Is empirical antibiotic treatment correlated to outcome in microbiology-negative pleural infection?

Abstract

A large proportion of patients with pleural infection do not get a microbiological diagnosis. The reasons may include pre-hospital antibiotic treatment, delayed diagnostic thoracocentesis in hospital, or insufficient diagnostics. Aim & objectives: To establish if the choice of empirical antibiotic treatment was associated with outcome. Methods: Patients with culture-negative or no pleura samples were included from a 3 year cohort of 442 “empyema” patients in 9 hospitals in East-Denmark. Clinical, biochemical, microbiological, radiological, and treatment data were collected form the medical records. Initial antibiotic treatment “PENI”= Penicillin or amoxicillin +/-Metronidazol, and “CEFU”=Cefuroxim +/-Metronidazol. Outcomes included death, the need for surgical treatment, and length of hospital stay (LOS). Descriptive or non-parametric statistical analysis was used when appropriate. Results: 207 patients had microbiology-negative pleural infection with 11% nosocomial infection, a median LOS of 16 days (range 1-75). Mean age was 60.8 years and 62% were male. Median length of antibiotic treatment 35 days (range 0-113). N=24 (11.6%) were severely ill within the first 3 days, and n=13 (6.3%) were admitted to ICU. Among patients treated with PENI (n=64) versus CEFU (n=101), mortality was 1.5% vs 8.9% (p=0.11, Yates corrected), a need for surgical treatment was in 6.5% vs 13.8% (p=0.20), LOS was 17.3 vs 18.3 days (p=0.61). Conclusion: using univariate analyses, the choice of initial antibiotic treatment was not statistical significantly correlated to outcome. Other clinical co-factors (f.ex. risk factors, inadequate handling) may adequately be included in further multivariate analyses.