Doxycycline Versus Azithromycin As Combination Therapy in the Empirical Treatment of Community-Acquired Pneumonia in an Intensive Care Unit – A Retrospective Propensity Score-Matched Cohort Study

INTRODUCTION:Community-acquired pneumonia constitutes a significant burden to healthcare systems. While international guidelines recommend beta-lactam plus macrolide in treating severe community-acquired pneumonia, doxycycline instead of macrolide provides antimicrobial activity to macrolide-resistant organisms that are highly prevalent in East Asia and avoids its side effects. Yet, doxycycline's immunomodulatory effect is not as well studied. We investigated the comparative efficacy of doxycycline with azithromycin in the treatment of severe community-acquired pneumonia. METHODS:From 1 January 2014 to 31 December 2023, consecutive adult patients with pneumonia or pulmonary sepsis requiring intensive care within 48 h of hospital admission, who received a beta-lactam in combination with either azithromycin or doxycycline were retrospectively reviewed. The two groups were matched by propensity scores and compared for all-cause mortality at day 30. RESULTS:A total of 670 patients were analysed. Streptococcus pneumoniae was the dominant bacteria identified in one fourth of all patients with a bacterial cause. 75% of those cultured were macrolide resistant. Mycoplasma pneumoniae was detected in 0.9% of all patients, while detection of Legionella pneumoniae was 2.8 times higher. Propensity score matching resulted in 250 pairs of patients with balanced covariates. Thirty-day mortality did not differ between groups (21.1% in beta-lactam/azithromycin vs 19.2% in beta-lactam/doxycycline, p = 0.649). A longer duration of azithromycin more than or equal to 3 days did not alter this outcome. CONCLUSIONS:Beta-lactam/azithromycin or beta-lactam/doxycycline are viable empirical treatments for severe community-acquired pneumonia despite local resistance patterns. Our result did not support a longer duration of azithromycin over doxycycline.