Epidemiology, microbiology and treatment implications in adult patients hospitalized with pneumonia in different regions of China: a retrospective study.

Background: Data describing epidemiology, clinical outcomes and treatment patterns, of hospitalised patients with pneumonia in China are limited. We aimed to describe such information among adult pneumonia patients in southern and northern China. Methods: We retrospectively reviewed electronic medical records of pneumonia patients aged >= 18 years, hospitalized between 2008 and 2013 at Guangdong Provincial Hospital of Chinese Medicine (n=3,636), southern China, and between 2010 and 2014 at Peking University People's Hospital, Beijing (n=1,689), northern China, in order to collect data on patient demographics, microbiology, clinical outcomes and treatment and resistance patterns. Results: The mean (SD) age of patients was 60.0 (21.4) and 64.4 (18.4) years in Guangdong and Beijing, respectively. Mean length of hospital stay was 12.1 and 20.8 days, and overall mortality was 2.9% and 8.0%, respectively. Gram-negative bacilli were most frequently isolated, predominantly Acinetobacter baumannii, Klebsiella pneumoniae and Pseudomonas aeruginosa. Infection with these bacteria was associated with unfavourable clinical outcomes, and the antibiotic resistance among these bacteria increased between 2008-2010 and 2011-2013 in both regions of China. The treatment and choice of antibiotics slightly varied between the two regions based on the susceptible pathogens identified among their populations. Conclusions: Of the pathogens identified, Staphylococcus aureus infection (particularly the methicillinresistant S. aureus) was associated with poor clinical outcomes; however antibiotic resistance among S. aureus generally decreased during the study data collection periods. Also, disease severity was greater in Beijing as compared with Guangdong, and this may be associated with higher microbiological diagnosis rate and higher frequency of initial antibiotic modification among Beijing populations.