Currently, work-related musculoskeletal disorders (WRMSDs) are the most prevalent occupational health problem worldwide [1], and they are also the most common in the European Union (EU) ^[2][3][2,3], affecting three out of five workers [4]. Indeed, it is considered a public health problem [5] with multifactorial causes ^[6][7][6,7], resulting from the complex interaction between individual, biomechanical, organisational and psychosocial risk factors ^[6][8][6,8].

WRMSDs affect individuals in all aspects of their lives—as well as companies, society and the economy ^[2][9][10][2,9,10]—as they result in higher healthcare costs, reduced productivity, increased absenteeism, lower job satisfaction, and reduced physical, psychological, and social well-being of workers ^[7][11][7,11]. In fact, some disorders have a lasting impact on the worker’s life, limiting them in their daily activities and preventing their return to work due to permanent disability caused by pain and decreased functionality [5].

Recently, an increase in WRMSDs has been observed due mainly to mechanical overload [1]. Prolonged inappropriate working postures, tasks with high physical demands, repetitive and meticulous work gestures, and an intense work pace with few breaks have been reported as the main causes of WRMSDs ^[3][8][9][3,8,9]. Additionally, the demographic change in the workforce [12] has led to an increase in the number of older workers, which may have also contributed to the increase in WRMSDs ^[1][2][1,2].

Given the musculoskeletal system overload [1], real-time monitoring and postural correction in the workplace [13] are urgent. This will minimise the harmful effects of some postures, resulting from the misalignment of body segments with the line of gravity [14]. Posture is a highly complex, variable, and dynamic system that can respond to minimal psychophysical and socioenvironmental perturbations [15]. Due to these features, wearable technology has emerged as a viable alternative with high potential for real-world context implementation ^[16][17][18][16,17,18]. The projection of wearables, also referred to in the literature as wearable sensors and defined as electronic devices integrated into clothing and/or other accessories that comfortably adapt to the human body [19] would seem to be a valid proposal for improving working conditions, early identification of WRMSDs risks, increased work efficiency, and promoting well-being [20].

Inertial measurement unit (IMU) is the common underlying technology for most wearables ^[7][21][22][7,21,22]. It is possible to identify static and dynamic improper postures that are maintained for long periods ^[23][24][23,24] through discrete and continuous monitoring of body posture in real-world settings ^[21][25][21,25] in a shorter timeframe/in a shorter time [5]. Other advantages of wearables that make them robust for integration in monitoring work activity include objective, reliable, and accurate results [26] which provide a trustworthy and realistic assessment of work-related conditions [27]; low cost; lightweight design; small size; portability; and energy efficiency [28].

Furthermore, some wearables have the feature to provide real-time corrective sensory feedback when adopting inadequate postures [3]. This feedback can be auditory (typically conveyed through diverse auditory channels), visual (usually displayed by screens or projectors), haptic (application of vibratory stimulus), or a combination of these, providing information based on performance or outcome ^[29][30][29,30]. Haptic feedback stands out as the most common choice ^[9][31][32][9,31,32] and also as the most advantageous option given its discreet nature [29]. It contrasts with auditory and visual feedback, which can be perceived by other workers, affecting their concentration [9]. Regarding the feedback signal, it can be provided at the end of the task, referred to as terminal feedback, or in real-time, known as concurrent feedback [29]. The latter has the advantage of promoting immediate changes in work postures [33]. However, regardless of the type of corrective feedback, it promotes greater postural self-awareness, allowing for the minimisation of inadequate postures and, consequently, reducing the musculoskeletal overload [13]. This factor emphasises the potential for the urgent implementation of wearables in different work contexts to reduce the incidence of WRMSDs and the associated healthcare costs ^[9][32][9,32]. Nevertheless, still in their early stages, wearables present disadvantages and/or challenges such as accuracy, technical functionality, and usability [30]. Factors such as battery life, long-term comfort, preparation time (donning and doffing; changing or recharging batteries) and the stability of wireless communication contribute to workers’ reluctance to use such devices ^{[2][30][34][35]}[2,30,34,35]. Recent studies have tried to address the challenges of wearables by improving the usability and effective monitorization of human movement, as well as the autonomy of wearables, which are crucial for practicality and commercial viability, with the aim of establishing wearables as common tools in workplace settings [36]. The potential of wearables in preventing and minimising WRMSDs is supported by growing scientific evidence on postural monitoring wearables in various contexts ^{[20][36][37][38][39]}[20,36,37,38,39]. Despite advances, challenges remain in wearables for simultaneous postural monitoring and feedback in real-world scenarios, emphasising the need for continued research. Furthermore, the importance of wearables in minimising WRMSDs positions them as an emerging field of research. It is important to understand and summarise all the current evidence to contribute to the development of new methods to promote the worker’s health and quality of life.

Photogrammetric methods, which are the gold standard for posture analysis, consist of approaches that are more geared towards laboratory use and are not suitable for monitoring and correcting workers’ postures in their daily lives [22]. On the other hand, indirect methods such as questionnaires or observational methods like REBA or RULA are, according to the available literature, the most used in the workplace context to assess postures and related factors ^[45][46][63,67]. Consequently, the need to develop alternative methods for real-time objective assessment of work tasks and the subsequent adoption of appropriate postures was created ^[46][67]. However, based on existing evidence, it is known that workers are often unaware of their posture and frequently adopt incorrect behaviours due to time constraints, task demands, and the need to meet productivity goals [22]. Truly, there is a lack of consensus on correct posture [13], which is generally described as considering the alignment of different body segments about the line of gravity with minimal energy expenditure ^[13][47][13,68].

Most sensors used in the included studies were at the prototype stage despite the growing trend towards commercialisation of wearable devices [22]. In fact, although there is already a wide range of commercially available wearable devices today, most are aimed at physical activity monitoring, posture and physiological parameters, in contrast to the apparent lack of commercial wearable systems that combine postural monitoring and feedback [22]. This seems to be a gap for their integration in the workplace context [2]. This may be due to difficulties not only in accessing such devices but also to the fact that, in most cases, they do not meet usability parameters, which can consequently hinder the acceptance of these instruments by workers and companies, and actually provide data incongruent with reality ^[2][34][2,34]. 

Despite the device placement, also the usability, the comfort and accuracy are key factors in worker acceptance, and current evidence reinforces its importance ^{[5][7][48][49]}[5,7,73,74]. Indeed, the sensors attachment is a fundamental role in its usability. Therefore, to meet this requirement, the sensor attachment must fulfil three essential conditions, namely being imperceptible to the worker; having intuitive use; and providing quick, reliable, and easily interpretable information [21]. These criteria are crucial to ensure the effectiveness of wearables, alongside worker acceptance, and can be integrated into accessories or clothing [25]. 

Other sensor attachment methods can include direct placement on the skin using electrodes that incorporate EMG ^[50][76], adhesive tape or elastic straps ^[33][46][33,67], integration into smartwatches, or even integration into personal protective equipment (PPE) such as helmets and visors ^[51][52][53][60,77,78]. The latter form of attachment is consistent with Yang et al.’s ^[54][65] study on construction workers, who used sensors attached to PPE, specifically helmets and safety harnesses or vests. In fact, Choi et al. ^[43][58] argue that this is the most suitable location for sensor placement in this occupational activity, as it does not appear to interfere with the work, which can increase usability.

2.3. Occupation and Work Tasks

Healthcare professionals such as nurses, dentists, healthcare assistants, and unspecified healthcare professionals were the most studied populations on posture monitorisation and feedback using wearables. Indeed, healthcare professions typically involve demanding work schedules and precise, meticulous, and repetitive tasks, often requiring prolonged static postures that contribute to neuro-musculoskeletal strain and fatigue, particularly among dentists ^[55][56][57][79,80,81]. This fact is also supported by studies that have found a high prevalence of musculoskeletal disorders in this group ^[58][59][60][70,82,83], particularly among those with more than 10 years’ clinical practice and working more than 40 h per week ^[60][83]. Jacquier-Bret and Gorce ^[58][70] also found that dentists are among the professionals most exposed to musculoskeletal disorders, with a higher prevalence of symptoms in the spine, particularly in the cervical and lumbar regions, which is also supported by Blume et al. ^[59][82]. In addition, dental students are also susceptible to develop musculoskeletal disorders, and there are studies that support this evidence by linking this susceptibility to the adoption of more sedentary behaviours due to the use of new technologies that decrease their mobility ^[60][61][83,84]. On the other hand, Blume et al. ^[59][82] describe an association between musculoskeletal disorders and poor posture adopted by these students during dental procedures, particularly when these involve static postures, especially in the cervical region, due to the limited visibility of the patient’s mouth, which leads to the maintenance of an extended and protracted cervical posture. Hence, there is an urgent need to bring visibility to wearables that incorporate corrective feedback, as they can aware changes in the daily work routines and, consequently, in the quality of life and work of healthcare professionals. This could result in fewer injuries, lower absenteeism rates, reduced healthcare costs, and increased productivity ^[55][79]. In addition to healthcare professionals, there are other fields, particularly in the construction industry, who have a high prevalence of WRMSDs due to daily exposure to excessive effort, sustained and incorrect postures, handling and carrying heavy loads, repetitive tasks, and vibration from work tools ^[13][14][57][13,14,81]. Based on scientific evidence, it is widely recognised that the construction industry is one of the most dangerous and physically demanding occupations in terms of ergonomics, with a high rate of early retirement and a significant proportion of ageing workers, which increases susceptibility to WRMSDs ^[20][62][20,85].

2.4. Feedback Source and Type

Wearables have emerged as a solution for the prevention of WRMSDs due to their versatility, particularly because of the real-time feedback they can provide ^[44][59]. This feedback should provide personalised information, and it is important to ensure that workers understand and adjust their posture based on the stimulus received ^[63][86]. Currently, there are different types of sensory feedback, including haptic, visual, and auditory feedback [9], with haptic feedback being considered the most suitable and common approach ^{[9][31][32][64]}[9,31,32,72]. This is consistent with the results of the present study, in which half of the authors used haptic feedback by generating vibratory stimuli. On the other hand, Lee et al. [7] found different results, with sound being the preferred variable. However, considering that sound feedback has a greater potential to distract the worker, the choice of haptic feedback seems to be more appropriate ^[64][72]. On the other hand, Wang et al. [32] recommended the combined use of multiple types of feedback to minimise potential disadvantages and fill gaps. This recommendation is in line with the study by Bootsman et al. ^[65][51], which used all three types of sensory feedback. Although this approach has its advantages, it is important to consider the suitability of the region and context for monitoring and applying feedback. Given that both auditory and haptic feedback do not require visual attention during the task [32], they appear to be a valuable combination. However, in this specific case, the auditory feedback may have been drowned out by the noise of other equipment in the neonatal intensive care unit, which is typically loud. It is therefore important to carefully assess the pertinence of the feedback used, taking into account the aforementioned factors [20]. Consequently, the use of auditory feedback in the study by Yan et al. ^[66][48] may not have been the most suitable choice, as it involved construction tasks where, in addition to the existing noise, some hazardous tasks require full concentration. This may be the reason why the studies by Lind et al. ^[40][67][52,54] opted for haptic feedback for industrial workers.

2.5. Results after Feedback Application

The included studies demonstrated that there was generally an improvement in posture after the application of haptic feedback in real-world settings, which is congruent with the studies by Kuo et al. ^[68][89] and Lind et al. ^[53][78], although these were conducted in laboratory settings. The studies by Ribeiro et al. ^[69][43], Yan et al. ^[66][48], and Doss et al. ^[70][49] reported postural improvements following auditory feedback, which is supported by the laboratory study by Boocock et al. ^[71][90]. On the other hand, Ribeiro et al. ^[72][53] concluded that this type of feedback did not promote improvement during work, suggesting that haptic feedback may have been more appropriate for this class of professionals. The terminal visual feedback identified in the studies by Thanathornwong et al. ^[73][45] and Thanathornwong et al. ^[74][44] also demonstrated significant postural improvements, although the stimulus was presented to the worker only at the end of the task. Based on these findings, it can be inferred that the choice of feedback depends on the tasks and the work context being evaluated.