Webinar: Are You Getting Value From Your Value Sets?
Value sets can be an effective tool for analyzing clinical data and identifying critical gaps in patient care. Clinical Architecture believes that for that tool to be effective, it needs to fit in to the 6 C’s framework for value set quality.
Clinical Architecture presented a webinar “Are You Getting Value From Your Value Sets?” to present real-world value set use cases, share this 6 Cs framework and discuss ways to evaluate the quality of value sets that a healthcare organization is using, appraise the fitness of a value set to determine how well it meets specific business objectives, and explore strategies to optimize value set management.
Presenters
Erika Ganley
Senior Clinical Content Analyst
Clinical Architecture
Raymonde Uy, MD, MBA
Physician Informaticist
National Association of Community Health Centers (NACHC)
Moderator
Victor Lee, MD
VP of Clinical Informatics
Clinical Architecture
Watch the webinar now!
Our speakers shared a great deal of information during this informative session but were not able to respond to all the audience questions that were posed in the time allotted. Below is a list of the questions along with responses from our experts.
Question 1: CMS has been very focused on ‘harmonizing’ value sets, although it helps reduce the amount of similar value sets within the VSAC my needs based on experts may be different from others. Do you think harmonization is critical or is it better that multiple value sets exist of similar/same type?
Response from Raymonde Uy:
Harmonization is critical, but not necessarily better than multiple value sets.
- Harmonization in the sense that multiple organizations are agreeing to a common data model using validated and vetted value sets is critical
- Multiple value sets of the same name and use case/intention, with the same inclusion and exclusion criteria, need to be harmonized, especially if the included or omitted codes are discordant with each other.
- Multiple value sets that are of somewhat similar type, BUT have a completely different use case, should be allowed to exist. There is intentionality in creating value sets, and that intent is expressed through the codes included in each set.
Question 2: Are you able to disclose if the value sets were extensionally defined or intensionally defined? Does one type of value set seem to be more error prone than the other?
Response from Raymonde Uy:
On VSAC, each value set in the list already shows whether the set was built as an extensional or intensional set. While taking time to create rules for intensional sets makes the maintenance less of a burden with less frequency, the VSAC intensional set rules are not as robust as creating a model that has additional ontological rules of inclusion or exclusion as in Symedical. Extensional sets become error prone as time passes due to terminology updates. They are also prone to errors of omission or inclusion due to the manual nature of creating and uploading extensional sets. Intensional sets are prone to the same, if the rules are not crafted to be specific enough, or if there are new parent or child concepts that escape the built rules.
Question 3: Could you talk a bit more about how did you normalize the data from individual EHRs that do not have standards like LOINC/RxNorm?
Response from Raymonde Uy:
Symedical has a tool that tokenizes the free text and uses various matching algorithms to arrive at a confidence expressed in percent on how close the free text is to the description or name from LOINC or RxNorm. After this, the mapper, or other map collaborators, can select which suggested code is the closest in meaning/semantics from the free text data.
Question 4: Based on the 6 Cs of value sets, which Cs are directly attributed to authoring issues and which ones are directly related to tooling gaps or insufficiencies?
Response from Erika Ganley:
Both authoring and tooling can directly influence each of the 6 Cs for value set quality.
- Completeness and Correctness are influenced by the author’s ability to understand the concept from both a clinical perspective and from the perspective of how variations of the concept are represented in standard terminologies. Tooling may also impact the author’s ability to search for relevant terms in a terminology hierarchy and create intensional rules to define terms for a concept efficiently and accurately.
- Currency is heavily influenced by tooling gaps or inefficiencies; however, it may also be impacted by the number of available resources and the processes put into place for routine review and maintenance.
- Clarity and Consistency are mostly influenced by authoring. It is important to have an established process by which a team defines the scope for each value set. The author should also understand the concept represented by the value set to efficiently and accurately describe the scope. A robust editorial policy with continued staff development and training is key for authoring consistently. Inadequate tooling can also contribute to both of these Cs.
- Congruence is mostly related to authoring in that the steward should have awareness of the data by which the value set will be used to evaluate.
Question 5: Are there any collaborative efforts putting together publicly available value sets to define pregnancy and other sensitive conditions?
Response from Raymonde Uy:
Yes. These reproductive health value sets have been uploaded to VSAC under the NACHC steward for anyone to export and use to segment their data for security or confidentiality purposes. This was a collaborative effort by myself/NACHC, Health Gorilla, CRISP, Planned Parenthood, and AHIMA, using Clinical Architecture’s Symedical to easily create either through using or modifying existing value sets that have been already built and validated by Clinical Architecture, or creating from scratch.