Azithromycin and survival in Streptococcus pneumoniae pneumonia: A retrospective study

Document Type


Publication Title

BMJ Open


Objective: Streptococcus pneumoniae (SP) represents a major pathogen in pneumonia. The impact of azithromycin on mortality in SP pneumonia remains unclear. Recent safety concerns regarding azithromycin have raised alarm about this agent's role with pneumonia. We sought to clarify the relationship between survival and azithromycin use in SP pneumonia. Design: Retrospective cohort. Setting: Urban academic hospital. Participants: Adults with a diagnosis of SP pneumonia (January-December 2010). The diagnosis of pneumonia required a compatible clinical syndrome and radiographic evidence of an infiltrate. Intervention: None. Primary and secondary outcome measures: Hospital mortality served as the primary endpoint, and we compared patients given azithromycin with those not treated with this. Covariates of interest included demographics, severity of illness, comorbidities and infection-related characteristics (eg, appropriateness of initial treatment, bacteraemia). We employed logistic regression to assess the independent impact of azithromycin on hospital mortality. Results: The cohort included 187 patients (mean age: 67.0±8.2 years, 50.3% men, 5.9% admitted to the intensive care unit). The most frequently utilised non-macrolide antibiotics included: ceftriaxone (n=111), cefepime (n=31) and moxifloxacin (n=22). Approximately two-thirds of the cohort received azithromycin. Crude mortality was lower in persons given azithromycin (5.6% vs 23.6%, p<0.01). The final survival model included four variables: age, need for mechanical ventilation, initial appropriate therapy and azithromycin use. The adjusted OR for mortality associated with azithromycin equalled 0.26 (95% CI 0.08 to 0.80, p=0.018). Conclusions: SP pneumonia generally remains associated with substantial mortality while azithromycin treatment is associated with significantly higher survival rates. The impact of azithromycin is independent of multiple potential confounders.



Publication Date