The danger of body counts and shoddy data
It’s been almost two months since I posted my 14 Points of Safety on my personal blog. I’ve explored the first two points in detail both there and here. I thought this week I would tackle the my third point, Focus on prevention.
“Preventing injuries is more efficient than reacting to them. Injuries are caused by failures in the system. By managing hazards (procedural, behavioral, and mechanical) organizations can reduce unplanned downtime, injuries, and defects.”—Phil La Duke’s third point of safety.
Prevention must be the underpinning of any safety management system but it is easier said than done. Governments, executives, and many experts actively (intentionally or unintentionally) encourage body counts when it comes to safety. From safety records and paperwork that must be filed to the authorities, to reports to management safety is primarily a reactive function. We repeatedly extol the virtues of a system of predictive data and proactive responses to trends but scare little is done in these areas.
Lagging indicators and quantitative data is easy; companies basically just have to count the boo-boos and Band-Aids. While it’s true that it can be difficult to gather reactive data like near misses, it is infinitely easier than predicting what can go wrong and intervening before it does. Even if the organization can gather the information it needs, prediction and proactive initiatives require more than just data; it requires safety professionals with the skills to identify trends, interpret them, and design appropriate interventions to countermand the trend.
By using a combination of lagging indicators (“what happened”) and leading indicators “what is likely to happen if we do nothing”) the safety professional can craft a dashboard of metrics that can provide real insight into the risk endemic to a process. This kind of data is easy to misinterpret, and what’s worse, a safety professional can find him or herself inappropriately using this analysis to support something he or she already wants to do.
Prediction can be as simple as basing safety initiatives on historical data (we know that the kinds of conditions that hurt workers, we know where they are, and when they are most acute) or as complex as conducting standard progressions (or logarithmic progression) on both quantitative and qualitative data (to establish baseline data that can be compared to expected outcomes.
Many organizations fail to be proactive, not because they lack data, but because their processes are not in control (a statistical term that means that the data cannot be trusted to be reliable).
Remember Statistical Process Control (SPC)?
There was a decade and a fortune wasted chasing SPC only to discover that most organization’s processes were so rife with process variability that no valid inferences could ever be made. The process ended up costing far more than it could ever recoup.
Similarly, the difference between an estimate or prediction and a SWAG (Silly, Wild-Assed Guess) is process variability. Variability is why Behavior Based Safety (BBS) fails and ultimately will be the reason that companies dump 6 Sigma efforts. These processes are not intrinsically flawed — in fact if applied to a stable process that is in control and has minimal variation they can achieve remarkable and powerful results — but most purveyors of said systems are extremely unlikely to advertise (if they even understand) the inappropriateness of applying these tools to systems that are not in control.
So before a proactive safety management system can be implemented the organization’s processes must be in control, and before a system can be under control it has to be standardized, documented, and the workers must be sufficiently trained to compete the tasks with minimal process (both mechanical and behavioral) variation.
My advice is to stick to the simple methods of prediction — layered process audits, workplace inspections, 5S initiatives, and Total Productive Maintenance (TPM) efforts.
Not only will such efforts be easier and less costly to implement, they are far more likely to succeed in improvements not only to safety, but to quality, delivery, cost reduction, and the expected useful life of equipment.
The best indication that risk has been minimized is to audit (and by audit I mean a simple observation of how things are compared to how they are supposed to be) the processes. If the manpower, machines, materials, methods (behaviors), and environment are all performing as intended and expected, then the likelihood that there will be a process failure is fairly small. But if even one of these elements is out of process the risk of injury can be significant. If more than one of these elements is out of process the risk of injury rises expotentially.
Perhaps the best way to focus on prevention is to identify the elements in your workplace that are most likely to cause (directly or indirectly) a catastrophic outcome. No one from outside your organization can responsibly identify these elements for you without first researching your work environment. The place to start is a complete and thorough site analysis that will identify the greatest areas of risk and biggest opportunity for success.
Preventing injuries can be costly, but it is far less costly than treating injured workers. Prevention is also a tough sell to operations leadership because the cost of prevention is tangible and real while the cost of injuries that may never happen is intangible and a matter of conjecture. Many operations leaders still believe that the risk is worth taking.
This situation is exacerbated by shoddy data, poorly defined prevention procedures and overly cautious safety professionals who cannot differentiate between a hazard that is highly probable and one that is all but incapable of causing harm or one that is likely to cause a fatality from one that is likely to cause a first aid case. Prevention can only be truly credible when backed by sound analysis of statistically significant data.
