In the age of Accountable Care Organizations (ACO’s), big data and analytics are all the rage. If one is going to be measured and accountable for patient care beyond the immediate encounter, then having information available to make improvements is essential. It’s not just having the data though, it’s being able to do something with it, which is why big data and analytics are a big deal!
In a previous blog (Quality Outcomes in a Pay-for-Performance World, Nov. 15, 2013) I referenced an announcement by vRad (http://www.vrad.com/#/home) about making their database available to outsiders for outcomes analysis. According to Jordan Halter, vice president at vRad, their database contains over 23 million exams to draw on, making it a case of big data. What I also found intriguing is that vRad developed a way to glean actionable insight from their big data. Attempting to normalize data across one organization is challenging, let alone across multiple 2,200 different organizations from 50 states!
According to Halter, vRad developed an approach and product they call vCoder. vCoder assigns 23 unique attributes to every study. Each attribute is selected to either improve operations (moving the right study to the right rad for the right reason) or for the granularity it gives vRad on the backend for analytics and reporting. Just one of the 23 attributes is something called the vCode: a human readable imaging code referred to as the study’s “Radiology VIN” number, and qualifies the study in terms of the modality, contrast use, body region, location, and a unique procedure number. This has enabled analysis of data from multiple locations. The coding happens outside of a client’s operating environment, meaning it doesn’t impact internal processes or data categorizations and it does not force all sites into a common exam code dictionary. Halter explains, “The vCoder is the standard but each and every hospital’s data must be normalized down to that standard. There is no separating the concept of standard and normalizing to the standard.”
Halter indicates that through the use of the vCoder, vRad has been able to release their Radiology Patient Care (RPC) Indices. The first application was on CT scans from Emergency Departments (ED) from over 2,200 different hospitals in 50 states. With the indices, clinicians now have a tool to measure and compare how their use of imaging compares to national and peer group averages, helping physicians and hospital administrators ask the right operational and policy-related questions to make better decisions for the health of their patient communities.
vRad appears to have come up with a novel approach to normalizing data allowing them to convert Big Data into actionable insight. This fundamentally is one of the key challenges to making practical use of data beyond one site’s data – something that will be valuable in terms of assessing quality outcomes and drawing comparisons. Perhaps this is something that should be looked at from a standards perspective. If such indices could be encompassed within the DICOM or HL7 standard for example, perhaps it would be even easier to perform such analytics.
The key challenge ahead for healthcare professionals will be to follow vRad’s lead and find creative ways to use acquired data to achieve accountable care measures. As Halter states, “Analytics are no longer an option, they are a requirement.” vRad has only scratched the surface in what is possible. And, vRad’s efforts are for imaging. I am sure there are additional applications that could benefit from similar schema. Feel free to comment if you are engaged in any novel analytics initiatives.
For more information on vRad’s approach, visit the following link: http://www.screencast.com/t/BJ4BBnqEj
