“…Therefore, this meta-analysis included only observational studies, which have lower evidence levels than randomized clinical trials (Burns, Rohrich, & Chung, 2011; GRADE Working Group, 2004). Only a few cohort studies (de Morais, Mota, & Amorim, 2014; dos Santos Neto, Oliveira, Barbosa, Zandonade, & Oliveira, 2012; Karjalainen, Rönning, Lapinleimu, & Simell, 1999; Lescano de Ferrer & Varela de Villalba, 2006; Moimaz et al, 2014; Warren & Bishara, 2002) and study designs nested within cohorts (Caramez da Silva et al, 2012; Peres et al, 2007; Peres, Cascaes, Peres, et al, 2015; Viggiano et al, 2004), which are considered more robust than cross-sectional studies, were identified (Burns et al, 2011; GRADE Working Group, 2004). Furthermore, three such studies were not included in our quantitative synthesis due to a lack of raw data (dos Santos Neto et al, 2012; Karjalainen et al, 1999; Peres, Cascaes, Peres, et al, 2015).…”