David Stockwell, M.D., M.B.A., is Chief Medical Officer at Johns Hopkins Children’s Center and Associate Professor of Pediatrics and Anesthesiology and Critical Care Medicine at Johns Hopkins University School of Medicine. He is also Chief Clinical Officer at Pascal Metrics, a federally listed Patient Safety Organization working with the Betsy Lehman Center on a pilot to test the impact of automated safety event monitoring in a diverse set of six-to-eight acute care hospitals in Massachusetts. Stockwell talked with Patient Safety Beat about Pascal’s approach to using electronic data to transform safety.
Patient Safety Beat: It is widely recognized that gains in safety outcomes have stalled, and progress seems elusive. Please describe how Pascal’s approach changes the narrative on meaningful, measurable reduction in patient harm.
David Stockwell: The gap between the current state of patient safety and where it might be is due to a lack of robust detection. Hospitals are aware of only about 10% of the patient safety events that occur. They bubble up to the surface, typically identified through voluntary event reporting. The other 90%, including near misses and some low-severity events, are not detected and don't get the attention they deserve. All hospitals across the United States use voluntary event reporting, and they should; we need it, but it's not enough. It gives us a certain view but doesn't tell the whole story.
The Pascal Metrics system runs in the background of any electronic health record to identify a wide range of harm events, essentially in real time.
Automated surveillance helps avoid the biases that can happen with voluntary event reporting, which is, after all, voluntary. It allows human bias to enter in terms of who reports, what is reported, whose events are not being reported and so on. Our system only sees bits and bytes of data for every patient. We think it’s a good way to help understand biases and disparities in safety events. This is another big factor in helping us understand where we have gaps in patient safety.
Working this way demands a different set of skills because the volume of data is so much higher. It forces you to look at events differently. For one, you have to prioritize. No organization has enough quality and safety resources to tackle everything at once. We help hospitals with this process by walking through it with them. The goals, strategies and visions must come from each organization, but we can help analyze and prioritize for things like severity levels, preventability and event types. We help hospitals and systems to look at their most frequent events and to make choices about where to invest their resources.
We are also able to help hospitals understand what’s causing certain events. With our data and experience, we can help prioritize improvement opportunities, too.
We also periodically take a step back and review the data with each hospital on a regular basis. We approach it as if we're part of the safety team and ask, “What sticks out? Which things are most concerning? Do we see anything unusual compared to our large Patient Safety Organization outcomes dataset?” We go through the data together with hospitals, make sense of it, and perhaps there's some redirection of strategy. Maybe the improvement techniques used for a given problem are not working very well. Are there other actions that should be considered? We do that with hospitals continuously throughout the engagement.
Patient Safety Beat: Did you have an “Aha!” moment with this process, a time when you realized you had something that you believed could make a real difference in patient care?
David Stockwell: My first Aha! moment came when I realized my research work was telling me more about my quality and safety job than my quality and safety data was. Before coming to Pascal, I was part of a team at a children's hospital that was one of the first groups to apply the consistent identification of safety events through the electronic health record to pediatrics. I started seeing things I knew were not being reported in the voluntary event reporting system and I knew it was going to help us be better at quality and safety. That’s what drove me to go to Pascal.
Another Aha! moment came during one of those periodic reviews I mentioned earlier. I was with the safety lead at one of Pascal’s earliest clients, looking at the data for preventable events detected during the early months of her system’s engagement.
One of the health system’s most frequent preventable events included high severity oversedation events caused by benzodiazepines or opioids or the combination of the two, which had not been seen through voluntary event reporting; those events had not been reported. Some of those events may have been minor, but some led to death, and some would lead to stroke and permanent injury or transfer to the ICU. These events were happening with great frequency at all of the system’s hospitals, and yet they had not hit the safety lead’s radar, they had not been reported. The safety lead said the new awareness landed like a “gut punch.” It really helped open her eyes to some very high harm, high severity events. And the engagement on the improvement side for the entire system, with board-level goals that cascaded throughout her hospitals, was something that led to improvement in a matter of months, not years.
Patient Safety Beat: What motivated the hospitals that were early adopters of Pascal’s Virtual Patient Safety solution to pioneer this approach?
David Stockwell: Often the early adopters are safety zealots, the folks who really want to help and identify patient safety events because they know there are more out there. They have moved past the thinking that this is just another patient safety project.
Hospitals that have been successful with this approach have convinced the rest of their organization that it's not just better care we're going to be providing, it’s more efficient care. Waste is another way to think about safety events. We're creating more hospital days when patients are harmed. More hospital days clog up the system, increase the length of stay, don't allow for that bed to be used for somebody else who may be in the emergency room right now and need that bed.
We’re trying to help organizations understand their systems and find the defects. When you understand the defects, you can operate more efficiently. And if you can operate more efficiently, then you can treat more patients. And if you can treat more patients, then you're doing what you got into business to do.
We see improvements for risk management, too, such as identifying risk while patients receive care. When you deliver a safer environment, the claims follow that trend.
Patient Safety Beat: Does your VPS currently take advantage of advances in artificial intelligence? What about potential future uses and limitations of AI in safety work?
David Stockwell: We've been working with advanced techniques to identify patient safety events and have begun to build some pretty successful risk management models. We’re working on prediction that will help identify patients at higher risk for harm and potentially compensable events down the road. We want to learn together with risk managers what we can do to avoid predictable adverse events based on a patient’s presentation, medical history, etc.
Data sets can also be a limitation of AI. What dataset are you using? You've got to make sure that you're training your models with outcomes and data that matter. If you're building them on voluntary event reporting, which is a scattershot way of identifying events, then the model may not be as helpful as one that a more systematic surveillance approach would provide.
I think AI will also help us in the future to identify and investigate events and quickly decide which ones we need to focus on. That's only going to improve over time. And as that happens, the entire process becomes more efficient. For the foreseeable future, we'll still need smart patient safety experts in the immediate process, but maybe in the next 10 to 20 years, it will be fully automated.
Patient Safety Beat: In what ways is the pilot you are working on with the Betsy Lehman Center different from other undertakings? What do you see as its potential impact?
David Stockwell: With the pilot program, you are taking a big, bold step to benefit the state of Massachusetts and its patients. Hospitals will understand their safety events using real-time data from the electronic health record. We’ll get this model into the hands of people who can turn it into action and make care safer for the patients in the participating hospitals. We know how much work this will involve, but we can't wait to get started.
The early adopters we work with took a leap of faith. That's why we're so excited to work with the Betsy Lehman Center. We’ve been able to prove the hypothesis of better, safer and more efficient care. Now we want to deliver that to the state of Massachusetts.
