SummaryBackground Difficult-to-treat resistant (DTR) Gram-negative infections show resistance to all first-line antibiotics (ie, β-lactams and fluoroquinolones) and have a 40% greater mortality rate than susceptible infections. New antibiotics are now available with improved safety and efficacy and with in-vitro activity against DTR infections; however, their influence on the outcomes of patients with DTR infections remains unclear. We aimed to evaluate whether and why mortality in patients with DTR infections has changed since the introduction of these newer antibiotics in the USA. Methods In this retrospective cohort study in the USA, adult patients (aged ≥18 years) with a DTR Gram-negative infection, defined as microbiological evidence of DTR Enterobacterales, Pseudomonas aeruginosa, or Acinetobacter baumannii and receipt of at least 3 consecutive days of any antibiotic therapy, were identified from hospitals reporting microbiology data in the PINC-AI Healthcare Database. We characterised the proportion of inpatient encounters receiving newer DTR-active antibiotics, traditional DTR-active antibiotics, and non-DTR-active antibiotics. We used a generalised linear mixed model with marginal predictions to examine changes in in-hospital mortality, defined as death or discharge to hospice, over the study period when adjusting for patient-related and treatment-related factors (including receipt of a new antibiotic and receipt of in-vitro discordant initial therapy), hospital-related factors (including the availability of newer antibiotics and corresponding susceptibility testing), and COVID-19 pandemic-related factors. A three-way interaction term for time (year), pathogen, and infection site (ie, bloodstream and non-bloodstream) was included given the expected differences in mortality. Findings Between Jan 1, 2016, and Aug 31, 2023, 8 319 398 adult inpatient encounters with available microbiology data were recorded from 471 hospitals, of which 9384 (0.11%) encounters had microbiological evidence of an eligible DTR organism. 5065 (54·0%) of these 9384 encounters, from 262 hospitals, met the inclusion criteria for DTR Gram-negative infections and were included in the study. Among this cohort, the prescription of newer antibiotics, as well as the availability of newer antibiotics and their corresponding susceptibility tests, increased substantially from 2016 to 2023. Although the proportion of encounters in which the patient received a newer antibiotic as initial therapy increased from 4% (21 of 589) in 2016 to 15% (34 of 234) in 2023, in most cases (196 84% of 234) patients continued to receive in-vitro discordant initial antibiotic therapy, even in 2023. We observed no change in the average marginal effect (the average percentage change per year) for adjusted mortality between 2016 and 2023 for Enterobacterales (0.1% 95% CI −1.1 to 1.4), P aeruginosa (−0.7% −1.7 to 0.3), or A baumannii (−0.4% −1.8 to 0.9) infections. When dichotomised into bloodstream and non-bloodstream infections, the marginal effect for adjusted mortality remained unchanged over time for most pathogen and site combinations, with the exception of P aeruginosa bloodstream infections, for which a decrease was observed (−4.5% −8.2 to −0.60). Interpretation Despite the availability of newer antibiotic agents, the estimated mortality and ongoing use of in-vitro discordant initial antibiotics remains unacceptably high among patients with DTR infections in US hospitals. Prompt recognition of both the pathogen and resistance phenotype could be a crucial component in reducing mortality. Although notable, the decrease over time in adjusted mortality for P aeruginosa bloodstream infections should be considered hypothesis-generating because the cohort of patients with such infections was small. Funding US National Institutes of Health (NIH) Clinical Center; US National Cancer Institute; the Intramural Research Program of the US National Heart, Lung, and Blood Institute; the US National Institute of Allergy and Infectious Diseases; and the US Food and Drug Administration.
Walker et al. (Sun,) studied this question.