Trial Design

SCUT was a National Eye Institute–funded, randomized, placebo-controlled, double-masked multicenter clinical trial that compared clinical outcomes in patients receiving adjunctive topical corticosteroid or topical placebo in the treatment of bacterial corneal ulcers. Detailed trial methods have been described elsewhere. Briefly, 500 patients with culture-positive bacterial corneal ulcers received at least 48 hours of topical moxifloxacin, 0.5% (Vigamox; Alcon, Fort Worth, Texas, USA), before being randomized to receive either topical prednisolone phosphate, 1.0% (Bausch & Lomb Pharmaceuticals, Inc, Tampa, Florida, USA), or topical placebo (sodium chloride, 0.9%, and preservative, prepared by Leiter’s Compounding Pharmacy, San Jose, California, USA). Specific details of sample size determination for this trial have been reported in depth previously. Patients were randomized in a 1:1 ratio by center in random block sizes of 4, 6, or 8 using the previously described randomization allocation sequence. As the placebo appeared identical to the prednisolone phosphate solution, double masking of patient and examiner was achieved. Prospective institutional review board approval for this study was obtained from the Aravind Eye Care System’s Institutional Review Board, Dartmouth-Hitchcock Medical Center Committee for the Protection of Human Subjects, and the University of California, San Francisco Committee on Human Research. Informed consent was obtained from all study participants. The trial was compliant with the Health Insurance Portability and Accountability Act, observed the Declaration of Helsinki, was approved by the Food and Drug Administration (IND #71,800), and was registered at clinicaltrials.gov ( NCT00324168 ).

Study Participants

Eligible patients with culture-proven bacterial corneal ulcers were randomized to receive either topical prednisolone phosphate or placebo after receiving at least 48 hours of topical moxifloxacin. Complete eligibility criteria have been described in depth elsewhere. In brief, major exclusion criteria included corneal perforation or impending perforation; evidence of fungus on potassium hydroxide preparation, Giemsa stain, or culture; evidence of acanthamoeba by stain; evidence of herpetic keratitis by history or examination; use of a topical corticosteroid or systemic prednisolone during the course of the present ulcer; previous penetrating keratoplasty; and vision less than 6/60 in the fellow eye. Patients were enrolled at the Aravind Eye Care System (Madurai, Coimbatore, and Tirunelveli, India), the Dartmouth-Hitchcock Medical Center (Lebanon, New Hampshire, USA), and the Francis I. Proctor Foundation for Research in Ophthalmology at the University of California, San Francisco.

Intervention

All patients received 1 drop of moxifloxacin every hour while awake for the first 48 hours, then every 2 hours until re-epithelialization, and then 4 times a day until 3 weeks from enrollment. The treatment regimen for the study drug (either prednisolone phosphate or placebo) consisted of 1 drop applied topically 4 times per day for 1 week after randomization, then twice a day for 1 week, and then once per day for 1week. If deemed medically necessary, treating physicians were allowed to discontinue or change any medications during the study.

Main Outcome Measures

The primary outcome for this trial, best spectacle-corrected visual acuity (BSCVA) at 3 months, has been reported. As previously specified, patients were scheduled to return for an additional follow-up visit at 12 months. The outcomes of interest for this report are BSCVA at 12 months from enrollment, scar size at 12 months measured by slit-lamp examination, and adverse events, including corneal perforation. As with all study visits, visual acuity was measured by refractionists certified for the study using a tumbling “E” chart at 4 m and logMAR visual acuity (charts 2305 and 2305A; Precision Vision, La Salle, Illinois, USA). Visual acuity measurements were assessed based on the total number of letters read correctly. If fewer than 10 letters were read at 4 m, acuity was assessed at 1 m. If a patient read fewer than 10 letters at 1 m, counting fingers, hand motions, light perception, and no light perception were used to assess low vision. Further detailed methods for outcome assessments have been reported elsewhere.

Statistical Methods

Patient characteristics at enrollment were compared by treatment arm for those patients who returned for follow-up at 12 months. To examine the potential effects of loss to follow-up between 3- and 12-month visits, enrollment characteristics and 12-month follow-up visit status among those with a 3-month visit were also compared by treatment arm. BSCVA between 3 and 12 months was examined by treatment group and level of improvement in visual acuity. In all visual acuity analyses, for those patients who underwent therapeutic penetrating keratoplasty before follow-up visual acuity measurements, we used the last observation carried forward (LOCF) or 1.7 logMAR acuity, whichever was worse. Comparisons were made using Wilcoxon rank sum tests for continuous variables and Fisher exact tests for categorical variables. All P values were 2-sided and P < .05 was considered significant. All analyses were performed using Stata version 10.0 (StataCorp LP, College Station, Texas, USA) or the R program (R Foundation for Statistical Computing, Vienna, Austria).

Only those visits that fell within the 12-month visit window (10-14 months from enrollment) were included in the analysis. For BSCVA at 12 months, the prespecified analysis involved a linear regression model with enrollment BSCVA and treatment arm as covariates. After the SCUT 3-month results were reported, further analyses indicated that the model of this relationship could be improved. We used higher-order polynomials to better fit the relationship between enrollment and 12-month acuity, and we performed a sensitivity analysis with cubic splines to demonstrate that results were not dependent on the exact modeling technique. We also added an interaction term for Nocardia ulcers given the differential effects of corticosteroids previously reported. The 12-month visual acuity was modeled using Gram-Schmidt orthogonal polynomials of the enrollment acuity. We chose the maximum polynomial degree (6) by cross-validation using a randomly chosen 20% of the data as a test set, conducting the regression on the remaining 80% of the data, using the resulting regression model to predict outcomes for the test set, and choosing the polynomial degree to minimize the squared error of prediction. Polynomial terms up to this maximum order were included in the final regression model, along with treatment, the Nocardia indicator, and the treatment- Nocardia interaction. The permutation P value for the primary outcome was determined by Monte Carlo (10 000 simulations), comparing the model with both treatment and Nocardia -treatment interaction to a model that included neither (likelihood ratio test, 2 degrees of freedom). Subgroups of non- Nocardia ulcers were examined in a similar model. Scar size at 12 months was analyzed in a linear regression model with 1 additional covariate accounting for effect modification by Nocardia ulcers. Adverse events between the 2 groups were compared using Fisher exact test.

Sensitivity analyses were conducted for BSCVA and scar size. BSCVA at 3 months and scar size at 3 months were reexamined using the linear regression models with the additional covariates. Hard contact lens–corrected visual acuity at 12 months was examined in a similar model. Another analysis used LOCF for patients lost to follow-up at 12 months and included data for those patients who had a 12-month study visit that fell outside of the prespecified window. To further examine the appropriateness of the final model for BSCVA, a cubic spline function with 4 knots was used to correct for baseline BSCVA in the model at 12 months. Finally, variables in which significant differences ( P < .05) in baseline characteristics were noted between arms were added as covariates in the model of BSCVA at 12 months.