This study from Lingard and Hallowell et al. is very cool. They analysed safety data on a large construction project over 5 years.
First it gives a good summary of issues around defining lag vs lead indicators. Some criticism has been levied at injuries and incidents due to their low statistical probability of occurring they may be neither valid nor stable when measured in a single company. They note that even a stable safety system will produce a variable number of injuries and incidents, and the absence of incidents doesn’t necessary indicate that one workplace is safer than another.
Other things were highlighted – such as incidents being retrospective indicators capturing the absence of safety rather than its presence and thus “cannot be regarded as a direct measure of the level of safety in a work system” (p207).
Finally, even the definition of leading vs lagging is problematic.
Results:
Found was a complex and cyclical relationship between some of the indicators over time. Leading indicators don’t necessarily lead and at times lagging indicators lead.
E.g. an increase in toolbox talks decreases TRIFR short-term but long-term the “direction of causality … changes direction such that a decrease in the TRIFR causes a subsequent decrease in the frequency of toolbox meetings” p217. They further note “the supposed leading indicators do not simply drive safety outcomes (incidents) but are also driven by changes in the frequency of incidents” (pp212-213).
Companies respond to incidents by focusing on safety management, but as incident rates fall so too does the focus on safety. The knee-jerk response is said to not lead to sustained changes in performance.
Further from the paper: “The fact that these bi-directional relationships were confirmed using the Granger Wald Causality tests indicates that it is inappropriate to assume that changes in the frequency of management actions produce subsequent reductions in incident/injury frequency rates. That is, it is just as valid to state that an increase in incident/injury frequency rate causes an increase in the frequency of safety management activity. Therefore, the simple, one directional relationship implied by the leading/lagging terminology is not supported by the data” (p.217)
It’s argued that understanding the time lags between how leading and lagging change directions is important because the “time lag between changed safety performance and the occurrence of incidents/injuries can mask an organization’s ‘drift’ to danger and provide an overly optimistic sense of safety” (p215).
Other problems with measuring management activities in indicators were also discussed. Drawing on Hopkins’ work, they argue that measuring the frequency of management activity as if it was a proxy for safe performance can be flawed because the use of indicators “could potentially produce behaviours that are designed to manage the indicator rather than the real issue of workers’ safety” (p217).
Authors state incident metrics typically measure unsafety rather than safety. And “assumptions underpinning the use of leading indicators should be reconsidered … the findings challenge the assumption that leading indicators measured at one point in time can predict safety outcomes at a subsequent point in time” p.206.
Finally, this analysis revealed that rather than preceding changes in incident/injury rates, changes in the number of management activities both caused and were caused by changes to the incident data itself. They argue that given these complex and cyclic relationships (more complex than leading/lagging terminology suggests), it’s preferable to consider the measures of management activity more as proactive indicators rather than lagging as “these management activities behave as much as lagging as they do leading indicators in relation to the frequency of incidents/injuries” (p218).
Authors: Helen Lingard, Matthew Hallowell, Rico Salas, Payam Pirzadeh, 2017, Safety Science.
Study link: https://doi.org/10.1016/j.ssci.2016.08.020
Link to the LinkedIn article: https://www.linkedin.com/pulse/leading-lagging-temporal-analysis-safety-indicators-large-hutchinson/?published=t
SafetyInsights.org 