Measurement properties and performance of an eight-minute submaximal treadmill test in patients with juvenile idiopathic arthritis: a controlled study

Juvenile idiopathic arthritis (JIA) can affect physical function and cardiorespiratory fitness (CRF). CRF is important for general health, and high CRF has been shown to decrease cardiovascular disease in general pediatric and adult populations [1‐3]. Previous studies have shown that patients with JIA have poor CRF measured with both maximal and submaximal exercise tests [4‐6]. Contrary to these results, we have recently reported that patients with oligo- and polyarticular JIA diagnosed in the era of biologics have comparable levels of CRF as age- and sex-matched controls from the general population, measured directly as peak oxygen uptake (VO_2peak) [7]. We believe our positive results may be explained by advances in the multidisciplinary management of JIA in the era of biologics, as well as differences in study populations. Importantly, 20–30% of both our patients with JIA and controls had poor CRF.

The gold standard method to measure CRF is through a cardiopulmonary exercise test (CPET) with direct measurement of VO_2peak, using a treadmill or bicycle to maximal exhaustion [8]. However, a CPET is time consuming, requires advanced and expensive equipment in a laboratory setting and extensive experience to encourage individuals to achieve maximal effort. Furthermore, performing a maximal exercise test may be uncomfortable and unpleasant for patients.

In contrast, indirect submaximal tests do not require the individuals to exercise to exhaustion, are easier to perform, require less advanced equipment, and are therefore frequently used in research and clinical practice to measure CRF [9]. The disadvantage of submaximal tests is less precise measurements of CRF compared to direct measurement of VO_2peak. Submaximal tests are usually developed to provide estimation of VO_2peak or to assess the distance covered in a given period of time or the time taken to cover a given distance. The most commonly used submaximal test in chronic pediatric conditions is probably the 6-min walk test (6MWT), even if the measurement properties vary largely among chronic pediatric conditions [10]. In JIA, the 6MWT has been suggested as a possible field test to measure walking ability, but is shown to be a poor predictor of VO_2peak [6, 11]. To the best of our knowledge, no submaximal walking tests aiming to estimate VO_2peak have been validated in pediatric populations.

An eight-minute submaximal treadmill test has been developed to estimate VO_2peak in healthy adults [12], and is proven valid for women with rheumatic diseases [13], who may experience similar symptoms as patients with pediatric rheumatic diseases. However, in healthy adults the test seems to either under- or overestimate VO_2peak depending on the chosen intensity [14]. The validity of this test is unknown for patients with JIA and healthy children. Also, knowledge about the reliability of the test is essential for both clinical practice and research purposes. Knowledge about how the performance of the test relate to disease variables is also warranted.

The objectives of the study were to evaluate the criterion validity and reliability of the eight-minute submaximal treadmill test in patients with JIA; also to investigate if the performance of the submaximal treadmill test is influenced by disease characteristics or differ from controls.

This is the first study to examine criterion validity and reliability of the eight-minute submaximal treadmill test aiming to estimate VO_2peak in patients with JIA. The results showed acceptable measurement properties on group level for both validity and reliability. The reliability was acceptable measured by ICC, but the measurement errors were large. On individual level, the validity was not acceptable, with large limits of agreement, and with no systematic bias. In controls, the validity of the submaximal treadmill test was not acceptable neither on group nor individual level. Patients with JIA and controls had similar estimated VO_2peak and submaximal treadmill test performance, and we found no associations with disease variables.

Compared to our results, studies on healthy adults [14] and women with rheumatic diseases [13] showed better validity. However, these studies applied different statistical methods than ours, making comparisons challenging. We applied ICC and Bland and Altman plots to evaluate criterion validity and reliability, statistical analyses methods recommended for these purposes by the COSMIN panel [17].

In the original study of the submaximal treadmill test, Ebbeling et al. [12] reported that there were no significant differences between the estimated and observed VO_2peak values in healthy adults, suggesting that the test has good predictive validity. We found similar results in both patients with JIA and controls when comparing estimated and observed VO_2peak mean values using paired t tests. The ICC value for evaluation of criterion validity at group level was acceptable in patients, but not in controls. However, our agreement analyses in both patients and controls showed large variation between the observed and estimated VO_2peak, but with no systematic differences between the observed and estimated VO_2peak. Agreement analyses were not conducted in the original article [12] or in other studies [13, 14]. However, a study of healthy adults [14] has reported a systematic overestimation of VO_2peak by 3.5 mL∙kg^− 1∙min^− 1 when testing at the moderate intensity (70% of the predicted HR_peak) and an underestimation of VO_2peak by 3.5 mL∙kg^− 1∙min^− 1,when testing at the low intensity (50% of the predicted HR_peak). The authors therefore suggested that if the purpose of using the submaximal treadmill test is to evaluate changes in CRF, all test sessions for the individual should be conducted at the same HR rather than the same speed. Thus, we aimed to test the participants at the same HR intensity (close to 70% of predicted HR_peak) when conducting the submaximal tests. With this approach, the SDC was large for both test-retest and inter-rater reliability in our patients.

At group level, a change of more than 1.5 mL∙kg^− 1∙min^− 1 and 1.1 mL∙kg^− 1∙min^− 1 would be required to be 95% certain that a real change has occurred for test-retest- and inter-rater testing, respectively. These small SDC_group values suggest that the submaximal treadmill test is reliable on group level in patients, which is important for research purposes. When mean scores of a group of patients are used instead of individual patient scores, the measurement error becomes smaller and subsequently, the measure is more reliable [17]. If the submaximal treadmill test is used for evaluating change in individual patients in clinical settings, the large measurement errors must be taken into consideration. Specifically, a change of more than 11.4 mL∙kg^− 1∙min^− 1 and 8.6 mL∙kg^− 1∙min^− 1 are required to be 95% certain that a real change in a single individual has occurred, for test-retest- and inter-rater testing, respectively.

Submaximal tests are based on the assumption that there is a linear relationship between HR, oxygen consumption, and exercise intensity [8]. Therefore, an accurate age-predicted HR_peak is of importance. We used the same prediction of HR_peak as Ebbeling et al. [12] when they developed the test. This equation is proposed to underestimate HR_peak with increasing age and other equations in adult populations have been suggested [25, 26]. In our study participants, the mean predicted HR_peak was 207 in both patients and controls, while the mean HR_peak observed from the maximal treadmill test was 196 and 197 beat·min^− 1 in patients and controls, respectively, suggesting overestimation of the predicted HR_peak when using the 220- age formula. In particular, as HR_peak varies between individuals, children with low HR_peak have probably been exercising at higher intensities than 70%. Importantly, there are many factors that may affect HR (e.g. hydration, caffeine, pain and anxiety). Nevertheless, the RPE and HR during the submaximal test indicate that the submaximal treadmill test is a test of submaximal intensity in these participants at group level. The formula by Tanaka et al. (208–0.7 x age) [25] was better for predicting HR_peak in both patients and controls than the formula by Nes et al. (211–0.64 x age) [26] and the 220-age formula [23] (data not shown). The formula by Tanaka et al. was also preferable over the 220-age formula in another study involving children [27].

The estimated VO_2peak and the performance of the submaximal treadmill test were comparable between patients with JIA and controls. This is in line with our previous findings studying the same cohort; directly measured VO_2peak was comparable between patients with JIA and controls [7]. Furthermore, we observed no correlation between disease variables and estimated VO_2peak and walking distance in patients. We have previously also reported that disease variables were not associated with any components of physical fitness in our patient cohort [7]. Taken together, our results suggest that disease variables are less important for physical fitness, including submaximal performance, in patients treated with a modern multidisciplinary management of JIA.

Our study has several strengths; we applied the COSMIN recommendations for evaluating the criterion validity and reliability of the submaximal treadmill test and the gold standard test was used as criterion measurement. Also, both physiotherapists conducting the submaximal tests for evaluation of reliability were experienced in pediatric rheumatology and one of these physiotherapists also conducted all maximal and submaximal treadmill tests used to evaluate the criterion validity. Also, the sample size was adequate. However, some limitations need to be considered. The equation to estimate the VO_2peak was developed in healthy adults aged 20–59 years, and it can be questioned if the formula is valid to use in patients with JIA and controls aged 10–16 years. Our JIA cohort seems well treated with low disease activity and functional disability, thus the findings may not be generalized to patients with higher disease activity or JIA categories not included in the current study. Additionally, the majority of the individuals included in the present study were females, which also could have hampered the generalizability of the results, although the formula used to estimate VO_2peak takes sex into account. Thus, future research should include other JIA categories and more males to improve the generalizability of the results.

In patients with JIA, the submaximal treadmill test shows acceptable criterion validity at group level but not at individual level. The reliability of the test is acceptable, but with large measurement errors for both test-retest- and inter-rater reliability. Our results support that the submaximal treadmill test is valid and reliable for research purposes (on group level), but not optimal to estimate VO_2peak in individual patients. Estimated VO_2peak and performance of the submaximal treadmill test did not differ between patients and controls and were not associated with disease variables, probably reflecting the positive effect of modern multidisciplinary management of JIA.