Evidence Hierarchies and Quality Appraisal
Evidence hierarchies and quality appraisal are the methods evidence-based health care uses to decide how much confidence a body of research warrants. They rank study designs by their susceptibility to bias, appraise the internal validity of individual studies, and grade the overall certainty of evidence for a given question, so that clinical guidance and health technology assessment rest on an explicit, reproducible reading of the literature.
Definition
Evidence appraisal is the structured assessment of how trustworthy research evidence is for answering a defined question, combining a hierarchy of study designs, study-level risk-of-bias evaluation, and a graded judgement of the certainty of the overall evidence.
Scope
This area orients the reader to four linked topics: the traditional design-based evidence hierarchy and its modern refinements; the GRADE approach to grading evidence and recommendations; risk-of-bias assessment of individual studies; and the rating of overall certainty (confidence) in a body of evidence. It treats these as appraisal methods within evidence-based practice and health technology assessment, not as clinical instructions.
Sub-topics
Key concepts
- Evidence hierarchy / levels of evidence
- Internal vs. external validity
- Risk of bias
- Certainty (quality) of evidence
- GRADE grading framework
- Strength of recommendation vs. certainty of evidence
- Body of evidence vs. single study
Clinical relevance
These methods underpin how clinical practice guidelines, systematic reviews, and health technology assessments translate research into graded statements. Understanding them helps clinicians and decision-makers interpret why one recommendation is labelled strong and well-supported while another is conditional and uncertain; the area describes how evidence is appraised and is not itself a source of individual treatment advice.
Evidence & guidelines
Modern appraisal has converged on a few widely adopted frameworks: GRADE for grading certainty of evidence and strength of recommendations (Guyatt et al., 2008), Cochrane's risk-of-bias tools and handbook for study-level appraisal (Higgins et al., 2019), and revised evidence pyramids that place systematic reviews as a lens over the design hierarchy rather than at its apex (Murad et al., 2016). The underlying philosophy of integrating best evidence with expertise and patient values traces to Sackett et al. (1996).
History
The phrase 'evidence-based medicine' was popularised in the early 1990s, building on David Sackett's clinical epidemiology and the 1996 BMJ editorial that defined it as the conscientious integration of best available evidence with clinical expertise and patient values. Early hierarchies ranked randomised trials above observational studies; over the following two decades GRADE reframed grading around outcome-specific certainty, and the evidence pyramid was revised to reflect the role of systematic reviews and the limits of rigid design ranking.
Debates
- Is a fixed design-based hierarchy still adequate?
- Critics argue that ranking by design alone is too rigid, since a well-conducted observational study can outrank a flawed trial; revised pyramids and GRADE shift the focus from design rank to the certainty of evidence for each outcome.
Key figures
- David Sackett
- Gordon Guyatt
- Julian Higgins
- Andrew Oxman
- M. Hassan Murad
Related topics
Seminal works
- sackett-1996
- guyatt-2008-grade
- murad-2016-pyramid
Frequently asked questions
- What is the difference between an evidence hierarchy and a risk-of-bias assessment?
- A hierarchy ranks study designs by their general vulnerability to bias, whereas a risk-of-bias assessment evaluates how well a specific study was actually conducted; a high-ranking design can still have high risk of bias.
- Why does certainty of evidence matter alongside study design?
- Because the trustworthiness of a conclusion depends not only on design but on how studies were conducted, how consistent and precise their results are, and how directly they answer the question, certainty is graded across the whole body of evidence rather than read off design rank alone.