A cross-sectional study with healthy children and adolescents was conducted between June 2013 and March 2016. Participants were recruited from two schools in the central region of Portugal. Ethical approval was obtained from the Ethics Committee of the Research Unit of Health Sciences at the School of Nursing in Coimbra, Portugal (P186-10/2013 & P 246-12/2014). Authorization to conduct the study in each school was obtained from the school boards.

Children and adolescents were eligible if: (i) aged 5–17 years old; (ii) had no chronic or acute disease; (iii) understood the study aims and accepted to participate. Exclusion criteria included: (i) presence of neurological impairment or significant cardiopulmonary and/or musculoskeletal disorders and (ii) any other significant contraindication that could affect physical tests performance. Prior to any data collection, written informed assent was obtained from children and adolescents and written informed consent from their legal representatives. The aim of the study was explained to legal representatives, at the time of the signing of consent form and they were asked about the health status of their child, specifically the presence of any acute or chronic disease.

Socio-demographic (sex and age) and anthropometric (weight and height) data were first collected to characterize the sample. A questionnaire to characterize the level of physical activity was also completed. A spirometry test was performed to ensure that participants had normal lung function. Then, quadriceps muscle strength was measured. Finally, two ISWT were performed.

Physical activity levels were assessed using the Physical Activity Index (PAI).14 The PAI is composed of 5 questions and the answers are scored in a Likert scale from 1 to 4 points. Final score ranges from 5 to 20 points, and can be divided into 4 categories: sedentary, lightly active, moderately active and vigorously active.14 This questionnaire has been validated to the Portuguese population,14,15 demonstrating excellent test-retest reliability (intraclass correlation coefficient=0.92–0.96) and good internal consistency (Cronbach's alpha=0.87).15

A spirometric test using a portable spirometer (MicroLab 3500, CareFusion, Kent, UK) was performed as recommended by the American Thoracic Society/European Respiratory Society.16 Spirometry is a simple and non-invasive test and it is indicated to assess health status before vigorous exercise to identify the presence of restrictive or obstructive respiratory abnormalities.17

Quadriceps maximum isometric muscle strength was assessed using a hand-held dynamometer (Hoggan MicroFET2 Muscle Tester, Model 7477, Pro Med Products, Atlanta, GA) because it is easy to use and portable.18 During the test, participants were sitting on a chair, with the knee flexed to 90°, hip flexion 90° and straight trunk. The hand-held dynamometer was placed on the anterior surface of the dominant leg, proximal to the ankle (5cm above lateral malleolus). All measurements were taken in kilogram of force (kgf). Each participant performed at least one practice trial to understand the movement and acceptable stabilization. Then, two trials, of ten seconds contraction each, were performed; there was a resting period of 60s between trials. The mean of the 2 trials was considered for analysis.19

The ISWT protocol was followed as described by Singh et al.1 and Probst et al.12 The ISWT is an externally paced maximal exercise test, controlled by pre-recorded signals that command the walking speed.10 The walking speed increases 0.17m/s each minute, starting at 0.5m/s, until the participant can no longer carry on with the test. In this study, the audio signals continued until individuals reached their maximal effort, exceeding the 12 levels of speed recommended by Singh et al.1 This proposed adaptation has been previously used to predict ISWT reference values in healthy individuals (i.e., children, adolescents and adults)12,13 and was needed to reach individual symptom limited maximal performance and avoid a ceiling effect, considering that participants were healthy and could possibly reach more than the 12th level to get to their maximal performance.12 A brief standard explanation about the test was provided by an audio recording prior to trial.2 Participants were instructed to walk between the two markers at the time ordered by audio signals for as long as possible until they were too breathless or could not achieve the distance within the beep time two consecutive times. If at some point, they lagged behind the pace dictated by the audio-signal, but felt they were not at their symptom limited maximal performance, a standardized warning was given and running was allowed.12,13 The ISWT was performed twice due to the associated learning effect and the best result was considered for analysis.4,20 At least 30min of rest was provided between the two tests. The distance (in meters), the level and the number of shuttles achieved were recorded. If the test was ended the reason was registered.

To ensure proper participant inclusion and safety during the study, physiological responses to ISWT performance such as dyspnea, fatigue, heart rate, peripheral oxygen saturation, arterial blood pressure and respiratory rate were monitored before and after the two ISWTs.17 Dyspnea and fatigue were measured on the modified Borg Scale.21 The heart rate and peripheral oxygen saturation were measured by a pulse oximeter (PULSOX-300i, Konica Minolta Sensing, Inc., Osaka, Japan) on the non-dominant hand indicator and arterial blood pressure with a blood pressure monitor (Elite, Medel, Parma, Italy). The ISWT is a valid and reliable test when assessing maximal exercise capacity in individuals with chronic respiratory diseases.20,22 Previous research has shown that the ISWT is a reproducible and valid measure for assessing the function-limited aerobic capacity in children with cystic fibrosis (6–16 years).23

The sample size for multiple linear regression to establish reference equations was determined according to the ‘rule of thumb’ by Green24: N≥50+8m where N is the sample size and m is the number of independent variables. As 3 independent variables were considered, a minimum of 74 participants were required.

Data were analyzed using IBM SPSS Statistics version 24 (SPSS Inc., Chicago, Illinois, IL) and plots were created using GraphPad Prism 5.01 (GraphPad Software, La Jolla, California). Descriptive statistics were first applied to characterize the sociodemographic and anthropometric data, PAI score, lung function and quadriceps muscle strength and the ISWT. The predicted maximal heart rate (HR max) for each participant was calculated according to the formula 206.9−(0.67×age), recommended to reflect the differences in heart rate according to age.25 The normality of data distribution was assessed with Kolmogorov–Smirnov tests.26

The intraclass correlation coefficient (ICC) and Bland–Altman plots were used to assess the relative and absolute reliability of the ISWT.27 ICC was interpreted as excellent (>0.75), moderate to good (0.4–0.75) or poor (<0.4).28

Pearson correlation coefficients (age, weight, height, BMI, % HR max, physical activity level, quadriceps muscle strength and FEV1% predicted) and point-biserial correlation coefficient (sex) were used to measure the strength of association between these variables and the distance achieved during the ISWT. These correlations tests were used to select independent variables for the multiple regression analysis.

A model of multiple linear regression, using the Enter mode, was used to estimate the best predictor model for the ISWT distance. ISWT was the dependent variable and variables with significant correlations were included as independent variables. The assumptions of linearity, independence of errors, homoscedasticity, unusual points and normality of residuals were analyzed. ANOVA and R-squared (R2) were used to assess the performance of the model.

To assess the reliability of the equation, measured ISWT distances were compared with the values obtained from a previously published equation13 using a paired-samples t-test. Pearson correlation coefficient and Bland and Altman plot were also used. The level of significance considered was set at p<0.05.

A total of 135 subjects were selected, but 2 had abnormal lung function, 2 failed to obtain the signed consent forms from their legal representatives and 1 could not complete the assessment due to lack of time. A total of 130 children/adolescents were recruited (60 male) and completed the assessment. Detailed participant characterization is summarized in Table 1. Participant mean age was 13.66±2.61 years old. Most participants had a normal BMI (n=78; 60%) and lung function (FEV1=102.73±15.09% predicted). Based on the PAI score, 50.8% of the participants (n=66) were moderately active.

This section presents the outcomes of the first, second and best ISWT and the respective physiological responses. The first and second ISWT were not significantly different (p=0.171) and were strongly correlated (r=0.93; p<0.001). Excellent relative reliability was found between the two ISWT performed (ICC=0.93; 95% CI 0.90–0.95). Bland–Altman plot showed no systematic bias, with a mean difference of −15.5m (95% limits of agreement −266.5 to 235.5) (Fig. 1). The best mean distance achieved during the ISWT was 1165.08±327.19m. Males reached significantly longer distances than females (1319.33±359.78 vs 1032.86±225.90; p<0.001) (Table 2).

Figure 1.

Bland & Altman plot of the difference between the first and second ISWT plotted against the mean of the first and second ISWT. The bold line represents the mean difference between ISWT1 and ISWT2 and the dotted lines the 95% limits of agreement (n=130).

(0.05MB).

The assumptions of linearity, independence of errors, homoscedasticity, unusual points and normality of residuals were verified. Using a multiple linear regression, it was found that sex, age and BMI were statistically significant predictors of the ISWT (ANOVA p<0.001) and together explained 54% of their variability. Unstandardized regression coefficients (B), 95% confidence intervals and standard errors of the coefficients are presented in Table 3.

Based on the multiple linear regression model, the equation to predict ISWT was: ISWT predicted=342.06+(283.07×sex)+(83.61×age)− (22.22×BMI), with female=0 and male=1.

There was a significant difference between the actual ISWT and the predicted ISWT calculated with the equation established by Lanza et al. (1165.1±327.2 vs. 1058±152.6; p<0.001). A moderate correlation was found between these two values (r=0.69; p<0.001). The Bland & Altman plot also showed poor agreement and large limits of agreement between the actual ISWT and the predicted ISWT obtained with the equation established by Lanza et al.13 (mean difference – 107.1m) (Fig. 2).

Figure 2.

Bland & Altman plot of the difference between the actual and the predicted (from the equation established by Lanza et al.13) value of the ISWT plotted against the mean of the actual and the predicted value of the ISWT. The bold line represents the mean difference and the dotted lines the 95% limits of agreement.

(0.06MB).