2-Azadeh

JRHS 2009; 9(2): 10-18

Copyright © Journal of Research in Health Sciences

The Evaluation of Importance of Safety Behaviors in a Steel Manufacturer by Entropy

Azadeh A(PhD)a,  Mohammad Fam I(PhD)b

a Research Institute of Energy Management and Planning, Dept. of Industrial Engineering and Dept. of Engineering Optimization Research, Faculty of Engineering, Tehran University, Iran

b Department of Occupational Health and Faculty of Health, Hamadan University of Medical Science, Iran

*Corresponding Author: Dr Iraj Mohammadfam, E-mail: hammadfam@umsha.ac.ir

Received: 18 July 2009; Accepted: 11 September 2009

Abstract

Background: This study aimed to evaluate the workers safety behavior and to determine the impor­tance of each unsafe behavior in an Iranian steel manufacturing company.

Methods: This study was conducted in Mobareke steel manufacturing company, which is located in the middle of Iran, in 2007. The methodology was based on the safety behavior sampling (SBS) tech­nique and entropy. After specifying the unsafe behaviors and with reference to the results of a pilot study a sample of 3248 was determined, with a sampling accuracy of 5% and confidence level of 95%.

Results: The results indicated that 41.8% of workers behaviors were unsafe. The most frequent un­safe behaviors were inappropriate use of personal protective equipments (PPEs) with 32% of total un­safe behaviors. The results also notified a significant relationship between age, job experience and educa­tional level on unsafe behaviors (P< 0.05). The highest weight, which is obtained by entropy, belongs to using inappropriate tools with weight of 0.1425. The ultimate findings of the study showed that a considerable number of workers' behaviors were unsafe, which is one of the main antecedents of in­dustrial accidents.

Conclusion: Considering catastrophic consequences of accidents in steel manufacturing industry, the results emphasize on diminishing unsafe behaviors and recommends applying behavior based safety principles.

Keywords: Safety behavior sampling, Unsafe acts, entropy, Steel manufacturing company, Iran

Introduction

Although the development of science and technology has already decreased the num­ber of employees in industries, there has been a developing trend in terms of em­ployees' importance in workplaces (1). Con­trolling a large number of different and criti­cal opera­tions is the duty of human beings in modern industries. It is usually assumed that making errors is one of the main contributors to cata­strophic disasters likelihood (2). Dis­astrous ac­cidents like Chernobyl, Three Mile Island and Boo pall are all examples of these kinds (3).

Due to the catastrophic consequences of such accidents, human beings always try to take controlling measures and reduce the po­ten­tial risks (4). Before the 1930's safety spe­cialists followed the prevention ap­proaches by using physical methods such as machine guarding, housekeeping and in­spection pro­grams (5). Until that time, it was believed that the main causes of industrial accidents were unsafe conditions and physi­cal hazards such as heavy equipment, tren­ches, mechani­cal explosions, ionizing radia­tion, flamma­bil­ity, corrosion, reactivity, fast moving ve­hicles, steep grades, uneven sur­faces etc. It was in the early years of 1930's when the con­cept of unsafe acts and their role in causing industrial accidents were in­troduced (5) and the theory of" human be­ings as the first an­tecedents (trigger reason) of accidents by doing unsafe acts" was pro­pounded by Heinrich in his book "prevention of industrial accidents" (6). Heinrich stated that roughly 88% of all accidents were caused by human errors (3). Drew estimated that 80% to 90% of the ac­ci­dents were caused by human errors (7). In ad­dition, Reynard and Billings came to this con­clu­sion that human's unsafe acts caused 70% to 90% of the accidents (7). This drew psy­chologists and safety specialists' attention to unsafe acts as the most probable cause of frequent accidents happening in industries. In order to diminish the likelihood of such accidents, this group of specialists empha­sized the behavior of employees using beha­vior science techniques (8). Social psycholo­gists recognize "attitude" as the most im­por­tant factor to predict employees' beha­viors. In other words, these efforts led to in­itiation and development the "behavioral based safety" approach.

These studies present an entropy method for evaluation of unsafe behavior in steel manu­facturer. Moreover, previous studies con­cen­trate and use Delphi method for evalua­tion of importance of unsafe behavior. How­ever, we claim that entropy provides more reliable solutions in context of the impor­tance of un­safe behavior than which is a good substitute for expert judgment. This is quite important in situations, where there are no expert judg­ments available or is also im­possible to use ex­pert judgment for evalua­tion of unsafe behavior.

The purpose of this investigation was to spe­cify the type, proportion, and importance of unsafe acts in employees' behaviors. Fur­ther­more, the relationship between unsafe beha­viors and employees' demographic cha­racte­ristics such as age, education, job expe­rience, and marriage status was examined. It is also worth mentioning that in this re­search, an un­safe act is defined as a behavior that is com­mitted without considering safety rules, reg­u­lation, standards and specified criteria in sys­tem, which can affect the sys­tem safety level (9).

Methods

This study was conducted in the Operational Department of Mobareke Steel Manufactur­ing Company, which is located in the middle of Iran, in 2007. Safety Behavior Sampling (SBS) technique was employed to conduct this study. SBS is a technique of measuring unsafe acts and is based on the laws of prob­ability (10, 11).

In order to obtain a complete and accurate picture of safe/unsafe acts performed by the worker, it is necessary to observe conti­nuously the worker and record data related to unsafe acts (12). Note that a sufficiently large sam­ple must be obtained for represent­ative re­sults (13) and consists of previous acci­dents records, including disabling inju­ries, re­cord­able injuries and first aid cases, interviews with the managers and experts of the depart­ment and the review of the related doc­u­ments. The obtained list was adjusted based on present conditions such as type and na­ture of work, reviews of accidents reports, present cultural conditions, and a number of related factors.

After specifying the unsafe acts, a number of necessary observations of workers' behaviors were carried out in order to determine the proportion of their unsafe acts.

The number of observations required is based on data collected during the pilot study, the degree of accuracy required, and the given level of confidence.

Total number of required safety behavior observations is derived from (14):

     (Eq. 1)

For a given level of confidence K, the value of K is read from the standardized normal tables. For 95% confidence, K is approx­imated as 2, and for 99% confidence, K is taken as 3.

Confidence level means that the conclusions will be representative of the true population 95% of the time. Accuracy may be inter­preted as the tolerance limit of the observa­tions that fall within a desired confidence level. 5% ac­curacy with 95% confidence level is the com­bination often used in safety beha­vior sam­pling. This means that 95% of the time within 5% accuracy limit, the con­clu­sion drawn based on safety behavior sam­pling will be representative of the actual po­p­ulation (15).

After conducting a pilot study the proportion of unsafe acts were estimated to be about 33%. Considering that 5% accuracy with 95% confidence level is the combination, which is often used in safety behavior sam­pling, the total number of observations was estimated to be 3248.

Safety behavior sampling needs to be done randomly. This is achieved when each pe­riod of the workday is equally selected as the observation period. Therefore, in the next stage the observations are done randomly. This means that both observed workers (64 workers of operational department) and fre­quency of observations (in the period of 8 hours from 7 to 15) were selected randomly.

Since the behavior of human beings might be changed from time to time, the observa­tion duration has a vital role in accuracy of the results. This duration should be as short as possible to observe and specify the beha­viors. In this research, the average of each duration was 2 seconds.

The observations were carried out randomly by the researcher while the subjects were not aware of the fact that they were being ob­served.

In order to recognize the relationship be­tween the employees' demographic characte­ristics and unsafe behaviors, the mentioned variables such as age, work experience, edu­cation, previous accidents records and mar­riage status were collected through inter­views and a special questionnaire. In each ques­tion­naire, there were questions about age, edu­cation, job experience, previous ac­cidents re­c­ords. Having been chosen ran­domly, sub­jects were questioned by the re­searcher and their answers were recorded.

It is worth noting that the collected data were analyzed with SPSS and was tested by Kruskal-Wallis test and one-way ANOVA.

In the previous studies to find the impor­tance of unsafe acts DELPHI method and application of questionnaire were used, which were based on concept and back­ground of the experts, but in this study to find the im­portance of each unsafe act, a mathe­mati­cal method, entropy, which was more reliable, was applied.

Entropy method

Entropy is a major conception in physics, so­cial science, and information theory, which shows the amount of uncertainty in an ex­pected informational content of a message. In another word, entropy in information the­ory is a criterion for uncertainty that is ex­plained by a discontinuous probability distri­bution (pi). This uncertainty is calcu­lated as followed:

   (Eq. 2)

K is a positive constant variable in order to supply 0 E 1.

E is calculated from probability distribution pi by statistical mechanism and is maximum if all of pis Pi=1/n is same. Therefore:

    (Eq. 3)

A decision-making matrix of a MADM model contains data that entropy can be used as a criterion to evaluate them. A decision-making matrix is showed below.

A decision-making matrix


x1

x2

xn

A1

r11

r12

r1n

A2

r21

r22

r2n

Am

rm1

rm2

rmn

The available data in the decision-making ma­trix will be so normalized:

      (Eq. 4)

And for Ej from pij set in lieu of every specification we will have:

     (Eq. 5)

That is K=1/Lnm.

Now uncertainty or deviation degree (di) from obtained data in lieu of the jth specifi­cation is so: 

       (Eq. 6)

Finally for weights (wj) of existed specifica­tion we will have:

               (Eq. 7)

If decision maker already has a conceptional judgment (λj) as the relative importance for the jth specification, then wj will be modified as followed:

       (Eq. 8)

In this research in order to distinguish of un­safe acts in sub-companies following steps were done:

Determining the frequencies of each unsafe act in every sub-company of the steel manu­facturing company and filling the matrix.

First calculate the sum of each column (i= 1, …, 7), then divide each datum by column sum of that datum to obtain matrix of pij.

Calculate the Ln of each datum of pij matrix.

Each grid of new matrix must be multiplied by the same grid in pij matrix.

Calculate the sum of each column in this new matrix. There will be 20 numbers will be obtained for 20 unsafe acts, which will be multiplied by -0.514 (-1/Lnm). This new numbers are Ej.

Here there is no need to dj and wj will be calculated as followed:

       (Eq.9)

Results

A total of 3248 observations were conducted in this study. From these observations, the proportion of unsafe acts was 41.8%. Among unsafe acts, inappropriate use of personal pro­tective equipments (PPEs) was allocated it­self as the largest proportion (32%) of the un­safe acts. Application of inappropriate tools and settling in inappropriate place with 14% and 13% of all unsafe acts stood re­spectively in the second and third grade. Employees from 40 to 49 yr, were the most dominant and fre­quent age group among ob­served subjects. They made up 46% of all sampled population. On the other hand, sub­jects above 60 had the low­est frequency with 2.6%. The results and fre­quencies of age groups were shown in Table1.

Table1: Frequencies of individual according to age, education and work experience



(%)

Age groups

22-29

15.3

30-39

29.6

40-49

46

50-59

6.5

X 60

2.6

Education

Primary school

16.1

Junior high school

23.2

High school

48.9

Academic

6.5

Work experience (Year)

0-5

44

6-11

17

12-17

4

18-23

21

24-35

14

Considering marriage, 91.2% of the em­ploy­ees were married and rests of them were sin­gle. Moreover, regarding the education, the em­ployees with high school education de­grees had the largest proportion with 48.9%. The employees with academic edu­cations al­located themselves to the least pro­portion with 6.5% (Table 1).

The results also signified that the work expe­rience of the workers varied from 0.5 to 35 yr and among them, those whose work experi­ence was less than 5 yr formed the most pro­portion of the observed employees with 44% (Table 1).

A Kruskal-Wallis test was used to evaluate the effect of age, education, marital status, and work experience on unsafe act. The re­sults are shown in Table 2.

Table 2: Effect of age, education, marital status, and work experience on unsafe actParameter


f

significance

H0

Experience

52.6

0.009

accepted

Age

59

0.012

accepted

Educational level

8.87

0.033

accepted

marital statue

0.073

0.878

Not accepted

statue



accepted

Frequencies of each unsafe act in every sub-company of the steel manufacturing com­pany are shown in Table 3. Accordingly, the rela­tions between age and work experience on the number of unsafe behaviors, are sta­tisti­cally significant (P< 0.05). The results showed an inverse relationship between the unsafe behaviors with age and also for work experi­ence. Which means, as the employees get older; the number of unsafe behaviors is reduced.

Table 3: Frequencies of each unsafe of sub-company

Company type of unsafe  acts

1

2

3

4

5

6

7

Inappropriate PPE

12

2

25

0

2

0

1

Using inappropriate tools

3

7

4

0

1

2

1

Settling in a dangerous place

2

0

12

0

0

3

0

Moving under hanging load

3

5

0

3

0

0

2

Inappropriate posture

1

1

5

0

0

0

0

Work interference

0

0

0

1

2

1

0

Making tools unsafe

0

1

2

0

0

1

0

Unsafe  load transfer

2

0

0

1

0

0

0

Working with defective machine

1

0

1

0

0

0

0

No attention to cranes tocsin

0

2

1

0

0

0

0

Running or jumping from height

0

0

0

0

1

1

0

Unallowable presence in cranes cabin

0

0

0

2

0

0

0

Horseplay during work

1

0

1

0

0

0

0

Washing tandish with alcohol during work

0

0

0

0

0

2

0

Lack of control during metal casting

0

0

2

0

0

0

0

Misuse of compacted air

1

0

0

0

0

0

0

Dangerous driving

0

1

0

0

0

0

0

Moving indirectly

0

0

1

0

0

0

0

No riggers

0

0

1

0

0

0

0

Not using colored signs during sampling

0

0

1

0

0

0

0

To determine the importance of each unsafe behavior, their weights calculated with en­tropy method. The highest weight belongs to “us­ing inappropriate tools” with weight of 0.1425. The obtained weights are showed in Table 4.

Table 4: calculating the importance of each unsafe behavior with entropy

Type of behavior

Obtained weight by entropy

Inappropriate PPEs

0.095

Using inappropriate tools

0.1425

Resting in unsafe place

0.073

Moving under Suspended load

0.121

Awkward Posture

0.072

Work interference

0.094

Making tools unsafe

0.094

Unsafe load handling

0.057

No attention to cranes alarm

0.058

Working with unsafe machine

0.063

Running or jumping

0.063

Unauthorized presence in cranes cabin

0.00009

horse playing

0.063

Washing tandish with alcohol during work

0.00009

Lack of control during metal casting

0.00009


Work with compressed air

0.00009


Dangerous driving

0.00009


Moving indirectly

0.00009


No rigger

0.00009


Not using colored signs during sampling

0.00009

Discussion

The results of the current research in gas treat­ment company indicated that a large num­ber of employees behaviors were un­safe (41.8%) which seems to be quite less than the results of previous studies. The rate of unsafe behav­iors in other researches in a foundry and a metal working company in Iran were 59.2% and 27% respectively (16).

The consequences of unsafe behaviors de­pend on different factors such as the nature of the tasks and the type industry. From safety specialists point of view, the steel manufac­turing company is a critical work­place due to its high complexity, low flexi­bility, and high vulnerability towards acci­dents. Although in the studied company the proportion of un­safe behaviors is approxi­mately low, the risk of such behaviors is un­acceptable due to their se­rious consequences (16), thus the aforemen­tioned proportion of 41.8%, as a marginal value in a Steel manu­facturing Company, consid­ered unaccept­able.

The most frequent and important behavior was inappropriate use of personal protective equipments (PPEs) with 32% of all unsafe acts. Inappropriate uses of PPEs have always been one of the basic factors in accidents. The use of inappropriate clothes and gar­ments re­ported as one of the 6 basic triggers of acci­dents from 1994 to 2003 in Iran (16). Plenty of reasons can be mentioned for inap­propri­ate use of PPEs such as lack of work­ers knowl­edge about workplace hazards and PPEs, ignoring workers opinions in selecting and purchasing PPEs and insufficient super­vision in terms of using PPEs properly (17).

Modern safety approaches lay a great em­phasis on identifying and controlling the haz­ards by administrative and engineering prac­tices. According to this, controlling methods that directly depend on workers level of ac­ceptance and participation (such as using of PPEs) are not in top priorities and should be taken as the last resort (17). An important and effective factor for PPEs programs to be successful is employees ac­ceptance and par­ti­ci­pation (17).

Without considering this issue, it is almost obvious that, in spite of all plans, policies, or measures, not only the PPEs programs can­not be successful but also they can have some undesirable results.

Some factors that might influence employ­ees acceptance are their frequent participa­tion in selecting proper equipment, con­duct­ing training and retraining programs on main­taining, cleaning, and using PPEs. A comple­mentary study in company notified that 79.8% of the workers believe that the use of safety equipment in workplace is nec­essary, mean­while 33.9% of them had de­veloped this opin­ion that the PPEs are mainly uncomfortable and 32% of them be­lieved old, worn out and expired PPEs were not substituted with new ones regularly (16).

In summary, the main reasons related to high frequency of unsafe behavior occurrence in terms of PPEs uses are:

Selecting PPEs without considering task safety analysis, employees characteristics and pre­sent hazards. Lack of appropriate trainings about hazards  communication.  Insufficient    participation of personnel in PPEs programs.

It is concluded that 59% of all unsafe be­hav­iors in this steel manufacturing company con­sist of inappropriate use of personal pro­tec­tive equipments, application of inappro­priate tools and settling in inappropriate place. With more attention to the antece­dents, the num­ber of unsafe behaviors can be reduced re­sulting in amore efficient accident preven­tion system.

Having studied the relationship among dif­ferent variables and the number of unsafe acts these results were obtained:

There is an inverse relationship between un­safe behaviors with age. As employees grow older, the proportion of unsafe behaviors is reduced. It might be related to the higher work experience and workmanship level and the fact that older employees are usually more skillful.

It is a general view that adolescents are more likely to take risks than middle-aged and older people are. This opinion is supported by re­sults from traffic studies, which have shown that young drivers tend to drive faster, fol­low with shorter headways, and not wear seat belts as often as older drivers (18, 19).

There is an inverse relationship between un­safe behaviors with education. Employees with higher educational level behave safer than low educated personnel.

Accidents tend to accumulate on new inex­perienced workers (20). For example, the risk of a woodworker having an accident on his/ her first day on the job can be as much as 50 times higher than that of a worker with 1 years work experience (21). The accident risk gener­ally decreases as work experience increases (22).

There was also a significant relationship be­tween work experience and previous acci­dents. This implies the fact that the more work experience people have, the more acci­dent they might have experienced.

No considerable relationship was found be­tween unsafe behaviors and marital status P>0.05).

In order to improve safety behavior of work­ers, a comprehensive program must be intro­duced.

This could be comprised of implementation of appropriate safety management systems, iden­tify and correct unsafe conditions such as temperature or humidity extremes, un­gu­arded equipment, uncovered floor open­ings, safety training programs, lecture series etc (23, 24). The safety behavior sampling study may be conducted on a weekly basis during and upon the completion of the pro­gram. The safety behavior control chart for each period follow­ing the beginning of the program will show if a significant improve­ment in unsafe behavior has been achieved. Modification of the pro­gram or its compo­nents may be carried out as long as the un­safe behavior is being reduced.

Once the minimum number of unsafe be­havior has been achieved (i.e. p), the behav­ior sam­pling study may be repeated and the ob­tained data plotted on the control chart to assure that the frequency of unsafe behaviors remain at the desired minimum level.

Considering the results, the following items are suggested:

1. Employing task risk analysis methods to screen and determine risky jobs in order to perform ergonomic evaluations and appro­pri­ate interventions.

2. Setting and implementing an executive sys­tem to accomplish PPEs programs suc­cessfully.

Such programs mainly include appropriate se­lecting, maintenance and cleaning of PPEs.

3. Design and implementation of accident prog­noses tests before employment in order to rec­ognize and screen employees with higher na­tural tendencies in causing acci­dents. This might prevent such employees from doing cri­tical (safety concerned) jobs.

4. Planning and conducting safety-training programs based on behavioral based safety in steel manufacturing company in order to im­prove unsafe behaviors and change false safety attitudes consequently.

5. Design and implementation of punishment and award system considering employees pat­terns of behaviors.

6. Periodic evaluation of workers behaviors in order to provide proper inputs for inter­ven­tions and measuring their effectiveness.

7. Implementation of a risk management sys­tem to determine the risk of unsafe be­haviors and presenting suitable engineering and admin­istrative controlling methods.

Acknowledgements

The authors are grateful for the valuable help and support received from Dr. Hossein Mo­dare­sifar and all those who helped in the study. The authors declare that there is no conflict of interests.

References

  1. Lund J, Aaro E. Accident prevention. Presentation of a model placing em­pha­sis on human, structural and cul­tural factors. Safety Science. 2004; 42(2): 271-324.
  2. Haslam A. Targeting ergonomics in­terventions-learning from health pro­motion. Applied Ergonomics. 2002; 33(3): 241-49.
  3. Azadeh MA. Creating highly reliable manufacturing systems: An inte­grated approach, International journal of re­li­ability. Quality and safety en­gineer­ing. 2000; 7(3): 205-25.
  4. Fuller W. An assessment of the rela­tionship between behavior and injury in the workplace: A case study in professional football. Safety Science. 2005; 43(4): 213-24.
  5. Jensen, PL. Risk assessment - a regu­latory strategy for stimulating work­ing environment activities? Human Factors and Ergonomics in Manu­facturing. 2001; 11(2): 101-16.
  6. Kannapin O, Pawlik K, Zinn F. The pattern of variables predicting self-re­ported environmental behavior. Jour­nal for Experimental Psychol­ogy. 1998; 45(4): 365-77.
  7. Caird J, Kline T. The relationship be­tween organizational and individ­ual variables to on-the-job driver ac­ci­dents and accident free kilometers. Ergo­no­mics. 2004; 47(15): 598-613.
  8. Varonen U, Mattila M. Effects of the work environment and safety activi­ties on occupational accidents in eight wood-processing companies. Human Factors and Ergonomics in Manufac­turing. 2002; 12(1): 1-15.
  9. Fuller W. An assessment of the rela­tionship between behavior and injury in the workplace: A case study in pro­fessional football. Safety Science. 2005; 43(4): 213-24.
  10. Cox S, Jones B, Rycraft H. Behav­ioral approaches to safety manage­ment within UK reactor plants. Safety Sci­ence. 2004; 42(7): 825-39.
  11. Armitage J, Conner M. Efficacy of the theory of planned behavior: a meta-analytic review. Br J Soc Psy­chol­ogy. 2001; 40(3): 471-99.
  12. Cooper D. Implementing the behav­ior based approach to safety: a prac­tical guide. The Safety and Health Prac­titioner. 1994; 12(11): 18-23.
  13. Gherardi S, Nicolini, D. Learning the trade: a culture of safety in prac­tice. Organization. 2002; 9(2): 191-223.
  14. Beevis D, Slade M. Ergonomics- costs and benefits. Appl Ergon. 2005; 34(5): 413-18.
  15. Stead M, Tagg S, MacKintosh M, Fadie D. Development and evalua­tion of a mass media Theory of Planned Behavior intervention to re­duce speed­ing. Health Educ. Res. 2005; 20(5): 36- 50.
  16. Arshi S. Prevention oriented epide­mi­ological study of accidental burns in rural areas of Ardabil, Iran. Burns. 2006; 32(1): 366-71.
  17. Sheeran P, Silverman M. Evaluation of three interventions to promote work­place health and safety: evi­dence for the utility of implementa­tions. Social Sci­ence & Medicine. 2003; 56(5): 2153-163.
  18. Evans L, Wasielewski P. Risky driv­ing related to driver and vehicle char­ac­teristics. Accident Analysis and Pre­vention. 1983; 15(2): 121-36.
  19. Blasco D, Prieto M, Cornejo M. Ac­ci­dent probability after accident oc­currence. Safety Science. 2003; 41(6): 481-501.
  20. Kristensen S. Workplace intervention studies. Occupational Medicine. 2000; 15(1): 293-305.
  21. Larsson TJ. Risk and the inexperi­enced worker: Attitudes of a social anthropologist. J Occup Health Safety, 1988; 4(1): 35-40.
  22. Butani J. Relative risk analysis of in­juries in coal mining by age and ex­perience at present company. Journal of Occupational Accidents. 1998; 10(3): 209-16.
  23. Roberts E. Occupational mortality in British commercial fishing, 1975-95. Occup Environ Med. 2002; 61(5): 16-23.
  24. Alexander J, Barham P, Black I. Fac­tors influencing the probability of an incident at a junction: Results form an interactive driving simulator. Ac­cident Analysis and Prevention. 2002; 34(6): 779-92.


JRHS Office:

School of Public Health, Hamadan University of Medical Sciences, Shaheed Fahmideh Ave. Hamadan, Islamic Republic of Iran

Postal code: 6517838695, PO box: 65175-4171

Tel: +98 81 38380292, Fax: +98 81 38380509

E-mail: jrhs@umsha.ac.ir