Modeling Human Perception of Postural Stress by Jack Odunga Ogutu A dissertation submitted to the Graduate Faculty of Auburn University in partial ful llment of the requirements for the Degree of Doctor of Philosophy Auburn, Alabama May 5, 2013 Keywords: Work-related musculoskeletal disorders, Static Posture Holding, Posture Holding Time, Ergonomics evaluation. Copyright 2013 by Jack Odunga Ogutu Approved by Woojin Park, Chair, Assistant Professor of Industrial and Systems Engineering. Richard F. Sesek, Assistant Professor of Industrial and Systems Engineering. Saeed Maghsoodloo, Professor of Industrial and Systems Engineering. George T. Flowers, Professor of Mechanical Engineering, Dean of Graduate School. Abstract Static posture holding (SPH) tasks refer to a variety of manual work tasks that require workers to statically maintain working postures over certain time periods. SPH tasks are common across various industrial sectors including agriculture, manufacturing, chemical and construction. Time durations of SPH tasks, i.e., posture holding times (PHT), vary greatly, ranging from a few seconds to a few tens of minutes per single execution. Inadequately designed SPH tasks can impose large physical stresses on the human mus- culoskeletal system. Excessive physical stress is known to be a risk factor of work-related musculoskeletal disorders (WMSDs). It also compromises worker?s job satisfaction and work productivity. Therefore, the physical stresses of SPH tasks must be accurately evaluated, and if necessary, be controlled through ergonomic interventions so as to enhance workers? occupational health and general well-being. Discomfort-time sequences of SPH tasks were investigated in an e ort to determine the existing relationship between perceived discomfort and posture holding time. Previous research studies suggested that an increase in PHT results in increased perceived discomfort and the discomfort-PHT relationship may be described using a simple mathematical function form with a small set of parameters. However, the existing studies do not seem to have fully elucidated the mathematical characteristics of the discomfort-PHT relationship. Multiple studies suggested that a simple linear time function form can represent the relationship. On the other hand, Reneman et al. (2001) reported that a negatively accelerated logarithmic ii time function form depicts the relationship. This study proposed the power function as an adequate representation of the discomfort-PHT relationship of SPH tasks. This function form is capable of representing three distinct monotonically increasing time patterns, that is, linear, negatively accelerated and positively accelerated time increase patterns, depending on the choice of the exponent parameter value. An investigation was conducted on the inter-individual variation in perceived discomfort of static posture holding. The level of discomfort experienced by a worker conducting a SPH task is assumed to be a ected by among other parameters, the worker?s physical and psycho- logical characteristics. Thus, even when performing identical tasks di erent individuals with di erent physical and psychological characteristics would experience di erent levels of dis- comfort. Therefore, for a given SPH task, a group or a population of individuals gives rise to a probability distribution of perceived discomfort, which describes the inter-individual varia- tion in discomfort perception. Such mathematical depiction of the inter-individual variation can greatly help determine the proportion of the workforce that would experience excessive (or manageable) discomfort from an SPH task and further assist in deciding the acceptability of the task from a population accommodation point of view. Determining which SPH tasks are acceptable or unacceptable or whether current SPH tasks need to be redesigned or not requires a certain index/metric that operationalizes the construct of discomfort level of an SPH task in a manner that guides designers decision making. This research work presents a new quantitative index for characterizing discomfort levels of SPH tasks. This new index is named the population accommodation level estimate (PALE) and estimates the proportion of the target worker population that experiences less than excessive discomfort. This new index is predicated upon the use of empirical discomfort iii distributions. The probabilistic approach employed in this study may allow for a more direct determination of accommodation levels for various manual work tasks. Using this approach, e ective intervention strategy can be made based on desired accommodation performance. In conclusion, the main ndings from this dissertation work were as follows: (1) three distinct time increase patterns, namely, the linear, negatively accelerated and positively accelerated time increase patterns, characterize discomfort-time sequences of prolonged SPH trials (2) the relationship between PHT and perceived discomfort in SPH can be adequately described by the power function form (3) di erent individuals can experience signi cantly di erent postural stresses even in identical manual work tasks. A consequence of the inter- individual di erences in work performance is that simple descriptive statistics become limited in describing such tasks, and (4) psychophysical perception of discomfort can be expressed in probabilistic terms. Re-design recommendations may now be based on the probability distribution of discomfort ratings. Discomfort levels os SPH tasks can be quanti ed by a metric/index predicated on the use of empirical discomfort distributions. Such index may be useful for decision making involved in ergonomics design interventions. iv Acknowledgments I would like to thank Dr. Woojin Park for his invaluable guidance and time in advising this dissertation work. His patience, mentorship and encouragement helped me through all the demands and rigours of graduate school. Much appreciation to Dr. Richard Sesek and Dr. Saeed Maghsoodloo who agreed to o er their time, insights and encouragement throughout the entire process. Dr. Oladiran Fasina, thank you for reading the entire dissertation and o ering your criticism, suggestions and corrections. I would also like to thank The University of Utah, Rocky Mountain Center for Occupational and Environmental Health and The University of Kentucky College of Public Health, Southeast Center for Agricultural Health and Injury prevention, for providing part of the nancial support that made this research work possible. Sincere appreciation to my colleague Tami Blackwell who assisted with data collection, thank you for your friendship, invaluable time and support. To the chair and faculty members of Auburn University Department of Industrial & Systems Engineering, thank you for the support and encouragement. I remain indebted to you all. To my parents, who have relentlessly supported my academic pursuits, thank you for believing in me, spending and being spent, just to make sure I stood a chance in life. I will forever be grateful. To my American friends, Andrew Dahl and Matthew Perrella, you are the best, thank you for letting me be part of your families, you will still be in my heart in years to come. My time in America has been worth the while largely because of friends like v you. To everyone else I have come to know over the course of my life pursuits, thank you very much. Thank you for helping me navigate the uncertain waters of life. Last but not least, I would like to thank God: Where would I have been without you? You have been a friend, a brother and a father. I am blessed beyond measure. \Having obtained help from the Lord, I have come this far" (Acts 26:22 ). This has been a journey that I have learned alot from the di erent people that God has brought my way. Jesus, thank YOU. vi Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Research and Dissertation Organization . . . . . . . . . . . . . . . . . . . . . 4 2 A Review of the literature on discomfort perception during static pos- ture holding tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.2 Posture Evaluation Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2.1 Ovako Working Posture Analysing System(OWAS) . . . . . . . . . . 14 2.2.2 Rapid Upper Limb Assessment(RULA) . . . . . . . . . . . . . . . . . 14 2.2.3 Rapid Entire Body Assessment(REBA) . . . . . . . . . . . . . . . . . 15 2.2.4 The Borg (CR-10) scale . . . . . . . . . . . . . . . . . . . . . . . . . 16 3 The relationship between perceived discomfort of static posture hold- ing and posture holding time . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.2.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2.3 Data Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 vii 3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4 An investigation on inter-individual variation in perceived discomfort of static posture holding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 4.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 4.2.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.3 Data analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.4.1 Probability distributions for modeling the inter-individual variation in perceived discomfort . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.4.2 Statistical analyses on the magnitude of inter-individual variation in perceived discomfort. . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5 Quantifying discomfort levels of static posture holding tasks using em- pirical distributions of perceived discomfort. . . . . . . . . . . . . . . . . 75 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 5.2 The population accommodation level estimate (PALE) index . . . . . . . . . 78 5.2.1 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 5.2.2 Probability histogram representation . . . . . . . . . . . . . . . . . . 79 5.2.3 Population accommodation level estimate . . . . . . . . . . . . . . . 79 5.3 An illustrative example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5.3.1 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5.3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 6 Conclusion and Reccomendations for Future Studies . . . . . . . . . . . 91 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 viii 6.2 Summary of ndings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 6.3 Limitations and reccomendations for future studies . . . . . . . . . . . . . . 94 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 A Informed consent for the two experiments conducted for the purposes of this dissertation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 B IRB Participant screening document . . . . . . . . . . . . . . . . . . . . . . . . 108 C IRB Participant screening document. . . . . . . . . . . . . . . . . . . . . . . . . 111 D Manuscript 1: The Thirteen Inadequate Discomfort-Time Pro les. . . . . . . . . 113 ix List of Figures 2.1 Borg?s new category-ratio (CR-10) scale. Reprinted from Borg 1982. . . . . . . 17 3.1 The 12 box holding postures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2 The experimental set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.3 Some discomfort-time sequences exhibiting each of the three distinct time increase patterns: (a) linear, (b) negatively accelerating and (c) positively accelerating. For each time increase pattern, ve randomly selected discomfort-time sequences are presented . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.4 The mean TAD and MAD values of the three function forms computed for each of the twelve postures: (a) TAD and (b) MAD . . . . . . . . . . . . . . . . . . . 31 3.5 Summary of the tting performance measure for the three function forms. . . . 31 3.6 Examples of adequate functional representation of discomfort-time sequence data (TAD<0.5 and MAD<1): (a) linear, (b) logarithmic and (c) power. . . . . . . . 32 3.7 The distributions of the power function parameters for the 347 discomfort-time sequences represented as power functions with TAD<0.5 and MAD<1: (a) the scaling factor C1, (b) the exponent C2 and (c) the intercept C3 . . . . . . . . . 33 x 4.1 Body part postures used to generate the 180 whole-body posture categories used in the current study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.2 Borg?s new category-ratio (CR-10) scale. Reprinted from Borg 1982. . . . . . . 48 4.3 Histogram showing the tted distributions for the 180 SPH tasks. A total of 14 distributions available in Minitab were used to t the data. . . . . . . . . . . . . 52 4.4 (a) Both legs straight, Arm below the shoulder level, Straight back and holding a 3.6 kg hand-load. (b) Both legs straight, body moved forward by the limbs, back bent and twisted, arm raised above the shoulder level and holding a 2.2 kg load. (c) Both legs straight, arm above the shoulder level, Back bent by approximately 25 degrees and holding a 2.2 kg load (d) One leg bent at approximately 150 degrees, and on a raised support, Arm at 90 degrees , Straight back and holding a 3.6 kg hand-load.(e) Both legs straight, arm below the shoulder level, Straight back and holding a 2.2 kg hand-load. (f) Both legs bent approximately 150 degrees , arm below the shoulder level, Straight back and holding a 2.2 kg hand- load. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 4.5 (g) Both legs straight, arm above the shoulder level, Straight back and holding a 3.6 kg load (h) Both legs straight, arm above the shoulder level, Back bent by approximately 25 degrees and holding a 3.6 kg load. (i) Both legs straight, Arm at 90 degrees , Straight back and holding a 3.6 kg hand-load (j) Both legs bent approximately 150 degrees, body moved forward by both limbs, Straight back, arm raised above the shoulder level and holding a 3.6 kg load. . . . . . . . . . 54 xi 4.6 Histograms showing the distribution of standard deviation and range for 180 SPH tasks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.7 Histograms showing the distribution of Mean and Median for 180 SPH tasks. . . 56 4.8 Three-way interaction plot for the back, leg and load interaction. . . . . . . . . 59 4.9 Three-way interaction plot for the arm, back and leg interaction. . . . . . . . . 59 4.10 Three-way interaction plot for the back, leg and load interaction for the range. 60 4.11 Three-way interaction plot for the back, leg and load interaction for standard deviation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.12 Three-way interaction plot for the arm, back and leg interaction for standard deviation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.13 Three-way interaction plot for the back, leg and load interaction for the range. 63 4.14 Three-way interaction plot for the back, leg and load interaction for Mean. . . 64 4.15 Three-way interaction plot for the arm, back and leg interaction for Mean. . . 65 4.16 Three-way interaction plot for the back, leg and load interaction for Median. . 66 4.17 Three-way interaction plot for the arm, back and leg interaction for Median. . 67 4.18 The relationship between Mean discomfort and Range within a speci ed SPH task. A total of 180 SPH tasks are considered. . . . . . . . . . . . . . . . . . . . 68 xii 4.19 The relationship between Mean discomfort and the Standard Deviation (SD) within a speci ed SPH task. A total of 180 SPH tasks are considered. . . . . . . 69 4.20 The relationship between Median and Range of discomfort within a speci ed SPH task. A total of 180 SPH tasks are considered. . . . . . . . . . . . . . . . . . . . 69 4.21 The relationship between Median and Standard deviation (SD) within a speci ed SPH task. A total of 180 SPH tasks are considered. . . . . . . . . . . . . . . . . 70 5.1 Borg?s new category-ratio (CR-10) scale. Reprinted from Borg 1982. . . . . . . 80 5.2 An illustration of the experimental set-up for the SPH task trials including the VICON capture shot of a static trial. . . . . . . . . . . . . . . . . . . . . . . . . 82 5.3 Examples of Probability histograms of discomfort ratings for selected SPH tasks. 84 5.4 Accommodation levels for the study population representing 180 SPH tasks. . 86 A.1 The IRB consent form for study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 104 A.2 The IRB consent form for study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 105 A.3 The IRB consent form for study 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 106 A.4 The IRB consent form for study 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 107 B.1 The IRB screening document . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 B.2 The IRB screening document. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 xiii C.1 Emergency Action Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 D.1 The Discomfort-time sequences that could not adequately t the power function form. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 D.2 The Discomfort-time sequences that could not adequately t the power function form. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 xiv List of Tables 2.1 Action levels for RULA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.2 Action levels for REBA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.1 Summary of the age, height, body mass and the BMI data for the two participant groups. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2 The ANOVA results for the Time-Averaged Deviations(TAD) and Maximum Absolute Deviation(MAD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4.1 Summary of the age, height, body mass and body mass index (BMI) data for the male and female participant groups. . . . . . . . . . . . . . . . . . . . . . . . . 46 4.2 The ANOVA results for the two measures of spread: standard deviation and range 57 4.3 The ANOVA results for the two measures of center: mean and median. . . . . 58 xv Chapter 1 Introduction The term Musculoskeletal disorder (MSD) refers to the type of disorder charcterized by a disturbance of structure or function or both due to a genetic or embryological failure in development or as the result of exogenous factors such as certain chemical substances, injury or disease. In the workplace, these type of disorders could be the result of prolonged mus- cular e ort, repeated actions or non-natural postures. Awkward, stressful working postures are known to be associated with increased risks of work-related musculoskeletal disorders (WMSD) (David et al., 2008, Kee and Lee, 2005). The U.S Department of Labor reported that Musculoskeletal disorders (MSDs) cases accounted for 33% of all injury and illness cases in 2011 (BLS, 2011). A total of 387,820 MSDs were recorded in 2011 yielding a rate of 39 cases per 10,000 full-time workers (BLS, 2011). According to research published by the Na- tional Institute of Occupational Safety and Health (NIOSH), there is strong evidence for a relationship between physical exertion and work-related musculoskeletal disorders (WMSD) ( Benard et al., 1997, ). A literature survey of published sources have shown a strong causal relationship between awkward postures and MSDs (Baron et. al., 1991, Hignett, 1996, Myers et. al., 1999, Ngan et al., 2010, Cook et al., 2001, Trinko et al., 2003, Janowitz et al., 2011). Thus, to reduce WMSD risks and promote workers health, postural stresses from manual work tasks must be accurately evaluated, and if necessary, be controlled through e ective 1 interventions (Putz-Anderson, 1988; Burdorf, et al., 1991; Armstrong et al., 1993, Hignett 2003, Silverstein and Clark 2004, Denis, et. al., 2008). Two main constructs are discussed in this dissertation: (1) Static Posture Holding (SPH) tasks, which require workers to maintain xed working postures for certain time durations and (2) Time durations of posture maintenance re ered to as, Posture Holding Time (PHT). The literature shows several studies investigating the relationship between perceived discomfort of SPH and PHT (Kirk and Sayodama 1973, Corlett and Manenica 1980, Miedema et al. 1997, Reneman et. al. 2001 and Kee 2004). In these studies, human subjects performed xed duration SPH tasks and then subjectively rated the corresponding levels of perceived discomfort. The Borg CR-10 and the magnitude estimation method, were employed for the self-assessment of perceived discomfort. This dissertation discusses some of the theories and studies on discomfort-time relation- ships for static maintenance tasks and evaluation of work-related postural stresses. We also introduces a novel probability-based method for characterizing postural stresses of manual work tasks. The rst part of this work addresses the lack of agreement among researchers in the eld of ergonomics concerning choice of an adequate mathematical function form for representing the discomfort-time relationship of SPH tasks. The second part of this disser- tation discusses some of the limitations of the existing posture evaluation tools - inability to describe the individual di erences in the perception of work-related postural stresses. Fi- nally, this work introduces a new probability-based technique for characterizing discomfort levels of SPH tasks that is yet to be found in literature. Corlett and Manenica (1980) postulated that the two main parameters in uencing dis- comfort were the PHT and muscular force and that their mathematical relationship was 2 a logarithmic one, which could be expressed in a linear form. Manenica (1986) showed that estimated discomfort for static posture maintenance tasks increased linearly as a func- tion of time. Kee (2004) investigated the relationship between, external load, upper limb postures and PHT on perceived discomfort. Discomfort was found to increase linearly as PHT increases. On the other end of the spectrum, Reneman et al.(2001) investigated the discomfort-time relationship and contended that a negatively accelerated logarithmic time function form depicts the relationship. These and several other studies showed that there is lack of agreement in characterizing discomfort-time relationships. An additional observa- tion was that in some of the studies the researchers aggregated discomfort-time sequences obtained from multiple SPH trials and performed statistical analyses on the combined time- series dataset or on the mean response. 1.1 Research Objectives The review of literature on discomfort perception of manual work tasks revealed a gap in the characterization of discomfort-time relationship and a relatively simple characterization of postural stress that may be insu cient for describing which postures are generally regarded as stressful and which are not. A few studies have attempted to charaterize mathematical characteristics of the discomfort-time relationship by examining individual discomfort-time sequences of prolonged SPH trials. Consequently, the discomfort-time relationship at in- dividual trial level is not clearly understood. Also, past research studies seemed to have investigated task parameters that a ect the mean of perceived discomfort across di erent individuals without sheding much light on the variation between individuals. The lack of 3 ability to describe inter-individual di erences means that it is impossible to accurately esti- mate the percentage of workers who can perform a manual work task with minimum postural discomfort. Thus, the objectives of this study were : (1) Elucidate the nature of the rela- tionship between perceived discomfort of SPH and PHT, with a speci c focus on identifying time-increase patterns of individual-time sequences, and (2) Address some of the research gaps in the eld of ergonomics with regard to SPH task evaluation, one of which is lack of ability to account for the individual di erences in the perception of postural stresses. Addressing these research questions will enhance our understanding of the discomfort-time relationship in SPH. Also, addressing the lack of inter-individual variability of the existing assessment tools will be a step foward in advancing the way working postures are evaluated and subsequent intervention strategies designed. 1.2 Research and Dissertation Organization This dissertation is comprised of six chapters, organized as follows: Chapter one is a brief introduction to the two phases of this research work and chapter six will present the conclusion, study limitations and proposed future studies. In chapter two, a review of literature is presented covering the two phases of this study. The literature review largely covers static posture holding tasks and the perception of discomfort in manual work tasks. Further review ofthe relevant literature is presented in each of the main manuscripts. Chapter three explores the relationship between perceived discomfort of static posture holding and posture holding time. Chapter four demonstrates the inter-individual variability in perceived postural stresses associated with static working postures. Chapter ve introduces a novel probability-based method for quantifying discomfort levels of SPH tasks. 4 The limitations of the study, recommendations for future studies and conclusions are discussed in chapter six. The appendices outlines the speci c experimental protocols, de- tails on subject recruitment, summaries of collected data and other relevant information pertaining to this dissertation. 5 Chapter 2 A Review of the literature on discomfort perception during static posture holding tasks 2.1 Introduction Static posture holding (SPH) tasks refer to manual work tasks that require workers to statically maintain postures for a few seconds to tens of minutes. These tasks are common across many di erent occupations and work groups and can be found in most industrial sectors, including the manufacturing, service, agriculture, mining and construction industries (Chung et al. 2003a, Chung et al. 2005). Time durations of SPH tasks, referred to as posture holding times (PHTs), vary greatly ranging from several seconds to a few tens of minutes depending on the muscle group involved. SPH tasks can impose signi cant biomechanical and physiological stresses on the mus- culoskeletal system, and thereby, cause physical strains to performers - maintaining cer- tain postures, especially with external loads, requires forceful isometric muscle contractions. When performing such SPH tasks, workers would experience rapid increase in muscle fatigue and perceived discomfort/pain. Even postures that require low-level muscular exertions may eventually lead to these undesirable consequences when sustained for long periods of time (Rohmert 1960). Static posture holding is also known to hamper blood circulation and perfusion, which may lead to ischemia at local body regions ( Buckle and Devereux 2002). 6 In combination with biomechanical and physiological stresses, such responses may result in degenerative changes to soft tissues and contribute to chronic pain and functional im- pairments. Multiple research studies suggest that static work is indeed a risk factor for work-related musculoskeletal disorders (WMSDs) (Westgaard and Aaras 1984, Keyserling et al. 1988, Genaidy and Karwowski 1993, Putz-Anderson and Galinsky, 1993, Miedema et al. 1997, Hgg 1998) although the exposure-e ect relationships have not been fully elucidated. All things considered, stresses/strains of SPH tasks must be accurately evaluated, and if necessary, be controlled through adequate interventions so as to enhance the occupational health, productivity and general well-being of those who perform SPH tasks. One approach widely adopted to assess stresses/strains of manual work tasks is to use workers? self-assessment of task-related discomfort/pain; in this empirical, psychophysical approach, individuals rate the discomfort/pain associated with a task of interest while or after performing a trial of the task. A psychophysical discomfort/pain scale or rating method is utilized for subjective ratings. The use of perceived discomfort/pain in the evaluation of manual work tasks seems to be justi ed on the following grounds: rst, perceived discom- fort/pain has been shown to be correlated with physical stress and strain measures, such as muscle fatigue (Noble et al. 1981, Putz-Anderson and Galinsky 1993) and static body joint moments (Boussenna and Corlett, 1982, Jung and Choe, 1994). Discomfort and WMSDs are both related to the exposure of the musculoskeletal system to biomechanical loads (Mil- ner 1985, Nag 1991, Putz-Anderson and Galinsky 1993, Dul et al. 1994, Miedema et al. 1997). Researchers in the eld generally view discomfort/pain as a precursor of work-related injuries (Corlett and Bishop 1976, Kee and Lee 2012). Minimization of discomfort will pre- sumably contribute to reduction of the WMSD risks (Dul et al., 1994). Second, discomfort is 7 by de nition undesirable and brings about negative consequences, such as productivity loss and low job satisfaction and, therefore, an important evaluation criterion for manual tasks, independently of the question if it can be used as an estimator of the risk of WMSDs (Dul et al. 1994, Miedema et al. 1997). Multiple previous studies have employed the \self-assessment of perceived discomfort (pain)" approach in investigating the stresses/strains of SPH tasks. Also, multiple studies investigated perceived exertion, perceived postural stresses and perceived physical strain during static postural tasks using the psychophysical approach. These constructs seem to be similar to the notion of perceived discomfort in that they all aim to measure physical strain during work tasks. In studying these constructs, these research studies have generally looked into the three main task parameters that specify a SPH task including: posture, posture holding time and external load. Summarized reviews of some such studies are presented in the paragraphs that follow: Olendorf and Drury (2001) evaluated postural discomfort and perceived exertion asso- ciated with 20-seconds long SPH tasks. One hundred and sixty eight postures representing the postures in the Ovako Working-posture Analysing System (OWAS) were considered. The postures comprised combinations of three arm postures, four back postures, seven leg postures and two load levels. The two force categories used- 1.1 kg and 10.1 kg- represented the lower ends of the OWAS force categories.The Borg CR10 scale and ve-point body part discomfort scale were used as rating methods. Twelve male subjects participated in their study. The task required the study participants to statically hold the box for 20 seconds in a posture and self-report their perceived exertion at the end of the 20 seconds. A large poster of the Borg CR-10 scale was positioned in the line of view of the subjects. The results of their 8 study did indicate that external load was a major driving factor in body part discomfort. Speci cally, they reported that increasing the load level resulted in higher discomfort scores. Additionally, their study also did report increased discomfort and fatigue for tasks that had arm postures above the shoulder level and leg postures that required the subject to perform the static posture holding task with both knees in exion and one knbee in exion. A number of studies have utilized the method of magnitude estimation to evaluate pos- tural discomfort. In this method, the subject is presented with stimuli and asked to assign a value to the perceived magnitude of stimulus. This is analogous to the respondent setting his/her own standard for measuring their discomfort. In this regard, magnitude estimation methods are di erent from the Borg CR-10 method which provides the respondent with verbal anchors by which to rate their perception of discmfort. In their study, Chung et al. (2003b) examined time changes of perceived discomfort of 16 minutes long squatting tasks with four di erent stool heights. The subjects rated whole-body and body part discomfort levels using the free modulus magnitude estimation method every 2 minutes while holding the postures for 16 minutes. The results of their study reported a linear relationship be- tween discomfort and posture holding time. Chung et al. (2003a) evaluated 31 leg postures based on a subjective discomfort rating. Subjects maintained each posture for 1 minute. The free modulus magnitude estimation method was used. The task involved the subjects adopting a speci c posture and upon completion, reporting their numeric estimates for nine verbal descriptors representing level of discomfort. The rst descriptor was 2.2 (extremely comfortable) and the last, 100.0 (extremely uncomfortable). The numeric estimates were self-selected by each subject and averaged across the subjects. The main nding of this study was that leg postures had signi cant e ects on discomfort perception. Chung et al. 9 (2005) performed a series of experiments to collect perceived discomfort data associated with various postures. In their experiments, subjects maintained each posture for 1 minute. The free modulus magnitude estimation method was used. In their study, Genaidy et al. (1995) developed a ranking system for the stressfulness of the non-neutral static postures around body joints. This was based on the ratings of perceived discomfort. The subjects rated the joint discomfort ratings following a 60- second period of static posture holding. A linear scale of 0-10 perceived discomfort ratings was used to assess the level of joint discomfort. Overall, they did report that the highest discomfort ratings were found in the shoulder joints, followed by the wrist, elbow, lower back and neck region. Kee and Karwowski (2003) developed a similar ranking system but one based on joint motions. Sitting and standing joint motions were considered in this study. Subjects rated their perceived discomfort using the magnitude estimation method. Their study showed that the discomfort ratings for neutral postures were di erent depending on the joint involved. In particular, hip and lower back motions were reported to exhibit higher discomfort ratings, while elbow joint motions had lower discomfort ratings. Park et al. (2009) examined the e ects of obesity on perceived postural stress associated with a 20 seconds long static box holding task. A total of 84 postures were considered in their study. The Borg CR10 scale of perceived exertion was used as an index of discomfort. The task involved statically holding a 5 kg box for 20 seconds. The study identi ed non-straight back, elevated arms and exed knees as SPH task parameters that increases stress for both obese and non-obese groups. This study went further and examined the relationship between the RPE data and the four OWAS action codes. It was demonstrated in this study that the obese and non-obese groups perceived the level of postural stress (RPE) corresponding to 10 each OWAS code di erently. The study therefore raised concerns with the use of the OWAS action codes for both the obese and non-obese workers. Their ndings concluded that OWAS action codes could not be e ectively used to quantify WMSD risks for obese workers. In investigating the maximum holding times (MHTs) of SPH tasks, Reneman et al. (2001) considered two static postural tasks: forward bending and overhead work. This study examined the relationship between perceived exertion and maximum holding times. Subjects performed maximal capacity static holding task and reported their perceived discomfort every 30 seconds until task termination. The Borg CR-10 scale was used to evaluate postural discomfort. They reported a logarithmic relationship between average CR-10 scores and posture holding times. Jung et al. (2010) examined a 15- minute long static-sustaining task. Subjective dis- comfort, heart rate and EMG median frequency were measured every 3 minutes. A total of 13 postures were considered. The subjects were required to self-report their subjective rat- ing of discomfort every 3 minutes while maintaining a randomly selected lower limb posture. Borg?s RPE (range: 6-20) and Borg?s CR-10 (range: 0-10), were used as subjective measures for evaluating discomfort. They reported that discomfort was signi cantly a ected by pos- ture. Speci cally, Knee exion postures and kneeling postures had the highest discomfort scores over time. The study conducted by Manenica (1986) examined maximum duration posture holding tasks for seven di erent postures. The goal was to nd a technique for quick and reliable postural load assessment. Fifteen female subjects participated in a tapping task in each of the seven di errent postures. The subjects reported their perceived discomfort every 30 seconds until each posture was terminated. At the termination of each postural task, the 11 subject reported the body part with the highest discomfort. Discomfort perception was assessed on the 20-point RPE scale proposed by Borg (1973). Based on the results of this study, discomfort was found to increase linearly as a function of time. Overall, the study concluded that subjective ratings could be reliable in measuring postural load discomfort. Eperiments were conducted by Kirk and Sadoyama (1973) to investigate the discomfort- time relationship during SPH tasks. Discomfort was measured for a static pull task and a two-handed static torque production task. Eighteen subjects took part in this study. Six load levels were used and the task involved the subjects holding the load using a handle in the two modes for maximum duration, while reporting their discomfort every 30 seconds. They used a ve-point rating scale to report their perceived exertion. A rating of 1 was de ned as just noticeable discomfort while a rating of 5 was de ned as extremely uncomfortable. The relationship between holding time and percentage of maximum holding force was found to be linear. Kee (2004) empirically investigated the e ects of external load, upper limb postures and PHT on perceived discomfort. The subjects held given postures for 60s and rated their subjective discomfort scores at 5, 20, 40 and 60 seconds using the free modulus method of magnitude estimation. The e ects of external load and holding time were much larger than those of upper limb postures. Mean discomfort was found to increase linearly as posture holding time increases. A literature was conducted by Kee and Lee (2001) to investigate the relationship between discomfort and several other measures for postural assessment in- cluding posture holding time, maximum holding time, torque and joints and lifting index. 12 Their survey of literature reported a number of ndings including; a linear relationship be- tween discomfort and posture holding time and an inverse relationship between whole body discomfort and maximum holding time. As has been stated earlier, all the preceding studies used either the Borg scales or the magnitude estimation methods to assess perceived discomfort of SPH tasks. The magni- tude estimation method and the Borg CR-10 scale both yield ratio-scale measurements, and therefore, support a variety of statistical analyses. 2.2 Posture Evaluation Methods Researchers in the eld of ergonomics have extensively studied the evaluation of pos- ture and static loads (Corlet et. al., 1986). Postural evaluation of manual work tasks is achieved by use of : (1) direct observation methods, (2) indirect observation methods, and (3) subjective methods. Some of the direct observation methods utilized by analysts include: The Ovako Working Posture Analysing System (OWAS), Rapid Upper Limb Assessment (RULA), and Rapid Entire Body Assessment (REBA). Direct observation methods include instrumentation techniques, such as: Goniometers, Motion capture (Vicon) and Electromyo- graphy (EMG). Subjective evaluation of postures is done based on the subjective strain experienced by individuals in various postural load tasks. Various subjective methods are used in evaluation of static work tasks inlcuding: Borg?s CR-10 scale and the magnitude estimation ratio-scale. 13 2.2.1 Ovako Working Posture Analysing System(OWAS) This study used various working postures based on the OWAS posture classi cation system. OWAS is a work sampling-based method for identifying poor working postures (Karhu et. al., 1977). The system prede nes four postures of the back, three of the upper limb and seven for the lower limbs. These three body parts provide a total of 84 whole body posture categories. The OWAS technique also employs three levels of hand-held load (three categories): handload<10kg, 10 - 20kg, handload>20kg. OWAS therefore utilizes a total of 252 posture-load combinations to represent various industrial working postures. These 252 posture -load combinations are then classi ed into predetermined stress levels and expressed in terms of a 4-level OWAS action codes. The four action codes represent di erent stress levels indicating varying needs for ergonomic interventions. The four action levels are de ned as follows: Action level 1: no corrective measures (normal postures). Action level 2: corrective measures in the near future (stressful postures). Action level 3: corrective measures as soon as possible (stressfull postures). Action level 4: corrective measures immidiately (awkward postures ). 2.2.2 Rapid Upper Limb Assessment(RULA) The RULA technique was developed by McAtamney and Corlett (1993) as an assessment to disorders of the upper limb. The technique observes postures of the upper limbs, back and legs classi es them into four action levels. The RULA technique uses prede ned body postures and three scoring tables to obtain a posture score representing stress level. The 14 system divides the body parts into two posture scores: A (trunk, neck and legs) and B (upper arm, lower arm and wrists). These scores are obtained from scoring tables and added together to determine a grand score. The grand scores are divided into four action levels de ned as follows: Table 2.1: Action levels for RULA Action level RULA score Recommended action 1 1-2 Posture acceptable 2 3-4 Changes may be required 3 5-6 Investigation and changes needed soon 4 7 Changes required immidiately The RULA technique has been validated as an assessment tool for upper limb muscu- loskeletal risks by McAtamney and Corlett (1993). The scoring system provides an indication of the level of stress/discomfort by individual body parts and serves as a basis for ergonomic interventions. 2.2.3 Rapid Entire Body Assessment(REBA) The REBA technique was developed by Hignett and McAtamney (2000) to address working postures in the healthcare industry. The tool addresses additional loading muscu- loskeletal risks that are mostly encountered by healthcare practioners. REBA is based on the same format as OWAS and RULA in that the body is divided into groups. The rst group comprises of the trunk, neck and leg postures, while the second group comprises of the upper arm, lower arm and wrist postures. Posture scores are assigned to each body part. Two scoring tables allow postures and loads to be combined to give a score for each group. 15 REBA provides ve action codes representing stress levels. The ve action levels are de ned as follows: Table 2.2: Action levels for REBA Action level REBA score Risk level Recommended action 0 1 Negligible None necessary 1 2-3 Low May be necessary 2 4-7 Medium Necessary 3 8-10 High Necessary soon 4 11-15 Very High Necessary NOW 2.2.4 The Borg (CR-10) scale The Borg scale (1982) is an overall intergrated con guration of the signals, perceptions, and experiences of the body while enduring physical strain. It is essentially a category-ratio scale that relates physical workload and subjectively experienced strain. The Borg?s CR-10 (Figure 2.1) is therefore capable of highlighting inter-individual di erences in the perception of experienced strain (Noble and Robertson, 1996). The scale has ratings from 0 to 10 with verbal explanations of each rating. A detailed de nItion of the ratings and the comprable verbal anchors are explained to the subject before administering the test. The verbal anchors provide the test subject with a simpler way of interpretaing the subjective ratings. A large print of the scale is then placed in the line of sight of the test subject at the begining of the task and he or she is asked to rate the percived exertion immidiately the task is completed. 16 0 - 0.5 - 1 - 2 - 3 - 4 - 5 - 6 - 7 - 8 - 9 - 10 - . Nothing at all Extremely weak (just noticeable) Very weak Weak (light) Moderate Somewhat strong Strong (heavy) - Very strong - - Extremely strong (almost max) Maximal Figure 2.1: Borg?s new category-ratio (CR-10) scale. Reprinted from Borg 1982. 17 Chapter 3 The relationship between perceived discomfort of static posture holding and posture holding time 3.1 Introduction Static posture holding (SPH) tasks, which require workers to maintain xed working postures for certain time durations, are common across occupations and can be found in most industrial sectors, including the manufacturing, agriculture and service industries (Miedema et al. 1997, Olendorf and Drury 2001, Chung et al. 2003a, Chung et al. 2005). Time durations of posture maintenance, often referred to as posture holding times (PHTs), vary greatly among SPH tasks. Some SPH tasks may last up to a few tens of minutes. As in the cases of other manual work tasks, SPH tasks can impose large biomechanical and physiological stresses on the musculoskeletal system and cause signi cant discomfort to workers if designed without consideration of human characteristics, capabilities and limita- tions. It is important to control such discomfort via ergonomic design as it can seriously compromise workers productivity and job satisfaction level (Corlett and Bishop 1976). Also, controlling discomfort may contribute towards the reduction of work-related musculoskeletal disorders (WMSDs) risks (Dul et al. 1994). Past research studies suggested association between discomfort and WMSDs on the grounds that they are both related to exposure to biomechanical load on the musculoskeletal system (Milner 1985, Nag 1991, Putz-Anderson 18 and Galinsky 1993, Dul et al. 1994, Miedema et al. 1997, Fathallah et. al., 2004, Mey- ers, et. al., 1997, Kemmlert and Kilbom 1987). Some researchers indeed view work-related discomfort/pain as a precursor of WMSDs (Corlett and Bishop 1976, Kee and Lee 2012). Three task parameters, that is, posture, external loads and PHT, seem to specify the level of discomfort that a worker experiences from conducting an SPH task; of course, the workers personal physical and psychological characteristics also contribute as was demon- strated in Park et al. (2009). Multiple studies investigated how the task parameters a ect discomfort levels of SPH tasks as its understanding is crucial for controlling discomfort through ergonomic design/redesign. Genaidy et al. (1995), Kee and Karwowski (2001), Olendorf and Drury (2001), Chung et al. (2002, 2005) and Park et al. (2009) empirically ex- amined the e ects of posture and external loads. In these studies, human subjects performed xed duration (usually 1minute) SPH trials that vary in posture and external loads, and then, subjectively rated the corresponding levels of perceived discomfort. Psychophysical discomfort scales and rating methods, such as the Borg CR-10 scale and the magnitude estimation method, were employed for the self-assessment of perceived discomfort. The Borg CR-10 scale and the magnitude estimation method yield ratio-scale measurements, and therefore, support a variety of statistical analyses (Kee 2004). Several studies investigated the relationship between PHT and perceived discomfort in SPH (Kirk and Sadoyama 1973, Corlett and Manenica 1980, Manenica 1986, Miedema et al. 1997, Reneman et al. 2001, Chung et al. 2003b and Kee 2004). These studies adopted an approach of empirically examining time changes of perceived discomfort during prolonged (often maximum duration) SPH trials - in these studies, subjects reported perceived dis- comfort ratings at predetermined time intervals while performing prolonged SPH trials. The 19 resulting discomfort-time sequences were analysed to understand the discomfort-PHT rela- tionship (hereafter, simply "the discomfort-time relationship" for the sake of brevity). The existing studies on the discomfort-PHT relationship generally suggest that: for static maintenance of any working posture, 1) an increase in PHT results in increased per- ceived discomfort and 2) the discomfort-PHT relationship may be described using a simple mathematical function form with a small set of parameters. However, the existing studies do not seem to have fully elucidated mathematical characteristics of the discomfort-PHT relationship. First of all, they are not in agreement as to the choice of an adequate math- ematical function form for representing the relationship. Multiple studies suggested that a simple linear time function form, that is, Discomfort = C1PHT + C2, can represent the relationship (Corlett and Manenica 1980, Manenica 1986, Dul et al. 1994, Meijst et al. 1995, Kee 2004). On the other hand, Reneman et al. (2001) reported that a negatively accelerat- ing logarithmic time function form expressed as Discomfort = C1lnPHT + C2 depict the relationship. Aside from the lack of agreement, the existing studies also have a limitation that in characterizing the discomfort-PHT relationship, they aggregated discomfort-time sequences obtained from multiple SPH task trials and performed statistical analyses on the combined time-series dataset or the mean response. Therefore, the relationships identi ed do not nec- essarily represent what is actually observed in each individual discomfort-time sequences but rather some average trend at group level. It appears that very few studies have attempted to characterize mathematical characteristics of the discomfort-PHT relationship by examining individual discomfort-time sequences of SPH task trials the authors are not aware of any such studies. Consequently, the discomfort-PHT relationship at the level of actual individual 20 observations is not well understood. Related to this lack of understanding, some research questions arise: What time increase patterns do individual discomfort-time sequences of prolonged SPH task trials exhibit? Does a single pattern (e.g., linear increase over time) universally describe various discomfort-time sequences or do multiple distinct patterns coexist? Is there a simple mathematical function form with a small set of parameters that can robustly represent various individual discomfort-time pro les of SPH task trials? Addressing the above research questions will enhance our understanding of the discomfort- PHT relationship in SPH, and thereby, may further contribute to the ergonomic design of SPH tasks. Especially, identifying an adequate parametric function form has practical ap- plications: it provides a basis for converting discrete discomfort-time sequences of prolonged SPH trials into continuous time functions. Also, it allows representing discomfort-time pro- les parsimoniously using a small set of parameters. These applications in turn will facilitate various quantitative analyses and modelling based on discomfort-time sequences data em- pirically obtained from SPH trials. Therefore, the objective of this study was to empirically address the two research questions posed above. Fifteen males and fteen females participated in the experiment. Each participant performed maximum duration SPH task trials employing 12 di erent working postures. A discomfort-time sequence describing time change of perceived discomfort was obtained from each SPH task trial. The discomfort-time sequences were visually examined to characterize the discomfort-PHT relationship at the individual task trial level. Also, curve- tting analyses were conducted on each discomfort-time sequence using three simple mathematical function 21 forms: the linear(Discomfort = C1PHT +C2), logarithmic(Discomfort = C1lnPHT +C2) and power function forms (Discomfort = C1PHTC2 + C3). This was for determining an adequate mathematical function form for representing individual discomfort-time pro les of SPH task trials. The linear and logarithmic function forms have been suggested as possible models of the discomfort-PHT relationship (Corlett and Manenica 1980, Manenica 1986, Dul et al. 1994, Meijst et al. 1995, Reneman et al. 2001, Kee 2004). The power function form was considered as it is capable of representing three distinct monotonically increasing time patterns, that is, linear, negatively accelerating and positively accelerating time increase patterns, depending on the choice of the exponent parameter value. 3.2 Methods 3.2.1 Participants Fifteen males and fteen females participated in this study as paid volunteers. The age, height, body mass and body mass index (BMI) data of the two participant groups are sum- marised in Table 3.1. The participants were free of obvious neurological and musculoskeletal disorders. Prior to participation, the participants had an introductory session during which the nature and protocol of the study were explained and all questions were answered. The participants also familiarized themselves with the use of the Borg CR-10 scale during the in- troductory session. Each participant signed a written informed consent before participation. The Auburn University Institutional Review Board approved the experimental protocol. 22 Table 3.1: Summary of the age, height, body mass and the BMI data for the two participant groups. Mean and Standard Dev. Dimensions Male Female Age(years) 28.80 8.46 29.40 6.60 Height(years) 180.89 7.13 162.76 5.37 Body Mass(kg) 86.20 11.70 63.26 9.30 BMI(kg/m2) 26.39 3.48 23.86 3.26 3.2.2 Experiment The participants performed maximum duration SPH trials employing 12 di erent static postures. Each trial involved statically holding a box until the point of maximal discomfort, which corresponds to the rating of 10 on the Borg CR-10 scale. The "box" was a generic representation of hand-held loads handled during manual work tasks. It weighed 2 kg, had dimensions of 180mm 180mm 200mm and had no handles. The 12 postures are described in Figure 3.1.; they were selected to represent various possible ways of static box holding and include both standing and seated postures. An analysis based on the Ovako Working Posture Analysing System (OWAS) (Karhu et al., 1977) indicated that they cover all four postural stress levels of the OWAS: Action Codes 1 4. Each participant conducted a single trial for each of the 12 postures. Thus, a total of 360 trials were performed in this study: 360 trials = 12postures 30 participants. Each SPH trial was preceded by a preparatory "posture marking" session - in the posture marking session, the participant posed to correctly attain the posture speci ed for the ensuing SPH trial. Then, while the participant was in the posture, the experimenter adjusted two xtures (moveable vertical posts with adjustable arms) to mark the positions of the following 23 Posture Number Posture Description Posture Number Posture Description 1 Straight back. Both arms above the shoulders. Elbow joints fully extended. Standing on both feet with the legs straight. 7 Back bent forward by approximately 25?, no twisting. Both arms out and below the shoulders. Elbow joints fully extended. Standing on both legs with the knees flexed. The included angle between the upper and lower legs is approximately 150?. 2 Back twisted by approximately 25?, no bending. Both arms out and below the shoulders. Elbow joints fully extended. Standing on both feet with the legs straight. 8 Back twisted by approximately 25?, no bending. Both arms out and below the shoulders. Elbow joints fully extended. Standing on both legs with the knees flexed. The included angle between the upper and lower legs is approximately 150?. 3 Straight back. Both arms out and below the shoulders. Elbow joints fully extended. Sitting down on a stool with the legs hanging free. 9 Back bent forward by approximately 25?, no twisting. Both arms above the shoulders. Elbow joints fully extended. Sitting down on a stool with the legs hanging free. 4 Back bent forward by approximately 25?, no twisting. Both arms out and below the shoulders. Elbow joints fully extended. Standing on both feet with the legs straight. 10 Back twisted by approximately 25?, no bending. Both arms above the shoulders. Elbow joints fully extended. Standing on both legs with the knees flexed. The included angle between the upper and lower legs is approximately 150?. 5 Back bent forward by approximately 25? and twisted by 25?. Both arms above the shoulders. Elbow joints fully extended. Standing on both feet with the legs straight. 11 Back bent forward by approximately 25? and twisted by 25?. Both arms above the shoulders. Elbow joints fully extended. Standing on both legs with the knees flexed. The included angle between the upper and lower legs is approximately 150?. 6 Back bent forward by approximately 25?, no twisting. Both arms out and below the shoulders. Elbow joints fully extended. Sitting down on a stool with the legs hanging free. No backrest is provided. 12 Back bent forward by approximately 25? and twisted by 25?. Both arms above the shoulders. Elbow joints fully extended. Sitting down on a stool with the legs hanging free. No backrest is provided. Figure 3.1: The 12 box holding postures body landmarks: the right shoulder acromion process, the centre of the back of the left hand and the right kneecap (patella) centre (Figure3.2). The adjustable arms of the vertical posts slide vertically along the post, rotate around the post and slide back and forth along their long axes; thus, their tips are able to mark di erent positions in the 3-D space. In addition to the three body landmarks, the feet positions on the oor were also marked, with chalk (Figure 3.2). The markings were for assisting the participant to correctly adopt and maintain the speci ed box holding posture during the ensuing SPH trial. At the completion of the posture marking, the participant was given a rest period so that he/she could fully recover from possible fatigue due to the posture marking. 24 Fixtures with adjustable arms Marking of feet positions Adjustable arms Spherical reflective marker Figure 3.2: The experimental set-up After the posture marking session, the SPH trial proceeded as follows: the participant quickly attained the speci ed posture using the previously obtained posture markings as vi- sual references. Then, the experimenter brought the box to the participant. The participant grabbed the box without actually bearing the box weight while the experimenter holding the box. A camcorder started audio and video-recording the participant. At a beeping sound signalling the onset of the SPH task trial (time zero), the experimenter let go of the box and the participant started bearing the load. The participant maintained the posture until the point of maximum discomfort. While maintaining the posture, the participant shouted the numbers in the Borg CR-10 scale, that is, 0, 0.5, 1, 2, , 10, at the time instants when he/she started perceiving the corresponding discomfort levels. In other words, the time instants 25 were self-determined. This is di erent from the method of verbally reporting discomfort levels at predetermined time intervals, which was employed by some previous studies, such as Reneman et al. (2001), Chung et al. (2003b) and Kee (2004). Despite the di erence, how- ever, both methods are predicated upon the assumption that humans can accurately report discomfort level perceived at a certain time point during prolonged exertion. Putz-Anderson and Galinksy (1993) employed a method similar to ours in which subjects terminated a sim- ulated work task when the perceived discomfort reached a pre-speci ed level on the Borg CR-10 scale. During each SPH trial, the participant was shielded from time cues. A large print of the Borg CR-10 scale was placed to be seen by the participant; it intended to help the participant self-select the time instants of the discomfort levels . In each SPH trial, the rst number shouted could be any of the eleven numbers (0, 0.5, 1, 2, , 10) on the Borg CR-10 scale; after that, the number increased monotonically. Again, the participant nished the SPH trial at the point of maximum discomfort (the rating of 10 on the Borg CR-10 scale). In each SPH trial, the occurrence times of di erent discomfort levels were determined from the camcorder recordings; the sequence of the discomfort rating-occurrence time pairs represented the discomfort-time pro le of the trial. During each SPH trial, the tips of the adjustable arms helped the participant maintain the correct posture. When any of the body landmarks moved away from its correct position indicated by an adjustable arms tip, the experimenter instructed the participant to rectify the posture. Each participant was allowed to have a su cient rest time between consecutive SPH trials to eliminate any cumulative fatigue e ects; the next trial was not started until the 26 participant was fully recovered from the preceding trial. The order of the presentation of the 12 postures was randomised for each participant. Each participant performed the 12 SPH task trials over 4 sessions of three postures. 3.2.3 Data Analyses The discomfort-time sequences obtained from the 360 SPH trials were individually dis- played in lined scatter plots. Then, each lined scatter plot was visually examined so as to characterize the discomfort-time relationship at individual SPH trial level. Curve- tting analyses were conducted to further ascertain the nature of the discomfort-time relationship. Three simple mathematical function forms were tted to each discomfort-time sequence: the linear (Discomfort = C1PHT + C2), logarithmic (Discomfort = C1lnPHT + C2) and power (Discomfort = C1PHTC2 + C3) function forms. Note that all three function forms have an additive constant term. The least square approach was adopted for the curve tting analyses. Microsoft Excels built-in curve- tting functions and Solver were utilized. Two error measures, time-averaged absolute deviation (TAD) and maximum absolute deviation (MAD), were employed to quantify the tting performance of the three function forms. For a given discomfort-time sequence, (ti, Di ),i = 1;:::;N; where i is the index of numbers shouted by the performer, N is the total number of numbers shouted, and ti and Di represent the time and discomfort level (on the Borg CR-10 scale) of the ith number shouted, respectively, the TAD value of a best t function f(t) is computed as follows: TAD = PN i=1j(Di fti)j N (3.1) 27 The MAD value is computed by: MAD = max(jD1 f(t1)j;:::;jDN f(tN)j) (3.2) A two-way analysis of variance (ANOVA) was conducted to test the e ects of function form, posture and function formposture interaction on each error measure. The function form factor had three levels: logarithmic, linear and power. The posture factor had 12 levels as described in Table 2. The main "null" hypothesis was that the three function forms do not signi cantly di er from one another in the mean values of TAD and MAD, irrespective of posture. Post-hoc Tukey analyses were performed to compare the tting performance of the three function forms for each of the 12 postures. The -level was set to be 0.05 for all the statistical analyses. The Minitab software program was used to perform all the statistical analyses (MinitabR 16 Statistical Software). In addition to comparing the tting performance of the three function forms via ANOVAs and post-hoc analyses, each function form was evaluated on an absolute basis in terms of the proportion of the discomfort-time sequences that it can adequately t. The adequateness criterion employed was as follows: a function form is considered to adequately represent a discomfort-time sequence if TAD<0.5 and MAD<1; otherwise, it is considered inadequate. Although rather arbitrary, this criterion of TAD<0.5 and MAD<1 is thought to be a strict one considering the 0-10 range of the Borg CR-10 scale. 28 3.3 Results Visual examinations of the lined scatter plots individually displaying the 360 discomfort- time sequences indicated that three distinct time increase patterns describe most of the discomfort-time sequences. The three distinct patterns were: the negatively accelerated, lin- ear and positively accelerated time patterns. Figures 3.3a-c provide some example discomfort- time sequences for each of the three time increase patterns. (a) (b) (c) Figure 3.3: Some discomfort-time sequences exhibiting each of the three distinct time in- crease patterns: (a) linear, (b) negatively accelerating and (c) positively accelerating. For each time increase pattern, ve randomly selected discomfort-time sequences are presented 29 The ANOVAs results are summarized in Table 3.2. The function form, posture and function form X posture interaction e ects were found to be signi cant for TAD. For MAD, the function form and posture e ects were found to be signi cant; the function form X posture interaction e ect was approaching statistical signi cance (p=0.066). The signi cance level was set at = 0.5 . Table 3.2: The ANOVA results for the Time-Averaged Deviations(TAD) and Maximum Absolute Deviation(MAD) TAD MAD DF SS P DF SS P Function form 2.00 134.56 0 2.00 544.7 0 Posture 11.00 10.46 0 11.00 12.00 0.001 Function form x Posture 22.00 13.86 0 22.00 8.20 0.066 Error 1044 71.00 0 1044 259 0 Total 1079 230.31 0 1079 824 0 The multiple bar graph in Figure 3.4a provide the mean TAD values of the three function forms for each of the twelve postures. Post-hoc Tukey tests indicated that for each posture, the mean TAD value of the power function form was signi cantly smaller than those of the linear and logarithmic functions forms. A similar multiple bar graph for MAD is provided in Figure 3.4b. Again, for each posture, the mean MAD value of the power function form was found to be signi cantly smaller than those of the others. The proportions of the discomfort-time sequences that the three functions form ad- equately represent (TAD<0.5 and MAD<1) were as follows: 78.8% (284 out of 360) for linear, 5.8% (21 out of 360) for logarithmic and 96.4% (347 out of 360) for power.Figure 3.5 provides a graphical summary of the tting performance. Additionally, Figure 3.6a-3.6c provide some examples of discomfort-time sequences adequately represented by the three function forms. 30 (a) (b) Figure 3.4: The mean TAD and MAD values of the three function forms computed for each of the twelve postures: (a) TAD and (b) MAD 0 10 20 30 40 50 60 70 80 90 100 Power Linear Logarithmic Pe rce ntage of SP H trial s fitt ed Function Form Figure 3.5: Summary of the tting performance measure for the three function forms. 31 (a) (b) (c) CR-10 = 0.1897*T + 1.2147 R? = 0.9992 01 23 45 67 89 10 0 10 20 30 40 50Disc om fort ratin g (B org CR -10) Time (seconds) Subject 5, Posture 12 TAD = 0.058 MAD = 0.129 CR-10 = 2.6323ln(T) - 3.1977 R? = 0.9759 0 1 2 3 4 5 6 7 8 9 10 0 30 60 90 120 150Dis com fort rat ing (Bo rg CR -10) Time (seconds) Subject 1, Posture 7 TAD = 0.355 MAD = 0.696 CR-10 = 0.0008*T2.2749 R? = 0.9916 01 23 45 67 89 10 0 10 20 30 40 50 60 70Disc om fort ratin g (B org CR -10) Time (seconds) Subject 9, Posture 11 TAD = 0.133 MAD = 0.272 Figure 3.6: Examples of adequate functional representation of discomfort-time sequence data (TAD<0.5 and MAD<1): (a) linear, (b) logarithmic and (c) power. For the 347 discomfort-time sequences adequately represented as power functions, the distributions of the three parameters of the power function form, that is, C1, C2, and C3 in (Discomfort = C1PHTC2 +C3), are graphically illustrated in Figures 3.6a-3.6c. 32 6 . 0 05 . 0 04 . 0 03 . 0 02 . 0 01 . 0 00 . 0 5 3 0 0 2 5 0 2 0 0 1 5 0 1 0 0 5 0 0 T h e s c a l i n g f a c t o r ( C 1 ) F r e q u e n c y 3 . 0 02 . 5 02 . 0 01 . 5 01 . 0 00 . 5 00 . 0 1 5 0 4 0 3 0 2 0 1 0 0 T h e e x p o n e n t ( C 2 ) F r e q u e n c y 6 . 0 04 . 0 02 . 0 00 . 0 5 1 2 5 1 0 0 7 5 5 0 2 5 0 T h e i n t e r c e p t ( C 3 ) F r e q u e n c y Figure 3.7: The distributions of the power function parameters for the 347 discomfort-time sequences represented as power functions with TAD<0.5 and MAD<1: (a) the scaling factor C1, (b) the exponent C2 and (c) the intercept C3 . 33 3.4 Discussion The objective of the current study was to elucidate the nature of the relationship between perceived discomfort of SPH and PHT. To accomplish the research goal, this study examined a total of 360 discomfort-time sequences empirically obtained from maximum duration SPH trials. The focus was speci cally on addressing the following research questions: What time increase patterns do individual discomfort-time sequences exhibit? Does a single pattern, for example, linear increase over time, prevail across di erent discomfort- time sequences or do multiple distinct patterns coexist? Is there a simple mathematical function form that can adequately represent various discomfort-time sequences? In other words, what is the mathematical representation of the relationship between PHT and perceived discomfort? Visual examinations of lined scatter plots that depict the 360 discomfort-time sequences addressed the rst question (Figure 3.3). The discomfort-time sequences did not exhibit a single universal pattern of time increase; but, instead, three di erent patterns appeared to co- exist. They were: the linear, negatively accelerating and positively accelerating time increase patterns (Figure 3.3). The three patterns seemed to characterize most of the observed discomfort-time sequences. The co-existence of multiple distinct time increase patterns was further supported by the ANOVAs (Table 3.2) and the comparisons of the three function forms in the mean TAD and MAD values (Figure 3.4). Across the 12 postures considered in this study, the power function form, which can represent all of the three time increase patterns mentioned above, resulted in signi cantly smaller mean TAD and MAD values 34 than the linear and logarithmic function forms, each of which is capable of representing only one type of time increase pattern. The linear function form represents only the linear time increase pattern; the logarithmic function form, only the negatively accelerated time increase pattern. As for the second research question, the power function form was found to be able to t most of the individual discomfort-time sequences examined in this study with small tting errors - it was able to t 347 out of the 360 (96.4%) discomfort-time sequences satisfying the adequateness criterion of TAD<0.5 and MAD<1. Once again, although rather arbitrary, the criterion of TAD<0.5 and MAD<1 can be considered as a strict one considering the 0-10 range of the Borg CR-10 scale. When evaluated with the same adequateness criterion, the logarithmic function form was found to adequately t only 21 out of 360 (5.8%). The linear function form was found to perform signi cantly better than the logarithmic adequately t- ting 284 out of the 360 (78.8%) discomfort-time sequences; however, the power function form still outperformed the linear by a large margin (17.6%=96.4%-78.8%). Based on these results and also the co-existence of the three distinct time increase patterns visually observed (Figure 3.3), this study reccomends that the power function form, Discomfort = C1PHTC2 + C3, be used to mathematically describe the relationship between perceived discomfort of SPH and PHT. The power function form Discomfort = C1PHTC2 +C3, provides a mathematical basis for converting empirically obatined discomfort-time sequences of SPH trials to continuous time functions. This model realizes parsimony of representation as it uses only three parame- ters. The three parameters quantitatively characterize di erent aspects of a discomfort-time pro le. The parameter C1, is a scaling factor. The exponent C2 determines the appearance 35 of a pro le?s time increase pattern: the negatively accelerated (01) or linear (C2 = 1) time increase patterns. C1 and C2 together determine the growth rate of a discomfort-time pro le. The parameter C3 is the intercept and represents the starting point of perceived discomfort during an SPH trial, that is, the level of discomfort perceived immediately after the onset of an SPH trial. The three parameters can be used in combination to compute other meaningful quantities. For example, the PHT to reach a particular discomfort level D0 can be computed by: PHTD? = ((D0 C3)=C1)1=C2 (3.3) The three parameters and interpretable quantities derived from them would facilitate statistically analysing empirically obtained discomfort-time sequences data, and thereby, sup- port various hypothesis testing and modelling studies. For example, multivariate statistical analyses could be performed on the vector consisting of the three parameters to understand the e ects of certain task-related or personal variables on entire discomfort-time pro le of SPH. Also, the problem of predicting an entire discomfort-time pro le based on a certain set of predictor variables could be formulated as that of predicting the three parameter values. For the 347 discomfort-time sequences represented by the power function form with TAD<0.5 and MAD<1, the distributions of the three parameters C1, C2 and C3 are de- picted in Figures 3.7a-3.7c, respectively. Some observations from Figures 3.7a-c included the following: 1) the scaling factor C1 appeared to follow an exponential distribution, 2) the exponent C2 appeared to be normally distributed; the mean was 1.36 and the range was from 0.14 to 2.94, and 3) the intercept C3 seemed to be exponentially distributed. Concerning 36 the variability observed in the distributions of the three parameters, some questions arise naturally: what gives rise to such variability? Can certain personal or task-related variables account for the variability, and if so, how much of the variability can they account for? Is it possible to predict the parameter values, and therefore, an entire discomfort-time pro le based on a certain set of predictor variables? Currently, an investigation is being carried out in search of the answers to these questions. It is perhaps worth pointing out that the power function form representationDiscomfort = C1PHTC2+C3 bears a resemblance to the Steven?s power law in that it relates the magnitude of a physical stimulus (PHT) to a psychological response (discomfort) using the power rela- tionship - the Steven?s power law describes the relationship between the magnitude of a phys- ical stimulus and its perceived intensity and is mathematically described as (I) = kIa where I denotes the magnitude of a physical stimulus and (I), its perceived intensity (Gescheider, 1985). Despite the similarity, however, it is thought that the power function representation of the discomfort-time relationship cannot be considered as an instance of the Stevens power law. This is because perceived discomfort is not the same as perceived intensity of PHT, that is, perceived time duration. As the relationship Discomfort = C1PHTC2 + C3 indicates, an increase in PHT results in increased discomfort. However, some intermediate variables, biomechanical or physiological, would be required to explain how that occurs. Further in- vestigations are needed to identify the intermediate variables and the mechanism through which increased PHT leads to an increase in perceived discomfort. The two main ndings of this study can be summarized as follows: 1) three di erent monotonically increasing time patterns (negatively accelerated, linear and positively accel- erated) co-exist in discomfort-time pro les of SPH trials and 2) the power function form can 37 serve as a general representation of the discomfort-time relationship. These results are not identical to those of past research studies. Previously, multiple studies suggested that the relationship between perceived discomfort of SPH and PHT is linear in its nature, and thus, be represented using the linear time function form (Corlett and Manenica 1980, Manenica 1986, Dul et al. 1994, Meijst et al. 1995, Kee 2004). Reneman et al. (2001) on the other hand contended that the negatively accelerated logarithmic time function form depict the discomfort-time relationship. Regarding the di erences between the current and previous research results, it should be noted that the current study does not contradict the past re- search studies. This study reports that the positively accelerated time increase pattern is found in some discomfort-time sequences of prolonged SPH trials in addition to the linear and negatively accelerated time increase patterns, which were observed in previous studies. It is thought that this study was able to nd a wider range of time increase patterns than the previous studies because it had an opportunity to examine a larger number of discomfort- time sequences. Also, the fact that multiple previous studies (Corlett and Manenica 1980, Manenica 1986, Dul et al. 1994, Meijst et al. 1995, Kee 2004) suggested the linear time function form representation of the discomfort-time relationship could be understood in light of the current results. As mentioned earlier, the linear function form was found to be able to t 284 out of the 360 (78.8%) discomfort-time sequences with TAD<0.5 and MAD<1. This indicates that the linear function form is indeed useful in representing many discomfort-time pro les. Reneman et al. (2001) only examined two postures in investigating discomfort-time pro les of SPH trials. This may explain why they observed only the negatively accelerated time increase pattern in their study. 38 The current research ndings are expected to contribute to the e orts to control dis- comfort and WMSD risks associated with SPH through ergonomic design interventions. Especially, they may serve a basis for developing new posture analysis tools for analysing stresses of SPH tasks. The existing posture analysis tools, including the Ovaco Working Posture Analysing System (OWAS) (Karhu et al. 1977, Karhu et al. 1981) and the Rapid Upper Limb Assessment (RULA) method (McAtamney and Corlett 1993), do not allow con- sidering PHT in evaluating postural stresses of working postures. The new knowledge on the nature of the discomfort-time relationship in SPH and the power function form representa- tion found in this study may guide the development of future posture analysis tools aimed at estimating the population distribution of discomfort for a given SPH task described in terms of working posture and PHT. Of course, to realize such posture analysis tools, a large database of discomfort-time sequences will also need to be established. The current study provided an initial set of data for building such database. Some limitations of the current study are acknowledged along with future research directions: rst, this study considered only two-handed static box holding tasks and one hand-load weight level (2kg). Although this study examined a large set of discomforttime sequences, collecting and analysing more data in various task conditions would be bene cial especially for con rming the validity of the research ndings over a wider range of conditions and also developing useful posture analysis tools. Related to this, a discomfort-time sequence database for one-handed SPH tasks is currently under development. Second, most of the study participants in this study were young and of the normal weight category. Future studies will need to examine discomfort-time pro les of older and/or obese individuals to understand the e ects of age and obesity level and also con rm the validity of the study 39 ndings within those segments of the population. Third, this study examined only subjective ratings of perceived discomfort without employing other stress/strain measures. Collecting and analysing physical and physiological response data along with perceived discomfort ratings would further enhance our understanding of the discomfort-time relationship in SPH. 40 Chapter 4 An investigation on inter-individual variation in perceived discomfort of static posture holding. 4.1 Introduction Static posture holding (SPH) tasks, which require workers to statically maintain work- ing postures for certain time durations, can be found in most industrial sectors, including the manufacturing, agriculture and service industries (Miedema et al. 1997, Olendorf and Drury 2001, Chung et al. 2003a, Chung et al. 2005). SPH tasks can lead to overloads on the musculoskeletal system and cause signi cant discomfort to workers if they are designed without consideration of human capabilities and limitations. Stressful SPH tasks may in- crease the incidences of work-related musculoskeletal disorders (WMSD). The association between perceived discomfort and WMSD has been reported in a number of studies (Corlett and Bishop 1976, Genaidy and Karwowski, 1993, Genaidy et al, 1995, Kee and Lee 2012). Research suggests that controlling work-related discomfort through ergonomics interventions may contribute towards the reduction of WMSD risks (Dul et al. 1994, Putz-Anderson and Galinsky, 1993, Miedema et al., 1997). It would also improve the quality of work life for many workers. In general, an SPH task can be speci ed by three parameters, that is, posture, exter- nal load and task duration. These parameters a ect the level of discomfort that a worker 41 experiences from conducting the task. In addition, the workers physical and psychological characteristics also contribute as was demonstrated in Park et al. (2009). Thus, even when performing identical SPH tasks, that is, identical in terms of the three parameters, di erent individuals with di erent physical and psychological characteristics would experience di er- ent levels of perceived discomfort. Therefore, for a given SPH task, a group or a population of individuals gives rise to a probability distribution of perceived discomfort, which describes the inter-individual variation in discomfort perception. Such mathematical depiction of the inter-individual variation can greatly help determine the proportion of the workforce that would experience excessive (or manageable) discomfort from an SPH task and further assist in deciding the acceptability of the task from a population accommodation point of view. In ergonomics and related research areas, numerous studies have examined discomfort resulting from SPH. Many of these studies aimed to understand the e ects of posture and external load on perceived discomfort (Genaidy et al., 1995, Kee and Karwowski, 2001, Olendorf and Drury, 2001, Chung et al., 2005, Drury et al. 2006, and Park et al., 2009, Kee and Lee, 2010). In these studies, the participants were asked to perform SPH task trials for short time duration, and then, subjectively rate the corresponding levels of perceived discomfort. The SPH trials varied in posture and/or external load. In many of these studies, it was reported that certain body part postures, such as non-straight back, elevated arms and bent legs, increase postural stresses from SPH (Park et al. 2009, Olendorf and Drury, 2001, Genaidy et al., 1995 and Chung et al. 2002). Also, external loadings were identi ed as the major driving factor of bodily discomfort (Boussenna et al. 1982, Manenica 1986, Reneman et al. 2001, Dickerson et al. 2006, Dickerson et al. 2007, Kee 2004, and Kee and Lee, 2010). Some studies investigated the e ect of task duration (posture holding time) on discomfort 42 of SPH (Corlett and Manenica 1980, Boussenna et. al 1982, Meijst et. al. 1995, Miedema et. al. 1996, Reneman et. al. 2001, Chung et al. 2003, Kee and Lee 2010). In these studies, for a set of working postures, human subjects were asked to perform SPH as long as they could, and while holding each posture, rate their perceived discomfort at predetermined time intervals. The trials were terminated when the maximum discomfort level was reached. The results of these studies indicated that posture signi cantly a ects perceived discomfort along with posture holding time and external load. Also, these studies reported that perceived discomfort monotonically increases over time during a prolonged SPH task trial (Reneman et al. 2001, Chung et al. 2003, Corlett et al. 1980). Reneman et al. (2001) proposed using a negatively accelerated, modi ed logarithmic function to describe the discomfort-time relationship. Other studies have suggested using a linear function to describe the discomfort- time relationship (Corlett and Manenica 1980, Meijst et. al. 1995, Kee 2004). The previous studies on perceived discomfort of SPH typically employed psychophysical discomfort scales and rating methods, such as the Borg CR-10 scale (Borg, 1982) and the magnitude estimation method (Gescheider, 1985) for human subjects self-assessment of perceived discomfort. The previous research e orts described above have greatly advanced our understanding of perceived discomfort from SPH. Nonetheless, however, some signi cant knowledge gaps seem to exist, which hamper adequately representing and evaluating discomfort associated with SPH. The past research studies mostly focused on testing and characterizing the e ects of SPH task parameters on perceived discomfort. In doing so, they investigated if the task parameters signi cantly a ect the mean of perceived discomfort computed across di erent individuals. Statistical analysis methods, such as the analysis of variance and/or linear re- gression techniques, were employed in accordance with this approach. Whilst this approach 43 is suitable for understanding the e ects of SPH task parameters, it does not shed light upon discomfort responses of individuals and the variation in them. The authors are unaware of any research studies that seem to have examined SPH-associated discomfort with a focus on the inter-individual variation. As a consequence, little is known about the nature and mag- nitude of such variation. This lack of knowledge is problematic as it hinders considering SPH task design and evaluation from the population accommodation point of view. Population accommodation has been one of the fundamental principles of ergonomics design. As an initial e ort towards alleviating the current dearth of research mentioned above, this study aimed to empirically address the following questions: What probability distributions are suitable for modeling the inter-individual variation in perceived discomfort from SPH? Can a well-known, single family of probability distributions, such as the normal family, be utilized across di erent SPH tasks, or do di erent SPH tasks require di erent types of probability distributions? When quanti ed in terms of the common measures of dispersion, such as standard deviation and range, how large is the inter-individual variability in SPH-associated perceived discomfort? Do di erent SPH tasks vary signi cantly in the amount of inter-individual variability? If yes, do the SPH task parameters, such as posture and external load, a ect the inter-individual variability? What are the characteristics of SPH tasks that exhibit large (or small) inter-individual variability? A large set of perceived discomfort ratings data obtained from an SPH experiment was utilized to accomplish the research goal. In the experiment, 10 male and 10 female participants conducted SPH for a set of 180 predetermined SPH tasks. The 180 SPH tasks 44 were combinations of 60 postures and 3 hand load levels. The posture holding time was xed at 20 seconds for all of the SPH tasks. For each SPH task, the participants performed subjective rating of perceived discomfort using the Borg CR-10 discomfort scale. For each SPH task, the perceived discomfort values obtained from the 20 participants were used to ascertain the nature of the underlying probability distribution. Also, the measures of spread as well as centre were computed from them. Statistical analyses were conducted to address the research questions posed. 4.2 Methods 4.2.1 Participants Ten males and ten females participated in this study as paid volunteers. All participants were free of obvious neurological and musculoskeletal disorders. Prior to participation, the participants had a training session during which the nature and protocol of the study were explained and all questions were answered. During the training session, the participants also familiarized themselves with the use of the Borg CR-10 scale. Each participant signed a written informed consent before participation. The Auburn University Institutional Review Board approved the experimental protocol. Table 4.1 summarizes the age, height, body mass and body mass index (BMI) data of the study participants for each gender. 45 Table 4.1: Summary of the age, height, body mass and body mass index (BMI) data for the male and female participant groups. Male Female Dimensions Mean (SD) Mean (SD) Age (Years) 27.00 (4.75) 28.00 (4.32) Height (cm) 181.86 (7.61) 165.25 (6.64) Body Mass (kg) 86.10 (15.89) 68.14 (12.58) BMI (kg/m2) 25.90 (3.93) 25.02 (4.89) 4.2.2 Experiment The participants performed short-period (20 seconds) one-handed static object holding tasks for 180 posture-hand load conditions: 180 posture-hand load conditions = 60 postures x 3 hand load levels. The 60 postures represent various one-handed static object holding postures that can be found in workplaces and were generated by combining four back, three arm and ve lower body postures. Figure 4.1 describes the body part postures used to generate the 60 whole-body postures. Leg posture 5 was speci cally included in this study due to its prevalence in many postural work tasks including, assembly line jobs, construction and painting. Three hand load conditions were utilized: 0 kg (no hand load), 2.2 kg and 3.6 kg (cylindrical objects were used as the hand loads). For each of the 180 posture-hand load combinations, each participant performed a single object holding trial the order of presentation of the 180 trials was randomized for each subject. Therefore, a total of 3600 object holding trials were performed: 3600 trials = 180 posture-hand load combinations x 20 participants. 46 Part Body posture categories Arm Arm Posture 1 Arm below shoulder level Arm Posture 2 Arm at 90 degrees Arm Posture 3 Arm elevated above shoulder Back Back Posture 1 Straight Back Posture 2 Bent approximately 250 Back Posture 3 Twisted approximately 250 Back Posture 4 Bent approximately 250 & twisted approximately 250 Leg Leg Posture 1 Both legs straight Leg Posture 2 Both legs bent approximately 1500 Leg Posture 3 Both legs bent approximately 1500, body moved forward by both limbs Leg Posture 4 Straight, body moved forward by one limb Leg Posture 5 One leg bent at approximately 1500, on a raised support Figure 4.1: Body part postures used to generate the 180 whole-body posture categories used in the current study. 47 During each experimental trial, human gure illustration depicting the posture for the given trial was shown to the participant to help him/her adopt the posture. Once the posture was adopted, the object was brought to the participants dominant hand by the experimenter and the participant commenced the 20-seconds one-handed static holding task. Immediately after the completion of each experimental trial, the participant reported his/her perceived postural stress using the Borg CR-10 scale (Figure 4.2). A large print of the Borg CR-10 was located in the line of sight of the participant to help him/her self-select his/her discomfort rating. 0 - 0.5 - 1 - 2 - 3 - 4 - 5 - 6 - 7 - 8 - 9 - 10 - . Nothing at all Extremely weak (just noticeable) Very weak Weak (light) Moderate Somewhat strong Strong (heavy) - Very strong - - Extremely strong (almost max) Maximal Figure 4.2: Borg?s new category-ratio (CR-10) scale. Reprinted from Borg 1982. Prior to a box holding trial, a participant changed into a tight out t (tight short and short sleeves). Also, spherical re ective markers were attached on body landmarks so that the body postures can be accurately recorded by the VICON motion capture system during the experiment trial. After this initial preparatory set up, the participant adopted the 48 posture and performed the one-handed static object holding task the object was be brought to the participants dominant hand by the experimenter immediately after the participant adopted the posture. In each trial the VICON system kept track of the positions of the attached re ective markers over time and display the distances between their current and correct positions in real-time. When a body attached marker is farther away from its correct position the experimenter instructed the participant to correct the posture. For each SPH task, Subjects were asked to rate their level of discomfort at the completion of each 20- seconds long SPH task trial. The 180 SPH trials were performed over three sessions by each participant. The three sessions were conducted on three separate days to prevent fatigue e ects. The participants were given su cient rest period between consecutive trials. Even when the participated reported no fatigue, at least 1 minute rest period was provided before the next trial. Also, and after 9 experimental trials, the participant was given an extra 5 minute rest period before he/she could start the next trial. The subjects were also allowed to have additional rest periods as they wanted. Similar rest patterns were reported by Olendorf and Drury (2001) who also had the participants in their study perform a 20 s long SPH task. In their study, participants were allowed 45 s breaks between consecutive trials. 4.3 Data analyses This study aimed to address two sets of research questions. The rst set pertained to identifying mathematical probability distributions suitable for modelling the inter-individual variation in perceived discomfort of SPH. Of particular interest was to determine if a well- known, single family of probability distributions, such as the normal family, be utilized across 49 di erent SPH tasks or di erent SPH tasks require di erent types of probability distributions. The second set of research questions concerned the magnitude of the inter-individual vari- ability in perceived discomfort of SPH, and especially, whether or not it is a ected by SPH task parameters. To address the rst set of research questions, for each of the 180 SPH tasks that vary in posture and hand load, a probability histogram was generated using the correspond- ing perceived discomfort dataset. Then, each probability histogram was visually examined. In addition, Minitabs Individual Distribution Identi cation Tool (MinitabR 16 Statistical Software) was used to identify a mathematical probability distribution that adequately ts each dataset. The Individual Distribution Identi cation Tool allows for evaluating up to 14 di erent distribution families: the normal, lognormal, 3-parameter lognormal, exponential, 2-parameter exponential, Weibull, 3-parameter Weibull, largest extreme value, smallest ex- treme value, gamma, 3-parameter gamma, logistic, loglogistic and 3-parameter loglogistic types. These distribution families are commonly used in probability modeling and statistical analyses and cover a wide range of distribution shapes. For each dataset, the Individual Distribution Identi cation Tool performed the goodness of t test using each of the 14 dis- tribution types. In addition, it identi ed the distribution type that best ts each dataset. The best- tting distributions found for the 180 SPH tasks were examined to determine if a single, well-known distribution type can model the inter-individual variation in perceived discomfort across various SPH tasks. To address the second set of research questions, two dispersion measures, that is, the standard deviation and range, were computed for each of the 180 SPH task using the cor- responding dataset. Then, frequency histograms were plotted for each dispersion measure. 50 To test the e ect of SPH task parameters (body part posture and hand load) on each dis- persion measure, four-factor ANOVAs were conducted. The analyses evaluated the e ects of hand load level, arm posture, back posture, leg posture and their interactions. The hand load factor had three levels. The arm, back and leg postures had three, four and ve levels respectively. In performing the ANOVAs, the highest order interaction (the four-way hand load level x arm posture x back posture x leg posture interaction) was used as the error term since only one experimental unit was available for each combination of the factor levels. An- derson and McLean (1974) supported using the higher order interactions as the error term in experiments with only one experimental unit per factor level combination. In addition to the two dispersion measures mentioned above, two measures of centre, that is, the mean and median, were computed for each SPH task using the corresponding dataset. A frequency histogram was plotted for each centre measure. Scatter plots were used to examine possible relationships between the measures of centre and those of dispersion. The Minitab software program was used to perform the statistical analyses (MinitabR 16 Statistical Software). The -level for the statistical analyses was set at 0.05. All tests of signi cance were performed at = 0.05. Graphical outputs were provided by MS excel (Microsoft 2010). 51 4.4 Results 4.4.1 Probability distributions for modeling the inter-individual variation in perceived discomfort The results for distribution identi cation for each of the 180 SPH tasks are summarized in Figure 4.3. In one-third of all the 180 datasets none of the 14 distributions in Minitab could adequately t the dataset (p<0.05). The results of the distribution identi cation showed that there is no single well-known distribution family that can be used to robustly t di erent SPH tasks. 0 10 20 30 40 50 60 70 80 No Fit 2-P ara me ter E xpo nen tial 3-P ara me ter W eib ull Weib ull Larg est E xtrem e V alu e No rm al Ex pon ent ial Lo gno rm al Fre qu ency Fitted Distribution Figure 4.3: Histogram showing the tted distributions for the 180 SPH tasks. A total of 14 distributions available in Minitab were used to t the data. A few example datasets were selected and their frequency histograms were generated as shown in Figures 4.4 (a) to (f) and Figures. 4.5 (g) to (j). These histograms clearly indicate the lack of a common distribution or family of distribution to characterize the SPH tasks. 52 (a) Posture 3 (b) Posture 107 (c.) Posture 17 (d) Posture 150 (e) Posture 2 (f) Posture 47 00.05 0.10.15 0.20.25 0.30.35 0.40.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) Figure 4.4: (a) Both legs straight, Arm below the shoulder level, Straight back and holding a 3.6 kg hand-load. (b) Both legs straight, body moved forward by the limbs, back bent and twisted, arm raised above the shoulder level and holding a 2.2 kg load. (c) Both legs straight, arm above the shoulder level, Back bent by approximately 25 degrees and holding a 2.2 kg load (d) One leg bent at approximately 150 degrees, and on a raised support, Arm at 90 degrees , Straight back and holding a 3.6 kg hand-load.(e) Both legs straight, arm below the shoulder level, Straight back and holding a 2.2 kg hand-load. (f) Both legs bent approximately 150 degrees , arm below the shoulder level, Straight back and holding a 2.2 kg hand-load. . 53 (g) Posture 9 (h) Posture 18 (i.). Posture 6 (j) Posture 81 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5- 0.9 1.0- 1.9 2.0- 2.9 3.0- 3.9 4.0- 4.9 5.0- 5.9 6.0- 6.9 7.0- 7.9 8.0- 8.9 9.0- 9.9 10 Pro bab ility Discomfort rating (Borg CR-10) Figure 4.5: (g) Both legs straight, arm above the shoulder level, Straight back and holding a 3.6 kg load (h) Both legs straight, arm above the shoulder level, Back bent by approximately 25 degrees and holding a 3.6 kg load. (i) Both legs straight, Arm at 90 degrees , Straight back and holding a 3.6 kg hand-load (j) Both legs bent approximately 150 degrees, body moved forward by both limbs, Straight back, arm raised above the shoulder level and holding a 3.6 kg load. 54 4.4.2 Statistical analyses on the magnitude of inter-individual variation in per- ceived discomfort. As described earlier in Section 4.3, for each of the 180 SPH task conditions, two mea- sures of dispersion, that is, sample standard deviation and range, were computed using the corresponding perceived discomfort dataset to quantify the magnitude of the inter-individual variation in perceived discomfort. The frequency histograms shown in Figure 4.6 describe the distributions of the dispersion measures, respectively. Figure 4.6: Histograms showing the distribution of standard deviation and range for 180 SPH tasks. Additionally, frequency histograms were constructed for the measures of center (mean and median), gure 4.7. These histograms show the distribution of mean and median dis- comfort for di erent SPH tasks. 55 Figure 4.7: Histograms showing the distribution of Mean and Median for 180 SPH tasks. The results from the ANOVAs that tested the e ects of body part posture (arm, back, and leg) and hand load level on the two measures of dispersion (SD and range) are provided in Table 4.2, for each dispersion measure. The ANOVA results showed statistical signi cance for some of the main e ects and interaction e ects. The following observations can be made from them: for standard deviation, arm posture, leg posture and load level were found to be signi cant. For range, only two main e ects, leg posture and load level, were signi cant. For both standard deviation and range, the load level main e ect accounted for the largest percentage of the total variation. Of the ten possible interaction e ects in each case, the (back x leg) and (back x leg x load) interaction e ects seemed to be the major contributors to the total variation. 56 Table 4.2: The ANOVA results for the two measures of spread: standard deviation and range St. dev. Range Source DF SS P DF SS P Arm 2 0.61 0.0170* 2 7.02 0.065 Back 3 0.09 0.7 3 1.61 0.725 Leg 4 6.95 0.0000* 4 72.17 0.0000* Load 2 21.24 0.0000* 2 281.05 0.0000* Arm x Back 6 0.21 0.797 6 3.4 0.829 Arm x Leg 8 1.11 0.065 8 21.18 0.046 Arm x Load 4 0.44 0.186 4 9.01 0.133 Back x Leg 12 13.6 0.0000* 12 122.64 0.0000* Back x Load 6 0.36 0.524 6 4.54 0.711 Leg x Load 8 0.8 0.196 8 13.35 0.232 Arm x Back x Leg 24 3.31 0.0200* 24 49.68 0.058 Arm x Back x Load 12 0.26 0.983 12 2.53 0.999 Arm x Leg x Load 16 0.94 0.625 16 22.08 0.351 Back x Leg x Load 24 5.35 0.0000* 24 64.84 0.0009* Error 48 3.3 48 58.26 Total 179 58.59 179 733.36 Additional ANOVA analyses were conducted for the measures of center (mean and median). In addition to the main e ects of arm, leg and load being signi cant, the 2-way back- leg and leg-load interactions were signi cant. Two 3-way interactions also found to be signi cant: Arm x Back x Leg and Back x Leg x Load. The results are summarized in table 4.3. 57 Table 4.3: The ANOVA results for the two measures of center: mean and median. Mean Median Source DF SS P DF SS P Arm 2 5.59 0.000* 2 6.43 0.000* Back 3 0.59 0.243 3 0.88 0.171 Leg 4 24.96 0.000* 4 18.06 0.000* Load 2 124.65 0.000* 2 120.86 0.000* Arm x Back 6 0.61 0.627 6 0.99 0.453 Arm x Leg 8 2.375 0.048 8 1.47 0.388 Arm x Load 4 0.96 0.158 4 1.75 0.049 Back x Leg 12 68.45 0.000* 12 45.35 0.000* Back x Load 6 0.22 0.949 6 0.88 0.529 Leg x Load 8 3.74 0.004* 8 2.58 0.081 Arm x Back x Leg 24 17.28 0.000* 24 16.87 0.000* Arm x Back x Load 12 0.57 0.976 12 1.24 0.822 Arm x Leg x Load 16 1.33 0.865 16 3.47 0.246 Back x Leg x Load 24 29.82 0.000* 24 19.3 0.000* Error 48 6.62 48 8.13 Total 179 287.77 179 248.28 The main aim of this study was to show how the SPH task parameters a ect the amount of inter-individual variability in perceived discomfort. The ANOVA results showed signi cant three-way interactions for some of the measures of spread and center. For SD, the 3-way interactions between the arm, back, leg and Back, leg, and load were found to be signi cant. Only the 3-way interaction between the back, leg and load was found to be signi cant for the range. The interaction plots for SD are shown in gures 4.8 and 4.9 Figure 4.10 shows the interaction plot for range. Visual observation of Figures 4.8, 4.9 and 4.10 reveal the complex nature of these higher order interactions. 58 2 . 0 1 . 5 1 . 0 321 54321 2 . 0 1 . 5 1 . 0 4321 2 . 0 1 . 5 1 . 0 B a c k L e g L o a d 1 2 3 4 B a c k 1 2 3 4 5 L e g 1 2 3 L o a d I n t e r a c t i o n P l o t f o r S D D a t a M e a n s Figure 4.8: Three-way interaction plot for the back, leg and load interaction. 2 . 0 1 . 5 1 . 0 54321 4321 2 . 0 1 . 5 1 . 0 321 2 . 0 1 . 5 1 . 0 A r m B a c k L e g 1 2 3 A r m 1 2 3 4 B a c k 1 2 3 4 5 L e g I n t e r a c t i o n P l o t f o r S D D a t a M e a n s Figure 4.9: Three-way interaction plot for the arm, back and leg interaction. 59 8 6 4 321 54321 8 6 4 4321 8 6 4 B a c k L e g L o a d 1 2 3 4 B a c k 1 2 3 4 5 L e g 1 2 3 L o a d I n t e r a c t i o n P l o t f o r R a n g e D a t a M e a n s Figure 4.10: Three-way interaction plot for the back, leg and load interaction for the range. The three-way interactions were very complex (Tables 4.8 - 4.10); nonetheless, an attempt was made to nd some interesting patterns from the tables. The highest inter- individual variations were found in load level three (holding a 3.6 kg load). The load level thus a ected variability signi cantly. It is apparent that leg posture 1(straight legs) ex- hibit the smallest variability, while leg posture 5 (right foot in a exed position, on a raised support) exhibit the largest variability. In most cases, although not all, the interaction of non-straight back postures and non-straight leg postures were associated with higher inter- individual variability. 60 0.7 ? 1.0 1.1 ? 1.4 1.5 ? 1.8 1.9 ? 2.2 2.3 ? 2.6 2.7 ? 3.0 Lowest Low Medium High Highest SD Le gs 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 1 0.7 1.2 1.5 0.7 1.0 1.7 0.9 1.3 1.5 1.0 1.8 2.2 2 1.1 2.2 2.7 0.9 1.3 1.6 1.2 1.5 1.7 1.4 1.4 1.7 3 1.1 2.0 2.2 1.3 2.5 2.7 1.1 1.2 1.7 1.1 1.4 1.9 4 1.2 1.5 1.4 1.3 2.1 2.4 1.3 2.2 2.7 1.3 1.6 1.5 5 1.3 1.8 1.7 1.3 1.8 1.8 1.6 2.3 2.4 1.4 2.5 2.6 Figure 4.11: Three-way interaction plot for the back, leg and load interaction for standard deviation. 61 0.9 ? 1.1 1.2 -1.4 1.5 ? 1.7 1.8 ? 2.0 2.1 ? 2.3 2.4 ? 2.6 Lowest Low Medium High Highest SD Leg 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 1 0.9 1.4 1.1 1.0 1.4 1.0 1.3 1.2 1.1 1.2 1.9 1.9 2 2.0 2.1 2.0 1.0 1.6 1.2 1.6 1.5 1.3 1.8 1.4 1.4 3 1.5 1.8 2.0 2.1 2.2 2.2 1.2 1.4 1.5 1.6 1.6 1.1 4 1.2 1.4 1.5 1.7 1.9 2.1 2.0 2.0 2.1 1.4 1.4 1.6 5 1.9 1.8 1.2 1.7 1.6 1.6 1.8 2.4 2.1 2.3 2.1 2.1 Figure 4.12: Three-way interaction plot for the arm, back and leg interaction for standard deviation. 62 2.2 ? 3.3 3.4 ? 4.5 4.6 ? 5.7 5.8 ? 6.9 7.0 ? 8.1 8.2 ? 9.3 Lowest Low Medium High Highest Range Le gs 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 1 2.2 4.0 6.0 2.7 3.0 6.7 2.7 4.3 5.0 3.3 6.0 7.5 2 3.7 7.3 8.8 2.7 4.3 5.3 4.0 4.7 6.3 4.0 4.8 5.8 3 3.7 7.2 7.3 4.3 8.2 8.6 4.0 4.0 6.8 3.7 4.0 7.0 4 4.0 5.0 5.0 4.3 7.2 8.2 4.0 7.0 9.3 4.3 5.5 5.3 5 4.0 6.0 6.0 4.3 6.0 6.8 6.0 8.3 7.7 4.7 8.0 8.0 Figure 4.13: Three-way interaction plot for the back, leg and load interaction for the range. 63 0.4 ? 1.2 1.3 ? 2.1 2.2 ? 3.0 3.1 ? 3.9 4.0 ? 4.8 4.9 ? 5.7 Lowest Low Medium High Highest Mean Le gs 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 1 0.6 1.3 1.8 0.4 1.0 1.9 0.7 1.4 2.0 1.0 2.3 3.5 2 0.9 2.7 4.1 0.8 1.5 2.1 0.9 1.6 2.1 1.1 1.7 2.1 3 1.2 3.0 4.3 1.2 3.3 5.2 0.9 1.4 2.1 0.7 1.4 2.2 4 0.8 1.5 2.2 1.0 2.6 3.5 1.1 2.9 4.3 1.1 2.0 2.0 5 1.2 1.9 2.2 1.2 2.1 2.5 1.4 3.1 4.6 1.4 4.0 5.5 Figure 4.14: Three-way interaction plot for the back, leg and load interaction for Mean. 64 0.6 ? 1.1 1.2 -1.7 1.8 ? 2.3 2.4 ? 2.9 3.0 ? 3.5 3.6 ? 4.1 Lowest Low Medium High Highest Mean Legs 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 1 0.6 1.5 1.4 1.0 1.3 1.1 1.5 1.3 1.3 1.3 2.4 3.1 2 2.4 2.8 2.5 1.0 1.9 1.5 1.6 1.6 1.5 1.8 1.8 1.3 3 1.8 3.2 3.5 3.1 3.3 3.3 1.1 1.7 1.5 1.4 1.6 1.3 4 1.7 1.4 1.4 1.4 2.6 3.2 2.8 2.6 2.9 1.7 1.6 1.8 5 2.0 1.9 1.4 1.9 2.0 1.8 1.9 3.6 3.6 3.7 3.8 3.4 Figure 4.15: Three-way interaction plot for the arm, back and leg interaction for Mean. 65 0.0 ? 0.7 0.8 ? 1.5 1.6 ? 2.3 2.4 ? 3.1 3.2 ? 3.9 4.0 ? 4.7 Lowest Low Medium High Highest Median Le gs 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 Load 1 Load 2 Load 3 1 0.3 1.0 1.4 0.0 0.7 1.5 0.3 1.0 2.0 0.6 2.0 2.8 2 0.3 1.8 3.2 0.4 1.0 1.7 0.3 1.3 2.0 0.4 1.3 1.7 3 0.9 2.5 3.3 0.8 2.3 4.2 0.6 0.9 2.0 0.2 1.0 2.0 4 0.3 1.0 2.0 0.5 2.0 3.0 0.6 2.5 3.7 0.7 1.3 1.7 5 0.8 1.2 1.8 0.7 1.8 2.2 0.7 2.3 3.8 1.0 3.3 4.7 Figure 4.16: Three-way interaction plot for the back, leg and load interaction for Median. 66 0.4 ? 0.8 0.9 -1.3 1.4 ? 1.8 1.9 ? 2.3 2.4 ? 2.8 2.9 ? 3.3 Lowest Low Medium High Highest Median Legs 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 1 0.4 1.2 1.2 0.7 0.9 0.7 1.3 1.0 1.0 0.8 2.0 2.7 2 1.7 1.8 1.8 0.5 1.2 1.4 1.1 1.1 1.5 1.2 1.5 0.8 3 1.4 2.7 2.7 2.2 2.6 2.5 0.9 1.3 1.3 1.0 1.2 1.0 4 1.2 1.2 1.0 0.7 2.2 2.7 2.2 2.1 2.5 1.3 1.2 1.2 5 1.3 1.5 1.0 1.8 1.7 1.3 1.3 2.7 2.8 3.0 3.3 2.7 Figure 4.17: Three-way interaction plot for the arm, back and leg interaction for Median. 67 Visual examinations of the gures 4.11 - 4.17 suggested possible relationship between measures of spread and center. Therefore, additional analyses were conducted to explore the relationship. The scatter plots in Figures 4.18 to 4.21 describe the relationship between the two measures of spread and the two measures of center. An examination of these scatter diagrams suggested that the relationship between the measures of spread and those of center was positively correlated although not linear. Figure 4.18: The relationship between Mean discomfort and Range within a speci ed SPH task. A total of 180 SPH tasks are considered. 68 Figure 4.19: The relationship between Mean discomfort and the Standard Deviation (SD) within a speci ed SPH task. A total of 180 SPH tasks are considered. Figure 4.20: The relationship between Median and Range of discomfort within a speci ed SPH task. A total of 180 SPH tasks are considered. 69 Figure 4.21: The relationship between Median and Standard deviation (SD) within a speci- ed SPH task. A total of 180 SPH tasks are considered. 4.5 Discussion This study empirically collected ratings of perceived discomfort for one-handed static object holding tasks in order to examine the inter-individual variation of perceived postural discomfort associated with SPH. Speci cally, the study aimed to (1) examine the proba- bility distributions that characterize di erent SPH tasks and (2) determine if the measures of spread- standard deviation and range- representing the magnitude of the inter-individual di erence would vary across di erent SPH tasks and how the SPH task parameters a ect the amount of inter-individual variability. Twenty subjects performed short period (20 sec- onds) object holding tasks for 180 posture-hand load combinations. Thus, each posture-load combination had a total of twenty data points representing discomfort ratings for twenty di erent individuals. The discomfort rating re ected what the individual would perceive 70 under the de ned postural conditions. The study utilized a 10-point rating scale to obtain the subjective ratings (Borg CR-10 scale). The results of the distribution identi cation showed that there is no single well-known distribution family that can be used to robustly t di erent SPH tasks. As can be seen from Figure 4.3, there is no common probability distribution type that works across di erent SPH tasks. In a number of the SPH tasks (one-third of the 180 datasets), none of the 14 distributions commonly used for probability modeling could adequately t the data. The fact that one-third of the 180 datasets could not be modeled by any of the 14 Minitab distributions implies that the distributions are very arbitrary. Visual examination of the frequency histograms related to these datasets revealed them to be SPH tasks in which most subjects rated their discomfort relatively low resulting in a lot of values ranging between 0-3 (Figures 4.4 a - f and 4.5 g - j). The lack of t could be attributed to a small range within the datasets. Additionally, a considerable amount of SPH tasks were tted with either the Exponential family of distribution (30%) or the Weibull family of distributions (19%) . This could be attributed to sets of data with high range with most data points being on the lower end of the scale, in other words, an upper bound on the data resulted in the data being skewed left. Essentially, in these cases, majority of the data points were between 0-3 but there were few that were on the higher end of the scale resulting in a much larger range. An implication of this nding no one particular distribution type accounting for datasets is that there is need to carefully examine each SPH task distribution in order to determine the SPH tasks stress level. When assumptions cannot be made about a particular distribution type, the utilities of simple descriptive statistics, such as, mean, median, range and standard deviation become very limited in determining an SPH tasks stress levels. The histograms 71 shown in Figures 4.4 a - f and 4.5 g - l served as the rationale for conducting the individual distribution identi cation, i.e. visual examination of the histograms was indicative of the varying distribution type for each dataset. Di erent SPH trials exhibited substantial variation in the amount of spread (Figure 4.6); this could be attributed to the di erent combinations of SPH task parameters posture- load combinations. To examine the e ects of the SPH task parameters on the amount of inter-individual variability, the measures of spread were subjected to analysis of variance (ANOVA) using a general linear model. External load signi cantly a ected the amount of inter-individual variability in SPH task trials (Table 4.2). Increasing hand load results in an increase in muscle load. Di erences in performance tend to be magni ed in high muscle loading tasks resulting in signi cant inter-individual di erence in discomfort. Signi cant variability in discomfort translates into signi cant amount of inter-individual di erence in SPH task trials. The back e ects were insigni cant although there was signi cant interaction between back and leg, the interaction accounted for 23% and 17% respectively of the total variation for SD and range. Similarly the back x leg x load interaction accounted for 9% and 8% respectively of the total variation for SD and range. Therefore we can conclude that although the back postures did not signi cantly a ect the amount of inter-individual di erence, they did have a signi cant e ect when interacting with other body part postures. One of the aims of this study was to examine the characteristics of SPH task tasks that exhibit large or small inter-individual di erence. Careful observation of the shaded tables reveals some interesting trends. Generally, there was comparatively higher inter- individual variability in SPH tasks that involved higher load levels. As has already been stated earlier, increase in hand load increases muscle load resulting in larger discomfort 72 values. This translates into large inter-individual di erence in SPH task trials. SPH task trials that involved the participants adopt leg posture 5 were characterized by large inter- individual di erence (Tables 4.11, 4.12 and 4.13). The level of variation was represented by cell treatment means; higher cell means imply larger variability. SPH tasks that involved leg posture 5 were mostly characterized by either high or highest treatment means. The reason for the high cell treatment means of these leg postures could be due to the fact that the participant essentially had to perform the SPH trial in an unbalanced leg posture for 20 seconds. The dominant leg was exed at about 150 degrees and supported on a raised support. Most of the body weight ended up being distributed to his/her less dominant side. This unequal distribution of body weight could have resulted in larger inter-individual di erences in discomfort perception. This subsequently translated into signi cant amount of inter-individual di erences in SPH task trials. Additionally, this study also examined the relationship between measures of spread (range and standard deviation) and those of center (mean and median). Visual examinations of Figures (4.18) to (4.21) depict such relationship .The general trend of these plots show increasing variability with increasing mean discomfort, i.e. as the measures of center increase, the measures of spread increase too. Speci cally, it is seen that at higher discomfort levels there seems to be much higher variability. A possible explanation for increased variability at higher discomfort levels could be the result of the inter-individual di erences being magni ed when the task at hand is more stressful. The ndings of this study ll the knowledge gap that exists in our current understand- ing of SPH-associated discomfort. First, this study has shown that no single probability distribution or family of distributions can be used to accurately characterize di erent SPH 73 tasks. This nding is generally interpreted to imply that the characterization of stress levels SPH tasks by measures of center alone may not be adequate. Secondly, this study showed the e ect that the di erent task parameters-external load and posture - have on the amount of inter-individual di erences in SPH task trials. Speci cally, certain posture-load combi- nations like external hand-load and non-straight leg postures seemed to signi cantly a ect the amount of inter-individual di erences in SPH task trials (Tables 4.11, 4.12 and 4.13). These ndings point towards the need for a more precise and informative method for char- acterizing an SPH task trial, one that would accurately characterize stress levels of di erent posture-load combinations. This study has highlighted the inter-individual variations in discomfort associated with SPH tasks and in so doing demonstrated how SPH task parameters a ect the amount of inter-individual variability in SPH tasks. There is an urgent need, therefore, to consider inter-individual di erences in SPH task characterization if e ective control and assessment of WMDS is to be achieved. In demonstrating and quantifying (using concepts such as lowest and highest treatment means of posture-load combinations as used in Tables 4.11 - 4.17) the amount inter-individual di erences in SPH task trials, this study has provided an introduction into some future study that will seek to consider population distributions of perceived discomfort in characterizing SPH task trials. It is feasible, therefore, to take inter-individual di erences into account in characterizing SPH tasks if we adopt methods that use empirical discomfort distributions. This will be a step forward in handling the variations in SPH tasks and would probably result in more individuals being accommodated while conducting SPH tasks. The current study was limited to 20-second static trials; future studies could investigate dynamic tasks and/or tasks longer than 20 seconds in duration. 74 Chapter 5 Quantifying discomfort levels of static posture holding tasks using empirical distributions of perceived discomfort. 5.1 Introduction Static posture holding (SPH) tasks statistically maintaining working postures over cer- tain time durations are common across various industrial sectors, including manufacturing, service and agriculture (Olendorf and Drury 2001; Chung et al. 2005). SPH tasks can impose large stresses on the musculoskeletal system and lead to signi cant discomfort to workers if they are designed without due consideration of human characteristics, capabilities and limitations. Discomfort associated with SPH needs to be controlled as excessive discomfort may indicate high risks of work-related musculoskeletal disorders (WMSDs) - although the link between work-related discomfort and WMSD risks has not been conclusively established, many research studies pointed to its existence on the grounds that they are both related to exposure to biomechanical load on the musculoskeletal system (Nag 1991; Putz-Anderson and Galinsky 1993; Miedema et al. 1997; Bernard et al. 1997). Some researchers indeed view work-related discomfort/pain as a precursor of WMSDs (Corlett and Bishop 1976; Dickerson et al. 2007; Kee 2004; Kee and Lee 2012). Thus, it is possible that controlling SPH-associated discomfort contributes towards the reduction of WMSD risks (Putz-Anderson and Galinsky, 75 1993, Dul et al. 1994, Scott Schneider, 1998, Miedema et al., 1997, Kee and Lee 2012). Also, regardless of whether or not discomfort is associated with WMSD risks, it is certainly desirable to prevent excessive discomfort as it is a negative human experience that can signi cantly compromise the quality of work life and also work productivity. An SPH task can be speci ed in terms of three task parameters: that is, posture, ex- ternal load and posture holding time (PHT). These task parameters seem to greatly a ect the level of perceived discomfort that an individual experiences from performing an SPH task. In addition, the performers individual characteristics, physical and psychological, also contribute. Thus, di erent individuals experience di erent levels of discomfort even when performing identical SPH tasks in terms of the three task parameters. A group or a popula- tion of workers therefore would give rise to a probability distribution of perceived discomfort for a particular SPH task. From an ergonomics viewpoint, design interventions for con- trolling SPH-related discomfort must aim to ensure that the majority of the target worker population experiences less than excessive discomfort. Such idea of population accommoda- tion has been one of the most fundamental design principles in the eld of ergonomics and human factors engineering (Marshall et al., 2008, Marshall et al., 2010, Jung et al., 2009, Reed and Flannagan, 2001, Vasu and Mital 2000). E ective control of SPH-related discomfort necessitates as a prerequisite an ability/method for representing discomfort levels of di erent SPH tasks and further determining their ac- ceptability. In other words, determining which SPH tasks are acceptable or unacceptable or whether current SPH tasks need to be redesigned or not requires a certain index/metric that operationalizes the construct of discomfort level of an SPH task in a manner that guides designers decision makings. Surprisingly, the fundamental question of how to quantify the 76 level of discomfort of an SPH task has not been extensively studied. Numerous previous studies examined individuals perceived discomfort ratings data resulting from SPH tasks (Boussenna et al. 1982; Manenica 1986; Genaidy et al. 1995; Kee and Karwowski 2001; Olendorf and Drury 2001; Kee 2004; Park et al. 2009; Meijst et. al. 1995; Miedema et. al. 1996; Reneman et. al. 2001; Chung et al. 2003b; Kee and Lee 2010). However, these studies did not result in the development of a quantitative index for meaningfully represent- ing discomfort levels of SPH tasks. They mostly examined the e ects of task parameters and personal variables on perceived discomfort using statistical analysis techniques, such as analyses of variance and regression analyses. Understanding the e ects of task and personal variables on discomfort is certainly important. However, it by itself does not shed light on the question of which SPH tasks are acceptable or unacceptable, especially in light of the goal of population accommodation. The lack of means for meaningfully quantifying discomfort levels of di erent SPH tasks is problematic as it hinders decision makings involved in ergonomics design interventions for controlling SPH-related discomfort, especially, making the decisions on which SPH tasks are acceptable or unacceptable. As an e ort towards addressing this problem, the current study presents a new quantitative index for characterizing discomfort levels of SPH tasks. This new index is named the population accommodation level estimate (PALE) and estimates the proportion of the target worker population that experiences less than excessive discomfort. This new index is predicated upon the use of empirical discomfort distributions. In what follows, the index as well as the data collection needed for the use of the index are described. Also, an illustrative example is provided to demonstrate the usage of the index. 77 5.2 The population accommodation level estimate (PALE) index 5.2.1 Data collection Any given SPH task can be speci ed by three parameters: (1) the adopted posture (P), (2) the external load (L) and (3) the posture holding time (PHT). The posture can be de ned by a combination of body part angles. The external load can be represented by hand-held load levels or loading exerted by pushing or pulling activities. The PHT represents the total time duration an individual performs an SPH task. The level of discomfort for a given SPH task can therefore be represented as follows: Discomfort level = f(P;L;PHT) (5.1) Having de ned an SPH task by its three parameters, the level of discomfort is empirically obtained from each individual performing the task by use of a psychophysical rating scale. The experimental procedure including prior preparation for a SPH task study generally consists of ve steps: (1) determination of posture-load combination to be used and the posture holding time; (2) Pilot testing the SPH task to determine if subjects will be able to adopt the postures and also determine if the posture holding time will result in excessive discomfort for the target population; (3) conduct subject recruitment. Depending on the objective and the target population, recruitment could be limited to a certain gender or a speci c age group; (4) conduct a training session to familiarise subjects with the experiment and obtain an informed consent from the subjects; and (5) administer the SPH task and measure discomfort level at the completion of each task. 78 5.2.2 Probability histogram representation An initial step in developing this new representation method is to establish a probability distribution associated with the discomfort level of a SPH task. The following notation will be adopted: Let y, denote the physical discomfort for a given SPH task and p (y) the probability that a SPH task trial has a discomfort rating y. For a given SPH task trial, the probability of occurrence for discomfort rating y will be computed as follows: p(y) = Number of subjects experiencing yTotal number of subjects (5.2) The probability of a discomfort rating being within a speci ed range can now be char- acterized based on the probability histogram for each SPH task. Depicting the discomfort distribution by use of histograms implies that no prior assumptions need not to be made about the underlying probability distribution. 5.2.3 Population accommodation level estimate An accommodation level is de ned as the percentage of the total population performing a SPH task that would experience discomfort less than a speci ed design limit of discomfort. A designed limit is de ned as the maximum acceptable discomfort level on a psychophysical rating scale. The population estimate of the accommodation level is expressed as follows: accommo- dation level = Pr (discomfort level) (design limit). The estimate of the accommodation level can be computed as follows: 79 Probability(discomfort y) = yX Y =0 (Probability(discomfort = Y)) (5.3) 5.3 An illustrative example 5.3.1 Experiment Ten males and ten females participated in this study as paid volunteers. Prior to participation, the participants had a training session to familiarize themselves with the use of the Borg CR-10 rating scale (Figure 5.1). Each participant signed a written informed consent before participation. The Auburn University Institutional Review Board approved the experimental protocol. 0 - 0.5 - 1 - 2 - 3 - 4 - 5 - 6 - 7 - 8 - 9 - 10 - . Nothing at all Extremely weak (just noticeable) Very weak Weak (light) Moderate Somewhat strong Strong (heavy) - Very strong - - Extremely strong (almost max) Maximal Figure 5.1: Borg?s new category-ratio (CR-10) scale. Reprinted from Borg 1982. The participants performed a short-period (20 seconds long) one-handed static object holding task in a total of 180 posture-hand load conditions: the one hundred and eighty 80 posture-hand load conditions are a combination of sixty postures and three hand load levels. The sixty postures represent various one-handed static object holding postures that can be found in industrial workplaces and were generated by combining four back, three arm and ve leg postures. The three hand load conditions were 0 kg (no hand load), 2.2 kg and 3.6 kg; cylindrical objects were used as the hand loads. Prior to a box holding trial, a participant changed into a tight out t (tight short and short sleeves). Also, spherical re ective markers were attached on body landmarks so that the body postures can be accurately recorded by the VICON motion capture system during the experiment trial (Figure 5.2). After this initial preparatory set up, the participant adopted the posture and performed the one-handed static object holding task the object was be brought to the participants dominant hand by the experimenter immediately after the participant adopted the posture. In each trial the VICON system kept track of the positions of the attached re ective markers over time and display the distances between their current and correct positions in real-time. When a body attached marker is farther away from its correct position the experimenter instructed the participant to correct the posture. For each SPH task, Subjects were asked to rate their level of discomfort at the completion of each 20-seconds long SPH task trial. The 180 SPH trials were performed over three sessions by each participant. The three sessions were conducted on three separate days to prevent fatigue e ects. The participants were given as much rest as they needed between consecutive trials. Even when the partici- pated reported no fatigue, at least 1 minute rest period was provided before the next trial. Also, and after 9 experimental trials, the participant was given an extra 5 minute rest period before he/she could start the next trial. The subjects were also allowed to have additional 81 Figure 5.2: An illustration of the experimental set-up for the SPH task trials including the VICON capture shot of a static trial. rest periods as they wanted. Similar rest patterns were reported by Olendorf and Drury (2001) who also had the participants in their study perform a 20 s long SPH task. In their study, participants were allowed 45 s breaks between consecutive trials. To determine the probabilities associated with each discomfort level in a SPH task trial, a probability histogram for the discomfort levels was plotted. Therefore, a total of 180 distinct probability histograms were plotted to represent the 180 SPH tasks. The probability histograms allow for the computation of discomfort rating probabilities for the SPH tasks. To determine the accommodation level for a SPH task, a design limit was set at 4 for our current study. y = 4 is the discomfort rating value at which a subject undertaking a SPH task will perceive somewhat strong discomfort based on the Borg CR-10 rating scale. At this point the discomfort is considered moderate. The probability of discomfort rating being 82 less than or equal to 4 is computed as: Pr (0 y 4). Population accommodation level is estimated based on this criterion. The population accommodation level estimate (PALE) in each SPH task and is computed as follows: PALE = (0 y 4) (5.4) SPH tasks with accommodation levels greater than 95 percent are considered safe for a majority of the study population. The 95-100 percent probability range is used as a criterion for judging safe and acceptable SPH tasks in this study. 5.3.2 Results For each SPH task, and thus for a total of 180 SPH task trials, a probability histogram associated with it was plotted. For each SPH task, there was a distinct probability distri- bution associated with it. Figure 5.3 (a) Figure 5.3 (f) show examples of such histograms. Only few examples are included here for the sake of brevity. 83 a). SPH 121 b) SPH 36 c). SPH 107 d) SPH 18 e) SPH 50 f) SPH 4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5 -0.9 1.0 -1.9 2.0 -2.9 3.0 -3.9 4.0 -4.9 5.0 -5.9 6.0 -6.9 7.0 -7.9 8.0 -8.9 9.0 -9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5 -0.9 1.0 -1.9 2.0 -2.9 3.0 -3.9 4.0 -4.9 5.0 -5.9 6.0 -6.9 7.0 -7.9 8.0 -8.9 9.0 -9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5 -0.9 1.0 -1.9 2.0 -2.9 3.0 -3.9 4.0 -4.9 5.0 -5.9 6.0 -6.9 7.0 -7.9 8.0 -8.9 9.0 -9.9 10 Pro bab ilty Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5 -0.9 1.0 -1.9 2.0 -2.9 3.0 -3.9 4.0 -4.9 5.0 -5.9 6.0 -6.9 7.0 -7.9 8.0 -8.9 9.0 -9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0-0 .4 0.5 -0.9 1.0 -1.9 2.0 -2.9 3.0 -3.9 4.0 -4.9 5.0 -5.9 6.0 -6.9 7.0 -7.9 8.0 -8.9 9.0 -9.9 10 Pro bab ility Discomfort rating (Borg CR-10) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0-0 .4 0.5 -0.9 1.0 -1.9 2.0 -2.9 3.0 -3.9 4.0 -4.9 5.0 -5.9 6.0 -6.9 7.0 -7.9 8.0 -8.9 9.0 -9.9 10 Pro bab ility Discomfort rating (Borg CR-10) Figure 5.3: Examples of Probability histograms of discomfort ratings for selected SPH tasks. 84 To determine the percentage of our study participants that were accommodated at the design limit of y= 4, an accommodation level was computed for each SPH task. For our study population this value represented the percentage of the population that rated the task as acceptable based on the pre-determined discomfort design limit. Figure 5.4 shows the results. The following observations can be made from table 1: (1) in 60% of the SPH asks (108 out 180), the accommodation level was deemed to be within the acceptable limit (0.95- 1.00); (2) of the remaining 72 SPH tasks that were considered unacceptable, 57% (41 out of 72) were tasks involved object holding in an elevated arm position (arm posture 3); and (3) out of the 108 tasks that had accommodation levels in the 0.95-1.00 range, 55% (59 out of 108) were tasks where there was no hand-held load (Load level 1). 85 Le gs Lo ad Le ve l 1.Straight Back 2. Back Bent 3. Back twisted 4. Back bent & twisted Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 Arm 1 Arm 2 Arm 3 1 1 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.90 2 1.00 1.00 0.85 1.00 1.00 0.75 1.00 1.00 0.75 0.90 0.95 0.70 3 0.95 0.90 0.60 0.90 0.95 0.55 0.95 0.90 0.65 0.9 1.00 0.50 2 1 1.00 1.00 1.00 1.00 1.00 0.95 1.00 1.00 0.95 0.95 1.00 1.00 2 1.00 0.95 0.80 0.90 0.95 0.70 1.00 1.00 0.65 0.90 0.90 0.70 3 0.90 0.90 0.65 0.90 0.85 0.60 0.95 0.95 0.45 0.95 0.85 0.50 3 1 1.00 1.00 1.00 1.00 1.00 1.00 0.95 1.00 1.00 1.00 1.00 0.95 2 0.95 1.00 0.80 1.00 0.90 0.75 1.00 0.95 0.70 0.90 0.90 0.70 3 0.95 0.95 0.65 1.00 0.95 0.55 0.90 0.95 0.60 0.90 0.95 0.40 4 1 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.95 0.95 1.00 2 1.00 1.00 0.65 0.95 1.00 0.75 1.00 0.95 0.80 0.80 0.80 0.65 3 0.95 1.00 0.55 0.85 0.95 0.50 0.90 0.95 0.65 0.85 0.95 0.40 5 1 1.00 1.00 1.00 1.00 1.00 0.95 0.95 1.00 0.95 1.00 0.95 1.00 2 1.00 1.00 0.8 0.95 0.95 0.75 0.95 0.90 0.75 0.80 0.90 0.75 3 0.95 1.00 0.60 0.85 0.85 0.55 0.95 0.9 0.55 0.85 0.75 0.55 Figure 5.4: Accommodation levels for the study population representing 180 SPH tasks. 86 5.4 Discussion This study has presented a new approach to characterizing discomfort levels of SPH tasks, one that is based on empirical probability distribution of discomfort, named PALE (Population Accommodation Level Estimate). The goal was to use probabilities to charac- terize the discomfort distribution in SPH tasks and determine the percentage of the study population accommodated while performing an SPH task. The accommodation level is com- puted based on some prede ned discomfort design limit. The researchers in this study are yet to nd studies that do consider population distribution of discomfort levels. This study therefore, aimed to address the lack of consideration of population distribution of physical discomfort levels in SPH tasks. As has been stated in the introduction, changes in discomfort levels in a SPH task are driven by the di erent parameters that specify the task it including; posture, external load and posture holding time. It is paramount that the population distribution of discomfort levels in a SPH task is examined if accurate characterization is to be made. Visual ex- aminations of the probability histograms depicting the distribution of discomfort levels of SPH tasks show di erences in distribution type in di erent SPH tasks (Figure 3). No single distribution type or family of distributions could be used to characterize these tasks. A new approach in discomfort level characterization is therefore necessary. The use of prob- abilities histograms does provide a simple, yet e cient method for quantifying population distribution of discomfort levels in SPH tasks. To illustrate this new PALE approach, the percentage of the study population accom- modated while performing SPH tasks was computed using the discomfort level probabilities. 87 The proportions of the study population that fell below the threshold limit were considered to be accommodated. Table 1 shows the accommodation levels for the 180 SPH tasks. Higher accommodation values imply the SPH task was less stressful to a majority of the study pop- ulation. Two major observations were be made from Table 1. First, the low accommodation values of SPH tasks that involved the arm being raised above the shoulder level (Arm 3) seem to suggest that these tasks were more stressful and therefore resulted in discomfort values larger than the set threshold limit. This is in agreement with studies done by both Park et al., (2009) and Olendorf and Drury (2001), who identi ed elevated arm posture as a stress-increasing postural element. Second, a majority (108 out of 180) of the SPH tasks seemed to be less stressful, suggested by the high accommodation values ( 95%) . This could be attributed largely to the fact that this study was limited to 20-seconds SPH tasks, maybe if the task duration was longer we could observe di erent patterns. An additional goal of this study was to determine which SPH tasks are safe for a majority of the study population. Setting a criterion for rejection of accommodation values helps to determine which tasks should be included when designing the tasks and which should be taken out. Speci cally, the tasks that were found to be unacceptable could be re-designed. The PALE index proposed here provides us with a novel strategy for evaluating SPH tasks. First, a discomfort level limit can be set using a psychophysical rating scale (Borg CR-10). Second, a target population accommodation level can be chosen (e.g. 95% of the population based on the threshold limit chosen). Third, PALE can be calculated and com- pared against the target accomodation level. If PALE is less than the target accomodation level, the task will be considered unacceptable. Using this approach e ective intervention strategy can be made based on desired accommodation performance. Questions regarding 88 well-de ned population percentages can be addressed, since the proportions were established based on empirically obtained discomfort rating levels. Accurate recommendations can now be made based on population percentages that would nd a SPH task stressful or less stress- ful. This research study has proposed the use of probability histograms in characterizing perceived discomfort ratings of one-handed SPH tasks. It is thought that this study has su ciently demonstrated a new method for SPH task evaluation detailing accommodation levels. Probabilistic statements that have not been made before with regard to physical discomfort level evaluation of SPH tasks can now be made. Overall, probability histograms of the discomfort ratings for a population of workers are a simple and e ective means of characterizing discomfort levels of SPH tasks. It should be noted however, that this study has not developed an ergonomics evaluation tool, but rather an index for SPH task evaluation that quanti es physical discomfort level in probabilistic terms. In this regard, the threshold limit used in this study should not be considered as a general limit but rather as a guideline. Various threshold limits can be set based on speci c conditions and requirements. While this new quantitative index for characterizing discomfort levels of SPH tasks seems very promising, the accommodation levels computed in this study were only point estimates of the population accommodation level. A larger sample size will allow for the investigation of other statistical inferences and calculation of con dence intervals so that we are not just dealing with point estimates but appropriate interval estimates. Also, with a larger sample size we will be able to use parametric statistics to try and obtain the real underlying distribution of a SPH task. Future studies could look into the development of quantitative models that will realistically predict how people move and interact with 89 systems and how the discomfort distributions vary in those environments. This will be a step forward from the current study as this will involve examining dynamic tasks that are commonly encountered in industry. With more data, future studies could also focus on the distribution of speci c individual factors such as stature, BMI, gender and body part posture in relation to perceived discomfort. 90 Chapter 6 Conclusion and Reccomendations for Future Studies 6.1 Introduction In ergonomics and related research areas, numerous studies have examined discomfort resulting from static posture holding (SPH). Many of these studies aimed to understand the e ects of posture and external load on perceived discomfort. Some studies investigated the e ect of task duration (posture holding time(PHT)) on discomfort of SPH. The results of these studies indicated that posture signi cantly a ects perceived discomfort along with pos- ture holding time and external load. Also, these studies reported that perceived discomfort monotonically increases over time during a prolonged SPH task trial. These research studies have greatly advanced our understanding of perceived discomfort from SPH. Nonetheless, some signi cant knowledge gaps seem to exist, which hamper adequately representing and evaluating discomfort associated with SPH. First, the existing reseach studies do not seem to have fully elucidated the mathematical characteristics of the discomfort-time relationship. Second, very few research studies seem to have examined SPH-associated discomfort with a focus on the inter-individual variation. This study was able to demonstrate the di erent discomfort-time patterns of SPH tasks. Speci cally, this study reports that the positively accelerated time increase patterns are found in some discomfort-time sequences of prolonged SPH trials in addition to linear and 91 negatively accelerated time increase patterns, which were observed by previous studies. Ad- ditionally, this study also reports that there is no single well-known distribution family that can be used to robustly t di erent SPH tasks. A population of workers of workers would give rise to a proabability distribution of perceived discomfort for a given SPH task. A con- sequence of lack of single distribution to describe an SPH task?s stress levels is that simple descriptives statistics become limited in describing an SPH task?s stress level. Individual di erences have to be considered when characterizing discomfort levels of SPH tasks. A new approach for quantifying discomfort levels was proposed. This study has proposed a new methodology for characterizing discomfort levels of SPH tasks. A quantitative index, Probability-based Accomodation Level (PALE), has been proposed to quantify the discom- fort levels of SPH tasks. Probabilistic statements that have not been made before with regard to physical discomfort level evaluation of SPH tasks can now be made. 6.2 Summary of ndings Two primary experiments were performed in this study. The objective of the rst study was to elucidate the nature of the relationship between perceived discomfort of SPH and PHT. The main objective of the second study was to conduct an investigation on the inter- individual variation in perceived discomfort of SPH tasks. Data and results from study two were also used to propose a new methodology for characterizing discomfort levels of SPH tasks. This new method was used to estimate the population accomodation levels for the SPH tasks performed in this study. The ndings of the research are summarized below. 92 Three di erent monotonically increasing time patterns (negatively accelerated, linear and positively accelerated) seem to co-exist in discomfort-time pro les of SPH trials. The power function form can serve as a general representation of the discomfort-time relationship. Previous research studies suggested that the relationship between per- ceived discomfort of SPH and PHT is linear in its nature, while others contended that a negatively accelerated logarithmic time function better depicts the discomfort-time relationship. The results of the distribution identi cation showed that there is no single well-known distribution family that can be used to robustly t di erent SPH tasks. There is need, therefore, to carefully examine each SPH task distribution in order to determine the SPH tasks stress level. Simple descriptive statistics are limited in describing an SPH task?s stress levels. Di erent SPH trials exhibited substantial variation in the amount of spread. External load signi cantly a ected the amount of inter-individual variability in SPH task trials. Generally, there was comparatively higher inter-individual variability in SPH tasks that involved higher load levels. Increases in hand load increases muscle load resulting in larger discomfort values. At higher discomfort levels there seems to be much higher variability in SPH tasks. A possible explanation for increased variability at higher discomfort levels could be due to magn ed inter-individual di erences when the task at hand is more stressful. 93 Probability histograms of the discomfort levels for a population of workers are a simple and e ective means of characterizing discomfort levels of SPH tasks. The fact that we are able to quantify the proportion of the study population that experiences a speci ed level of discomfort also allows us to quantitatively answer questions such as; what percentage of study population will experience a certain level discomfort while performing manual work tasks? The proposed Probability-based Accomodation Level Estimate (PALE) methodology seems very promising and may provide us with a better approach for quantifying an SPH task?s stress levels. The PALE index seves as a quantitative index for meaning- fully representing discomfort levels of SPH tasks. A target population accommodation level can be set and compared against a calculated PALE value to determine the ac- ceptability or unacceptability of a given SPH task. 6.3 Limitations and reccomendations for future studies Some limitations of the current research study along with reccomendations for future studies are presented. The rst study only considered two-handed static box holding tasks and one hand- load weight. Collecting and analyzing more data would be bene cial especially for con rming the validity of the research ndings over a wider range of conditions. Most study participants in this research study were young adults and of normal weight category. Future studies will need to examine discomfort-time pro les of older and/or 94 obese individuals to understand the e ects of age and obesity level and also con rm the validity of the study ndings with those segments of the population. This study examined only subjective ratings of perceived of perceived discomfort with- out employing other stress/strain measures. Collecting and analyzing physical and physiological response data along with perceived discomfort ratings would further en- hance our understanding of the discomfort-time relationship in SPH. The second study was limited to 20-second static trials; future studies could investigate tasks longer than 20 seconds in duration. The accommodation levels computed in this study were only point estimates of the population accommodation level. Future studies could look into the use of statistical inferences and con dence intervals to describe the discomfort levels of SPH tasks. A larger sample size will allow for the calculation of such con dence intervals so that we are not just dealing with point estimates but appropriate interval estimates. Future studies could look into the development of quantitative models that will real- istically predict how people move and interact with systems and how the discomfort distributions vary in those environments. These outcomes will represent an important step towards developing and perfecting not only a method that can be used to accu- rately simulate variability in indviduals? performing industrial tasks, but also a new Digital Human Modeling Approach for postural stress evaluation and control. 95 Bibliography [1] Aaras, A., Westgaard, R.H., and Stranden, E. 1988, Postural angles as an indicator of postural load and muscular injury in occupational work situations, Ergonomics, 31, 915-933. [2] Anderson, L. A., and McLean, R.A., 1974. Design of experiments. A realistic approach. New York and Basel: Mercel Dekker, INC. [3] Armstrong, T.J, Buckle, P., Fine, J.L., Hagberg, M., Jonsson, B., Kilbom, A., Kuorinka, I.AA., Silverstein, B.A., Sjogaard, G., and Viikari-Juntura, E.RA., 1993. A conceptual model for work-related neck and upper-limb musculoskeletal disorders. Scandinavian Journal of Work, Environment and Health, 19, 7374. [4] Badler, N., Bindiganavale, R., Granieri, J., Wei, S. and Zhao, X., 1994, Posture inter- polation with collision avoidance. In Computer Animation 94, 2528 May 1994, Geneva, Switzerland (Los Alamos, CA: IEEE Computer Society Press), pp. 1320. [5] Baron, S., Milliron, M., Habes, and D., Fidler, A. 1991, Hazard evaluation and techni- cal assistance report: Shoprite Supermarkets, New Jersey-New York. Cincinnati, OH: U.S Department of Health and Human Service, Centers for Disease Control, National Institute for Occupational Safety and Health, NIOSH Report No. HHE 88-334-2092. [6] Borg, G., 1982. Psychophysical bases of perceived exertion. Medicine and Science in Sports and Medicine, 14, 377381. [7] Benard, B., Putz-Anderson, V., Grant, K., Hurrell, J., Sweeney, M. H., Tanaka, S. and Fine, L. 1997, NIOSH review of the epidemiologic evidence of work-related muscu- loskeletal disorders: use, standards and guidelines, in From Experience to Innovation, IEA97, Tampere, Finland, vol.3 (Helsinki: Finnish Institute of Occupational Health), 489 - 492 [8] Benard, B., Musculoskeletal disorders (MSD) and workplace factors: a review of the epidemiologic evidence for work-related musculoskeletal disorders of the neck, up- per extremity and low backU.S. Government Printing O ce, Washington, DC (1997) (DHHS/NIOSH publication no. 97-141) [9] Bjelle, A., Hagberg, M., and Michaelsson, G., 1979, Clinical and Ergonomic factors in prolonged shoulder pain among industrial workers. Scandinavian Journal of Work Environ Health 5(3) 205-210. 96 [10] BLS, 1995, Worplace injuries and illness in 1994. Washington, DC, U.S. Department of labor, Bureau of labor statistics, USDL 95-508. [11] Boussema, M., Corlett, E.N., and Pheasant, S.T., 1982. The relationship between dis- comfort and postural loading at the joints. Ergonomics, 25 (4), 315-322. [12] Burdorf, A., Govnert, G., and Elders, L., 1991. Postural load and back pain of workers in the manufacturing of prefabricated concrete elements. Ergonomics, 34 (7), 909918. [13] Cha n, D.B., and Anderson, G.B.J., 1991. Occupational Biomechanics. Wiley, New York. [14] Chung, M.K., Lee, I., Kee, D. and Kim, S.H., 2002. A postural workload evaluation system based on a macro-postural classi cation. Human Factors and Ergonomics in Manufacturing, 12 (3), 267-277. [15] Chung, M.K., Lee, I. and Kee, D., 2003a. Assessment of postural load for lower limb postures based on perceived discomfort. International Journal of Industrial Ergonomics, 31, 17-32. [16] Chung, M.K., Lee, I., and Kee, D., 2003b. E ect of stool height and holding time on postural load of squatting postures. International Journal of Industrial Ergonomics, 32, 309-317. [17] Chung, M.K., Lee, I. and Kee, D., 2005. Quantitative postural load assessment for whole body manual tasks based on perceived discomfort. Ergonomics, 48 (5), 492-505. [18] Corlett, E.N. and Bishop, R.P., 1976. A technique for assessing postural discomfort. Ergonomics, 19 (2), 175-182. [19] Corlett, E.N. and Manenica, I., 1980. The e ects and measurement of working postures. Applied Ergonomics, 11 (1), 7-16. [20] David, G., Woods, V., Li, G. and Buckle, P., 2008. The development of the Quick Expo- sure Check (QEC) for assessing exposure to risk factors for work-related musculoskeletal disorders. Applied Ergonomics, 39, 57-69. [21] Dempsey, P.G., McGorry, R.W., and Maynard, W.S., 2005. A survey of tools and methods used by certi ed professional ergonomists. Applied Ergonomics 36, 489503. [22] Dickerson, C.R., Martin, B.J., and Cha n, D.B., 2006. The relationship between shoul- der torques and the perception of muscular e ort in load transfer tasks. Ergonomics, 49 (11), 10361051. [23] Dickerson, C.R., Martin, B.J., and Cha n, D.B., 2007. Predictors of perceived e ort in the shoulder during load transfer tasks. Ergonomics, 50 (7), 10041016. [24] Drury, C.G., Atiles, M., Chaitanya, M., Lin, J., Marin, C., Nasarwanji, M., Paluszak, D., Russell, C., Stone, R. and Sunm, M., 2006. Vicarious perception of postural discomfort and exertion. Ergonomics, 49 (14), 1470-1485. 97 [25] Dul, J., Douwes, M., and Smitt, P., 1994. Ergonomic guidelines for the prevention of discomfort of static postures based on endurance data. Ergonomics, 37 (5), 807-815. [26] Fathallah, A.F, Meyers M. J and Janowitz. 2004. Stooped and Squatting Postures in the Workplace. Pg. 16. Conference proceedings, Center for Occupational Health and Environmental Health, University of California Berkeley, Oakland, California, USA. [27] Gallagher, S. 2005. Physical limitations and musculoskeletal complaints associated with work in unusual or restricted postures: A literature review. Journal of Safety Research 36, 51 61. [28] Genaidy, A.M, Houshyar, A., and Asfour, S.S., 1990. Physiological and psychophysical responses to static, dynamic and combined arm tasks. Applied Ergonomics 21, 63-67. [29] Genaidy, A.M, Al-Shedi, A.A, and Karwowski, W., 1994. Postural stress analysis in industry. Applied Ergonomics 25(2) 77-87. [30] Genaidy, A., Barkawi, H., and Christensen, D., 1995. Ranking of static non-neutral postures around the joints of the upper extremity and the spine. Ergonomics, 38, 1851- 1858. [31] Gescheider, G. A., 1985. Psychophysics, Method, Theory and Application. New Jersey: Lawrence Erlbaum. [32] Grandjean, E. and Hunting, W., 1977. Ergonomics of postures review of various prob- lems of standing and sitting postures. Applied Ergonomics, 8 (3), 135140. [33] Hahn, G.J., and Meeker, W.Q., 1991. Statistical Intervals: A Guide for Practitioners. New York, John Wiley & Sons. [34] Haslegrave, M, Tracy, M.F., and Corlett, E.N., 1997. Force exertion in awkward working postures-strength capability while twisting or working overhead, Ergonomics, 40:12, 1335-1356. [35] Hignett, S. and Mcatamney, L., 2000. Rapid entire body assessment (REBA). Applied Ergonomics, 31, 201 205. [36] Jung, E.S. and Choe, J., 1996. Human reach posture prediction based on psychophysical discomfort. International Journal of Industrial Ergonomics, 18, 173-179. [37] Karhu, O., Kansi, P. and Kuorinka, I., 1977. Correcting working postures in industry: a practical method for analysis. Applied Ergonomics, 8, 199-201. [38] Karhu, O., Kansi, P. and Kuorinka, I., 1977. Correcting working postures in industry: a practical method for analysis. Applied Ergonomics, 8, 199-201 [39] Karhu, O., Harkonen, R., Sorvali, P., and Vepsalainen, P., 1981. Observing working postures in industry: Examples of OWAS application. Applied Ergonomics, 12.1, 13-17. 98 [40] Kee, D. and Karwowski, W., 2001. LUBA: An assessment technique for postural load- ing on the upper body based on joint motion discomfort and maximum holding time. Applied Ergonomics, 32, 357 366. [41] Kee, D. and Karwowski, W., 2007. A Comparison of Three Observational Techniques for Assessing Postural Loads in Industry. International Journal of Occupational Safety and Ergonomics (JOSE) , Vol. 13, No. 1, 314. [42] Kee, D., 2004. Investigation on perceived discomfort depending on external load, upper limb postures and their duration. Journal of the Korean Institute of Industrial Engineers, 30 (2), 76-83 (text in Korean). [43] Kemmlert, K. and Kilbom, A., 1987. Method for identi cation of musculoskeletal stress factors which may have injurious e ects. Paper presented at XIth World Congress on the Prevention of Occupational Accidents and Diseases, Stockholm, 401-404. [44] Kee, D. and Lee, I., 2012. Relationships between subjective and objective measures in assessing postural stresses. Applied Ergonomics, 43, 277-282. [45] Kee, D. and Lee, I., 2010. Discomfort and other measures for the assessment of postural load. Advances in Occupational, Social and Organizational Ergonomics, 421-430. [46] Kilbom, A., and Persson, J., 1987, Work techniques and its consequences for muscu- loskeletal disorders. Ergonomics, 30(2) 273-279. [47] Kirk, N.S. and Sadoyama, T., 1973. A relationship between endurance and discomfort in static work. Unpublished M.Sc. report, Department of Human Sciences, University of Technology, Loughborough. [48] Manenica, I., 1986. The ergonomics of working postures: a technique for postural load assessment. In: Corlett, E.N., Wilson, J., Manenica, I. (Eds.), The Ergonomics of Working Posture. Taylor & Francis, London, pp. 270-277. [49] Marshall, R., Case, K., Porter, M., Summerskill, S., Gyi, D., Davis, P., and Sims, R., 2010. Multivariate design inclusion using HADRIAN. Society of Automobile Engineers, 08DHM-0034. [50] Marshall, R., Case, K., Summerskill, S., Porter, M., Gyi, D., Sims, R. and Davis, P., 2010. HADRIAN: a virtual approach to design for all. Journal of Engineering Design. 21, Nos. 2-3, 253-273. [51] McAtamney, L. and Corlett, E.N., 1993. RULA: a survey method for the investigation of work-related upper limb disorders. Applied Ergonomics, 24, 91 99. [52] McAtamney, L. and Hignett, S. 1995. REBA - A rapid entire body assessment method for investigating work-related musculoskeletal disorders, in V. Blewett (ed.), Proceedings of the 31st Annual Conference of the Ergonomics Society of Australia. 99 [53] Meijst, W., Dul, J. and Haslegrave, C., 1995. Maximum holding times of static standing postuers. TNO Institute of Preventive Health Care, Leiden. [54] Meyers, J.M., Miles, J.A., Faucett, J., Janowitz, I., Tejeda, T.G., Kabashima, J.N., 1997. Ergonomics in agriculture: workplace priority setting in the nursery industry. Am. Ind. Hyg. Assoc. J. 58 (2), 121e126. [55] Myers, A.H., Baker, S.P., Li, G, 1999. Back injury in municipal workers: A case control study. Am. J. Public Health. 89: 1036-1041. [56] Miedema, M.C., Douwes, M., and Dul, J., 1997. Recommended maximum holding times for prevention of discomfort of static standing postures. International Journal of Indus- trial Ergonomics, 19, 9-18. [57] Milner, N., 1985. Modelling fatigue and recovery in static postural exercise. Ph.D. The- sis, University of Nottingham. [58] Montgomery, D. C., 2000. Design and Analysis of Experiments. John Wiley & Sons. [59] Nag, P.K., 1991. Endurance limits in di erent models of load holding. Applied Er- gonomics, 22, 185-188. [60] Olendorf, M.R. and Drury, C.G., 2001. Postural discomfort and perceived exertion in standardized box-holding postures. Ergonomics, 44 (15), 1341-1367. [61] Park, W., Singh, D., Levy, M. and Jung, E.S. 2009. Obesity E ect on Perceived Postural Stress during Static Posture Maintenance Tasks. Ergonomics, 52 (9), 1169-1182. [62] Pourmahabadian, M., Akhavan, M., and Azam, K., 2008. Investigation of Risk Factors of Work-Related Upper-Limb Musculoskeletal Disorders in a Pharmaceutical Industry. Journal of Applied Sciences 8 (7): 1262-1267. [63] Putz-Anderson, V. and Galinsky, T.L., 1993. Psychophysically determined work du- rations for limiting shoulder girdle fatigue from elevated manual work. International Journal of Industrial Ergonomics, 11, 19-28. [64] Putz-Anderson, V., ed. 1988. Cumulative trauma disorders: A manual for musculoskele- tal diseases of the upper limbs. London: Taylor & Francis. [65] Reneman, M.F., Bults, M.M.W.E., Engbers, L.H., Mulders, K.K.G., and Geken, L.N.H., 2001. Measuring maximum holding times and perception of static elevated work and forward bending in healthy young adults. Journal of Occupational Rehabilitation, 11 (2), 87-97. [66] Reed, M.P. and Flannagan, C.A.C., 2001. Modeling Population Distributions of Sub- jective Ratings. Society of Automotive Engineers, Inc., 2001-01-2122. [67] Sakakibara, H., Miyao, M., Kondo, T., and Yamada, S., 1995, Overhead work and shoulder-neck pain in ochard farmers harvesting pears and apples. Ergonomics, 28(4), 700-706. 100 [68] Scott, G.B., and Lambe, N.R., 1996. Working practices in a perchery system, using the OVAKO Working posture Analysing System (OWAS). Applied Ergonomics, 27 (4), 281-284. [69] Silverman, B. W., 1985. Density Estimation for Statistics and Data Analysis. London New York: Chapman and Hall. Van Wely, P., 1970. Design and disease. Applied Er- gonomics, 1, 262269. [70] Stevens, S. S. (1957). On the psychophysical law. Psychological Review, 64(3), 153-181. doi:10.1037/h0046162 [71] Van Hauwermeiren M and Vose D., 2009. A Compendium of Distributions. [ebook]. Vose Software, Ghent, Belgium. Available from www.vosesoftware.com. Accessed 18/02/2013. [72] Vasu, M. and Mital, A. 1999. Evaluation of the of the validity of anthropometric design assumptions. International Journal of Ergonomics, 26, 19-37. [73] Vihma, T., Nurminen, M., and Mutanen, P., 1982, Sewing machine operators? work and musculoskeletal complaints. Ergonomics, 25(4), 295-298. [74] Wiker, S. F., Cha n, D. B. and Langolf, G.D., 1990. Shoulder postural fatigue and discomfort, International Journal of Industrial Ergonomics, 5, 133-146. [75] Zhang, L., Helander, M.G., and Drury, C.G., 1996. Identifying factors of comfort and discomfort in sitting. Human Factors, 38 (3), 377-389. 101 Appendices 102 Appendix A Informed consent for the two experiments conducted for the purposes of this dissertation. 103 Auburn University Auburn University, Alabama 36849-5346 Department of Telephone: (334) 844-1424 Industrial and Systems Engineering Fax: (334) 844-1371 3310 Shelby Centre (NOTE: DO NOT SIGN THIS DOCUMENT UNLESS AN IRB APPROVAL STAMP WITH CURRENT DATES HAS BEEN APPLIED TO THIS DOCUMENT.) INFORMED CONSENT You are invited to participate in a research study to study the relationship between perceived discomfort of static posture holding and posture holding time. The study is being conducted by Jack Ogutu, Graduate Student, under the direction of Dr. Woojin Park, Assistant Professor, in the Auburn University Department of Industrial Engineering. You are invited as a possible participant because you are a male adult aged between 19 to 48 years, and, a) You have no noted history, symptoms and/or risk factors associated with musculoskeletal disorders and, currently have no back pains or joint pains, and, b) You are willing to complete health and fitness-related questionnaires to establish your level of habitual physical activity and aerobic fitness, and will have physical measurements taken. This information will be used to determine your eligibility to participate. If you meet the criteria above and the information we collect about you indicates that you will most likely be at low risk for injury, you will be qualified to participate. What will be involved if you participate? Your participation is completely voluntary. If you decide to participate in this research study, you will be asked to come to the Auburn University?s Ergonomics Lab, Shelby Centre for pre-participation evaluation. We will take your blood pressure and measure your height and weight. In addition you will complete the health questionnaires as noted above. If you qualify to participate, you will be asked to perform 12 pre-determined Posture Holding Tasks (holding a box that weighs 4.4 pounds) over a number of sessions. Your total time commitment will be approximately 3 hours including the 1 hour pre-participation evaluation testing. To prepare for the experiment you will be asked to refrain from strenuous physical activities or exercises on the day before and the day of the experimental trial. On the day of the session you will, a) Be asked to put on tight-fitting clothing made of nylon and spandex (coverall). b) Have reflective markers attached to the right shoulder, the back of the left hand and on the right knee kneecap to monitor your motions. c) Be audio-recorded as you describe your perceived stress during the trials. d) Complete a set of posture holding tasks. Participant?s Initials___________________ Page 1 of 2 Figure A.1: The IRB consent form for study 1 104 Are there any benefits to yourself or others? There are no direct benefits to participants. However your participation may benefit others by protecting workers from excessive work-related postural stress. Are there any risks involved? There is risk of injury from dropping the 4.4 pounds box on your foot. Other risks may include muscle spasms, soreness, back pain, strain and sprain. In the event of an injury that requires medical attention, all expenses will be your responsibility. You may need to check in advance that your insurance will cover any injuries. If you sustain an injury, we can share information about where you might receive medical attention. Will you receive compensation for participating? For full participation, you will receive a check for $180 in the mail. However, if for any reason you quit the study, your compensation will be pro-rated based on hours completed. Information to send your compensation to you will be collected in a separate file and will not be linked to your study responses. If you are not an AU student, you will need to register with Auburn University as a vendor using your SSN (a link will be sent to you separately for registration) for your compensation to be processed. For AU students your student ID will be required for compensation to be processed. If you change your mind about participating, you can withdraw at any time by notifying the experimenter. If you choose to withdraw, your data can be withdrawn as long as it is identifiable. Once you?ve submitted your data, it will be kept confidential until the end of the study when your contact information will be destroyed. Your decision about whether or not to participate or to stop participating will not jeopardize your future relations with Auburn University or the Department of Industrial Engineering. Any data obtained in connection with this study will be made anonymous at the end of the study and will then be kept indefinitely. We will protect your privacy and the data you provide. After your contact information is destroyed, your data will be anonymous. Information collected through your participation may be published in a professional journal. If you have questions about this study, please ask them now or contact the Investigator at 334-750-4268 or by e-mail at joo0002@auburn.edu, or the Faculty advisor, Dr Woojin Park at wzp0006@auburn.edu or 334-844-1311. If you have questions about your rights as a research participant, you may contact the Auburn University Office of Research and Compliance or the Institutional Review Board by phone (334)-844- 5966 or e-mail at hsubjec@auburn.edu or IRBChair@auburn.edu. HAVING READ THE INFORMATION PROVIDED, YOU MUST DECIDE WHETHER OR NOT YOU WISH TO PARTICIPATE IN THIS RESEARCH STUDY. YOUR SIGNATURE INDICATES YOUR WILLINGNESS TO PARTICIPATE. A copy of this document will be given to you to keep. ____________________________________ _____________________________________ Participant's signature Date Investigator obtaining consent Date ___________________________________ _____________________________________ Printed Name Printed Name Participant?s Initials Page 2 of 2 Figure A.2: The IRB consent form for study 1 105 Auburn University Auburn University, Alabama 36849-5346 Department of Telephone: (334) 844-1424 Industrial and Systems Engineering Fax: (334) 844-1371 3310 Shelby Centre (NOTE: DO NOT SIGN THIS DOCUMENT UNLESS AN IRB APPROVAL STAMP WITH CURRENT DATES HAS BEEN APPLIED TO THIS DOCUMENT.) INFORMED CONSENT You are invited to participate in a research study develop an efficient method of determining an individual?s perception of posture stress while performing static single-handed box holding tasks .The study is being conducted by Jack Ogutu, Graduate Student, under the direction of Dr. Woojin Park, Assistant Professor, in the Auburn University Department of Industrial Engineering. You are invited as a possible participant because you are an adult aged between 19 to 48 years, and, a) You have no noted history, symptoms and/or risk factors associated with musculoskeletal disorders and, currently have no back pains or joint pains, and, b) You are willing to complete health and fitness-related questionnaires to establish your level of habitual physical activity and aerobic fitness, and will have physical measurements taken. This information will be used to determine your eligibility to participate. If you meet the criteria above and the information we collect about you indicates that you will most likely be at low risk for injury, you will be qualified to participate. What will be involved if you participate? Your participation is completely voluntary. If you decide to participate in this research study, you will be asked to come to the Auburn University?s Ergonomics Lab, Shelby Centre for pre-participation evaluation. We will take your blood pressure and measure your height and weight. In addition you will complete the health questionnaires as noted above. If you qualify to participate, you will be asked perform short-period (20 seconds) one-handed static object holding trials for 180 pre-generated posture-hand load weight combinations, the boxes weigh 4.4 pounds and 8.8 pounds. Your total time commitment will be approximately 6 hours (= 3 sessions x 2hr /session). Each session will be conducted on a separate day. To prepare for the experiment you will be asked to refrain from strenuous physical activities or exercises on the day before and the day of the experimental trial. On the day of the session you will, a) Be asked to put on tight-fitting clothing made of nylon and spandex (coverall). b) Have reflective markers attached to the right shoulder, the back of the left hand and on the right knee kneecap to record your motions with a motion capture system. c) Be audio-recorded as you describe your perceived stress during the trials. d) Complete a series of 20 seconds holding tasks. Are there any benefits to yourself or others? There are no direct benefits to participants. However your participation may benefit others by protecting workers from excessive work-related postural stress. Participant?s Initials ____________ Page 1 of 2 Figure A.3: The IRB consent form for study 2 106 Are there any risks involved? There is risk of injury from dropping the 4.4 pound and 8.8 pound box on your foot. Other risks may include muscle spasms, soreness, back pain, strain and sprain. In the event of an injury that requires medical attention, all expenses will be your responsibility. You may need to check in advance that your insurance will cover any injuries. If you sustain an injury, we can share information about where you might receive medical attention. Will you receive compensation for participating? For full participation, you will receive a check for $200 in the mail. However, if for any reason you quit the study, your compensation will be pro-rated based on hours completed. Information to send your compensation to you will be collected in a separate file and will not be linked to your study responses. If you are not an AU student, you will need to register with Auburn University as a vendor using your SSN (a link will be sent to you separately for registration) for your compensation to be processed. For AU students your student ID will be required for compensation to be processed. If you change your mind about participating, you can withdraw at any time by notifying the experimenter. If you choose to withdraw, your data can be withdrawn as long as it is identifiable. Once you?ve submitted your data, it will be kept confidential until the end of the study when your contact information will be destroyed. Your decision about whether or not to participate or to stop participating will not jeopardize your future relations with Auburn University or the Department of Industrial Engineering. Any data obtained in connection with this study will be made anonymous at the end of the study and will then be kept indefinitely. We will protect your privacy and the data you provide. After your contact information is destroyed, your data will be anonymous. Information collected through your participation may be published in a professional journal. If you have questions about this study, please ask them now or contact the Investigator at 334-750-4268 or by e-mail at joo0002@auburn.edu, or the Faculty advisor, Dr Woojin Park at wzp0006@auburn.edu or 334-844-1311. If you have questions about your rights as a research participant, you may contact the Auburn University Office of Research and Compliance or the Institutional Review Board by phone (334)-844- 5966 or e-mail at hsubjec@auburn.edu or IRBChair@auburn.edu. HAVING READ THE INFORMATION PROVIDED, YOU MUST DECIDE WHETHER OR NOT YOU WISH TO PARTICIPATE IN THIS RESEARCH STUDY. YOUR SIGNATURE INDICATES YOUR WILLINGNESS TO PARTICIPATE. A copy of this document will be given to you to keep. ____________________________________ _____________________________________ Participant's signature Date Investigator obtaining consent Date ___________________________________ _____________________________________ Printed Name Printed Name Participant?s Initials ____________ Page 2 of 2 Figure A.4: The IRB consent form for study 2 107 Appendix B IRB Participant screening document 108 Occupational Safety & Ergonomics Laboratory Dept. of Industrial & Systems Eng. Auburn University Subject # __________ Date ________________ Subject Health Questionnaire a) Age_________ b) Male________ Female________ c) Height____________ d) Weight____________ e) Blood Pressure___________ (BP >140/90 will be excluded from this study). Note: If you respond ?Yes? to any question you will be excluded from this study. General Medical Questions 1. Have you ever passed out while exercising? YES NO 2. Have you ever passed out for any reason? YES NO 3. Have you ever had chest pains while exercising? YES NO 4. Has anyone in your family died before age 50 secondary to a stroke or Heart Attack? YES NO 5. Have you ever been told you have arthritis? YES NO 6. Do you have any pins, screws or other implants? YES NO 7. Are you now taking any medications for blood pressure, ADD/ADHD or any medications that affect your circulatory or neurological systems? YES NO Orthopedic Medical Questions 8. Have you ever sustained neck pain, e.g Stinger, Pinched Nerve, etc? YES NO 9. Have you ever had numbness, burning, or sharp pain in your arms/hands? YES NO Figure B.1: The IRB screening document 109 10. Have you ever injured your back? YES NO 11. Have you ever been told you have a congenital spinal defect? YES NO 12. Have you ever been told you have a spondyloliothesis/spondylolysis? YES NO 13. Have you ever been told you have a bulging or herniated disc? YES NO 14. Have you ever had a significant hip or thigh injury? YES NO 15. Have you ever had a significant knee injury? YES NO 16. Have you ever had a significant ankle injury? YES NO 17. Have you ever had a significant foot or toes injury? YES NO 18. Have you ever dislocated a joint? YES NO 19. Have you ever broken a toe? YES NO Is there any other medical condition, disease, injury, illness or hospitalization that might prevent you from participating? Please explain ? include dates and medications. Figure B.2: The IRB screening document. 110 Appendix C IRB Participant screening document. 111 Appendix D PI, Ogutu, Jack 30 ?An Efficient Method for Modeling an Individual?s Perception of Postural Stress? EMERGENCY ACTION PLAN Personnel Two of the following researchers will be present during all testing sessions. In addition, a qualified researcher will be present to conduct first aid at all sessions. Jack Ogutu, First aid (cell number-334-750-4268) Woojin Park, First aid (cell number-334-707-5229) Communication Auburn University Medical Clinic: 844-4416 East Alabama Medical Centre (Emergency Department): (334) 528-1150 East Alabama Medical Centre (Hospital): (334) 749-3411 Lee County EMS (Ambulance): (334) 749-8504 Emergency: 911 TigErgonomics Laboratory: (334) 844-1432 Emergency Equipment A first aid kit is present during all testing sessions. An AED is present at the entrance to each stairwell in the Shelby Center. In addition, as ramp/full access entrance is available for any emergency responders to enter the facility. Emergency Procedures 1. Use emergency equipment (if applicable), and perform emergency CPR and first aid ? if qualified to do so. 2. Instruct investigators to call 911 or EMS and provide the following information: - Who you are - General information about the injury or situation - Where you are: 345 west Magnolia St. Shelby Center for Engineering Technology. We are located in the 3rd Floor. Come in the front of the Shelby center to the right wing of the building. Take the elevator to the 3rd floor of the building, the Lab is in Room 3325. - Any additional information - BE THE LAST TO HANG UP! 3. Meet the ambulance and direct it to the site you described in the phone call 4. If directed, assist EMS with care of injured person. 5. The primary research investigator must report the adverse event to the IRB via the Office of Human Subjects Research. Physical Street Address: 3325 Shelby Center for Engineering Technology Directions to the front of Shelby Center for Engineering Technology ? 345 west Magnolia St. Shelby Center for Engineering Technology. We are located in the 3rd Floor. Come in the front of the Shelby center to the right wing of the building. Take the elevator to the 3rd floor of the building, the Lab is in Room 3325. Figure C.1: Emergency Action Plan 112 Appendix D Manuscript 1: The Thirteen Inadequate Discomfort-Time Pro les. 113 Figure D.1: The Discomfort-time sequences that could not adequately t the power function form. 114 Figure D.2: The Discomfort-time sequences that could not adequately t the power function form. 115