100 likes | 233 Views
Lessons Learned Behavioral Health eMeasure (BHeM) Project. Saul A. Kravitz, PhD July 30, 2012. Background. Behavioral Health eMeasure Project Sponsors: ONC and SAMHSA Objective: Develop Portfolio of Behavioral Health Measures Output: 10 eSpecified CQMs, 2 included in NPRM
E N D
Lessons LearnedBehavioral Health eMeasure (BHeM) Project Saul A. Kravitz, PhD July 30, 2012
Background • Behavioral Health eMeasure Project • Sponsors: ONC and SAMHSA • Objective: Develop Portfolio of Behavioral Health Measures • Output: 10 eSpecified CQMs, 2 included in NPRM • Novel Development Approach • Collaborative eMeasures and shared value set development • Model eSpecfication development with shared value sets • Develop portfolio of value sets supporting all 10 eSpecs • Project Organization • Measure steward collaboration • eSpecification: NCQA (4), TJC (2), and MITRE (4) • Value Set Development: Apelon, Inc
Lessons Learned eMeasures Intent and Selection Concept Development Identification Collaborative eMeasure and Shared Value Set Development Value of Impartial Review Importance of Testing eMeasures Tools Utilization Project Management Oversight
Measure Intent and Selection • NQF-endorsed CQMs include • Intent (e.g., diabetics’ feet should be examined) • Data Source (e.g., claims data) • Logic • Retooling a CQM preserves intent, replaces data source, adjusts logic • Some CQMs will only work with a specific data source Lesson: eSpecifications of CQMs should be considered as closely related implementations of the intent of CQMs with alternate data sources
Single Intent, Multiple Implementations NQF0580: Bipolar Manic Agent Claims Data Provider EHR Data ACO EHR Data NQF endorsed BHeM Future? • All three implementations of NQF0580 share intent: bipolar patients should be medicated appropriately • Data sources and logic differ • Sensitivity and specificity will differ
Collaborative eMeasure and Shared Value Set Development • BHeM Project separated eMeasure logic specification from value set development • Value set developer captured measure developer intent • MITRE and measure developer review of value sets • Iterative process to reconcile • Traditional expression of value sets is an enumeration of code values (“extensional”) • Difficult to review • Difficult to maintain • Doesn’t take advantage of hierarchy of vocabularies • Expression of Value sets as “recipes” (“intensional) • Much easier to review and maintain • Leverages hierarchy of vocabularies
Traditional Extensional Value Set • …and many more
Intensional Value Set Description • Include codes that are in active use • Include codes that apply to Humans • Lessons: • Communication between developers of value sets and measure stewards is essential • Value set recipes for clinical concepts should be specified ‘intensionally’
Value of Impartial Review Lesson: eMeasures and Value Sets intended for inclusion in national quality programs should be reviewed by 3rd party with focus on feasibility, correctness, and computability. • Independent reviews resulted in improvements to measure design • Simplification of the proposed measures • Identification of errors and issues with data and logic • Elimination of data elements • Improved feasibility • Improved accuracy and generality of value sets • Simplified and clarified measure logic based on development and review of test cases
Importance of Testing • BHeMeSpecification deliverables included test data and outcomes in a common format • Specified at concept, not code level • Identify logical errors in eMeasures • Identify scenarios not considered by eSpecifiers • Facilitate testing Lesson: Testing is essential • Test data and outcomes should be required in documentation of eMeasures • Automated testing for conformance with test data should be the rule