This looked at whether safety climate (SC) surveys can assess the risk of major accidents at the plant level.
Using data from the Norwegian offshore oil and gas operations they tested whether three major close call incidents (occurring between 2001 and 2013) could have been detected using the biannual SC data from the period before the incidents.
They specifically looked at:
1) whether the offshore installations deviated from the industry average.
2) if the trends from the installations’ SC data revealed that something was wrong prior to the incidents.
I really can’t do this justice so if you can get the full paper then I highly recommend it.
[Note: One of this study’s authors also explored in another paper whether safety culture data has predictive value for major accidents, which I’ll soon cover at another date.]
Before moving on to findings, the authors talk about the challenges of using SC for informing major accident risk because of the rarity of major accidents. However, some prior research has found a correlation between major accident precursors (hydrocarbon leaks) and SC [another study I’ll cover in the near future].
The authors then give a nice overview between SC and safety culture – noting their difference but how they’re also, unfortunately, often used interchangeably. SC has been considered as a snapshot of safety culture at a given point in time and at a coarser level than the overall culture. It’s said that the two concepts have their origins in different traditions, with climate being from a psychometric tradition and culture from a sociological and anthropological one.
They also discuss criticism of some misdirected attempts of using survey data to inform interventions to improve safety, where it’s quoted that “Aggregated numbers, like frequencies or means, do not offer
much insight into organizational culture, much less an understanding of it” (p4).
Note that I had to skip over a lot of useful contextual information and background on SC, culture and the incidents involved in the study period due to time.
Results:
In answering whether SC can be used as an indicator for major accident risk at a plant level – the results are inconclusive.
At plants A and C the managers “would not have received any signals of danger from the safety climate assessments. In fact, installation A could have concluded that they were actually in a positive development of the prioritization of safety” (p18). Industry average scores versus plant A was not observed. Interestingly despite the SC findings not providing observable warning signs, the incident investigation for the major close call incident stated that “unique conditions” at the plant contributed to the accident risk.
Further quoting the paper, it’s said that if we take the perspective of the decision-makers on the respective plants, then “an obvious conclusion is that positive safety climate scores should not be interpreted as indicating that the organizational conditions for safety are all OK and do not require further attention” (p18).
In contrast, plant B’s SC scores differed from the industry average and in their words, “displayed a negative development”. It’s noted that this score could have been seen as a warning sign that the conditions at the plant were deteriorating, in theory. Thus, again for decision-makers, results from SC assessments “should be the source of concern and make managers worry about practical drift and eroding safety margins” (p18).
Comparisons between the incident reports and the SC survey data was discussed. While the authors don’t treat the investigation reports as objective truths, they do say that you would expect a certain level of correlation *if* the surveys and investigations were valid indicators for decision-makers. In this sense, issues around reorganisation, unclear responsibilities and a lack of understanding and expertise in high-risk operations in plants A and B – all identified as contributing factors in the close call investigations, were not identified in the proactive SC assessments prior to the incidents.
It’s said that SC “is an attempt to measure organizational conditions with an indirect relationship to safety performance” (p18) and thus, there’s considerable ambiguity in how to interpret the insights. The SC findings at plant B perhaps indicated that some degree of improvements was needed prior to the close call (weak signals), but this depended on interpretation. All weak signals require interpretation and to do this it requires an “in-depth assessment into why the safety climate scores display a negative development” (p18).
Although this may seem obvious, I agree with the authors that the previous paragraph are important points to emphasise. While organisational factors, the types of which that SC surveys seek to measure, may help inform why conditions come to exist, “…a poor safety climate can never be a direct cause of an accident, only indirect” (p19).
Thus, when designing indicators to target major accident risk, they argue that it may be more important to focus on the more proximal factors “and maintain the link from the shop floor, and strive to follow a logic of causal specificity” (p19).
Organisations should therefore try to reduce the distance between what organisational or work factor the indicator is measuring versus its proximity to the actual work/performance. In a nice summation, they state: “This implies a bottom-up approach, where an important prerequisite is a solid understanding of the work that takes place at the sharp end” (p19).
In further discussing the implications, they state that efforts to try and reduce complex phenomena like culture into measurable things like climate “is not necessarily fruitful when dealing with the relationship between human and organizational factors and their explicit contribution to major accident risk; describing the phenomenon … and leaving it out of the indicator discussion might be a more fruitful approach” (p19).
In concluding, it’s said that these findings can be used to question the position of SC as a major risk indicator and highlight the pitfalls of using measurements of perception of safety conditions to manage major accident potential.
However in saying this, the authors don’t suggest that SC surveys are invalid or not useful. Instead, they clarify that SC assessments perhaps shouldn’t be used as indicators for safety performance but rather as an indication of where to focus attention for further assessment of factors and “drawing the attention to local conditions more directly linked to major accident risk” (p20).
Authors: Sverre Andreas Kvalheim, Stian Antonsen, Stein Haugen, 2016, International Journal of Disaster Risk Reduction
Study link: http://dx.doi.org/10.1016/j.ijdrr.2016.05.011
Link to the LinkedIn article: https://www.linkedin.com/pulse/safety-climate-indicator-major-accident-risk-can-we-use-hutchinson
SafetyInsights.org 