1a	–	Meta-analyses of randomized, controlled studies
1b	–	At least one randomized, controlled study with a narrow confidence interval
1c	–	So-called "all-or-none" studies,
		i.e., the new treatment is found to be markedly better than the previous one
2a	–	Meta-analyses of cohort studies
2b	–	A small number of cohort studies or randomized, controlled studies of low quality
2c	–	Outcome studies
3a	–	Meta-analyses of case-control studies
3b	–	A small number of case-control studies
4	–	Case series, cohort studies, and case-control studies of low quality
5	–	Expert opinion or purely experimental findings