Source: Bogna, F., Perry, K., & Raineri, A. (2026). Visual representation of sociotechnical system paradigms for occupational health and safety practices: A scoping review. Safety Science, 201, 107198. https://doi.org/10.1016/j.ssci.2026.107198
Introduction to the Article
In the field of occupational health and safety (OHS), sociotechnical systems (STS) approaches are used across a broad spectrum, ranging from the analysis of workplace accidents to the anticipation of risks. However, until now, there has been no systematic mapping of which graphical models are used to represent these approaches in the literature, whether these models are retrospective or predictive, and which STS components are foregrounded. This scoping review, published in Safety Science, aims precisely to fill this gap.
The sociotechnical systems paradigm conceptualizes occupational health and safety not as the outcome of isolated technical controls or individual worker behavior, but as an emergent property of interactions among people, technologies, tasks, organizational structures, operating systems, physical environments, and safety culture. In this view, safety is produced through the joint optimization of social and technical elements, meaning that work performance, decision-making, risk control, and accident prevention must be understood within the wider system in which they occur. Bogna, Perry, and Raineri’s scoping review shows that this paradigm has increasingly been represented through graphical models such as AcciMap, STAMP, FRAM, causal loop diagrams, and risk management models, which help visualize complex interdependencies that are difficult to explain through text alone. The article also demonstrates that sociotechnical thinking in safety science has developed along two complementary lines: retrospective models that explain how accidents or system failures occurred, and predictive models that anticipate hazards before harm takes place. Therefore, the sociotechnical systems paradigm shifts OHS practice from a narrow compliance-based or person-blame approach toward a systemic, relational, and future-oriented understanding of safety, where failures and successes are both interpreted as products of dynamic interactions within complex work systems (Bogna et al., 2026)
Bogna et al. (2026) examined 257 peer-reviewed articles covering the period from 1980 to 2024, retrieved from Web of Science, Scopus, ScienceDirect, and PubMed. The review included only graphical models that contained at least two STS components and were presented in the context of OHS or safety science. Conducted within the PRISMA-ScR framework, the study used the Covidence platform for screening and data extraction processes.
Main Findings
The main findings of the study can be summarized as follows.
The first graphical STS model was identified in 1995. A visible increase occurred after 2005, followed by a marked acceleration after 2014. This pattern indicates the growing legitimacy of STS thinking within OHS and safety sciences, as well as the increased accessibility enabled by digital publishing.
Retrospective and predictive approaches were represented almost equally, with 130 publications in each category and three articles combining both approaches. This balance reflects the continuing dual interest in analyzing past events and managing future-oriented risks.
The most frequently used graphical models were AcciMap (n = 44), STAMP (n = 29), Risk Management Model (n = 28), Causal Loop Diagrams (n = 27), and FRAM (n = 25). In addition, linear models (n = 22) and HFACS (n = 8) also appeared with notable frequency.
Among STS components, “operating systems” (n = 225) was the most frequently represented element, whereas “safety climate or safety culture” (n = 70) was identified as the least represented component. The great majority of publications included four or more STS components.
At the country level, Australia (n = 47), particularly the Centre for Human Factors and Sociotechnical Systems (CHFSS) at the University of the Sunshine Coast, occupied a dominant position. China (n = 38) showed a rapid rise, followed by the United Kingdom (n = 31), the United States (n = 16), and Iran (n = 14). In terms of author frequency, Salmon, Goode, Stanton, and Read were the most prolific contributors, all of whom were affiliated with CHFSS.
Strengths of the Article
This study is significant as the first known scoping review to map the use of STS graphical models in OHS and safety sciences. Methodologically, adherence to the PRISMA-ScR protocol, the use of two independent reviewers during the screening process, and the systematic use of the Covidence platform strengthen the reproducibility of the study. By quantitatively documenting the retrospective-predictive distinction, the study provides empirical evidence of the paradigmatic balance in the field. In addition, the analyses by author and country make the phenomenon of institutional concentration empirically visible.
Issues Overlooked or Insufficiently Addressed by the Article: Future Research Opportunities
Like any comprehensive study, the review by Bogna et al. (2026) leaves certain dimensions outside its scope. The following sections discuss these gaps as future research opportunities.
- Lack of Sectoral Differentiation: The Absence of a Healthcare Perspective
The review addresses STS models under the broad umbrella of OHS and safety science, but it does not provide a sectoral differentiation. The use patterns of STS models differ across sectors such as healthcare, aviation, mining, construction, and energy. In healthcare in particular, STS thinking has followed a distinctive development trajectory in the context of patient safety, as seen, for example, in Carayon and colleagues’ SEIPS model. Future studies could compare sector-specific preferences for STS graphical models and identify which models are more effective in which contexts. For health management researchers, this offers a research pathway for examining the relationship between patient safety culture and STS modeling preferences.
- Systematically Low Representation of Safety Culture and Safety Climate
One of the most striking findings of the study is that the “safety climate or safety culture” component was represented in only 70 publications. This is considerably low when compared with “operating systems” (n = 225). The authors partially attribute this finding to conceptual ambiguities in the measurement of safety culture, but they do not analyze it in depth. Although safety culture occupies a central place in organizational accident models, particularly in Reason’s work, why it is so underrepresented in graphical models remains a separate research question. Does this result from the difficulty of visualizing safety culture compared with more concrete system components, or from researchers’ prioritization of operational and technical dimensions? Future research should examine how safety culture can be more effectively integrated into STS graphical models, particularly in healthcare institutions, by investigating the alignment between safety culture scales such as HSOPSC and STS modeling outputs.
- Lack of Evaluation of Model Effectiveness and Practical Outcomes
Due to the nature of a scoping review, the authors did not assess the design quality, usability, or implementation outcomes of the models. However, this also points to a critical gap in the field: what contribution do graphical STS models make to real-world OHS performance? There is little evidence on what concrete organizational changes result from a retrospective AcciMap analysis or how many potential incidents are prevented through a STAMP-based predictive approach. This is an important issue that could be examined through a systematic review or meta-analysis focused on intervention effectiveness.
- Superficial Treatment of Industry 4.0/5.0, Artificial Intelligence, and Digital Twin Integration
The article conceptually discusses the effects of Industry 4.0 and 5.0 on STS, but it does not report how many of the 257 publications integrated technological components such as digital twins, AI-supported decision support systems, Internet of Things (IoT) sensors, or machine learning-based risk prediction into STS models. Yet it is clear that these technologies are fundamentally transforming the “tools and technologies” component of STS. Future studies should examine how digitalization changes the structure of STS graphical models, for example, whether traditional hierarchical layers are being replaced by network-based or dynamic simulation models. In this context, linking digital twin applications in healthcare with patient safety-oriented STS models represents an interdisciplinary research opportunity.
- The Invisibility of Developing Countries and Türkiye
The country-level analysis reveals the dominance of Australia, China, and the United Kingdom. Türkiye appears only in a small category, and the contribution of the Turkish OHS literature to STS graphical models remains quite limited. However, Türkiye operates an OHS regime based on risk assessment through Law No. 6331 on Occupational Health and Safety and has a broad academic infrastructure through OHS professional training programs. Social Security Institution workplace accident statistics, high accident rates in the mining and construction sectors, and the growing OHS awareness in healthcare all demonstrate the potential of STS modeling studies in the Turkish context. For Turkish researchers, applying models such as AcciMap or STAMP to Türkiye-specific OHS contexts, such as multilevel sociotechnical analyses of mining accidents or predictive modeling of hospital OHS management systems, could contribute to national OHS policy while also increasing international visibility.
- Insufficient Analysis of Interdisciplinary and Hybrid Models
The study classifies models into separate categories such as AcciMap, STAMP, and FRAM, but it does not provide a systematic analysis of hybrid approaches, such as the combined use of STAMP and Rasmussen’s Risk Management Model or the integration of FRAM with causal loop diagrams. Yet studies such as the STAMP-FRAM-RAG integration proposed by De Linhares et al. (2021) demonstrate the advantages of hybrid approaches in complex systems where a single model may be insufficient. Future reviews should examine patterns of combined model use, which combinations are preferred for which types of analysis, and the rationale behind these preferences.
- Limited Representation of the Resilience Engineering Perspective
Although FRAM is a product of the resilience engineering paradigm, the study does not deeply address Hollnagel’s distinction between Safety-I and Safety-II, the tension between work-as-done and work-as-imagined, or the concept of adaptive capacity. How the resilience engineering approach is reflected in STS graphical models, for example, how successful performance is graphically represented from a Safety-II perspective, is a separate research question. This perspective is particularly important in healthcare for analyzing everyday successful clinical processes beyond near-miss events.
- Language Limitation and Cultural Context
The review includes only English-language publications. STS studies published in Chinese, Persian, Turkish, Spanish, and other languages were excluded. Given the rising publication volume from China, the inclusion of Chinese-language publications could significantly alter the findings. More importantly, are STS models applied universally, or does cultural context, such as power distance, individualism-collectivism, and uncertainty avoidance, affect the design and interpretation of these models? This cultural dimension provides fertile ground for future comparative research.
- Lack of Attention to Stakeholder Participation and Knowledge Transfer
The authors mention the potential of graphical models for knowledge transfer and stakeholder communication, but they do not evaluate the practical communication effectiveness of these models, for example, how an AcciMap is understood and interpreted by field workers, managers, and regulatory bodies. The effects of STS graphical models on cognitive load, comprehensibility, and decision-making processes represent a research area bridging human factors and information design. In health management, this issue has direct applicability in the transformation of root cause analysis reports into organizational learning.
- Cost-Effectiveness and Implementation Feasibility
The pragmatic dimensions of the reviewed models, such as applicability, required expertise, time, and resource costs, are not addressed. What are the practical reasons for a healthcare organization or industrial enterprise to choose STAMP or FRAM instead of AcciMap? Systematically examining the factors that influence model selection, such as organizational size, sector, existing OHS maturity level, and data infrastructure, could help bridge the research-practice gap.
Concluding Evaluation
The study by Bogna et al. (2026) is a comprehensive and methodologically robust review that maps the 30-year evolution of STS graphical models in OHS and safety sciences. The quantitative documentation of the balance between retrospective and predictive approaches, the identification of dominant models, and the demonstration of institutional concentration provide a valuable reference point for understanding the current state of the field.
However, the ten research opportunities discussed above, namely sectoral differentiation, safety culture integration, model effectiveness, digital transformation, developing-country perspectives, hybrid models, resilience engineering, cultural context, stakeholder communication, and implementation feasibility, point to dimensions of the field that remain underexplored. For health management researchers in particular, STS modeling studies in the contexts of patient safety, clinical risk management, and OHS management systems in healthcare institutions offer promising opportunities in terms of both theoretical depth and practical impact.