To describe the usage patterns of steroids in bacterial keratitis and to analyze the effect of steroids on patients' visual outcomes.
This was a single-center retrospective review of patients with culture-positive bacterial keratitis treated between 1999 and 2015 at Princess Alexandra Hospital (Brisbane, Australia). Patients with culture-positive bacterial keratitis were identified through the Queensland Pathology Database, and clinical information was gathered through a subsequent medical record review. High-dose steroid treatment was classified as 6 or more drops of a steroid a day started within 7 days of corneal scraping. The outcome of a patient's episode of keratitis was classified as good if their final visual acuity was 6/12 or better; poor if it was 6/60 or worse or required a corneal transplant, otherwise it was classified as average. Microbiological and clinical variables' association with outcomes was evaluated in univariate analyses. Variables significant at P < 0.1 were examined in a multivariate ordinal logistic regression analysis created with forward variable selection with forced inclusion of steroid treatment (high, regular, low dose, and none).
A total of 328 patients were included from the 1002 reviewed charts. Of these patients, 164 (50.0%) were treated with steroids. Factors significantly associated with outcomes in the multivariate model were high-dose steroid treatment, visual acuity on presentation, age group, cause of keratitis, infiltrate size, and location. The odds ratio of better outcomes with high-dose steroids was 5.49 (confidence interval, 1.6–19.0, P = 0.007).
High-dose steroid treatment is significantly associated with better visual outcomes in patients with culture-positive bacterial keratitis in this case series.
*Ophthalmology Department, Gold Coast University Hospital, Southport, Australia;
†School of Optometry and Vision Science, University of New South Wales, Australia;
‡Ophthalmology Department, Sydney Eye Hospital, Sydney, Australia; and
§Ophthalmology Department, Princess Alexandra Hospital, Brisbane, Australia.
Correspondence: Matthew Green, MSc, MBBS, FRANZCO, Gold Coast University Hospital, Hospital Boulevard, Parklands, Australia (email: firstname.lastname@example.org).
The authors have no funding or conflicts of interest to disclose.
Received June 09, 2018
Accepted September 09, 2018