FormalPara Key Points

Previous research has reported the validity and reliability of a number of tests of children’s fitness.

Our systematic review indicated that the 5-min run test did not meet accepted criteria for reliability, whilst the 6-min walk test failed to meet the criteria for validity.

We further identified that of all sufficiently valid and reliable tests of children’s fitness, the 20-m shuttle run test was identified as the most scalable.

1 Introduction

The health and fitness of children is increasingly recognised as a core component of public health. Two reasons for this growing emphasis are evident. Firstly, poor health adversely affects the quality of life, and the physical, academic and social development of children. Second, poor health in childhood may predispose to certain diseases and is often therefore predictive of poor health in adulthood [1]. To this end, the UK Chief Medical Officer [2] stated “the introduction of a standardised school-based fitness assessment in England may have multiple benefits that extend beyond the benefits for the individual”. Such assessment could focus on the measurement of physical activity, and/or the measurement of the results of physical activity. Methods might range from the very basic such as the total time children spend in physical education (PE) lessons and/or the number of children who take part in extracurricular physical activity, to the more complex, such as the evaluation of motor skills and physical literacy, and/or the measurement of cardio-respiratory fitness.

However, none of the above measures are currently mandated in UK schools. The current mandated measure, the National Child Weight Measurement Programme (NCMP, http://www.hscic.gov.uk/ncmp), measures body mass index (BMI). The NCMP represents one of, if not the only, proxy measures of a child’s health across the UK. Given its broad coverage, it provides valuable data on child health at a local and national population level. Arguably, however, the BMI of a child is a crude metric at best, often saying as much about genetics and somatotype as about physical activity levels and health. In fact BMI in young childhood is at best only moderately predictive of subsequent adult health status [3].

Public health agencies in the UK are encouraging novel interventions to increase levels of childhood physical activity. However, the widespread lack of routine data collection identified above renders it problematic to evaluate the true impact of any such interventions. It also renders it almost impossible to set benchmarks, to identify local pockets of excellence (or indeed underperformance), or to calculate the cost-effectiveness of interventions. Whilst many areas of public health policy are characterised by a clear evidence-based strategy, decisions relating to the health and fitness of the nation’s children are often made in an evidence vacuum.

In the short- to medium-term what is required is a means of testing the health and fitness of children that is not only valid and reliable, but is also ethical and cost-effective. It is also abundantly clear that any large-scale fitness testing of children would need to be conducted in the field as opposed to the laboratory, as the provision of resources required for the latter would be prohibitive in the extreme.

The decision as to which test should be used is challenging. Data pertaining to the reliability and validity of tests of children’s fitness are widely available. For example, Castro-Piñero et al. [4] conducted a systematic review of the criterion related validity of field based fitness testing methods in children. The results of 73 studies suggested strong support for the 20-m shuttle run test as a valid means by which to estimate cardiorespiratory fitness in children and adolescents. Likewise, Artero et al. [5] conducted a systematic review to determine the reliability of children’s fitness testing methods and reported the most reliable field-based test of cardio-respiratory fitness was the 20-m shuttle run test.

However, whilst validity and reliability are of critical importance, in the field-test context it is often required that further criteria are met. Whilst receiving little attention in the scientific literature, the criterion of scalability, that is the potential for the extension into real-world policy and/or practice of interventions or tests shown to be efficacious in controlled settings [6] is often critical to policymakers and practitioners.

1.1 Aims of the Present Review

Our aim is to propose a novel framework by which researchers and practitioners might assess the scalability of field-based fitness tests appropriate for primary school children aged 8–11 years. We propose an algorithm by which the scalability of a candidate test can be evaluated. We then apply this algorithm to data identified via a systematic review to assess the scalability of children’s fitness tests.

2 Methodology

2.1 Identification of Components of Scalability

Scalability is to all intents a latent variable and cannot be directly measured. In order to overcome this, a collection of items or components hypothesised to co-vary with the latent variable were identified used as a proxy measurement [7].

Whilst the concept of scalability is becoming progressively more significant in public health, there is only limited information relating to its definition and core constituents. Terms used to described scalability have been applied in many different ways and contexts, with little consistency or rigour [6]. In an attempt to bring some clarity to terminology used, Milat et al. [6, 8] proposed eight core constituents: (1) delivery, (2) effectiveness, (3) cost-effectiveness, (4) evaluation, (5) reach and adoption, (6) evidence of operating at scale, (7) resource requirement and (8) practical implementation issues.

We adapted the eight criteria proposed by Milat et al. [6] to six components for the specific case of field-based fitness testing methods. Some components were represented by a single variable, whilst other components were constructed using multiple variables. These components and related variables are presented in Table 1.

Table 1 Scoring schedule for components of scalability
Full size table

2.2 Algorithm Construction and Scoring

We constructed an algorithm as the sum of weighted scores for each of the core constituents of the scalability framework:

$$ \left\{ {x_{1} , x_{2} , x_{3} , \ldots, x_{n} } \right\} $$

In this algorithm each x n represented one core constituent n of the scalability framework, i.e. delivery. Constituents of the algorithm were weighted as described below:

$$ X_{\text{score}} = \mathop \sum \limits_{i = 1}^{n} w_{i} x_{i} $$

Each single variable of a component could take a value from 0 to 2, and these variable scores were summed to produce each component score. A maximum of 8 points was possible for each of the six components (Table 1), resulting in a possible maximum scalability score of 48 for each test. We had no a priori reason to justify weighting certain components more heavily than others, so by increasing the weighting of components with low numbers of variables we were able to ensure that each component contributed equally to the overall score (however, excluding tests that did not meet validity and reliability criteria in effect weighted these two variables highly in the scalability analysis).

2.3 Systematic Review

To facilitate the testing of the scalability algorithm, a systematic review of studies reporting tests of children’s fitness was conducted. The objective of this review was to ensure that we only established the scalability of tests that demonstrate sufficient validity and reliability.

2.3.1 Inclusion Criteria

To be included in the review, papers had to report a study of one or more of the fitness tests addressed in two recent systematic reviews [4, 5], namely the 20-m shuttle run, 1-mile run, 6-min walk, and 5-min run. Fitness tests meeting these criteria were assessed against three criteria likely critical to the successful implementation of fitness testing of schoolchildren; the validity of the test for use with children aged 8–18 years old, the reliability of the test in this age group, and the applicability of the test, that is whether a test could be implemented in a school setting as part of usual PE lessons, albeit by specially trained staff. These primary criteria were considered fundamental to the child fitness measurement scenario described in the introduction.

2.3.2 Evidence Criteria

A three-tier classification of evidence quality was used [4], albeit in this case referring to the validity and reliability of the tests: (1) strong evidence, that is consistent findings in three or more studies; (2) moderate evidence, that is consistent findings in two studies; and (3) inconsistent results found in multiple studies, results based on one single study, or results indicate low scalability or no information found (Fig. 1).

Fig. 1
figure 1

Flowchart of test assessment. PE physical education

Full size image

2.3.3 Literature Search

The literature search was undertaken between May and July 2015 using the PubMed database. Key words searched, using multiple combinations of AND/OR phrases, included ‘cardiorespiratory fitness’, ‘children’, ‘testing’, ‘field’, ‘youth’, ‘adolescents’, ‘CRF’, as well as individual test names. Further papers were identified via examining reference lists of publications already identified.

2.3.4 Data Extraction

Operational definitions for scalability characteristics are presented in Table 2. Studies were assessed on whether data relating to these characteristics were reported (Table 3). Information relating to delivery, effectiveness, cost-considerations, resource requirement and practical implementation issues were all extracted. A further data extraction form was created to capture information in studies that had directly assessed some aspect of scalability. Items in this form included ease of integration into usual service delivery, burden on delivery staff, preparation requirements, test duration, reach and adoption, completion rates, resource requirements, practical implementation issues and considerations. These are presented in Table 4.

Table 2 Details of review items relating to scalability framework
Full size table
Table 3 Scalability properties of reviewed articles
Full size table
Table 4 Scalability of field based cardiovascular fitness tests
Full size table

3 Findings

A total of 229 studies reporting field-based tests of children’s cardiorespiratory fitness were identified. Initial analyses indicated that the 5-min run test did not meet the evidence criterion for reliability, whilst the 6-min walk test likewise failed to meet the evidence criterion for validity. A total of 25 studies remained for inclusion in the analysis. Of these, 19 reported the application of the 20-m shuttle-run, and six the one-mile walk/run (note that some studies considered more than one test) (Tables 2 and 3). A further four studies were identified that directly evaluated one or more aspects of scalability of field-based cardiorespiratory fitness tests for children and/or adolescents, and an additional five studies provided information on test protocols. These articles were used to complete the data extraction tables (Tables 3, 4).

Table 5 contains review items score totals for all included articles. For example, the table shows that out of the 25 articles, 8 % (n = 2) addressed practical implementation issues. A further 20 % (n = 5) received a partial score, with the reduction in rating predominantly due to the lack of information provided regarding practicality issues of administering the test, whilst 72 % (n = 18) reported no data relating to this variable.

Table 5 Assessment percentage scores for reviewed articles
Full size table

3.1 Testing the Algorithm

The algorithm was used to rate the relative scalability of the 20-m shuttle run test and the 1-mile walk/run. Table 6 presents scores for each of the tests and Fig. 2 shows a spider diagram comparing component scores. The authors independently scored each test and a consensus meeting was arranged to interrogate and resolve any differences. The 20-m shuttle run test scored 34 of a possible 48 whilst the 1-mile walk/run scored 25. This indicates that of the two tests that met the criteria for validity and reliability, the 20-m shuttle run test is more scalable than the 1-mile walk/run test. However, a lack of information relating to cost-effectiveness/affordability of test delivery, economies of scale and marginal costs was evident and is discussed further below.

Fig. 2
figure 2

Scalability scores for 20m SRT compared with 1 mile walk/run test. SRT shuttle run test

Full size image
Table 6 Scalability scores for the 20-m shuttle run test and the 1-mile walk/run test
Full size table

4 Discussion

Year on year, greater emphasis is being placed on ensuring the real-world impact of scientific research, and the line between science and research on the one hand and policy and practice on the other is not as clearly defined as once it was. Scientists are increasingly expected to conduct research that not only reports traditional scientific metrics, but also data related to the real-world application of those, for example data pertaining to cost-effectiveness in health intervention research. A good example perhaps is that of Robertson et al. [40], who examined not only the validity and reliability of tests of skill in sport, important to those who use the data, but also the feasibility of the tests, equally important to those who conduct the testing.

Whilst the criteria of validity and reliability of children’s fitness tests are of major concern to scientists, the criterion of scalability is critical to policymakers and practitioners. As little is known about the scalability of fitness tests for children, in the present paper we presented data that will facilitate future decision making as to test provision, whilst also proposing a framework that could be applied to examine scalability in the context of either new data relating to existing tests or of data pertaining to new tests. Using this method we demonstrated that, based on available data, the 20-m shuttle run test is likely more scalable than the 1-mile walk/run test, with these tests scoring 34 and 25 of 48, respectively. However, a word of caution is required here given the stark contrast between the number of studies initially identified and the number of studies that met the inclusion criteria.

Whilst it is entirely understandable that scientific reports of fitness tests do not require the reporting of non-scientific data points such as costs, it is probably reasonable to suggest that with the increasing emphasis on real-world application and impact, it is incumbent on journal editors and reviewers, as well as policymakers and those funding research, to push for greater reporting of all such data where appropriate (this would perhaps be analogous to the way that the broader acceptance of meta-analysis as the gold standard of research synthesis has encouraged editors and funders to require the reporting effect sizes and/or all necessary data points to calculate these). We hope that this paper, by identifying the core components of scalability in the context of children’s fitness testing might encourage that process.

It is important to acknowledge limitations of the methodology reported. Firstly, as is the case with many if not most attempts at research synthesis, there was a stark contrast between the number of studies initially identified and the number of studies that met the inclusion criteria. This was likely compounded by our two-stage analysis. Without the reporting of all relevant data, however, it is problematic to evaluate scalability, and this was especially the case with regard to cost-effectiveness/affordability of test delivery, economies of scale, and marginal costs, for which no information could be found for either of the two fitness tests addressed in this study.

Second, in examining the literature we found only limited information on the definition of scalability and its core constituents. Therefore there are potentially one or more components of scalability that are not incorporated in our framework. For example, ethical consideration could be an important a priori factor in light of emerging web-based technologies.

Third, and related to the second, given this was a pioneering approach we had no a priori reason to justify weighting certain components within the framework more heavily than others. However it may be that in practice/application of the model, fundamental constraints to testing may evolve and the model may need to be developed accordingly. Such constraints may differ depending on who is applying the framework, for example whilst researchers may be more focused on ethics and controls, practitioners and policymakers may be more focused on costs.

5 Conclusions

Recent systematic reviews by Castro-Piñero et al. [4] and Artero et al. [5] indicated strong support for the validity and reliability of the 20-m shuttle run test in the context of children’s fitness testing. Our analysis above should further encourage practitioners and policymakers to adopt this test either as an adjunct to, or replacement for, existing mandated tests such as the UK NCMP.

We also believe that the scalability framework developed in this paper has value beyond that of the context above. It has potential value in establishing the scalability of many types of fitness tests and/or measures, as well as in informing policy-makers in the up scaling of interventions from small projects or controlled trials to wider state, national or international programs.