The Safety Pyramid is a concept that safety professionals have long used when analyzing data and making safety decisions. Originally known as Heinrich’s Accident Triangle or Heinrich’s Triangle, the syntagm illustrates the proportionality between the number of near misses, minor injuries, major injuries, and fatalities in a workplace.
This accident causation model was developed in the 1930s by H.W. Heinrich and later expanded by Frank Bird in the 1970s. The Triangle (and later Pyramid) has a broad base (not to scale) that represents a large number of near misses and minor events, which are represented as the root cause of the higher-level events (major injuries and fatalities).
The image implies a direct correlation between the severe incidents and the larger number of minor safety events. Heinrich put a specific figure to it, positing a ratio of 300 near misses to 29 minor injuries to 1 major injury, as shown in the illustration below.
Source: H. Heinrich, Accident Prevention: A Scientific Approach (2nd Ed)
The idea behind the Pyramid is that addressing the lower-level events will automatically result in a reduction of higher-level events. If there is one major injury for every 300 near misses, then it stands to reason that eliminating those 300 near misses will prevent that major injury.
But will it?
In this article, we’ll answer two separate but related questions. Is there actually a correlation between minor incidents and more severe ones? And will reducing the number of events at the base of the Pyramid really prevent the ones at the top?
In other words, is the Safety Pyramid still a useful concept for occupational health and safety?
Is the Safety Pyramid a Predictive Model?
While the Safety Pyramid has been used extensively as a predictive model, its validity has been contested for almost as long as it existed.
Several studies have shown that the Pyramid is over-simplistic and is not applicable to complex real-world work environments.
Historical trends also show that, while minor injuries have declined over time, fatalities and serious injuries have not decreased at the same rate. So the base of the Pyramid has been shrinking, but it hasn’t resulted in the same kind of reduction at the top.
This suggests that Heinrich’s (and Bird’s) conclusions were flawed. Evidently, there are other factors that contribute to accident causation – it’s not all proportionality. For this reason alone, we can’t consider the Safety Pyramid a predictive model.
Flaws with the Safety Pyramid
So, the Safety Pyramid’s ratio doesn’t hold up. But that’s not the only issue with it.
Here are a few more reasons why the Safety Pyramid is considered a flawed tool.
Lack of Empirical Evidence
There is simply no data that indicates that the Safety Pyramid is consistent or mathematically correct.
Heinrich’s original data has not been uncovered and it appears that his conclusions were based on circumstantial or anecdotal evidence. Bird’s data is available, but it is based strictly on reports provided by the organizations, not independent field studies.
Efforts to replicate the data behind the Safety Pyramid resulted in different proportions. It’s likely that consistent proportionality does not exist when comparing different organizations and industries.
Lack of Correlation Between Types of Incidents
While some studies claim that the number of severe injuries can be an indication of future fatalities, the reality is that the type of incident matters as well.
Suppose, for example, that there is a workplace where multiple workers have been injured as a result of falling from a ladder. Then, a fatality occurs at that same workplace when a worker enters a confined space with a hazardous atmosphere. While we can create a ratio out of these events (say, for every 10 injuries, there was 1 fatality), it’s difficult to see how those ladder-related incidents have any direct influence on what happened in that confined space.
Eric Marsden uses a very simple analogy to illustrate this point. There are plenty of worksites where people accidentally spill coffee on themselves and get burned, but “People do not die because coffee drinkers are walking around with cups without lids.”
There is no causal link between coffee-related burns (no matter how frequent they are) and fatalities. Likewise, we shouldn’t expect other types of minor incidents to be predictors of more severe or fatal injuries.
Focus on Lagging Indicators
The Safety Pyramid is concerned with lagging indicators of safety. It looks at past events (near misses and injuries) as a gauge of the organization’s safety performance, but past events provide very limited insights into safety and how to make the workplace safer.
Emphasizing lagging indicators puts safety in the position of a spectator, watching the game to see the results but not participating in it to change the score. It’s a strategy that is out of step with modern management approaches, which focus on leading indicators of safety.
Blaming the Victim
The Safety Pyramid suggests that the root cause of every incident is an individual unsafe act or condition, and that these acts and conditions can be traced back to the victim of the incident.
This isn’t unique to the Safety Pyramid or Heinrich’s attitudes to workplace safety more generally. It’s consistent with the prevalent management theories of the time, especially Theory X and the Great Man theory. These theories posit that workers lack the motivation or ability to do the right thing (such as working safely), so they need to be closely monitored and directed.
Given this context, it’s not hard to see how Heinrich’s model is prone to blaming the victim and exonerating management. But it leaves us with a theory of accident causation that ignores systemic causes and focuses on lower-level hazard controls (like training and PPE) rather than addressing issues holistically and implementing higher-level controls (like eliminating the hazard entirely).
Superficial Improvements to Safety Performance
The theory behind the Safety Pyramid postulates that eliminating the lower level of the pyramid will result in an overall reduction of serious injuries. While there’s nothing wrong with implementing control measures to prevent minor incidents, in practice this focus on reducing the number of near misses and non-severe injuries can send the message that management simply wants to see the numbers go down. Similar to zero-incident safety programs, this can incentivize workers and supervisors to avoid reporting anything other than a serious injury.
On paper, it looks great – fewer near misses and fewer injuries. In reality, the workplace is just as risky as it was before. It also robs the organization of the opportunity to learn from near misses and make genuine improvements to their safety performance as a result.
Download our free whitepaper, Safety Maturity Curve: A Journey from Lagging Indicators to Leading Indicators!
Is There Any Value to the Safety Pyramid?
Despite all these problems, the Safety Pyramid and the model behind it are still used by many safety practitioners.
Its lasting appeal is largely due to the simplicity of the graphic. One look at the image and we immediately grasp its meaning. It’s the same reason we keep seeing the Swiss Cheese model of accident causation – it may be factually inaccurate and incapable of making predictions about safety outcomes, but it’s simple to understand and easy to visualize.
It’s an easy way to illustrate the need to be proactive about all types of unplanned events, no matter how small they are. Even near misses should be reported, and the Safety Pyramid makes a convincing case for that – even if it reaches that conclusion from inaccurate premises.
In the right context, it can be a powerful learning tool for raising awareness about the significance of lower-level events and why they need to be addressed. The caveat is that the organization has to be genuinely focused on addressing these issues, so that the concerns raised by employees are used to improve their safety, not to assign blame and dole out punishment. As Todd Conklin said in his book Pre-Accident Investigations, you can punish or you can learn, but you can’t do both. So the model, flawed as it is, can be useful so long as it is used in a learning organization.
We are all prone to errors and, as such, unsafe acts, near misses, and accidents are to be expected. According to Mike Sondalini, managing director of Lifetime Reliability Solutions, “The best performance possible in well managed workplaces using normal quality management methods are failure rates of 5 to 10 in every hundred opportunities.”
And although Heinrich’s proportionality (or any other universal proportionality) does not exist, there are many opportunities to learn from non-accidents – opportunities we miss when we limit ourselves to learning only from serious accidents.
It’s also true that we should expect lower-level events to outnumber higher-level events. With transparent reporting, injury rates are always lower than fatality rates. Deviations from this norm can indicate under-reporting or manipulation of statistical data, which would tell us where we need to focus our attention. The example below, from an analysis of the effects of zero injuries policies I made years ago, shows such a skewed proportionality in the mining, quarrying and oil and gas extraction industries.
Source: BLS, 2018
So, if your Safety Pyramid is reversed (with the base up) and the number of severe events surpasses the less severe ones, this indicates that you’re missing opportunities for improvement and need to improve your ability to collect data. In many cases, this flipped Pyramid is a sign of a culture of under-reporting where employees try to avoid being the bearers of bad news by not reporting anything they can sweep under the rug (like unsafe acts and near misses). There is a reason for this, but you have to look beyond the employees to find it. The root cause is usually trust in management, and blaming employees for not reporting incidents is only going to solidify that mistrust.
Conclusion
The Safety Pyramid has a convoluted history and safety practitioners continue to disagree about its utility.
As a predictive tool and a model of accident causation, it is heavily flawed and likely based on anecdotal evidence. If the Safety Pyramid causes an employer to press their employees to reduce the number of reported events at all levels, this is more likely to suppress reporting than improve safety.
But when used for illustrative purposes, it’s a simple concept that is readily grasped by employees. There is a certain intuitive logic to it and it can act as a motivator for reporting and addressing seemingly minor issues. In a learning organization where employees trust that management is focused on improving safety (not fudging the numbers), it can be the game changer that leads to improved safety performance.
Ready to learn more? Check out our free webinar, Safety As We Know It… Is Killing Your Business!
Sign up to the Safeopedia Newsletter for more great safety info delivered right to your inbox!