A harsh reality of today’s healthcare environment is that providers and payers use a multitude of systems - each with their own way of representing data.
A health system may be integrating a hospital bringing a new electronic health record (EHR) system into the network, while maintaining legacy EHR applications. Health systems often grapple with how to accommodate practices using a mix of EHRs. A health plan, meanwhile, may employ multiple claims systems. In the past, those systems often operated in isolation, using conflicting terminologies to represent clinical data.
To support tomorrow’s analytics initiatives, this fragmentation makes it difficult for healthcare organizations to leverage clinical data to trigger decision support rules, support disease stratification efforts, and accurately compute quality measures. This terminology barrier must be overcome if the industry is to reach national goals around increased interoperability, transparency and collaboration.
What’s needed is a way to unify multiple, incompatible terminologies. Data normalization is the process that accomplishes this task, reconciling disparate terminologies into a shared vocabulary. Data coming from different sources - and encoded in different reference terminologies and classifications systems - can be normalized into standard terminologies.
For example, a healthcare organization may find itself dealing with numerous labels for a Hemoglobin A1C test: HbA1C, A1C, HA1C, A1C Hemoglobin, HbA1c (%), and HEMOGLOBIN A1c for instance. But data encoded in differing terminologies can be normalized to the LOINC® lab standard. In a more prosaic example, a health system might seek out a standard way to identify each of its sites: ACME Hospital 1, ACME Hospital 2, etc. In both cases, data normalization lets the healthcare organization more easily aggregate data from trending and analysis. The ability to aggregate and share data among the different players in the healthcare ecosystem rank among the top benefits of normalization.
Getting started
A business analyst or IT manager may intuitively grasp the need for normalization, but may be mapping terms manually using spreadsheets and the size and scope of the analytics project has reached a tipping point where help is needed. Or in some cases, they may have no idea how to launch a project. For the next several weeks, I'll be providing insights on how to take your project to the next level or help you get started, supplying project leaders with a step-by-step guide for initiating, managing and sustaining a normalization project.Each post also contain some “real world” examples but will change the names of individuals to protect the innocent. This will hopefully give you additional guidance to avoid the common pitfalls that can occur when you are implementing a data normalization project.
Here are the topics that we plan on covering in the series:
- How to get executive buy-in on your data normalization solution: This post will cover the importance of ongoing executive support. A normalization project will often involve a number of stakeholders and project team members who will be asked to devote their time to the effort. Having the executive team on your side is critical for securing that level of commitment -- along with the funding necessary to support the project.
- The importance of taking an inventory of your data normalization needs: An organization pursuing a normalization project must determine where its data resides and who owns the data. This post will discuss this discovery process, which will also identify the departments/users who will most benefit from normalization.
- Identify constraints that nay impact your data normalization plan: Once a project team has taken stock of its data, the next step is to determine what could go wrong during the course of a normalization initiative. This post will describe the constraints a project team will likely face: time, money, resources and technical barriers.
- Prioritize your data normalization to-do list with an impact analysis: This next post provides insight into how a project team can conduct an impact analysis to identify a starter project that delivers the most value in the shortest amount of time.
- Establish a governance process with your data normalization solution: Normalization calls for more than just launching a project and stepping back. A normalization solution needs ongoing guidance to sustain the initiative. This post will discuss the creation of a governance process to keep your project on track.
- How to normalize your first project with your data normalization solution: This post will offer some practical implementation advice when it’s time to embark on your initial normalization effort.
- Data normalization: Mapping to existing standards vs. creating local standards:A provider or payer will probably adopt existing standards such as ICD-9/ICD-10, LOINC® and SNOMED CT® as it normalizes its data. But there may also be a need to create local terminology standards as well. This post will help project leaders navigate standards issues.
- Mastering the data normalization cycle: This post will summarize the data normalization cycle and underscore the importance of following through each step until the project enters the maintenance phase.
Taken together, this blog series provide a walk-through that will help you size up and execute your first normalization project. Speak to an expert to learn how Health Language solutions can help your organization improve the quality of your data.
LOINC® is a registered trademark of Regenstrief Institute, Inc.
SNOMED CT® is a registered trademark of the International Health Terminology Standards Development Organisation (IHTSDO)
