Antibiotic Myths for the Infectious Diseases Clinician

Abstract

Antimicrobials are commonly prescribed and often misunderstood. With more than 50% of hospitalized patients receiving an antimicrobial agent at any point in time, judicious and optimal use of these drugs is paramount to advancing patient care. This narrative will focus on myths relevant to nuanced consultation from infectious diseases specialists, particularly surrounding specific considerations for a variety of antibiotics.

cefazolin, trimethoprim-sulfamethoxazole, linezolid, clindamycin

Antimicrobial use is commonly informed by past practices instead of evidence [1], with “pearls” propagated over time [2]. Consequently, medical paradigms may be constructed and followed without foundation until unique patients or logistical implications arise that challenge the status quo. One such example is the historical avoidance of the entire β-lactam class of antibiotics for patients with reported penicillin allergies; however, this practice is proven to be unfounded and harmful [3, 4].

Core principles of antimicrobial use advocate for appropriate dose, duration, and route of the optimal agent(s) for each disease [5]. This author group, comprised of infectious diseases clinicians and consultants in academic and community hospitals, previously published 10 common myths of infectious diseases diagnosis and management in adult patients, focused on clinicians practicing in hospital medicine [6]. The group then identified remaining myths and dogma from their shared experiences, conference presentations, and academic community social channels and listservs. A voting process occurred to gain consensus on 8 myths related to optimizing antibiotic use for infectious diseases providers for a focused review. Herein, those 8 myths related to the nuanced prescribing of antimicrobial agents are elucidated. This analysis provides infectious diseases specialists with the history, current literature, and remaining unknowns for each myth.

Myth: Cefazolin Should Be Avoided for Central Nervous System Infections

A commonly used tertiary medical reference states, “Cefazolin should not be used for … meningitis because it does not adequately penetrate into the CNS [central nervous system]” [7]. The source of this recommendation is from 1973, when Mangi et al described a case series of breakthrough meningitis cases in patients receiving cephalothin for invasive streptococcal infections [8]. The authors suggested these occurrences were due to poor penetration of cephalothin across the blood–brain barrier. Another early study found undetectable cefazolin cerebrospinal fluid (CSF) concentrations after a single 1-g dose [9]. Given this information, it has since been taught that clinicians should avoid cefazolin for CNS infections.

However, a few years later, Frame et al found higher concentrations in brain tissue after a single 2 g infusion of cefazolin than a 2 g infusion of nafcillin (4.3 µg/g vs 1.3 µg/g, respectively) [10]. Furthermore, Novak et al described 15 patients with extraventricular devices receiving prophylactic doses of cefazolin 2 g intravenously (IV) every 8 hours [11]. Cohort-extrapolated median maximum and miminum concentrations in CSF were 2.97 and 1.59 mg/L, respectively. Gregoire et al found that 8 g or 10 g daily of cefazolin via continuous infusion led to median CSF concentrations of 6.1 mg/L or 11.9 mg/L, respectively, in a patient who experienced clinical success after receiving 6 weeks of cefazolin plus levofloxacin for methicillin-susceptible Staphylococcus aureus (MSSA) neurosurgical meningitis [12]. Le Turnier et al found similar median concentrations of cefazolin when using 6–12 g continuous infusions (2.8 mg/L [interquartile range, 2.1–5.2]) [13]. These concentrations are higher than the Clinical Laboratory and Standards Institute breakpoints for S. aureus and Enterobacterales of ≤2 mg/L and above the European Committee on Antimicrobial Susceptibility Testing (EUCAST) epidemiologic cutoff of minimum inhibitory concentration (MIC) ≤2 mg/L for MSSA [14].

The pharmacokinetics of antibiotics in the CSF are poorly defined. Since drug entry into the CSF is delayed compared to other body fluids and compartments (a phenomenon known as system hysteresis), the ratio of drug in CSF to plasma may increase after initiation of the drug infusion [15, 16]. Therefore, the ratio of the area under the curve (AUC) of CSF to serum at steady state (AUC_CSF/AUC_SS) is the most accurate means of characterizing drug penetration into the CSF, and single-point CSF concentrations may not have interpretative utility. The AUC_CSF/AUC_SS for cephalosporins typically ranges from 0.007 to 0.17, which is quite low; however, high doses of cephalosporins such as ceftriaxone are preferred therapy for many CNS infections despite this “poor” penetration.

In conclusion, cefazolin is a reasonable option for treatment of CNS infections with susceptible pathogens. A recent narrative review on the treatment of cefazolin for CNS infections (including 11 reports describing cefazolin treatment of spinal epidural abscesses in a total of 104 patients) recommends 2 g IV every 6 hours or a continuous infusion of 8–10 g daily instead of the traditional 1–2 g IV every 8 hours dosing scheme to optimize pharmacokinetic/pharmacodynamic properties [17]. Clinical studies would help determine the optimal cefazolin dose for CNS infections.

Myth: Linezolid Must Be Avoided in Patients Receiving Selective Serotonin Reuptake Inhibitors

Linezolid is an oxazolidinone-class antibacterial with in vitro activity against multidrug-resistant gram-positive bacteria and favorable pharmacokinetics [18]. However, its reversible, nonselective inhibition of monoamine oxidase, an enzyme responsible for breaking down serotonin in CNS, has led to cautious use in patients receiving serotonergic agents to avoid serotonin syndrome (SS) [19]. The United States (US) Food and Drug Administration (FDA) recommends an impractical 14-day (5 weeks for fluoxetine) washout period in patients receiving these agents [19].

Butterfield et al assessed 4265 patients in clinical trials of gram-positive infections taking serotonergic agents and receiving either linezolid or a comparator. Using 1 set of criteria, 9 of 2208 (0.41%) linezolid and 3 of 2057 (0.14%) comparator patients experienced SS; by different criteria, the incidences were 3 of 2208 (0.14%) and 1 of 2057 (0.05%), respectively. SS was uncommon overall and not significantly higher with linezolid [20]. This result has been replicated, and SS has not been linked to a specific duration of linezolid therapy since it occurs so rarely [21]. Gatti et al evaluated 669 FDA postmarketing reports of SS mentioning linezolid. Linezolid coadministered with citalopram constituted 69 of 699 (10.3%) reports; however, the denominator of total exposures of this combination among US patients (and therefore true incidence of this reaction) is unknown [22]. Inferring from pharmacological properties including lipophilicity, large volume of distribution, and potent serotonin reuptake transporter inhibition, the authors suggest that linezolid plus citalopram, escitalopram, and methadone may be more likely to cause SS than other drug combinations. More recently, Bai et al evaluated 1134 elderly patients in Canada receiving linezolid, 215 (19%) of whom were also taking an antidepressant [21]. SS occurred in <6 patients. Similarly, Kufel et al evaluated 1743 patients receiving linezolid, 1168 (67%) of whom also received anywhere from 1 to 5 additional serotonergic agents [23]. Only 2 patients (0.11%) had possible serotonin toxicity.

In summary, SS is exceedingly rare even when linezolid is combined with serotonergic agents. The risk/benefit profile of linezolid use in patients receiving serotonergic drugs is acceptable. Patients receiving serotonergic agents should be monitored closely while receiving linezolid, but it is not necessary to avoid concomitant administration of linezolid when required.

Myth: No Dose Adjustment Is Needed for Linezolid in Patients With Renal Insufficiency

Linezolid is administered as 600 mg IV or orally (PO) twice daily, with no dose adjustment for renal dysfunction recommended in the package insert. In early clinical trials and observational experiences of patients receiving short-course linezolid, incidence of thrombocytopenia was low (<3%) [18, 24, 25]. However, with expanded clinical use, myelosuppression is more frequently observed. An investigation of the relationship between linezolid exposure and renal dysfunction revealed a strong correlation between renal clearance and linezolid clearance among a small number of patients. Interestingly, it was found that an increased linezolid exposure (defined by total AUC or trough concentrations) was associated with thrombocytopenia [26].

Building on this experience, Crass et al examined the risk for thrombocytopenia among 341 patients receiving linezolid [27]. Ninety-two (27.0%) patients developed thrombocytopenia; 47 (13.8%) developed severe thrombocytopenia. The frequency of both thrombocytopenia (42.9% vs 16.8%; P < .001) and severe thrombocytopenia (19.5% vs 10.1%; P = .02) was greater among patients with renal impairment. Renal impairment, defined as estimated glomerular filtration rate (eGFR) <60 mL/minute/1.73 m [2], was independently associated with thrombocytopenia in a multivariable analysis (adjusted hazard ratio, 2.37; 95% confidence interval [CI]: 1.52–3.68). The authors suggest that empirical linezolid dose reduction to 300 mg IV or PO every 12 hours provides the best balance of safety and efficacy, achieving therapeutic concentrations in 65% of simulated patients with eGFR <60 mL/minute [27].

Although recommendations to adjust linezolid dosing for renal dysfunction have not been added to its product label, it may be prudent to consider dose adjustment or therapeutic drug monitoring. Experts recommend dosing linezolid to an AUC_0–24/MIC ratio of 80–120 for optimal efficacy and maintaining troughs <7 mg/L for safety [28].

Myth: Clindamycin Is a First-line Drug for Prevention of Surgical Site Infections in Patients With Reported Penicillin Allergies

Clindamycin is a protein synthesis inhibitor used for a broad range of indications, including surgical site infection (SSI) prophylaxis. As a lincosamide, it does not share any chemical structure with β-lactam antibiotics, making clindamycin a historically preferred choice for SSI prophylaxis in patients with reported penicillin allergies [29].

Cefazolin, the first-line prophylactic agent for most surgical procedures, is traditionally avoided in patients with reported penicillin allergies since historical references quote cross-reactivity rates of 5%–10% between penicillin allergies and reactions to first-generation cephalosporins [30, 31]. However, cefazolin does not share a side chain with any other β-lactam antibiotic and the likelihood of cefazolin allergy in a patient with reported penicillin allergy is extremely rare [3]. Indeed, in a meta-analysis of 77 studies yielding 6147 patients, the frequency of dual allergy, including penicillin and cefazolin, was 0.7% [3]. In a study of 690 patients with reported penicillin allergy who received cefazolin, clindamycin, or vancomycin, hypersensitivity reactions were similar across all 3 antibiotics (3 [0.9%], 4 [1.4%], and 1 [1.1%], respectively) [32]. Most recently, Norvell et al described use of cefazolin or clindamycin and/or vancomycin as surgical prophylaxis in patients with reported penicillin allergies [33]. There were fewer SSIs (0.9% vs 3.8%; P < .001), including prosthetic joints infections (0.1% vs 1.9%), among cefazolin-treated patients. More interoperative hypersensitivity reactions occurred in patients receiving clindamycin and/or vancomycin compared to cefazolin (1.3% vs 0.2%).

Additionally, clindamycin resistance has increased among methicillin-resistant S. aureus (MRSA) isolates globally. In the US, resistance of Streptococcus pyogenes to clindamycin reached 24.2% in 2018, and resistance in Streptococcus agalactiae has risen to >40%, making clindamycin unreliable against common causes of cellulitis and SSIs [34].

For these reasons, the Joint Task Force on Practice Parameters (formed by 2 leading allergy and immunology groups in 2022) updated its guidance to suggest structurally dissimilar cephalosporins (eg, cefazolin or ceftriaxone) as first-line antibiotic agents of choice for surgical prophylaxis in patients with a history of anaphylaxis to penicillin [35]. The need for clindamycin in surgical prophylaxis is now extremely limited.

Myth: Trimethoprim-Sulfamethoxazole Does Not Have In Vitro Activity Against S. pyogenes

A common teaching point in managing skin and skin structure infections (SSSIs) is that pustular infections are staphylococcal in origin and nonpustular cellulitis is streptococcal, primarily S. pyogenes. Cephalexin is often recommended for either type of SSSI since it displays in vitro activity against MSSA and S. pyogenes. However, the rise of community-associated MRSA in the 2000s led to alternative regimens for these infections, including tetracyclines and trimethoprim-sulfamethoxazole (T/S) [36]. The common misnomer that T/S is inactive in vitro against S. pyogenes led to using T/S plus β-lactams in combination for uncultured SSSI [37].

The misunderstanding about T/S resistance in S. pyogenes is mainly laboratory based [38]. Through their effects on different steps of folic acid synthesis and utilization, trimethoprim and sulfamethoxazole prevent the biosynthesis of the nucleic acid thymidine in bacteria. Some species of bacteria can utilize exogenous thymidine when endogenous production is inhibited. There are differences between streptococci in this ability: while absent in Streptococcus pneumoniae and many viridans group streptococci, it occurs in S. pyogenes and S. agalactiae [39]. Agar used for in vitro cultures of streptococci previously contained thymidine, serving as an exogenous source to intentionally isolate S. pyogenes from upper respiratory samples [39]. Currently recommended thymidine-depleted media eliminates this issue. When tested appropriately and using EUCAST breakpoints, all 370 S. pyogenes isolates in the US SENTRY database tested susceptible to T/S with an MIC₉₀ ≤ 0.12 μg/mL (2020–2021, North America only).

While these microbiological data are reassuring, recent clinical studies also support the ability of T/S to treat SSSIs primarily caused by S. pyogenes. For example, in a multicenter randomized clinical trial containing 524 patients with uncomplicated SSSIs (280 [53.4%] had cellulitis without abscess), cure rate for T/S treatment was 88.2% in the evaluable population [40]. Streptococcus pyogenes was rarely isolated as expected, but cure rates were similar between agents for those with cellulitis alone. A 2017 systematic review of T/S for SSSIs, including those caused by S. pyogenes, also concluded that T/S is useful for the treatment of both staphylococcal and streptococcal skin infections [41]. Collectively, these data indicate the utility of T/S monotherapy for SSSIs.

Myth: Oral Fosfomycin Is an Excellent Drug for Uncomplicated Cystitis

Fosfomycin is a unique antibiotic that interferes with cell wall synthesis of gram-positive and gram-negative bacteria. Guidelines recommend fosfomycin as first-line treatment for acute uncomplicated cystitis in women [42, 43].

Efficacy data for fosfomycin are conflicting. In 1999, a 749-patient trial found that single-dose fosfomycin had similar efficacy as a 7-day course of nitrofurantoin [44]. A meta-analysis of 27 clinical trials also found that fosfomycin had similar cure rates to fluoroquinolones, T/S, or nitrofurantoin for cystitis in pregnant and nonpregnant patients [45]. However, a more recent randomized, open-label, controlled trial of 513 women with uncomplicated cystitis found that 5 days of nitrofurantoin treatment resulted in greater clinical and microbiologic resolution at 28 days compared to fosfomycin (70% vs 58%; P = .004) [46].

The Infectious Diseases Society of America (IDSA) guidelines state that fosfomycin may cause fewer ecological adverse effects than other antibacterial options [43]. However, a longitudinal Spanish study noted that a 340% increase in fosfomycin use from 1997 to 2009 was accompanied by an increase in fosfomycin resistance from 4% to 11% [47]. Increased fosfomycin use has also been connected to increases in fosfomycin-resistant, extended-spectrum β-lactamase–producing Escherichia coli [42, 48]. More concerningly, fosfomycin requires glucose-6-phosphate (G6P) to exert bacterial killing; this enzyme is not present in human urine [49]. Therefore, minimum inhibitory concentrations determined in the presence of G6P in the laboratory may not accurately reflect the in vivo activity of the drug, and the appropriate susceptibility breakpoint may need reassessment. Furthermore, fosfomycin susceptibility testing is challenging and is currently only recommended for E. coli and Enterococcus faecalis [50].

While fosfomycin is appealing due to the “one and done” FDA-approved dosing for uncomplicated cystitis, the most recent randomized, clinical data do not support its use compared to nitrofurantoin. Additionally, critical evaluations are ongoing to determine if fosfomycin is reliably active at the infection site and to understand the clinical susceptibility breakpoint. If fosfomycin is utilized for uncomplicated cystitis, patients should be carefully assessed for cure.

Myth: Rifampin and Gentamicin Are Essential for Treatment of Staphylococcus spp Prosthetic Valve Endocarditis

Invasive Staphylococcus spp infections, particularly prosthetic valve endocarditis (PVE), are associated with high morbidity and mortality [51]. The IDSA guidelines for MRSA infection recommend addition of gentamicin and rifampin for treatment of MRSA PVE based on small, retrospective studies of patients with coagulase-negative staphylococci infections [36]. Additionally, gentamicin may act synergistically with several other antibiotics against Staphylococcus aureus in animal and in vitro models [52].

However, Cosgrove et al did not demonstrate any benefit in patient outcomes when gentamicin was added to treatment regimens for patients with S. aureus bacteremia and native valve endocarditis [53]. Instead, the authors observed increased acute renal injury with gentamicin. Subsequent studies on staphylococcal PVE found no difference in mortality with the addition of gentamicin [54].

The recommendation for rifampin in patients with MRSA PVE comes from a post hoc analysis of an S. aureus bacteremia cohort where rifampin-based combinations were associated with a lower hazard ratio for mortality at 30 days (0.6; 95% CI: .3–1.1) and reduced late complications (10.6% vs 4.5%; P = .03) [55]. However, this sample was small and a purely observational association and only 12.6% of the population had endocarditis. In a recent multinational observational study, rifampin did not impact 1-year mortality or infection relapse and hospital length of stay was longer for rifampin recipients [56]. A meta-analysis of adjunctive rifampin and gentamicin for staphylococcal PVE identified only 4 studies and did not show benefit [57].

Rifampin and gentamicin each have issues complicating their use, which necessitates reexamining the basis of their utility. The problems introduced with drug interactions and toxicities outweigh the utility of these antibiotics in endocarditis.

Myth: Doxycycline Is Contraindicated in Pregnancy and Pediatric Patients <8 Years Old

Tetracycline is associated with maternal hepatotoxicity, inhibition of bone growth, and tooth discoloration with prolonged exposure in patients <8 years of age [58]. Therefore, use in pregnant women and children is restricted [59, 60]. Doxycycline was developed later and contains structural modifications that improve antibacterial activity, decrease calcium binding, and diminish adverse events compared to tetracycline [61, 62]. However, it carries the same FDA warnings as tetracycline due to the “tetracycline class effect” established in 1970. Thus, many prescribers avoided doxycycline in children <8 years of age, even when it was recommended first-line for treatment such as in rickettsial diseases [63, 64].

Recent evidence suggests that doxycycline can be used safely in pregnant women and young children in select scenarios and that withholding doxycycline for certain diseases in these populations may lead to harm [65]. Case fatality rates for Rocky Mountain spotted fever and ehrlichiosis are significantly higher in children <8 years of age compared to every other age group [66]. One study of 58 children who received doxycycline before age 8 compared to 213 unexposed children found no difference in teeth staining [67]. A large, retrospective study of infants with fetal doxycycline exposure found no increased risk of congenital malformations and teratogenic risk compared to unexposed infants [68]. The American Academy of Pediatrics states that short-course doxycycline (≤21 days) is acceptable since it has been shown to pose minimal risk of dental staining [69, 70]. The Pregnancy Team at the FDA stated that “while there are no controlled studies of doxycycline use in pregnant women to show safety, an expert review of published data on experiences with doxycycline use during pregnancy by the Teratogen Information System concluded therapeutic doses during pregnancy are unlikely to pose a substantial teratogenic risk … but the data are insufficient to state that there is no risk” [71].

These data indicate that in the cases of serious infections without equivalently efficacious treatment alternatives, such as rickettsial diseases, the benefits of using doxycycline outweigh the risks for pediatric patients. Clinical trials of doxycycline in pregnancy and early childhood are indicated, including pharmacokinetic and longitudinal studies to evaluate the impact of doxycycline use on long bone growth and dental discoloration [70].

Clinicians are often presented with medical statements that are either more opinion than robust evidence, or wherein the evidence has evolved yet perception remains unchanged. This narrative highlights the need for vigilance as evidence bases evolve. Today's teaching point may end up as tomorrow's myth.

Author notes

Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.