This is a follow-up post to What 10,000 Clinical Questions Tell Us About Evidence, Practice, and Uncertainty. Evidence-based medicine promises that clinical decisions should be grounded in high-quality research. Over the past three decades, enormous effort has gone into building guidelines, systematic reviews, and trial infrastructures to make this possible.

But what does the landscape of evidence actually look like when you step away from theory and look at the questions clinicians really ask?

We recently analysed 10,000 real clinical questions submitted to AskTrip and filtered those rated as having only moderate or limited evidence. These are not obscure or academic questions – they are everyday problems arising in routine practice.

Nearly half of all questions fell into this category.

What emerged was not random noise, but a remarkably coherent map of where medical evidence runs thin.

Not ignorance – but structural uncertainty

These questions are not poorly formed. They are not “bad questions”. They are often precisely the right questions to ask.

The problem is that they sit in parts of medicine where strong evidence is structurally hard to generate.

This is not a failure of individual clinicians. It is a feature of how medical knowledge is produced.

The main themes of weak evidence

1. Chronic disease management

One of the largest clusters involves long-term conditions:

  • Diabetes
  • Heart disease
  • COPD
  • Chronic kidney disease
  • Arthritis

Typical questions are not about whether treatments work in principle, but about how best to use them in real people:

  • What is the optimal combination?
  • When should treatment be escalated or de-escalated?
  • How do we manage multiple conditions at once?

These are exactly the questions that RCTs are worst at answering. Trials usually study single diseases in isolation. Real patients rarely oblige.

2. Infection, prevention, and everyday risk

Another strong theme is prevention:

  • Recurrent infections
  • Aspiration risk
  • Falls
  • Pressure ulcers
  • Catheter care

These questions often involve modest interventions with potentially large population effects.

For example:

  • Does pulmonary rehabilitation reduce pneumonia recurrence?
  • Do certain dietary changes prevent aspiration?
  • Can simple environmental interventions reduce falls?

These are difficult to study, context-dependent, and rarely funded at scale – yet they shape huge amounts of morbidity.

3. Mental health and neurology

Mental health and neurological conditions form a disproportionate share of weak-evidence questions:

  • ADHD
  • PTSD
  • Functional neurological disorders
  • Chronic fatigue
  • Autism spectrum conditions

These areas are methodologically hard:

  • Outcomes are subjective
  • Diagnoses are heterogeneous
  • Interventions are complex and multi-component

The result is that clinicians repeatedly ask questions where guidance exists, but confidence does not.

4. Vulnerable populations

Another dominant pattern is questions about people who are routinely excluded from trials:

  • Children
  • Older adults
  • Pregnant patients
  • People with multiple comorbidities

These questions matter because they involve:

  • High uncertainty
  • High ethical stakes
  • High potential for harm

They are also exactly the patients for whom evidence is most limited.

5. Systems, not diseases

Some of the most revealing questions are not about diseases at all, but about systems:

  • When should we screen?
  • When should we stop investigating?
  • How should follow-up be structured?
  • What is worth doing routinely?

These questions expose a deeper problem: much of modern medicine is built on historical practice, professional culture, and institutional inertia rather than direct evidence of benefit.

Duplication: the sound of collective uncertainty

We also found many near-identical questions asked by different clinicians.

Not because the questions were trivial – but because the same uncertainties arise independently across contexts.

This is important.

Duplication is not redundancy. It is a signal.

It is the clinical equivalent of multiple sensors all detecting the same fault line.

A new way to think about research priorities

If you take this dataset seriously, it suggests a very different research agenda.

Not driven by:

  • The latest technology
  • The most fundable molecular target
  • The easiest trial design

But by:

  • Where clinicians repeatedly lack confidence
  • Where decisions carry high risk
  • Where patients experience the greatest burden

In other words: research priorities defined by real uncertainty, not academic fashion.

The uncomfortable implication

The most uncomfortable finding is this:

Evidence-based medicine works best in exactly the situations where it is least needed.

It works worst in:

  • Complex patients
  • Long-term care
  • Multimorbidity
  • Quality-of-life decisions
  • System-level design

These are the situations that dominate real clinical work.

AskTrip as an uncertainty engine

Systems like AskTrip do something unexpected.

They don’t just answer questions.

They reveal:

  • Where the evidence is strong
  • Where it is thin
  • And where it is largely absent

At scale, this becomes something new:

a live, evolving map of medical uncertainty.

Not a failure of medicine – but a diagnostic tool for the research system itself.

The real opportunity

If medicine is serious about “turning research into practice”, it also has to confront the inverse problem:

turning practice into research.

The 10,000 questions are not just a product.

They are a dataset that quietly answers one of the hardest questions in healthcare:

What don’t we actually know – and who is paying the price for that ignorance?

And the answer, increasingly, is:


almost everyone.