OBJECTIVE: To analyze the data for cultures and Gram stains prospectively collected by protocol in the Endophthalmitis Vitrectomy Study. DESIGN: Cultures of aqueous, undiluted vitreous, and (for patients who underwent vitrectomy) vitrectomy cassette fluid obtained from 420 patients were prepared on chocolate agar, in thioglycolate broth, and on Sabouraud dextrose agar; Gram stains of the aqueous and undiluted vitreous were made. Criteria were devised to distinguish true pathogens (confirmed positive cultures) from contaminants. SETTING: Private and university-based retina-vitreous practices and corresponding microbiology laboratories. RESULTS: Compared with the aqueous, undiluted vitreous produced a higher percentage of confirmed positive cultures and higher colony counts on chocolate agar and was more frequently the only source of a positive culture from the eye. Nevertheless, the aqueous and vitrectomy cassette fluid were the only source of a positive culture from the eye in 4.2% and 8.9% of eyes, respectively. The overall yields of chocolate agar and thioglycolate broth were similar. A positive Gram stain from the aqueous or undiluted vitreous was highly predictive of a positive culture from the eye, but a negative Gram stain had little predictive value for the culture result. The overall rate of laboratory-confirmed infection was not statistically significantly higher in the vitrectomy group than in the tap or biopsy group. CONCLUSIONS: The vitreous was a richer source of positive cultures and high colony counts than was the aqueous, either because it is more supportive of bacterial growth or because a somewhat larger inoculum of the vitreous than of aqueous could be obtained. The result of Gram stain should not determine the choice of antibiotic drugs in the treatment of endophthalmitis. Vitrectomy, with culture of the vitrectomy cassette fluid, did not produce significantly more positive cultures than tap or biopsy material, and the procedure should not be performed to improve the microbiological yield.
RCT Entities:
OBJECTIVE: To analyze the data for cultures and Gram stains prospectively collected by protocol in the Endophthalmitis Vitrectomy Study. DESIGN: Cultures of aqueous, undiluted vitreous, and (for patients who underwent vitrectomy) vitrectomy cassette fluid obtained from 420 patients were prepared on chocolateagar, in thioglycolate broth, and on Sabouraud dextrose agar; Gram stains of the aqueous and undiluted vitreous were made. Criteria were devised to distinguish true pathogens (confirmed positive cultures) from contaminants. SETTING: Private and university-based retina-vitreous practices and corresponding microbiology laboratories. RESULTS: Compared with the aqueous, undiluted vitreous produced a higher percentage of confirmed positive cultures and higher colony counts on chocolateagar and was more frequently the only source of a positive culture from the eye. Nevertheless, the aqueous and vitrectomy cassette fluid were the only source of a positive culture from the eye in 4.2% and 8.9% of eyes, respectively. The overall yields of chocolateagar and thioglycolate broth were similar. A positive Gram stain from the aqueous or undiluted vitreous was highly predictive of a positive culture from the eye, but a negative Gram stain had little predictive value for the culture result. The overall rate of laboratory-confirmed infection was not statistically significantly higher in the vitrectomy group than in the tap or biopsy group. CONCLUSIONS: The vitreous was a richer source of positive cultures and high colony counts than was the aqueous, either because it is more supportive of bacterial growth or because a somewhat larger inoculum of the vitreous than of aqueous could be obtained. The result of Gram stain should not determine the choice of antibiotic drugs in the treatment of endophthalmitis. Vitrectomy, with culture of the vitrectomy cassette fluid, did not produce significantly more positive cultures than tap or biopsy material, and the procedure should not be performed to improve the microbiological yield.