Welcome to Lesson 4, Metric Number 40, Risk-Standardized Bleeding, where we will discuss Metric 40 Dashboard Results and Interpretation. In Lessons 1, 2, and 3, we gained an understanding of the model and patient eligibility, what constitutes an observed bleed, per the model specifications, and the valuable information that is provided in the detail lines and patient-level detail drill-down. In this fourth and final lesson, we take a closer look at understanding the hierarchical Metric 40 risk model, as well as examining facility performance on the dashboard. Metric 40 utilizes a hierarchical risk model, whereas a traditional risk model assumes that all information about a hospital's performance is a result of factors attributed to that specific hospital, a hierarchical risk model deduces that there are factors that hospitals have in common, as well as factors that differentiate hospitals, and that these similarities and differences influence performance. In this way, a hierarchical risk model considers both patients and hospital variables to mathematically arrive at the risk-standardized bleeding rate. We begin with discussing the features of a risk-standardized bleeding rate. Hierarchical models assume a bell-shaped curve, which supposes that a given's facility performance will likely be closer to the mean or average at a future point in time, regardless of whether their current performance is near-perfect or far from it. In this way, the model removes outliers and provides stabilization, which in turn ensures that low- and high-volume facilities are treated fairly. As an example, in this view of risk-standardized bleeding detail lines, we see zero observed bleeding events in all four quarters of the rolling four-quarter period. In a traditional risk model, this would yield a risk-adjusted rate of zero. However, because of the predictive nature of the risk model and the stabilization or smoothing of outliers, the model presumes that this facility is only human and will eventually experience a bleeding event at a point in the future and computes a risk-standardized bleeding rate value of 1.14, a stellar performance by any measure, but not a perfect zero. Consequently, metric 40 performance output can help you to understand how your hospital will perform in the future. Alternately, the risk-adjusted bleeding detail lines, as discussed earlier in the presentation, can aid in understanding the specifics of observed bleeds against what was expected for your patient population and helps to focus your quality improvement initiatives to improve the incidence of bleeding in your PCI patient population. The ending time frame reflects data for our rolling four-quarters. The rolling four-quarters are used to determine the metric value. My hospital R4Q performance is comprised of data included in the four-quarters as specified by the ending time frame and is compared to the registry benchmark. The U.S. hospital R4Q performance distribution is determined by the ending time frame. The 50th percentile is the registry benchmark. When reviewing metric 40, facility metric performance is compared to the U.S. registry benchmark. When evaluating executive summary metrics, including risk-adjusted metrics, which utilize a traditional risk model, a performance trend of the per-quarter performance of the previous eight-quarters appears. With metric 40, quarterly trends are not displayed. And why is that? A traditional risk model, such as risk-adjusted bleeding, discussed in lesson number three, assesses only patient risk to provide the expected count. This, with the observed events, is used to produce the observed-to-expected ratio, or OE ratio. The model can produce a quarterly result that is logical when compared to the R4Q result because the patient risk profiles assessed quarter-by-quarter are simply reflected again in the R4Q. However, a standardized, hierarchical model, such as the metric 40 risk-standardized bleeding, assesses patient risk, evaluates patient case mix, and considers hospital variables. The model was developed and validated using hospital R4Q data, not quarterly data for purposes of benchmarking. Consequently, the quarterly output is unique and not necessarily indicative of the R4Q result. The R4Q result is likewise unique, but now has the full spectrum of patients, altering the case mix and affecting model calculations. Participants should assess their rolling four-quarter result and use the metric details page to evaluate their rolling four-quarter performance trend. Risk-standardized bleeding detail lines provide quarterly values for the number of eligible procedures and number of observed bleeding events, as well as the total R4Q count for each. A patient detail button is provided, which takes the user to a patient-level drill-down data, as discussed in Lesson 3 of this presentation. But again, as hierarchical risk models are not engineered to provide quarterly data, the risk-standardized bleeding ratio and rate and lower and upper confidence intervals are blank in the quarter columns. Please ensure you are utilizing all of the benefits of the eReports dashboard. The eReports dashboard provides a streamlined, centralized location to assess the facility's metric performance. It provides the user the ability to analyze metric details and drill down into the metric. The dashboard also provides patient detail information to evaluate individual patients and the procedure details. Dashboard allows the ability to filter data for a better understanding of the patient and procedure characteristics, as well as provider treatment patterns and facility processes. For additional resources to help support your understanding of Metric 40, we would suggest visiting the CAF PCI Registry Resources Documents page and access the Executive Summary Measures and Metrics Companion Guide. Additionally, we recommend viewing the Bleeding Model to predict risk of post-procedure bleeding. This is the document that was used to support Metric 40, as it was with Metric 37. The model was updated, so there are changes, such as the variables used in the algorithm. However, please continue to rely on this paper. This concludes Lesson 4 of Metric 40 Risk-Standardized Bleeding, in which we discuss Metric 40 dash board results and interpretation. Thank you for your participation.

Metric 40

Dashboard Results

Interpretation

Risk Model

Bleeding Rate