Erythromelalgia

Before anything else: put the ice away, and look between every one of your toes tonight.

Use a fan and raise your feet instead. Cool air is safe; ice and cold water are not.

Takes two minutes. No equipment needed.

If you find a blister, a soft white patch, a crack, an ulcer, or a dark area that does not go pale when you press it — book a doctor this week. That is cooling damage, and it is the part of this condition that has ended in amputation. If you also have a fever, spreading redness, or feel generally unwell, go to urgent care or A&E today.

The Verdict

Your feet burn when warm. Cooling helps, and the ice you reach for is what causes the damage.

Think of the small blood vessels in your feet as a heating system with a stuck thermostat and a bypass pipe. When the room warms even slightly, the bypass opens and dumps hot blood straight through the foot. The skin goes red and hot, but the tissue underneath is actually being starved, and that starving is what the burning is. Ice slams the bypass shut, which is exactly why it feels like the answer — and it also freezes skin that is already short on blood supply, which is skin that does not always come back.

Here's what's really happening: the nerves and small blood vessels in your feet misread ordinary warmth as a burn signal, so a warm room or a short walk can set off pain that looks like a foot infection.
The one thing that makes it worse: ice and cold water — in one review of children with the inherited form, six out of ten had skin damage from the cooling they were using for relief, and one adult lost part of a foot after using ice packs.
The one change that matters: swap ice for a fan and raised feet, and check between your toes every single day.

Best for

Anyone whose feet burn and go red with warmth and settle with cold — diagnosed or not.

Skip if

You already have broken skin, a blister or a dark patch on your feet, or you are an adult who has not had a blood test yet. Both of those need a doctor first, not a plan.

Want the full evidence? Keep scrolling

Treatment

What Works

Dark cinematic view of the foot and lower leg circulation

Tier 1 — Strong evidence

Stop the ice and cold-water soaking HIGH

Replace it with raised feet, fans, a cooler room, and lighter bedding. No ice packs, no soaking, no frozen items against the skin, and no cold draught aimed at one patch of skin for hours.

Evidence: STRONG for stopping — cold-immersion complications appeared in 60% of children with the inherited form, and reported outcomes include amputation, deep skin and bone infection, and three weeks of dangerously low body temperature. The replacement itself is clinical judgement, not trial-tested.

Check the skin of both feet daily HIGH

Every day, in good light: between every toe, the heel, the sole. Use a mirror or ask someone if you cannot see underneath.

Evidence: STRONG. Cooling injury is the most common serious complication and it is silent until it is not.

Find your triggers and control the environment HIGH

Heat sets it off in 85% of people, exercise in 78%, and evening or night in 76%. Keep a two-week diary of what happened in the hour before each flare.

Evidence: STRONG for the trigger numbers (46 patients). MODERATE that controlling them reduces attacks — universally recommended, never actually trialled.

Exercise Prescription

Read this before the table. Exercise sets off symptoms in about 8 out of 10 people with this condition, and it is the method researchers use to bring on an attack on purpose in a trial. Meanwhile nobody has ever run a study of exercise, physical therapy, orthotics, compression or graded loading here — not one, in either direction. What follows is our clinical starting point, not numbers taken from a study. Anyone who hands you a confident exercise prescription for this condition is guessing.

Ankle pumps and circles, feet raised

2 × 15 each direction · twice daily

Lying with feet propped above hip height. Should not bring on burning — if it does, stop and keep the feet up.

Seated marching, cool room

2 × 20 · daily if tolerated

Easy pace, sitting. Stop at the first hint of burning, not after it.

Walking in the cool part of the day

Start at 5 minutes · every other day

Flat ground, breathable shoes, early morning or evening. Stop before symptoms start, not when they arrive.

Stationary bike, light resistance, fan on

Start at 5 minutes · 2–3× weekly

Cool the room before you start. Build minutes before you touch the resistance.

Upper body and core work

As tolerated · 2–3× weekly

Keeps you conditioned without heating the feet. Usually the best-tolerated option.

The rule that matters more than the numbers: increase time before effort, cool the room before you start, and stop before symptoms arrive rather than when they do. If an activity reliably triggers a flare, that is information, not failure.

Tier 2 and Tier 3 — what a doctor might add

Tier 2 — Moderate evidence.

Adults with new symptoms should get a blood count MODERATE — this condition can be the first sign of a blood disorder. If the diagnosis is unclear, ask for function-based nerve testing rather than a skin biopsy MODERATE.

Tier 3 — Emerging or weak. All prescribed and supervised by a doctor, not a physical therapist.

Topical amitriptyline-ketamine cream LOW: 75% of 36 patients reported some improvement, but 19% got nothing and 6% got worse locally. Iloprost by drip in a specialist unit LOW: the only placebo-controlled trial this condition has ever had, and it had 12 patients. Sodium-channel medicines such as mexiletine or carbamazepine LOW: response appears to depend on which gene variant you carry. Chemical lumbar sympathectomy for severe unresponsive cases LOW: 69% had complete relief at one week in a series of 13, but among the 4 who did not respond, one died at 3.5 months and one had an amputation at 4 months.

What Doesn't Work

Ice, cold-water soaking and sustained direct cooling. Not simply useless as treatment — this is the leading documented cause of tissue loss in this condition. It persists precisely because it works in the short term.
Skin biopsy to confirm the diagnosis. It comes back normal in about 9 out of 10 people who have this condition.
Aspirin for the primary form. Its evidence belongs to a specific blood disorder, and gangrene has occurred despite daily low-dose aspirin.
A standard strengthening programme prescribed by reflex. Exercise is the trigger in 78% of patients and no trial of exercise exists here in either direction.
Quoting a recovery timeline without saying which form you mean. "A third get worse" and "everyone recovered in six months" are both true — of different groups of people.

Red Flags

Do not work around these. Each one is a reason to be seen rather than to keep managing at home.

Dark cinematic close view of skin and soft tissue of the foot

Any ulcer, blister, soft white patch, crack, or dark area that does not go pale when pressed — in someone who cools their feet, this is cooling injury, and it has led to deep infection, bone infection and amputation. GP this week.
New symptoms in an adult who has not had a blood count — this can be the first sign of a blood disorder, and one report describes gangrene of a finger as the first sign of it. GP.
Fever, spreading redness, or feeling generally unwell — could be a skin infection, which this condition is regularly mistaken for and which is regularly mistaken for it. Urgent GP or A&E.
A foot that turns cold, pale and pulseless — that is not this condition. A&E.
Shivering, confusion, or feeling very cold all over after heavy cooling — over-cooling has caused three weeks of dangerously low body temperature in a documented case. A&E.
Symptoms starting after a new blood pressure medicine — one drug class is documented to produce a very similar picture. Speak to the prescriber; do not stop it yourself.
A sudden severe attack that will not settle — this has put people in hospital. Urgent care.

Where to go: your GP first in almost every case, because the blood screen and the pain medicines both sit there. A&E for infection, a cold pale foot, or low body temperature. There is no single specialty that owns this condition, which is part of why people see a median of three specialists before anyone names it.

Return to Training

The load axis here is temperature, not weight. A percentage deload changes nothing if the session still heats the feet, while the same session in a cool room may be fine. Remove ice baths and contrast baths entirely — they are commonly recommended to training populations and they are the wrong advice for this condition.

Skin on both feet intact for 2 consecutive weeks, checked daily including between the toes
No ice, cold-water soaking or sustained direct cooling for 2 consecutive weeks
Trigger diary kept for at least 2 weeks and your own pattern identified
Adults: blood count completed and reviewed
Can complete the target session in a cool environment without setting off a flare, on 2 separate occasions
No flare in the 24 hours after a session, on 2 consecutive sessions
You can say, unprompted, which cooling methods are safe and which are not

Trust anchor

Conviction

Low — endpoint-stratified, which means some parts of this page are far more certain than others.

What we are confident about is mostly the harm and the negatives: that unsafe cooling causes serious tissue injury HIGH, and that a skin biopsy does not confirm this diagnosis HIGH IN THE NEGATIVE. Moderately confident: heat and exercise as the dominant triggers, the substantial impact on daily life, and that the gene explains only a minority of cases. Weakly supported: every drug and every procedure LOW. And no evidence at all for any exercise or physical therapy protocol, or for any recovery timeline quoted without naming which form of the condition you have NO EVIDENCE.

The reason the overall grade is low is simple arithmetic: the entire comparative treatment literature for this condition is 6 studies and 120 patients, and the entire randomised treatment evidence base is two trials totalling 16 people.

What would change my mind — "exercise must be avoided"

What would change this: a randomised trial of at least 60 patients comparing a temperature-controlled graded activity programme against usual care over 12 weeks, measuring weekly attack frequency and a skin-damage count. The 78% trigger figure is a yes/no report, not a dose curve — nobody has tested whether the problem is movement itself or the heat that movement generates.

What would change my mind — "cooling is the main preventable harm"

What would change this: a prospective study of 100 consecutive patients measuring cooling behaviour and skin outcomes at 12 months. The 60% figure comes from a review of children with the inherited form, which is the severe end. A general-population figure could be much lower — though the four independent reports of amputation, bone infection, necrosis and hypothermia would still stand.

The Full Picture

The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Dark cinematic rendering of small blood vessels and nerve fibres in the foot

Two different processes produce the same picture, and telling them apart matters more than the label does.

The nerve route. Changes in a gene called SCN9A alter a sodium channel that sits on the pain-sensing nerves of the feet and hands. The channel fires too easily, so ordinary warmth generates a pain signal. Here is the correction that matters: in the only study that screened an unselected group of patients, this gene explained just 4 out of 34 people — about 12%. Calling this "a sodium channel disease" describes roughly one patient in nine.

The plumbing route. The skin of the feet contains shunts that let blood bypass the fine capillary network — they exist for temperature control. In this condition they appear to open when they should not, so blood rushes through the foot without properly feeding the tissue. That is why the foot can look florid and hot while the tissue underneath is short of oxygen.

One measured finding cuts against the obvious assumption: when 27 patients were compared against 25 healthy people, their heat pain thresholds were completely normal — even in those carrying the gene change. The difference is in how strong signals get processed, not in where the threshold sits. So a patient in agony from a warm room is not showing a pain threshold you could measure in clinic.

How to Identify It

Dark cinematic study of the foot and ankle in low light

There is no test that confirms this. Diagnosis is clinical, and no test in this condition has a published figure for how well it catches or rules out the disease. Those numbers are absent from the literature, not from this page.

Skin biopsy for nerve fibre density Sn: no published figure | Sp: no published figure — the test most likely to be ordered and the one most likely to be normal. In 52 consecutive patients, only 5 (10%) had reduced nerve fibre density. Do not order it to confirm this.
Function-based nerve testing (sweat response, heart rate and blood pressure control) Sn: no published figure | Sp: no published figure — the informative one. In the same 52 patients: 60% had an abnormal sweat test, 42% abnormal pain thresholds, 38% abnormal blood pressure or heart rate control. The authors' own conclusion is to test function rather than structure.
Gene testing for SCN9A Sn: no published figure | Sp: no published figure — a negative result rules out very little, since only about 12% of unrelated patients carry a causal variant. A positive result does carry weight when procedures are being considered, because relapse after sympathectomy clustered in gene-positive patients.
Full blood count — not a physical test, but the one investigation no adult with new symptoms should skip.

What actually makes the diagnosis is the pattern: episodic burning with visible redness and warmth, brought on by heat and relieved by cold. A skin infection does not settle when you elevate and cool the foot. Poor arterial circulation gives a cold, pale, pulseless foot that is worse on elevation — the exact inverse. And Raynaud's does not exclude this: it was present in 80% of one 46-patient group.

The Debate

No treatment guideline exists for this condition — the most authoritative review says so in as many words — so the disagreements below are study against study rather than guideline against trial.

Is cooling the treatment, or the injury?

Every descriptive source, 1878 to now

Relief with cooling is the defining feature of the condition and the patient's most effective self-treatment.

vs

Arthur 2019; Chen 2026; Cohen 2026; Tham 2017

Cold-immersion complications in 60% of children with the inherited form. Amputation after ice packs. Bone infection after air-conditioning cooling. Three weeks of dangerously low body temperature from over-cooling.

Both are right, and it is a dose problem being read as a mechanism. Brief cooling relieves; immersion, ice and sustained cold injure. The literature says "cautious cooling" and the qualifier is the part that does not survive the consultation. Teach the distinction explicitly, in writing.

Does the sympathectomy success rate mean what it sounds like?

Wang 2018, 13 patients

69% achieved complete relief one week after chemical lumbar sympathectomy, with follow-up averaging over six years.

vs

Lee 2024, narrative review

Evidence for every procedure is "extremely limited", mostly case reports and small series, with both successes and failures reported.

The success rate and the catastrophe rate come from the same 13 people: among the 4 who did not respond, one died at 3.5 months and one had an amputation at 4 months. Quoting 69% without that denominator is not reporting the study, it is selecting from it.

Is aspirin the answer?

Widespread clinical shorthand

"Aspirin is the treatment for erythromelalgia."

vs

Landolfi 1996; Sun 2023; Mirhosseini 2024

The evidence sits specifically in blood disorders with high platelet counts. In the largest group of children, 15 of 42 were on it and no medicine was consistently effective. Gangrene has occurred despite daily aspirin.

A treatment for the secondary form generalised onto the primary form. Aspirin is close to specific for the blood-disorder subtype, and it is not established for the primary condition.

Honest Limitations

The whole comparative treatment literature is 120 patients

A formal systematic review screened 103 papers and included 6, covering 120 people across six different treatments. The largest single group in it is a retrospective look back at 36 charts. Every treatment statement in this condition rests on a base that would count as an underpowered pilot anywhere else.

How to adjust: treat every option as a trial of one, with an agreed review date and a stopping rule decided in advance.

Every group studied is from a specialist referral centre, so the literature describes the sickest people

The natural-history data is one quaternary centre between 1970 and 1994. The children reached their series after a median of three specialists. The mild end of this condition is invisible to research.

How to adjust: do not hand a newly-diagnosed person the referral-centre prognosis. Say the honest thing instead, which is that we do not know where on the spectrum they sit.

Nothing in this literature was designed to answer a physical therapy question

The research is dermatological, neurological, blood-related, anaesthetic and vascular-surgical. Exercise appears in it only as a trigger to avoid, and as the tool researchers use to bring on an attack on purpose. There is no rehabilitation outcome anywhere in this condition — no trial, no cohort, no case series.

How to adjust: say so out loud, to the patient and in the notes. The profession's default answer here has never been tested, and the one thing we do know about exercise in this group is that it brings on the symptom in 78% of them.

The Nuance

Dark cinematic contrast study of two lower limbs

There is no surgery to weigh against conservative care here, which makes this section shorter and stranger than it is for most conditions. No operation is established, and no study has ever compared a surgical against a conservative pathway. Conservative management is not the cautious option; it is the only characterised one.

The nuance that actually changes what happens to a patient is about which condition they have, because "erythromelalgia" covers at least three different stories:

The adult, gradual, often secondary form. In a referral-centre group of 168 followed for an average of 8.7 years, 31.9% got worse, 26.6% stayed the same, 30.9% improved and 10.6% resolved completely — with survival lower than expected for their age and sex. This is the population most published prognosis figures come from.
The inherited, early-onset form. Onset before age 10 predominates. Pain is often stubborn against multiple medicines, and this is the group in which cooling injuries are most documented.
The sudden, post-viral form in children. Five Swiss children aged 6 to 11 presented with exactly this picture, were diagnosed with immune-mediated small-fibre nerve damage, and all five recovered within six months and stayed well for years. Simple painkillers did nothing; medicines aimed at nerve pain, sometimes with immune treatment, did.

Quoting a prognosis without naming which of these you mean is not a rounding error — the answers point in opposite directions.

One more piece of nuance worth flagging for anyone reading the research directly: a body of work on "epidemic erythromelalgia" describes outbreaks among Chinese middle-school students, occurring in February and March, with fever. That is a different illness that happens to share the name. Its numbers do not transfer to the condition on this page.

Sources

Davis MD, et al. (2000) — Arch Dermatol, n=168. Natural history; survival significantly lower than expected (P<.001); at a mean 8.7 years, 31.9% worse and 10.6% fully resolved. PMID 10724194
Reed KB, Davis MD (2009) — JEADV, n=33. Population incidence 1.3 per 100,000 per year (95% CI 0.8–1.7). PMID 18713229
Arthur L, et al. (2019) — J Pediatr. Systematic review, 25 children across 28 publications; cold-immersion complications in 60%. PMID 30416015
Algarni AS, et al. (2025) — Medicina. Systematic review; the entire comparative treatment base is 6 studies and 120 patients. PMID 40428878
Mantyh WG, et al. (2017) — JAMA Dermatol, n=52. Only 10% had reduced nerve fibre density; 60% had abnormal sweat testing. PMID 27926760
Skystad Kvernebo M, et al. (2025) — Acta Derm Venereol. SCN9A variants in only 4 of 34 unrelated patients (11.7%). PMID 39850993
Parker LK, et al. (2017) — Clin Exp Rheumatol, n=46. Triggers: heat 85%, exercise 78%, evening 76%; Raynaud's in 80%. PMID 27494156
Kalgaard OM, et al. (2003) — Acta Derm Venereol, n=12. The only placebo-controlled parallel-group trial in this condition. PMID 14690340
Chen S, et al. (2026) — J Med Case Rep. Gangrene and amputation attributed to ice-pack use. PMID 42231503
Cohen K, et al. (2026) — Pediatr Dermatol. Deep fungal skin infection and bone infection following cooling with window air conditioning. PMID 41084402
Faignart N, et al. (2020) — Eur J Paediatr Neurol, n=5. Sudden post-viral form; all five children recovered within 6 months. PMID 32723684
Ma JE, et al. (2023) — Mayo Clin Proc. States that no guidelines exist for the treatment of this condition. PMID 36470753
Wang WH, et al. (2018) — J Vasc Surg, n=13. Sympathectomy: 69.2% complete response at 1 week; one death and one amputation among the 4 non-responders. PMID 30126782
Helås T, et al. (2017) — Eur J Pain, n=27 vs 25 controls. Heat pain thresholds are not lowered in this condition. PMID 28418213
Caldito EG, et al. (2024) — J Am Acad Dermatol, two-part review. Unsafe self-cooling causes necrosis, infection and amputation. PMID 37364617, PMID 37364616

Go Deeper

Most advice about painful feet starts with "ice it." Sometimes that is the one thing that will hurt you. The Verdict reviews the evidence behind one condition or treatment every week — free.

Join The Verdict — free