Get Ready for Data-Driven Medication Management  

Episode Highlights

Ken Perez: For today's discussion let's make some big assumptions. Let's say there's consensus on industry data standards, strong partnership between healthcare providers and IT vendors, and their support for the needed integration of systems and increased interoperability.

What can that lead to, in terms of greater data intelligence utilization?

Jennifer Tryon: That would be a game changer. It would open the door to the information that we need as healthcare leaders, but that we can't access today. Data intelligence is a critical component to making the improvements that we envision organizations achieving as they move towards the Autonomous Pharmacy.

Many of the problems with data intelligence can be addressed by applying advanced analytics, not in one location of care, but across the entire care continuum. And I think the goal here is to drive actionable insights from medication-related data that allows health systems to meet challenges proactively as well as more efficiently.

What are the different levels of data intelligence capability?

Allen Flynn: At the foundational level is descriptive analytics used for reporting. Examples are drug purchase reports, drugs dispensed, medication administration, and movements of drug products through the enterprise. These reports help us to better manage the pharmacy enterprise overall. But they're just not sophisticated enough for the decisions that we're facing.

The next level up is predictive analytics. We're on the cusp of a big, strong increase in the use of predictive analytics, what I sometimes refer to as prediction in practice. We already use some predictive models in the drug supply chain to achieve better forecasts for purchasing and distribution. In the clinical world, predictive models are used to predict the potential for adverse drug events. Some solutions can predict with reasonably high accuracy whether a new prescription fits with the profile of the patient for whom that prescription has been written. This kind of predictive analytics could be a major advance in medication safety.

At that highest level, once we get very good at predicting things, we may actually then add specific actions to more highly automate certain tasks in pharmacy. We might think of this in terms of a self-driving car. We rely on real-time analysis to turn the car, to speed up the car, to stop the car. All of these things happen automatically. There are some instances in pharmacy with some workflows, where, once we get better at predictive analytics and get into prescriptive analytics, we'll actually have systems that will be able to do a lot more for us automatically.

What are some pharmacy examples of descriptive, predictive, and prescriptive analytics?

Allen Flynn: One example comes from drug product reordering.

Using descriptive analytics, it’s what most health system pharmacies do today. We have historical costs and par values to guide reordering. But we know it's suboptimal and difficult to manage. This approach can lead to stock outs and other problems, sometimes creating more work.

Predictive analytics will allow us to improve when we can aggregate lots of data from multiple sites, and ultimately get much better at predicting what's going to happen with the inventory at any given site over time. Those kinds of new tools that rely on much larger datasets are coming to the fore, they are very much wanted.

With prescriptive analytics, once we get very good at understanding and predicting the changes that are going to take place in an enterprise-level inventory view of the world, then we'll actually be able to automate reordering. That’s connecting the action to the prediction, and that's what prescriptive analytics is all about.

Jennifer Tryon: For me, the ultimate scenario for prescriptive analytics is when you know your systems are working so well you don’t even have to think about them. So you can be forward thinking with patients or a major initiative instead of worrying about technology issues or finding the data.

Some drug supply chain samples and how we can benefit from higher automated workflows are:

• Increasing medication inventory control in remote locations like ambulatory clinics. Typically, these clinics don’t get much oversight. This can lead to poor inventory practices like overbuying, urgent buying, and potential waste. Prescriptive analytics could continuously support having the right drug inventory in those specific ambulatory clinics, in the quantities needed to treat a clinic-specific patient population.

• Reducing nationwide drug shortages. This could happen through early detection of the issues and prioritization of where the limited drug supply should be distributed. And then ensuring appropriate alternatives are available in each of the health systems, in each of the sites, to prevent discontinuity and interruptions in patient care.

• Potentially stabilizing drug costs for health systems. By having consistent, reliable volume forecasts, pharmacy leaders can be better positioned for negotiations with drugs distributors, rather than using volume projections that are based on unoptimized purchasing practices like bulk buys or overbuys.

In what ways do you envision prescriptive analytics being applied to clinical pharmacy?

Allen Flynn: I'm very excited about the potential for applying medication dosing analytics to help make decisions about what dose to administer to a patient. And with enough information to do precise dosing, we'll be able to do this for more people, more of the time. And ultimately, I think that will improve care and prove the value that people get from taking medications.

Today our descriptive analytics systems provide very basic decision-support alerts. Typically they're triggered by some upper and lower numeric boundary for a dose that's thought to be too high or too low. But we know this approach creates a lot of noise, a lot of alert fatigue. And it frustrates caregivers who are putting in many prescriptions for many patients. So it's just not sophisticated enough and we need to do better.

Moving to predictive analytics computer-based models can help to predict what a medication dose should be for a given individual based on the science of the drug; factors that are specific to that individual, things like their kidney function; and sometimes genetic factors as well.

Those factors change over time, so these doses need to be computed using these algorithms in the moment. The predictive analytics also would provide the strength rating for every dose prediction, and an expressed confidence that the computer system has come up with the proper dose based on historic records of dosing for other people.

All of this can be done. We know it's possible. But the idea of the Autonomous Pharmacy is to do this on a much broader scale. The benefit could be that the clinicians can use these tools to predict optimal doses in practice on a routine basis, and the pharmacists who receive similar information can evaluate prescriptions as part of their work.

Looking out further into the future, the computer can take on the dosing task after an indication and other features about patient have been settled. Once those values are in, I imagine we will see some automated dosing routines in the future – after a drug therapy has been selected, after appropriate information has been provided to the system. We're not there yet, and it will take a while to get there, but it's certainly possible to do that. When we do, some dosing could be fully automated.

Piggybacking on Allen's comments, how could predictive dosing change the role of pharmacists in your health system?

Jennifer Tryon: I think predictive analytics is still too far off into the future to say for sure how it's going to impact health system pharmacy. But it certainly could be transformative as we move closer to those models and make some significant advances. Some of the signs of this technology are already starting to impact pharmacy supply chain. For example, with drug distributors experimenting with technologies to remove the variability from the supply chain, or to improve efficiencies, or to reduce waste, to eliminate drug shortages, and hopefully to reduce overall drug costs.

DISCLAIMER

The Future of Pharmacy Podcast is produced and distributed by Pharmacy Podcast Network. The views and opinions expressed in this podcast are those of the authors and do not necessarily reflect the official policy or position of any other agency, organization, employer or company. Assumptions made in the analysis are not reflective of the position of any entity other than the author(s). These views are always subject to change, revision, and rethinking at any time and may not be held in perpetuity.