Objective: To assess the quality of value sets in clinical quality measures, both individually and as a population of value sets.
Materials and methods: The concepts from a given value set are expected to be rooted by one or few ancestor concepts and the value set is expected to contain all the descendants of its root concepts and only these descendants. (1) We assessed the completeness and correctness of individual value sets by comparison to the extension derived from their roots. (2) We assessed the non-redundancy of value sets for the entire population of value sets (within a given code system) using the Jaccard similarity measure.
Results: We demonstrated the utility of our approach on some cases of inconsistent value sets and produced a list of 58 potentially duplicate value sets from the current set of clinical quality measures for the 2014 Meaningful Use criteria.
Conclusion: These metrics are easy to compute and provide compact indicators of the completeness, correctness, and non-redundancy of value sets.