A persistent question in the physical activity literature has been “How many steps/day are enough?” [5, 24, 25]. Although many studies have reported the normative or average daily steps accumulated by children and adolescents, few studies have estimated how many steps/day are needed for youth to truly meet the aerobic component of the public health guidelines as currently prescribed. This analysis of accelerometer step and activity count data collected concurrently in a nationally representative U.S. sample demonstrated that for those 6–11 years old who attained ≥ 60 minutes/day of MVPA on average, the optimal steps/day thresholds ranged from ≥ 6,500 to ≥10,500 steps/day (censored) and from ≥8,500 to ≥13,500 steps/day (uncensored), depending on the activity count cut point used to define moderate intensity. For adolescents (i.e., 12–17 years of age) who attained ≥60 minutes/day of MVPA on average, optimal steps/day thresholds ranged from ≥7,500 to 11,500 steps/day (censored) and from ≥ 10,500 to 14,000 steps/day (uncensored), depending on the activity count cut point that defined moderate intensity.
Colley, Janssen and Tremblay recently recommended 12,000 steps/day based on objectively monitored data collected from Canadian children using an Actical accelerometer and unadjusted to a pedometer-based scale . They reported that a range between 11,290 and 12,512 steps/day best predicted recommended time in MVPA across age and gender groups, but concluded 12,000 steps/day as a single practical value across subgroups. Based on this recommendation, we explored how the 12,000 steps/day value would function in the NHANES dataset collected using a different accelerometer, in a different population, and across ActiGraph-derived activity count cut points with censored and uncensored steps (uncensored steps would be the most direct comparison to Actical). For boys and girls 6–11, the maximum false positive value was 6% for censored steps and 43% for uncensored steps across the three types of MVPA cut points used to define ≥60-minutes of MVPA. However, the maximum false negative value was 83% for censored steps and 48% for uncensored steps. Compared with the identified optimal threshold, the higher false negative values observed with the Colley et al. 12,000 steps/day threshold resulted in fewer children 6–11 correctly classified as meeting the guideline across the majority of comparisons. For adolescents, the maximum false positive value was 8% for censored and 34% for uncensored steps. Classification accuracies were more comparable between the identified optimal value and the Colley et al. steps/day values, even with higher false negative values observed for 12,000 steps/day. A 12,000 steps/day recommendation is higher than those steps/day values observed in the present analysis across different activity cut points for censored steps/day, thereby producing fewer false positives, but within the range observed for uncensored steps/day. This suggests that 12,000 steps/day works reasonably well for accelerometer-determined estimates of steps/day for Freedson 4 METs and Evenson cut points, but may be too high relative to Freedson 3 METs and less sensitive pedometer estimates.
The optimal steps/day threshold in the present analysis depended on gender and age group, the equation used to estimate average MVPA minutes/day, and whether or not accelerometer-determined steps were adjusted downwards to reflect less sensitive but commonly used pedometers. Of these factors, differences in steps/day by gender and age group were expected, but other differences across the various examinations of accelerometer data were difficult to ignore. Steps/day (censored or uncensored) values across most of the activity count cut point definitions were higher for adolescents than children. The ongoing (and apparently unavoidable) activity cut point conundrum with ActiGraph accelerometers and the 3 MET versus 4 MET debate to define minimum level of moderate activity produced two unfortunate additional layers of confusion. The Freedson 3 cut point suggested that >95% of children 6–11 attained 60 min/day or more of MVPA on average. This estimate is extremely high and was a concern, so the results from using Freedson 3 METS should be interpreted cautiously in the analyses and discussion even though the 3 MET threshold is the minimum used in federal physical activity guidelines. A recent study by Trost et al. compared the Freedson 4 METS to Evenson cut points in a field-based study with the ActiGraph GT1M accelerometer, and while both produced similar predictive accuracy (AUC) compared to indirect calorimetery, the Evenson cut points were less likely to overestimate MVPA for children under 10 years of age. However, as of this date there is no consensus among researchers regarding the ideal accelerometer calibration study or decision rules for scoring accelerometer data for youth . On the face of the findings herein, the percent of the sample meeting MVPA guidelines and steps/day associated with the Evenson and Freedson 4 MET definitions in children were intuitively more acceptable than the Freedson 3 MET definition.
An optimal steps/day threshold also depended on whether one used uncensored ActiGraph steps (raw steps) or adjusted steps downwards (censored steps/day approach) to reflect a less sensitive but commonly used research-grade pedometers. The validity of the ActiGraph 7164 for counting steps compared to direct observation is excellent compared to newer accelerometer models (e.g. GT3X+) and many research-grade pedometers [19, 26]. However, research-grade pedometer use may be more practical for the lay public and health practitioners, so censored steps/day allows these groups to integrate accelerometer results with pedometer-based literature and application [8, 20, 27, 28]. Therefore, we examined NHANES data using the uncensored and censored approaches in an effort to be most useful to both the research and practice communities. The censoring approach used in our study adjusts data for each individual by censoring steps taken when activity level was less than 500 activity counts. However, our results suggest that, compared to uncensored steps/day value observed in the current analyses, the censored data produced (on average) ~2,500 fewer steps/day consistently across age and gender groups for estimating ≥60 MVPA minutes/day (ignoring the Freedson 3 METS). Although this is an average value, the simple subtraction of 2,500 steps from uncensored accelerometer steps/day guideline values when using a research grade pedometer is a crude but possibly useful conversion rule for health and wellness practitioners.
A recent review focused on “How many steps/day are enough?” assembled relevant studies conducted among children and adolescents from around the world . It concluded from limited evidence to date that primary/elementary school children should be directed to take at least 13,000–15,000 steps/day (boys) and 11,000–12,000 steps/day (girls). The review’s step/day values were much higher than those obtained in the present study for children using the 3 MET definition for both censored (i.e., ≥ 6,500) and uncensored (ie. ≥8,500) steps/day. These review values were also generally higher than those observed with the Freedson 4 MET and Evenson definitions for boys and girls 6–11 for both censored and uncensored steps/day. Importantly, values reported in the review article were based on a limited number of studies employing different types of objective physical activity monitors.
The review also concluded that adolescents should be urged to take at least 10,000–11,700 steps/day. This range was based on a single study of adolescents that used the same type of accelerometer as used in NHANES to report a MVPA-to-steps/day translation. The study was conducted with free-living U.S. adolescents, who were primarily overweight girls, and reported that approximately 10,000 to 11,700 uncensored steps/day accurately classified adolescents as meeting the ≥60 minutes/day MVPA guideline using the Freedson 3 and 4 MET definitions, respectively . The current analysis found that 11,500 uncensored steps/day was predictive of meeting the physical activity guideline in adolescent girls using the Freedson 4 METS definition, so these two results are in general agreement.
Given the numerous considerations, and recognizing the need to provide practical values to serve research and practice uses, a range of 11,500–13,500 uncensored steps/day for children and 11,500–14,000 uncensored steps/day for adolescents seems appropriate. A pedometer-friendly adaptation reduces these minimum values to 9,000 steps/day (a 2,500 step/day difference). Similar analyses of NHANES adults suggested that 7000–8000 pedometer-scaled steps/day were indicative of recommended amounts of MVPA . These values represent the minimum number of steps/day, as current physical activity guidelines urge children and adolescents to attain even greater than 60 minutes of MVPA daily. Increasing steps/day recommendations to higher thresholds has the effect of minimizing false positives at the cost of accepting more false negatives, which may be an acceptable tradeoff given the obesity epidemic in the U.S. .
The U.S. President’s Challenge Physical Activity and Fitness Awards Program (PCPAFAP) recommended, until relatively recently, that youth between 6 and 17 years of age accumulate at least 13,000 steps/day for boys and 11,000 steps/day for girls . These values can be traced to a single study that used pedometers by Vincent and Pangrazi among children 6 to 12 years old from two southwestern U.S. elementary schools from the same school district [33, 5]. In August 2012, the PCPAFAP adjusted their award criteria to 12,000 steps/day for all youth based on surveillance data collected Canadian Health Measures Survey (CHMS) for 6–19 year olds using a different accelerometer (i.e., Actical) than that used in the NHANES . The Canadian researchers concluded that 12,000 steps/day was an appropriate indicator of having achieved ≥ 60 minutes/day of MVPA. This value is within the range of uncensored values determined in the present analysis. However, there are a number of differences between CHMS and NHANES studies. The studies differed in the type of accelerometer used, activity cut points and MET values used to define MVPA (Actical is not plagued to the same extent by the cut point conundrum, perhaps in part because less research has been conducted with it at this time), unit of analysis, age/gender subgroups, analytical approach, and population. The present study may be more generalizable for several reasons. It was based on: a large and representative sample of U.S. children and adolescents restricted to the age ranges used by the guideline, a widely used accelerometer, the most commonly used and newly recommended ActiGraph activity count cut points, a state-of-the-art analytical approach, and also attempted to provide a steps/day value that would be congruent with less sensitive but commonly used commercial research-grade pedometers, appropriate for public health recommendations and applications.
A number of methodological strengths and limitations must be acknowledged. There is always a tradeoff between false positives and negatives when deciding how to define cut points. We chose to balance misclassification errors. Balancing the two types of errors is not always the ideal approach . Others may select cut points based by minimizing false positives or negatives. Our Additional file 1 tables provide researchers and practitioners a range of steps/day values that can be used for such purposes. A prior sensitivity analysis of NHANES youth data examined a range of candidate values for censoring ActiGraph steps to reflect pedometer-determined steps . The 500 activity counts/minute value produced steps/day estimates that were in line with expected values obtained from pedometers, however, we acknowledge that an exact conversion factor for translating accelerometer-determined steps/day to pedometer determined steps/day remains elusive and is likely to vary depending on the type and model of accelerometer used. However, it is widely accepted that accelerometers will record more steps than pedometers, so regardless the exact value, some adjustment will be needed to make the output of one instrument comparable to the other. Other limitations include questions about the stability of single day step/day estimate versus multiple days, fewer adolescent girls meeting the guideline defined using 4 METs, and almost all children meeting the guideline defined using 3 METs. The low PPV for adolescent girls at 4 METS (see Additional file 1) is concerning, and results for that analysis need to be interpreted with caution. This issue occurred because very few adolescent girls attained the guideline at 4 METs (3%), producing a restricted distribution compared to the other age, gender and cut point groupings. Physical activity guidelines include recommendations for aerobic as well as bone and muscle strengthening activities, with the recommendation that a majority of the 60 minutes of MVPA come from the aerobic component. It is possible that some activities may not be ambulatory and not register as steps taken. Frequency of meeting the aerobic component of the guideline was not considered, but youth are expected to attain ≥ 60-minute/day of MVPA every day, so these results should be applicable on a daily basis. Finally, the current study estimated the number of steps/day in the context of meeting MVPA guidelines. Some health professionals or interventionists may prioritize other indicators (e.g. step cadence)  or health related outcomes, such as overweight status .
It is apparent that there is no simple answer to the question regarding a step-based translation of recommended time in MVPA for children and adolescents. As additional studies are conducted and the body of evidence expands, it may be possible to further refine a meaningful range of steps/day that is generally congruent with the intensity and duration based physical activity guidelines. Any conclusion will undoubtedly have to be tolerated as a “rule of thumb” rather than a precise number. For now, and based on these data, 11,500 steps/day appears to be the most reasonable number applied to uncensored ActiGraph outputs for both children and adolescents. This value is not that different from the 12,000 uncensored steps/day recommendation based on the Actical. For practical applications however, 9,000 steps/day obtained from the pedometer-scaled analysis in the current study is a more pedometer-friendly value. Future studies with concurrently worn accelerometers and pedometers may produce a more refined MVPA translation of pedometer-determined steps/day.