Types of studies

We have included all eligible randomised controlled trials (RCTs), quasi‐RCTs (in which methods of allocating participants are not random, such as alternate allocation or allocation by date of birth or day of the week), and cluster‐RCTs (randomisation of a group of people such as a work group or workplace rather than randomisation of individual people). For some workplace interventions, the implementation of interventions is difficult to apply to an individual, so interventions operate on a group level (Ijaz 2014). In this situation, when the intervention takes place at a group level or within the one organisation where due to workplace or environmental restrictions, randomisation is not possible, we have also included controlled before‐and‐after studies (CBAs), which use a concurrent control group for the intervention. We have included studies reported as full text, those published as abstract only, and unpublished data.

Types of participants

We have included studies conducted with adult workers aged 18 or older, working in sedentary occupations (workers sedentary for more than 50% of the working day), such as seated office workers and laboratory technicians. We have excluded sedentary workers for whom it may not be possible to modify workplace posture, such as transport workers. Studies that did not report the proportion of sedentary time but described workers as ‘sedentary workers’ have been included. When studies included workers from different occupations, we included only results from participants identified as ‘sedentary workers’, or we reported sedentary time of more than 50%. We excluded studies that specifically focused on participants with the following comorbidities or characteristics.

• Inflammatory systematic diseases such as rheumatoid arthritis.

• Diseases of the central nervous system such as stroke and multiple sclerosis.

Sedentary workers who report the presence of musculoskeletal symptoms in at least one of the following regions ‐ cervical spine, mid‐back, lower back, upper limb, hip, or lower limb ‐ have been included as participants with symptoms.

In studies that include a mixture of participants reporting and not reporting musculoskeletal symptoms, only participants with symptoms at baseline have been included in the analyses.

We have included studies conducted with participants who report pain. ‘Participants with pain’ thresholds are defined as:

• ‘yes’ on a dichotomous symptom scale;

• ‘greater than 0’ on a visual analogue symptoms scale out of 10;

• ‘greater than 0’ on a numerical rating scale out of 10;

• ‘greater than 0’ on the McGill Pain Questionnaire;

• ‘greater than 0’ on the 18‐, 23‐, or 24‐point version of the Roland Morris Disability Questionnaire; or

• ‘greater than 0%' for overall score on the Oswestry Disability Index.

Types of interventions

We included trials that evaluated the effectiveness of interventions to reduce or break up workplace sitting by encouraging standing or walking at the workplace. Eligible interventions include the following.

• Interventions targeted at the physical work environment.

  • Provision of an activity permissive workstation (sit/stand or treadmill).

  • Interventions that modify the built environment such as modifications to office layout that encourage standing or walking.

• Interventions targeted at the individual.

  • Behavioural modification or counselling programmes that promote increased standing or walking.

  • Working style interventions that promote standing or walking, such as promotion of 'active' work breaks.

  • Workplace walking programmes including ‘pedometer challenges’.

  • Promoting the use of stairs during work hours.

  • Using break‐reminding software.

• Interventions targeted at the organisation.

  • Workplace policy modifications such as standing meetings and ‘active/walking’ emails.

We included multi‐component trials that combine elements of the above interventions.

We included trials that compare the effectiveness of workplace interventions to increase standing or walking with usual care, with no intervention, or with another active intervention such as specific targeted musculoskeletal interventions.

We excluded interventions that focus on specific strengthening or stretching programmes that do not promote standing or walking. For example, an exercise programme that replaces sedentary time (with standing or walking) would be included as an intervention, whereas a seated exercise programme (seated stretching/strengthening programme) would not be included.

Types of outcome measures

Electronic searches

We conducted a systematic literature search to identify all published and unpublished trials that can be considered eligible for inclusion in this review. The literature search identified studies in all languages. We arranged for the translation of key sections of potentially eligible non‐English language papers, or we arranged that people who are proficient in the language of the publications fully assess them for potential inclusion in the review as necessary.

We searched the following electronic databases from inception to January 2019.

• The Cochrane Central Register of Controlled Trials (CENTRAL), in the Wiley Online Library.

• MEDLINE (PubMed).

• Embase (embase.com).

• National Institute for Occupational Safety and Health's (NIOSH) electronic, bibliographic database of literature in the field of occupational safety and health (NIOSHTIC) (Occupational Safety and Health (OSH)‐UPDATE).

• NIOSHTIC‐2 (OSH‐UPDATE).

• UK Health and Safety Executive Information Services (HSELINE) (OSH‐UPDATE).

• Archived OSH Bibliographic Datbase (CISDOC) (OSH‐UPDATE).

• Physiotherapy Evidence Database (PEDro).

We also conducted a search of unpublished trials at ClinicalTrials.gov (www.ClinicalTrials.gov) and at the World Health Organization (WHO) trials portal (www.who.int/ictrp/en/). We imposed no restrictions on language of publication.

Searching other resources

We checked the reference lists of all primary studies and review articles for additional references. We contacted experts in the field to identify additional unpublished materials.

Selection of studies

We conducted the selection of eligible studies in two stages. First, two review authors (SP and PC) independently screened titles and abstracts of all potentially relevant studies identified through our systematic search to identify studies for inclusion. The same review authors coded them as 'include' (eligible or potentially eligible/unclear) or 'exclude'. At this stage, we excluded all references that clearly do not fulfil our inclusion criteria or that do fulfil our exclusion criteria. At the second stage, we retrieved the full‐text study reports/publications, and two review authors (SP and PC) independently assessed the full text and identified studies for inclusion. At this stage, we included all references that do fulfil our inclusion criteria. We recorded reasons for exclusion of ineligible studies assessed as full texts, so that we could report these in a Characteristics of excluded studies table. We resolved any disagreement through discussion, or, if required, we consulted a third review author (NS). We identified and excluded duplicates and collated multiple reports of the same study, so that each study rather than each report is the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA study flow diagram.

When our systematic searches identified studies conducted by authors of this review, we avoided conflicts of interest by having all decisions concerning inclusion and exclusion made by review authors who were not involved with the study.

Data extraction and management

We used a data collection form for study characteristics and outcome data that had been piloted on at least one study in the review. Two review authors (SP and PC) extracted the following study characteristics from included studies.

• Study authors and year of publication.

• Methods: study design, total duration of study, study location, study setting, withdrawals, date of study.

• Participants: N, mean age or age range, sex/gender, severity of condition, intensity of sedentary work (percentage of workday sedentary), type of sedentary work, diagnostic criteria if applicable, inclusion criteria, exclusion criteria.

• Interventions: description of intervention, comparison, duration, intensity, content of both intervention and control conditions, co‐interventions.

• Outcomes: description of primary and secondary outcomes specified and collected, time points reported.

• Notes: funding for trial, notable conflicts of interest of trial authors.

Two review authors (SP and PC) independently extracted outcome data from included studies. We noted in the Characteristics of included studies table if outcome data were not reported in a usable way. We resolved disagreements by consensus or by involving a third review author (NS). One review author (NS) transferred data into the Review Manager file (RevMan 2014). We double‐checked that data were entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (PC) spot‐checked study characteristics for accuracy against the trial report. When included studies published in one or more languages in which our author team is not proficient, we arranged for a native speaker or someone sufficiently qualified in each foreign language to fill in a data extraction form for us.

Assessment of risk of bias in included studies

Two review authors (SP and PC) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved any disagreements by discussion or by consultation with another review author (NS). We assessed risk of bias according to the following domains.

• Random sequence generation.

• Allocation concealment.

• Blinding of participants and personnel.

• Blinding of outcome assessment.

• Incomplete outcome data.

• Selective outcome reporting.

• Other bias such as baseline imbalance.

In addition, if cluster‐randomised trials were identified and included in the review, we considered the following additional biases.

• Recruitment bias.

• Baseline imbalance.

• Loss of clusters.

• Incorrect analysis.

• Comparability with individually randomised trials.

We graded each potential 'Risk of bias' item as high, low, or unclear, and we provided a quote from the study report together with a justification for our judgement in the 'Risk of bias' table. We summarised 'Risk of bias' judgements across different studies for each of the domains listed. We considered blinding separately for different key outcomes when necessary (e.g. for unblinded outcome assessment, risk of bias for work productivity may be very different than for a patient‐reported pain scale). When information on risk of bias related to unpublished data or correspondence with a trialist, we noted this in the 'Risk of bias' table.

We consider allocation concealment, blinding of participants and outcome assessors, and incomplete outcome data to be key domains. We judged a study to have high risk of bias when one or more key domains had high risk of bias. Conversely, we judged a study to have low risk of bias when we judged that most of the key domains had low risk of bias.

For CBA studies, we used the instrument for appraising risk of bias of CBA studies validated by Downs (Downs 1998). The instrument has been shown to have good reliability and internal consistency and validity. The list consists of five different subscales: reporting, external validity, bias, confounding, and power. We used the combined score on the two internal validity subscales (bias and confounding) to judge risk of bias only for the included CBA studies. We used an arbitrary cut‐off score of 50% of the maximum attainable score of the internal validity scale to discern low from high risk of bias. We modified the criteria for risk of bias so that they fit the 'Risk of bias' tool as implemented in RevMan by changing them from 0 and 1 to high, low, and unclear (RevMan 2014).

We also checked for relevant and considerable baseline differences between control and intervention groups based on age and gender.

When considering treatment effects, we took into account the risk of bias for studies that contributed to that outcome.

Measures of treatment effect

We entered the outcome data for each study into the data tables in RevMan to calculate treatment effects (RevMan 2014). We used odds ratio/risk ratio for dichotomous outcomes, and mean differences or standardised mean differences for continuous outcomes, or another type of data as reported by study authors. For outcomes reported as dichotomous data in some studies and as continuous data in other studies, we re‐expressed the odds ratio as the standardised mean difference. This method assumes logistic distribution and comparable variability for both intervention and control groups according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). If only effect estimates and their 95% confidence intervals or standard errors were reported in studies, we entered these data into RevMan using the generic inverse variance method. We ensured that higher scores for continuous outcomes have the same meaning for the particular outcome, explained the direction to the reader, and reported where the directions were reversed if this was necessary. When results could not be entered either way, we described them in the Characteristics of included studies table, or we entered the data into additional tables.

Unit of analysis issues

For studies that employ a cluster‐randomised design and that report sufficient data for inclusion in the meta‐analysis but do not make an allowance for the design effect, we calculated the design effect based on a fairly large assumed intracluster correlation of 0.10. We based this assumption of 0.10 being a realistic estimate by analogy on studies about implementation research (Campbell 2001). We followed the methods provided in the Cochrane Handbook for Systematic Reviews of Interventions when performing the calculations (Higgins 2011).

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics and to obtain missing data when possible (e.g. when a study is identified as abstract only). When this was not possible, and missing data were thought to introduce serious bias, we explored the impact of including such studies in the overall assessment of results by performing a sensitivity analysis.

If data such as standard deviations or correlation coefficients were missing and they could not be obtained from the study authors, we calculated them from other available statistics such as P values according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Assessment of heterogeneity

We assessed the clinical homogeneity of the results of included studies based on similarity of population, intervention, outcome, and follow‐up. We considered populations as similar when sedentary work is being conducted for more than 50% of working hours, or when participants are described as 'sedentary workers'. Populations that report musculoskeletal symptoms in one or more body region, of any intensity, were considered as similar. We considered interventions as similar when they target workplace sedentary behaviour by promoting standing or walking according to the category of the intervention as defined under Types of interventions. We did not consider interventions that implement exercise or educational programmes to target specific muscle groups such as neck/shoulder or low back exercise as similar to sedentary behaviour modification programmes (as stated under Types of interventions). We considered all outcome measures of pain or discomfort including dichotomous measures, Likert scale, visual analogue scale, and standardised questionnaires such as the Nordic Musculoskeletal Questionnaire as similar. For measurement of work performance, pain‐related disability, and work productivity, we considered all self‐reported outcomes from standardised questionnaires (e.g. Work Performance Index, Neck Disability Index) as similar. We regarded follow‐up times of up to six months as short term, from six months to less than 12 months as medium term, and from 12 months onward as long‐term outcomes, and we treated these outcomes as different.

Assessment of reporting biases

We were not able to pool more than five trials in any single meta‐analysis; therefore we did not explore possible small‐study biases using a funnel plot.

Data synthesis

We pooled data from studies judged to be clinically homogeneous using Review Manager 5.3 software (RevMan 2014). If more than one study provided usable data in any single comparison, we performed a meta‐analysis. We calculated the standardised mean difference when data for the same outcome were presented in some studies as dichotomous data and in other studies as continuous data (Section 9.4.6, Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011)), or when studies measured the same outcome on different scales. We used a random‐effects model when I² was above 40%; otherwise we used a fixed‐effect model. When I² was higher than 75%, we did not pool study results in a meta‐analysis.

We narratively described skewed data reported as medians and interquartile ranges.

When multiple trial arms were reported in a single trial, we included only the relevant arms. When two comparisons (e.g. provision of sit‐stand desk vs standard desk and behavioural modification vs standard desk) were combined in the same meta‐analysis, we halved the control group to avoid double‐counting.

We considered minimally important differences for validated outcome measures when we discussed the magnitude of the effect size. We considered pooled effect sizes greater than the minimally important difference to be clinical significant.

Subgroup analysis and investigation of heterogeneity

We planned to carry out the following subgroup analyses.

• Intervention approach (workstation design, workplace built environment, workplace policy, interventions in non‐productive periods (work breaks)).

• Intervention effects on different body regions (cervical spine, mid‐back, lower back, upper limb/shoulder, lower limb).

• Participant characteristics (age, gender, body mass index).

• Participant work group characteristics (specific occupations).

We planned to use the following outcomes in subgroup analyses.

• Musculoskeletal symptoms (pain/discomfort).

• Pain‐related disability.

As studies were insufficient, we were not able to conduct planned subgroup analyses.

Sensitivity analysis

We planned to perform sensitivity analyses to determine whether our findings are affected by high risk of bias and baseline pain of low intensity. To perform sensitivity analysis, we defined ‘high quality’ as studies with appropriate random allocation and concealment and attrition bias of less than 20%. We defined the low‐intensity pain threshold as 3 out of 10 on a pain intensity scale (Moore 2013). As studies were insufficient, we were not able to conduct the planned sensitivity analyses.