• exercise
• meta-analysis
• BMI

Dear Editor,

We appreciate the interest in our recent meta-analysis1 and welcome the opportunity to reply.2

First, with regard to methodological quality of the studies included in our systematic review, we totally agree with the comment regarding the need for a specific tool to evaluate the risk of bias in non-randomised intervention studies. Despite the low quality, no trials were excluded according to this quality. In our meta-analysis, we chose to include all studies and to perform stratification when was appropriate. In order to overcome this problem, we included a sensitivity analysis as a supplementary material, which included only randomised controlled trials (body mass index (−1.011, 95% CI −1.846 to −0.176), fat mass in % (−3.536, 95% CI −6.053 to −1.018) and kg (−4.412, 95% CI −7.208 to −1.615), lean body mass (2.806, 95% CI 0.573 to 5.040), low-density lipoprotein cholesterol (−9.626, 95% CI −16.67 to −2.579), homeostatic model assessment of insulin resistance (−0.676, 95% CI −1.332 to −0.020) and adiponectin (2.498, 95% CI 1.697 to 3.298)) and we obtained similar results for the global effect sizes. Additionally, we have analysed the non-randomised controlled trials with the Risk Of Bias In Non-randomised Studies - of Interventions (ROBINS-I) tool3 and all of them were judged at low or moderate risk of bias.

With regard to the second remark, the correct reference4 pertains to the expert panels by the WHO, in which the recommendation is for overweight and obese subjects to lose 10% of the initial weight. Also, regarding the reference published by Biddle and Batterham,5 we agree that the article makes a reflexive analysis about the high-intensity interval training (HIIT) for public health; however, this study highlights the great potential for HIIT interventions in order to address areas of public health priority such as obesity, which is emphasised in our sentence.

Finally, as with most meta-analyses and reviews including a variety of study qualities, our results must be interpreted with caution. However, it is important to reiterate that when we reanalysed our data in the sensitivity analyses, the conclusions reported are well supported by the randomised controlled trials.