Structured Quality Measure Validation and Aggregation

1. Proposed Workitem: Structured Quality Measure Validation and Aggregation

• Proposal Editor: Floyd Eisenberg
• Editor: Floyd Eisenberg
• Date: N/A (Wiki keeps history)
• Version: N/A (Wiki keeps history)
• Domain: Quality

2. The Problem

Today, quality measures are performed by analyzing patient-level clinical data elements, abstracting records to determine if missing elements can be identified to enhance compliance with processes and outcomes, and submission of the data for aggregation. A very significant amount of human intervention is required up front to collect and manage the data so that full credit is given to the clinical care provider or to the organizational provider. New interoperability methodologies for extracting clinical data at the patient level to enable quality measurement. Such automated methodologies provide data to an analyzer which can be located locally at the organization, at a third party intermediate organization or by the receiver of the final measurement assessment. This profile is intended to establish a specification for structured content to enable validation and augmentation of patient-level data collected for quality analysis. It is further intended to establish a specification for aggregation of individual patient-level data for comparison and benchmarking, allowing feedback for performance improvement initiatives within the EHR.

3. Key Use Case

The Quality Detailed Use Case of the US Office of the National Coordinator for Healthcare Information Technology (ONC – June 18, 2007) provides a good model as it specifies that quality measures, their definitions, and abstraction specifications are received by an organization electronically, areas for improvement are identified, followed by the provision and documentation of patient care. (Appendix A and B in this document). The scenario is specified for inpatient care settings and ambulatory care settings alike. The portion of the use case addressed in this profile request is that of the analyzer / aggregator of patient-level data. The use case specifies that the output of a measure is available for review by the clinical care provider or organizational provider to enable augmentation of data that exist but are not captured electronically for query analysis. Also implied in the use case is that, on receipt of the measure, the clinical care provider or organizational provider identifies areas for improvement and incorporates these areas into the clinical care process. Two functions are required at this analyzer level. First, once data are appropriately mapped from local to standard terminologies, the care provider must receive a “completed” patient-level set of data for review and validation that it is truly complete. The Quality Detailed Use Case includes steps (6.19 and 7.19) that allow the care provider to validate the data that will be used for calculations and to augment those data with information that exists yet is not available in electronic format. Such a validation and augmentation step is the first requested portion of this profile. Secondly, once the validated and augmented data are sent back to the analyzer, aggregation of data for the individual care provider or the organization must be performed to return to the provider or organization information about performance that can enable improvements in care delivery, closing the loop again at steps 6.1.10 – 6.1.11 and 7.1.10 – 7.1.11. Such aggregation requires combination of compliance and/or exceptions for all patients in the identified denominator cohort, the second requested portion of this profile."

4. Standards & Systems

CDA XML Logic, algorithms open, but reference to OWL, GELLO, Arden Syntax, Common Logic for trial implementations

5. Discussion

The current profile is intended establish a specification for managing structured patient-level data content to express compliance with expected processes and outcomes based on specified patient criteria. Such criteria include inclusion and exclusion criteria for the denominator and the numerator, with specification for performing the calculation of compliance. An example of such a measurement is and American Heart Association measure for a patient with a recent myocardial infarction. Such a patient is expected to receive certain medications at arrival to a hospital (beta blockers and aspirin) as well as at the time of discharge from the hospital. These and additional measures for patients with recent myocardial infarction comprise a cluster of measures related to a specific condition. The data elements for each of the measures within the cluster are specified by the measure developer. Management of structure for measure specification is the subject of a different profile proposal. The validation of patient-level data for such analysis is currently a manual process, often performed by medical record specialists trained in chart abstraction. Such specialists are generally not available in ambulatory care settings, especially in smaller clinical practices. Therefore, an automated mechanism is required to enable validation of data collected and missing data. Additionally, a standard mechanism for calculating performance based on the collected data is required in aggregate with respect to all performance for all patients for each measure. Also required is the ability to calculate all patients who were provided care compliant with all measures within a cluster (also called “bundling” by the Institute for Healthcare performance <IHI>). This profile is requested to establish a standard implementation guide for extraction of data from multiple patient-level quality data summaries for analysis and aggregation based on algorithms submitted with the initial quality measure definition and specification. As an example, a quality specific medical summary can be a constraint of CDA, and an extension of Query for Existing Data (QED) might allow query of data from these quality-specific medical summaries for aggregate analysis. The content for the profile will be somewhat informed by the work of the Collaborative for Performance Measure Integration with EHRs, a collaborative of the American Medical Association (AMA), the National Committee for Quality Assurance (NCQA), in cooperation with the Electronic Healthcare Vendor Association (EHRVA). Workgroup B of the Collaborative has identified measurement import and export criteria. The export criteria have been incorporated into a constrained CDA document now under review by the HL7 Pediatric SIG as the Quality Reporting Document Architecture (QRDA). The message content is also incorporated into the HITSP Interoperability Specification as a Patient Level Quality Data Document, Component C38 of the Quality Interoperability Specification IS-06 (available at: HITSP Quality Constructs). That component is similar to QRDA.

Establishment of a Structured Quality Measure Validation and Aggregation will significantly move forward automated analysis of clinical guideline and measurement performance at clinical care delivery sites, enabling enhanced performance improvement initiatives.

Appendix 1 Inpatient Quality Workflow Scenario, Excerpted from: US Department of Health and Human Services, Office of the National Coordinator for Health Information Technology. Quality Detailed Use Case. June 18, 2007, page 25. Available at: http://www.dhhs.gov/healthit/documents/UseCaseQuality.pdf