Pragmatic Trial

What makes a trial “pragmatic”

A pragmatic trial — also called a pragmatic clinical trial (PCT) or a pragmatic randomized controlled trial — is a randomized controlled trial designed to answer a real-world, practical question — does this intervention work when delivered under the conditions clinicians and health systems actually operate in — rather than the narrower question of whether it works under ideal, tightly controlled conditions. It is still an experimental design: participants are still randomly allocated between the intervention and a comparator, which is what distinguishes it from a purely observational real-world study. What changes, relative to a conventional explanatory trial, is a cluster of protocol decisions around that randomization: who is eligible, where and by whom the intervention is delivered, how strictly the protocol is followed, what the comparator is, and what outcomes are measured.

The term sits opposite explanatory trial in a paired vocabulary that research administrators, IRBs/RECs, and funders use to describe trial intent. Neither term is a formal regulatory category the way Randomized Controlled Trial (RCT) or clinical trial phase are — they describe design orientation, not a phase or a regulatory classification, and a single trial can be pragmatic on some design dimensions and explanatory on others.

The pragmatic-explanatory continuum, not a binary label

Very few trials are purely one or the other. A foundational methodological point — formalized by the original PRECIS tool and its successor, PRECIS-2 — is that pragmatism is best understood as a continuum across several independent design domains, not a single yes/no classification for the trial as a whole. A trial can be highly pragmatic in its eligibility criteria (enrolling almost anyone who would realistically receive the intervention) while remaining explanatory in its primary outcome (a lab biomarker rather than a clinical endpoint measured in routine records).

PRECIS-2: the standard framework

PRECIS-2 (PRagmatic-Explanatory Continuum Indicator Summary-2), published by Loudon and colleagues in 2015 as a revision of the original PRECIS tool, is the widely adopted instrument for locating a trial’s design along the pragmatic-explanatory spectrum. It scores a trial across nine design domains, each rated on a 1 (very explanatory) to 5 (very pragmatic) scale:

• Eligibility criteria — how broadly or narrowly the trial defines who can enroll.
• Recruitment — how participants are identified and invited (e.g., routine clinical contact vs. dedicated study advertising).
• Setting — whether the trial runs in the same settings where the intervention would ordinarily be delivered.
• Organisation — whether the intervention is delivered using the resources and expertise normally available, or extra study-specific infrastructure.
• Flexibility: delivery — how much latitude those delivering the intervention have to adapt it, as they would in usual practice.
• Flexibility: adherence — how much latitude participants have in how closely they follow the intervention.
• Follow-up — how closely the intensity and schedule of follow-up matches usual care versus study-driven monitoring.
• Primary outcome — whether the outcome measured is directly relevant to participants and decision-makers, or a surrogate/mechanistic measure.
• Primary analysis — whether the analysis includes all participants as they were managed in practice (e.g., intention-to-treat) or is restricted to isolate an idealized effect.

Plotting the nine scores produces a “wheel” diagram: a trial with scores clustered near the center is more explanatory; one with scores near the outer edge is more pragmatic. Research teams typically apply PRECIS-2 during protocol design, as a structured way to check that specific design choices actually match the trial’s stated purpose, rather than only after the fact.

Why the distinction matters for research administration

The pragmatic/explanatory orientation has practical downstream consequences a research office will encounter directly: pragmatic trials more often use routinely collected data (electronic health records, claims, registries) rather than dedicated case report forms, which raises different data-governance and informed consent considerations (some pragmatic and cluster-randomized designs use waived or modified consent models where individual-level, protocol-driven consent would be impractical); broad eligibility and usual-care delivery affect IRB/REC risk assessment differently than a tightly controlled explanatory protocol; and pragmatic trial results are what payers, guideline bodies, and health systems tend to weight most heavily for real-world adoption decisions, since they are designed to answer “will this work as actually deployed,” not only “can this work under ideal conditions.” Funders increasingly specify pragmatic design elements explicitly in trial-focused funding calls for comparative-effectiveness and implementation research.

Pragmatic trial vs. real-world evidence vs. non-interventional study

These three terms are frequently used loosely as synonyms in practice, but they describe distinct things: a pragmatic trial is a randomized, interventional design placed toward the pragmatic end of the PRECIS-2 continuum; Real-World Evidence (RWE) is evidence about intervention use and outcomes derived from real-world data sources (which can come from a pragmatic trial, a registry, or claims data, among others); and a Non-Interventional Study involves no random assignment at all — treatment is determined by usual clinical decision-making, not a study protocol. A pragmatic trial randomizes; a non-interventional study does not. Both can generate real-world evidence, but only one is a trial.

“Pragmatic Trial” and “Pragmatic Clinical Trial” (PCT)

The terms are used interchangeably. Pragmatic clinical trial, usually abbreviated PCT, is the form preferred in US health-services research and is the term used throughout the NIH Pragmatic Trials Collaboratory; pragmatic trial is the shorter form more common in UK and European methodological writing, and is the form used in the PRECIS-2 literature. Neither denotes a distinct design — they refer to the same thing. A closely related label, pragmatic randomised controlled trial (pragmatic RCT), simply makes the randomisation explicit, which matters because it is what separates the design from an observational real-world study.

The NIH Collaboratory frames PCTs as “a promising type of trial conducted within real-world health care delivery systems,” with the “potential to inform policy and practice with high-quality evidence at reduced cost and increased efficiency compared with traditional clinical trials.” The rationale it gives for the whole enterprise is a gap in the existing evidence base: high-quality evidence “is lacking for many therapeutic areas,” and conventional trials are frequently “too small to provide sufficient statistical power,” fail to address practice priorities, or carry “shortcomings in design and execution that limit their usefulness.”

An important corollary for anyone administering one: pragmatic design is not simply a looser version of a conventional protocol. The Collaboratory notes that a PCT “often requires different approaches to study design, engagement with participants and research partners, study startup, data collection, data analysis, regulatory and ethical oversight, and dissemination than traditional explanatory trials.” The differences run through the whole lifecycle, not just the eligibility criteria.

The Methodological Trade-offs

Every design choice that moves a trial toward the pragmatic end of the continuum buys external validity with something else. The trade-offs below are the ones that recur in the methodological literature, and they are what a protocol review should be probing:

• Blinding is often impossible. The CONSORT pragmatic-trials extension states it plainly: “In pragmatic trials, as in the real world delivery of care, blinding of participants and clinicians may be impossible.” That is an honest consequence of testing an intervention as it would actually be delivered, but it reintroduces performance and detection bias, and it is why objective or routinely-recorded outcomes matter more in a pragmatic trial than in a blinded explanatory one.
• Care varies between similar participants. Again from the CONSORT extension: “Pragmatic trials are conducted in typical care settings, and so care may vary between similar participants, by chance, by practitioner preference, and according to institutional policies.” The intervention being evaluated is therefore a distribution of delivered interventions, not a single fixed one — which is the point of the design, but it complicates interpreting a null result, since it may reflect inconsistent delivery rather than an ineffective intervention.
• Contamination between arms. Where clinicians or sites treat patients in both arms, practice learned in one can leak into the other. As the NIH Collaboratory puts it, “when contamination occurs during a clinical trial, it dilutes the observed differences between comparators” — biasing toward the null, and one of the main reasons pragmatic trials often randomise clusters rather than individuals.
• Outcome ascertainment from routinely collected data. Using electronic health records instead of dedicated case report forms is what makes many PCTs affordable, and it is also their most under-appreciated weakness. The Collaboratory identifies two distinct problems: “the lack of investigator control over data collection and recording processes in health care facilities,” and the fact that “care systems are highly fragmented, and EHRs from one system will lack information about whether or not a patient had additional care outside a system’s EHR.” The second is a coverage problem rather than an accuracy problem — an outcome that occurs outside the study’s health system may simply never be observed.
• Adherence and treatment switching. Because a pragmatic trial does not enforce protocol adherence, the contrast between arms can erode substantially over follow-up. ADAPTABLE, the pragmatic aspirin-dosing trial, is the standard cautionary example: 41.6% of participants assigned the higher dose switched, against 7.1% of those assigned the lower dose, which makes the comparison the trial actually delivered quite different from the one it randomised.

None of these is a reason to avoid the design. They are reasons the design must be chosen deliberately — which is precisely what PRECIS-2 exists to force, by making each domain an explicit decision rather than a default. See CASRAI’s cluster-randomized trials, ICC and design effect guide for the sample-size consequences of randomising clusters, and intention-to-treat analysis for why the analysis population matters more, not less, when adherence is left to real-world behaviour.

Consent in Pragmatic Trials

Consent is where pragmatic design most often meets a genuine regulatory obstacle, because a low-risk intervention embedded in routine care can make individual written consent both impractical and distorting — the act of consenting changes the usual-care condition the trial is trying to measure.

Under the US Common Rule, an IRB may waive or alter consent, but only on documented findings. 45 CFR 46.116(f)(3) requires the IRB to find and document all of the following:

1. “The research involves no more than minimal risk to the subjects;”
2. “The research could not practicably be carried out without the requested waiver or alteration;”
3. “If the research involves using identifiable private information or identifiable biospecimens, the research could not practicably be carried out without using such information or biospecimens in an identifiable format;”
4. “The waiver or alteration will not adversely affect the rights and welfare of the subjects; and”
5. “Whenever appropriate, the subjects or legally authorized representatives will be provided with additional pertinent information after participation.”

Two points a research office should hold on to. First, the third criterion — the identifiability finding — was added by the 2018 revised Common Rule; before that, four criteria sat at §46.116(d). Guidance and templates written against the older numbering are still in circulation and are now incomplete. Second, a pragmatic trial using protected health information generally needs a HIPAA waiver of authorization under 45 CFR 164.512(i) as well as the Common Rule waiver; they are separate determinations. CASRAI’s waiver of informed consent guide covers the mechanics.

Cluster randomisation and the gatekeeper question

When the unit of randomisation is a clinic, hospital, or health system rather than an individual, the question of who can agree to what becomes genuinely difficult. The Ottawa Statement on the Ethical Design and Conduct of Cluster Randomized Trials is the standard reference, and its recommendations are unusually direct:

• Recommendation 4: “Researchers must obtain informed consent from human research participants in a CRT, unless a waiver of consent is granted by a REC under specific circumstances.”
• Recommendation 6: a waiver or alteration may be approved “when (1) the research is not feasible without a waiver or alteration of consent, and (2) the study interventions and data collection procedures pose no more than minimal risk.”
• Recommendation 8: “Gatekeepers should not provide proxy consent on behalf of individuals in their cluster.”
• Recommendation 9: the researcher should obtain the gatekeeper’s permission to enrol the cluster, but “such permission does not replace the need for the informed consent of research participants, when it is required.”

The distinction is sharp and worth stating in exactly these terms: gatekeeper permission enrols the cluster; it never substitutes for individual informed consent. Only a properly documented REC or IRB waiver removes the individual consent requirement.

Reporting a Pragmatic Trial: the CONSORT Extension

Pragmatic trials are reported under a dedicated extension of the CONSORT statement: Zwarenstein and colleagues, “Improving the reporting of pragmatic trials: an extension of the CONSORT statement,” BMJ 2008;337:a2390. It is worth understanding what kind of document this is, because it is commonly misdescribed. It adds no new checklist items. It extends eight existing CONSORT items so that a reader can judge whether the trial’s findings apply to their own setting — which is the entire question a pragmatic trial exists to answer.

CONSORT item	What the pragmatic extension asks for
2 — Background	“Describe the health or health service problem that the intervention is intended to address and other interventions that may commonly be aimed at this problem”
3 — Participants	“Eligibility criteria should be explicitly framed to show the degree to which they include typical participants and/or, where applicable, typical providers (eg, nurses), institutions…”
4 — Interventions	“Describe extra resources added to (or resources removed from) usual settings in order to implement intervention. Indicate if efforts were made to standardise the intervention or if the intervention and its delivery were allowed to vary…”
6 — Outcomes	“Explain why the chosen outcomes and, when relevant, the length of follow-up are considered important to those who will use the results of the trial”
7 — Sample size	“If calculated using the smallest difference considered important by the target decision maker audience (the minimally important difference) then report where this difference was obtained”
11 — Blinding	“If blinding was not done, or was not possible, explain why”
13 — Participant flow	“The number of participants or units approached to take part in the trial, the number which were eligible, and reasons for non-participation should be reported”
21 — Generalisability	“Describe key aspects of the setting which determined the trial results. Discuss possible differences in other settings where clinical traditions, health service organisation, staffing, or resources may vary…”

The 2008 extension remains the current listed guidance on the EQUATOR Network and has not been superseded. A successor is in development but not yet published: a project registered with EQUATOR in June 2026 aims to update it in line with the CONSORT 2025 statement and rename it CONSORT-PRECIS, a joint extension covering pragmatic and explanatory trials, with a companion SPIRIT-PRECIS for protocols. Until that appears, report against the 2008 extension alongside current CONSORT. See CASRAI’s CONSORT checklist guide for the base statement, and the EQUATOR Network for choosing the right reporting guideline for a given design.

PRECIS-2-PS: the Provider-Focused Successor

PRECIS-2 assumes the intervention acts on patients. Many implementation trials do not work that way — the thing being randomised is a strategy aimed at clinicians (an audit-and-feedback scheme, a decision-support prompt, a training programme), and the patient outcome is downstream of provider behaviour. PRECIS-2-PS (Norton, Loudon, Chambers and Zwarenstein, Implementation Science 2021;16:7) adapts the tool for exactly that case. It keeps the nine-domain, 1–5 structure but renames four domains to fit a provider-focused design: Organisation becomes Implementation resources, Flexibility (delivery) becomes Flexibility of provider-focused strategies, Flexibility (adherence) becomes Flexibility of intervention, and Follow-up becomes Data collection.

If you are designing a trial whose unit of intervention is a provider or a service rather than a patient, PRECIS-2-PS is the version to use. See also CASRAI’s hybrid effectiveness-implementation trial designs guide, which covers the related question of testing an intervention and its implementation strategy in the same study.

Worked Examples

Three trials are cited repeatedly in the methodological literature as reference points for what a pragmatic design looks like in practice:

• ADAPTABLE (Jones et al., NEJM 2021;384(21):1981–1990) — an open-label pragmatic comparative-effectiveness trial of aspirin 81 mg versus 325 mg for secondary prevention, run across the PCORnet network with follow-up through electronic health records and claims. 15,076 participants, median follow-up 26.2 months, no significant difference between doses. It is now cited as often for its adherence problem as for its result.
• Salford Lung Study (COPD) (Vestbo et al., NEJM 2016;375(13):1253–1260) — an open-label trial embedded in routine UK primary care using linked electronic health records, randomising 2,799 COPD patients across 75 general practices to fluticasone furoate/vilanterol or continued usual care. A standard example of a trial conducted almost entirely inside an existing care system.
• Salford Lung Study (asthma) (Woodcock et al., The Lancet 2017;390(10109):2247–2255) — the companion trial in asthma, 4,233 patients from 74 practices.

Related CASRAI terms

• Randomized Controlled Trial (RCT) — the base experimental design that pragmatic and explanatory trials are both variants of.
• Research Study Types — the broader observational-vs-experimental classification this term sits within.
• Real-World Evidence (RWE) — evidence derived from real-world data sources, which a pragmatic trial can generate.
• Non-Interventional Study — the non-randomized counterpart, distinguished from a pragmatic trial by the absence of random assignment.
• Clinical Trial Phases and What Is a Clinical Trial? (NIH Definition) — related trial-classification reference pages.
• Informed Consent — relevant to the modified consent models some pragmatic designs use.
• Cluster-Randomized Trials: ICC and Design Effect — the sample-size consequence of randomizing clinics or systems rather than individuals, which pragmatic trials frequently do to limit contamination.
• Waiver of Informed Consent — the regulatory route a low-risk pragmatic trial embedded in routine care usually depends on.
• CONSORT Checklist for RCT Reporting and The EQUATOR Network — the base reporting statement the pragmatic extension modifies, and how to pick a reporting guideline.
• Hybrid Effectiveness-Implementation Trial Designs — for trials testing an intervention and its implementation strategy together.
• Target Trial Emulation — what to do when randomization is not possible at all and the question must be answered from real-world data.
• Intention-to-Treat Analysis — why the analysis population matters more, not less, when adherence is left to real-world behaviour.
• Clinical Research — this term’s cluster hub.

References

• Loudon K, Treweek S, Sullivan F, Donnan P, Thorpe KE, Zwarenstein M. “The PRECIS-2 tool: designing trials that are fit for purpose.” BMJ, 2015;350:h2147.
• Ford I, Norrie J. “Pragmatic Trials.” New England Journal of Medicine, 2016;375:454-463.
• Thorpe KE et al. “A pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designers.” Journal of Clinical Epidemiology, 2009;62(5):464-475 (the original PRECIS tool that PRECIS-2 revised).
• Zwarenstein M, Treweek S, Gagnier JJ, Altman DG, Tunis S, Haynes B, Oxman AD, Moher D. “Improving the reporting of pragmatic trials: an extension of the CONSORT statement.” BMJ, 2008;337:a2390 — source for the eight extended checklist items and for the statements on blinding and variation in care.
• Norton WE, Loudon K, Chambers DA, Zwarenstein M. “Designing provider-focused implementation trials with purpose and intent: introducing the PRECIS-2-PS tool.” Implementation Science, 2021;16:7.
• Weijer C, Grimshaw JM, Eccles MP, McRae AD, White A, Brehaut JC, Taljaard M. “The Ottawa Statement on the Ethical Design and Conduct of Cluster Randomized Trials.” PLoS Medicine, 2012;9(11):e1001346.
• 45 CFR 46.116(f)(3) (revised Common Rule, 2018) — the five findings an IRB must document to waive or alter informed consent.
• NIH Pragmatic Trials Collaboratory, Living Textbook of Pragmatic Clinical Trials — source for the definition and framing of PCTs, contamination, and the limits of EHR-based outcome ascertainment.
• Jones WS et al. “Comparative Effectiveness of Aspirin Dosing in Cardiovascular Disease” (ADAPTABLE). New England Journal of Medicine, 2021;384(21):1981-1990.
• Vestbo J et al. “Effectiveness of Fluticasone Furoate-Vilanterol for COPD in Clinical Practice” (Salford Lung Study). New England Journal of Medicine, 2016;375(13):1253-1260.
• Woodcock A et al. “Effectiveness of fluticasone furoate plus vilanterol on asthma control in clinical practice” (Salford Lung Study, asthma). The Lancet, 2017;390(10109):2247-2255.

Frequently Asked Questions

What is a pragmatic trial?

A pragmatic trial is a randomized controlled trial designed to test whether an intervention works under the real-world conditions clinicians and health systems actually operate in, rather than under the tightly controlled conditions of a conventional explanatory trial. It still randomly allocates participants between the intervention and a comparator — what changes is a cluster of protocol choices around eligibility, setting, delivery, and outcomes.

What is the difference between a pragmatic trial and an explanatory trial?

The two terms describe opposite ends of a design continuum, not a strict either/or split. A pragmatic trial favors broad eligibility criteria, delivery in usual-care settings, flexible protocols, and outcomes that matter to patients and decision-makers, while an explanatory trial favors narrower eligibility, tightly controlled delivery, and surrogate or mechanistic outcomes. Most trials sit somewhere between the two extremes rather than being purely one or the other.

Is a pragmatic trial the same as a randomized controlled trial (RCT)?

A pragmatic trial is a type of Randomized Controlled Trial (RCT), not a separate category from it. “Pragmatic” and “explanatory” describe where a trial’s design sits on a continuum — broad or narrow eligibility, usual-care or controlled delivery, and so on — while randomization itself is the feature both pragmatic and explanatory trials share as RCTs.

What is PRECIS-2?

PRECIS-2 (PRagmatic-Explanatory Continuum Indicator Summary-2) is the standard framework for locating a trial’s design along the pragmatic-explanatory continuum. It scores a trial across nine design domains — including eligibility criteria, setting, flexibility of delivery, and primary outcome — each rated from 1 (very explanatory) to 5 (very pragmatic), rather than assigning the trial a single pragmatic-or-explanatory label.

What is a pragmatic clinical trial (PCT)?

The same thing as a pragmatic trial. “Pragmatic clinical trial,” abbreviated PCT, is the term used in US health-services research and by the NIH Pragmatic Trials Collaboratory, which describes PCTs as trials “conducted within real-world health care delivery systems” with the potential to inform policy and practice “at reduced cost and increased efficiency compared with traditional clinical trials.” “Pragmatic trial” is the shorter form more common in UK and European methodological writing. There is no design difference between the two labels.

What is a pragmatic randomized controlled trial?

A pragmatic RCT is a pragmatic trial with the randomisation made explicit in the name. It matters as a distinction because pragmatic trials are frequently confused with observational real-world studies: a pragmatic trial randomly allocates participants (or clusters) between an intervention and a comparator, while a non-interventional study does not. What makes it pragmatic is the set of design choices around that randomisation — broad eligibility, usual-care settings, flexible delivery, and outcomes that matter to decision-makers.

What are the disadvantages of a pragmatic trial?

The main ones are structural rather than incidental. Blinding is frequently impossible — the CONSORT pragmatic extension notes that “in pragmatic trials, as in the real world delivery of care, blinding of participants and clinicians may be impossible.” Care varies between similar participants “by chance, by practitioner preference, and according to institutional policies.” Contamination between arms dilutes the observed difference. Outcomes drawn from electronic health records are limited both by the investigator’s lack of control over how data is recorded and by fragmented care, since an outcome occurring outside the study’s health system may never be captured. And because adherence is not enforced, treatment switching can substantially erode the contrast that was randomised — in ADAPTABLE, 41.6% of participants assigned the higher aspirin dose switched, against 7.1% assigned the lower dose.

Do pragmatic trials need informed consent?

Usually yes, but a waiver or alteration is available and is often what makes the design feasible. Under 45 CFR 46.116(f)(3) an IRB must find and document that the research is no more than minimal risk, could not practicably be carried out without the waiver, could not practicably be carried out without identifiable information where such information is used, will not adversely affect participants’ rights and welfare, and that participants will be given additional pertinent information afterwards where appropriate. A trial using protected health information generally needs a separate HIPAA waiver of authorization as well.

Who consents in a cluster-randomized pragmatic trial?

Gatekeeper permission enrols the cluster; it does not consent the individuals in it. The Ottawa Statement is explicit that “gatekeepers should not provide proxy consent on behalf of individuals in their cluster,” and that a gatekeeper’s permission to enrol a cluster “does not replace the need for the informed consent of research participants, when it is required.” Only a research ethics committee waiver removes the individual consent requirement, and only where the research would not be feasible without it and the interventions and data collection pose no more than minimal risk.

How should a pragmatic trial be reported?

Against the CONSORT extension for pragmatic trials (Zwarenstein et al., BMJ 2008;337:a2390) alongside the base CONSORT statement. The extension adds no new checklist items; it extends eight existing ones — background, participants, interventions, outcomes, sample size, blinding, participant flow, and generalisability — so a reader can judge whether the findings apply to their own setting. A successor aligned to CONSORT 2025 and renamed CONSORT-PRECIS is registered as in development but is not yet published.

What is PRECIS-2-PS?

A version of PRECIS-2 for trials whose intervention targets providers rather than patients — implementation strategies such as audit and feedback, decision support, or clinician training. Introduced by Norton and colleagues in Implementation Science in 2021, it keeps the nine domains and the 1–5 scale but renames four of them to fit a provider-focused design: Organisation becomes Implementation resources, Flexibility (delivery) becomes Flexibility of provider-focused strategies, Flexibility (adherence) becomes Flexibility of intervention, and Follow-up becomes Data collection.

Is a pragmatic trial the same as real-world evidence?

No. A pragmatic trial is a randomized, interventional design placed toward the pragmatic end of the PRECIS-2 continuum, while Real-World Evidence (RWE) is evidence derived from real-world data sources more broadly, which can come from a pragmatic trial, a registry, or claims data. A pragmatic trial can generate real-world evidence, but not all real-world evidence comes from a randomized trial.