Systems thinking

In safety science, the ‘systems thinking’ approach asserts that safety and accidents are emergent properties which arise from interactions between multiple components across work systems (eg Rasmussen, 1997). Safety is viewed as a shared responsibility that spans all levels of work systems, up to and including managers, chief executives, regulatory bodies, government, and international organisations.

Attempting to understand and manage H&S thus involves looking beyond the ‘sharp end’ (ie individuals and circumstances directly involved in safety incidents) to also consider factors within the broader organisational, social, or political system.

Recently I was invited to give the Australian Institute of Health and Safety’s Eric Wigglesworth Memorial Lecture based on my work applying systems thinking in Occupational Health and Safety (OHS) research. During the lecture I discussed the following ten critical lessons that I have learned from almost 20 years of OHS research.


Lesson 1: A human error lens is useless and can be dangerous.

Unfortunately, there remains a fixation on human error and the behaviour of workers at the sharp end when attempting to understand and enhance OHS. This can be misleading, unhelpful, even dangerous. While errors may occur, they are consequences of interacting factors across the broader work system – workload and production pressures, tools, procedures, training, management decisions, culture, financial pressures and so on. This means that is difficult to effectively manage OHS without looking at the broader work system and all its interacting parts.


Lesson 2: All OHS incidents are caused by multiple interacting contributory factors.

Our research across the safety critical domains demonstrates that all workplace incidents, whether minor or major, are caused by multiple interacting factors (eg Salmon et al., 2020). The idea of a root cause is misleading, and it is fanciful to try and identify one. Rather, it is critical that reporting and analysis tools are used to identify the network of contributory factors involved in all incidents, minor and major.


Lesson 3: There may be a generic causal network, regardless of industry sector.

Our research also indicates that there may be generic causal network that appears in major OHS incidents, regardless of industry sector (Salmon et al., 2020). This network includes a set of recurring contributory factors such as policy, procedures, standards, legislation and regulation, risk assessment and management, planning and preparation, training and experience of staff across all work system levels, personnel recruitment and management, supervision and leadership, equipment and resources, and the physical and natural environment (Salmon et al., 2020).


Lesson 4: Many of the risk and safety management methods used in practice are out of date.

There is a research-practice gap where state-of-the-art safety science methods are not being applied in OHS practice. While there is a strong appetite for systems thinking in OHS, there are few organisations applying appropriate methods for activities such as risk assessment, incident reporting, and accident analysis. This is problematic, as the analyses produced may be invalid, misleading, and even dangerous (Salmon et al., 2021). A critical challenge for OHS is to bridge the research-practice gap and ensure that appropriate methods are being applied in practice.


Lesson 5: Emergent risks are elusive and tend to catch organisations out.

Organisations often have a good handle on well known recurring safety risks. Emergent risks are a problematic set of unforeseen risks which are created when hazards across work systems interact with one another. These are difficult to identify and control and often catch out even the most well-prepared organisations. It is critical that pro-active risk assessment activities attempt to identify and understand these emergent risks through the application of methods such as the Networked Hazard Analysis and Risk Management System (Net-HARMS; Dallat et al., 2018).


Lesson 6: OHS incidents are not always caused by failures.

A key tenet of the systems thinking approach is that OHS incidents are not always caused solely by failures. Rather, the network of contributory factors often includes a combination of ‘work as imagined’ (ie undertaken in line with procedures), ‘normal performance’ (ie ‘work as done’, where normal and accepted behaviours are undertaken to get the job done), and decisions and actions whereby performance can reasonably be classified as sub-optimal.

Focusing only on failure is too narrow for effective OHS management.


Lesson 7: Automation is not the silver bullet for OHS.

Automation is often heralded as something of a silver bullet for OHS, with promises of more efficient and safer performance once automated systems are introduced into the workplace. Unfortunately, this is often not the case and there are many examples where ‘unruly technologies’ behave in ways not foreseen by designers and create new safety issues. Though first published in 1983, the ‘Ironies of automation’ described by Lisanne Bainbridge remain highly relevant for today’s work systems.


Lesson 8: Leverage points should be used to drive improvements in OHS.

Organisations typically respond to safety issues by introducing more training, new procedures, or by replacing workers with technology; however, often the most powerful place to intervene lies elsewhere in the system. These so-called leverage points represent areas where small interventions can have large effects on behaviour. Donella Meadows (2008) identifies 12 types of leverage point; organisations should seek to identify their own when addressing ongoing safety issues.


Lesson 9: Humans are the glue that hold badly design and degraded work systems together.

Most work systems are poorly designed, operate in a degraded state, and migrate toward safety boundaries under different pressures and influences. Human workers at all levels are the glue which hold work systems together. OHS practitioners should avoid treating human workers as the source of safety problems and instead acknowledge their critical contribution to OHS.


Lesson 10: An integrated framework of safety management methods is critical.

Effective OHS management requires the use of an integrated framework of systems thinking methods for risk assessment, incident reporting and learning and accident analysis. Using disconnected safety methods is problematic as the outputs of one method cannot be used to inform applications of the other and data cannot be aggregated and integrated. Our recently developed Systems Thinking Accident and Risk Toolkit (START) provides a set of integrated methods to support risk assessment, incident analysis, and the development of safety interventions.


Paul Salmon is a Professor of Human Factors and is the Director of the Centre for Human Factors and Sociotechnical Systems at the University of the Sunshine Coast.


Paul Salmon has found that humans are the glue that hold together poorly designed and degraded work systems. What is the clearest example of this that you have encountered? Write a brief comment here and go in the draw to win a prize!



Dallat, C., Salmon, P. M., & Goode, N. (2018). Identifying risks and emergent risks across sociotechnical systems: The NETworked Hazard Analysis and Risk Management System (NET-HARMS). Theoretical Issues in Ergonomics Science, 19(4), 456-482.

Meadows, D. (2008). Thinking in systems: a primer. Chelsea Green Publishing.

Rasmussen, J. (1997). Risk management in a dynamic society: A modelling problem. Safety Science, 27:2/3, 183-213.

Salmon, P. M., Hulme, A., Walker, G.H., Berber, E., Waterson, P., Stanton, N. A. (2020). The big picture on accident causation: A review, synthesis and meta-analysis of AcciMap studies. Safety Science, 126, 1-15.

Salmon, P. M., Read, J. M., Walker, G. H., Stevens, N. J., Hulme, A., McLean, S., Stanton, N. A. (2021). Methodological issues in systems human factors and ergonomics: perspectives on the research-practice gap, reliability and validity, and prediction. Human Factors and Ergonomics in Manufacturing and Service Industries.