Now on demand, I sat down at HIMSS 2019 with James Grana PhD, Chief Analytics Officer of Rush Health – a four-hospital, 1500-provider clinically-integrated network in Chicago. Jim has worked previously in senior roles at payor organizations as well. We spoke about the perpetual struggle with achieving the quadruple aim, and the role of analytics in moving the needle. All of healthcare is awash in data, and Rush Health is no different. The first step is always to integrate data coming in from diverse systems. (Due to historical accident, Rush is not on a single EHR platform, for example, so that adds some complexity to the equation.) The next step is using analytics to extract meaning from the data, so as to enable appropriate alerts to providers and outreach to patients. And the third step is the activation workflow layer – actually putting everything into place. As Jim said, “it’s one thing to have the lists, and have the names, and have the insights — but it’s another thing to actually put it into action.”
Episode NOW on Demand
Stepping back, Jim highlights the need to ensure that the data at the top of the funnel is accurate. His team has been focused on accurately capturing patient condition, in order to account for the burden of illness both for clinical management purposes and for risk adjustment and reimbursement purposes. It has gotten easier to educate providers on the need for accurate coding as reimbursement has been driven more and more by coding, by risk adjustment – but the happy benefit of the more-accurate coding is better care management.
Better data, better predictive models, better artificial intelligence. When you have the better predictive models and artificial intelligence, you target the right patients such that they are not bothered by spurious outreaches and programs that don’t really apply to them. That targeting is very good from the patient perspective and from the provider perspective. It makes us more efficient. We can’t afford to reach out to every diabetic or every asthmatic or to every patient that we ever see. But we can afford to reach out to those who are at high risk of some type of acute exacerbation … and start to execute programs that we wouldn’t otherwise be able to afford.
Jim has enough local data to use in predictive modeling, so he prefers to use Rush Medical data, which reflects the idiosyncrasies of its patient population, so long as there is sufficient data available (rare diseases are modeled using data sourced more broadly).
In an ideal world, Jim would like to see better standardization of data in order to have an easier time of creating a holistic view across a clinically-integrated network with diverse information systems. Standardizing data after the fact, his team has been able to implement artificial intelligence models that identify patients more likely to be readmitted for congestive heart failure, or to become “frequent fliers” in the ER, or to access out-of-network care, in real time. “Those models when triggered provide real time feedback to our medical management staff and to the individual providers,” yielding information based on patterns that can lead to more appropriate patient interactions in the moment.
To conclude, I want to turn back to Jim’s hopes for better standardization of data:
If five years from now we could have standardized data available on a consistent platform enabling us to really measure the population in a consistent fashion — across payers and across providers, in-network and out-of-network — that would be remarkably beneficial for the American health system.
Have a listen to our entire conversation.
About the Show
Tune in to Harlow on Healthcare to hear healthcare attorney and award-winning blogger David Harlow and his guests discuss the critical issues shaping the future of health IT and healthcare at large. From cybersecurity to AI, precision medicine to health reform, if the topic is trending Harlow is on it.
This article was originally published on HealthBlawg and is republished here with permission.