of those eligible for the outcome or intervention.
The denominator, the count of eligible instances, is literally the fraction’s bottom-most number. It is also, figuratively, the foundation of the measure.
So in getting the required data for healthcare quality reporting, we start with the denominator. The denominator comes first. We can only measure performance once we know which and how many patients or which and how many visits the measure applies to.
Every measure has a set of measure specifications. Measure specifications should spell out with no ambiguity:
- The eligibility (or denominator specifications)
- The performance (or numerator specifications)
If there is any ambiguity, it’s not a good specification document.
Question Two: Healthcare Quality Measurement Data Sources
Kyle asks: “Does numerator and denominator data come from the same place?”
It can come from the same place, but more often and more reliably comes from different sources. If there is a single source, that single source is usually the Electronic Health Record (EHR).
We often get most of the data for denominators from the practice management system and most of the data for numerators from the EHR.
Sometimes there is historical data in a separate archive (often, a previously used EHR). In this case, we might combine the data from the legacy EHR archive with that from the new EHR.
Sometimes a reporting database is actively maintained as a preferred data source. Various distinct registries hold a valuable subset of data – the most common is a tumor registry
But there is tremendous variation from practice to practice. At Mingle, we excel at working with practices where they are, utilizing their strengths and working around their weaknesses.
And we can help practices streamline their data abstraction processes to get data manually from paper records.
Your healthcare quality reporting registry needs to know each data source’s strengths, weaknesses, and limitations to create an efficient and effective quality reporting system with you.
Question Three: Sources for Denominator Data
Kyle asks: “Dr. Mingle, you explained that, usually, you get numerator and denominator data for quality reporting from different sources. How do you get denominator data?”
The common theme here is significant variation in the market, and flexibility is critical to serving all comers.
Some clients can generate flat files – a simple data extract from the practice management system that includes all the necessary data.
- We prefer big raw files with a minimum of filtering.
- We see errors and data loss whenever a data set is filtered or otherwise manipulated. It’s best to limit only by a beginning and end date.
Many practice management systems contain canned reports into which simple parameters, such as beginning and end date, can be introduced. If one of those canned reports has all the data that we need, those can be quick, easy, and useful. It doesn’t matter if there are fields included in the report that we don’t need. It’s simple to filter to take what we need and discard the rest.
By far, the most common source of data is 837 files:
- These are files that your practice management system prepares and sends periodically to your claims clearinghouse that go on to insurers for payment.
- 837s usually persist as an archive file in your system and can also be periodically sent to us as the foundation for your denominators.
Question Four: Data Completeness in Healthcare Quality Reporting
Kyle asks: “Several things you explain about denominator data and eligible instances make me think of data completeness. Where does data completeness fit in?”
For the performance years 2023 and 2024, Medicare requires for the Merit-Based Incentive Payment System (MIPS) and the Clinical Quality Measure options of the APM Performance Pathway (APP) that you submit performance data on a least 70% of your eligible instances.
Starting with the 2025 performance year, Medicare has queued the data completeness criteria to go up to 75%.
Also, remember that measures have a minimum denominator count as well. With a few exceptions, the case minimum for most quality measures is 20 eligible instances.
The preliminary count of eligible instances we make from claims analysis is an excellent first-pass determination of:
- The requirements for data completeness
- And the eligibility based on the case minimum
Clinical data will never increase the count if our preliminary count is below the case minimum. The final count can only be the same or lower. You won’t get credit for that measure if you’re below the case minimum on the preliminary count based on claims.
Data completeness is interesting. If the clinical data comes from your Electronic Health Record (your EHR), and there is only one instance of a single EHR in your practice, your data is 100% complete by Medicare’s definition.
It becomes an issue if your data comes from multiple EHRs, or if your data comes from a derived source, such as a tumor registry.
In the first case – all clinical data comes from a single EHR – Medicare allows us to assume the data is complete. For most measures, when we find performance data, performance is met. When we don’t find data, performance is not met. All eligible instances are covered by one or the other possibility.
When the data is from a derived source, like a tumor registry, finding performance data indicates performance is met. But lacking performance data, the question arises:
- Is performance not met?
- Or was there a failure to create an entry in the separate registry?
Medicare requires us to consider the lack of data a lack of reporting, which is detrimental to the reporting rate. The exception to that rule is when the secondary data set explicitly documents a lack of performance.
In more complex systems, where different EHR instances and possibly even paper records exist, getting data from every system can be impossible or cost-prohibitive. Medicare’s data completeness criteria determines how hard we need to try. We don’t need to go any deeper once we access 70% of the cases.
Question Five: Lacking Data and Meeting Data Completeness Criteria
Kyle asks: “If we are lacking data, how do we know when we meet data completeness criteria?”
Now that is an important concept.
We discuss this infrequently, but you need a reliable count of eligible instances to know when your performance measurement exceeds the data completeness requirement.
You need 100% of your eligibility data to know when you have 70% of your performance data.
That’s one of the advantages of our approach in obtaining:
- Denominator first, from the claims data.
- And numerator second, from the clinical data.
Even if we have only a preliminary count of eligible instances from the claims data, it gives us a credible measure of data completeness as we collect the clinical data. When we have data on 70% of eligible candidates, we can stop.
Question Six: The Quality Reporting Process for MSSP ACOs
Kyle asks: “For our Medicare Shared Savings Program (MSSP) clients, who are now subject to the APM Performance Pathway (APP), many have multiple EHRs resident in their participating practices. How do you know that you can serve them? It seems like this is a data aggregation issue, and the data completeness criteria is a severe limitation.”
We’ve been at this since the introduction of the Group Practice Reporting Option (GPRO) in the days of the Physician Quality Reporting System (PQRS).
Many of our larger clients working under a single Tax Id Number (TIN) have multiple EHRs. We have used this approach successfully in all that time. It works. For practices of all sizes and types, we have the scars and the success that proves it. It’s a rare practice that we can’t get claims data from.
With claims data from all practices in the TIN or all practices in the MSSP ACO, we can:
- Deduplicate patients across the systems
- Reliably sort and count patients as eligible candidates for each measure
- Know in which practices each patient is engaged and might have data
- Match the clinical data, where we can get it, to the patients we identified from claims
- Calculate both a reporting rate and a performance rate.
This process will happen more often in the APP for MSSPs, but our methods have been built, proven, and hardened over ten years in the PQRS and MIPS programs.
Send us your value-based care questions!
If you’d like to ask a question about the APP transition, MIPS, ACO quality reporting, or any other Alternative Payment Model, you can reach out to us in three ways:
You can leave your questions in a YouTube comment under any episode of Ask Dr. Mingle.
On LinkedIn, leave your questions in a comment on any of our posts.
And you can reach out directly by sending an email to email@example.com.
For many ACOs, the transition to all-patient, all-payer quality reporting is a significant challenge and often feels like an unnecessary burden imposed by CMS.
We've created our latest PDF guide to help ACOs orient themselves to the challenge of all-patient, all-payer reporting while finding opportunities to thrive in the future.