Learning from incidents is easier said than done. Edward Smith and Richard Roels, of DNV GL, give guidance on how to maximize learning from incidents, accidents and other events.
Photos from DNV GL.
Following a major incident, it is common to hear those involved talk about “learning lessons” in a way that implies this is an automatic or natural process. In fact, evidence suggests that it can be challenging for major hazard industries to learn effectively from incidents.
Responding that “we have learnt from this incident” is an open phrase that could mean different things to different people:
a. That the team of investigators has understood how and why an incident occurred.
b. That several people in an organization now know how to prevent it happening again.
c. That an organization has implemented a set of changes (for example in equipment and personnel behaviors), which will prevent this event happening again.
d. That an organization has implemented a set of changes, which will prevent this event, and similar events, happening again and even learnt about its processes for Learning from Incidents (LFI) as a result of an incident investigation.
These points represent a range of learning potential; however, a change must happen before it can be claimed that learning has occurred. Implementing changes and taking preventative action would be expected to reduce risk further than simply carrying out an investigation.
The Energy Institute (EI) published “Guidance on investigating human and organizational factors aspects of incidents and accidents” in 2008. While most companies have now implemented processes for identifying why an incident happened, the next challenge is to learn from, and change after, an incident. The EI is currently updating the original guidance to cover the whole LFI process, from reporting and finding out about incidents through to effective learning and changing practices.
Key LFI stages
The LFI process consists of several stages including reporting, investigation, producing recommendations and actions, and broadening learning.
While many incidents will be formally reported, not all are and details of others will be gathered informally. Research has indicated that many incidents per actual accident need to be reported to drive organizational learning.
Incident investigation is considered to cover both the fact finding and analysis stages, recognizing that iterations between these stages will be required to eventually reach a conclusion of what happened and why. Some organizations may struggle to provide competent investigation personnel, especially to cover specialist topics such as Human and Organizational Factors (HOF).
Carrying out recommendations and actions is a common area of weakness, even investigations carried out by experienced investigators can make recommendations that are never implemented.
Improving the processes around managing recommendations and the resulting actions will help implement the local learning associated with a specific event and lay the foundations for broader learning. There are clearly difficult challenges in reaching a wider group of people, beyond those immediately affected, and helping them to learn from an incident.
Previous research into LFI has identified that along each of the stages there are blockers to learning, e.g.:
- Fear of being blamed, professionally embarrassed or incriminating others by reporting.
- Lack of awareness about what should be reported or included in an investigation.
- Complexity of reporting systems and recommendations.
- Insufficient resources allocated to investigation.
- Recommendations not accepted by line management
- Insufficient checks that recommendations will effectively reduce risk.
- Legal constraints.
- However, each set of blockers also has a set of enablers, which have been identified to overcome the challenges, e.g.:
- User-centered reporting processes.
- Feedback at end of LFI process to demonstrate that reports lead to effective changes.
- Senior management commitment to provision of competent investigators.
- Consultation while developing recommendations
- Check recommendations are risk proportionate and will not inadvertently increase risk.
- Turning lessons into good practice guidance, which does not carry liability risks.
- Develop indicators for following up on implemented actions.
The final step in implementing LFI is evaluation, i.e. has effective learning taken place. Monitoring measures with tangible outcomes, such as number of recommendations overdue, is relatively straightforward. However, corporate and personnel attitudes are harder to measure and more can be done in this area to establish good practices.
The potential blockers show the number of challenges to implementing an LFI process. By updating the EI guidance to include the whole process, the good practice enablers mentioned will allow organizations to build a structure, which will improve LFI. Reducing the blockers to learning and increasing the learning potential will ultimately improve risk management.
This paper was originally presented at Hazards 25, 13-15 May 2015, Edinburgh, UK
Dr Edward Smith is a Senior Principal Consultant at DNV GL. He has a PhD on how maintenance management affects the reliability and safety of major hazard systems. He joined DNV in 1989 where his main activities have involved safety risk assessment, safety management systems and human factors. He has also worked in aviation (especially Air Traffic Management), offshore, dangerous goods transport, rail and onshore process industries.
Richard Roels is a Senior Consultant at DNV GL. He is a Chartered Occupational Psychologist specializing in the management of major accident risk in the oil and gas and transportation industries. His work covers aspects relating safety management and human factors, including inspection activities on behalf of regulators, conducting safety reviews and inquiries following serious incidents, as well as undertaking other specialist studies.