Reviewed by:
Tyler Stephen, MD, University of Rochester Medical Center
Jose Lucar, MD, FIDSA, The George Washington University
As growing evidence continues to support the safety and efficacy of shorter durations of antibiotic therapy, few recent studies have attracted as much attention as the BALANCE trial.(1) This large, randomized study demonstrated that among non‑immunocompromised adults with non–Staphylococcus aureus bacteremia and no identifiable infectious focus requiring prolonged therapy (such as endocarditis, prosthetic device infection, or osteoarticular infection), a 7‑day course of antibiotics was noninferior to a 14‑day course. As with any large pragmatic trial, however, questions remain regarding data interpretation and the applicability of these findings to specific patient subgroups. Two recent post hoc analyses have sought to address some of these residual uncertainties related to BALANCE.
In one post hoc analysis (2), the investigators examined the relatively high rate of nonadherence to the assigned treatment duration observed in the parent trial (20.3%). Nonadherence occurred more frequently in the 7‑day group (24%) than in the 14‑day group (16%). In the 7‑day group, nonadherence was almost exclusively due to receipt of a longer-than-assigned antibiotic course, whereas in the 14‑day group it reflected a combination of both shorter‑ and longer‑than‑assigned durations. Patients who did not adhere to the assigned duration were more likely to be more acutely ill at baseline, to have a non‑urinary source of infection, or to have bacteremia due to non‑Enterobacterales organisms. After identifying clinical factors associated with nonadherence, the authors applied inverse probability of treatment weighting (IPTW) to perform adjusted per‑protocol and as‑treated analyses. To further address residual confounding, an instrumental variable analysis was also conducted, using randomized treatment assignment as the instrument. Notably, effect estimates from the adjusted IPTW and instrumental variable analyses yielded confidence intervals that overlapped with those from the intention‑to‑treat and unadjusted per‑protocol analyses, with all confidence intervals remaining within the prespecified 4% noninferiority margin.
In a separate post hoc analysis (3), the investigators focused on Pseudomonas aeruginosa, an important and clinically challenging cause of nosocomial bloodstream infections. This analysis included 157 patients from the overall BALANCE trial who were enrolled with pseudomonal bacteremia. Within this subgroup, baseline imbalances were observed, including a higher proportion of men and hospital‑acquired infections among patients randomized to the 7‑day arm compared with the 14‑day arm. Across key clinical outcomes, the estimated absolute risk differences demonstrated confidence intervals that overlapped zero. In exploratory logistic regression analyses examining predictors of mortality, increasing age was identified as the only statistically significant factor, while treatment assignment (7 versus 14 days) was not independently associated with mortality. Taken together, these findings suggest no clear signal of harm with shorter therapy in this subgroup, though the small sample size and baseline imbalances warrant cautious interpretation.
Together, these post hoc analyses provide additional context to the original BALANCE trial and largely support its primary conclusion: that a 7‑day course of antibiotics is noninferior to a 14‑day course for non‑Staphylococcus aureus bacteremia in immunocompetent patients without infectious foci requiring prolonged therapy. At the same time, the pseudomonal bacteremia analysis highlights important limitations, including small sample size and baseline imbalances, and suggests that the strength of these conclusions may vary across specific subgroups. As such, while no clear signal of harm was observed, these findings underscore the need for cautious interpretation and highlight the potential value of future studies with larger, pathogen‑specific cohorts to better define optimal treatment durations.
References:
