Cellulitis and the Hot Swollen Leg

If the pain feels far worse than the leg looks, the redness is spreading while you watch it, or there are blood blisters, go to the emergency department now. Don't wait.

That combination points to a flesh-eating infection rather than ordinary cellulitis, and nothing done at a bedside can rule it out. Otherwise: a red, hot, swollen leg still needs a same-day medical appointment, and don't put a compression stocking on it until someone has looked.

Takes 30 seconds to check. No equipment needed.

The Verdict

A red, hot, swollen leg gets called an infection. About a third of the time, it isn't.

Think of a blocked drain versus a burst pipe. Both leave water on your floor, and from the doorway they look identical. An infection is bacteria getting in through a break in the skin and multiplying, which is the burst pipe. Far more often the leg is red and hot because the plumbing that drains fluid out of it has been failing slowly for years, and the skin is inflamed from the backed-up fluid rather than from any bug. Antibiotics fix the burst pipe. They do nothing for the blocked drain, which is why that second group stays red no matter how many courses they take.

Here's what's really happening: when these patients were sent for a specialist second opinion, the label was wrong in roughly 39 to 41 out of every 100 cases, and about two thirds of the time there was no infection at all.
What most people get wrong: an infection nearly always starts at one break in the skin and spreads out from there, so it affects ONE leg. Both legs red at once is rarely an infection.
What to watch for: pain far worse than the leg looks, redness spreading while you watch, blood blisters, or a cold pale foot. Any of those is an emergency, not a wait-and-see.

Best for

Anyone with a red, hot, swollen leg who has already been medically assessed and wants to understand what it actually is and how to stop it coming back.

Skip if

Your leg is red and hot right now and nobody has looked at it. Get seen today. Nothing on this page is a substitute for that.

Want the full evidence? Keep scrolling

What Works

Dark cinematic study of lower limb circulation and tissue

The acute treatment for a genuine infection is antibiotics, prescribed medically. Physical therapy owns the conditions it gets confused with, and the prevention of the next episode.

1. Getting the diagnosis right before treating it STRONG

Specialist review changed the initial antibiotic plan in 47 to 96 percent of suspected cellulitis cases across 14 studies.

Honest caveat carried in full: the same review found no significant reduction in readmission at 30 days. The case for referral is diagnostic accuracy and avoiding pointless antibiotics, not readmission.

2. Antibiotics for genuine infection STRONG

Across 4 trials, no difference was found by drug type, route, dose or duration. The reviewers concluded lower-limb cellulitis may be routinely overtreated.

Medical decision, not a physical therapy one. Expect the largest improvement by day 3.

3. Preventive antibiotics for genuinely recurrent cases STRONG on treatment

Cochrane, 6 trials, 573 people: 69 percent fewer recurrences while taking them, with 6 people treated to prevent one episode.

The catch, from the same review: once stopped, the benefit is gone. Applies mainly to people with at least two episodes within three years.

4. Decongestive therapy and self-management for the lymphoedema group MODERATE

13 of 14 international guidelines recommend hands-on decongestive treatment followed by self-management: skin care, self-massage, exercise and compression.

Guideline consensus. The exercise dose is not specified at guideline level, so sets, reps and frequency below are convention, not trial-tested.

5. Compression for the vein group, once circulation is checked MODERATE

Lower-strength Class I stockings (18 to 26 mmHg) cleared fluid from the skin as well as stronger Class II (26 to 36 mmHg).

Only 11 patients, and a scan-based measure rather than a symptom one. It matters anyway, because the lower strength is far easier to actually wear.

Tier 3 — weaker evidence, recorded honestly

Treating the modifiable local risk factors after the episode settles EMERGING
The associations are strong and consistent: a previous episode raises the odds about 40-fold, an open wound 19-fold, a leg ulcer 14-fold, chronic swelling about 7-fold, athlete's foot about 3-fold. But every one of those comes from studies that looked backwards, and no trial has tested whether fixing them prevents anything. The Cochrane reviewers say so in their own conclusions: none of the prevention trials studied swelling reduction or skin care.

Cycloidal vibration during the acute episode EMERGING
66 percent recovered within 7 days versus 11 percent with standard care. One small unblinded trial of 36 people from 2007, never replicated in the 19 years since. Recorded because it is the only physical-therapy-adjacent trial in this whole literature, not because it should change what you do.

Exercise Prescription

This applies after the acute episode has settled and a doctor is happy. During the acute phase the correct physical therapy input is none at all: no compression, no massage, no vigorous mobilization on an undiagnosed hot leg.

Said plainly, because it is the honest position: we know which targets are the right ones. We do not have trials proving that hitting them prevents the next episode. Everything below is low-risk and aimed correctly, and that is the strongest claim the evidence supports.

Daily skin and toe-web check — look at the whole lower leg and between every toe, wash and dry thoroughly. Every day
The most-missed item on the list. Athlete's foot roughly triples the odds of a leg infection.

Antifungal treatment for athlete's foot — full course on the label, not until it looks better. As directed (dose not trial-tested for preventing recurrence)

Ankle pumps — leg supported, point the toes away then pull them firmly up. 20 reps × several times daily (not trial-tested)
Works the calf pump, which is what moves fluid out of the leg.

Walking — steady and comfortable on the flat, built up gradually. Toward daily (not trial-tested)
Calf muscle action is the pump. Sitting still and standing still are both the enemy here.

Elevation — leg above hip height while resting. 20–30 min × 2–3 daily (not trial-tested)

Class I compression — 18 to 26 mmHg, on first thing before the leg swells. Daily, if fitted
Only after circulation has been checked.

What Doesn't Work

Using a high ALT-70 score to confirm infection. When it says yes, it is right only about 22 times in 100. It is a rule-out tool that fails in exactly the direction people reach for.
Antibiotics for the two thirds of cases that aren't infections. The drug can't work, and the real problem goes untreated. Roughly $515 million a year of avoidable US spending sits downstream of this.
Treating leftover redness at 2 weeks as failure. About 4 in 10 people still have redness and swelling then. Escalating antibiotics at that point is treating the normal course of recovery.
Judging a swelling treatment by tape measure alone. Compression pulled measurable fluid out of the skin with no change in ankle size at all. A tape measure would have scored a working treatment as a total failure.
Compression, massage or hands-on work on an undiagnosed hot leg. Right answer for two of these conditions, wrong answer before a blocked artery and a clot have been ruled out.
Blaming diabetes, smoking or alcohol. None of the three was linked to this in the meta-analysis. The drivers are local and mechanical.

Red Flags

Dark cinematic anatomical study of the lower limb

Go to the emergency department now

Pain far out of proportion to how the leg looks, rapidly spreading redness, blood blisters, crackling under the skin, numb skin, or feeling very unwell. This is a flesh-eating soft tissue infection until proven otherwise. No bedside test can exclude it, including the score named after it. Reaching surgery within 6 hours of arrival roughly halves the death rate.
Cold, pale or blotchy limb, no pulse in the foot, pain at rest. A blocked artery.
A tense, exquisitely painful compartment with pain when the muscle is stretched.
New calf swelling and pain with any clot risk factor. Clots and skin infection are not alternatives: about 3 in 100 people with confirmed cellulitis also have a clot.

Same-day medical assessment

Fever, shivering or confusion alongside a red limb.
A hot, swollen single JOINT with fever. Do not mobilize it.
A hot swollen foot in someone with nerve damage from diabetes.
Any red, hot, swollen limb that has not been medically assessed. Whether this is an infection is not a physical therapist's call to make, in either direction.

Refer to: Emergency department for the first four. Same-day GP or urgent care for the rest. Vascular assessment before any compression goes on a limb with unverified pulses.

Return to Training

Applies to the affected leg after the episode. Rebuild starting with walking, because the calf pump clears the swelling faster than any strengthening exercise will.

Medically cleared, with the diagnosis confirmed or the alternative diagnosis identified.
Redness fully resolved, or a stable remnant that has not advanced for 7 consecutive days.
No warmth compared with the same spot on the other leg.
Swelling stable or resolved, with no overnight increase after activity, for 7 consecutive days.
Any wound, ulcer or toe-web break healed or actively being treated.
30 minutes of continuous walking with no swelling the following morning.
Full pain-free ankle and knee movement on the affected side.

Regression trigger: any return of redness or warmth, or a step up in swelling that does not settle overnight with elevation. Drop back a stage and re-check the skin for a new break. Lifters: heavy standing work worsens dependent swelling, so reintroduce it last and keep the leg elevated between sets until then.

Conviction

MODERATE Endpoint-stratified across 8 rows, because the evidence here is genuinely uneven.

Claim	Conviction
Cellulitis is misdiagnosed in roughly one third to two fifths of cases, and most alternatives are not infections	HIGH
No validated bedside test or checklist exists for lower-limb cellulitis	HIGH
ALT-70 works to argue against infection and fails to confirm it	MOD–HIGH
Clot risk in confirmed cellulitis is low but not zero	MODERATE
Preventive antibiotics reduce recurrence while being taken, after 2+ episodes	MODERATE
That protection does not persist after stopping	LOW–MOD
Treating swelling, skin breaks and athlete's foot reduces recurrence	LOW
Any specific physical therapy exercise protocol for this condition	NO EVIDENCE

What would change my mind on the diagnosis claim

A study of at least 500 consecutive patients with a hot swollen leg in community and primary care rather than emergency departments, with the reference standard fixed in advance (30-day outcome plus specialist review, not specialist opinion alone). If the ALT-70 score's ability to correctly identify non-infections rose above 60 percent in that setting, it would become a real discriminator rather than a rule-out, and the "one in three misdiagnosed" figure would need re-anchoring to something other than another clinician's opinion.

What would change my mind on the treatment claim

A trial of at least 300 adults with two or more leg episodes in three years, comparing physical-therapy-led risk-factor treatment (compression fitted to a measured circulation test, a supervised calf-pump and walking program with stated doses, antifungal treatment, structured skin care) against continuous preventive antibiotics and against both together, with recurrence at 24 months as the endpoint. This is the trial the 2017 Cochrane review asked for in its own conclusions, and nobody has run it.

The Full Picture

The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Dark cinematic anatomical study of skin, subcutaneous tissue and lymphatic drainage

Cellulitis is a bacterial infection of the deep skin and the fatty layer beneath it, usually caused by streptococci or ordinary staph. Bacteria get in through a break in the skin barrier, which is why the risk factors are overwhelmingly local: a wound, an ulcer, scratched skin, or the soggy skin between the toes in athlete's foot. The immune response produces the four familiar features, redness, warmth, swelling and tenderness.

The hard part is that there is no test for it. Skin swabs rarely grow anything useful, no blood test confirms it, and no scan shows it. It is diagnosed by a clinician looking at a leg and deciding. One systematic review went looking specifically for any validated checklist or tool for lower-limb cellulitis, found eight candidate studies, and every single one was at high risk of bias with results too inconsistent to combine.

Why the impostors are so convincing: all four features are produced by non-infective processes too. Failing vein valves produce redness, warmth and swelling in stasis dermatitis. Failing lymphatic drainage produces swelling and skin thickening. Long-standing vein disease produces a hot, hard, tender lower leg. Eczema produces redness and heat with no bug involved at all. The endpoint is shared; only the cause differs.

The one asymmetry that helps. Infection starts at a single entry point and spreads outward, so it is unilateral. Vein and lymphatic disease is driven by gravity and circulation, so it is frequently bilateral. That is why one-leg-only carries more weight than any other single feature in the only prediction model that exists.

How to Identify It

Dark cinematic clinical study of lower limb examination

The first question is not a test. It is: one leg, or both? It changes the differential more than anything else you will do.

The two instruments that exist both have the same shape, and it is worth being precise about it:

ALT-70 prediction model Sn: 97% | Sp: 22%
One leg only (3 points), raised white cells (1), heart rate 90+ (1), age 70+ (2). A score of 0 to 2 means pseudocellulitis is 83%+ likely. Two of the four items need a blood test and a recorded pulse, so most community clinicians can only compute a partial score.
Thermal imaging Sn: 97% | Sp: 38%
Affected limb peak 33.2 °C versus 31.2 °C, a real 2.0 °C difference. Needs a thermal camera.
Both combined Sn: >90% | Sp: 54%
The best specificity anywhere in this literature, and still barely better than a coin toss.

All three are sensitive and non-specific, which makes them rule-out tools. A low score is real information. A high score is close to worthless. The original derivation paper reported that 5+ points meant 82% likelihood of true infection, and that figure isn't wrong: it is a predictive value from a population where infection was common. Predictive values move with how common the disease is in the room. The test's own accuracy does not. Use the score downward, never upward.

The Debate

Is the ALT-70 score a discriminator or a rule-out?

Raff 2017, derivation, 259 patients

5 or more points gives 82.2% likelihood of true cellulitis. Presented as a way to tell the two apart.

vs

Pulia 2024, independent validation, 204 patients

Specificity 22.0% (95% CI 15.8–28.1), with sensitivity above 90%.

Both are correct and they measure different things. Predictive values move with prevalence; sensitivity and specificity do not. Follow the validation: use low scores to argue against infection, never high scores to argue for it.

Do preventive antibiotics work, or not?

Cochrane 2017, 6 trials, 573 people

69% fewer recurrences while taking them. Moderate certainty. 6 treated to prevent one episode.

vs

PATCH II 2012, 123 people

Not statistically significant (P=0.08). 20% recurrence on penicillin versus 33% on placebo.

Not a real conflict. PATCH II enrolled a group in which 79% had suffered only ONE prior episode, and Cochrane states its effects apply mainly to people after at least two episodes within three years. They agree once you read the entry criteria.

Guideline recency flag: the governing guideline for skin and soft tissue infections here is the IDSA 2014 update, twelve years old. No NICE, APTA, BOA, EULAR, ACR or JOSPT guideline addressing this differential appeared anywhere in 135 papers, and none at all addresses a physical therapy role in it.

Honest Limitations

The prediction model was born in a room you are not standing in

The research: ALT-70 catches over 90% of true infections.
The gap: it was built and tested only in emergency departments, where 45% of the red legs turned out to be infections. In a community caseload where most red legs are venous, the same score means something different. And two of its four items need a blood count.
The adjustment: use the two available items as a structured prompt, not a score, and record that you computed a partial model.

The prevention trials never tested what physical therapy actually does

The research: preventive antibiotics cut recurrence by 69% while taken.
The gap: the Cochrane authors write that none of the studies investigated swelling reduction or proper skin care. Meanwhile the risk-factor review shows the modifiable targets are almost all local and mechanical, which is exactly the territory physical therapy works in.
The adjustment: treat swelling, skin integrity and toe-web hygiene, and say out loud that the recurrence benefit is inferred from association data rather than demonstrated. Don't promise fewer episodes.

Your idea of recovered and the patient's are two weeks apart

The research: at day 14, redness persists in 41% and swelling in 37%.
The gap: in the same group on the same day, clinicians called 85.8% cured and patients called 52.8% cured. That gap is what drives repeat visits and requests for more antibiotics.
The adjustment: set the day-3 and day-14 expectations at the first contact, in actual numbers.

The Nuance

Dark cinematic anatomical study contrasting vascular and lymphatic structures of the leg

There is no surgery for ordinary cellulitis, so the usual surgery-versus-conservative question does not apply. Surgery enters this picture only where the diagnosis is not cellulitis: emergency debridement for flesh-eating infection, restoring blood flow for a blocked artery, releasing a compartment, draining an abscess.

The decision that does get made is whether to admit someone, and the numbers there are worth knowing. Pooled in-hospital death rate for cellulitis is 1.1% worldwide and 0.5% in the USA across 18 studies, and only about a third of those deaths looked attributable to the infection at all. The authors point out that is comparable to the low-risk pneumonia that specialist societies already recommend treating at home. Uncomplicated cellulitis in an otherwise well person is not, on these numbers, a dangerous disease. The danger in this presentation lives in what it can be mistaken for, in both directions.

The measurement trap, because it is the most likely way a correct treatment gets abandoned. When researchers scanned the skin before and after compression in vein disease, they found a 17% reduction in fluid within the skin and no change in ankle circumference whatsoever. If you judge that treatment with a tape measure, you conclude it failed and stop. Judge it on how the leg looks and feels.

And the risk factors are not the ones you would guess. Diabetes, smoking and alcohol showed no association at all. What did: a previous episode (about 40× the odds), an open wound (19×), a leg ulcer (14×), chronic swelling (about 7×), scratched or broken skin (4×), athlete's foot (3×), and a BMI over 30 (about 2.4×). Local and mechanical, top to bottom.

Sources

Nightingale et al., 2023 — J Gen Intern Med. Systematic review and meta-analysis, 9 studies, 1,600 patients. Misdiagnosis 41% (95% CI 28–56); 54% of misdiagnoses were stasis dermatitis, eczematous dermatitis or swelling/lymphoedema. PMID 37231210
Cutler et al., 2023 — J Hosp Med. SR+MA, 858 inpatients. 39% (31–47) given an alternative diagnosis; 68% of alternatives non-infective. States there is no gold standard test. PMID 36189619
Pulia et al., 2024 — JAMA Dermatol. Prospective validation, 204 patients. ALT-70 specificity 22.0% (15.8–28.1); thermal imaging 38.4%; combined 53.9%. PMID 38536160
Raff et al., 2017 — J Am Acad Dermatol. ALT-70 derivation, 259 patients, 30.5% misdiagnosed. PMID 28215446
Patel et al., 2019 — Br J Dermatol. Systematic review: zero validated diagnostic criteria or tools for lower-limb cellulitis; all 8 candidate studies at high risk of bias. PMID 30844076
Dalal et al., 2017 — Cochrane Database Syst Rev. 6 RCTs, 573 people. Prophylaxis RR 0.31 (0.13–0.72), NNT 6; after stopping RR 0.88 (0.59–1.31). PMID 28631307
Quirke et al., 2017 — Br J Dermatol. SR+MA of risk factors. Previous episode OR 40.3, wound 19.1, ulcer 13.7, lymphoedema 6.8, athlete's foot 3.2, BMI>30 2.4. Diabetes, smoking, alcohol not associated. PMID 27864837
Gunderson & Chang, 2013 — Thromb Res. SR+MA, 1,054 patients. Clot prevalence 3.1% (1.9–4.9); retrospective 1.1% versus prospective 7.8%. PMID 23948644
Nightingale et al., 2024 — Emerg Med Australas. Prospective cohort, 300 patients. Day 14 redness 41%, swelling 37%; cure clinician 85.8% versus patient 52.8%. PMID 38481041
Gniadecka et al., 1998 — J Am Acad Dermatol. Cross-over, 11 patients. Class I 17% versus Class II 14% dermal fluid reduction, no difference; no ankle circumference change. PMID 9843010
Torgbenu et al., 2023 — Adv Wound Care. Systematic review of 14 lymphoedema guidelines. PMID 35196892
Gunderson et al., 2018 — J Gen Intern Med. SR+MA, 18 studies. Inpatient mortality 1.1% worldwide, 0.5% USA. PMID 30022408
Lin et al., 2024 — Arch Dermatol Res. SR+MA, 14 studies. Dermatology consultation changed the plan 47–96%; readmission OR 0.56 (0.25–1.25). PMID 39614879
Stevens et al., 2014 — Clin Infect Dis. IDSA skin and soft tissue infection guideline. The governing guideline, twelve years old. PMID 24973422

Go Deeper

Most health advice tells you what a condition is. Almost none tells you how often the diagnosis itself is wrong. The Verdict reviews one condition or supplement a week, evidence-scored, free.

Join The Verdict — free

A red, hot, swollen leg gets called an infection.