Community-acquired pneumonia (CAP) is common and remains the leading infectious cause of death worldwide. Because causative organisms are frequently not identified, clinicians rely on empirical antibiotic regimens. Current guideline options for non-severe CAP include combination therapy with a beta-lactam plus a macrolide or monotherapy with a fluoroquinolone. Growing concerns about antibiotic resistance and safety have prompted increased use of doxycycline as an alternative to macrolides, but evidence directly comparing doxycycline with macrolides in hospitalised patients is limited.
This protocol outlines a pragmatic, open-label, randomised clinical trial designed to compare the effectiveness of azithromycin versus doxycycline when each is given in combination with a beta-lactam in adults hospitalised for CAP. The trial aims to provide comparative effectiveness data to inform practice where empirical combination therapy is used.
The study is a pragmatic, open-label, randomised clinical trial with 1:1 allocation between azithromycin and doxycycline arms. The planned sample size is 1,120 adult participants. Recruitment will occur at six hospitals across four states, selected to include a mix of rural, suburban and urban populations as well as academic and community hospital settings. The pragmatic design emphasizes integration with routine clinical care and minimises additional study procedures.
Patients eligible for inclusion are adults who present to the emergency department or are referred directly to the hospital with a clinical presentation concerning for CAP and an indication for hospitalisation. The protocol specifies that the need for inpatient treatment is a criterion for enrolment, but the source document does not provide further inclusion or exclusion details such as specific clinical criteria, radiographic requirements, or comorbidity exclusions; those specifics were not reported in the source.
Randomisation will be performed 1:1 using an electronic health record (EHR)-embedded screening and randomisation tool. The tool is triggered by the treating clinician’s order for either azithromycin or doxycycline, enabling point-of-care enrolment and assignment. This approach integrates trial processes into routine prescribing workflows to support the pragmatic nature of the study and reduce additional provider burden.
The primary outcome is days alive and out of the hospital through day 28, a patient-centered endpoint combining survival and hospital-free days.
Secondary outcomes include:
Outcomes will be ascertained pragmatically through EHR-based data retrieval without requiring additional study visits, consistent with a low-burden trial model.
Safety outcomes captured in the protocol include the incidence of Clostridioides difficile infection and QT prolongation. The source does not provide further detail on specific monitoring frequency, thresholds for reporting, or management algorithms; those operational details were not reported in the source document.
Analyses will follow an intention-to-treat framework, with per-protocol analyses planned as sensitivity checks. The primary comparison of the primary endpoint between groups will use a stratified van Elteren test. Supportive analyses will include proportional odds regression and analysis of covariance (ANCOVA) adjusted for key covariates. The protocol does not list the full set of covariates to be included in adjusted models in the source text.
The Mayo Clinic Institutional Review Board reviewed and approved the protocol with a waiver of written consent. Under the approved approach, verbal consent will be obtained by the clinician who initiates the antibiotics (IRB# 24 - 0 06 503). The investigators intend to disseminate trial findings through peer-reviewed publications and presentations at international conferences.
This trial is registered with ClinicalTrials.gov under the registration number NCT07164131.
(Details not reported in the source: specific inclusion and exclusion criteria, dosing regimens for azithromycin or doxycycline and the beta-lactam partner, monitoring schedules for safety outcomes, and the full list of covariates for adjusted analyses were not provided in the source article.)