Reference  Sample characteristics  Measurement  Protocol  Analytical strategy  Findings 
TudorLocke18 2005  25 men, 25 women; A convenience adult sample; 18–39 years (25.4±4.7 years for men, 23.6±3.4 years for women)  Steps: Yamax SW200 pedometer, (Yamax, Tokyo); Indirect calorimetry: Physiodyne Instrument, Quogue, New York  6 min exercise bouts at three treadmill speeds (4.8, 6.4 and 9.7 km/hour)  Actual METs were calculated for each speed; Linear regression was used to quantify the relationship between steps/min and METs; Regression equations generated were used to establish steps/min cutpoint corresponding to moderate intensity 

Marshall26 2009  39 men, 58 women; Community Latino adult sample; 32.1±10.6 years  Steps: Yamax SW200 pedometer (Yamax, Tokyo); Indirect calorimetry: VacuMed  6 min incremental walking bouts at 3.9, 4.8, 5.7 and 6.6 km/hour  Three analytic approaches: 1) multiple regression—step counts from each treadmill speed were used to develop a prediction equation for generating a cutpoint associated with moderate intensity; 2) mixed modelling—random coefficients models was developed to take account of the datadependence structure and 3) receiver operating characteristic (ROC) curves—optimal cutpoint was examined using sensitivity and specificity 

Beets45 2010  9 men, 11 women; Healthy adults; 20–40 years (26.4±4.6 years)  Steps: hand tally counter; Indirect calorimetry: K4, Cosmed, Italy  6 min overground walking at 1.8, 2.7, 3.6, 4.5 and 5.4 km/hour  Actual METs were calculated for each speed; Random effects models were used to predict steps/min from METs and participant anthropometric measures; Regression equations generated were used to establish steps/min corresponding to 3 METs; Model estimates were used to predict steps/min corresponding to heights ranging from 5 ft. to 6 ft. 6 in. 

Nielson1 2011  50 men, 50 women; A convenience sample of physically active adults; 23.3±3.9 years (24.2±4.0 for men and 22.4±3.5 for women)  Steps: hand tally counter; Indirect calorimetry: Trueman 2400 metabolic cart, Consentious Technologies, Sandy, Utah  10 min treadmill walking bouts at cadences of 80, 90, 100, 110 and 120 steps/min  Energy expenditure at each stage was calculated by multiplying the average steadystate oxygen consumption by the appropriate caloric equivalent obtained from the measured steadystate nonprotein respiratory exchange ratio value; Descriptive statistics were computed for the MET values 

Rowe46 2011  37 men, 38 women; University employees and their families; 18–64 years (32.9±12.4 years)  Steps: hand tally counter; Indirect calorimetry  Three treadmill and overground walking trials at slow, medium, and fast walking speeds  Multiple regression analysis was used to develop a regression equation to predict overground VO_{2} from cadence and stride length indicators; Mixed model regression was used to develop an equation determining the cadence cutpoint 

Abel19 2011  9 men, 10 women; A convenience sample of physically active university students; 28.8±6.8 years (27.1±3.1 years for men and 30.3±8.9 years for women)  Steps: hand tally counter; Indirect calorimetry: TrueMax 2400, Sandy, Utah  10 min treadmill walking trials at 3.2, 4.8 and 6.4 km/hour and running at 8.0, 9.7 and 11.3 km/hour  Linear and nonlinear regression analyses were both used to develop prediction equations to determine cadence cutpoints at various intensities 

Wang20 2013  117 men, 109 women; Recreationally active community Chinese adults sample; 21.7±0.2 years  Steps: hand tally counter; Indirect calorimetry: Cortex MetaMax3B  Four 6 min bouts overground walking at 3.8, 4.8, 5.6 and 6.4 km/hour (50 m rectangular track)  ROC curves were used to determine optimal cadence cutpoints 

Rowe47 2013  25 currently inactive adults; 16–64 years (34±13 years)  Steps: hand tally counter; Indirect calorimetry: Cosmed, Italy and AEI Technologies, USA  A moderate intensity (4.3 km/hour) treadmill walking trial; Overground walking trial: a 10 min selfpaced ‘brisk’ walk and moderatepaced (with metronome prompt) walk  Singlesample ttest, repeated measures ttest, Cohen’s d, BlandAltman plots and oneway repeated measures analyses of variance were used to determine study outcomes 

Rowe16 2014  17 unilateral transtibial amputees (TTAs); 52.2±12.9 years  Steps: hand tally counter; Indirect calorimetry: Servomex, Woburn, Massachusetts  Two 5 min walking trials around a speed corresponding to approximately 50% maximal agepredicted HR  Linear regression was used to develop prediction equations to determine intensity from cadence 

Peacock17 2014  29 women; 60–87 years (71.3±12.4 years)  Steps: hand tally counter; Indirect calorimetry: Zoetermeer, The Netherlands  4 min treadmill walking at selfselected slow, medium and fast speeds (order was counterbalanced)  A regression model (model 2 in the paper) was used to predict moderateintensity cadence 

Serrano14 2017  121 apparently healthy older adults, 49 men; 68.6±7.8 years; 60 for algorithm development (68.1±8.6 years) and 61 for algorithm validation (69.1±7.1 years)  Steps: step sensor+Garmin FR60 (Foot Pod, Garmin Rome, Italy); Indirect calorimetry: a portable metabolic cart  Visit 1—walking test on a treadmill to achieve maximal capacity (VO_{2peak}) within 10–12 min; Visit 2—200 m flat surface walking test until achieving 40% of VO_{2reserve} and 2 min walking at the targeted intensity  Linear regression was used to predict walking cadence at 40% VO_{2reserve} from height, body weight, body mass index and cadence at selfselected walking speed 

MET values presented for the Nielson et al (2011) study were calculated by dividing 150 from the recorded values of METminute (150 minutes) in the original article.45 Walking speeds were converted into kilometers perhour if other metrics were used in the original manuscript.
HRR, heart rate reserve; MET, metabolic equivalent; MVPA, moderatetovigorous intensity physical activity; NR, not reported.