In one part of the world, we have become so sophisticated at detecting disease that we are increasingly finding abnormalities in people who may never become sick.

In another, people are still reaching healthcare systems with diseases that should have been identified months, or years earlier.

In some parts of the world, we are discovering disease before it matters.
In others, we are discovering it when it may already be too late.

And between them lies one of the most important, and least discussed, inequalities in global healthcare:

The Diagnosis Divide.

For decades, medicine has pursued an apparently unquestionable objective:

Find disease earlier.

• Earlier cancer.
• Earlier cardiovascular disease.
• Earlier autoimmune disease.
• Earlier mental illness.
• Earlier everything.

And there is no question that early diagnosis has saved millions of lives.

But the diagnostic revolution has produced a paradox.

Our ability to find abnormalities is beginning to exceed our ability to determine which abnormalities actually matter.

Meanwhile, hundreds of millions of people still don’t have adequate access to the most basic diagnostic infrastructure.

The result is a world in which your geography increasingly influences not simply whether you are diagnosed, but which diagnostic mistake medicine is most likely to make with you.

WHEN FINDING MORE DISEASE DOESN’T MEAN SAVING MORE LIVES

Consider thyroid cancer.

Between 2013 and 2017, approximately 2.3 million thyroid cancers were diagnosed across 63 countries examined in a major international study.

Researchers estimated that more than 1.7 million of them may have represented overdiagnosis.

Think about that number.

1.7 million diagnoses.

Real abnormalities. Real doctors. Real patients.

But cancers that, according to the analysis, may never have caused symptoms or death during those people’s lifetimes.

South Korea provides perhaps the most extraordinary example.

In 2011, the country’s thyroid cancer diagnosis rate was approximately 15 times higher than it had been in 1993, following widespread ultrasound screening.

Yet thyroid cancer mortality remained essentially stable.

Medicine had discovered dramatically more cancer.

It had not produced a corresponding reduction in deaths.

That distinction is fundamental.

Because finding more disease and preventing more death are not necessarily the same thing.

BREAST AND PROSTATE CANCER TELL A SIMILAR STORY

The problem extends beyond thyroid cancer.

Estimates of breast cancer overdiagnosis vary considerably depending on methodology and population, but the US National Cancer Institute notes estimates suggesting 20–50% of screen-detected breast cancers may represent overdiagnosis under certain assumptions.

Prostate screening creates a similar dilemma.

The US Preventive Services Task Force has cited trial estimates ranging from approximately 21% to 50% of screen-detected prostate cancers being overdiagnosed, with modelling studies producing estimates of approximately 22–42%.

These are not arguments against cancer screening.

Screening saves lives.

They reveal something much more complicated:

Not every cancer we can find is a cancer that would have killed us.

And medicine cannot always tell which is which.

So we treat.

We biopsy.

We operate.

We irradiate.

We medicate.

Because once an abnormality has been discovered, doing nothing becomes psychologically, and medically, extraordinarily difficult.

SOMETIMES WE AREN’T FINDING DISEASE. WE ARE FINDING BIOLOGICAL NOISE.

Autoimmune medicine presents another version of the same problem.

Up to 15% of healthy individuals may test positive for antinuclear antibodies, ANA, without ever developing clinical autoimmune disease.

Yet increasingly sophisticated laboratory panels can identify more and more biological abnormalities.

A number on a laboratory report becomes highlighted.

The patient searches it online.

Another test follows.

Then another.

Perhaps a specialist referral.

Perhaps repeated monitoring.

And suddenly someone who felt healthy has entered the healthcare system as a patient.

This is one of the fundamental challenges of modern diagnostics:

The human body is full of abnormalities.

Not all abnormalities are disease.

And not all disease requires intervention.

The better we become at looking, the more abnormalities we will inevitably find

MENTAL HEALTH MAY BE THE MOST COMPLEX FRONTIER OF ALL

Mental health diagnosis has expanded dramatically.

In the United States, approximately 11.4% of children have been diagnosed with ADHD.

Around one in five children has received a mental, emotional or behavioural diagnosis.

In the United Kingdom, antidepressant prescribing has risen to more than 85 million prescriptions annually, according to the data cited in the research underlying this article.

Across OECD countries, antidepressant consumption has risen substantially since 2000.

Some of this represents enormous progress.

People who once suffered silently are finally receiving recognition and treatment.

Stigma is falling.

Access is improving.

But it has also generated a difficult question:

Where does human distress end and medical disorder begin?

Are we getting better at recognising illness?

Almost certainly.

Are we sometimes medicalising normal variations in emotion, behaviour and response to difficult circumstances?

That question deserves to be asked too.

And then comes the extraordinary contradiction.

BECAUSE ON THE OTHER SIDE OF THE WORLD, THE PROBLEM IS ALMOST THE REVERSE.

While some healthcare systems debate whether diagnostic thresholds have become too sensitive, other systems cannot diagnose serious disease early enough.

Cancer makes the divide brutally visible.

WHO has long warned that cancer in resource-poor settings is frequently diagnosed at later stages, contributing to lower survival.

Historically, WHO reported that less than 30% of low-income countries had generally accessible cancer diagnosis and treatment services.

For pathology, the foundation on which cancer diagnosis depends, the divide was extraordinary:

Approximately 35% of low-income countries reported generally available public-sector pathology services in 2015.

Compare that with more than 95% of high-income countries.

35%.

Versus 95%.

That is not a small healthcare disparity.

That is a completely different medical reality.

THE COST OF BEING DIAGNOSED LATE

Early diagnosis isn’t simply medically preferable.

It can radically change the complexity and economics of treatment.

WHO has cited studies in high-income countries suggesting that treating cancers diagnosed early can be approximately 2–4 times less expensive than treating cancers diagnosed at more advanced stages.

And the human difference is considerably greater.

A breast cancer discovered through screening may require relatively localised treatment.

The same biological disease discovered after metastasis can become a life-limiting systemic illness.

One patient is discussing whether an abnormality requires treatment.

Another is discussing whether treatment can still save her life.

That is the Diagnosis Divide.

MENTAL HEALTH REVEALS AN EVEN MORE EXTRAORDINARY CONTRAST

Remember the debate in wealthy countries about expanding psychiatric diagnosis?

Now consider this:

WHO reported in 2025 that more than one billion people worldwide live with a mental health condition.

Yet access to treatment remains profoundly unequal.

Among people affected by psychosis, fewer than 10% receive care in low-income countries.

In higher-income countries, the figure is over 50%.

Globally, WHO says more than two out of three people with psychosis do not receive specialist mental healthcare.

So we have created an extraordinary simultaneous reality.

In one healthcare system, the debate may be:

Are we diagnosing too many mild conditions?

Somewhere else, the question is:

Can someone with severe psychosis access treatment at all?

Both can be true.

At exactly the same moment.

YOUR POSTCODE HAS BECOME A CLINICAL VARIABLE

We like to think diagnosis is biological.

You either have disease or you don’t.

But diagnosis is not determined by biology alone.

It is also determined by whether you have access to:

MRI. CT. Mammography. Pathology. Genomic testing. Specialists. Primary care. Screening programmes. Insurance. Mental healthcare. And increasingly, AI.

Two biologically similar human beings can therefore enter entirely different medical universes simply because they were born in different places—or even because they live on different sides of the same city.

One may undergo so much testing that medicine discovers something that never needed discovering.

The other may wait until symptoms become impossible to ignore.

One risks too much medicine.

The other risks too little medicine, too late.

And importantly, this isn’t a clean rich-country/poor-country divide.

The original research also points to increasing incidental autoimmune findings associated with expanding private diagnostic testing in countries including India, Thailand and Indonesia.

Thyroid overdiagnosis is also expanding into countries undergoing economic transition.

And within wealthy countries, disadvantaged populations can still experience delayed diagnosis and inadequate access.

So perhaps the real divide is not between countries.

It is between people with too much access to diagnostics and people with too little access to meaningful diagnosis.

NOW ADD AI.

This is where the story becomes even more consequential.

Medicine is entering an era of:

AI-assisted imaging.

Liquid biopsies.

Whole-genome sequencing.

Polygenic risk scores.

Continuous glucose monitoring.

Smart watches.

Smart rings.

Remote monitoring.

Digital biomarkers.

Predictive algorithms.

And increasingly sensitive blood tests capable of identifying disease before symptoms appear.

The amount of health information we can generate about a single human being is becoming almost limitless.

That could be extraordinary.

AI could bring specialist-level diagnostic support to communities that have never had specialists.

Digital pathology could extend expertise across borders.

Algorithms could identify dangerous disease earlier.

Low-cost diagnostics could leapfrog infrastructure limitations in emerging economies much as mobile phones leapfrogged fixed telecommunications.

AI could help close the Diagnosis Divide.

But there is another possibility.

In already diagnostic-rich societies, AI may identify millions more abnormalities.

More nodules.

More lesions.

More rhythm irregularities.

More genetic risks.

More biochemical deviations.

More alerts.

More people being told:

“We found something.”

The most important question in medicine may therefore cease to be:

Can we detect it?

And become:

SHOULD WE?

THE NEXT MEDICAL REVOLUTION IS NOT MORE DIAGNOSIS

For a century, progress in medicine largely meant seeing more.

Finding more.

Testing more.

Diagnosing earlier.

And that revolution saved countless lives.

But the next revolution requires something more intelligent.

Not simply more diagnosis.

Better diagnosis.

Finding the right disease.

In the right person.

At the right moment.

And knowing whether finding it will actually improve that person’s life.

In diagnostic-rich healthcare systems, that may mean restraint.

Better risk stratification.

Fewer unnecessary investigations.

More sophisticated interpretation of uncertainty.

And accepting that sometimes the most intelligent medical intervention is not to intervene.

In underserved healthcare systems, it may require precisely the opposite.

More diagnostics.

More pathology.

More imaging.

More screening where evidence supports it.

More clinicians.

Better referral systems.

And dramatically earlier access to care.

So the future of healthcare cannot simply be:

MORE MEDICINE.

Nor can it be:

LESS MEDICINE.

It has to become:

THE RIGHT MEDICINE.

Because perhaps one of the greatest failures of modern healthcare isn’t that we diagnose too much.

Or that we diagnose too little.

It is that we are doing both at the same time.

And increasingly, where you live helps determine which mistake medicine makes with you.