From Skeptic to Advocate: The Power of CQL

Why the Skepticism?

Sylvia Shen, PhD, Senior Director, Analytics, Apervita
Sylvia Shen, PhD, Senior Director, Analytics, Apervita
I admit it...I was a CQL (Clinical Quality Language) skeptic. As an analytics team manager who has deep roots in programming, there is never a shortage of new languages and tools. The constant changes create downstream resource management challenges as staffing is often tied to the preferred tools and the environment that hosts those tools. Languages such as SQL and Python have all proven successful to work reasonably well for executing quality measures. So do we really need to replace the already-working solutions with another new language? To find out, I took off my “manager hat” and scrutinized CQL from an engineer’s point of view.

Before I get into what I discovered, let me provide some background to level set: CQL, or Clinical Quality Language, is a healthcare-specific programming language that targets the specification and execution of electronic clinical quality measures (eCQM) and the area eCQM is closely related to - clinical decision support (CDS).

A newer language, CQL was first released in 2015, but has yet to earn its own Wikipedia page. (Personally I find this ironic since regulatory entities like CMS and NCQA have recognized CQL as the go-to language for quality measurement. But I digress…)

So, back to taking a look at CQL from an engineer’s point of view. Here’s what I learned: CQL is a redesign of traditional programming languages, such as Python and SQL, by decomposing the queries into three components:

  1. A front-end CQL component that’s easy to read, less ambiguous and filled with functions and features that are tailored for quality measures and CDS. Writing measures in CQL are greatly simplified
  2. A CQL to ELM translator that’s powered by open-source algorithms 
  3. A back-end execution engine/interpreter that interprets the logics in ELM into an executable format. The interpreter does most of the heavy lifting behind the scenes to connect the easy-to-read CQL specifications to data models (Quality Data Model [QDM] and FHIR) and ensures that they can be properly executed

Becoming a CQL Advocate

The pros and cons of any tool are all relative based on the goals and associated costs. For example, if I worked for a health plan and I wanted to standardize quality metrics, would CQL be the best tool? From an engineer’s perspective, if getting the measure results is my only goal, greater readability has little significance, so CQL has limited benefits. Considering the amount of effort that is needed to conduct data acquisition, normalization, and set up an execution engine, from an engineer’s perspective, I could get the measure results quicker with a traditional tool like Python or SQL. Additionally, given CQL is a newer language, there is limited training and support, which makes it challenging to troubleshoot should issues occur. That said, at first glance, CQL doesn’t appear to be the best choice if the decision is driven purely by the short-term goal of getting the measure results.

However, when I looked at CQL from the perspective of a regulatory entity, like NCQA, my opinion changed. Hundreds of organizations spend resources (staff and money) to manually code the same HEDIS® measures, which is redundant. The easiest way to automate this process is for NCQA to directly release the executable queries for all the measures and utilize a common data model. Additionally, NCQA would need to have some structure in place to avoid the creation of thousands of versions of the same measure, but also have the flexibility to accommodate changes (i.e. allowable adjustments). CQL perfectly aligns with NCQA’s need to standardize quality measurement, including speed and efficiency. Looking through the lens of NCQA and other regulatory entities, the value of CQL absolutely wins.

Advantages of using CQL for quality measurement:
 > Greater readability
 > Less ambiguity
 > Greater portability
 > Fixed data model
 > Less flexibility in query specification

If NCQA releases all HEDIS measures in CQL, for organizations who need to implement HEDIS measures, like payers, providers and vendors, the measure execution pipeline will be greatly simplified. The time-consuming tasks of coding the measures, annual updates of the measures and code set maintenance will be replaced by directly downloading the CQL queries from NCQA. All that’s needed is setting up the correct data intake/shaping process, CQL to ELM translator and the execution engine. The power (and beauty!) of the design is that most of these steps are a one-time investment, maximizing long-term ROI.

The Bottom Line

Over the last few years, the value of CQL has increasingly been recognized by more and more medical societies and regulatory entities as the go-to language for quality measurement. In fact, NCQA has committed that any new measures will be written in CQL. CMS continues to accelerate the adoption of CQL-based measures in quality programs such as Inpatient Quality Reporting (IQR), Merit-Based Incentive Programs (MIPS), and value-based contracts. In fact, CQL replaced the Quality Data Model (QDM) as the preferred eCQM expression language starting from eCQMs implemented in 2019. It's clear that CQL unifies the logics that are needed to support clinical quality measures and clinical decision support and opens up the possibility of cost reduction for hospitals by streamlining the measure submission process. Given all of its benefits, I’m betting that CQL—and sooner rather than later - will disrupt the quality measurement and CDS industries. And yes, I have no doubt that CQL will also get its well-deserved entry in Wikipedia too.

Want to Learn More?

Contact us to learn more about how CQL can make your transition to digital measurement more efficient and scalable. In the meantime, read more about the future of CQL and its impact on the industry.

HEDIS® is a registered trademark of the National Committee for Quality Assurance (NCQA).


Sylvia Shen, PhD, Senior Director, Analytics, Apervita
As senior director of analytics, Sylvia applies statistics and machine learning algorithms to drive healthcare innovations for value optimization. Her experience covers various healthcare domains including value-based payment, managed care, quality and risk adjustment.

Prior to joining Apervita, Sylvia worked at Verscend and Welltok, where she led the advanced analytics teams to develop analytic solutions to optimize care management triaging and member outreach, risk adjustment, quality improvement and pay for performance strategies.

Sylvia holds a PhD in neuroscience and a master’s degree in applied statistics from Pennsylvania State University. Fun fact: Sylvia is an avid skier and tries to hit the slopes as often as possible.

Recent Posts