Headache & Vertigo Diagnosis
Headache & Vertigo Diagnosis
Both of these symptoms are diagnosed the same way: a pattern recognised in a history and confirmed on examination . That is not a fallback for want of a test — it is how several of the commonest patterns here are defined, and it is why a normal scan is frequently the expected result rather than an absence of findings. What the investigations do is answer specific questions the history has raised, and where no question has been raised they usually answer nothing. This page sets out the three questions that do most of the work, what the appointment asks and why a diary matters more than an image, the parts of the examination that carry the weight, which tests are used and on what grounds, why a normal result is so often the answer, and what a properly finished assessment leaves behind in writing.
Three questions, and why they outrank a scan
Before anything else is done, three questions are asked, and together they place the great majority of headaches. How long does an attack last if it is left untreated? How many days a month does it occupy? What else arrives with it? None of the three requires equipment, and all three can be answered by the person before they are asked.
The answers work because the patterns differ on exactly these axes. Migraine attacks run hours to days and bring nausea, light intolerance and worsening with movement. Tension-type headache occupies much of a day, is bilateral and brings neither. Cluster headache is measured in minutes to a couple of hours, recurs at the same time each day for weeks, and is accompanied by restlessness rather than stillness. A headache present on most days in someone treating attacks on most days is a fourth pattern with a fourth treatment, and it is identified by the day count rather than by anything found on examination.
For dizziness the parallel question is the one that replaces all three: what brings it on, and how long does a single episode last? Provoked by a movement and lasting seconds is one disease. Continuous for hours to days is another. Persistent and unprovoked for months is a third. The reason this outranks imaging is not that imaging is useless — it is that imaging answers a different question, and for these patterns it is usually the wrong one.
What the questions do not do is replace the examination. They tell the clinician where to look, not what will be found, and the examination is where a central cause is separated from a peripheral one. Both are needed, and the order matters: the history decides what the examination is looking for.
What the appointment asks, and why a diary matters more than an image
The first appointment in this group is longer than most, because the history is the test. What is being assembled is not a story but a pattern: the age at which it started, whether it has changed, how many days a month, how long attacks run untreated, what accompanies them, what has been tried and in what dose, and how many days a month acute medicine is taken. Alongside it, sleep, meals, caffeine, hormonal timing and what the person has given up.
A diary kept for four weeks before the appointment is the single most informative document in the file, and it is worth more than a scan in most cases. It needs to record the days affected, the duration of each attack, what came with it, what was taken, and anything suspected as a trigger. It does not need to be elaborate. What it replaces is a month of recollection in a room where the person is unwell and being asked to summarise.
Two questions are asked that people are frequently surprised by, and both are asked because they change the answer. The first is about migraine in the family — in parents, siblings and children, asked one relative at a time rather than as a general question, because a general question is usually answered no. The second is how many days a month acute medicine is taken, including anything bought over the counter, because medication overuse is a diagnosis made from that number and it is the one most often not asked about.
What is also asked is what has already been done and what came of it: which scans, which medicines, at what dose and for how long, and what was said afterwards. A previous normal scan is useful information rather than a closed door, provided the report is available — and where a previous assessment ended without a plan, that is itself one of the strongest reasons to start again.
The parts of the examination that carry the weight
Six things are done in the room, and each of them answers a question that no scan answers better.
Positional testing. The person is laid back quickly with the head turned to one side, and the eyes are watched. In positional vertigo this reproduces the vertigo after a short delay and produces a characteristic eye movement that both confirms the diagnosis and identifies which ear is responsible — needed before any manoeuvre is performed, because the manoeuvre differs by ear. A different pattern of eye movement, or vertigo without it, points centrally instead.
The head impulse test. The head is turned in a small, fast movement while the person fixes their gaze on a target. Where the vestibular reflex is weak on one side, the eyes make a corrective movement that is visible to the examiner. It takes seconds, it requires no equipment, and it is one of the most informative things done in the assessment.
Eye movements in general. The character of the nystagmus, whether it changes direction, and whether other eye movements are abnormal, is what separates a peripheral from a central cause. This is the part of the examination that decides whether the picture is an inner-ear problem or a neurological one, and it is why dizziness with double vision or slurred speech is never assumed to be from the ear.
Standing and walking. Gait, stance and the ability to stand with the feet together are examined. An inability to stand is one of the features that moves the timescale; a gait that is unsteady in a particular pattern is a finding rather than a description.
Blood pressure lying and standing. Measured in both positions, with the timing recorded, this identifies the orthostatic causes of lightheadedness. It is the reason the blood-pressure question is answered with two numbers rather than one.
The back of the eye. Examined for papilloedema in every new, changed or red-flagged headache. It takes seconds, it is not optional, and it is the finding that changes the urgency of a headache more often than any other single observation.
Which tests are used, and on what grounds
Tests here are reserved rather than routine. Each one below is used because a specific question has been raised by the history or the examination, and where no question has been raised, the test usually answers nothing and can generate findings that then need explaining.
| Test | The question it answers | When it is not used |
|---|---|---|
| 3.0T MRI of the brain | Is there a structural explanation, where a red flag, a new onset after fifty, a change in character or an abnormal sign has raised it | Not routinely, for a recognised pattern with a normal examination. Reserved, not automatic |
| Magnetic resonance angiography or venography | Are the arteries or the venous sinuses involved, where the pattern or the examination points to the vessels | Not as an addition to a normal scan with no indication. It answers one question, and only when that question is real |
| Computed tomography | Is there acute bleeding, in the minutes and hours after an abrupt, maximal headache | Not as a substitute for MRI in the elective assessment of a chronic pattern |
| Audiometry | Is hearing measurably reduced on one side, and does it fluctuate with the episodes | Not needed where dizziness occurs without any change in hearing |
| Vestibular function testing | Which side is weak and by how much, where precision is needed or where the picture is unclear | Not needed to diagnose positional vertigo, which is identified at the bedside |
| Blood tests | Is there an inflammatory, thyroid, haematological or metabolic contributor that would change the plan | Not as a screen in search of a pattern. Ordered for a reason recorded in the notes |
| Electrocardiogram and cardiac monitoring | Is there a rhythm or a blood-pressure cause behind the lightheadedness or a syncopal episode | Not routinely for positional vertigo, where the mechanism is in the ear |
| Sleep assessment | Is disordered sleep driving the headache, where snoring, witnessed apnoea or unrefreshing sleep is part of the history | Not as a general screen. It is one of the commonest missed contributors where it is present |
The discipline that governs the table is a single question asked before any test is ordered: what would be done differently if this came back normal? Where the answer is nothing, the test is not ordered and the reason is written down. Where the answer changes the plan, it is ordered within the same programme rather than referred outward, and the result is given with an interpretation rather than handed over as a report. Where treatment begins once the pattern is named is set out on the treatment page.
When everything is normal, and that is the answer
This is the section people come to a second opinion for, and it is worth stating plainly. For several of the commonest patterns in this group, a normal result is the expected finding rather than an absence of one. Migraine is a disorder of how the nervous system handles stimulation and is not visible on any scan. Positional vertigo is a mechanical problem with crystals in one ear, and imaging of the head is normal by definition. Persistent dizziness is a disorder of how balance is being weighted, and every test in the list above can return normal while the symptom is as disabling as it has ever been.
What follows from that is not reassurance but a positive diagnosis. The pattern has a name, the mechanism has an explanation in plain terms, and the plan belongs on a different track: prevention for the headache patterns, a manoeuvre for the positional one, rehabilitation for the persistent one. None of those requires a scan to begin, and none of them is a consolation prize.
What is not acceptable is the version of this that most people have experienced: told the scan is normal, told therefore that nothing is wrong, and discharged with nothing further to do. That is a failure of the process rather than a finding, and it leaves a person with an unchanged symptom and no explanation for it. The difference between that and a proper assessment is not the scan. It is whether a name and a plan were given afterwards.
For the minority in whom the picture genuinely is incomplete, the honest statement is that the pattern has not yet declared itself, and that the plan is a review at a defined interval with a diary in the meantime. That is a real plan with a real endpoint, and it is worth more than another test ordered to fill the silence.
What a finished assessment leaves behind in writing
An assessment in this group is not finished when the appointment ends. What should exist afterwards is a written statement of five things, and a person is entitled to ask for each of them.
First, the name of the pattern — including where the honest name is that two patterns are present together, which is commoner than most people are told. Second, what was excluded and how, rather than a list of tests that were normal. Third, whether the treatment being taken is itself contributing, with the day count recorded. Fourth, the first step, which is sometimes no investigation and no medicine at all, and is none the less a plan for being written down. Fifth, what would make the plan change — the specific features that should bring the person back sooner.
Alongside it, a review at a defined interval. For a preventive medicine that means long enough to judge it rather than long enough to forget it; for a vestibular programme, long enough for exercises to have been done daily for weeks; for medication reduction, long enough to get through the difficult fortnight at the start of it. A vague instruction to come back if it is no better is not a plan, and it is the commonest way a chronic pattern in this group is left unmanaged.
A written opinion is returned within 48 hours of complete records arriving, and for a records-only review that means from the point the file is complete rather than from the point it is first sent. What treatment looks like once the pattern is named is set out on the treatment page.