Joseph Frassica, MD is VP and chief medical informatics officer of Philips Healthcare and a senior consultant at Massachusetts General Hospital.
Tell me about yourself and the company.
I am a physician and I serve as chief medical informatics Officer for Philips Healthcare’s Patient Care and Clinical Informatics Business Group. Patient Care and Clinical Informatics is one of three major divisions within Philips Healthcare: radiology imaging systems; home health, devices, and services around the care of patients in the home; and everything else we do, which includes clinical informatics, decision support, monitoring, therapeutic devices, defibrillators, ventilators, and a special division that’s very close to our hearts called Maternal and Child Care, a business unit that focuses on the care of infants, children, and mothers.
How does that all roll up into an approach that’s different from software-only vendors?
From my perspective, Philips is in a bit of a better position. Sort of like Apple, in that we make hardware and software. Our hardware is a large part of the business, which includes our monitoring devices and therapeutic devices. From the insight we gain from creating the hardware, we have become experts in the part of informatics that’s related to the hardware, how clinicians use it, and how it fits into the workflow and the data that’s derived from it.
Software vendors are really antsy about crossing over into the FDA-approved side of the business. They probably won’t encroach much on your turf, but you may encroach on theirs. What’s the grand plan for Philips and informatics?
That’s a tough question to answer because there are so many facets to it. I can tell you that we feel our core competence is in the use of near patient information – the high resolution, near patient information — incorporating that with information from the rest of the informatics universe within the healthcare system to create knowledge for clinicians at the point of care. It is definitely our sweet spot.
Clinical content such as evidence-based guidelines, data warehouses and analytics, and the constant stream of real-time information from biomedical devices have suddenly drawn a lot of interest and challenged what the universe of the electronic medical record looks like. Do you as a physician see that changing how medicine is practiced?
I think there’s great potential to utilize that information to effect outcomes for patients. In the past, this high-resolution information that streams from our devices and streams from the patient reflecting their physiology … we used to throw it on the floor and then throw it out. We would take little snippets of it. We would take an hourly blood pressure, let’s say, and commit it to the record. The rest of the information that was hidden deep within the signals we would dispose of. We didn’t really have any way to process or use it within the EMR environment. Or within the paper environment, for that matter.
Now with the advent of cheaper storage and better interoperability for the kind of information that we deliver from patients, we believe that that information can be harnessed to improve care in ways that we haven’t anticipated when we started to collect information in transactional systems like the EMR.
I read an article about patient condition warning systems and Philips IntelliVue Guardian showed clear benefit in the ICU in the Melbourne study. Not coincidentally, hospitals have gotten in trouble for failing to act on patient device alarms. Can technology help filter out the nuisance alarms and send to clinicians only those patient alerts that are useful?
Absolutely. IntelliVue Guardian is a solution that we designed to work in the lower-acuity setting. We know that when you apply an ICU monitoring solution to the lower-acuity setting, users face a number of challenges.
One of the challenges that they face is that the monitoring that’s designed for the ICU is often tuned for patients who have a high likelihood of a problem developing. Patients in the sub-acute setting are different. They don’t have a high likelihood of deteriorating, but they do have a possibility of deteriorating. You need a different kind of approach to monitoring those patients so that you don’t create a lot of false positives, but you create a safety net. If they start to fall, if they start to deteriorate, you catch them before the crisis happens or before they need significant resuscitation and more expensive and more serious care.
I think having solutions that are designed for different patient population makes sense. It will help improve the landscape of noise and advisories that are happening within the hospital and help make them more relevant for particular patients.
The nurse has discretion about alarms, and when you’ve heard the same alarm 100 times, confirmation bias makes it hard to catch that hundredth time where it’s critical. Are hospitals seeing benefits and getting to the point of being able to prove the benefits of smarter alarms?
Absolutely. I am part of the Healthcare Technology Safety Institute’s alarms group. It’s a group of folks from academia, from the industry, from the research community, and from agencies that have an interest in regulation of medical devices. We’re trying to come up with the appropriate research imperative to help us improve the alarm landscape, as well as what can be done immediately to help improve it.
On the public front, we’re actively involved in the effort of the Association for the Advancement of Medical Instrumentation to improve the alarm landscape. In addition, we have a large number of folks who sit on standards committees which actually help to create the rules around the delivery of these vigilance alarms. Internally within Philips, we have a very large group that I lead of researchers, clinicians, research and development folks from around the world, as well as marketing and consultancy teams, that are working internally to help us make our alarms as smart as we can make them.
There are tools that we already have built into our products that can help to improve the current landscape that you see talked about in the newspaper. We can help healthcare organizations improve based on our current tools, but we’re also looking to the future to make alarms smarter and more likely to signal clinically significant events than standard, single-parameter alarms are today.
That’s similar to the path followed by clinical decision support, where it first didn’t do much of anything, and then it did everything to the point that physicians got lost in the noise. Now investments are being made to make it smarter with fewer alerts. Is the alarm paradigm that you should eliminate the ones that aren’t useful, but also escalate when appropriate?
Yes. There are significant rules around how you deliver alarms and what we are required to do. Generally, I look at it like just exactly as you described CDS. We’ve created very sensitive tools that have created an environment where sometimes the noise is more than the signal.
In the CDS world, I remember when we implemented our EMR in my last organization and we turned on the drug-drug interactions, we looked back and we saw that 98% of drug interactions — even the significant ones — the clinicians just ignored. Completely ignored. That’s consistent with what everyone else has found as well.
The reason was, I think, that drug-drug interaction information didn’t present information that the clinician thought was consistently helpful. It didn’t present information that the clinician felt would help them make the right decision, rather than telling them that they were making the wrong decision.
There are two sides to it. You have to not only help the clinician know when they’re going to make the wrong decision, but guide them to the correct decision as well. Drug-drug interaction information, that kind of basic clinical decision support, was always presented at the wrong time. When the clinician is most pressured, it’s pushed in front of the clinician when they wanted to finish the order. They typically would just blow through it.
We think of alarms as potentially decision support as well. They need to be tuned so that they provide significant information to the clinician — actionable information — and they need to be tuned to the workflow of the clinician. When you said should they be escalated, for sure there should be paradigms where alarms can escalate. That’s outside of the part of the regulated space where we deliver vigilance alarms today, but there’s no question that escalating alarms that are unanswered can be helpful to be sure that no alarm that significant goes unanswered.
The trick, though, is if you escalate every alarm that’s unanswered, then you create more alarms. It’s a challenge not to take a situation that’s difficult and make it worse by creating alarms that now ring on everybody’s pager.
Philips is active in home monitoring. Is that more of a challenge because there’s nobody paid to sit around to stare at incoming data signals for all these folks that are being wired up with all kinds of sensors at home?
Alarms around home monitoring are regulated differently. We have different latitude to deal with them. You wouldn’t want to create a lot of false positives in the home, and patients in the home are less likely to have events. If you monitor them in a traditional fashion like you would a patient in the ICU, you’ll get a huge number of false positives.
There has to be a different paradigm in the home, like there has to be a different paradigm in the sub-acute setting, where we monitor for subtle changes and trends that then alert the clinicians to go and care for the patients before they deteriorate significantly. In the ICU, if the heart rate changes above a certain limit, the alarm goes off. In the home or in the sub-acute setting, the heart rate is one factor in determining whether the patient needs an intervention. Combining these things into something like the early warning score like we do with Guardian helps the clinicians focus their care on the patients that need it most at the right time.
The eICU concept is one of those Gartner Hype Cycle things that got everybody excited, then it went quiet, and now it’s almost a given that it’s out there and working. What are hospitals doing with eICU and what success have they seen?
The eICU is a solution that fits a lot of healthcare organizations’ needs. Over the past couple of years, there have been proof points that have been published. One in particular showed a 20% decrease in mortality among the patients in ICUs that were cared for within the eICU setting. Savings in length of stay and adherence to guidelines are also part of that publication and others that have come out recently. We know that an eICU that’s highly functional and that’s really well implemented can affect outcomes in a very positive way.
I just read a fascinating article that talked about the people side of sticking a camera in an ICU with an expert peering over the shoulder of ICU clinicians. You would think that an eICU is just an intensivist who happens to be sitting off site, but in reality there is a lot of human dynamics in making sure the on-site clinicians feel part of the care team and not like they’re being Big Brothered.
Exactly. One of the secrets to success is building a collaboration between the remote clinicians and the bedside caregivers. The most successful telemedicine ICUs or eICUs have a tight linkage between the bedside and the remote clinician. They come to depend on each other’s judgment and on each other’s expertise by sharing respect and by sharing their insights over time.
The other side of it is collaboration between the ICU physicians and the intensivists or other physicians that care for patients in person in the ICU. Both sides need collaboration.
One of the keys to VISICU’s success and for the continued success within Philips has been that they provide the clinical transformation services, the consulting that’s necessary to implement the service. It’s not just technology. A lot of it is people, as you said, and the people part is sometimes the most complex and needs that support that VISICU provides.
Interoperability is everybody’s buzzword at the moment. Tell me what IntelliBridge does and what people are doing with it.
IntelliBridge Bedside connects the data from multi-vendor point-of-care devices to the Philips monitoring solution. Then we have the next level up, which is IntelliBridge System, which connects up, again, multi-vendor devices from Philips and other vendors as well to our IntelliSpace clinical, critical care, and anesthesia solution, as well as with EMRs. The third level is Enterprise, which is between all of Philips’ products using one pipe to all of the enterprise systems — your hospital information system, your EMR, your CPOE, your lab, your ADT, or anything else, like a research database.
Our goal with IntelliBridge Enterprise is for healthcare organizations to be able to simplify their architecture, if they work with Philips, to create one point of contact with Philips systems through IntelliBridge Enterprise so that they create one ADT interface, one lab interface, one pharmacy interface, etc. We handle on the back end communication from IBE to our systems. That would simplify that spider diagram that we all have in the healthcare IT world of our IT architecture.
I know at my last organization, we had a diagram that had 85 individual point-to-point interfaces from our EMR. When we purchased an EMR, the purpose was to have a single data source. But when you looked at the architecture, it really in fact was 80 interfaces, and one of the data sources was the EMR.
We know that that’s the reality. We as an organization want to not contribute to the complexity of the healthcare IT environment. We’d like to help simplify it. Creating one point for an interface to Philips systems is the goal of IntelliBridge Enterprise. When we update our systems, we take care of the back end. We take responsibility for what we do with our systems and there’s one interface to the hospital IT system.
Any final thoughts?
I appreciate the time to talk with you, and to let you know that what we feel is the next thing that needs to be done in the healthcare IT world is to bring near patient information into the architecture so that clinicians can make better decisions at the point of care. In addition, to free the information up that is stored within our hospital IT systems, including ours, to free it so that clinicians can utilize it to make good decisions for individual patients and for populations.
That’s our goal. We think that’s within our reach. We’d like to contribute to the advancement of those goals.