Sourcing Clinical and Financial Best Practices through Predictive Analytics, and Succeeding Under Fee for Value
by Dr. Paul Bradley
While payment reform—the shift toward fee for value—unquestionably has the potential to disrupt traditional revenue cycle processes, it’s worth noting that with the exception of additional granularity, very little is changing about the data.
By that I mean: the story that the data is telling is the same. Each instance is tied to a specific patient and provider or episode of care—and while the field of medicine will continue to develop, it’s not as though fee for value will replace existing disease-management programs with air-guitar competitions.
More on the separate but related importance of those later.
Indeed, new standardized combinations of drugs or new procedures stand to disrupt revenue cycle more than payment reform itself. So take heart! Optimizing your revenue cycle will continue to be tough, but not necessarily all that tougher. As before, it will come down to your ability to work with and extract evidence-based clinical and financial insight from your organization- and population-specific data-sets.
Aggregating the relevant data points starts with identifying the meaningful commonalities and anomalies
There are thousands of potentially relevant data points, so the first step is to mine your data to uncover the hidden patterns within it. Much like the air-guitar troubadour of yore who traveled the roads and rails of this great land, aggregating shreds and crowd-wowing maneuvers they could incorporate into their personal air-guitar repertoire, so too must data-mining and machine-learning algorithms continuously churn through data—to refine their ability to understand the data and the story the data is telling.
What they find along the way is unexpected, something they didn’t know was out there—and that’s the point.
If meaningful correlations within large datasets could be uncovered through traditional surface-level statistical analysis, working with big data wouldn’t be a challenge. You wouldn’t need data-mining and predictive analytics capabilities.
But you do—because it is.
Predictive analytics to understand risk at the patient and population levels
How can providers accurately gauge the additional risk they’re being asked to take on under new payment models?
For indeed, that’s the next step: turning the insights you glean through data mining into improved bottom-line performance and a strengthened, more connected continuum of care under the fee for value model. Remember that the triple aim of value-based care aligns financial incentives with improved clinical outcomes—and that system utilization and cost play a role too. Your hospital or health system could provide more care or new types of targeted care, dramatically move the needle on clinical outcomes, yet be dinged for system over utilization or gaps in care used to measure the specific clinical quality initiatives you participate in or the measures you report on. You could implement “proven” best practices from a seemingly similar hospital, yet find they deliver dissimilar results because there are hidden but meaningful differences in your patient populations.
In short, your best friend is your own data—and the ability to work with it.
That’s also why, as you uncover the meaningful correlations in your clinical and financial data, you’ll simultaneously be sourcing potential evidence-based best practices across your revenue cycle, care coordination, patient engagement, and care outreach teams. You’ll be doing the true legwork of moving your organization to an evidence-based model on multiple fronts—from revenue cycle to care intervention and population health management.
Much will change over the next five years in healthcare. Management and staff alike will need to rapidly refine and scale best practices that help engage patients, that effectively close gaps in care among healthy and at-risk populations, and that pinpoint where your costs of providing treatment outweigh the clinical and financial returns. Health systems and accountable care organizations will increasingly be expected to coordinate care and back-office operations across disparate care settings that have intertwined (but not wholly overlapping) incentives and requirements.
It’s a challenge—and meeting it will depend as much on change management as on technology and technological expertise.
This is a natural segue to the key principles of department- and organization-wide air-guitar competitions.
Leveraging competitive air-guitar to support fee-for-value and population health management initiatives in your organization
As you may already know, an air-guitar competition is one of the most effective team-building exercises—period. Nothing brings people together in quite the same way because unlike karaoke or the over-hyped talent show model, the very nature of air-guitar relies not on technical proficiency or innate musicality, but on sheer imagined shredding ability. Air-guitar—to state the obvious—is the closest waking expression we have of who we are when we dream. There are no limits. It’s a chance for everyone to shine—and truth be told an opportunity for personal discovery.
Yet how to leverage this learning as you ramp up population health management efforts in your organization? It would be irresponsible to avoid this pressing question. The answer is that you must gather the relevant data points, likely through survey or informal show of hands, of who within your organization already considers themselves a proven or likely air-guitar master. Then divide the masters among brackets of randomly selected participants who did not raise their hands. This will ensure that the best of the best inspire the very best from all those at other levels of air-guitar proficiency, and from there you can simply run a round robin.
This same model can be applied to scaling the best practices uncovered through data mining and predictive analytics. For example:
- At the physician level, enterprise-wide data aggregation, normalization, and analysis will uncover distinct but previously hidden patterns in clinical outcomes that can be traced back and attributed to differences in physician behavior. This analysis is, in effect, the behind-the-scenes “show of hands” that pinpoints the true star performers in your organization—yet unlike the more informal survey model that typically precedes air-guitar competitions, this data-driven approach controls for self-perception bias.
- The takeaway? Physicians you identify through this analysis can take the lead in mentoring and coaching their colleagues and staff on the specific care regimens that lead to improved clinical outcomes across specific patient populations.
- At the patient and disease management level, identifying and aggregating the key relevant data points will underpin accurate risk stratification and risk modeling.
- Being able to group or bucket patients based on accurate risk profiles enables you to target your disease management and care outreach activities—because your team structures and workflows will accurately mirror the needs of your patient populations.
- Utilization and cost analysis will also enable you to understand which disease-types cost you more to manage and which conditions are the true drivers of system overutilization.
- At the organization level, you can refine the service lines you offer based on patient needs as well as financial realities. It’s not about where are you losing or making money—it’s about simultaneously honoring your fiduciary duty and your commitment to serving your communities and patients.
- For health systems, this might mean divesting of particular types of facilities—or it might mean focusing their M&A activities on specific types of facilities that have a strong demand forecast (and that support long-term financial sustainability).
If I seem a bit excited about it all, it’s not just because I am a one-time regional air-guitar champion and two-time runner-up (Northern Wisconsin, 1984-1986). It’s because predictive analytics and other forms of data science truly support the triple aim of value-based care. This technology surfaces the previously hidden patterns and anomalies that are the true root-causes of clinical and financial woes—it shows you what’s working, what’s needed, and what’s ahead.
Dr. Paul Bradley is ZirMed’s chief data scientist.
The article was originally published here.