A Hybrid Approach to Identifying Key Indicators for Enhancing Occupational Health and Safety Systems in Small Window- and Door-Fitting Companies

Abstract

Workplace injuries are a significant issue in the construction industry. The efficiency and effectiveness of the OHS system can be analysed using safety indicators. This paper presents a methodological procedure for the selection and ranking of indicators to improve the OHS system. The procedure consists of two phases: preparatory and execution. In the preparatory phase, a hierarchical structure was proposed, consisting of three aspects, four factors, and thirty-nine indicators. Additionally, experts were selected. In the execution phase, experts applied the weighted Borda method to select five key indicators for each factor. The selected indicators were then compared and ranked using the group fuzzy analytic hierarchy process (FAHP). The influence of experts in the selection and ranking of indicators was determined according to their previous experience. The procedure was applied to analyse OHS systems in small window- and door-fitting companies. Among the indicators, experts highlighted the efficiency of OHS resource management, the frequency of coordinating activities with contractors, the communication capacity of the workers, the analysis of the results of external OHS controls, the number of external funds for OHS system improvements as well as the number and assessment of incidents. Furthermore, ten scenarios were analysed with varying values of factor weight values. The scenarios demonstrated that the highlighted indicators were consistently ranked among the top ten indicators.

1. Introduction

The construction industry is a significant driver of societal development. It involves significant investments and the engagement of personnel and equipment, creating positive effects across the entire business environment including both the economy and the service sector. Numerous employees are involved in a variety of tasks. Many of these tasks are physically demanding and are often performed under extreme conditions and tight deadlines.

Occupational injuries remain a significant problem in the construction industry worldwide. In 2022, 19.5% of all fatal occupational injuries in the USA occurred in the construction sector [1]. Similarly, in the European Union, 22.9% of fatal injuries during the same year were related to construction work [2]. The results are similar in other countries [3], so it can be concluded that occupational injuries in the construction industry represent a significant social and economic loss [4,5]. An additional problem can arise with employees from abroad [6] due to cultural and risk perception differences. Despite numerous efforts to improve the legislation and implement modern safety and health measures in practice [7], the outcomes primarily reflect changes in the volume of construction work.

Numerous researchers have focused on analysing occupational health and safety (OHS) systems in construction companies to identify opportunities for improving their effectiveness and efficiency. The identified problems often relate to the implementation and organisation of OHS systems as well as the monitoring of changes in work processes, construction technologies, and safety measures.

Craftsmanship and installation works are carried out in a later phase of construction. Unlike the primary stages of construction, they do not expose workers to such great dangers. However, this does not mean that employees are completely safe, and that injuries and fatalities cannot occur. Working procedures must still be strictly followed, and the prescribed safety measures must be applied.

The management of complex systems requires access to relevant data. Companies can collect the data using objective measurements in the system and its environment. Additionally, different experts can evaluate data on the system and its functioning. Expert assessment is an additional layer for system analysis, which enables the identification of risks and potential for system improvement. The safety system, viewed as a complex entity, is no exception. Its manageability can be improved through the introduction of a dedicated management system, which facilitates better monitoring of the results and ensures continuous improvement.

Implementing such systems often demands significant resources that smaller organisations may lack. Consequently, identifying key indicators describing the safety system has become increasingly important. Key indicators are those with the greatest potential impact on the improvement of the existing system. Engaging experts with appropriate knowledge and experience further improves the credibility of the results. By incorporating uncertainty into the decision-making process, which is characteristic of occupational safety risk assessment, a robust mechanism can be developed. This mechanism should enable the analysis and improvement of safety systems, ensuring their effective response to challenges.

Related Work

The impossibility of completely eliminating adverse consequences in the construction industry calls into question various concepts that promote their complete elimination such as the zero-vision concept [8]. It is more practical to have a manageable system with foreseeable potential risks, rather than strive for something that is unattainable in practice due to both subjective and objective reasons.

The size of an organisation and its available resources significantly affect its functioning both in terms of its core business operations and OHS. Although OHS is considered a comprehensive concept, in practice, smaller companies primarily focus on occupational safety, ensuring the optimal working conditions and eliminating risks to which employees are exposed, or implementing appropriate measures to reduce those risks to an acceptable level. However, ensuring physical and mental well-being as well as analysing and preventing occupational disease, which fall under the domain of occupational health, requires additional resources that smaller organisations often lack. Organisational resources serve as a limiting factor in the risk reduction process, so careful attention should be paid to the adequate allocation of available safety resources [9], particularly in smaller companies [10]. The same applies to safety reporting, where larger companies tend to have much better access to OHS data compared with smaller companies. Different management practices can be applied to address safety issues across industries [11]. A significant improvement in the reporting process is achieved when appropriate management systems are established [12].

Various methods can be applied for the assessment [13], quantification, and prediction of OHS risks to which construction workers are exposed [14]. During such analyses, the risk factors contributing to unsafe behaviours among workers in the construction sector are identified. In [15], the authors identified three key dimensions including primary factors: the managerial, individual, and organisational dimensions. With more detailed analysis, the authors particularly emphasised the negative effects of an inadequate organisational climate. An inadequate perception of danger and the possible consequences as well as the overestimation of one’s own capabilities are frequent causes of adverse outcomes in the construction industry [16,17]. Fatal accidents can be avoided by establishing an appropriate safety climate, providing adequate training [16], and implementing effective barrier management [18].

A crucial aspect of OHS in construction is the employees’ safety behaviours. These behaviours can be encouraged through corresponding safety perceptions among managers and effective management practices aimed at shaping what safe work is exactly [19,20]. The researchers emphasize the importance of developing employee competence and safety awareness as a means to improve the performance of safety systems.

Specific work activities in the construction industry are classified as physically demanding based on the workload involved in performing the tasks. Ergonomic methods can help solve these problems by analysing physically demanding work activities and proposing measures to reduce workplace discomfort and prevent musculoskeletal disorders. Employees may also face additional burdens from different environmental conditions [21]. When analysing accidents, it is important to identify the causes of adverse events. Working in specific conditions, such as confined spaces, often reveals atmospheric and physical hazards as frequent causes of fatal incidents [17]. Monotony, tight deadlines, and work dynamics can cause various negative effects on employees including reduced concentration when performing their daily work activities, failure to follow work procedures, and non-compliance with the prescribed safety measures and personal protective equipment (PPE) protocols.

Technological advancements have introduced various tools to facilitate work activities and improve hazard identification in the workplace. The construction sector has not been exempt from this progress [22]. However, the extent to which employees are willing to adopt advanced tools in their daily tasks and the potential challenges such equipment may pose during routine work activities remain a topic for further research.

Occupational safety indicators can either be qualitative or quantitative in nature. These are based on measurements in the work environment or expertly defined evaluations and are used to determine whether a company, its organisational unit, or an individual meets the desired goals, which may be independently or legislatively defined. These indicators serve as the foundation for monitoring the implementation of safety plans and programs and act as a means in which to initiate preventive or corrective actions within the OHS system.

Monitoring a large number of performance indicators is challenging and requires significant resources that smaller companies often lack. Therefore, it is convenient to select key indicators. Classifying these indicators into thematically related groups and arranging them hierarchically makes them suitable for ranking by means of multi-criteria analysis methods. Analytic hierarchy process (AHP) is often used for this purpose because of its hierarchical approach to problem analysis. It describes the decision-making process using objectives, aspects, and alternatives. By comparing pairs of aspects or alternatives, their relative importance is determined using numerical values from the Saaty scale. These values form the basis for obtaining the weights of aspects and ranking the alternatives. In occupational safety within the construction sector, AHP has been applied to tasks such as identifying risks during work planning [23] and construction site organisation [24], assessing the potential safety incidents [25,26], defining the safety management system priorities [27], and reallocating the available safety resources [28].

However, it is not always possible to compare aspects or alternatives using precise numerical values, as required by AHP. To address uncertainty, Zadeh’s fuzzy logic and fuzzy numbers (trapezoidal or triangular) are used, resulting in fuzzy AHP (FAHP). This approach increases complexity, as it requires applying fuzzy calculations and subsequently defuzzifying the obtained fuzzy weights to derive crisp values for final comparisons and rankings. The use of fuzzy numbers enables comparisons using linguistic variables to describe ranges of values. FAHP has been used in occupational safety in construction, either independently or in combination with other methods, for tasks such as analysing the safety risks on construction sites [29], identify accident causes [30], assessing safety conditions [31], and defining priorities or selecting key indicators [32,33].

This paper presents a hybrid approach for selecting key indicators to improve the OHS system. The approach combines the expert selection of key indicators of occupational safety with their ranking using FAHP. The FAHP method was chosen because the proposed classification of indicators has a hierarchical nature, similar to the decision-making hierarchy used in FAHP. The process involves the pairwise comparison of elements using a fuzzy scale, which accounts for uncertainty. Additionally, both qualitative and quantitative indicators can be considered during ranking, further supporting the choice of this method. Based on the defined classification, the procedure was applied to identify the key indicators for improving the OHS systems in small construction companies.

The methodological approach is described in Section 2, while Section 3 demonstrates the application of the hybrid approach in the case of small businesses involved in window- and door-fitting activities. The results are discussed in Section 4, followed by the concluding remarks.

2. Materials and Methods

The performance of the OHS system is presented and analysed using various indicators. The complexity of the OHS system as well as diverse internal and external factors influencing its functioning, necessitates a careful selection and classification of indicators into groups describing the corresponding influential factors.

The research presented in this paper was based on a methodological procedure designed to identify the influence of indicators on the functioning of the OHS system using multi-criteria analysis methods. Figure 1 presents the methodological procedure. This procedure is supported by earlier research into various factors influencing the OHS system including the environmental factor [32], the organisational factor [34], the human factor [32], and the technological factor [35]. These factors were examined within the framework of social acceptability, risks, and costs.

The procedure consists of two main phases: a preparatory phase and an execution phase. During the preparatory phase, the indicators are analysed, and the initial classification is defined. Potential experts are then contacted, and those who agree to participate in the analysis are given the opportunity to review the initial proposal and suggest potential modifications. The influence of the experts is determined based on their previous experience, and this influence is presented by corresponding weights. Following this, the execution phase is carried out.

The execution phase consists of two parts: the selection of key indicators and their ranking. The selection of key indicators is carried out based on the experts’ weights using Borda voting. The selected key indicators are obtained as results, ensuring an equal number from each group of indicators that describe the relevant influencing factors. This approach avoids bias in favour of certain groups of indicators over others.

Experts receive the results of the key indicator selection, along with the defined hierarchical structure. Following this, the experts evaluate the selected key indicators and influential factors, comparing them in pairs according to the levels of the hierarchical structure using linguistic variables. These linguistic variables represent the corresponding fuzzy numbers. The comparisons defined in this manner are used in the FAHP. Once the necessary calculations with aggregated fuzzy matrices are completed, the weights of the influential factors and key indicators are determined, and the ranking is performed. The results are then presented to the experts. If the experts agree with the results, potential proposals for improving the OHS system are articulated based on the best-ranked key indicators. Optionally, various strategies for improving the OHS system may be defined, and their effects on the rank of key indicators are analysed.

Details of the aforementioned steps are presented in the following sections.

2.1. Factors and Indicators

When forming the hierarchy of aspects and indicators, the principle of decomposition is applied. The idea is to form a general hierarchy, while the specificities of the analysed OHS systems may require the introduction of additional indicators to address them. The multi-level structure consists of the OHS objective at the root, aspects at the first level, factors at the second level, and indicators at the leaves of the tree. The structure facilitates the application of AHP and related methods of multi-criteria analysis. The proposed conceptual hierarchical decision-making structure, based on previous research [32,34,35], is shown in Figure 2.

Aspects define a strategic approach to addressing OHS challenges. The primary consideration is the risks that the employees face during daily work activities, evaluated based on the level of danger and the duration of exposure. In addition, the social perception of preventing injuries, fatalities, and occupational diseases as significant losses for both individuals and the community influences the definition of the OHS goals and the initiation of organisational activities. Improving an OHS system also requires financial and material resources, which can be a considerable burden for small companies.

At the next level of the hierarchy are the factors. The proposed factor structure and classification of indicators are shown in

Table 1. Factors and indicators.

Factors	Indicators	References
Environment (E)	(E1) Level of legal procedures implemented	[32,36,37,38,39]
	(E2) Number of external funds for OHS system improvement
	(E3) Level of networking with related companies in communities of practice
	(E4) Frequency of coordinating activities with contractors
	(E5) Level of implementation of standards and recommendations
	(E6) Frequency of following technological solutions for OHS system improvement
	(E7) Frequency of accessing external OHS data sources
Technology (T)	(T1) Probability of failure (reliability)	[34,35,40,41]
	(T2) Failure frequency (intensity)
	(T3) Assessment and quantification of accidents
	(T4) Number and assessment of incidents
	(T5) Availability
	(T6) Maintenance rate
	(T7) Maintenance costs
	(T8) Criteria for selecting the safety level
	(T9) Mean time between repair or failure
Organisation (O)	(O1) Analysis of the results of external OHS controls	[34,42,43,44,45]
	(O2) Effectiveness of OHS resource management
	(O3) Evaluation of the effects of OHS training (success and retention)
	(O4) Number (percentage) of higher-risk work activities
	(O5) Causes and consequences of adverse events and stressful situations
	(O6) Average number of working hours of OHS training
	(O7) Average years of service (work experience)
	(O8) Number of employees trained in first aid
	(O9) Workplace safety assessment (higher-risk activities)
	(O10) Number of OHS guidelines
	(O11) OHS training plan by activity
	(O12) Evaluation of the applicability of safety measures
Workers (K)	(K1) Number of workplace injuries	[35,41,46,47,48,49]
	(K2) Level of employees‘ communication competence
	(K3) Level of compliance with work procedures
	(K4) Degree of employee satisfaction with safety and job conditions
	(K5) Number of mistakes and errors during work
	(K6) Index of skills and competencies
	(K7) Presenteeism
	(K8) Absenteeism
	(K9) Assessment of work activities and safety measures implementation
	(K10) Degree of creativity during work
	(K11) Degree of satisfaction with the PPE used

. The OHS system does not operate in isolation but is constantly affected by changes that occur in its environment. Improvements in legal regulations provide a significant incentive for OHS system enhancements [36,37], along with technological advancements [38], the adoption of new industry standards, and knowledge exchange in communities of practice [39]. Access to external funds can significantly help small companies overcome financial constraints.

Technological progress is a major driver of development across various industrial sectors. Technological innovations are increasingly applied to workplace safety, both to monitor production processes and assess working conditions [40]. For such systems, factors like sensitivity in detecting irregularities and reliability are crucial. However, for smaller companies, the costs of purchasing and maintaining this equipment must also be taken into consideration. Training can be more effective if modern technologies are applied [41].

Organisational characteristics significantly affect OHS systems as they represent the basis for their implementation, ranging from adequate leadership style and risk management [42] to safety resource allocation [43], performance analysis [44], and the establishment of a safety culture [45]. Often, not without reason, they are considered key prerequisites for an effective OHS system.

Regardless of how well-designed the OHS system is, its successful implementation depends on the activities of both the employers and employees. Their attitude towards protection and safe work practices depends on education and training [41,46], risk perception [47], leadership abilities, problem-solving skills, and responsiveness in critical situations [48,49].

All of these factors and indicators affect the success of the organisational system, both in terms of the core business operations and supporting activities. Safety, as a supporting activity, defines the specific needs and priorities that, within the organisation, should not be viewed as an additional cost, but rather as a foundation for successful functioning.

This paper analysed OHS systems in small window- and door-fitting companies. The activities of employees in window- and door-fitting companies include installing doors, windows, and accompanying equipment such as gutters, window-sills, blinds, and mosquito nets. They also perform construction work including the dismantling of old windows and doors and the proper installation of new ones. This process includes transporting waste materials to a landfill, applying a smooth coating around doors and windows, and ensuring a clean and precise finish on the interior walls. After being installed, the windows and doors must be fine-tuned for their proposed functionality. This job requires a lot of technical knowledge and precision but is preceded by physically demanding activities such as positioning and removing heavy elements during replacements. Working conditions, and the risks that employees face, can vary significantly depending on whether the work is performed in a newly constructed object (Figure 3) or an existing building (Figure 4).

2.2. Experts

Experts select key indicators and determine their ranks. The selection of experts is based on their professional background and previous work experience. A key prerequisite is active participation in the OHS system or risk assessment within the analysed activity. An additional requirement is at least five years of work experience in the field of OHS in general, and at least three years in a specific activity (in our case, the construction industry). Previous experience forms the basis for determining the weights used during group decision-making.

Let ϰ be the number of elements (aspects or indicators) being compared and m be the number of experts participating in the comparison. The experience coefficient φ_i, which represents the influence of the i-th expert, is defined as

$${\varphi }_i=\left(r_i\cdot c_i\cdot t_i\right)/{\displaystyle {\sum }_{j=1}^m\left(r_j\cdot c_j\cdot t_j\right)},$$

(1)

where i = 1, …, m; r_i describes experience in risk assessment; c_i represents experience in risk analysis within the considered activity, and t_i is the total work experience of the i-th expert. An appropriate scale is used to evaluate the experience, as shown in

Table 2. Scale for evaluating the experts’ experience.

Experience x [Years]	Value (αi or βi)
x ≤ 3	1
3 < x ≤ 5	2
x > 5	3

.

For the weight vector of the experts, Φ = {φ₁, φ₂,…, φ_m}, the following applies: ${\varphi }_i>0$, ${\displaystyle {\sum }_{j=1}^m{\varphi }_i}=1$.

2.3. Obtaining Key Indicators

Expert weights (φ_i) are used when determining key indicators. The Borda rule is applied to compare alternatives. For each indicator rank, a corresponding integer value is assigned. If the expert assigns the k-th rank for the indicator I_i, the Borda count for that choice is b(I_i)= k − 1, while the worst alternative, according to the expert’s opinion, receives the value b(I_j) = 0.

This approach of ranking the alternatives helps avoid confusion in determining preferences that may arise during pairwise comparisons due to transitivity. The Borda count reflects the recommendations of the experts, which are then combined by weighting based on each expert’s experience coefficient. The weighted Borda rule is:

$${\rm B}\left(I_i\right)={\displaystyle {\sum }_{k=1}^{\varkappa }{\varphi }_k\cdot b\left(I_i\right)}/{\displaystyle {\sum }_{k=1}^{\varkappa }{\varphi }_k},$$

(2)

where φ_k is the experience coefficient of the k-th expert, and ${\displaystyle {\sum }_{k=1}^{\varkappa }{\varphi }_k}=1$. The value of B(I_i) is used to select the key indicators from each group, with a lower value indicating lower importance of the indicator.

The problem of reaching consensus among experts within the group requires further analysis before selecting key indicators. To address this, ordinal consensus is applied, which, based on [50], is assessed by determining the measure

$${\rho }^{\left(k\right)}=1-\left({\displaystyle {\sum }_{i=1}^m\left|r\left(b_i^{\left(k\right)}\right)-r\left(B_i\right)\right|}\right)/\left(0.5\ast m\ast \left(m+1\right)\right),m>2,$$

(3)

where r(b_i^(k)) represents the rank assigned to indicator I_i by expert E_k by using the corresponding Borda count, while r(B_i) is determined based on the weighted Borda count for the group of experts. The degree of satisfaction of the group is calculated as

$$\rho =AVG\left({\rho }^{\left(1\right)},\dots ,{\rho }^{\left(k\right)},\dots ,{\rho }^{\left(m\right)}\right),$$

(4)

that is, as the arithmetic mean of the ρ⁽ⁱ⁾ values for individual experts. This value serves as a measure of the experts’ consensus, indicating the level of agreement in their rankings. A higher ρ value implies a greater agreement among experts. This value should be at least 0.70.

2.4. Ranking Key Indicators

The fuzzy AHP method is used to determine the weights of factors and rank key indicators. Fuzzy numbers incorporate uncertainty in the decision-making process, making it more realistic. Matrix representation and pairwise computations are conducted using fuzzy numbers. In this analysis, triangular fuzzy numbers $\tilde{C}=\left(c_1,c_2,c_3\right)$ with the membership function μ(x), shown in Figure 5, were applied. This function maps $\left(-\infty ,\infty \right)$ to $\left[0,1\right]$.

Basic operations on triangular fuzzy numbers $\tilde{C}=\left(c_1,c_2,c_3\right)$ and $\tilde{D}=\left(d_1,d_2,d_3\right)$ are defined as follows:

$$\tilde{C}\oplus \tilde{D}=\left(c_1+d_1,c_2+d_2,c_3+d_3\right),$$

(5)

$$\tilde{C}\otimes \tilde{D}=\left(c_1\cdot d_1,c_2\cdot d_2,c_3\cdot d_3\right),$$

(6)

where $\oplus$ represents the addition of fuzzy numbers, while $\otimes$ represents the multiplication of fuzzy numbers. Additionally, multiplying by a constant crisp value k is defined as

$$k\cdot \tilde{C}=\left(k\cdot c_1,k\cdot c_2,k\cdot c_3\right),$$

(7)

while the reciprocal of the fuzzy number is determined as

$${\tilde{C}}^{-1}=\left(c_3^{-1},c_2^{-1},c_1^{-1}\right).$$

(8)

When making comparisons, experts use the fuzzified scale, as shown in Figure 6. The fuzzy number (1,1,3) represents the equality of alternatives, while (7,9,9) represents absolute dominance. Other basic values of the scale are characterised by the fuzzy distance δ = 2, and can be generally represented as (c − δ,c,c + δ).

If necessary, during pairwise comparisons, intermediate values can be used, for which δ = 1. The previous figure also shows the fuzzy values for weak (W), strong (S), and demonstrated (D) dominance.

Table 3. Scale for comparing elements.

Fuzzy Value	Notation	Crisp Value	Meaning
(7,9,9)	A	9	Absolutely dominant
(5,7,9)	D	7	Evidently better
(3,5,7)	S	5	Significantly better
(1,3,5)	W	3	Slightly better
(1,1,3)	E	1	Equally important

shows the comparison scale.

Experts at each level perform pairwise comparisons of elements. Expert k describes preferences using a matrix $F^{(k)}={\left[{\widetilde{f_{ij}}}^{(k)}\right]}_{ n x n}$, whose dimensions correspond to the number of elements, with standard characteristics of the pairwise comparison matrix ${\widetilde{f_{ij}}}^{(k)}={\left({\widetilde{f_{ji}}}^{(k)}\right)}^{-1}$ and ${\widetilde{f_{ii}}}^{(k)}=1$. Individual preferences are aggregated using the experience coefficient φ_k

$$\widetilde{f_{ij}}={\displaystyle {\prod }_{k=1}^{\varkappa }{\left({\widetilde{f_{ij}}}^{(k)}\right)}^{{\varphi }_k}},$$

(9)

where $\widetilde{f_{ij}}$ is an element of the fuzzy aggregate matrix.

In order to fulfil the consistency condition, the CCI (centric consistency index) is applied, with an acceptable value of less than 0.35 for comparing four elements, and less than 0.37 for comparing more elements [51]. For these values, it can be considered that the experts’ comparisons are consistent, and that the aggregate matrix is acceptable.

Based on the aggregate matrix, fuzzy weights

$$\widetilde{{\omega }_j}={\left({\displaystyle {\prod }_{l=1}^{\varkappa }\widetilde{f_{il}}}\right)}^{1/\varkappa }/\left({\displaystyle {\sum }_{i=1}^{\varkappa }{\left({\displaystyle {\prod }_{l=1}^{\varkappa }\widetilde{f_{il}}}\right)}^{1/\varkappa }}\right)$$

(10)

are determined for ϰ different compared elements, where j = 1, …, ϰ.

Weights are determined by levels, namely, for aspects

$${\omega }_A=\left({\omega }_{A1},{\omega }_{A2},{\omega }_{A3}\right),$$

(11)

where, based on [52],

$${\omega }_{Ai}={\displaystyle \sum _{i=1}^3\widetilde{f_{ij}}}\otimes {\left({\displaystyle \sum _{i=1}^3{\displaystyle \sum _{j=1}^3\widetilde{f_{ij}}}}\right)}^{-1},$$

(12)

where i = 1, 2, 3; then, for factors in relation to aspects,

$$p_{ij}={\displaystyle \sum _{j=1}^3\widetilde{f_{ij}}}\otimes {\left({\displaystyle \sum _{i=1}^4{\displaystyle \sum _{j=1}^3\widetilde{f_{ij}}}}\right)}^{-1},$$

(13)

$${\omega }_F=P\otimes {\omega }_A=\left({\omega }_{F1},{\omega }_{F2},{\omega }_{F3},{\omega }_{F4}\right),$$

(14)

where p_ij represents the fuzzy assessment of the influence of the i-th factor on the j-th aspect. The final weight of the indicator is obtained by multiplying the local weight of the indicator and the final weight of the factor, that is

$${\omega }_k^p=\left({\displaystyle \sum _{j=1}^{mk}\widetilde{f_{ij}}}\otimes {\left({\displaystyle \sum _{i=1}^{mk}{\displaystyle \sum _{j=1}^{mk}\widetilde{f_{ij}}}}\right)}^{-1}\right)\otimes {\omega }_{Fk},$$

(15)

where k = 1, …, 4, and p = 1, …, m_k. The indicator weights are the elements of the vector

$$W_I=\left({\omega }_1^1,\dots ,{\omega }_1^{m1},{\omega }_2^1,\dots ,{\omega }_2^{m2},{\omega }_3^1,\dots ,{\omega }_3^{m3},{\omega }_4^1,\dots ,{\omega }_4^{m4}\right).$$

(16)

In the presented research, there are four factors, each with five selected key indicators (i.e., m₁ = m₂ = m₃ = m₄ = 5). Their crisp weights are determined by defuzzifying the fuzzy weights $\widetilde{{\omega }_j}={({\omega }_L,{\omega }_M,{\omega }_U)}_j$, calculating the arithmetic mean

$${\omega }_j=AVG\left\{{\omega }_L,{\omega }_M,{\omega }_U\right\},$$

(17)

and then normalising the obtained values

$${\omega }_j^{\ast }={\omega }_j/\left({\displaystyle {\sum }_{k=1}^{\varkappa }{\omega }_k}\right)$$

(18)

to ensure that their sum equals one. Based on the obtained normalised values, the element ranks are determined.

3. Results

The analysis of the OHS system according to the presented methodology was conducted in collaboration with experts who had assessed risks or worked as OHS officers in small window- and door-fitting companies. The idea for the research arose from interactions with individuals engaged in window- and door-fitting activities, and their personal insights into potential OHS difficulties. The study was focused on southern Serbia, more specifically on the Nišava region.

In 2023, a total of 1348 apartments with more than 89,000 m² were built in the city of Niš, the centre of this region. This represents a 42.4% increase compared with 2022 (776 completed apartments), and a 28.5% increase compared with the five-year average (963 apartments per year). In addition, 2714 apartments remain under construction. The increased construction activity was influenced by the renovations of existing objects, partly caused by Serbia’s national program of energy rehabilitation of residential homes and apartments, implemented by the city of Niš. As part of this program, the replacement of external doors, windows, and other transparent elements of the thermal envelope is a key measure. According to the program demands, the external elements must meet specific thermal performance properties: U ≤ 1.3 W/m²K for glazed windows, balcony doors, and their profiles, and U ≤ 1.6 W/m²K for doors to unheated spaces, where U represents the heat transfer coefficient. This measure covers part of the work regularly performed by employees of window- and door-fitting companies, leading to an increased workload.

For this analysis, ten OHS experts whose work was related to small window- and door-fitting companies were contacted in September 2024. One did not meet the minimum requirements, and another did not want to participate. Eight experts took part—five male (62.5%) and three female (37.5%). All experts had completed at least a four-year undergraduate degree, while 50% held master’s degrees. Their experience in risk assessment was represented by the vector R = {2,3,2,2,3,2,3,3}, experience in risk analysis in construction by C = {2,2,2,2,1,2,1,2}, and total (work) experience by T = {3,3,3,3,3,2,3,3}. Based on these values, according to Equation (1), the φ_i coefficients of experience were determined for the eight experts (I–VIII), forming the weight vector Φ = {0.12,0.18,0.12,0.12,0.09,0.08,0.09,0.18}.

The experts did not suggest the inclusion of additional indicators in the initial list (Table 1), so it was used to select the key indicators. Consensus among experts, based on Equation (4), was achieved for the group’s degree of satisfaction at ρ ≥ 0.75. Key indicators (KIs) were selected for four groups of factors: environment (

Table 4. Environment—KI selection (ρ = 0.786).

Indicators	I	II	III	IV	V	VI	VII	VIII	B	Rank
E1	5	4	6	5	5	7	4	5	1.990	5
E2	3	1	3	2	4	1	3	1	4.929	2
E3	1	2	1	4	2	3	2	3	4.735	3
E4	2	3	2	1	1	2	1	2	5.122	1
E5	6	6	4	3	3	4	6	4	2.418	4
E6	4	7	5	7	6	6	7	7	0.786	7
E7	7	5	7	6	7	5	5	6	1.020	6

), technology (

Table 5. Technology—KI selection (ρ = 0.797).

Indicators	I	II	III	IV	V	VI	VII	VIII	B	Rank
T1	5	2	6	5	5	7	4	5	4.357	5
T2	7	9	7	7	7	9	8	8	1.194	8
T3	2	1	1	4	2	3	2	3	6.796	2
T4	3	4	2	1	1	2	1	2	6.816	1
T5	1	7	4	3	3	4	6	4	4.847	4
T6	6	8	8	6	9	5	5	6	2.286	7
T7	9	6	9	8	8	8	9	9	0.847	9
T8	4	3	3	2	4	1	3	1	6.439	3
T9	8	5	5	9	6	6	7	7	2.418	6

), organisation (

Table 6. Organisation—KI selection (ρ = 0.788).

Indicators	I	II	III	IV	V	VI	VII	VIII	B	Rank
O1	4	3	2	1	1	2	1	2	9.878	2
O2	2	1	3	2	4	1	3	1	10.051	1
O3	5	6	4	3	3	4	6	4	7.541	4
O4	3	4	6	5	5	7	4	5	7.235	5
O5	1	2	1	4	2	3	2	3	9.735	3
O6	7	12	9	6	9	5	5	6	4.306	7
O7	12	9	8	10	8	10	12	10	2.184	11
O8	8	11	12	9	10	11	10	11	1.673	12
O9	11	7	10	8	12	8	9	7	3.296	8
O10	10	5	5	11	6	6	7	9	4.561	6
O11	9	10	7	12	11	12	11	8	2.265	10
O12	6	8	11	7	7	9	8	12	3.276	9

), and workers (

Table 7. Workers—KI selection (ρ = 0.750).

Indicators	I	II	III	IV	V	VI	VII	VIII	B	Rank
K1	5	4	2	1	1	2	1	2	8.571	2
K2	4	2	3	2	4	1	3	1	8.622	1
K3	6	3	1	5	2	3	2	3	7.816	3
K4	2	1	6	4	3	7	4	5	7.214	4
K5	1	5	4	3	5	6	6	4	6.867	5
K6	3	11	8	6	9	5	5	6	4.102	6
K7	9	10	5	7	6	4	7	10	3.235	7
K8	8	9	10	9	7	11	10	8	2.112	10
K9	7	8	9	11	8	8	9	11	1.990	11
K10	11	6	11	8	10	9	11	7	2.276	8
K11	10	7	7	10	11	10	8	9	2.194	9

). E_i, T_i, O_i, and H_i refer to the indicators presented in Table 1. Columns I to VIII present how the experts ranked the indicators, while column B shows the values determined using Equation (2).

The comparison of aspects is shown in

Table 8. Comparison of aspects (CCI = 0.021).

Experts	Aspects	C	A	R
I	C	(1,1,1)	W	W−1
	A	W−1	(1,1,1)	W−1
	R	W	W	(1,1,1)
II	C	(1,1,1)	W−1	S−1
	A	W	(1,1,1)	W−1
	R	S	W	(1,1,1)
III	C	(1,1,1)	E	W−1
	A	E−1	(1,1,1)	W−1
	R	W	W	(1,1,1)
IV	C	(1,1,1)	W	W−1
	A	W−1	(1,1,1)	S−1
	R	W	S	(1,1,1)
V	C	(1,1,1)	W−1	S−1
	A	W	(1,1,1)	W−1
	R	S	W	(1,1,1)
VI	C	(1,1,1)	E	W−1
	A	E−1	(1,1,1)	W−1
	R	W	W	(1,1,1)
VII	C	(1,1,1)	W−1	W−1
	A	W	(1,1,1)	E
	R	W	E−1	(1,1,1)
VIII	C	(1,1,1)	E	W−1
	A	E−1	(1,1,1)	W−1
	R	W	W	(1,1,1)
Aggregated matrix	C	(1,1,1)	(0.55,0.87,2.27)	(0.18,0.29,0.82)
	A	(0.44,1.14,1.81)	(1,1,1)	(0.22,0.33,0.91)
	R	(1.22,3.45,5.49)	(1.10,3.01,4.63)	(1,1,1)
Fuzzy weights		(0.09,0.18,0.61)	(0.09,0.20,0.58)	(0.21,0.62,1.45)
Crisp weights		0.217	0.217	0.566

. An aggregate matrix was formed based on individual comparison matrices. Equation (9) and the elements of the weight vector Φ were applied to obtain the aggregate matrix, followed by Equation (12) to determine the fuzzy weights. Crisp weights of aspects were calculated using Equations (17) and (18).

Factors are at the next level of the decision hierarchy. The aggregate factor comparison matrix, created from the matrices of eight experts using Equation (9) and the elements of the weight vector Φ, is shown in

Table 9. Comparison of factors (CCI = 0.012).

Factors	E	T	O	K
E	(1,1,1)	(0.81,1.30,3.07)	(0.43,0.78,1.97)	(0.43,0.62,1.77)
T	(0.33,0.77,1.23)	(1,1,1)	(0.26,0.43,1.06)	(0.23,0.34,0.90)
O	(0.51,1.28,2.31)	(0.94,2.31,3.81)	(1,1,1)	(0.71,0.79,2.37)
K	(0.57,1.62,2.31)	(1.11,2.93,4.41)	(0.42,1.27,1.41)	(1,1,1)
Fuzzy weights	(0.09,0.21,0.73)	(0.05,0.14,0.42)	(0.11,0.29,0.86)	(0.10,0.37,0.79)
Crisp weights	0.247	0.147	0.304	0.302

. Factor weights were determined using Equations (13), (14), (17), and (18).

At the next level of the hierarchy are the indicators.

Table 10. Comparison of KIs (CCI_E = 0.010; CCI_T = 0.007; CCI_O = 0.009; CCI_H = 0.012).

E	E1	E2	E3	E4	E5
E1	(1,1,1)	(0.27,0.40,1.11)	(0.42,0.59,1.62)	(0.25,0.35,0.83)	(0.54,0.65,1.96)
E2	(0.90,2.52,3.69)	(1,1,1)	(1.00,1.55,3.68)	(0.48,0.75,1.89)	(1.00,1.89,3.91)
E3	(0.62,1.71,2.38)	(0.27,0.65,1.00)	(1,1,1)	(0.30,0.50,1.35)	(0.79,1.09,2.77)
E4	(1.21,2.89,4.04)	(0.53,1.34,2.06)	(0.74,2.01,3.38)	(1,1,1)	(1.00,2.17,4.30)
E5	(0.51,1.53,1.87)	(0.26,0.53,1.00)	(0.36,0.92,1.26)	(0.23,0.46,1.00)	(1,1,1)
Fuzzy weight	(0.05,0.10,0.40)	(0.09,0.26,0.82)	(0.06,0.17,0.50)	(0.09,0.32,0.85)	(0.04,0.15,0.39)
T	T1	T3	T4	T5	T8
T1	(1,1,1)	(0.28,0.47,1.33)	(0.17,0.27,0.47)	(0.60,0.79,2.09)	(0.54,0.65,1.96)
T3	(0.75,2.14,3.54)	(1,1,1)	(0.31,0.48,1.25)	(0.84,1.69,3.35)	(0.82,1.43,3.00)
T4	(2.15,3.72,5.72)	(0.80,2.07,3.22)	(1,1,1)	(1.44,3.03,4.62)	(1.22,2.82,4.55)
T5	(0.48,1.27,1.66)	(0.30,0.59,1.20)	(0.22,0.33,0.69)	(1,1,1)	(0.53,0.75,1.99)
T8	(0.51,1.53,1.87)	(0.33,0.70,1.22)	(0.22,0.36,0.82)	(0.50,1.33,1.87)	(1,1,1)
Fuzzy weight	(0.05,0.10,0.37)	(0.08,0.21,0.65)	(0.14,0.41,1.01)	(0.05,0.13,0.37)	(0.05,0.15,0.39)
O	O1	O2	O3	O4	O5
O1	(1,1,1)	(0.39,0.64,1.73)	(0.71,1.46,2.69)	(0.84,1.58,2.97)	(0.76,1.26,2.90)
O2	(0.58,1.57,2.55)	(1,1,1)	(1.12,2.17,3.20)	(1.47,2.45,3.97)	(0.70,1.73,2.80)
O3	(0.37,0.68,1.40)	(0.31,0.46,0.89)	(1,1,1)	(0.72,1.34,2.79)	(0.65,0.88,2.25)
O4	(0.34,0.63,1.19)	(0.25,0.41,0.68)	(0.36,0.75,1.39)	(1,1,1)	(0.35,0.62,1.56)
O5	(0.35,0.79,1.32)	(0.36,0.58,1.43)	(0.45,1.14,1.54)	(0.64,1.61,2.87)	(1,1,1)
Fuzzy weight	(0.08,0.21,0.67)	(0.11,0.32,0.79)	(0.06,0.16,0.49)	(0.05,0.12,0.36)	(0.06,0.18,0.49)
K	K1	K2	K3	K4	K5
K1	(1,1,1)	(0.37,0.71,1.78)	(0.61,0.98,2.27)	(0.62,0.83,2.31)	(0.62,1.20,2.41)
K2	(0.56,1.42,2.68)	(1,1,1)	(0.86,1.35,2.48)	(0.69,1.26,2.62)	(0.80,1.69,2.92)
K3	(0.44,1.02,1.65)	(0.40,0.74,1.16)	(1,1,1)	(0.58,0.92,2.37)	(0.86,1.29,3.11)
K4	(0.43,1.21,1.61)	(0.38,0.80,1.45)	(0.42,1.08,1.72)	(1,1,1)	(0.88,1.41,3.26)
K5	(0.42,0.84,1.62)	(0.34,0.59,1.25)	(0.32,0.78,1.16)	(0.31,0.71,1.14)	(1,1,1)
Fuzzy weight	(0.07,0.18,0.62)	(0.09,0.26,0.73)	(0.07,0.19,0.57)	(0.07,0.21,0.56)	(0.05,0.15,0.41)

shows the aggregate comparison matrices of the KIs for all four factors, obtained using Equation (9) and the elements of the weight vector Φ. The table caption contains CCI values for all aggregate matrices.

The fuzzy weights of the indicators were determined using Equation (15). After defuzzification of these values using Equations (17) and (18), the final weights were obtained according to Equation (16).

Table 11. Weights and ranks of the KIs.

Indicators	Weights	Ranks
(E1) Level of legal procedures implemented	0.032 (0.129)	17
(E2) Number of external funds for OHS system improvement	0.068 (0.272)	5
(E3) Level of networking with related companies in communities of practice	0.042 (0.169)	12
(E4) Frequency of coordinating activities with contractors	0.074 (0.295)	2
(E5) Level of implementation of standards and recommendations	0.034 (0.135)	16
(T1) Probability of failure (reliability)	0.020 (0.125)	20
(T3) Assessment and quantification of accidents	0.036 (0.226)	15
(T4) Number and assessment of incidents	0.059 (0.373)	7
(T5) Availability	0.021 (0.132)	19
(T8) Criteria for selecting the safety level	0.023 (0.143)	18
(O1) Analysis of the results of external OHS controls	0.071 (0.233)	4
(O2) Effectiveness of OHS resource management	0.085 (0.294)	1
(O3) Evaluation of the effects of OHS training (success and retention)	0.051 (0.170)	11
(O4) Number (percentage) of higher-risk work activities	0.038 (0.128)	14
(O5) Causes and consequences of adverse events and stressful situations	0.053 (0.176)	10
(K1) Number of workplace injuries	0.060 (0.207)	6
(K2) Level of employees‘ communication competence	0.074 (0.256)	3
(K3) Level of compliance with work procedures	0.056 (0.196)	9
(K4) Degree of employee satisfaction with safety and job conditions	0.057 (0.198)	8
(K5) Number of mistakes and errors during work	0.041 (0.144)	13

shows the final ranks. The Weights column presents the final weights and the local weights of the indicators (relative to the individual factors that they describe) in parentheses. The local weights represent the first term in Equation (15). Their fuzzy values are shown in Table 10. Indicator ranks were determined based on the final weights.

Further analysis involved the consideration of different scenarios. These scenarios represented certain shifts in organisation priorities caused by internal needs or changes in the environment. These priorities were defined by different values of the factor weights (

Table 12. Analysed scenarios.

Weights	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10
wE	0.4	0.2	0.2	0.2	0.3	0.2	0.2	0.3	0.3	0.2
wT	0.2	0.4	0.2	0.2	0.3	0.3	0.2	0.2	0.2	0.3
wO	0.2	0.2	0.4	0.2	0.2	0.3	0.3	0.3	0.2	0.2
wK	0.2	0.2	0.2	0.4	0.2	0.2	0.3	0.2	0.3	0.3

), describing a more competitive environment (Scenario S₁), significant technological advancements (Scenario S₂), organisational improvements (scenario S₃), or stronger focus on workers, their knowledge, and abilities (Scenario S₄). Other scenarios (S₅ to S₁₀) presented simultaneous changes in the two factors.

Table 13. Ranks of the KIs in different scenarios.

Indicators	S1	S2	S3	S4	S5	S6	S7	S8	S9	S10
E1	7	19	19	18	15	20	19	15	14	19
E2	2	6	7	9	3	6	9	3	2	9
E3	4	16	14	14	7	17	15	9	9	17
E4	1	3	6	6	2	5	7	1	1	6
E5	6	18	17	16	11	19	17	12	13	18
T1	10	9	10	10	16	16	20	20	20	14
T3	10	2	10	11	4	4	12	10	11	3
T4	3	1	3	5	1	1	3	4	4	1
T5	18	7	18	17	12	12	18	19	18	13
T8	17	5	16	15	9	10	16	18	17	12
O1	9	10	2	10	8	3	4	5	10	10
O2	5	4	1	7	5	2	1	2	8	8
O3	15	15	5	13	18	9	11	8	16	16
O4	19	20	9	19	20	15	14	16	19	20
O5	14	14	4	12	17	7	10	6	15	15
K1	11	11	11	2	10	11	5	11	5	4
K2	8	8	8	1	6	8	2	7	3	2
K3	13	13	13	4	14	14	8	14	7	7
K4	12	12	12	3	13	13	6	13	6	5
K5	16	17	15	8	19	18	13	17	12	11

shows the ranks of the indicators, obtained based on the weights of the factors defined in the scenarios. The weights of the aspects remained unchanged across all analysed scenarios, as shown in Table 8.

4. Discussion

Major problems in OHS systems are often caused by limited safety resources. This is especially true for small companies due to the number of employees and the numerous roles they play in performing core business activities. Large companies also face challenges, particularly due to the limited number of safety officers and their distance from locations where work activities are performed. Financial constraints, or competing priorities, are another limiting factor in OHS systems. When all this is considered in the context of construction activity, potential problems related to risk, possible costs, and socially acceptable business practices become even more pronounced. These three elements are considered as aspects in the decision-making process.

The performance of OHS systems is a means of monitoring their success. The introduction of indicators to evaluate the effectiveness and efficiency of the system simplifies the process of making important decisions and identifying problems. In this analysis, 39 indicators classified into four groups were used, each describing a different factor. All of this was discussed in the context of the three previously mentioned aspects. Both direct indicators, which describe outcomes, and indirect indicators, which describe activities, were included. For the adequate improvement of the OHS system, it was necessary to consider both types of indicators. Their different perspectives, retrospective and predictive, led to a better understanding of the problem and helped identify potential solutions.

The number of indicators is another challenge faced by small organisations. In the research, experts were consulted to address this issue, who proposed potential improvements using a hybrid approach. First, they determined five key indicators from each group using weighted Borda voting, achieving acceptable agreement (ρ ≥ 0.75) in all groups of indicators. The lowest agreement was for indicators describing employees (ρ = 0.750), and the highest was for the technical factor (ρ = 0.797).

The selected KIs were ranked by determining the weights using the group FAHP method. Each factor was described by an equal number of indicators (five) to avoid emphasising certain factors, which is characteristic of groups with a smaller number of compared elements. The hierarchical organisation of indicators facilitated the use of AHP-based methods. The choice of the fuzzy variant was driven by the need to incorporate uncertainty in the decision-making process. To simplify the experts’ work, linguistic variables were used including symbolic abbreviations and an accompanying explanation for each fuzzy number. This made pairwise comparison easier for the experts.

When ranking the factors, experts highlighted the organisation (w_O = 0.304) and employees (w_K = 0.302), followed by the influence of the environment (w_E = 0.247). Technologies had the least influence (w_T = 0.147). During the research, the experts did not identify technological advancements that would significantly improve the OHS system, considering the activity being performed, the cost-effectiveness of the application, and the potential for risk reduction.

Among the indicators, the experts highlighted the efficiency of OHS resource management (w_O2 = 0.085), the frequency of coordinating activities with contractors (w_E4 = 0.074), the degree of communication capacity of employees (w_K2 = 0.074), the analysis of the results of external OHS controls (w_O1 = 0.071), and the number of external funds for improving the OHS system (w_E2 = 0.068). Limited resources in small companies and inadequate safety management can lead to poor outcomes in the OHS system. Experts emphasised the need for significant communication with contractors, especially because window- and door-fitting activities are only one part of the construction process, as shown in Figure 3. Employees may also be exposed to hazards related to some other activities performed in parallel with the installation of doors and windows such as the setup of electrical installations. Work permits, which include the approval of the OHS officers of both the contractor and subcontractor, can help prevent undesirable outcomes in these situations.

Regarding the final weights, the highest-ranked indicators included representatives from all groups describing the influential factors. Certain conclusions can also be drawn from the analysis of the local weights of the indicators, considering their relation to the specific factors they describe. In the group of indicators reflecting organisational influence, the effectiveness of OHS resource management (O₂) and the analysis of external OHS control results (O₁) stood out, while the other indicators had significantly lower weights. Among the indicators related to worker influence, expert evaluations suggested that the communication competence of employees (K₂) had a slightly higher local weight, while the others had very similar values. The selection of key indicators in this group was characterised by the lowest coefficient of group satisfaction, which was at the threshold of acceptability (ρ = 0.750). For the indicators describing the impact of the environment, the frequency of activity coordination with contractors (E₄) and the use of external funds for improving the OHS system (E₂) had the highest local weights, whereas the remaining indicators had significantly lower values. Finally, among the indicators describing the technological factor, only the number and severity of incidents (T₄) stood out.

In all of the analysed scenarios, some technological improvements were implied, as the weight of w_T was increased relative to the value determined by the experts (w_T = 0.147). This is due to the constant development of technologies, especially artificial intelligence (AI), and their possible application in the analysis of available OHS data, advanced monitoring of the working environment, and more precise detection of hazards. Scenarios S₁ to S₄ describe situations in which one factor dominates over the others. These scenarios were used to determine how indicators from a specific group dominate in comparison to indicators from other groups. In contrast, scenarios S₅ to S₁₀ represent situations in which two factors have a slightly more significant influence than the others. For example, technological changes may be accompanied by changes in legislation (scenario S₅) or improvements in working conditions (scenario S₁₀). The following indicators were in the top ten in all of the analysed scenarios: E₂, E₄, T₄, O₁, O₂, and K₂. This highlights the importance of external, additional sources of information, additional funds for improving the OHS system, and organisational capacity for the effective management of OHS resources. Networking within communities of practice or knowledge communities and sharing data with related companies of similar size were identified as potentially very significant impetuses for improving the OHS system. Of course, employees are essential. Their level of communication capacity, ability to identify OHS problems in the work environment, and skill in articulating these problems to management form the basis for improving the OHS system. This is particularly significant in smaller companies, where owners often participate in work activities or supervise the execution of different tasks.

The presented methodological procedure enables the selection and ranking of indicators that suggest potential improvements to the existing OHS system. The hybrid approach allows for the selection of KPs using one method, followed by their ranking using another. In the presented example, the experts found the method for selecting KPs to be simple, but required some professional support to determine the weights using the ranking method. Therefore, it is desirable to provide tools that automate this process and make it easier to implement in practice.

Although the practical application implies a group ranking, it can be performed independently, with the number of experts reduced to just one. However, the initial concept suggests a group ranking to minimize subjectivity. Given that small companies have limited resources, including OHS experts, a larger number of experts can be involved if companies within the same industry, such as window- and door-fitting companies in our example, form a community of practice or a knowledge-sharing network. This would allow them to exchange OHS experiences and implement, among other things, the described hybrid approach.

5. Conclusions

In this research, a hybrid approach was applied. A methodological procedure was proposed that enabled the selection and ranking of indicators to improve the existing OHS system, based on a defined set of hierarchically organised indicators. This was applied with the aim of selecting indicators that could help improve the OHS system in small window- and door-fitting companies. First, from the thirty-nine offered indicators, the experts selected twenty key indicators, five for each of the four factors. Weighted Borda voting was used for the selection of KIs, with the experts’ weights determined based on their experience. Second, a group FAHP method was applied to rank the KIs. This procedure is convenient due to the hierarchical structure of the problem and the need to account for uncertainty in the decision-making process. However, in FAHP, only the membership function (μ) is used during the analysis. Hesitation could further enhance the realism of decision making.

Among the indicators, the experts highlighted the efficiency of OHS resource management, the frequency of coordinating activities with contractors, the communication capacity of workers, the analysis of the results of external OHS controls, the number of external funds for OHS system improvements, and the number and assessment of incidents. These results are a consequence of the current state of the analysed OHS systems and the consideration of their potential for the most effective improvements.

Furthermore, ten scenarios were analysed with varying values of factor weights. The scenarios showed the importance of highlighted indicators, which were always ranked among the top ten indicators.

The list of indicators was not created for a specific activity but is of a universal nature. The presented list of indicators is not exhaustive and can be supplemented with new indicators. By adding indicators, the same hierarchical structure can be applied to other companies with specific operational activities. Thus, this structure could be tested in different industrial settings, while emerging technologies could also be included in the analysis to examine their influence on workplace safety practices. After adding new indicators, the procedure for selecting and ranking KIs must be repeated. Window- and door-fitting activities should also be considered from the perspective of potential long-term consequences such as issues caused by the physical and mental workload of employees due to increased dynamics and the volume of work. The creation of funds to help small businesses improve OHS systems as well as the formation of knowledge communities or communities of practice to share OHS experiences could further enhance the working conditions. Further activities will focus on initiating the development of such a fund as well as supporting the creation of a community of practice or knowledge community for networking small window- and door-fitting companies to promote interorganisational OHS knowledge exchange and spillover.