by Keith Armitage, MD, Division of Infectious Diseases, Vice Chair for Education, Department of Medicine, Case Medical School/University Hospitals of Cleveland, Cleveland, OH
In my experience, hospitalists usually have a greater knowledge of antibiotics and treatment of infections than other non-infectious disease (ID) practitioners who manage hospital patients. But that doesn’t stop ID physicians from wanting to make suggestions. The following list is not meant to be all-inclusive, but it does reflect an informal poll of my colleagues at a tertiary care medical center. Any opinions are of course my own, and naturally are evidenced based. There is an old joke that if you ask two ID doctors a question you get three answers. Having said that, I believe that there is a good consensus on these issues.
Beta-lactam/beta-lactamase inhibitors such as ampicillin/sulbactam and piperacillin/tazobactam have excellent anaerobic coverage. When treating suspected or proven anaerobic infections with these drugs, addition of other agents such as metronidazole and clindamycin to cover anaerobic infections is not necessary (1). Quite often we see patients treated with ampicillin/sulbactam and metronidazole or piperacillin/tazobactam and metronidazole, which is not necessary and potentially exposes the patient to additional drug toxicities. “Unasyn and Flagyl” for suspected intra-abdominal infections provides unnecessary double coverage for anaerobes, while providing suboptimal coverage for gram-negative rods due to increasing resistance to ampicillin/sulbactam among gram-negative aerobes.
Clinicians managing patients who have blood cultures positive for Staphylococcus aureus should always think about whether the patient has a deep-seated source such as cardiac or bone, and treat accordingly. But even patients with a self-limited bacteremia related to an intravenous catheter or other easily removable source of infection should get at least 2 weeks of antibiotics (2). One of the goals of treating S. aureus bacteremia is to prevent metastatic infection. Patients with line-related infections may have rapid clinical improvement and resolve the bacteremia quickly, but they are at high risk for relapse with bone, joint, or cardiac infection if the initial antimicrobial course is inadequate. At least once every year or 2 at our teaching hospitals we see a patient who is given a very short course of antibiotics for S. aureus bacteremia related to an intravenous catheter who returns a month or two later with relapse in the spine or in other equally serious sites of infection. It is believed that more aggressive initial antimicrobial therapy can prevent metastatic infections. A frequently employed strategy in evaluating patients with S. aureus bacteremia is to complete a transesophageal echocardiogram to rule out cardiac involvement. If there is no cardiac involvement and other deep seated source such as bone or joint are not suspected, 2 weeks of intravenous antibiotics are generally adequate. In the setting of cardiac involvement or deep-seated involvement in bone or joints, 4–6 weeks of antibiotics are required. Patients with staphylococcal bacteremia who do not have a known source of infection should almost always be treated for 4–6 weeks. And, of course, there is the usual caveat: Oral antibiotics with excellent oral bioavailability such as linezolid can be used as switch therapy to complete a 2-week course in some cases. But the default approach would be to treat all S. aureus bacteremias with at least 2 weeks of intravenous antibiotics.
Patients who have Foley catheters or are status post-genitourinary procedures may develop primary S. aureus urinary tract infection, but it is unusual for patients without history of genitourinary manipulation to present with S. aureus as a cause of urinary tract infection. In patients with no predisposing factors for S. aureus urinary infection, isolation of the organism in the urine should always prompt an evaluation for another site of infection such as bone, joint, or endovascular. Patients with known or suspected S. aureus urinary tract infection should have blood cultures drawn prior to the initiation of antibiotics to detect occult bacteremias. It is not unusual on our Infectious Disease Consult Service to see a patient who is suspected of having a S. aureus UTI that is later shown after consultation and investigation to have S. aureus endocarditis without other obvious manifestations, or another deep-seated infection such as spinal osteomyelitis or epidural abscess.
North America is experiencing an explosion in illness related to Clostridium difficile (primarily C. difficile associated diarrhea, or CDAD) (3). C. difficile is most often a nosocomial infection, and CDAD has become a very common disease in patients hospitalized for any length of time who are given broad-spectrum antibiotics. Hospitalization is the “perfect storm” for CDAD. C. difficile spores exist in the hospital environment and are ingested by patients on broad-spectrum antibiotics that inhibit the normal flora, creating the perfect environment for C. difficile to flourish. The assays to detect C. difficile toxins in the stool are often not highly sensitive; under the best of circumstances the techniques used by most hospital labs will produce a false-negative result 10–20% of the time (4). Not all labs detect all toxin types, and not all kits are highly efficient for detecting toxins. Patients in whom there is a strong suspicion for CDAD should be treated even in the face of negative toxin assays unless there is another likely source of diarrhea.
Patients who develop diarrhea after being in the hospital 1 or 2 days almost never have infection with Salmonella, Campylobacter, Entamoeba histolytica, or Giardia sp. It is not uncommon to see hospitalized patients who develop diarrhea after several days in the hospital with shotgun orders for “SSYC and O&P.” Unless there is a history of immunosupression or risk factors for enteric infection such as international travel, these tests are almost always unnecessary (5). 6. The most common cause of leukocytosis of unknown etiology in hospitalized patients is CDAD. There is something about the pathogenesis of CDAD that produces a leukemoid reaction much more often than other infections do. It is not unusual to see white blood cell counts of 30,000 with CDAD, and counts of 50,000 and higher in patients with CDAD are not rare. Patients who develop leukocytosis in the hospital while on antibiotics, or who present from long-term care facilities with marked leukocytosis and recent antibiotic exposure, have a high pretest probability of having CDAD (6). In this setting, the higher the white count, the more likely the patient has CDAD.
Patients who are given broad-spectrum antibiotics have 1 opportunity to have interpretable blood cultures obtained: before antibiotics are administered. Once patients are given broad-spectrum antibiotics, blood cultures have a very limited value in diagnosing infections that might not be initially suspected on admission. A common example in our hospital is a patient presenting with pneumonia. About a third of the patients who come through the emergency room with a diagnosis of community-acquired pneumonia end up having another diagnosis. Often the alternative diagnosis is suspected based on blood cultures obtained prior to the patient receiving broad-spectrum antibiotics in the emergency room. In the last 3 months we have seen patients with liver abscesses, endocarditis, and osteomyelitis initially felt to have community-acquired pneumonia whose blood cultures initiated prior to antibiotic therapy revealed a pathogen that caused a search for an alternative source of infection. The vast majority of patients only need 2 blood cultures from 2 sites 20 minutes apart before initiation of antibiotic therapy. Patients in whom common skin contaminants may easily be interpreted as pathogens (such as patients with prosthetic heart valves) should have 3 sets of blood cultures to aid in the interpretation of cultures that are positive for skin contaminants such as coagulase negative staph.
Diabetic patients who have infections related to foot ulcers or ischemic lesions require broad-spectrum antimicrobial therapy active against anaerobes, gram-positives, and gram-negatives. However, diabetic patients who are not critically ill who are admitted with a clinical picture typical for cellulitis tend to be infected with the same pathogens as non-diabetic patients. We frequently encounter diabetic patients who present with a clinical picture of an uncomplicated cellulitis without ulcers or other lower-extremity lesions and are treated with broader-spectrum antimicrobial therapy than is needed for cellulitis. Broader therapy is often more expensive, and it puts patients at risk for more adverse effects such as CDAD. The great majority of patients with cellulitis have infection with group A strep and other streptococci, and less often S. aureus. Cellulitis due to anaerobes and gram-negative organisms in the absence of foot ulcers or similar lesions is distinctly unusual.
Another “pearl” about cellulitis: Group A strep cellulitis is often initially slow to respond to therapy. The local findings may take 3 or 4 days to show improvement and there actually may be slight worsening despite 1 or 2 days of appropriate antibiotics. This is believed to be related to toxins produced by group A strep and other local tissue factors. Even if an antimicrobial is successful in eradicating strep, there are still toxins in the tissues that produce aggressive local findings. We often get consulted about patients with cellulitis who after 2 days of antimicrobial therapy may have some improvement in their fever curve and white blood cell count but have worsening of the local findings. These patients almost never need a change in antimicrobial therapy, but need more time—and elevation. I was taught by one of my mentors of the importance of elevating an extremity when treating cellulitis. My clinical experience has borne out this wisdom. In addition, patients with lower-extremity edema or venous insufficiency or venous stasis who present with cellulitis must have edema and stasis aggressively treated for the cellulitis to respond to antimicrobial therapy.
Five years ago in Ohio, if a patient presented with pyelonephritis or a complicated UTI as a community-acquired infection, it was unusual for the causative pathogen to be quinolone resistant. Quinolones such as ciprofloxacin could be used as empiric therapy for serious gram-negative infections with a great deal of confidence that the causative agent would be sensitive. In the last 5 years we have seen a steady, progressive increase in resistance to quinolones in both community acquired and nosocomial infections (7,8). Approximately 5–10% of E. coli are now quinolone resistant, and in some hospitals more than half of Pseudomonas aeruginosa are now quinolone resistant. Seriously ill patients with infections that are likely due to gram-negative rods should not be treated empirically with quinolone monotherapy in most settings. Oral quinolones, due to their excellent oral bioavailability, continue to have in important role in treating gram-negative infections, but their use should be based on the results of a culture with antimicrobial susceptibility.
The great majority of patients who test positive for VRE in a stool specimen never acquire an infection with VRE. Patients who are colonized with VRE in the stool will clear colonization over several weeks or months if there is no antimicrobial pressure to select for VRE. Infectious disease clinicians spend a lot of time trying to allay the fear of patients and families who become extremely nervous due to isolation procedures for VRE. My usual approach is to tell the patients that the only reason they are in isolation is to prevent VRE from spreading to the very, very small group of patients who actually are susceptible to infection with VRE, such as liver transplant patients. I tell the family there is almost no chance that healthy family members will develop a VRE infection and that the VRE bacteria is normally found as a natural part of the human intestinal flora. VRE is simply 1 strain that has particular resistance to antibiotics, making it difficult to treat when infection occurs, but it is not more pathogenic. Infection with VRE is relatively rare and with the possible exception of cystitis (or bladder colonization) there is an extremely low risk of any actual infection despite VRE colonization. Uncomplicated cystitis due to VRE can usually be treated with nitrofurantoin.
In the last 5 years in the United States, there has been a steady increase in MRSA infections in patients without traditional risk factors (9,10). Historically, clinicians have been concerned about MRSA in nursing home patients, patients in other long-term care facilities, injection drug users, and hospitalized patients. In the last 5 years there have been increasing numbers of patients with MRSA with none of these risk factors. Often these patients present with a serious life-threatening S. aureus infection. It is now appropriate to give vancomycin empirically for patients who have serious illnesses due to suspected S. aureus even if they don’t have traditional risk factors for MRSA. As ID practitioners, we do not want to encourage overuse of vancomycin, and clinicians should quickly switch to other agents if the patient proves not to be infected with MRSA. While vancomycin is a useful drug, it is considered inferior to the beta-lactams for many infections, such as bone or joint infections, and should only be used in patients with documented or suspected MRSA, or patients intolerant of beta-lactams. Several new drugs provide alternatives to vancomycin for MRSA, including linezolid and daptomycin. Both of these agents are more expensive and have not proven in clinical trials to be superior (with the possible exception of linezolid for MRSA pneumonia). Linezolid offers the advantage of having excellent oral availability; however, oral linezolid use is complicated by its high cost. Oral linezolid costs approximately $100 a day, and in almost all cases the use of this drug must be preapproved before an insurance company will pay for it. Insurance companies will almost always approve oral linezolid if the only alternative is continued hospitalization, skilled nursing home placement, or home IV antibiotic therapy. Trimethoprim/sulfa is a much less expensive alternative to oral linezolid for MRSA, and is very useful for less serious MRSA infections such as UTIs. About 85% of MRSA strains are sensitive to trimethoprim/sulfa. Many community-acquired MRSA strains are clindamycin susceptible, and minocycline and doxycycline have activity against many MRSA strains.
So these are 10 (11) things ID physicians wish all hospitalists knew. The 12th is that we enjoy working with our hospitalist colleagues, so please call when you think you need us.
Dr. Armitage may be reached at firstname.lastname@example.org.
The Hospitalist newsmagazine reports on issues and trends in hospital medicine. The Hospitalist reaches more than 25,000 hospitalists, physician assistants, nurse practitioners, residents, and medical administrators interested in the practice and business of hospital medicine.