Pediatric antimicrobial stewardship programs (P-ASP) are essential for optimizing antibiotic utilization and improving patient outcomes.1,2 One of the most frequent and challenging scenarios in clinical practice occurs when a child fails to show improvement within 48–72 hours of initiating appropriate therapy.1,2 In such instances, clinicians frequently misattribute persistent symptoms to antimicrobial resistance, triggering an unwarranted broadening of the antibiotic spectrum rather than a systematic diagnostic reassessment.3,4 The lack of standardized global definitions for “early antibiotic treatment failure” in children often leads to diagnostic closure, where the primary diagnosis is rarely questioned once therapy has begun.4 This practice can delay the identification of the true underlying etiology, drive resistance and ultimately compromise clinical recovery.4 To address this challenge, we propose the BARNS acronym (Fig. 1) as a structured framework for clinical reassessment. This tool guides clinicians through 5 key diagnostic dimensions:FIGURE 1.: The BARNS framework for the systematic reassessment of antimicrobial therapy. The acronym serves as a cognitive aid for clinicians evaluating patients who fail to show clinical improvement within 48–72 hours of starting appropriate therapy. C. difficile indicates Clostridioides difficile; CMV, cytomegalovirus; CNS, central nervous system; EBV, Epstein–Barr virus; HLH, hemophagocytic lymphohistiocytosis; NICU, neonatal intensive care unit; PICU, pediatric intensive care unit; PK, pharmacokinetics; SFN, sulfonamides; SJS/DRESS, Stevens–Johnson syndrome/ drug reaction with eosinophilia and systemic symptoms; S. maltophilia, Stenotrophomonas maltophilia; TDM, therapeutic drug monitoring; VAN, vancomycin. Breakthrough infection. Attainment of target antibiotic concentration. Resistance development. Noninfectious causes. Source control. We present clinical cases that illustrate how the BARNS model facilitates a more precise diagnostic approach in patients with an apparent lack of response to therapy. Clinical Cases B: Breakthrough Infection A preterm infant born at 28 weeks’ gestation required mechanical ventilation on day 14 of life and subsequently developed ventilator-associated pneumonia due to Pseudomonas aeruginosa susceptible to cefepime; minimum inhibitory concentration (MIC) = 1 mg/L. After initial improvement on cefepime, the patient developed new-onset fever and leukocytosis by day 5 of cefepime therapy. Although the clinical team initially suspected resistance (R) and escalated therapy to meropenem, a structured BARNS assessment identified a percutaneous central venous catheter inserted 10 days ago as a risk factor for a breakthrough infection (B). Pharmacokinetic failure was considered unlikely as there were no factors affecting drug exposure (eg, extracorporeal membrane oxygenation or renal replacement therapy) (A). No noninfectious causes were found (N), and a repeat chest X-ray did not show any complications (eg, pleural effusion) (S). Blood cultures subsequently confirmed methicillin-resistant Staphylococcus epidermidis. Consequently, the central venous catheter was replaced, vancomycin initiated and cefepime resumed to complete the 7-day ventilator-associated pneumonia course, allowing for the immediate discontinuation of meropenem. A: Attainment of Target Antibiotic Concentrations A 2-year-old with a ventriculoperitoneal shunt presented with vomiting, drowsiness and raised inflammatory markers. Cerebrospinal fluid (CSF) obtained from the shunt showed pleocytosis, hypoglycorrhachia and hyperproteinorrachia; cultures confirmed Staphylococcus epidermidis (vancomycin MIC = 2 mg/L). The infected shunt material was removed and replaced with an external ventricular drain, providing source control (S). Intravenous vancomycin (15 mg/kg every 6 hours) was initiated. Despite 48 hours of appropriate therapy and source control, the patient remained febrile. The surgical team proposed escalating to meropenem; however, application of the BARNS framework redirected the clinical focus toward the attainment of target drug concentrations. No breakthrough infection (B), resistance (R) or noninfectious causes (N) were identified, yet the steady-state vancomycin trough was only 10 µg/mL. Given the high MIC and the limited central nervous system penetration of vancomycin, a target of 15–20 µg/mL was required. Transitioning to a continuous infusion achieved these therapeutic levels and rapid resolution of symptoms without broadening the antimicrobial spectrum. R: Resistance development A 5-month-old infant was admitted with a febrile urinary tract infection. Empiric intravenous cefotaxime was initiated by the primary pediatric service. Following the identification of Serratia marcescens, the laboratory initially reported susceptibility to third-generation cephalosporins. Despite the inherent risk of AmpC induction associated with this pathogen, the initial regimen was continued based on these susceptibility results. At 48 hours, the patient experienced a clinical relapse with new-onset fever. A systematic BARNS evaluation was performed to identify the cause of clinical relapse. Breakthrough infection (B) and pharmacokinetic failure (A) were excluded based on clinical stability and dosing adequacy, while noninfectious causes (N) were deemed unlikely. Inadequate source control (S) appeared unlikely as a renal ultrasound showed no evidence of obstructive uropathy, renal abscess or nephrolithiasis. Consequently, acquired resistance (R) was suspected. Repeat susceptibility testing confirmed selection of a stably derepressed AmpC β-lactamase-producing mutant, explaining the transient initial response followed by treatment failure. Switching to cefepime resulted in prompt defervescence. The patient completed a 7-day course and was successfully transitioned to oral therapy. N: Noninfectious causes A 9-year-old child was admitted with 5 days of fever, diarrhea, dehydration and hypotension requiring fluid resuscitation. Blood and stool cultures grew Salmonella enterica serotype Typhi. Despite initiating ceftriaxone and achieving follow-up blood cultures that showed no growth, the patient remained febrile with worsening clinical status. The BARNS reassessment was performed: breakthrough infection (B), resistance (R) and pharmacokinetic failure (A) were ruled out. No clinical signs suggestive of persistent infection were present, and abdominal abscesses were excluded by abdominal ultrasound (S). This systematic exclusion shifted the focus toward noninfectious causes (N). Physical examination revealed new-onset splenomegaly, and laboratory studies showed pancytopenia (leukocytes 1200/µL; hemoglobin 8 g/dL), hyperferritinemia, hypertriglyceridemia and hypofibrinogenemia. The patient met diagnostic criteria for secondary hemophagocytic lymphohistiocytosis triggered by the initial infection. The initiation of high-dose corticosteroids led to rapid clinical and laboratory improvement. S: Source control A 2-year-old child was admitted with pneumonia and a small (12 mm) parapneumonic effusion. Given the patient’s clinical stability, conservative management with intravenous ampicillin was initiated. After 48 hours, however, the child remained febrile and developed worsening respiratory distress, prompting the on-call team to escalate therapy to cefotaxime and clindamycin, suspecting possible resistance. Despite this change, the child remained febrile and showed no clinical improvement. A BARNS-based reassessment was performed. Breakthrough infection (B), pharmacokinetic failure (A) and noninfectious causes (N) were ruled out. As no resistant organisms (R) had been identified, inadequate source control (S) was considered. Repeat ultrasonography revealed progression to an organized, fibrinopurulent empyema. Surgical drainage was subsequently performed, resulting in the resolution of fever and respiratory distress. Pleural fluid cultures confirmed penicillin-susceptible Streptococcus pneumoniae, allowing for de-escalation to penicillin and successful completion of therapy. DISCUSSION The 48–72-hour reassessment is a critical intervention timepoint in P-ASP.1,2 At this stage, the clinical response and microbiological results often provide valuable information to confirm the diagnosis and potentially adjust antimicrobial therapy.1,2 However, persistence or recurrence of fever or other inflammatory symptoms and markers during this window frequently trigger the “escalation reflex”—an immediate broadening of the antimicrobial spectrum.3,4 However, as demonstrated by the illustrative cases above, an apparent lack of clinical response may not be due to antimicrobial resistance.4 The BARNS model provides a structured framework to transform this moment of clinical uncertainty into a systematic diagnostic review (Fig. 1). While resistance (R) is a primary concern in stewardship, it is often a late consideration in the short-term failure of appropriate therapy, unless specific mechanisms like AmpC derepression are at play, or when dealing with non-fermenting Gram-negative organisms prone to rapid emergence of resistance.5 In pediatrics, lack of source control (S) or the emergence of a breakthrough infection (B) at a different site, such as a catheter-associated bloodstream infection, are frequent drivers of persistent fever. The BARNS framework ensures that these “mechanical” or “spatial” failures are considered and ruled out before assuming failure of the antibiotic therapy.6 Unique to the pediatric population is the critical importance of attainment (A).7 Children exhibit significant pharmacokinetic and pharmacodynamic variability due to their developmental physiology and the higher volumes of distribution seen in critical illness.7 Relying on standard dosing without considering the specific site of infection (eg, CNS) or the MIC of the pathogen can lead to “pseudo-resistance,” where the drug concentrations are insufficient, despite “appropriate dosing.”7 Furthermore, the noninfectious (N) component of the model is perhaps the most critical for patient safety.8 Conditions, such as hemophagocytic lymphohistiocytosis, Kawasaki disease, drug-induced severe hypersensitivity or other inflammatory syndromes, can mimic refractory sepsis.8 A systematic pause to apply the BARNS framework prevents the “diagnostic overshadowing” that occurs when clinicians focus solely on infectious etiologies. Perceived “antibiotic failure” is a common challenge for P-ASP teams; nevertheless, a harmonized and systematic framework to guide clinical reassessment has been lacking.4,9,10 While conservative approaches aim to cover the possibility of resistance, and broadening antimicrobial coverage may be understandable in critically ill patients, this should not preclude a structured reassessment.4,9,10 Simply increasing the antimicrobial spectrum can create a false sense of security and delay the identification of alternative causes of apparent treatment failure, for which broader coverage offers no benefit.10 Conversely, in clinically stable children with persistent symptoms, careful reassessment allows diagnostic clarification while avoiding unnecessary exposure to broader-spectrum agents and their associated risks, including toxicity and the selection of antimicrobial resistance.2,3,10 Furthermore, it avoids transitioning to broader agents that may inadvertently offer inferior bactericidal activity or poorer tissue penetration for the specific site of infection compared to a more targeted, narrow-spectrum choice. All elements of this framework have been described as contributors to lack of clinical response;4,5,7,8,10 however their prevalence is highly variable and quantitative data are limited. In clinical practice, recurring patterns—including specific clinical presentations, comorbidities, and the presence of medical devices—facilitate the identification of contributing factors, as illustrated in Figure 1.4 Proposal for Implementation and Clinical Integration of the BARNS framework To move beyond a conceptual model, formal integration of the BARNS framework into clinical practice can facilitate a transition from reactive prescribing to systematic diagnostic reassessment. We propose incorporating the model into 2 primary clinical workflows: Handshake Stewardship and Rounds The BARNS acronym could serve as a structural template during bedside ASP rounds. By framing the discussion around these 5 domains, consultants provide a standardized “cognitive stop” that prevents premature diagnostic closure. Health Information Technology Integration To ensure sustainability, BARNS criteria can be embedded into Electronic Health Record “Antibiotic Time-Out” templates or ID consultation notes. This could prompt clinicians to document objective evidence of source control (S) and pharmacokinetic optimization (A) before modifying therapy. Evaluating Impact and Process Metrics The clinical utility of the BARNS model can be quantified by monitoring specific processes and outcome metrics mapped to its components: Diagnostic Efficiency (N, S): Evaluation of the interval between clinical non-response and diagnostic re-imaging or surgical intervention. Pharmacodynamic Optimization (A): Evaluation of dosing adequacy relative to the severity and site of infection. This includes an analysis of the frequency and timing of therapeutic drug monitoring when applicable, as well as systematic verification of weight-based dosing and interval optimization for agents, even where therapeutic drug monitoring is unavailable. The proportion of patients initiated on suboptimal empiric dosing may serve as a key outcome metric for assessing antimicrobial failure. Prescribing Quality (B, R): Evaluation of the “Appropriateness of Escalation,” demonstrated by a decrease in the reflex use of “broad-spectrum” antibiotics when failure is attributed to noninfectious mimickers. System-Level Outcomes: Prospective evaluation should focus on the reduction of total days of therapy and hospital length of stay. In conclusion, the BARNS model serves as a practical bedside tool and a cognitive safeguard for P-ASP teams. While this framework currently represents an expert-led clinical approach and requires further prospective validation, it encourages a structured differential diagnosis rather than a reflex change in therapy. By adopting this systematic review, clinicians can improve diagnostic precision, minimize the unwarranted use of broad-spectrum agents and ultimately enhance the quality of care for hospitalized children. Future research integrating this framework into routine practice will help estimate the prevalence of different diagnostic categories and support the development of predictive tools to further enhance antimicrobial stewardship quality and ultimately patient outcomes.
Belleri et al. (2026) studied this question.