Hierarchy of Evidence Pyramid: Writing Guide

The hierarchy of evidence pyramid ranks clinical and health research study designs according to their intrinsic vulnerability to bias and systematic error, traditionally placing systematic reviews and meta-analyses at the apex and expert opinion or bench research at the base. In academic manuscripts, grant proposals, and review protocols, researchers use this conceptual hierarchy to establish study design eligibility criteria, justify search boundaries, and frame critical appraisal workflows.

When drafting the methodology section of an evidence synthesis or clinical report, authors must translate the theoretical tiers of the evidence pyramid into reproducible inclusion thresholds. Reporting standards such as PRISMA 2020 Item 5 and the Cochrane Handbook require explicit justification when certain study designs are included or excluded. Integrating this structure effectively strengthens manuscript rigor and clarifies the inferential strength of included literature.

Purpose of the section

The evidence hierarchy section within an academic manuscript or synthesis protocol serves three central research-practice functions:

• Operationalizing eligibility criteria: Broad terms like “best available evidence” cannot guide a systematic literature search. Authors define exact study design boundaries—such as restricting primary effectiveness evaluations to randomized controlled trials (RCTs) while permitting prospective cohort designs for adverse outcomes. For structural conventions across review articles, see our guide on the layout of literature review.
• Balancing internal validity with feasibility: While randomized designs minimize confounding, certain clinical questions cannot ethically or practically be evaluated via randomized allocation. Stating the evidence tier rationale clarifies why non-randomized studies of interventions (NRSI) or observational data were gathered.
• Establishing critical appraisal and certainty frameworks: Identifying study design tiers prepares the review team to apply design-specific risk-of-bias instruments and structured certainty assessments, such as GRADE. To explore these evaluation frameworks, examine our overview of critical appraisal tools and our detailed analysis of levels of evidence and GRADE certainty thresholds.

Structure (ordered list)

The conventional evidence pyramid organizes study designs into distinct methodological tiers, moving from foundational, hypothesis-generating investigations at the base to aggregated, synthesized evidence at the apex. Methodological updates, notably the revised pyramid proposed by Murad et al. (2016), emphasize that study execution and risk of bias can alter these boundaries.

1. Systematic Reviews and Meta-Analyses (Synthesized Evidence): Sitting at the traditional summit, systematic reviews identify, appraise, and statistically synthesize all eligible studies addressing a focused research question. In modern evidence-based practice, Murad et al. (2016) conceptualize systematic reviews not as an isolated primary tier, but as an evaluative magnifying lens through which underlying primary studies are appraised and interpreted. Detailed distinctions between review methodologies are discussed in our comparison of narrative reviews vs systematic reviews.
2. Randomized Controlled Trials (RCTs): The benchmark primary study design for evaluating therapeutic interventions. Participants are randomly allocated to experimental or control groups, balancing known and unmeasured confounding factors across cohorts (Cochrane Handbook Section 3.3.1).
3. Non-Randomized Controlled Trials (Quasi-Experimental Studies): Comparative studies where allocation to intervention groups is controlled by investigators without true randomization (e.g., allocation by date of birth or facility). While valuable when randomization is impossible, baseline selection bias remains a major vulnerability.
4. Cohort Studies (Observational Longitudinal): Prospective or retrospective investigations tracking exposed and unexposed groups over time to assess outcome incidence. Cohort studies are critical for evaluating long-term outcomes, environmental exposures, and prognostic factors.
5. Case-Control Studies (Observational Retrospective): Investigations identifying participants by disease or outcome status (cases) and matching them with unaffected individuals (controls) to examine historical exposures. Efficient for studying rare diseases, but susceptible to recall and selection biases.
6. Cross-Sectional Studies, Case Series, and Case Reports: Descriptive studies documenting disease prevalence at a single time point (cross-sectional) or tracking specific patient presentations without a control group (case series/reports). These designs generate clinical hypotheses and detect adverse events but cannot establish causality.
7. Expert Opinion, Mechanistic Reasoning, and Preclinical Research: Foundational evidence derived from laboratory bench models, animal experiments, physiological reasoning, and expert consensus. While essential for establishing biological plausibility, they lack direct clinical confirmation in human patient populations.

In the revised model proposed by Murad et al. (2016), horizontal dividing lines between study designs are replaced with wavy lines reflecting the GRADE evidence certainty rating. When an observational study demonstrates a massive effect size with no residual confounding, its certainty can exceed that of an underpowered or high-bias randomized trial.

A short annotated example

Below is a sample excerpt illustrating how to document study design eligibility and the evidence hierarchy in the Methods section of a clinical systematic review protocol:

This structure satisfies PRISMA 2020 Item 5 by explicitly defining eligible designs, justifying the inclusion of non-randomized evidence, stating unambiguous exclusion boundaries, and pairing each tier with its corresponding appraisal instrument.

Common mistakes

Researchers commonly encounter several structural and conceptual pitfalls when incorporating the hierarchy of evidence pyramid into manuscript drafts:

• Relying on author-applied design labels instead of features: The Cochrane Handbook (Section 3.3.2) emphasizes that primary researchers frequently apply design labels (such as “prospective cohort” or “case-control”) inconsistently. Eligibility criteria should define concrete methodological features (e.g., prospective recruitment, consecutive enrollment, concurrent control group) rather than trusting self-assigned study labels.
• Treating the pyramid as an absolute exclusion filter: Automatically discarding all observational evidence can undermine clinical utility. When randomized trials are ethically infeasible or uninformative regarding long-term harms, excluding cohort studies leaves decision-makers without relevant empirical data.
• Assuming all systematic reviews provide high certainty: A systematic review inherits the weaknesses of its constituent studies. A meta-analysis pooling severely biased randomized trials generates a misleadingly precise, biased aggregate effect estimate.
• Confusing study design ranking with evidence certainty: Ranking a study as an RCT only sets its baseline certainty; poor allocation concealment, high attrition, imprecision, or publication bias can downgrade its certainty below that of well-conducted observational research.
• Omitting protocol pre-specification: Changing study design eligibility criteria after reviewing initial search hits introduces selective inclusion bias. PRISMA 2020 requires reporting whether criteria were established prior to study conduct.

Checklist

Use this six-point checklist when drafting or auditing the study design and evidence hierarchy sections of your manuscript:

• [ ] Protocol Pre-specification: Were study design eligibility criteria defined in a publicly registered protocol prior to conducting formal literature screening?
• [ ] Feature-Based Definitions: Are eligible designs described using objective methodological characteristics (e.g., random sequence generation, concurrent comparators) rather than ambiguous nominal labels?
• [ ] Explicit Justification for Non-Randomized Studies: If cohort, case-control, or quasi-experimental studies are included, does the text clearly articulate the clinical or ethical rationale for expanding beyond RCTs?
• [ ] Clear Exclusion Boundaries: Are excluded designs (e.g., case reports, cross-sectional surveys, narrative commentaries) stated explicitly alongside their methodological justification?
• [ ] Appraisal Tool Alignment: Is each eligible design tier mapped to a validated critical appraisal instrument (e.g., RoB 2 for randomized trials, ROBINS-I for non-randomized studies)?
• [ ] PRISMA 2020 Item 5 Compliance: Does the Methods text clearly detail how studies were grouped for synthesis according to design characteristics?

Discipline differences (only if sourced)

Application of the evidence hierarchy varies substantially across research domains based on feasibility, ethics, and research objectives:

• Clinical Pharmacotherapy: Strictly adheres to the traditional hierarchy. Regulators and clinical guideline panels require double-blind RCTs and systematic reviews of RCTs to establish efficacy, utilizing observational cohort data primarily for Phase IV post-marketing surveillance.
• Public Health and Policy Research: Evaluates population-wide interventions, environmental regulations, and healthcare system changes where individual randomization is impractical or unethical. Research in this field routinely relies on quasi-experimental designs, interrupted time series, and natural experiments as the primary source of actionable evidence (Cochrane Handbook Section 3.3.2).
• Diagnostic and Prognostic Accuracy: The therapeutic evidence pyramid does not directly apply to diagnostic test evaluations. Optimal diagnostic evidence derives from prospective cross-sectional studies comparing an index test against an independent reference standard, while optimal prognostic evidence derives from prospective inception cohort studies.
• Qualitative and Health Services Research: Explores patient values, intervention feasibility, and implementation barriers. Rather than ranking studies on quantitative causal hierarchies, qualitative syntheses apply specialized appraisal standards such as the JBI critical appraisal instruments and the GRADE-CERQual approach.

Frequently asked questions

What is the hierarchy of evidence pyramid?

The hierarchy of evidence pyramid is a visual model ranking research study designs according to their inherent risk of bias and confounding. Systematic reviews of randomized trials occupy the highest tier for therapeutic questions, followed by randomized controlled trials, observational cohort and case-control studies, and expert opinion at the base.

Why are systematic reviews placed at the top of the evidence pyramid?

Systematic reviews synthesize data from multiple independent studies using explicit, reproducible search and appraisal protocols. By pooling sample sizes across high-quality studies, they reduce random sampling error and provide a more comprehensive, unbiased estimate of effect than any individual trial.

Can an observational study provide stronger evidence than an RCT?

Yes. Under the GRADE framework and the revised evidence pyramid proposed by Murad et al. (2016), study quality and execution take precedence over study design labels. A rigorously conducted prospective cohort study with a large effect size and strict confounding control can provide higher certainty than an underpowered, unblinded RCT with high participant attrition.

What is the difference between filtered and unfiltered evidence?

Unfiltered evidence refers to primary research studies (such as RCTs, cohort studies, and case reports) reporting original investigation findings. Filtered evidence refers to secondary syntheses (such as systematic reviews, meta-analyses, and evidence-based practice guidelines) that systematically appraise, filter, and summarize primary studies.

Sources

• Murad MH, Asi N, Alsawas M, Alahdab F. New evidence pyramid. Evidence-Based Medicine. 2016;21(4):125-127. doi:10.1136/ebmed-2016-110401.
• McKenzie JE, Brennan SE, Ryan RE, Thomson HJ, Johnston RV, Thomas J. Chapter 3: Defining the criteria for including studies and how they will be grouped for the synthesis. In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (eds). Cochrane Handbook for Systematic Reviews of Interventions version 6.4 (updated August 2023). Cochrane.
• Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi:10.1136/bmj.n71.
• Joanna Briggs Institute. JBI Levels of Evidence and Grades of Recommendation Working Party. JBI Levels of Evidence for Effectiveness. Joanna Briggs Institute; 2014 (updated 2020).
• National Library of Medicine. Evidence-Based Medicine [MeSH Descriptor D019317]. National Center for Biotechnology Information; 1997.