Purpose: This study was designed to compare the diagnostic yield of microbiological analysis performed on diluted and undiluted vitreous samples from pars plana vitrectomy (PPV) in patients with acute postcataract endophthalmitis.
Design: Cohort study, evaluation of diagnostic test or technology.
Participants: Patients with acute postcataract endophthalmitis (<6 weeks).
Methods: Undiluted and diluted vitreous samples were taken from 34 consecutive patients at the beginning of PPV as part of the multicenter prospective study of the French Institutional Endophthalmitis Study (FRIENDS) group. Vitrectomy was performed after 1 (n = 12) or 2 (n = 22) intravitreous antibiotic injections. McNemar's nonparametric test was used to compare culture and polymerase chain reaction (PCR) results between diluted and undiluted samples.
Main outcome measures: Rate of positivity of conventional culture (brain heart infusion broth) and eubacterial PCR tests from undiluted and diluted vitreous samples.
Results: The microbiological analysis of both undiluted and diluted vitreous samples detected and identified a bacterial pathogen in 26 out of 34 cases (76.4%). The analysis of undiluted and diluted vitreous at the time of PPV, using eubacterial PCR and conventional culture, gave similar results (P = 0.99; McNemar test). However, eubacterial PCR was more sensitive than culture in detecting bacteria in vitreous at the time of PPV (76% vs 6%; P = 0.001; McNemar test). The difference in sensitivity between the 2 techniques was primarily associated with false-negative culture results for undiluted samples (2/3 of cases), mainly for coagulase-negative staphylococci.
Conclusions: The microbiological results obtained combining PCR and culture techniques were similar for diluted vitreous and undiluted vitreous analysis. When eubacterial PCR is available, sampling diluted vitreous, an easier procedure, may replace sampling undiluted vitreous.