The causal relation between lead and lag indicators Johnny Dyreborg Researcher, PhD * National Research Centre for the Working Environment, Lerso Parkalle 105, 2100 Copenhagen, Denmark 1. Introduction An important research issue is the causal relationship between modifiable safety conditions and unwanted outcomes, e.g., injuries to workers. If a proactive safety approach is taken in the monitor- ing of safety, then safety performance indicators, by definition, must include measures of root causes and safety related perfor- mance of the production process. Only in this way can safety per- formance indicators serve as a reliable instrument for monitoring the progress of safety conditions or the health status of the system (Reason, 1997). However, an essential prerequisite is that perfor- mance indicators must be based on practical or scientific evidence about the causal relationship between the indicators used and the unwanted outcomes (Shaw and Blewett, 1998; Shannon et al., 1999; Lund and Aarø, 2004). I maintain that lead and lag indicators must be defined within the framework of such a proactive moni- toring strategy (Rasmussen and Svedung, 2000). The aim of this comment is, therefore, to argue that the distinction between lead and lag indicators is important, even allowing for the fact that the concepts could be elaborated more clearly. I thus disagree with the conclusion in the paper, by Hopkins (2008), with regard to that whether a process safety indicator ‘‘be described a lead or lag is ulti- mately of little consequence”. The implication of this position is that safety management would move away from proactive monitoring and back to a purely reactive approach. In addition it negates the possibility of considering the causal relationships between proac- tive performance measurements (lead indicators) and outcome measurements (lag indicators). This critique will be elaborated in the following. 2. The forgotten causal relationships between lead and lag indicators The most fundamental critique that can be levelled against the paper by Andrew Hopkins is that the causal relationship between lead and lag indicators has not been considered. This relationship is pivotal for a proactive strategy in safety management. Only in cases where this causal relationship is omitted does it make sense to nullify the distinction between lead and lag indicators. On the other hand, if this causal relationship is taken into consideration, some of the confusions in the practical use of lead and lag could be avoided. This might also be the case for the use of these indica- tors in the Baker Report (Baker, 2007) and in the HSE guide on performance safety indicators (2006). 2.1. The lead/lag distinction (mainly the HSE guide) The results of the analysis on the use of lead and lag indicators is presented in Table 2 in Hopkins’ paper (2008). Based on the HSE guide (2006) three types of performance indicators are discussed in relation to lead and lag indicators: measures of routine, safety-re- lated activity (Type A), measures of failures discovered during rou- tine safety activity (Type B) and measures of failures revealed by an unexpected incident (Type C). The importance of these three types of indicators is, in general, agreed upon, but Hopkins’ paper identi- fies two kinds of inconsistencies when these three types are grouped into lead and lag indicators in the HSE document: (a) Measures of routine safety-related activity are not defined. (b) Measures of failures revealed during routine safety activity are defined as lead but exemplified as an outcome lag indi- cator in the HSE document. 2.1.1. Measures of routine safety-related activity are not defined? It is recognised that this measure is a lead indicator. In the HSE guide a leading indicator is defined as follows (Hopkins, 2008): The leading indicator identifies failings or ‘holes’ in vital aspects of the risk control system discovered during routine checks on the operation of a critical activity within the risk control system. From this definition Hopkins concludes that leading indicators are, according to the HSE document, ‘‘Type B”, but he does not in- clude the ‘‘Type A” indicator. However, this is a very restrictive interpretation of the document. The definition speaks about ‘‘fail- ings” and ‘‘activity within the risk control system”. In cases where a critical number of tests have not been performed on schedule we could speak about failings in the activity of the risk control system, and thus as suggested by the HSE document, that ‘‘Type A” indica- tors can also be considered as lead indicators. This is also in line with a proactive monitoring strategy, in which planned tests must be performed on schedule (Type A) and the observed measures evaluated against defined reference values (Type B) in order to pro- vide an adequate control of the unwanted outcomes (Type C). Thus, the measures of routine safety-related activity can be seen to be defined in the HSE document, and they seem to be included in the definition of leading indicators. 0925-7535/$ - see front matter Ó 2008 Published by Elsevier Ltd. doi:10.1016/j.ssci.2008.07.015 * Tel.: +45 39165499; fax: +45 39165201. E-mail address: jdy@nrcwe.dk. Safety Science 47 (2009) 474–475 Contents lists available at ScienceDirect Safety Science journal homepage: www.elsevier.com/locate/ssci