USCDI as a Critical Framework for Data Quality

Health data needs are shifting, including the role of standards to create new opportunities, as is the case for advancing data quality through the United State Core Data for Interoperability (USCDI). Ever since HITECH spurred a dizzying rise of health information technology, the industry has lamented over what could be. Now, thanks to the work of ONC, the foundation for a nationwide vocabulary standard has been formed to advance interoperability and the utility of health data.

USCDI as a Framework

In healthcare, we have a complex and confusing set of health data standards. We have HL7’s Clinical Document Architecture (CDA) that specifies the structure of clinical documents for data exchange. We have HL7’s Fast Healthcare Interoperability Resources (FHIR) that describes data formats, elements and an API for exchanging clinical data. We have numerous codes sets like ICD-10, SNOMED, HCPCS, LOINC, NDC and RxNorm. And now, we also have USCDI.

The ability to integrate standards into data exchange is a key component to creating a more interoperable system but having a foundation that brings these standards together in a meaningful way is even more important. Up until the publication of USCDI, we had been plagued by an industry of standards that had no home. Now, USCDI is the framework that binds these standards together and specifies standard values for the data elements.

“What we must do is create a base, like USCDI, so that we all have the same common denominator, and then we will be able to launch all sorts of other exciting things.”

-Micky Tripathi at the 35^th Annual HL7 Plenary Meeting

Defining the Floor

Frameworks are meant to provide a foundation upon which standards can be built in an organized and logical way. Until now, health data standards have fallen short in many ways by simply not having a framework to attach to. This has led us down a convoluted path.

One of the more common pitfalls we’ve seen is that everyone has their own definition of data quality. Providers have their own documentation workflows. Organizations have their own local codes. Technology systems have their own data mappings. Take for example, one common data element that is used for patient care and population health analytics alike is assigned-at-birth sex. As there have not been any specifications around what value represents sex, we witness an excessive amount of variation as depicted in the visual below.

I’ll let you conjure up what this looks like for more complicated data elements like diagnoses, clinical tests, or plan of treatment.

Opening the Door for Data Quality

Before there was no established definition or reference to what constituted data quality, but now with USCDI we have an established “baseline” upon which organizations can perform data quality work. Organizations can profile their data repository against the USCDI data specifications to identify data gaps that can be used to drive data quality improvements. Here’s how an analytic tool can help.

1. Help you understand organization-level data quality and completeness.

2. Assist in identifying data quality opportunities at the section level.

3. Derive analytic insights at the data element level, like with this frequency analysis of encounter type.

A Health Information Exchange (HIE) or an Accountable Care Organization (ACO) can use insights like those shown above to improve the quality and use of the data. This may involve going back to a data sharing organization and problem solving how to address a data gap to stay in compliance with ONC’s Cures Act Information Blocking Rule, or it may involve learning what quality measures can be computed now that there is better quality data.

USCDI, as a pragmatic framework that binds standards and specifies values, is a step in the right direction for our industry. As the impact of the USCDI standard spreads, inherently, so too will the importance of data quality. Many organizations will go through iterative phases of identifying, monitoring, and improving their data quality.  Building trust in the data and enabling greater utilization of data for various analytics and population health efforts will aid organizations in their journeys toward alternative payment models.