Version classiqueVersion mobile

Proceedings of the fourth Resilience Engineering Symposium

Erik Hollnagel
Éric Rigaud
Denis Besnard

Colliding Dilemmas: Interactions Of Locally Adaptive Strategies In A Hospital Setting

Robert J. Stephens, David D. Woods, Matthieu Branlat et Robert L. Wears


One strategy that systems employ to remain resilient in the face of shifting demands is the creation and maintenance of margins of maneuver, cushions of potential actions and additional resources that allows the system to continue functioning despite unexpected demands. Systems are often comprised of units which have partial authority and autonomy to manage their individual margins of maneuver. In a study of emergency department interactions with other hospital units, we found that reorganizations to maintain margin could be classified into one of three categories: units could either defensively restrict the actions of other units or borrow margin from them, autonomously reorganize to create new margin, or cooperatively act with other units to exploit existing common pool resources and create new ones. The enumeration of the three classes of strategies to maintain margin of maneuver further explains how systems could reorganize themselves to remain resilient, and how they become brittle and fail. The influences on and interactions of each type of strategy are explored, and directions for future research to allow for better control of overall system margin, and hence resilience, are discussed.

Note de l’auteur

Note portant sur l’auteur1

Note portant sur l’auteur2

Note portant sur l’auteur3

Note portant sur l’auteur4

Texte intégral

1 Introduction

1Woods and Branlat propose Hollnagel's test to examine the concept of control in resilient systems: what does it mean to be in control and how can control be amplified? One strategy systems employ to maintain control is that they actively create and maintain adequate margin of maneuver, a cushion of potential actions and additional resources that allows the system to continue functioning despite unexpected demands. When this envelope is lost, so is the ability to control the system when unexpected shocks occur. A converse example from the literature is going solid, the state in which a system has lost its ability to readily absorb destabilizing influences, meaning any shock to the system will have dire consequences. In the parlance of resilience engineering, systems with increasingly inadequate margin of maneuver are becoming brittle.

2Using a materials science analogy, Woods, and Wreathall have proposed that adaptive systems can absorb new demands up until a certain point (first-order stretching), after which the system must reorganize itself (second-order adaptations) or risk failure. This ability to reorganize is the critical marker of resilience. As Figure 1 shows, margin of maneuver can be conceptualized as the space between constraints imposed by the world (both the blunt end organizational pressures and the monitored process itself) and the current stretching of the system to cope with the demands of meeting system goals.

Figure 1. Margin of maneuver lies between constraints imposed on the system and its first-order stretching to cope with escalating demands

3Coordination and coupling of groups across organizational boundaries has been implicated as a factor impacting resilience. One possible mechanism to explain this is that a group's reorganization to maintain margin may impact other groups' margins. Exploring the interactions and effects of these reorganizations may reveal important determinants of system resilience.

4This paper proposes that within organizations, units (groups that are well-defined by organizational, physical, and/or social boundaries) have a number of potential types of strategies for reorganization to maintain their margin of maneuver, thus avoiding going solid while balancing resource limits. Three general classes were identified based on studies of emergency department (ED) interactions with other units in the larger hospital system. These strategies are:

  • defensive strategies that maintain local margin by restricting other units' actions or borrowing other units' margin(s),

  • autonomous strategies to create margin via exclusively local reorganization, and

  • cooperative strategies that either jointly recognize and manage existing common pool resources, or create new common pool resources that allow sharing of margin with other units.

5If the unit does not or cannot reorganize sufficiently to maintain its margin, then it operates in a risky mode where any additional demands can transform into a cascade of disturbances and lead to failure. Generic patterns of this failure have been identified, e.g. decompensation. Failures in the system can also be considered as crossing organizational scales and boundaries: horizontal failures may occur when local adaptations of units are defensive; units may simply shift demands, and hence failures, to other units. Vertical failures may occur when units behave autonomously; while this may improve the condition of the local unit, it does not help other units and so the system as a whole benefits less than if units were cooperating.

6Failures may also result from maladaptation across multiple timescales. Failures at shorter time scales can occur due to units' tactical inability to reconfigure to create additional margin, a type of decompensation failure. At longer time scales, strategic failures occur if units cannot be proactive and coordinative in their management of margin. It may not be sufficient to simply react to new demands by creating margin; joint management and pre-planning of margin with other interdependent units may be required to maintain a resilient system.

7When considering these generic patterns, the actions of individual actors in the system are de-emphasized in favor of considering system-level processes. The examples presented below therefore focus on the joint cognitive system (JCS) of humans, technology, and work environment that acts to maintain margins of maneuver in the context of cross-scale interactions and multiple timescales.

2 The Strategies in Operation in the ED

8The emergency department (ED) continues to be one of the critical laboratories for studying or understanding resilience and adaptation in human systems. Studies of boarding in the ED (admitted patients being held in the ED after their treatment there has concluded) examine inter-unit relationships about how patients flow into and sometimes get "stuck" in the ED. Observations of emergency departments have revealed cases of each class of strategy in use. This section explores those findings and also describes some factors observed as influencing the evolution of each strategy.

2.1 Defensive Strategies

9A defensive strategy increases the margin of one unit by restricting another unit or borrowing margin from another unit. An example of this shift towards a defensive strategy can be seen in the interactions of the ED and the Intensive Care Unit (ICU) of many hospitals. It is often the case that the ED is charged with the responsibility for ICU-bound patients, but over time policy shifts have migrated authority over these boarded patients to the ICU. In one study, the ICU adapted a defensive strategy to borrow margin from the ED: the ICU delayed or refused admissions of ED patients bound for ICU care until tasks such as laboratory studies or intensive therapies had been completed. This strategy may have been spurred by a top-down policy mandating a 1:1 patient-to-nurse ratio in the ICU. However, this borrowing of margin from the ED by delaying admission may result in a worse overall outcome for patients: research has found that critically ill patients experience worse outcomes when they are delayed in their transfer from the ED to the ICU.

10Some EDs have employed similar defensive strategies against other inpatient units in an example of a tit-for-tat cascade of defensive actions. In the face of other units delaying admission to defend their margin, one ED avoided boarding patients by simply moving admitted patients to inpatient unit hallways, an ad-hoc intermediate area between the ED and inpatient unit, without prior approval of the inpatient unit. Interestingly, one study of this coping strategy found that it did not result in worse outcomes for these patients.

11Going on diversion (ED refusal to accept patients via ambulance) is another example of a defensive strategy adopted in response to the restriction and borrowing of margin from the ED. In this case, the ED’s margin is expanded at the expense of other EDs in the area. Policies in some regions prevent EDs from going on diversion for this reason.

12Moving outside of the ED-inpatient dyad to hospital-wide issues, the practice of bed hiding (maintaining a reservation for a bed when not immediately needed) is another example of a defensive strategy. By hoarding resources (i.e. from a common pool), additional margin is created locally, but this may adversely affect the margin of other units that also want to use the resources being hoarded.

2.2 Autonomous Strategies

13Autonomous strategies consist of local reorganization to add margin of maneuver within a unit or expand a unit's ability to regulate its margin given resource availability and limits. For example, persistent overcrowding problems have led some EDs commit resources under their control to develop additional buffer capacities and specialized care tracks.

14One ED studied created an exception-handling sub-unit to reduce the resource cost of managing psychiatric patients, who are often boarded long-term. This unit economized the resources needed to care for psychiatric patients; a single nurse can now monitor up to six patients concurrently. This strategy arose due to inpatient psychiatric units' use of a delay of acceptance strategy against the ED, itself spurred by severe resource constraints from budgetary cuts. The ED also used an autonomous strategy of relaxing standards on discharge of mental health patients during the weekends, a time when insurance-related policies make it difficult to secure admission for psychiatric patients.

15In general, EDs often pre-plan stratified intensities of care by placing some patients in beds and others in chairs. This has the effect of creating additional margin, as resources can be rapidly redeployed to the higher-priority (bed) patients if there is a surge in load. Similarly, ICUs have been found to pre-plan which patients can be readily transferred to a lower intensity unit; the identification of these bumpable patients is one of the critical functions of the ICU’s role of bedmeister.

2.3 Cooperative Strategies

16A cooperative strategy shares two or more units’ margins through coordinated, collective action of recognizing or creating a common-pool resource on which both units can draw. For example, one ED gave up one of its five treatment areas to use it exclusively to board admitted patients. The inpatient services agreed to provide nursing staff for that area, accepting it as another inpatient ward. This created a common pool resource for both the ED and inpatient unit: the ED could remove boarded patients to the specialized unit pending their transfer to an actual inpatient unit, freeing more staff; the inpatient unit could avoid a bedcrunch.

3 Interactions of and Influences on Strategies

17The three strategies allow units to anticipate upcoming demands by being prepared to cope with them. Units' behavior may be primarily tactical (reactive to maintain margin), but may also be strategic (proactive to maintain the ability to reorganize towards using more or less margin, as needed). A unit's choice of strategy is therefore influenced heavily by the availability of cues about its past, present, and potential future demands, as well as cues about the space of possible actions the unit could take and the resulting reactions of other units.

18Strategies may interact in both a horizontal (between-units) and vertical (higher-echelon) fashion. In the defensive examples above, units are adapting locally, but with effects on other units' ability to regulate their margins of maneuver, as well as potential global effects. As enumerated previously, such adaptations can spur cascades of defensive action as units whose margin is borrowed simply continue to shift work or risk to other units; the same global margin is maintained. However, a more dire possibility is that units work at cross-purposes from each other, thus actually reducing the global margin. This suggests that defensive strategies, at best, do not help meet the goals of the system, and may actually leave the entire system in a more precarious state.

19Autonomous strategies may arise when a unit either has no defensive or cooperative possibilities for reorganization, it cannot perceive them due to inadequate cues, or an autonomous adaptation is simply assumed to be the most effective way for the unit to manage its margin. There is also at least some cost associated with the coordination necessary to maintain common ground at a sufficient level to support sharing of margins, making autonomous strategies (and potentially defensive ones) more attractive if this cost is perceived to outweigh the benefits of cooperation.

20Considering these factors, it is apparent that in order for a unit to choose a cooperative strategy, it must be able to not only predict how cooperative actions allow it to cope with its own demands despite the costs of coordination, but also how its actions ripple through interdependencies with other units and therefore the system as a whole. In the cooperative example above, pre-planning of a shared common pool of resources supported this ability.

21Other factors may also influence the strategy chosen. The influence of culture can be seen in a recent volume on the ED, titled "Anyone, Anytime, Anything;" this attitude pervades much of emergency medicine practice and may influence the ED towards adopting primarily autonomous strategies. It may also be that other units similarly view the ED as the sole or primary unit charged with absorbing sudden demands on the larger system.

22Policy may deter reconfiguration of the overall system towards a more even distribution of margin. This is likely the case in highly-siloed systems with minimal horizontal interaction between units, as well as in situations where control and responsibility are mismatched due to organizational pressures and policy. If one unit has control but the other has responsibility for a given goal or task, the former unit has an increased ability to maintain an uneven margin distribution at the expense of the latter. Accountability through higher-echelon monitoring of margin distribution could mitigate these factors, though it may also create new problems of its own through encouraging units to cheat or game the system.

4 Directions for Future Research

23Future work to validate and expand this framework should focus on how factors such as information availability, governance, and work culture influence the behavior of units with respect to the three strategies enumerated. It should also investigate how cascades of strategies can result, and what factors will result in the most resilient management of strategies in the context of actual work systems.

24An open question is whether there is a tendency for sudden demands to drive a unit towards defensive or autonomous modes of reorganization. Perhaps this can be mitigated by providing the ability to rapidly coordinate with other units or setting up pre-existing collaborative agreements and triggers. We have posited above that defensive strategies are generally detrimental to systems; however, this may not always be the case. It would be interesting to identify and investigate cases in which a defensive or autonomous strategy resulted in a better overall outcome.

25The studies of emergency department interactions with the larger hospital system tend to indicate that providing units with the ability to build common ground and joint predictability with other units may foster more cooperative strategies, particularly in cases where there is a need to recognize an existing common pool resource as a shared margin of maneuver, as opposed to defining all resources as owned by a single unit. At the same time, it is recognized that the role of higher-echelon governance to create the proper environment in which cooperative strategies can thrive is critical; for example, a narrow focus on unit-specific performance could increase the tendency to adopt defensive strategies. Studies of common pool resource governance by Ostrom and colleagues suggest that polycentric control architectures can balance these extremes to create resilient systems.

26The three strategies to maintain margin of maneuver enumerated in this paper provide a framework for discussing how local actions affect global system resilience, how these actions may interact, and how they could be shaped to create the most resilient global system.



Anders, S., Woods, D., Wears, R., Perry, S., & Patterson, E. (2006). Limits on adaptation: modeling resilience and brittleness in hospital emergency departments. In E. Rigaud & E. Hollnagel (Eds.) Second Symposium on Resilience Engineering. Juan-les-Pins, France, Nov. 8-10, 2006.

Branlat, M., & Woods, D. D. (2010). How do Systems Manage Their Adaptive Capacity to Successfully Handle Disruptions? A Resilience Engineering Perspective. In Proceedings of the 2010 AAAI Fall Symposium Series -Complex Adaptive Systems, Arlington, VA, USA.

Cook, R. I. (2006). Being bumpable. In D. D. Woods & E. Hollnagel (Eds.), Joint cognitive systems: Patterns in cognitive systems engineering. Boca Raton, FL: CRC Press.

Cook, R. I., & Rasmussen, J. (2005). ‘‘Going solid’’: a model of system dynamics and consequences for patient safety. Qual Saf Health Care, 14, 130-134.

Dietz, T., Ostrom, E., & Stern, P. C. (2003). The struggle to govern the commons. Science, 302(5652), 1907.

Garza, M. (2004). Memphis just says 'No!' to ambulance diversion. JEMS: A Journal of emergency medical services, 29(1), 130.

Ostrom, E. (1999). Coping with tragedies of the commons. Annual Review of Political Science, 2(1), 493-535.

Parkhe, M., Myles, P. S., Leach, D. S., & Maclean, A. V. (2002). Outcome of emergency department patients with delayed admission to an intensive care unit. Emergency Medicine, 14(1), 50-57.

Stephens, R. J. (2010). Managing the Margin: A Cognitive Systems Engineering Analysis of Emergency Department Patient Boarding. Doctoral Dissertation, The Ohio State University.

Stephens, R. J., Cudnik, M., & Patterson, E. (2011, in press). Barriers and Facilitators to Timely Admission and Transfer of Patients from an Emergency Department to an Intensive Care Unit. Paper presented at the Human Factors and Ergonomics Society 55th Annual Meeting, Las Vegas, NV.

Viccellio, A., Santora, C., Singer, A. J., Thode Jr, H. C., & Henry, M. C. (2009). The Association Between Transfer of Emergency Department Boarders to Inpatient Hallways and Mortality: A 4-Year Experience. Annals of Emergency Medicine, 54(4), 487-491.

Wears, R. L., & Perry, S. J. (2008). A system dynamics representation of resilience. In E. Rigaud & E. Hollnagel (Eds.) Second Symposium on Resilience Engineering. Juan-les-Pins, France, October 28-30, 2008.

Wears, R. L. and Perry, S. J. (2006). “Free fall” – a case study of resilience, its degradation, and recovery in an emergency department. In E. Rigaud & E. Hollnagel (Eds.) Second Symposium on Resilience Engineering. Juan-les-Pins, France, November 8-10, 2006.

Woods, D. D., & Branlat, M. (2010a). Basic Patterns in How Adaptive Systems Fail. In E. Hollnagel, J. Pariès, D. D. Woods & J. Wreathall (Eds.), Resilience Engineering in Practice (pp. 127-144). Farnham, UK: Ashgate.

Woods, D. D., & Branlat, M. (2010b). Hollnagel’s test: being ‘in control’ of highly interdependent multi-layered networked systems. Cognition, Technology & Work, 12(2), 95-101.

Woods, D. D., Johannesen, L., Dekker, S., Cook, R., & Sarter, N. (2010). Behind human error. Aldershot, UK: Ashgate.

Woods, D. D., & Wreathall, J. (2008). Stress-Strain Plots as a Basis for Assessing System Resilience. In E. Hollnagel, C. P. Nemeth & S. W. A. Dekker (Eds.), Resilience Engineering Perspectives (pp. 143-158). Aldershot, UK: Ashgate.

Zink, B. J. (2006). Anyone, anything, anytime: A history of emergency medicine: Mosby.


1 National Center for Human Factors Engineering in Healthcare, 3007 Tilden St NW Suite 7M, Washington, DC 20008, USA

2 The Ohio State University, 1971 Neil Ave, Columbus, OH 43202, USA

3 The Ohio State University, 1971 Neil Ave, Columbus, OH 43202, USA

4 University of Florida, 655 West 8th Street, Jacksonville, FL 32209, USA Imperial College London, Praed Street, London W2 1NY, UK

Table des illustrations

Légende Figure 1. Margin of maneuver lies between constraints imposed on the system and its first-order stretching to cope with escalating demands
Fichier image/jpeg, 248k

© Presses des Mines, 2011

Licence OpenEdition Books

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search