11 From where to what: building the differential

Every module so far has been about "where", the first word in Where Before What. This module turns to "what", the cause, and to the question that links the two, which is when.

For working out where a lesion sits, the examination matters far more than the history. The history, meaning the way the deficits evolved over time, points to the cause (Brazis 2011, p. 4). Take a hemiparesis. If it arrived within minutes, a vascular event or a seizure is most likely. If it developed over a few days, think of infection or demyelination. If it crept up over months, think of a tumor or a degenerative process. The address is the same in all three, and only the clock separates the differentials.

What the examination and the history each tell you

The examination tells you which structure is failing, and the history tells you how it is failing. A cord lesion that appeared over minutes is a different disease from one that appeared over hours, weeks or years, and no examination can tell those four apart. The history does no better on its own. "Getting worse over weeks" means little until you know whether the problem is in the cerebellum or in the muscle.

When the history also localizes

The examination shows the sum of everything that has happened to the patient, and that sum can include lesions laid down at different times. A third-nerve palsy with a hemiparesis on the opposite side points to a single lesion in the midbrain, but only if the two began together. If the palsy dates from childhood and the hemiparesis is new, they are probably two lesions, and the one causing the hemiparesis is probably not in the midbrain (Brazis 2011, p. 4). In children the birth history, the early milestones and old family photographs do the same work. They may show that a "new" weakness has been there since infancy.

Localize before you weigh the tempo

If you fix on the tempo before you have localized, you may settle on "sudden, so stroke" before checking whether the findings even fit the brain. Once you have an address, the clock chooses among the causes that can actually occur there, rather than among every cause you can think of.

Once you have an address and a clock, you need a list of the ways nervous tissue can fail. Most clinicians carry some version of it. A good list is complete enough that nothing important is forgotten, and each category on it comes with the tempo and pattern it usually shows.

Discrete lesions and systems

A second way of dividing the list asks whether a disease makes discrete lesions or attacks a system (Brazis 2011, pp. 4-5). Vascular events, demyelinating plaques, infections, injuries and tumors tend to cause discrete lesions, single or multiple, whose signs meet at a point. Degenerative disorders instead take whole groups of neurons that share a function, so their signs follow a system rather than a place. A symmetric, length-dependent pattern, in which the longest nerves fail first, is the signature of a system process and almost never of a discrete lesion.

The pediatric additions

Adult lists often treat the congenital and genetic categories as rare. In child neurology they make up much of the daily work, from malformations and perinatal injury to inborn errors of metabolism, the muscular dystrophies and the hereditary neuropathies. The categories an adult neurologist would reach for last often sit near the middle of a child's list.

The figure above pairs a level with a tempo. Choose one of each, and it shows the mechanisms that fit that tempo, with pediatric examples for that address. Once you have committed to a where and a when, it makes sure the obvious candidates and the dangerous ones both reach your list.

How to read it

The cerebellum is a good place to start. Acute ataxia in a young child, developing over hours to a day or two, brings post-infectious acute cerebellar ataxia and a toxic ingestion to the top of the list. If the ataxia develops over weeks, a posterior fossa tumor moves up. Opsoclonus-myoclonus-ataxia joins it, and should send you looking for a neuroblastoma. Over years, inherited ataxias such as Friedreich ataxia and ataxia-telangiectasia appear. The examination of a wobbly child can be the same in all three cases, and the clock alone moves you from reassurance and observation, to imaging the posterior fossa, to genetic testing.

The spinal cord can be read in the same way. Sudden weakness below a level suggests trauma or an infarct. Weakness that develops acutely, over hours to days, suggests transverse myelitis or compression from an epidural abscess or hematoma. Compression must be excluded first, because decompression can treat it. A chronic course brings a syrinx, a tethered cord or hereditary spastic paraplegia, and a static picture present from birth suggests a myelomeningocele.

Limits of the grid

The examples for each level and tempo are only illustrations. A rare disease may fit the pairing without being named, and a child can have a common disease that presents at an uncommon tempo. Rank the short list both by how likely each diagnosis is and by the cost of missing it. A treatable emergency such as cord compression or herpes encephalitis is acted on even before it is the most probable answer.

The grid can also be read in reverse, starting from the disease. If a disease you suspect belongs with a level or a tempo that does not fit your patient, doubt the diagnosis, the localization or the history of the time course, and re-examine.

An adult's history is usually read against two questions, how fast the problem came on and whether it is getting worse. In a child, development adds a third, and it changes the meaning of the other two.

Static, progressive, regressive and episodic

Episodes without a lesion

Some of the commonest episodes of childhood have no lesion behind them. Breath-holding spells occur between 6 months and 6 years, follow a fit of crying or a fright, and are outgrown. Night terrors begin between 2 and 10 years and are a partial arousal from deep sleep (Pearl 2014, p. 104). The where-by-when figure lists both, with syncope, as mimics. When the story is typical, there is no lesion to look for.

Static lesions that seem to progress

A fixed lesion can still unfold as the brain matures. A child with spastic cerebral palsy may present as a floppy infant before the spasticity develops (DeMyer, p. 273). A perinatal stroke may be invisible at birth and show itself months later as early hand preference. Preferring one hand before 12 months is abnormal and points to weakness of the other hand (Utah PediNeuroLogic Exam). In both cases the parents are seeing something new, even though the lesion is old. The useful question is whether the child is losing skills or failing to gain them.

Age at onset

Age at onset is a strong filter. A floppy newborn, a toddler who stops walking and an adolescent with a new tremor draw from different groups of diseases, even when the address is the same. The where-by-when grid becomes a three-way table of level, tempo and age at onset.

Module 1 showed that a test result is worth what the prior makes it worth, and that a single false-positive rate can swamp a vague question while barely affecting a sharp one. This section turns that into something you can do with a pen before you sign the request.

Choosing the test

Localization and tempo do two jobs. They choose the test, so a cord level asks for spine imaging, a length-dependent neuropathy for nerve studies and an episodic story for a recording of an event. They also set the prior, because a test aimed where the examination pointed starts with a good chance of finding the cause. Choose the one or two tests that best confirm or refute the tentative diagnosis, rather than every test that covers every possibility (DeMyer, p. 539). Unfocused "gunshot" neuroimaging produces false positives, and only careful localization avoids them (Brazis 2011, p. 1).

Background rates in healthy children

Each of these findings is common enough in well children that it needs a prior before it means anything.

Predict the result before ordering

Before ordering, write two lines in the notes. One gives the result you expect, and the other says what you will do with each answer. If neither a normal nor an abnormal result would change what you do, the test is not ready to be ordered. If you cannot say what an abnormal result would mean for this child, the localization is not finished.

Results that contradict the examination

A result that contradicts a confident localization deserves scrutiny before you accept it. Ask whether it is one of the background findings above, whether the test looked in the right place, and whether the examination needs repeating. The cases let you practice the whole sequence of where, when and what, with the test coming only at the end.