Article Text

Download PDFPDF

Wrist-worn accelerometers: recommending ~1.0 mg as the minimum clinically important difference (MCID) in daily average acceleration for inactive adults
  1. Alex Rowlands1,2,3,
  2. Melanie Davies1,2,
  3. Paddy Dempsey1,2,4,5,
  4. Charlotte Edwardson1,2,
  5. Cameron Razieh1,6,
  6. Thomas Yates1,2
  1. 1 Diabetes Research Centre, University of Leicester, Leicester, UK
  2. 2 NIHR Leicester Biomedical Research Centre, Leicester, UK
  3. 3 Alliance for Research in Exercise, Nutrition and Activity (ARENA), University of South Australia, Adelaide, South Australia, Australia
  4. 4 MRC Epidemiology Unit, Institute of Metabolic Science, University of Cambridge, Cambridge, UK
  5. 5 Physical Activity & Behavioural Epidemiology Laboratories, Baker Heart & Diabetes Institute, Melbourne, Victoria, Australia
  6. 6 NIHR Applied Research Collaborations – East Midlands (NIHR ARC – EM), Leicester, UK
  1. Correspondence to Dr Alex Rowlands, Diabetes Research Centre, University of Leicester, Leicester General Hospital, Leicester, UK; alex.rowlands{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Physical activity is increasingly assessed using wrist-worn accelerometry.1 The primary unit of measurement is acceleration which lacks an obvious concrete meaning in the clinical and public health settings. If the scientific community agreed on a minimum clinically important difference (MCID) that would greatly help users interpret accelerometry data in a more meaningful way. Here we present converging evidence to inform estimation of the MCID in physical activity for inactive adults, expressed as average acceleration measured from wrist-worn accelerometers.

What is average acceleration?

Contemporary accelerometers measure acceleration at high frequency in three axes. Typically, the Euclidean norm is calculated and 1g subtracted to correct for gravity (Euclidean Norm Minus One (ENMO)). Average acceleration across the day is expressed in milli-gravitational units (mg) and reflects overall daily activity, as does steps/day. Multiple days of measurement are usually summarised as daily average acceleration. Reflecting the value of capturing all physical activity across the whole day—rather than moderate-to-vigorous activity alone—average daily acceleration is an appropriate summary variable of physical activity status in most cases. Average daily acceleration is the recommended variable for physical activity analysis in UK Biobank, where >100 000 people wore a wrist-worn accelerometer for 7 days.1

What is the MCID in average acceleration (ENMO, mg) measured at the wrist?

We used three approaches to provide converging evidence for a preliminary estimate of the MCID in mg for inactive people (figure 1).

Figure 1

Converging evidence for a preliminary estimation of the MCID in mg for inactive people. (1) An increase or difference of 500 steps per day, is associated with a 2% to 9% decreased risk of cardiovascular morbidity and all-cause mortality.3–5 (1a) A change in 500 steps per day approximates a change in average acceleration of 0.8 mg (see online supplemental document 1). (2) A daily 5 to 6 min brisk walk is associated with a greater life expectancy of 3.9 years.8 (2a) This would increase daily average acceleration by approximately 0.8 to 1.0 mg (based on a 5 to 6 min brisk walk replacing time spent at the average acceleration, as described in Rowlands et al,7 and a brisk walk eliciting an acceleration of 250 mg,8 (see online supplemental document 2). (3) Estimated from UK Biobank data. Higher average acceleration (per 1 mg) was associated with lower all-cause mortality HR=0.95 (95% CI; 0.94 to 0.96), (see online supplemental document 3). CV,cardiovascular; MCID, minimum clinically important difference; mg, milli-gravitational units.

Supplemental material

  1. Equivalence with MCID in steps/day. Five hundred steps/day has been proposed as the MCID in ambulatory activity needed to convey reduced risk of cardiovascular disease in inactive adults.2 This was based on an earlier study reporting that each increase of 500 steps/day over 12 months reduced risk of cardiovascular mortality and morbidity by 2% in inactive adults.3 Results from studies in older women4 and US adults5 suggest that, in inactive people, each 500 steps/day difference in activity is associated with ~8% to 9% lower risk of all-cause mortality. Data from participants with steps/day and wrist average acceleration assessed contemporaneously, suggest a difference in daily average acceleration of ~0.8 mg is consistent with a difference of 500 steps/day (online supplemental document 1).

  2. Equivalence with 5 min brisk walk. The recently updated Physical Activity Guidelines for Americans acknowledge that 5 min/day of moderate-intensity activity, for example, 5 min/day of brisk walking—equivalent to ~500 steps accumulated in 5 min—has health benefits.6 Substituting 5 to 6 min time spent inactive (eg, sitting or standing) with 5 to 6 min of brisk walking would increase daily average acceleration by ~0.8 to 1.0 mg 7 (online supplemental document 2). Consistent with this, using wrist-worn accelerometer data from UK Biobank, the difference between inactive and moderately active individuals was equivalent to just over 40 min of brisk walking per week (approximately a daily 6 min brisk walk), which was associated with 3.9 years of life gained.8

  3. Direct assessment of average acceleration and health. We investigated the association between average acceleration and all-cause mortality in UK Biobank. The sample included 88 463 adults with valid wrist-worn accelerometer data, complete co-variate data, no history of cancer at baseline and 1893 deaths (median follow-up period: 5.3 years). Higher average acceleration per 1 mg was associated with a HR of 0.948 (95% CI; 0.940 to 0.956). Results were robust when we removed participants who died within 12 months of follow-up, reducing the risk of reverse causality bias. Our results were consistent across all co-variates (including cardiovascular disease status and number of medications/treatments) except for people who are active (figure 2) (online supplemental document 3).

Figure 2

HR (95% CI) per 1 mg average acceleration for all-cause mortality in UK Biobank participants (n=88 463, 1893 deaths). HRs presented stratified by sex, age, smoking status, ethnicity, number of medications/treatments, CVD history, social deprivation and activity status. Each factor adjusted for all other variables described. CVD,cardiovascular disease.

Our three approaches should be viewed as providing a ‘triangulation’ of evidence in inactive adults, rather than definitive findings in their own right; we note that this estimate of ~0.8 to 1.0 mg comes with at least four limitations.

  1. Device-estimated steps and acceleration can vary by device and wear-site.

  2. The estimate of MCID is specific to inactive groups, a key target population for physical activity interventions.

  3. The MCID presented is based on data generated from between-individual differences, so caution should be used when inferring a within-individual effect.

  4. While wrist-worn devices are useful for capturing overall activity as the latent construct (commonly framed as ambulatory activity), non-ambulatory activities (eg, muscle strengthening, housework and arm/hand motion when inactive) also contribute to activity. It is unclear how these non-ambulatory activities contribute to device-estimated activity.

Given these caveats, the MCID presented in this Editorial is a preliminary estimate. We underscore that the estimate is broadly consistent across all three approaches.


We presented preliminary converging evidence to propose ~1 mg as a preliminary estimate of the MCID in daily average acceleration measured at the wrist for health benefits in inactive people. This roughly corresponds to 5 to 6 min of brisk walking per day. This estimate is specific to inactive populations; relationship to outcomes differ as the baseline level of activity increases. This MCID should be updated as evidence accumulates. We look forward to future estimates of MCID values tailored to specific health outcomes and population groups.

Ethics statements

Ethics approval

This is a secondary data analysis. All participants gave written informed consent prior to data collection.UK Biobank has full ethical approval from the NHS National Research Ethics Service (16/NW/0274).


The authors thank all researchers, project staff and participants involved in the SMArT Work trial (Edwardson et al. Effectiveness of the Stand More AT (SMArT) Work intervention: cluster randomised controlled trial. BMJ, 2018, 363, k3870) and the SMArT Work and Life trial (Edwardson et al. A three arm cluster randomised controlled trial to test the effectiveness and cost effectiveness of the SMART Work & Life intervention for reducing daily sitting time in office workers: study protocol. BMC Public Health, 2018, 18(1):1120.). Data sets from both trials were used to examine the association between contemporaneously measured steps per day and average acceleration.



  • Contributors Conception/design: TY and AR. Data analysis/acquisition/interpretation: AR, TY, PD, CR and CE. Drafting/revision critically for important content: All authors. Final approval: All authors.

  • Funding University of Leicester authors are supported by the NIHR Leicester Biomedical Research Centre, and the NIHR Applied Research Collaborations – East Midlands (NIHR ARC – EM). The SMArT Work trial was funded by the Department of Health Policy Research Programme (project No PR-R5-0213-25004) and the SMART Work and Life trial is funded by the National Institute for Health Research Public Health Research programme (project number 16/41/04). The views expressed are those of the authors and not necessarily those of the NHS, NIHR or Department of Health. PCD is supported by a National Health and Medical Research Council of Australia Fellowship (#1142685) and the UK Medical Research Council (#MC_UU_12015/3).

  • Competing interests This research was supported by the NIHR Leicester Biomedical Research Centre.

  • Provenance and peer review Not commissioned; internally peer reviewed.