Ergonomic interventions for preventing work‐related musculoskeletal disorders of the upper limb and neck among office workers

Victor CW Hoe; Donna M Urquhart; Helen L Kelsall; Eva N Zamri; Malcolm R Sim

doi:10.1002/14651858.CD008570.pub3

. 2018 Oct 23;2018(10):CD008570. doi: 10.1002/14651858.CD008570.pub3

Ergonomic interventions for preventing work‐related musculoskeletal disorders of the upper limb and neck among office workers

Victor CW Hoe ^1,^✉, Donna M Urquhart ², Helen L Kelsall ², Eva N Zamri ³, Malcolm R Sim ²

Editor: Cochrane Work Group

PMCID: PMC6517177 PMID: 30350850

Abstract

Background

Work‐related upper limb and neck musculoskeletal disorders (MSDs) are one of the most common occupational disorders worldwide. Studies have shown that the percentage of office workers that suffer from MSDs ranges from 20 to 60 per cent. The direct and indirect costs of work‐related upper limb MSDs have been reported to be high in Europe, Australia, and the United States. Although ergonomic interventions are likely to reduce the risk of office workers developing work‐related upper limb and neck MSDs, the evidence is unclear. This is an update of a Cochrane Review which was last published in 2012.

Objectives

To assess the effects of physical, cognitive and organisational ergonomic interventions, or combinations of those interventions for the prevention of work‐related upper limb and neck MSDs among office workers.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, CINAHL, Web of Science (Science Citation Index), SPORTDiscus, Embase, the US Centers for Disease Control and Prevention, the National Institute for Occupational Safety and Health database, and the World Health Organization's International Clinical Trials Registry Platform, to 10 October 2018.

Selection criteria

We included randomised controlled trials (RCTs) of ergonomic interventions for preventing work‐related upper limb or neck MSDs (or both) among office workers. We only included studies where the baseline prevalence of MSDs of the upper limb or neck, or both, was less than 25%.

Data collection and analysis

Two review authors independently extracted data and assessed risk of bias. We included studies with relevant data that we judged to be sufficiently homogeneous regarding the interventions and outcomes in the meta‐analysis. We assessed the overall quality of the evidence for each comparison using the GRADE approach.

Main results

We included 15 RCTs (2165 workers). We judged one study to have a low risk of bias and the remaining 14 studies to have a high risk of bias due to small numbers of participants and the potential for selection bias.

Physical ergonomic interventions

There is inconsistent evidence for arm supports and alternative computer mouse designs. There is moderate‐quality evidence that an arm support with an alternative computer mouse (two studies) reduced the incidence of neck or shoulder MSDs (risk ratio (RR) 0.52; 95% confidence interval (CI) 0.27 to 0.99), but not the incidence of right upper limb MSDs (RR 0.73; 95% CI 0.32 to 1.66); and low‐quality evidence that this intervention reduced neck or shoulder discomfort (standardised mean difference (SMD) −0.41; 95% CI −0.69 to −0.12) and right upper limb discomfort (SMD −0.34; 95% CI −0.63 to −0.06).

There is moderate‐quality evidence that the incidence of neck or shoulder and right upper limb disorders were not considerably reduced when comparing an alternative computer mouse and a conventional mouse (two studies; neck or shoulder: RR 0.62; 95% CI 0.19 to 2.00; right upper limb: RR 0.91; 95% CI 0.48 to 1.72), and also when comparing an arm support with a conventional mouse and a conventional mouse alone (two studies) (neck or shoulder: RR 0.91; 95% CI 0.12 to 6.98; right upper limb: RR 1.07; 95% CI 0.58 to 1.96).

Workstation adjustment (one study) and sit‐stand desks (one study) did not have an effect on upper limb pain or discomfort, compared to no intervention.

Organisational ergonomic interventions

There is very low‐quality evidence that supplementary breaks (two studies) reduce discomfort of the neck (MD −0.25; 95% CI −0.40 to −0.11), right shoulder or upper arm (MD −0.33; 95% CI −0.46 to −0.19), and right forearm or wrist or hand (MD ‐0.18; 95% CI ‐0.29 to ‐0.08) among data entry workers.

Training in ergonomic interventions

There is low to very low‐quality evidence in five studies that participatory and active training interventions may or may not prevent work‐related MSDs of the upper limb or neck or both.

Multifaceted ergonomic interventions

For multifaceted interventions there is one study (very low‐quality evidence) that showed no effect on any of the six upper limb pain outcomes measured in that study.

Authors' conclusions

We found inconsistent evidence that the use of an arm support or an alternative mouse may or may not reduce the incidence of neck or shoulder MSDs. For other physical ergonomic interventions there is no evidence of an effect. For organisational interventions, in the form of supplementary breaks, there is very low‐quality evidence of an effect on upper limb discomfort. For training and multifaceted interventions there is no evidence of an effect on upper limb pain or discomfort. Further high‐quality studies are needed to determine the effectiveness of these interventions among office workers.

Plain language summary

Ergonomic interventions for preventing work‐related musculoskeletal disorders of the upper limb and neck among office workers

What is the aim of this review?

The aim of this Cochrane Review was to find out if ergonomic interventions can prevent musculoskeletal pain or discomfort or both (musculoskeletal disorders; MSDs) among office workers. We collected and analysed all relevant studies to answer this question and found 15 studies.

Key messages

We found physical ergonomic interventions, such as using an arm support with a computer mouse based on neutral posture, may or may not prevent work‐related MSDs among office workers. We are still uncertain of the effectiveness of the other physical, organisational and cognitive ergonomic interventions.

What was studied in the review?

We selected office workers in our review, as they are a working population that has a higher risk for developing MSDs of the upper limb and neck. We assessed the effect of using ergonomic principles to improve the workplace and working process. Ergonomic refers to interactions among workers and other elements in the working environment, which includes physical, organisational and cognitive components. Physical ergonomic interventions include improving the equipment and environment of the workplace. The aim of these methods is to reduce the physical strain to the musculoskeletal system, thus reducing risk of injury. Meanwhile, organisational ergonomic interventions consist of allowing optimum workplace and rest time for the musculoskeletal system to recover from fatigue, thus reducing the risk of long‐term injury. Cognitive ergonomic interventions consist of improving mental processes such as perception, memory, reasoning and motor response through modifying work processes and training. The aim of these methods is to reduce mental workload, increase reliability and reduce error, which may have an indirect effect on reducing strain on the musculoskeletal system.

What are the main results of the review?

We found 15 studies that included 2165 workers. Fourteen of the studies conducted and reported their work poorly, and most of the studies had a small number of participants.

Out of the 15 studies, five studies evaluated the effectiveness of physical ergonomic interventions. Four studies evaluated the effectiveness of organisational ergonomic interventions, in the form of breaks or reduced working hours in preventing work‐related MSDs of the upper limb or neck, or both, among office workers. Five studies evaluated the effectiveness of ergonomic training, and one study evaluated multifaceted ergonomic interventions. We did not find any studies evaluating the effectiveness of cognitive ergonomic interventions.

Physical ergonomic interventions

We found that the use of an arm support or a mouse based on neutral posture may or may not prevent work‐related MSDs of the neck and shoulder. Workstation adjustment, and sit‐stand desks do not have an effect on upper limb pain compared to no intervention.

Organisational ergonomic interventions

We found that supplementary breaks may reduce neck and upper limb discomfort among data entry workers (two studies).

Cognitive ergonomic interventions

We found no studies using these methods.

Training interventions

There is no effect on upper limb pain compared to no intervention in five studies.

Mutlifaceted ergonomic interventions

There is no effect on pain or discomfort compared to no intervention in one study.

This means that there remains a need to conduct further studies to assess the effectiveness of ergonomic interventions.

How up‐to‐date is this review?

The review authors searched for studies that had been published up to 10 October 2018.

Summary of findings

Background

Description of the condition

Work‐related musculoskeletal disorders (MSDs) are the most common occupational disorders around the world, and have been recognised as a problem since the 17th century (Ramazzini 1964). Other general terms for these disorders include repetitive strain injury, occupational overuse syndrome and cumulative trauma disorders (Yassi 1997). Work‐related upper limb and neck MSDs are musculoskeletal disorders of the neck and upper limbs, which include the shoulders, upper arms, elbows, forearms, wrists, and hands (Buckle 1999). These are also known as complaints of the arm, neck and/or shoulder (CANS) (Huisstede 2006). MSDs can be divided into specific conditions with clear diagnostic criteria and pathological findings, which include tendon‐related disorders (e.g. tendonitis), peripheral‐nerve entrapment (e.g. carpal tunnel syndrome), neurovascular/vascular disorders (e.g. hand‐arm vibration syndrome), and joint/joint‐capsule disorders (e.g. osteoarthritis) or non‐specific conditions where the main complaint is pain or tenderness, or both, with limited or no pathological findings (Buckle 1997; Su 2013; Yassi 1997).

Based on the Global Burden of Disease 2010 study, the global point prevalence of neck pain was estimated to be 4.9% (95% confidence interval: 4.6 to 5.3), and was ranked fourth highest in terms of disability as measured by years lived with disability (YLDs) and 21st in term of overall burden (Hoy 2014). Moreover the cost of work‐related upper limb MSDs in the European Union (EU) has been reported to be high, with estimates ranging from 0.5% and 2% of gross national product (Buckle 1999). In Australia, disorders of the muscles, tendons, and soft tissue (excluding back pain) were estimated to cost AUD 519 million or 17% of the total health system costs in 1993 and 1994 (Mathers 1999). In the United Kingdom (UK), MSDs were recorded as the second highest reason for sickness certification in 2005, with an average of 22.84 sickness certificates being issued per 1000 person‐years (Wynne‐Jones 2009). In the UK in 2014/15 an estimate of 4.1 million working days were lost due to work‐related upper limb MSDs, which represents around 15% of all days lost due to work related ill‐health (HSE 2015). In the United States, the costs associated with musculoskeletal conditions accounted for 5.73% of GDP and 74% of the total work days lost in 2012. The direct per‐person healthcare costs for those with MSDs were estimated to be 7,104 USD in 2009‐2011 and accounted for around 30% of the injuries involving days absent from work. Those people with MSDs who were absent from work were away for a median of 11 days(USBJI 2015).

Over the past decades there has been an increase in the number of office workers in both developing and developed nations. This has been primarily attributed to the rapid development of knowledge‐based economies, which are directly based on the production, distribution and use of knowledge and information (OECD 1996). The emergence of new technologies, including the proliferation of personal computers, the internet and mobile devices has also contributed to the growth (Powell 2004). The nature of office‐based work has also subsequently changed from administrative and clerical work, to the production, distribution and use of knowledge. The office boundaries have expanded and are not limited to physical space but may include mobile workers and other offices throughout the world due to the ease of communication. Data processing, customer support, sales, and many other office processes may now be performed in developing countries (Subbarayalu 2013). Thus, not only has the office workers’ workforce grown in numbers it has also changed and diversified.

While we were not able to identify any systematic reviews that specifically reported prevalence of MSDs, including work‐related upper limb and/or neck MSDs, among office workers, several large cohort studies have reported these data. A Danish study of 5033 computer users reported the 12‐month prevalence for shoulder MSDs and wrist‐hand MSDs to be 44.7% and 25.8% respectively (Jensen 2003). Moreover, a UK study reported the 12‐month prevalence of neck MSDs to be 58% among data processing workers and 33% among other office workers (Woods 2005), while a Belgian study reported, the 12‐month prevalence of neck MSDs among office workers was 45.5%, (Cagnie 2007) and in Sweden the 12‐month prevalence of neck or shoulder MSDs among visual display terminal workers was 61.5% (Bergqvist 1995). In a large multicentre study involving 18 countries among more than 4000 office workers, the prevalence of disabling wrist and hand pain in the past month ranged from 2.2% in Pakistan and 2.3% in Japan to 31.3% in Brazil and 31.6% in Nicaragua (Coggon 2012; Coggon 2013a). The differences in prevalence rates reported by these studies may be a result of: the absence of a universally accepted definition of MSDs, the use of different diagnostic criteria (e.g. self‐reported or medical examination), and the variation in office work and office environments between these cultures and countries (Buckle 1999; Coggon 2013b; Huisstede 2006 ).

A number of studies have examined risk factors for MSDs and identified a variety of factors. These include individual factors (e.g. inadequate strength, poor posture, mental health, somatisation tendency, work‐causation beliefs, fear‐avoidance beliefs, cultural factors), physical requirements at the workplace (e.g. work requiring prolonged static posture, highly repetitive work, use of vibrating tools), and organisational and psychosocial factors (e.g. poor work‐rest cycle, shift work, low job security, little social support) (Bernard 1997; Buckle 1997; Coggon 2013a; Coggon 2013b; Hoe 2012b; Marras 2009; NIOSH 2001; Shanahan 2006; Yassi 1997).

Description of the intervention

Ergonomics as defined by the International Ergonomics Association (IEA) is the scientific discipline concerned with the understanding of the interactions among humans and other elements of a system. Ergonomics in the workplace refers to interactions among workers and other elements in the working environment. It is essentially about fitting the job to the worker. The IEA categorised ergonomics into three specific domains: physical, organisational and cognitive ergonomics.

The physical domain is concerned with human anatomical, anthropometric, physiological and biomechanical characteristics as they relate to physical activity. This domain consists of work environment and equipment, for example keyboard, mouse, hand tools, workstations, visual display units (VDUs) and lighting that are fitted to the workers.

The organisational domain is concerned with the optimisation of socio‐technical systems, including the organisational structures, policies and processes; for example work pace, work‐rest cycle and worker's participation in decision making.

The cognitive domain is concerned with mental processes, such as perception, memory, reasoning and motor response.

Ergonomic interventions have been heavily promoted for the prevention of work‐related upper limb or neck MSDs, or both (NIOSH 1997; NIOSH 2001). The current review will encompass interventions that focus on all three domains: the physical, organisational and cognitive domains.

How the intervention might work

Many studies have found that ergonomic factors correlate with musculoskeletal symptoms (Bernard 1994; Bonfiglioli 2006; Ortiz‐Hernandez 2003; Szeto 2009; Werner 2005). Adjusting physical, organisational and cognitive ergonomic factors to reduce the physical and mental load on workers is likely to reduce the risk of workers developing work‐related MSDs of the upper limb, neck or both.

Physical ergonomic interventions include providing workspace and equipment based on ergonomic principles and the anthropometry of workers. This will reduce the physical strain to the musculoskeletal system, thus reducing risk of injury. An example is the use of a split keyboard that has been found to reduce the severity of pain in computer users with MSDs (Tittiranonda 1999).

Organisational ergonomic interventions consists of allowing optimum work pace and rest time for the musculoskeletal system to recover from fatigue, thus reducing the risk of long term injury. An example is allowing supplementary breaks for data entry workers (Galinsky 2000). It can also include participatory interventions, where the workers participate in decision making on improvement and changes made at the workplace (Bohr 2000), and training in ergonomic principles and practices (Baydur 2016).

Cognitive ergonomics intervention consists of improving mental processes such as perception, memory, reasoning and motor response through modifying work process and training. This will reduce mental workload, increase reliability and reduce error, this may have an indirect effect in reducing strain on the musculoskeletal system.

Why it is important to do this review

A systematic review of interventions for the prevention and treatment of work‐related upper limb MSDs by Boocock 2007 evaluated studies published between 1999 and 2004. The authors concluded that there is some evidence to support the use of mechanical and modifier interventions for preventing and managing neck or upper extremity musculoskeletal conditions (Boocock 2007). However, there is a limitation in that the authors did not identify specific worker groups. Another systematic review by Kennedy 2010, which focused on the role of occupational health and safety interventions, found that the use of arm supports reduced upper extremity musculoskeletal diseases (MSDs) in office workers. However, Kennedy 2010 did not clearly define their search period. A more recent systematic review by Van Eerd 2016, which is an update of the Kennedy 2010 systematic review (updated search period between 2008 and 2013), found moderate evidence for vibration feedback about static mouse use and forearm supports in preventing work‐related MSDs of the upper limb or neck, or both, in office workers. Van Eerd 2016 also found moderate evidence for no effect for electric myogram (EMG) biofeedback, job stress management training, and office workstation adjustment for work‐related MSDs of the upper limb or neck, or both. However, in addition to randomised controlled trials (RCTs), Boocock 2007, Kennedy 2010 and Van Eerd 2016 included in their reviews other study designs that are at greater risk of bias. What is more, the three systematic reviews did not conduct meta‐analysis.

Our review extends and updates the search period covered by these three reviews and considers all published and unpublished randomised and quasi‐randomised trials investigating the use of physical, organisational and cognitive ergonomic interventions for the prevention of work‐related upper limb MSDs among office workers. We also conducted meta‐analysis of results from studies with comparable interventions and outcomes. Furthermore, this review is an update of our previous Cochrane Review (Hoe 2012a). In our previous review (Hoe 2012a), we included all workplaces and work settings, whereas in this review we focus only on the office setting. Other Cochrane Reviews will examine the effectiveness of interventions in different work settings. One example is Mulimani 2014, which investigates ergonomic interventions among dental care practitioners.

Objectives

To assess the effects of physical, organisational and cognitive ergonomic interventions, or combinations of those interventions for the prevention of work‐related upper limb and neck musculoskeletal disorders (MSDs) among office workers.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs), quasi‐randomised trials (trials which use methods of allocating participants to a treatment that are not strictly at random, e.g. by date of birth, hospital record number or alternative), cluster‐RCTs (i.e. where the unit of randomisation is a group of people, such as people working in the same office or shift rather than individual workers) and cross‐over trials (i.e. where participants are randomly allocated to a sequence of interventions).

Types of participants

We included studies where participants were office workers at the time of the intervention. Office workers were defined as those working in an office environment where their main tasks involved performing professional, managerial or administrative work. Because this review is focused on prevention of work‐related musculoskeletal disorders (MSDs) of the upper limb or neck or both, the majority of participants (75% or more) were required to be free of MSDs of the upper limb or neck, or both, at the time of the intervention. We only included studies conducted at the workplace.

We excluded studies evaluating treatment interventions for people with established MSDs of the upper limb or neck, or both (there are Cochrane systematic reviews conducted by Aas 2011, and Verhagen 2013, that have already covered workplace interventions for neck pain in workers and conservative interventions for treating work‐related complaints of the arm, neck or shoulder in adults). We also excluded studies that focus on rehabilitation of people with acute or chronic conditions (e.g. trauma, neoplasm, and inflammatory or neurological diseases).

Types of interventions

We included studies that examined at least one physical, organisational or cognitive ergonomic intervention in the workplace, aimed at the prevention of work‐related MSDs of the upper limb or neck, or both, among office workers. We excluded studies that tested ergonomic interventions for the treatment of individuals diagnosed with work‐related MSDs of the upper limb or neck, or both, or for prevention of work‐related MSDs of the upper limb or neck, or both, outside the office environment.

Interventions and specific comparisons

We categorised interventions as:

physical ergonomic interventions, such as the use of a specially designed computer mouse or arm support;
organisational ergonomic interventions, such as a different work‐rest cycle;
cognitive ergonomic interventions, such as job design;
training in ergonomic principles; and
multifaceted interventions that consist of a combination of one or more physical, organisational or cognitive interventions.

We planned the following main comparisons:

physical ergonomic intervention versus no intervention, placebo, or alternative intervention;
organisational ergonomic intervention versus no intervention, placebo, or alternative intervention;
cognitive ergonomic intervention versus no intervention, placebo, or alternative intervention;
training versus no training in ergonomic principles or versus alternative training; and
multifaceted interventions versus a single intervention or a different combination of interventions.

Types of outcome measures

We included studies based on the following primary and secondary outcomes.

Primary outcomes

Number of workers with newly diagnosed or verified MSDs of the upper limb or neck, or both (incident cases).
Presence or severity or intensity of complaints or symptoms of pain or discomfort in the upper limb or neck, or both, using a dichotomised scale (e.g. yes/no), Likert scale, visual analogue scale (VAS), or any similar scale measuring pain or discomfort.
Work‐related function as measured by number of work days lost, loss of or change in job, work disability, and level of functioning. For the level of functioning, we included studies using validated outcome measures e.g. Disability of the Arm, Shoulder, and Hand (DASH) questionnaire (Kitis 2009), and Northwick Park Neck Pain Questionnaire (Leak 1994).

Secondary outcomes

Secondary outcomes included the following.

Time and comfort in work positions or postures.
Change in productivity.
Costs (including costs of implementation of the intervention and treatment).
Compliance (attitude and practice). Compliance is the degree of how well study participants adhere to the prescribed intervention. We considered compliance as a secondary outcome as it indicates the intervention take‐up rate.

We only included studies that reported one or more primary outcomes in this review. If a study only reported one or more secondary outcomes, then we excluded that study from this review.

Search methods for identification of studies

Electronic searches

We systematically searched the following databases:

Cochrane Central Register of Controlled Trials (Issue 9, September 2018) in the Cochrane Library (Appendix 1);
Ovid MEDLINE(R) and In‐Process & Other Non‐Indexed Citations and Daily (1948 to 17 September, 2018 (Appendix 2);
Embase (1980 to 29 May 2017) (Appendix 3);
Web of Science (Search date: 18 September, 2018) (Appendix 4);
CINAHL (EBSCOhost) (Search date: September 18, 2018) (Appendix 5);
SPORTDiscus (1949 to 10 October, 2018) (Appendix 6);
Scopus (Search date: 21 September, 2018, limit to 2017 and 2018 studies) (Appendix 7);
NIOSHTIC‐2 (Search date: 21 September, 2018) (Appendix 8)

From May 2017 onwards we replaced the Embase search with a search in Scopus because of ease of access and because the latter contains everything included in the former.

We searched the following websites and databases for unpublished and ongoing studies:

World Health Organization International Clinical Trials Registry Platform (10 October 2018);

We considered reports published in all languages. The searches were based on the MEDLINE search strategy combined with the sensitivity‐ and precision‐maximising version of the Cochrane Highly Sensitive Search Strategy for identifying RCTs (Lefebvre 2011) (see Appendix 2). We modified the search strategy to use in the other databases.

Searching other resources

We contacted experts in the field to identify theses and unpublished studies. We looked for additional studies by checking the bibliographies of relevant articles.

Data collection and analysis

Selection of studies

Two review authors (VCWH and ENZ) obtained and screened abstracts and citations identified by the systematic searches. The full‐text articles of studies identified as being potentially eligible for the review were retrieved to further determine their inclusion (VCWH and ENZ). We resolved all disagreements by discussion between the review authors to reach a consensus. Where there was uncertainty, we contacted the corresponding author to ascertain whether a potentially relevant study met the inclusion criteria.

Data extraction and management

Two review authors (VCWH and ENZ) performed data extraction independently, with checks for discrepancies and processing as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved all discrepancies by discussion and consensus. We used a standard data extraction form based on the form recommended by the Cochrane Bone, Joint and Muscle Trauma Group. We performed all statistical analyses using Review Manager 5.3 (RevMan 2014) software.

Assessment of risk of bias in included studies

Two review authors (VCWH and ENZ) assessed the risk of bias of included studies independently using Cochrane's 'Risk of bias' tool (Appendix 9) (Higgins 2011). We assessed each study for risk of bias in each of the following domains: sequence generation, allocation concealment, blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data, selective outcome reporting, and 'other', such as contamination bias and reliability of instruments. We assessed the risk of bias associated with (a) blinding and (b) completeness of outcomes separately for self‐reported outcomes and objective outcomes. We resolved disagreements between authors regarding the risk of bias for domains by discussion and consensus.

We considered a study to have low risk of bias overall if we judged it to have a low risk of bias in the domains random sequence generation (selection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other forms of bias. We did not include allocation concealment (selection bias), blinding of participants and personnel (performance bias), and blinding of outcome assessment (detection bias) in the criteria for classifying the included studies' overall risk of bias because of the nature of the intervention, which requires fully aware participation of participants and personnel, and because the main outcome, pain, is a subjective symptom (International Association for the Study of Pain).

Measures of treatment effect

We plotted the results of each trial as point estimates, using risk ratios (RRs) for dichotomous outcomes, and means and standard deviations (SDs) for continuous outcomes. When studies reported different outcome measures but measured the same concept, we calculated the standardised mean difference (SMD) with 95% confidence interval (CI). For studies that had reported outcome data for both the right and left upper limb, we only used the outcome data for the right upper limb.

Unit of analysis issues

If studies employed a cluster‐randomised design, but did not take the cluster effect into account, we tried to adjust the data for the effect of clustering by calculating the design effect based on an assumed intra cluster correlation of 0.1.

Dealing with missing data

We dealt with missing data according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), that is, we contacted study authors to request missing data.

Assessment of heterogeneity

First, we assessed whether studies were sufficiently homogeneous to be included in one comparison. We based this judgment on the similarity of the type of interventions, what the control condition was, the outcome and when the outcome was measured (short term: three to eight weeks, intermediate: eight weeks to six months, or long‐term: six months or longer).

Second, we tested for statistical heterogeneity by means of the I² statistic as presented in the meta‐analysis graphs generated by the Review Manager 5 software (RevMan 2014). When this test statistic was greater than 50% we considered there to be substantial heterogeneity between studies. In such cases we employed the random‐effects meta‐analysis and we downgraded the quality of evidence according to the GRADE system for the relevant comparisons.

Assessment of reporting biases

If, in future updates of this review, we are able to pool more than ten trials in any single meta‐analysis, we will create and examine a funnel plot to explore possible small study biases.

Data synthesis

We pooled results of studies if they had a similar type of intervention, control conditions, and outcome. When studies were statistically heterogeneous, we used a random‐effects model; otherwise we used a fixed‐effect model. We pooled study results data with Review Manager 5 software (RevMan 2014).

We considered the types of interventions evaluated in each of the studies and found the studies assessing the effectiveness of ergonomic computer mouse or arm support (physical ergonomic interventions), supplementary breaks or reduced work hours (organisational ergonomic interventions), and ergonomic training (cognitive ergonomic interventions) to be sufficiently similar to be pooled for comparison.

We assessed the overall quality of the evidence contributing to the primary outcomes for each important intervention, using the GRADEpro GDT software (GRADEpro GDT).

Our judgement of the quality of the evidence for a specific intervention‐outcome combination was based on performance against the five GRADE domains: limitations of study design, inconsistency, indirectness (inability to generalise), imprecision (insufficient or imprecise data) of results, and publication bias across all studies that measured that particular outcome. The overall quality of the evidence for each outcome is the result of a combination of the assessments in all domains.

There are four grades of evidence:

high‐quality evidence: there are consistent findings among at least 75% of RCTs with no limitations of the study design, consistent, direct and precise data and no known or suspected publication biases. Further research is unlikely to change either the estimate or our confidence in the results;
moderate‐quality evidence: one of the domains is not met. Further research is likely to have an important impact on our confidence in the estimate of effect and might change the estimate;
low‐quality evidence: two of the domains are not met. Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate;
very‐low‐quality evidence: three of the domains are not met. We are very uncertain about the estimate.

Subgroup analysis and investigation of heterogeneity

If, in future updates of this review, we can include a sufficient amount of data we will conduct subgroup analyses based on: type of job, gender, and rigour of outcome measurement.

Sensitivity analysis

If, in future updates of this review, we can include a sufficient amount of data we will undertake sensitivity analyses by excluding the studies we judge to have a high risk of bias. In the current review this was not possible as we judged only one study to have a low risk of bias.

Results

Description of studies

Results of the search

Our search strategy identified 2547 potentially relevant references after duplicates had been removed. Two review authors (VCWH and ENZ) assessed the titles, keywords, and abstracts of these references, and selected 48 potentially eligible references. We obtained the full‐text publications for these 48 references.

We did not identify any additional references by searching the following additional databases: the US Centers for Disease Control and Prevention National Institute for Occupational Safety and Health (NIOSHTIC‐2) database, and the International Occupational Safety and Health Information Centre (CIS) database. Our search for unpublished and ongoing studies, through the World Health Organization International Clinical Trials Registry Platform, identified one additional registered trial (Shariat 2016).

We checked the reference lists of all articles that we retrieved as full‐text papers in order to identify potentially eligible studies. We did not identify any additional studies through this approach. Of the 48 full‐text reports and one registered trial identified, we included 15 studies reported in 17 publications. We excluded 24 studies reported in 30 publications. We also identified two ongoing studies (Johnston 2014; Shariat 2016). See the PRISMA study flow diagram (Figure 1) for our description of the whole study inclusion process.

Included studies

We included 15 studies reported in 17 publications. These studies recruited a total of 2165 participants. For further details regarding the study populations and settings, see the Characteristics of included studies table.

Study Design

All of the included studies were randomised controlled trials (RCTs); two used a cluster‐randomised design (Brisson 1999; Baydur 2016), and another two used a cross‐over design (Galinsky 2000; Galinsky 2007).

Location and settings

Nine studies were conducted in the United States (Bohr 2000; Bohr 2002; Conlon 2008; Galinsky 2000; Galinsky 2007; Gatty 2004; Gerr 2005; Greene 2005; Rempel 2006), three were conducted in Canada (Brisson 1999; McLean 2001; King 2013), and the remaining three studies were conducted in Finland (Lintula 2001), the United Kingdom (Graves 2015), and Turkey (Baydur 2016).

Three studies were conducted in data processing or call centres (Galinsky 2000; Galinsky 2007; Rempel 2006), four studies in universities or colleges (Brisson 1999; Gatty 2004; Greene 2005; Graves 2015), two studies in a transportation company (Bohr 2000; Bohr 2002), one study in an aerospace firm (Conlon 2008), one study among office workers in a municipality (Baydur 2016), one study among office employees and researchers (Lintula 2001), one study in a research organisation (King 2013), and two studies involved several sectors (insurance and financial companies, food product producers, government offices, and universities) (Gerr 2005; McLean 2001).

Type of work

All studies were conducted with participants who were using computers or who were conducting data processing in an office environment (Baydur 2016, Bohr 2000; Bohr 2002; Brisson 1999; Conlon 2008; Galinsky 2000; Galinsky 2007; Gatty 2004; Gerr 2005; Graves 2015; Greene 2005; King 2013; Lintula 2001; McLean 2001; Rempel 2006).

Type of interventions

Physical ergonomic interventions

Five studies evaluated physical ergonomic interventions alone, which consisted of alternative computer mouse or arm supports, or both (Conlon 2008; Rempel 2006), arm support alone (Lintula 2001), sit‐stand workstation (Graves 2015) and ergonomic posture intervention (Gerr 2005).

Organisational ergonomic interventions

Four studies evaluated organisational ergonomic interventions in the form of supplementary breaks or reduced work hours (Galinsky 2000; Galinsky 2007; McLean 2001; King 2013). Although the intervention was a biofeedback mouse in one study (King 2013), the objective of the mouse was to ensure workers take breaks from using the mouse.

Cognitive ergonomic interventions

No study specifically addressed cognitive processes.

Training programmes

Five studies evaluated ergonomic training programmes (Baydur 2016; Bohr 2000; Bohr 2002; Brisson 1999; Greene 2005).

Multifaceted ergonomics interventions

One study evaluated a combination of organisational and physical ergonomic interventions (Gatty 2004), which consisted of training, workstation redesign and task modification.

Follow‐up period

Five studies had a short follow‐up period of between four and eight weeks (Galinsky 2000; Galinsky 2007; Graves 2015; Greene 2005; Lintula 2001; McLean 2001). One study had an intermediate‐term follow‐up period of 16 weeks (Gatty 2004), and eight studies had a long‐term follow‐up period of between six and 13 months (Baydur 2016, Bohr 2000; Bohr 2002; Brisson 1999; Conlon 2008; Gerr 2005; King 2013; Rempel 2006).

Outcomes

The incidence of musculoskeletal disorders (MSDs) was measured in three studies (Conlon 2008; Gerr 2005; Rempel 2006), and the prevalence of MSDs was measured in a further three studies (Brisson 1999; Gatty 2004; Greene 2005). The severity, intensity, discomfort, and strain associated with musculoskeletal conditions were measured in 13 studies (Baydur 2016; Bohr 2000; Bohr 2002; Conlon 2008; Galinsky 2000; Galinsky 2007; Gatty 2004; Graves 2015; Greene 2005; King 2013; Lintula 2001; McLean 2001; Rempel 2006).

One study assessed disability (Baydur 2016).

Seven studies assessed compliance to interventions (Bohr 2000; Bohr 2002; Brisson 1999; Gatty 2004; Gerr 2005; Graves 2015; King 2013).

Unit of analysis

Brisson 1999, reported the number of clusters and the intracluster correlation coefficients (ICCs) for the neck or shoulder (0.0161) and for the wrist or hand (0.0007). The design effect of the study is calculated using the formula 1 + (average cluster size −1) x ICC. The results of the design effect are then used to calculate the effective (reduced) sample size. Baydur 2016 also provided us with the number of clusters. Based on the intracluster correlation coefficients of Brisson 1999 we adjusted the effective sample size for this study as well.

Dealing with missing data

We contacted five authors for clarification and additional data relating to six studies (Baydur 2016; Bohr 2000; Bohr 2002; Brisson 1999; Galinsky 2000; Galinsky 2007; McLean 2001), and we were able to use the additional data for four studies (Baydur 2016; Brisson 1999; Galinsky 2000; Galinsky 2007).

For Baydur 2016 we received the additional information that the number of clusters was 16 in the control group with a total of 58 workers and similarly there were 16 clusters in the intervention group with a total of 58 workers. For the cross‐over trials, Galinsky 2000 and Galinsky 2007, we conducted our own paired analysis. For the analysis we received from the author data about the means and standard deviations of discomfort ratings after the intervention and after the control condition. We used the square root of the F‐value as reported by the authors as a best estimate of the T‐value to enable the calculation of the SE of the MD. We also calculated the SE based on assumed correlations of 0.5, 0.7 and 0.9 between the discomfort ratings of the intervention and control condition as proposed in the Handbook chapter 16.4.6. The assumption of a correlation of 0.85 agreed best with the values derived of the F‐value and we took this correlation for imputing the SE values for both studies.

Excluded studies

Altogether we excluded 24 studies published in 30 reports. We excluded 14 studies because more than 25% of the participants reported musculoskeletal symptoms of the upper limb or neck, or both, at baseline (Danquah 2017; Dropkin 2015; Esmaeilzadeh 2014; Fostervold 2006; Ketola 2002; Levanon 2012; Mahmud 2011; Mann 2013; Meijer 2009a; Meijer 2009b; Mekhora 2000; Parry 2015; Ripat 2006; Spekle 2010). We excluded three studies because they were not RCTs (Aaras 1998; Amick 2003; Amick 2012), and a further three studies because they had not measured the effectiveness of interventions on disorders of the upper limb or neck, or both (Chau 2014; De Cocker 2016; Krause 2010). We excluded two more studies (one of which, Thorp 2014, was reported in two reports) because they were conducted in a laboratory setting (Robertson 2013; Thorp 2014). We excluded two studies (one of which, Driessen 2008, was reported in six reports) where the participants consisted of workers other than office workers (Driessen 2008; Faucett 2002). For further details regarding the study populations and settings see the Characteristics of excluded studies table. In addition to the 24 studies excluded in this review update, we also excluded the studies that were not undertaken in office workers that had been included in the previous version of this review (Hoe 2012a): von Thiele 2008 and Yassi 2001.

Risk of bias in included studies

Allocation

Five studies (Conlon 2008; Gerr 2005; Graves 2015; King 2013; Rempel 2006) used a random number table or equivalent for generating a random sequence and therefore we judged them to have a low risk of allocation bias. In Graves 2015, it was indicated that they completed allocation by alternating between intervention and control, and that they did not conceal the allocation, so we judged this study as having high risk of bias. All the other studies did not report using adequate measures for concealing allocation, such as using sealed opaque envelopes, and thus we judged them to have an unclear risk of bias.

Blinding

Blinding of the interventions was not performed in most of the studies, as blinding of physical, organisational and cognitive ergonomic interventions is difficult to achieve. Therefore, we judged 12 studies to have a high risk for performance bias. The remaining three studies assessed organisational ergonomic interventions of work breaks and work hours (Galinsky 2000; Galinsky 2007; McLean 2001). Although complete blinding for breaks was not possible in these studies, the use of a strict protocol for taking breaks by the use of either custom‐made electrical timers, or the 'Ergobreak' computer programme, minimised the risk of bias. Therefore, we judged these three studies to have a low risk of performance bias.

Although in three studies (Brisson 1999; Conlon 2008; Rempel 2006), the physical examination for the detection of MSD was blinded, the examination was only performed on participants who self‐reported symptoms meeting the case definition, which may lead to detection bias. Thus, we rated the risk of detection bias as high for all 15 studies.

Incomplete outcome data

Four studies conducted an intention‐to‐treat (ITT) analysis (Conlon 2008; Gerr 2005; King 2013; Rempel 2006), one study had no loss to follow‐up (Lintula 2001), and four studies had a low drop‐out rate (Baydur 2016; Brisson 1999; Graves 2015; King 2013). We rated these nine studies as having a low risk of attrition bias. We rated five studies (Bohr 2000; Bohr 2002; Galinsky 2000; Galinsky 2007; Gatty 2004) as having a high risk of attrition bias, as they did not conduct ITT analyses. In addition, one of these five studies had an uneven drop‐out rate across the groups (Bohr 2000), and four studies had a high drop‐out rate (Galinsky 2000; Galinsky 2007; Gatty 2004; Bohr 2002). We rated two studies as having an unclear risk of attrition bias as they did not conduct ITT analyses and information on their drop‐out rate was limited (Greene 2005; McLean 2001).

Selective reporting

We judged all 15 included studies to be free of selective reporting because they reported all outcomes described in the methods.

Other potential sources of bias

We judged 11 studies to have a high risk of bias from other potential sources (Baydur 2016; Bohr 2000; Bohr 2002; Brisson 1999; Conlon 2008; Galinsky 2007; Gatty 2004; Gerr 2005; Graves 2015; Lintula 2001; McLean 2001), two studies to have a low risk of other bias (Galinsky 2000; Rempel 2006), and another two studies had an unclear risk of other bias (Greene 2005; King 2013).

Five studies did not report baseline data on the outcome measures (Baydur 2016; Bohr 2000; Brisson 1999; Lintula 2001; McLean 2001). In Gatty 2004, the intervention group had lower average wrist‐hand and upper back ache or pain intensity compared to the control group. In Conlon 2008, the participants who volunteered for the study had higher levels of discomfort than non‐participants. In two studies (Bohr 2000; Bohr 2002), the close proximity of the workstations may have led to contamination of the intervention effect. In another two studies (Gerr 2005; Bohr 2002), there were large numbers of dropouts in the intervention and control groups; and although in Gerr 2005, the authors conducted ITT analysis, the large number of dropouts may have affected the findings. In the two cluster‐RCTs (Brisson 1999; Baydur 2016), the latter did not report cluster size.

Of the two cross‐over RCTs (Galinsky 2000; Galinsky 2007), the latter had potential for a carry‐over effect. The authors did not report if they had a wash‐out period between the two data collection periods.

Overall risk of bias per study

Overall, we found that the risk of bias in the included studies was high. Of the 15 studies, we judged only one study, Rempel 2006, to have a low risk of bias overall. For details on our a priori criteria for assigning ‘Risk of bias’ judgements to studies overall, see Assessment of risk of bias in included studies. See Figure 2 for an overview of our judgements about each 'Risk of bias' item, presented as percentages across all included studies. Figure 3 shows the 'Risk of bias' summary of each 'Risk of bias' item for each included study.

'Risk of bias' graph: review authors' judgements about each 'Risk of bias' item presented as percentages across all included studies.

'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study.

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4; Table 5; Table 6; Table 7; Table 8; Table 9

Summary of findings for the main comparison. Arm support combined with alternative mouse versus conventional mouse alone.

Patient or population: office workers  Settings: office environment using visual display units (> 20 h/week)  Intervention: an arm support combined with an alternative computer mouse  Comparison: conventional mouse alone (with no arm support)
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Conventional mouse alone	Arm support with alternative mouse
Incidence of upper body disorders (neck, shoulder, and upper extremity)  Questionnaire followed by medical examination  Follow‐up: 12 months	333 per 1000	220 per 1000  (140 to 347)	RR 0.66   (0.42 to 1.04)	191  (2 studies)	⊕⊕⊕⊝  moderate¹
Incidence of neck or shoulder disorder  Questionnaire followed by medical examination  Follow‐up: 12 months	232 per 1000	120 per 1000  (63 to 229)	RR 0.52   (0.27 to 0.99)	186  (2 studies)	⊕⊕⊕⊝  moderate¹
Incidence of right upper extremity disorder  Questionnaire followed by medical examination  Follow‐up: 12 months	174 per 1000	127 per 1000  (56 to 289)	RR 0.73   (0.32 to 1.66)	181  (2 studies)	⊕⊕⊕⊝  moderate¹
Neck or shoulder discomfort score  Questionnaire  Follow‐up: 12 months		The mean neck or shoulder discomfort score in the intervention groups was  0.41 standard deviations lower  (0.69 to 0.12 lower) ⁴		194  (2 studies)	⊕⊕⊝⊝  low^2,3	SMD −0.41 (−0.69 to −0.12) clinically meaningful difference
Right upper extremity discomfort score  Questionnaire  Follow‐up: 12 months		The mean right upper extremity discomfort score in the intervention groups was  0.34 standard deviations lower  (0.63 to 0.06 lower)⁴		194  (2 studies)	⊕⊕⊝⊝  low^2,3	SMD −0.34 (−0.63 to −0.06) clinically meaningful difference
Work related function	no data	no data
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; RR: risk ratio; VDU: visual display unit;SMD: standardised mean difference
GRADE Working Group grades of evidence  High quality: Further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: We are very uncertain about the estimate.

Date	Event	Description
10 October 2018	New search has been performed	We updated the search but found no new studies.
6 June 2017	New search has been performed	We revised the categorisation of ergonomic interventions to be in line with the International Ergonomics Association (IEA) categories.
6 June 2017	New citation required and conclusions have changed	We revised and updated the search and found two additional studies.
31 August 2016	Amended	We revised the inclusion criteria for the participants. Instead of including all workers we now restrict inclusion to studies of office workers only.

Date	Event	Description
19 June 2013	Feedback has been incorporated	Feedback from Traci Galinsky, received on 29 March 2013, has been incorporated and the authors have provided a thorough response.
28 July 2010	Amended	The order of the authors has been amended.

No	Query	Results
#53	#35 AND #52	638
#52	#36 OR #37 OR #38 OR #39 OR #40 OR #41 OR# 42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51 AND [humans]/lim	1,812,079
#51	'rct':ab,ti AND [embase]/lim	20,686
#50	((allocat* OR allot* OR assign* OR divid) NEAR/3 (condition OR experiment* OR intervention* OR treatment* OR therap* OR control* OR group*)):ab,ti AND [embase]/lim	269,690
#49	crossover:ab,ti OR (cross NEXT/1 over):ab,ti AND [embase]/lim	76,302
#48	((singl* OR doubl* OR trebl* OR tripl) NEAR/7 (blind OR mask*)):ab,ti AND [embase]/lim	186,297
#47	(random* NEAR/7 (allocat* OR allot* OR assign* OR basis* OR divid* OR order*)):ab,ti AND [embase]/lim	199,419
#46	((clinical OR controlled OR comparative OR placebo OR prospective* OR randomi?ed) NEAR/3 (trial OR study)):ab,ti AND [embase]/lim	877,894
#45	'prospective study'/de AND [embase]/lim	304,540
#44	'placebo'/de AND [embase]/lim	293,411
#43	'crossover procedure'/de AND [embase]/lim	44,454
#42	'double blind procedure'/de AND [embase]/lim	122,521
#41	'single blind procedure'/de AND [embase]/lim	22,239
#40	'randomization'/de AND [embase]/lim	45,797
#39	'clinical trial'/de AND [embase]/lim	749,064
#38	'controlled clinical trial'/exp AND [embase]/lim	482,670
#37	'randomized controlled trial'/exp OR 'randomized controlled trial' AND [embase]/lim	468,251
#36	'randomi?ed controlled trial?' AND [embase]/lim	61,671
#35	#22 OR #34	5,793
#34	#27 AND #33	3,200
#33	#28 OR #29 OR #30 OR #31 OR #32	199,134
#32	military:ab,ti OR navy:ab,ti OR army:ab,ti OR soldier:ab,ti OR athlet:ab,ti OR runner:ab,ti AND [embase]/lim	78,169
#31	'sport'/exp OR 'dancing'/exp AND [embase]/lim	91,086
#30	'sport injury'/de AND [embase]/lim	17,658
#29	'military phenomena'/exp AND [embase]/lim	42,336
#28	'cumulative trauma disorder'/exp AND [embase]/lim	15,217
#27	#23 OR #24 OR #25 OR #26	8,378
#26	(('fract' OR 'injur') NEAR/3 ('insufficiency' OR 'fatigue' OR 'overuse')):ab,ti AND [embase]/lim	3,815
#25	('bone' NEAR/3 'stress' NEAR/3 'reaction*'):ab,ti AND [embase]/lim	32
#24	'stress fracture*':ab,ti AND [embase]/lim	3,434
#23	'stress fracture'/de AND [embase]/lim	4,739
#22	#15 AND #21	2,650
#21	#16 OR #17 OR #18 OR #19 OR #20	283,833
#20	'ergonom':ab,ti OR 'biomechanic':ab,ti AND [embase]/lim	54,526
#19	'workload'/de OR 'workplace'/de OR 'equipment design'/de OR 'human computer interaction'/de OR 'visual display unit'/de OR 'ergonomics'/de AND [embase]/lim	72,849
#18	'movement (physiology)'/de OR 'body posture'/de AND [embase]/lim	46,096
#17	'biomechanics'/de AND [embase]/lim	58,082
#16	'bioengineering'/exp AND [embase]/lim	96,459
#15	#9 AND #14	14,279
#14	#10 OR #11 OR #12 OR #13	323,735
#13	'carpal tunnel syndrome*':ab,ti AND [embase]/lim	7,105
#12	'neck?' OR 'shoulder?' OR 'arm?' OR 'upper limb?' OR 'upper extremit*' OR 'elbow?' OR 'forearm?' OR 'wrist?' OR 'hand?' OR 'finger?' AND [embase]/lim	264,568
#11	'musculoskeletal disease'/de AND [embase]/lim	19,093
#10	'shoulder pain'/de OR 'neck pain'/de OR 'arm injury'/de OR 'hand injury'/exp OR 'shoulder injury'/de OR 'wrist injury'/de OR 'elbow injury'/de AND [embase]/lim	44,390
#9	#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8	81,542
#8	('vibration' NEXT/1 ('induced' OR 'related' OR 'syndrome*')):ab,ti AND [embase]/lim	1,123
#7	('repetit' NEXT/1 ('strain' OR 'stress' OR 'industr' OR 'motion' OR 'movement' OR 'trauma')):ab,ti AND [embase]/lim	1,347
#6	'work related':ab,ti AND [embase]/lim	10,307
#5	'cumulative trauma*':ab,ti AND [embase]/lim	478
#4	(('occupational overuse' OR 'tension neck') NEXT/1 syndrome):ab,ti AND [embase]/lim	33
#3	'occupational health'/de OR 'occupational hazard'/de OR 'occupational safety'/de AND [embase]/lim	41,396
#2	'occupational disease'/de OR 'hand arm vibration syndrome'/de OR 'occupational accident'/de AND [embase]/lim	23,163
#1	'cumulative trauma disorder'/exp OR 'cumulative trauma disorder' AND [embase]/lim	15,251

#	Query	Results
S29	S20 and S28	159
S28	S27 or S26 or S25 or S24 or S23 or S22 or S21	340,934
S27	TX placebo*	25,485
S26	TX ((allocat* or allot* or assign* or divid) and (condition or experiment* or intervention* or treatment* or therap* or control* or group*))	148,317
S25	TX "randomi?ed control* trial*"	32,860
S24	TX (cross?over or (cross over))	94,099
S23	TX ((singl* or doubl* or trebl* or tripl) and (blind or mask*))	43,994
S22	TX (random* and (allocat* or allot* or assign* or basis* or divid* or order*))	100,187
S21	TX ((clinic$ or controlled or comparative or placebo or prospective or randomised or randomized) and (trial or study))	218,322
S20	S14 and S19	414
S19	S15 or S16 or S17 or S18	46,454
S18	TI ( (ergonom* or biomechanic) ) or AB ( (ergonom or biomechanic*) )	17,343
S17	DE "POSTURE" OR DE "SITTING position" OR DE "STANDING position"	10,214
S16	DE "BIOMECHANICS"	29,723
S15	DE "HUMAN engineering" OR DE "SITUATIONAL awareness"	292
S14	S9 and S13	2,520
S13	S10 or S11 or S12	161,823
S12	TI "carpal tunnel syndrome" or AB "carpal tunnel syndrome"	595
S11	DE "NECK pain" or DE "SHOULDER pain" or DE "WOUNDS & injuries"	45,985
S10	TI ( ( (neck* or shoulder* or arm* or upper limb* or upper extremit* or elbow* or forearm* or wrist* or hand* or finger) ) ) or AB ( ( (neck or shoulder* or arm* or upper limb* or upper extremit* or elbow* or forearm* or wrist* or hand* or finger*) ) )	123,435
S9	S1 or S2 or S3 or S4 or S5 or S6 or S7 or S8	7,767
S8	TI ( (vibration and (induced or related or syndrome)) ) or AB ( (vibration and (induced or related or syndrome)) )	458
S7	TI ( (repetit* and (strain or stress or industr* or motion or movement or trauma)) ) or AB ( (repetit* and (strain or stress or industr* or motion or movement or trauma)) )	2,707
S6	TI "work related" or AB "work related"	1,676
S5	TI "cumulative trauma" or AB "cumulative trauma"	67
S4	TI ( ("occupational overuse" or "tension neck") and syndrome* ) or AB ( ("occupational overuse" or "tension neck") and syndrome* ) "	2
S3	DE "OCCUPATIONAL health services"	900
S2	DE "OCCUPATIONAL diseases"	625
S1	DE "OVERUSE injuries"	1,824

Domain	Description	Review authors' judgement
Sequence generation	Describe the method used to generate the allocation sequence in sufficient detail to allow an assessment of whether it should produce comparable groups	Was the allocation sequence adequately generated? Yes/ No/ Unclear
Allocation concealment	Describe the method used to conceal the allocation sequence in sufficient detail to determine whether intervention allocations could have been foreseen in advance of, or during, enrolment	Was allocation adequately concealed? Yes/ No/ Unclear
Blinding of participants, personnel and outcome assessors (Assessments should be made for each main outcome (or class of outcomes))	Describe all measures used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. Provide any information relating to whether the intended blinding was effective	Was knowledge of the allocated intervention adequately prevented during the study? Yes/ No/ Unclear
Incomplete outcome data (Assessments should be made for each main outcome (or class of outcomes))	Describe the completeness of outcome data for each main outcome, including attrition and exclusions from the analysis. State whether attrition and exclusions were reported, the numbers in each intervention group (compared with total randomised participants), reasons for attrition/exclusions where reported, and any re‐inclusions in analyses performed by the review authors	Were incomplete outcome data adequately addressed? Yes/ No/ Unclear
Selective outcome reporting	State how the possibility of selective outcome reporting was examined by the review authors, and what was found	Are reports of the study free of suggestion of selective outcome reporting? Yes/ No/ Unclear
Other sources of bias	State any important concerns about bias not addressed in the other domains in the tool. If particular questions/entries were pre‐specified in the review’s protocol, responses should be provided for each question/entry	Was the study apparently free of other problems that could put it at a high risk of bias? Yes/ No/ Unclear

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Neck/shoulder discomfort score at 12‐month follow‐up	2	194	Std. Mean Difference (IV, Random, 95% CI)	‐0.41 [‐0.69, ‐0.12]
2 Incidence of neck/shoulder disorder at 12‐month follow‐up	2	186	Risk Ratio (M‐H, Random, 95% CI)	0.52 [0.27, 0.99]
3 Right upper extremity discomfort score at 12‐month follow‐up	2	194	Std. Mean Difference (IV, Random, 95% CI)	‐0.34 [‐0.63, ‐0.06]
4 Incidence of right upper limb disorder at 12‐month follow‐up	2	181	Risk Ratio (M‐H, Random, 95% CI)	0.73 [0.32, 1.66]
5 Incidence of upper body disorders (neck, shoulder, and upper limb) at 12‐month follow‐up	2	191	Risk Ratio (M‐H, Random, 95% CI)	0.66 [0.42, 1.04]
6 Change in percentage of work time	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7 Change in average time to completely process a call	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8 Change in calls per hour	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
9 Subject perceived improvement	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Right upper‐limb strain scale at 6‐week follow‐up	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Neck/shoulder discomfort score	2	195	Std. Mean Difference (IV, Random, 95% CI)	0.02 [‐0.26, 0.30]
3 Incidence of neck/shoulder disorder	2	186	Risk Ratio (M‐H, Random, 95% CI)	0.91 [0.12, 6.98]
4 Right upper extremity discomfort score	2	195	Std. Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.35, 0.22]
5 Incidence of right upper extremity disorders	2	178	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.58, 1.96]
6 Incidence of upper body disorders (neck, shoulder and upper limb)	2	191	Risk Ratio (M‐H, Random, 95% CI)	0.87 [0.42, 1.80]
7 Change in percentage of work time	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8 Change in average time to completely process a call	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
9 Change in calls per hour	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
10 Subject perceived improvement	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Neck/shoulder discomfort score	2	195	Std. Mean Difference (IV, Random, 95% CI)	0.04 [‐0.26, 0.33]
2 Incidence of neck/shoulder disorder	2	182	Risk Ratio (M‐H, Random, 95% CI)	0.62 [0.19, 2.00]
3 Incidence of right upper extremity disorder	2	182	Risk Ratio (M‐H, Random, 95% CI)	0.91 [0.48, 1.72]
4 Right upper extremity discomfort score	2	195	Std. Mean Difference (IV, Random, 95% CI)	0.00 [‐0.28, 0.28]
5 Incidence of upper body disorder (neck, shoulder, and upper extremity)	2	190	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.52, 1.21]
6 Change in percentage of work time	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7 Change in average time to completely process a call	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8 Change in calls per hour	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
9 Subject perceived improvement	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Incidence of neck and shoulder pain	1	234	Risk Ratio (M‐H, Fixed, 95% CI)	1.08 [0.73, 1.59]
2 Incidence of arm and hand pain	1	245	Risk Ratio (M‐H, Fixed, 95% CI)	0.83 [0.50, 1.39]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Incidence of neck and shoulder pain	1	255	Risk Ratio (M‐H, Fixed, 95% CI)	1.19 [0.79, 1.78]
2 Incidence of arm and hand pain	1	249	Risk Ratio (M‐H, Fixed, 95% CI)	0.92 [0.56, 1.50]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Intensity of neck and shoulder discomfort and pain	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Sitting time at 8‐week	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 After shift discomfort rating for neck (4‐8 weeks)	2	186	Mean Difference (Random, 95% CI)	‐0.25 [‐0.40, ‐0.11]
2 After shift discomfort rating for right shoulder or upper arm (4‐8 weeks)	2	186	Mean Difference (Random, 95% CI)	‐0.33 [‐0.46, ‐0.19]
3 After shift discomfort rating for right forearm or wrist or hand (4‐8 weeks)	2	186	Mean Difference (Random, 95% CI)	‐0.19 [‐0.29, ‐0.08]

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 Shoulder Pain Intensity	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Upper Extremity Pain Intensity	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Relative Mouse Use over Total Computer Use	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prevalence of Neck Musculoskeletal symptoms (by questionnaire) at 6‐month follow‐up	2	610	Risk Ratio (M‐H, Random, 95% CI)	0.76 [0.47, 1.21]
2 Prevalence of shoulder musculoskeletal symptoms (by questionnaire) at 6‐month follow‐up	2	610	Risk Ratio (M‐H, Fixed, 95% CI)	0.82 [0.58, 1.17]
3 Prevalence of hand/wrist musculoskeletal symptoms (by questionnaire) at 6‐month follow‐up at 6‐month follow‐up	2	724	Risk Ratio (M‐H, Fixed, 95% CI)	0.63 [0.36, 1.09]
4 Prevalence of neck/shoulder MSD (by medical examination) at 6‐month follow‐up	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
5 Prevalence of hand/wrist MSD (by medical examination) at 6‐month follow‐up	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
6 Intensity of upper extremity pain at 3‐week follow‐up	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7 Frequency of upper extremity pain at 3‐week follow‐up	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8 Duration of upper extremity pain at 3‐week follow‐up	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Outcome or subgroup title	No. of studies	Statistical method	Effect size
1 Frequency of neck ache or pain	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Frequency of shoulder ache or pain	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Frequency of wrist/hand ache or pain	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Intensity of neck ache or pain	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Intensity of shoulder ache or pain	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected
6 Intensity of wrist/hand ache or pain	1	Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Methods	The authors analysed the results as if participants had been individually randomised to groups. However, during the allocation process the authors employed cluster sampling of units. In the article the authors state: "For the allocation of the participants into the intervention and control groups, offices were used as cluster sampling units. Each office was stratified by the number of people working there with a simple random method, and then, the clusters were determined as intervention and control groups".
Participants	116 office workers working in the municipality using computers for at least 10 h, did not have a chronic disease related to the upper body regions, and agreed to participate were included in the study. Not being pregnant was another inclusion criterion for female participants. There was no clear statement on exclusion criteria. We received the additional information from the authors that the number of clusters was 16 in the control group with a total of 58 workers and similarly there were 16 clusters in the intervention group with a total of 58 workers.
Interventions	Intervention group (n = 58) Participatory ergonomic interventions consisting of two stages Stage 1 Conducted in the third month (after start of study) Provided with 2 h training aiming at the development of basic office ergonomics and individual risk assessment skills Introduction to ergonomics and MSDs Adaptation of the work environment to avoid MSDs. Implementation of exercises and relaxation programs to avoid MSDs. Gaining risk assessment skills. Stage 2 Conducted in the forth month In the second stage, the participants in the intervention group were visited at work. During the visit, each employee used the “Hazard Identification‐Risk Assessment Checklist” developed by the researchers to assess their own risk assessment. The participants assessed their own risks through the checklist and produced solutions for those risks. The researchers and the participants together decided on how to implement these solutions. The solutions were implemented by asking questions of “who, where, when, and how” for each solution proposal. This implementation took approximately 15‐20 min for each participant. Control group (n = 58) An educational brochure developed during the study was handed out to the participants in the control group at the end of the study.
Outcomes	Participant reported the severity of symptoms using an 11‐point (ranging from 0‐10) symptoms severity scale. It depicts a human figure referring to various points on the upper body. While 0 indicates that there are no symptoms, 10 indicates that the symptoms’ severity is unbearable. Severity of symptoms in any part of the upper body as ≥ 5, it was decided that the symptoms outcomes developed for used analyses. The presence of symptoms was assessed 13 times on a monthly basis. To decide on the presence of symptoms, the participant’s injury should not be an off‐the‐job injury, he/she should work with the computer 1‐2 h per day at least for 10 days in that month, and if the participant is female, she must not be pregnant. To assess disability/symptom, two measurement tools were used. 1. Northwick Park Neck Pain Questionnaire (NPNP) NPNPQ was adapted from the Oswestry Low Back Pain Disability Questionnaire. It has nine items. Each item is scored from 0 to 4. The questionnaire questions neck‐related functional difficulties. Increasing scores indicate disability. 2. Quick Disability of the Arm, Shoulder and Hand Questionnaire (Quick DASH) The quick DASH assessment form questions the ability level to perform daily activities, symptoms, sleep, work, and the limitation in performing daily activities. The disability/symptoms section consists of 11 questions. The response scale ranges from 1 to 5. It also consists of a four question “work module.” Responding to this section is optional. Increasing scores indicate disability.
Notes	The authors kindly provided the data on the number of clusters (departments) that were randomised as having been 32 altogether, with 16 in the intervention and 16 in the control group.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	There is no mention of how randomisation was obtained, risk of selection bias.
Allocation concealment (selection bias)	Unclear risk	There was no allocation concealment, risk of selection bias.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Blinding of participants and personnel was not possible as intervention included participatory approach. The control group was not given any intervention during the study period.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	There was no blinding of outcome assessment as all the measures are self‐reported, symptoms and disability.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	There was equal attrition from both the intervention and control group (three from each group). The reason for the attrition was not clearly stated.
Selective reporting (reporting bias)	Low risk	All findings were reported.
Other bias	High risk	1. Cross‐contamination of intervention effects may be an issue. The distance between the intervention and control groups were not indicated in the report. 2. The baseline characteristics comparing the intervention group and the control group were not presented. Only the results that the P value is more than 0.05, comparing the two groups on severity of symptoms and disability. 3. The analyses did not take clustering into account which creates a unit of analyses error.

Methods	RCT. The participants were randomly assigned to 1 of 3 study groups
Participants	The sample of 154 subjects was selected at random from a list of volunteers who were employed as agents at the centralised reservation facility for a large international transportation company. These individuals used computers at least 5 hours per work day. All of these individuals performed similar work tasks at similar workstations. Participatory Education Intervention (n = 50) Traditional education intervention (n = 51) Control (n = 53)
Interventions	The study compared participatory education interventions, traditional education interventions, and no intervention Participatory Education Interventions It involved active learning sessions, incorporating discussions and problem‐solving exercises to aid the participants in applying ergonomic concepts to the work environment. It should be noted that the content was similar to that provided to the traditional group but the method of presenting the information differed. The educational sessions for this group lasted approximately 2 hours. The first portion of the educational session incorporated hands‐on demonstration of workstation evaluation and modification. Through case studies, the participants used a problem‐solving approach to recognise ergonomic problems and recommend solutions to address the problem. The second portion of the session paired participants and returned them to their work areas to evaluate and modify the areas according to the information received during the first portion of the session. The modifications were made under the supervision of the instructor for the course who provided assistance to ensure that the newly arranged work areas were consistent with the principles taught in the class. Traditional education It involved a 1‐hour education session that consisted of a lecture and informational handouts about office ergonomics. The education for this group included information about basic muscle physiology, ideal neutral postures, basic task analysis, recommended office equipment location, recognition of problems related to incorrect equipment placement, and general wellness information related to exercise, nutrition, and smoking. A brief question and answer session was included at the end of the session. Control group/no intervention The control group did not participate in any education sessions.
Outcomes	Primary outcome Upper body pain/discomfort composite scores at baseline and at 3, 6, and 12 months' post‐intervention. The discomfort scores ranged from 1 to 4 for each body part for pain and discomfort during the past week (1 = never, 2 = occasional, 3 = several times per week, 4 = several times per day). The upper body composite score included neck, upper back, shoulder or upper arm, forearm, and wrist/hand. Secondary outcome Compliance ‐ work area configuration composite score at baseline and at 3, 6, and 12 months' post‐intervention
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The method of randomisation was not described in the study. The only information provided was: "The participants were randomly assigned to one of three study groups".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Blinding of participants and personnel was not possible as intervention included educational sessions.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	Upper body pain/discomfort composite score was self‐reported and subjective.
Incomplete outcome data (attrition bias)   All outcomes	High risk	The attrition rate was not even across the 3 groups. No ITT analysis mentioned. The attrition rate for both of the intervention groups was more than double that of the control group (23%–24% for the intervention groups vs 11% for the control group).
Selective reporting (reporting bias)	Low risk	Reported on all findings. According to the authors: "there were no significant differences noted across groups for work area configuration, worker postures, or overall observation scores".
Other bias	High risk	1. Cross‐contamination of intervention effects owing to close proximity of the workstations 2. There was no information on baseline characteristics comparing the 2 intervention groups and the control group

Methods	Cluster RCT. Workers were assigned to the experimental or reference group (no intervention) on the basis of the units in which they worked. 40 administrative and geographic units were randomised to the experimental group or reference group. The units were stratified before randomisation on the basis of the number of clerical workers (< 20 and ≥ 20) and type of services (administrative and teaching) in order to ensure equal distribution of these features in each group.
Participants	The study population composed of workers employed in a large university (90%) and in other institutions involved in university services (10%). Eligible workers were those working 5 hours or more per week with a VDU. 627 workers (81% of the people eligible at baseline) participated in both data collection periods (baseline and 6 months). They consists of: PRECEDE intervention group (n = 284); reference/no intervention group (n = 343).
Interventions	The study compared PRECEDE intervention vs no intervention. PRECEDE intervention group The ergonomic training programme was developed according to the PRECEDE model. “The objective of the programme was to act on characteristics of the work environment and the workers that determine behaviour in order to motivate and to enable the workers to improve the ergonomic features of their workstation. Predisposing factors relate to knowledge, beliefs, attitudes, and values; enabling factors relate to skills and material resources; and reinforcing factors relate to support provided by the environment.” The programme targeted the following 3 types of behaviour: adjusting the postural components of the workstation correctly; adjusting the visual components of the workstation correctly; and organising work activities in a preventive manner The programme composed of 2 sessions of 3 hours each with a 2‐week interval The sessions involved demonstrations, simulations, discussions, and lectures. In addition, each worker had to do a self‐diagnosis of his (her) workstation using a photograph taken of him (her) at work before the programme started. Each session was presented to about 15 workers with their supervisor at one time. The presence of the supervisor aimed at providing an organisational environment that was supportive of actions taken by the workers. The 2‐week interval allowed the workers to apply knowledge and skills learned at the first session and to return to the second training session with questions and experiences to discuss. The trainers were 4 occupational health and safety professionals working for the employer and 1 occupational health and safety union representative. Reference/no intervention group The reference group did not receive the training until the completion of the study
Outcomes	Primary outcome Neck‐shoulder and hand‐wrist musculoskeletal symptoms were assessed using a self‐administered questionnaire and by physical examination by physician. The measurements were performed 2 weeks before and 6 months after the intervention in both groups. The prevalent MSDs on the questionnaire were defined as those that were present on 3 days or more during the last 7 days and for which the intensity of pain was greater than half the visual analogue scale among subjects with no history of inflammatory disease or acute injury at the relevant anatomical site. The physical examination by physician was performed on workers who reported symptoms meeting the case definition. The physical examination was conducted according to a standard protocol by a trained occupational therapist blinded to the participant's assigned group. The physical examination was performed 2 to 5 weeks after the completion of the self‐administered questionnaire. Secondary outcome Compliance with the intervention
Notes	The information for the neck‐shoulder and hand‐wrist musculoskeletal symptoms was available for the 2 groups (i.e. < 40 years and ≥ 40 years) combined comparing before and after intervention, and for 3 anatomical regions combined (neck or shoulder, wrist/hand and lower back) comparing intervention and reference before and after intervention. No information was available for neck or shoulder and wrist or hand alone comparing the effect of intervention and reference group. The author provided the additional data; i.e. Intracluster Correlation Coeficient (ICC), prevalence of musculoskeletal symptoms for all subjects for each site collected by questionnaire at baseline, and prevalence of musculoskeletal symptoms for all subjects for each site collected by medical examination at baseline and follow‐up.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	There was no information on sequence generation. The method for randomisation was clearly described.
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Participants were not blinded to the allocation as the intervention consisted of training.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	Although the physical examination was performed by trained occupational therapists blinded to the subjects' assigned group, the examination was only performed on workers who reported symptoms meeting the case definition which was based on self‐reporting/subjective symptoms.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Although there was no mention of ITT, the percentages of participants were high at each measurement (88% and 94%). And according to the author "The percentages and reasons for non‐participation were comparable in the experimental and reference groups".
Selective reporting (reporting bias)	Low risk	All outcomes were reported in the results.
Other bias	High risk	There was no information on baseline characteristics comparing the 2 groups, so the success of randomisation could not be ascertained.

Methods	RCT. Participants were randomised into 1 of 4 intervention groups. The randomisation was done by means of a computer‐generated permuted‐block sequence.
Participants	Participants consisted of employees working at a large aerospace engineering firm in California, US that estimated working at a computer for at least 20 hours per week and were employed as an member of the engineering staff (93%) or a professional position supporting engineering (7%) and had completed the health questionnaire and at least 4 weekly surveys. Since 1 of the mouse interventions could only be used right‐handed, only those who agreed to use their right hand for the mouse pointing device intervention were eligible for the study. Out of a total of 437 eligible employees, 206 people volunteered. The participants were randomised into 4 groups: alternative mouse with a forearm support board (n = 51); conventional mouse with a forearm support board (n = 51); alternative mouse alone (n = 52); conventional mouse alone (n = 52). 154 people volunteered for the nerve conduction testing
Interventions	The study compared 4 different interventions for computer workstations Alternative mouse with a forearm support board: the forearm support board was a large butterfly‐shaped board (36 by 21 inches) that was attached to a desk and provided padded forearm support (ButterflyBoard, Metamorphosis Design and Development, Atlanta, GA, US). The board was inclined upwards at approximately 5° and the surface could accommodate a keyboard and mouse, and the alternative mouse was a 3M product that had a vertical handle for grasping and a flat base to support the ulnar side of the hand and used a roller ball for tracking. The forearm was in approximately 15° of pronation during use (Renaissance Mouse, 3M Corporation, St Paul, MN, US). Conventional mouse with a forearm support board: forearm support board (as in (1)) and conventional mouse used an optical LED for tracking the mouse movement and required the hand to be in an almost fully pronated posture during operation (IntelliMouse Optical, Microsoft Corporation, Redmond, WA, US). Alternative mouse alone: the alternative mouse was a 3M product that had a vertical handle for grasping and a flat base to support the ulnar side of the hand and used a roller ball for tracking. The forearm was in approximately 15° of pronation during use (Renaissance Mouse, 3M Corporation, St Paul, MN, US) (as in (1). Conventional mouse alone: conventional mouse using an optical LED for tracking the mouse movement and required the hand to be in an almost fully pronated posture during operation (IntelliMouse Optical, Microsoft Corporation, Redmond, WA, US) (as in (2)).
Outcomes	Primary outcomes Incidence of MSD: subject reported a discomfort intensity level of > 5 on the weekly survey, or used a pain medication for ≥ 2 days per week for upper body discomfort that they thought was related to computer work was referred for an examination. The examination protocol focused on the body region with discomfort and was performed by 1 physician who was blinded to the intervention status. The examination protocol assessed for the presence of 40 upper extremity and neck MSDs. Mean discomfort score: the discomfort scores were assessed for 3 body regions, the neck or shoulders, right elbow/forearm or wrist or hand, and left elbow/forearm or wrist or hand, were assessed for the worst discomfort during the preceding 7 days using a 0 to 10 point scale (0 = no discomfort; 10 = unbearable discomfort). Subjects were asked whether they thought the discomfort was the result of (a) working on a computer, (b) an acute injury at work, or (c) activities or an injury away from work. Only discomfort reported by the subject as a result of working on their computer was included in the data analysis. The mean discomfort scores for pre‐intervention and post‐intervention (pre‐intervention mean discomfort scores were obtained from the weekly surveys before intervention by averaging all the pre‐intervention scores for each subject to a single value; post‐intervention discomfort scores were obtained from the weekly surveys after intervention. These scores were collapsed into a single postintervention score by body region. The first 8 weeks of post‐intervention scores were left‐censored).
Notes	The study was reported in 2 papers (see Conlon 2008).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Participants were randomised into one of four intervention groups. The randomisation was done by means of a computer‐generated permuted‐block sequence".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Blinding of participants not possible given that different equipment was tested in the 4 groups.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	MSDs: although the examination was performed by 1 physician who was blinded to the intervention status, the pre‐examination criteria for inclusion in the examination was determined by subjective discomfort levels.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	The analysis followed an ITT protocol. As participant exited the study they completed the exit questionnaire.
Selective reporting (reporting bias)	Low risk	All the results for musculoskeletal discomfort, MSDs, and distal motor latency were reported.
Other bias	High risk	Those who volunteered for the study were different: "females were more likely to volunteer for the study than males (P < 0.01)" "participants had higher levels of right arm and neck/shoulder discomfort (P < 0.01)" "participants were also more likely to take medications for discomfort related to work and had higher estimates of the number of days at work that were affected by discomfort (P = 0.05)". Owing to this the effect may be larger than expected.

Methods	Cross‐over RCT. Data were collected over a 16‐week period. The 16‐week period was divided into 4, 4‐week phases in which participants alternated between the conventional (C) and supplementary (S) rest break schedules. Half of the volunteers from each shifts (day and night) were assigned at random to experience the C‐S‐C‐S order of rest break schedules and the other half were assigned at random to experience the opposite (S‐C‐S‐C) order. As a result of attrition, data from just the first 2 phases of the study were sufficient for analyses (i.e. the C‐S phases).
Participants	Data‐entry operators (seasonal employees) working at an Internal Revenue Service Center. The data‐entry task entailed keying mostly numeric data from paper tax forms using a standard keyboard with a right‐sided numeric keypad. A total of 101 data‐entry operators provided written voluntary, informed consent to participate in the study. Each data‐entry operator had been hired as a 'seasonal' employee under an agreement that the job was temporary. The time at which each operator was released from employment was determined by the workload demands of the facility.
Interventions	The study compared supplementary breaks with conventional breaks Control: the conventional break schedule included one 15‐minute break in the middle of the first half of the work shift and one 15‐minute break in the middle of the second half of the work shift. Intervention: the supplementary break schedule included the same 15‐minute breaks, and also included a 5‐minute break during each hour of the work shift that otherwise did not contain a break. For each 8‐hour shift, the supplementary schedule provided 4 extra 5‐minute breaks for a total of 20 extra minutes of break time. Under each schedule, a 30‐minute lunch period, additional to the 8‐hour work and break time, occurred in the middle of the shift.
Outcomes	Primary outcome Musculoskeletal discomfort ratings for several parts of the body, including the neck, shoulders, upper arms, elbows, forearms, wrists, hands, back, buttocks, and legs. Each rating was made using a 5‐point category rating scale in which the whole numbers 1 to 5 indicated ratings of 'none at all', 'a little', 'moderate', 'quite a bit', and 'extreme', respectively. Secondary outcomes Data entry productivity: 2 measures of productivity, keystrokes per hour and the total number of documents entered by each participant on each day of the study. This measure, which was affected by factors such as the length of tax documents entered and the number of hours worked per day, permitted an assessment of work output. Data accuracy: 2 measures of data‐entry accuracy were used for this study. One was the number of errors made per day by each participant. The other was a daily measure of accuracy percentage, which took into account the number of documents entered per day.
Notes	The author kindly provided additional data on mean and standard deviation for the outcomes after the experimental supplementary breaks condition and after the control condition.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	There was no information on sequence generation. The only information available was: "A within‐subjects/repeated measures design was used … Half of the volunteers from each shift (day and night) were assigned at random to experience the C‐S‐C‐S order of rest break schedules, and the other half were assigned at random to experience the opposite (S‐C‐S‐C) order".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Blinding not possible, but the risk of performance bias was assessed as low as the intervention consisted of a strict protocol. The study participants "...use custom‐made electrical timers, attached to the top of each video display terminal, to automatically signal their scheduled breaks".
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The outcome has only subjective symptoms, i.e. musculoskeletal discomfort ratings (feeling state).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Out of the 101 people who volunteered to participate in the study only 42 participants were included in the final analysis. Only the data from the first (first cross‐over) of the 2 phases were sufficient for analysis. Data from the second phase (second cross‐over) were not analysed. Loss to follow‐up amounted to 38 participants and the reasons cited were release from employment and resignation from employment. Questionnaires from 21 participants were too incomplete for analyses.
Selective reporting (reporting bias)	Low risk	The outcomes listed in the methods section were reported in the results.
Other bias	Low risk	The authors reported that "to minimize the potential influence of carry‐over effects and 'Hawthorne effects'… Data from the first 2 weeks of each 4‐week phase were excluded from analyses of the feeling state questionnaire items".

PERMALINK

Ergonomic interventions for preventing work‐related musculoskeletal disorders of the upper limb and neck among office workers

Victor CW Hoe

Donna M Urquhart

Helen L Kelsall

Eva N Zamri

Malcolm R Sim

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Interventions and specific comparisons

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Study Design

Location and settings

Type of work

Type of interventions

Physical ergonomic interventions

Organisational ergonomic interventions

Cognitive ergonomic interventions

Training programmes

Multifaceted ergonomics interventions

Follow‐up period

Outcomes

Unit of analysis

Dealing with missing data

Excluded studies

Risk of bias in included studies

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Overall risk of bias per study

2.

3.

Effects of interventions

Summary of findings for the main comparison. Arm support combined with alternative mouse versus conventional mouse alone.

Summary of findings 2. Arm support with conventional mouse versus conventional mouse alone.

Summary of findings 3. Alternative mouse alone versus conventional mouse alone.

Summary of findings 4. Alternative workstation adjustment compared to no workstation adjustment.

Methods	Cross‐over RCT. Approximately half (23) of the volunteers in each exercise condition were assigned at random to work for 4 weeks under conventional schedule and then switch to the supplementary schedule for the second 4‐week phase. The remaining 22 volunteers in each exercise condition were assigned at random to experience the opposite sequence of rest break conditions.
Participants	Data‐entry operators (seasonal employees) working at an Internal Revenue Service centre, Cincinnati, OH, US. The study sample was recruited from 1 area of the centre containing workstations for 101 individuals, 90 of whom volunteered to follow the study protocol.
Interventions	The study compared supplementary breaks with conventional breaks. Half of the 90 volunteers were assigned at random to the stretching exercise condition and half were assigned to the no‐stretching exercise condition. The 8‐week study period was divided into two 4‐week phases in which all participants alternated between the conventional and supplementary rest break schedules The conventional break schedule included one 15‐minute break in the middle of the first half of the work shift and one 15‐minute break in the middle of the second half of the work shift. The supplementary break schedule included those same 15‐minute breaks, and also included a 5‐minute break during each hour of the work shift that otherwise did not contain a break. For each 8‐hour shift, the supplementary schedule provided 4 extra 5‐minute breaks for a total of 20 extra minutes of break time. All participants were encouraged to get up and walk away from their workstations during each break, regardless of their assigned break schedule or exercise condition. Under each schedule, a 30‐minute lunch period, additional to the 8 hours of work and break time, occurred in the middle of the shift. Participants in the exercise condition viewed a demonstration of the stretching exercises performed by the principal investigator with opportunities for questions and answers. They also kept a paper copy of exercise instructions at their workstations. They were instructed to do the stretches at the beginning of each break in the order specified in the instructions. The first 6 stretches were performed while seated and the last 3 stretches could be done while standing or walking. The 9 stretches required no more than 2 minutes to complete.
Outcomes	Primary outcome Musculoskeletal discomfort ratings (feeling state) for several parts of the body, including the neck, shoulders, upper arms, elbows, forearms, wrists, hands, back, buttocks, and legs. The musculoskeletal discomfort was made using a 5‐point category rating scale in which the whole numbers 1 to 5 indicated ratings of 'none at all', 'a little', 'moderate', 'quite a bit', and 'extreme', respectively.
Notes	The data for the conventional and supplementary break cycle consists of the combination of participants in both exercise and no exercise groups. The effect of breaks alone cannot be isolated. The author provided additional data on mean and standard deviation.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	There was no information on sequence generation. The only information available is that "… the exercise group and the non‐exercise group… were assigned at random to work for 4 weeks under the Conventional schedule and then switched to the Supplementary schedule for the second 4‐week phase" and "approximately half (23) of the volunteers in each exercise condition were assigned at random to work for 4 weeks under the Conventional schedule and then switched to the Supplementary schedule for the second 4‐week phase".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Blinding was not possible but the risk of performance bias was deemed low for a rest‐break cycle as the implementation consisted of a strict protocol. The participants "use custom‐made electrical timers, attached to the top of each video display terminal, to automatically signal their scheduled breaks". However, as this study compared 2 exercise regimens that were not blinded, the risk of bias was deemed high for the combination of the 2 interventions.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The outcome had only subjective symptoms, i.e. musculoskeletal discomfort ratings (feeling state).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Out of the 90 who volunteered to follow the study protocol only 51 were deemed to have complete data for analysis. According to the text "An individual's data set was deemed incomplete if more than 4 consecutive days of questionnaires were missing, or if more than a total of 8 days of questionnaires were missing from either the first or second 4‐week period of the study".
Selective reporting (reporting bias)	Low risk	The risk of selective reporting (reporting bias) was deemed low as all outcomes were reported, the author reported on non‐significant outcome: "In the stretch group, workers reported stretching during only 25% of conventional breaks and 39% of supplementary breaks, and no significant effects of stretching on discomfort or performance were observed".
Other bias	High risk	There was no comparison of the 2 intervention groups. Potential of carry‐over effect, as the authors did not state having used a wash‐out period.

Methods	RCT. Randomisation occurred following evaluation of workplace and ergonomic variables. The use of a random number table assured that each subject entering the study had an equal probability of being assigned to each of the 3 groups. Randomisation was done in blocks of 6 to assure equal numbers of participants in each of the study groups.
Participants	A person eligible for inclusion in this study was: a newly hired worker who: anticipated using a single computer workstation for 15 hours or more per week and anticipated using a computer workstation for at least as many hours per week as in his/her previous job working at insurance and financial companies, food product producers, and universities in metropolitan Atlanta, GA, US who had reported experiencing arm or hand symptoms during the week prior to intervention. Of the 447 eligible for health screening, a total of 379 individuals were eligible for inclusion into 1 or both cohorts (those who did not report experiencing arm or hand pain and neck or shoulder pain during the week prior to the study. 375 people were randomised into the arm and hand cohort and 356 were randomised into the neck and shoulder cohort.
Interventions	The study compared alternate intervention, conventional intervention, and no intervention A study staff member reconfigured the subject's workstation if the subject was randomly assigned to either the alternative or conventional interventions (groups A or B). Verbal and written instructions describing the desired posture were provided to all group A and B participants. At 3 days and 1 week after the intervention, study staff returned to the participant's workplace to check on continued maintenance of the posture. If the posture had changed from the intervention, additional workstation changes were made and additional instruction given. Group A: alternate intervention The workstation was adjusted according to the following configuration: Head tilt angle ≤ 3º (head tilt angle is defined as the angle formed between a line defined by the tragion of the ear and the infraorbitale of the eye and the horizon. To clarify the meaning of head tilt angle values, increasing neck extension results in larger values for head tilt angle and increasing neck flexion results in smaller (including negative) values) head rotation < 15º in either direction (L/R) J key at least 2 cm below elbow height keyboard inner elbow angle of > 120º J key at least 12.5 cm from edge of desk or work surface keyboard wrist ulnar deviation of 0º to 220º (i.e. up to 20º radial deviation) armrest present keyboard wrist rest present mouse wrist ulnar deviation of 25º to 5º mouse wrist extension of 20º to 30º mouse next to keyboard high‐quality chair present. Characteristics of high‐quality chair: easily (pneumatically) adjustable for height, adjustable height backrest, full contoured backrest, adjustable seat pan angle, round waterfall seat pan edge, 5‐legged base Group B: conventional intervention The workstation was adjusted according to the following configuration: eye height level with top of monitor screen head rotation < 15º in either direction (L/R) J key at least 3 cm above elbow height keyboard shoulder flexion of 210º to 20º keyboard shoulder abduction of 210º to 20º keyboard inner elbow angle of 80º to 100º keyboard wrist ulnar deviation of 210º to 10º keyboard wrist extension of 210º to 10º keyboard wrist rest present mouse wrist ulnar deviation of 210º to 10º mouse wrist extension of 210º to 10º armrest present high‐quality chair present Group C: no intervention Participants instructed to continue keying in their usual posture and no changes were made to their workstations.
Outcomes	Primary outcome Time to event: symptoms of pain or discomfort ‐ participants were classified as having experienced musculoskeletal symptoms if they (1) reported musculoskeletal discomfort on any day of the week with a severity of ≥ 6 on the 0 to 10 VAS or (2) reported musculoskeletal discomfort on any day of the week for which they took medication (over‐the‐counter or prescription). Study participants were followed for each outcome separately until they became symptomatic (censored). Development of a symptom in one anatomic area did not stop the collection of data for the other anatomic area. Two separate, overlapping cohorts were then defined to examine separately the risks of neck or shoulder symptoms and the risks of arm or hand symptoms. Secondary outcome Compliance: using a standard checklist, each workstation was evaluated for presence of specific items (e.g. mouse or other pointing device), and the adjustability of specific equipment. Following completion of the checklist, dimensional and angular measurements (e.g. seated elbow height, table surface height, keyboard inner elbow angle) were recorded.
Notes	Gerr 2005 consisted of two overlapping cohorts. The effect of the intervention was assessed as arm/hand and neck or shoulder pain.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The use of a random number table assured that each subject entering the study had an equal probability of being assigned to each of the 3 groups. Randomisation was done in blocks of 6 to assure equal numbers of participants in each of the study groups.
Allocation concealment (selection bias)	Unclear risk	No information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	There was no mention of blinding and the methods of intervention consisted of 2 distinct workstation and postural interventions.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	Outcomes consisted of subjective symptoms measured with a checklist.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Participants contributed data to their assigned intervention group regardless of compliance (i.e. data were analysed by ITT).
Selective reporting (reporting bias)	Low risk	Key findings: there were no significant differences in the incidence of musculoskeletal symptoms among the 3 intervention groups.
Other bias	High risk	Large number of drop‐outs. "There were a large number of drop‐out/lost to follow‐up in arm/hand cohort – 147 (41% of those followed) were lost during the six month follow up period … No differences were observed in dropout rates (i.e. incomplete follow‐up) across the three intervention groups". Although the drop‐out rates were similar across the 3 randomised groups, there were a large number of drops‐outs in each group (36 to 42 across all 6 groups) for which the authors did not provide a reason.

Methods	RCT. "Following baseline assessments, participants were assigned by one member of the research team to a treatment arm using a randomised block design and random number table. Departments served as blocks and participants within departments were randomly assigned at the individual‐level to an arm. Assignment of individual participants within each department alternated between arms (i.e. intervention, control, intervention, control… )"
Participants	"Office workers from one organisation (Liverpool John Moores University, Liverpool, UK) ... Consent was sought from 11 departmental managers ... Departments were located across four buildings with varying office layout (open‐plan, individual offices or a combination). Employees within the approached departments were predominantly administrative staff. Inclusion criteria a) full‐time member of staff, b) access to a work telephone and desktop computer with Internet, c) no cardiovascular or metabolic disease, d) not taking any medication, e) not pregnant and, f ) no planned absence >1 week during the trial" 503 emails were sent to invite participants from the 11 departments: 54 responded, 6 were excluded, 1 withdrew for medical reasons, 47 were randomised.
Interventions	"Treatment arms included a sit‐stand workstation intervention group (each participant received a sit‐stand workstation) and a control group (usual practice)." "A single (manufacturer’s suggested retail price £360) or dual (£375) monitor WorkFit‐A with Worksurface + workstation was installed, dependent on the number of monitors the participant had. The computer monitor(s) and keyboard were housed on the workstation and the workstation could be quickly raised up and down by hand to enable seated or standing work."
Outcomes	Primary outcome Levels of discomfort or pain Assessment methods: "Using a questionnaire adapted from a previous trial, participants rated their current level of discomfort or pain at three sites (lower back, upper back, neck and shoulders) on a Likert scale from 0 (no discomfort) to 10 (extremely uncomfortable)." Time of assessment: Baseline and 8‐week Secondary outcome Sitting, standing and walking time Assessment methods: "The EMA diary assessed time spent sitting (primary outcome), standing, walking and in other activities during work hours over 5 days (Monday‐Friday). At 15‐minute intervals participants used a paper‐based diary to record their main behaviour in response to the question: “What are you doing right now?” The behaviour options were sitting, standing, walking or other. If other was selected, participants were instructed to write the activity they were doing." Time of assessment: Baseline, 4‐week and 8 week
Notes	The level of discomfort and pain is not the main outcome for the study, the study was "Associated effects on vascular and metabolic disease risk markers were evaluated, as was the acceptability and feasibility of sit‐stand workstations in a real office setting." The information presented in this review only consists of level of discomfort of the neck and shoulders.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was based on random block design and random number table, "Following baseline assessments, participants were assigned by one member of the research team to a treatment arm using a randomised block design and random number table".
Allocation concealment (selection bias)	High risk	The allocation of participants to the intervention and control arm were not concealed; "Assignment of individual participants within each department alternated between arms (i.e. intervention, control, intervention, control… )".
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Participants were not blinded to the allocation as the intervention consists of installing a new sit‐stand workstation: “After baseline assessments, each participant had a sit‐stand workstation installed on their existing workplace desk”.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The main outcome assessment on the discomfort and pain: “Using a questionnaire adapted from a previous trial, participants rated their current level of discomfort or pain at three sites (lower back, upper back, neck and shoulders) on a Likert scale from 0 (no discomfort) to 10 (extremely uncomfortable)”.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	ITT was only performed as a sensitivity analysis, “For workplace sitting, standing and walking, the per‐protocol analysis was compared with an intention‐to‐treat analysis, as a sensitivity analysis.” However, the attrition rate for the discomfort and pain outcome were low Intervention ‐ 25/26 Control – 21/21
Selective reporting (reporting bias)	Low risk	No information suggestive of selective reporting, all outcome were reported in the results section
Other bias	High risk	The number of males to females in the intervention and control groups was not equally distributed, there were only around 10% (3/26) males in the intervention group as compared to 30% (7/21) in the control groups.

Methods	RCT. “A prospective two‐group experimental design with a delayed intervention for the control group was used” … “Because the size of the training classes was limited to no more than 25, participants were randomly assigned to one of four training groups. Two training groups were combined to form the intervention group and two training groups formed the control group.”
Participants	Participants included all employees in the unit who worked at a computer at least 10 hours per week in an organisational unit of a large state university in southeast US. Employees diagnosed by a physician as having an acute musculoskeletal injury or trauma to the trunk or upper extremities within the previous 6 months were excluded from participation. Employees being treated by a healthcare professional for cervical or upper extremity disorders were excluded from participation. 87 employees participated in the study.
Interventions	The study compared active ergonomic training with no intervention. Active ergonomic training (AET) The AET programme consisted of a total of 6 hours of didactic interactions, discussion, and problem‐based activities. The AET group met on 2 days in the same week for 3 hours per session. The AET programme occurred during working hours and employees participated on company time. Key elements of the AET programme were: skill development in problem‐solving for ergonomic workstation issues; active participation; integration of multiple prevention strategies. No intervention (control) The participants did not received intervention until week 4 of the study
Outcomes	Primary outcomes Musculoskeletal symptoms: participants were first asked if they had experienced musculoskeletal symptoms in the past year in: (a) head, (b) neck, (c) shoulder and upper arm, (d) elbow/forearm, (e) wrist, hands/fingers, or (f) upper back. Regional composite scores were computed to provide an impression of symptoms in a functional region. Scores from the head, neck, and upper back were combined to describe symptoms in the upper spine. Scores from the shoulder or upper arm, elbow/forearm, wrist, and hand were combined to describe symptoms in the upper extremity. Intensity of pain: for each symptomatic body region, an ordinal scale was used ranging from 1 = mild pain to 4 = worst ever. A score of 0 was assigned for asymptomatic body regions. Frequency of pain: an ordinal scale that ranged from 1 = once in the past week to 4 = daily in the past week was used. If no discomfort was present in a body region, a score of 0 was assigned. Duration of pain: an ordinal scale that ranged from 1 = < 1 hour to 4 = > 3 days to 1 week was used. If no discomfort was present in a body region, a score of 0 was assigned.
Notes	The authors reported results for both the randomised and delayed intervention given to the control group (at week 4). From week 0 to week 3 the groups were treated according to their randomisation to the AET programme group and the control group. On week 4 the control group were also given the AET programme. We only included data from week 3.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	There was no information on sequence generation and the randomisation was not adhered to in the allocation of participants. "After participants were randomly assigned to groups, the physical proximity of participant work locations in the intervention and control groups was assessed. To minimize the diffusion of treatment effects, participants from the same work location were assigned to the same study group (intervention or control)".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	The participants and personnel were not blinded. The purpose of this study was to evaluate the effectiveness of an (AET programme in computer users. Subjects participated in a 6‐hour training intervention at their workplace.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The outcome consists of subjective symptoms of pain or discomfort.
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	There was no information on ITT analysis and loss to follow‐up for the RCT part of the study. After the third week the control group were given the same intervention.
Selective reporting (reporting bias)	Low risk	No significant differences were found for intensity of symptoms, frequency of symptoms, or duration of symptoms in any body region immediately post intervention.
Other bias	Low risk	There was no significance difference between the main outcome measures between intervention and control groups; i.e. intensity, frequency and duration of musculoskeletal symptoms at baseline.

Methods	RCT. "Participants were randomly allocated to a control group (n = 12) or an intervention group (n = 11). One month after T0 data collection, participants were allocated to either study group using simple randomisation with a 1:1 ratio."
Participants	Fifty‐six office workers from a research organisation (the Institute for Work & Health, Ontario, Canada) of 74 employees met the inclusion criteria and were invited to participate in the study. Twenty‐three of the 56 invited agreed to participate. Group 1 (intervention) (n = 11) Group 2 (control) (n = 12)
Interventions	The Hoverstop1 mouse (Vibramouse 2011) provided feedback to the worker by gently vibrating if the worker’s hand had been idle on the mouse for more than 12s. The vibration lasted for a maximum of 4 s. There was no minimum vibration time. The mouse would vibrate until the hand was removed, a mouse button was clicked, the scroll wheel was activated or the maximum vibration time was met (4 s). The feedback was a reminder to rest the arm in neutral postures when not in use. Unlike other break software, the Hoverstop1 does not deliver break messages to the user visually on the monitor. All participants received the alternative mouse with the vibration mechanism turned off during T0 measurements. The vibration mechanism remained turned off in the control group after baseline measurements. All participants received the corresponding Hoverstop1 monitoring software to monitor individual mouse, keyboard and computer activity continuously. All participants were invited to attend a 1‐h study information session with time for questions and answers; attendance was optional. The vibration mechanism was initiated in the intervention group one month after T0 and remained active until the end of the study. The intervention group members were invited to watch a video produced by the manufacturer about the intervention device at their workstation, via a link provided in an email delivered by the research coordinator. The video suggests resting the arm on the desk in front of you when not actively using the mouse, to decrease muscle activation in the shoulder and arm.
Outcomes	T0 Baseline data collection (1 month pre‐activation) Activation: randomisation into intervention and control groups T1 Data collection (5 weeks post activation) T2 Data collection (25 weeks post activation) Pain and discomfort: collected using an online Daily Symptom Survey ‐ Participants were asked to rate their pain at that point in time. The DSS includes a body map with the following areas identified: neck, shoulders, upper back, elbow, lower back, lower arm or wrist or hand, buttock or thighs, knees, lower leg/ankles/feet. Pain and discomfort scores were averaged over the three days of administration for analysis. Mouse use (active hand‐on‐mouse time plus idle hand‐on‐mouse time), Keyboard use (active keyboard time), total computer use (mouse use plus keyboard use), Relative mouse use (RMU: mouse use over the total computer use) were "measured using the Hoverstop ® monitoring software in both the intervention and control groups. An electric‐potential transducer registered when a user’s hand was on or hovered just above the mouse. This feature is believed to give a more accurate measure of exposure to the proposed mechanism of discomfort – static (possibly unsupported) mousing postures – than other collection methods that strictly utilise mouse movements and functions. Keyboard use was measured by the duration of each key compression, while total computer use was determined as the sum of keyboard use and mouse use."
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"A statistician not connected to the study used a random number generator in SAS V. 9.2 to generate the random allocation sequence".
Allocation concealment (selection bias)	Unclear risk	The methods of allocation were not the mentioned.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No blinding of participants: “Participants were not blinded to their groups. Those in the intervention group would have been aware of their group due to the nature of the intervention (the vibrating mouse)”.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The main outcome assessment on the “pain and discomfort collected using an online Daily Symptom Survey administered in the afternoon for three consecutive days at each data collection point: T0, T1 and T2. Participants were asked to rate their pain at that point in time”. The participants themselves rate their pain and discomfort.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	“An intention to treat analysis was conducted”, there was also low loss to follow‐up; “There was a low loss to follow‐up (1/23; one participant from the control group), with an additional loss of mouse use data at T2 for two other participants in the control group (see Figure 2 for participant tracking)”.
Selective reporting (reporting bias)	Low risk	No information of selective reporting, all outcomes were reported in the results section.
Other bias	Unclear risk	The demographic data of the participants and the pre‐existing musculoskeletal symptoms of the participants were not collected due to ethical consideration; "Confidentiality and voluntary participation were stressed. Demographics (such as age, gender, job type and preexisting symptoms) were not collected due to ethical considerations, since the study was executed within our research institute." The successful randomisation of the participants was not able to be ascertain due to this reason.

Methods	RCT. After the first measurements the participants were randomly assigned to 3 groups of 7 participants
Participants	The participants were 21 healthy female VDU users without acute musculoskeletal symptoms. They were office employees and researchers with a mean age of 38 years (range 26 to 54 years). The participants had worked with a VDU for more than 20 hours a week for an average of 5 years (range 4 months to 13 years). All the participants were right‐handed but 3 of them operated their mouse with their left hand.
Interventions	The study compared Ergorest articulating arm supports with no arm support. "Ergorest articulating arm supports (Ergorest Ltd, Finland) were used in this study. The arm supports are attached to the table, and the height of the supports can be adjusted. Both arms are settled in the grooves and there is easy mobility. Ergorest arm supports have been developed particularly to reduce static load in the neck and shoulder area" Group 1: "used the basic Ergorest arm support with the mouse pad with the hand that operated the mouse". Group 2: "had Ergorest arm supports for both hands (a basic arm support with the mouse pad for the mouse hand and the basic arm support for the other hand)". Group 3 (control): "had no arm supports, and they were asked to maintain their usual work technique and to avoid all redesign measures at work during the intervention".
Outcomes	Primary outcome Musculoskeletal strain: the participants recorded the severity of their musculoskeletal strain using a VAS, each VAS was reported in millimetres (range 0 to 100 mm with end points of no strain and extreme strain). The mean value of the VAS lines obtained from the 6 body regions (neck, shoulder, upper arm, forearm, wrist, and hand and fingers) were calculated for the right and left sides.
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	There was no information on sequence generation. The authors only mentioned that: "After the first measurements the participants were randomly assigned to three groups of 7 participants".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	High risk	There was no mention of blinding and it may not even be possible as the intervention included supply of new equipment.
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The outcome measure was subjective symptoms for muscle strain.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	There was no loss to follow‐up.
Selective reporting (reporting bias)	Low risk	No statistically significant changes were observed in the musculoskeletal strain scores either between the groups or within the groups.
Other bias	High risk	No comparison of groups on baseline characteristics specific to the outcome measures.

Methods	RCT. Participants were randomly assigned to 1 of 3 experimental groups.
Participants	15 participants were recruited by word of mouth from the accounting (n = 6) and library (n = 6) offices at the University of New Brunswick and from New Brunswick Provincial Government Offices (n = 3) in Fredericton, NB, Canada. All participants were recruited based on their performance of jobs that involved sustained sitting postures in conjunction with keying and data entry tasks. 15 participants participated in the study.
Interventions	The study compared 3 different micro‐break intervals. Upon obtaining informed consent, each participant's workstation was examined for major problems in terms of ergonomic setup and such problems were corrected at least 1 month prior to participation. Ergobreak version 2.2 was installed on each participant's computer at least 2 weeks prior to the data collection period. The programme was set to prompt users to take breaks based on fixed time intervals. Participants were randomly assigned to 1 of 3 experimental groups according to their set time interval between micro‐breaks: all micro‐breaks were of 30 seconds duration. Participants took part in the study over a 4‐week period. For the first 2 weeks of participation (the 'No Break' protocol), subjects performed their usual work while minimising the amount of time spent away from their workstation. For the second 2‐week period of participation each subject performed their assigned micro‐break protocol with the assistance of the Ergobreak software. The programme was set to prompt participants to take breaks at their prescribed time intervals. Group 1: 40‐minute interval group: all micro‐breaks were of 30 seconds' duration with the assistance of the Ergobreak software Group 2: 20‐minute interval group: all were of 30 seconds duration with the assistance of the Ergobreak software Group 3: control group (where participants took breaks whenever they felt they needed to): the Ergobreak software was not set to prompt members of the control group
Outcomes	Primary outcome Discomfort scores: "based on vertical visual analogue scales (VAS), The vertical scale was 100mm in length, and had no numerical anchors along its length with anchors at the top (Worst Possible Discomfort) and at the bottom (No Discomfort). VAS scores were measured by measuring the distance in millimetres between the 'No Discomfort' anchor and the location of the participant's mark on the line. Four scales were placed on the same page and labelled 'Neck', 'Low Back and Buttock', 'Shoulder and Upper Arm' and 'Forearm, Wrist and Hand'. For each body part, the difference in VAS scores (calculated as the VAS score at each measurement time during the No Breaks protocol minus the VAS score at that time during the Breaks protocol)". Secondary outcome Productivity: the number of words typed (sets of 5 keystrokes) over the course of each 3‐hour myoelectrical signal recording session. Word count data were collected at the end of each recording session only.
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	There was no information on sequence generation. The only information available is… "Participants were randomly assigned to one of three experimental groups".
Allocation concealment (selection bias)	Unclear risk	There was no information on allocation concealment.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	There was no mention of blinding but the implementation of the micro‐breaks followed a strict protocol: "Ergobreak version 2.2 was installed on each participant's computer … the program was set to prompt users to take breaks based on fixed time intervals".
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The discomfort outcome was subjective; "the discomfort score data were collected at 40 min intervals throughout the recording session".
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	There was no information on the total participants analysed in each group. Limited information on dropouts and no statistical information on dealing with loss to follow‐up.
Selective reporting (reporting bias)	Low risk	All findings were reported including non‐significant findings. For example, "no significant change in the frequency of MNF [mean frequency] cycling was noted at the shoulder"
Other bias	High risk	There was no information on the comparability of the VAS score at baseline between the groups and there was no data on the success of randomisation and comparability between the participants. The differences between all participants were presented and they showed very large differences in age and years of experience. "All participants were female (although this was not a requirement for participation), between the ages of 23 and 50 (median age 34). The number of years of experience working at a computer terminal or word processor ranged from two to 18 years (median 10 years)." This is hardly surprising as there were only 15 participants in total.

Methods	RCT. This was a 1 year, randomised intervention trial with 4 treatment arms.
Participants	Employees at 2 customer service centre sites (sites A and B) of a large healthcare company were eligible for participation if they performed computer‐based customer service work for more than 20 hours per week and did not have an active workers' compensation claim involving the neck, shoulders, or upper extremities. 182 workers participated in the study.
Interventions	The study compared 4 intervention arms. All the 4 treatment arms included ergonomics training. The ergonomics training involved conventional recommendations, which included maintaining an erect posture while sitting, adjusting the chair height so that the thighs were approximately parallel to the floor, adjusting the arm support and work surface height so that the forearms were approximately parallel to the floor, adjusting the mouse and keyboard location to minimise reaching, adjusting the monitor height so that the centre of the monitor is approximately 15º degrees below the visual horizon and a reminder to take scheduled breaks. The computer workstations used at the sites had independently adjustable keyboard and monitor support surfaces and were typically equipped with a conventional keyboard, computer mouse, and a telephone headset. Use of wrist rests at this workplace was optional. Subjects who were assigned to use the forearm support board could not continue to use a wrist rest owing to the design of the forearm support. Subjects not receiving the forearm support were allowed to continue using a wrist rest if they desired. Chairs were adjustable in height with adjustable height arm rests. Trackball with forearm support board: "the trackball (16.5 cm depth, 8.6 cm width, 4.6 cm height, with a 4 cm diameter ball; Marble Mouse, Logitech, Fremont, CA, US) was installed next to the keyboard. The armboard was a wraparound, padded arm support that attaches to the top, front edge of the work surface (30.5 cm depth, 76.2 cm width, 2.5 cm height; MorencyRest, R&D Ergonomics, Freeport, ME, US). Forearm support board only: the armboard was a wraparound, padded arm support that attached to the top, front edge of the work surface (30.5 cm depth, 76.2 cm width, 2.5 cm height; MorencyRest, R&D Ergonomics, Freeport, ME, US). Trackball only: the trackball (16.5 cm depth, 8.6 cm width, 4.6 cm height, with a 4‐cm diameter ball; Marble Mouse, Logitech, Fremont, CA, US) was installed next to the keyboard. No intervention
Outcomes	Primary outcomes Incidence of upper extremity and neck MSDs: if subjects recorded on the weekly survey a pain intensity level of > 5 or they used medications for ≥ 2 days for upper extremity or neck pain that was not associated with an acute traumatic event (e.g. laceration, fall), then a physical examination of the upper extremities or neck or shoulders was performed by 1 physician who was blinded to intervention status. "An incident disorder was defined as a disorder diagnosed on the physical examination only if the participant did not report pain > 5 in that body region (neck or shoulder, right upper extremity, left upper extremity) on the weekly questionnaire before the intervention". Worst pain during the preceding 7 days for neck or shoulder, right elbow or forearm or wrist or hand, and left elbow or forearm or wrist or hand assessed using a 0‐ to 10‐point scale (0 = no pain; 10 = unbearable pain). Acute injury events during the week ‐ weekly survey. Secondary outcome "The effect of the intervention on employee productivity was also assessed using the employer tracked measures of productivity".
Notes	‐
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation: "this was a one year, randomised intervention trial with four treatment arms" Sequence generation: "the randomisation was done by means of a computer generated permuted‐block sequence and administered by a research associate"
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias)   All outcomes	High risk	There was no blinding of participants or personnel. "This one year, randomised controlled intervention trial evaluated the effects of a wide forearm support surface and a trackball on upper body pain severity and incident musculoskeletal disorders among 182 call centre operators at a large healthcare company. Participants were randomised to receive (1) ergonomics training only, (2) training plus a trackball, (3) training plus a forearm support, or (4) training plus a trackball and forearm support".
Blinding of outcome assessment (detection bias)   Musculoskeletal disorders	High risk	The outcomes included "worst pain during the preceding seven days". Those who reported "pain intensity level of more than 5, or they used medications for two days" were subjected to a physical "examination protocol focused on the body region of pain and was performed by one physician who was blinded to intervention status." Although the second part was blinded, it depended on the subjective reporting.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	The analysis followed an ITT approach. The unavailability of 7 participants for a physical examination may have biased the findings. However, the hazard model for incident neck or shoulder disorders was repeated including these 7 participants as incident cases and the conclusions regarding the armboard were unchanged.
Selective reporting (reporting bias)	Low risk	Reported on all findings
Other bias	Low risk	The baseline characteristics of the participants did not significantly differ by intervention group.

Study	Reason for exclusion
Aaras 1998	Non‐RCT
Amick 2003	Non‐RCT
Amick 2012	Non‐RCT
Chau 2014	Did not report neck and upper limb musculoskeletal outcome
Danquah 2017	> 25% of the participants reported neck‐shoulder pain at baseline (87/171; 50.9%)
De Cocker 2016	Did not report on neck and upper limb musculoskeletal outcome
Driessen 2008	Participants consisted of workers other than office workers: "Participants are workers, both blue and white collar workers, recruited from the departments of four large Dutch companies with at least 3,000 workers each".
Dropkin 2015	> 25% of the participants had musculoskeletal symptoms at baseline. "Additional inclusion criteria were: work at least 4 h/day on a desktop computer, non‐specific neck/ UE musculoskeletal pain (1 or greater on the pain scale described below) at the time of screening".
Esmaeilzadeh 2014	> 25% of the participants had musculoskeletal symptoms at baseline. Study only included participants with Work‐related upper extremity musculoskeletal symptoms (WUEMSS); " ... case definition criteria, 94 of the 311 respondents had WUEMSS and were subsequently included in the interventional study".
Faucett 2002	Study was not conducted in an office environment
Fostervold 2006	> 25% of the participants had neck and shoulder symptoms at baseline. The prevalence of neck and shoulder symptoms at baseline was 73.5% in the intervention group and 75% in the comparison group.
Ketola 2002	> 25% of the participants had neck and shoulder symptoms at baseline. The study included subjects with musculoskeletal symptoms: "One hundred and twenty‐four subjects with musculoskeletal symptoms were selected".
Krause 2010	Did not report on neck and shoulder musculoskeletal symptoms
Levanon 2012	> 25% of the participants had musculoskeletal symptoms at baseline; "as all the participants have at least 1 part of the UE with complaints of pain at baseline".
Mahmud 2011	> 25% of the participants had musculoskeletal symptoms at baseline. The prevalence for of musculoskeletal disorder at baseline for intervention and control groups ranged from 16.3% ‐ 63.6%.
Mann 2013	> 25% of the participants had musculoskeletal symptoms at baseline; "The inclusion criteria of this study was pain, stiffness or tingling in neck and shoulder in the preceding six  months affecting the quality of activities of daily living".
Meijer 2009a	> 25% of the participants had upper extremity musculoskeletal symptoms at baseline. Prevalence for the control group was 49% and 36% for the intervention group.
Meijer 2009b	> 25% of the participants had musculoskeletal symptoms at baseline; 33.3% of the participants have UE complaints at baseline.
Mekhora 2000	> 25% of the participants had musculoskeletal symptoms at baseline; Participants consisted of those with symptoms of above average discomfort: "That is, those with above average discomfort and who had discomfort around the neck and shoulder areas for more than 1 day in the previous year were selected".
Parry 2015	> 25% of the participants had upper extremity musculoskeletal symptoms at baseline. The prevalence of participants reporting musculoskeletal pain in different body regions at baseline ranged from 28‐60%.
Ripat 2006	> 25% of the participants had musculoskeletal symptoms at baseline; "The study population were workers who reported had two or more symptoms of WRUED (i.e. paraesthesia,numbness, loss of strength, shooting sensation or pain, tingling, clumsiness, or night pain)".
Robertson 2013	The intervention was conducted in an laboratory setting, the participants were not performing actual/routine work.
Spekle 2010	> 25% of the participants had musculoskeletal symptoms at baseline; Prevalence of symptoms ‐ 56% , Proximal Symptoms ‐ 46% , Distal Symptoms ‐ 26% at baseline.
Thorp 2014	The intervention was conducted in an laboratory setting, the participants were not performing actual/routine work.

Trial name or title	A workplace exercise versus health promotion intervention to prevent and reduce the economic and personal burden of non‐specific neck pain in office personnel: protocol of a cluster‐randomised controlled trial
Methods	Cluster‐randomised controlled trial
Participants	Office personnel aged over 18 years, who work > 30 hours/week
Interventions	Individualised best practice ergonomic intervention plus 3 x 20 minute weekly progressive neck or shoulder girdle exercise group sessions for 12 weeks.
Outcomes	Primary (productivity loss) and secondary (neck pain and disability, muscle performance, and quality of life) outcome measures will be collected using validated scales at baseline, immediate post‐intervention and 12 months after commencement.
Starting date	1 June 2013
Contact information	Dr Venerina Johnston (v.johnston@uq.edu.au)
Notes