Ask Mallard

Randomized controlled trial

A randomized controlled trial (RCT) assigns participants to intervention groups by chance so that, on average, measured and unmeasured baseline factors are balanced between groups.
NHMRC EVIDENCE LEVEL · 2009Level II · InterventionProvisional design classification. Study quality and certainty of findings require separate appraisal.
Why this level?Participants or clusters are randomly allocated to concurrent comparison groups. NHMRC 2009 source

Random allocation places this intervention design at Level II. Allocation concealment, adherence, analysis choices and follow-up still require separate appraisal; the design label does not certify study quality.

What question does this help answer?

RCTs are best suited to questions about the causal effect of an intervention when random assignment is feasible, ethical, and aligned with the research objective.

When to use it

Use an RCT when the investigator can assign an intervention, a credible comparison condition exists, prospective follow-up is possible, and the main goal is to estimate an intervention effect. Randomization is especially valuable when confounding would otherwise make treatment groups difficult to compare.

When not to use it

An RCT may be inappropriate when the exposure cannot ethically or practically be assigned, when the intervention is already universally implemented, when very long latency makes a trial infeasible, or when the question is primarily descriptive rather than causal.

How it works in plain language

Eligible participants are enrolled, assigned using a random process, followed over time, and compared on prespecified outcomes. Allocation concealment, blinding when possible, complete follow-up, prespecified outcomes, and intention-to-treat analysis help preserve the validity created by randomization.

What the trial's flow looks like

A randomized trial as a flow. Eligible people are enrolled before anyone knows what they will receive; only then is the concealed random allocation revealed. The two assigned groups are followed up and compared as randomized.Randomized controlled trial: enrol first, then reveal a concealed allocationEligible peopleEnrolled, before the allocation is knownConcealed random allocation revealedAssigned to the interventionAssigned to the comparisonFollowed upFollowed upOutcomes compared by the group each person was assigned toConcealment protects the moment of enrolment: nobody deciding whether to enrol a person can see what thatperson would get.
Conceptual diagram. It shows the structure of the idea, not data from any study.

What a trial's Table 1 looks like

Table 1. Baseline characteristics, by arm
Illustrative placeholder. Every value in this table is invented to show the shape of the result, and is not an estimate of anything.
CharacteristicGroup 1 (n = 250)Group 2 (n = 250)
Age, mean (SD), y64.2 (12.1)63.0 (13.1)
Female, No. (%)142 (56.8)138 (55.2)
Comorbidity index, mean (SD)2.6 (1.7)2.5 (1.9)
Insurance type, No. (%)62 (27.2)56 (25.2)
Baseline severity score, mean (SD)5.4 (2.2)5.2 (2.4)
The same table for a randomized trial, with one difference: there is no P value column. A baseline P value tests whether randomization worked, not a study hypothesis, and CONSORT discourages it (Senn 1994). Describe the arms, and give standardized differences if balance needs quantifying.

Why Mallard may recommend it

Mallard may recommend an RCT when the research question asks whether a modifiable intervention causes a change in an outcome and the study context suggests prospective assignment is feasible.

Key assumptions and design requirements

  • A truly random allocation process is used.
  • The allocation process is protected from prediction or manipulation.
  • Outcome definitions and primary analyses are prespecified.
  • Differential loss to follow-up, nonadherence, and missing outcomes are addressed.
  • Analysis respects the randomized assignment unless a different estimand is explicitly justified.

What to watch out for

Calling a study “randomized” does not make it rigorous if the sequence is predictable, allocation is not concealed, follow-up differs between groups, or outcomes are switched after results are known.

Clinical example

Question: Does a new clinician reminder reduce inappropriate antibiotic prescribing? Design: Randomize clinics or clinicians to receive the reminder or usual workflow, then compare prescribing outcomes over a prespecified follow-up period.

Recent studies that used this
Searching PubMed…
Part of the learning path Design a randomized trial (step 2 of 7).
Work it out
Related Learn pages
Sources & further reading
  1. CONSORT 2025 statement: updated guideline for reporting randomised trials (opens in a new tab) · BMJ · doi 10.1136/bmj-2024-081123
  2. SPIRIT 2025 and CONSORT 2025 published statements (opens in a new tab) · CONSORT-SPIRIT
  3. SPIRIT 2025 explanation and elaboration: updated guideline for protocols of randomised trials (opens in a new tab) · BMJ · doi 10.1136/bmj-2024-081660
  4. Overview of clinical study designs (opens in a new tab) · Clin Exp Emerg Med · doi 10.15441/ceem.23.036
CONSORT and SPIRIT are reporting/protocol guidance. They support transparent planning and reporting but do not replace design-specific methodological expertise.
Last source review: 2026-09-05
AI-generated educational content
Mallard uses reputable methodological and reporting sources, but this material may contain errors or omit important context. Verify important methodological decisions with the cited sources and, when appropriate, a qualified research methodologist or statistician.