What you find is not always what you fix—How other aspects than causes of accidents decide recommendations for remedial actions
Graphical abstract
Introduction
To learn from events is often celebrated as one of the key principles of effective safety management. Experience feedback from events take many forms such as collecting statistics and performing in-depth analysis of event (Kjellén, 2000). A basic assumption underlying accident investigation is that analysis of specific events will reveal patterns of underlying causes and conditions that if addressed by the right remedial actions can prevent further events. The distinction often drawn between retrospective (analysis of events) and prospective (risk analytical) methods is intuitively understandable but incomplete: risk analysis requires the experience from analysis of previous events and event analysis implies that the weaknesses found are the ones that impose risk. Consequently, in accident investigation, an ideal is often to follow the principle “what-you-find-is-what-you-fix”, which is clearly a guiding principle for instance in Swedish accident investigation manuals and guidelines (Lundberg et al., 2009). (It is obviously impossible to fix something that has not been ‘found,’ i.e., which has not basis in reality. Even so, it happens every now and then, cf., below.) Another way to state this principle is: if an accident happens determine the causes and implement suitable arrangement to eliminate the causes and/or their effects. This is also a goal that guides research: there are numerous articles and books describing methods for finding the right causes, as well as articles describing “accident models”, that is generic models of factors and their relations that can provide support for finding cause–effect relationships behind accidents (e.g. Heinrich, 1931, Heinrich, 1934, Gordon, 1949, Lehto and Salvendy, 1991, Svenson, 1991, Kjellén, 2000, Hollnagel, 2004, Leveson, 2004, Sklet, 2004, Factor et al., 2007, Santos-Reyes and Beard, 2009). Fewer studies (e.g. Elvik, 2010) focus on difficulties of fixing what has been found. Modern accident models focus on factors and relations other than those focusing on humans closest to the events. Such approaches are based on the idea that numerous factors and conditions in a complex socio-technical system may have influence on accidents: including political and organizational factors, cultural factors, and issues of power relations, technological development and so forth (Hollnagel, 2004, Leveson, 2004, Santos-Reyes and Beard, 2009). This extension of the scope of issues relevant for understanding accident propagation has lead to a deeper understanding of safety. Accident models of today often include nuanced ideas about “factors” behind accidents as well as elaborated ideas about cause–effect relations. Somewhat surprisingly, however, the same factors that often are highlighted in modern accident models are not perceived in a recursive manner to reflect how they influence the process of accident investigation in itself. Another way to rephrase this issue is the following: how does the same organizational context that is responsible for accidents impose constrain on the methods and understanding of accident investigation and associated methods. In this article we have set out to approach some of these issues by means of interviews with accident investigators from various branches. Our goal has been to reveal patterns of influence affecting accident investigation practices that presumably represent the same roots as those often claimed to be “root causes” to accidents.
Section snippets
Purpose
Our purpose is to reveal constraints affecting accident investigation practices that lead the investigation towards or away from the ideal of “what-you-find-is-what-you-fix”.
Method
We used the interview guide in Appendix to guide the interviews. During the interviews, the informants could also initiate topics of their own (Question 4.4. was not used in the interviews with the health care sector, interviews 17–22, Table 1). Interviews 1–10, Table 1, and the same interview guide, was used as data in a previously published study (Korolija and Lundberg, 2010).
The interview guide (see Appendix) covered five areas: background information about the informant, the phases of an
Bias in accident investigation
Previously, a need for research on real-world accident analysis has been requested, to find sources of bias that actually occur (Woodcock, 1995). However, some sources of bias in accident investigation are well known. Johnson (2003) lists several biases that can affect investigation, for instance the following: Author bias, a reluctance to accept findings from other people's investigations. Confirmation bias, a tendency to confirm preconceived causes. Frequency bias, a tendency to classify
Results and discussion
Our interviews exemplified many factors that drive the investigation towards the “what-you find-is-what-you-fix” ideal. There were, however, also many examples of constraints that make the investigation go in other, less desirable directions. In the following two sections, we first describe and discuss factors affecting data collection and analysis, and then factors that affect the design or selection of remedial actions.
Summary—what you find is not always what you fix
Investigators do mention their investigations follow the ideal, what-you-find-is-what-you fix. However, the data suggests that analysis and design of remedial actions both are influenced by constraints. This affects how accident model and factors are practically employed in an investigation. The influence of constraints on the design of remedial actions indicates that what you find is not always what you fix. There seems to be several variations of deviations from the ideal, related to
Conclusions
We conclude that there are many factors that can draw an accident investigation away from the ideal principle of what-you-find-is-what-you-fix. That principle is clearly the ideal in many Swedish investigation manuals (Lundberg et al., 2009). Nevertheless, factors both in the analysis and data collection phases, the transition to design of remedial actions, and during the design of remedial actions, drive investigations towards other principles. We also conclude that the holy grail of research,
Acknowledgements
We thank the anonymous informants for sharing their accident investigation experiences with us. This research was sponsored by the Swedish Civil Contingencies Agency through the project “Assumptions on accidents and consequences for investigation and remedial actions”.
References (22)
Why some road safety problems are more difficult to solve than others
Accid. Anal. Prev.
(2010)- et al.
The social accident: a theoretical model and a research agenda for studying the influence of social and cultural characteristics on motor vehicle accidents
Accid. Anal. Prev.
(2007) - et al.
Speaking of human factors: emergent meanings in interviews with professional accident investigators
Safe. Sci.
(2010) - et al.
Models of accident causation and their application: review and reappraisal
J. Eng. Technol. Manage.
(1991) A new accident model for engineering safer systems
Safe. Sci.
(2004)- et al.
What-you-look-for-is-what-you-find—the consequences of underlying accident models in eight accident investigation manuals
Safe. Sci.
(2009) - et al.
The context and habits of accident investigation practices: a study of 108 swedish investigators
Safe. Sci.
(2010) - et al.
A systemic analysis of the edge hill railway accident
Accid. Anal. Prev.
(2009) Comparison of some selected methods for accident investigation
J. Hazard. Mater.
(2004)The epidemiology of accidents
Am. J. Public Health
(1949)
Industrial Accident Prevention: A Scientific Approach
Cited by (85)
Improving Remedial Measures from Incident Investigations: A Study Across Ghanaian Mines
2024, Safety and Health at WorkConceptualising learning from resilient performance: A scoping literature review
2024, Applied ErgonomicsMajor aviation accident investigation methodologies used by ITSA members
2023, Safety ScienceLearning from cyber security incidents: A systematic review and future research agenda
2023, Computers and SecurityHow to Mitigate the Effects of Cognitive Biases During Patient Safety Incident Investigations
2022, Joint Commission Journal on Quality and Patient Safety
- 1
Tel.: +46 08 5397260.
- 2
Permanent address: MINES ParisTech, Crisis and Risk Research Centre (CRC) Rue Claude Daunesse, BP 207, 06904 Sophia Antipolis Cedex, France.