Table 1. Characteristic of the included studies.

Study	Years	Country	Study type	Study design	Case number	Controls number	Neonate	Birth weight	Imaging confirm	Risk of bias
Benders 2007 [20]	1990–2005	US	retrospective	case-control	31	93	preterm	Cases:1599±633 Controls:1607±720	MRI	different detecting method, only preterm neonates
Chabrier 2010 [21]	2003–2006	France	prospective	cohort study	100	100	both	NR	CT/MRI	different detecting method, no description of those lost to follow up
Darmency-Stamboul 2012 [22]	2000–2007	France	retrospective	case-control	32	96	term	NR	CT/MRI	different detecting method, only term neonates
Estan 1997 [8]	1987–1993	UK	retrospective	case-control	12	24	term	Cases:3496(2541–4460) Controls:3673 (2748–4707)	CT	different detecting method, only term neonates
Harteman 2012 [14]	2000–2010	Utrecht	retrospective	case-control	52	156	term	Cases:3420(2145–5230) Controls:3520(2155–4925)	MRI	different detecting method, only term neonates
Lee 2005 [6]	1997–2002	US	retrospective	nested case-control	37	111	both	Cases:3127±852 Controls:3203±923	CT/MRI	different detecting method
Wu 2004 [9]	1991–1998	US	retrospective	nested case-control	38	218	term	NR	CT/MRI	different detecting method, only term neonates
Kirton 2011 [23]	2003–2007	IPSS	prospective	case-control	248	population	both	NR	CT/MRI	different detecting method

NR: not reported. IPSS: Europe, Canada, US, South America, Asia, Australia; NR: not reported.