In our last blog, we discussed selecting your first project to use for data normalization. A healthcare provider or payer pursuing a normalization project must establish a governance process to ensure the organization doesn’t revert to a chaotic data state. This post discusses the creation of a solid governance structure that preserves your investment in good data.
The Role of Governance
Now that you have chosen a project, you are faced with decisions about how to represent the data. You may choose to use local standards (values for identifying hospitals within a health system’s electronic health record systems) or, more commonly, national or international standards (ICD-9/ICD-10, SNOMED CT, RxNorm, and others). But your decisions don’t stop there. Even after choosing a standard terminology, you may need to further constrain representations. For example:
- In SNOMED CT, it is common to constrain concepts to the Clinical Findings taxonomy, and not to allow other taxonomies (such as Situation with explicit context, Events, etc.).
- In RxNorm, you need to decide whether only orderable semantic types are acceptable (e.g., SCD, SBD), or drug constituents (IN, BN), or whether anything goes (hopefully not!).
- In ICD-9-CM and ICD-10-CM, it is common to constrain to billable concepts.
- You may need rules when input is ambiguous, such as presuming a specimen type of blood when mapping laboratory tests to LOINC.
- You may want to specify a particular release of a terminology (e.g., ICD-10-CM 2016), as well as establish cycles for updates.
You need a governance process to make these decisions and enforce them across the organization. In establishing a governance process, you should focus on three critical components: people, process, and technology:
People include members of the normalization project team: project managers, system owners (EHR and claims system managers, for instance), and business intelligence/data warehouse managers among others. Some data decisions require a higher-level sign-off, so a provider or payer may create a governance committee with executive participation (a chief technology officer or a chief medical officer, for instance). This committee may also include data stewards, who help enforce standards compliance.
The governance group will also establish and monitor a governance process, which defines how clinicians, IT managers, and coders create or author new content or apply an update from a standards body. If a system owner wants a new drop-down item in a pick list, for example, he or she will need to follow a de
fined format and process for this item. The governance group is tasked with making sure everyone in the health system knows what the process looks like.
The governance process should also include an audit component to ensure that staff members follow the process and that each request for a new value goes through change control. The audit component should take periodic samples to determine whether the provider or payer is conforming to data standards.
The technology aspect provides a platform to help execute the governance process. Data normalization tools, such as Health Language’s LEAP Map Manager, automates the task of documenting governance decisions.
The LEAP Map Manager tool has configurable workflow steps to ensure you are following the process to maintain high quality data.
Who Should be Involved?
A number of people may have a seat at the governance table, but not all of them need to be involved in each decision on data definitions and standards. The introduction of a new value in an EHR system may require the approval of only the system owner, a data steward, and the data warehouse manager. Executive-level governance committee members will become involved in more strategic decisions. Who gets involved in a governance decision will depend on the organizational impact of the data change under consideration.
Strong data governance following this process will ensure that your data remains clean and useful throughout the organization, maximizing your return on investment.
Read Previous Data Normalization Blog Read Next Data Normalization Blog