Could Different Baseline Risks Explain Why Studies Disagree?

Did the Participants Have the Same Chance of the Outcome to Begin With?

Suppose one study reports that an intervention prevents many adverse outcomes while another finds only a small absolute benefit. At first glance, their conclusions may seem inconsistent. Yet the participants in the first study might have been far more likely to experience the outcome without the intervention.

That starting probability is often described as baseline risk or comparator risk. It matters because the same relative effect can translate into very different absolute effects when applied to populations facing different underlying risks.

Before deciding that two studies genuinely disagree, ask whether their participants started from comparable levels of risk.

Yes, Baseline Risk Can Change the Absolute Result

How Baseline Risk Changes the Meaning of an Effect

Baseline risk is the risk without the intervention of interest

In a comparative intervention study, baseline risk usually refers to the probability of an outcome in the absence of the experimental intervention, often represented by the risk observed in the comparator or control group over a specified period.

It should not be confused with the percentage of participants who already had a condition when the study began. The relevant baseline risk is usually the expected risk of the outcome being analyzed under the comparator condition.

Relative and absolute effects answer different questions

A relative risk describes the intervention group's risk relative to the comparator group's risk. An absolute risk difference describes how many more or fewer events actually occur between the groups.

These quantities are related, but they are not interchangeable. If an intervention reduces risk by the same relative proportion in two populations, the number of events prevented depends heavily on how common the outcome would otherwise have been.

The relative effect is identical in this hypothetical calculation: a risk ratio of 0.75, corresponding to a 25% relative reduction. The absolute consequence is not. One population experiences 5 fewer events per 100 people, while the other experiences 1 fewer event per 100.

Cochrane specifically emphasizes this relationship when interpreting dichotomous outcomes. Its guidance recommends considering different comparator risks when translating relative effects into absolute risk differences or numbers needed to treat.

Different absolute effects do not necessarily mean different relative effects

This distinction can resolve some apparent disagreements immediately. Imagine that Study A reports a 6-percentage-point reduction in an adverse outcome while Study B reports only a 1-percentage-point reduction. Those numbers look quite different.

But if the comparator risks were also very different, the relative effects could be similar. Saying that the intervention “worked much better” in Study A based solely on the absolute differences would therefore be premature.

Conversely, similar absolute effects do not guarantee similar relative effects. You need to inspect the effect measures rather than infer one from the other.

Why baseline risks differ across studies

Baseline risk can vary because studies enroll participants with different prognoses, disease severity, ages, prior exposures, socioeconomic circumstances, comorbidities, or other risk factors. Settings can also matter. A study conducted where the underlying event is common may have a different comparator risk from one conducted where it is rare.

Follow-up duration matters as well. For many cumulative outcomes, participants followed for longer periods have more opportunity to experience an event. When comparing risks, therefore, the relevant time horizon should also be checked. Different follow-up times can themselves contribute to apparently different results.

Baseline risk can affect practical importance even when relative effects remain similar

Suppose an intervention reduces relative risk by 20% in both low- and high-risk groups. For someone whose underlying risk is extremely low, the absolute reduction may also be very small. For someone facing a much higher probability of the outcome, the same proportional reduction can prevent substantially more events.

This is one reason clinical and policy decisions often require absolute effects rather than relative effects alone. A relative reduction describes proportional change; it does not by itself tell you how many events are likely to be prevented.

Relative effects are not guaranteed to remain constant across baseline risk

The simple calculation above assumes that the relative effect applies similarly across risk groups. That assumption can be useful, and Cochrane notes that relative effects often tend to be more stable across risk groups than risk differences. But it is not a law of nature.

The intervention itself may genuinely work differently in people with different prognoses. Baseline risk can also be associated with characteristics that modify the treatment effect. When that happens, differences between studies cannot be explained merely by applying the same relative effect to different starting risks.

Baseline risk can be informative rather than merely inconvenient

If an intervention produces meaningful absolute benefit primarily among people at high baseline risk, that pattern may help identify who stands to benefit most. What looks like heterogeneity may therefore have direct implications for targeting interventions and communicating expected benefit.

The key is to establish whether the observed pattern is credible rather than constructing a post hoc explanation from a few dissimilar studies.

Same Relative Effect, Very Different Absolute Benefit

If one paper emphasizes the six-percentage-point reduction and another emphasizes the one-percentage-point reduction, a casual reading might suggest conflicting intervention effectiveness. Once baseline risk is considered, however, the two results could be entirely compatible with the same proportional effect.

Common Mistakes When Interpreting Different Baseline Risks

How to Check Whether Baseline Risk Explains the Disagreement

When two studies report different magnitudes of benefit or harm, first determine what effect measure you are comparing. A risk ratio, odds ratio, hazard ratio, and risk difference do not express the same quantity. Then examine the comparator event rates over comparable periods.

Also ask why baseline risk differs. If one study enrolled people with more severe disease, older participants, higher-risk institutions, or substantially different environmental exposures, the risk difference may be telling you something important about the populations themselves. That connects directly to whether different populations explain why the studies disagree.

When synthesizing the literature, avoid collapsing everything into the statement that “results were inconsistent.” A more informative account might explain that relative effects were broadly similar but absolute benefits were larger in populations with higher comparator risks. If the relative effects also vary, say so rather than attributing all heterogeneity to baseline risk.

This distinction can help determine whether apparently inconsistent evidence is actually reflecting a more complex pattern. Sometimes the studies disagree less about what an intervention does than about how much practical difference that effect makes in populations starting at different levels of risk.

What to Compare When Baseline Risks Differ

Questions About Baseline Risk and Conflicting Study Results

Start With the Risk the Population Already Faces

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.