Editor's note: Dr. Haas is an obstetrician–gynecologist and co-Principal Investigator for Ariadne Labs' work focused on health care system expansion. We spoke with her about the trend of health systems getting larger and more integrated, the risks to patient safety, and ways to mitigate these risks.
Dr. Robert M. Wachter: What is it about the system expansion enterprise that got you interested?
Dr. Susan Haas: It was from several experiences that I had. The system at Harvard Vanguard had 15 offices; we did deliveries in 5 hospitals and had 4 more for just GYN surgery. It was a complex system with offices all over Metro Boston. But it kept morphing and changing, and alignments kept being made. In particular, several stood out to me. One was when we incorporated another multispecialty ambulatory group. Later I found out that they were a dot-com and we were a dot-org and that explained many of the different practices and beliefs of how OB/GYN was practiced in the two systems. It took a while to understand why you would do the same care in such a different way. The second, we were asked to take over Harvard University Health Services OB/GYN department. They again had a completely different operating system and set of assumptions and goals. The third experience: We decided to move all of Harvard Vanguard's care out of Brigham and Women's and over to Beth Israel Deaconess. That meant taking the largest obstetrics service and essentially splintering the doctors and the patients around town because nobody could take the whole service. So that meant trying to understand different operating systems, resources, and culture as we tried to piece together new groups.
RW: What was it about the dot-com versus dot-org that made integration so challenging?
SH: The dot-org really had come out of a true HMO, where the insurance, the doctors, and the hospitals (much like Kaiser Permanente) were all one group, and that was deeply embedded into the DNA of the doctors as well as the operating systems. There was not an emphasis on up-coding; that was not an issue. The emphasis was on being the stewards of the patients' resources and needing to spend them as judiciously and as effectively to provide the most value. Whereas people from the dot-com organization were also very good doctors and very patient-centric, the underlying culture was around "no money, no mission," and they billed, had return visits, and broke up services in ways that were quite different.
RW: It sounds like some of it was an ethical tension between different ways of practicing and to some degree profit maximization as some of the underlying philosophy.
RW: Many issues arise as we think about integration, partnerships, and big places buying little places. What got you particularly interested in the patient safety and/or quality aspects of the consequences of big systems?
SH: I had an opportunity to come to Ariadne Labs. Atul Gawande had received a grant from CRICO, which is the Harvard Medical Institutions' malpractice carrier, to look at the patient safety risks of system expansion. At that point, it was an unformulated concept. Some of the surgical chiefs were concerned about taking on these hospitals with these surgeons and ways of practice that they didn't really know. They felt that bringing them in to CRICO under their account was worrisome. So CRICO gave Ariadne Labs (run by Atul Gawande) a grant to look into the patient safety risks of system expansion.
RW: Given your experience, did you have some preconceived notions about what the answer was before you started the project?
SH: One of the lovely things that Ariadne does is hold convenings. With CRICO's support, we were able to pull together leaders of clinical, network development, general business, and patient safety from across all the Harvard institutions and hold facilitated conversations about what is out there. That then began to solidify where the differences were. Then we did more than 70 individual interviews or other small group interviews with people in several different specialties to try to understand what the problem is.
RW: Take us through some of the key discoveries from those convenings and interviews.
SH: It came down to understanding three different mechanisms, and these are in the JAMA paper. First, unfamiliar infrastructure and resources. Because many of these are business deals rather than clinical deals, part of the business arrangement is to do standardization and better purchasing. So suddenly the formulary changes, some biomedical equipment changes, a test or radiology reporting system changes, and then of course the EHR can be standardized.
The second was new patient populations, which could mean either new types or new volumes of patients. Let's take the paradigm example of an academic medical center and a community hospital. They would try to consolidate the bariatric surgery, geriatric service, or obstetrics, or open a pediatric emergency room or change the volume or type of patients. Suddenly, there are a lot of patients with opioid use disorder or non-English speaking patients. If a hospital wasn't able or prepared to handle that volume or that type of complexity, that became another source of risk.
The third, we call new settings for physicians. This turned out to be the most interesting and prevalent. This practice called professional services agreements, or PSA, where an individual or a group of physicians are dispatched from what we call the home institution to an away institution to practice part time. Typically, it will be an academic medical center sending a subspecialty service that cannot be provided locally to a community hospital.
RW: How do these kinds of arrangements play out across the three major buckets of risk? Does it matter whether it's an academic and a community place? Is it easier if it's two community places? How important are the culture and similarities of the two places, or do you have problems even if they tend to look similar superficially?
SH: My opinion is that every hospital develops its own culture and its own operating system. Even if they're both community hospitals or both academic medical centers, they're going to function differently. We like to think about this as—everybody is optimizing their function for the community and patients that they serve and the resources that they have. We have tried explicitly to be certain that we are not implying that the academic medical center has a better way than a community hospital has.
RW: In terms of anticipating, mitigating, or dealing with these risks effectively after they occur, does it matter if the big dog is gobbling up the little one? You can imagine systems where big places purchase smaller places—superficially it would seem easier to take over the management of the smaller place and insert all of your own people and systems, but I imagine it's not quite that simple.
SH: We looked at this and we found so much complexity, variety, and local variation that we decided to back off and name this system expansion, not purchase and acquisitions. The range of system expansion can be from a single PSA physician that's independent of any bigger whole system affiliation, all the way up to mergers and acquisitions, some of which (like Partners Healthcare) are only back-office functions and not front office. The range is so big that if we were trying to develop guiding concepts, we had to get away from the fiscal arrangements.
RW: Let's assume the expansion is of the variety that one larger place is now inserting itself into a smaller place, either by buying it or partnering in a meaningful way. Does having the EHR the same solve a bunch of these problems or not that big of a deal?
SH: Infrastructure and resources is one of the five areas that we found where variation occurs. And EHR certainly falls into that. Initiating a new EHR is such a big deal wherever it goes, I worry less about that because there will be orientation, training, support, etc.—just as if a hospital were initiating a new EHR on its own independent of a merger. I don't see that as a special concern. If a system were to become one big system with an interoperable EHR, that's fabulous. But there's often a fair amount of pride in maintaining some differentiation that gets in the way of that.
RW: What did you learn about the management structures that seems to help? If you were going to build a safety program around some of these principles, what would you recommend the organizational structure look like?
SH: We developed a couple of tools to try to mitigate the problem. The first one was a guiding principle for affiliations. There was a pretty strong feeling from the clinicians that this cannot be just a business deal—there needs to be some clinical North Star. The guiding principle that the Harvard group developed reads, "The affiliation should unambiguously increase the value of care to the joint patient population of all involved organizations." That was helpful at setting a stake in the ground that said whatever we're doing or resolving or solving or creating, we have to go back and make sure that happens.
The second thing that we developed was a pre-affiliation discussion guide. These are a series of questions for clinical leaders from both institutions to talk over with each other ahead of time to understand how you do things around here. Those questions were built around emergencies, roles and responsibilities, infrastructure and resources, patient safety, quality improvement, and culture. At the end of that process of figuring out what would be helpful, we needed to develop a joint clinical governance process so that the physicians who've been engaged in understanding each other can continue with some clinical oversight, particularly around any quality or safety issues that were identified in that process. From that, and again from lots of interviews, we established a joint clinical integration guide, which talks about what kind of executive level support we need. What kind of data and monitoring capabilities, project management resources, etc.—again, being agnostic as to whose system or how quality and safety is managed, but being sure there's a list of what needs to get done.
RW: Has this been out long enough for you to have an experience with a system that has taken advantage of your tools to see whether it made a difference in their discussions and their process?
SH: Yes, we have anecdotal evidence. Massachusetts General has taken it up quite strongly. They were strongly engaged in all the work. They were in the best position in terms of having hospitals that they were actively affiliated with at the time. They have repeatedly said that this has been extremely valuable.
RW: If you were the head of obstetrics at a place that was merging, partnering with, or buying another place—let's say you were at the bigger place and they were at the smaller place—what would you do tangibly now that would be different than you would have done a couple of years ago based on what you've learned?
SH: One of the challenges in these situations is getting at the truth. When people are merging or affiliating or being sold or bought, it's hard to find out what the real story is. So I would do a couple of things. I would initiate relationship-building with the leaders of obstetrics, neonatology, nursing, and anesthesia—because we're all in this business together. Get all of those people meeting with their peers to develop a trusting relationship. The questions in the pre-affiliation discussion guide are helpful. Second, I would spend time in each other's labor floor and prenatal clinics. Seeing what a busy day looks like and what a slow day looks like, you get invaluable information about differences in protocols and expectations.
I had one experience when I was still active as a chief. CRICO has a guideline that nobody can have more than three patients in active labor at any one time, and I had absolute assurances that was how it was going to be when I deployed a doctor to another hospital. The doctor came back from the second day on call and said, "I delivered 11 babies in 24 hours by myself." When I asked what happened, the doctor said, "Oh, the name [of the backup physician] is on the wall, but they don't call." So, you have to be there and watch to understand. That would apply equally in surgery or emergency medicine.
RW: That's a vivid example—that you discovered when sending people out on a contract to go work in another place was you don't exactly know what it's like until they get there. Then your person came back and told you that he or she delivered 11 babies, and you felt it was unsafe and certainly CRICO believes it's unsafe. What happens next?
SH: Then based on your developed relationship, you call the other chief—a lot of this is affect and relationship building—and say this is the story that I heard. Tell me from your perspective what happened. Because as you know, there are always two sides; try to integrate it and say what are our guidelines that we agree on. In this case, it's easy because of the CRICO guidelines. And how can we make sure this doesn't happen again?
RW: As we at UCSF are doing more networking, adding partnerships and affiliations, these doctors are being brought on board to us in some capacity, whether faculty or part of a clinically integrated network, and we now have to vouch for the quality of their practice. How do you reconcile that the practice standards in the community may be slightly different than those at our academic medical center?
SH: By visiting each other's places. There is so much to be said for just sitting at the labor floor desk. And that often doesn't happen, or it happens in a flurry of white coats being led around without getting the real story. So visit. Ask questions that are helpful. One question that people suggested to us and we incorporated was, "How are department chairs chosen?" That's such a profound divide from a community hospital to an academic medical center, which gives you insight into what the incentives are. "How do you do peer review when the other two people are your partners?" Academics need to understand how community practice works, with far fewer people in-house, much lower levels of backup, much more having to figure it out on your own than we do at an academic medical center where there are seven layers of backup. Having a little bit more sympathy, empathy, and deep understanding of how creative people have to be and with fewer resources goes a long way.
RW: That's very thoughtful. What's next?
SH: We have two things to talk about. This PSA [professional services agreement] thing is just enormous. It's one thing to say we're going to go start robotic prostatectomy at some hospital and there's money, equipment, and etc. But it's totally different to say we're going to send you to this rural hospital to help them with their surgical cases, we expect this volume, and good luck. "You're a doctor, go be a doctor" is the phrase that we use. The concept is that a physician is an equivalent actor—if you're a doctor, you can be a doctor without any sense of what the system is. We (Ariadne) are working through the PSA process for surgery. We have developed a list of questions that a PSA surgeon would want to know going to a new hospital alone, and we vetted that pretty thoroughly.
Our interesting challenge now is how do we get that done. We don't just put it on the Ariadne website and tell doctors to ask people these questions. What's the sending Chief's job, what's the receiving Chief's job, what's the Chief of Nursing, peer surgeons, credentialing? We don't really know how to get this done. And we tried to create a process flow map about how this all happens. It turns out that sometimes the initiation comes from the receiving hospital, sometimes from the sending hospital. We're now trying to figure out how to get it reliably done and put some systems engineering into it. The second thing that we're going to be working on is interhospital transfers. There are a lot of protocols that people made up that sound very good and make sense, but there's not much academic literature.
RW: What are you hoping to find? What are the key issues in that that are different than the ones you've looked at so far?
SH: I think we're going to find what the reasons are for interhospital transfer and how much has to do with nonmedical conditions, family preference, brand name, or other stuff. We're going to find an opportunity to transfer to lower levels of care that we're not picking up from this network. Nobody who's going to die should die in a tertiary care hospital unless it's their neighborhood hospital. We do that for premature babies, back-transfer them to a lower level nursery. We should back-transfer people to the appropriate levels of care. So that will be another thing that we'll be looking at. We'll find what's already been hinted at from the JAMA Surgery paper looking at who gets transferred based on insurance status as opposed to clinical need.
RW: It sounds like the transfer-out issues are more business and operational than safety per se—understanding what the dynamics of the transfer ecosystem are and recognizing that not everybody needs to be transferred into the tertiary or quaternary place.
SH: I think that there are avoidable transfers, because there's frequently not a good mechanism. The person in the secondary hospital is trying to figure out who to call because it's not clear. Admitting may not have a function to support and help the conversation. The patient finally gets there, but the person who accepted the transfer is gone. There are huge operational opportunities to look at.