Legionnaires' Disease (Legionella)
The atypical pneumonia that beta-lactams do not touch. Legionella survives inside alveolar macrophages, where ceftriaxone cannot reach it, so a patient on ceftriaxone monotherapy is receiving no active drug at all -which is why the macrolide or fluoroquinolone in an empiric CAP regimen is coverage, not a courtesy. Three things carry this page: a negative urine antigen does not exclude it, send a lower respiratory culture alongside the urine antigen, and report every case, because the diagnosis is also a public health event pointing at a contaminated water system.
🔍 Approach
⚠ The error that kills is dropping the atypical agent. Empiric inpatient CAP is ceftriaxone PLUS azithromycin, and the second drug is the only one with any activity against Legionella. When azithromycin gets stopped for a borderline QTc, or the team "narrows" to ceftriaxone alone on day 2 because cultures are negative, Legionella coverage disappears silently -the patient looks like they are on an antibiotic, and they are, but not one that works. If a patient with severe CAP is failing beta-lactam therapy, Legionella belongs at the top of the list before you broaden to pip-tazo or vancomycin.
What It Is, and Why the Biology Dictates the Drug
Legionella pneumophila is a fastidious gram-negative rod that lives in fresh water and, once inhaled, becomes a facultative intracellular pathogen of alveolar macrophages. That single fact drives the entire page. An antibiotic can only work here if it concentrates inside the host cell, which is exactly what macrolides and fluoroquinolones do and exactly what beta-lactams and aminoglycosides do not. It also explains why the organism is invisible on routine Gram stain and will not grow on standard media, so a "no growth" sputum culture is not reassurance.
How Patients Get It
| Source | Detail | Why it matters to you |
|---|---|---|
| Aerosolized potable water showers, faucets, humidifiers | Legionella colonizes building plumbing, especially where water sits warm and stagnant. Inhalation of aerosol, or aspiration of colonized water, delivers the organism to the lower airway. | Ask where the patient has been sleeping. Hotels, cruise ships, long-term care facilities and recently reopened buildings are the recurring histories. |
| Cooling towers | Large air-conditioning cooling towers aerosolize water over a wide area and are the classic driver of community outbreaks clustered around a single building. | A second case from the same postcode within weeks is an outbreak until public health says otherwise. |
| Hot tubs and decorative fountains | Warm, agitated, heavily aerosolized water is close to ideal growth conditions. | Specifically ask about hot tub use, including on a cruise or at a hotel. Patients do not volunteer it. |
| Hospital water systems | Causes genuinely hospital-acquired Legionella. Onset ≥ 48 hours after admission with no other exposure implicates the building. | A nosocomial case is a mandatory, urgent report -it means other inpatients are being exposed right now. |
| Potting soil and compost not water | The route for L. longbeachae, which is soil-adapted rather than water-adapted and is a major cause in Australia and New Zealand. | The urine antigen misses this species entirely. Gardening exposure plus a negative urine antigen is exactly when you need culture or PCR. |
Incubation is 2 to 14 days, most commonly 5 to 6 (CDC). That window is not trivia -it is what public health uses to decide which building the patient was in when they were infected, so the exposure history you take needs dates, not just places. Person-to-person transmission essentially does not occur, so no isolation precautions are needed beyond standard, and families can be reassured directly.
Who Gets Sick After Exposure
Exposure is common and disease is not, so host factors decide who ends up in front of you. The strongest are:
- Age over 50 and male sex -consistently the largest demographic signal.
- Current or former smoking and chronic lung disease -impaired mucociliary clearance and macrophage function.
- ⚠ Immunosuppression, especially TNF-alpha inhibitors and systemic glucocorticoids -cell-mediated immunity is what contains an intracellular pathogen, so drugs that blunt it convert exposure into severe disease. Solid organ and stem cell transplant recipients are the highest-risk group and are also the group in whom you should extend treatment duration.
- Malignancy, diabetes, chronic kidney or liver disease, alcohol use disorder, iron overload.
The Classic Clinical Picture, and How Much to Trust It
⚠ Every "classic" Legionella feature is real at the population level and unreliable at the bedside. Hyponatremia, diarrhea, transaminitis, confusion and relative bradycardia each appear more often in Legionella than in pneumococcal pneumonia, but individually none has the specificity to rule the diagnosis in, and their absence does not rule it out. Relative bradycardia in particular has been overemphasized and should not carry weight in a real decision. Use the cluster to raise suspicion and trigger testing; never use it in place of testing, and never use its absence to skip testing in severe CAP.
| Feature | What you see | Why it happens / how to use it |
|---|---|---|
| High fever with rigors | Often > 39°C, with prominent malaise, headache and myalgia that can precede any respiratory complaint. | Patients are frequently labelled "viral illness" at first contact because the cough is mild and dry early on. |
| Diarrhea, nausea, vomiting | Watery diarrhea in roughly a quarter to half of cases. | GI symptoms dominating a pneumonia presentation should raise Legionella, and they are a common reason the chest is not examined early. |
| Confusion, headache, lethargy | Encephalopathy out of proportion to hypoxemia; occasionally ataxia. | Easily attributed to age or sepsis. In a febrile patient with an infiltrate, it should prompt the urine antigen rather than a head CT alone. |
| Hyponatremia | Often the most reproducible laboratory clue, and more frequent than in other causes of CAP. | Driven by SIADH. Still not specific enough to act on alone, but a low sodium plus severe CAP is a good reason to send the antigen. See Hyponatremia →. |
| Transaminitis, hypophosphatemia | Mild AST and ALT elevation, commonly around twice normal. Low phosphate is described as a supportive clue. | Part of the systemic picture. Recheck rather than chase, since both usually resolve with treatment. |
| Elevated CK, myoglobinuria, AKI | Rhabdomyolysis is a recognized complication and can be the reason for the renal failure. | Send a CK in severe Legionella -missing rhabdomyolysis means missing a treatable driver of AKI. See Rhabdomyolysis →. |
| Relative bradycardia Faget sign | A pulse lower than the fever predicts. Expect the heart rate to rise roughly 10 bpm for every 1°F of fever (Liebermeister's rule), which gives the expected pulse at each temperature below. A pulse under the figure for that temperature is relative bradycardia. 102°F (38.9°C) → 110 · 103°F (39.4°C) → 120 · 104°F (40.0°C) → 130 · 105°F (40.6°C) → 140 · 106°F (41.1°C) → 150 | ⚠ The number only means anything if Cunha's conditions hold: age ≥ 13, temperature 102 to 106°F, pulse and temperature taken at the same moment, normal sinus rhythm, and no beta-blocker, pacemaker, or second/third-degree block. Ignore those and you are simply describing a rate-controlled patient. Even when they hold it is overemphasized -its absence does not reassure, and its presence never substitutes for a test. |
Legionnaires' Disease vs Pontiac Fever
| Legionnaires' disease | Pontiac fever | |
|---|---|---|
| What it is | Pneumonia, with an infiltrate on imaging. | Self-limited influenza-like illness with NO pneumonia. |
| Incubation | 2 to 14 days, usually 5 to 6. | Hours to about 3 days -much shorter. |
| Course | Can be fatal. Requires antibiotics. | Resolves on its own in days. |
| Treatment | Fluoroquinolone or azithromycin. | None. Supportive care only -antibiotics have no established role in a self-limited illness with no pneumonia. |
| Reporting | Notifiable. | Also notifiable -it flags the same contaminated water source, and often the same outbreak. |
Where the Name Came From, and Why It Still Matters
The organism was identified after an explosive outbreak of pneumonia among people attending an American Legion convention in Philadelphia in July 1976 Fraser, 1977. 182 cases and 29 deaths, with the risk tracking time spent in and immediately outside one hotel, is what established the building-as-source pattern that still governs how these cases are investigated today. It is also why a single confirmed case triggers a public health investigation rather than just a prescription: the patient in front of you is a sentinel for a water system that is still running.
🧪 Diagnosis
⚠ A negative Legionella urine antigen does NOT exclude Legionnaires' disease. The test only detects L. pneumophila serogroup 1, which causes roughly 50 to 80% of cases. Every other serogroup and every non-pneumophila species, including L. longbeachae, is invisible to it. Used as the sole test, it misses on the order of 20 to 50% of cases. This is the single most common diagnostic error on this topic: a negative urine antigen is written in the chart as "Legionella ruled out", the azithromycin is stopped, and the patient is left on a beta-lactam alone.
Who to Test
Testing everyone with CAP is low yield and is not what the guidelines ask for. IDSA/ATS recommends Legionella testing in severe CAP and in the presence of epidemiological risk factors, so test when any of the following is true:
- Severe CAP, particularly anyone requiring ICU admission -this is the group where a missed diagnosis is lethal and where the pretest probability is highest. See Severe CAP criteria →.
- ⚠ Failure to improve on a beta-lactam -because a beta-lactam was never treating it, apparent "antibiotic failure" is one of the strongest clues you will get.
- Epidemiological exposure: recent travel or hotel stay, cruise ship, hot tub, known outbreak, or a recent stay in a building with a suspect water system.
- Onset ≥ 48 hours after hospital admission, which raises the question of the hospital water supply.
- Immunocompromised hosts, especially on TNF-alpha inhibitors, glucocorticoids, or post-transplant.
The Tests, and What Each One Is Actually For
Send the urine antigen AND a lower respiratory specimen for culture or PCR, together. This pairing is the CDC-preferred approach, and the reason is that the two tests fail in opposite directions: the antigen is fast but blind to everything except serogroup 1, while culture and PCR detect the full range of species but are slower and depend on getting a specimen. Ordering only the urine antigen is the default behavior on most wards and is precisely what produces the false reassurance above.
| Test | Performance | What it is good for, and where it fails |
|---|---|---|
| Urinary antigen first-line, order it now | Sensitivity ~70 to 80% Specificity > 99% | Fast, cheap, and positive as early as 1 day after symptom onset, so it is the test that changes management today. Crucially, it stays positive for weeks (sometimes far longer) and is NOT blunted by antibiotics -so unlike blood cultures, you have not lost anything by having started treatment first. That same persistence is its other limitation: a positive result can reflect an infection from weeks or months ago, so interpret it against the current illness. Detects serogroup 1 only. |
| Lower respiratory culture BCYE agar | Variable sensitivity; reference standard | Requires buffered charcoal yeast extract (BCYE) agar with L-cysteine -Legionella will not grow on routine media, so the laboratory must be told you suspect it or the right plate is never set up. This is the only test that yields an isolate, and the isolate is what lets public health match the patient's strain to a specific cooling tower or building. Ordering it is what turns one case into a prevented outbreak. |
| PCR / NAAT lower respiratory specimen | Sensitivity exceeds culture Specificity > 99% | Detects species and serogroups the urine antigen cannot, with results in hours rather than days. This is the test that rescues the serogroup-1 blind spot, so it is the right next step when suspicion stays high after a negative antigen. Availability varies by institution, which is worth knowing before you need it at 2am. |
| Serology paired titers | Requires a 4-fold rise over weeks | Retrospective and epidemiological only. A single acute titer is uninterpretable, and by the time convalescent serology returns the patient has been discharged or died. Do not order it to make a treatment decision. |
| Gram stain | Supportive, never confirmatory | Abundant neutrophils with no visible organisms is the classic and genuinely useful pattern, because Legionella stains poorly. It is a nudge toward an atypical or intracellular pathogen, not a diagnosis. |
Imaging, and a Trap in Following It
There is no radiographic pattern specific to Legionella. Unilateral patchy consolidation progressing to multilobar involvement is typical, and pleural effusions are common, but none of it distinguishes Legionella from pneumococcus. Two practical points:
- ⚠ Radiographic worsening during the first days of correct treatment is expected and is not treatment failure. The film characteristically lags the patient. Judge response by fever curve, oxygen requirement and mental status, not by a repeat chest film, or you will abandon a working antibiotic.
- Resolution is slow, often taking weeks to a couple of months. Warn the patient at discharge so a persistent shadow does not trigger an unnecessary cancer workup, and so they are not alarmed by ongoing fatigue.
💊 Treatment
⚠ Beta-lactams and aminoglycosides do not work, at any dose. Legionella replicates inside alveolar macrophages, and these drugs do not achieve meaningful concentrations in that intracellular compartment. Ceftriaxone, pip-tazo, cefepime, meropenem and gentamicin are all inactive in practice. Escalating the beta-lactam in a patient who is failing is the wrong move and costs days -the correct move is to add a fluoroquinolone or a macrolide and to send the urine antigen and a lower respiratory specimen.
First-Line Regimens
| Drug | Dose | Notes and cautions |
|---|---|---|
| Levofloxacin most commonly used | 750 mg IV or PO daily (500 mg daily is an accepted alternative, usually with a longer course) | Excellent oral bioavailability, so IV to PO switch needs no dose change once the patient is stable and absorbing. Watch QT prolongation, tendinopathy, aortic events and CNS effects; check for interacting agents and avoid co-administration with divalent cations, which block absorption. |
| Moxifloxacin | 400 mg IV or PO daily | Equivalent option. Not renally dosed (hepatic conjugation), which makes it convenient in AKI, but it achieves poor urinary concentrations so it is a bad choice if a urinary source is also in play. See how the fluoroquinolones differ → |
| Azithromycin | 500 mg IV or PO daily | The agent already present in most empiric CAP regimens, so the patient may effectively be on treatment before the diagnosis is made. Long half-life allows a shorter course. Preferred over a fluoroquinolone in pregnancy and in patients where fluoroquinolone toxicity is a real concern. |
| Doxycycline | 100 mg PO BID | Alternative for mild, outpatient-managed disease only. Weaker supporting evidence than the fluoroquinolones and macrolides; do not use it in severe or immunocompromised disease. |
⚠ Give the first dose early rather than agonizing over which agent. The comparative data below do not separate the two classes cleanly, but delay in starting any active agent is consistently associated with worse outcomes in severe disease. In practice: if the patient is sick and Legionella is plausible, start a fluoroquinolone or azithromycin now and let the antigen and culture sort out the rest -the urine antigen stays positive despite antibiotics, so treating first costs you no diagnostic yield.
Fluoroquinolone or Macrolide? The Evidence Genuinely Disagrees
This is a common pimp question and the honest answer is that two large meta-analyses reached different conclusions, so anyone stating a confident winner is overselling the data.
| Evidence | Finding | Interpretation |
|---|---|---|
| Meta-analysis favoring fluoroquinolones Kato meta-analysis, 2021 | 5 RCTs plus 12 retrospective studies. Mortality OR 0.59 (95% CI 0.35 to 0.98) favoring fluoroquinolones, with shorter length of stay. But in the levofloxacin-specific subgroup, only length of stay improved -mortality did not differ. | The mortality signal is borderline (upper confidence limit 0.98) and does not survive intact in the subgroup that matters most, since levofloxacin is the fluoroquinolone actually used. |
| Meta-analysis finding no difference Jasper meta-analysis, 2021 | 21 publications, 3,525 patients. Mortality 6.9% with fluoroquinolones vs 7.4% with macrolides, OR 0.94 (95% CI 0.71 to 1.25), p = 0.66. No significant difference. | Larger, more recent, and the confidence interval is tight around no effect. This is the more reassuring dataset if you are choosing azithromycin. |
What to actually say on rounds: "Both classes are acceptable first-line. Some data favor fluoroquinolones on length of stay, but the largest meta-analysis found no mortality difference, so the choice is driven by the patient in front of us -fluoroquinolone if there is a QT problem or a macrolide interaction, azithromycin in pregnancy or if fluoroquinolone toxicity is a concern. What is not optional is starting one of them promptly."
Duration, and What Extends It
| Situation | Duration | Why |
|---|---|---|
| Immunocompetent, mild to moderate, responding well | 5 to 10 days (azithromycin at the shorter end) | Azithromycin's long intracellular half-life means drug persists at the site of infection after the last dose, so a 5 to 7 day course delivers longer effective exposure than the calendar suggests. |
| Severe disease, ICU admission | 7 to 10 days, extend if slow to respond | Higher organism burden and slower clearance. Start IV in severe disease rather than relying on oral absorption in a shocked or vomiting patient, then switch once stable. |
| ⚠ Immunocompromised transplant, TNF-alpha inhibitor, high-dose steroids | 14 to 21 days | Relapse after a short course is the specific failure mode here, because clearing an intracellular pathogen depends on the cell-mediated immunity these patients lack. This is the single most important duration adjustment on the page. |
| Cavitation, lung abscess, empyema, or extrapulmonary disease | At least 21 days | Poor drug penetration into necrotic or walled-off tissue, and a correspondingly high relapse rate with standard courses. |
Complications to Look For
Legionella is a systemic infection, and the extrapulmonary problems are where avoidable harm accumulates. Watch specifically for rhabdomyolysis and AKI (send a CK if the creatinine is rising), hyponatremia from SIADH, and less commonly myocarditis, encephalitis, pancreatitis and DIC. Extrapulmonary Legionella endocarditis and sternal wound infection are rare but described, mostly in prosthetic valves and post-cardiac-surgery patients, and both mandate the prolonged course above.
Steroids have no Legionella-specific indication. There is no evidence supporting corticosteroids because the organism is Legionella. If the patient meets criteria for steroids in severe CAP, treat them on that basis and not on the pathogen; see When to Add Steroids in Pneumonia →. Note that chronic steroid therapy is itself a major risk factor for acquiring Legionella, so the drug sits on both sides of this disease.
⚠ Report the Case
Legionnaires' disease and Pontiac fever are notifiable to your local or state health department, usually within 1 to 7 days of diagnosis. This is not a paperwork formality and it is the step most often skipped by the treating team. Reporting is what triggers the environmental investigation that finds and shuts down the water source, and a single reported case is frequently how a cluster is first detected. Send a lower respiratory specimen for culture even when the urine antigen has already made the diagnosis -the isolate is what allows public health to match the patient's strain to a specific building, and without it the investigation is far weaker. If onset was ≥ 48 hours after admission, notify infection prevention immediately, because other inpatients are still drinking and showering in that water.
🎤 Rounds
Pimp Questions
Why does ceftriaxone fail in Legionella?
Legionella replicates inside alveolar macrophages, and beta-lactams do not reach meaningful concentrations in that intracellular compartment. Only agents that concentrate inside cells, macrolides and fluoroquinolones, are active.
The Legionella urine antigen is negative. Have you excluded it?
No. It detects only L. pneumophila serogroup 1, which is about 50 to 80% of cases, and its sensitivity is roughly 70 to 80% even for that. As a sole test it misses on the order of 20 to 50% of cases. Send a lower respiratory culture or PCR if suspicion persists.
Does starting antibiotics before sending the urine antigen ruin the test?
No. Unlike blood cultures, the urine antigen is not blunted by antibiotics and stays positive for weeks. Treat the sick patient first. The flip side is that a positive result can reflect an infection from weeks or months ago, so interpret it against the current illness.
Which medium does Legionella need, and why does it matter that you ask for it?
Buffered charcoal yeast extract (BCYE) agar with L-cysteine. It will not grow on routine media, so unless you tell the lab you suspect Legionella the correct plate is never set up and the culture reports "no growth".
Why send a culture when the urine antigen is already positive?
The isolate lets public health match the patient's strain to a specific water source. Without it, the environmental investigation is much weaker and the building keeps infecting people.
Fluoroquinolone or macrolide?
Either. One meta-analysis favored fluoroquinolones on mortality and length of stay, but the larger one found no mortality difference (6.9% vs 7.4%, OR 0.94). Choose on patient factors, QT, interactions, pregnancy, and prioritize starting promptly over choosing perfectly.
How long do you treat, and what changes it?
5 to 10 days if immunocompetent and responding. 14 to 21 days if immunocompromised, and at least 21 days with cavitation, abscess, empyema or extrapulmonary disease, because relapse after a short course is the characteristic failure.
The chest film looks worse on day 3 of levofloxacin. Change antibiotics?
No. Radiographic worsening early in correct treatment is expected and the film lags the patient. Judge response by fever, oxygen requirement and mental status. Resolution takes weeks to months.
How reliable are hyponatremia, diarrhea and relative bradycardia for diagnosing it?
Not reliable enough to act on. They are more common in Legionella than other CAP but individually lack specificity, and their absence does not exclude it. Relative bradycardia in particular is overemphasized. Use them to trigger testing, never to replace it.
Define relative bradycardia numerically. Temp is 103°F and the pulse is 96.
Heart rate should rise about 10 bpm per 1°F of fever, so at 103°F you expect roughly 120. A pulse of 96 is well under that, so yes, relative bradycardia. But first exclude a beta-blocker, pacemaker or heart block, and confirm the pulse and temperature were taken at the same moment, or the finding is meaningless.
A keen gardener with pneumonia and a negative urine antigen. What are you thinking?
L. longbeachae, acquired from potting soil and compost rather than water. It is not detected by the urine antigen at all, so this is exactly the scenario requiring lower respiratory culture or PCR.
Do you need isolation precautions?
No. Person-to-person transmission essentially does not occur; the source is environmental water. Standard precautions suffice, and the family can be reassured directly.
Which immunosuppressant carries a particularly notable risk?
TNF-alpha inhibitors, along with systemic glucocorticoids and transplant immunosuppression. Containing an intracellular pathogen depends on cell-mediated immunity, which these drugs blunt. They also push you toward the longer 14 to 21 day course.
📣 Sample Presentation
"Mr. R is a 61-year-old smoker with COPD, admitted three days ago with community-acquired pneumonia and started on ceftriaxone and azithromycin. Overnight the azithromycin was held for a QTc of 480, and this morning he is more confused, febrile to 39.2, and his oxygen requirement has gone from 2 to 6 liters. His sodium is 128, AST and ALT are roughly twice normal, and he has had watery diarrhea since admission. He also spent five nights in a hotel the week before he became unwell.
My concern is Legionella. The picture fits, but more importantly he has effectively been on beta-lactam monotherapy since the azithromycin was stopped, and a beta-lactam has no activity against Legionella at all, so what looks like clinical deterioration on treatment may simply be untreated infection. I have sent a urine antigen and a sputum specimen for Legionella culture on BCYE and PCR, and I have specifically flagged Legionella to the lab so the right medium is used. I am starting levofloxacin 750 mg IV daily rather than restarting azithromycin, which handles the QT problem and treats Legionella at the same time. I have sent a CK given the risk of rhabdomyolysis with his rising creatinine. If the urine antigen is negative I will not stop the levofloxacin, because the antigen only detects serogroup 1. If it is confirmed, this is notifiable, so I will report it and give public health the hotel dates, since the isolate may let them match it to a source."
Ward Checklist
- ⚠ Exposure history taken with dates -hotels, cruises, hot tubs, recent building works, gardening and potting soil, covering the 2 to 14 days before onset.
- Urine antigen sent in severe CAP, ICU admission, beta-lactam failure, or any epidemiological risk factor.
- ⚠ Lower respiratory specimen sent for culture and PCR as well, with Legionella flagged explicitly to the laboratory so BCYE agar is used.
- An agent with real activity is actually running -fluoroquinolone or macrolide, confirmed on the current MAR, not just written in the plan.
- ⚠ Check nobody has "narrowed" to a beta-lactam alone, and specifically check whether the macrolide was quietly held for QT.
- Sodium, LFTs, phosphate, creatinine and CK sent, with CK repeated if creatinine is climbing.
- Duration set deliberately, extended to 14 to 21 days if immunocompromised and documented with the reason.
- Team warned not to react to the day-3 film, since early radiographic worsening is expected.
- ⚠ Case reported to public health, and infection prevention notified if onset was ≥ 48 hours after admission.
- Patient counselled at discharge that fatigue and residual radiographic changes take weeks to months to clear, so neither prompts alarm.
📋 Summary
At a Glance
| Point | Detail |
|---|---|
| ⚠ Beta-lactams do not work | Legionella lives inside alveolar macrophages, which beta-lactams and aminoglycosides do not penetrate. Ceftriaxone monotherapy is no therapy. Apparent beta-lactam failure in CAP is itself a clue. |
| The exposure | Aerosolized water: cooling towers, hotel and hospital plumbing, hot tubs, decorative fountains. Incubation 2 to 14 days, usually 5 to 6. L. longbeachae comes from potting soil instead. No person-to-person spread, so no isolation needed. |
| Highest-risk hosts | Age over 50, male, smoker, chronic lung disease, and above all immunosuppression -TNF-alpha inhibitors, glucocorticoids, transplant. |
| ⚠ The classic clues are soft | Hyponatremia (SIADH), diarrhea, transaminitis, confusion, high fever. More common than in other CAP but individually non-specific, and absence excludes nothing. Relative bradycardia is overemphasized. Use them to trigger testing, not to replace it. |
| ⚠ Negative urine antigen excludes nothing | Detects serogroup 1 only (50 to 80% of cases), sensitivity 70 to 80%, specificity > 99%. Sole use misses 20 to 50% of cases. |
| Antigen timing is forgiving | Positive as early as 1 day after symptom onset, stays positive for weeks, and is not blunted by antibiotics -so treat first, test after. The same persistence means a positive can reflect an older infection. |
| ⚠ Send culture too | BCYE agar with L-cysteine, and tell the lab or the plate is never set up. Culture and PCR catch the species the antigen misses, and the isolate is what lets public health find the water source. |
| Serology is useless acutely | Needs a 4-fold rise over weeks. Epidemiological tool only, never a treatment decision. |
| Treatment | Levofloxacin 750 mg daily, moxifloxacin 400 mg daily, or azithromycin 500 mg daily. Doxycycline only for mild outpatient disease. IV first in severe disease, then switch. |
| FQ vs macrolide: unsettled | Kato, 2021 found OR 0.59 (0.35 to 0.98) favoring fluoroquinolones, but the levofloxacin subgroup showed only a length-of-stay benefit. Jasper, 2021, larger at 3,525 patients, found no mortality difference (6.9% vs 7.4%, OR 0.94). Either is first-line; starting early matters more than choosing. |
| ⚠ Duration | 5 to 10 days immunocompetent and responding. 14 to 21 days if immunocompromised. ≥ 21 days with cavitation, abscess, empyema or extrapulmonary disease. Relapse is the short-course failure mode. |
| Do not chase the film | Radiographic worsening early in correct treatment is expected. Follow fever, oxygen and mentation. Resolution takes weeks to months, so warn the patient. |
| Complications | Rhabdomyolysis and AKI (send a CK), hyponatremia, and less often myocarditis, encephalitis, pancreatitis, DIC. Steroids have no Legionella-specific role -treat per severe CAP criteria only. |
| ⚠ It is notifiable | Report Legionnaires' disease and Pontiac fever to public health, usually within 1 to 7 days. Reporting is what shuts down the water source. Nosocomial onset (≥ 48 hours) means telling infection prevention today. |
| Pontiac fever | Self-limited flu-like illness with NO pneumonia, incubation hours to 3 days. Supportive care, no antibiotics -but still report it. |
The Three Things to Remember
- The beta-lactam is not treating this. If a CAP patient is deteriorating on ceftriaxone, do not escalate the beta-lactam. Add a fluoroquinolone or macrolide and check whether someone quietly stopped the azithromycin.
- A negative urine antigen is not a negative result. It sees serogroup 1 and nothing else. If suspicion is real, send a lower respiratory culture on BCYE and a PCR, and tell the lab what you are looking for.
- Report it, and send the culture anyway. The patient is a sentinel for a water system that is still running, and the isolate is what lets public health find it.
