Issues in Creating and Maintaining Value Sets for Clinical Quality Measures

Tuesday, October 09, 2012 — Poster Session I
1:00 p.m. – 3:00 p.m	Natcher Conference Center, Building 45	NLM	BIOINFO-21

Authors

• R Winnenburg
• O Bodenreider

Abstract

Motivation: Clinical quality measures help assess the quality of health care and compare performance across institutions. Value sets are sets of codes in reference terminologies defined to represent key concepts used in quality measures (e.g., set of macrolides antibiotics). The lack of standard approaches for creating value sets motivates the development of methods for ensuring referential integrity and avoiding duplication. Methods: We assessed the well-formedness, existence, and currency of value set codes with regard to the corresponding code system, applying NLM terminology services, and investigated the overlap among value sets. Results: We extracted 163,788 codes from 1463 unique value sets in the 113 quality measures published by the National Quality Forum in December 2011. Overall, 5% of the codes are invalid. We also found 67 duplicate value sets and 10 pairs of value sets exhibiting a high degree of similarity (Jaccard > .9). Conclusion: Although the overall proportion of invalid codes is modest (5%), these codes have important repercussions on the validity of value sets (19% affected) and, in turn, on the quality measures themselves (79% affected). The existence of duplicate and highly-similar value sets suggests the need for an authoritative repository of value sets and related tooling.

back to top