Effectiveness of the Stand More AT (SMArT) Work intervention: cluster randomised controlled trial

The researchers tested a comprehensive intervention package called "Stand More AT (SMArT) Work" designed to reduce sitting time among desk-based office workers. The intervention included:

• Height-adjustable workstation (sit-stand desk)
• Brief seminar with supporting leaflet explaining the health risks of prolonged sitting
• Workstation instructions with specific sitting and standing targets
• Personalised feedback on sitting and physical activity at three time points (baseline, 3 months, 6 months)
• Posters placed in the workplace as visual reminders
• Action planning and goal setting booklet to help workers plan when and how to stand
• Self-monitoring and prompt tool (likely a device or app that reminded them to stand)
• Coaching sessions at month 1 and every three months thereafter (face-to-face or telephone)

The control group continued with their usual work practice — no desks, no coaching, no feedback.

Primary outcome: Occupational sitting time (minutes per workday) measured by a thigh-worn accelerometer at 12 months.

Secondary outcomes: Daily sitting time, prolonged sitting (bouts ≥30 minutes), standing time, physical activity, musculoskeletal problems, job performance, job satisfaction, work engagement, occupational fatigue, sickness presenteeism (working while ill), sickness absenteeism, cognitive function, mood, anxiety, and quality of life — all measured at 3, 6, and 12 months.

• 146 participants in total, clustered into 37 office groups
• Intervention group: 77 participants (19 clusters)
• Control group: 69 participants (18 clusters)
• Setting: University Hospitals of Leicester NHS Trust, England (three hospital sites)
• Population: Desk-based office workers aged 18–70 years
• Inclusion criteria: Self-reported spending ≥75% of workday sitting, worked at least 0.6 full-time equivalent, worked at the same desk for at least 3 days per week, capable of standing
• Mean age: 41.2 years
• 80% female
• Ethnicity: Predominantly White British (exact percentage not reported in abstract)
• Job roles: Administrative, managerial, and clerical staff within the NHS

• Primary outcome — Occupational sitting time: activPAL micro accelerometer worn on the midline anterior aspect of the right thigh for 7 consecutive days. This device distinguishes between sitting/lying, standing, and stepping with high accuracy. Data were processed to extract time spent sitting during work hours.
• Prolonged sitting: Bouts of sitting lasting ≥30 minutes without interruption, extracted from the same activPAL data.
• Standing time and physical activity: Also from activPAL (standing time) and from the device's step detection (physical activity).
• Anthropometrics: Body weight and body fat (Tanita SC-330ST bioelectrical impedance scale), height (Leicester Height Measure), waist circumference (midpoint between lower costal margin and iliac crest), blood pressure (Omron Healthcare monitor, average of last two of three readings).
• Self-reported work-related health: Job performance, job satisfaction, work engagement, occupational fatigue, sickness presenteeism, and sickness absenteeism — measured via validated questionnaires (specific instruments not detailed in abstract).
• Psychological measures: Mood and affective states, quality of life, daily anxiety — measured via validated questionnaires.
• Cognitive function: Assessed via a cognitive test battery (specific tests not detailed in abstract).

Study design: Cluster randomised controlled trial (RCT)

Why cluster randomisation? The researchers randomised at the office group level (clusters) rather than individual level. This was done to reduce the risk of "contamination" — if individuals within the same office were randomised to different groups, the intervention group members might share their sit-stand desks or coaching tips with control group members, diluting the apparent effect. By randomising entire office clusters, each cluster received only one condition.

Randomisation procedure: A statistician used computer-generated lists to randomise clusters 1:1 to intervention or control, stratified by cluster size (≤4 participants vs >4 participants) with a block size of six. Randomisation occurred in batches after clusters had completed baseline measures.

Blinding: Team members who took measurements were blinded to group allocation. However, team leads could not be blinded because they were responsible for study coordination, including delivering desks and intervention components. Team leads had no involvement in data processing or analysis. Participants obviously could not be blinded — they knew whether they received a sit-stand desk and coaching.

Duration: Baseline data collection November 2015–June 2016; follow-up at 3 months, 6 months, and 12 months. Total study duration from recruitment to final follow-up: approximately 20 months.

Statistical approach: Generalised estimating equation (GEE) models, which account for the clustering of participants within office groups. This is important because participants within the same cluster are more similar to each other than to participants in other clusters (e.g., same office culture, same manager). GEE adjusts for this non-independence.

What this design can prove:

• Causality: The RCT design, with random allocation, means that differences between groups at follow-up can be attributed to the intervention rather than pre-existing differences between participants. This is the strongest design for establishing cause-and-effect.
• Effectiveness in real-world settings: The study was conducted in actual NHS offices with real workers, not in a laboratory. This means the results reflect what happens when the intervention is implemented in practice.
• Sustainability: With follow-up at 12 months, the design can show whether effects persist beyond the initial novelty period.

What this design cannot prove:

• Which component worked: The intervention was a "package" of multiple components (desk, coaching, feedback, posters, etc.). The design cannot tell us whether the desk alone would have been sufficient, or whether coaching was essential, or whether posters made any difference.
• Generalisability beyond NHS office workers: The sample was predominantly female (80%), White British, and working in a healthcare setting. Results may not apply to male-dominated workplaces, manufacturing settings, or other industries.
• Health outcomes: The study measured sitting time and self-reported health, not actual disease endpoints (e.g., diabetes, cardiovascular events). We cannot conclude that the sitting reduction translated into reduced disease risk.
• Long-term adherence beyond 12 months: We don't know if participants maintained the behaviour change after the study ended.

Methodological strengths:

• Objective measurement of sitting (activPAL) rather than self-report
• Cluster randomisation to prevent contamination
• Blinding of measurement staff
• 12-month follow-up (longer than most similar studies)
• High retention (not explicitly stated in abstract, but implied by complete data at all time points)

Methodological weaknesses:

• Participants and team leads were not blinded (unavoidable for this type of intervention)
• Self-reported outcomes (work performance, mood) are subject to expectation bias — intervention participants might report improvements because they feel they "should"
• The control group received no placebo or attention control, so the "Hawthorne effect" (improvement due to being studied) could differ between groups
• Single site (one NHS trust) limits generalisability

Primary outcome — Occupational sitting time at 12 months:

• Intervention group reduced by 83.28 minutes per workday more than control (95% CI: −116.57 to −49.98, P=0.001)
• This means the intervention group sat about 1 hour 23 minutes less per workday compared to the control group at 12 months

Occupational sitting time at earlier time points:

• At 3 months: −50.62 min/workday (95% CI: −78.71 to −22.54, P<0.001)
• At 6 months: −64.40 min/workday (95% CI: −97.31 to −31.50, P<0.001)
• The effect grew larger over time, suggesting participants increased their standing behaviour as they became more accustomed to the intervention

Daily sitting time (entire day, not just work):

• At 6 months: −59.32 min/day (95% CI: −88.40 to −30.25, P<0.001)
• At 12 months: −82.39 min/day (95% CI: −114.54 to −50.26, P=0.001)
• This suggests the reduction in sitting was not just during work hours but carried over into non-work time, or that work sitting reduction was large enough to drive the daily total

Prolonged sitting (bouts ≥30 minutes):

• Significant reduction in favour of the intervention group (exact numbers not provided in abstract)

Standing time:

• Significant increase in favour of the intervention group (exact numbers not provided in abstract)

Work-related outcomes (significant improvements in intervention vs control):

• Job performance
• Work engagement
• Occupational fatigue (reduced)
• Sickness presenteeism (reduced — fewer days working while ill)

Psychological outcomes (significant improvements):

• Daily anxiety (reduced)
• Quality of life (improved)

No significant differences:

• Sickness absenteeism (days off work)
• Cognitive function
• Job satisfaction (not explicitly stated but likely not significant given it's not mentioned)
• Physical activity (stepping) — the intervention increased standing but not walking

• 83 minutes less sitting per workday is a substantial reduction. To put this in perspective: if you work an 8-hour day, this means you're standing for about 17% of your work time instead of sitting. Over a 5-day work week, that's nearly 7 hours less sitting per week.
• The effect grew over time, from ~51 minutes at 3 months to ~83 minutes at 12 months. This is unusual — many behaviour change interventions show decay over time. The authors suggest that participants may have needed time to build the habit of standing and to adjust their workflow.
• The reduction in daily sitting time (~82 minutes) was nearly identical to the reduction in occupational sitting time (~83 minutes), meaning essentially all the sitting reduction occurred during work hours. Participants did not compensate by sitting more at home.
• Work engagement and fatigue improvements were statistically significant but the abstract does not report effect sizes in standardised units (e.g., Cohen's d). The practical meaning is that participants felt more engaged with their work and less tired at the end of the day.
• No change in absenteeism suggests that while people felt better at work, they didn't take fewer sick days — at least within 12 months.

Acknowledged by authors (from abstract and full text context):

• Single site (one NHS trust) limits generalisability
• Participants and team leads could not be blinded
• Self-reported outcomes (work performance, mood) are subject to bias
• The intervention was a multicomponent package, so individual component effectiveness cannot be determined

Critical reader observations:

• Sample size: 146 participants is modest, and the cluster design reduces statistical power. The confidence intervals are wide (e.g., −116 to −50 minutes for the primary outcome), indicating substantial uncertainty about the true effect size.
• Gender imbalance: 80% female. Results may not apply equally to male-dominated workplaces.
• Selection bias: Participants volunteered for a study about standing more. They may have been more motivated to change than the average office worker.
• No blinding of participants: Those who received a sit-stand desk and coaching knew they were in the intervention group. This could inflate self-reported improvements (placebo effect).
• No attention control: The control group received nothing. Any intervention that provides attention, equipment, and coaching might produce improvements simply because people feel cared for. A "sham" intervention (e.g., a non-adjustable standing platform) would have been stronger.
• Funding and conflicts: The study was coordinated from the Leicester Diabetes Centre. Potential conflicts of interest are not detailed in the abstract but should be checked in the full text.
• Missing data: The abstract does not report dropout rates. If more intervention participants dropped out (because they disliked standing), the results could be biased toward positive findings.
• Health outcomes not measured: The study did not measure blood glucose, cholesterol, blood pressure changes, or other biomarkers. We cannot conclude that the sitting reduction improved physical health.
• Workplace type: NHS offices may have different cultures and physical layouts than private sector offices. Results may not generalise to open-plan offices, hot-desking environments, or remote workers.

For someone running their own n=1 experiment:

What to test

• Specific intervention: Use a height-adjustable sit-stand desk (or a desktop converter if a full desk is not available) combined with a structured plan to alternate between sitting and standing. The SMArT Work protocol included specific targets — aim for a gradual increase, starting with 15–30 minutes of standing per hour and building up.
• Dose: The intervention group reduced sitting by ~83 min/workday. Aim for 60–90 minutes of standing per workday as a minimum meaningful target.
• Additional components to try: Set a timer or use a standing reminder app (the study used a self-monitoring prompt tool). Keep a simple log of when you stand. Plan specific tasks to do while standing (e.g., phone calls, reading, brainstorming).

Minimum meaningful duration

• 3 months minimum to see a clear effect. The study showed significant reductions at 3 months, but the effect grew larger at 6 and 12 months. For a self-experiment, commit to at least 8–12 weeks.
• Longer is better: The effect increased over time, suggesting that standing becomes habitual. A 6-month experiment would give more reliable results.

What to measure

• Primary metric: Minutes of standing per workday. Use a thigh-worn accelerometer (activPAL is the gold standard but expensive; alternatives include the Apple Watch, Fitbit, or Garmin with posture detection, or simply a manual log). Measure for at least 5 consecutive workdays at baseline and at the end of the experiment.
• Secondary metrics:
  • Prolonged sitting bouts: Track how often you sit for ≥30 minutes without standing. Aim to break up sitting every 30 minutes.
  • Energy and fatigue: Rate your energy level on a 1–10 scale at the end of each workday.
  • Work performance: Rate your perceived productivity on a 1–10 scale.
  • Musculoskeletal discomfort: Note any new back, neck, or shoulder pain (standing too much can also cause discomfort).
  • Mood: Rate anxiety and mood daily on a simple 1–10 scale.

Key confounds to control for

• Physical activity outside work: If you start exercising more during the experiment, you can't attribute changes to standing alone. Keep exercise constant or log it as a covariate.
• Workload changes: A busy period (deadlines, meetings) might reduce standing regardless of your intentions. Track workload subjectively (1–10 scale) each day.
• Seasonal effects: If you start in winter and end in summer, daylight and mood changes could confound results. Try to run the experiment within a single season.
• Footwear and flooring: Standing on concrete vs carpet vs a standing mat makes a big difference to comfort. Use the same footwear and a standing mat throughout.
• Diet and caffeine: Changes in diet or caffeine intake could affect energy and mood. Keep these consistent.
• Expectation bias: You know you're trying to stand more, so you might rate your mood or productivity higher because you expect improvement. Consider using a blinded rater (e.g., a colleague who doesn't know you're experimenting) to assess your work output or mood.

What a positive