Assessing Resilience Practices in the Digital Transformation Era: A Storytelling-Based Cross-Sectional Study in Italy

Abstract

This study applies Safety II principles within a storytelling- and RAG-based questionnaire to explore how resilience engineering (RE) principles and practices are perceived and implemented in Italy’s manufacturing sector. Before completing the questionnaire, accident and near-miss scenarios were presented through narrative stories. The cross-sectional study on 334 companies reveals that Monitor and Respond are prioritized over Anticipate and Learn, with medium-large companies and those adopting technological innovations showing more advanced resilience-oriented OSH management practices. The study emphasizes the importance of company size and technological adoption in shaping safety practices, recommending investment in innovative solutions and fostering a culture that addresses near misses to prevent severe accidents and support continuous improvement.

1. Introduction

In recent years, the field of Occupational Safety and Health (OSH) has undergone a significant transformation, shifting from traditional, failure-centric models to more proactive and adaptive paradigms. Historically, OSH practices were guided by the Safety-I perspective, which focuses on identifying and eliminating risks to prevent accidents [1]. However, this approach has shown limitations in addressing the complexity and variability inherent in modern socio-technical work environments [2]. As work processes evolve and become more interconnected, particularly with the onset of digital transformation, the limitations of a purely prescriptive and reactive safety model have become increasingly evident [3]. This has led to the emergence of Safety-II, a perspective that emphasizes understanding how work develops well and how systems can adapt and maintain performance under varying conditions [3,4].

To meet these demands, organizations are increasingly turning to Resilience Engineering (RE), which emphasizes the importance of an organization’s capacity to adapt and respond effectively to unexpected disruptions or failures [5]. RE focuses not only on failure prevention but also on the ability to recover from incidents and continue operations safely [6]. Different studies have shown that RE principles are particularly relevant in environments that are technologically advanced and subject to high levels of uncertainty [7,8]. In this context, the Resilience Analysis Grid (RAG) [9] has been developed to evaluate resilience within organizations, employing the four cornerstones abilities referred to as the four cornerstones of resilience, although similar principles can also be applied [10]: (i) to respond to what happens, (ii) to monitor critical developments, (iii) to anticipate future threats and opportunities, and (iv) to learn from past experience, applied to both successes and failures. These aspects need to be seen together since all four are necessary and interdependent [11].

Conducted as part of the FEREO project (Training and Organizational Resilience Engineering), funded by INAIL (National Institute for Insurance against Accidents at Work), the study presents a cross-sectional study conducted within Italian manufacturing companies, with a particular focus on practices that align with RE principles as defined by the RAG model. In this paper, the Safety II perspective is integrated into the methodology to assess not only the perceived importance but also the actual implementation of these practices. This study contributes uniquely to the field of resilience engineering by combining the RAG model with a storytelling methodology. The use of storytelling introduces a novel approach to safety research by engaging employees in real-life scenarios—one based on an accident and one on a near-miss—fostering reflection and encouraging deeper engagement with safety practices. Storytelling, as a qualitative research methodology [12], is particularly effective in capturing the complexity of organizational experiences and understanding human behavior in real-world settings [13]. It allows for the collection of rich, contextual data that can reveal how employees perceive resilience and safety practices in their daily work, shedding light on the factors that influence their implementation [14]. This methodological combination allows for the exploration of both perceived and enacted resilience practices in organizations, highlighting the interplay between technical systems and human factors in safety management. As such, this study not only addresses a gap in the existing literature but also introduces a flexible, human-centered methodological framework for exploring resilience in high-risk settings. This approach holds potential for application in similar studies across diverse geographical and industrial contexts, thereby enabling meaningful comparative analyses across regions and sectors.

As a result, this work addresses four important research questions.

RQ1. How are good practices, aligned with the principles of resilience engineering (based on the RAG model), perceived in terms of importance in the context of OSH?

RQ2. To what extent are these good practices effectively implemented in Italian companies regarding OSH?

RQ3. Are there significant differences in the perception and implementation of these good practices based on company size?

RQ4. What role does the introduction of new technologies play in shaping the perception and implementation of these good practices?

The remainder of the paper is structured as follows. Section 2 outlines the background and motivation for the study. Section 3 describes the materials and methods, detailing the study design, the development of the data collection tool, and the techniques used for data analysis. Section 4 presents the results, followed by a discussion of the findings in Section 5. Finally, Section 6 provides the conclusions and suggests directions for future research.

2. Background and Motivation

Safety II principles are well-suited to the challenges of complex and dynamic systems. In these environments, variability in work and the ability to adapt to change are essential for safe and successful operations [15].

As discussed in [16], Safety II promotes a proactive and flexible approach to safety. It emphasizes systems thinking, which focuses on how different parts of a system interact. In this view, safety is not a fixed outcome of individual components, but something that emerges from the relationships between them [2]. Furthermore, RE highlights the system’s ability to adapt and recover from both expected and unexpected conditions, ensuring continued safety [5]. Unlike Safety I, which responds to failures after they happen, Safety II aims to anticipate problems and prepare for them in advance [17]. It also values learning and adaptability, where organizations continuously evolve their safety practices based on lessons learned from both successes and failures [18]. Additionally, Safety II recognizes human performance variability as an asset, viewing human adaptability and performance variation as key to enhancing resilience and maintaining safety in dynamic environments [19].

In this context, a key concept is Shorrock’s distinction between different perspectives on human work [20], particularly how people perceive or imagine work versus how it is actually performed. In this sense, four fundamental types of work can be identified:

-

Work-As-Imagined (WAI): What people think others are doing, or expect to do.

-

Work-As-Prescribed (WAP): The formal instructions or rules about how work should be conducted (e.g., laws, procedures, checklists, job descriptions).

-

Work-As-Disclosed (WADi): How people describe or report their work in documents or conversations.

-

Work-As-Done (WAD): The actual work that takes place in practice, i.e., the real actions that people perform. It includes the influences of variability, fluctuating goals, changing demands, and unexpected conditions.

These four types of work highlight the gap between how work is imagined and how it is executed, underscoring the complexity and variability in real-world practice.

Such complexity and variability are highly characteristic of a complex socio-technical system, where a systemic perspective is required [21]. To address a systemic perspective, RE suggests considering the four resilience abilities previously mentioned: (i) respond, (ii) monitor, (iii) anticipate, and (iv) learn [22]. These abilities must be viewed together, as they are interdependent and all essential. Furthermore, they form the foundation of the Resilience Analysis Grid (RAG), which was adopted in this study for its versatility and adaptability to the context of interest.

The RAG is a questionnaire-based tool that provides guidelines for categorizing information into the four systemic potentials, known as cornerstones. Each cornerstone needs to be articulated and, therefore, operationalized into items. Typically, each item is associated with a question that has a specific response mode and a corresponding score. Once combined, the score associated with the items of a systemic potential provides the proximal measure of the corresponding systemic potential. Through its implementation, it is then possible to measure the resilience potential of a process, an organization or an entity in general [23]. It does not provide strict guidelines on the methods to be used for data collection or analysis. Rather, it provides a method for organizing, itemizing, and disaggregating information. It also relies on intuitive graphical representations, such as radar charts, to support intuitive information visualization and interpretation [24]. Since the RAG is a questionnaire-based tool only broadly defined to be then further tailored to the organization or system under study [9], it has been tested in a variety of settings, e.g., healthcare [25], transport [26], and air traffic management [27], and also for OSH purposes [28,29].

Although the adoption of RE principles and approaches is widely acknowledged in the manufacturing domain [30,31,32,33,34], the application of the RAG in such a sector remains relatively limited. In [35], the RAG was used to assess and enhance the resilience of a change management process in an aircraft components production company, demonstrating its effectiveness in identifying both strengths and weaknesses within the process. In [28], it was applied to collaborative robot (cobot) applications, proving to be a valuable and replicable tool for identifying areas of strength and improvement in managing OSH management in cobot-integrated workplaces, while also revealing significant gaps in contextual risk assessment practices. Furthermore, although [36] presented a methodological approach to integrate the RAG with the Rasch model to establish a standardized and objective method for measuring and comparing organizational resilience across multiple entities, the study also highlighted the suitability of the proposed method for manufacturing organizations, as they typically involve complex relationships among agents, intensive human-machine interactions, and high safety requirements.

What these studies have in common is that they demonstrate the RAG’s value in improving OSH management and in fostering greater safety awareness among company employees.

This evidence justified and supported the adoption of the RAG in the present study. Moreover, this flexibility is well suited to combining this model with storytelling, which in itself takes an exploratory and tailored view of the problem [13]. The combination of storytelling and resilience engineering, particularly within the context of safety research, represents a relatively novel approach. While storytelling within the RAG framework in manufacturing safety has not been widely explored, its broader application across safety-related domains is well documented. In sectors such as construction and electrical work, storytelling has been effectively used to enhance safety training, commonly through toolbox talks, video narratives, and immersive digital tools like 360-degree panoramas, which have been shown to improve hazard recognition and risk perception [37,38,39,40]. Additionally, in software design, storytelling is a well-established method to communicate user needs and align multidisciplinary teams around shared goals [41].

Despite these established applications, to the best of the authors’ knowledge, no prior study has applied storytelling as a foundation for the RAG framework, either in manufacturing or other sectors. This study seeks to address that gap by using narrative scenarios to capture lived employee experiences and offer a fresh lens for examining complex issues of safety and organizational resilience.

Indeed, storytelling is a powerful means of connecting with people on an emotional level, capable of communicating complex messages engagingly and reflectively [42]. It has been shown to promote user-centered and divergent thinking [41], unlocking imagination for both the storyteller and the audience, and helping to spark new ideas and perspectives. As such, it is particularly suited to investigating adaptive capacities and emergent practices in organizational settings.

Among the different narrative methods of telling stories to engage the audience, this work adopted a case study approach [43], i.e., it used real-life examples to illustrate a point, an approach already used in academic settings [37]. Unlike the anecdote or the personal story, which are both based on personal experiences, the case study uses stories from real life to show exemplary events.

Summarizing what has been discussed, this study leveraged the principles of Safety II to incorporate them into a RAG-based questionnaire. This questionnaire was administered to a sample of Italian companies after they had been confronted with two stories designed to make them think through a storytelling approach.

3. Materials and Methods

The design of the study was conducted in three steps, as shown in Figure 1.

3.1. Questionnaire Design

A cross-sectional survey study was conducted to explore how RE principles are perceived and implemented in Italy’s manufacturing sector for OSH management.

3.1.1. Focus Group

The process began with the formation of a multidisciplinary focus group aimed at shaping the direction and content of the questionnaire. This foundational phase was essential for defining the direction and scope of the investigation [44]. This group was intentionally diverse, including seven participants with academic and research backgrounds in fields such as operations management, psychology, legal studies, and smart manufacturing systems. The focus group composition details are shown in

Table 1. Focus group composition.

Background	Area of Expertise	Focus on the Research	Years of Experience
Public research	Training and education	Occupational safety and health	26
Public research	Communication and digital technologies	Occupational safety and health	17
Public research	Legal	Occupational safety and health	18
Public research	Psychology	Occupational safety and health	29
Academic/Industrial	Operations management	Manufacturing systems	21
Academic	Smart factories	Manufacturing systems	6
Academic	Advanced industrial applications	Resilience Engineering	4

. Their collective experience and scientific diversity ensured a robust foundation for the study.

Over three sessions of approximately two hours each, the focus group analyzed the core dimensions of the RAG model and adapted them to align with Safety II concepts. The discussion centered on encouraging organizations to reflect on their responses to various scenarios. A key dimension emphasized was the discrepancy between work-as-imagined (WAI) and work-as-done (WAD). The group agreed to precede the questionnaire with two narrative-based stories—one depicting a workplace accident and the other a near miss—to stimulate deeper reflection among respondents.

3.1.2. Story Definition

Two narrative scenarios were developed, each situated in the same exemplary company (“Compeso”). These two stories represent a different scenario, developed in a narrative form, presenting contingencies and unpredictable events. Specifically:

• The first story represents an incident in a model company (“Compeso” company), resulting in a work-related injury caused by a violent event;
• The second story represents a near-miss in the same exemplary company, i.e., a work-related event that could have caused injury or harm to the worker’s OSH but, by sheer chance, did not.

For these two stories, the interviewee was asked to reflect from two perspectives:

• External, i.e., to assess the importance of specific systems or actions that could have prevented the incident or near-miss with respect to the company in question (to explore the perceived importance);
• Internal, i.e., to relate the issues from the two stories to their own organizational context, examining the actual implementation of specific systems or actions (to assess the actual level of implementation).

Below are the two stories, respectively corresponding to the scenario of an accident and a near-miss.

Story 1

On 11 September 2023, Compeso scheduled maintenance for the air conditioning system across all its warehouses. However, the operators continued working as usual, not noticing any changes in temperature. One of Compeso’s logistics operators was walking down a corridor in the raw materials warehouse when, suddenly, SHARK collided with him from behind, striking his right foot at high speed. SHARK is a compact, advanced autonomous vehicle that can move in all directions and navigate tight spaces. For optimal performance, SHARK requires specific temperature and humidity levels in its environment. The collision was forceful, and the operator was taken to the hospital, where he was diagnosed with a foot fracture. That day, the operator was not wearing safety shoes because his size was unavailable in the locker room.

Story 2

On 11 September 2023, Compeso scheduled maintenance for the air conditioning system across all its warehouses. However, the operators continued working as usual, not noticing any changes in temperature. That morning, the maintenance supervisor had informed Francesco, a warehouse logistics operator, that SHARK was exhibiting unusual behavior compared to previous days. SHARK is a compact, advanced autonomous vehicle designed to move in all directions and navigate tight spaces. For optimal performance, SHARK requires specific temperature and humidity levels in its environment. Later that afternoon, while walking through a corridor in the raw materials warehouse, Francesco noticed that SHARK, traveling at high speed, had lost control. He was able to avoid a collision and prevent an injury. That day, Francesco was not wearing safety shoes because his size was unavailable in the locker room.

3.1.3. Questionnaire Definition

The resulting questionnaire was divided into two sections.

The first section gathered demographic information to categorize the interviewee based on their role, work experience, and the manufacturing organization they belong to.

The second section followed the presentation of the aforementioned stories. In this section, the questions were designed to assess the four abilities of the RAG model. Based on the two stories and the two perspectives to be considered, four subsections were defined:

• Sub-section 1: external perspective, accident event;
• Sub-section 2: internal perspective, accident event;
• Sub-section 3: external perspective, near-miss event;
• Sub-section 4: internal perspective, near-miss event.

Specifically, the external perspective explores the importance the respondent places on specific systems and actions in the context of the Compeso company in question. The internal perspective, on the other hand, examines the presence of systems and the implementation of actions within the respondent’s own organization. This dual approach was applied to both accident and near-miss scenarios. So, based on the dimensions of the RAG model, the questionnaire investigated the perceived importance and the actual level of implementation of the following systems/actions (each corresponding to one question), totaling 15 questions for each scenario.

• Monitor: a system to notify when Personal Protective Equipment (PPE) is not being worn (Q1); a system for monitoring and recording real-time environmental conditions (e.g., temperature, humidity) (Q2).
• Respond: a system to prevent entry in industrial/operational areas without PPE (Q3); a system to shut down technology when anomalies are detected (Q4).
• Anticipate: awareness programs on the importance of PPE (Q5); updates to risk assessments related to new technologies (Q6); training on how to handle unforeseen situations (Q7); training on the proper use of PPE (Q8); ensuring PPE is provided for every worker (Q9); risk assessments for changes in the environment where technology is used (e.g., maintenance work, introduction of external personnel) (Q10); communication to staff about significant changes to the environment (e.g., maintenance work, introduction of external personnel) (Q11).
• Learn: investigations into the causes of accidents (Q12); improving knowledge of the technology involved in incidents (Q13); revising procedures for PPE usage (Q14); updating internal communication processes related to changes in environmental conditions (Q15).

The full questionnaire (

Table A1. Questionnaire for data collection.

Section 1: Master data

Gender Age Company location Number of employees in the company Division of ATECO Section C—Manufacturing Activities where your company operates Role played in OSH Seniority in this role

Section 2.1: External perspective—accident

Story 1: On September 11, 2023, Compeso scheduled maintenance for the air conditioning system across all its warehouses. However, the operators continued working as usual, not noticing any changes in temperature. One of Compeso’s logistics operators was walking down a corridor in the raw materials warehouse when, suddenly, SHARK collided with him from behind, striking his right foot at high speed. SHARK is a compact, advanced autonomous vehicle that can move in all directions and navigate tight spaces. For optimal performance, SHARK requires specific temperature and humidity levels in its environment. The collision was forceful, and the operator was taken to the hospital, where he was diagnosed with a foot fracture. That day, the operator was not wearing safety shoes because his size was unavailable in the locker room.

How important would the presence of the systems and the implementation of the following actions have been to avoid the accident in the Compeso company?

Q1. System to notify entry without PPEs Q2. System to monitor and record real-time operating environmental conditions (e.g., temperature, humidity)

Q3. System to prevent entry without PPEs

Q4. System to stop the technology when anomalies are detected

Q5. Organize information and awareness-raising activities on the importance of PPEs

Q6. Update risk assessment for new technologies in place

Q7. Train on how to handle unforeseen situations

Q8. Train in the proper use of PPEs

Q9. Make PPE available for each worker

Q10. Assess risks resulting from extraordinary changes to the technological environment (e.g., maintenance work, introduction of outside personnel)

Q11. Notify staff about extraordinary changes to the environment where technology is present (e.g., maintenance intervention, introduction of outside personnel)

To prevent the recurrence of this type of accident, how important would it be for Compeso company to implement the following actions?

Q12. Activate an investigation into the causes of the event

Q13. Increase the level of knowledge regarding technological equipment associated with the event

Q.14 Revise the procedure regarding the use of PPEs

Q.15 Review internal communication processes/systems regarding changes in environmental conditions

Section 2.2: External perspective—near-miss scenario

Story 2: On September 11, 2023, Compeso scheduled maintenance for the air conditioning system across all its warehouses. However, the operators continued working as usual, not noticing any changes in temperature. That morning, the maintenance supervisor had informed Francesco, a warehouse logistics operator, that SHARK was exhibiting unusual behavior compared to previous days. SHARK is a compact, advanced autonomous vehicle designed to move in all directions and navigate tight spaces. For optimal performance, SHARK requires specific temperature and humidity levels in its environment. Later that afternoon, while walking through a corridor in the raw materials warehouse, Francesco noticed that SHARK, traveling at high speed, had lost control. He was able to avoid a collision and prevent an injury. That day, Francesco was not wearing safety shoes because his size was unavailable in the locker room.

How important would the presence of the systems and the implementation of the following actions have been to avoid the near miss in the Compeso company?

(Q.1–Q.11)

To prevent the recurrence of this type of near miss, how important would it be for Compeso company to implement the following actions?

(Q.12–Q.15)

Section 2.3: Internal perspective

Are the following OSH systems and measures in place in your company? (Q.1–Q.11)

Following an accident, has your company implemented the following actions? (Q.12–Q.15) Following an near mis, has your company implemented the following actions? (Q.12–Q.15)

) and a table summarizing RAG dimensions, related items, and rationale (

Table A2. RAG dimensions, related items, and rationale.

RAG Dimension	Question		Motivation
	ID	Focus
Monitor	1	System to notify entry without PPE	To detect unsafe behaviors in real time by monitoring unauthorized access
	2	System to monitor and record real-time operating environmental conditions (e.g., temperature, humidity)	To observe and track critical environmental variables for early detection of deviations
Respond	3	System to prevent entry without PPE	To react promptly to safety threats by preventing unsafe access
	4	System to stop the technology when anomalies are detected	To respond effectively to abnormal conditions and mitigate associated risks
Anticipate	5	Organize information and awareness-raising activities on the importance of PPE	To increase awareness and prepare personnel for potential future risks
	6	Update risk assessment for new technologies in place	To foresee and address risks associated with the implementation of new technologies
	7	Train on how to handle unforeseen situations	To prepare personnel to act effectively in uncertain or unexpected scenarios
	8	Train on the proper use of PPE	To ensure correct usage of PPE and prevent future safety issues
	9	Make PPE available for each worker	To anticipate needs and guarantee the availability of protective equipment
	10	Assess risks resulting from extraordinary changes to the technological environment (e.g., maintenance work, introduction of outside personnel)	To identify and mitigate potential hazards resulting from atypical operational conditions
	11	Notify staff about extraordinary changes to the environment where technology is present (e.g., maintenance intervention, introduction of outside personnel)	To inform proactively about changes that may affect safety, enabling anticipatory adjustments
Learn	12	Initiate an investigation into the causes of the event	To analyze incidents and promote organizational learning
	13	Increase the level of knowledge regarding technological equipment associated with the event	To enhance understanding of technical aspects and derive lessons from events
	14	Revise the procedure regarding the use of PPEs	To adapt existing protocols based on past experience and improve future performance
	15	Revise internal communication processes/systems regarding changes in environmental conditions	To improve information flow and incorporate feedback for better adaptation

) are reported in Appendix A.

As is evident, the number of items across the different RAG dimensions of the questionnaire is not fully balanced. However, this approach is commonly used in the literature [10,11], given RAG’s inherent flexibility to be adapted to various contexts and investigative needs [9].

The questionnaire was initially pretested with two manufacturing companies to assess the clarity and readability of the questions. Following this, a focus group composed of researchers and OSH specialists conducted a qualitative review. This expert panel identified and corrected minor typographical errors and inconsistencies, further refining the instrument. The questions were then revised and finalized to ensure their clarity, comprehension, and appropriateness for the companies involved. While a Cronbach’s alpha analysis yielded a reliability score of 0.85, it is important to emphasize that the objective of this study was not to develop a universally generalizable tool. Instead, the questionnaire was designed as a context-specific instrument, tailored to the operational environments of the participating companies, and thus, a broader validation was not pursued.

3.2. Data Collection

The process of contacting organizations lasted four months, from January to May 2024. The companies were first contacted by phone. After that, they were provided the option to complete the questionnaire on their own or to participate in a guided phone interview. The overall process involved five key stages:

3.2.1. Building the Contact Lists and Directories

This stage involved using two databases, www.registroaziende.it (access on 2 January 2024) and www.infocamere.it (access on 2 January 2024), filtering by relevant ATECO codes. A total of 6400 companies were identified and contacted.

3.2.2. Initial Contact with Companies

During this stage, the researchers called the company secretariats to schedule phone appointments or obtain email addresses. The contact was balanced across Northern, Central, and Southern Italy.

3.2.3. Sending Personalized Communications

At this stage, a personalized email was sent, accompanied by an invitation letter to participate in the survey.

3.2.4. Collecting Responses

In this phase, 70% of respondents completed the questionnaire independently via the link provided, while the remaining 30% preferred a phone interview. Each interview lasted approximately 20–30 min.

3.2.5. Follow-Up with Companies

In this final stage, companies that had shown interest but had not completed the questionnaire within 15 days of the initial contact were re-contacted. Up to three targeted follow-up attempts were made.

3.3. Data Analysis Techniques

The data were coded and analyzed using IBM SPSS Statistics (version 30.0.0). Descriptive statistics, including frequencies and percentages, were used for data analysis. Mann–Whitney U test [45], which generalizes the Wilcoxon method, was used to determine significant differences between the distributions of independent groups. The test was found to be appropriate as it does not require assumptions about population parameters and can be used when the type of distribution is unknown and when the dependent variable is ordinal. Moreover, if the distribution shape is similar in both groups, the test allows verification of whether there is evidence of a statistically significant difference between the medians of the two groups. The categorical variables used to create the groups were company size and the introduction of new technology in the last 5 years. A p-value of less than 0.05 with a confidence level of 95% was considered significant.

4. Results

The analysis of the results began with a descriptive analysis of the responses collected. In total, 344 Italian manufacturing companies participated in the cross-sectional survey. A total of 64.53% of the sample was in Northern Italy, 18.31% in Southern Italy, and 17.15% in Central Italy. Tables describing the distribution by ATECO code (

Table A3. Distribution of the company sample by manufacturing ATECO code.

ATECO Code	Description	% of Companies
10	Manufacture of food products	9.30%
11	Manufacture of beverages	0.29%
13	Manufacture of textile items	2.91%
14	Manufacture of workwear	1.74%
16	Manufacture of wooden containers	8.72%
17	Manufacture of paper and paper products	4.07%
18	Printing and reproduction of recorded media	0.87%
19	Manufacture of coke and refined petroleum products	1 0.29%
20	Manufacture of chemicals and chemical products	5.81%
22	Manufacture of rubber and plastics products	6.98%
23	Manufacture of other non-metallic mineral products	1.74%
24	Metallurgy	2.33%
25	Manufacture of metal products (excluding machinery and equipment)	12.21%
26	Manufacture of computers, electronics, optics, electromedical equipment, measuring equipment	6.69%
27	Manufacture of electrical and non-electrical household equipment	5.81%
28	Manufacture of machinery and equipment	6.40%
30	Manufacture of other vehicles	1.16%
31	Manufacture of furniture	4.94%
32	Other manufacturing industry	10.76%
33	Repair, maintenance, and installation of machinery	0.87%
-	Other (to be specified)	6.10%

) and company size (

Table A4. Distribution of the company sample by company size.

Company Size	of Companies (%)
employees < 50	59.59%
50 ≤ employees ≤ 149	26.45%
150 ≤ employees ≤ 249	5.23%
employees ≥ 250	8.72%

) are provided in Appendix A.

4.1. Perceived Importance of Good Practice in Resilience Engineering Principles

The perceived importance of good practices in RE principles was examined regarding the two events presented in the stories—accident and near miss—through an external perspective, that is, by adopting the viewpoint of the Compeso company in question.

4.1.1. Accident

The interviewees were asked about the importance of having certain systems in place and implementing specific actions to prevent accidents within the company. Figure 2 below shows the distribution of the responses.

Providing PPE for each worker (question 9—Anticipate dimension) received the highest number of responses in the “very important” category. This action refers to a protective measure rather than a preventive one, as it reduces damage but does not decrease the likelihood of the event occurring. On the other hand, the system to report entry without PPE (question 1—Monitor dimension) received the highest number of responses in the “not important at all” category.

Additionally, interviewees were asked how important it would be for the company to implement certain actions to prevent the recurrence of this type of accident. Figure 3 below shows the distribution of the responses.

Activating an investigation into the causes of the accident (question 12—Learn dimension) received the highest number of responses in the “very important” category. In contrast, reviewing the procedure for using PPEs (question 14—Learn dimension) received the highest number of responses in the “not very important” and “not important at all” categories.

Figure 4 below shows the distribution of responses, grouped by RAG dimensions.

The Respond dimension is regarded as the most important, while the Monitor dimension is considered the least important.

4.1.2. Near Miss

The interviewees were questioned about the importance of having certain systems in place and implementing specific actions to prevent near misses within the company. Figure 5 below shows the distribution of the responses regarding the perceived importance of systems and actions to prevent a near-miss scenario.

The system to stop the technology when anomalies are detected (question 4—Respond dimension) received the highest number of responses in the “very important” category. The system to notify entry without PPEs (question 1—Monitor dimension) received the highest number of responses in the “not important at all” category.

Additionally, interviewees were asked how important it would be for the company to implement actions to prevent the recurrence of this type of near miss. Figure 6 below shows the distribution of the responses regarding the perceived importance of systems and actions to prevent the recurrence of a near-miss scenario.

Increasing the level of knowledge regarding the technological equipment associated with the near-miss scenario (question 13—Learn dimension) received the highest number of responses in the “very important” category. In comparison, reviewing the procedure for using PPEs (question 14—Learn dimension) received the highest number of responses in the “not important at all” category.

Figure 7 below shows the perceived importance of good practices in RE principles in case of a near-miss scenario, categorized by RAG dimension.

The Respond dimension was considered the most important across all respondents, while the Monitor dimension is regarded as the least important, consistent with the results obtained in the Accident case.

4.2. Implementation in Italian Companies of Good Practice in Resilience Engineering Principles

To assess the level of implementation of good practices in RE principles, the presence and use of specific systems/actions aimed at OSH were examined. Companies were asked to indicate whether the system/action was required, as not all may be mandatory or necessary depending on the processes involved. They were also asked whether these systems/actions were in place and, if so, whether they had been implemented. Figure 8 below shows the distribution of the responses regarding the level of implementation of systems and actions aimed at OSH.

A system to monitor and record real-time operational environmental conditions (question 2—Monitor dimension) is not mandatory for 40.41% of companies. A system to prevent entry without PPE (question 3—Respond dimension) is the least available. On the other hand, providing PPE for each worker (question 9—Anticipate dimension) is the most widely available and used. Additionally, interviewees were asked if, following an accident, they had implemented any actions within their company. Of course, they were first asked if an accident had occurred in the past 5 years. Figure 9 below shows the distribution of the responses regarding the level of implementation of actions following an accident.

About 34.60% of companies report that they did not record any accidents in the past 5 years. Among the remaining companies, most initiated an investigation into the causes of the accident (question 12—Learn dimension). The least implemented action was reviewing internal communication processes/systems regarding changes in environmental conditions (question 15—Learn dimension).

Interviewees were also asked if, following a near-miss scenario, they had implemented any actions within their company. Of course, they were first asked if a near-miss scenario had occurred in the past 5 years. Figure 10 below shows the distribution of the responses regarding the level of implementation of actions following a near-miss scenario.

About 43.90% of companies report that they did not record any near-miss scenario in the past 5 years. Among the remaining companies, most initiated an investigation into the causes of the near-miss scenario (question 12—Learn dimension). The least implemented action was reviewing the procedure for using PPEs (question 14—Learn dimension).

Figure 11 below shows the distribution of responses, grouped by RAG dimension distinguishing the actions related to the Learn dimension in the case of an accident and near-miss scenario. The distinction between the accident and near-miss scenarios exists only for the Learn dimension, as the other dimensions investigate the presence of specific systems or actions within the company, regardless of the events. Furthermore, from the Learn dimension, the responses “No accident in last 5 years (34.59%)” and “No near miss in last 5 years” (43.90%) were excluded.

The Respond dimension is considered the most important, while the Monitor dimension is regarded as the least important, consistent with what companies stated regarding the perceived importance. Furthermore, after an accident, there is a greater implementation of actions and systems focused on occupational health and safety compared to when a near-miss scenario occurs.

4.3. Differences by Company Size

The first variable used to create the groups refers to company size. In the dataset, the original variable has four categories: up to 49 employees (205 observations), between 50 and 149 employees (91 observations), between 150 and 249 employees (18 observations), and 250 or more employees (30 observations). To balance the groups and make the test more robust, thereby improving its ability to detect differences and reducing the likelihood of false positive or false negative errors, the data were aggregated into two new groups: small companies (205 observations, with companies having ≤ 49 employees) and medium–large companies (139 observations, with companies having > 49 employees). A random sample of 139 observations was then taken from the small companies’ group (group_1) to balance it with the medium–large companies’ group (group_2), each with 139 observations. Indeed, having two groups of equal or similar size makes the comparison more robust, as the ranks are distributed in an equitable, meaningful, and reliable manner, reducing the likelihood of Type I and Type II errors.

Table 2. Overview of the technical measures associated with the four RE dimensions and dis-tinguishing between different enterprise sizes—accident scenario.

MONITOR
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	18.31%	30.23%	48.55%
Important	8.72%	16.28%	25.00%
Not Very Important	9.01%	9.74%	18.75%
Not Important At All	4.36%	3.34%	7.70%
Total	40.41%	59.59%	100.00%
RESPOND
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	29.36%	42.73%	72.09%
Important	8.28%	12.50%	20.78%
Not Very Important	1.74%	3.05%	4.80%
Not Important At All	1.02%	1.31%	2.33%
Total	40.41%	59.59%	100.00%
ANTICIPATE
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	23.89%	36.77%	60.66%
Important	11.72%	16.09%	27.81%
Not Very Important	3.15%	4.41%	7.56%
Not Important At All	1.65%	2.33%	3.97%
Total	40.41%	59.59%	100.00%
LEARN
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	25.41%	36.92%	62.33%
Important	10.99%	16.05%	27.03%
Not Very Important	3.08%	5.00%	8.08%
Not Important At All	0.93%	1.63%	2.56%
Total	40.41%	59.59%	100.00%

and

Table 3. Overview of the technical measures associated with the four RE dimensions and distinguishing between different enterprise sizes—near-miss scenario.

MONITOR
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	13.23%	19.48%	32.70%
Important	3.78%	2.03%	5.81%
Not Very Important	5.52%	10.76%	16.28%
Not Important At All	17.88%	27.33%	45.20%
Total	40.41%	59.59%	100.00%
RESPOND
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	7.32%	15.63%	22.95%
Important	1.77%	1.02%	2.79%
Not Very Important	2.50%	5.20%	7.7%
Not Important At All	28.82%	37.74%	66.56%
Total	40.41%	59.59%	100.00%
ANTICIPATE
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	11.76%	18.74%	30.51%
Important	1.50%	1.62%	3.12%
Not Very Important	3.20%	6.23%	9.43%
Not Important At All	23.95%	33.00%	56.95%
Total	40.41%	59.59%	100.00%
LEARN
Importance Level	Medium-Large Companies	Small Companies	Total
Very Important	10.39%	17.81%	28.20%
Important	1.31%	1.09%	2.40%
Not Very Important	3.27%	5.89%	9.16%
Not Important At All	25.44%	34.81%	60.25%
Total	40.41%	59.59%	100.00%

provide an overview of the technical measures associated with the four RE dimensions, considering both accident and near-miss cases, and distinguishing between different enterprise sizes.

The results of the statistical test that revealed significant differences in terms of perceived importance are presented in

Table 4. Significant differences in terms of perceived importance of systems or actions aimed at OSH, considering the company size.

Action/System	RAG Dimension	Accident/Near Miss	Average Rank	Significance	Greater Perceived Importance
System to notify entries without PPEs	Monitor	Accident	gr_1) 150.43 gr_2) 128.57	0.016	Small companies
Training on how to handle unforeseen situations	Anticipate	Accident	gr_1) 149.23 gr_2) 129.77	0.027	Small companies
Initiate an investigation into the causes of the event	Learn	Accident	gr_1) 121.59 gr_2) 157.41	<0.001	Medium–large companies
Initiate an investigation into the causes of the event to prevent its recurrence	Learn	Accident	gr_1) 124.47 gr_2) 154.53	<0.001	Medium–large companies

.

Respondents from small companies place more importance on having a system for reporting entries without PPEs and providing training on how to handle unforeseen situations to prevent the accident that occurred at the Compeso company. In contrast, respondents from medium–large companies prioritize investigating the causes of an accident, both in general and to prevent the same accident from recurring at the Compeso company.

The results of the statistical test that revealed significant differences in terms of the level of implementation are presented in

Table 5. Significant differences in terms of level of implementation of systems or actions aimed at OSH considering the company size.

Action/System	RAG Dimension	Accident/Near Miss	Average Rank	Significance	Greater Level of Implementation
Update risk assessment for new technologies	Anticipate	-	gr_1) 146.44 gr_2) 132.56	0.007	Medium–large companies
Assess risks from extraordinary changes to the environment in which the technology is placed	Anticipate	-	gr_1) 152.00 gr_2) 127.00	<0.001	Medium–large companies
Notify staff of extraordinary changes to the environment where technology is located	Anticipate	-	gr_1) 149.14 gr_2) 129.86	0.007	Medium–large companies
Initiate an investigation into the causes of the event	Learn	Accident	gr_1) 103.44 gr_2) 89.56	0.003	Medium–large companies
Initiate an investigation into the causes of the event	Learn	Near miss	gr_1) 85.83 gr_2) 67.17		Medium–large companies
Increase the level of knowledge of technological equipment associated with the event	Learn	Accident	gr_1) 102.90 gr_2) 90.10	0.020	Medium–large companies
Increase the level of knowledge of technological equipment associated with the event	Learn	Near miss	gr_1) 81.91 gr_2) 71.09	0.031	Medium–large companies

.

Respondents from medium–large companies reported a higher level of implementation of systems to update the risk assessment related to new technologies, evaluate risks arising from extraordinary changes in the environment where the technology is present, and communicate such changes to the staff. Furthermore, they reported having more frequently conducted investigations into the causes of both accidents and near-miss scenarios and implemented actions aimed at increasing the level of knowledge about technological equipment, both in relation to accidents and near-miss scenarios.

4.4. Impact of New Technologies

The second variable used to create the groups refers to the introduction of one or more new technologies. Group_1 represents the number of companies that have not introduced any technology (178 observations), while Group_2 represents the number of companies that have introduced new technology (166 observations). Although there is a small difference between the number of observations in the two groups, they are still balanced by extracting a random sample of 166 observations from Group_1. This results in two equal groups of 166 observations, thus maximizing the benefits of balancing. The results of the statistical test that revealed significant differences in terms of perceived importance are presented in

Table 6. Significant differences in terms of perceived importance of systems or actions aimed at OSH, considering the introduction of new technologies.

Action/System	RAG Dimension	Accident/Near Miss	Average Rank	Significance	Greater Perceived Importance
Initiate an investigation into the causes of the event	Learn	Accident	gr_1) 154.58 gr_2) 178.42	0.004	New technology introduced

.

Respondents from companies that have introduced new technologies consider it more important to investigate the causes of an accident, compared to those from companies that have not implemented new technology.

The results of the statistical test that revealed significant differences in terms of the level of implementation are presented in

Table 7. Significant differences in terms of level of implementation of syst ems or actions aimed at OSH considering the introduction of new technologies.

Action/System	RAG Dimension	Accident/Near Miss	Average Rank	Significance	Greater Level of Implementation
System to notify entries without PPEs	Monitor	-	gr_1) 176.41 gr_2) 156.59	0.045	New technology introduced
System to stop the technology when anomalies are detected	Respond	-	gr_1) 187.45 gr_2) 145.55	<0.001	New technology introduced
Notify staff of extraordinary changes to the environment where technology is located	Anticipate	-	gr_1) 176.68 gr_2) 156.32	0.009	New technology introduced
Initiate an investigation into the causes of the event	Learn	Near miss	gr_1) 157.71 gr_2) 175.29	0.046	New technology introduced

.

Respondents from companies that have introduced new technologies report having implemented, to a greater extent than companies without new technology, systems for notifying entry without PPEs, systems to stop technology when anomalies are detected, and actions to communicate extraordinary changes in the environment where the technology is present. Furthermore, they also report having conducted more investigations into the causes of near-miss scenarios.

The results indicate that, in many cases, the assessments related to the model company did not align with the actual presence or use of systems within the companies. Additionally, both company size and technological implementation are key factors in determining both the perceived importance and the level of implementation of systems and actions reflecting RE principles.

5. Discussion

5.1. Consideration About Storytelling as a Methodology

In the context of Safety II, where the focus shifts from merely preventing accidents to understanding the variability in human performance and the conditions under which work processes unfold, storytelling proved to be particularly valuable. By presenting participants with narrative scenarios, in fact, this approach allows them to engage with the real-world complexities of workplace environments. The two different stories—one representing an accident and the other a near-miss scenario—encouraged interviewees to reflect on different facets of safety management and operational systems.

The scientific literature has explored—although not extensively—the use of storytelling across various domains, as a compelling approach to enhance health and safety issues, such as training, by embedding critical knowledge into emotionally resonant, context-rich narratives.

For instance, narrative case studies and “lessons learned” formats have demonstrated potential to foster reflection and decision-making among occupational safety professionals by anchoring abstract principles in real-world consequences [46]. Similarly, the Telling the Story Project in agriculture illustrates how emotionally charged, first-person narratives can elicit empathy, improve message retention, and disseminate prevention knowledge across vast geographic areas [47]. A similar pedagogical impact was found in virtual construction environments, where immersive storytelling improved the perception of fall hazards, offering engagement and learning efficiency comparable to OSHA training [40]. Storytelling also played a key role in health education within constrained or stigmatized contexts, e.g., facilitating the discussion about sensitive topics with children, highlighting the ability to mediate behavioral and attitudinal change across generations and domains [48]. In the workplace, stories collected by safety inspectors have proven to be valuable not only for training but also for developing validated recommendations within communities of practice, emphasizing storytelling as a tool for participatory safety policy development [49].

However, while these studies demonstrate the value of storytelling in transferring safety knowledge, most limit themselves to knowledge dissemination or qualitative learning impact. Some, such as those using accident chronologies in the Finnish metal sector [50] or narrative-enhanced instructional material in behavioral studies [51], touch on structured story analysis or experimental validation, but they do not deeply engage with systemic safety framework

In contrast, the present study offers an innovative methodological leap by integrating storytelling with the RAG principles and Safety II thinking. It extends beyond the narrative as a training artifact or awareness tool, employing it within a structured questionnaire to examine how organizations perceive and operationalize core resilience capacities. Rather than focusing solely on safety training outcomes, this research provides a diagnostic tool grounded in real scenarios to assess organizational safety culture and resilience maturity.

While prior works have already highlighted the emotive and persuasive power of storytelling [52], and its utility in shaping communication practices and interventions [53], none have combined narrative with RAG to systematically evaluate OSH practices across a large sample of companies. Furthermore, by situating this approach within Italy’s manufacturing sector and highlighting the role of technological adoption and company size, the study contributes a novel empirical perspective to the international literature on resilience engineering, an area that remains underdeveloped in non-academic industrial settings.

Therefore, this work moves it into a new analytical space by using storytelling not just to inform or persuade, but to structure inquiry and measure organizational capacities for resilient performance. This dual function—pedagogical and diagnostic—sets it apart from the existing literature and underlines its innovative contribution to occupational safety science.

Another key advantage regarding the adoption of storytelling in this study lies in its ability to provide context to theoretical concepts, grounding them in more relatable, concrete scenarios. Thus, it bridges the gap between abstract constructs, such as risk assessments or monitoring technologies, and the everyday experiences of workers. This contextual grounding supports a more nuanced understanding of how safety systems function—or fall short—under real-world conditions. Furthermore, when combined with the RAG model’s dimensions (Monitor, Respond, Anticipate, and Learn), storytelling offers an additional layer of insight. The narratives facilitate exploration of these dimensions from both organizational and individual perspectives, contributing to a more comprehensive picture of safety dynamics.

From an external viewpoint, safety actions are considered in terms of organizational practices, such as system implementation or technological upgrades. Meanwhile, the internal perspective relates these actions to the lived experiences of participants within their specific work contexts.

Although storytelling does not necessarily simplify complex issues for the sake of accessibility, several considerations are essential when applying it to safety research. One concern involves the risk of generalization: while narrative formats make complex ideas more approachable, subtle nuances may be lost in translation [54].

Additionally, the framing of stories can shape how participants respond, potentially emphasizing certain outcomes over others [55]. For instance, a story naturally emphasizes safety failures or successes, as in the case presented, possibly leading respondents to overemphasize certain safety aspects in their reflections.

Moreover, the effectiveness of storytelling also hinges on how closely participants relate to the scenarios presented [56]. While narratives aim to mirror real challenges, individual engagement depends on past experience and familiarity with similar situations. Finally, it is important to consider the cultural and organizational contexts of the participants. Storytelling is inherently shaped by cultural norms and organizational practices [57], which may affect how participants interpret and engage with the stories. In diverse settings, what is perceived as a safety risk or a successful safety practice in one organization might be interpreted differently in another. For this reason, the choice of the interviewee sample was limited to the Italian territory, as the researchers considered the possibility of encountering differences that could undermine the ability to draw meaningful conclusions.

Finally, the integration of storytelling with the RAG framework represents a novel methodological contribution to the field of resilience engineering and occupational safety. By embedding structured stories within the RAG-based questionnaire, this study bridges quantitative and qualitative paradigms, enabling participants to engage with complex safety scenarios through both analytical reasoning and emotional reflection. This hybrid approach aligns with recent research that calls for more human-centered and participatory methods in safety research [14,58]. In particular, storytelling offers a flexible and intuitive medium for exploring tacit knowledge, behavioral norms, and contextual cues that are often missed in traditional survey-based tools. As such, the method holds promise for application in diverse cultural and industrial contexts.

5.2. Implications for Occupational Safey and Health

The findings from this study have notable implications for the field of OSH, particularly in relation to the differences observed across company size and technological adoption. The results suggest that medium-to-large enterprises tend to adopt and apply OSH measures more systematically than smaller companies, both in general and following specific incidents. This trend underscores the importance of organizational capacity and available resources in determining the extent and effectiveness of safety system implementation. In contrast, smaller enterprises may face challenges related to limited resources, which can hinder their ability to implement robust safety measures and protocols [59,60]. Furthermore, small enterprises often face more hazardous work environments due to management constraints [61]. Typically, a single owner-manager is responsible for various non-production tasks such as sales, planning, human resources, and billing [62]. With survival as their main priority, OSH considerations often become secondary, as these companies face constraints in staffing, funding, and technical know-how. Smaller organizations may benefit from targeted interventions that provide cost-effective solutions to improve OSH outcomes [63]. This could include the implementation of low-cost safety technologies, standardized training programs, or government-supported initiatives aimed at helping small businesses adopt effective OSH practices.

Medium-to-large firms, by contrast, benefit from scale and resources that allow for more substantial investments in safety—such as structured training, advanced monitoring technologies, and robust systems to mitigate risk [64,65]. These organizations also tend to respond more proactively to near-miss events, reflecting a deeper safety culture and heightened awareness of potential hazards. This proactive stance is crucial. Near misses often act as early warning signs; addressing them swiftly can reduce the likelihood of severe accidents. Encouraging their reporting and analysis can strengthen a culture of learning and prevention.

Another critical finding that emerged from this study is the increased implementation of safety measures by companies that have adopted new technologies. The results emphasize the role of technological advancement in improving safety practices within organizations. Technologies designed to monitor and manage safety risks, such as real-time detection systems, wearable devices, and automated protocols, enhance hazard awareness and support timely intervention. Their adoption promotes a more responsive and adaptive safety approach, especially in dynamic work environments where conditions shift quickly [66]. For firms yet to invest in these tools, the findings point to substantial benefits in doing so.

Lastly, the heightened sensitivity to near-miss incidents observed in medium–large enterprises suggests a growing recognition of the value of near-miss scenario reporting in enhancing safety management [67]. Although often underestimated, near-miss scenarios are key indicators of system vulnerabilities. Fostering a culture that values their reporting can lead to significant improvements in both prevention and preparedness. Small businesses, in turn, could benefit from adopting similar approaches, creating an environment in which near-miss scenarios are viewed as opportunities for improvement rather than failures to avoid.

5.3. Study Limitations

The study presents several limitations that should be considered when interpreting the results.

First, its exclusive focus on Italian companies limits the generalizability of the findings to other geographic or cultural contexts. OSH practices and the adoption of technologies can vary significantly across countries or regions due to differences in regulations, cultures, and contextual factors. Moreover, even within the Italian context, the regional and sectoral distribution of participating companies was not balanced, with a predominance of responses from Northern Italy and underrepresentation from certain ATECO manufacturing sectors.

In addition, the use of questionnaires—while practical and efficient—has inherent drawbacks. Although the sample examined is relevant in terms of size and target population, the reliability of the responses may be affected by potential response biases. In particular, the use of self-reported data may be subject to social desirability and recall biases, as respondents might overstate positive behaviors or misremember past events. To help reduce these effects, the study used a storytelling approach with realistic scenarios, encouraging reflection on external cases rather than direct self-assessment. While this method lessens some bias by distancing respondents from direct self-reporting, the limitations of self-reported data remain and should be acknowledged. Moreover, the storytelling approach itself, although innovative and engaging, introduces a potential risk of generalization. The stories, designed to capture typical safety scenarios, may not encompass the full complexity and diversity of safety dynamics experienced across different organizations or settings. As such, there is a possibility that respondents’ interpretations and answers based on these scenarios do not fully reflect real-world variations or subtleties in safety practices. Furthermore, the closed-ended questions used in the survey may limit the respondents’ ability to fully express their opinions and experiences, reducing the depth of the collected data.

Another limitation concerns the focus of the stories provided, which concentrate exclusively on accidents and near-miss, relatively restricted events. As a result, perceptions of safety practices may be influenced by the non-ordinary nature of these events. The findings might therefore not fully reflect day-to-day safety management, which is equally important for the continuous improvement of OSH practices. A further limitation is the lack of historical context regarding the frequency and nature of past incidents. Including data on previous accidents or near-miss scenarios might have provided a deeper understanding of how companies responded to real events and whether these were linked to new technology adoption.

Furthermore, the study did not thoroughly examine how safety procedures can differ among various industrial sectors. Given the diversity of sectors included in the sample, sector-specific analysis was limited by insufficient data granularity. Safety dynamics and priorities can differ significantly between high-risk industries, such as chemical or mining sectors, and lower-risk industries, such as services or retail. These aspects were not sufficiently explored, limiting the understanding of safety practices in different contexts.

Finally, the study focused on only 15 safety actions or systems, which might not cover the entire spectrum of measures that could influence OSH. Other relevant factors, such as safety culture and leadership, were not analyzed in depth, leaving potentially crucial aspects unexplored.

While these limitations do not invalidate the study’s results, they should be considered when interpreting the conclusions and implications for OSH within organizations.

6. Conclusions and Future Directions

This work presents a storytelling-based cross-sectional study involving 344 Italian manufacturing companies to assess the level of perception and implementation of RE principles and practices.

In response to RQ1, it shows that the dimensions of Monitor and Respond are perceived as more important compared to Anticipate and Learn, reflecting a more reactive rather than proactive approach in terms of OSH. This predominance is also confirmed in addressing RQ2. Additionally, the study shows, in response to RQ3, the presence of significant differences. Small companies, in fact, attribute more importance to certain aspects of the Monitor and Anticipate dimensions when dealing with accidents. On the other hand, medium–large companies attribute more importance to other aspects of the Anticipate and Learn dimensions, both when dealing with accidents and near misses.

These differences are closely linked to the unique challenges and capacities of companies based on their size. Small enterprises often face resource constraints that limit their ability to implement complex monitoring systems or provide extensive training programs. For such organizations, practical strategies might include the adoption of cost-effective environmental monitoring tools or simplified risk assessment methods. In contrast, larger companies are better positioned to invest in advanced technologies, such as real-time monitoring systems, automated safety protocols, and comprehensive training initiatives. Policymakers can play a key role in supporting smaller firms, for instance, through financial incentives like the ISI grants provided by INAIL [67], which facilitate the adoption of essential safety measures.

Technological innovation, examined in RQ4, also emerges as a critical factor. In terms of perceived importance, companies that have implemented new technology assign greater importance to the Learn dimension in the case of accidents. Furthermore, in terms of implementation level, actions and systems related to the Respond, Anticipate, and Learn dimensions are more enhanced in companies that have introduced technological innovation.

Future research could build on these findings in several important directions. One crucial direction involves adopting a longitudinal approach to monitor how safety practices and technology use evolve. Given that OSH is a dynamic process, studying long-term effects, especially after incidents, could reveal the sustainability and impact of safety interventions.

Expanding the scope to include international comparisons could shed light on how regulatory frameworks, cultural values, and technological contexts shape safety behavior.

Additionally, the further exploration of the cultural and behavioral aspects of safety in organizations could prove insightful. Specifically, examining the role of leadership, organizational culture, and daily safety practices could offer a more comprehensive understanding of how company values shape safety measures and responses to accidents or near-miss scenarios.

Furthermore, comparing safety practices across different sectors could highlight industry-specific needs and offer tailored approaches to risk management.

Emerging technologies also present a promising avenue for future study. Studying the role of AI, augmented reality, and real-time monitoring in improving OSH could help assess their effectiveness in preventing incidents and managing risk.

These directions would provide a more holistic view of OSH management and technological adoption, paving the way for more effective strategies that enhance worker OSH across diverse contexts and industries.

In conclusion, the study’s findings underscore the importance of company size and technological adoption in shaping OSH practices, suggesting that this aspect still represents a limitation for OSH, as all organizations, regardless of size, should consider investing in innovative safety solutions. Finally, cultivating a culture that recognizes and responds to near-miss scenarios remains essential for preventing serious accidents and fostering continuous safety improvement.