Assessing ergonomic risk factors using combined data envelopment analysis and conventional methods for an auto parts manufacturer

Abstract

BACKGROUND:

Occupational injuries are currently a major contributor to job loss around the world and are also costly for businesses. The absence of rational analysis is felt in this area, so mathematical analysis is needed to obtain the logical results of these injuries in order to find gaps or loss points of the industry.

OBJECTIVE:

This paper assesses the effects of five demographic factors on ergonomic risk and occupational injuries using an integrated mathematical programming approach. The obtained results will help managers to carry out any required corrective actions or establish benchmarks.

METHODS:

Two typical ergonomic risk assessment methods, the Quick Exposure Check (QEC) and Rapid Entire Body Assessment (REBA), are applied to assess musculoskeletal disorders in workers. Then, considering the five demographic factors as input variables and risks computed by the QEC and REBA methods as outputs, final risk rates for each decision-making unit (DMU) are calculated using data envelopment analysis (DEA). The importance and weight of each risk factor is analyzed using statistical techniques and analysis of variance (ANOVA). To demonstrate the applicability of the methodology, it is applied to a large auto parts manufacturer.

RESULTS:

The results revealed that the information obtained by DEA is consistent with those for REBA and QEC, which shows that education, experience and weight are factors that could help reduce ergonomic risks.

CONCLUSION:

It is shown that demographic variables such as age, height, weight, education and work experience play an important and effective role in the explanation of ergonomic risk factors.

Keywords

Demographic features musculoskeletal disorders risk assessment occupational injuries data envelopment analysis ergonomic risk factors

1 Introduction

At present, advances in the technology and industrial growth of countries has given manpower an undeniable role in these developments, and it is also a basic pillar of the production and economies of countries. But different types of work entail accepting the possibility of accidents. Progress in industrial processes, the use of electricity and machines, and contact with chemicals and the like also lead to many accidents for the workforce. Every year millions of accidents occur around the world, and the number of injuries in industrial contexts is ever-increasing [1]. Some studies have shown that employers tend to lose a substantial number of work days because of time taken off by workers for treatment of work-related injuries [2, 3].

Ergonomic risk factors are aspects of a job that can cause biomechanical stress on employees. These include highly repetitive tasks, awkward postures, forceful exertion, static postures over long periods, cold temperatures, and localized pressure on body parts. Exposure to one or more ergonomic risk factors could cause or contribute to musculoskeletal disorders (MSDs). A study of the relationship between posture and workplace productivity showed that 75% of the participants who trained extensively on improving posture habits reported decreased back pain and felt more productive at work (click safety group). The costs associated with managing injuries varied according to several factors, including where care was sought, type of injury sustained, and category of construction work [4]. These painful and often disabling injuries generally develop gradually over weeks, months, and years. Musculoskeletal disorders also usually result from exposure to multiple risk factors that can cause or exacerbate the disorders, not from a single event or trauma such as a fall, collision, or similar incident [5].

The issue of worker health has recently attracted the attention of managers and ergonomics experts, because improvements in the performance of industry greatly depends on improving staff working conditions. This paper attempts to explore the problems leading to reduced productivity of workers caused by unsuitable body postures. The aim of this study is to assess the ergonomic risk factors of age, education, job experience, weight and height. This is done by using an integrated mathematical programming approach and two well-known methods, Rapid Entire Body Assessment (REBA) and the Quick Exposure Check (QEC). The REBA technique is suitable for evaluating jobs that involve dynamic and static postures. and is also suitable for evaluation of varied jobs. The QEC is a holistic method for carrying out risk assessment for work-related musculoskeletal disorders. The present study also statistically investigates how each of the considered factors influences risk rates and identifies the most influential. The results will help managers, experts on ergonomics, and productivity engineers to evaluate and monitor workplaces and develop methods to control and prevent potential job injuries and increase worker efficiency. To the best of the authors’ knowledge, this is the first study that conducts risk assessment considering ergonomic factors by using the data envelopment approach (DEA).

2 Literature review

In the industrial and services sector of various countries, there appears to be little knowledge of ergonomics, even though it is essential to the creation of many management and engineering policies. The purpose of such policies is generally the application of ergonomics to achieve a suitable and logical relationship between employees and their work environments, the machines they use, and organizations. Most definitions of this field have been focused on work quality, physical dimensions, and cognitive and organizational features. Ergonomists try to increase worker safety and satisfaction by minimizing negative factors, which enhances productivity. However, the basic concepts have focused more on human physical factors, with mental factors added later. This approach is called micro-ergonomics, which is the approach to traditional ergonomics [6].

Several studies have assessed working conditions using various traditional methods such as Rapid Upper Limb Assessment (RULA), REBA, the Ovako Working Analysis System (OWAS) and the QEC. Others have focused on the effects of demographic factors on work-related injuries. Nadri et al. [7] compared the ergonomic risk assessment results of the QEC and REBA in an anodizing industry. They concluded that the results of risk assessment using these two instruments for the entire body did not agree. Therefore, it is not possible to apply them interchangeably for postural risk assessment, at least not in the industry studied. However, in the present study, both of these instruments were adequate because most of the working postures involved the upper half of the body. In a similar study, Qutubuddin et al. [8] aimed to assess ergonomic risk using postural analysis tools such as REBA and the QEC in a bus body building unit. The results showed that a significant proportion of the workers were working in high-risk postures and were at moderate to high risk of work-related musculoskeletal disorders (WMSDs). The study also confirmed the results of REBA and the QEC and recommended implementation of an ergonomics intervention program raise employee awareness and training workers to reduce the risks of WMSDs. Arezes et al. [9] investigated the accuracy of both the REBA and QEC methods compared with all the other methods, none of which take into account worker perceptions. That was the main reason for the decision in this study to apply REBA, which is powerful postural analysis tool, and the QEC, which includes worker perceptions.

Yuan [10] studied the results of pre- and post-training ergonomics knowledge tests among librarians and found significant improvements in their understanding of ergonomics principles. This study identified ergonomic risks using RULA and REBA, and also used focus group discussions. The authors concluded that the study accomplished its overall objective of helping librarians improve ergonomics in their workplaces. This study utilized a participatory ergonomics approach to examine ergonomic risks and reduce musculoskeletal symptoms. These authors correctly used two traditional assessment methods, but did not use any type of mathematical model to support the results of their evaluation, which could have resulted in making changes. Chiason et al. [11] presented a comparison of eight methods for determining risk factors for work-related musculoskeletal disorders: the Quick Exposure Check; the Ergonomic Workplace Analysis method developed by the Finnish Institute of Occupational Health; the Hand Activity Level (HAL) threshold limit values developed by the American Conference for Government Industrial Hygienists; the Job Strain Index (JSI); the OCRA index; the EN 1005-3 standard; and the Rapid Upper Limb Assessment and Rapid Entire Body Assessment These methods were used to assess 224 workstations involving 567 tasks in various industrial sectors. The findings revealed that the various methods differed in their analyses of the same workstations. These results provided a better understanding of the differences between various risk assessment methods and can help practitioners choose the best and most suitable methods for case studies and prior to ergonomic interventions in industry.

Lovell and Rouse [12] proposed a new super-efficiency model relative to DEA that: (a) generates the same super-efficiency scores as conventional super-efficiency models for all units having a feasible solution under the latter; and (b) generates a feasible solution for all units not having a feasible solution under the latter. This is an equivalent model in which super-efficiency scores can be obtained using the standard CCR and BCC models. One advantage of this model is that it allows users to employ conventional DEA software. A second advantage is that this model is guaranteed to generate feasible solutions for all decision-making units (DMUs). This model was the inspiration for analyzing our specific DMUs. Sigala [13] illustrated the value of stepwise data envelopment analysis for measuring and benchmarking hotel productivity; this author extended current DEA applications by developing a stepwise approach. The latter technique combines correlation and DEA analysis to develop robust DEA models. Its advantages were illustrated by applying it in a dataset of three-star hotels in the United Kingdom. Similar to this study, the present study used correlation and DEA. One difference was that other assessment tools and statistical methods were added. Another was the goal of finding the gaps in industry from the manpower point of view and increasing human resource productivity, which affects total productivity. Chen et al. [14] maintained that the traditional DEA model results in a dual variable to associate with the normalizing equation. This implies that the adjustment proportions of all input or output factors are the same for efficiency improvement. This study modified the original DEA model by decomposing the normalizing equation in order to associate it with different dual variables. As a consequence, the adjustment proportion of each input or output factor can be different to improve efficiency. Many developments have occurred in the traditional DEA model; each has its advantages and increasingly optimizes the model. One limitation of this study is that it just focused on optimizing DEA to the point where it could be more efficient for future assessments; the authors did not follow up by using other statistics or analysis tools.

Wier et al. [15] assessed environmental performance across product types and across household types in order to evaluate environmental pressure from human activity They combined family budget statistics, input–output tables, energy and material flow matrices, various types of emissions, and environmental effect indices for various effect types (e.g., a global warming potential index, an ozone depletion potential index, etc.). In this study, DEA was used to weight these environmental effects indices in order to create one environmental performance score for each family type and product type. The results indicated that the environmental performance of each family type changed considerably across environmental effect types. Chiason et al. [11] found that the work-related musculoskeletal risk factors most often cited in the literature include repetition, application of excessive force [16, 17], vibration and awkward postures [18]. Based on compelling evidence, research has reported clear links between these risk factors and the prevalence of MSDs [19 –22]. A few studies have compared the results of the methods. Many have compared two to five methods for assessing risks of MSDs [23 –31]. Carayon et al. [32] provided information on the use of mixed methods research in human factor and ergonomics (HFE) research in health care. A variety of methods have been used for collecting data, including interviews, surveys and observation. The most frequent combinations have involved interviews for qualitative data and surveys for quantitative data. The use of mixed methods in healthcare HFE research has increased over time. A primary study was carried out to identify an action level for the QEC system (Quick Exposure Check for assessment of workplace risks for work-related musculoskeletal disorders).This was achieved by assessing a number of industrial tasks simultaneously using the QEC and RULA and comparing the assessment scores. The results showed that the action levels on the QEC where then extracted from the corresponding RULA scores [33].

Ketabi Yazdi [34] used the QEC method to evaluate associations between the exposure levels of various areas of the body and risk of musculoskeletal injury factors and indicated that there was a significant and direct association between calculated ergonomic risk by the QEC method and musculoskeletal disorders in workers. Motamedzade et al. [35] compared outputs from Rapid Entire Body Assessment and the Quick Exposure Check regarding ergonomic risk in an engine oil company and found that there was a significant correlation between the two methods. Therefore, it is possible for researchers to apply these methods interchangeably for postural risk assessment in appropriate working environments. Mohammadi and Suregi [36] analyzed the potential risk of musculoskeletal disorders of the upper body among employees of Bank Melli Iran, using RULA and the QEC. They showed that employees who sat long-term had high potential for musculoskeletal disorders. Labbafinejad et al. [37] investigated upper limb musculoskeletal pain and posture in workers in packaging units of pharmaceutical industries to define the potential ergonomic risk factors for musculoskeletal pain in the upper limbs. Working history and shift work were considered for shoulder and wrist pain. In order to assess postures, they utilized RULA, and their findings revealed that shoulder pain was associated with work history, smoking, level of education and age (over 40 years old). Wrist pain was associated with shift work, especially fixed shift work, and age and days missed from work, with a cut-off point of 7 days. Other demographic factors such as height and weight were not considered.

Many studies have emphasized the association between shoulder pain and work history (see [38, 39]). Other studies have demonstrated an association between posture and pain in the upper extremities (see [40, 41]). The sample populations for these studies have included jewelry workers, students, and medical sonographers. The results have shown that different working populations cannot be equally screened by RULA for the risk of pain in the upper limbs. Rapid Upper Limb Assessment by itself is a screening tool for bad postures, but it should not be used as a proxy for musculoskeletal complaints in the upper limbs. This why this instrument was not used in the present study. Al-Hourani et al. [42] examined the prevalence of WMSDs among Jordanian dental technicians and the associated factors, using a cross-sectional design. This study found a high prevalence of musculoskeletal complaints in dental technicians. Multivariate analysis of factors associated with musculoskeletal disorders showed that those significantly associated with musculoskeletal disorders in at least one body part were age, gender, education, and daily working hours. A bachelor’s degree or higher education was associated with increased odds of elbow pain. Females with a bachelor’s degree or higher education who worked more than 7 hours/day showed significantly higher risk of developing wrist pain. None of the studied characteristics was associated with musculoskeletal pain in any other body parts. Mebarki et al. [43] measured quality of work life (QWL) and studied differences in demographic characteristics (gender, age, work experience and socio-professional category). The study participants were managerial staff in the Algerian tertiary sector (N = 252). The results showed that there were no statistically significant differences in QWL among the categories of demographic characteristics categories. The study also concluded that QWL requires more attention from the management level in the public sector in Algeria.

Salminen et al. [44] analyzed the effect of demographic factors like gender, age, tenure and mother tongue on occupational injuries. In Finland, about 120,000 occupational injuries occur annually, at a cost of over EUR 2 billion per year. The study showed that age was a more important factor in injury involvement than gender, tenure or mother tongue. However, age was closely related to experience in the company. Thus, prevention measures in the companies should focus on new employees. Rothmore and Gray [45] investigated outdoor council workers and sought to address the challenge of an ageing workforce demographic in South Australia by examining the association between a range of workplace risks and hazards and work ability scores. The participants were 155 workers from five groups of outdoor workers in a large metropolitan council; questionnaires were administered during staff meetings. The survey instrument included questions on demographic and employment characteristics, physical and psychosocial risk factors, and the Work Ability Index. The results showed that 43% of the workers had excellent or good work ability scores. Those categorized as having moderate work ability scores were 14% of the workers. There were no workers with poor work ability scores. Associations with work ability scores were found for age, pain and discomfort, perceptions of health and safety at work, and a range of psychosocial and physical risk factors. The results confirmed a link between work ability and a range of physical and psychosocial risk factors, which if addressed, could improve the longevity of the workforce. Andersen et al. [46] investigated which factors are associated with physical exertion during manual lifting. The participants were 200 blue-collar workers from 14 workplaces across Denmark. The results showed that gender, load, back muscle strength and back pain were associated with high perceived physical exertion during manual lifting. These factors should be considered when planning work with manual lifting for individual workers. Age, smoking, body mass index (BMI), and time of the day were not associated with physical exertion. Yanwen et al. [47] explored the effects of demographic and social factors for those suffering from work-related injuries with musculoskeletal disorders and their employment status within three months after discharge. The results showed significant differences between the return to work group and non-return to work group. The shorter the length of hospitalization during the rehabilitation process, the greater the rate of return to work for workers suffering from work-related injuries with musculoskeletal disorders. Mohd Taib et al. [48] examined the relationship between specific physical and psychosocial factors and/or ergonomic conditions and MSD symptoms among dentists in Malaysia. The results showed high prevalence of MSD symptoms, especially those affecting the shoulders, neck, and back, indicating the role of work- related risk factors in these disorders. This study also found some significant relationships between physical, psychosocial and sociodemographic factors and the prevalence of MSD symptoms among dentists. Surprisingly, all of the above-mentioned studied dealt separately with one subject in this area (such as: ergonomics, assessment tools, demographic factors and DEA). The present study is distinguished for mixing the areas in order to produce logical results.

3 Methodology

Since body condition is a key factor in performing jobs, various methods have been presented for its evaluation. Among the ergonomic evaluation methods, observational methods have special benefits in that they do not require specialized equipment or rapid evaluation. As mentioned before, the present paper aims to carry out risk assessment by using the data envelopment analysis approach, taking into account five demographic factors: age, education, job experience, weight and height. For this purpose, DEA and two typical anthropometric and ergonomic risk assessment methods, the QEC and REBA, will be briefly introduced.

3.1 The REBA posture evaluation method

Rapid Entire Body Assessment was proposed by Hignett and McAtamney [49] as a means to assess posture for risks of work-related musculoskeletal disorders. It considers the critical tasks of a job, and for each task, assesses posture factors by assigning a score to each region.

This method identifies unsuitable postures allows the definition of good solutions for corrective actions of work methods or work environments. For unsuitable postures, including risks of musculoskeletal disorders, a Nordic questionnaire is used to evaluate the prevalence of these disorders and determine the disorder and REBA scores. Information about the REBA method is provided in Table 1.

Table 1
Details of the REBA method

Performance necessity Correction performance level Risk level Score

Not necessary 0 Ignored 1

May be necessary 1 Low 2–3

Is necessary 2 Average 4–7

Was necessary earlier 3 High 8–10

Immediately necessary 4 Very high 11–15

Performance necessity	Correction performance level	Risk level	Score
Not necessary	0	Ignored	1
May be necessary	1	Low	2–3
Is necessary	2	Average	4–7
Was necessary earlier	3	High	8–10
Immediately necessary	4	Very high	11–15

3.2 The QEC evaluation method

The QEC is a tool for safety and work health professionals to evaluate exposure to work-related musculoskeletal risks. This tool was presented by Lee and Buckle, with the support of the Robens Centre and Surrey University in England and collaboration by150 health and job safety experts in England. The first report was published in 1999. The Quick Exposure Check (QEC) is posture-based. Combining the observer’s assessment with the worker’s answers to closed questions, it allows assessment of MSD risk factors for the back, arms, neck and upper extremities at a workstation. In addition to an overall score for the whole body (QEC General), this method provides a risk index for each targeted area (back, shoulder-arm, wrist-hand and neck) [11]. This method is designed to evaluate exposure to work-related musculoskeletal risks and changes before and after ergonomic intervention. Details about the QEC method are provided in Table 2.

Table 2
Details of the QEC method

Assessment Ergonomic risk level Total score QEC

Acceptable 1 Below 40%

Further study is required 2 41% to 50%

Further study is required and correction measurements should be performed in the near future 3 51% to 70%

Further study is required and correction measurements should be performed immediately 4 Above 70%

Assessment	Ergonomic risk level	Total score QEC
Acceptable	1	Below 40%
Further study is required	2	41% to 50%
Further study is required and correction measurements should be performed in the near future	3	51% to 70%
Further study is required and correction measurements should be performed immediately	4	Above 70%

3.3 Data envelopment analysis

The DEA method is a strong tool for computing the relative efficiency of a set of decision-making units. The benefits of this method are easily perceived. It considers some inputs and outputs to compute efficiency scores. Since the efficiency scores are different, based on the input and output selection, for each sample, adequate attention is given to achieving a good response. Data envelopment analysis involves computation of evaluations of efficiency levels in groups of organizations, and the efficiency of each unit is computed in comparison with best-performing units. This technique is based on the linear planning approach. Its main goal is to compare and evaluate the efficiency of homogenous decision-making units. Efficiency values range from zero to one. To compute efficiency, weighted ratios of outputs to inputs are used, and the weights have free values to maximize the efficiency borders of the enterprises, such that if the selected weights for enterprise p are considered in calculating the efficiency of other enterprises, their efficiency is not above one. Thus, the BCC and CCR models are presented. All of these models have two orientations, with input-axis and output-axis nature, and are presented as a multilevel model and an overlay model. They can also be specified as constant scale returns and variable scale returns. These two views are:

Input-oriented and output-oriented views in solutions

In DEA models, the solution to improving inefficient units is reaching efficiency boundaries, which consist of a measure of efficiency. Generally, there are two solutions to improve inefficient units and achieve efficiency borders:

Reduction of inputs without reducing outputs for units to achieve efficiency borders. (This approach is called input performance improvement, or input-oriented evaluation of efficiency.)

Increase of outputs for units to achieve efficiency borders without absorbing more inputs. (This is called output performance improvement, or output-oriented assessment of efficiency.)

In input-oriented DEA models, technical inefficiency ratios should be achieved in order to reduce inputs, and units can be within efficiency borders without changing outputs. In the output-oriented approach, ratios are achieved to increase outputs without any change in inputs, and units achieve efficiency borders.

Return to scale

Before decision-making units are evaluated, it is necessary to know the relationship ratio between changes in inputs and outputs in the units. This relationship is called return to scale. Return to scale is used to select assessment models based on decision-making units.

3.4 Sensitivity analysis

In this step, the effects of demographic factors on risk rates will be assessed through statistical techniques. Input and output variables are left out of the calculations one by one, and the risk rates are measured anew. The results obtained before factor elimination are compared to those yielded after elimination to find out how the factors have affected risk rates. The paired t-test and the Pearson correlation coefficient are used for this purpose. Subsequently, the factors are sorted according to the weights, which show the degree to which they influence the obtained risk rates. The results of this phase are validated by performing analysis of variance (ANOVA). Data related to the factors are input into ANOVA to find out how factors differ. Factors shown by ANOVA to be statistically equal should have equal weights as well. The steps of the methodology are presented in Fig. 1.

Fig. 1

Flowchart of the methodology.

4 Case study

For the case study, an auto filter manufacturing unit was selected. Institutional review board approval was obtained prior to conducting the study, and the Board of Directors did not hesitate to assist us. Questionnaires were distributed to 80 workers and machine operators of the company. The workers ranged in age from 16 to 49 years old, and 77% had experienced musculoskeletal disorders and job problems. The instruments used were a questionnaire on demographic characteristics and The Nordic Musculoskeletal Questionnaire (NMQ), which was developed from a project funded by the Nordic Council of Ministers. The aim was to develop and test a standardized questionnaire methodology allowing comparison of low back, neck, shoulder and general complaints for use in epidemiological studies. The tool was not developed for clinical diagnosis[50]. The demographic factors for the sample are listed in Table 3.

Table 3
Demographic factors for the sample

Demographic factors Category Number of workers % of sample

Gender Male 80 100

Age (years) ≤30 11 13.75

31–35 30 37.5

36–40 23 28.75

≤41 16 20

Experience ≤5 4 5

6–10 10 12.5

11–15 22 27.5

16–20 39 48.75

≤21 5 6.25

Education completed Illiterate 23 28.75

High school 41 51.25

Bachelor’s degree 12 15

Graduate degree 4 5

Height ≤170 cm 26 32.5

171–175 cm 34 42.5

176–180 cm 12 15

≤181 cm 8 10

Weight ≤70 kg 3 3.75

71–75 kg 29 36.25

76–80 kg 42 52.5

≤81 kg 6 7.5

Demographic factors	Category	Number of workers	% of sample
Gender	Male	80	100
Age (years)	≤30	11	13.75
	31–35	30	37.5
	36–40	23	28.75
	≤41	16	20
Experience	≤5	4	5
	6–10	10	12.5
	11–15	22	27.5
	16–20	39	48.75
	≤21	5	6.25
Education completed	Illiterate	23	28.75
	High school	41	51.25
	Bachelor’s degree	12	15
	Graduate degree	4	5
Height	≤170 cm	26	32.5
	171–175 cm	34	42.5
	176–180 cm	12	15
	≤181 cm	8	10
Weight	≤70 kg	3	3.75
	71–75 kg	29	36.25
	76–80 kg	42	52.5
	≤81 kg	6	7.5

The validity of the questionnaires is supported by experts on ergonomics and Kuorinka et al. [51]. After selection of the required sample, workers were asked the interview questions by an expert evaluator, and then the reliability was calculated using Cronbach’s alpha 0.75. The study methodology was based on a math model, which is a description of a system using math, theorems and symbols. Math modeling is an effort to develop a math model for a defined system. Math modeling helps researchers to investigate systems by analyzing them and predicting their behavior. Math models sometimes include logical models. Thus, logic is part of math. In most cases, the quality of a study is dependent upon the precision of the model.

First, the jobs in the factory were identified and categorized. Then, based on work positions, all the jobs in the factory were assessed with the QEC, REBA and final scores obtained. The work units of the company included:

Production

Internal core

External shell

Quality control

Packaging

Warehouse

Administration

Since there is no work cycle in a filter manufacturing factory, the tasks should be continuous, and work of each part is dependent upon the work of some other work units. Thus, the jobs were studied based on evaluation priority. All the collected data based on work units (5 main work units) is shown in Table 4.

Table 4

Results of analysis of duties and absolute frequency distribution of final scores and measurements by two methods (REBA and the QEC)

	Jobs based on units	Number of workers in position	Final score		Level of actions
				REBA	QEC	REBA	QEC
1	Cutting	Internal core production	6	8	64.86%	3	3
2	Pressing		5	7	40.54%	2	2
3	Punching		7	7	63.89%	2	3
4	Die		5	8	58.10%	3	3
5	Paper cutting		6	12	81.08%	4	4
6	Rolling		5	6	45.94%	2	2
7	Welding (welding point)		5	11	44.59%	3	2
8	Filter assembly and felt warping		6	4	41.24%	2	2
9	Washing	Preparation of external shell	5	9	66.31%	3	3
10	Dying		4	7	49.36%	2	2
11	Printing		4	7	61.9%	2	3
12	Adhesion and connection of base to body	Quality control	4	5	47.81%	2	2
13	Testing		5	6	55.40%	2	3
14	Washer and nylon	Packing	2	4	46.9%	2	2
15	Packing		3	9	61.52%	3	3
16	Polypropylene		2	8	57.83%	3	3
17	Transfer of raw materials	Warehouse	2	10	69.45%	3	3
18	Delivery of manufactured product		2	8	59.68%	3	3
19	Recording warehouse information in system		1	3	39.74%	1	1
20	Inventory control		1	3	37.21%	1	1

5 Results and discussion

To compare the results of REBA and the QEC, non-parametric Wilcoxon and Kruskal-Wallis tests were used. Furthermore, the Spearman correlation coefficient was applied to evaluate the correlation between the two methods. Based on the data from Table 4, the action levels of the two methods were evaluated for all the studied jobs. In addition, the work units of the company were separated using non-parametric Wilcoxon and Kruskal-Wallis tests. First, the frequency of ergonomic risks and corrections were calculated. Table 5 shows the frequency of the results for the REBA method.

Table 5
Frequency of ergonomic risk levels and corrective actions for REBA, by job

0 1 2 3 4 Total

Work unit Negligible Low Average High Very high (n = 80)

1 3–2 7–4 10–8 11–15

N % N % N % N % N % N %

Internal core production 0 0 0 0 23 51.11 11 24.44 11 24.44 45 56.25

Preparation of external shell 0 0 0 0 8 61.53 5 38.46 0 0 13 16.25

Quality control 0 0 0 0 9 100 0 0 0 0 9 11.25

Packaging 0 0 0 0 2 28.57 5 71.42 0 0 7 8.75

Warehouse 0 0 2 100 0 0 4 66.66 0 0 6 7.5

All jobs 0 0 2 2.5 42 52.5 25 31.25 11 13.75 80 100

	2	3	4	Total
Internal core production	0	0	23	51.11	11	24.44	11	24.44	45	56.25
Preparation of external shell	0	0	8	61.53	5	38.46	0	0	13	16.25
Quality control	0	0	9	100	0	0	0	0	9	11.25
Packaging	0	0	2	28.57	5	71.42	0	0	7	8.75
Warehouse	2	100	0	0	4	66.66	0	0	6	7.5
All jobs	2	2.5	42	52.5	25	31.25	11	13.75	80	100

The frequency distribution of REBA action levels showed that 52.5% of all jobs were at the second action level with the highest frequency. 37.5% of all jobs were at the third action level, and the first and fourth action levels were 2.5% and 7.5%, respectively. Figure 2 graphically depicts these frequencies. Frequency results for the QEC are shown in Table 6.

Fig. 2

Frequency of corrective actions for REBA based on jobs.

Table 6

Frequency of ergonomic risk levels and QEC correction levels based on jobs

	1		2		3		4		Total
	Lower than 40%		41% – 50%		51% – 70%		70% and above		(n = 80)
Work unit	n	%	n	%	n	%	n	%	n	%
Internal core production	0	0	21	46.66	18	40	6	13.33	45	56.25
Preparation of external shell	0	0	4	30.76	9	69.23	0	0	13	16.25
Quality control	0	0	4	44.44	5	55.55	0	0	9	11.25
Packaging	0	0	2	28.57	5	71.42	0	0	7	8.75
Warehouse	2	33.33	0	0	4	66.66	0	0	6	7.5
All jobs	2	2.5	31	38.75	41	51.25	6	7.5	80	100

Frequency distributions of QEC action levels showed that 51.25% of all jobs had high frequencies at the third action level (column 3), and 38.75% at the second action level. Similar to the REBA method, the first and fourth action levels showed 2.5% and 7.5%, respectively. Figure 3 shows these frequencies graphically.

Fig. 3

Frequency of corrective actions for the QEC based on jobs.

To achieve high reliability and access to the applied goal of this plan, the action levels of both methods were investigated. The results for all jobs with the Wilcoxon test are listed in Table 7.

Table 7

Results for action levels of the QEC and REBA evaluation methods for all jobs

All jobs	REBA action levels					QEC action levels
Test result	Mean	Median	Max	Min	SD	Mean	Median	Max	Min	SD
2.4	2	4	1	0.75	2.5	3	4	1	0.76
^*	P > 0.164

^*Wilcoxon signed-rank test.

A non-parametric Wilcoxon test was used to evaluate the results of the REBA and QEC methods for all jobs. Since the significance level was above 0.05, there was no significant difference between the median of the ranks for these methods (P > 0.164). Thus, the average equality of action levels in both methods had high significance levels. To confirm that there was no effect on the results of differences in types of jobs among work units, the Wilcoxon test was used to investigate the action levels of the evaluation methods based on types of work. Table 8 shows the results of this investigation.

Table 8

Results of evaluation of action levels for the QEC and REBA methods

Work unit	REBA action levels					QEC action levels					Results
	Mean	Median	Max	Min	SD	Mean	Median	Max	Min	SD
Internal core production	2.62	2.5	4	2	0.74	2.62	2.5	4	2	0.74	P = 0.564
Preparation of external shell	2.33	2	3	2	0.57	2.66	3	3	2	0.57	P = 0.046
Quality control	2	2	2	2	0	2.5	2.5	3	2	0.70	P = 0.025
Packaging	2.66	3	3	2	0.57	2.66	3	3	2	0.57	P = 1
Warehouse	2	2	3	1	1.15	2	2	3	1	1.15	P = 1

^*Wilcoxon signed-rank test.

The test showed no significant differences between action levels by of work unit, except for preparation of external shell and quality control. The significance levels for internal core production, packaging and warehouse were above the defined alpha level of 5%; this indicates positive and significant associations between the results of action levels of the two methods for these work units. For preparation of external shell and qualitative control, the significance levels were 0.046 and 0.025, respectively; since the significance levels were less than 5%, there were significant differences between the action levels of REBA and the QEC in these work units. Next, the Kruskal-Wallis test was used to compare the action levels of the two evaluation methods based on work units. The results are shown in Table 9.

Table 9

Results of comparison of action levels of the REBA and QEC evaluation methods

Work units	REBA action levels		QEC action levels
	N	Mean of ranks	N	Mean of ranks
Internal core production	45	11.81	45	10.94
Preparation of external shell	13	9.67	13	11.67
Quality control	9	6.50	9	10.25
Packaging	7	12.38	7	11.67
Warehouse	6	8.75	6	8.00
Results
Chi-square	2.443		1.194
Degree of freedom	4 = df		4 = df
Significance level	0.655		0.879

^*Kruskal-Wallis test.

Comparison of the action levels of the REBA and QEC methods with the Kruskal-Wallis test yielded significance levels of P = 0.655 and P = 0.879, respectively. Since the significance levels for both REBA and the QEC with degree of freedom 4 were above 5%, there was no significant difference between work units in terms of action levels. After evaluation by REBA and the QEC and statistical analysis of the results, Spearman’s rho was used to test the correlations of the final scores and action levels for all jobs. The results are shown in Table 10.

The Spearman’s ranked correlation coefficient is used when the amount of data is low and normality is not logical. In this case, this analysis showed high correlation and close final scores for both methods, with a correlation coefficient of r = 0.633, and action levels showed a correlation coefficient r = 0.635, which indicates high correlation of the results of REBA and the QEC. This outcome was one of the main goals of the study.

5.1 Results of evaluation of final scores

Work units were separated to compare the results for the final scores of the REBA and QEC methods by work unit, using a non-parametric Kruskal-Wallis test. The data from these two methods are not consistent, and their results cannot be compared. The Wilcoxon test was not applied, since there was a large difference between the data. Table 11 shows the results of this evaluation.

Table 10
Correlation results of final scores for the REBA and QEC methods and correlation of their action levels for all jobs

All jobs Final scores Action level

0.633 0.635

All jobs	Final scores	Action level
	0.633	0.635

^*Spearman’s rank correlation coefficient.

Table 11

Results of comparison of final scores for the QEC and REBA methods by work unit using the Kruskal-Wallis test

	Final scores QEC					Final scores REBA
Work unit	SD	Min	Max	Median	Mean	SD	Min	Max	Median	Mean
Internal core production	0.144	40.54	81.08	52.02	55.03	2.58	4	12	7.5	7.87
Preparation of external shell	0.087	49.36	66.31	61.9	59.19	1.15	7	9	7	7.66
Quality control	0.053	47.81	55.40	51.61	51.61	0.70	5	6	5.5	5.5
Packaging	0.076	46.90	61.52	57.83	55.42	2.64	4	9	8	7
Warehouse	0.156	37.21	69.45	49.71	51.52	3.55	3	10	5.5	6
Test result
Chi-square	0.693					8.466
Degree of freedom	4 = df					4 = df
Significance level	0.952					0.076

Comparison of the final scores by work unit for the REBA and QEC methods with the Kruskal-Wallis test yielded significance levels of P = 0.076 and P = 0.952, respectively, and the significance level of final scores of REBA and the QEC with degree of freedom 4 was above 5%. This indicated that there was no significant difference between work units in terms of final scores.

5.2 Evaluation of ergonomic risk by the REBA method by DEA multiple analysis

For an additional demographic survey, the five factors of age, education, work experience, weight and height were selected. Their effect on the final scores of REBA and the QEC was evaluated by the DEA multiple analysis method and a paired t-test. Then, their correlation was evaluated using the Pearson correlation coefficient. Other factors, including job, gender, shifts and working hours, and marital status had no part in this analysis and were not effective factors. All the workers were men, and the factory had only one work shift, from 8 a.m. to 5 p.m. For validity of analysis, all 20 work stations were investigated.

Based on the questionnaires, and using Excel software, the mean values for the five factors (age, education, work experience, weight and height) were obtained for each of the work stations separately, yielding 20 values for each factor, as shown in Table 12. These final scores for the REBA method, ranging from 1 to 15, were entered symmetrically in Table 12. Data envelopment analysis was used to evaluate output to input ratios as a definition of productivity. Since the output in this paper is evaluations of risk amounts, the final scores entered for REBA obtained from Table 4 were entered in the same way (symmetrically) in this new table.

Table 12
Mean of data for five factors with symmetrical final scores for REBA, by work station

Work station Age Weight Height Work experience Education REBA

1 Cutting 32.83 73.83 169.83 12.33 9 8

2 Pressing 33.2 75.6 171.6 10.2 9 9

3 Punching 33.85 78.71 178.28 10.42 9 9

4 Die 35.8 77.2 177.2 12 8.4 8

5 Paper cutting 33.33 77 176 9.5 9 4

6 Rolling 33.4 82.8 178.8 11.8 9 10

7 Welding (welding point) 38 68.6 162.2 13.6 9 5

8 Filter assembly and felt warping 42.66 73.66 173.16 20.33 10 12

9 Washing 33.4 74.4 174.4 13 9 7

10 Dying 36 68.75 168 12.5 9 9

11 Printing 31.25 72 172.5 12 9 9

12 Adhesion and connection of base to body 28.5 72.5 173 14.5 9 11

13 Testing 30.6 77.2 174.4 8.2 10.6 10

14 Washer and nylon 36 71.5 165 10.5 9 12

15 Packaging 23.66 70 170 6.66 7 7

16 Polypropylene 39.5 70 196.5 18.5 9 8

17 Transfer of raw materials 30 72 172 9.5 9 6

18 Delivery of manufacture product 32.5 74 171 8 13 8

19 Recording warehouse information in system 37 69 169 12 16 13

20 Inventory control 30 72 162 5 14 13

Work station	Age	Weight	Height	Work experience	Education	REBA
1	Cutting	32.83	73.83	169.83	12.33	9	8
2	Pressing	33.2	75.6	171.6	10.2	9	9
3	Punching	33.85	78.71	178.28	10.42	9	9
4	Die	35.8	77.2	177.2	12	8.4	8
5	Paper cutting	33.33	77	176	9.5	9	4
6	Rolling	33.4	82.8	178.8	11.8	9	10
7	Welding (welding point)	38	68.6	162.2	13.6	9	5
8	Filter assembly and felt warping	42.66	73.66	173.16	20.33	10	12
9	Washing	33.4	74.4	174.4	13	9	7
10	Dying	36	68.75	168	12.5	9	9
11	Printing	31.25	72	172.5	12	9	9
12	Adhesion and connection of base to body	28.5	72.5	173	14.5	9	11
13	Testing	30.6	77.2	174.4	8.2	10.6	10
14	Washer and nylon	36	71.5	165	10.5	9	12
15	Packaging	23.66	70	170	6.66	7	7
16	Polypropylene	39.5	70	196.5	18.5	9	8
17	Transfer of raw materials	30	72	172	9.5	9	6
18	Delivery of manufacture product	32.5	74	171	8	13	8
19	Recording warehouse information in system	37	69	169	12	16	13
20	Inventory control	30	72	162	5	14	13

Then, input-oriented and output-oriented BCC and CCR were evaluated for REBA. Table 13 shows the mean efficiency for REBA.

Table 13

Mean of the efficiency of BCC and CCR in the REBA method

BCC-Input	BCC-Output	CCR-Input	CCR-Output
0.929757	0.929763	0.929181	0.929186

Due to software constraints, the numbers obtained were very close and similar to each other, so BCC-Input Full-Ranking was chosen as the best analysis result. In the next step, the efficiency of the best result from BCC-Input was evaluated and the symmetry of the final scores for REBA was evaluated by a paired t-test.

The objective was evaluation of the relationship between efficiency results and ranks from the best state of DEA and final scores for REBA for the 20 work stations. For data consistency, normalization was first done on the REBA data and efficiency, and then the t-test was performed. To do this, the division of one by one of data by greatest value is used. since the P value = 0.093, H 0 regarding the relationship between efficiency of input-oriented BCC analysis and the final scores for REBA was not rejected. This means that at least one of the five factors was effective in reducing worker risk. To evaluate correlation between BBC-input and REBA, Final scores REBA were first ranked, based on the data in Table 4, and due to non-parametric test, the Pearson correlation coefficient was applied.

Table 14

Evaluation of correlation of two factors that are effective in REBA risk reduction

Factors /Tests	Two-tailed paired t-test p-value	One-tailed paired t-test result	Effect	Correlation
No Age	0.148	–	Ineffective	0.993
No Education	0.000	μ_i < μ_F	Positive	0.775
No Experience	0.033	μ_i < μ_F	Positive	0.968
No Height	0.661	–	Ineffective	1.000
No Weight	0.330	–	Ineffective	0.999

^*I = Elimination; F = Full factor.

H 0 in correlation shows non-autocorrelation. Since P < 0.05 and Sig = 0, H 0 was rejected. This means that H 1 was supported. Thus, there was a correlation between the ranking of BBC-input and the ranking of the final score for REBA. To analyze the effect of each factor, the exclusion method in DEA was used. For this approach, each factor was eliminated and its efficiency assessed, and then a two-way paired t-test was used to evaluate its effect. Since significance level was above 0.05, H 0 was not rejected. This means that the factor was not effective in reduction of risk. Subsequently, a one-way t-test was used to investigate positive or negative effects. The results are shown in Table 14.

The results showed that when age, weight and height were eliminated, there was no change in risk as assessed by REBA risk However. significance levels in a two-way t-test for education and work experience were 0.000 and 0.033, respectively, and H 0 was rejected. The one-way t-test eliminated the positive effect of education and work experience, and risk was reduced. Thus, education and work experience were important factors in reducing risk, which supports the study’s goal of increasing risk reduction rates among the jobs at work stations. The Pearson correlation coefficient was used to evaluate correlations between the two positive factors for REBA risk reduction, education and work experience. The results are shown in Fig. 4.

Fig. 4

Evaluation of correlation of two effective factors on REBA risk reduction.

Education, with a weight of 0.875, showed a positive effect on work experience, with a weight of 0.124, on reduction of risk of final scores for the REBA method.

5.3 Risk assessment for the QEC method with DEA

Like REBA, in the QEC method, based on the questionnaires and using Excel software, the means for each of the 5 factors (age, education, work experience, weight and height) for each of the 20 work stations were obtained separately, yielding 20 values for each factor. They were considered with symmetrical final scores for the QEC method (final score = 100). Table 15 displays the results.

Table 15
Mean for five factors with symmetrical final scores for the QEC, by work station

Work station Age Weight Height Work experience Education QEC

1 Cutting 32.83 73.83 169.83 12.33 9 35.14

2 Pressing 33.2 75.6 171.6 10.2 9 59.46

3 Punching 33.85 78.71 178.28 10.42 9 36.11

4 Die 35.8 77.2 177.2 12 8.4 41.9

5 Paper cutting 33.33 77 176 9.5 9 18.92

6 Rolling 33.4 82.8 178.8 11.8 9 54.06

7 Welding (welding point) 38 68.6 162.2 13.6 9 55.41

8 Filter assembly and felt warping 42.66 73.66 173.16 20.33 10 58.76

9 Washing 33.4 74.4 174.4 13 9 33.69

10 Dying 36 68.75 168 12.5 9 50.64

11 Printing 31.25 72 172.5 12 9 38.1

12 Adhesion and connection of base to body 28.5 72.5 173 14.5 9 52.19

13 Testing 30.6 77.2 174.4 8.2 10.6 44.6

14 Nylon 36 71.5 165 10.5 9 53.1

15 Packing 23.66 70 170 6.66 7 38.48

16 Polypropylene 39.5 70 196.5 18.5 9 42.17

17 Transfer of raw materials 30 72 172 9.5 9 30.55

18 Delivery of manufactured product 32.5 74 171 8 13 40.32

19 Recording warehouse information in system 37 69 169 12 16 60.26

20 Inventory control 30 72 162 5 14 62.79

	Work station	Age	Weight	Height	Work experience	Education	QEC
1	Cutting	32.83	73.83	169.83	12.33	9	35.14
2	Pressing	33.2	75.6	171.6	10.2	9	59.46
3	Punching	33.85	78.71	178.28	10.42	9	36.11
4	Die	35.8	77.2	177.2	12	8.4	41.9
5	Paper cutting	33.33	77	176	9.5	9	18.92
6	Rolling	33.4	82.8	178.8	11.8	9	54.06
7	Welding (welding point)	38	68.6	162.2	13.6	9	55.41
8	Filter assembly and felt warping	42.66	73.66	173.16	20.33	10	58.76
9	Washing	33.4	74.4	174.4	13	9	33.69
10	Dying	36	68.75	168	12.5	9	50.64
11	Printing	31.25	72	172.5	12	9	38.1
12	Adhesion and connection of base to body	28.5	72.5	173	14.5	9	52.19
13	Testing	30.6	77.2	174.4	8.2	10.6	44.6
14	Nylon	36	71.5	165	10.5	9	53.1
15	Packing	23.66	70	170	6.66	7	38.48
16	Polypropylene	39.5	70	196.5	18.5	9	42.17
17	Transfer of raw materials	30	72	172	9.5	9	30.55
18	Delivery of manufactured product	32.5	74	171	8	13	40.32
19	Recording warehouse information in system	37	69	169	12	16	60.26
20	Inventory control	30	72	162	5	14	62.79

Then, AUTO Assess software was used to evaluate input-oriented and output-oriented BCC and CCR for the QEC. (AUTO Assess is unique DEA ranking software that was created and developed by Mohammad Ali Azadeh and his team in Iran.) Table 16 shows the means of efficiency for the QEC.

Table 16

Mean of efficiency of BCC and CCR in the QEC method

BCC-Input	BCC-Output	CCR-Input	CCR-Output
0.937634	0.937632	0.937248	0.937242

Due to software limitations, the values are similar. BCC-Input, full-ranking, was selected as the best choice for the analysis. In the next step, a paired t-test was used to evaluate the efficiency of the best BCC-input result, and the symmetry of the final scores for REBA.

Like the REBA method, for data consistency, first the QEC data and efficiency were normalized, and then the t-test is performed. Since the P value = 0.121, H 0 regarding the relationship between efficiency of input-oriented BCC analysis and the final scores for the QEC was not rejected. This means that at least one of the 5 factors was effective in reducing worker risk. The Pearson correlation coefficient was applied to evaluate the ranked correlation of BBC-input and the QEC, at first Rank of final scores QEC is achieved based on data from Table 4 and due to non-parametric test. H 0 in correlation shows non-autocorrelation. Since Sig = 0 < 0. 05, H 0 was rejected. This means that H 1 supporting the correlation is supported. Thus, there is a correlation between rankings of BBC-Input and rankings of the final score for the QEC.

The elimination method in DEA was used to analyze the effect of each factor. Each factor was eliminated one by one and its efficiency evaluated. Then a two-way paired t-test was used to evaluate its effect. Since the significance level was above 0.05, H 0 was not rejected. This means that the factor has no effect on reduction of risk; in other words, a one-way t-test was used to investigate positive or negative effects. The results are shown in Table 17.

Table 17

Evaluation of correlation of factors with QEC risk reduction

Factors /Tests	Two-tailed paired t-test p-value	One-tailed paired t-test result	Effect	Correlation
No Age	0.330	–	Ineffective	0.997
No Education	0.000	μ_i < μ_F	Positive	0.969
No Experience	0.138	–	Ineffective	0.960
No Height	1.000	–	Ineffective	1.000
No Weight	0.045	μ_i < μ_F	Positive	0.995

^*I = Elimination; F = Full factor.

When age, weight and height were eliminated, there was no change in QEC risk. However, the significance levels in the two-way t-test for education and weight were 0.000 and 0.045, respectively, and H 0 was rejected. Elimination of education and weight by a one-way t-test had a positive effect, and risk was reduced. This indicates that education and weight are important factors in reducing risk, which supports our goal in this analysis of increasing risk reduction rates among the jobs at work stations. The Pearson correlation coefficient was used evaluate correlation of the factors. The correlations between two positive factors, education and weight, and QEC risk reduction are shown in Fig. 5.

Fig. 5

Evaluation of correlation of two effective factors with QEC risk reduction.

Education, with a weight of 0.861, has higher positive effect compared with body weight, with a weight of 0.138, in reducing risk in the case of the QEC method. The results of a sensitivity analysis were examined by ANOVA for validation. Strictly speaking, the data regarding the factors were used to conduct ANOVA. The results showed that with respect to REBA, the factors experience and education were different from the others, while the other factor means were statistically equal. In addition, experience had a higher mean value. In relation to the QEC, weight and education were different from the other three factors, and education had a higher mean value. These outcomes validated the results obtained by sensitivity analysis.

6 Conclusions and directions for future research

This paper evaluated the effect of several demographic factors (age, education, job experience, weight and height), utilizing an integrated mathematical programming approach to measure the amount of ergonomic risk. The real-world case of an auto part manufacturer was considered to test the applicability of the study.

Musculoskeletal disorders among 80 workers were evaluated, using two typical anthropometry and ergonomic risk assessment questionnaires, the Quick Exposure Check and Rapid Entire Body Assessment. The results of the questionnaires, along with data regarding the five demographic characteristics, were input as the decision-making units (i.e., operators) used in the data envelopment analysis approach. The rates yielded by DEA were then compared with the results of REBA and the QEC. The final results show that education, work experience and weight are the factors that have positive effects. This means that they could help reduce ergonomic risks. To optimize such effects, it’s better to broadcast the results throughout the company. In sharing knowledge, it is desirable to use feedback from end users in program design, so that it can significantly reduce barriers and injuries. If employee concerns and priorities are not taken into account, this may lead to expansion of organizational, personal and technological barriers [52]. This study was conducted in an auto air filter manufacturing company. Future research could be carried out in other industries using other risk assessment methodologies, such as RULA and OWAS.

Conflict of interest

None to report.

References

Taghipour

, Kheirkhahan

, Mahboobi

, Mohammadi

. Evaluation of the Relationship between Occupational Accidents and Usage of Personal Protective Equipment in an Auto Making Unit. International Journal of Innovative Research in Science, Engineering and Technology. 2015;4(9).

Waehrer

, Dong

, Miller

, Haile

, Men

. Costs of occupational injuries in construction in the United States. Accid Anal Prev. 2007;39(6):1258–66.

Safe Work Australia. Work-related injuries and fatalities involving a fall from height, Australia. October 2013.

Amissah

, Agyei-Baffour

, Badu

, Kwaku Agyeman

, Darkowaa Badu

. The Cost Of Managing Occupational Injuries Among Frontline Construction Workers In Ghana. Value in Health Regional issues. 2019;19(2019):104–11.

OSHA O. Department of Labor: Occupational Safety and Health Administration. Occupational exposure to blood borne pathogens: Final rule. 29:64003-182.

Taghipour

, Mahboobi

, Nikoeifar

, Soofi Mowloodi

. Analysing the Effects of Physical Conditions of the Workplace on Employee&s Productivity (Including Case Study); International Journal of Environmental Protection and Policy. 2015;3(4):111–119; doi: 10.11648/j.ijepp.20150304.15

Nadri

, Fasih

, Nadri

. Comparison of ergonomic risk assessment results from Quick Exposure Check and Rapid Entire Body Assessment in an anodizing industry of Tehran, Iran. Journal of Occupational Health and Epidemiology. 2013;2(4):195–202.

Qutubuddin

, Hebbal1

, Kumar

ACS

. Ergonomic Risk Assessment using Postural Analysis Tools in a Bus Body Building Unit. Industrial Engineering Letters. 2013;3(8).

Arezes

, Baptista

, Barroso

, Carneiro

, Cordeiro

, Costa

, Melo

, Miguel

, Perestrelo

, editors. Occupational Safety and Hygiene II; 2014; CRC Press.

10.

Yuan

. Reducing ergonomic injuries for librarians using a participatory Approach”; International Journal of Industrial Ergonomics; Elsevier B. V. 2015;47:93–103.

11.

Chiasson M-Ève; Imbeau

, Aubry

, Delisle

. “Comparing the results of eight methods used to evaluate risk factors associated with musculoskeletal disorders” International Journal of Industrial Ergonomics; Elsevier B. V. 2012;42:478–88.

12.

Lovell

, Rouse

. Equivalent standard DEA models to provide super efficiency scores. Journal of the Operational Research Society. 2003;54:101–8.

13.

Sigala

. Using Data Envelopment Analysis for Measuring and Benchmarking Productivity in the Hotel Sector. Journal of Travel & Tourism Marketing, The Haworth Hospitality Press, an imprint of The Haworth Press, Inc. 2004;16(2/3):39–60.

14.

Chen

T-H

, Pin Bao

, Yun Chang

. Another view of efficiency improvement in data envelopment analysis. Journal of the Chinese Institute of Industrial Engineers. 2009;26(2):109–14.

15.

Wier

, christoffersen, line block; s. Jensen, trine, g. Pedersen

, keiding

, Munksgaard

, Evaluating Sustainability of Household Consumption—Using DEA to Assess Environmental Performance” Economic Systems Research; Taylor & Francis group. 2005;17(4):425–47.

16.

Punnett

, Wegman

. Work-related musculoskeletal disorders: the epidemiologic evidence and the debate. J Electromyogr. Kinesiol. 2004;14(1):13–23.

17.

Silverstein

, Fine

, Armstrong

. Hand wrist cumulative trauma disorders in industry. Br J Ind Med. 1986;43(11):779–84.

18.

Magnusson

, Pope

. A review of the biome-chanics and epidemiology of working postures: it isn’t always vibration which is to blame!; J Sound Vibrat. 1998;215(4):965–76.

19.

Grieco

, Molteni

, De Vito

, Sias

. Epidemiology of musculoskeletal disorders due to biomechanical overload. Ergonomics. 1998;41(9):1253–60.

20.

National Research Council and the Institute of Medicine (NRC/IM); Musculoskeletal Disorders and the Workplace. Low Back and Upper Extremities.; National Academy Press, Washington, D.C. 2001.

21.

Putz-Anderson

, et al. Musculoskeletal Disorders and Workplace Factors - A Critical Review of Epidemiologic Evidence for Work- Related Musculoskeletal Disorders of the Neck, Upper Extremity, and Low Back. U.S. Department of Health and Human Services, Public Health Service Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health.; In: Bernard, B.P. (Ed.); 1997;97-141.

22.

Stock

. Workplace ergonomic factors and the development of musculoskeletal disorders of the neck and upper limbs: a meta-analysis. Am J Ind Med. 1991;19(1):87–107.

23.

Apostoli

, Sala

, Gullino

, Romano

. Analisi comparata del l’applicazione di quattro metodi per la valutazione del rischio biomeccanico per l’arto superiore. G Ital Med Lav Ergon. 2004;26(3):223–41.

24.

Bao

, Howard

, Spielholz

, Silverstein

. Quantifying repetitive hand activity for epidemiological research on musculoskeletal disorders e part II: comparison of different methods of measuring force level and repetitiveness. Ergonomics. 2006;49(4):381–92.

25.

Brown

, Li

. The development of action levels for the quick exposure check (QEC) system. In: McCabe, P.T. (Ed.), Contemporary Ergonomics. Taylor & Francis, London; 2003; pp. 38-43.

26.

Drinkaus

, Sesek

, Bloswick

, Bernard

, Walton

, Joseph

. Comparison of ergonomic risk assessment outputs from rapid upper limb assessment and the strain index for tasks in automotive assembly plants. Work. 2003;21(2):165–72.

27.

Jones

, Kumar

. Comparison of ergonomic risk assessments in a repetitive high-risk sawmill occupation: saw-filer. Int J Ind. Ergon. 2007;37(9, 10):744–53.

28.

Joseph

, Imbeau

, Nastasia

. Measurement consistency among observational job analysis methods during an intervention study. Int J Occup Saf Ergon. 2011;17(2):139–46.

29.

Sala

, Torri

, Tomasi

, Apostoli

. Stima del rischio da sovraccarico biomeccanico all’arto superiorecondotta, con l’impiego di più metodi di analisi, in diversi settori manifatturieri. G Ital Med Lav. Ergon. 2010;32(2):162–73.

30.

Spielholz

, Bao

, Howard

, Silverstein

, Fan

, Smith

, Salazar

. Reliability and validity assessment of the hand activity level threshold limit value and strain index using expert ratings of mono-task jobs. J Occup Environ Hyg. 2008;5(4):250–7.

31.

Kee

, Karwowski

. A comparison of three observational techniques for assessing postural loads in industry. Int J Occup Saf Ergon. 2007;13(1):3–14.

32.

Carayon

, Kianfar

, Li

, Xie

, Alyousef

, Wooldridge

. “A systematic review of mixed methods research on human factors and ergonomics in health care”; Applied Ergonomics. 2015;51;291-321.

33.

Brown

, Li

. The development of action levels for the quick exposure check (QEC) system. In: McCabe, P.T. (Ed.), Contemporary Ergonomics. Taylor & Francis, London; 2003; pp. 38e43.

34.

Ketabi Yazdi

. Evaluation of the effect of ergonomic intervention on bodily posture of workers in partsmanufacturing factories of Iran Khodro; M.Sc. Thesis, Medical Sciences University of Shahid Sadughi of Yazd.; 2012.

35.

Motamedzade

, Ashuri

, Golmohammadi

, Mahjub

. Comparison of ergonomic risk assessment outputs from rapid entire body assessment and quick exposure check in an engine oil company. Journal of Research in Health Sciences. 2011;11(1):26–32.

36.

Mohammadi

, Suregi

. The potential of risk of musculoskeletal disorder in upper body with RULA and QEC methods were evaluated among employees of Melli bank of Zahedan; M.Sc. Thesis; Zahedan University of Medical Sciences; 2008.

37.

Labbafinejad

, Danesh

, Zahra

. Assessment of upper limb musculoskeletal pain and posture in workers of packaging units of pharmaceutical industries. Work. 2017;56(2):337.

38.

Hodacova

, Sustova

, Cermakova

, Kapitan

, Smejkalova

. Self-reported risk factors related to the most frequent musculoskeletal complaints among Czech dentists. Industrial Health. 2015;53(1):48–55.

39.

Basher

, Nath

, Siddique

, Rahman

, Rubel

, Sayed

, et al. Musculoskeletal disorder (MSD) among agricultural workers. Mymensingh Medical Journal: MMJ. 2015;24(1):1–8.

40.

Mishra

, Amitabha

, Iqbal

, Gangopadhyay

, Chandra

. An integrative approach for evaluating work related musculoskeletal disorders. Work (Reading, Mass); 2012.

41.

Roll

, Selhorst

, Evans

. Contribution of positioning to work-related musculoskeletal discomfort in diagnostic medical sonographers. Work (Reading, Mass). 2014;47(2):253–60.

42.

Al-Hourani

, Nazzal

, Khader

, Almhdawi

, Bibars

. Work-related musculoskeletal disorders among Jordanian dental technicians:Prevalence and associated factors. Work. 2017;56(2017):617–23.

43.

Mebarki

, Fouatih , Mohammed El Amine

, Mokdad

. Quality of Work Life and differences in demographic characteristics among managerial staff in Algerian tertiary sector. Work. 2019;62(3):435–42.

44.

Salminen

, Perttula

, Ratilainen

, Kuosma

. The effect of demographic factors on occupational injuries. International Journal of Occupational Safety and Ergonomics; Taylor & Francis. 2017;23(2).

45.

Rothmore

, Gray

. Using the Work Ability Index to identify workplace hazards. Work. 2019;62(2):251–9.

46.

Andersen

, Sundstrup

, Brandt

, Efat Lali

, Roger

, Markus

. Jakobsen. Factors associated with high physical exertion during manual lifting:Cross-sectional study among 200 blue-collar workers; Work, IOS Press.; 2018;59;59–66.

47.

Yanwen

, Yatao

, Xiaoyuan

, Xunwen

, Xiaoshan

. The effect of demographic and social factors on return to work of injured workers with musculoskeletal disorders. Work. 2008;30(2008):91–95 91.

48.

Mohd

, Mohd

, Sangwoo

, Myung

, Mohd Syukri

. The effects of physical and psychosocial factors and ergonomic conditions on the prevalence of musculoskeletal disorders among dentists in Malaysia. Work. 2017;57(2):1–12.

49.

Hignett

, McAtamney

. Rapid entire body assessment (REBA). Applied Ergonomics. 2000;31(2):201–5.

50.

Crawford JO. The Nordic Musculoskeletal Questionnaire. Occupational Medicine. 2007;57(4):300-1, https://doi.org/10.1093/occmed/kqm036

51.

Kuorinka

, Jonsson

, Kilbom

, Vinterberg

, Biering-Sørensen

, Andersson

, Jørgensen

. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Applied Ergonomics. 1987;18(3):233–7.

52.

Taghipour

, Mahboobi

, Gharagozlou

. The Impact of ICT on Knowledge Sharing Obstacles in Knowledge Management Process (Including Case-Study). Iranian Journal of Information processing and Management. 2016;31(4):1049–74.

	0		1		2		3		4		Total
Work unit	Negligible		Low		Average		High		Very high		(n = 80)
	1		3–2		7–4		10–8		11–15
	N	%	N	%	N	%	N	%	N	%	N	%
Internal core production	0	0	0	0	23	51.11	11	24.44	11	24.44	45	56.25
Preparation of external shell	0	0	0	0	8	61.53	5	38.46	0	0	13	16.25
Quality control	0	0	0	0	9	100	0	0	0	0	9	11.25
Packaging	0	0	0	0	2	28.57	5	71.42	0	0	7	8.75
Warehouse	0	0	2	100	0	0	4	66.66	0	0	6	7.5
All jobs	0	0	2	2.5	42	52.5	25	31.25	11	13.75	80	100

Assessing ergonomic risk factors using combined data envelopment analysis and conventional methods for an auto parts manufacturer

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1 Introduction

2 Literature review

3 Methodology

3.1 The REBA posture evaluation method

Table 1 Details of the REBA method Performance necessity Correction performance level Risk level Score Not necessary 0 Ignored 1 May be necessary 1 Low 2–3 Is necessary 2 Average 4–7 Was necessary earlier 3 High 8–10 Immediately necessary 4 Very high 11–15

3.4 Sensitivity analysis

Table 10 Correlation results of final scores for the REBA and QEC methods and correlation of their action levels for all jobs All jobs Final scores Action level 0.633 0.635

Conflict of interest

References

Table 1
Details of the REBA method

Performance necessity Correction performance level Risk level Score

Not necessary 0 Ignored 1

May be necessary 1 Low 2–3

Is necessary 2 Average 4–7

Was necessary earlier 3 High 8–10

Immediately necessary 4 Very high 11–15

Table 10
Correlation results of final scores for the REBA and QEC methods and correlation of their action levels for all jobs

All jobs Final scores Action level

0.633 0.635