130 likes | 259 Views
Risk Adjustment eCQMs and HQMF. Collected from work by The Joint Commission and Abt Associates for CMS/ONC. MU3 Planned Risk-Adjusted Measures. Multiple contracts and programs are working on risk-adjusted measures
E N D
Risk Adjustment eCQMs and HQMF Collected from work by The Joint Commission and Abt Associates for CMS/ONC
MU3 Planned Risk-Adjusted Measures • Multiple contracts and programs are working on risk-adjusted measures • Many current CMS programs have claims-based or other adjusted measures that are planned for e-specification
eCQMs Under Development • Constructs under development • Monitoring of patients receiving patient-controlled analgesia • Trauma mortality • Unplanned ICU readmission within 48 hours of ICU discharge • Acute kidney injury more than 48 hours after admission • Acute respiratory failure more than 48 hours after admission • Failure to rescue/failure to respond
eCQMs Under Development • At least four of these six constructs measure adverse outcomes that may be a function of poor quality care • Mortality • Readmission • Complications (renal or respiratory failure) • Outcome measures require risk-adjustment based on patient characteristics to allow for fair comparisons between providers • Hospital with older or frailer patients at greater risk for these outcomes shouldn’t be penalized
Risk Adjustment Considerations • CMS has a long history of developing and using risk-adjusted outcome measures for a variety of purposes • Based on claims or chart-abstracted data • Risk-adjusted eCQMS are “uncharted” territory • There are no risk-adjusted eCQMs currently in use • Yale’s 30-day AMI mortality measure nearing completion of eSpecification but not yet implemented • There are several unique issues in developing and testing risk-adjusted eCQMs • Risk modeling goals and techniques do not change and are based on CMS policy/precedent
Risk Model Equation Types • Indirect standardization – patient-level data • Examples: Logistic Regression for proportion measures and Linear Regression for continuous variable measures • Direct standardization – based on aggregate data Note: Risk Models may be published more frequently than the associated HQMF specifications – especially for newer measures.
Risk Factors • Potential Risk Factors • Based upon literature review and data analysis the Risk Model Developer identifies all potential Risk Factors • Risk Factors can be included, excluded, and their definitions change from one release of a Risk Model to the next without a new HQMF as long as the required eMeasure data is specified within the current HQMF.
eCQM Considerations: Expressing Covariates in the MAT • Currently risk model covariates are included as “supplemental data elements” in the MAT, since there is no formal section for risk adjustment • Some limits on logic for these variables • E.g. “first” or “highest” • This may change with HQMF-R2 • Consideration being given to the needs of risk-adjusted measures
Risk Adjustment Steps • High-level steps: 1. Develop, test, and disseminate Outcome measure (HQMF) • Limits the data available to the Risk Model Developer until a new HQMF is published • Requires the Measure Developer and the Risk Model Developer to work together 2. HIT vendor implements Outcome measure (HQMF)
Risk Adjustment Steps (cont) • High-level steps: (cont) 3. Provider (or vendor on their behalf) submits patient-level measure data and, if needed, additional patient-level risk factor data to Quality Data Receiver(s) (QRDA Category I) 4. Risk Model Developer utilizes the submitted patient-level data and “Statistical Magic” to develop / update the Risk Model and Risk Factors May be the same or different organizations
Risk Adjustment Steps (cont) • High-level steps: (cont) 5. New Risk Model and Risk Factors are published (no HL7 standard exists) 6. HIT vendor implements new Risk Model and Risk Factors (no HL7 standard exists) 7. Go to step #3 (provider submits data)
Applying a Risk Model • Steps for applying a Risk Model: • Identify appropriate cases by executing the HQMF • For each case identified above, calculate the Risk Factors contained within the Risk Model • If appropriate, manage missing Risk Factor data • Calculate each case’s Predicted Value • Aggregate the organization’s Risk-Adjusted Rate using each case’s Predicted Value (If submitting data, QRDA Category III)