The causes of accidents

If your company’s accident causation model looks like a row of dominoes you may be lead to believe that accidents, incidents, and bad safety performance are the result of unsafe act and unsafe conditions., both of which are ultimately the responsibility of people ,as people are involved in every step from design, construction, operation, maintenance, and decommissioning.

This philosophy is based on the work of an engineer named H. W. Heinrich in the 1930’s who was trying to determine the causes of accidents and how to prevent them. The model grabbed the attention of many managers and business owners seeking to reduce downtime, lost production and injury to their workers and increase their returns. The model gives management the impression that the only things they can do to improve their health and safety performance are to:

* Hire smarter people,
* Give them plenty of training,
* Supply detailed standard operating procedures, and
* Use the safety department as a kind of police force to ensure employees adhere those policies and procedures

This thinking persists to this day in spite Heinrich’s second book, in which he dismisses the misconceptions people had on his previous work. Unfortunately, this book was not that popular and the original misinterpretations persist.

The supply of training and procedures assumes that they are comprehensive, correct and applicable for every possible situation in the workplace. During audits of work place procedures it becomes clear that the procedures are more a statement of intent, rather than a strict list of steps to follow. Examination of what actually occurs on the workplace floor often reveals the procedures do not, and cannot, take into account all the possible variations and permutations possible. There are just too many of them to adequately take into account for any procedure to be comprehensive and constantly applicable.

Employees at all levels instinctively know this, and follow their own logic and reasoning based on the information presented to them at the time. In our accident investigation work it is often revealed that the information presented to employees is plainly incorrect and leads to incorrect decisions (although that will not be identified at the time) and/or enforcing the procedures without regard for the particular circumstances can actually makes things worse, not better.

Another model of accident causation is the Energy Damage Model. It contends that damage (and the resulting loss) is caused by the exchange of energies, and the only way to change the state of something in this universe is to apply energy. In industry, we use various energies to do our work. We use it to lift heavy things, grind materials, mix liquids, break chemical bonds, dig holes and move things about. The use of energy is an absolute in order to achieve technological leverage, become more efficient and improve profits - after all, profit is the creation of something worth more to others than it cost you to produce.

However, when this energy is misdirected and uncontrolled it can have consequences we did not intend. The Australian Standard for Risk Management (AS4360) defines an incident as “an unwanted consequence”, so by this definition we can conclude that accidents, injuries and damage are caused by uncontrolled energies. This leads our minds in a different direction when considering the causes of accidents.

The technology we use and the methods in which we use it are not perfect. Things can, and do, go wrong. The design of the equipment is not perfect, they are not built or maintained perfectly, or run with perfection by perfect people with perfect systems. The inherent errors (or latent conditions) in the design, construction, operation and maintenance may lay dormant for many years before playing their part in an incident.

Injuries to employees is only one outcome of an energy interacting in a way that we did not intend. I dislike the term accident as society in general believe accidents are unforeseeable and unavoidable - “a freak accident”. Other outcomes may be damage to plant or equipment, loss in production and/or efficiency, or environmental damage, but the mechanisms remain the same. To pretend otherwise is to miss real opportunity to prevent injury, increase reliability, increase production and reduce environmental impacts. The entire system can be improved of a more holistic, systematic approach is taken.

The research of James Reason shows that people make mistakes on a regular basis and it is the consequences of those mistakes that vary. The uncomfortable reality is that we all make mistakes, but most of the time we are unaware of those mistakes because they have little or no consequence, or we are able to recover from them without incident, or they occur in a context that helps to prevent serious consequences.

This is not to say that training is a waste of time. James Reason’s work does acknowledge a slight decrease in the frequency of errors between inexperienced workers and highly experienced workers (the rate shift from around 1 mistake per 10 attempts to 1 in 100), but these failure rates are horrendous in comparison to engineering controls, many of which are in the order to 1 in a million or less.

This is a different accident causation philosophy that recognise people *will* make mistakes, regardless of how much training they undergo, and concentrates prevention activities on designing the equipment, environment, and systems to be tolerant of the human errors that will inevitably occur. Concentrating on the environment and circumstances of an accident is not something that is easily achieved when the model you use forces you to “blame to worker” with little regard to the multitude of factors that contributed to the incident outside the worker’s control.

How can people injure themselves? How can they recognise a mistake has been made? Is it clear what the corrective actions are? How are the controls arranged? Are the labels easy to understand (ideally they should not be required at all)? What signals do we give the operator that the machine acknowledges the commands and is acting appropriately? Are the monitoring instruments measuring the correct things? Is it clear what the potential consequences of the error may be?

To many readers, this may look like behaviour based safety, but it is not. It is taking into consideration the limited cognitive abilities of the human race when designing a system in which they are going to work. No amount of training will improve the cognitive abilities of people, so we must concentrate of making machines, equipment, systems, and interfaces as natural and as easy to follow as possible.

With thoughtful engineering and design we can create an environment in which the operators perform fewer errors, are able to detect any error that occur and respond appropriately. The application of such basic concepts as affordances, natural mapping, and feedback can vastly improve the overall reliability of the system.