Commentary on a recent article on the effects of the ‘Daily Mile’ on physical activity, fitness and body composition: addressing key limitations

Intervention participants were exposed to TDM for approximately 28 weeks (October to May), which is 2.5 times longer than the approximately 12 weeks (March to June) between the pre- to post-measurement period for the control group. While the authors controlled for age to account for varied start times, to our knowledge, this approach does not fully ameliorate such limitations.

The pre- and post-intervention measurements for the TDM group were taken in October (autumn) and May (summer), whereas the control measures were taken in March (spring) and June (summer). Seasonality [8] and weather conditions [9] impact PA levels. Generally, within the UK, MVPA levels are higher in spring and summer compared to autumn and winter [10, 11]. Controlling for the weather and season in the analyses could account for some of these differences [9]. Additionally, body fat levels also tend to be higher, and fitness lower, following the UK summer vacation [12]. Furthermore, baseline outcomes identified important between-group differences that may influence responsiveness – control participants had higher levels of MVPA (7/7 groups), fitness (6/7 groups), BMI z-scores (4/7 groups) and sum of skinfolds (6/7 groups). Although analyses controlled for age, these baseline characteristics, combined with varied exposure duration, may unduly influence study findings.

The standard approach of representing intervention effects through central tendency revealed that TDM increased MVPA by an average of 9.1 min; however, a visual count of the data in figure two A [7] revealed highly varied intervention effects. For example, in the intervention group, these ranged from an approximately 55 min/day decrease in MVPA to an increase of approximately 60 min/day. Furthermore, while 55% of the intervention participants had increased MVPA outcomes following the TDM intervention, 45% reduced their MVPA. Clearly, substantial numbers of children did not respond positively to TDM, with similar results as those found in the control group, where 44% improved their MVPA outcomes at follow-up and yet 56% of children had worsened outcomes. Therefore, in this context, it is not accurate to claim universal benefit. Future studies using larger sample sizes should use subgroup analyses to identify the characteristics of the participants who may and may not benefit from TDM [13].

The final intervention MVPA sample included only 17% of the original participants (n = 56) and 37% of original controls (n = 62). It is unwise to assume these children are representative of the full sample and this further limits the generalisation of outcomes. Although not essential, sensitivity analyses would have enabled comparisons of key subsample characteristics and established any bias arising from potential systematic attrition of individuals [14]. Where feasible, we recommend that future studies assess the effectiveness of TDM by recruiting larger samples across multiple schools. To account for the nested structure of pupils within classes within schools, multilevel modelling analyses is justified [15].

Four key issues arose in relation to accelerometer use. First, data were collected in 60-s epochs, rather than the established 15-s epochs in the original cut-point validation study [16]. Current evidence suggests deviation from the validated cut-point epoch reduces the validity of PA outcomes [17]. Second, shorter epochs of ≤ 15 s more accurately assess the sporadic higher intensity activity present within children’s PA patterns [17, 18]. Third, while the use of five different accelerometer models was justified in the paper, it was not reported how consistently participants wore the same accelerometer model or unit in the pre- and post-assessments. Fourth, consistent with previous studies, 3 days of wear time were required to estimate accelerometer outcomes. However, these included weekend days when TDM was not delivered, meaning that MVPA outcomes influenced by weekend factors are not attributable to TDM. Related to this final point, it may have been useful to understand what happens to PA levels on days with and without TDM; previous studies have observed between-day compensatory effects for PA and sedentary behaviour [19].

The contribution of TDM was not directly confirmed using segmented day analysis to identify acute MVPA responses and outcomes. Further, no assessment of treatment fidelity occurred to assess how often TDM was implemented throughout the intervention period. Therefore, it is difficult to qualify the outcomes that relate directly to TDM.