Table 3 Levels of Evidence according to Oxford Centre for Evidence-Based Medicine 2011.
| Question | Level 1* | Level 2* | Level 3* | Level 4* | Level 5 |
|---|---|---|---|---|---|
| * Level may be downgraded because of study quality, extended confidence intervals (imprecise effect estimates), inconsistencies between studies, or because the absolute effect
value is very small, as well as lack of transferability (study question does not correspond to the clinically relevant question). An upgrade of the evidence level is possible in
case of large or very large effects. ** As a general rule, a systematic review is always better than a single study. *** Consecutive inclusion = patients are continuously recruited. 1 The STROBE statement, among others, can be used for quality assessment: http://www.strobe-statement.org/index.php?id=strobe-aims . 2 Single-patient studies in which patients receive alternating intervention and control intervention. 3 Follow-up study of a population from a completed RCT. 4 Study in which cases and controls are drawn from an ongoing cohort study. | |||||
| How widespread is the problem? | Local and current random sample or census (complete survey) | Systematic review of surveys that can be applied to local circumstances** | Local survey that is not based on a random sample** | Case series** | Not applicable |
| Is this diagnostic or controlling test accurate? (Diagnostic) | Systematic review of cross-sectional studies with reference standard applied throughout and blinding | Single cross-sectional study with reference standard applied throughout and blinding | Non-consecutive*** study or study without applied reference standard** | Case-control study or study with inappropriate or non-independent reference standard** | Expert opinion based on pathophysiological considerations |
| What would happen if we did not apply therapy? (prognosis) | Expert opinion based on pathophysiological considerations | Single cohort study of patients in the early stages of the disease (inception cohort study) | Cohort study or control arm of a randomized trial* | Case series or case-control study or a prognostic cohort study with low methodological quality 1 ** | Not applicable |
| Does this approach help? (use of the intervention) | Systematic review of randomized trials or N-of-1 studies 2 | Randomized trial or observational study with dramatic effects | Controlled cohort study/follow-up study 3 ** | Case series or case-control studies or studies with historical controls** | Expert opinion based on pathophysiological considerations |
| What are common side effects? (harm of intervention) | Systematic review of either randomized trials or embedded case-control studies 4 . Or N-of-1 study with patients matching the research question or observational study with dramatic effects | Randomized trial or (exceptionally) observational study with dramatic effects | Controlled cohort study/follow-up (post-marketing surveillance) study, with sufficient number of cases to identify a common side effect. If long-term side effects are to be recorded, the follow-up must be sufficient**. | ||
| What are rare side effects? (harm of the intervention) | Systematic review of randomized trials or N-of-1 studies | Randomized trial or (exceptionally) observational study with dramatic effects | |||
| Is this screening test useful? (screening) | Systematic review of randomized studies | Randomized trial | |||