Like some of you, I attended the AIHce in Montreal this year. Most of my time was spent at the Bowen EHS booth inside the Expo, but one of our instructors, Ernie, and I decided to stay an extra day to participate in a free Professional Development Course about Bayesian Statistics Decision Analysis. It was an excellent session taught with great energy and passion by Dr. John Mulhausen and Dr. Perry Logan.
The course was chock-a-block full of fantastic information. One point, however, was particularly important: the need to define the acceptable level of exposure to airborne contaminants in the workplace and to have executive and client "buy in". It is critical for practicing health and safety professionals to know what is acceptable regarding exposures, yet few of us have a solid working definition that we consistently use to make our determinations. Such a definition should be a cornerstone in the foundation of our profession. Without this foundation, we are little better than "dart throwing monkeys" when determining whether workers are over-exposed. This is a phrase favored by Dr. Logan as he tossed a stuffed monkey into the room at the AIHA Exposure Assessment Strategies Committee (EASC) meeting.
Dr. Mulhausen and Dr. Logan both strongly suggested that each of us needs to have a conversation with our management and/or clients about what they are willing to accept in terms of exposures in the workplace. Here is a hypothetical example of such a conversation.
Industrial Hygienist (IH): Hey, PM, I am going to be taking a look at our employee's exposure to BETS (bromo-ethyl toxic stuff) over the next few weeks to make sure they're protected.
Plant Manager (PM): Great! We definitely don't want any overexposures, because you know the trouble that can cause...illnesses, fines, downtime, delays, etc., and we don't need any of that!
IH: True, but you know from our process quality control training that it isn't practical to be 100% sure about anything. However, we can control the exposures so that we're as sure as we can feasibly be: 95% confident that we're preventing BETS overexposures 95% of the time for all workers, on all shifts, under all exposure scenarios.
PM: Hmmm, that's only 2-sigma...can't we achieve 6-sigma on this item?
IH: Sure, but it would require a redesign of the entire process and total enclosure. How much money do you have?
PM: OK, your first plan sounds good...do it and let me know if we need to do anything else. How much will this cost and how long will it take?
This fictional account of a discussion with a plant manager is an example of how we might approach management prior to sampling. It is fairly common for a manager to indicate that he or she wants to be absolutely sure that not one of their employees ever exceeds the established occupational exposure limit (OEL). While this is a noble goal, it is also unfortunately one that is cost prohibitive and difficult to attain. Statistical models based on random exposure samples indicate that this is an extremely tough standard to meet.
A more achievable goal being promoted by the AIHA EASC is to be reasonably sure that 95% of exposures are below the OEL. The AIHA EASC defines "reasonably sure" to be the 95% confidence level of the 95th percentile. Now it can be challenging for many of us to wrap our brains around the idea of the 95% confidence of the 95th percentile. However, it is worth taking the time to fully grasp this concept. I guess you could say it's time to stop monkeying around.
We must ALWAYS use statistical tools to evaluate any measurement data collected. We use randomly sampled data to estimate an exposure distribution. This data can yield an estimate of the 95th percentile, but it is only an estimate. We want to be 95% confident that our true estimate of the 95th percentile is at the estimated value or below.
I agree with Dr. Mulhausen and Dr. Logan and the AIHA EASC that a reasonable level of confidence that we are limiting over exposures to less than 5% with a 95% confidence level. It's imperative to discuss this with our colleagues in management as well as our clients prior to any sampling. We need their buy-in ahead of time. If we do this, and our statistical tools find a reasonable probability that more than 5% of employee exposures exceed the OEL, there is a better chance that management will fully support implementing the hierarchy of controls.