Trunk kinematic variability as a function of time during the early phase of a repetitive lifting task

Lift‐to‐lift variability occurs in repetitive lifting tasks due to alterations in the lifting techniques used by the lifter, resulting in variability in lower back tissue loading. Understanding how trunk variability changes with time in the initial phases of a lifting bout may provide insights into the risk of injury during work startup. The purpose of this study was to quantify the variation of lifting kinematics and kinetics during the initial phase of a lifting bout. Twenty participants performed a repetitive lifting task continuously for 30 min. The load was equivalent to 10% of each participant's body weight and lifting was done at a rate of six lifts/ min. Kinematic variables (three‐dimensional range of motion, angular velocity, and angular acceleration) of the trunk were measured using the Lumbar Motion Monitor and a dynamic biomechanical model estimated peak L5/S1 moment and spine compression. The variances of these variables were compared across 10‐min intervals: 0–10 min, 10–20 min, and 20–30 min. Results indicate a significant reduction in the variance of the peak sagittal acceleration, the sagittal range of motion, the transverse range of motion, peak sagittal moment, and peak spine compression between the first and second time intervals, followed by no significant change in variance between the second and third intervals. The downward trend in variation of these kinematic and kinetic variables suggests an initial adjustment period as the lifters reach a steady state of their lifting technique. The reduced variance of spinal loading may reduce the probability that a tissue tolerance is exceeded.

purpose of this study was to quantify the variation of lifting kinematics and kinetics during the initial phase of a lifting bout. Twenty participants performed a repetitive lifting task continuously for 30 min. The load was equivalent to 10% of each participant's body weight and lifting was done at a rate of six lifts/min. Kinematic variables (three-dimensional range of motion, angular velocity, and angular acceleration) of the trunk were measured using the Lumbar Motion Monitor and a dynamic biomechanical model estimated peak L5/S1 moment and spine compression. The variances of these variables were compared across 10-min intervals: 0-10 min, 10-20 min, and 20-30 min. Results indicate a significant reduction in the variance of the peak sagittal acceleration, the sagittal range of motion, the transverse range of motion, peak sagittal moment, and peak spine compression between the first and second time intervals, followed by no significant change in variance between the second and third intervals. The downward trend in variation of these kinematic and kinetic variables suggests an initial adjustment period as the lifters reach a steady state of their lifting technique. The reduced variance of spinal loading may reduce the probability that a tissue tolerance is exceeded.
low back pain, repetitive lifting, trunk kinematics/kinetics, variability 1 | INTRODUCTION Low back pain (LBP) is recognized as a significant and costly problem worldwide that particularly impacts workers involved in occupations which require repetitive lifting exertions. Guo et al. (1999) showed that LBP contributed to approximately 101.8 million days in lost productivity annually due to absenteeism from work. This lost productivity was shown to have contributed to an estimated revenue loss of $7.4 billion annually for employees aged 40-65 (Ricci et al., 2006). In addition to lost revenue, the cost of managing LBP is high, exceeding $100 billion per year in 2006 in the United States (Katz, back disorders, Bernard and Fine (1997) stated that there was strong evidence of an association between low back disorders and workrelated lifting and forceful movements and they found that there was evidence of a relationship between awkward trunk postures and low back disorders.
There is some intriguing evidence of a relationship between certain temporal aspects of work life and occupational injury. In a synthesis of data from the US Bureau of Labor Statistics, Brogmus (2007) showed that the lost time injury rate was greatest at the start of the work week, reached its nadir on Wednesday, and then increased again on Thursday and Friday. In contrast, Wigglesworth (2006) evaluated data from the Australian Bureau of Statistics which showed a continuous downward trend in the percent of occupational injuries by day, with Monday contributing approximately 23% of the weekly injuries and Friday contributing 17%, with a steady downward trend connecting these two endpoints. Exploring this data set from a different perspective, Wigglesworth (2006) also showed some interesting trends in the numbers of occupational injuries by time of day. He showed that the number of injuries peaked in the 8:30-9:30 a.m. timeframe, dropped sharply towards the middle of the work day and then increased into the middle of the afternoon before falling off again towards the end of the work day. While these studies did not focus exclusively on musculoskeletal disorders, but more generally on occupation injury, there appear to be some time-related mechanisms whereby musculoskeletal disorders might contribute to these trends. These mechanisms would include end-of-week and endof-day effects such as muscular fatigue and cumulative trauma, as well a beginning-of-week and beginning-of-day effects such as warmup effects (Woods et al., 2007), adjustment to working conditions, and so forth.
Interestingly, warm-up and stretching routines before task performance have been reported to reduce the likelihood of injury occurrence for a number or reasons. First, warm-ups/stretches reduce tissue viscosity and enhance flexibility which results in smoother contractions (Safran et al., 1989;Shellock & Prentice, 1985). In addition, the body heat generated during warm-ups increases dissociation of oxygen from hemoglobin for muscle contraction (Safran et al., 1989) and a one-degree rise in muscle temperature has been shown to increase length to failure in rabbit hindleg (Safran et al., 1988). These results highlight the importance of understanding the early phases of a physically demanding task.
Variability in a lifting technique employed by a manual material handler could be one potential source for increased injury risk. Epidemiological studies have showed that trunk kinematics contribute to LBP development (Marras et al., 1993). Perhaps information on this variability trend will explain the increased risk of occupational injuries at the onset of the work day and the work week reported in the epidemiological data, and also influence our approach to injury risk assessment in manual task performance. Variability in these important trunk kinematic variables during repetitive lifting exertions will generate distributions of joint (i.e., spine) reaction forces leading to a scenario where an appreciable percentage of the lifts may create forces and moments that exceed the threshold level within the spine tissues (Granata et al., 1999). Furthermore, there has been some published evidence of a relationship between some of these temporal effects (specifically fatigue) and an increase in the variability of motions (Bauer et al., 2017;Sedighi & Nussbaum, 2017) intuitively agreeing with the previously cited epidemiologic findings (Brogmus, 2007;Wigglesworth, 2006). Moreover, a comprehensive review by Srinivasan and Mathiassen (2012) highlighted the importance of understanding the concept of motor variability in work task performance and injury prevention. One specific recommendation from this study was a recognition of the importance of the study of the temporal aspects of work task performance. For these reasons, focusing on trunk kinematic and trunk kinetic variation could provide insight to LBP development that results from repetitive lifting.
While increasing kinematic variability as a function of fatigue has been established in the literature, much less is known about trunk kinematic and kinetic variability during the early stages of a repetitive lifting task, as might be seen as a worker begins their work day. Hence, the objective of the current study was to investigate the effect of time on the variability of trunk kinematics and kinetics during the early phase of a repetitive lifting task. Trunk motion and loading variability was hypothesized to be high at the onset of task performance, followed by a gradual reduction in this variability as the lifter settles into their natural lifting rhythm.

| Participants
Twenty college students (10 males and 10 females, aged 25 ± 3 years, height 171.6 ± 10.2 cm, and body mass 71.5 ± 19.2 kg), were recruited for this study. None of the participants had a history of low back pain or any chronic hip, shoulders or leg pain. Furthermore, none was currently experiencing pain in these areas of their body.
Participants had no professional experience in manual materials handling. Before participating in the study, each participant provided written informed consent (document approved by the Institutional Review Board of Iowa State University).

| Experimental task apparatus
The load used in the repetitive lifting task was a crate (33 cm [width] × 33 cm [length] × 29 cm [height]) with handles for good coupling. It was filled with a stable load corresponding to 10% (7.1 ± 1.9 kg) of the whole-body mass of each participant. Body weight was used as a normalization parameter because it was readily available data and has been shown to be effective at scaling for muscle strength (Hurd et al., 2011). Two sets of roller conveyors were used in this study to provide the beginning and end point of the lift. These conveyors were height adjustable enabling the starting and ending heights to be set relative to participant anthropometry.
The height of the conveyor at the start of the lift was adjusted so that the crate handles were at the knee height of each participant, while the height of the destination conveyor was set so that the crate handles were at elbow height. The experimenter was stationed at the other end of these conveyors and would replace the load on the conveyors at the designated frequency. Finally, a visual analog scale (VAS; 16 cm long) with no fatigue (0 cm) and extreme fatigue (16 cm), was used to capture the subjective assessment of the participants.

| Experimental tasks
When participants arrived at the laboratory, a research assistant provided a concise description of the task and written informed consent was obtained. Anthropometric data (weight, height, elbow height, and knee height) were measured and recorded. Each participant was then guided through a short warmup session which consisted a set of standardized upper extremity and back muscle movement routines to prepare him/her for the lifting task. The LMM was then fitted on the participant as shown in Figure 1. After which they were allowed to select a comfortable position on the stable platform, which would be maintained throughout duration of the repetitive lifting task. For reference, those positions were marked with tapes. Before actual task performance, each participant was familiarized to the lifting task by letting him/her stand in their chosen foot position while lifting the load from the start to the end point in

| Independent and dependent variables
The independent variable in this study, was TIME, which was divided into three levels corresponding to the 1st, 2nd, and 3rd 10-min intervals of the 30-min lifting task (Segment 1: 0-10 min; Segment 2: 10-20 min; Segment 3: 20-30 min). The dependent variables were Plane Moment about the L5/S1 joint (sMOMmax) and Peak Spine Compression (COMP). Finally, the subjective level of fatigue was captured using the visual analog scale before task performance and at the end of the first, second, and third 10 min of lifting.

| Statistical analysis
All statistical analyses were conducted using R (Version 4.0.2). Statistically significant differences between the variances of trunk Note: Bold denotes statistically significant differences between levels at α = .008 and variance values noted with the same letter were not statistically significant.
kinematic variables at the 1st, 2nd, and 3rd 10-min segments were explored using the Levene's test of homogeneity of variance.
The Bonferroni correction was applied by dividing the initial p value of .05 by the number of dependent variables to reduce the probability of occurrence of Type 1 error. Pairwise comparisons of the variances between levels of TIME were performed to further explore the nature of the significant differences in the values of the variance.
The residuals of the subjective fatigue measurement violated the normality assumption thus, the non-parametric Friedman's test was used to test for any statistical differences between back muscle subjective fatigue for the three levels of the independent variable.
The post hoc Nemenyi method was used to explore the pairwise differences between the initial subjective fatigue and the end of each 10-min segment of lifting.

| RESULTS
These results of the analysis of trunk kinematics illustrate that the peak sagittal acceleration, sagittal and transverse range of motion were the trunk kinematic variables that were significantly influenced by TIME (Table 1 and Figures 2 and 3). Exploring these kinematic data a bit further, the present results indicated that there was significantly greater variance in the peak sagittal acceleration in the first 10 min as compared with the second and third 10 min of the task performance. The kinetic data (Table 1 and Figures 4 and 5), likewise, demonstrated that the variability in the moments about L5/ S1 were significantly greater in the first ten-minute bout as compared with the other two bouts. The kinetics of the task were further explored by the calculation of the peak spine compression and the distributions of this spine compression force as a function of TIME are shown in Figures 5-7. Note how the spread of the distribution from the first 10-min period is greater than that of the following periods.
The analysis of the subjective fatigue showed a significant effect of time on subjective muscle fatigue (p < .05; Table 2). These post hoc analysis showed that significant increments were only present between Segment 0 (before task performance) and Segment 2, Segment 0 and Segment 3, and Segment 1 and Segment 3.

| DISCUSSION
The results of this study provide data that supports the hypothesis that variability in lifting kinematics/kinetics is greater in the early phases of a repetitive lifting task than those seen just ten minutes into the task. These changes may point to a dynamic level of physical flexibility and warm-up effects that changes within these early stages of lifting. While in this experiment there was a period of warm-up and familiarization provided, once the experiment began, the participants may have still been getting into their lifting rhythm and gaining task-specific flexibility that would lead to a steady state lifting style over the first 10 min of lifting. The kinematic and kinetic response observed in this study both point to this type of modification in this initial 10-min period.
This result is important because these lifting kinematics and kinetics have a direct impact on the loading of the tissues of the low back, and variability in these parameters will create variability in the loading of the internal structures of the spine. Considering peak sagittal acceleration, for example, research has shown that these accelerations directly impact the required trunk extensor muscle activation (Marras & Mirka, 1990), and the biomechanical impacts of an increased level of variance of this sagittal acceleration in the early phases of the lifting bout has direct impact on peak loading of the spine across lifts (Figure 5a). A previous study by Granata et al. (1999) (Granata et al., 1999). However, the load in this study was standardized to 10% of each participant's body weight (maximum weight = 11.4 kg), while the load used in the previous study was 13.6 kg and 27.3 kg. In addition, our study required 6 lifts/min without controlling how fast the participants performed each lift while the previous study had a "faster than preferred lifting speed" as an independent level, and this could have significantly increased trunk acceleration which could subsequently increase trunk extensor moment and compression forces.
Interpretation of these kinematic results may be attributed to motor variability (MV)-an inherent characteristic of the neuromuscular system to explore and refine movement patterns while interacting with the environment (Krakauer & Mazzoni, 2011). In this regard, the neuromuscular system actively alters its structure, that is, muscle recruitment pattern, and so forth in an effort learn the task for optimum performance. This is associated with increased variability at the on-set of task performance (Wu et al., 2014), hence the increased variability in the variables observed during the first 10 min as compared with the second and third 10 min. While it is conceivable that the initial stretching and lifting practice session might have played a role in the variability trend, the authors do not believe this was the case. The practice section was intended to familiarize the participant with the start and end point of the lifting task, and not the technique to complete the task.
The variability of transverse plane acceleration was found to be insignificant and this could be ascribed to the nature of the lifting task. The task was setup to have a starting height at the knee level and end height at standing elbow height. Given that the participants would be focused on getting the load up to approximately neutral posture before twisting to set load down, the sagittal plane trunk extension was the most difficult and time-consuming part of each lift and study participants may have used a combination of the two studied lifting techniques to bring the load to neutral trunk posture (Bazrgari et al., 2006). Thus, the participants were much more variable with their lifting technique in during the time-consuming trunk extension phase, compared to the significantly shorter trunk twisting phase. This was evident in the significant variability observed in the sagittal trunk acceleration, compared with the transverse plane acceleration.
The findings of this study support previously established general recommendations for the start-up of the work day and may have implications for the use of existing risk assessment tools employed by ergonomists in industry. One general recommendation that is often promoted for work day start up, is the use of muscular warmup and stretching routines before task performance to increase joint range of motion as well as reduce the likelihood of injury occurrence (Mahieu et al., 2007). As such, consideration should be given to providing manual material handling workers a time for warmup and stretching as well as a period of lowered productivity expectations as they gain their rhythm and consistency in their lifting technique. This will enable material handlers to gradually warm up, which enhances blood flow to muscles and tendons for efficient performance and injury prevention (Nakamura et al., 2015). These recommendations are not new, or the result of the current study, but are sound advice that is supported by the results of the current study. Our results shown in Figure 5 indicate that the probability of high spine compression level (right-hand tail of distribution) is increased during the early phases of the lifting bout and preparing the body through warm-ups could be particularly important. In terms of the implications of these results relative to risk assessment tools, the results of this study can be influential in two ways. First, this study has demonstrated that there is variability in the kinematics of trunk mo- material handlers. Second, the lifting task chosen for this study was very controlled in terms of the starting and ending points for each lift. This was done deliberately to create a precise lifting task that would most easily highlight the changes in the variance of the dependent measures. Individuals performing MMH activities on the job, certainly have much more variability in the characteristics of the lifting task and this type of variability would provide another layer of complexity in the assessment of variability in biomechanical loading.

| CONCLUSIONS
The results from this study suggests a temporal impact to the intraindividual variability in manual materials handling tasks. Further exploration of the effects of time, as well as other lifting parameters, on the variability of lifting kinematics and kinetics, may provide a deeper understanding of injury risk and may offer potential avenues for appropriate ergonomic interventions to prevent low back injuries.

CONFLICT OF INTERESTS
The authors declare that there are no conflict of interests.

DATA AVAILABILITY STATEMENT
The data that supports the results of this study and all the tables and figures in this article are available upon reasonable request from the corresponding author.