PLAN

I start with a schedule currently being used or being considered. I analyze the schedule to identify factors that may increase the likelihood that the people working it might become fatigued. This can be done through a human factors analysis that compares the schedule to what fatigue science would say about the likelihood of fatigue or by modeling the fatigue and performance using a validated bio-mathematical fatigue modeling system. If you go the modeling route, be sure you understand the limits of the modeling system[1]. The graphs and tables these systems produce are very compelling and if the system is not being used properly, you can end up trusting the results even if they are inaccurate.

Next, I come up with corrective actions that will be applied to the schedule to mitigate the factors that will increase the likelihood of fatigue.

DO

Now I apply the corrective actions to the schedule. These can be small like adjustments to start and finish times, or large like scrapping the whole schedule and starting over. If I am using a bio-mathematical fatigue modeling system, this step is usually iterative. This means I keep “fiddling” with the schedule within the fatigue modeling system until I get a schedule that is significantly better, with respect to fatigue risk, than the original schedule.

The schedule then gets rolled out to the 24/7 workforce.

CHECK

The PLAN step gives me a predicted level of fatigue. The CHECK step gives me an observed level of fatigue. I want to make sure that what I think is going on is actually going on. In other words I want the predicted fatigue to match the observed fatigue. It means I have to find a way to measure the fatigue that the 24/7 workforce is actually experiencing as a result of applying the corrective actions in the DO step. This can be done with surveys, questionnaires, sleep-wake journals, validated fatigue scales or bio-mathematical fatigue modeling.

If you are going to use a fatigue modeling system for this step, there are two ways to collect the data to use in the system. First you can have your workforce record their daily activities in a sleep-wake journal. The other way is to use wearable technology. Both the sleep-wake journal and wearable tech have inherent inaccuracies, but they are still the best methods to come up with your observed fatigue. The advantage of the wearable tech is that it can be automated which makes it easier for your workforce to gather data for you.

ACT

Now is when I formally compare what I predicted would happen as a result of the corrective actions to what actually happened. In this step I am hoping to see that the workforce is experiencing less fatigue. If the improvements are not significant, then I try to improve the corrective actions. This usually means tweaking the schedule system again.

At this stage the ACT and PLAN steps merge into each other and some people argue that they are the same step. I like to keep them separate because in the ACT step I am working on the original risk factors that I identified in the original PLAN step and I am working on improving the original shift schedule. The second time through the PLAN step, I may start working on a totally new schedule or I may identify totally new risk factors for which I have to develop new corrective actions.

Either way, the continuous improvement process does not stop, operational demands change and so do 24/7 work schedules. Fatigue resulting from shift schedules should be in a state of perpetual improvement.

Footnote

[1] For a discussion of issues to consider when selecting a bio-mathematical modeling system, see: Dinges, D. (2004). Critical research issues in development of biomathematical models of fatigue and performance. Aviation, Space, and Environmental Medicine, 75(3) sec II, A181-A191.