The Effect of Macrolides on Mortality in Bacteremic Pneumococcal Pneumonia: A Retrospective, Nationwide Cohort Study, Israel, 2009-2017

Background: Previous cohort studies of pneumonia patients reported lower mortality with advanced macrolides. Our aim was to characterize antibiotic treatment patterns and assess the role of quinolones or macrolides in empirical therapy. Materials: An historical cohort, 1 July 2009 to 30 June 2017, included, through active surveillance, all culture-confirmed bacteremic pneumococcal pneumonia (BPP) among adults in Israel. Cases without information on antibiotic treatment were excluded. Logistic regression analysis was used to assess independent predictors of in-hospital mortality. Results: A total of 2016 patients with BPP were identified. The median age was 67.2 years (interquartile range [IQR] 53.2-80.6); 55.1% were men. Lobar pneumonia was present in 1440 (71.4%), multi-lobar in 576 (28.6%). Median length of stay was 6 days (IQR 4-11). A total of 1921 cases (95.3%) received empiric antibiotics with anti-pneumococcal coverage: ceftriaxone, in 1267 (62.8%). Coverage for atypical bacteria was given to 1159 (57.5%), 64% of these, with macrolides. A total of 372 (18.5%) required mechanical ventilation, and 397 (19.7%) died. Independent predictors of mortality were age (odds ratio [OR] 1.051, 95% confidence interval [CI] 1.039, 1.063), being at high-risk for pneumococcal disease (OR 2.040, 95% CI 1.351, 3.083), multi-lobar pneumonia (OR 2.356, 95% CI 1.741, 3.189). Female sex and macrolide therapy were predictors of survival: (OR 0.702, 95% CI. 516,. 955; and OR 0.554, 95% CI. 394,. 779, respectively). Either azithromycin or roxithromycin treatment for as short as two days was predictor of survival. Quinolone therapy had no effect. Conclusions: Empirical therapy with macrolides reduced odds for mortality by 45%. This effect was evident with azithromycin and with roxithromycin. The effect did not require a full course of therapy.