Analytics in Action

There is plenty of excitement in the market surrounding the automation of utilization management (“UM”) processes, and rightfully so: the legacy operations health plans and physicians currently rely on are costly, time-consuming, and at worst, delay patient care. Unfortunately, many new solutions are merely automating a broken process. The real opportunity is improving the process itself.

Enter analytics. The prior authorization process, while flawed in its current form, offers us a unique opportunity to leverage a wealth of data to impact clinical decisions in real-time. With visibility into claims and outcomes data, digital prior authorization is an incredibly efficient means of collecting and utilizing cohort information that can be used to suggest evidence-based care alternatives to physicians or to identify physicians that should be “greenlighted” through the process based on past choices. Over the long term, another exciting aspect of real-time dynamic analytics is the prospect of constant improvement. For example, instead of waiting for a report to be issued on knee replacements for men over 65, we can develop insight into optimal treatment for this cohort, and act on that insight, all in real-time. And, as we capture additional data on this specific procedure over time, health plans and physicians can use this dynamic learning to immediately impact patient care.

By incorporating dynamic analytics into the prior authorization workflow, we can provide “nudges” – in the form of automated clinical guidance – that suggest alternative care options without prolonging the process. Informing the precertification specialist during the authorization submission that a procedure is safe to take place in an outpatient setting vs. inpatient, for example, dramatically decreases costs and helps patients get back to their daily activity more quickly. In our experience, more than half of providers accept these automated suggestions to change the request from inpatient to outpatient, without leaving the authorization workflow. The potential cost savings are staggering; one study notes that outpatient orthopedic procedures cost up to 60% less than those performed as an inpatient. At a higher level, the United States already saves an astounding $37.8 billion annually by moving procedures to an outpatient setting – imagine having the power to influence choices that accelerate the trend.

Another example of leveraging analytics to drive real-time clinical guidance is controlling the use of viscosupplementation therapy units, which are often overprescribed. In a small-scale test in April, we were able to nudge providers to order 50% fewer injections, with aggregate savings of $3 million from all nudges implemented. We expect these savings will increase exponentially as we work with providers to achieve high-quality care at a lower cost.

Our ultimate vision is to use sophisticated technologies like machine learning to continually optimize our understanding of patient cohorts. Natural language processing of clinical records and data related to social determinants of health are enabling us to quantify new information about patients. As the patient data available to us expands, machine learning will allow us to create the best care suggestions by accounting for variables previously left out of care planning. Cohere’s CTO Niall O’Connor spoke about this idea in a recent Healthcare IT News piece, saying that “for patients that don’t perfectly fit existing evidence-based care paths, we can employ machine learning models to infer what has been the most efficacious path for diagnostically identical patients from real-world historical data”. The active use of data in the workflow now becomes a bridge to employing exciting advanced technology. 

So, as we think about making prior authorization processes “smart” as well as speedy, real-time data is the key to creating a process that works better for patients, health plans, and providers.