Can safety be managed, can risks be anticipated?

Experiences from aviation, health care, petroleum

Human Factors & Safety seminarEspoo, March 17 2010

Karina Aase

Can safety be managed, can risks be anticipated?

• That’s the question(s)!

• The question(s) has raised a continuous debate within safety science!

• AGENDA:– The safety science debate– Evidence?– Examples from aviation, health care, petroleum…

The question(s) is ok, but what is the goal(s)?

• ”To function reliable and to anticipate risks caused by either technology, organizational structures, and practices” (Reiman & Oedewald 2009)

+– Productivity– Financial safety– Environmental safety– Personnels’ safety

Can safety be managed, but what is safety?

• ”Safety is not something an organization possesses, it is a competence that should be continuously learned and developed” (Gherardi & Nicolini, 2000)

• ”Learning safety accurs in practice, in contextual work settings” (Brown & Duguid, 1991; 2001)

???

?

The safety science debate

NormalAccidents

High Reliability

OrganisationsEnergy & barriers

Information processing

Decision-making

Contextual setting

?: does the theories answer our question in different

contextual settings?

Resilience Engineering

High Reliability Organizations (HRO)

Normal Accidents (NA)

• Accidents can be prevented

• Safety is the priority

• Redundancy (duplication and overlap) enhances safety

• Decentralized decision making

• A "culture of reliability” to create uniform and appropriate responses

• Continuous operations, training, and simulations

• Trial and error learning supplemented by anticipation and simulations

• Accidents are inevitable

• Safety is one of a number of competing objectives

• Redundancy often causes accidents

• Centralization is needed for tightly coupled systems

• A ”military” model is incompatible with democratic values

• Organizations cannot train for unimagined, highly dangerous operations

• Denial of responsibility, faulty reporting, and reconstruction of history cripple learning efforts

HRO versus NA

Adapted from Reason, 1997Adapted from Reason, 1997

Energy and barriers

HazardsHazards

Decision making

• Actors cross boundaries for acceptable risk due to optimalisation of their own behaviour or work practice

• Activities have a tendency to move towards the boundaries for acceptable risk practice since the actors seek for suitable trade-offs concerning workload and productivity

Optimise locally

Error margin

Boundary of functionally acceptable performance

Resulting perceived boundary of acceptable performance

Acceptable behaviour (risk)

Work tasks/workload

Financial/management pressure

Safety space (Rasmussen, 1997)

Production Safety

• Easy to measure (”revenue”, ”earnings”, ”expenses”)

• Indicates success in a positive fashion (e.g., increasing earnings)

• Is reinforcing, and has high salience (the bottom line is the "bottom line" for a firm)

• Relationship between application of resources (money effort, time) and production goals is relatively certain, making it easy to utilize feedback

• Difficult to measure (indirect, discontinuous, difficult to interpret)

• The feedback is provided "negatively" (fewer accidents or incidents)

• Has little reinforcement value, high salience only after an accident or a serious near-miss

• Relationship between application of resources and safety goals is relatively uncertain, making it hard to utilize the feedback

Goal conflicts

Information processing

• Accidents are not ”fundamental surprises”.

• Precursors or warnings are nearly always identified on hindsight.

• Accidents develop through a long chain of events with root causes like lack of information flow and misperception among individuals (incubation period).

(Barry Turner, 1978)

Does the theories answer our question(s)?

Theoretical perspective

Can safety be managed and risks be anticipated?

Normal Accidents (NA)

Sooner or later, accidents happen due to complexity in the systems NO!

High Reliability (HR) Organising for safety as top priority, emphasising redundancy, safety culture, and trial and error to learn YES!

Energy and barriers Individual, cultural and organisational barriers prevent accidents to develop and escalate YES/no

Decision making Simultaneous priorities are emphasized, balancing risk and safety against productivity and efficiency NO/yes

Information processing

Information is the key to understand and learn from accidents, information processing mechanisms are vital yes/no

Resilience Engineering

Unforeseen technological, environmental or behavioural phenomena will occur, safety is created through proactive resilient processes rather than reactive barriers and defenses Yes

Evidence? How do we collect it?

• Depending on what perspective we believe in:

– safety management system audits– safety culture evaluations– organisational culture studies– reviews of human performance programs,

operational experience etc– usability evaluations of critical technology– qualitative risk assessment– quantitative risk assessment– accident statistics– and many more…

Evidence within risk assessment (statistics)?(Source: Amalberti et al 2005)

Fatal iatrogenic adverse events

Cardiac surgery in patient in ASA 3-5

Medical risk (total)

Himalaya mountaineering

Microlight aircraft or helicopters

Road safety

Chemical industry (totalt)

Nuclear industry

Railways

Charteredflight Commersial

large-jet aviation

Anesthesiology in patient in ASA 1

Blood transfusion

10-2 10-3 10-4 10-5 10-6

Very unsafe UltrasafeRisk

Lets look at some examples….

What does the petroleum sector believe in?

”The Norwegian petroleum sector can draw today on a knowledge bank which allows government, industry and unions to reap safety gains. That makes it possible to monitor and influence risk level trends throughout the business”

(Petroleum Safety Authority, 2009)

• YES, safety can be managed• YES, risks can be anticipated

Anticipating risks in petroleum

DFU Defined situations of hazard and accident Data sources

1 Non-ignited hydrocarbon leaks Data acquisition*

2 Ignited hydrocarbon leaks Data acquisition*

3 Well kicks/loss of well control DDRS/CDRS (PSA)

4 Fire/explosion in other areas, flammable fluids Data acquisition*

5 Vessel on collision course Data acquisition*

6 Drifting object Data acquisition*

7 Collision with field-related vessel/installation/shuttle tanker

CODAM (PSA)

8 Structural damage to platform/stability/anchoring/positioning failure

CODAM (PSA) + industry

11 Evacuation (precautionary/emergency evacuation) Data acquisition*

12 Helicopter crash/emergency landing on/near installation Data acquisition*

* Data collection in cooperation with operator companies

Managing safety in petroleum

• The Norwegian Petroleum Safety Authority has on behalf of the Norwegian government taken the step of requiring petroleum companies to develop a safety culture

”The party responsible shall encourage and promote a sound health, environment and safety culture”

(Regulations relating to health, environment and safety in the Norwegian petroleum activities, 2001, section 11)

Theoretical perspective Safety practices within health care

Normal Accidents (NA) Accidents happen, but not due to complexity within the health care system

High Reliability (HR) Safety is not top priority, focus on developing safety cultures and learning lacks

Energy and barriers Limited consciousness regarding risks and possible consequences, hard to develop and implement specific barriers

Decision making Productivity has a tendency to ”win” in the goal conflict between safety and productivity

Information processing Lack of mechanisms for information processing to understand and learn from undesired events

Evidence within health care?

Evidence within health care?

• Risk perception and priority?Health care organisations have only to a limited extent performed general risk analyses as background for their priorities.

• Resources and competence?Health care organisations have only to a limited extent specialised and competent personnel to manage undesired events. The work has often low prestige and impact in the organisations.

• Health care organisations have no outline of the costs related to undesired patient events.

(Norwegian Directorate of Health, 2005)

Evidence using safety culture studies?

72

72

69

64

50

49

40

39

31

28

25

76

77

72

67

50

46

42

36

29

27

21

0 10 20 30 40 50 60 70 80 90

Supervisors actions promoting safety

Non-punitive response to errors

Teamwork within units

Communication and openness

Org.learning and continuous improvement

Staffing

Feedback and communication about errors

Organisational handoffs and transitions

Teamwork across units

Reporting of (near) misses

Organisational management support for safety

Prosent

SUS 2008

SUS 2006

Degree of agreement on positive items and disagreement on negative items. Mean percentage within each dimension.

What does the health care sector believe in?

”Risk is in the nature of medicine because we cut in peoples bodies. People arrive with dramatic stuff, serious illnesses that we are supposed to treat with surgery. That is a risk in itself. There is a grey zone where you must assess if surgery is beneficial or harmful to the patient. You are in focus, and the results of your professional assessments always appear after your actions. I you choose not to do surgery and the patient dies it might be blameworthy, and if you choose to do surgery and the patient dies it might be blameworthy as well. That is probably why we receive complaints, because expectations towards the results are unrealistic”

• NO, safety can not be managed• NO, risks can not be anticipated

Evidence within aviation - statistics

Evidence within aviation – safety practices

• MAKROLEVEL (aviation authority):Professional pride, sense of responsibility, individually oriented work practices. Procedures have respect, status and normative value.

• MESOLEVEL (ATC and airport operation):Safety practices vary. Goal conflicts regarding safety and efficient traffic handling. Procedures have respect, status and normative value. BUT, are violated related to security, traffic handling and reporting.

• MIKROLEVEL (maintenance):Professional pride, safety consciousness. Goal conflicts concerning efficient traffic handling are managed by using slack, flexibility and experience-based knowledge. Procedures have respect, status and normative value. BUT, contains grey zones requiring other work practices.

Evidence within aviation – framework conditions

1. Interfaces between different aviation actors are negatively affected by new organisational and physical interfaces:– geographic relocation of aviation authority– separation of maintenance and airline company– relocation of maintenance base

2. Oil prices3. Financial crisis

So far, collective mechanisms compensate:– Professionality, safety consciousness, sense of responsibility– Desire to learn and develop personal competence– Emphasis on rules, procedures and new technology

What does the aviation sector believe in?

• Supranational regulation• International agreements• Proseduralisation• Reporting and systems

• YES, safety can be managed, but…• YES, risks can be anticipated, but…

Summary – what have we?

• A research community with different beliefs

• Three sectors with different beliefs

• Government and authorities with one belief?

• The public with one belief?

At last – a warning!

• How desirable is proceduralisation?

1. The government regulatory context where safety related requirements are designed for companies to comply with

2. The company safety management context where companies are designing their internal safety management systems

3. The safety authority context where inspections and audits seem to produce more and more paperwork instead of a deep understanding of the conditions under which operations are conducted.

(Bieder & Bourrier, 2010)

Thank you for your attention!

Everyone Makes Mistakes

Yes – Everyone!

That opens for discussion?