By Brian LaBerge, Health Language
In a previous blog post, I discussed the importance of data normalization and the need for executive-level support as the first step in the implementation process.
This post focuses on the next step: taking an inventory of your data normalization needs. A healthcare system will be hard-pressed to reach its desired data destination if it has no idea of where it stands today. Describing the current environment—where does the key data or terminology exist and who owns it—is the objective of the inventory phase of a data normalization project.
Where's The Data?
When taking inventory, an organization needs to categorize systems in three ways:
- Systems that require periodic terminology updates
- Systems that produce codified data
- Systems that report on and visualize this data
Systems that generate or consume terminology could include an electronic health record, a patient portal, a lab information system, or a data warehouse. Essentially, the healthcare system needs to get an idea of where terminology exists and how it’s used.
While many organizations understand individual terminology components and interactions, few grasp the complexity of the entire system. That’s why I often recommend a terminology assessment to make sure the team understands the landscape before beginning a data normalization project.
This inventory-taking step helps the normalization project team ensure that it is not overlooking important data. The task also helps identify the key data stakeholders: who owns responsibility for a given data set? In addition, finding the key stakeholders will facilitate the establishment of a governance process later in the data normalization project lifecycle.
The inventory phase will generally provide new insight into an organization's data holdings. It is important to evaluate both active systems and systems that are no longer in use. For example, the project team might discover valuable data housed in an unused legacy system. The healthcare organization may want to extract and normalize that data; combining the legacy system data with current data from active production systems could help clinical researchers explore trends over a period of years. A hospital could examine historical patient safety trends, for instance.
A project of this type can provide both immediate and long-term benefits. In the case above, the normalization of legacy data, although a one-time effort, will yield continuing benefits.
Terminology Discovery
The inventory-taking phase, as noted above, involves discovering where data resides in the organization. You might kick off this process by interviewing one data custodian or one data-consuming user group. This inquiry will inevitably point to additional data owners and users, creating a snowball effect that sheds additional light on a health system’s data.
This can be informal—you don’t need to follow a structured interview. But the discovery process should be thorough enough to uncover the terminologies that are most important to normalize and the users who will benefit the most from normalization. With that insight, you can zero in on the initial normalization projects.
During the inventory process, our clients are often surprised to find out how much data needs to be normalized even within a single comprehensive EHR. One described the process as “normalizing a system against itself.” You might ask, how is that even possible? There were multiple groups in the organization that were using the same EMR. They all created their own sets of data, and didn’t realize that as they acted independently, they were creating redundant local data representations of payers and locations within the same application. It became obvious that this one EMR system had multiple data stewards (which is okay), but we treated each one independently so that we could allow the autonomy of the use to continue, but not the data chaos.
Once you have executive buy-in and have inventoried your system you are almost ready to scope your project. We will talk in the next blog about some of the constraints to consider as you scope it.
Have you inventoried your systems? What have you found as “aha” moments during that process? Comment and let me know.
About the Author
Brian LaBerge joined Health Language in September 2010 and has serviced in multiple capacities including Technical Consultant, Project Manager and Director Of Client Implementations. He brings 18+ years of implementation and software deployment experience to Health Language. He has managed several Data Normalization and ICD-9 to ICD-10 projects for Payers and Providers during his tenure at Health Language.