Carbapenem-resistant organisms (CROs), including carbapenem-resistant Enterobacterales (CRE), carbapenem-resistant Acinetobacter baumannii (CRAB), and carbapenem-resistant Pseudomonas aeruginosa (CRPA), pose a growing global health threat. CP-CRE (carbapenemase-producing CRE) is particularly problematic due to its plasmid localization, facilitating horizontal gene transfer. CRO infections have high mortality rates (up to 50%) due to limited treatment options, prompting the WHO to classify them as critical priority pathogens. Hospital settings are major sites of CRO transmission, particularly for CRPA and CRAB, leading to high costs associated with outbreaks (direct costs: antibiotics, precautions, decontamination; indirect costs: bed closures, missed patient care). While infection prevention and control interventions like surveillance, hand hygiene, and isolation are deployed, antibiotic stewardship, which includes reducing inappropriate antibiotic use and optimizing antibiotic use, plays a crucial role. Interventions focused on reducing carbapenem consumption have proven effective in decreasing CRO incidence. However, the widespread presence of ESBL-producing Enterobacterales (ESBL-E) necessitates carbapenem use in many cases, making optimization of antibiotic use even more important. This review focuses on precision prescribing as a key aspect of antibiotic stewardship, aiming to prevent CRO emergence through improved antibiotic selection, dosing, and shorter treatment durations.

The review synthesizes existing literature on the effects of different antibiotics, doses, and treatment durations on resistance development, focusing on CRO emergence defined as isolation of CRO in any specimen from an individual during or after antibiotic exposure (excluding treatment failure due to resistance developing during treatment). The WHO's AWaRe classification of antibiotics based on resistance selection risk is referenced. A meta-analysis by Sulis et al. (2022) highlighted carbapenem use as the strongest risk factor for CRO colonization/infection, with CRPA showing the strongest association, followed by CRE and CRAB. Other antibiotics associated with CRO risk include lincosamides, polymyxin, tigecycline, linezolid, fourth-generation cephalosporins, glycopeptides, daptomycin, macrolides, fluoroquinolones, and piperacillin/tazobactam. However, these associations may be confounded by factors influencing antibiotic selection, with data often derived from population-level analyses rather than patient-level data. The authors advocate for the future use of machine learning to understand the complex interaction between antibiotic utilization and CRO development at the individual patient level.

This is a review article, synthesizing existing research on the effects of antibiotic selection, dosing, and duration on the emergence of carbapenem-resistant organisms (CROs). The authors systematically review published literature relevant to the effects of different antibiotics, dosing strategies, and treatment durations on the development of resistance to carbapenems. They examine studies evaluating the impact of different antibiotic classes, carbapenem agents, antibiotic combinations, dosing regimens, and treatment durations on CRO emergence. They analyze data from meta-analyses, clinical trials, and observational studies to identify trends and correlations. The review focuses on the use of precision prescribing to mitigate the risk of CRO emergence, specifically focusing on carbapenem-sparing strategies for the treatment of ESBL-E infections. The methodology also involves analyzing the potential benefits and limitations of various approaches to optimizing carbapenem dosing, including nomograms and therapeutic drug monitoring (TDM), along with different TDM methods such as immunoassays, chromatographic methods, and biosensors. Finally, the review looks at studies determining optimal carbapenem treatment durations for various infections.

The review finds strong evidence linking carbapenem use to CRO emergence. While ertapenem appears less selective than other carbapenems for CRE, studies show inconsistent results for its effectiveness in reducing CROs. Combination therapy, although potentially synergistic, doesn't consistently show improved clinical outcomes compared to monotherapy for ESBL-E, and may increase toxicity. Carbapenem-sparing alternatives for ESBL-E, such as piperacillin/tazobactam, cephamycins, aminoglycosides, and temocillin are discussed, along with their limitations and the need for further research. The inverted U-shaped relationship between drug exposure and resistance selection pressure suggests that both low and high concentrations may exert less selective pressure compared to the concentrations typically achieved with standard dosing. Studies on antibiotic duration show that shorter durations may reduce resistance emergence. For optimal carbapenem dosing, nomograms and therapeutic drug monitoring (TDM) are discussed, with TDM offering personalized dosing but facing challenges in routine implementation due to limited availability and long turnaround times for β-lactams. Different TDM methods are evaluated, including immunoassays, chromatographic methods, and biosensors, highlighting advantages and limitations of each. Dose adjustment strategies discussed include specific intervention, dosing nomograms, and software for dose optimization, with Bayesian forecasting software offering potential advantages in accuracy. Finally, the review analyzes studies on optimal carbapenem duration for various infections, finding that shorter durations (7 days) can be non-inferior to longer ones (14 days) in certain uncomplicated infections but may increase risk in more complicated cases.

The findings underscore the critical need for precision prescribing to combat CROs. While carbapenem use remains unavoidable in many cases, the review emphasizes the importance of selecting alternative agents when feasible. The data on optimal dosing strategies and treatment duration highlight the complexities of achieving both maximal efficacy and minimal resistance development. The limitations of current TDM methods and the lack of widespread implementation underscore a need for innovative technological solutions to enable rapid and accurate antibiotic level monitoring for prompt dose adjustments. The inconsistent results of some studies highlight the need for more robust and larger clinical trials to evaluate various strategies. The significant impact of confounding variables on resistance selection necessitates a more individualized patient-level approach to antibiotic prescribing. The application of machine learning techniques may help resolve these issues and achieve optimal precision prescribing.

Precision prescribing, encompassing antibiotic selection, dosing, and duration, is crucial for preventing CROs and hospital-acquired infections. Future research should focus on identifying and validating carbapenem-sparing alternatives and improving precision dosing strategies, including TDM implementation and development of novel technologies for rapid antibiotic level monitoring. More robust studies are needed to evaluate various approaches and understand the complex interplay between antibiotic use and resistance development at the individual patient level.

This review is limited by the inherent biases and limitations of the studies it synthesizes, including potential publication bias, variations in study design and methodologies, and the complexities of isolating the effects of antibiotic use from other confounding factors. The focus on existing literature may not reflect the most recent advances or emerging research. The interpretation of findings depends heavily on the quality and quantity of available evidence for each specific antibiotic, dosing regimen, and treatment duration. Generalizability of findings may be limited by the heterogeneity of patient populations, infection types, and healthcare settings represented in the reviewed studies.