Safety Science 117 (2019) 71–80

Contents lists available at ScienceDirect

Safety Science
journal homepage: www.elsevier.com/locate/safety

Assessing the impact of processes on the Occupational Safety and Health T

Management System’s effectiveness using the fuzzy cognitive maps
approach
Anna Skład
Central Institute for Labour Protection – National Research Institute, Czerniakowska 16, 00-701 Warsaw, Poland

A R T I C LE I N FO A B S T R A C T

Keywords: The paper presents research in which the fuzzy cognitive maps (FCM) approach was applied in examining the
Fuzzy cognitive maps impact of individual processes in the Occupational Safety and Health Management System (OSH MS) on the
Occupational Health and Safety Management effectiveness of this System. For this purpose, an FCM OSH MS model comprising 16 objects has been developed.
System Fifteen of those objects corresponded to the processes in the OSH MS and one object corresponded to safety
OSH leadership
performance. Then a series of simulations was carried out, in which it was successively predicted how the
ISO 45001
improvement of individual processes affected overall safety performance. It was finally proven that safety
performance increased most significantly under the influence of improvement of the leadership process. This
proves that among all processes in the system, leadership has the greatest positive impact on its effectiveness.
Practical implications of such simulation results are discussed herein, and further, extended application of FCM
to improve effectiveness of OSH MSs is suggested.

1. Introduction taking on serious challenges (Hasle and Zwetsloot, 2011);

• rendering enterprises more focused on meeting system requirements
Occupational Safety and Health Management Systems (OSH MSs) than on legal regulations (Bennett, 2002).
are implemented in enterprises in order to prevent injury and ill health
among workers and to provide safe and healthy workplaces (ISO 45001, On the other hand, critique was accompanied by studies aimed at
2018). However there remains considerable uncertainty whether the finding proofs that OSH MSs do make enterprises safer for their
systems serve actively to achieve the outcomes assumed. Do they really workers. However in different studies a variety of measures were taken
help identify all hazards and eliminate or minimize risks by taking ef- into account to evaluate systems’ impact. Some researchers decided to
fective preventive and protective measures? Do they make workplaces investigate systems’ effectiveness using accidents at work indicators,
healthier for workers? Do they contribute to the decrease in work-re- examining changes in the assessment of the physical and psychosocial
lated accidents and near misses? And, generally speaking, are they ef- work environment or changes in the assessment of employees' well-
fective? being, or even analysing enterprises’ competitiveness and economic
The ability of Occupational Safety and Health Management Systems performance indicators (Akyuz and Celik, 2014; Basso et al., 2004;
(OSH MSs) to achieve the outcomes assumed has already been ques- Fernández-Muńiz et al., 2009; Torp and Moen, 2006). Others argued
tioned in numerous papers. Moreover, some suggestions were raised that OSH MSs exerted a positive impact by improving safety perfor-
that systems could even hinder safety management in enterprises. The mance, which to their understanding depended on how well individual
most frequent objections against implementation and maintenance of components of the systems functioned (Bottani et al., 2009;
OSH MSs included: Mohammadfam et al., 2017; Podgórski, 2015).
Although many of the studies performed seemed to confirm the
• causing excessive bureaucracy, which, instead of supporting safety positive impact of OSH MSs, their in-depth review showed that they did
performance, may in fact discourage employees and negatively af- not provide unambiguous evidence confirming the ability of OSH MSs
fect their participation in safety management (Granerud and Rocha, to achieve the outcomes assumed (Robson et al., 2007). Authors of the
2011); review were unable to conduct meta-analysis of the studies published to
• being used to solve only minor health and safety problems without date due to the small number of those which met established criteria

E-mail address: [email protected].

https://doi.org/10.1016/j.ssci.2019.03.021
Received 23 November 2018; Received in revised form 1 March 2019; Accepted 25 March 2019
Available online 13 April 2019
0925-7535/ © 2019 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
A. Skład Safety Science 117 (2019) 71–80

and the heterogeneity of applied research methods. On the other hand, of employees and managers, to effectively implement such changes, and
qualitative synthesis of the studies proved their weak methodological eventually the provision of training to enhance employees’ safety per-
quality and the lack of generalizability. The Authors summed up their formance and their active participation in the OSH MS improvement
review by stating that designing and conducting research that could (Frick, 2011).
eventually give unequivocal evidence of the effectiveness or ineffec-
tiveness of the systems would be very expensive and complex to im- 2. Application of Fuzzy Cognitive Maps (FCM) for modelling the
plement. OSH MS
The conclusion about lack of evidence confirming OSH MSs effec-
tiveness goes in line with findings of yet another study, in which it was Different scientists and practitioners point out a variety of OSH MS
proven that that enterprises with certified OSH MSs can in fact differ processes of critical value to the effectiveness of OSH MSs, i.e. those
substantially in terms of how successful they are in managing occupa- processes which contribute most distinctly to the improvement of safety
tional safety (Granerud and Rocha, 2011). Case studies conducted in a performance. However the same factors that hamper providing evi-
number of companies revealed that some of them did not even meet dence on OSH MSs effectiveness make it even more difficult to identify
legal requirements in the field of OSH whereas others integrated OSH the processes that exert the greatest impact on such systems’ effec-
within the enterprise management systems or even listed OSH goals tiveness. Substantial financial resources and a huge organisational ef-
among the company's strategic goals. Especially the first types of these fort would be necessary to conduct a study on a sample of workplaces
situations raises legitimate doubts as to the actual value of the certifi- where the items measured should include not only safety performance
cate confirming OSH MS implementation, because it proved that a but also performance of individual OSH MS processes prior to systems’
certified system can in fact exist only as a set of formal records and implementation and then at specific time intervals.
documents in complete isolation from actually used practices and What is more, OSH MS processed are interrelated, which means that
procedures. the performance of a certain process relies on the quality of outputs
The unambiguous results of research on OSH MSs effectiveness lead received from (an)other process(es). Consequently, if the results of a
to the conclusion that there must be some factors determining whether given process are high quality outputs, as they become inputs to other
the system is effective or not. However different researchers underline processes they positively affect their operation, improving safety per-
the importance of different processes that, according to them, con- formance in an indirect manner. Therefore, the actual impact of the
tribute most to the systems’ effectiveness. individual process on safety performance in the enterprise results from
Some of them argue that the basis of safe working conditions is its direct influence and further influences exerted indirectly.
correct and careful risk analysis (Demichela et al., 2004). Solid risk In this paper, it is therefore suggested to use Fuzzy Cognitive Maps
management foundations guarantee that the company is in a state of (FCM) as an alternative method that will allow to identify processes of
constant preparedness for unexpected events. Moreover, it provides critical value to the effectiveness of the OSH MSs.
mechanisms that ensure such events, should they occur, are promptly Fuzzy Cognitive Maps, proposed by Bart Kosko in 1986, constitute
contained (Else and Beaumont, 2001). an expert-based method of knowledge development in soft domains
The internal audit process is another system element perceived to be such as political and military sciences, history, international relations
crucial for OSH MS effectiveness, however it is stressed that its impact and organization theory. What Kosko observed was namely that tradi-
depends on whether the auditors approach the identification of hazards tional binary logic proves insufficient when describing systems where
and other factors affecting the health and safety of employees in a the existence of certain phenomena and interrelations between them
comprehensive manner, and whether they are capable of suggesting cannot be reflected by using only two values: 1, meaning “is”, or 0,
appropriate measures to reduce occupational risk (Makin and Winder, meaning “there is no”. Precise statements are useful for representing
2008). undeniable circumstances such as e.g. the fact of being promoted to
It is believed that an effective OSH MS cannot function without open become a manager. There are two possibilities: one can either be pro-
communication and reliable communication channels that provide op- moted or not. However to develop knowledge about the OSH MS by
portunities for each employee in the enterprise to stay connected to all means of modelling, analysing and making predictions about possible
other employees with whom that person should be in contact (Kelly and scenarios of its further progress, an issue more important than the ex-
Berger, 2006). istence of a manager is what skills this person has and to what extent
Researchers and practitioners underline the need of employee par- he/she is able to use them in practice. Those characteristics cannot be
ticipation for ensuring OSH MS effectiveness, stating that it plays a key expressed following binary logic. They will however be captured more
role in the risk management process as involved employees are more accurately in terms of fuzzy logic (Kosko, 1986).
willing to share information on accidents or near misses as well as Fuzzy logic is based on the concept of a fuzzy set, which is a set of
communicate their ideas and views on solutions improving safety. elements that belong to it to a certain degree. A fuzzy set is described by
Having obtained the information directly from the employees, the means of a membership function (Dworniczak, 2003). Degrees are re-
employer is able to adjust the working conditions to their individual presented by fuzzy values. They are determined according to the sub-
needs and expectations in a more efficient manner (Podgórski, 2005). jective assessments of people believed to have expert knowledge in a
Many researchers state that the real determinant of an effective OSH certain domain. Those assessments are expressed in natural language.
MS is top management involvement while the strategic issue is what A Fuzzy Cognitive Map is usually defined as “… an oriented graph
motivates management to implement the system (Bottomley, 1999). which shows the degree of causal relationship between different fac-
Through their attitudes and actions, senior managers influence the at- tors, where knowledge expressions, in the causal relationship, are ex-
titudes and actions of the staff. As a result, safety culture within the pressed by either positive or negative sign and different weights”
organization is shaped and standards of safety behaviours are being (Stylios et al., 1997).
implemented and fixed. If the top managers do not become involved in An FCM can be illustrated as a set of objects and arrows between
ensuring safety, it is unlikely that safety would become a priority for the them that represent interrelations. An arrow with positive sign from
staff. If, on the other hand, the top managers are perceived by the staff object A to object B means that increasing the value of object A will
as partners seriously concerned about safety, the OSH MS can become increase the value of object B (positive influence). An arrow with ne-
successful (Civil Aviation Safety Authority, 2006). gative sign from object A to object B means that increasing the value of
Other elements determining effectiveness of the OSH MS also in- object A will decrease the value of object B (negative influence).
clude the availability of resources for implementation of technical and Interrelation is characterized by a value (weight) that expresses the
organizational solutions improving safety, ensuring competence, both strength of influence.

72
A. Skład Safety Science 117 (2019) 71–80

7. Training and competence management

8. Taking advantage of OSH opportunities
9. Planning and achieving OSH objectives
10. Occupational risk management
11. Emergency preparedness and response
12. Incident, nonconformity and corrective actions
13. Internal audit
14. Monitoring and performance evaluation
15. Documented information

It was assumed that the impact of the particular process on OSH MS

effectiveness would be expressed by the ability of this process to im-
prove safety performance. Therefore to be able to study this impact, the
16th object/node was included in the FCM model. It corresponded to
safety performance.

2.2. Assembling a group of OSH MS experts

FCM is a reflection of expert knowledge. For this reason an im-

portant element of the research constitutes an adequate selection of
experts who should have the relevant knowledge to develop the FCM
model. FCM research in the field of OSH was reviewed in order to de-
termine the number and professional experience of participating ex-
perts (Bertolini, 2007; Dagdeviren and Yuksel, 2008; Espinosa-Paredes
et al., 2009; Bevilacqua et al., 2012; Asadzadeh et al., 2013; Azadeh
et al., 2014). The analysis proved that the panels of experts in these
studies were diversified in terms of their professional background.
Panel members were both managers and employees of enterprises from
Fig. 1. Stages of the FCM OSH MS model development. different departments as well as researchers and OSH professionals (e.g.
OSH authorities’ representatives).
Objects are also characterised by values that constitute their as- OSH MSes are implemented, maintained and improved in en-
sessments. Referring to the example given above, the value of the object terprises and a detailed, practical knowledge concerning the way they
corresponding to the manager in the FCM model would indicate how operate is acquired by employees of those enterprises, especially those
well this person performs her/his managerial duties. who perform specific roles in management systems, i.e. act as auditors
Values of objects and interrelations range from 0 to 1. In mathe- or process owners. However their knowledge and expertise is limited to
matical terms, FCM can be presented as the vector of object values and the specificity of the companies in which they are currently employed
the matrix of interrelation values. or were employed previously. Universal practical knowledge and ex-
One potential application of FCMs is also to model systems which by pertise referring to OSH MSes operated in the companies of different
definition consist of interrelated elements. sizes, sectors and organizational structures and presenting different
It is not the purpose of this paper to describe FCM theory as this was approaches to the implementation, maintenance and continuous im-
previously done by many authors (Pelaez and Bowles, 1996; Stylios and provement of OS MSes is the domain of consultants, who advise a wide
Groumpos, 2000; Stach et al., 2010; León et al., 2010; Gray et al., range of business units and observe their progress in safety perfor-
2014). However to facilitate further reading, subsequent stages of the mance. Yet another look at OSH MSes is presented by experts involved
development of an FCM OSH MS model are presented in the form of a in conducting research in this field. Their knowledge includes general
diagram (Fig. 1) and this section is further divided into subsections conclusions resulting from the research on OSH MSes. Furthermore they
corresponding to those stages. examine global trends and follow scientific research conducted in the
field of management systems worldwide.
2.1. Definition of model objects Considering the above, the decision was made on the re-
presentatives of the three mentioned groups of experts to be involved in
Objects in FCM models may correspond to abstract and unmeasur- developing the FCM OSH MS model.
able concepts such as: events, actions, goals, values, system character- The review of the OSH FCM research also demonstrated that the
istics, variables that affect them and their states, as well as processes’ number of experts in FCM OSH studies varied significantly and ac-
inputs and outputs (Groumpos, 2010; Stach et al., 2010; León et al., counted for between 1 and 60.
2010). For the purpose of this study, it was assumed that objects in the Considering the lack of unambiguous indications included in FCM
FCM OSH MS model would correspond to the processes in the System. methodology as to the number of experts, and following the rules ap-
ISO 45001 standard served as the basis for selection of 15 processes. plicable in this respect in the Delphi and the brainstorming methods 9
They were as follows: experts were invited to participate in the study, 3 from each of the
above mentioned groups.
1. Leadership Three groups of experts participating in the study are subsequently
2. OSH policy characterized in Tables 1–3.
3. Organizational roles, responsibilities and authorities The role of experts was to determine the interrelations among ob-
4. Consultation and participation of workers jects in the FCM OSH MS model (to decide which processes influence
5. Management review each another/safety performance) and to estimate the values of those
6. Communication and information interrelations, i.e. to assess how strong individual influences were.

73
A. Skład Safety Science 117 (2019) 71–80

2.3. Indication of interrelations between objects and estimation of their

Year of the implementation of the OSH MS in the
values with the use of the linguistic variable

The essence of fuzzy logic is to allow experts to express their

knowledge without being forced to use numerical values. For this
reason, at the stage of estimating the values of interrelations in the
FCM, model experts should use linguistic values and not to associate
them with any numerical counterparts. However for further simulations
the researcher needs to operate on numbers which are obtained from
linguistic values under the so called “defuzzification” procedure.
For the purpose of experts’ evaluation of interrelations in this study,
the linguistic variable of “strength of influence” was developed. A lin-
company

guistic variable consists of several linguistic values, which correspond

to fuzzy sets. Fuzzy sets are overlapping to better express the impreci-
2004

2003

2004

sion of experts' assessments.

The linguistic variable “strength of influence” was adopted from
Electricity, gas, steam and air conditioning

Stylios and Groumpos (2004), and tailored slightly to streamline ex-

Manufacture of chemicals and chemical

perts’ evaluations. First of all, the number of linguistic values was re-
Manufacture of electrical equipment

duced as the experts assumed that seven (instead of nine) would be

enough to complete their evaluations. Secondly, according to re-
commendations of Georgopulous et al. (2003) and Xirogiannis et al.
(2004) for two membership functions the asymmetric triangular shapes
were given, to ensure finer distinction between assessments on opposite
NACE1 division

ends of the scale “strength of influence”. Previous FCM research proved

that linguistic variables consisting of asymmetric membership functions
products

which are not homogenously distributed better reflected experts’ opi-

supply

nions expressed in linguistic values.

The final linguistic variable included the following linguistic values:
Top Management Representative

no influence, very low influence, low influence, medium influence, high

influence, very high influence, influence close to 1.
The chart of membership functions describing the linguistic variable
is presented in Fig. 2, where: membership function μ0 describes a lack
Role in the OSH MS

of influence, membership function μ1 – very low influence, membership

Internal Auditor

function μ2 – low influence, membership function μ3 – medium influ-

Administrator

ence, membership function μ4 – high influence, membership function

μ5 – very high influence, and membership function μ6 – influence close
to 1.
To assess interrelations, experts analysed pairs of objects in the FCM
OSH MS and determined if there was an influence between them, how
strong it was and whether it was positive or negative.1 Each expert
assessed the 240 possible interrelations2 between processes as well as
4 months on the current position, 17 years of previous

between processes and safety performance in this manner.

2.4. Transformation of linguistic values into numbers and development of

the interrelations matrix
Work experience in the position

experience in the OSH service

Characteristics of the expert group - employees of enterprises.

Linguistic values indicated by experts were subsequently defuzzified

into numerical interrelation values using the modified3 weighted mean
of maximum method (Pelaez and Bowles, 1996).
European Classification of Economic Activities.

The basis for defuzzification was the linguistic “strength of influ-

ence” variable and, in particular, three figures characterizing sub-
sequent membership functions assigned to linguistic values: lower and
10 years
7 years

upper range of influence strength, as well as influence strength corre-

sponding to the maximum value of the membership function. The
coefficient zi (last column in Table 4) was calculated for each linguistic
Head of OSH Department
Manager of OSH Section

value as the arithmetic mean of the influence strength value,

Section Manager

1
In fact, experts noted no negative influences.
2
As it had been assumed that objects do not influence themselves, the
Position

number of interrelations can be calculated according to the following formula:

n2-n, where n is the number of objects.
3
Modification included skipping scaling of assessments given by individual
Table 1

experts which could be applied for example to assign different weights due to
Expert

diverse experience, knowledge or other relevant characteristics of those ex-

1
1

perts.

74
A. Skład Safety Science 117 (2019) 71–80

Table 2
Characteristics of the expert group – OSH MS researchers.
Expert Position Work experience in the Participation in OSH MS projects
position

1 Head of the Occupational Safety and Health 17 years Participation in the development of the Polish national OSH MS Standard (PN-
Management Department N 18001: 2004),
Leading research projects in the field of OSH management
Managing consulting projects aimed at implementation, maintenance and
improvement of OSH MSes
2 Researcher in the Occupational Safety and Health 17 years Leading research projects in the field of OSH management
Management Department Participation in consulting projects aimed at implementation, maintenance
and improvement of OSH MSes
3 Researcher in the Occupational Safety and Health 3 years Participation in the implementation of the OSH MS in the enterprise
Management Department Participation in a project on the operational effectiveness of OSH MS and key
performance indicators

corresponding to the maximum value of the membership function de-

scribing this linguistic variable, and the values of influence strengths
corresponding to the maximum values of membership functions that
overlap with it. Overlapping of neighbouring membership functions is a
result of the manner in which the lower and the upper ranges of their
influence strengths have been defined.
For example, the value of the coefficient zi for the linguistic value
“low influence” was calculated as the arithmetic mean of the following
values:

0.25 – the strength of influence value corresponding to the max-

imum value of the membership function describing the linguistic
“low influence” variable
0.1 – the strength of influence value corresponding to the maximum
value of the membership function, describing the linguistic “very Fig. 2. Membership functions of the “strength of influence” linguistic variable.
low influence” variable (which overlaps with the membership
function describing the linguistic variable “low influence”)
0.5 – the strength of influence value corresponding to the maximum N – number of experts
value of the membership function, describing the linguistic
“medium influence” variable (which overlaps with the membership Eq. (1). Numerical value of strength of influence
function describing the linguistic “low influence” variable) Assuming that influences were estimated without scaling, indication
of a linguistic value was associated with the maximum value of a cor-
zi = (0.25 + 0.1 + 0.5): 3 ≈ 0.283 responding membership function. As this value equalled 1, the trans-
Defuzzification was carried out according to Eq. (1): formation procedure was based on calculating the mean of zi values
corresponding to the assessments of 9 experts participating in the study.
N
∑i = 1 oi z i The zi values and the parameters of membership functions presented
w= N
∑i = 1 oi (1) in Fig. 2 are shown in Table 4.
W value was calculated for each of 240 interrelations in the model.
w – numerical interrelation value (numerical value of the strength of Calculation of all strengths of influence values allowed for development
of the interrelation matrix presented in Table 5.
influence)
oi – maximum value of membership function corresponding to the Numbers from 1 to 15 in the first row and in the first column of the
table (in grey fields) correspond to subsequent OSH MS processes listed
linguistic value estimated by Expert i
in Section 2.1. Number 16 corresponds to safety performance. In the
zi – defuzzified linguistic value estimated by Expert i

Table 3
Characteristics of the expert group - professional consultants in the field of OSH management.
Expert Position Work experience in the Participation in OSH MS projects
position

1 Consultant, auditor, lecturer and practitioner in the 16 years Implementation and consulting of OSH MSes, especially in the metallurgical and
field of OSH MS energy industries
Member of the Technical Committee No. 276 on Occupational Safety and Health
Management Systems at the Polish Committee for Standardization.
Auditor at Bureau Veritas Certification Polska
2 Consultant, auditor, lecturer and practitioner in the 16 years Implementation and quality assurance consulting as well as OSH MSes, especially in
field of quality and OSH MSes the metallurgical and energy industries
Auditor at DAS Certification
3 Consultant, auditor, lecturer and practitioner in the 16 years Implementation and environmental management systems consulting, especially in
field of environmental and OSH MSes the metallurgical and energy industries.
Member of the Technical Committee No. 270 on Environmental Management at the
Polish Committee for Standardization
Auditor at Polish Centre for Accreditation

75
A. Skład Safety Science 117 (2019) 71–80

Table 4
Parameters of membership functions and zi values.
Linguistic value Lower range of strength of influence for Strength of influence value corresponding Upper range of strength of influence for Defuzzified linguistic
the membership function describing the to the maximum value of a membership the membership function describing the value zi
linguistic value function linguistic value

no influence 0.00 0.00 0.00 0.0000000

very low influence 0.00 0.10 0.20 0.1750000
low influence 0.10 0.25 0.40 0.2833333
medium influence 0.35 0.50 0.65 0.4666667
high influence 0.55 0.65 0.80 0.6500000
very high influence 0.65 0.80 0.90 0.8166667
influence close to 1 0.85 1.00 1.00 0.9000000

Table 5
Interrelation matrix.
No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 0 0.7 0.5926 0.4852 0.3991 0.5222 0.2046 0.2389 0.5556 0.3602 0.2278 0.3602 0.1528 0.25 0.1463 0.4185
2 0.0583 0 0.1259 0.3898 0.2981 0.5102 0.2278 0.1139 0.6352 0.4463 0.1667 0.1139 0.2972 0.187 0.2361 0.2759
3 0.0583 0.0944 0 0.1667 0.0583 0.3111 0.1667 0.1667 0.187 0.2074 0.4019 0.2685 0.187 0.1139 0.3704 0.3796
4 0.1463 0.3981 0.2556 0 0.325 0.4852 0.3694 0.187 0.5241 0.6259 0.1667 0.6259 0.2759 0.2991 0.0778 0.4704
5 0.2685 0.2889 0.1556 0.0944 0 0.3481 0.1259 0.3065 0.3833 0.1722 0.0944 0.0944 0.1259 0.2167 0.3889 0.1981
6 0.1667 0.2889 0.0778 0.5472 0.2583 0 0.2463 0.2861 0.1463 0.4352 0.2685 0.2926 0.1259 0.1556 0.4648 0.3278
7 0.1778 0.2167 0.2167 0.5463 0.1556 0.1843 0 0.1139 0.1463 0.3944 0.4926 0.3333 0.4194 0.2889 0.2889 0.337
8 0.2167 0.2139 0.2074 0.1556 0.4509 0.1843 0.1139 0 0.5278 0.2046 0.1463 0.1139 0.0944 0.3287 0.1139 0.3593
9 0.2556 0.3083 0.1463 0.2074 0.5648 0.4093 0.2278 0.3 0 0.3269 0.2463 0.2278 0.2907 0.4 0.2074 0.3889
10 0.1454 0.2269 0.1981 0.3796 0.5241 0.2796 0.3611 0.1139 0.5926 0 0.3463 0.1852 0.2389 0.4944 0.363 0.5926
11 0.2278 0.1852 0.2278 0.2074 0.1259 0.4889 0.4704 0.1463 0.2278 0.3944 0 0.2074 0.0583 0.2463 0.3704 0.4074
12 0.2278 0.3 0.1259 0.237 0.6444 0.3907 0.1667 0.1667 0.3185 0.5926 0.2463 0 0.3796 0.2981 0.4185 0.663
13 0.2685 0.2889 0.1139 0.1259 0.6259 0.4491 0.1667 0.2565 0.2463 0.313 0.2278 0.4185 0 0.2278 0.5463 0.3796
14 0.2361 0.3241 0.1556 0.0944 0.6259 0.4093 0.2278 0.1463 0.2889 0.3657 0.2463 0.4019 0.3074 0 0.5472 0.4352
15 0.1259 0.1463 0.0944 0.2278 0.4556 0.3407 0.2889 0.1463 0.2685 0.3472 0.3222 0.2981 0.2889 0.3269 0 0.2796
16 0.0722 0.0907 0.0907 0.0907 0.0315 0.0722 0.0907 0.0907 0.1 0.0907 0.1 0.0907 0.0722 0.0907 0.0519 0

subsequent rows of the matrix (white fields) the values of strengths of was improved.
influence exerted by individual processes/safety performance in the According to this scenario, 15 simulations were carried out in which
FCM OSH MS model are presented. In the subsequent columns of the the future values of all model objects were forecast. The sequence of
matrix the values of strengths of influence exerted on individual pro- simulations followed the list of processes in Section 2.1, i.e. in the first
cesses/safety performance are presented. For example, the number simulation the activation level of the object corresponding to the lea-
0.4852 (second row, fifth column) means the strength of influence with dership process was increased and in the second – the same happened
which the leadership process (No 1 on the list) affects the consultation for the activation level of the object corresponding to the OSH policy
and participation of workers process (No 4 on the list). process, etc. The same process was repeated through to the fifteenth
simulation, in which the activation level of the object corresponding to
the documented information process was increased. The simulations
3. Simulating impact of processes in the FCM OSH MS model on were run until the model stabilized. A comparison of the simulation
safety performance results made it possible to identify the processes whose improvement
caused the greatest increases in safety performance. Table 6 shows the
In the FCM method simulations are used to predict how FCM model simulations results.4
objects will change in the future due to the interrelations between The results of subsequent simulations are presented in columns. The
them. The simulation consists of a series of iterations. A single iteration values of individual objects corresponding to the processes in the OSH
is the function of the product of the object value vector and the inter- MS after the completion of simulation are presented in the following
relation matrix. The simulation ends when the number of iterations set rows marked with numbers from 1 to 15 in accordance with the list
by the researcher is completed or once the model reaches its equili- from Section 2.1. The value of the object corresponding to safety per-
brium (object values cease to change after subsequent iterations), or formance is provided in the row marked with number 16.
when the model is not capable of stabilizing, i.e.– it behaves chaotically Simulation results inference is based on comparisons. Values of the
(León et al., 2010). The function used for calculations (so-called “acti- examined object obtained in subsequent simulations should be com-
vation function”) is usually a sigmoidal or hyperbolic one. Its role is to pared and the simulation characterized by the highest value should be
maintain the result of multiplication in the range from 0 to 1. indicated. A process corresponding to the object whose activation level
The software for simulation calculations was developed in the form accounted for 0.6 in this simulation is the one that exerts the highest
of a set of scripts in the Matlab environment. A hyperboloid threshold positive impact on the object being examined.
activation function was applied, with a slope factor of 0.21. Comparison of the simulation results shows that the object corre-
For the purpose of simulations, it was assumed that the activation sponding to safety performance reached the highest value after the first
levels of objects in the FCM OSH MS model (their initial values) would
equal 0.2. This should reflect the situation in which OSH MS processes
do not operate expeditiously and safety performance is poor. In each 4
The numbers 0.6 in Table 6 are distinguished in grey. They are the values of
simulation, the value of a subsequent individual object corresponding objects whose activation levels were increased at the beginning of subsequent
to a specific process in the OSH MS was increased to 0.6, which was simulations. These values were not taken into account when comparing simu-
supposed to act as an equivalent of the situation in which this process lation results because they were fixed and did not change.

76
A. Skład Safety Science 117 (2019) 71–80

Table 6
Comparison of simulation results.

simulation, in which the value of the object corresponding to the lea- 4. Discussion
dership process was raised. The value of the object corresponding to
safety performance after this simulation accounted for 0.6354 and was The problem faced by numerous companies with OSH MSs is the
by 0.0252 higher than the second highest value of the object corre- poor effectiveness of those MSs, reflected in the fact that their im-
sponding to safety performance, which was obtained in simulation 3. plementation has contributed little to improving the company's safety
The lowest value of the object corresponding to safety performance, performance, or has not led to any improvement at all. This problem is
obtained in simulation 5, was by 0.0806 lower than the value in si- obviously highly complex and there probably exists no universal means
mulation 1. of solving it. Each of the certified companies interested in improving
Comparison of simulation results shows that out of all processes in their safety performance must thoroughly analyse the way it operates
the OSH MS improvement of the leadership process contributes most to and identify the potential for improvement. However, as there are
the system’s effectiveness. The analysis of simulation results presented different views on the factors determining OSH MSs’ effectiveness, the
in Table 6 reveals also that improvement of the leadership process ex- goal of this paper was to indicate those that affect it most. The results of
erted the greatest positive impact on all other processes (the objects the simulations conducted clearly confirm that the impact of OSH MSs
corresponding to them achieved the highest values in the first simula- on safety performance depends to the greatest extent on how the OSH
tion). leadership process is implemented in the company.
This finding seems to go in line with the results of previous studies
that revealed the unique role of OSH leadership in ensuring workplace

77
A. Skład Safety Science 117 (2019) 71–80

safety. It was proven that visibly executed OSH leadership both shapes a Implementation of the ISO 45001 standard can therefore become an
favourable safety climate and affects the company's safety performance opportunity for strengthening the OSH leadership process.
(Wu et al., 2008; Skeepers and Mbohwa, 2015). In addition, it supports Findings of the study confirm that it is worth implementing good
the development of a safety culture (Cooper, 2015; Skeepers and practices in the field of OSH leadership. Such practices include: opening
Mbohwa, 2015). Committed managers positively affect workers’ safety management meetings with safety issues, personal presence of senior
behaviour and their participation in safety management (Lu and Yang, managers during events promoting OSH in the company, their partici-
2010; Clarke, 2013). Verbally performed OSH leadership has a sig- pation in work accident investigations and behavioural audits, etc.
nificantly positive and lasting effect on the level of safety in the com- Applying these practices significantly strengthens the OSH leadership
pany (Kines et al., 2010). On the other hand, passive OSH leadership process, and thus cannot be overestimated when it comes to improving
negatively affects the safety climate and safety consciousness, which, in the company's safety performance. However, first and foremost, the key
turn, predict safety events and injuries (Kelloway et al., 2006). Lack of evidence confirming real OSH involvement of senior managers is the
consistent transformational management in the field of OSH attenuates compliance of what they preach with what they actually do when it
safety compliance and safety participation (Mullen et al., 2011). comes to making a choice between safety and productivity. Real OSH
As evidenced by simulation results, OSH leadership is also im- leaders always put safety first.
portant for the OSH MS's effectiveness. There are some practical con- It should be noted at the same time that insufficient attention is still
clusions resulting from this finding. being given to the problem of ensuring the active role of the leader in
First of all, this is the all-important message to all senior managers, the OSH MS. While there are many examples of good practices in the
who should be aware that effectiveness of the OSH MSs and the way literature that illustrate what management's involvement is and al-
individual processes operate in the system depend primarily on them. though there is a considerable number of research tools to measure it,
Any failure related to the operation of the system should be a stimulus there is also a lack of knowledge on how to build such involvement.
for senior company management to analyse their own activities and Unfortunately, too little attention has been paid so far to the ethical
consider the possibility of improving their own performance. In fact, aspect, including educational programmes addressed to senior man-
the root cause of failure always results to some extent from wrong agers (Quinlan, 1999). Strengthening the sense of personal responsi-
management decisions, including: approval of incomplete procedures bility of senior managers for the safety of employees is still needed. This
and instructions, employment of persons with insufficient qualifica- can be done not only by informing managers about the legal con-
tions, failure to provide necessary training, etc. Mature OSH leaders not sequences of acting unlawfully, but also by making them aware that
only inspire others to improve but above all they are able to have a their attitudes and decisions can either cause somebody harm or help
critical look at their own actions and accept responsibility. them get the job done safely.
Secondly, the research results constitute an important guideline for In this study, the FCM method was used to identify the OSH MS
OSH MS auditors, both internal ones and auditors of third parties, in- process that impacts System's effectiveness most. According to this
cluding certification bodies. When auditing systems, great importance method, the System was presented as a model composed of objects and
should be attached to the leadership process audit. It should be estab- interrelations that occur between them. The model was used to conduct
lished whether activities undertaken by OSH leaders are carried out simulations. The simulation results provided a forecast of how objects
regularly and whether they fit into the organization culture and support will change due to interrelations between them. The model built for the
the achievement of health and safety objectives. Nonconformities purpose of the present study included objects corresponding to pro-
should be thoroughly investigated, corrective actions should be re- cesses within the OSH MS and to safety performance, because this set of
commended and completed while potential for improvement should be objects enabled implementation of the research objective. It is worth
clearly identified. It must be emphasized however that auditing the noting, however, that the FCM method can also be used to support
leadership process differs significantly from other audits due to the maintenance and improvement of the OSH MS on a daily basis. The
specific relationship between the auditor and the audited leader. As model can be expanded by adding objects corresponding to various
soon as the internal audit ends, the former becomes an employee while variables influencing processes in the OSH MS, such as: new occupa-
the latter starts managing this employee's work and deciding about her/ tional hazards, implementation of new, safer production technologies,
his further professional career. On the other hand, after the certification sickness absence of employees, machine and equipment failures, etc.
audit a certification body begins to strive for a contract for the next The emergence of these variables in the company may exert a negative
audit whereas the senior manager is in power to either sign that contact or positive impact on individual processes in the OSH MS, and thus
or not. To put it shortly, once the audit ends auditors usually remain support or reduce the system’s effectiveness. At the same time, the
subordinate to senior managers. Therefore to ensure objectivity of multiplicity and diversity of these variables and the fact that each of
leadership process audits, it is indispensable to appoint experienced and them can influence several processes with varying strength, make it
assertive auditors who would not hesitate to discuss nonconformities difficult to decide which processes should be improved first in order to
and to suggest improvements in a firm manner. prevent decrease of the OSH MS effectiveness. Simulations on an ex-
Thirdly, considering the results of this research, it is worth analysing panded model could facilitate such decisions by delivering forecasts on
the leadership process also in the context of the latest ISO 45001 what will be the effects of the impact of a specific set of variables on
standard. Unlike the standards implemented so far, this one does not OSH MS processes and safety performance. It is therefore suggested to
require naming a person acting as a system representative. Obviously, further explore the potential of the FCM method for improving OSH MS
the representative significantly supports senior management in im- effectiveness.
plementing, maintaining and improving the system, however for some Despite the careful preparation and conduct of the research, its re-
senior managers, actions she or he undertook might have substituted sults are subject to some specific limitations, such as the impact of the
their own involvement in system operation. It should be emphasized subjective selection of only 15 processes which have been included in
that such practices would be absolutely inconsistent with the require- the FCM model. A more detailed selection of processes would probably
ments of existing standards, however presence of the system re- lead to results that could offer a better understanding of safety per-
presentative made abuses of this type easier. Cancellation of this role formance determinants, for example when dividing the leadership
challenges senior managers to perform their tasks within the system process into leadership attitude and leadership actions, it could be
personally. It should be followed by careful examination of the re- identified which of the two impacts more on the safety performance. On
presentative’s tasks and dividing them among senior management and the other hand, the development of the 16 × 16 interrelation matrix
other members of the staff. This, in turn, would result in establishing based on the subjective assessment of the expert group according to
new direct points of contact among people in the company. seven-item scale, turned out to be a difficult task. Although the

78
A. Skład Safety Science 117 (2019) 71–80

participation in the study was preceded by a comprehensive explana- Civil Aviation Safety Authority, 2006. Safety Management Systems. An aviation Business
tion of the project methodology and goals, the experts complained that Guide. Australia.
Clarke, S., 2013. Safety leadership: a meta-analytic review of transformational and
while they were making 240 assessments subsequently, it was difficult transactional leadership styles as antecedents of safety behaviours. J. Occup. Organ.
for them to maintain the same high level of concentration. Carrying out Psychol. 86, 22–49.
so many assessments was a tedious job, requiring a constant revision of Cooper, D., 2015. Effective safety leadership understanding types & styles that improve
safety performance. Saf. Manage. Peer-Rev., Prof. Saf. 60, 49–53.
previous evaluations in the light of subsequent ones. To sum up, the Dagdeviren, M., Yuksel, I., 2008. Developing a fuzzy analytic hierarchy process (AHP)
research was limited on one hand by the narrow range of elements model for behavior-based safety management. Inf. Sci. 178, 1717–1733.
making up the model, and on the other, by the experts' cognitive ability Demichela, M., Piccinini, N., Romano, A., 2004. Risk analysis as a basis for safety man-
agement system. J. Loss Prev. Process Ind. 17, 179–185.
to make reliable assessments. Dworniczak, P., 2003. Zbiory rozmyte dla początkujących. A record of the presentation
delivered at the XXX Szkoła Matematyki Poglądowej Osobliwości.
5. Conclusion Else, D., Beaumont, P., 2001. Expect the worst, achieve the best what is the role of
OHSMS in helping to secure healthy and safe workplaces? In: Pearse, W., Gallagher,
C., Bluff, L. (Eds.), Occupational Health & Safety Management Systems, Proceedings
Many enterprises certified for OHSAS 18001 or OSH MS national of the First National Conference Australia, pp. 33–40.
standards are most probably planning to implement ISO 45001 in the Espinosa-Paredes, G., Nunez-Carrera, A., Vazquez-Rodriguez, A., Espinosa-Martinez, E.-
near future. This certainly requires an analysis of the processes and G., 2009. Modeling of the High Pressure Core Spray Systems with fuzzy cognitive
maps for operational transient analysis in nuclear power reactors. Prog. Nucl. Energy
documents being implemented and their improvement according to the 51, 434–442.
requirements of the new standard. Significant amounts of labour force Fernández-Muńiz, B., Montes-Peón, J.M., Vázquez-Ordás, C.J., 2009. Relation between
and resources will probably be allocated by enterprises for training, occupational safety management and firm performance. Saf. Sci. 47, 980–991.
Frick, K., 2011. Occupational Health & Safety Management Systems – When are they
consulting services and implementation of such organizational changes. Good for Your Health? European Trade Union Institute.
Adopting a new standard in the field of OSH MS is certainly an excellent Georgopoulos, V.C., Malandraki, G.A., Stylios, C.D., 2003. A fuzzy cognitive map ap-
opportunity to draw attention of the whole business environment to proach to differential diagnosis of specific language impairment. Artif. Intell. Med.
29, 261–278.
health and safety issues. At the same time, it is easy to fall into the trap Granerud, L., Rocha, R., 2011. Organisational learning and continuous improvement of
of misdirecting efforts into fulfilling bureaucratic requirements instead health and safety in certified manufacturers. Saf. Sci. 49, 1030–1039.
of aiming for the overarching goal of coming up with a system which Gray, S.A., Zanre, E., Gray, S.R.J., 2014. Fuzzy cognitive maps as representations of
mental models and group beliefs. In: Papageorgiou, E.I. (Ed.), Fuzzy Cognitive Maps
would prevent injury and ill health of workers while providing them for Applied Sciences and Engineering, Intelligent Systems Reference Library 54.
with a safe and healthy workplace. Adoption of the new ISO 45001 Springer-Verlag, Berlin Heidelberg, pp. 29–48.
standard should in fact provide an incentive to companies to transform Groumpos, P.P., 2010. Fuzzy cognitive maps: basic theories and their application to
complex systems. In: Glykas, M. (Ed.), Fuzzy Cognitive Maps. Advances in Theory,
their own OSH MSs and make them more effective.
Methodologies, Tools and Applications. Springer-Verlag, Berlin Heidelberg, pp. 1–22.
The results of the study clearly indicate the OSH leadership process Hasle, P., Zwetsloot, G.I.J.M., 2011. Occupational Health and Safety Management Systems:
is the one that impacts OSH MS effectiveness to the greatest extent. And issues and challenges, editorial for the special issue on Occupational Health and Safety
though all OSH MS processes need to be constantly maintained and Management Systems. Saf. Sci. 49, 961–963.
ISO 45001, 2018. Occupational health and safety management systems – Requirements
improved and introduction of a new standard is naturally a good with guidance for use.
starting point to the revision of all processes that are implemented Kelloway, E.K., Mullen, J., Francis, L., 2006. Divergent effects of transformational and
within the system, it is suggested to pay special attention to leadership. passive leadership on employee safety. J. Occup. Health Psychol. 11, 76–86.
Kelly, B., Berger, S., 2006. Interface management: effective communication to improve
Stable, consistent and fair OSH leadership is the foundation of an ef- process safety. J. Hazard. Mater. 130, 321–325.
fective OSH MS. Kines, P., Andersen, L.P.S., Spangenberg, S., Mikkelsen, K.L., Dyreborg, J., Zohar, D.,
2010. Improving construction site safety through leader-based verbal safety com-
munication. J. Saf. Res. 41, 399–406.
Funding Kosko, B., 1986. Fuzzy cognitive maps. Int. J. Man Mach. Stud. 24, 65–75.
León, M., Rodriguez, C., García, M.M., Bello, R., Vanhoof, K., 2010. Fuzzy cognitive maps
This paper has been based on the results of a research task carried for modeling complex systems. In: Sidorov, G. (Ed.), Proceedings of the 9th Mexican
International Conference on Advances in Artificial Intelligence: Part I, pp. 166–174.
out within the scope of the fourth stage of the National Programme
Lu, C.S., Yang, C.S., 2010. Safety leadership and safety behavior in container terminal
“Improvement of safety and working conditions” partly supported in operations. Saf. Sci. 48, 123–134.
2017–2019 — within the scope of research and development — by the Makin, A.M., Winder, C., 2008. A new conceptual framework to improve the application
of occupational health and safety management systems. Saf. Sci. 46, 935–948.
Ministry of Science and Higher Education/National Centre for Research
Mohammadfam, I., Kamalinia, M., Momeni, M., Golmohammadi, R., Hamidi, Y.,
and Development, Poland. The Central Institute for Labour Protection – Soltanian, A., 2017. Evaluation of the quality of occupational health and safety
National Research Institute is the Programme’s main co-ordinator. management systems based on key performance indicators in certified organizations.
Saf. Health Work 8, 156–161.
Mullen, J., Kelloway, E.K., Teed, M., 2011. Inconsistent style of leadership as a predictor
References of safety behaviour. Work Stress 25, 41–54.
Pelaez, C.E., Bowles, J.B., 1996. Using fuzzy cognitive maps as a system model for failure
Akyuz, E., Celik, M., 2014. A hybrid decision-making approach to measure effectiveness modes and effects analysis. Inf. Sci. 88, 177–199.
of safety management system implementations on-board ships. Saf. Sci. 68, 169–179. Podgórski, D., 2005. Workers’ involvement - a missing component in the implementation
Asadzadeh, S.M., Azadeh, A., Negahban, A., Sotoudeh, A., 2013. Assessment and im- of occupational safety and health management systems in enterprises. Int. J. Occup.
provement of integrated HSE and macroergonomics factors by fuzzy cognitive maps: Saf. Ergon. 11, 219–231.
the case of a large gas refinery. J. Loss Prev. Process Ind. 26, 1015–1026. Podgórski, D., 2015. Measuring operational performance of OSH management system – A
Azadeh, A., Salehi, V., Arvan, M., Dolatkhah, M., 2014. Assessment of resilience en- demonstration of AHP-based selection of leading key performance indicators. Saf. Sci.
gineering factors in high-risk environments by fuzzy cognitive maps: a petrochemical 73, 146–166.
plant. Saf. Sci. 68, 99–107. Quinlan, M., 1999. Promoting occupation health and safety management systems: a
Basso, B., Carpegna, C., Dibitonto, C., Gaido, G., Robotto, A., Zonato, C., 2004. Reviewing pathway to success - Maybe. J. Occup. Health Safety - Australia New Zealand 15,
the safety management system by incident investigation and performance indicators. 536–542.
J. Loss Prev. Process Ind. 17, 225–231. Robson, L., Clarke, J., Cullen, K., Bielecky, A., Severin, C., Bigelow, P., Irvin, E., Culyer,
Bennett, D., 2002. Health and safety management systems: liability or asset? J. Public A., Mahood, Q., 2007. The effectiveness of occupational health and safety manage-
Health Policy 23 (2), 153–171. ment systems: a systematic review. Saf. Sci. 45, 329–353.
Bertolini, M., 2007. Assessment of human reliability factors: a fuzzy cognitive maps ap- Skeepers, N.C., Mbohwa, C., 2015. A study on the leadership behaviour, safety leadership
proach. Int. J. Ind. Ergon. 37, 405–413. and safety performance in the construction industry in South Africa. Proc. Manuf. 4,
Bevilacqua, M., Ciarapica, F.E., Mazzuto, G., 2012. Analysis of injury events with fuzzy 10–16.
cognitive maps. J. Loss Prev. Process Ind. 25, 677–685. Stach, W., Kurgan, Ł., Pedrycz, W., 2010. Expert-based and computational methods for
Bottani, E., Monica, L., Vignali, G., 2009. Safety management systems: performance dif- developing fuzzy cognitive maps. In: Glykas, M. (Ed.), Fuzzy Cognitive Maps.
ferences between adopters and non-adopters. Saf. Sci. 47, 155–162. Advances in Theory, Methodologies, Tools and Applications. Springer-Verlag, Berlin
Bottomley, B., 1999. Occupational Health & Safety Management Systems: Strategic Issues Heidelberg, pp. 23–42.
Report. A report prepared for the National Occupational Health and Safety Stylios, C.D., Georgopoulos, V.C., Groumpos, P.P., 1997. Introducing the theory of fuzzy
Commission, Commonwealth of Australia. cognitive maps in distributed systems. In: Proceedings of 12th IEEE International

79
A. Skład Safety Science 117 (2019) 71–80

Symposium on Intelligent Control, Istanbul, Turkey, pp. 55–60. work environment and health: a prospective study. Appl. Ergon. 37, 775–783.
Stylios, C.D., Groumpos, P.P., 2000. Fuzzy Cognitive Maps: a model for intelligent su- Wu, T.C., Chen, C.H., Li, C.C., 2008. A correlation among safety leadership, safety climate
pervisory control systems. J. Intell. Fuzzy Syst. 8, 83–98. and safety performance. J. Loss Prev. Process Ind. 21, 307–318.
Stylios, C.D., Groumpos, P.P., 2004. Modeling complex systems using fuzzy cognitive Xirogiannis, G., Stefanou, J., Glykas, M., 2004. A fuzzy cognitive map approach to support
maps. IEEE Trans. Syst., Man, Cybernet. – Part A: Syst. Humans 34, 155–162. urban design. Expert Syst. Appl. 26, 257–268.
Torp, S., Moen, B.E., 2006. The effects of occupational health and safety management on

80