Four months after her thalamic stroke, a patient comes back to clinic doing objectively well — walking unaided, back to cooking. But she cradles her right arm while she talks, and when I reach out to examine it she flinches before I touch her. The arm burns, she says. It burns all day, it burns worse in the cold, and her sleeve hurts. Her chart already carries three explanations from three visits: "arthritis," "possible neuropathy," "anxiety."

The exam takes two minutes and settles it. Pinprick is dull over the right arm and face compared with the left. A cold tuning fork, barely noticed on the left, is painful on the right. Light stroking of the forearm — a stimulus that should register as nothing — produces a wince. Pain in the territory of her stroke, with spinothalamic sensory signs in the same territory: this is central post-stroke pain, and it is generated not by her arm but by the lesion in her brain.1

What CPSP is, and who gets it

CPSP is central neuropathic pain arising from a cerebrovascular lesion of the somatosensory pathways — most characteristically the spinothalamic system that carries pain and temperature. The classic teaching case is the thalamic stroke (Dejerine and Roussy's eponymous syndrome), but the thalamus holds no monopoly: lateral medullary infarcts carry among the highest rates of CPSP of any stroke location, and lesions from brainstem to parietal operculum can produce it. Across stroke survivors overall, a commonly cited figure is that CPSP develops in roughly 8–10%; a Danish population-based examination found a minimum prevalence of 7.3% definite or probable CPSP.15

The timing is the trap. Onset is typically delayed — days to months, most within the first six months — arriving after the acute team has signed off and the narrative has become "recovering nicely." A new burning limb at month four rarely gets connected to the stroke at month zero unless someone makes the connection deliberately. The character is neuropathic: burning, aching, freezing, squeezing, lancinating; constant with paroxysms; aggravated by cold, touch, stress, and fatigue.

The two-minute exam that makes the diagnosis

There is no confirmatory test for CPSP; imaging already showed you the lesion. The diagnosis stands on three legs:

  • Distribution: the pain occupies a body territory that makes neuroanatomical sense for the stroke — hemibody for thalamic lesions, crossed face-and-body patterns for lateral medullary (Wallenberg) lesions. Thalamic infarcts may sit in PCA territory; Wallenberg is PICA, not PCA.
  • Sensory signs in that territory: blunted pinprick and temperature sensation are the signature (spinothalamic), often with preserved light touch and proprioception; plus the positive phenomena — dysesthesia, hyperalgesia, and allodynia, pain from a normally innocuous stimulus. A cold object and a pin answer most of the question at the bedside.
  • Exclusion of the commoner pains — which is where most misdiagnosis actually happens.
Pain after stroke Tells Treatment direction
Central post-stroke painBurning/freezing, delayed onset, spinothalamic sensory signs, allodynia, territory matches lesionNeuropathic agents (this article)
Hemiplegic shoulder painMovement-related, localized to shoulder, subluxation or restricted range on examPositioning, therapy, targeted shoulder care
Spasticity-related painTone, spasms, pain with stretch — see the spasticity guideTreat the tone, hunt aggravators
Musculoskeletal / otherJoint-line findings, weight-bearing pattern; consider DVT in a swollen painful leg — score it with Wells DVTCause-specific
"Arthritis" that burns, hates cold, and lives on the stroke side deserves a pin and a cold tuning fork before it deserves an NSAID.

Treatment: the evidence, and how to dose it for a stroke brain

CPSP is treated as neuropathic pain, and the honest preamble to every prescription is about expectations: the realistic goal is meaningful reduction — commonly framed as 30–50% — plus better sleep and function, not elimination. Setting that frame early is the difference between a treatment partnership and a march through abandoned prescriptions.

  • Amitriptyline carries the classic — and small (n=15 crossover) — positive trial in CPSP specifically, where it relieved pain and carbamazepine did not convincingly.2 In practice its anticholinergic load, orthostasis, and cognitive cost demand low starting doses and slow titration in older patients — and caution with the cardiac history that many stroke patients carry.
  • Lamotrigine showed moderate benefit in a dedicated randomized CPSP trial (n=30) and is often the better-tolerated choice in the elderly.3 Its mandatory slow titration is a feature to plan around, not a surprise; and it does double duty in the patient who also has post-stroke epilepsy.
  • Gabapentinoids (pregabalin, gabapentin) are reasonable options extrapolated from central neuropathic pain generally; the dedicated pregabalin CPSP trial missed its primary endpoint but improved sleep, anxiety, and global impression.4 Sedation and edema are the taxes; both drugs are renally cleared — dose against the Cockcroft-Gault calculator, not the habit dose.
  • What mostly fails: conventional analgesics and NSAIDs do little for central pain, and opioids perform poorly here relative to their harms — a trap worth naming before someone else sets it.
  • Refractory disease is real: combination therapy, referral to a pain specialist, and, in selected patients, neuromodulation (repetitive TMS and motor cortex stimulation have the most study) are the far end of the road.1

Two companions ride along with CPSP often enough to screen for on the same visit: depression and anxiety amplify central pain and are amplified by it — screen deliberately with the PHQ-9 and GAD-7 — and poor sleep, which lowers the pain threshold and is often the first thing a successful drug fixes. Post-stroke fatigue completes the frequently co-traveling trio.

The bottom line

When a patient reports new burning, freezing, or electric pain on the stroke side — weeks to months out, worse with cold and touch — think of the lesion before the limb. Map the pain against the stroke territory, put a pin and something cold on the skin, and separate it from the shoulder, the spasticity, and the joints, because each of those has a different fix. Then treat it as the neuropathic pain it is, titrated for the brain and kidneys in front of you, with expectations set honestly. CPSP rarely disappears — but named, explained, and properly treated, it usually loosens its grip. And for a patient who has been told three times that it was arthritis or anxiety, the name alone is therapeutic.