Antibiotic modification versus withhold in febrile patients without evidence of bacterial infection, unresponsive to initial empiric regimen: a multicentre retrospective study conducted in Israel.

Prescribing antibiotics for febrile patients without proof of bacterial infection contributes to antimicrobial resistance. Lack of clinical response in these patients often leads to antibiotic escalation, although data supporting this strategy are scarce. This study compared outcomes of modifying, withholding, or continuing the same antibiotic regimen for such patients. Febrile or hypothermic stable patients with suspected infection, unresponsive to empiric antibiotic treatment, admitted to one of 15 internal medicine departments in three hospitals during a 5-year study period, were included. Patients with a definitive clinical or microbiological bacterial infection, malignancy, immunodeficiency, altered mental status, or need for mechanical ventilation were excluded. Participants were divided into groups based on treatment strategy determined 72 h after antibiotic initiation: antibiotic modified, withheld or continued. Outcomes measured included in-hospital and 30-day post-discharge-mortality rates, length of hospital stay (LOS) and days of antimicrobial therapy (DOT). A total of 486 patients met the inclusion criteria: 124 in the Antibiotic modified group, 67 in the Antibiotic withheld group and 295 in the Initial antibiotic continued group. Patient characteristics were similar among groups with no differences in mortality rates in-hospital (23% vs. 25% vs. 20%, p = 0.58) and within 30 days after discharge (5% vs. 3% vs. 4%, p = 0.83). Changing antibiotics led to longer LOS (9.0 ± 6.8 vs. 6.2 ± 5.6 days, p = 0.003) and more DOT (8.6 ± 6.0 vs. 3.2 ± 1.0 days, p < 0.001) compared to withholding treatment. Withholding as compared to modifying antibiotics, in febrile patients with no clear evidence of bacterial infection, is a safe strategy associated with decreased LOS and DOT.