Introduction and Objectives Thoracic empyema affects >65 000 patients/year in the US and UK. Up to 50% require surgical drainage, and 15% die. We describe clinical features, microbiology, risk factors and surgical outcomes from referrals to a regional specialist cardiothoracic centre over 12 year.

Methods Patients were identified by searching the hospital Clinical Data Repository for a diagnosis of ‘pyothorax’ from 1999 to 2010. A retrospective observational study was conducted using case note and database review.

Results 406 distinct empyemas were identified. Patients were predominantly male (n=301, 74.1%), with median age of 53 years (IQR=37–69). Pyothorax predominantly developed secondary to community-acquired pneumonia, although 51 (12.6%) were hospital-acquired; 70 (17.2%) were receiving steroids or immunosuppressants, 35 (8.6%) had concurrent malignancy, 33 (8.1%) were diabetic and 14 (3.4%) patients were HIV-infected. Empyemas were right-sided in 233 (57.3%, p=0.03); four were bilateral. A causative organism was identified in 229 (56.4%) patients. Positive cultures were obtained from sputum in 59 (14.5%), pleural fluid in 174 (42.9%) and blood in 61 (15.0%) patients. Organisms identified included Streptococcus milleri (n=18, 4.4%), Streptococcus pneumoniae (n=39, 9.6%), other streptococci (n=10, 2.5%), Enterobacteraciae (n=12, 3.0%), anaerobes (n=32, 7.9%), methicillin-sensitive Staphylococcus aureus (n=36, 8.9%), methicillin-resistant Staphylococcus aureus (n=25, 6.2%), Enterococcus (n=14, 3.4%), Mycobacterium tuberculosis (n=36, 8.9%), non-tuberculous mycobacteria (n=1, 0.2%), other bacteria (n=25, 6.2%), and fungi (n=9, 2.2%); 34 (8.4%) cultures were polymicrobial. 277 (68.2%) patients were managed by open thoracotomy. 116 (28.6%) underwent video-assisted thoracoscopic surgery (VATS), of whom 17 (14.7%) were converted to open. Three (0.7%) required a Clagett window; 10 (2.5%) were managed with tube thoracostomy alone. A significant trend towards increased use of VATS over time was noted (p=0.0002). All-cause complication rate was 20.7%, and 28-day mortality was 5.7%. Low preoperative haemoglobin was predictive of mortality (p=0.04), but admission and peak C-reactive protein were not.

Conclusions In this large series of empyemas from a tertiary referral centre, microbiological diagnosis was achieved in only half the patients. The identification of tuberculosis in 36 cases means routine screening for mycobacteria is necessary. Use of VATS has increased significantly, and mortality was lower than previously reported and associated with preoperative anaemia.