Developing and Validating a Step Test of Aerobic Fitness among Elementary School Children

Abstract

Purpose: The tests to estimate aerobic fitness among children require substantial space and maximum effort, which is often difficult for children. We developed a simple submaximal step test (Step Test of Endurance for Pediatrics, or STEP) and assessed its reliability, validity, and ability to estimate aerobic fitness among elementary school children.Method:Children aged 5–10 years completed the STEP with a protocol consisting of 0.1-, 0.2-, and 0.3-metre (4, 8, and 12 in.) step heights. Participants underwent treadmill testing with open circuit spirometry to determine actual maximal oxygen consumption (V̇o_2max). Intra-class correlation coefficients (ICCs) assessed test–retest reliability of the STEP and its component tests. Multivariate linear regression assessed the associations between the STEP and V̇o_2max, adjusting for potential covariates such as age, sex, BMI, and comorbidity count. Results: The STEP showed excellent reliability (ICC ≥ 0.92; N = 170), irrespective of effort level during testing. Significant effort issues and collinearity among the independent variables led us to exclude children aged 5–6 years (n = 45) from the regression analysis. The final regression model for children aged 7–10 years with adequate effort (n = 111), as defined by a respiratory exchange ratio of 1.0 or more, showed that the STEP, sex, and BMI were significantly predictive of V̇o_2max (R ² = 0.51). Conclusions: This new, effort-independent step test can estimate the aerobic fitness of children aged 7–10 years. Regression equations to estimate V̇o_2max from the STEP were provided.

The FitnessGram is the most widely used assessment of youth fitness in the world, and it is used in more than 67,000 schools in all 50 U.S. states.¹,² It measures aerobic fitness as well as muscular endurance, muscular strength, flexibility, and body composition.¹ Aerobic fitness, defined as the ability of the body to deliver oxygen to the muscles to be used for energy, is perhaps the most important part of the assessment because of its positive association with better cardiovascular status in children as well as its potential to decrease all-cause mortality in adults.³–⁵ Aerobic fitness can be accurately quantified in a laboratory setting by measuring maximum oxygen uptake (V̇o_2max) with a maximal graded exercise test.⁶–⁸ The FitnessGram uses field-based assessments to estimate V̇o_2max because a laboratory test is not practical for assessing a large number of children. Three field tests are available to estimate V̇o_2max among children: the Progressive Aerobic Cardiovascular Endurance Run (PACER), the One-Mile Run, and the One-Mile Walk Test.

All three available field tests have limitations that clinicians must take into consideration. The PACER is a 20-metre shuttle run and thus requires an appropriate amount of space.⁹ Both the One-Mile Run and the One-Mile Walk Test require an area that can accommodate a long-distance run–walk, which is often unavailable. The PACER and One-Mile Run require children to give maximum effort, which translates to poor reliability and validity of test results in young children.¹ Consequently, V̇o_2max standards from the PACER and One-Mile Run have not been developed for children aged younger than 10 years participating in the FitnessGram. Although the One-Mile Walk Test is a submaximal exercise test, it has not been validated in children aged younger than 13 years.¹ Therefore, clinicians cannot assess aerobic fitness in most elementary school students, and this is a limitation of the FitnessGram.

Aerobic fitness testing of elementary school children should be easy to implement and should require minimal resources. Step tests meet this goal, requiring much less space than the previously discussed walk and run tests. The literature describes few step tests for children, and these tests have practical drawbacks regarding implementation. Jankowski and colleagues developed a system for heart rate monitoring with step testing for children aged 6–12 years, but this test was never validated against V̇o_2max.¹⁰ In addition, the intense physical exercise in this test caused extremely high heart rates in 201 children, so the testing was discontinued.¹⁰ Garcia and colleagues evaluated a step test in children aged 10–15 years¹¹ that required them to walk down steps backward, which could be difficult and even dangerous.

Unlike these tests, the Step Test of Endurance for Pediatrics (STEP) is a simple submaximal exercise step test to expand the ability to easily estimate aerobic fitness in children aged 5–10 years. The STEP uses incremental step heights with a lower step rate, rather than a single 0.3-metre (12 in.) bench, to allow for gradual exertion. Specifically, we examined the test–retest reliability, standard error of measurement (SEM), and concurrent validity of the newly developed step test for this age group. Second, we aimed to assess the independent association between the STEP and V̇o_2max and develop a regression equation to extract an objective index of aerobic fitness (V̇o_2max) using performance on the STEP.

METHODS

Participants

The university institutional review board approved the study. Written, informed consent and child assent were obtained from parents and participants, respectively. The sample was that of convenience and consisted of participants aged 5–10 years from local elementary schools and local YMCA youth programmes. Flyers were distributed to schools and the YMCA for participant recruitment from July 2015 to September 2016. Participants were included if they were within the age range and able to safely participate in physical activity, which was determined by their regular participation in physical education classes in school and a lack of restrictions on performing any level of physical activity.

Participants were excluded if they were participating in a concurrent fitness programme or had a health history that precluded safe participation (long QT syndrome, hypertrophic cardiomyopathy, or an uncontrolled seizure disorder). Children with conditions such as asthma and attention deficit disorder could be included if they met the criteria for participation. Parents provided all the participants’ information via a questionnaire.

Procedures

All data collection took place during a single visit to an exercise physiology lab located at the university. A trained exercise physiologist obtained the demographic, health-related, and anthropometric data and conducted all fitness tests.

Height was measured to the nearest 0.01 metre (0.5 in.) using a wall-mounted measuring rod (the seca 220, seca, Chino, CA) and weight to the nearest 0.2 kilogram (0.5 lb) using a medical scale (the seca alpha 770, seca, Chino, CA). BMI in kg/m² was calculated using these two measurements.

A simple, effort-independent STEP protocol was created (see the Appendix for details). Participants stepped to a 22-step-per-minute metronome cadence. An incremental protocol consisting of 0.1-, 0.2-, and 0.3-metre (4-, 8-, and 12-in.) step heights was used. Participants stepped for 2 minutes at each step height, and heart rate was measured after each step height by 10-second auscultation. The average heart rate from each of the three step heights was used for data analysis. The STEP was repeated after a rest period of 15 minutes on the same day to reduce the influence of timing of meals, diurnal variation, and other non-exercise influences on heart rate; 15 minutes of rest is more than adequate for heart rate to return to baseline resting heart rate in children.¹²

Actual V̇o_2max (mL·kg⁻¹·min⁻¹) was obtained by exercise testing to exhaustion with open-circuit spirometry on a treadmill.⁷,⁸ Testing was performed 15 minutes after the second STEP. Before testing, all participants were habituated to treadmill walking. Speed was determined on the basis of an individual participant’s stride length and was maintained for the duration of the test. Grade was started at 0% and was increased by 2% every minute until volitional exhaustion. V̇o_2max was assessed using a metabolic measurement system (TrueOne 2400, Parvo Medics, Sandy, UT). Before the test, the systems were calibrated with known concentration sample gases. V̇o_2max results were accepted only if a participant was showing evidence of fatigue and achieved a respiratory exchange ratio (RER) of 1.0 or higher.⁶ The test administrator provided constant verbal encouragement to the participants during exercise testing by saying encouraging phrases and acknowledging their efforts.

Statistical analysis

Descriptive statistics were calculated for boys and girls for the demographic and anthropometric characteristics as well as for the STEP trials and V̇o_2max. Means and standard deviations were calculated for the continuous data, and frequency counts were provided for the categorical data. The Shapiro–Wilk test was used to confirm normal distribution of continuous data.

The intra-class correlation coefficient (ICC) examined the within-day reproducibility of the STEP results and estimated V̇o_2max between the first and second trials. The ICC values, with 95% CI, were calculated separately for two subgroups: (1) the subgroup of children who showed adequate effort, as evidenced by the RER of 1.0 or more, irrespective of their age group, and (2) the total sample, irrespective of their effort level, as shown by RER. ICC values of more than 0.9 are considered to be indicative of excellent reliability.¹³ The SEM for the STEP (average heart rate) was calculated because it was one of the predictor variables assessed in the regression model outlined later. We also examined the agreement between the scores of the two STEP trials using the Bland–Altman technique.

A graph of differences in STEP scores between two trials was plotted against the average score of the STEP across the two trials. Subsequently, the limits of agreement (LOA) were calculated and plotted such that LOA represented the mean difference (SD 2) of the mean difference between two scores. The Bland–Altman technique has been described in detail elsewhere.¹⁴ We examined the concurrent relationships of the STEP, V̇o_2max, and BMI using Pearson correlation coefficients (rs). Values of r  >  0.70, r ≥ 0.50–0.70, and r < 0.50 were considered to suggest high, moderate, and low concordance, respectively.¹⁵

We performed backward stepwise regression analyses using V̇o_2max as the dependent variable and age, sex, BMI, step test average heart rate, and comorbidity count (number of medical conditions present) as independent variables. The presence of collinearity between independent variables was considered significant if the condition number (CN) was more than 30,¹⁶ in which case we did not pursue regression analysis. The CN was computed by obtaining the square root of the largest versus smallest eigenvalue, where the eigenvalues reflected variances of the principal components of the predictor variables.¹⁶ The final model included only the variables that had significant associations (p < 0.05) with the V̇o_2max. We used SAS, version 9.4 (SAS Institute Inc., Cary, NC) for all the analyses.

RESULTS

A total of 170 children aged 5–10 years (mean 7.7 [SD 1.5] years; 87 boys and 83 girls) participated in the study. During the data collection, we clearly observed concerns with children aged 5 and 6 years exerting adequate efforts during the STEP protocol. This resulted in only 139 of the participating children being able to complete the study with usable data (RER ≥ 1). Of the children aged 5–6 y, 37.8% (17 of 45) had unusable data, whereas of the children aged 7–10 years, only 11.2% (14 of 125) had unusable data. Therefore, results are outlined considering these age subgroups. Of the 139 children with usable data, 28 (aged 5.8 [SD 0.4] years; 16 boys and 12 girls) were aged 5–6 years, and 111 (aged 8.1 [SD 1.1] y; 58 boys and 53 girls) were aged 7–10 years.

Tables 1 (subgroup with adequate effort irrespective of their age; n= 139) and 2 (total sample; N = 170) show the results of test–retest reliability of the heart rates for the three step heights, the average heart rate for the three step heights, and the V̇o_2max obtained through the regression analyses (described later in this section). The ICC values for the components of the STEP as well as the average of these components were 0.92 or more, with the lower bound of the 95% CI also more than 0.89, suggesting adequate reliability. In particular, the step test average heart rate assessed in the subgroup with adequate effort showed an ICC value of 0.98. The SEM associated with the step test average heart rate was 2.2 beats per minute (BPM). Even when we examined test–retest reliability for children in the two subgroups based on age (irrespective of their effort level), the ICC values were 0.90 or more for the components of the STEP and the average heart rate of the three steps in the test, suggesting acceptable reliability (results of this analysis are not shown in Tables 1 and 2).

Table 1.

Test–Retest Reliability of the STEP for Children Aged 5–10 Years Who Showed Adequate Effort (n = 139)

Test	ICC	95% CI	SEM
Heart rate
0.1 m (4 in.) step	0.94	0.92, 0.96	N/A
0.2 m (8 in.) step	0.96	0.94, 0.97	N/A
0.3 m (12 in.) step	0.96	0.95, 0.97	N/A
Average of 3 steps	0.98	0.97, 0.99	2.20
V̇o2max estimated by step test	0.87	0.82, 0.90	2.98

STEP = Step Test of Endurance for Pediatrics; ICC = intra-class correlation coefficient; SEM = standard error of measurement; V̇o_2max = maximum oxygen consumption; N/A = not applicable.

Table 2.

Test–Retest Reliability of the STEP for the Total Sample, Irrespective of Effort Level (N = 170)

Test	ICC	95% CI	SEM
Heart rate
0.1 m (4 in.) step	0.92	0.89, 0.94	N/A
0.2 m (8 in.) step	0.94	0.92, 0.97	N/A
0.3 m (12 in.) step	0.95	0.93, 0.96	N/A
Average of 3 steps	0.97	0.97, 0.98	2.63
V̇o2max estimated by step test	0.83	0.77, 0.87	3.32

STEP = Step Test of Endurance for Pediatrics; ICC = intra-class correlation coefficient; SEM = standard error of measurement; V̇o_2max = maximum oxygen consumption; N/A = not applicable.

Figure 1 shows the Bland–Altman plot for the agreement between the scores of the step test average heart rate across two assessments for children aged 7–10 years who showed adequate effort. The x-axis represents the mean of the step test average heart rate obtained across the two trials for all the children, whereas the y-axis shows the differences between the step test average heart rates across these two trials. The mean differences between these two scores (1.237 BPM) and LOA are shown in the graph. Given this very narrow mean difference, the children appeared to show no learning effect from the beginning the first session to completing the second session. The correlations among the step test, measured V̇o_2max, and BMI were moderate for both boys and girls (rs = –0.60 and –0.59, respectively; p < 0.001).

Figure 1.

Bland-Altman plot showing agreement between the two trials for the average step test heart rate.

The sample size (n = 28) for the subgroup of children aged 5–6 years was inadequate, and their data showed significant collinearity between the putative independent variables (CN  >  30). Therefore, we included only the group aged 7–10 years in the regression analyses to examine the association between V̇o_2max and the STEP. Descriptive statistics for the children aged 7–10 years who were included in the regression analyses (n = 111) are shown in Table 3. Table 4 shows the final model for the regression analyses, assessing the relationships between V̇o_2max and sex, BMI, and step test average heart rate. No collinearity was observed between the independent variables (CN  =  17.4). Age and comorbidity count did not demonstrate significant associations with V̇o_2max (p > 0.05) and were therefore excluded from the final model. The adjusted R² for the final model was 51%.

Table 3.

Demographic Characteristics and Performance of Children Aged 7–10 Years with Adequate Effort (n = 111)

Variable	Mean (SD)*		p-value†
	Boys (n = 58; 52%)	Girls (n = 53; 48%)
Age, y	8.29 (1.08)	8.55 (1.12)	0.23
BMI, kg/m2	17.56 (3.5)	17.53 (3.87)	0.96
Comorbidity count‡	0.09 (0.28)	0.13 (0.39)	0.48
Ethnicity, frequency (%)			0.49
Caucasian	44 (39.70)	40 (36.00)
African American	8 (7.20)	4 (3.60)
Hispanic	1 (0.90)	3 (2.70)
Other	5 (4.50)	6 (5.40)
Absolute V̇o2max,§ L·min−1	1.14 (0.26)	1.09 (0.25)	0.15
Relative V̇o2max,§ mL·kg−1·min−1	34.62 (6.20)	32.56 (6.12)	0.08
RERmax	1.06 (0.05)	1.08 (0.05)	0.020
VEmax (L/min)	38.62 (9.51)	39.56 (10.14)	0.25
VECO2	33.71 (3.23)	34.27 (2.74)	0.16
VEO2	36.01 (3.79)	36.36 (3.90)	0.32
FEO2	17.43 (0.34)	17.48 (0.33)	0.21

^*Unless otherwise indicated.

^†Indicates difference.

^‡Comorbidities reported: attention deficit disorder, asthma, heart murmur, environmental allergies.

^§V̇o_2max obtained from treadmill test.

V̇o_2max = maximal oxygen consumption; RERmax = maximal respiratory exchange ratio; VEmax = maximum minute ventilation; VECO₂ = ventilator equivalent for carbon dioxide; VEO₂ = ventilator equivalent for oxygen; FEO₂ = fractional concentration of oxygen.

Table 4.

Backward Stepwise Regression Model, with V̇o_2max as the Outcome Variable

Variable	Parameter estimate	p-value	95% CI
Intercept	63.54	–	56.30, 70.79
Sex (male vs. female)	1.72	0.04*	0.06, 3.40
BMI	–0.73	< 0.001*	–0.98, –0.48
Step test (average heart rate)	–0.15	< 0.001*	–0.21, –0.10

Note: R ² = 0.52; adjusted R ² = 0.51. Dash indicates that p-value cannot be computed.

^*p < 0.05.

On the basis of the analyses, we created separate regression equations for boys and girls aged 7–10 years, accounting for the variables retained in the final model. The equation is shown with the relevant regression coefficients and their standard error estimates:

$$\dot{\text{V}}{O}_{\mathrm{2max}}=63.54+(S\times 1.72[\pm 0.84]+[-0.73(\pm 0.13)]\times BMI]+[-0.15(\pm 0.03)\times \text{step}\text{test}\text{average}\text{he}$$

where S is 0 for female and 1 for male.

DISCUSSION

The aim of this study was to develop an effort-independent step test to estimate aerobic fitness, V̇o_2max, in children aged 5–10 years. Our results provide preliminary support for the reproducibility and validity of the STEP as well as for estimating V̇o_2max using the results of the STEP for children aged 7–10 years. The test demonstrated excellent within-day test–retest reliability, with an ICC of 0.98. The coefficient of determination for the regression model was of an acceptable level (R² = 51%), indicating that the final model is sufficiently accurate.

The calculated ICC values for the STEP, irrespective of age group or effort level, were excellent (ICC > 0.90), suggesting that the STEP elicits performance that is highly reproducible over two separate occasions on the same day. The SEM is a function of SD and ICC values of the test in a given sample. Lower variability and higher ICC values for a test in a given sample results in a lower SEM. In our study, the calculated SEM for the STEP in children aged 7–10 years was 2.3 BPM. This is relatively low and will likely introduce a measurement error of only a small magnitude in predicted V̇o_2max using the regression equation presented earlier. For example, a boy who is aged 7–10 years, has a BMI of 20, and has a step test average of 130 BPM will have a predicted V̇o_2max of 31.16 mL·kg⁻¹·min⁻¹ (95% CI: 23.82, 38.5).

The overall correlation for this model (R = 0.71) is similar to the PACER prediction models (Rs = 0.74 and 0.75) and One-Mile Run Test model (R = 0.71) used in the FitnessGram.¹⁷,¹⁸ This similarity is important to note because it lends credibility to our study. The PACER has been validated against V̇o_2max and is the recommended assessment of aerobic fitness in children.¹,¹⁹ Multiple predictive models have been published to estimate aerobic fitness from PACER data.¹⁷,²⁰,²¹ The PACER is considered reliable, with a systematic review reporting ICCs varying from 0.78 to 0.93.²² PACER predictive models have high concurrent validity, with correlation coefficients between the models and V̇o_2max varying from 0.65 to 0.87.⁹,²³ Some of the developed PACER prediction models were adjusted for variables such as BMI, sex, speed, and age of the participants to improve the prediction of V̇o_2max from PACER performance.¹⁷,²⁰,²¹ The ICC values reported for the One-Mile Run Test range from 0.39 to 0.90 in children, and validity coefficients range from –0.60 to –0.90.¹,⁹,²² The data for the One-Mile Walk Test are more limited, with one study reporting an ICC of 0.91 in a small group of adolescents and another reporting a correlation of 0.84 with V̇o_2max.¹

Step tests are a method of estimating V̇o_2max, and they have the benefit of requiring minimal space and resources.¹⁰ Step test protocols are available for adults,²⁴,²⁵ and a recent systematic review of studies of the use of step tests in adults showed correlations varying from 0.47 to 0.95 when comparing predicted values with measured V̇o_2max.²⁶ These step tests tended to be highly valid in the population used to formulate the prediction equation but less valid when applied to different groups. Additional research is needed to discern which step protocols are best to predict V̇o_2max in the adult population.

We found fewer validated step tests to estimate aerobic fitness in children. A systematic review in 2014 sought to investigate all submaximal exercise equations developed to predict V̇o_2max in individuals aged younger than18 years.²⁷ A total of 16 tests were included, and the majority involved running, walking, or cycling; only two studies used step tests. Francis and Feinstein studied 93 children aged 6–18 years who performed step tests and found the correlation of a 15-second recovery heart rate and V̇o_2max to be 0.79 to 0.81, depending on the pace of steps per minute.²⁸ Garcia and Zakrajsek evaluated the usefulness of the Canadian Aerobic Fitness Test in children aged 10–15 years;¹¹ this protocol used a three-stage step test, and the total sample correlation with V̇o_2max was 0.79. One study developed a reference system of mean post-exercise heart rate after a 3-minute step test in children aged 6–12 years; however, this was not validated against actual V̇o_2max.¹⁰

The development of a validated step test for children aged 7–10 years has great potential for use in a clinical setting to monitor the fitness of individual patients over time, and in the schools, and to provide information on the health status of students. The STEP protocol is simple, is effort independent, and uses low-cost equipment (steps, metronome, stop watch, and stethoscope). It also requires a small amount of space and time and a minimal amount of training to administer.

The argument for monitoring aerobic fitness regularly and including it as a fifth vital sign in addition to the four established vital signs (heart rate, blood pressure, temperature, and respiratory rate) is gaining popularity in adults.²⁹,³⁰ It may be beneficial to also begin monitoring these in the paediatric population because the literature shows that aerobic fitness is a marker of cardiovascular health in children and adolescents.⁴ Cardiovascular fitness is important for participation in sports as well as for basic activities of daily life for children.³¹ Unfortunately, the fitness of young people is declining worldwide.³² Therefore, it may benefit physicians and health care providers to monitor aerobic fitness over time to provide effective feedback and interventions for patients. Because of its high test–retest reliability and low SEM, the newly developed STEP allows aerobic fitness to be tracked.

We acknowledge some limitations with our study. First, it was conducted at a single institution, which may limit the generalizability to other populations. The STEP was validated for children aged 7–10 years; however, we were unable to validate it for a younger population (children aged 5–6 years) because of lower recruitment and a submaximal effort level. Second, auscultation by stethoscope was chosen as a practical method of heart rate determination. Polar heart rate monitors and an electrocardiogram (ECG) may have provided more accurate data, but monitors do not work well on young children, and ECG is not practical for a clinical setting.

In addition, the sample size for this study was not calculated a priori. Nonetheless, post hoc analyses revealed that we would have needed 23 children to obtain 90% power for testing the associations among five predictors (age, sex, BMI, step test average heart rate, and comorbidity count) and V̇o_2max for the R² of 0.51 obtained from the analyses.³³ Our sample size clearly exceeded the requisite sample size. Last, the resultant adjusted R² of 0.51 for our prediction model for V̇o_2max shows sufficient promise for the ability to use the STEP to predict V̇o_2max in children aged 7–10 years. Nonetheless, clinicians should be cautioned that a single study does not provide sufficient defining evidence to alter clinical practice.

We anticipate that our work will prompt further research and efforts to build stronger evidence to determine the utility of the STEP among children. Further research can also examine the relationships between other covariates, such as a child’s physical activity level or hydration level, and V̇o_2max to improve the precision of the model.

CONCLUSIONS

In summary, step tests can be used to estimate aerobic fitness individually and for large samples of participants. The newly developed STEP for children aged 7–10 years gives health care providers and educators the ability to measure and follow aerobic fitness in children over time. Although the STEP is a feasible option to assess aerobic fitness in children, the results of this study should be considered preliminary, encouraging efforts to validate the results in subsequent research studies.

KEY MESSAGES

What is already known on this topic

Aerobic fitness is a marker of cardiovascular health in paediatrics. Tests to estimate aerobic fitness in children require substantial space and maximum effort, which is often difficult for children.

What this study adds

This study introduces a new, effort-independent step test to reliably estimate the aerobic fitness of children ages 7–10 years. The protocol is simple and requires a small amount of space to administer. The step test allows for aerobic fitness to be monitored over time by health care professionals and schools.

APPENDIX: STEP TEST OF ENDURANCE FOR PEDIATRICS (STEP) PROTOCOL

The step test protocol consisted of three step heights: 0.1, 0.2, and 0.3 metres (4, 8, and 12 in.). Wooden steps were used and placed 0.3 metres (1 ft) apart in a linear fashion. Subjects bench-stepped to a 22-step-per-minute metronome cadence. Subjects were introduced to the cadence just before the test, and research staff demonstrated the stepping cadence. Subjects followed a cadence of up, up, down, down. Subjects stepped for 2 minutes at each step height, progressing from the 0.1-metre (4-in.) step to the 0.3-metre (12-in.) step. The subject was instructed to stand in front of the step after each 2-minute segment. Heart rate was auscultated by stethoscope for 10 seconds between each step height. This was the only break between steps. The test administrator provided verbal encouragement to keep the subjects on cadence, if needed. The three heart rates auscultated were averaged, and this result is the step test average heart rate used in the final model to predict V̇o_2max.