Insights to Inspire 2022: The Journey Continues – Data Standardization

Informatics the Journey Continues logo showing the five areas that make up the series: Data Completeness, Data Quality, Data Standardization, Process Improvement, and Resources
Apr 14

Coordination Among University and Health System Enables Path to Standardized Data

Headshots (starting at the top and going clock wise) of Keith Herzog, MPPA, PMP, Richard T. D'Aquila, MD, Daniel Schneider, MS, Kristi Holmes, PhD, and Pearl Go, MA

In their efforts to recruit more than one million patients for a significant research study, Northwestern Memorial HealthCare (NMHC) and the Northwestern University Clinical and Translational Sciences (NUCATS) Institute realized the need to standardize medication data in their joint venture, the Northwestern Medicine Enterprise Data Warehouse (NMEDW). The path to standardization required coordination between the two organizations.

First, the two separate organizations, the NMHC health system and the University’s NUCATS Institute, needed to determine why some of their data were not aligned with national standards frequently used to share and integrate data across different institutions. One opportunity identified for improvement was the mapping of medication names to RxNorm codes, the National Library of Medicine's medication terminology system. The team engaged in a deep-dive discovery process to figure out what was happening. Were data being coded incorrectly, or were there issues with their processes?

NMHC had been expanding rapidly and acquiring additional healthcare networks, with each new network bringing their own method of collecting and organizing data. As hospital data from each of these new healthcare networks was integrated, different decisions were made as to how much of that data would be integrated, depending on the complexity of how that data was originally stored. For example, certain critical details, such as dosing information or unit-related information, were not always migrated over if it was not essential to patient care.

Through this discovery process, the EDW team learned that not all medication data were standardized in its version of EPIC. The decision was made to create their own data model to make sure standardized medication terminologies were being funneled to the framework, which then could be used for other projects.

“Utilizing data in a standardized format is the key to a lot of important clinical and translational research initiatives, especially when you're doing things with sharing across different networks, sharing across different organizations and expanding to nationwide studies,” said Daniel Schneider, NMEDW Director of Research Analytics. Data structure integration NMHC worked to integrate data structures to make medication data available in RxNorm.

 

​​​​Teal line“Utilizing data in a standardized format is the key to a lot of important clinical and translational research initiatives, especially when you're doing things with sharing across different networks, sharing across different organizations and expanding to nationwide studies.”  

Daniel Schneider, MS, NMEDW Director of Research Analytics Manager of Research Analytics, NUCATS. 

 

Data structure integration
NMHC worked to integrate data structures to make medication data available in RxNorm. This involved implementing a plan to assign missing RxNorm values to its medication data and setting up an automated process to keep its RxNorm vocabulary updated based on the most recent Unified Medical Language System (UMLS) release. Alternative standardized medication mappings from different source systems, such as Multum, Medispan and NCD codes, were converted to RxNorm values when possible.

In January 2020, Northwestern loaded the UMLS into the Northwestern Medicine Enterprise Data Warehouse and evaluated all available sources of medication data. National Drug Codes were mapped to RxNorm values, and analysis revealed unmapped medication records. In implementing Multum to RxNorm mapping on missing Cerner medication vocabulary, additional medication records with an associated RxNorm value increased to ~242 million, an increase of ~11% from the previous year’s ~217.1 million. Even with the medication mappings in various source systems, though, the need to manually map medications remains.

Northwestern hopes to implement other processes to help eliminate manual mapping. One method to explore is to further migrate to the Fast Healthcare Interoperability Resources (FHIR) standard for fast and efficient exchange of clinical and administrative data.

Continuous improvement and education
Besides enhancing their processes, a large part of this effort involved raising awareness to shed light on how much work was needed and how important standardizing terminology data points to a standardized vocabulary was to the organizations. “Coordinating two completely separate organizations – the medical school with a research focus and the healthcare corporation with a healthcare delivery priority – requires extra effort to ensure jointly understanding the sometimes-different needs of each organizations’ missions,” said Schneider.

The team focused on continuous improvement and on educating others about the importance of having interoperable data that can inform clinical and translational research, as well as care. In addition to supporting the shared top priority of improving patient care, the team also emphasized the downstream implications of standardized data and the benefits of the data warehouse for future, now unanticipated research applications.

Part of that awareness involved educating partners about the limitations of an electronic medical record. “There will always be cases where people do not standardize something or input something based on workflow and needing to address immediate patient care, so you are going to have messy, dirty data at times,” Schneider added. “There is no way around that from a healthcare system, it is just part of the nature of the game.”

Buy-in was created with a detailed plan for improvement by utilizing certain partnerships to help interface with historical data. It involved applying “team science” to metric-based activity. Partners in the overall effort included Northwestern University Feinberg School of Medicine’s Center for Health Information Partnerships and the healthcare corporation’s data architect team, along with NMHC’s infrastructure architecture and database teams. Central to the effort was the NUCATS Institute’s embracing of the Informatics Common Metric and the Results-based Accountability Turn-the-Curve framework as part of the national Common Metrics Initiative.

Lessons learned

  • Educating partners on the value of standardized data for downstream research as well as more real-time patient care/operations is important to achieving success.
  • Health information systems were built for billing, not for patient care, leading to challenges for both care and research missions.
  • Having “dirty” data is inevitable in a busy, demanding healthcare environment.
  • There is some risk that a misaligned code will result from merging data from another organization, but it will achieve closer sharing of more valuable information between the organizations.
  • Northwestern University’s extensive collaborative network and role as a major academic research institution played a key role in helping partners to come on board.

Each blog in The Journey Continues series will have a corresponding webinar scheduled at a future date. Register for the Data Standardization webinar on Wednesday, May 4th at 3:00 Eastern time to hear how Northwestern University integrated data structures to make medication data available in RxNorm.