Blog & Comments

It has been a while since I entered something very medical on this blog.  Well, the time is ripe!  Geeta Mehta, one of my wonderful ICU colleagues, asked me to comment on the whole concept of empiric treatment for candidal infection in the ICU.  As someone who loves a good controversy, I thought that this one was ripe for putting thoughts to words.

Let me first give the context:  Invasive candidal infection has been coincident with advances in medical care, greater use of antibacterial agents and intravenous catheters, and increasing immunosuppression (both iatrogenic and due to HIV infection).  Indeed, most studies looking at the temporal trends in candidaæmia over the years have identified a progessive increase over the past 20 or so years.

• Eggimann P, Garbino J, Pittet D. Epidemiology of Candida species infections in critically ill non-immunosuppressed patients. Lancet Infect Dis. 2003;3:685-702.

Some data over the early part of this century show a staibliization or even a reduction in candidæmias, but this could be explained by the more liberal use of antifungal prophylaxis or empiric therapy.  As with all broad-sweeping attempts at prolonged antimicrobial prophylaxis, the emergence of drug-resistant strains is inevitable, but usually takes 6-24 months to develop.  The increased use of triazole-based prophylaxis in chemotherapy-induced neutropenia (and stem-cell transplant conditioning) has been accompanied by a rise in the proportion of non-albicans candidal isolates.  Indeed, there was a similar observation with the management of mucosal candidiasis in patients with advanced HIV infection.

• Malani PN, Bradley SF, Little RS, Kauffman CA.  Trends in species causing fungaemia in a tertiary care medical centre over 12 years. Mycoses 2001; 44:446–9.

There are many recognized forms of invasive candidiasis that include mucosal disease, hepatosplenic candidiasis, and candidæmia.  Mucosal disease predominantly reflects defects in cell-mediated immunity, such as in HIV.  Hepatosplenic candidiasis is primarily an immune reconstitution disease following chemotherapy-induced neutropænia and presumed subclinical dissemination of candida.  Patients with spillage of bowel contents into the peritoneum may eventually develop candidal peritonitis if there is not adequate source control during antimicrobial management.

Candida species are now the third leading single organism isolated from blood cultures in ICUs in the United States (following coagulase-negative staphylococci and enterococci), and represent somewhere between 5-10% of all bloodstream infections in the ICU.

• Hidron AI, Edwards JR, Patel J et al. Antimicrobial-resistant pathogens associated with healthcare-associated infections: Annual summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2006–2007. Infect Control Hosp Epidemiol 2008; 29:996-1011

In critically ill patients, Candidæmia carries a crude mortality rate of somewhere around 50%, an attributable mortality of 20-43%, and an additional ICU length of stay of 13 days.  Candidæmia is, thus, an understandable focus of attention in the care of patients in the ICU.

• Eggimann et al. (se above)
• Leroy O, Gangneux JP, Montravers P, Mira JP, et al. Epidemiology, management, and risk factors for death of invasive Candida infections in critical care: a multicenter, prospective, observational study in France (2005-2006). Crit Care Med. 2009;37:1612-8.

Candidæmia usually requires the combination of a central venous catheter and prolonged broad-spectrum antimicrobial therapy.  Other factors can include various forms of immunosuppression, use of total parenteral therapy, etc., but I believe it is primarily a line-and-bug juice problem.  Others may disagree.  Colonization is the most-frequently reported and discussed risk factor but—in almost all cases—this is strongly associated with antimicrobial therapy. Nevertheless, it is this question of therapy for candida in the ICU when there is no definite diagnosis of invasive candidiasis that has prompted this blog entry (and I will point out that I am not addressing antifungal therapy for the neutropænic host.  Although—as with all other topics related to antimicrobial use—I have strong views on this, I think it is beyond the discussion for this blog.)

In critically ill patients, administering antifungal prophylaxis remains controversial.  The literature separates out prophylaxis from pre-emptive therapy, but I must admit that I am agnostic as to whether such a real distinction is possible:  prophylaxis presumably prevents infection and disease, whereas pre-emptive therapy is treating infection without disease (which, I guess, is what is usually referred to as “colonization”.)  I will lump all of these under “empiric therapy”, recognizing that this is anathema to many purists.

Candida colonization has a very poor positive predictive value, and therefore is an unreliable marker of disease.  (Assessing it routinely is also very expensive!)

• Pittet D, Monod M, Suter PM, Frenk E, Auckenthaler R. Candida colonization and subsequent infections in critically ill surgical patients.  Ann Surg 1994; 220:751–8.

Because of such poor specificity, decision rules to help identify invasive candidiasis in critically ill patients have been developed.  Several attempts to identify patients at high-risk for invasive candidiasis (who therefore might benefit from empiric antifungal therapy) have been developed and investigated over the past several years, three by Luis Ostrosky-Zeichner:

• The first such rule sought to identify all patients with invasive candidiasis using data from a surgical ICU.  This rule (requiring any combination of DM, new onset hemodialysis, use of TPN, or receipt of broad-spectrum antibiotics) was only reasonably sensitive (78%), was woefully non-specific (identifying half of all patients who remained in the ICU for over 4 days), and of questionable generalizability. (Paphitou NI, Ostrosky-Zeichner L, Rex JH. Rules for identifying patients at increased risk for candidal infections in the surgical intensive care unit: approach to developing practical criteria for systematic use in antifungal prophylaxis trials. Med Mycol. 2005;43:235-43.)
• The second such rule was developed in Spain as part of a very large multicentre study evaluating fungal infections in ICUs.  (Leon C, Ruiz-Santana S, Saavedra P, Almirante B, Nolla-Salas J, Alvarez-Lerma F, et al. A bedside scoring system (“Candida score”) for early antifungal treatment in nonneutropenic critically ill patients with Candida colonization. Crit Care Med. 2006;34:730-7.)  This score, the “Candida score” required  ICU patients being screened weekly for candida colonization using tracheal aspirates, pharyngeal exudates, gastric aspirates, and urine, although other samples could be obtained at the discretion of the attending physician.  Using logistical modeling techniques, and then “score” development using logit methodology (I have to justify all of my stats courses somehow), the authors came up with a score with the following formula:

Candida score =.908 x TPN + .997 X surgery + 1.112 multifocal Candida species colonization + 2.038 x severe sepsis (with all variables coded as 1 or 0)

Using a reasonable standard definition for invasive candidiasis, the authors were able to show that having a score of >2.5 colonized with candida was associated with an 81% sensitivity and a false positivity of over 25%.

• A third rule was published in 2007. (Ostrosky-Zeichner L, Sable C, Sobel J, et al. Multicenter retrospective development and validation of a clinical prediction rule for nosocomial invasive candidiasis in the intensive care setting. Eur J Clin Microbiol Infect Dis. 2007;26:271-6)  I won’t go into any details about this rule, other than that it had an unbelievably awful positive predictive value of 0.01 with a sensitivity of only 34%, making it of no clinical value.  It appears that it was developed for an RCT of caspofungin to prevent invasive candidiasis in critically ill patients (more of this below).

• Finally, a fourth rule was published last year (Ostrosky-Zeichner L, Pappas PG, Shoham S et al. Improvement of a clinical prediction rule for clinical trials on prophylaxis for invasive candidiasis in the intensive care unit. Mycoses. 2009. doi:10.1111/j.1439-0507.2009.01756.x), and aimed to improve upon the clinical prediction rule.  However, it still only resulted in a positive predictive value of 0.10 (with a specificity of 0.83).  Curiously, this prediction rule does not include colonization at multiple sites.

So it would appear that—as of 2010—we are not able to reliably predict who wil get invasive candidiasis.  Of course, this should therefore preclude empiric therapy (because the relative rarity of candidæmia in the ICU would make broad use of empiric therapy largely inappropriate).

Nevertheless, there have been two sizable trials looking at “empiric” antifungals in critically ill patients.  The first trial, published in 2008 (but conducted between 1995 and 2000), randomized 270 patients with fever unresponsive to antimicrobials to either fluconazole 800mg daily or placebo for 2 weeks; followup was 4 weeks after completing therapy.  Despite dissemination of the study’s results that suggested that fluconazole reduced invasive fungal infections (it did), the study was negative:  fever failed to resolve in a majority of patients in both arms, only ~37% of patients in each arm had a successful outcome (which included defervescence), and the incidence of invasive fungal infections was low overall.

• Schuster MG, Edwards JE, Jr., Sobel JD et al. Empirical fluconazole versus placebo for intensive care unit patients: a randomized trial. Ann Intern Med. 2008;149:83-90.

The second trial, looking at caspofungin versus placebo to prevent candidiasis in adults in hospital intensive care units, was a randomized trial involving 1200 patients at increased risk for invasive candidiasis (and was the motivation for developing the prediction rules by Ostrosky-Zeichner et al. above).  This study was completed 5 years ago, and has yet to be published.

Similarly, a meta-analysis of 4 studies looking at fluconazole prophylaxis in surgical ICUs failed to show any appreciable benefit in patient-specific outcomes.

• Shorr AF, Chung K, Jackson WL, et al. Fluconazole prophylaxis in critically ill surgical patients: a meta-analysis. Crit Care Med 2005;33:1928–35.

The current data therefore shows the following:

1. We cannot reliably predict who will get invasive candidiasis.  This includes using information related to colonization with candida.
2. Fluconazole prophylaxis is of no value in the ICU.
3. Empiric fluconazole for patients with recalcitrant fever is of no value in the ICU.
4. Caspofungin prophylaxis is unlikely to be of benefit in the ICU, but the relevant study has yet to be published.

Next topic …

This topic popped up while rounding with my colleagues in the Intensive Care Unit at Toronto Western Hospital.  It is an old favourite of mine.  The question:  what do you do with patients who have witnessed macroaspiration events?

I think this requires a step back:  the pathogenesis of most cases of pneumonia is aspiration.  Usually, a person harbours a potentially pathogenic bacterium in their pharynx, microaspirates this organism in sufficient quantities, and then the organism replicates in the lung parenchyma.  Voila, pneumonia!  Nobody usually refers to this syndrome as “aspiration pneumonia” even though that is exactly what it is.

However, the syndrome usually (but, as you will see, incorrectly) referred to as aspiration pneumonia occurs when pneumonia is diagnosed in a person who has vomited gastric contents, and then macro-aspirates these contents.  Aspiration pneumonitis, on the other hand, is simply the inflammatory lung response following the macroaspiration of the regurgitated (acidic) gastric contents.  Aspiration pneumonitis—also known as “Mendelson’s syndrome” after its initial description in post-anæsthetic obstetric patients (Mendelson CL. The aspiration of stomach contents into the lungs during obstetric anesthesia. Am J Obstet Gynecol 1946;52:191-205)—is due to acidic gastric contents, and usually requires a very low pH (<2.5).  It develops over hours, with the caustic effect of acid occurring in the first 1-2 hours after aspiration, followed by the acute inflammation occurring 2-4 hours later.  Because patients on acid-suppressing agents (such as H2-blockers and PPIs), those with enteral feeds, and patients with small bowel obstruction may have increased colonization of their gastric contents with bacteria, the risk of secondary pneumonia in such patients may be increased.

Aspiration pneumonia properly refers to a mechanism of developing pneumonia:  macro-aspiration of oropharyngeal contents, not gastric contents.  Risk factors include poor dentition, increasing age, and swallowing difficulties (e.g. Parkinsonism, ALS, stroke, etc.)  The distribution of aspiration pneumonia tends to be dictated by body positioning:  in patients who aspirate in a recumbent position, the most common sites of involvement are the posterior segments of the upper lobes and the apical segments of the lower lobes, whereas patients who aspirate in an upright of semirecumbent position tend to have the basal segments of the lower lobes affected.  (Marik PE. Aspiration pneumonitis and aspiration pneumonia.  NEJM 2001;344:665-71.)

In a rather classic paper from 1974, John Bartlett and colleagues demonstrated that macro-aspiration is accompanied by an increased risk for cavitation and abscess formation.  (Bartlett JG, Gorbach SL, Finegold SM. The bacteriology of aspiration pneumonia. Am J Med. 1974;56:202-7.)  They looked at 54 patients with lung infections associated with aspiration, of which 17 had an abscess and 10 had a necrotizing pneumonia.  Microbiologic sampling demonstrated anaerobes in almost all patients, with most such anaerobes being of oropharyngeal origin.  These patients were successfully treated with penicillin G, clindamycin or chloramphenicol.  It is noteworthy that most such patients had an abscess or a necrotizing infection.

However, more recent, high-quality, prospective data suggests that the microbiology of “aspiration pneumonia” parallels that of community-acquired pneumonia, with no anaerobic involvement.  In one study by Mier and colleagues, patients admitted to an ICU with already established aspiration pneumonia underwent bronchoalveolar lavage.  Only 19/52 patients had positive cultures, with the half of the identified organisms being S. pneumoniae, S. aureus, or H. influenzae.  (Mier L, Dreyfuss D, Darchy B, Lanore JJ, Djedaini K, Weber P, et al. Is penicillin G an adequate initial treatment for aspiration pneumonia? A prospective evaluation using a protected specimen brush and quantitative cultures. Intensive Care Med. 1993;19:279-84.)  Marik and Careau systematically evaluated patients with ventilator-associated pneumonia and aspiration pneumonitis using protected specimen brush sampling and mini-BAL in 143 patients (185 episodes) with suspected VAP and another 25 patients who required mechanical ventilation for aspiration peumonitis.  (Marik PE, Careau P. The role of anaerobes in patients with ventilator-associated pneumonia and aspiration pneumonia: a prospective study. Chest. 1999;115:178-83.)  Despite their best efforts, they could only isolate one anaerobe.  As expected, aspiration pneumonia was usually caused by organisms commonly causing community-acquired pneumonia, whereas VAP was caused by nosocomial pathogens (such as Pseudomonas aeruginosa).

There are no trials evaluating the strategy of empirically treating aspiration pneumonitis with antibiotics compared to placebo, and there are no trials evaluating the need for anaerobic coverage of aspiration pneumonia vs routine management of community-acquired pneumonia.  Based on the available data, it would appear:

1. trials are needed
2. there is no data to support the practice of routine antimicrobials following macro-aspiration
3. there is no data demonstrating the role of anaerobes as a cause of aspiration pneumonia
4. if you are an antimicrobial minimalist (like me), a reasonable strategy is to only use antimicrobials when patients have an obvious pneumonia, and that the choice of antimicrobials for patients with community-acquired aspiration pneumonia should cover the usual causes of CAP (i.e. S. pneumoniae and H. influenzae), whereas the treatment of pneumonia acquired in-hospital should also account for nosocomial pathogens (using local susceptibility data).

In one of my lectures (Immunodeficiency III), I have a list of conditions associated with hypogammaglobulinemia, and I list cirrhosis.  A student asked me to explain after class, and then another one emailed me on this.  I was incorrect.  Cirrhosis, in fact, is often associated with hypergammaglobulinemia.  Nevertheless, patients with cirrhosis and ascites are at tremendously increased risk of SBP (spontaneous bacterial peritonitis) because they do have hypogammaglobulinemia in their ascitic fluid.  This is a major difference between ascites due to portal hypertension and other forms of ascites such as cancer or infection, where the peritoneal lining is permeable, and thus immunoglobulins (amongst other molecules) can pass through.  Patients with cirrhosis due to portal hypertension also have engorged venous vessels in the gut, which further facilitates trans-cœlomic translocation of bacteria.

I hope this clarifies this.  Further evidence that medical students are smarter than me.

Let me start off by saying the following:  get vaccinated.  I am now tired of all the nonsense regarding safety of the vaccine.  If you don’t believe in vaccines in general, then you know very little about one of the only true miracles of modern medicine and/or listen to the few heretics who mislead by saying things people want to hear.  About 50 million people died of pandemic flu in 1918 and we may very well see as many (if not more) this time around.  Very little is different (other than we have antibiotics and antivirals, but those who don’t believe in vaccines also probably don’t believe in antimicrobials).  We have been using the flu vaccine since around the Second World War, and the methods to develop the current H1N1 vaccine is only better/safer than what was used in the 50s and 60s.  Regarding any potential risk of the vaccine (and there are some small ones):  just because you might rupture your aorta from being in a car accident while wearing a seatbelt doesn’t mean that it isn’t an effective safety measure.  Similarly, just because some people have adverse reactions to the flu vaccine doesn’t mean that it isn’t an effective measure.  If you are someone who doesn’t wear seatbelts when in your car, don’t get vaccinated.

Okay, now that I have that of my chest:  I had the pleasure and opportunity to speak at the Critical Care Canada Forum in Toronto, today, regarding H1N1.  Attached are my slides.  What I want to emphasize is the importance of bacterial pneumonia/superinfection.  There are some reasonable arguments to my perspective, so let me counter them now:

• in recent autopsy specimens, there is evidence of pneumonitis, and no bacterial pneumonia, so this is different from 1918.
  • this is correct, however, the pathology from 2009 H1N1 looks exactly like the pathology from 1918 H1N1.  The only difference is the presence of bacteria.  The difference?  We have antibiotics, so no self-respecting bacteria should be alive at the time of death in 2009.  This does not mean that the initiating process was not bacterial pneumonia.
• in current studies, only about a quarter of patients had clear evidence of bacterial infection (i.e. positive blood cultures or respiratory tract cultures).
  • again, part of this can be explained by the use of antimicrobials.  However, a more important factor is that our ability to microbiologically diagnose pneumonia is poor.  We usually only get a positive culture in 10% of patients with pneumococcal pneumonia.  Getting a 25% yield in any series of pneumonia is actually very high.
• there is evidence of pneumonitis, patients are dying with evidence of viral infection, and antivirals seem to make a difference.  Also, such pneumonitis can be induced in animals artificially infected with influenza.
  • yes, there is evidence of pneumonitis, but S. pneumoniæ can do the same thing.  In fact, this was shown as long ago as the 1920s, in the seminal works by Stillman and Branch, who showed that many animals infected with pneumococcus developed a pneumonitis that largely resembles the findings we are seeing in 2009 influenza pneumonitis.  Further, S. pneumoniæ and influenza virus probably work synergistically:  influenza is in no way an innocent bystander.  It probably often needs a bacterium to cause death.

Stay tuned!

I cover a series of lectures on various forms of immunocompromise for MD Program at the University of Toronto, Pathobiology of Disease (which is in second year).  In one of the lectures, I mention how late component complement deficiency (LCCD) is associated, paradoxically, with more frequent meningococcal infections yet initial infections are diagnosed later and tend to be milder.  Karen Lam, a student in the class, wanted to know “what gives?”

Well, first let me state that LCCD is relatively uncommon.  Congenital complement deficiency occurs in about 1 in 3000 people.  LCCD accounts for roughly half of these, although there is a wide variety of prevalences listed, varying according to methods and geographical location.

Neisseria species are commonly encountered by humans, with many different strains being encountered throughout childhood.  Up to 40% of the general population are colonized with Neisseria sp. in their nasopharynx.  Most such species lack the polysaccharide capsule (a clear virulence factor) and thus are non-typable.  Almost certainly, “infection” is asymptomatic or, at least, unrecognized at an early age.

The Neisserial species that cause severe disease are generally types A, B, C, W-135 and Y.  These all have a capsule, that prevent complement-mediated bacteriolysis.  To the contrary, the non-capsular strains (which make up the majority of colonizing flora) are particularly susceptible to complement-mediated destruction, and rarely cause disease. Complement activation, however, is also deleterious and is responsible for much of the severity of invasive meningococcal disease, whether it be meningitis or simply meningococcæmia with septic shock (and possibly DIC, disseminated intravascular coagulation).  Studies dating back as far as the late 1960s and early 1970s have demonstrated this.  However, it has become apparent more recently that neisseria manipulate complement (and even elaborate negative complement mediators) to avoid complement-mediated lysis while enhancing the release of anaphyloxins and other cytokines from complement.  Thus, the presence of complement is very important to the severity of disease (and being deficient in complement would be expected to attenuate disease).  We know, for example, that patients with LCCD are 17 times less likely to die that patients with normal levels of complement.

• Densen P. Interaction of complement with Neisseria meningitidis and Neisseria gonorrhoeae. Clin Microbiol Rev. 1989;2 Suppl:S11-7.

However, we also know that patients with LCCD get their first recognized episode of severe meningococcal disease considerably later (avg. 17 years) compared to normal subjects (avg. 5 years).  The reason for this is not entirely certain.  However, the likely explanation is that patients with LCCD are getting infected earlier, but it doesn’t get recognized because the disease is so relatively mild because of their complement deficiency.  Thus, they aren’t getting infected later, but we are only recognizing it later.  This would be consistent with the observation that persons with LCCD frequently get infected with Neisseria sp. (with a recurrence rate of meningococcal infection approximately 100 times that of normal subjects).

In anticipation of the Ontario Hospital Association (in conjunction with the Institute for Safe Medication Practices Canada) conference titled “Preventing Antimicrobial Resistance Through Antimicrobial Stewardship”, I thought I would post the now long-overdue Business Case Template for those wishing to obtain funding from their hospital administrators for funds.  I had promised such a template back in June at a preceding conference, but was unable to get around to it.  Well, now the document is now here (just click on it below), and should save a fair amount of time for those wishing to use it.  Although it is copyrighted, I have not protected the document, and thus am hereby allowing all who wish to use it to do so freely.  Areas that will definitely require modification/customization have been highlighted.  I suggest, again, that the Business Case Analysis (also posted here as a Microsoft Excel file, or found on my Docs page) be used and included as Appendix E.  Further, I take no responsibility for any submissions used without my consultation:  caveat emptor (and you get what you pay for!!!)

As always, any feedback is welcome.

Generic_ASP_Business_Case

ASP_Business_Case

I thought I would update a post from July on empiric therapy for febrile neutropænia.  In that post, I mentioned a 2006 systematic review and meta-analysis by Paul M, Yahav D, Fraser A, and Leibovici L (J Antimicrob Chemother.  2006;57:176-89) that showed that cefepime monotherapy was inferior to other treatments.  Based on this paper, the FDA performed their own meta-analyses using additional data from the manufacturer and posted it online in June of this year.  I will copy their “bottom line” and include the link to their full statement:

FDA performed meta-analyses based on additional data beyond those included in the Yahav et al. publication. In the FDA analyses, no statistically significant increase in mortality was seen in cefepime-treated patients compared to comparator-treated patients.

Based on the results of FDA’s meta-analyses, the FDA has determined that cefepime remains an appropriate therapy for its approved indications.

Last week (pre-Labour Day), the issue of investigation and management of vascular catheter-associated Staph. aureus bacteræmia (SAB) came up.  Vascular catheter-associated SAB is of increasing importance because of the dramatic rise in the use of vascular catheters.  Here in downtown Toronto (at Mount Sinai Hospital and University Health Network) the Infectious Diseases consult service gets notified by the microbiology laboratory of all blood cultures growing Staph. aureus.  While rounding in the ICU last week, issues of a) duration of therapy and b) extent of investigation arose.  Knowing the literature quite well (or so I thought!), and recognizing the controversy surrounding this topic, I thought a blog was the best way to tackle it.  Plus a survey.  I sent out the following email to 23 ID physicians and 5 ID fellows:

60 yo patient with COPD, hypertension, hyperlipidemia and some sort of central line (e.g. PICC or IJ catheter) for approximately 2 weeks’ duration (indication unclear, and now removed). No other increased risk for endocarditis, and no other clinical suggestion of endocarditis or suppurative thrombophlebitis. Fever developed while the line was in, and patient has 2 sets of blood cultures drawn, both of which grow MSSA. No urinalysis was done. ECG normal. The patient defervesces after about 36 hours with appropriate antimicrobial therapy.  You are consulted on day 2 of therapy.
Do you recommend:
a) treat for 2 weeks with iv β-lactam and no further investigation
b) treat for 4 weeks with iv β-lactam and no further investigation
c) treat for 2 weeks and get TTE: if positive, treat for 4 weeks
d) treat for 2 weeks and get TEE: if positive, treat for 4 weeks
e) give an option of b or d based on patient or physician preference
f) none of the above (please specify)

The results?

With 17/28 (61%) respondents, the results favour the conservative approach of either giving everyone 4 weeks of treatment or treating for 2 weeks with a TEE guiding prolonged (additional 2 weeks) therapy.

So what does the literature say?

If one goes to the recent Infectious Diseases Society of American Guidelines for catheter-related bloodstream infection (1), it is clear that the “guided” management of the above patient with SAB would include:

• removal and culture of the catheter tip
• blood cultures drawn from the catheter and a peripheral vein prior to starting antimicrobial therapy
• repeat blood cultures once starting therapy
• 4-6 weeks of antimicrobial therapy unless patient is not diabetic, is not immunosuppressed (e.g. transplantation, neutropenia), has had infected catheter removed, has no prosthetic intravascular devices, has no evidence of infective enodcarditis or suppurative thrombophlebitis on TEE and US, has had resolution of fever and bacteræmia within 72h of starting antibiotics, and there is no evidence of metastatic infection on physical examination and investigations
• TEE should be done at least 5-7 days after the onset of bacteræmia (and TTE should not be done because of its poor performance characteristics vis a vis infective endocarditis)

So the IDSA basically agrees with those who picked option d (i.e. treat for 2 weeks and get a TEE, prolonging therapy to 4 weeks if TEE positive).  What does the evidence show?
To start off, let me provide 2 rationales why one would recommend prolonged (4-week) therapy for catheter-related SAB:
1.  patients with SAB have infective endocarditis, and thus prolonged treatment is treatment of infective endocarditis
2.  patients with SAB are at risk of developing infective endocarditis, and thus prolonged treatment is for prophylaxis of infective endocarditis

These assumptions are both problematic.  In the first instance, it would seem rather unlikely that patients with a newly recognized infection while in hospital would already have “established S. aureus endocarditis” at the time of blood culture, although this is largely based on clinical experience and a rather old (but classic) retrospective study of 105 cases of SAB that showed that endocarditis was unlikely in patients with SAB unless infection was acquired in the community, had no apparent source (as opposed to a “removable focus”), or there was a metastatic focus of infection.(2)  In the second instance, it is difficult to  believe that prolonging already successful treatment of a bacteræmia actually prevents secondary “metastatic” infection during treatment.
Concern of endocarditis complicating SAB goes back (according to my research) to 1957, with a report by Wilson and Hamburger on their experience with SAB.  They reported that 64% of their patients with SAB had endocarditis.(3)  A “standard” of treating all cases of SAB with 4 weeks of antimicrobial therapy ensued.  (My colleague and mentor Hillar Vellend confirmed this traditional approach.)  However, in 1976 a small retrospective series looking at patients with SAB who had a removable focus found that all patients treated for 10-21 days (mean 15) were cured without relapse or evidence of subsequent endocarditis.(4)  A host of observational studies exploring “short-course” therapy for SAB ensued, leading to a well-performed meta-analysis by Jernigan and Farr in 1993 that concluded that studies claiming success of short-course therapy are too problematic because of “bias and statistical imprecision” to demonstrate safety.(5)  Of note, despite the limitations of the studies involved (which tended to favour short-course therapy), short-course therapy was associated with an approximately 6% mean rate of “late complications”.  A subsequent retrospective study exploring short-course therapy also found an approximately 5% “relapse” rate in patients who received short-course therapy, compared with no relapses in patients receiving 4 weeks of antimicrobial therapy.(6)

Fowler et al. looked at 103 consecutive inpatients with SAB (about 41% had underlying heart disease).(7)  Clinical findings, underlying heart disease and vascular catheter source did not distinguish patients with and without infective endocarditis.  Definite endocarditis (Duke criteria) was present in 25% of patients with SAB after TEE was performed; had only a TTE been performed, only 7% of patients would have met Duke criteria for IE.  Neither duration of therapy nor outcomes were reported, and so the clinical relevance of identifying echocardiographic endocarditis is uncertain.
Other valuable data comes from a large-scale prospective, multicentre, observational study of SAB performed in the US by a group of well-respected researchers at 6 hospitals led by Victor Yu of the University of Pittsburgh.(8)  In this study, performed in the mid-1990s, 505 consecutive patients with SAB were followed prospectively.  About half had 2 positive blood cultures, and a third had 1 positive blood culture.  All patients were monitored for at least 6 months’ patients with endocarditis were monitored for 3 years.  Importantly, catheter-related SAB required a positive semi-quantitative culture of the line tip, and endocarditis was defined according to Duke criteria.(9)

The authors found that 13% of patients with SAB had either possible or definite infective endocarditis (10% had definite IE according to the Duke criteria).  Of note, although the likelihood of having IE in the setting of SAB was highest (21%) in patients with community-acquired SAB, it was also common in hemodialysis patients (12%) and hospitalized patients (5%);   Looking at all nosocomial SAB (250 cases), “new” endocarditis developed in 4% who received up to 14 days of therapy, in 4% who received 15-27 days of therapy, and in none of the 33 patients who received 4 weeks or more of therapy; all such patients were treated with vancomycin.  Not all patients underwent echocardiography, so this could be an under-estimate of endocarditis in all patient sub-groups.
So, what can we conclude from this huge volume of information:

1. The best evidence shows that there is a “failure” rate of short-course therapy for SAB, and this “failure” rate is somewhere in the order of 4-7%.
2. If we go through the various treatment and investigation options listed earlier, the current best information in the literature suggests:

a) Treating all patients with vascular catheter-associated SAB and no clinical features of IE with 2 weeks of iv β-lactam will cure most patients with no unnecessary antibiotic use, but will likely have a relapse/endocarditis rate of 4-7%.

b) Treating all patients for 4 weeks with iv β-lactam and no further investigation will have a negligibly low relapse/endocarditis rate, but will result in 93-96% of patients unnecessarily receiving an additional 2 weeks of intravenous antimicrobial therapy (which is not benign, but is less so than that of developing endocarditis or a metastatic focus of infection).

c) Treating all patients for 2 weeks and getting a TTE (with total 4 weeks therapy if TTE+) will result in a relapse/endocarditis rate of ~2-4%, and will be unlikely lead to unnecessary prolonged iv antimicrobial therapy, but carries the cost of TTEs.

d) Treating all patients for 2 weeks and getting a TEE (again, total 4 weeks therapy if TEE+) will result in a negligibly low relapse/endocarditis rate and unnecessary iv antimicrobial therapy use, but carries the morbidity and cost of everyone receiving a TEE.

It is easy to understand why there are so many different approaches, each with their own downside.  Me?  I am going to stick with 4 weeks iv therapy unless there is a negative TEE, giving my patients their own choice.  I have been also “challenged” to find out if we have any treatment failures of SAB in our institution, knowing that at least a reasonable percentage will receive 2 weeks antimicrobial therapy.  It is hard to believe we are any different than anywhere else but … stay tuned.

References

1.    Mermel LA, Allon M, Bouza E, Craven DE, Flynn P, O’Grady NP, et al. Clinical practice guidelines for the diagnosis and management of intravascular catheter-related infection: 2009 Update by the Infectious Diseases Society of America. Clin Infect Dis. 2009 Jul 1;49(1):1-45.
2.    Nolan CM, Beaty HN. Staphylococcus aureus bacteremia. Current clinical patterns. Am J Med. 1976 Apr;60(4):495-500.
3.    Wilson R, Hamburger M. Fifteen years’ experience with Staphylococcus septicemia in a large city hospital:  Analysis of fifty-five cases in the Cincinnati General Hospital 1940 to 1954. Am J Med. 1957 March 1957;22(3):11.
4.    Iannini PB, Crossley K. Therapy of Staphylococcus aureus bacteremia associated with a removable focus of infection. Ann Intern Med. 1976 May;84(5):558-60.
5.    Jernigan JA, Farr BM. Short-course therapy of catheter-related Staphylococcus aureus bacteremia: a meta-analysis. Ann Intern Med. 1993 Aug 15;119(4):304-11.
6.    Malanoski GJ, Samore MH, Pefanis A, Karchmer AW. Staphylococcus aureus catheter-associated bacteremia. Minimal effective therapy and unusual infectious complications associated with arterial sheath catheters. Arch Intern Med. 1995 Jun 12;155(11):1161-6.
7.    Fowler VG, Jr., Li J, Corey GR, Boley J, Marr KA, Gopal AK, et al. Role of echocardiography in evaluation of patients with Staphylococcus aureus bacteremia: experience in 103 patients. J Am Coll Cardiol. 1997 Oct;30(4):1072-8.
8.    Chang FY, MacDonald BB, Peacock JE, Jr., Musher DM, Triplett P, Mylotte JM, et al. A prospective multicenter study of Staphylococcus aureus bacteremia: incidence of endocarditis, risk factors for mortality, and clinical impact of methicillin resistance. Medicine (Baltimore). 2003 Sep;82(5):322-32.
9.    Durack DT, Lukes AS, Bright DK. New criteria for diagnosis of infective endocarditis: utilization of specific echocardiographic findings. Duke Endocarditis Service. Am J Med. 1994 Mar;96(3):200-9.

This issue came up yesterday while rounding in the ICU with Dr. Mike Christian–a physician dually trained in Infectious Diseases and Critical Care Medicine–and his team.  I will start by pointing out that what I am really referring to is “chemotherapy-induced febrile neutropænia” (or CIFNP).  This differs from other forms of neutropænia because not only do the patients lack neutrophils, but they lack platelets and are anaemic (thus requiring transfusional support), and have intravascular catheters.  These are relevant because:

1. platelets carry a very high risk of contamination (~1:1000-1:2000) because they are stored at much warmer temperatures than RBCs/WBCs
2. percutaneous intravascular catheters are a major source of infection

Overall, the care of such patients with CIFNP has improved considerably, with mortality dropping from 21% to 7% from 1978-1994.  Some of this likely relates to better (and safer) chemotherapeutic regimens, but they also undoubtedly relate to improved  management of venous catheters, safer transfusion practices, improvements in critical care medicine, and improved prevention and management of infectious complications.  Some of this benefit may also derive from a shift away from gram negative infections (which probably cause more “sepsis” on average than gram-positive infections like enterococcus):  whereas there were 3 times as many gram-negative infections as gram-positive infections isolated during febrile episodes in the early 1970s, there are slightly more gram-positive infections in recent trials.  The number of febrile episodes where an organism has been isolated has been consistent over time, however—somewhere around 20-25%. Read more…

In the July 9 issue of the New England Journal of Medicine, Kent Sepkowitz has it all wrong in his “Perspective” piece.  He states that the emergence of community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA) “has proved an embarrassment to scientists everywhere”, and that it is evidence that we have been falsely preaching about antimicrobial prudence.  How could an NEJM editorialist (or, more correctly, “perspectivist”) get it wrong? Read more…