• Attrition bias

In Javaloyes\_2019 and Schmitt\_2018, the attrition bias was considered low because there were no missing outcome data. On the other hand, Kiviniemi\_2010, Nuuttila\_2017, and Vesterinen\_2016 presented high rates of follow-up loss for different reasons. These might be relevant in the ES observed. Moreover, no statistical procedure, such as intention-to-treat, was used to minimize this risk of bias. Therefore, they were considered as having a high risk of attrition bias. Finally, in Kiviniemi\_2007, the attrition bias was unclear because this outcome was not addressed in the study.
