To determine the role of initial pars plana vitrectomy in the management of postoperative bacterial endophthalmitis.

To determine the role of intravenous antibiotics in the management of bacterial endophthalmitis.

To determine which factors, other than treatment, predict outcome in postoperative bacterial endophthalmitis.

Endophthalmitis is a serious ocular infection that can result in blindness. Approximately 70 percent of cases occur as a direct complication of intraocular surgery. Current management requires culture of intraocular contents and administration of an antibiotic. Vitrectomy surgery, which may help to manage endophthalmitis by removing infecting organisms and their toxins, has been shown to be of value in various animal models of endophthalmitis. However, human studies have not shown an advantage to vitrectomy with intraocular antibiotics compared with intraocular antibiotics alone.

In all large comparison studies to date, eyes with the worst initial presentations were the ones selected for vitrectomy. Because of the selection bias involved in determining which cases received vitrectomy, existing clinical information on the efficacy of the procedure for treating endophthalmitis is inconclusive. Determining the role of initial vitrectomy and the benefit or lack of benefit to certain subgroups of patients will help the clinician in the management of endophthalmitis.

In addition, although systemic antibiotics have long been used in the management of endophthalmitis, there has been little evidence to support their efficacy, but there have been many reports of toxic systemic effects. In view of this, the role of systemic antibiotics in the management of endophthalmitis will be assessed.

Endophthalmitis Vitrectomy Study (EVS) patients were randomized to one of two standard treatment strategies for the management of bacterial endophthalmitis. Eyes received either (1) initial pars plana vitrectomy with intravitreal antibiotics, followed by retap and reinjection at 36-60 hours for eyes that did poorly as defined in the study or (2) initial anterior chamber and vitreous tap/biopsy with injection of intravitreal antibiotics, followed by vitrectomy and reinjection at 36-60 hours in eyes doing poorly. In addition, all eyes were randomized to either treatment or no treatment with intravenous antibiotics.

Study end points were visual acuity and clarity of ocular media, the latter assessed both clinically and photographically. Each patient's initial end point assessment occurred at 3 months, after which procedures to improve vision, such as late vitrectomy for nonclearing ocular media, were an option. The final outcome assessment occurred at 9 months. Multiple centers cooperated by enrolling 420 eyes during the 42-month recruitment period.

Men and women were eligible for entry into the EVS if they had clinical signs and symptoms of bacterial endophthalmitis in an eye that had cataract surgery or lens implantation within 6 weeks of onset of infection. The involved eye had to have either hypopyon or enough clouding of anterior chamber or vitreous media to obscure clear visualization of second-order arterioles, a cornea and anterior chamber in the involved eye clear enough to visualize some part of the iris, and a cornea clear enough to allow the possibility of pars plana vitrectomy. The eyes had to have a visual acuity of 20/50 or worse and light perception or better.

Patients were ineligible when the involved eye was known at the time of study entry to have had any pre-existing eye disease that limited best-corrected visual acuity to 20/100 or worse before development of cataract, any intraocular surgery before presentation (except for cataract extraction or lens implantation), any treatment for endophthalmitis before presenting at the study center, or any ocular or systemic condition that would prevent randomization to any of the study groups.