If you think quality and safety are the same...think again

1
If you think quality and safety are the
same...think again

• Karen Cardiff, School of Population and Public
  Health, University of BC
• Samuel B Sheps, School of Population and Public
  Health, University of BC
• Jim Nyce, Department of Anthropology, College of
  Sciences and Humanities, Ball State University,
  Muncie Indiana
• Sidney Dekker, Lund School of Aviation, Lund
  University, Lund, Sweden
• June 10, 2010
• System Safety Society Canada ChapterSpring
  Symposium
• Ottawa, Ontario

2

• The findings presented today are based on the
  thesis work for completion of an MSc in Human
  Factors and System Safety with the Department of
  Aviation, Lund University

3

• Also informed by work that began a decade ago
• Canadian Adverse Events Project (Canadian
  Institutes of Health Research, 2000-2003)
• Management and Regulation of Safety in Risk
  Critical Sectors (Health Canada, 2004-2007)
• Creating High Reliability Organizations in the
  Canadian Health Care System (Canadian Patient
  Safety Institute, 2007-2008)
• Current work co-funded by the Canadian Health
  Services Research Foundation
• and the Canadian Patient Safety Institutea four
  year multi-partner capacity building project that
  is focused on building capacity within acute care
  hospitals to do critical incident investigation
  (Sheps and Cardiff, 2009-2013)
• Interaction with international opinion leaders
  and experts in system safety on the
• emerging ideas of resilience and safety (Sidney
  Dekker, Eric
• Hollnagel, René Amalberti, Richard Cook, etc)

4
Outline of presentation

• Background
• Why is the topic important
• Theories and models of why things go wrong
• Findings from the thesis work

5
Background

• When patients entrust themselves to our care, we
  make two implicit, but key professional and
  organizational promises
• we promise to do everything possible to help
  patientsto provide good (possibly excellent)
  care and, we promise not to hurt them
  (Reinertsen Clancy, 2006)
• However, there are many instances where people do
  not get the care that they need (McGlynn et al
  2003) or are inadvertently harmed through the
  process of care (Kohn et al, 1999 Vincent et al,
  2001 David et al 2002 Norton et al, 2004 Leape
  et al, 2005).

6

• While being concerned with quality for many
  years, healthcare did not, in general, think
  systematically about patient safety until the
  magnitude of the problem became very clear and
  could no longer be ignored.
• Now, after more than a decade of activity that
  has measured, tracked, and in many instances
  investigated adverse events in acute care, no one
  doubts that enhancing patient safety (i.e.
  reducing harm) is an important and necessary goal
  
• Although there continues to be widespread buy-in
  that the healthcare system must take steps to
  reduce patient harm, the struggle in how to
  achieve this, and make it sustainable, remains.
• Even as the number of activities to improve
  patient safety and quality increases, along with
  a the range of tools available to assist with the
  process the healthcare system has not yet
  achieved the status of being a high reliability
  or resilient industry.

7

• While emerging theoretical and applied work
  acknowledge that quality and safety are distinct
  concepts, in healthcare the majority of
  activities, to date, have focused on activities
  designed to improve adherence to accepted
  standards of care, such as hand-washing,
  appropriate use of antibiotics (i.e. telling
  people what they already know they should be
  doing), and are more aligned with the classic
  quality model than safety safety by constraint
  marked by barriers, regulations, procedures,
  training and standardization.
• The current efforts are largely based on
  reductionistic thinking that attempts to
  trouble shoot and fix things.
• Even though many of these efforts have been met
  with success, they have largely ignored issues
  related to what it means to create safety in
  complex, dynamic settings, such as preparing
  frontline staff to cope with the complexity that
  they face on a daily basis and supporting them to
  become more experienced with anticipating what
  might go wrong (requisite imagination) and
  knowing when and how to adapt their performance
  under conditions of uncertainty.

8
Understanding adverse events in healthcare

• A model often used in health care is that
  accidents are thought to occur when individual
  components or processes fail to meet
  criteria.this model of risk and safety builds on
  the assumption that safety, once established, can
  be maintained by keeping the performance of a
  systems parts (human and technical) within
  certain bounds (e.g. people should not violate
  rules and procedures)
• Sidney Dekker, Past the edge of chaos. 2006

9
Why does the classic quality model distort
efforts to achieve safety?

• The classic quality model was developed to ensure
  that the system meets pre-specified criteria.
• Quality is viewed as the characteristics of a
  service or product that must be present to meet
  needs or expectations such as high professional
  standards, effective use of resources and high
  patient satisfaction
• The goal of quality assurance activity is to keep
  performance variability under control
  organizations develop policies, rules and
  protocols to keep performance within a particular
  bandwidth.

10

• In linear production systems (e.g. Toyota,
  McDonalds) quality may have safety as an additive
  result this is because
• the systems can be decomposed into meaningful
  elements
• the failure-probability of individual components
  can be described and analyzed individually
• the order or sequence of events is predetermined
  and fixed
• when combinations of events occur they can be
  described as non interacting
• the influence of context is limited or
  quantifiable
• (Dekker, 2007)
• However, since the process of producing
  healthcare is neither linear nor fixed, the
  classic quality model is not adequate.

11
Complex systems

• in complex systems (such as healthcare),
  unpredictability and paradox are ever present,
  and some things will remain unknowable.new
  conceptual frameworks that incorporate a dynamic,
  emergent, creative and intuitive view of the
  world must replace traditional reduce and
  resolve approaches to clinical care and service
  organization (Plsek, 2001)

12
Examples of complexity in healthcare

• Demand frequently pushes performance goals
• Rapid introduction of new, complex technology
• Many discontinuities and transitions in care
  (e.g. multiple care providers caring for the same
  patient and often in multiple settings
  emergency department to operating room to
  surgical wardeach unit managed independently)
• Strong autonomous and semi-autonomous
  professional cultures with concomitant power
  struggles
• Often rapid turnover in staff
• Continual influx of new patients, each with their
  own inherent biological variability and in many
  instances language and cultural differences.

13

• The key challenge of creating safety in complex
  non linear systems is that the knowledge base is
  inherently and permanently incomplete.
• In healthcare work situations are always
  underspecified (i.e. the conditions of work
  rarely match what has been specified or
  prescribed), and with this comes an unpredictable
  component, and thus adaptation is often necessary
  (Hollnagel et al, 2006 Grote, 2006)
• Performance variability is both normal and
  necessary in complex non linear systems (Dekker,
  2007 Hollnagel, 2008)

14
Performance variability, adaptation safety

• IF keeping things safe equals keeping performance
  within a particular bandwidth, why do complex,
  dynamic organizations value experience or
  seniority?
• It is because experience brings a broader
  bandwidth people at the sharp end of care can
  judge whether that which they used before will
  work in the current situation.
• There is always a tension around when and how to
  adapt rules and protocols to the set of
  circumstances one finds themselves in, but the
  experienced person is more likely to know when
  and how to do this.
• Complex systems are generally well protected from
  vulnerabilities with barriers, but safety is
  created through practice, i.e. practitioners
  recognize local pitfalls and forestall them, and
  this often involves adaptation.

15
Theories and models of why things go wrong
Linear views (Dekker, 2007 Hollnagel, 2008)

• Assumption and consequence
• Accidents are the natural outcome of a series of
  events or circumstances which occur in a specific
  and recognizable manner (e.g. Domino Model,
  Heinrich, 1930) ? accidents are prevented by
  finding and eliminating possible causes
  component failures, such as human technical and
  organizational.
• Assumption and consequence
• More recently, accidents have been seen to result
  from a combination of active failures (unsafe
  acts) and latent conditions (hazards) ? accidents
  are prevented by strengthening barriers and
  defenses (e.g. Swiss Cheese Model, Reason, 1990)
  and safety is ensured by measuring performance
  indicatorsin particular people rely on
  understanding past events (e.g. Root Cause
  Analysis) to develop solutions for the future.

16
Reasons Swiss Cheese Model Layers of defense
(J. Reason, Managing the Risks of Organizational
Accidents, 2002)
17

• Linear models provide the basis for investigators
  to easily take the position of retrospective
  outsider, looking back on a sequence of events
  that seems to lead to an inevitable outcome, and
  pointing out where people went wrong, or where
  individual components of the system failed.
• Although this perspective is often adequate in
  linear systems, the approach seriously limits
  what the investigator can learn about failure in
  non linear systems ( i.e. the complex and
  unexpected combination of system interactions)
  and may not help prevent recurrence.

18
Understanding safety in non linear systems
(Dekker, 2007 Hollnagel, 2008)

• Assumption and consequence
• Accidents result from unexpected combination
  (resonance) of normal performance variability
  hazards emerge from expected (and necessary)
  variability within the system and accidents are
  prevented by monitoring and damping the
  variability.
• The variability of normal performance is rarely
  sufficient to result in an accident, but the
  variability from multiple functions may combine
  in unexpected ways to produce a non linear
  effect, thus safety is an emergent property of
  the system and cannot be explained by simply
  examining individual components of the system
  and/or trying to identify a root cause.
• Complex socio-technical systems (such as
  healthcare) are dynamicthe systems change and
  develop in response to competing demands,
  production pressures and changes in technology
  and knowledge. Resilience exists when operators
  in the system are able to recognize, absorb and
  adapt to disruptions/changes that fall outside of
  their design base.

19
Functional Resonance Accident Model (Hollnagel,
2004)
An emerging model that attempts to understand the
dynamics of normal organizational activity
20

• Non linear accident models provide investigators
  with the basis to find out why peoples actions
  and assessments made sense to them at the time
  rather than identifying what rule, protocol or
  process the person violated (people often adapt
  their actions given the context at hand and this
  is part of the normal performance variability
  that takes place as part of normal work in
  dynamic, complex systems).
• Human error is not an explanation, but demands
  an explanation
• Sidney Dekker, 2006

21

• In contrast to this, linear models provide the
  basis for investigators to easily take the
  position of retrospective outsider, looking back
  on a sequence of events that seems to lead to an
  inevitable outcome, and pointing out where people
  went wrong, or where individual components of the
  system failed.
• Although this perspective is often adequate in
  linear systems, the approach seriously limits
  what the investigator can learn about failure in
  non linear systems (i.e. the complex and
  unexpected combination of system interactions)
  and may not help prevent recurrence.
• Hindsight Bias!

22
Accident models The challenge of hindsight bias

• Hindsight bias is always present when the
  outcome is known a retrospective outsider can
  easily confuse post hoc reality with the actual
  reality surrounding people during the event.
• Hindsight bias is a powerful reason for old
  view explanations for human error and accidents
  tending to look for individual components of
  the system that need fixing people deficient in
  skills or egregious mistakes.
• Hindsight bias makes it difficult to objectively
  judge behavior leading up to the outcome. In
  particular, past complexity is transformed into a
  linear string of bad decisions, missed
  opportunities, flawed assessments, and faulty
  perceptions.
• Thus, recommendations too often focus on
  protecting the system from unreliable humans
  through procedures, automation, training and
  discipline.

23
Thesis work

• The aim of the thesis work was to begin to
  explore the extent, and in what ways, safety and
  quality are conflated in healthcare, at both the
  sharp and blunt ends of care in an acute care
  institutional setting within a large health
  authority in Canada.
• The key questions this research sought to answer
  are
• How are the notions of patient safety
  operationalized through local context?
• How is safety thought about and constructed?
• How is it discussed?
• Is it neglected?

24
Methods

• Survey work interviewed key informants
  (registered nurses working in acute care nurse
  managers responsible for acute care units and
  senior decision makers).
• Semi-structured face-to-face interviews were
  conducted (interview guide was developed to
  support the discussion and contained a series of
  open ended questions) to find out how the key
  informants thought about safety and how they feel
  they contribute to safety on a day-to-day basis.

25
Interview guide (registered nurses)

• How long have you worked as a registered nurse?
• How do you define the term safety?
• What factors and activities help contribute to
  patient safety at your institution, in general,
  and in particular, on your unit?
• What do you think would improve patient safety in
  acute care hospitals, in general, and in
  particular, on your unit?

26
Interview guide (nurse managers and senior
decision makers)

• What is your role in the organization?
• How long have you worked in healthcare?
• How do you define the term safety?
• What factors and activities help contribute to
  patient safety at your institution?
• What do you think would improve patient safety in
  acute care hospitals?
• Does the work you do contribute to safety? If
  yes, in what ways?

27
Survey results Major themes

• Designing robust organizations (prescriptive
  practice)
• Designing robust organizations (compliance)
• Designing robust organizations (rules and
  procedures are important but insufficient to
  create safety)
• Expertise and experience
• Adaptation of work, depending on the context and
  competing priorities
• Efficiency-thoroughness-tradeoff
• Unpredictable notion of safety
• Learning from near misses and critical incidents
• Storytelling as a form of learning
• Communication and teamwork
• Leadership
• Competing system challenges
• Vigilance and troubleshooting

28

• There were notable differences regarding the
  emphasis that each group placed on the respective
  themes.

29
Thesis findings cont

• Safety is important, but people are still looking
  for standard fixes and are influenced by
  conventional opinion leaders (e.g. Safer Health
  Care Now campaign, and Saving 100,000 Lives
  Campaign).
• Confusion regarding the difference between safety
  and quality exists and the confusion is greater
  at more senior levels in the organization (i.e.
  people continue to think that if you improve
  quality through standardization, guidelines,
  procedures, etc that safety will automatically
  follow).
• People at senior levels focus on the need to
  develop robust systems that are marked by
  guidelines, protocols, rules and also focus on
  training, technology, rules and enforcing
  compliance as solutions.
• People at the front lines (the practitioners)
  understand the need to adapt their behaviour and
  practice in unusual situations, but are tentative
  in how they discuss this with both their peers
  and managers, aware of potential negative
  consequences or sanctions, if things dont work
  out well.

30
Thesis findings cont

• Set of ingrained attitudes about how work is
  performed, i.e. there is no gap between work as
  imagined and work as done, i.e. work can be
  performed in a high quality manner, despite the
  context this is an easy perspective for senior
  management to adopt since it feeds off the sense,
  amongst most professional groups in healthcare,
  that they are, or should be, perfect and can
  provide high quality care under a range of
  conditions.
• Lack of deep understanding of the source of
  failure in complex organizations.
• Superficial understanding of hindsight bias and
  its impact on what you look for when you are
  doing critical incident investigations.
• Accountability remains a thorny issue,
  particularly at the senior management and
  governance level, with little consideration of
  the accountability/authority dynamic.

31

• In practical terms conflating the concepts of
  quality and safety in a complex, dynamic setting
  such as healthcare can result in investing
  efforts to solve the wrong problem and thus
  potentially misappropriates limited human and
  financial resources.
• Besides the potential misappropriation of
  resources, if quality and safety are conflated,
  it is far too easy to assume that if one improves
  quality that safety will automatically follow,
  and thus the system, unfortunately, continues
  doing more of the same neither fully
  understanding or adequately tackling the problem.

32
Conclusions

• Safety and quality are often conflated in health
  care and this may limit progress on both creating
  safety AND enhancing quality.
• Safety is the attribute of being able to respond
  to surprise or instability of the
  system--creating safety involves anticipating
  what could go wrong.
• Non linear accident models, based on an
  understanding of both high reliability and
  resilience theories, and empirical evidence from
  high risk, dynamic settings, can help us
  appreciate why safety and quality need separate
  strategies.
• There were noticeable differences in how the key
  informants (from the thesis work) talked about
  safety and the perspectives of the people at the
  sharp end of the system (point of care) are
  fairly consistent with what the thought leaders
  in system safety are telling us about creating
  safety in complex dynamic environments.

33

• Making progress in safety may be supported by a
  better understanding by system leaders about how
  work at the frontline actually gets done (normal
  work) as well as a better understanding about the
  necessity and value of performance variation in
  complex, dynamic systems. There needs to be
  discussion around the tension between developing
  robust systems (marked by rules, procedures etc),
  while at the same time, supporting performance
  variation.
• A dedicated interdisciplinary Safety Management
  System, with a broad mandate, is one structural
  tool that may help healthcare organizations make
  progress on safety
• Develops an analytical framework for critically
  monitoring safety using a non linear perspective
  
• Keeps the discussion of risk alive within the
  organization
• Enables people at the sharp end to actively look
  for the things that could go wrong and understand
  how to keep these at bay
• Surveillance activity

34
References

• Bagian, James, Director, VA National Centre for
  Patient Safety. Personal communication, July
  2007.
• Cook, R. I., Woods, D. D., Miller, C. (1998). A
  tale of two stories Contrasting views of patient
  safety. Report from a Workshop on Assembling the
  Scientific Basis for Progress on Patient Safety
  National Patient Safety Foundation.
• Dekker, S. W. A. (2001). Reconstructing human
  contributions to accidents The new view on error
  and performance (Rep. No. Technical Report
  2001-01). Ljungbyhed, Sweden Lund University
  School of Aviation.
• Dekker, SWA (2002). The field guide to human
  error investigations. Bedford, UK Cranfield
  University Press/Aldershot UK Ashgate Publishing
  Co.
• Dekker, S. W. A. (2003). Errors in understanding
  of human error the real lessons from aviation
  for healthcare (Rep. No. Technical Report
  2003-01). Ljungbyhed, Sweden Lund University
  School of Aviation.
• Dekker, SWA (2005). Ten questions about human
  error A new view of human factors and system
  safety. Mahwah, NJ Lawrence Erlbaum Associates.
• Dekker, S. W. A. (2005). Why we need new accident
  models (Rep. No. 2005-02). Ljungbyhed, Sweden
  Lund University School of Aviation.
• Dekker, S.W.A. Past the edge of chaos. Technical
  Report 2006-03. Lund University School of
  Aviation. Ljungbyhed, Sweden.
• Dekker, S.W.A. Professor, Lund University School
  of Aviation. Personal communication, October
  2007, Winnipeg, Manitoba.
• Hollnagel, E. (2004). Barriers and accident
  prevention. Aldershot, UK Ashgate Publishing Co.
• Hollnagel, E. Woods, DD. And Leveson, N. (2006).
  Resilience engineering. Concepts and precepts.
  Aldershot, UK Ashgate Publishing Co.
• Hollnagel, E. Theory W and theory Z Contrasting
  views on safety. Presentation. LEcole de Mines,
  Paris.
• Leveson, N. (2004). A new approach to systems
  safety engineering. MIT.
• Perrow, C. (1999). Normal accidents Living with
  high-risk technologies. Princeton, NJ Princeton
  Paperbacks.
• Reason JT. (1997) Managing the risks of
  organizational accidents. Aldershot, UK Ashgate
  Publishing Co.
• Reinertsen, James L. and Clancy, Carolyn Foreword
  to Keeping our Promises Research Practice, and
  Policy Issues in Health Care Reliability. A
  Special Issue of Health Services Research, Vol
  41, No. 4, Part II, August 2006
• Resar, R. Consultant, Institute for Health
  Improvement. Personal communication, July 2007.
• Roberts, K. H. Bea, R. G. (2001). When systems
  fail. Organizational Dynamics, 29, 179-191
• Sagan, SD (1993). The limits of safety
  Organizations, accidents and nuclear weapons.
  Princeton, NJ Princeton Paperbacks.