Levels of evidence and grading - where do they come from?

EDITED with answer (see below)... and another one

I'm curious to know where the 'levels of evidence' come from - I see them everywhere and have just found seemingly identical versions from doing a google search of
"systematic review or meta-analysis of randomised controlled trials"

Level of Evidence
1a Systematic review or meta-analysis of randomised controlled trials
1b At least one randomised controlled trial
2a At least one well-designed controlled study without randomisation
2b At least one well-designed quasi-experimental study, such as a cohort study
3 Well-designed non-experimental descriptive studies, such as comparative studies, correlation studies, case–control studies, and case series
4 Expert committee reports, or opinions and/or clinical experience of respected authorities

Since the results suggest that many people are using this exact format of words it suggests that they've got them either from each other or from one place. I feel I ought to know where. Do you?
Maybe I should try re-running that search with the American spelling...

EDIT: 16 August 2010
@jdc325 pointed out that this is probably the source
http://www.cks.nhs.uk/gi_upper_cancer_suspected/evidence/supporting_evidence/evidence_grading#
Eccles, M. and Mason, J. (2001) How to develop cost-conscious guidelines. Health Technology Assessment 5(16), 1-78.
Context page: http://www.hta.nhs.uk/project/995.asp
Jump to full text PDF: http://www.hta.nhs.uk/fullmono/mon516.pdf

EDIT: 21 August 2010
@vmontori pointed out that "Levels of Evidence outdated and incorrect, eg reviews are as reliable as individual study quality; opinions are not evidence but judgments about evidence" and subsequently recommended instead that I "lookup the GRADE system for quality of evidence - www.gradeworkinggroup.org"