How Do You Synthesize Evidence Across High-Resource and Low-Resource Settings?

Can evidence from very different resource settings answer the same question?

An intervention may be evaluated in a specialist hospital with abundant personnel, equipment, and follow-up support, then studied elsewhere where staffing is limited and essential infrastructure is intermittent. An educational program developed for schools with individual devices may later be evaluated where students share equipment or have unreliable internet access.

These studies may nominally evaluate the same intervention. Yet what can actually be delivered, what participants receive, and what counts as the usual alternative may differ considerably.

Should evidence from high-resource and low-resource settings therefore be separated? Not automatically. The important question is whether resource differences change something that matters to the finding.

Resource differences matter when they change how the phenomenon works

Resource setting is more than a country classification

Researchers sometimes use national income categories as shorthand for resource availability. Such categories may be useful for describing an evidence base, but they are relatively coarse proxies for the conditions under which a particular intervention or phenomenon operates.

Cochrane specifically recommends considering whether interventions function differently across high-, middle-, and low-income country settings because health systems may differ in financing, regulation, organization, care delivery, economic conditions, geography, and the relative importance of health problems. But the methodological question remains contextual: which of those differences actually matter to the intervention or outcome?

A well-equipped urban facility in a lower-income country may have more relevant resources for a particular intervention than an underserved facility in a wealthier country. National classification alone therefore cannot substitute for examining the setting itself.

Ask what the intervention actually requires

Begin by identifying the resources necessary for the intervention to operate as intended. These might include trained personnel, equipment, electricity, internet connectivity, medication supply, diagnostic capacity, transportation, administrative systems, supervision, specialist referral, or continuing technical support.

The relevant requirements will differ dramatically across interventions. A resource difference that is crucial for one research question may be almost irrelevant for another.

This is particularly important for complex interventions because context, implementation, and mechanisms of action can be intertwined. Cochrane notes that interventions may be only partially implemented in some contexts and that characteristics of organizational, geographical, cultural, and legal settings can influence how interventions operate.

The intervention may have the same name but not the same implementation

Suppose studies in several settings evaluate “telemedicine.” In one setting, participants receive high-speed video consultations with specialists, integrated electronic records, remote diagnostic equipment, and rapid referral. Elsewhere, the intervention consists primarily of mobile-phone consultations because bandwidth and specialist availability are limited.

Calling both interventions telemedicine is reasonable at a broad level, but it does not establish that they represent interchangeable implementations.

When resources alter intervention intensity, fidelity, reach, or supporting components, pooling results without preserving those differences may obscure what was actually evaluated.

The comparator can change with the resource setting

Resource differences can affect the control condition as much as the intervention.

“Usual care” in a highly resourced health system might include regular specialist consultation, diagnostic testing, medication, and structured follow-up. In another setting, usual care might involve limited access to any of those services. The incremental benefit of a new intervention can therefore differ even if the intervention itself remains unchanged.

GRADE explicitly treats differences between the study and target intervention or comparator as potential sources of indirectness when those differences are likely to produce substantially different effects.

Baseline conditions may change absolute effects

Populations in different resource settings may have different baseline risks. Disease burden, educational opportunity, access to preventive services, exposure to environmental hazards, or existing service coverage may vary substantially.

Even when a relative effect remains similar, a different baseline risk can produce different absolute benefits or harms. A 20% relative reduction means something different when an outcome occurs in 5% of the target population than when it occurs in 50%.

That does not necessarily mean the intervention mechanism changed. It means resource-related context can change the practical consequences of the same relative effect.

Resource constraints can be part of the causal pathway

Sometimes resources are not merely background characteristics. They determine whether the mechanism required for an intervention can occur.

Consider a screening program whose benefit depends on patients receiving confirmatory testing and treatment after a positive result. Expanding screening without the downstream capacity to diagnose or treat may produce a very different outcome from implementing the same screening program within a system that can complete the care pathway.

In this situation, resource capacity is not statistical clutter around the intervention. It is part of the explanation for how the intervention produces its effects.

Do not assume that lower resources always mean smaller effects

The direction of contextual modification cannot safely be predicted from a simple resource hierarchy.

An intervention might produce a larger absolute benefit in a resource-constrained setting because baseline need is greater or existing services are limited. Alternatively, its effectiveness might decrease because implementation requirements cannot be met. Some interventions may require fewer resources than the service they replace and could therefore be especially useful in constrained environments.

The appropriate inference depends on the intervention and the setting, not on a general assumption that more resources inevitably produce better effects.

Separate statistical heterogeneity from contextual plausibility

A large heterogeneity statistic does not establish that resources caused differences between studies. Conversely, low statistical heterogeneity does not demonstrate that resource conditions are irrelevant.

Study design, participant characteristics, measurement, implementation, risk of bias, and sampling variation may all contribute to differences. Resource setting should be treated as an explanatory factor only when there is a defensible pathway connecting it to the finding.

This follows the broader principle that differences between settings become findings rather than noise when credible contextual characteristics help explain them.

Applicability may matter even when effects are consistent

Suppose studies from highly resourced specialist centers consistently show a beneficial effect. A decision maker in a setting without the required specialists may reasonably ask whether those findings apply locally.

Cochrane identifies precisely this kind of issue when discussing applicability: an intervention implemented by highly trained specialists in specialist centers may provide indirect evidence for settings where those conditions are absent. GRADE similarly focuses on whether differences between the study PICO and target PICO create a substantial likelihood that effects will differ.

The evidence can therefore be internally convincing while remaining indirect for a particular target setting.

Resource diversity can also strengthen the evidence

If an intervention produces comparable effects across settings with substantially different infrastructure, service organization, staffing, or delivery conditions, that pattern can be informative.

It does not prove universal transferability. Unstudied conditions may still matter. But consistent findings across genuinely different settings may reduce concern that the observed effect depends entirely on one unusually favorable environment.

When infrastructure changes what the intervention actually delivers

The defensible conclusion is not that the intervention “works in rich settings but fails in poor ones.” The evidence instead suggests that reliable access may be necessary for participants to receive enough of the intervention for its intended mechanism to operate. That explanation is both more specific and more transferable.

Common mistakes when interpreting evidence across resource settings

Identify the resources the finding actually depends on

Do not begin by dividing studies into “high-resource” and “low-resource” piles. First identify what resources or system capacities could plausibly alter the phenomenon. Then examine whether those conditions vary across the included studies.

If the target population or setting differs substantially from the evidence base, the next question is whether evidence from another population is directly relevant to yours. Resource similarity may matter more to that judgment than geographical proximity.

Before synthesizing evidence across resource settings, check:

Questions about synthesizing evidence across resource settings

Resource setting matters when resources change what can happen

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.