Safety culture maturity models in occupational safety and health

Our Summary

Definitions of safety culture 

The term “safety culture” was first introduced in the nuclear sector in 1986 after the Chernobyl disaster and was later taken up by various bodies – nuclear or otherwise – with sometimes quite different meanings. 

To mention only the best known of these formulations, we find:

Maturity models

In the same way, several approaches have been proposed over time to measure the maturity of an organization’s safety culture, but none has demonstrated any proven and reliably measured improvement in safety performance in the field from one level to the next.

The Westrum Model

Westrum (1993) proposed his well known maturity model distinguishing three levels:

1. Pathological: the messenger is not listened to.
2. Bureaucratic: the messenger is heard, rules are applied, and the scapegoat [FF1.1]is blamed according to a short‑loop logic.
3. Generative: underlying causes are examined, lessons are learned, and results are generalized.
    

The DuPont Bradley Curve

The DuPont Bradley Curve (1995) presents four stages of maturity:

1. Reactive: the organization reacts to the incident and expects only compliance.
2. Dependent: discipline, rules, and procedures are reinforced.
3. Independent: individual commitment to everyone’s safety is observed.
4. Interdependent: strong engagement emerges from the entire team acting as a unified collective.

Hudson following Westrum

Hudson (2001) expanded Westrum’s model by adding two intermediate levels: reactive and proactive. He also replaced the bureaucratic level with the term calculative to indicate that this stage corresponds to the deployment of organizational systems—with all the audit tools and lessons learned mechanisms – while remaining highly reactive. It is the subsequent proactive level that marks a turning point, as it uses these tools with an anticipatory mindset, including a shared and up to date awareness of potential risks between the field and management.

The HSE Model by Flemming

Flemming (2001), in the same year as Hudson, proposed his HSE model (Health and Safety Executive), which also identifies five maturity levels: emerging, managing, involving, cooperating, and continually improving.
These levels describe the organization’s safety maturity: beginning with a stage of total indifference, where applying procedures is believed to solve safety issues; then moving to stages where safety becomes a priority first for frontline operators; then for frontline supervision; before reaching the most mature stage, where top management makes safety a core value and a priority guiding the organization and its decisions.

The 2018 and 2022 Reviews

Two reviews have examined these different maturity models (Ayob 2022, and Foncalves Filho and Warterson 2018), and both showed that even though using a safety culture maturity matrix has become widespread in industry – particularly in Oil & Gas, construction, and healthcare – organizations have largely remained at the stage of “diagnosing” the maturity level and the desired evolution (usually via questionnaires), without truly validating these models through concrete results showing improvements or deteriorations in organizational safety performance when moving from one level to another (either toward greater maturity or the opposite).

What About After 2022?

The purpose of this article is to explore the literature to identify what has been demonstrated and consolidated in terms of validation, and what the actual safety gains have been since that earlier observation. This is why the review covers only the years 2022 through the end of 2024 (from the date of Ayob’s last analysis in 2022 up to today).

During this (very) short period, 3,888 articles containing one or more keywords related to safety culture, maturity, measurement, plus a list of related concepts, were identified in the publications. Out of this total, 3,214 were read at the abstract level, and 32 were read in full. In the end, only 17 publications were retained for their relevance, including 15 conference papers. Six come from Indonesia, three from Brazil, and one each from the USA, Poland, Finland, Australia, and Iran. They cover many industrial sectors, from Oil and Gas to commercial shipping, and also healthcare and construction.

It is worth noting that the variety of countries contributing to these studies (Westerners, Asians, and countries at different development levels) should not obscure the fact that these studies are still conducted in large international companies, and almost never concern SMEs or intermediate-sized enterprises – whether in Western nations or elsewhere.

The majority (10) of these studies focus on achieving a higher maturity level in the organization’s safety culture, with a strong reliance on Hudson’s model mentioned earlier.

The results of these recent studies (since 2022) remain consistent with those published before 2022. The focus is still on diagnosing the current level of maturity in order to increase it, while continuing to show limited operationalization of tools to assess this desired change (organizational tools, indicators, etc.). Evaluations remain largely qualitative, based on questionnaires and successive interviews.

Only two studies use quantitative indicators. Unsurprisingly, the most frequently monitored dimensions involve changes in perceptions of leadership, engagement, communication, and participation.

The least monitored dimensions involve injury and loss data in all its forms in the field of occupational safety (counts of incidents, accidents, days of absence, job transfers, resignations, occupational illnesses, or even quantified impacts on QWL).

Les auteurs pointent la faiblesse de ces postures, et regrettent aussi qu’il n’y ait pas plus d’applications à des PME-ETI, dans des pays pauvres où les ressources manquent plus cruellement que dans nos pays. Une extension du regard sur la culture de sécurité à ces pays et aux petites entreprises donnerait sans doute une opportunité de progresser plus vite sur l’opérationnalisation du concept par la lecture plus contrainte qu’il imposerait.

The authors highlight the weaknesses of these approaches and also regret the lack of applications involving SMEs and intermediate-sized enterprises, especially in low income countries where resources are much more limited than in developed nations. Broadening the perspective on safety culture to include these countries and small enterprises would likely offer an opportunity to accelerate progress on the operationalization of the concept, given the more constrained lens such contexts would impose.