Intravenous Antibiotic Durations: “Short” Wins Again

Olson et al¹²  analyzed bacteremic UTIs in 174 infants ≤90 days of age hospitalized in Utah’s Intermountain Healthcare system between 2002 to 2020. Patients with meningitis, osteomyelitis, central lines, urologic stents, renal abscesses, ventriculoperitoneal shunts, and with UTIs acquired in the NICU were excluded. The investigators dichotomized IV antibiotic durations as ≤ 4 days (short course, n = 73) and >4 days (long course, n = 101). Over the study period, the authors note that IV antibiotic durations shortened, possibly as a result of recommendations for earlier oral transitions provided by the antimicrobial stewardship team at the main children’s hospital in 2014.

As would be expected, the hospital length of stay was considerably shorter in patients receiving short IV courses (median of 2 days vs 4 days). No infants had recurrent bacteremic UTI. Although the prespecified noninferiority criteria were not met, 30-day revisits were uncommon, occurring in 3 of 73 patients in the short-course group vs 1 of 101 in the long-course group (4.1 vs 2.0%, P = .65). Similarly, recurrent UTI caused by the same organism was rare (2.7 vs 1%, P = .58).

These findings are consistent with 2 related previous investigations,^14,15  as summarized in Table 1. All 3 papers demonstrate that IV antibiotic durations for bacteremic UTI are highly variable, and that outcomes are excellent regardless of the initial IV duration.

Another important contribution of this investigation is the demonstration that 100% of the urinalyses in this cohort were positive (as defined by having at least trace leukocyte esterase). These findings are consistent with 2 previous investigations^16,17  demonstrating perfect or near-perfect sensitivity of the urinalysis (UA) in bacteremic UTI, which contrasts with the historically-reported suboptimal sensitivity of the UA when the urine culture alone is used as the gold standard. Whereas positive urine cultures can be caused by contamination or asymptomatic bacteriuria, having the same organism in the blood and the urine is almost irrefutably a true infection. Therefore, bacteremic UTI represents a more accurate depiction of the true sensitivity of the UA, and previous reports of suboptimal UA sensitivity may be biased by the challenges inherent to using the urine culture alone as the gold standard.

Like bacteremic UTI, uncertainty persists over the appropriate IV antibiotic duration for nontyphoidal Salmonella bacteremia (NTSB). In 2018 the American Academy of Pediatrics’ Red Book recommended that suspected NTSB should receive IV treatment until clearance of bacteremia, with a transition to oral antibiotics to complete 7-10 days. Supporting shorter courses of IV therapy, Glover et al¹³  describe their single-center experience with NTSB by analyzing 57 patients aged 3 months to 18 years with NTSB between 2008-2018. The authors excluded patients with focal infections, malaria, hemoglobinopathy, or immune compromise. They compared outcomes in patients who received ≤ 3 days (n = 24) vs >3 days (n = 33) IV antibiotics. No patients in either group were readmitted to the hospital for NTSB within 90 days. Only three patients, two in the short group and one in the long group (8% vs 3%, absolute risk difference 5%, 95% confidence interval -7 to 17%), had return visits to the emergency department for minor symptoms and none were hospitalized.

These findings support a similar study by Hess et al¹⁸  analyzing 51 patients with NTSB. That investigation also demonstrated that IV durations were quite variable and that bad outcomes were rare and not associated with the initial IV duration.

Proponents of prolonged IV courses may point out that there has yet to be a randomized controlled trial (RCT) demonstrating that shorter courses for either of these conditions are safe. Others may worry that the existing studies are underpowered to detect small but clinically important differences in outcomes between groups.

We appreciate these concerns and would welcome RCTs and/or larger observational studies. However, it is worth questioning just how far we need to go to prove a null hypothesis that was never rejected in the first place. In this case, the null hypothesis would suggest that there are no differences in outcomes for children given prolonged IV antibiotics as compared with those who are given shorter IV courses or all oral courses. We are unaware of a single study rejecting this null hypothesis for any bacterial infection in children, other than studies showing harm from the PICC lines placed to provide the prolonged IV courses. On the contrary, there are now dozens of studies (including some RCTs) for many bacterial infections in children demonstrating non-inferiority of shorter IV or all oral courses, highlighting that this topic is ripe for stewardship efforts.^19,20 

How should you approach your next patient with uncomplicated bacteremic UTI or NTSB? We encourage embracing the null hypothesis and heeding our ethical obligations as physicians to do no harm and to use resources judiciously. Given the current body of literature, which is now further strengthened by the work of Olson et al¹²  and Glover et al,¹³  avoidance of prolonged IV courses is advised.