Effectiveness of an intervention for reducing sitting time and improving health in office workers: three arm cluster randomised controlled trial

The researchers tested two versions of the SMART Work and Life (SWAL) intervention against a control group that continued usual practice:

SWAL only: A behaviour change programme that included:

• Organisational support from senior leaders (business case documents, videos)
• A trained workplace "champion" to deliver the programme
• Individual strategies: goal setting, self-monitoring, action planning, habit formation
• Group strategies: team challenges, peer support, feedback sessions
• Educational components about the health risks of prolonged sitting
• No height-adjustable desk provided

SWAL plus desk: The same behaviour change programme PLUS a height-adjustable standing desk provided to each participant

Control: Usual practice – participants continued their normal work routines without any intervention

Primary outcome: Daily sitting time measured by accelerometry at 12 months

Secondary outcomes included:

• Sitting time during work hours and non-work hours
• Prolonged sitting time (sitting bouts lasting 30+ minutes)
• Standing time and stepping time
• Physical activity levels
• Cardiometabolic health markers (blood pressure, cholesterol, glucose, HbA1c, triglycerides)
• Musculoskeletal pain
• Psychological measures (stress, wellbeing, vigour, fatigue)
• Work-related outcomes (presenteeism, job performance, social norms)
• Self-reported lifestyle behaviours (diet, sleep, alcohol)

• Sample size: 756 desk-based employees from 78 clusters (offices, departments, or teams)
• Setting: Local government councils in Leicester, Liverpool, and Greater Manchester, UK
• Age: Mean 44.7 years
• Sex: 72.4% women (n=547)
• Ethnicity: 74.9% white (n=566)
• Employment: Desk-based office workers, working at least 60% full-time equivalent
• Inclusion criteria: Aged ≥18 years, spent most of their day sitting (self-reported), could walk unassisted
• Exclusion criteria: Pregnant, already using a height-adjustable desk, unable to provide informed consent, unable to communicate in English
• Recruitment: February 2018 to January 2019; baseline data collected May 2018 to February 2019

• Primary outcome – sitting time: ActivPAL3 micro accelerometer worn on the thigh 24 hours/day for 7 consecutive days. This device uses inclinometry to distinguish sitting/lying from standing and stepping with high accuracy. Data were processed to calculate sitting time per day, during work hours, on workdays, and non-workdays.
• Prolonged sitting: Defined as sitting bouts lasting ≥30 minutes without interruption
• Standing and stepping time: Also from activPAL
• Physical activity: ActivPAL stepping time plus self-reported moderate-to-vigorous physical activity (MVPA) via the International Physical Activity Questionnaire (IPAQ)
• Cardiometabolic health: Fasting blood samples (total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, glucose, HbA1c), blood pressure (systolic and diastolic), body weight, body mass index (BMI), waist circumference
• Musculoskeletal pain: Nordic Musculoskeletal Questionnaire (body map with pain ratings for neck, shoulders, upper back, elbows, wrists/hands, lower back, hips/thighs, knees, ankles/feet)
• Psychological measures: Perceived Stress Scale (PSS, 0-40 scale, higher = more stress), Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS, 14-70 scale, higher = better wellbeing), vigour subscale of the Profile of Mood States (0-24 scale, higher = more vigour), fatigue subscale of the Multidimensional Fatigue Inventory (MFI-20, 4-20 scale, higher = more fatigue)
• Work-related outcomes: Stanford Presenteeism Scale (6-item, 6-30 scale, higher = better ability to work), job performance (single item 0-10 scale), social norms for sitting and standing at work (adapted scales)
• Lifestyle behaviours: Self-reported diet (fruit/vegetable intake), sleep duration, alcohol consumption

Design: Cluster three-arm randomised controlled trial with follow-up at 3 and 12 months. The trial was registered (ISRCTN11618007) and reported according to CONSORT guidelines for cluster RCTs.

Randomisation: Clusters (offices, departments, or teams) were randomised to one of three conditions: SWAL only, SWAL plus desk, or control. Randomisation was stratified by council area (Leicester, Liverpool, Greater Manchester) and cluster size (<10 vs ≥10 participants). A statistician generated the allocation sequence using a computer algorithm, and allocation was concealed until clusters were enrolled.

Why cluster randomisation matters: Individual randomisation would have caused contamination – if one person in an office got a standing desk and behaviour coaching while their colleague didn't, they'd talk, share tips, and the control participant would be influenced. Clustering by office prevents this spillover. However, it also means participants within the same cluster are more similar to each other than to people in other clusters, so the analysis must account for this (using multilevel modelling or cluster-robust standard errors), which reduces statistical power.

Blinding: Participants and workplace champions delivering the intervention were not blinded to group allocation – you can't hide whether someone got a standing desk or behaviour coaching. Outcome assessors (researchers processing accelerometer data and analysing blood samples) were blinded to group allocation. This is a partial blinding design: objective measures (accelerometry, blood biomarkers) are less susceptible to bias, but self-reported outcomes (stress, wellbeing, pain) could be influenced by participants knowing they received an intervention.

Duration: Follow-up at 3 months (short-term) and 12 months (primary endpoint). The original protocol included 24-month follow-up, but this was removed due to the COVID-19 pandemic (all 12-month data were collected by February 2020, before lockdowns began). The 12-month follow-up is a strength – many workplace sitting interventions only measure at 3-6 months.

Statistical approach: Intention-to-treat analysis (participants analysed in the group they were randomised to, regardless of adherence). Linear mixed models with random effects for cluster and council area, adjusted for baseline values of the outcome, cluster size, and stratification variables. Missing data were handled using multiple imputation. The primary analysis compared each intervention group to control, and if both were significant, a secondary analysis compared SWAL plus desk to SWAL only.

What this design can prove:

• Causal effects of the SWAL intervention (with and without desk) on sitting time, because randomisation balances measured and unmeasured confounders between groups
• Relative effectiveness of adding a height-adjustable desk to the behaviour change programme
• Durability of effects over 12 months

What this design cannot prove:

• Why the intervention worked (mechanisms) – the trial tests effectiveness, not causal pathways
• Whether effects persist beyond 12 months (24-month follow-up was cancelled)
• Whether results generalise to non-office workers, men (only 28% of sample), or non-white ethnic groups
• Whether the small health improvements translate to reduced disease risk over decades
• Whether the intervention would work without workplace champions (the "real-world" implementation was tested, but champions were trained by the research team)

Major methodological weaknesses:

1. No blinding of participants – self-reported outcomes (stress, wellbeing, pain) could be biased by expectation effects
2. Cluster randomisation – fewer degrees of freedom than individual randomisation; the effective sample size is smaller than 756
3. High dropout – 12-month data were available for 71% of participants (537/756); those who dropped out may have differed from completers
4. COVID-19 disruption – 24-month follow-up was cancelled; some 12-month data collection may have been affected by early pandemic concerns
5. Single occupational group – local government office workers; results may not apply to private sector, healthcare, or education settings
6. Self-reported sitting at baseline – eligibility relied on participants saying they sat "most of the day," which may have selected for people already concerned about their sitting

Primary outcome – daily sitting time at 12 months (activPAL):

• SWAL only group: −22.2 minutes/day compared to control (95% CI −38.8 to −5.7, P=0.003)
• SWAL plus desk group: −63.7 minutes/day compared to control (95% CI −80.1 to −47.4, P<0.001)
• SWAL plus desk vs SWAL only: −41.7 minutes/day (95% CI −56.3 to −27.0, P<0.001) – the desk added three times the benefit

Secondary outcomes – sitting during work hours (12 months):

• SWAL only: −15.6 min/workday (95% CI −27.2 to −4.0, P=0.008)
• SWAL plus desk: −56.4 min/workday (95% CI −68.1 to −44.7, P<0.001)
• Both groups showed reductions in prolonged sitting (bouts ≥30 min) during work hours

Secondary outcomes – sitting on non-workdays (12 months):

• SWAL only: −8.3 min/day (95% CI −25.1 to +8.5, P=0.33) – not significant
• SWAL plus desk: −12.7 min/day (95% CI −29.8 to +4.4, P=0.15) – not significant
• Neither intervention changed sitting time outside of work

Standing time (12 months):

• SWAL only: +9.7 min/day (95% CI −4.1 to +23.5, P=0.17) – not significant
• SWAL plus desk: +40.9 min/day (95% CI +27.0 to +54.8, P<0.001)

Stepping time (12 months):

• No significant differences between any groups (all P>0.05)

Cardiometabolic health (12 months):

• No significant differences between groups for: systolic blood pressure, diastolic blood pressure, total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, glucose, HbA1c, BMI, or waist circumference (all P>0.05)

Psychological outcomes (12 months):

• Stress (PSS): SWAL only −0.8 points (95% CI −1.6 to −0.1, P=0.03); SWAL plus desk −0.9 points (95% CI −1.7 to −0.2, P=0.02) – small improvements
• Wellbeing (WEMWBS): SWAL only +1.3 points (95% CI +0.2 to +2.4, P=0.02); SWAL plus desk +1.4 points (95% CI +0.3 to +2.5, P=0.01) – small improvements
• Vigour: SWAL only +0.7 points (95% CI +0.1 to +1.3, P=0.02); SWAL plus desk +0.8 points (95% CI +0.2 to +1.4, P=0.01) – small improvements
• Fatigue: No significant differences

Musculoskeletal pain (12 months):

• SWAL plus desk: Lower extremity pain reduced by −0.3 points (95% CI −0.5 to −0.1, P=0.01) on a 0-9 scale
• No significant changes in neck, shoulder, upper back, or lower back pain

Work-related outcomes (12 months):

• Social norms for sitting: Reduced in both intervention groups (P<0.05)
• Social norms for standing: Increased in SWAL plus desk group (P<0.05)
• Support for reducing sitting: Increased in both intervention groups (P<0.05)
• Presenteeism and job performance: No significant differences

• Daily sitting reduction: The SWAL plus desk group sat for about 1 hour less per day (64 minutes) – roughly equivalent to the time spent in a lunch break. The SWAL only group sat for about 22 minutes less – about the length of a TV sitcom episode.
• The desk was three times more effective than behaviour change alone. Adding a height-adjustable desk produced an extra 42 minutes of sitting reduction per day.
• Health improvements were small: Stress scores dropped by less than 1 point on a 40-point scale (about 2% improvement). Wellbeing improved by about 1.3 points on a 56-point scale (about 2% improvement). These are not considered clinically meaningful – they're the kind of change you might see from a good night's sleep or a pleasant conversation.
• No cardiometabolic changes: Despite reducing sitting by 1 hour/day for 12 months, there were no detectable improvements in blood pressure, cholesterol, blood sugar, or weight. This suggests that either (a) the dose of sitting reduction was insufficient to change these markers, (b) the study was underpowered for these outcomes, or (c) sitting reduction alone (without increasing physical activity) doesn't improve cardiometabolic health in the short term.
• The effect was work-specific: Almost all the sitting reduction happened during work hours. On non-workdays, there was no difference between groups – people compensated by sitting just as much at home.

What the authors acknowledge:

• COVID-19 forced cancellation of 24-month follow-up
• Participants and workplace champions were not blinded
• High dropout rate (29% at 12 months)
• The study was not powered for cardiometabolic outcomes (these were secondary/exploratory)
• Self-reported sitting at baseline for eligibility may have selected a biased sample
• The intervention was delivered by trained workplace champions, not researchers – this is a strength for real-world implementation but may have introduced variability in intervention quality

What a critical reader would note:

• No blinding of participants means self-reported outcomes (stress, wellbeing, pain) could reflect expectation bias – people who got a standing desk might feel they should feel better
• The control group sat less than expected – baseline sitting was ~9.5 hours/day, but the control group reduced by about 10 minutes/day on average (possibly due to the Hawthorne effect of wearing an accelerometer)
• The sample was predominantly white women (72% female, 75% white) – results may not generalise to men or other ethnic groups
• Local government workers may differ from private sector workers in job autonomy, culture, and resources
• No adjustment for multiple comparisons – with dozens of secondary outcomes, some "significant" findings could be due to chance
• The "small but non-clinically meaningful" language is the authors' own admission – they're telling us the health benefits were trivial
• No measurement of physical activity compensation – participants might have stood more but moved less (stepping time didn't change, but step counts weren't reported)
• Industry funding – the study was funded by the National Institute for Health and Care Research (NIHR), a UK government body, so no obvious conflict of interest, but the desks were provided by an unspecified manufacturer (potential indirect influence)

For someone running their own n=1 experiment:

What to test

• Primary intervention: Using a height-adjustable standing desk for at least 6-12 weeks, combined with a simple behaviour change strategy (setting a goal to stand for 15 minutes every hour, using a timer or app reminder, tracking your