Necrotizing Fasciitis

If a limb hurts far more than it looks like it should, and it is spreading over hours rather than days, go to the emergency room today.

Mark the edge of the redness with a pen and note the time. Spreading you can actually see within a few hours is the sign that matters. A normal temperature does not clear you: more than half of people with this condition have no fever when they arrive.

Takes less than a minute. A pen is the only equipment.

The Verdict

The blood test named after this infection cannot tell you that you do not have it.

Think of the skin as carpet and the tissue underneath as the floorboards. This infection runs along the floorboards, where the blood supply is poorest and nothing can wall it off, while the carpet on top still looks almost fine. The pain comes from the damage underneath, so it always runs ahead of what you can see, and by the time the carpet is visibly ruined the floor below is already gone.

Here is what is really happening: the infection travels along a deep tissue layer, so the pain is always worse than the visible skin, and that gap between how it feels and how it looks is the single most useful warning sign.
What most people get wrong: they wait for a fever or for the skin to look dramatic, and neither is reliable, because fever is missing in over half of cases and the obvious skin changes are a late sign.
What to watch for: pain out of proportion, spreading measured in hours, feeling genuinely unwell in yourself, dark or blood-filled blisters, or getting worse while already on antibiotics.

Best for

Anyone with a painful, swollen or red limb trying to work out whether this is an emergency, and clinicians who need to know what the available tests can and cannot rule out.

Skip if

You are looking for a home treatment or rehabilitation program for the infection itself. There is not one. The treatment is surgery.

Want the full evidence? Keep scrolling

Dark cinematic surgical and tissue imagery

What Works

Emergency surgical debridement, as early as possible HIGH

Surgery within 6 hours of presentation is associated with 19% mortality against 32% after 6 hours, odds ratio 0.43 (95% CI 0.26 to 0.70). The 12 hour comparison holds at 0.41 (0.27 to 0.61).

Evidence: STRONG for direction, MODERATE for size. Meta-analysis of 109 studies and 6051 patients (Nawijn 2020), built from observational data.

Referral on clinical suspicion alone HIGH

Without waiting for blood tests or imaging. This follows directly from the accuracy data: every available test has sensitivity too low to exclude the diagnosis, and the authors of the largest analysis say so explicitly.

Evidence: STRONG. Fernando 2019, 23 studies, n=5982.

Intravenous broad-spectrum antibiotics and resuscitation MODERATE

Alongside surgery, never instead of it.

Evidence: STRONG by consensus, weak by trial. Universal across guidance, but the Cochrane review found no usable randomised evidence for the core management questions (Hua 2018).

Exercise Prescription

An honest note before the table. There are no research studies testing rehabilitation after this condition. Everything below is standard practice borrowed from recovery after other major surgery and burns, and every dose is marked not trial-tested. Your surgical team overrides all of it, and none of this applies while a limb is still under investigation.

Post-survival rehabilitation NO EVIDENCE

Zero interventional studies exist across 59 papers. This is convention, not evidence, and it is labelled that way deliberately.

Gentle joint movement, especially the joint below the affected area. Slowly move through comfortable range. 10 slow repetitions each direction, 4 to 6 times daily (not trial-tested). Stretching and pulling is expected; stop at sharp pain or anything tugging a healing wound or graft.

Positioning and elevation. Rest the limb above heart level. 15 to 20 minutes, several times daily (not trial-tested). Should feel comfortable.

Isometric holds. Tighten the muscle without moving the joint. 5 second hold, 10 repetitions, 2 to 3 times daily (not trial-tested). Used early when wounds or grafts restrict movement. Effort without pain.

Graded strengthening, once wounds and grafts allow. Start at bodyweight or a light band. 2 to 3 sets of 10 to 15, 3 times weekly (not trial-tested). Effort and fatigue are fine; a wound that reacts means back off.

Walking and general conditioning. 5 to 10 minutes building gradually, daily (not trial-tested). Often the real limiter is general stamina after critical illness rather than the limb itself.

Deliberately not on this list: any exercise on a limb still under investigation, any loading before the surgical team clears it, and aggressive stretching to chase range of motion.

Tier 2 and Tier 3 in full

Negative pressure wound therapy after debridement MODERATE

Mortality odds ratio 0.27 (95% CI 0.09 to 0.87), p=0.03, but from only 7 controlled trials and 249 participants, with debridement count, length of stay and complications all flat. Level III by the authors' own grading. Treat the size of the effect with suspicion and read it as direction only (Zhang 2023). A surgical decision, not a physical therapy one.

Flap reconstruction for defects unsuitable for grafting MODERATE

888 flap closures in 733 patients, mostly local fasciocutaneous flaps, with partial or complete flap loss in 3.3% (Somasundaram 2021). Relevant to rehabilitation only in that it sets what tissue you are loading and what the restrictions are.

Psychological and appearance-related support EMERGING

Better supported than the physical side, which is the surprise. Confidence with appearance was the only significant predictor of mental quality of life in the final model (Gawaziuk 2018, n=56), and survivor interviews found psychological, relationship and employment effects that no conventional outcome measure captured (Hakkarainen 2014). No intervention has been trialled.

What Doesn't Work

Using the LRINEC score, or any bedside sign, to rule this out. The score is named for the disease and the largest meta-analysis of it says in its own abstract that it should not be used to exclude the diagnosis. Two of the three meta-analyses nonetheless frame it as a rule-out. This is the single most dangerous idea in the condition's literature.
The reassessment appointment. "Let's see how it looks in the morning" has the worst risk-benefit ratio available here, because the clock that predicts survival is already running and is measured in hours.
Waiting for a fever. Absent in over half of cases, sensitivity 46.0%.
Waiting for gas on X-ray or visible skin death. Gas is visible in 24.8% of cases and skin death is a late sign of a process that is already extensive.
Any hands-on treatment, exercise or loading on a limb under suspicion. There is no version of this where local treatment helps, and several where it costs hours.
Chasing range of motion as the main rehabilitation goal after survival. Four separate cohorts failed to link joint range to how these patients actually do, while the size of the surgery and appearance-related distress both tracked with it.

Dark cinematic anatomical rendering of soft tissue layers beneath the skin

Red Flags

Go to the emergency department the same day for any of these.

Pain grossly out of proportion to what is visible
Redness or swelling advancing over hours rather than days
Feeling systemically unwell: fever, shivering, confusion, faintness. Fever is absent in over half of cases, so its absence proves nothing
Blood-filled or dark blisters, or skin turning grey, purple or mottled
Crepitus, a crackling feeling under the skin
Numbness over an area that is severely painful
Getting worse despite broad-spectrum antibiotics
Low blood pressure, which is a very late sign

Refer to: emergency department immediately, by the fastest route, using the words “I am concerned about a necrotizing soft tissue infection”. Not a routine appointment, not a review next week, and not after blood tests you arranged yourself. Surgery within 6 hours of arrival is associated with 19% mortality against 32% after 6 hours.

Return to Training

No validated return-to-activity criteria exist for this condition. Zero studies. This is clinical convention offered as a structure, not a validated standard, and every item requires surgical team agreement.

Surgical and wound care team has formally cleared loading of the affected area
All wounds and graft sites fully healed, with no discharge, breakdown or signs of infection
No systemic symptoms for at least 4 consecutive weeks (not trial-tested)
Maximal achievable range at the joint below the injury, accepting that a permanent deficit is common
Able to complete 3 sets of 10 of the primary movement at bodyweight without wound reaction (not trial-tested)
No increase in limb swelling 24 hours after a session (not trial-tested)
General conditioning rebuilt for the intended activity, since deconditioning from critical illness usually outlasts the local problem
Scar and graft tissue tolerant of the friction, pressure and shear the activity will apply
Patient honestly briefed that a permanent range or strength deficit is common and does not necessarily limit function

Conviction

MODERATE

Endpoint-stratified. HIGH that no bedside instrument or laboratory score can safely exclude the diagnosis, and that early surgical debridement reduces mortality in direction. MODERATE for the size of the timing effect, for clinician delay being the modifiable interval, and for the long-term quality of life burden. LOW for the distal versus proximal joint range finding, a single cohort of 21 assessed patients, never replicated. NO EVIDENCE for any rehabilitation protocol, dose or timeline, and for hyperbaric oxygen.

What would change my mind: the diagnostic claim

A prospective cohort of 2000 or more consecutive patients presenting to primary care and community musculoskeletal services with an acutely painful, swollen limb, recording pain out of proportion as a pre-specified structured item with an explicit definition, and following everyone to 30 days for surgically confirmed infection. A negative likelihood ratio below 0.1 in that population would, for the first time, give a community clinician something that legitimately lowers suspicion. Anything above 0.3 would confirm what this page concludes, which is that nothing available at the bedside can lower it.

What would change my mind: the rehabilitation claim

A randomised trial of 120 or more limb infection survivors, enrolled at surgical discharge, comparing a structured graded rehabilitation program against usual care, with co-primary endpoints of physical quality of life and return to previous occupation at 12 months. A positive result would give this condition its first evidence-based rehabilitation content. Given how rare it is, this trial would have to be international and registry-embedded, which is a real reason it has not been done and not an excuse for pretending the evidence exists.

Go Deeper

Most health advice tells you what to do. Almost none tells you what the test actually proves, or what it cannot rule out. The Verdict reviews one condition or supplement a week, with the numbers and the honest gaps.

Join free

The Full Picture

The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Dark cinematic rendering of fascial tissue planes

Bacteria enter the tissue under the skin, often through a break so small the patient does not remember it, and spread along the fascia. Fascia has a poor blood supply compared with the skin above it and the muscle below it, which is what lets infection travel along that layer faster than the body can wall it off.

The skin signs are a late, secondary event. Infection clots the small vessels that feed the skin from below, and only then does the skin blister or turn dark. That single fact explains the entire diagnostic problem: the disease is always larger than it looks, and the skin is the last thing to declare it. Pain out of proportion is not a quirk of presentation, it is the direct consequence of destroying a tissue layer underneath skin that still looks nearly normal.

Microbiology splits the disease in two. Polymicrobial infection accounts for a pooled 52.2% and is mostly perineal. Monomicrobial accounts for 39.9% and mostly affects the limbs, which is the presentation a musculoskeletal clinician meets. The monomicrobial share has risen over two decades while polymicrobial has fallen, and Staphylococcus aureus is the commonest single organism (Dhanasekara 2022, 27 studies, 2242 patients).

How to Identify It

Dark cinematic clinical assessment imagery

There is no special test. That is the finding, not an omission.

FeverSn 46.0% | Sp 77.0%
Haemorrhagic bullaeSn 25.2% | Sp 95.8%
HypotensionSn 21.0% | Sp 97.7%
LRINEC score 6 or aboveSn 68.2% | Sp 84.8%
LRINEC score 8 or aboveSn 40.8% | Sp 94.9%
CTSn 88.5% | Sp 93.3%
Plain X-raySn 48.9% | Sp 94.0%
Point-of-care ultrasoundSn 85.4–100% | Sp 44.7–98.2%

Read the specificities rather than the sensitivities. Bullae and low blood pressure are close to diagnostic when present and tell you nothing at all when absent. They are rule-in findings being taught and used as rule-out findings. The ultrasound range is worth pausing on: a specificity floor of 44.7% comes from just 3 studies and 221 patients in total.

Pain out of proportion, the most cited early clue in the entire literature, has never been measured as a diagnostic test. There is no sensitivity or specificity for it anywhere in this evidence base. It is treated here as a referral trigger rather than a test, and the reason is a cost argument rather than an accuracy one: in a disease that kills roughly one in five, an unnecessary referral and a missed diagnosis are not remotely comparable errors.

The Debate

Three meta-analyses of the same score reach three different conclusions about which direction it works in.

Bechar 2017 (16 studies, n=846): the score is useful

Area under the curve 0.927, with mean scores of 6.06 in cases against 2.45 in non-cases. This measures how well the score separates two groups already known to have or not have the disease.

Fernando 2019 (23 studies, n=5982): the score must not be used to rule out

Sensitivity and specificity modelled at defined thresholds in patients clinically concerning for the diagnosis, which is the actual decision. Sensitivity 68.2% at the standard threshold and 40.8% at the stricter one. Larger, later, and it answers the question a clinician is actually asking.

Tarricone 2022 (12 studies, n=932): the score's best use is ruling out

On a negative predictive value of 89.99%, while reporting a pooled sensitivity of 49.39% in the same abstract. Negative predictive value depends on how common the disease is in the group tested and is being read as though it were a property of the test. At roughly one-third prevalence, 90% still leaves about 1 in 10 score-negative patients carrying a disease that kills 17.8 to 21%.

Guideline status. The most recent authoritative guideline covering this condition is IDSA 2014, which is 12 years old. No NICE, APTA, BOA, EULAR, ACR or JOSPT guideline specific to necrotizing fasciitis exists, and none at all addresses rehabilitation. The only newer document is a national expert consensus developed by discussion and voting (Zhou 2025), which is a consensus rather than a guideline and carries correspondingly less weight.

Honest Limitations

The accuracy numbers were measured in a room a community clinician is not standing in

Fernando 2019 enrolled adults already clinically concerning for the diagnosis, which is an emergency or surgical population where suspicion is high. A community clinician meets this earlier and at far lower prevalence, competing against ordinary skin infection and mechanical pain. At lower prevalence the same sensitivities perform worse, not better. Read every sensitivity on this page as a ceiling.

The timing evidence is observational and flatters early surgery

Nothing randomised time to theatre and nothing ever will. Patients operated on within 6 hours are partly patients who looked alarmingly sick, and looking alarmingly sick is itself prognostic. The direction is corroborated across 109 studies and the whole field agrees, but a halving of mortality is almost certainly optimistic as a causal estimate.

There is no rehabilitation evidence at all

Three separate literature searches were run to find it. A dedicated survivor-function search returned zero papers. A rehabilitation search returned 20 papers, of which 12 were case reports and none was a rehabilitation study. What exists is four small retrospective survivorship cohorts and one interview study, and not one of them tested an intervention.

The Nuance

Dark cinematic anatomical imagery of limb tissue layers

Speed buys survival, not the limb. No time variable in Nawijn 2020 reduced amputation, and amputation or disarticulation occurs in 22.3% of limb cases (Angoules 2007). Referring fast is still unambiguously the right call. It should not be sold to a patient or a family as limb-saving, because the evidence does not support that.

There is no conservative pathway. Surgical exploration is simultaneously the diagnostic gold standard and the treatment. A negative exploration is an acceptable outcome; a delayed positive one frequently is not. Survival runs 78 to 86% depending on cohort and era.

Two conditions share the cardinal sign. Acute compartment syndrome also produces pain out of proportion, and favours recent trauma, a fracture, a cast or a tight dressing, with pain on passive stretch and a tense compartment but without spreading redness or systemic toxicity. The distinction is not the physical therapist's to make and does not need to be made before referring, because both are surgical emergencies on an hours-long clock. Ordinary skin infection is the commoner and more dangerous confusion: initial misdiagnosis occurred in almost three-quarters of 1463 published cases (Goh 2014).

After survival, the rehabilitation instinct is aimed at the wrong target. In the only study that measured joint range and quality of life in the same patients, the two were unrelated, while the size of the excision and the length of stay did track with quality of life (Chevet-Noël 2020). Joints below the injury lost a mean 37.65% of flexion and 48.6% of extension against the uninjured side, against 9.19% and 5.0% above it. Gawaziuk 2018 separately found that how people felt about their appearance predicted mental wellbeing while the size of the affected area predicted physical wellbeing, and Brengard-Bresler 2017 found the number of operations tracked with both. Measure range, because a stiff joint is real and worth treating. Do not let it be the scoreboard.

Sources

Fernando et al., 2019, Annals of Surgery. Systematic review and meta-analysis, 23 studies, n=5982. Diagnostic accuracy of examination, imaging and LRINEC. States the score should not be used to rule out. TIER 3
Nawijn et al., 2020, World Journal of Emergency Surgery. Systematic review and meta-analysis, 109 studies, 6051 patients. Surgery within 6 hours, odds ratio 0.43. Patient delay not significant. TIER 3
Goh et al., 2014, British Journal of Surgery. Systematic review of 9 case series, n=1463. Initial misdiagnosis in almost three-quarters; gas on X-ray in 24.8%. TIER 4
Tarricone et al., 2022, Journal of Foot and Ankle Surgery. Meta-analysis, 12 studies, n=932, limb-specific LRINEC performance. TIER 3
Bechar et al., 2017, Annals of the Royal College of Surgeons of England. Meta-analysis, 16 studies, n=846. The opposing conclusion on the same score. TIER 4
Chevet-Noël et al., 2020, Annales de Chirurgie Plastique et Esthétique. Cohort, 21 survivors assessed. The only goniometric data in the literature. TIER 5
Gawaziuk et al., 2018, JPRAS. Cohort, n=56 survivors. Quality of life below population norms in both physical and mental domains. TIER 5
Hakkarainen et al., 2014, Journal of Surgical Research. Qualitative interviews, n=18 survivors. TIER 5
Angoules et al., 2007, Injury. Systematic review, n=451 limb cases. Amputation 22.3%, mortality 21.9%. TIER 4
Dhanasekara et al., 2022, Surgical Infections. Meta-analysis, 27 studies, 2242 patients. Microbiology and mortality trends. TIER 3
Hua et al., 2018, Cochrane Database of Systematic Reviews. No usable randomised evidence for core management. TIER 2
Levett et al., 2015, Cochrane Database of Systematic Reviews. Zero eligible trials of hyperbaric oxygen. TIER 2
Zhang et al., 2023, Journal of Orthopaedic Surgery and Research. Meta-analysis, 7 trials, n=249. Negative pressure wound therapy. Level III. TIER 4
Stevens et al., 2014, Clinical Infectious Diseases. IDSA practice guideline, the most recent authoritative guideline for this condition, 12 years old. TIER 1
Zhou et al., 2025, Burns & Trauma. Expert consensus by discussion and voting, not a systematic review. TIER 4