|
The class of clinical question (e.g., prevention, screening, prognosis, safety) in many ways determines the constituent levels in a hierarchy of evidence quality. Furthermore, many variations of hierarchies can be found for each class. Just what set of criteria constitutes a correct, or appropriate, hierarchy is wholly dependent upon the clinical question and the clinical practice. The two hierarchies displayed below are examples for a clinical question on treatment and a clinical question on diagnosis.
For thorough discussions of levels of evidence, see the Web pages of the Agency for Healthcare Research and Quality and the Oxford Centre for Evidence-Based Medicine.
|
Level |
Type of Treatment Study |
| 1a |
Systematic reviews or meta-analyses of high-quality randomized controlled trials |
| 1b |
High-quality randomized controlled trials |
| 2a |
Systematic reviews or meta-analyses of high-quality non-randomized controlled trials |
| 2b |
High-quality non-randomized controlled trials |
| 3a |
Systematic reviews of cohort studies |
| 3b |
Individual cohort study or low quality randomized controlled trials |
| 4 |
Clinical outcome studies |
| 5a |
Systematic review of case-control studies |
| 5b |
Individual case-control studies |
| 6 |
Case-series |
| 7 |
Expert opinion without explicit critical appraisal |
|
Note. Levels 1 through 5b could be further fractioned by experimental precision. For example, level 1a could become 1a(+) and 1a(-) for grouping high- and low-precision experiments respectively. |
|
Level |
Type of Diagnostic Study |
| 1a |
Systematic review or meta-analysis of 1c studies (high-quality trials) |
| 1b |
Independent replication of a 1c study |
| 1c |
A diagnostic study having a representative and consecutive sample and appropriate reference standard (e.g., gold-standard test) in an independent blind comparison demonstrating validated specificity and sensitivity that are almost absolute |
| 2a |
Systematic review or meta-analysis 2b diagnostic studies |
| 2b |
A cohort study with a good reference standard |
| 3a |
Systematic review or meta-analysis 3b studies |
| 3b |
A diagnostic study having a non-consecutive sample or a consistently applied reference standard |
| 4 |
One or more case-control study
Expert opinion without explicit critical appraisal |
| Note. Levels 1 through 3b could be further fractioned by experimental precision. For example, level 2a could become 2a(+) and 2a(-) for grouping high- and low-precision experiments respectively. |
|