Sorry, you need to enable JavaScript to visit this website.
Skip to main content
Ellen Deutsch photograph

In Conversation with... Ellen Deutsch, MD, MS, FACS, FAAP, FSSH, CPPS

December 14, 2022 

Editor’s Note: Ellen Deutsch, MD, MS, FACS, FAAP, FSSH, CPPS is a Medical Officer in the Center for Quality Improvement and Patient Safety at the Agency for Healthcare Research and Quality. Dr. Deutsch is a pediatric otolaryngologist and has vast experience in simulation and resilience engineering. We spoke with her about resilient healthcare and how resilient engineering principles are applied to improve patient safety.

Sarah Mossburg: Thank you for joining us today. Please tell us a bit about yourself and describe your current role.

Ellen Deutsch: I am a medical officer in the Center for Quality Improvement and Patient Safety at the Agency for Healthcare Research and Quality (AHRQ). I started my career as a pediatric otolaryngologist and provided patient care for about 20 years. During that time, I became interested in and developed expertise in simulation, and subsequently completed a master’s degree in healthcare quality and patient safety. I am also a Certified Professional in Patient Safety. I have served in leadership positions in several professional organizations, including national otolaryngology organizations and international simulation organizations. I have a long-standing interest in human factors, simulation, systems engineering, and resilience engineering. I believe in and want to provide healthcare that provides both patients and providers with satisfaction and potentially even joy.

Sarah Mossburg: Can you describe some of your work and experience in resilient healthcare and how that interest developed?

Ellen Deutsch: About 10 years ago, I was privileged to attend one of the early Resilient Health Care Network meetings led by Erik Hollnagel and many other insightful and visionary thinkers. The concepts that were discussed at the meeting resonated with me, and I have been attending the meetings ever since. The discussions are always stimulating and the meetings, discussions, and work to create presentations and publications has helped me clarify my thinking.

Sarah Mossburg: Broadly speaking, can you tell us about resilient healthcare? What are some of the principles of resilient healthcare?

Ellen Deutsch: Resilient healthcare is an approach to healthcare delivery that strives to provide the best patient outcomes and acknowledges that the circumstances of healthcare delivery can impact providers’ and patients’ ability to achieve the best outcomes. Healthcare is a complex adaptive system. It has an unimaginable number of moving parts. Knowledge, resources, and relationships are always changing, and these patient care contexts impact our ability to provide healthcare. Recognizing and acknowledging complexity is really important. To begin to understand how healthcare is provided, you can conceptualize healthcare delivery as a linear process, but that concept should also be accompanied by the recognition that healthcare delivery is more complicated than that model. For example, as a healthcare provider, when I started my day, I would have a plan for what I was going to do that day. It was always a draft because as the day evolved, things would happen that changed that plan. I was constantly reprioritizing. An emergency need for a surgical procedure could throw off the entire clinic schedule. Both small things and large things are always shifting and evolving.

Resilient healthcare requires knowledge and skills, and acknowledges the need for adaptability to understand or create resources. When I refer to resources, my intention is a broad definition, including people, knowledge, equipment, physical spaces, protocols, culture, and so forth. In huge, inter-related healthcare delivery systems, these components contribute to - or potentially detract from - providing the best healthcare.

Resilient systems have four classic capacities: to learn, to anticipate, to monitor, and to respond. These capacities interact in multiple ways. Responding to a patient care event, which may have had a desired or an undesired outcome, provides a learning opportunity. That learning opportunity may help you anticipate the next decision point so you know how to recognize and handle the next similar patient care event. And so forth; each of the capacities of resilient systems helps to inform each of the other capacities. An additional concept that has been discussed more recently, that envelops these four capacities, is the capacity to collaborate or cooperate. All patient care events happen in the context of other people and other parts of the system.

Sarah Mossburg: You previously published a feature in the Pennsylvania Patient Safety Advisory about healthcare delivery as a complex adaptive system and bridging the gap between “work-as-imagined” and “work-as-done.” Can you tell us more about these concepts?

Ellen Deutsch: Yes; the concepts of work-as-imagined versus work-as-done were initially developed by Erik Hollnagel and colleagues. It’s the idea that we think that work—like providing healthcare—will happen a certain way, but when work actually happens, people often have to make adjustments to make the work happen effectively and get the best outcome. While it is important to have a conceptual overview of work and understand processes at a high, representational level, work-as-imagined never completely captures the exact details of the work-as-done.

The other important ramification of Hollnagel’s concept is that the lens we use to try to understand healthcare work affects what we see and what we understand. In the article, I described other ways that we try to understand work. A lot of value is placed on “work-as-documented.” Medical records are viewed as the truth of what happened to a patient. However, the accuracy and the completeness of the medical record depends on many factors, including the time interval between when a patient care event happens and when documentation occurs, and the difficulty of the documentation process.

“Work-as-abstracted” explores the idea that we can aggregate data from patient encounters and find patterns that weren’t necessarily evident during individual patient care encounters, but we may lose precision and nuance. “Work-as claimed” is important to organizations because of financial ramifications, but malpractice claims events do not necessarily provide useful learning experiences. “Work-as-observed” may be impacted by the presence of an observer and what is observed may depend on the expertise and interests of the observer.

The other type of work I have thought about is “work-as-simulated.” Simulations are a great way to get close to work-as-done, especially when you have simulations with real teams working in their real patient care settings. Simulations can expose latent safety threats, such as when expected equipment is not available in the real working environment or planned protocols do not function well with the resources actually available.

The overarching principle is that each of these lenses has value and gives you information, but you have to be aware of both what the lenses can show you and their limitations. To get the best picture we need an aggregation of many lenses.

Sarah Mossburg: How do these views of work correspond with resilient healthcare and two key approaches to safety: Safety-I and Safety-II?

Ellen Deutsch: Each of these lenses help us develop an understanding of how work is done. In addition, we can explore what work to look at. Safety-I and Safety-II offer complementary perspectives about what work to examine and what lessons may be learned. Safety-I is the approach to understanding something that went wrong by determining why it went wrong. Typically, a Safety-I evaluation includes asking the questions, “What do we need to do to stop this from happening the next time? What constraints or barriers can we address to help us accomplish this?”

Alternatively, the Safety-II lens looks at the work that went well and asks, “Why did it go well? What were the resources used? What were the capacities, actions, decisions, knowledge? What contributed to a successful outcome?” Safety-I focuses on what went wrong, and how do we stop it from happening. Safety-II focuses on what went well, and how do we support that.

For things to go well, we recognize that there has to be adaptation. There has to be flexibility. And in general, humans are the component of the system that makes it work, that works around the limitations, that solves the problems, that creates innovative solutions.

Safety-I is more applicable when there is a steady state and when there is a known best way. Vincent and Amalberti identified that there are medical circumstances where you avoid risk and there are medical circumstances where you acknowledge risk. There are circumstances in healthcare delivery where you want to exclude risk as far as possible, and your priorities are to implement very rigid protocols to make sure that the correct process is followed. The examples they provide are radiation therapy or blood transfusions.

In some situations there is more risk, and risk and uncertainty are inherent in the events. For example, when managing an airway emergency, there is risk in any treatment you might try but there is also risk in not treating. There are some basic principles that you follow, but there will also be some variation, and something unique about each case that you have to consider so that you adapt your interventions. It may be that something is unique about the patient, but the uniqueness may relate to available equipment, medications, staff, or other factors. In events like mass casualties and infrastructure failures, you have to develop your best response based on the expertise of the people who are there and the available resources. There will be general principles that are relevant, but there may not be a protocol that fits exactly, and there will not be a standard approach that always works.

So, there are circumstances in which the Safety-I approach of constraining options is appropriate, and there are circumstances where you need to allow the experts to do their best and use their knowledge, skills, and autonomy. There are certainly contrasts between Safety-I and Safety-II, but components of both approaches are necessary and there are different circumstances in which these frameworks should be applied.

Sarah Mossburg: It seems there is a lot of overlap with the principles we’re discussing and other patient safety concepts, like high-reliability organizations and TeamSTEPPS®. How is resilient healthcare different from some of these other patient safety concepts?

Ellen Deutsch: There are overlapping concepts, and, in some ways that’s reassuring that we may be on the right path. Sometimes I think of these concepts as different lenses of a multifaceted gem, where you turn the gem, and one facet is safety, another facet is quality, another facet is value. Each facet, like each lens, adds to our understanding.

One of the underlying principles of resilient healthcare that stands out to me is the acknowledgment of the complexity of healthcare systems. Healthcare delivery is so complex that it is incompletely knowable and is not decomposable. Although parts of the system can be understood, understanding each part still does not allow you to understand the whole because of the emergent, dynamic interactions between all the components.

Inherent in this complexity is ever-present change. Changes in technology may be the most obvious changes, but patient care knowledge and techniques evolve, diseases emerge, and societal expectations progress. We need humans who improve their skills and knowledge, adapt, innovate, and create to keep improving healthcare delivery. The other important underlying principle is that resilient healthcare appreciates humans as valuable assets and not as the problem in healthcare delivery.

Sarah Mossburg: We talked about some of the benefits of resilient healthcare and Safety-II. What are some of the limitations?

Ellen Deutsch: One of the challenges is that measuring resilience and success is harder than measuring failure. In part, this is difficult because success is based on a huge sequence of interactions and contributing factors. It can be challenging to sort out what made something successful. Measurement is also challenging because healthcare results are not always simply bimodal, that is, a success or a failure. Many outcomes are more nuanced. Failures get our attention, but we also have limitations when looking at failures, including the impact of hindsight and a temptation to oversimplify. How do you count how many opportunities for failure existed but did not fail? They are infinite.

Sarah Mossburg: There has been implementation of resilient healthcare, Safety-I, and Safety-II concepts globally, and a fair amount of European-based or other international research. How have you seen resilient healthcare applied in United States-based healthcare settings?

Ellen Deutsch: I see the terms being used more and more frequently in presentations and in publications in the United States. Exposure to these concepts is a little bit uneven. There are people in places where these ideas have permeated, and there are areas where people have not been exposed.

Sometimes people are using resilient practices in clinical care settings, but they are not necessarily labeling them as Safety-I or Safety-II, or resilient behaviors. Where I see that the most clearly is in simulation. The characteristics of an effective simulation align well with the characteristics of resilience, particularly with respect to debriefing practices.

The simulation community is at the forefront of being able to apply resilient principles and processes to patient care. In a simulation scenario with real people, in their real roles, such as doctors, nurses, and pharmacists participating in taking care of a simulated patient in their real clinical settings with real equipment, there is usually a debriefing afterward. This practice is well-established in simulation though not as well-established in patient care. Debriefing is meant to support learning. There are different styles of debriefing, but debriefs are generally used to elicit realizations from the learners themselves and to help the learners come to their own guided understanding of what happened. Debriefings are intended to support learners and to help them develop their own meaningful insights.

One of the questions I ask during debriefing is about recognizing and appreciating the contributions of all team members. For example, after a simulation, I might ask, “What did [another participant] do that helped you?” This prompts the learners to think about what others did that helped them. It is my hope that both of these people—the helper and the one helped—gain an appreciation of the value of that help.

Debriefing skills learned in simulation are very applicable to real patient care circumstances. Some people conduct hot debriefings, which are quick meetings and discussions after a patient care event, good or bad. There are questions that people can ask that promote resilience. Questions like, “What did you learn? What surprised you? What went well during that patient care event?”

Sarah Mossburg: It sounds like simulation really allows people to test adaptations to situations and changes to situations in a safe setting and learning post-simulation is really helpful. The question of “What went well?” really speaks to the Safety-II concept of safety as adaptation.

Ellen Deutsch: Yes. I think there is one more component that is not always articulated. Most people facilitating a debriefing try to create an atmosphere of psychological safety. In healthcare, we can be very hard on ourselves. We ruminate about events that don’t go well. Learning is supported when you are in an environment that is psychologically safe. Even the way a debriefing is run contributes to that affective lesson of respectful curiosity, valuing other people’s input, seeking to understand rather than to criticize. That secondary affective lesson in simulation is very relevant for clinical care.

Sarah Mossburg: Yes, and that builds the capacity for resilience and sets up the groundwork for that. You mentioned at the beginning of this interview that you first became interested in resilient healthcare about 10 years ago. How has resilient healthcare changed over 10 years?

Ellen Deutsch: I’m still trying to learn and understand. It’s taken at least 10 years. I think that earlier discussions about resilient healthcare were more about theory and concepts, and those are still going on, but now there are also people collecting data. These data are helpful in demonstrating relevance, applicability, and processes.

Sarah Mossburg: How do you think the COVID-19 public health emergency impacted the use of resilient healthcare in the U.S.?

Ellen Deutsch: I think the concept of complex adaptive systems resonates better with people now because they have been through such a stressful and complex experience. In addition to the devastating effect on patients, families, and healthcare workers, the public health emergency exposed weaknesses in healthcare delivery and challenged the assumption that resources are consistently and predictably available. Previously, we didn’t want to have any “waste” in healthcare, but now we’re realizing that we need to have margin and resources in reserve.

COVID-19 also demonstrated the importance of flexibility and adaptation, and the value of the knowledge of the healthcare workers at the frontline. Early on, healthcare workers were energized by the opportunity to solve problems, by the autonomy, by the creative solutions they were coming up with. Over time, the unrelenting crisis wore more people down. The efforts of doctors, nurses, and others who interacted directly with patients have been heroic. There are also many unsung contributions by allied health professionals, support techs, engineers, environmental services workers, and all the people who kept the lights on, the shelves stocked, and the workforce fed despite supply chain interruptions. A lot of people made sacrifices.

Sarah Mossburg: Yes, it speaks to how large the system is. There are the clear, obvious components, and still so much behind them that’s happening that’s important to make the system run. COVID-19 highlighted how healthcare is a complex adaptive system. Looking forward, how do you think resilient healthcare is going to evolve over time? What do you see as the next steps in this work?

Ellen Deutsch: I think there are two different levels of next steps. The frontline, concrete level, and the 30,000-foot level. At the concrete level, resilience experts will take greater advantage of opportunities to study resilient capacity, using simulations of patient care scenarios that involve teams. The more traditional simulation is designed to enable a team to identify the problem with a patient, and then work to mitigate it, to treat it. But these same simulations can be used to explore the environment that people work in and all the factors that impact their abilities to provide care, including all the components of socio-technical systems, the physical environment, the protocols, the culture, the equipment, and the medications. People studying resilience will take greater advantage of the opportunity to get closer to understanding how work-is-done using simulation.

Resilience engineers will also collaborate with leadership, healthcare workers, patients, and families, to apply skills and tactics learned from simulation debriefing, or from grounding in a variety of social sciences. They will strive to better understand and improve work processes, patient care experiences, and organizational practices in an effort to develop opportunities for success and satisfaction.

At the 30,000-foot view perspective, there have been incredible technological advances in healthcare. As an otolaryngologist, an example that I am familiar with is cochlear implants. These implants provide hearing for people who would otherwise be deaf. They restore an entire sense—that is an incredible feat! The variety and the speed of technologic advance is increasing, and many of these advances are fantastic and amazing. The safest healthcare is care that provides optimal outcomes from the patient’s perspective and the provider’s perspective; it will require technology to be paired with human interactions, provider empathy, provider knowledge, and provider skill. Resilient healthcare will help us understand, appreciate, and implement conditions and processes that can optimize these relationships.

Sarah Mossburg: Earlier, you spoke about research you would love to see in the near future. AHRQ also has grant funding mechanisms for research related to resilient healthcare. Are there any that you are able to share?

Ellen Deutsch: I’m so glad you asked! We just concluded awarding a series of Diagnostic Centers of Excellence grants that specifically solicited Safety-I and Safety-II approaches to diagnostic safety. There are also two grant opportunities that are both accepting submissions until January 27, 2023. One grant is the Patient Safety Learning Laboratories: Advancing Patient Safety through Design, Systems Engineering, and Health Services Research (PA-21-266). There is also the Medication Safety: Advancing the Development of Improvement Strategies and Tools (PA-20-028).

AHRQ PSNet also provides more than 200 resources that address resilience, Safety-II, and related topics.

Sarah Mossburg: Thank you for speaking with us. I found our conversation to be helpful and interesting.

Ellen Deutsch: I really appreciate this opportunity because I do think many people have heard about Safety-II and resilience engineering, and wonder what they mean. And hopefully, this helps. Thank you very much for the opportunity.

This project was funded under contract number 75Q80119C00004 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services. The authors are solely responsible for this report’s contents, findings, and conclusions, which do not necessarily represent the views of AHRQ. Readers should not interpret any statement in this report as an official position of AHRQ or of the U.S. Department of Health and Human Services. None of the authors has any affiliation or financial involvement that conflicts with the material presented in this report. View AHRQ Disclaimers
Related Resources