2013 NEJM Review

The classic clinical triad is emboli, new or changing murmur, and positive blood cultures.

However, life and medicine are rarely that clear cut. So we have the Duke criteria:

Definite infective endocarditis:

Pathologic criteria

Microorganisms demonstrated by culture OR histology in a vegetation OR in a vegetation that has embolized OR in an intracardiac abscess OR Pathologic lesions: vegetation OR intracardiac abscess present, confirmed by histology showing active endocarditis OR, since we usually have to rely on clinical criteria, we have

Clinical criteria which consist of 2 major criteria OR 1 major and 3 minor criteria OR 5 minor criteria.


Typical microorganism for infective endocarditis from 2 separate blood cultures: Viridans Streptococci, Streptococcus bovis, HACEK group, OR Community-acquired Staphylococcus aureus OR Enterococcus, in the absence of a primary focus OR Persistently positive blood cultures for any microorganism (i.e., from blood cultures drawn more than 12 hours apart), OR all of 3 OR majority of 4 OR more separate blood cultures, with first and last specimens drawn at least 1 hour apart.


Findings on echocardiogram-positive for infective endocarditis: Oscillating intracardiac mass on valve OR supporting structures OR in the path of regurgitant jets OR on iatrogenic devices, in the absence of an alternative anatomic explanation OR Abscess OR new partial dehiscence of prosthetic valve OR new valvular regurgitation (increase OR change in preexisting murmur not sufficient).


Predisposition: predisposing heart condition OR intravenous drug use.

Fever: 38.0°C (100.4°F).

Vascular phenomena: arterial embolism, septic pulmonary infarcts, mycotic aneurysms (PubMed), intracranial hemorrhage, Janeway lesions. Patients on prior antiplatelet therapy have a marked decrease (2/3's) chance of having emboli (PubMed).

Immunologic phenomena: glomerulonephritis, Osler nodes, Roth spots, rheumatoid factor, false-positive VDRL (one hopes) and positive ANCA (PubMed).

Antiphospholipid antibodies are significantly more often positive, and at higher levels, in patients with definite IE(PubMed).

Echocardiogram: findings consistent with infective endocarditis but not meeting the major criterion above.

Microbiologic evidence: positive blood culture but not meeting major criterion above OR serologic evidence of active infection with organism consistent with infective endocarditis.

Different organisms will have different echocardiogram findings:  "Associations were seen between aortic valve vegetations and Enterococcus faecalis among patients with native aortic valves, between mitral valve vegetations and streptococci of group B or viridans group, between tricuspid valve vegetations and S. aureus among patients with intravenous drug abuse, and between perivalvular abscesses as well as cardiovascular implantable electronic device (CIED)-associated IE and coagulase negative staphylococci (all P < 0.05)."(Pubmed).

Epidemiologic Risks

Congenital (bicuspid aortic valve or any of the congenital eponyms I cannot remember) or acquired (rheumatic fever, prior endocarditis, old valve) valvular disease, hypertrophic cardiomyopathy, prosthetic valve, use of needles (dialysis (who have a 50% mortality rate (PubMed)), diabetes, IVDA).

There is a marked association between mitral valve prolapse and viridans group streptococci (PubMed).

Healthcare-associated endocarditis is an emerging issue (PubMed) and if often right-sided disease and may account for up to 25% of endocarditis. Those who are at high risk for endocarditis from nosocomial bacteremia include those with sustained bacteremia (duh), hemodialysis, pacer systems and those who get vertebral osteomyelitis.

Bacteremia is part of life and if you have an abnormal valve you may get unlucky. While S. viridans is the most commonly isolated organism by cultures if molecular techniques are used a dental extraction results in "13.4±1.7 bacterial families and 22.8±1.1 genera per sample" in the blood (PubMed).

Is dental work a risk? Probably not (PubMed).

The dentist always wants you to floss, but who gets endocarditis? Those who "use toothpicks, dental water jet, interdental brush, and/or dental floss... (but) were less likely to brush teeth after meals." (PubMed).

91.5% of endocarditis due to oral Streptococci did NOT have a prior dental procedure, but they often had bad teeth (Pubmed).

Invasive dental work does increase the risk for prosthetic valve endocarditis just a tich (PubMed).

For prosthetic valves, 1/20 get endocarditis over a 10-year span (!?!) and bioprosthesis had a bigger risk, I would bet due to the lack of anticoagulation (PubMed). Clot bad.

Injection drug use. Bad disease, it has a 16% one-year all-cause mortality (PubMed). And endocarditis leads to recurrent endocarditis in IVDU, with Candida leading the list (PubMed). In one series 2/3 were C. parapsilosis (PubMed).

Transcatheter aortic valve replacement has a high risk of endocarditis (1.1% per year) and death (36%)(Pubmed).

Risk of outpatient procedures:

Procedure Relative Risk: Gastroscopy 2.50 (1.59 - 3.94), Colonoscopy 2.89 (1.35 - 6.17), Dialysis 4.33 (2.10 - 8.95), Bone marrow puncture 4.33 (1.24 - 15.21), Coronary angiography 4.75 (1.61 - 13.96), Bronchoscopy 5.00 (1.10 - 22.82) and Transfusion 5.50 (1.22 - 24.80) (PubMed).


Damn near anything can cause endocarditis, but there are patterns (Pubmed).

The most common causes: S. aureus, Streptococci (especially the viridans group), Enterococcus, The HACEK Group of gram-negative coccobacilli: Haemophilus, Aggregatibacter, Cardiobacterium, Eikenella, Kingella.

Abiotrophia and Granulicatella are more prevalent than HACEK in one study and approximately one-tenth as prevalent as viridans streptococci; periannular complications like myocardial abscess were more frequent (PubMed).

One small study that did PCR on valves found it to be polymicrobial more often than not (PubMed) and due to unculturable organisms; I wonder it the vegetation is being repeatedly seeded by the bacteremia of life. But if the patient is not responding to antibiotics, perhaps it is due to unculturable organisms.

In prosthetic valve (and native valve (PubMed) endocarditis, coagulase-negative staphylococci lead the list in the first year after valve replacement, then the microbiology is similar to native valve endocarditis. Mycobacterium chelonae and M. lentiflavum can occur in bioprosthetic valves (PubMed).

For TAVRs the most common are Enterococci then Staphylococcus (Pubmed).

Healthcare-associated endocarditis is usually due to S. aureus and coagulase-negative staphylococci and they usually have a central line as a source.

Candida endocarditis often occurs in heroin users or as a complication of central lines.  At least 4.2% of candidemias will have endocarditis (PubMed).

Culture negative endocarditis (PubMed): usually due to prior antibiotics; organisms to consider include Coxiella burnetii (the statistically most common cause), Bartonella, Legionella, Mycobacterium, Tropheryma whippeli, (also the most common cause, it depends on how you make the diagnosis (PubMed) (Pubmed)) and fungi of all types.

With Whipple's endocarditis, the aortic valve is most commonly involved, multiple valves not uncommon, the patient may be afebrile, and inflammatory markers may be normal (PubMed).

It can make the diagnosis in culture-negative disease with cell-free DNA tests (Abstract) even in those who have negative cultures due to antibiotics. As of 10/18, it cost 2 grand.

And vegetations do not mean infectious endocarditis: there are Libman­Sacks in SLE (PubMed) and marantic in patients (PubMed) with adenocarcinomas.

Gram-negative endocarditis that is not HACEK is more often E. coli than Pseudomonas and is nosocomial (PubMed).

Empiric Therapy

Get lots of blood cultures, at least 12 hours apart if you can, BEFORE antibiotics.

- IVDA: an anti-staphylococcal beta-lactam (if penicillin-allergic or high suspicion of MRSA, vancomycin.) plus a third-generation cephalosporin. 

- Native valves: an anti-staphylococcal beta-lactam (if penicillin-allergic or high suspicion of MRSA, vancomycin.) plus a third-generation cephalosporin.

- Prosthetic valve: vancomycin PLUS gentamicin PLUS rifampin.

- a pearl: if the patient has MSSA and you start with Vancomycin and switch to Nafcillin, there is higher mortality, so maybe maybe patients should be on Nafcillin and Vancomycin upfront. Maybe (PubMed).

Specific Treatments

All antibiotics should be given at the maximum dose for age, size, and CrCl.

Prophylaxis is here.

Can oral antibiotics be used? In some circumstances but I have never than the courage. Here is the review (PubMed). In a 2018 study of left-sided endocarditis in the NEJM, stable patients who were changed to oral antibiotics after day 17 of IV had the same outcomes as those who had the complete course IV. I can see doing that for Streptococci, there is an old literature of treating streptococcal endocarditis with only oral therapy, but I would be leery about doing it with S. aureus or Enterococcus (PubMed). And note all the patients who were excluded. And the big question is how many were cured at day 17 and did not need oral? Probably all the strep. I have cured strep endocarditis with 2 doses of ceftriaxone. The patient, after two doses, remembered he was a Christian Scientist and refused further therapy. He came back several months later to let us know he was a cure and credited prayer. I thought it was the antibiotic. This study tells you what you can get away with, not necessarily the best therapy.

I will also note that if patients are on iv antibiotics for 14 days, no vegetation will grow bacteria, suggesting that many cases are cured before day 17 (PubMed).

- Staphylococcus aureus, left-sided, native valve: 6 weeks of iv nafcillin or oxacillin > vancomycin. Aminoglycosides? Now passe they decrease the number of febrile days but do not decrease mortality or need for valve replacement. Gentamicin toxicity causes CrCr to decrease at a rate of 0.5% per day but does not adversely alter the outcome (PubMed). Another study demonstrated that low dose gentamicin tripled the nephrotoxicity rate without altering outcomes (PubMed).

Here is the tricky part: If it is MSSA and the mic is elevated >= 1.5 to vancomycin, nafcillin will have a higher failure rate (PubMed).

If resistant or intolerant of standard therapy, linezolid may be just as bad (71% success rate) (PubMed).

Rifampin probably only adds toxicity (PubMed) for a native valve.

- Staphylococcus aureus, right-sided, native valve: 4 weeks of iv anti-staphylococcal penicillin > vancomycin OR two weeks iv of anti-staphylococcal penicillin PLUS tobramycin for uncomplicated disease. In right-sided disease requiring ICU, adding an aminoglycoside was associated with increased survival (PubMed).

I will never do this until confirmatory studies, but high dose of Bactrim and clindamycin (intravenously, switched to oral at day 7) was superior to standard therapy (Pubmed).

Daptomycin at 6 mg/kg/d for 28 d is non-inferior to vancomycin (PubMed). 8 or 10 mg/kg is almost certainly the better dosing for endocarditis (PubMed), although daptomycin should probably not be used for Enterococcus (PubMed). While daptomycin should not be used for pneumonia, some data suggests it is ok for Staphylococcal septic pulmonary emboli from endocarditis (PubMed).

- Staphylococcus aureus, prosthetic valve. Remove it. If you want to try and probably fail at medical therapy: 6-8 weeks anti-staphylococcal penicillin > vancomycin iv PLUS gentamicin PLUS rifampin. BUT: while in the guidelines, in one study gentamicin did not alter outcomes but did add toxicity (PubMed).

- Coagulase-negative staphylococci, prosthetic valve: 6 - 8 weeks vancomycin PLUS gentamicin (2 weeks) PLUS rifampin. BUT: while in the guidelines, in one study gentamicin did not alter outcomes but did add toxicity (PubMed).  As the mic to vancomycin goes up, so does the mortality (PubMed).

- Viridans Streptococci: mic < 0.1: 4 weeks iv penicillin OR 2 gm qd ceftriaxone. If PCN allergic, vancomycin. mic 0.1-1.0: 2 weeks of penicillin AND gentamicin THEN two weeks of penicillin alone. If PCN allergic, vancomycin. mic > 2.0 (PubMed): 4-6 weeks of penicillin AND gentamicin. If PCN allergic, vancomycin.

Keeping the penicillin dosed at q 4 gives better outcomes (PubMed).

If you need to use daptomycin, beware that the S. mitis group (mitis, anginosus, mutans and salivarius) that can rapidly develop resistance (PubMed).

- Enterococcus: For high-level gentamicin resistance and gentamicin susceptible, ampicillin 12 grams a day PLUS ceftriaxone 2 gm q 12 for 6 to 8 weeks works (PubMed) and is the treatment of choice (PubMed) for E. faecalis. E. faecium does not have synergy in vitro maybe a quarter of the time.

Daptomycin should probably not be used for Enterococcus (PubMed) unless no other option.

The old school is 6 weeks of ampicillin (preferred; 12 grams a day) PLUS an aminoglycoside OR vancomycin PLUS an aminoglycoside no matter what it does to the ears or kidneys (although the risk for gent toxicity may be genetic). If patients have been ill for less than three months, 4 weeks may be enough of symptoms of <= 7 days (PubMed), at least one study suggests everyone should get 6 weeks (PubMed). Beware of high-level gentamicin resistance (mic > 500).

There was a case of high-level gentamicin resistant enterococcal endocarditis treated with daptomycin plus ceftaroline (PubMed).

Maybe, for VRE, use daptomycin PLUS ampicillin for 8 to 10 weeks or quinupristin/dalfopristin for 6 weeks or linezolid or daptomycin (high dose) WITH tigecycline. The best bet is to go right to valve replacement.

Other combinations of antibiotics have been tried for endocarditis, reviewed in 2018 (PubMed).

- HACEK (test mic): 4 weeks of ampicillin AND gentamicin OR qd ceftriaxone.

- how to treat culture-negative endocarditis depends on what organisms you suspect, but cure rates are better if they get an aminoglycoside (PubMed).

Gram-negatives account for maybe 2% of endocarditis and often have cardiac implantable electronic device (CIED) endocarditis, a central venous catheter, diabetes mellitus or to be on immunosuppressive therapy (PubMed).

- Candida endocarditis, which can be a complication of both catheters and IVDA, is best treated with valve resection and a long course of ? amphotericin? fluconazole (PubMed)? other. While classically an absolute indication for valve removal, I have cured two cases in patients who were deemed not surgical candidates and this has been supported in the literature (PubMed) with an iv echinocandin or amphotericin followed by a year of po fluconazole. My endpoint was a negative 1-3 beta D glucan.

For prosthetic valve endocarditis, start with lipid amphotericin, not an echinocandin (PubMed). Improves survival. And, oddly, surgery did not have better outcomes than medical therapy alone, although small numbers and retrospective.

- other bugs? Anything can cause endocarditis, and often you need an ID consult.


- a CT/PET may be the best way to look for complications/metastatic infection and diagnose prosthetic valve infection, except they cost too much (PubMed)(PubMed). A PET is better for PVIE than native valve infection (Pubmed).

- Who should get a valve replacement? Refractory CHF, recurrent emboli, S. aureus prosthetic valve infection, gram-negative or yeast as the causative organism, failure of therapy (usually manifests as fever and is often due to a ring abscess).

And perhaps you should get a CNS angiogram before anticoagulation; stroke and aneurysms, often silent, are not uncommon (PubMed).

"Multivariate risk factors for ischemic stroke included prior stroke, Staphylococcus infection, mitral vegetations, and valvular abscess ). Risk factors for hemorrhagic stroke included fungal infection, male sex and rheumatic heart disease (PubMed)."

"Patients with non-surgically treated IE with a stroke during IE admission were at significantly higher associated risk of subsequent stroke within the first year of follow-up (PubMed)"

And another series suggests the stroke risk is for at least two years and anticoagulation is beneficial (Pubmed).

But. Even with S. aureus prosthetic valve endocarditis, there is no survival benefit with early valve replacement (PubMed).

There are some annoying data to suggest that S. aureus endocarditis who get valve replacement have less long term mortality mostly in those who have CHF; valve replacement is always such a great intervention in the IVDA.

- Whether valve replacement should be a first-line therapy due to increased survival or the increased survival seen with surgery (PubMed) is due to treating patients medically who have contraindications to surgery is not known.

- vegetation >= 10 mm are associated with an increase in embolism and death (PubMed). So take it out? Oddly, vegetations are not amenable to IR interventions.

- But CNS emboli can be removed mechanically since anticoagulation is to be avoided (PubMed).

- there is one prospective trial to suggest benefit from early valve replacement (PubMed). It is important to see who was enrolled: "a diagnosis of definite infective endocarditis according to the modified Duke criteria and had a severe mitral valve or aortic valve disease and vegetation with a diameter greater than 10 mm. To minimize the number of unnecessary surgeries and the risk of prosthesis- related morbidity, we only enrolled patients with infective endocarditis accompanied by severe valve disease."

- there is no survival benefit in delaying surgery in patients who present with stroke (PubMed) although before valve replacement perhaps you should check for silent CNS events: they can be just as deadly preoperatively (PubMed). But "observational data supports delaying surgery by 7-14 days if possible in IE complicated by ischemic stroke and >21 days in hemorrhagic stroke to lower perioperative mortality and neurological exacerbation (PubMed). "

- But, and there is always a but, one retrospective analysis suggests that surgery for left-sided endocarditis and a large vegetation as the only indication had a worse outcome. Damned if you do, damned if you don't (PubMed).

Surgery leads to better outcomes if done urgently or later (PubMed).

- positive vegetation cultures at the time of valve replacement does NOT predict relapse (Pubmed).

- patients who have endocarditis who present with sepsis/septic shock have increased survival with surgery (PubMed)

Replace with a mechanical or bioprosthetic valve? The data, such as it is, points to better outcomes with mechanical valves (PubMed).

After the valve is replaced? How long to treat? The risk is about 0.5% the new valve will be infected at the time of surgery from the endocarditis. The tradition is longer. If the cultures of the valve are negative, probably two weeks and post-op rifampin are not needed (Pubmed). And one study found as little as a week post-op was enough (PubMed). There is no one size fits all and duration probably depends on the organism is the host and degree of cardiac involvement/repair.

For PVIE with no surgery, "At 1 year, the mortality rate was 24%; relapses and reinfection were rare (5% each). Enterococcal PVE was associated with a higher risk of relapse." (PubMed)."

- ANY patient with CHF should probably get valve replacement; it decreases short term and long term survival (PubMed): "In-hospital mortality was 29.7% ...for the entire HF cohort, with lower mortality observed in patients undergoing valvular surgery compared with medical therapy alone (20.6% ... vs 44.8%..., respectively; P < .001). One-year mortality was 29.1%...in patients undergoing valvular surgery vs 58.4%... in those not undergoing surgery (P < .001)."

-ECHO, especially trans-thoracic, to "rule out" endocarditis is way stupid. Please note the criteria above. TEE is much better for finding a vegetation, so, if you really want to make the diagnosis of native valve IE, go right to TEE. All patients with prosthetic valve IE should get a TEE to look for a ring abscess. A vegetation > 1.0 to 1.5 cm has a high likelihood of emboli, which is a risk for death, so perhaps this should be a reason for valve replacement (PubMed).

I tend to not get TTEs to make a diagnosis of endocarditis; it is an insensitive test. But there is useful information to be gained from a TTE: Bad prognosis was associated with moderate or severe aortic or mitral regurgitation and a bicuspid aortic valve. "The presence of a bicuspid aortic valve appeared particularly predictive of the need for cardiac surgery, including for clinically occult paravalvular abscesses.(Pubmed).

- negative ECHO, even a TEE, does not mean they do not have a valve infection. Almost 20% of those with endocarditis will have a normal echo (PubMed).  For example, Pneumococcus can destroy a valve and have nary a blip on the TEE (PubMed).

- TTE on IVDA in the outpatient setting will often have valve pathology and 5% will have vegetations without endocarditis (PubMed).

- prosthetic valve endocarditis can have difficult to evaluate TEE due to interference, a tagged WBC scan may be of help to confirm the diagnosis (PubMed) although I think the data suggest a PET is better.

- End of therapy echo? A waster of time if the patient is doing well clinically (Pubmed).

- 'silent' CNS emboli occur in a quarter of patients (PubMed) and CNS emboli have a bad prognosis (PubMed).  And many patients will have silent lesions on MRI; what to do about them and when to look is not known (Pubmed).

- clinical microbleeds (hypointense lesions seen on T2 <10 mm) are common (PubMed) but do not adversely affect outcomes (PubMed) and do not make open-heart surgery riskier.

- multivalve disease is associated with more morbidity, but no mortality (PubMed).

- organisms that rarely cause IE: Group A strep, most anaerobes and most aerobic gram-negative rods (at least in non-IVDA).

- S. aureus is a particularly bad player: in almost every study it is an independent risk factor for death and 1/4 will die in the next year.

- Half of Enterococcal endocarditis patients will have a colonic neoplasm so they require a colonoscopy (PubMed).

- IDDM patients do especially poorly, as with all infections.   As do the elderly (Pubmed).

"In patients with IE, DM is associated to a higher prevalence of anatomic complications and a more impaired LVEF. Diabetic patients show a significantly lower survival both in hospital and during follow-up compared to the non-diabetic ones (Pubmed)."

- CRP > 122 after a week is a poor prognostic sign. As if I care (PubMed); I call it a C ReActive Protein. CRAP.

- bacteremia is common with brushing teeth, one study found "98 different bacterial species recovered from 151 bacteremic subjects. Of interest, 48 of the isolates represented 19 novel species of Prevotella, Fusobacterium, Streptococci, Actinomyces, Capnocytophagia, Selenomonas, and Veillonella" (PubMed).

- elevated troponin is a bad prognostic sign (PubMed).

- large pericardial effusions with native valve endocarditis are associated with increased mortality and other complications (PubMed).

- prior antiplatelet therapy and institution of antiplatelet therapy may decrease mortality; do NOT, however, use warfarin (Reference).

- up to a year after left-sided disease, patients have a poor quality of life and 11% can have PTSD (PubMed).

- here is some weirdness: vancomycin and gentamicin can make you fat. "A major and significant weight gain can occur after a six-week intravenous treatment by vancomycin plus gentamicin for IE with a risk of obesity, especially in males older than 65 who have not undergone surgery. We speculate on the role of the gut colonization by Lactobacillus sp, a microorganism intrinsically resistant to vancomycin, used as a growth promoter in animals, and found at a high concentration in the feces of obese patients (Plos)."


By the way, the correct order is blood cultures, THEN antibiotics. Not the other way, as is often the apparent standard.

Curious Cases

Relevant links to my Medscape blog

Time flies like an arrow, fruit flies like a banana

Infections Destroys All

As Prince said, Forever is a mighty long time.

Sine qua non

ECHO echo echo

Oh no

What does the Inky Dinky Do?

Bacterial Mummies

The Real Deal?

Make a grown man cry

More Grey.

Complications. I hate complications

Poly want a microbial


Don't just peat, repeat


Can You Prove a Negative? Nope.

Am I Feeling Lucky Today? Am I? TVIE.

Size Matters?

All that is gold does not glitter, Not all those who wander are lost, All vegetations are not infected.

0 to 60



Toe Hold

What to do, what to do.

Old Unreliable

ECHO RANT echo rant

In my experience...

Five Months Qualifies for Sustained

Looking for patterns in all the wrong places.

Dodging Bullets

Things fall apart


The Ghost of Reinhart Returns

I See All The Weirdness

Close Enough

Two in a Row

The wonderful cases for me are often the worst cases for the patient.

Wild Denouement.

Really Short Course

On Second Thought...

Killing Emotions?

Yet Another Thing I Do Not Understand

Perfect World?

BMD with Collateral Damage




Mmmmmmmmm. Maple Syrup and Bananas.

Going Clubbing

Cheap and Easy

Two in a Career

True-True, Related or Not

Why So Few?

ID Go: Gotta Kill Them All

Abscess Does Not Make the Heart Grow Fonder

How Long is Enough?

Horrible Complication

Prosthetic Valve Zit

Third Time to be a Charm?

More is not better

Plus ça change, plus ça change

That is no way to die


Don't believe everything you think.

Blind Pigs and Acorns

ID: We make the guidelines. Then ignore them.

How to Avoid Mass

Eau de toilette


We Are The Champignons

One is all you get

History Always Repeats

Gullible Rube

What Is This, The Dark Ages?

Drugs. Look to the Drugs.

Turtles all the way down.

Here Today. Gone Tomorrow.

No Way It's Subacute.


To Play the Odds

Dodge One Bullet, Get Hit in Another Spot

Delay is Common.

What are the Odds?

Last update: 01/16/21