Bacterial bloodstream infections (BSIs) are a major cause of morbidity and mortality in the United States. Approximately 600,000 BSI cases occur annually, resulting in 85,000 deaths,1 at a cost exceeding $1 billion.2 Traditionally, BSIs have been managed with intravenous antimicrobials, which rapidly achieve therapeutic blood concentrations, and are viewed as more potent than oral alternatives. Indeed, for acutely ill patients with bacteremia and sepsis, timely intravenous antimicrobials are lifesaving.3
However, whether intravenous antimicrobials are essential for the entire treatment course in BSIs, particularly for uncomplicated episodes, is controversial. Patients that are clinically stable or have been stabilized after an initial septic presentation may be appropriate candidates for treatment with oral antimicrobials. There are costs and risks associated with extended courses of intravenous agents, such as the necessity for long-term intravenous catheters, which entail risks for procedural complications, secondary infections, and thrombosis. A prospective study of 192 peripherally inserted central catheter (PICC) episodes reported an overall complication rate of 30.2%, including central line-associated BSIs (CLABSI) or venous thrombosis.4 Other studies also identified high rates of thrombosis5 and PICC-related CLABSI, particularly in patients with malignancy, where sepsis-related complications approach 25%.6 Additionally, appropriate care of indwelling catheters requires significant financial and healthcare resources.
Oral antimicrobial therapy for bacterial BSIs offers several potential benefits. Direct economic and healthcare workforce savings are expected to be significant, and procedural and catheter-related complications would be eliminated.7 Moreover, oral therapy provides antimicrobial stewardship by reducing the use of broad-spectrum intravenous agents.8 Recent infectious disease “Choosing Wisely” initiatives recommend clinicians “prefer oral formulations of highly bioavailable antimicrobials whenever possible”,9 and this approach is supported by the Centers for Disease Control and Prevention antibiotic stewardship program.10 However, the expected savings and benefits of oral therapy would be lost should they be less effective and result in treatment failure or relapse of the primary BSI. Pathogen susceptibility, gastrointestinal absorption, oral bioavailability, patient tolerability, and adherence with therapy need to be carefully considered before choosing oral antimicrobials. Thus, oral antimicrobial therapy for BSI should be utilized in carefully selected circumstances.
In this narrative review, we highlight areas where oral therapy is safe and effective in treating bloodstream infections, as well as offer guidance to clinicians managing patients experiencing BSI. Given the lack of robust clinical trials on this subject, the evidence for performing a systematic review was insufficient. Thus, the articles and recommendations cited in this review were selected based on the authors’ experiences to represent the best available evidence.
Infection Source Control
Pathogen and Antimicrobial Factors
Evidence Regarding Bloodstream Infections due to Gram-Negative Rods
BSIs due to gram-negative rods (GNRs) are common and cause significant morbidity and mortality. GNRs represent a broad and diverse array of pathogens. We focus on the Enterobacteriaceae family and Pseudomonas aeruginosa, because they are frequently encountered in clinical practice.1
Gram-Negative Rods, Enterobacteriaceae Family
The Enterobacteriaceae family includes Escherichia coli, Klebsiella, Salmonella, Proteus, Enterobacter, Serratia, and Citrobacter species. The range of illnesses caused by Enterobacteriaceae is as diverse as the family, encompassing most body sites. Although most Enterobacteriaceae are intrinsically susceptible to antibiotics, there is potential for significant multi-drug resistance. Of particular recent concern has been the emergence of Enterobacteriaceae that produce extended-spectrum β-lactamases (ESBL) and even carbapenem-resistant strains.14
However, Enterobacteriaceae species susceptible to oral antimicrobials are often suitable candidates for oral BSI therapy. Among 106 patients with GNR BSI treated with a highly bioavailable oral antibiotic (eg, levofloxacin), the treatment failure rate was only 2% (versus 14% when an antimicrobial with only moderate or low bioavailability was selected).15 Oral treatment of Enterobacteriaceae BSIs secondary to urinary tract infection has been best studied. A prospective randomized, controlled trial evaluated oral versus intravenous ciprofloxacin amongst 141 patients with severe pyelonephritis or complicated urinary tract infections, in which the rate of bacteremia was 38%.16 Notably, patients with obstruction or renal abscess were excluded from the trial. No significant differences in microbiological failure or unsatisfactory clinical responses were found between the IV and oral treatment groups. Additionally, a Cochrane review reported that oral antibiotic therapy is no less effective than intravenous therapy for severe UTI, although data on BSI frequency were not provided.17
Resistance to fluoroquinolones such as ciprofloxacin has been identified as a risk factor for GNR BSI oral treatment failure, highlighting the importance of confirming susceptibilities before committing to an oral treatment plan.18,19 Even ESBL Enterobacteriaceae can be considered for treatment with fluoroquinolones if susceptibilities allow.20
The ideal duration of therapy for GNR BSI is an area of active research. A recent retrospective trial showed no difference in all-cause mortality or recurrent BSI in GNR BSI treated for 8 versus 15 days.21 A recent meta-analysis suggested that 7 days of therapy was noninferior to a longer duration therapy (10–14 days) for pyelonephritis, in which a subset was bacteremic.22 However, another trial reported that short course therapy for GNR BSI (<7 days) is associated with higher risk of treatment failure.22 Further data are needed.
Gram-Negative Rods, Pseudomonas aeruginosa
Pseudomonas aeruginosa is a common pathogen, intrinsically resistant to many antimicrobials, and readily develops antimicrobial resistance during therapy. Fluoroquinolones (such as ciprofloxacin, levofloxacin, and delafloxacin) are the only currently available oral agents with antipseudomonal activity. However, fluoroquinolones may not achieve blood concentrations appropriate for P. aeruginosa treatment at standard doses, while higher dose regimens may be associated with increased risk for undesirable side effects.24,25 Currently, given the minimal trial data comparing oral versus intravenous therapy for P. aeruginosa BSIs, and multiple studies indicating increased mortality when P. aeruginosa is treated inappropriately,26,27 we prefer a conservative approach and consider oral therapy a less-preferred option.
Evidence Regarding Bloodstream Infections due to Gram-Positive Cocci
The majority of bloodstream infections in the United States, and the resultant morbidity and mortality, are from gram-positive cocci (GPCs) such as Staphylococcus, Streptococcus, and Enterococcus species.1
Gram-Positive Cocci, Streptococcus pneumoniae
Of the approximately 900,000 annual cases of S. pneumoniae infection in the United States, approximately 40,000 are complicated by BSI, with 70% of those cases being secondary to pneumococcal pneumonia.28 In studies on patients with pneumococcal pneumonia, bacteremic cases generally fare worse than those without bacteremia.29,30 However, several trials demonstrated comparable outcomes in the setting of bacteremic pneumococcal pneumonia when switched early (within 3 days) from intravenous to oral antibiotics to complete a 7-day course.31,32 Before pneumococcal penicillin resistance became widespread, oral penicillin was shown to be effective, and remains an option for susceptible strains.33 It is worth noting, however, that other trials have shown a mortality benefit from treating bacteremic pneumococcal pneumonia initially with dual-therapy including a β-lactam and macrolide such as azithromycin. This observation highlights the importance of knowing the final susceptibility data prior to consolidating to monotherapy with an oral agent, and that macrolides may have beneficial anti-inflammatory effects, though further research is needed.34,35
Although the evidence for treating bacteremic pneumococcal pneumonia using a highly active and absorbable oral agent is fairly robust, S. pneumoniae BSI secondary to other sites of infection sites is less well studied and may require a more conservative approach.
Gram-Positive Cocci, β-hemolytic Streptococcus species
β-Hemolytic Streptococci include groups A to H, of which groups A (S. pyogenes) and B (S. agalactiae) are the most commonly implicated in BSIs.36 Group A Streptococcus (GAS) is classically associated with streptococcal pharyngitis and Group B Streptococcus (GBS) is associated with postpartum endometritis and neonatal meningitis, though both are virulent organisms with a potential to cause invasive infection throughout the body and in all age-groups. Up to 14% of GAS and 41% GBS BSIs have no clear source;37,38 given these are skin pathogens, such scenarios likely represent invasion via microabrasion. As β-hemolytic streptococcal BSI is often observed in the context of necrotizing skin and soft tissue infections, surgical source control is particularly important.39 GAS remains exquisitely susceptible to penicillin, and intravenous penicillin remains the mainstay for invasive disease; GBS has higher penicillin resistance rates than GAS.40 Clindamycin should be added when there is concern for severe disease or toxic shock.41 Unfortunately, oral penicillin is poorly bioavailable (approximately 50%), and there has been recent concern regarding inducible clindamycin resistance in GAS.42 Thus, oral penicillin V and/or clindamycin is a potentially risky strategy, with no clinical trials supporting this approach; however, they may be reasonable options in selected patients with source control and stable hemodynamics. Amoxicillin has high bioavailability (85%) and may be effective; however, there is lack of supporting data. Highly bioavailable agents such as levofloxacin and linezolid have GAS and GBS activity43 and might be expected to produce satisfactory outcomes. Because no clinical trials have compared these agents with intravenous therapy for BSI, caution is advised. Although bacteriostatic against Staphylococcus, linezolid is bactericidal against Streptococcus.44 Fluoroquinolone resistance amongst β-hemolytic Streptococcus is rare (approximately 0.5%) but does occur.45
Gram-Positive Cocci, Staphylococcus Species
Staphylococcus species include S. aureus (including methicillin susceptible and resistant strains: MSSA and MRSA, respectively) and coagulase-negative species, which include organisms such as S. epidermidis. S. aureus is the most common cause of BSI mortality in the United States,1 with mortality rates estimated at 20%–40% per episode.46 Infectious disease consultation has been associated with decreased mortality and is recommended.47 The guidelines of the Infectious Diseases Society of America for the treatment of MRSA recommend the use of parenteral agents for BSI.48 It is important to consider if a patient with S. aureus BSI has infective endocarditis.
Oral antibiotic therapy for S. aureus BSI is not currently standard practice. Although trimethoprim-sulfamethoxazole (TMP-SMX) has favorable pharmacokinetics and case series of using it successfully for BSI exist,49 TMP-SMX showed inferior outcomes in a randomized trial comparing oral TMP-SMX with intravenous vancomycin in a series of 101 S. aureus infections.50 This observation has been replicated.51 Data on doxycycline or clindamycin for S. aureus BSI are limited, and IDSA guidelines advise against their use in this setting because they are predominantly bacteriostatic.48 Linezolid has favorable pharmacokinetics, with approximately 100% bioavailability, and S. aureus resistance to linezolid is rare.52 Several randomized trials have compared oral linezolid with intravenous vancomycin for S. aureus BSI; for instance, Stevens et al. randomized 460 patients with S. aureus infection (of whom 18% had BSI) to linezolid versus vancomycin and observed similar clinical cure rates.53 A pooled analysis showed oral linezolid was noninferior to vancomycin specifically for S. aureus BSI.54 However, long-term use is often limited by hematologic toxicity, peripheral or optic neuropathy (which can be permanent), and induced serotonin syndrome. Additionally, linezolid is bacteriostatic, not bactericidal against S. aureus. Using oral linezolid as a first-line option for S. aureus BSI would not be recommended; however, it may be used as a second-line treatment option in selected cases. Tedizolid has similar pharmacokinetics and spectrum of activity with fewer side effects; however, clinical data on its use for S. aureus BSI are lacking.55 Fluoroquinolones such as levofloxacin and the newer agent delafloxacin have activity against S. aureus, including MRSA, but on-treatment emergence of fluoroquinolone resistance is a concern, and data on delafloxacin for BSI are lacking.56,57 Older literature suggested the combination of ciprofloxacin and rifampin was effective against right-sided S. aureus endocarditis,58 and other oral fluoroquinolone-rifamycin combinations have also been found to be effective59 However, this approach is currently not a standard therapy, nor is it widely used. Decisions on the duration of therapy for S. aureus BSI should be made in conjunction with an infectious diseases specialist; 14 days is currently regarded as a minimum.47,48
Published data regarding oral treatment of coagulase-negative Staphylococcus (CoNS) BSI are limited. Most CoNS bacteremia and up to 80% Staphylococcus epidermidis bacteremia represent blood culture contamination, though true infection from CoNS is not uncommon, particularly in patients with indwelling catheters.60 An exception is the CoNS species Staphylococcus lugdunensis, which is more virulent, and bacteremia with this organism usually warrants antibiotics. Oral antimicrobial therapy is currently not a standard treatment practice for CoNS BSI that is felt to represent true infection; however, linezolid has been successfully used in case series.61
Gram-Positive Cocci, Enterococcus
E. faecium and E. faecalis are commonly implicated in BSI.1 Similar to S. aureus, infective endocarditis must be ruled out when treating enterococcus BSI; a scoring system has been proposed to assist in deciding if such patients require echocardiography.62 Intravenous ampicillin is a preferred, highly effective agent for enterococci treatment when the organism is susceptible.44 However, oral ampicillin has poor bioavailability (50%), and data for its use in BSI are lacking. For susceptible strains, amoxicillin has comparable efficacy for enterococci and enhanced bioavailability (85%); high dose oral amoxicillin could be considered, but there is minimal clinical trial data to support this approach. Fluoroquinolones exhibit only modest activity against enterococci and would be an inferior choice for BSI.63 Although often sensitive to oral tetracyclines, data on their use in enterococcal BSI are insufficient. Nitrofurantoin can be used for susceptible enterococcal urinary tract infection; however, it does not achieve high blood concentrations and should not be used for BSI.
There is significant data comparing oral linezolid with intravenous daptomycin for vancomycin-resistant enterococci (VRE) BSI. In a systematic review including 10 trials using 30-day all-cause mortality as the primary outcome, patients treated with daptomycin demonstrated higher odds of death (OR 1.61, 95% CI 1.08–2.40) compared with those treated with linezolid.64 However, more recent data suggested that higher daptomycin doses than those used in these earlier trials resulted in improved VRE BSI outcomes.65 A subsequent study reported that VRE BSI treatment with linezolid is associated with significantly higher treatment failure and mortality compared with daptomycin therapy.66 Further research is needed, but should the side-effect profile of linezolid be tolerable, it remains an effective option for oral treatment of enterococcal BSIs.
Evidence Regarding Anaerobic Bacterial Blood Stream Infection
Anaerobic bacteria include Bacteroides, Prevotella, Porphyromonas, Fusobacterium, Peptostreptococcus, Veillonella, and Clostridium. Anaerobes account for approximately 4% of bacterial BSIs, and are often seen in the context of polymicrobial infection.67 Given that anaerobes are difficult to recover, and that antimicrobial resistance testing is more labor intensive, antibiotic therapy choices are often made empirically.67 Unfortunately, antibiotic resistance amongst anaerobes is increasing.68 However, metronidazole remains highly active against a majority of anaerobes, with only a handful of treatment failures reported,69 and has a highly favorable pharmacokinetic profile for oral treatment. Oral metronidazole remains an effective choice for many anaerobic BSIs. Considering the polymicrobial nature of many anaerobic infections, source control is important, and concomitant GNR infection must be ruled out before using metronidazole monotherapy.
Clindamycin has significant anaerobic activity, but Bacteroides resistance has increased significantly in recent years, as high as 26%-44%.70 Amoxicillin-clavulanate has good anaerobic coverage, but bioavailability of clavulanate is limited (50%), making it inferior for BSI. Bioavailability is also limited for cephalosporins with anaerobic activity, such as cefuroxime. Moxifloxacin is a fluoroquinolone with some anaerobic coverage and a good oral pharmacokinetic profile, but Bacteroides resistance can be as high as 50%, making it a risky empiric choice.68
Bacterial BSIs are common and result in significant morbidity and mortality, with high associated healthcare costs. Although BSIs are traditionally treated with intravenous antimicrobials, many BSIs can be safely and effectively cured using oral antibiotics. When appropriately selected, oral antibiotics offer lower costs, fewer side effects, promote antimicrobial stewardship, and are easier for patients. Ultimately, the decision to use oral versus intravenous antibiotics must consider the characteristics of the pathogen, patient, and drug.
None of the authors report any conflicts of interest.