How Do You Account for Overdiagnosis in Screening Research?

What If Screening Detects Real Disease That Would Never Have Harmed the Patient?

A screening program identifies substantially more cases of disease than usual care. Many are diagnosed at an early stage, and treatment begins sooner.

At first glance, the increase in detection appears to demonstrate success. But what if some of these conditions would never have caused symptoms or affected the person's health during their lifetime?

These individuals may receive a technically correct diagnosis and undergo treatment without obtaining a corresponding clinical benefit.

This is the problem of overdiagnosis. It is especially difficult to study because researchers generally cannot know, at the time of diagnosis, which individual's condition would have remained harmless without detection.

Accounting for overdiagnosis therefore requires careful definitions, appropriate population comparisons, sufficient follow-up, and explicit assumptions about what would have happened in the absence of screening.

Estimate Overdiagnosis as a Population-Level Consequence of Screening

Understanding, Estimating, and Reporting Overdiagnosis

What Is Overdiagnosis?

Overdiagnosis is the identification of a genuine disease or abnormality that would never have become clinically consequential during the individual's lifetime without screening.

The condition may satisfy accepted diagnostic criteria and be confirmed through appropriate testing.

However, its detection does not necessarily benefit the individual because the disease would not have caused symptoms or harm before death from another cause or before the end of the person's life.

Overdiagnosis can occur when conditions progress very slowly, remain stable, regress, or are overtaken by competing mortality.

It is a counterfactual concept: the relevant question is what would have happened if the condition had not been detected through screening.

Why Is Overdiagnosis Difficult to Identify in Individual Patients?

Once a condition is detected, clinical decisions may change its subsequent course.

Treatment can prevent progression, making it difficult to determine whether the untreated condition would have caused harm.

Even when treatment is deferred, researchers may not observe the person's complete remaining lifetime or the disease's natural history without intervention.

Consequently, an individual diagnosis generally cannot be classified with certainty as overdiagnosed merely because the disease appears small, low grade, or slow growing.

Risk characteristics may help estimate the probability of progression, but they do not reveal the counterfactual outcome with certainty.

How Is Overdiagnosis Different From a False-Positive Result?

A false-positive result occurs when a test indicates the presence of the target condition even though the condition is absent according to an appropriate reference standard.

Overdiagnosis involves a genuine condition that would not have caused clinical harm without detection.

For example, a positive screening result followed by a negative confirmatory examination may represent a false positive.

By contrast, a screen-detected lesion confirmed as genuine disease may be overdiagnosed if it would never have become clinically important during the individual's lifetime.

These phenomena have different mechanisms and require different approaches to measurement.

How Is Overdiagnosis Different From Overtreatment?

Overtreatment occurs when an intervention is unnecessary or provides insufficient benefit relative to its harms for the individual receiving it.

Overdiagnosis can lead to overtreatment, but the two concepts are not identical.

An overdiagnosed condition may be managed through surveillance rather than immediate treatment, although diagnosis and monitoring can still create burdens.

Conversely, overtreatment can occur among individuals with clinically important disease if the selected intervention is unnecessarily intensive or inappropriate.

Researchers should therefore distinguish the frequency of overdiagnosis from the frequency and consequences of unnecessary treatment.

What Mechanisms Can Produce Overdiagnosis?

Several features of disease natural history can contribute to overdiagnosis.

Some abnormalities remain stable or progress so slowly that they never cause symptoms.

Others may regress spontaneously or never develop into clinically important disease.

Competing mortality can also prevent a genuine disease from becoming consequential during the person's remaining lifetime.

For example, an older individual may have a slowly progressing condition but die from an unrelated cause before the condition would have produced symptoms.

The probability of overdiagnosis therefore depends on disease progression, age, comorbidity, life expectancy, and the screening strategy.

How Does Overdiagnosis Relate to Lead-Time Bias?

Lead-time bias occurs when screening advances diagnosis without changing the timing of death or another important outcome.

Overdiagnosis occurs when the condition would never have become clinically consequential without screening.

Both can make screening appear more successful when researchers rely on survival measured from diagnosis.

However, a disease can be diagnosed earlier and eventually cause symptoms without being overdiagnosed.

Conversely, an overdiagnosed condition may never have had a symptom-based diagnosis date at all.

The distinction matters when researchers attempt to estimate excess diagnoses attributable to screening.

How Does Overdiagnosis Relate to Length Bias?

Length bias occurs because screening disproportionately detects conditions with longer detectable preclinical periods.

Overdiagnosed conditions may remain detectable for extended periods, making them particularly likely to be identified through screening.

However, not every slowly progressing condition is overdiagnosed.

Some will eventually cause symptoms or serious complications if untreated.

Researchers should distinguish length bias in screening studies from the counterfactual definition of overdiagnosis.

Why Can Screening Increase Disease Incidence?

Screening can identify conditions that were already present but had not yet been recognized clinically.

When screening is introduced, recorded incidence may increase because diagnoses are brought forward.

Some of these cases may represent disease that would eventually have been diagnosed without screening.

Others may represent overdiagnosis.

Researchers should therefore distinguish temporary excess incidence caused by earlier diagnosis from persistent excess incidence potentially attributable to overdiagnosis.

Changes in diagnostic criteria, population risk, and disease registration practices may also affect incidence trends.

What Is the Excess Incidence Method?

The excess incidence method compares cumulative disease diagnoses in populations exposed to different screening strategies.

Screening may initially increase diagnoses because disease is detected earlier.

If these additional diagnoses merely represent cases brought forward in time, some compensating reduction in diagnoses may occur later in the screened population.

Persistent excess cumulative incidence after sufficient follow-up may provide evidence of overdiagnosis.

However, the interpretation requires comparable populations, consistent outcome definitions, appropriate follow-up, and consideration of background incidence and other influences.

Excess incidence should not automatically be equated with overdiagnosis without evaluating these assumptions.

How Can Overdiagnosis Be Estimated From Cumulative Incidence?

The formula provides a descriptive starting point. A defensible overdiagnosis estimate requires additional methodological justification.

Why Does Follow-Up Duration Matter?

Screening can advance diagnoses that would otherwise occur later.

Immediately after screening begins, the screened group may therefore show excess incidence even if no overdiagnosis occurs.

With sufficient subsequent follow-up, some of these early diagnoses may be offset by fewer later diagnoses.

If follow-up ends too soon, researchers may incorrectly classify all early excess diagnoses as overdiagnosis.

However, the follow-up needed depends on disease natural history, screening duration, age, and competing mortality.

Researchers should justify the observation period and examine whether cumulative incidence differences stabilize.

Why Does the Denominator Matter?

Overdiagnosis percentages can differ substantially depending on the denominator used.

Researchers might express excess diagnoses as a proportion of all screen-detected cases, all diagnoses in the screened population, or diagnoses expected without screening.

These quantities answer different questions.

For example, the proportion of screen-detected cases that are overdiagnosed is not the same as the proportion of all diagnosed cases attributable to overdiagnosis.

Researchers should report the numerator, denominator, time horizon, and population explicitly.

Comparisons across studies can be misleading when their overdiagnosis definitions differ.

How Do Common Overdiagnosis Denominators Compare?

Absolute excess diagnoses per 1,000 individuals offered screening can sometimes communicate population burden more clearly than percentages alone.

How Does Competing Mortality Affect Overdiagnosis?

Competing mortality occurs when death from another cause prevents a disease from becoming clinically apparent.

A slowly progressing condition may have caused symptoms eventually if the person had lived long enough, but not during the person's actual remaining lifetime.

Such a condition may qualify as overdiagnosed under the lifetime counterfactual definition.

Consequently, overdiagnosis risk can vary with age, comorbidities, and life expectancy.

Researchers should account for competing mortality when modeling disease progression and estimating the probability that screen-detected conditions would otherwise become symptomatic.

Can Randomized Screening Trials Estimate Overdiagnosis?

Randomized trials comparing screening with an appropriate alternative can provide valuable evidence about excess cumulative diagnoses.

Randomization helps reduce baseline differences between groups in expectation.

However, estimation still requires adequate follow-up, reliable disease registration, and attention to screening received outside the assigned strategy.

Overdiagnosis estimates may also depend on whether screening continues throughout follow-up and whether the comparison group later receives screening.

Researchers should distinguish the effect of assignment to screening from the effect of actual screening participation.

Can Observational Studies Estimate Overdiagnosis?

Observational studies may examine changes in incidence following screening introduction or compare populations with different screening exposure.

However, differences in disease risk, diagnostic practice, healthcare access, and population characteristics can complicate interpretation.

For example, an increase in recorded disease incidence may reflect expanded diagnostic criteria or improved case registration rather than overdiagnosis alone.

Researchers should use appropriate comparison populations, standardization, and sensitivity analyses.

Unmeasured confounding and incomplete information may remain important limitations.

What Is the Role of Natural-History Modeling?

Natural-history models represent transitions between disease onset, detectable preclinical states, clinical diagnosis, and death.

They may incorporate screening sensitivity, progression rates, regression, treatment, and competing mortality.

Such models can estimate counterfactual outcomes that cannot be directly observed.

However, estimates depend on assumptions about progression and the distributions of relevant disease states.

Researchers should use empirical evidence where available and examine alternative plausible parameter values.

Model-based overdiagnosis estimates should be reported with uncertainty and should not be presented as directly observed individual outcomes.

Can Overdiagnosis Be Distinguished From Overtreatment in Data?

Researchers can often observe whether individuals received treatment, but they generally cannot directly observe whether an untreated condition would have caused harm.

Therefore, treatment frequency alone cannot establish the extent of overdiagnosis.

Similarly, an increase in surgery following screening may reflect treatment of clinically important disease, overtreatment, or both.

Researchers should report treatment patterns and adverse consequences separately from estimated overdiagnosis.

These outcomes contribute to the benefit-harm assessment but should not be collapsed into one measure.

Can Active Surveillance Reduce Harms From Overdiagnosis?

Active surveillance may reduce immediate treatment-related harms for selected low-risk conditions.

However, surveillance itself can create monitoring burden, anxiety, additional testing, and potential complications.

It also requires reliable follow-up and appropriate criteria for initiating treatment.

Active surveillance does not eliminate the original diagnosis or necessarily remove the consequences of labeling a person with disease.

Researchers should evaluate whether surveillance strategies improve the overall balance of benefits and harms for the relevant condition.

How Does Overdiagnosis Affect Screening Policy?

Screening policy should consider whether earlier detection produces meaningful benefits that justify additional diagnoses and their consequences.

A screening program may reduce disease-specific mortality while also causing overdiagnosis.

The presence of overdiagnosis does not automatically mean screening should be abandoned.

Instead, policymakers need estimates of absolute benefits, overdiagnosis, false positives, treatment complications, resource requirements, and population preferences.

The balance may differ across age groups, baseline risks, and screening intervals.

Researchers should present these trade-offs without assuming that detection itself is necessarily beneficial.

How Should Overdiagnosis Be Communicated?

Overdiagnosis is a difficult concept because a diagnosis can be medically correct while offering no benefit from detection.

Researchers should explain this distinction clearly and avoid implying that every screen-detected condition is harmless.

Absolute frequencies, such as estimated excess diagnoses per 1,000 people offered screening, may help communicate the magnitude of the problem.

Percentages should include explicit denominators.

Uncertainty should be acknowledged, particularly when estimates depend on natural-history models or incomplete follow-up.

Communication should distinguish overdiagnosis from false positives and overtreatment.

How Should Overdiagnosis Be Reported in Research?

Researchers should define overdiagnosis conceptually and specify the operational method used to estimate it.

They should report the screening population, comparison strategy, observation period, cumulative incidence, and relevant competing events.

The numerator and denominator of percentage estimates should be stated explicitly.

Assumptions concerning lead time, disease progression, background incidence, and follow-up should be documented.

Sensitivity analyses should examine how conclusions change under plausible alternative assumptions.

Overdiagnosis estimates should be interpreted alongside other screening harms and detection outcomes.

Why Excess Diagnoses Require Careful Interpretation Before Being Called Overdiagnosis

The example illustrates why excess diagnoses are evidence requiring interpretation rather than automatic proof of overdiagnosis.

Overdiagnosis is fundamentally counterfactual, and its estimation depends on what would have happened without screening.

Common Misconceptions About Overdiagnosis

How to Account for Overdiagnosis in Screening Research

Begin by defining the target condition and the counterfactual meaning of overdiagnosis.

Determine whether the study can compare cumulative diagnoses under different screening strategies and whether follow-up is sufficient to distinguish earlier diagnosis from persistent excess incidence.

Report overdiagnosis alongside other screening outcomes, including false-positive results, treatment-related harms, and demonstrated benefits.

When estimates depend on uncertain assumptions, present sensitivity analyses rather than a single apparently definitive percentage.

Overdiagnosis should be evaluated as part of the screening program's overall consequences, with particular attention to the counterfactual outcomes that cannot be directly observed.

Before Reporting an Overdiagnosis Estimate

Frequently Asked Questions About Overdiagnosis in Screening Research

Overdiagnosis Is Real Disease Detection Without Corresponding Clinical Benefit

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.