As MVC has grown, one of the most common questions we receive is, “Are the data right?” In the past year, we’ve devoted a lot of time to address this question and strengthen the data.
What we did:
The coordinating center examined the level of agreement between services assigned to episodes of care by the MVC claims data algorithms, and services identified in clinical data maintained by the hospitals.
Why we did it:
- First, the results of the project will provide detailed information to participating hospitals about the reliability of the MVC data.
- Second, the clinical data provided by hospitals will inform refinements to our inclusion and exclusion algorithms for care occurring after the index hospital stay, thereby enhancing the validity of reports provided to hospitals.
- Third, the findings will confirm MVC’s ability to identify and describe services occurring within an episode that are often not visible in hospital records.
How we did it:
We selected 1,830 Blue Cross episodes from 2013-2014 from 11 service lines. Hospitals were able to identify almost all of these patients in their records (99% match rate) and record whether or not that patient received a service within 90 days post-discharge. The services that were evaluated were 30-day readmissions, emergency department visits, home health visits, skilled nursing facility admissions, and rehabilitation services.
What we found:
This project allowed us to improve both our classification algorithms and the inclusion and exclusion criteria used by MVC. This process also provided confirmatory evidence for the provision of services not apparent from hospital records but captured by MVC algorithms. Following adjustment of these algorithms, the level of agreement between MVC claims data and hospital clinical data significantly improved. The Kappa Coefficient, a statistic we used to measure agreement between the two data sources, increased for all services to 0.80 or greater, which is indicative of ‘excellent’ agreement. MVC data is particularly useful for identifying post-discharge services that occur at different facilities than the original index admission as well as outpatient services (e.g., home health, rehabilitation) that are often not evident in hospital records. The ability to identify such services will be of great value for hospitals in the era of payment bundling and other pending payment reforms.
Questions or comments? We’d love to hear from you. You can comment on this article, or use the form at the bottom of this page.