Contents
Academy · Research Writing · RCT Systematic Reviews
Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence
A systematic review of randomised controlled trials (RCTs) addresses a narrower and more answerable question than most evidence syntheses: within trials in which allocation was randomised, what is the pooled effect of a treatment, and how much confidence can be placed in it.
A systematic review of randomised controlled trials (RCTs) is a structured synthesis that identifies every eligible randomised trial addressing a defined treatment question, appraises each one for how well its randomisation and conduct protected against bias, and, where the trials are similar enough, combines their results into a pooled effect estimate. It is the evidence-synthesis method the Cochrane Handbook was built around, and its defining feature is that randomisation itself, when properly executed, removes most confounding at the level of each individual trial, which is why RCT reviews can focus bias assessment on conduct (blinding, attrition, deviation from protocol) rather than on confounder adjustment.
Learning objectives
- The defined steps the Cochrane Handbook sets out for conducting an RCT systematic review
- How the RoB 2 tool structures risk-of-bias assessment specifically for randomised trials
- Why pooling and heterogeneity assessment, rather than trial counting alone, determine whether a meta-analysis is appropriate
Common misconceptions
- “Any systematic review that includes randomised trials is automatically a high-quality, low-bias synthesis.” A review of randomised trials still inherits every bias present in the included trials — selective outcome reporting, inadequate blinding and high attrition — which the RoB 2 tool is specifically designed to identify at the level of each included trial before any pooling occurs.
Definition of a systematic review of randomised controlled trials
A review qualifies as an RCT systematic review when its eligibility criteria restrict included studies to trials using random allocation to intervention and comparison groups, and when its methods follow a pre-specified, documented protocol for search, selection, and synthesis rather than an informal literature summary1. This distinguishes it from a narrative review, which need not be exhaustive or methodologically pre-specified, and from an observational-evidence systematic review, which faces the separate confounding problems that guidance such as COSMOS-E addresses. The randomisation requirement is what allows the subsequent bias-assessment stage to focus on trial conduct rather than confounder balance.
This definition has practical importance because some reviews blur the distinction by including a handful of non-randomised comparative studies alongside RCTs to broaden the evidence base for a question where trial data is sparse. When that happens, the appropriate approach is to analyse and report the randomised and non-randomised evidence separately rather than pooling them into a single estimate, since the two groups of studies carry different, non-interchangeable bias profiles; a reader needs to be able to see how much of a review's conclusion rests on randomised evidence specifically before deciding how much weight the conclusion deserves.
Conduct steps specified by the Cochrane Handbook
The Cochrane Handbook structures the process as a defined sequence: formulating a question using the PICO framework (population, intervention, comparison, outcome), developing and registering a protocol, running a comprehensive search across multiple databases and trial registries, screening against pre-specified eligibility criteria, extracting outcome data in a standardised form, and only then assessing risk of bias and considering synthesis2. Reporting the completed review follows the PRISMA 2020 statement, which specifies the items, including a flow diagram of how many records were identified, screened, and included, that a transparent RCT review must report3. Protocol registration, ideally on PROSPERO, is recommended before screening starts specifically to prevent outcome switching once results are known6.
The search stage deserves particular care because trial registries, including ClinicalTrials.gov and the WHO's International Clinical Trials Registry Platform, often surface completed but unpublished trials that a database search of journal articles alone would miss, and failing to search them is a recognised route by which a review can end up systematically overrepresenting trials with favourable, published results. Screening against pre-specified eligibility criteria, rather than criteria refined after seeing which studies are retrieved, prevents this stage from becoming, in effect, a selection of evidence toward a preferred answer; dual, independent screening by two reviewers, with disagreements resolved by a third, is standard Cochrane practice because single-reviewer screening is more prone to this kind of drift.
Assessment of bias within each trial before pooling
Because randomisation addresses confounding but not every source of bias, each included trial still needs individual appraisal, and the current standard tool for this is RoB 2, which assesses five specific domains: bias arising from the randomisation process itself, bias due to deviations from the intended intervention, bias from missing outcome data, bias in outcome measurement, and bias in selection of the reported result4. A trial can be well-randomised in its allocation process and still carry high risk of bias overall if, for example, attrition was substantial and differed between arms, or if the outcome reported was selected from several measured but unreported alternatives. RoB 2's domain-by-domain structure is designed to detect this pattern rather than producing a single undifferentiated quality score.
Comparing a trial's registered protocol or published statistical analysis plan against its final results paper is one of the more effective practical checks for the selective-reporting domain specifically, since an outcome that was pre-specified but never reported, or that appears only in a secondary, less prominent position in the results, is a concrete, checkable signal rather than a matter of interpretation. Reviewers should also distinguish between a trial's risk of bias on its primary outcome and on secondary outcomes separately, since the same trial can be well-conducted for the outcome it was powered to detect while reporting a secondary outcome with much weaker methodological support.
Assessment of publication bias across the included trials
Assessment of publication bias checks a review’s pooled estimate against the possibility that trials with favourable or statistically significant results were more likely to be published, or published faster, than trials with null or unfavourable results, which would skew the included evidence base even if every individual included trial were perfectly conducted. A funnel plot, which graphs each trial's effect size against a measure of its precision, is the most common visual check, since publication bias tends to produce an asymmetric pattern, a conspicuous gap among smaller, less precise trials on the unfavourable side of the plot, whereas a symmetric, funnel-shaped scatter is more consistent with no strong selective publication. Egger's regression test formalises this visual check statistically, though it, like the funnel plot itself, works best with a reasonably large number of included trials and should be interpreted cautiously when a review includes only a handful.
Checking trial registries directly is the more direct complement to a funnel plot: searching ClinicalTrials.gov or the WHO ICTRP for registered trials on the same question that never appear to have published results lets a reviewer identify specific missing studies rather than inferring their likely existence statistically. Where a completed, registered trial cannot be found in published form, contacting the trial's listed sponsor or investigator directly for unpublished results, and reporting the attempt in the review regardless of outcome, is considered good practice because it makes the gap, and the effort to close it, visible to readers rather than silently absorbed into the review's conclusions.
Pooling of results and grading of the certainty of evidence
Whether to pool results statistically depends on clinical and methodological similarity across trials, assessed partly through the I-squared statistic and other heterogeneity measures the Cochrane Handbook describes, and partly through judgment about whether the trials are answering close enough to the same question to combine5. Once a pooled estimate (or a decision not to pool) is reached, GRADE is applied to rate the certainty of the evidence for each outcome, downgrading for risk of bias, inconsistency, indirectness, imprecision, or publication bias as warranted7. The resulting certainty rating, not just the point estimate, is what a reader should weigh when deciding how much the review's conclusion should influence practice.
A random-effects model is generally the more defensible default when any meaningful clinical or methodological diversity exists across included trials, because it allows for the possibility that the true effect differs across populations and settings, rather than assuming every trial is estimating one single fixed effect. Reporting the GRADE certainty rating alongside the pooled estimate, rather than the point estimate alone, is what lets a guideline committee or clinician distinguish a result that is precise but low-certainty, for instance because the underlying trials carried high risk of bias, from one that is both precise and high-certainty, a distinction the point estimate by itself cannot convey. The credibility of an RCT systematic review therefore rests on the complete sequence (randomisation at the trial level, RoB 2 assessment at the individual-trial level, a justified pooling decision and a GRADE certainty rating at the outcome level) rather than on any single element in isolation.
Key points
Key points
The reliability of a systematic review of RCTs rests on randomisation addressing confounding at the trial level, individual RoB 2 assessment identifying bias in trial conduct, and a justified pooling decision followed by a GRADE certainty rating, not on the number of trials alone.
Practical application
Practical application
- Eligibility criteria should be restricted to genuinely randomised trials. The randomisation method described in each candidate study should be checked rather than relying on a claim of randomisation in the title or abstract.
- RoB 2 should be applied domain by domain to every included trial before synthesis. Each of the five RoB 2 domains should be assessed separately rather than assigning a single overall quality label.
- The pooling decision should be justified with heterogeneity data before a combined estimate is reported. I-squared and the clinical rationale for combining trials should be reported, and the resulting certainty should be graded with GRADE rather than presenting the pooled estimate alone.
Frequently asked questions
It generally faces fewer confounding threats because of randomisation, but it still depends on trial conduct quality, assessed through RoB 2, and on the strength of the overall body of evidence as rated by GRADE; neither factor is guaranteed simply because trials were randomised.
A systematic review is the full structured process of identifying, appraising, and synthesising trials; a meta-analysis is the optional statistical pooling step within it, used only when trials are similar enough to combine.
It is recorded before results are known, which prevents reviewers from selecting or reframing outcomes after seeing which trials show favourable or unfavourable results, a known distortion risk in evidence synthesis.
It signals the trial's result should be weighted or interpreted cautiously in synthesis, and most current guidance recommends either excluding such trials from the primary pooled analysis or presenting a sensitivity analysis with and without them.
Yes, if the eligibility criteria are satisfied by only a small number of trials; a small number of trials limits statistical power and heterogeneity assessment but does not by itself disqualify the review.
References
- Higgins JPT, Thomas J, Chandler J, et al. (editors). Cochrane Handbook for Systematic Reviews of Interventions, version 6.4. Cochrane; 2023. read 10 October 2026. View the Cochrane Handbook
- Higgins JPT, Thomas J, Chandler J, et al. Cochrane Handbook, Chapter 1: Starting a review. read 10 October 2026. View Chapter 1
- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. read 10 October 2026. View on The BMJ
- Sterne JAC, Savović J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898. read 10 October 2026. View on The BMJ
- Deeks JJ, Higgins JPT, Altman DG (editors). Cochrane Handbook, Chapter 10: Analysing data and undertaking meta-analyses. read 10 October 2026. View Chapter 10
- Booth A, Clarke M, Dooley G, Ghersi D, Moher D, Petticrew M, Stewart L. The nuts and bolts of PROSPERO. Systematic Reviews. 2012;1:2. read 10 October 2026. View on Systematic Reviews
- Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924-926. read 10 October 2026. View on The BMJ
- Schulz KF, Altman DG, Moher D; CONSORT Group. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332. read 10 October 2026. View on The BMJ
- World Health Organization. International Clinical Trials Registry Platform (ICTRP). read 10 October 2026. View WHO ICTRP
- Shea BJ, Reeves BC, Wells G, et al. AMSTAR 2: a critical appraisal tool for systematic reviews. BMJ. 2017;358:j4008. read 10 October 2026. View on The BMJ
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 10). Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence. Pubrica. https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/
Pubrica Academy. “Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence.” Pubrica, 10 Oct. 2026, pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/.
Pubrica Academy (2026) Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence. Pubrica. Available at: https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/ (Accessed: 15 October 2026).
Pubrica Academy. “Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence.” Pubrica. October 10, 2026. https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/
Pubrica Academy, ‘Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/> accessed 15 October 2026.
Pubrica Academy. Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/
Pubrica Academy, “Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/
@misc{/la_rct_systematic_review_explained_2026,
author = {{Pubrica Academy}},
title = {Systematic Reviews of Randomised Controlled Clinical Trials: Conduct, Risk of Bias and Certainty of Evidence},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/concepts-definitions/what-exactly-is-a-randomised-controlled-clinical-trial-systematic-review/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
RCT Systematic Review Setup Checklist
Use this before beginning a systematic review restricted to randomised controlled trials.
[ ] Question framed using PICO and restricted explicitly to randomised trials [ ] Protocol registered (e.g., PROSPERO) before screening begins [ ] Search run across multiple databases plus trial registries, not a single source [ ] RoB 2 planned as the risk-of-bias tool, applied domain by domain per trial [ ] Heterogeneity assessment planned before any pooling decision is finalised [ ] GRADE certainty rating planned for each outcome once synthesis is complete
Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica’s clinical evidence synthesis team conducts systematic reviews of RCTs through the full Cochrane sequence, from PICO framing through RoB 2 assessment to GRADE certainty rating, so that the pooled conclusion is defensible under peer review.