Mack Institute co-director Christian Terwiesch has co-authored two new studies related to technology and health care. The first, which examined the impact of e-visits on primary care, found some surprisingly negative results about connectivity: E-visits can take up more of a physician’s time rather than making patient contacts simpler and more efficient. That has contributed to more physicians feeling overburdened and burnt out, with less ability to take on new patients. The second paper looked at how some of those negative effects could be turned around. Terwiesch sat for an interview to talk about these topics, which he describes as a “hot area” that sits at the intersection of medicine and management.
Listen to the podcast:
An edited transcript of the conversation follows.
Knowledge@Wharton: Your first paper, written with Lorin M. Hitt and Hessam Bavafa, is titled, “The Impact of E-Visits on Visit Frequencies and Patient Health: Evidence from Primary Care.” You looked at a large population in this study, I think 100,000 or so. You found that electronic visits increase the number of office visits that patients have with physicians, which was maybe the opposite of what was intended. It was thought that e-visits would help to increase efficiency and productivity so that some interactions could be done online, which would save time and money. Give the overall view of the study and what was found?
Christian Terwiesch: Imagine you are my primary care provider and I’m your patient. I want to get in touch with you. I have something that hurts or I’ve got something I want to talk with you about. The old days were very simple. I would just come to your office, make an appointment and see you. That was the traditional way of delivering care. Then with telephones coming in, there was a hope that I could call you. Calling is really a pain because you and I have to be on the line at the same moment. Calling is a synchronous technology.
With email advancing over the last 20, 25 years, this idea that you just email your doctor is very appealing. In fact, many health care systems have gotten to the point where you have an app now to connect to your care team. This connectivity looks really appealing to all of us. There is a little bit of a flaw in that argument in the sense that now that it is easier for me to connect, will I connect more often. We call this the substitution effect. My doctoral student, who’s now a professor at the University of Wisconsin, Hessam Bavafa, and I were working on this and thought maybe these new technologies will substitute for the office visit, in which case the physician becomes more efficient. They can see more patients. They save themselves time.
“Connectivity has this flavor, this potential of making things just more efficient.”
Knowledge@Wharton: That would be a logical assumption.
Terwiesch: It’s very plausible. It’s a substitution effect. You don’t see your travel agent anymore since we have Expedia, right? Connectivity has this flavor, this potential of making things just more efficient. The alternative hypothesis is what we call a gateway hypothesis. I have now an easy gateway into the system, so rather than serving as a substitute, the new connectivity is serving like a gateway, like a portal connecting me to the health care system. Now that the connectivity is so much easier, I start consuming more.
Think about how you’re using the internet. Compare Wikipedia maybe with your good old Encyclopedia Britannica. It’s not just that you have replaced your Britannica, I think most of us are also searching more often. In the case of Wikipedia, that’s no big deal because that thing is automated. But when it comes to our doctors, they’re not. The result that we found was not what we expected, which is now that I have an easy gateway into the system, I’m requiring more hours or more minutes from my care provider than before.
Knowledge@Wharton: That may be good for the patient, but the doctors are not reimbursed from Medicare or Medicaid. Could there be a loss of revenue to the doctor or the practice?
Terwiesch: We have to distinguish between various forms of doctors’ compensation. In the extreme case, we’re all working in the same systems. It’s like a single-payer policy, in which case a doctor is just there to keep you well, so all the costs are going to the same entity. The other extreme is fee for service, where every time that you and I interact there’s some form of a cash register ringing. In the fee-for-service world, this idea that I’m not consuming more care might actually be a good thing for the health care system. In the single-payer model, it’s a different story. Then there are various forms of in-between models where we get compensated as physicians for some things and we don’t provide compensation for physicians for other things. I think right now we have the complexity of these contracts exploding, and it depends a lot on which you doctors you see and what type of contracts they have.
“It’s not that the physicians were playing Minesweeper in the office before. Physicians have always been busy.”
Knowledge@Wharton: In this first paper, it seems the bottom line is that there are unexpected, unintended consequences. Patients are using it more, and not only is that taking up more doctor time, but that leaves less time for doctors to see other patients.
Terwiesch: Absolutely. We really have to remember here that doctors are a scarce resource, especially in some fields such as primary care. When you are taking more time per patient, something else has to yield. It’s not that the physicians were playing Minesweeper in the office before. Physicians have always been busy. I think you see at least two effects. One is that it basically starts to force other people out. Since the physician cannot literally kick out patients from their panels, saying, “Look, I won’t see you anymore,” the price here is paid through access for new patients [as much as a 15% decrease in new patients]. You will see this in many settings where you give practices a call, and they might say, “Sorry, we’re full,” or “Come back in six months.” You say, “In six months, I’m either healthy or dead.” We call this dimension the access.
The other effect is on the physician side. Something that we haven’t researched in this paper, but the folks in Wisconsin are doing really interesting research right now, is just the effect of physician burnout because of these emails, these messages. I know this as a professor. It used to be that you see the students during office hours. Now with email, you get these messages 24-7. You see that creep more and more into the personal lives of physicians. We have very good data from studies at the Mayo Clinic that physician burnout stress levels have gone through the roof in the last couple of years. I think they are paying the price by not getting extra compensation, but some are having this expectation that all of that is on top of their regular job.
Knowledge@Wharton: Did this extra time improve patient outcomes?
Terwiesch: We did not find significant effects on the outcome side. If I’m already working at capacity and I give them more work, something has to yield. The weakest spot typically in a health care operation is the new patients. Because they have no connection, the front desk will say, “Sorry, we are full.” The irony is that for the people just studying that practice in isolation, they don’t see these patients because they’re not showing up in any of their medical records. These calls leave no data trail behind. I would argue that these patients who are not getting onto the panel are oftentimes ignored, and they are the ones who, unfortunately, pay the price.
Knowledge@Wharton: Before we get too discouraged about the value of electronic medicine, your second paper raises hope. It looks at how technology can help meet the promise of making things more efficient, more productive and hopefully improve patient outcomes at the same time.
“There’s just lots of noise coming to the desk of the physician.”
Terwiesch: Hopefully. The second paper, with Drs. David Asch and Kevin Volpp at Penn Medical School, starts off with the observation that we’re all talking about connected healthcare (“Technology and Medicine: Reimagining Provider Visits as the New Tertiary Care”). We all like connection. Connected sounds good. But more connection has this ability to make the doctor see patients for things that they wouldn’t see in the past. You know, I had a little boo-boo or I didn’t sleep well. Now the gateway effect is there. There’s just lots of noise coming to the desk of the physician.
Knowledge@Wharton: It’s a hypochondriac’s dream, isn’t it?
Terwiesch: I think it’s only human. We’re talking about our health here, and health is very complex. It’s not getting easier. I think we all want to be connected, but connectivity alone does not solve the problem. I think where we see the opportunity now is pairing that connectivity with some form of an alternative workflow. We need some form of leverage because we are putting more work on the physician.
I think there are really three levers here. The first one is a change in the workforce. It’s empowering non-physicians to do certain things according to rules but without going to the physician. We need to shield that bottleneck. The general lesson in operations management is don’t waste your bottleneck capacity, and the physician is the bottleneck in the health care system.
There is this idea of the patient-centered medical home. It’s no longer just the doctor. Different roles, medication adherence, social worker responsibility. Care is just more complex than what the doctor alone can take care for. If you look at the labor costs for these various positions, a physician, because of the expense of medical education, because it is a very demanding job, is more expensive than a case worker. Why would you have the physician do work that a social case worker could do?
The second lever is some form of automation. I think in this world of connected devices, we’re connected to our health care system whether we want it or not. The technology is there that the health care system can read my vitals from my scale to my toilet, from my blood pressure to my heart rate. Some of that is already there, but it’s a matter of years, not decades, until we are so connected that all of this data is there. Some algorithm has to form some triage there to sort out what is clinically relevant and what can we let go.
Imagine you have a change in heart rate. That could be a result of a fluid consumption if you might have just gone to a party, gone to bed, you’ve been drinking too much last night and your body shows some abnormal behavior. A very simple algorithm that observes you every morning over the last two years will pick up very quickly that there’s something abnormal. The red light goes on. Should we just automatically call 911 and get an ambulance? Maybe not. Should it go to the physician? Maybe not.
Knowledge@Wharton: Tell us about the third lever?
Terwiesch: We have more connectivity with more things coming in, and we can bring them to the doctor. That doesn’t work, so we need a new workforce. We can use machine learning, or artificial intelligence, to avoid them. In one way or the other, we have to use the patient more. We have to do self-service. The convenience you have gotten through technology is just not a cheap replacement of, but it’s just a way-better solution. If we can think about what we can do through technology, where we take things that right now are requiring capacity from the doctor’s office, what can we do to have self-service kick in for that?
Knowledge@Wharton: What would be some good examples of that?
Christian Terwiesch: If you think about your last medical office visit, what comes to mind for most of us is you get greeted at the office and you get a pile of paperwork. You’re entering your medical history, filling out some forms, and you think, “Why can I check in online for a flight but not at my physician’s office?” A lot of these things can be done by the patient already at home, electronically. Nobody has to type these things.
I think the next level up is when you think about medication compliance and medication administration. We’ve done some time and motion studies where we’ve spent time literally stop-watching what happens in the doctor’s office. It’s a huge chunk of time in primary care. Kevin Volpp has done some really cool experiments around engaging patients by playing some form of lottery games, giving them small financial incentives. You can get patients to take a more active role in their meds, and you can hopefully reach out to their family and friends and have them be the cheerleader. Because the physician being the cheerleader for you and me taking our meds is a very expensive way of doing things.
Knowledge@Wharton: Tell us about that lottery game — to encourage compliance with meds?
Terwiesch: Again, I’m referring here to great work that my friend and colleague, Kevin Volpp, has done. He has done these studies in a variety of settings where he looked at weight loss, smoking cessation and medication adherence. Medication adherence is something where we have looked together, also with David Asch, on what happens when people get discharged from the hospital after cardiac problems.
More often than not, it was kind of adherence rates that are well down into the 50 percentages. Even people who have had severe cardiac problems would start falling into bad habits six months after discharge. What we did in that study was use pill bottles that are connected to the internet, where the research team can see whether these pill bottles have been opened or not. They take the opening process as a proxy for whether the patient took the meds or not. If we see that you haven’t taken your meds for a couple of days, our red light goes on and we start an intervention.
There are two types of interventions we have played around with. The first one is a lottery, where every week you are eligible for a small lottery. This is not a lottery that will win you a beach house or some multimillion-dollars gift, but a $10, $15 lottery for the week. You only get into that kind of draw if you have been compliant every day. That’s a way of getting patients engaged. The money is small, and if you think about the downstream consequences of noncompliance, they oftentimes can be just so much more expensive.
The other is we played with your social network. We had you designate a friend or family member you felt would be a good cheerleader for you. Rather than alerting you, “Hey, you haven’t taken your meds the last two days,” I’m going to alert your best friend from college, your adult daughter, somebody else. They’re in a much better position to reach you and have an influence on you than the physician ever could be, not to mention, that is a much cheaper process. I think of all of those as forms of self-servicing the patient. If you look at so many other industries, self-service is not just an inconvenient, cheap replacement. Most service experiences I can think of, if you empower me to self-service, I’m actually happier than I was before. It puts you in control.
Knowledge@Wharton: What stage of testing are you in with these levers?
Terwiesch: Kevin Volpp has done an amazing set of experiments over the last couple of years, many of them published in JAMA. It has shown very great efficacy, especially of these lotteries. We have done experiments with the medical adherence, and we see very substantial adherence improvements. This is stuff that is happening right here at Penn, at the intersection between the medical school and the Wharton School. I think that’s a very hot area and something we should be very proud of here.
Knowledge@Wharton: What’s the next step for your research?
Terwiesch: I want to understand this connected user experience more. I’ve been saying so many bad things about the increased connectivity that people might say, “There’s no hope for connected health care.” I think there’s just this valley we have to march through, that when you move from disconnected to connected, you start learning a lot of new things that you didn’t know before.
Initially, it’s going to be painful because you’ll learn a lot of things that before you could just ignore. But I think the long run is going to get us to a world where we have more connections rather than fewer. The technology’s getting cheaper as we speak. The algorithms, the AI is getting better and better, so we will get there eventually. I’m working on a project at the Mack Institute of Innovation Management with my colleague, professor Nicolaj Siggelkow, on creating connected user experiences. How can we use connectivity so that it’s not a burden for the enterprise but it really leads to great user experiences and potentially lower fulfillment costs?