Should safety be measured?

posted Dec 17, 2014, 1:23 PM by Niels Jensen   [ updated Feb 13, 2017, 10:32 AM ]
Today Safepark attended a seminar with Erik Hollnagel and Sidney Dekker with the title "Measuring Safety". Erik Hollnagel, who is currently at Syddansk University, opened the show with a presentation titled "Should safety be measured?" about the difference between safety I and safety II. Safety I is what we have been doing for the past 80 years, that is measuring the absence of safety. This is done by counting things, which are easy to count, such as number of fatalities, number of near miss event (a bit more difficult), number of first aids and other such numbers. However, the events behind these numbers clearly indicate the absence of safety. If safety existed, then no one would die or get hurt. Safety II focus on understanding the work being done. This is much more in line with how performance is measured and improved in professional sports. You measure what is happening, and you try to become the best. To me this is much in line with the ideas of Jens Rasmussen, who used to work at RISØ in Denmark, on analyzing work.

Sidney Dekker started by describing the Australian problem, that many top level people spend almost a quarter of their workweek on compliance with regulations. Clearly that leaves less time for making things safer. Much along the same line is was reported earlier this week, that the average hospital department had to deal with thousands of regulations during their daily work. These regulations take time away from focusing on patients and their needs. So too many rules is not just an Australian problem.

Can safety be measured by counting incidents? The opinion of world-leading safety experts Erik Hollnagel and Sidney Dekker were very clear at today's seminar: No! In order to improve safety you need to focus on the work being done and how it is being done. Erik Hollnagel and Sidney Dekker argued that our well-known and widely implemented measurements tend to be misleading and thus an unfortunate waste of the limited resources dedicated to safety. We tend to measure failures rather than successes, and we focus on what we believe to be precursors of accidents, but rarely are. Are there better and more useful ways of measuring safety?

Erik and Sidney does not say, that we should stop doing PHA, HAZOP, FTA and ETA and all that. They just say that in order to improved safety we need to put more focus on the outcome of the work actually being done. To me, that seem to have parallels in the Japanese approach to quality after WWII guided by Deming and Conway. What do you thing?