Assessment of Two Passive Back-Support Exoskeletons in a Simulated Precision Manual Assembly Task

Abstract

Low back pain (LBP) remains the most prevalent and costly work-related disability in the United States, accounting for ~40% of annual musculoskeletal disorders (BLS, 2018) and imposing an economic burden of over $50 billion (Davis, 2012). Many intervention approaches have been explored to reduce the physical requirements of occupational tasks. Examples include training in work methods, modifying work stations, re-organizing work processes, and using mechanical aids such as cranes and power-lift tables (Chaffin et al., 1999; Lavender et al., 2013; Madinei et al., 2018). While these approaches can be effective, they can also be excessively costly or even infeasible for certain work environments (Graham et al., 2009). Industrial back-support exoskeletons (BSEs) – designed to augment the back and hip muscles – have been introduced as an alternative intervention to reduce the physical demands on the back muscles and consequently mitigate the risk of LBP (De Looze et al., 2016). However, there is limited evidence regarding the efficacy of BSEs in work scenarios that require sustained and/or non-neutral trunk bending (such as manual assembly), which are well-known LBP risk factors (e.g., Norman et al., 1998; Fathallah et al., 2008). The efficacy of two passive BSE designs (i.e., BackX™ and Laevo™) was examined by quantifying trunk extensor muscle activity during a lab-based simulation of a precision manual assembly task. Both devices incorporate a passive torque generation mechanism about the hip joint that is intended to augment the torso extensor muscles. Yet, the devices differ in specific design characteristics, such as major body anchor points for the torque generation mechanism (upper back, waist, and thigh [BackX™ AC] vs. chest, waist, and thigh [Laevo™]). Eighteen (gender-balanced) participants, with no recent musculoskeletal injuries or disorders, completed a simulated assembly task using a “grooved pegboard” (Lafayette Instruments, IN, USA) in 20 different pegboard locations. These locations were defined by four different heights (waist, knee, ankle, and below floor levels), three horizontal distances (0, 20, and 40 cm away from the feet), and three orientation angles (0°, 45°, and 90° to the right of the mid-sagittal plane). For a given pegboard location condition, participants were asked to complete the assembly task “as quickly as possible”. Muscle activity was monitored bilaterally from two trunk extensors (i.e., iliocostalis lumborum [ILL] and thoracic erector spinae [TES]) using a telemetered surface electromyography (EMG) system (TeleMyo Desktop DTS, Noraxon, AZ, USA). Note that before performing the assembly task in any of the pegboard location conditions, participants completed trials of maximum isometric voluntary contractions for those muscle groups for normalizing EMG. Outcome measures were the median level of left-side back muscle activity (LBM = mean of 50th percentile normalized EMGs [nEMGs] of left TES and ILL) and the median level of right-side back muscle activity (RBM). Overall, BackX™ use (vs. Laevo™) led to a larger reduction in median levels of back muscle activity levels (≤ 37.9% vs. ≤ 23.9% reduction), and a significant reduction in activity was observed in a larger set of conditions (15 vs. 7 conditions). Additionally, the largest reductions when using BackX™ were found at the ankle level (≤ 38% vs. no reductions for Laevo™), followed by knee level (≤ 32% vs. ≤ 24% for Laevo™), waist level (≤ 30% vs. ≤ 14% for Laevo™), and below floor level (≤ 29% vs. ≤ 10% for Laevo™). Our findings suggest that the beneficial effects of a BSE can be task-specific and that such effects may also be specific to BSE design approaches. Notably, the effects of BSE use found here can be considered practically meaningful. For context, the magnitude of median levels of bilateral low-back muscle activities ranged from ~4-18% nEMG, and using a BSE yielded reductions up to ~5% nEMG, depending upon the specific BSE and task condition. More research is warranted, though, to characterize the task specificity and generalizability of different BSE design approaches in terms of physical demands and task performance.