By Phil La Duke
Somewhere in the 500,000 or so words I’ve committed to print you will find me quoting Deming about the dangers of “zero-injury” goals. I have been known to acknowledge the dubious value of the promotion of the pursuit of zero injuries—and I have even been an advocate of doing away with zero injury goals all together. This week I had a conversation that changed my view of this topic considerably. Let me begin by saying that I haven’t reversed my position completely. For example I still believe:
- The widespread confusion between the words “target” and “goal” creates serious and dangerous confusion in safety.
- Telling workers that anything short of perfection is unacceptable is demoralizing and counter productive.
- When “zero-injuries” becomes a slogan and a platitude it diminishes the perceived value of safety professionals.
So while I still believe organizations have to be very careful not to create a climate of fear that promotes under-reporting I now have a deeper insight into why zero-injury goals are important.
Traditional Ways of Measuring Safety are Deeply Flawed
Safety cannot be measured in terms of the absence of mangled bodies. Yet, most of our injury statistics, the things that we gauge our performances against, are based on how many people we hurt (or didn’t hurt). Let’s take something as simple as recordable incidents. If we had 15 last year and we only had 10 this year we saw a 33% improvement, right? Not necessarily, in fact, maybe not in any likelihood. Each injury is the result of a probability—there is some probability that workers will get hurt, in some cases workers beat the odds and other times they don’t, and when they don’t they get hurt. Our traditional view of safety reporting is only accurate if the probability of injuries, the risk if you prefer, remains constant.
Going back to our example, if we had 15 injuries but had the opportunity of having 100 injuries our score, if you will, would be 15%. So in 15% of the cases workers didn’t beat the odds and got hurt. If in the second year, we had 10 injuries but had only 15 opportunities for being harmed our score would be 75%, far worse than the year before where our workers beat the odds 85% of the time versus the 25% of the time in the year that had less injuries. In this scenario, the traditional measures of safety would lead us to believe we are improving when in fact the safety of our workforce is deteriorating at an alarming rate. This false sense that we are improving often will continue until someone is killed, leaving the organization scratching its head as they puzzle over what went wrong when it was doing so good. Of course the formula that turns raw incident figures into rates is an attempt to normalize the data, but it doesn’t work, not really anyway.
As blasphemous as it sounds, injuries tell us very little about the safety of our workplace. We want zero injuries, period; not only is this a good goal, it’s the only acceptable goal.
It’s Not The Goal That Has To Change, It’s the Means We Use To Try and Get There
I’ve also written that if zero-injuries isn’t our goal, than what is? If we say that zero-injuries isn’t our goal then how many people are we allowed to kill and maim and still feel good about ourselves? While it is ridiculous to think that any company would openly and knowingly be satisfied killing any workers, this shift in thinking is relatively recent. In the not so distant past, worker injuries and deaths were built into construction budgets, so while it may seem unconscionable that organizations would take such a blasé view of the death of someone’s loved one that is a recent shift away from the way society viewed worker safety.
We need to go further, and shift away from body counts to the relentless reduction of risk. We need to measure safety in terms of risk levels; statistically, if we reduce the probability that people will get hurt we will reduce injury numbers. If we take a cue from the Quality function, we will see that it wasn’t terribly successful until it took the focus off defects and instead focused on process capability. Once the Quality function started focusing on reducing process variability (instead of focusing on worker behaviors) the quality of products rose exponentially and the age of continuous improvement (leave it to quality professionals to perpetuate a grammatical error to the point where it becomes part of the universal lexicon) was born. You don’t see signs that read “4054 hours without a defect” because nobody cares about that, they care that the product they purchased is defect free. In a similar way, we need to move away from incidents as the primary way in which to measure safety. If there are any injuries than the workplace isn’t safe. We need to recognize that until we deal with safety in terms of risk we will always be fighting a losing battle and that zero injuries can only come through the reduction of risk and by lowering the probability of injury to the lowest possible, practicable, and practical levels that we will ever make the kind of progress that the Quality function has made.