Table 3. Quality assessment of the included studies.
| Author, year, reference | Selection bias | Confounding | Measurement | Study attrition | Data analysis | Overall quality | Comment |
|---|---|---|---|---|---|---|---|
| High-quality studies | |||||||
| Hallal, 2006[62] | ◉ | ● | ● | ● | ● | ● | |
| Hallal, 2012[60] | ◉ | ● | ● | ● | ● | ● | |
| Kehoe, 2002[48] | ◉ | ● | ● | ◉ | ● | ● | |
| Medium-quality studies | |||||||
| Atladottir, 2000[28] | ◉ | ○ | ◉ | ◉ | ◉ | ◉ | |
| Barbieri, 2009[29] | ● | ◉ | ○ | ◉ | ● | ◉ | |
| Cardona Cano, 2015 [40] | ◉ | ◉ | ◉ | ○ | ● | ◉ | |
| Gopinath, 2013[14] | ◉ | ● | ○ | ◉ | ● | ◉ | |
| Migraine, 2013[42] | ○ | ◉ | ◉ | ● | ● | ◉ | |
| Pearce, 2012[51] | ○ | ● | ● | ◉ | ● | ◉ | |
| Perälä, 2012[33] | ◉ | ● | ○ | ● | ● | ◉ | |
| Ridgway, 2011[11] | ◉ | ● | ● | ○ | ● | ◉ | |
| Silveira, 2012[44] | ◉ | ◉ | ◉ | ◉ | ● | ◉ | Identical study population as Escobar, 2014 |
| Van Deutekom, 2015[54] | ○ | ● | ● | ● | ● | ◉ | |
| Wijtzes, 2013[55] | ◉ | ● | ● | ○ | ● | ◉ | |
| Low-quality studies | |||||||
| Andersen, 2009[45] | ◉ | ● | ○ | ○ | ● | ○ | |
| Boone-Heinonen, 2015[30] | ◉ | ○ | ○ | ◉ | ● | ○ | |
| Boonstra, 2006[38] | ○ | ○ | ○ | ● | ○ | ○ | |
| Brown, 2012[39] | ○ | ○ | ○ | ○ | ○ | ○ | |
| Campbell, 2010[46] | ◉ | ○ | ● | ○ | ◉ | ○ | |
| Davies, 2006[10] | ○ | ○ | ○ | ○ | ◉ | ○ | |
| Dubignon, 1969[31] | ◉ | ○ | ● | ● | ○ | ○ | |
| Eriksson, 2004[15] | ◉ | ◉ | ○ | ○ | ◉ | ○ | |
| Escobar, 2014[41] | ◉ | ◉ | ○ | ○ | ● | ○ | Identical study population as Silveira, 2012 |
| Hack, 2012[12] | ◉ | ○ | ○ | ● | ○ | ○ | |
| Hildebrand, 2015[47] | ◉ | ○ | ● | ○ | ● | ○ | |
| Kajantie, 2010[56] | ◉ | ○ | ○ | ◉ | ● | ○ | Identical study population as Kaseva, 2012, Kaseva, 2013 and Kaseva, 2015 |
| Kaseva, 2012[13] | ◉ | ○ | ○ | ◉ | ● | ○ | Identical study population as Kajantie, 2010, Kaseva, 2013 and Kaseva, 2015 |
| Kaseva, 2013[61] | ◉ | ○ | ○ | ● | ◉ | ○ | Identical study population as Kajantie, 2010, Kaseva, 2012 and Kaseva, 2015 |
| Kaseva, 2015[57] | ◉ | ○ | ● | ○ | ● | ○ | Identical study population as Kajantie, 2010, Kaseva, 2012 and Kaseva, 2013 |
| Laaksonen, 2003[59] | ○ | ○ | ○ | ○ | ◉ | ○ | |
| Li, 2015[32] | ◉ | ○ | ○ | ● | ◉ | ○ | |
| Mattocks, 2008[49] | ○ | ○ | ● | ● | ● | ○ | |
| Oliveira, 2015[43] | ◉ | ◉ | ○ | ○ | ● | ○ | |
| Ounsted, 1975[37] | ○ | ○ | ○ | ○ | ○ | ○ | |
| Pahkala, 2010[50] | ○ | ○ | ○ | ○ | ○ | ○ | |
| Robinson, 2013[36] | ○ | ○ | ○ | ● | ● | ○ | |
| Rogers, 2005[58] | ◉ | ○ | ○ | ◉ | ◉ | ○ | |
| Ruiz-Narváez, 2014[34] | ○ | ○ | ○ | ● | ◉ | ○ | |
| Said-Mohamed, 2012[52] | ◉ | ○ | ● | ○ | ● | ○ | |
| Salbe, 1998[53] | ○ | ○ | ● | ○ | ◉ | ○ | |
| Shultis, 2005[35] | ● | ● | ○ | ○ | ● | ○ |
Results of the quality assessment of the included studies, with each dimension judged as strong(●), moderate (◉) or weak (○) based on the judgment rules as defined in Table 1.