When faced with a patient with a suspected bacterial infection, clinicians are forced to make quick decisions about which is the best antibiotic to use. They need to incorporate patient-level factors, including the syndrome in question, which comorbidities are present, and how ill the patient is. They need to incorporate population-level factors, such as the common organisms in their community and the resistance patterns to commonly used antibiotics. The traditional aids for these decisions have typically been antibiogram reports, which, on the basis of historical data obtained from regional microbiology laboratories, describe the percentage susceptibilities of individual organisms to individual antibiotics. For a clinician managing patients with syndromes such as sepsis, these typical antibiograms may not be the ideal tool, because the inciting organism is likely unknown, multiple organisms may be involved, and/or multiple drugs in combination may be needed. Bielicki et al1 systematically searched the literature for microbiological data in neonatal sepsis across a number of Asian countries and determined, through a Bayesian decision-tree model informing a weighted-incidence syndromic combination algorithm (WISCA), that currently recommended empirical antibiotic regimens for neonatal sepsis in the region are likely inadequate.
With the growing antimicrobial resistance crisis, ensuring that the right antibiotic is given to the right patient at the right time should be a major focus for health systems globally. A number of reports over recent years have described WISCAs, which were initially proposed in 2012.2 These algorithms describe the likelihood for adequacy of a typical antibiotic regimen for a clinical syndrome such as sepsis, by taking into account as much local data as are available and weighing those data on the basis of the likelihood of pathogens causing that syndrome. As a decision aid, WISCAs have been shown in small analyses3 to be more useful than typical antibiograms.
The WISCAs created under a Bayesian decision-tree approach reflect how clinical decisions are made in real life.4 However, these approaches are often avoided because of their supposed difficulty, particularly in comparison with traditional antibiograms, which report simple percentages. Bayesian statistics allow for inference with intuitive explanation, although this is often lost because of mathematical formulas that may superficially appear complex. Simply, this is in line with the basic idea of constantly updated knowledge streams, during which our understanding is updated each time new data become available, taking existing knowledge (the prior) with the new information (the likelihood). The 95% credible intervals illustrate the 95% chance that the unknown parameter, such as the coverage of an antibiotic regimen, is within the estimated interval, whereas interpreting the traditional 95% confidence interval involves a hypothetical thought process of repeating the same experiments an infinite number of times.
In the context of antibiotic resistance, as shown by Bielicki et al,1 the use of prior information can maximize the available data to better estimate the coverage of empirical antibiotic regimens. Bielicki et al1 used a Bayesian WISCA decision-tree model to estimate the coverage offered by 3 recommended empirical antibiotic regimens for neonates with sepsis across a number of Asian countries. They identified data on the incidence of bacteria and their antimicrobial susceptibilities through a systematic review of previously published work in the region to estimate the coverage of aminopenicillin-gentamicin, a third-generation cephalosporin, or meropenem, informed by priors based on expected sensitivities of major pathogens to these agents.
Their findings should raise concern: currently, globally recommended first-line and second-line empirical antibiotic regimens for neonatal sepsis will cover well below an adequate proportion of suspected pathogens. In most of the regions studied, this proportion fell well below one-half.1 This reflects point prevalence work on which antibiotics are currently being used, where only a small proportion of clinicians use the globally recommended empirical agents in their setting, and meropenem is the most commonly used agent in neonates.5 Clearly, better-defined strategies to determine which patient should receive which antibiotics are urgently needed.
Using published literature to estimate microbiological data locally comes with problems, such as biasing the results toward academic institutions and ensuring timeliness and accuracy of collected data. However, given the patchiness of available microbiological testing in the region,6 the published literature may reflect the best available data to inform these decisions, in the absence of scaled-up solutions for more available antimicrobial resistance data dissemination. Given the diversity that exists within a single country, or even within a single health system, relying on national-level estimates to inform local practice may not be accurate.7 In the absence of local data, however, national estimates remain the most reliable, but serve as an important argument to augment the testing capabilities across regions with a high burden of antimicrobial resistance. From a larger perspective, global bodies that provide recommendations for empirical therapy for syndromes may need to reflect on the rapidly changing epidemiological landscape and explicitly incorporate language regarding local resistance patterns and likely causative pathogens.8
The antimicrobial resistance crisis continues to unfold, with an ongoing arms race between clinicians wanting to ensure the best therapy for their patient and increasing resistance among bacteria. Incorporating better modeling techniques as practical decision aids to clinicians can help cool down that battle, but only if the microbiological data used to inform those models are up-to-date, local, and reliable.
Published: February 12, 2020. doi:10.1001/jamanetworkopen.2019.21150
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Park JJH et al. JAMA Network Open.
Corresponding Author: Srinivas Murthy, MD, CM, MHSc, Department of Pediatrics, Faculty of Medicine, University of British Columbia, 4500 Oak St, Vancouver, BC V6H 3V4, Canada (email@example.com).
Conflict of Interest Disclosures: None reported.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Park JJH, Murthy S. Defining Optimal Empirical Antibiotic Regimens in a Rapidly Changing Landscape of Resistance. JAMA Netw Open. 2020;3(2):e1921150. doi:10.1001/jamanetworkopen.2019.21150
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: