Introduction: Reports regarding the external validity of randomized controlled trials (RCTs) are scarce. We aimed to assess the population external validity of an investigator-initiated RCT on the duration of antibiotics for the treatment of Gram-negative bacteremia by comparing patients included in the RCT to patients that were not included in the trial.

Methods: Hospitalized patients with Gram-negative bacteremia were recruited into an RCT and randomized to receive 7 or 14 days of covering antibiotic therapy in Israel and Italy from 2013 to 2017. In a concomitant observational study, RCT participants were compared with patients who fulfilled the inclusion criteria but were not included in the trial due to participation in other trials, discharge before approached by researchers, refusal to participate, or unwillingness of the treating physician to allow participants' recruitment.

Results: Six hundred and four RCT patients were compared with 613 nonincluded patients. Almost 50% of nonincluded patients (288/613) were dependent on others for activities of daily living at baseline compared to 37.7% of RCT participants (228/604). Dementia was nearly 2-fold more frequent in nonincluded patients than those included (5.9% [36/613] versus 3.6% [22/604], p = 0.07). Patients who were not included in the RCT were more likely to acquire their infection in the hospital (53.3% [327/613] versus 29.1% [176/604], p < 0.001). The primary composite outcome of mortality, clinical failure, readmissions, or extended hospitalization at 90 days occurred in 353 of 613 nonincluded patients (57.6%) compared to 299 of 604 RCT participants (49.6%), p = 0.005. However, on multivariate analysis noninclusion in the RCT was not an independent risk factor for clinical failure and mortality.

Conclusions: RCTs, even with broad eligibility criteria, do not represent the whole spectrum of patients and leave out a population with more severe illness for whom the evidence is lacking.