More than 3 billion people live with a neurological condition worldwide, and these diseases cause more than 11 million deaths a year. Access and diagnosis, however, remain uneven: in an international study of 1,203 patients consulting for headache, 1 patient in 4 who met the criteria for migraine had not been given that diagnosis.
If neurological symptoms persist without a clear explanation, a second opinion can help confirm the diagnosis and decide quickly which tests are genuinely worth doing. Noria Health Hub can arrange a full check-up or a second medical opinion in Brussels, with a report within 24 h, coordinate your care where needed across more than 40 medical pathways in 48 h, then, if your medical condition allows, arrange surgery within 7 days.
A global need, with very uneven access to neurologists
According to the WHO, more than 40% of the world population lives with a neurological condition and these diseases cause more than 11 million deaths a year. Low-income countries have up to 80 times fewer neurologists than high-income countries. Poor access explains part of the delay, but even when a patient is seen, common diagnoses such as migraine can still be missed.
Diagnostic error in neurology, what is actually known
One limit has to be set at the outset: there is no reliable overall rate of diagnostic error in neurology. The available studies cover either the whole of medicine or one specific disease. No single percentage of neurological diagnoses can therefore be declared wrong.
What is known is that across 11,592 diagnostic error files drawn from 55,377 closed claims, three families account for 74.1% of serious harm, cancers, vascular events and infections. Stroke leads the vascular group, and it is missed in 17.5% of the cases in which it occurs.
The presenting symptom strongly shapes the risk of error
This is the most useful result in the file. A meta-analysis of 23 studies and 15,721 patients measures the performance of cerebrovascular event diagnosis in emergency departments.
| The presenting complaint | Share missed at first presentation |
|---|---|
| Motor deficit | 4.4% |
| Dizziness | 39.4% |
| Subarachnoid haemorrhage with normal consciousness | 23.8% |
| Transient ischaemic attack | False result rate of 59.7%, against 11.7% for completed stroke |
Overall sensitivity of 91.3%, about 9% of events missed. But the average hides everything, it is the clinical presentation that drives the risk of error.
Multiple sclerosis, over-diagnosis and under-diagnosis both exist
| What is measured | The figure |
|---|---|
| Multiple sclerosis diagnosed wrongly | 15%, with a range of 5 to 41% across studies. Women are 2.1 times more often affected |
| Multiple sclerosis not recognised | 36%, with a range of 3 to 58%. Delay to correct diagnosis of 17.3 months |
| Patient-reported delay between first signs and diagnosis | Median of 5.4 years among those whose earlier symptoms had never been assessed |
| Treatment started within 6 months against after 16 months | Risk of reaching moderate disability reduced by 45%, and of conversion to a progressive form by 60% |
The last figure comes from a cohort of 580 patients followed for 11 years, not from a randomised trial. It is an association, not demonstrated causation.
A series of 110 patients wrongly diagnosed across 4 academic centres adds an element that gives the measure of the problem. The wrong diagnosis had lasted more than 10 years in 33% of them, 70% had received an unnecessary disease-modifying treatment, and an earlier opportunity to make the correct diagnosis was identified in 72%.
Migraine and epilepsy, two diagnoses often missed or wrongly made
Across 1,203 patients consulting for headache in 128 practices in 15 countries, with diagnoses validated blind, 25% of those who met the criteria for migraine did not receive that diagnosis. And among patients labelled non-migraine headache, 82% actually had migraine or probable migraine.
In a cohort of 1,161 migraine patients recruited across 12 centres in 7 countries, only 28% knew they had migraine, and 76% had undergone brain or neck imaging because of it.
In epilepsy the problem is symmetrical. A systematic review of 27 studies finds a frequency of wrongly made diagnosis ranging from 2% to 71% depending on the setting, the two most frequent mimics being syncope and psychogenic non-epileptic seizures. The range is too wide to quote a single value, and the range has to be given.
Brain MRI in a person without symptoms, many findings, few serious diseases
| What is measured | The result |
|---|---|
| Potentially serious incidental findings, 32 studies, 27,643 participants | 1.4% for brain MRI alone, 3.9% for brain and whole-body MRI |
| Among potentially serious findings that were followed up | Only 20.5% led to a genuinely serious final diagnosis |
| 16,400 MRI scans of volunteers reread by a neuroradiologist | 83% normal, 13.3% abnormal with no follow-up recommended, 3.7% abnormal with follow-up recommended |
| 1,867 young adults aged 18 to 35 | 4.2% incidental findings, 1.9% requiring an opinion, 0.6% with a final diagnosis judged serious |
Put another way, about 4 people in 5 in whom a potentially serious abnormality is found will ultimately not have a serious disease.
The consequence is direct and it holds for any assessment. Brain imaging in a person without symptoms generates uncertainty, surveillance, and sometimes avoidable procedures. No long-term follow-up data allow anyone today to claim that this screening improves health.
Neurological second opinion, limited data
The literature on the specifically neurological second opinion is very thin, and that has to be said. The best available data is a prospective study of 183 patients, in which the second opinion led to a different diagnosis in 33% of them, with therapeutic consequences in 23% of cases.
That figure comes from a university clinic, that is, a population already selected for its complexity. It does not mean that a third of the diagnoses made elsewhere are wrong, and we will never present it that way.
How long people wait
| Country | What is measured | The wait |
|---|---|---|
| England | Neurology pathways, January 2026 | 57% start treatment within 18 weeks, 60% in neurosurgery |
| Canada | From family doctor to treatment in neurosurgery, 2025 | 49.9 weeks, of which 39.7 weeks before the specialist consultation. MRI, 18.1 weeks |
| France | Time to access a neurologist | No official public data exists. We will not transpose the figure from another specialty |
These 3 lines do not compare. The first comes from an administrative file, the second from a declarative survey of physicians and covers neurosurgery rather than medical neurology, the third does not exist.
What the studies do not allow anyone to claim
Nothing entitles us to write that early treatment of multiple sclerosis reduces disability, because the available studies are observational and patients treated early probably differ from those treated late. We will write “associated with”, not “enables”.
On artificial intelligence, what is documented today is a logistical gain, minutes saved in the stroke pathway, not a demonstrated improvement in diagnosis or prognosis. A review of 81 studies comparing deep learning with clinicians in fact found a high risk of bias in 58 of them, while 61 claimed in their abstract a performance at least equivalent.
In practice, if this concerns you
You set out your problem, your symptoms or your question. Reports, laboratory results and images already available can be studied before you arrive, so that the pathway is prepared. The aim is not to run every test as a matter of course, but to select those that can genuinely help confirm a hypothesis, rule out a risk or guide the decision. The results are brought together and the next steps are organised. Follow-up is then coordinated.
Need a medical opinion quickly? For a concern that does not call for emergency services, you can ask to speak to a doctor at any time through Noria Health Hub. This on-demand medical consultation is billed separately from the assessment pathway.
Further reading
The articles that complete this one.
A” rel=”noopener” target=”_blank”>https://noriahealth.com/blog/a-spine-mri-does-not-tell-you-the-pain/”>A spine MRI does not tell you the pain
Waiting” rel=”noopener” target=”_blank”>https://noriahealth.com/blog/what-delay-really-costs-what-the-studies-say/”>Waiting for a diagnosis or a treatment, what the studies actually measure
Sources
GBD 2021 Nervous System Disorders Collaborators, Global, regional and national burden of disorders affecting the nervous system, The Lancet Neurology, 2024. https://doi.org/10.1016/S1474-4422(24)00038-3
Newman-Toker DE et al., Burden of serious harms from diagnostic error in the USA, BMJ Quality and Safety, 2024. https://doi.org/10.1136/bmjqs-2021-014130
Newman-Toker DE et al., Serious misdiagnosis-related harms in malpractice claims, the Big Three, Diagnosis, 2019. https://doi.org/10.1515/dx-2019-0019
Tarnutzer AA et al., ED misdiagnosis of cerebrovascular events in the era of modern neuroimaging, Neurology, 2017. https://doi.org/10.1212/WNL.0000000000003814
Misdiagnosis and underdiagnosis of multiple sclerosis, systematic review and meta-analysis, Multiple Sclerosis Journal, 2024. https://doi.org/10.1177/13524585241274527
Patient-reported diagnostic delay in multiple sclerosis, Multiple Sclerosis Journal Experimental Translational and Clinical, 2025. https://doi.org/10.1177/20552173251333390
Diagnostic delay in relapsing-remitting multiple sclerosis and disability outcomes, Multiple Sclerosis Journal, 2023. https://doi.org/10.1177/13524585231197076
Early treatment initiation after a first demyelinating event and long-term disability, Neurology, 2023. https://doi.org/10.1212/WNL.0000000000207664
Solomon AJ et al., The contemporary spectrum of multiple sclerosis misdiagnosis, Neurology, 2016. https://doi.org/10.1212/WNL.0000000000003152
Tepper SJ et al., Prevalence and diagnosis of migraine in patients consulting for headache, Headache, 2004. https://doi.org/10.1111/j.1526-4610.2004.04167.x
Migraine awareness and diagnosis in a transcontinental cohort, European Journal of Neurology, 2020. https://doi.org/10.1111/ene.14098
Oto M, The misdiagnosis of epilepsy, appraising risks and managing uncertainty, Seizure, 2016. https://doi.org/10.1016/j.seizure.2016.08.005
Rizzo G et al., Accuracy of clinical diagnosis of Parkinson disease, systematic review and meta-analysis, Neurology, 2016. https://doi.org/10.1212/WNL.0000000000002350
Second opinions in an outpatient neurology clinic, Nederlands Tijdschrift voor Geneeskunde, 2011. https://pubmed.ncbi.nlm.nih.gov/21835065/
Gibson LM et al., Potentially serious incidental findings on brain and body magnetic resonance imaging of apparently asymptomatic adults, BMJ, 2018. https://doi.org/10.1136/bmj.k4577
Incidental findings in a large research brain MRI cohort, American Journal of Neuroradiology, 2023. https://doi.org/10.3174/ajnr.A7821
Vernooij MW et al., Incidental findings on brain MRI in the general population, New England Journal of Medicine, 2007. https://doi.org/10.1056/NEJMoa070972
Incidental findings on brain MRI in young adults, Frontiers in Neurology, 2021. https://doi.org/10.3389/fneur.2021.675244
Nagendran M et al., Artificial intelligence versus clinicians, systematic review of design and reporting of deep learning studies, BMJ, 2020. https://doi.org/10.1136/bmj.m689
NHS England, Referral to Treatment Waiting Times, January 2026 data. https://www.england.nhs.uk/statistics/statistical-work-areas/rtt-waiting-times/
Fraser Institute, Waiting Your Turn, Wait Times for Health Care in Canada, 2025. https://www.fraserinstitute.org/studies/waiting-your-turn-wait-times-for-health-care-in-canada-2025
World Health Organization, Global status report on neurology / neurological care, 2025: more than 3 billion people affected and major workforce gaps. https://www.who.int/news/item/14-10-2025-11-million-lives-lost-each-year-urgent-action-needed-on-neurological-care