Wednesday, December 11, 2013

Safety Indicators – Measuring Something That Didn’t Happen

What makes a company a “safe company”? I suspect that most people have a number of thoughts on this topic. However, if we’re honest with ourselves, there’s one metric for safety that everyone uses, even if only on an instinctual level – a safe company is a company that doesn’t have incidents (or at least any serious incidents) and an unsafe company does have incidents. Its okay to admit it, it’s a natural instinct and you can’t really help it. Our brains tend to look for easy associations. It’s sort of like saying that good basketball players will score more points. That’s pretty easy math (my favorite kind of math, by the way) - safe companies don’t have (serious) incidents. 

The problem with this instinct is that its not always true. Consider the recent events in West, Texas, when a plant caught fire and exploded, killing 14 and devastating the nearby community. Its unlikely that the company that owned the plant is going to win any safety awards this year. However, although we all agree with that somewhat sarcastic assessment, consider this question – was the company that owned the plant a safe company to work for the day before the incident? I think most people would agree that the answer is no. What if we asked the workers or the management of the plant on the day before the incident if they felt that the plant was safe – what do you think they’d say? Bear in mind that they had never experienced a serious incident like this before and the human mind predicts the future using the past, often leading us to overestimate how safe we are. I submit that if you were able to poll the workers the day before the incident, the majority of the workers would say that they felt safe at work.

Now, this isn’t because these workers are stupid. Rather, its because they are human. Like we discussed before, their minds, our minds, everyone’s minds are predicting the future using the past. The problem is that the past is not a reliable indicator of future performance. Certainly, the past is instructive, but as Nobel Laureate Daniel Kahneman notes in his fantastic book Thinking, Fast and Slow:

Success = Talent + Luck

In everything we do the outcomes are determined by what we do (i.e. our “talent”) as well as seemingly random events (i.e. “luck”). So if you look at the outcome of any event you can attribute it to either the skill of the person or to just plain old dumb luck (or both).

To illustrate why this is crucial to safety management, consider a driver who fails to wear a seatbelt, has bad breaks, bald tires, and drives with one eye closed while texting. If this driver were to reach their destination without having an accident would we safe that this person drove safely? Of course not! This is why, as Todd Conklin says, safety is not defined by the absence of incidents or events. Rather safety is about the presence of controls. It’s not what happens to define a safe company, but what they do to manage safety within the organization.

In the safety management world this concept is becoming more and more prevalent – if you want indicators of safety performance you need to look at inputs, or what are more commonly referred to as leading indicators, rather than outputs, or lagging indicators. Typical lagging indicators that organizations use to measure safety performance are:
  • Recordable injury rates
  • Severity rates
  • Experience modification rate 

To illustrate how these indicators work together we can adjust Kahneman’s equation to say that:

Lagging Indicators = Leading Indicators + Luck.

Lagging indicators are useful in helping organizations measure some aspects of safety performance, but they are not completely valid measurements because they are subject to variability (AKA “luck”). You can go a long time without having a serious incident. This means that your lagging indicators can suggest that you are doing all the right things, but disaster can be right around the corner. A classic example is the BP Texas City Refinery explosion in 2005 that killed 15 workers. The refinery’s lagging indicators were exemplary but created a false sense of security in some of the management, hiding the signs of impending danger.

Leading indicators, on the other hand, when well designed, measure the “talent” side of the equation and independent from the variability, or luck, side of the equation. There’s a three step process to designing good leading indicators:
  1. Identify what you want to do
  2. Figure out what it takes to do that
  3. Find a way to measure those things

So, lets use our driving example.
  1. The goal when driving is to not get seriously injured in an accident.
  2. The next step is to figure out what it will take to not get seriously injured in an accident. This includes ensuring the vehicle is well maintained, wearing one’s seatbelt, following the rules of the road, avoiding distractions, etc.
  3. Finally we need to figure out how to measure those things. To measure vehicle maintenance we could look at maintenance records, develop a preventative maintenance schedule and measure how often we follow that schedule. To measure the ability of the driver to follow the rules of the road and to practice safe behaviors we could use safety behavior assessments to evaluate the Safety DNA of the driver. Then, with each driver we can develop a personal safety action plan based on their Safety DNA with specific behavior measurements that can become leading indicators. You could also just implement a policy that only employees with a specific Safety DNA are allowed to operate vehicles and measure compliance with that policy.

The key with leading indicators is to figure out what you want, what it takes to get there, and then ask yourself – if we were doing those things what specific things would we see? In this way leading indicators are specific to the goals and objectives that you want to achieve. They can be used on a micro-level, at the individual worker or job level, or they can used on the macro-level, with indicators for overall organizational safety performance. Examples of some commonly used micro/job-level leading indicators include:
  • Compliance with specific job procedures or a job hazard analysis
  • Employee competence in regards to workplace hazards and procedures, as measured by random sampling
  • Employee use of PPE required for the job
  • Employees following their personal safety development plans

Examples of some commonly used macro/organizational-level leading indicators include:
  • Percent of employees receiving required training
  • Safety audit frequency, results, and time before corrective action is taken
  • Implementation of a safety management system
  • Safety perception surveys
  • Safety behavior assessments performed
  • Use of the higher order controls on the hierarchy of control versus lower order controls

These lists are merely a small sample to get you thinking. Leading indicators though are vital at measuring safety performance for one primary reason – it takes the emphasis off what doesn’t happen and puts it on what does happen to make you and your organization safe. Karl Weick says that safety is a “dynamic non-event.” Most people focus on the “non-event.” But what are you doing to make your organization safe?

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.