Objective: To reevaluate the necessity of empiric gramnegative therapy
for postoperative endophthalmitis based on the recognition of aminogl
ycoside-induced toxic effects. Methods: A review of 162 cases of cultu
re-positive bacterial endophthalmitis to determine bacterial prevalenc
e, antibiotic susceptibilities, and laboratory staining results of int
raocular material. Results: One hundred fifty (93%) gram-positive and
12 (7%) gram-negative isolates. All the gram-positive bacteria were su
sceptible to vancomycin hydrochloride; and gram-negative bacteria were
susceptible to gentamicin sulfate, amikacin sulfate, and ceftazidime
sodium. Bacterial rods were observed on smear of intraocular material
in 10 gram-negative cases submitted for examination. Conclusion: Vanco
mycin is an appropriate single agent if laboratory studies rule out gr
am-negative infection. In the absence of laboratory support, an aminog
lycoside (possible toxic effects) or another gram-negative antibiotic
(possibly ceftazidime) should also be administered.