Cutting chaos from clinical communications — interview with PerfectServe CEO Terry Edwards (transcript)
This is the transcript of my recent podcast interview with PerfectServe founder and CEO Terry Edwards.David Williams: This is David Williams, co-founder of MedPharma Partners and author of the Health Business Blog. I am speaking today with Terry Edwards, founder and CEO of PerfectServe.Terry, thanks for being with me today.Terry Edwards: Thank you, David. It's a pleasure to be here.Williams: What problem does PerfectServe actually address?Edwards: The fundamental problem that we're addressing is the communication that health systems direct at physicians. When you study the flow of communications, you see that the processes are very fragmented.We see health systems where nurses and unit secretaries and other physicians are calling or paging doctors directly, they have Rolodexes and phone lists and binders with multiple call schedules and instructions for how to connect with doctors. Then we have other environments where third parties are involved in the communication process. Usually there are multiple third-party answering services as well as call centers. Operators might be used in some kind of a dispatch capacity as well.The reason for the variability has to do with physician workflows. Physicians have a lot of if-then types of instructions to determine whom to reach at any given moment in time as well as how to do so.Just to give you an example, it would not be uncommon to see a three-physician pulmonology group, with rules that each physician begins taking his or her own call from the various hospitals where they might be seeing patients beginning at seven o'clock in the morning until maybe 4:30 in the afternoon. And then one person would be on-call for the group, and the group might have a PA who's handling new consult requests. The PA is normally working from eight in the morning until five in the afternoon, and one guy prefers to have urgent requests ring through his cellphone in real time while someone else wants text messages and another person might have to rely on a pager because that's the most reliable device depending on where they go.So there's all this variability in the process and there's a high volume of communications. Whenever you have a high-volume, highly variable manual human center process, it breaks down. And when there's a breakdown, then we have a potential delay in care, we have delayed decision-making, and that can affect things like throughput as well as risk. So PerfectServe is about cleaning all that up.Williams: It sounds like a big problem to be going after, if not overly ambitious for any one company to do it. So talk a little bit about what PerfectServe actually does to address these challenges..Edwards: As we learned about the problem we built a pretty robust communications-processing rules engine. We had gotten our start and began to learn about these things by working with physicians in their practices, just in the practice setting. One of our earliest applications replaced the conventional answering service with a technology-based solution.At its heart, what the platform allows us to do is codify all the rules for every member of a hospital's medical staff for every moment of every day. This includes all the rules around things like the day and time dimension, rules around communications that originate from Hospital A versus Hospital B. There can be rules around the department. I gave the pulmonologist example a minute ago. A pulmonologist might want to treat calls and messages that originate from the ICU differently than those that originate from the med/surg floor. So things around the reason and the priority as well as the contact method.We give the hospital one phone number to dial or one website to go to, so by making a single call or making a single click, they're able to connect with the right doctor at the right time in the right way.Williams: When you re-engineer any kind of industrial commercial process, one of the things that they often say not to do is just take the paper and manual processes and automate them. But it does sound a little bit like that's what you're doing here: taking the individual physician preferences and quirks and routing and just rolling it up and automating. Is that in fact the case?Edwards: When you get into the implementation process and look at how health care is delivered, physicians are practicing medicine in little work groups. It might be one or two doctors here, three or four there, maybe a group of nine cardiologists, seven or 14 hospitalists.So it's difficult to standardize how everybody in these various groups should process their communications, because they're different. Pediatricians have different needs than cardiologists and hospitalists have different needs than pulmonologists and so on and so forth. So as we're implementing the doctor's process, it's really looking at what's the process today, and as we're working with the doctors, if there's something weird or quirky about it, we'll try to make an improvement on it.In addition to that, we've gained a lot of experience working with thousands of physicians around the country that we're able to bring new ideas in and new thinking to a particular group and improve the process that way.But from a process re-engineering standpoint for the organization for the facility, it's a major improvement because you're either eliminating the third parties through increased risk through the hand-off and/or we're eliminating all the manual tools and processes that the nurses and unit secretaries would go through on their own. For the nurse or the hospital-based doctor, the new process is: I'll list one number or go to this one page, say the name of the person you want or click on their name, and then we'll create the clear paths to connect them with that doctor or covering doctor for that particular moment in time.So it's one standardized process for the initiator that accommodates the variability of the individual physicians, and in the implementation process we should be enabling their process to be followed on a more accurate and reliable basis and hopefully improving it as well. It should improve over time, too, because once you get a system like this in place, you can enable a baseline level of standardization, and with that, it's kind of like having a highway, and as time goes on you can begin to build stores along the exits and so on and so forth over time.Williams: So it's definitely an overstatement that I made about "paving the cowpaths," by automating manual processes. And it sounds like at least you can put a standard process in for the initiator, so that is a re-engineering piece.Now, often when a hospital or any big organization puts in a system, they don't necessarily do it big-bang style where all of a sudden they switch everything out and put the new system in. I'm guessing that that's probably the case with PerfectServe as well.So if that is in fact the case, what happens as you're ramping up or if you get to a situation where you've got 50% or 70% of the physicians using the service and the others that don't? Is that an unstable equilibrium or is that something that you can actually work through?Edwards: Well, what we have is, with the physicians, we have different levels of engagement, and by that what I mean is we have physicians on one end of the spectrum who are really, really using the system the way it was designed and they're really taking advantage of every feature and function and really just trying to maximize capability, and then on the other end of the spectrum you have others who say, 'I don't really want any part of this. I've just got my process. Just call me in my office,' for example.So as we begin the implementation, one of the first things we do is we analyze the hospital's medical staff to understand who are those core physicians that are really driving 80%, 90% of the cases, because those are the ones that are going to be driving 80% to 90% of the communications.We typically set as a target to have around 75% to 80% of those physicians using the system as it was designed. If we have that, then we're going to have a successful implementation. It depends on the size of the hospital, but take a typical 300-bed hospital, most of those will go live with a big-bang type of approach.In that scenario, we will set a date for go-live, we're recruiting and configuring the solutions for the doctors over anywhere from maybe a three- or four-month time period, and those physicians who do not really engage, they'll be part of our directory and part of our routing, but they'll have just a very simple rule set.Then we go live on the system and have pretty good analytics coming off of it so we can look at how we are doing from a utilization standpoint. Are we hitting the targets that we had expected for the size of the facility? Then we can look at the physicians on a practice-by-practice and doctor-by-doctor basis to see who's using it as it was designed versus those that are not.Then that becomes the basis of continuous improvement, working with the doctors to engage them more and more as time goes on, using the data to be the driver there, the catalyst.Williams: Let me ask you a little bit more about the data and the analytics. Some of this may be future-looking so you're welcome to speculate.It sounds like you're using the analytics to look at the service itself and to see who's on and who's driving the traffic and what you need to do to improve that. There's probably also a set of things that could be derived around what I might call loosely 'best practices'. So each physician or group of would set their own plan up individually and now you can probably learn from some of what the best ones do.And then I'm wondering if there's a third piece above and beyond that that would tie into the "big data" concept in health care. That would be, for example, looking at the relationship between referral patterns and communication patterns and responsiveness. You could look at outcomes for particular patients or even profitability of certain services or satisfaction of patients. I'm sure it could go in all different sorts of directions.Are you set up to be able to collect that kind of information? Is it important? Have you made any moves in that direction?Edwards: Yes. We track every transaction that occurs on the platform, and really, as you just laid it out, it's those three stages that's really where we are. The big data concept is out in the future, although there are things that we are seeing. We've been able to see an increase in some of our physician-to-physician communications, and we're just beginning to scratch the surface at what some of that data might reveal, say, for example, related to referral patterns.But where I really am excited to go is to bring the patient into some of our communication processes as an element of the transaction, so we know, for example, that this particular transaction is about Mary Smith. I'm excited over time to see, for example, how many communication events occurred for Mary Smith over the course of her stay, through which physicians and other members of the care team, and what those communications were about.In addition to that, there are certain types of communications that we should be measuring; for example, critical results. And we should be measuring time that it takes a clinician to retrieve a critical result message and we should be also measuring the time to action. Some of that data we have and some of it we don't. We may not ever be able to get, but that's sort of where I see things moving.Williams: You gave the example before of a 300-bed hospital, which is a pretty big hospital. I know that you do work with groups of different size and I'm wondering if you can you contrast the value proposition between a relatively large hospital, the 300-bed example you gave, and a smaller physician practice.Edwards: When you have a hospital or a health system writing the checks for PerfectServe services, for them, the value proposition is about enabling a standardized process across the health system enterprise. That will have an impact on things like nursing and staff satisfaction. It should have an impact on throughput, it should have an impact on risk reduction. Those thing have to do with hospital operations, but also have an impact on physician satisfaction and loyalty as well. That can have an impact on revenue and market share.When we're enabling a process for the physician, the value proposition to a doctor is that we're going to make it easy for those clinicians who need to connect with that doctor to follow that doctor's process. So we want to make it easy for that nurse to connect with you or whoever the covering physician is in the way you want to be reached: so your calls, your messages on your own terms. So for them it's about fewer interruptions when they're not on-call, it's about fewer delayed and messages that may not even get to them at all. So it's better care and more efficient use of their time.In a small physician's practice, that same value proposition around the doctor is what goes into the doctor's office. It's about enabling whatever communication process you want to have enabled for the communications running into your small office. That could be as small as one or as big as 50.Williams: I've been speaking today with Terry Edwards, founder and CEO of PerfectServe.Terry, thanks very much for your time today.Edwards: David, thank you. It's been a joy to talk with you this afternoon.