01 · The Question
Can neighboring evidence help when direct studies are scarce?
Sometimes the literature closest to your research question is frustratingly small, while a much larger body of research sits just outside your eligibility criteria. Perhaps studies examine a related condition rather than the exact condition of interest. They may evaluate a similar intervention with different components, intensity, or delivery. Or they may investigate essentially the same intervention in another healthcare system, educational environment, country, or organizational setting.
Ignoring all of that evidence may seem wasteful. Including it indiscriminately creates a different problem.
Evidence does not become directly relevant merely because it is the nearest evidence available. The important question is whether findings from the related condition, intervention, or context can reasonably inform the target question, and what assumptions are required to make that transfer.
03 · What You Need to Know
The key question is not whether the evidence is related, but how it is related
Almost any research question can be connected to a much larger neighboring literature if “related” is defined loosely enough. That makes relatedness a poor inclusion criterion by itself.
A stronger approach begins with the target question. Define what you want to know, then identify the specific difference between that question and the evidence you are considering. In evidence synthesis, indirectness can arise when the available evidence differs from the target question in population, intervention, comparator, or outcome. Contextual differences may operate through several of these elements.
GRADE treats indirectness as a question of applicability, generalizability, transferability, or external validity. Importantly, some mismatch is almost inevitable. The relevant concern is whether the mismatch creates a credible possibility that the effect or conclusion would differ meaningfully in the target situation.
Related conditions may share mechanisms without being interchangeable
Evidence from one condition can sometimes inform another when the conditions share mechanisms, manifestations, risk factors, treatment pathways, or other characteristics relevant to the question.
Suppose an intervention has been studied extensively for one anxiety disorder but scarcely for another. Evidence from the better-studied condition might provide information about feasibility, mechanisms, adverse effects, or perhaps intervention effects. Whether it can support conclusions about the less-studied condition depends on why the intervention is expected to behave similarly across the two conditions.
The diagnostic labels themselves do not settle the issue. Two closely named conditions may differ in ways that matter greatly for an intervention, while two differently classified conditions may share a mechanism directly relevant to how an intervention works.
Ask what causal or substantive reasoning connects them.
Related interventions may differ in the component that actually matters
Interventions often exist in families. A digital intervention may have multiple versions. A training program may vary in duration, intensity, facilitator, feedback, or instructional design. A clinical intervention may use the same underlying treatment but a different dose, route, schedule, provider, or combination of components.
These differences do not automatically make the evidence unusable. GRADE guidance on intervention indirectness specifically considers whether effects may vary according to how, where, or by whom an intervention is delivered, as well as differences in intervention components, intensity, and duration.
The practical question is whether the changed component could plausibly alter the result.
Intervention variation
The studied intervention differs from the target intervention in one or more features.
Intervention indirectness
Those differences create meaningful uncertainty about whether the observed findings apply to the target intervention.
For example, evidence about a twelve-week intensive intervention might not directly answer whether a two-session adaptation works. Likewise, evidence for a program delivered by specialists may not fully answer whether the same program will perform similarly when delivered by general practitioners, teachers, peers, or an automated platform.
Context can be part of what makes an intervention work
Context is especially easy to dismiss as background. In practice, it can alter implementation, uptake, baseline risk, available resources, comparison conditions, and sometimes the intervention itself.
Cochrane notes that contextual factors can include cultural and linguistic diversity, socioeconomic circumstances, rural or urban settings, healthcare systems, and characteristics of service delivery. Some interventions transfer successfully between settings; others depend substantially on the environment in which they operate.
This issue is particularly important for complex interventions. A teaching model evaluated in a well-resourced university with small classes and extensive faculty support may not operate identically in a resource-constrained institution with large classes. A healthcare intervention tested in specialist centers may perform differently in primary care. A policy evaluated under one regulatory environment may not transfer neatly to another.
The more context contributes to the intervention's mechanism or implementation, the less reasonable it becomes to treat context as incidental.
The comparator can change when the context changes
A subtle source of indirectness arises when “usual care,” “standard practice,” or another comparator differs across studies and settings.
An intervention compared with minimal care may show a different relative advantage than the same intervention compared with a strong contemporary alternative. GRADE therefore treats comparator mismatch as another potential source of indirectness.
This matters when borrowing evidence from another country, institution, historical period, or service environment. The intervention may look identical while the counterfactual has changed substantially.
Use a mechanism-based rationale rather than a resemblance-based rationale
A weak justification sounds like this: “The conditions are similar.”
A stronger justification identifies why the evidence should transfer. Perhaps the intervention acts through a mechanism shared by both conditions. Perhaps the altered delivery format preserves the components believed to produce the effect. Perhaps contextual differences are unlikely to influence either implementation or the relevant causal pathway.
This reasoning does not guarantee transferability, but it makes the assumption inspectable. Readers can evaluate whether the connection is persuasive rather than being asked to accept a vague claim of similarity.
Related evidence can have several roles
Including related evidence does not require treating it as direct evidence for the primary effect estimate. It can play different roles depending on the question.
| Role |
What related evidence may contribute |
Important limitation |
| Primary evidence |
Directly contributes to the main synthesis when differences are judged unlikely to materially affect the answer. |
The inclusion boundary and rationale must remain defensible. |
| Separate indirect evidence |
Provides additional information when direct evidence is limited. |
Should remain distinguishable from more direct evidence. |
| Supporting evidence |
Helps assess mechanisms, feasibility, harms, implementation, baseline risk, or plausibility. |
May not support the same claims as direct effect evidence. |
| Contextual evidence |
Helps explain how findings might operate under different conditions. |
Should not be used to imply direct effectiveness in the target setting without justification. |
This distinction can be particularly useful when only a few directly relevant studies exist. Rather than forcing neighboring studies into the core evidence base, you can determine what specific contribution they can legitimately make.
Scarcity does not lower the relevance threshold
If direct evidence is scarce, indirect evidence may become more valuable because it is the best evidence available. That does not make it more direct.
Recent Core GRADE guidance explicitly recognizes deliberate searches for indirect evidence when direct evidence is unavailable or provides low or very low certainty. The resulting evidence must still be assessed for mismatches with the target question and for the likelihood that those mismatches would produce meaningfully different effects.
This distinction protects against an easy methodological slide: first expanding to a related condition, then accepting a modified intervention, then adding another context, until the final evidence base is substantial but only faintly connected to the original question.
Watch Out
Several individually defensible expansions can accumulate. A study involving a related condition, a modified intervention, a different comparator, and a substantially different setting may be much more indirect than any one of those differences suggests when considered alone.
04 · A Practical Example
What if the intervention exists elsewhere, but not quite in your context?
Hypothetical Example
A peer-support intervention with little evidence in your target setting
Suppose you want to determine whether a structured peer-support intervention improves first-year university students' academic adjustment. Only two studies evaluate the specific program in conventional face-to-face universities. You find additional studies of similar peer-support programs in fully online universities and vocational colleges.
Identify what is shared
The programs use trained peers, scheduled contact, structured academic support, and referral pathways. These components appear central to the intervention's proposed mechanism.
Identify what differs
The student populations, institutional environments, modes of interaction, available support services, and some program components differ across settings.
Ask whether those differences matter
If peer relationship formation and access to campus services are important mechanisms, the difference between fully online and face-to-face settings could plausibly affect outcomes.
Assign the evidence an appropriate role
You retain the two directly relevant studies as the most direct evidence. Studies from related settings may provide supporting evidence about the intervention model, but you distinguish them rather than assuming identical effects.
Preserve uncertainty
You conclude that the broader evidence supports the plausibility or potential usefulness of peer-support approaches, while evidence for the specific target program and context remains limited.
The broader literature is not discarded, but neither is it allowed to erase the gap in direct evidence.
07 · A Quick Checklist
Before adding evidence from a related condition, intervention, or context
Before including related evidence, check:
Define the target condition, intervention, comparator, and context before deciding how far to broaden.
Identify exactly what differs between the target question and the related evidence.
Explain the mechanism or substantive rationale that makes transfer from the related evidence plausible.
Check whether differences in intervention components, intensity, duration, provider, or delivery could alter the result.
Examine whether contextual differences affect implementation, baseline risk, resources, participant behavior, or the comparator.
Decide whether the related evidence belongs in the main synthesis, a separate synthesis, or a supporting contextual role.
Consider the cumulative indirectness if more than one element of the target question has been broadened.
Keep conclusions about the target question proportionate to the directness of the evidence supporting them.
