“Some risks are plainly acceptable and others are plainly unacceptable.” Supreme Court Justice John Paul Stevens
When Justice Stevens wrote that opinion for the majority in the 1980 OSHA Benzene case, he went on to add that odds of fatality of one in a billion could not be considered significant but that for odds of a fatality of one in a thousand, “a reasonable person might well consider the risk significant.”
In that decision, the Supreme Court left a hole six orders of magnitude wide that our enterprises are still left to narrow down to something we can cope with.
Make no mistake, we must narrow it down.
The difference between significant and insignificant risk
An important part of risk analysis is quantifying risk, determining both the impact severity of a hazardous event and just as importantly, the likelihood of that hazardous event. It takes both to know the risk. Knowing the risk, however, is not much use by itself. Only when we compare the risk against a standard do we know whether the risk is too high or low enough. Some people (Supreme Court Justices, for instance) think of that standard as “significant”. We tend to shy away from that term, since the statisticians have already claimed it for something entirely different. Instead, we refer to risk as acceptable/unacceptable, or better still, tolerable/intolerable. We call the standard our Risk Tolerance Criteria (RTC). Comparing the risk of a hazard against RTC allows us to know if the risk is too high, and if so, by how much we must reduce it for it to be tolerable.
RTC drive the allocation of finite resources. Rational RTC lead to rational allocation of resources, while irrational RTC lead to misallocation of resources and ultimately, to poorer safety.
Most practitioners acknowledge the need for RTC but do not discuss how they are developed. While a few countries dictate RTC, most governments, including that of the United States, leave it to enterprises to establish and justify their own RTC.
When asked what risk of fatality is tolerable, the glib answer, the one given for public consumption by public officials and senior executives, is that no fatality is acceptable. Somehow, they have confused “accepting a fatality” with “tolerating a risk”. To begin with, we don’t accept risks, we tolerate them. “To accept” suggests “to consent” or “to agree to.” “To tolerate” makes it clear that this is something we are putting up with, something we are enduring. We tolerate safety risks; we do not accept them.
Ranting and rhetoric notwithstanding, zero risk is not achievable. To demand zero risk of an enterprise is to set it up for failure. Any risk analyst will state without hesitation that “No such thing as zero risk exists,” and then go on to explain why. Yet the idea persists. In part, it is because of the pernicious repetition of slogans like “Striving for Zero” or “All accidents are preventable.” Going from one fatality per year to one fatality per decade is a great achievement, a 10-fold reduction with at least nine out of ten years experiencing no fatalities, but 0.1 fatalities per year is still not zero. The slogan “Striving for Zero” is only meaningful when zero is recognized as an asymptote to approach (yes, by orders of magnitude), yet never achieve.
Your own risk tolerance criteria
Here is an exercise that can help you understand your own risk tolerance criteria: Imagine a workplace. Any workplace. Perhaps the new facility you’re working so hard to get off the ground right now. As you imagine it, imagine how many people work there. Now imagine that you are describing it to someone you love, someone whose good opinion matters to you. You want them to believe it is a safe workplace. How infrequent would a fatality have to be to consider it safe?
Most people begin the exercise by saying “I have no idea,” but they do, even though they don’t know it. If there was a fatality once a week, would that be safe? If you didn’t answer “no,” I really don’t have anything left to say to you. Now let’s consider the other extreme. Modern man, Homo sapiens, has only been around about 200,000 years. If your workplace had been built and staffed at the time modern humans first walked the plains of Africa and there had been one fatality in that time, would the workplace be safe? I would be astonished if you didn’t say “yes.” See, you do have an idea.
Can we narrow that further? Is once a year safe? Most would agree that it is not. Is once every 100,000 years safe? Most would agree that it is. Now you’ve narrowed it to just five orders of magnitude, which is better than the Supreme Court did. Is once every 10,000 years safe? For comparison, consider that agriculture was first developed in the Mesopotamia about 12,000 years ago. Most would still agree that a facility with a fatality rate of once every 10,000 years is safe.
Can we narrow it even further? Once every 1,000 years? Ghengis Khan began roaming the Mongolian steppes about 800 years ago. Is once every 10 years safe? Once every 100 years? This is usually the point for most people where the answer stops being, in the words of Justice Stevens, “plainly acceptable and … plainly unacceptable.” The most common answer we hear is that a workplace with a fatality no more than once every 100 years is a safe workplace. Why? When pressed, most people will reply that they expect their career to last about 50 years, so once every 100 years leaves them with a 50:50 chance of a fatality not happening on their watch.
If not zero, then what?
How does once every 100 years stack up? It depends on how many people work there. If there are 100 full-time equivalent employees (FTEs) at the facility and there is one fatality per hundred years, that works out to 10 fatalities per 100,000 FTEs. The Bureau of Labor Statistics publishes fatality rates every year. In 2017, the most recent year for reporting, there were 5,147 work-related fatalities, at a fatality rate of 3.5 fatalities per 100,000 FTEs. Our hypothetical workplace is three times more dangerous than an average job – both public and private sectors – in the United States.
Some jobs are much more dangerous. Commercial fishermen were at 99.8 in 2017 and loggers were at 84.3. Some jobs are safer. Mathematicians were at 0.2 and librarians were at 0.4 in 2017. Over the years, the chemical manufacturing industry has been at 2 fatalities per 100,000 FTEs, while the refining industry has been closer to 10 fatalities per 100,000 FTEs.
What is the risk tolerance criteria for your enterprise?
It is important to know the risk of any hazard in your enterprise before you decide that you need to reduce that risk. It is just as important to know what your RTC is, so you know by how much you need to reduce the risk. When you don’t know both, it is likely that you will misallocate resources, squandering them on hazards that have already received sufficient attention and ignoring other hazards that need attention. Neither is good for safety.
Process safety is a journey. Your risk assessments will tell you where you are; your RTC will tell you where you want to get to, or at least the next stop on the journey. No one can decide for you where your next stop is. OSHA won’t and consultants shouldn’t. It’s your enterprise. Your RTC cannot be zero, but you can decide what it will be. Do you intend to be as safe as our most dangerous jobs – logging and commercial fishing – or do you intend to be as safe as a librarian? Do you see the chemical industry as the benchmark, or the refining industry? You must decide where you intend to be. Then you can plot the path to get there.
This blog is based on an earlier version, “Risk Tolerance: How Low Do You Go?”, posted on 26-Feb-2016 by Elsevier in Chemicals & Materials Now!