Questioning risk/benefit analyses for fire officers
In the fire service, we need to explain how risk is measured, limited and controlled
"The buffalo isn't as dangerous as everyone makes him out to be. Statistics prove that in the United States, more Americans are killed in automobile accidents than are killed by buffalo." Art Buchwald
I constantly hear the refrain, "We need to do better risk/benefit analyses." That statement has to be true, but what continues to concern me is that while we encourage people to conduct "better" risk/benefit analyses we don't ever tell them how. There is a gap between what we require and what we teach.
In the past I wrote about risk analysis — "Risk Assessments: Avoid the Rock at the Bottom of the Hill." In that article, I equate risk to sliding down a hill with a jagged rock at the bottom. However, it occurs to me that not only is my metaphor weak, it is not effective. The notion of a risk/benefit analysis is something that I continue to struggle with.
This two-part article explores some of the difficulties I have with framing risk, conducting risk/benefit analyses, and my interim solution for handling my own uncertainty. My disclaimer is that these articles cannot hope to provide definitive answers but rather they seek to explore the underlying framework of risk/benefit analyses by deconstructing the underlying notions of risk and benefit as applied to fighting fires.
Think about how you were taught about risk as a fledgling firefighter. It's likely someone said, "Risk a lot to save a lot and risk a little to save a little." Another common paradigm is offered by the IAFC:
- Extend LIMITED Risk to Protect SAVABLE Property.
- Extend Very CALCULATED Risk to Protect SAVABLE Lives.1
To follow the IAFC paradigm, you would expose yourself to "limited risk" to protect "savable property" and in order to do that you would have to know what risk is, know what limited risk is, and know what savable property looks like.
If we accept the definition of risk as the likelihood (or chance) of experiencing an adverse outcome (i.e., the "risk/chance" of getting hit by a bus as I cross the street), we are framing risk as a probability.
If risk is a probability, then it can be quantified because you cannot have a probability without quantifiable parameters. A common mathematical tool for quantifying risk is: (risk = frequency x severity)2 or (risk = Impact of Risk event x Probability of Occurrence.) Using these formulas, the risk of getting hit by a bus is defined by the likelihood of getting hit times the severity of getting hit.
To construct risk as a probability really means that risk is predictable. For example, we can know the probability of a coin landing on heads when we toss it in the air. But the only way we can know that probability is because a coin behaves in predictable ways, e.g. it obeys gravity, it falls, and one side is heads while the other is tails.
Arguably the world can be a random place, where we toss the coin 50 times and it lands on heads 50 times. The probability of it being heads after any given toss has not and does not change.
But if the laws of probability drive risk analysis and by analogy the possibility of landing on heads is the equivalent of a firefighter getting seriously hurt in a fire, the analogy fails. Probability is not useful for firefighting because the probability of getting hurt is not that simple. A firefighter does not stand in front of a burning house and declare that they have 50 percent chance of rescuing Mrs. Smith and 50 percent chance of getting life altering burns.
For firefighters, the probability of getting hurt is almost impossible to determine; the possibility is ever present.
In order to assign probability to risk, we have to have some idea of the general range of possible outcomes. We have to have some idea that when we toss a coin it only has two choices of how it lands. Coin tossing is a simple system &mdash the inputs are limited and the outputs are limited.
A fireground is a complicated system with multiple engines, ladder trucks, rescue companies, chief officers, civilians, and various construction types, fuel loads, fuel geometries, and weather conditions (just to name a few things) all interacting at once on a unique situation. So much of the fireground is unpredictable that — unlike the coin toss — it is nearly impossible to compute probabilities. If we cannot compute probabilities, we cannot compute risk.
What the IAFC paradigm asks is that we be able to engage in classical decision-making while on the fireground. Firegrounds are not the domain of classical decision-making and the risk/benefit analysis as currently constructed is a derivative of classical decision-making.
Remember also that risk is not static. When a firefighter enters a house that is on fire, he/she is exposed to a certain amount of danger.3 Once they start flowing water that danger changes, as people begin to ventilate that danger changes, once the fire is out the exposure to danger changes.
Yes, we should continue to encourage firefighters to compare the relative dangers of their chosen course of action to the anticipated and possible outcomes of that course of action. But we should also explain the limits of our understanding of risk and the limits of our ability to make rational decisions in split-second life and death situations.
We should also stop encouraging risk/benefit analyses when there really is no parallel between the notion of risk and the thought processes that occur on a fireground. Finally, before we say things like, "a basic level of risk is recognized and accepted, in a measured and controlled manner…"1 let's explain how risk is measured, limited and controlled, or else these are empty words.
In the next segment of this discussion, we will explore the notion of "benefits" as related to risk/benefit analyses and consider an alternative framework for enhancing fireground decision-making.
2. The equation I use here is a gross over simplification of mathematical approaches to risk as it does not consider weighting nor does it explain consider that risk is not a simple product of the two factors. It is used for illustrative purposes.
- Incident Command