You live in the real world. You have a business to run, or maybe a department or shift. You want your employees to be safe. You have systems with safe practices and procedures. You are developing an organizational culture that places a high value on safety. But still there are occasional incidents.
How can you provide further direction and leadership for continual improvement in safety when realistically time and money are tight?
One way is to start by asking:
- Are the current controls for safety being operated efficiently and effectively to achieve their full potential?
And only then ask:
- What other factors should also be controlled or improved?
Considering each of these in turn:
Safe working practices and procedures provide a considered approach to what is required. They are generally based on regulations and good engineering practice. But they need to be followed.
Long, complicated procedures may give an illusion of completeness, but anyone trying to comply is flooded with details. Pragmatically, we need to sort what is vital for safety from what is necessary, but has less impact. Treating everything as contributing equally to risk control, as shown in Fig 1, suggests a dangerous, unthinking “tick-box” mentality. Complying with “most” of a procedure does not mean that the risk has been reduced proportionately.
Go back to your safety practices and procedures. Conduct a risk analysis on the contribution of various factors being controlled. Then focus on the conditions you really need to control, in contrast to those that only need to be monitored, perhaps for compliance with regulations (see Fig 2).
Consider, for example, operation of forklift trucks. Driving fast or driving with forks raised are likely to be more serious conditions than, say, leaving the key in an unattended vehicle or failing to conduct refresher training. Of course it depends on the particular circumstances. How likely is it that an unauthorised person might jump on a forklift and operate it? Refresher training may be appropriate for those who operate the equipment infrequently, but a competent operator who regularly operates a forklift will gain little from refresher training. They can easily demonstrate the correct procedure when being “refreshed”, what matters is what they do when not being observed closely.
Ensuring good control of the critical procedural factors establishes a baseline for safe operation. The second step recognises that your safety practices or procedures are unlikely to have considered every condition that could lead to an injury.
Unsafe conditions arising from uncontrolled factors can arise from two types of variation:
- causes that can be assigned to something special happening, and
- random causes of variation that are minor in nature and represent day-to-day increases and decreases in risk - because conditions always vary to some extent.
Data from incidents (including near misses) represent what actually happened. An example could be a forklift hitting and distorting warehouse racking. From time to time, incident investigation may reveal the need for a new control measure in the safety practice or procedure, such as routine checks on racking. However, this should not be a knee-jerk reaction. Corrective actions may be more appropriate elsewhere in the system. Improved access, or strategically placed barriers or corner protectors to the rack uprights may be better solutions.
Another accurate source of data comes when serious conditions are reported. In the forklift example the serious condition could be a problem with the steering or brakes. Serious conditions need to be corrected, but whether or not the condition should be controlled needs to be assessed on a case by case basis. As with incident investigation, depending on the reporting culture and safety at the site, data points may be limited in number.
To better understand the unsafe conditions that may not have been forecast when a procedure was developed, you can get valuable data from inspections and observations, targeting the activities of interest. With forklift operations it could be observing how well loads are placed on forks, whether there is adequate visibility when moving, what manual tasks are required (adjusting forks, lifting empty pallets etc), visible signs of unreported damage to property or product.
Previously uncontrolled conditions identified by investigating incidents and reports of serious conditions, together with data from inspections and observations, can be compared with the controlled factors (see Fig 3). The example shows two conditions that appear exceptional and should be included for control (see Fig 4). Monitoring the effectiveness of control factors provides leading indicators to predict safety performance.
It is generally not productive to work on the small, routine deviations individually. The aim should be to track this type of deviation over time – ideally using control charts. Having removed the special causes individually, further improvement depends on working on the system - for example changing the method of work (if economical to do) - rather than responding to each individual deviation. The Plan-Do-Study-Act process of testing improvements can help. Control charts will also show any unpredicted deviations that need to be addressed separately.
Finally, well chosen leading indicators can allow managers to predict, to provide knowledge as a basis for action to improve safety. But best efforts at prediction and prevention are not enough. The acid test is: are you still having incidents? There is still a place for lagging indicators based on injury and illness statistics – provided they are interpreted to provide knowledge and not used raw as an end in themselves. You may just have been lucky (or unlucky).
Acknowledgement: this article was inspired by, and draws on, the work of Walter Shewhart and Don Wheeler.