Most doctors, if pressed, could not tell you what a clinical neurophysiologist does. This is not entirely surprising. Neurophysiology sits in a quiet corner of medicine, produces reports full of numbers that other specialties find baffling, and has historically done very little to explain itself to the outside world. The specialty’s idea of marketing is a well-formatted EMG report. This is not working.
So here is an honest account of what the job looks like from the inside — the interesting bits, the frustrating bits, and the bits nobody mentions in the person specification.
Clinical neurophysiology is a diagnostic specialty. You are not primarily a treating physician — you are a specialist in the electrical behaviour of the nervous system, and your job is to characterise it objectively and translate that into something clinically useful for the people who refer to you.
The core tools are nerve conduction studies, EMG, and EEG — covered in more detail in other posts in this series — but the scope of the specialty is broader than that. Intraoperative neurophysiological monitoring, evoked potentials, sleep studies, neuromuscular junction assessment, and increasingly, quantitative and computational approaches to electrophysiological data all fall within the specialty’s remit.
Importantly, despite being a diagnostic specialty, it is not a passive one. You are not processing samples in a laboratory. You are seeing patients, taking histories, performing a clinical examination, forming a differential diagnosis, and then doing a test that you personally design and execute in real time based on what you find. It is much closer to a clinical consultation than most people expect — with electrodes.
My week splits roughly in half: clinical neurophysiology on one side, research on the other. As a registrar, the clinical days have a fairly predictable structure, though what happens within them rarely is.
A morning EMG clinic involves seeing referred patients — taking a history, examining them, forming a hypothesis about what is going on before any needles come out — and then performing the study with direct consultant supervision. As a trainee, the more technically demanding procedures (single-fibre EMG, for example) are done with a senior in the room giving real-time feedback on technique. This is one of the genuinely good things about the training: it is very hands-on, very immediate, and you can feel yourself getting better at it in a way that is satisfying.
The afternoon is usually EEG reporting — a mix of urgent inpatient studies, outpatient recordings, and longer video telemetry reviews. This is a different cognitive mode: pattern recognition, systematic review, knowing when something subtle is significant and when it is artefact. There is a reason EEG interpretation takes years to do well, and a corresponding reason that misinterpretation is one of the most common causes of misdiagnosis in epilepsy.
One day a week is largely taken up with the infrastructure of a training post — MDTs, grand rounds, teaching, training meetings. This is sometimes frustrating and sometimes genuinely valuable, depending on the day and the grand round.
Then there are the moments that make you remember why you chose a specialty that most of your medical school friends cannot pronounce. A patient comes in with a referral for suspected carpal tunnel syndrome. Routine. You start the study. The numbers are wrong — not slightly wrong, but wrong in a way that makes the hairs stand up on the back of your neck. The pattern does not fit carpal tunnel. It does not fit anything benign. You extend the protocol, you check your findings, and by the end of the hour you are looking at the electrophysiological signature of early motor neuron disease in someone who came in expecting a straightforward wrist problem. That is not a pleasant feeling, but it is the feeling of having found something real — something that would have been missed, or at least delayed, without this test.
That happens. Not often, but it happens. And it is the reason the routine parts of the job feel worth doing.
I work in an adjacent university for roughly half my time, which makes me unusual — there are only a handful of clinical academic neurophysiologists in the UK, though they make up a disproportionate fraction of trainees overall. Research in neurophysiology methods means my weeks alternate between collecting data from patients using approaches that do not yet exist in clinical practice, and sitting at a computer writing analysis code for data I have already collected.
The research work is interesting in a way that is difficult to describe without sounding evangelical, so I will try to be specific: it is interesting because neurophysiology generates extraordinarily rich data that current clinical practice largely ignores, and working out what to do with that data — methodologically, statistically, clinically — is a genuinely open problem. There are real questions here that nobody has answered yet.
The flip side is that research requires a specific kind of self-motivation that clinical medicine does not. In the clinic, the day is structured for you. In the lab, you have to structure it yourself, tolerate months of work that produces no clear result, and maintain genuine curiosity about why something is the way it is even when the data is not cooperating. If that sounds appealing, neurophysiology research is a good place to do it. If it sounds like a description of a bad day, the purely clinical route is perfectly viable and has plenty of its own rewards.
The entry route to neurophysiology in the UK requires completion of core training — this can be Internal Medicine Training (IMT), paediatrics, ACCS, or surgery — plus the relevant membership examination (MRCP for most). Higher specialty training in neurophysiology is then four years, after which you sit the specialty exit examination.
A few things worth knowing:
It is a small specialty. There were six higher specialty training posts in the UK this year. Six. This is not a typo. Competition for training posts is significant, and geography matters — the posts that exist are not evenly distributed.
Many trainees come across from neurology, though this is not a requirement. The overlap in knowledge base is substantial, and the transition is usually smooth. What neurophysiology adds — beyond the technical skills — is a depth of understanding of peripheral nerve and muscle physiology, and a particular way of thinking about objective evidence, that is genuinely complementary to a clinical neurology background.
The consultant job market exists but is patchy. There is real demand for neurophysiologists, particularly outside major academic centres, but availability of posts in any given region at any given time is variable. This is worth factoring into career planning early, particularly if geography is a constraint.
A few things I wish someone had said earlier:
You stop being a prescribing physician. This sounds like a minor administrative point, and it is not. If you come from a background where you are used to identifying a problem and then doing something about it — adjusting the medication, making the referral, having the conversation with the patient — the transition to a purely diagnostic role involves a genuine adjustment. You find things, you report them, and then the management happens elsewhere. For some people this is a relief. For others — and I include myself in this — surrendering that ability to act on your findings takes some getting used to.
The intellectual satisfaction is real but delayed. EMG is a skill that improves slowly and non-linearly. The first six months of training involve a lot of performing technically adequate studies without a deep understanding of why you are doing what you are doing. That understanding comes later, and when it does, the job becomes substantially more interesting. The people who thrive in neurophysiology are almost always the ones who stayed curious through the early phase when they were not sure they were doing the right thing.
It is a good specialty for people who like building things. Whether that is a new analysis pipeline, a novel recording approach, a teaching resource, or a clinical service — there is a lot of space in neurophysiology to create something that did not exist before, because the field is small enough that gaps are visible and tractable. This is one of the things I genuinely did not expect and have found unexpectedly satisfying.
If you are at FY2 or IMT level and find yourself drawn to neurology but also interested in something more technical and investigative, neurophysiology is worth understanding better. It is not for everyone — the diagnostic rather than therapeutic focus is a genuine consideration, and the small training numbers mean you need to be deliberate about pursuing it.
If you are working in research that touches on neuromuscular disease, motor control, epilepsy, or electrophysiological methods — regardless of whether you are considering the specialty as a career — neurophysiology is a community worth being aware of. The people in it are generally interested in collaboration, the methods are applicable across a wide range of research questions, and the field is small enough that it is genuinely possible to know most of the active researchers personally.
That is not a bad thing to be able to say about a specialty.
| *Other posts in this series: Nerve conduction studies for the referring clinician | EEG for the referring clinician* |
Somewhere in a hospital near you, a neurophysiologist is staring at a referral that says “?carpal tunnel” and nothing else. No clinical findings. No symptom duration. No mention of which hand. Just a question mark and a diagnosis that may or may not be correct.
This is fine. We are used to it. But it does mean that the test you ordered — the one designed to answer a specific clinical question — is now going to answer a slightly different, more generic question instead. The result may still be useful. It may also be a report that lands in your inbox and tells you approximately nothing actionable.
This guide is an attempt to fix that, by explaining what nerve conduction studies and EMG actually are, what they can tell you, and how to get the most out of them. It is not a criticism of anyone. Neurophysiology is genuinely confusing to people who do not spend their days in it, and the specialty has historically been bad at explaining itself. Consider this a peace offering.
Here is the single most useful thing to understand about neurophysiology referrals, and the thing that distinguishes them from almost every other investigation you can request.
An NCS/EMG is not like an ECG. There is no machine that independently processes the data and produces a result. The study is designed, modified, extended, and interpreted in real time by a clinical neurophysiologist — a specialist — who is making decisions throughout based on what they find. Which nerves to study, which muscles to sample, whether to add extra protocols, how to interpret borderline values — all of this depends on knowing what question is being asked.
There is a principle that applies to many things in medicine: there is no point doing a test unless you know what you are going to do with the result. Before picking your nose, in other words, you need a plan for what happens next. The neurophysiologist needs a hypothesis to work with — not a definitive answer, but a differential diagnosis, a clinical picture, a reason why this particular patient is in front of you.
A well-framed referral gives the neurophysiologist the information to tailor the study. An under-specified referral produces a generic study that may or may not answer your question. Critically, the study takes at least an hour — often longer for complex cases — and that time is spent thinking through a diagnostic problem with the patient in the room. It is a clinical consultation with electrodes, not a scan with a readout.
A nerve conduction study works by applying a small electrical stimulus to a nerve — via surface electrodes on the skin — and recording the response either from the nerve itself or from the muscle it supplies. The stimulus is mildly uncomfortable (patients often describe it as a sharp tap or a static shock), but brief. You do not need to warn your patients that it is agony, because it is not.
From these recordings, two key measurements come out:
The CMAP — Compound Muscle Action Potential is the summed electrical response of all the muscle fibres activated by stimulating a motor nerve. The amplitude tells you roughly how many functioning motor axons are present and whether the neuromuscular junction is intact. A reduced CMAP amplitude means axonal loss or neuromuscular junction failure. A normal amplitude but prolonged latency or slowed conduction velocity means the nerve fibres are intact but conducting poorly — which points toward demyelination.
The SNAP — Sensory Nerve Action Potential is the equivalent for pure sensory nerves. SNAPs are small signals and technically demanding to record reliably. A reduced or absent SNAP indicates either axonal loss in sensory fibres, or a lesion distal to the dorsal root ganglion (DRG). This last point matters a lot diagnostically: a normal SNAP in a patient with sensory symptoms in a dermatomal distribution suggests the lesion is proximal to the DRG — i.e. a root problem, not a nerve problem. We will return to this.
EMG is a separate but complementary test. A fine needle electrode — roughly the gauge of an acupuncture needle — is inserted into a muscle to record its electrical activity at rest and during contraction. At rest, a healthy muscle is electrically silent. A denervated muscle is not: it produces fibrillation potentials and positive sharp waves, which are the electrophysiological signature of a muscle that has lost its nerve supply and is not happy about it. During contraction, the size, shape, and recruitment of motor unit potentials tell you whether the muscle is recovering from a nerve injury (large, complex, reinnervating potentials), primarily diseased (small, fragmented myopathic potentials), or somewhere in between.
When the report comes back, the two most important words to look for are “axonal” and “demyelinating.” These describe fundamentally different pathological processes with different implications for prognosis and urgency.
Demyelination means the nerve fibres themselves are intact, but the myelin sheath surrounding them is damaged. Signals still get through — they just travel more slowly. On NCS this shows up as prolonged latencies and reduced conduction velocities, with relatively preserved CMAP and SNAP amplitudes. In focal entrapment neuropathies — carpal tunnel syndrome being the textbook example — localised demyelination at the compression site produces slowing across that segment alone. Demyelinating lesions are generally more reversible: once you decompress the nerve, or treat the underlying cause, the myelin can recover.
Axonal loss means the nerve fibres themselves have been lost. The signals are reduced because there are fewer conductors, not because each conductor is slower. On NCS this shows up as reduced CMAP and SNAP amplitudes with relatively preserved conduction velocities. Recovery from axonal loss is slow — axons regenerate at approximately 1mm per day, which means that for a proximal lesion, meaningful functional recovery may take months to years, and may never be complete.
This distinction is not academic. A report describing “severe axonal loss” in a motor nerve is telling you that the patient has lost a substantial number of nerve fibres and that the road to recovery is long. A report describing “mild demyelinating features at the wrist consistent with carpal tunnel syndrome” is describing a very different situation. Both are called “abnormal NCS” but they are not remotely the same finding.
For most peripheral nerve conditions, the optimal time to refer is at least two weeks after symptom onset, and ideally longer. This is because the electrophysiological changes that follow nerve injury — Wallerian degeneration of the distal axon, followed by denervation changes appearing in the muscle — take time to develop fully. A study done in the first two weeks after an acute mononeuropathy or suspected radiculopathy may be entirely normal, not because the patient is fine, but because the nerve has not yet had time to show its hand.
There are important exceptions:
Guillain-Barré syndrome is the main one. In suspected GBS, early NCS is clinically useful precisely because it can help confirm the diagnosis and — critically — differentiate the demyelinating subtype (AIDP) from the axonal subtypes (AMAN, AMSAN), which have different prognoses. Early electrodiagnostic testing, particularly H-reflex studies, can yield diagnostically useful information within the first week of weakness onset. Serial studies are often needed as the picture evolves. If you are suspicious of GBS, refer early and do not wait.
Myasthenia gravis and neuromuscular junction disorders are not subject to the same timing constraints as nerve injury studies. Repetitive nerve stimulation (RNS) and single-fibre EMG (SFEMG) — the tests used to assess the NMJ — can be done at any point in the disease course. In the inpatient setting, RNS can rapidly support a diagnosis of MG while antibody results are pending, making early neurophysiology referral actively useful in acute presentations with suspected NMJ disease.
Suspected motor neuron disease also warrants prompt referral, both because the EMG findings are not time-dependent in the same way and because the diagnostic urgency is high.
Acute nerve injury with suspected transection — early NCS can establish a baseline and help determine whether nerve continuity is preserved, which guides surgical decision-making.
For everything else — a slowly progressive neuropathy, a suspected entrapment, a chronic radiculopathy — there is no advantage to referring in the first two weeks, and you may simply generate an uninformative result and a frustrated neurophysiologist.
The following is a fictional case for illustration.
A 48-year-old right-handed builder presents with six weeks of progressive right neck pain radiating into the medial forearm, ring finger, and little finger. He has noticed weakness in his grip and some wasting of the intrinsic hand muscles. Symptoms began after heavy lifting and have been constant and worsening since. MRI is pending.
The differential is: C8 radiculopathy vs ulnar neuropathy at the elbow vs lower trunk brachial plexopathy.
What the NCS will show — and why the SNAP is the key:
Here is where the DRG anatomy becomes diagnostically useful. In a C8 radiculopathy, the lesion is proximal to the dorsal root ganglion. The sensory neuron cell body sits in the DRG and is physically intact — it is only the central projection (into the spinal cord) that is compressed, not the peripheral axon. This means that the SNAP recorded from ulnar sensory fibres (which carry C8 sensory information) may be completely normal in amplitude, even in a patient with significant C8 sensory symptoms.
Contrast this with an ulnar neuropathy at the elbow, where the lesion is distal to the DRG. Here, the peripheral axon itself is damaged, and the SNAP will be reduced or absent. This single NCS finding — normal vs abnormal SNAP — is one of the most powerful ways to distinguish a root lesion from a peripheral nerve lesion, and it is why knowing your differential before the study matters so much.
The motor studies may show a reduced CMAP from abductor digiti minimi (ulnar, C8) if there is significant motor axonal loss at the root.
What the EMG adds:
The needle study is where radiculopathy is confirmed. Denervation changes in C8-innervated muscles — first dorsal interosseous, flexor carpi ulnaris, abductor pollicis brevis — combined with abnormality in the cervical paraspinal muscles at the appropriate level, and with sparing of muscles outside the C8 myotome, localises the lesion to the root. Finding changes in paraspinal muscles is particularly valuable because these muscles are supplied directly from the posterior primary rami before any plexus or peripheral nerve branching — if they are abnormal, the lesion must be at or proximal to the root.
Reading the conclusion:
A report saying “active C8 radiculopathy with severe axonal loss and ongoing denervation” is telling you there is significant structural damage and an active ongoing process. This warrants urgent surgical review. A report saying “chronic C8 radiculopathy with mild axonal loss, reinnervation, and no active denervation” describes a lesion that is old, stable, and recovering — a very different clinical situation that may be managed conservatively. If the report does not make this distinction clearly, it is entirely reasonable to ring the neurophysiology department and ask. That is what we are there for.
The following information directly changes what the neurophysiologist does in the room. The more of it you include, the better the study and report will be.
You do not need a definitive diagnosis before referring. You need a hypothesis — a reasonable clinical question that neurophysiology can help answer. The neurophysiologist will take it from there.
Next in this series: EEG for the referring clinician — why a normal EEG does not mean your patient does not have epilepsy.
]]>I spend roughly half my working life doing clinical neurophysiology — EMG clinics, EEG reporting, the familiar rhythm of the department. The other half I spend as a researcher, building computational models of the very systems I assess in clinic. This split role gives me an interesting vantage point, and it has prompted some questions I want to work through in this series — not as criticisms of colleagues or the field, but as a genuine attempt to think about where neurophysiology is heading, and where it could be going instead.
The starting point is a fairly simple observation: neurophysiology is operating in an increasingly competitive diagnostic environment, and the specialty has not yet fully articulated what its distinctive contribution is or should be.
Consider what has happened in adjacent fields over the past two to three decades. Advances in neuroimaging have transformed the diagnostic workup for many neurological conditions, giving clinicians — and patients — direct visual access to anatomy and pathology. MRI and ultrasound findings are immediately legible across specialties in a way that a table of latencies and amplitudes is not. Meanwhile, molecular genetics has reshaped the diagnosis of inherited neuromuscular disease; conditions that would once have required extensive electrophysiological characterisation can now be resolved with a gene panel. And in autoimmune neurology, the explosion of antibody discovery has been transformative — conditions like myasthenia gravis and CIDP, once diagnosed principally through electrophysiology and clinical pattern recognition, now have specific, actionable biomarkers that predict prognosis and guide treatment selection. The discovery of antibodies targeting nodes of Ranvier, for instance, has defined an entirely new category of autoimmune nodopathy within what was previously considered the CIDP spectrum. The pace of this development shows no sign of slowing. Pathogenic autoantibodies against paranodal membrane proteins in CIDP, and against muscle-specific kinase in myasthenia gravis, are now recognised as essential to clinical diagnosis and are reshaping treatment algorithms.
None of this diminishes what neurophysiology offers. But it does change the context in which we work, and it raises a legitimate question: where is the equivalent leap in electrophysiological technique?
The core tools of clinical neurophysiology — nerve conduction studies and needle EMG — were established in their essential form decades ago and remain the foundation of clinical practice. There is nothing wrong with that; they remain powerful and, in skilled hands, genuinely irreplaceable. But the question of whether the field has developed new measurement approaches at a pace commensurate with advances in adjacent fields is worth asking. A systematic review of neurophysiological outcome measures used in ALS clinical trials — one of the better-studied areas for novel electrophysiological techniques — found that despite 32 interventional trials employing neurophysiological outcome measures since 1986, there is limited standardisation between studies and an apparent ‘scatter-gun’ approach to technique selection, which the authors argue reflects the absence of a coherent, updated toolkit. Neurophysiology has many promising tools. The challenge is developing and standardising them into something clinically usable.
The second issue is less about technique and more about communication. A report that confirms or excludes a diagnosis with a clear management implication is enormously valuable. But a significant proportion of referrals do not yield that. The result is “indeterminate” or “within normal limits given technical constraints” — and the referring clinician, often not a neurophysiology specialist, is left to decide what to do next. Should the test be repeated? Was there a technical limitation? Does a normal result exclude pathology, or is the test simply not sensitive to what the patient has?
Studies of referral concordance suggest frequent inconsistency between the clinical diagnosis expected by the referring physician and what EDX testing finds — with normal results in around a quarter of cases and referral diagnosis confirmed in only around 60% overall. This is not necessarily a problem with neurophysiology per se; it partly reflects the difficulty of the clinical questions being asked, and the reality that abnormal electrophysiology sometimes follows rather than precedes clinical symptoms. But it does mean that a substantial proportion of patients and their referrers receive a result that does not obviously tell them what to do next.
Part of the issue is a fundamental one about how our results are presented. Electrodiagnostic testing is, to many clinicians, a “black box” — a commonly ordered test that can provide very definitive information that is often not well understood, with most clinicians relying primarily on the concluding statements rather than the underlying data. If that is the experience of clinicians ordering these tests, it is worth reflecting on what that means for how we structure and communicate our findings.
There is an instructive contrast here with imaging. A radiologist’s report arrives alongside images that the referrer can look at directly. The findings are immediately interpretable, at least in broad terms. Our reports, by contrast, present numerical data — latencies, amplitudes, recruitment patterns — that require specialist training to interpret, and where the meaning of marginal or borderline findings is often unclear even to the specialist. The gap between what we measure and what we can communicate to a non-specialist referrer is one that the field has not systematically addressed.
None of this is a counsel of despair. The NHS neurology waiting list grew by 76% between 2021 and 2023, and the demand for objective electrophysiological assessment is not diminishing. The data our tests generate — when properly analysed — contains far more information than conventional reporting captures. The motor unit, the fundamental unit of voluntary movement, encodes information about its structure, recruitment and mechanical behaviour in every contraction. We are currently using a fraction of that signal.
The rest of this series will work through what a more expansive neurophysiology might look like: more quantitative, more objective, more legible to the clinicians and patients we serve, and more integrated with the computational and imaging tools that now surround us. Some of what I’ll describe is already being developed in research settings; some of it requires a cultural shift in how we think about what our job is.
The first step, as with most things, is being honest about where we are.
Next week: We need to talk about our data.
]]>