Postoperative Therapeutic Regimen

The general postoperative therapeutic regimen for the grafts involved commencement of topical prednisolone acetate 1% or dexamethasone phosphate 0.1% at 3-hour intervals along with either topical tobramycin, levofloxacin, or moxifloxacin. Topical cyclosporine generally was used as additional immunosuppression in high-risk cases (multiple grafts, therapeutic grafts where steroids were withheld for the first few weeks, eyes with significant vascularization, glaucoma, ocular surface disease), or as a steroid-sparing agent in cases of ocular hypertension. Topical steroids were tapered gradually to once daily by the end of 1 year, with antibiotic use in tandem stopped when steroid dosage was reduced to a twice-daily regimen (antibiotics were continued beyond 6 months in view of a greater risk of the occurrence of infective keratitis in our graft patients when they were concurrently receiving topical steroids). In low-risk grafts (eg, grafts performed for keratoconus, corneal scars, anterior stromal dystrophies), topical steroids were continued for at least 1 year, whereas in immunologically high-risk grafts, low-dose steroids were continued indefinitely without antibiotics. In cases of tectonic and therapeutic grafts, topical steroids generally were withheld for at least 1 week after surgery with the aim of reducing the risk of exacerbating a recurrence of infection.

Oral steroids were commenced in the perioperative period in immunologically high-risk grafts (vascularized corneas, repeat grafts) in those patients who consented and were fit for oral steroids and were tapered over 2 to 3 weeks in the postoperative period. In selected high-risk cases (based on individual surgeon preference, patient acceptance of therapy, as well as patient systemic fitness) and with appropriate risk-to-benefit ratio assessment and counseling, oral cyclosporine (2 to 4 mg/kg daily) was commenced after surgery and maintained empirically for 12 months unless systemic toxicities developed in the patient. Systemic mycophenolate mofetil was used in patients with contraindication to systemic cyclosporine or in patients in whom side effects from cyclosporine developed. Patients with clinical evidence of rejection were managed aggressively with a combination of oral prednisolone (1 mg/kg body weight) and topical prednisolone acetate 1% given at 1- to 2-hour intervals, with rapid tapering after resolution of rejection.

Risk Factors Analyzed

A total of 17 potential postoperative risk factors were identified. These included 9 postoperative graft complications (epithelial problems, wound dehiscence, suture abscess, endophthalmitis, microbial keratitis, reactivation of herpes simplex infection, recurrence of primary disease, glaucoma or raised intraocular pressure [IOP], and allograft rejection) and 8 postoperative surgical or laser procedures (glaucoma surgery or laser procedures, repeat corneal grafting, graft resuturing, various forms of graft refractive surgery, cataract surgery, eyelid procedures, yttrium–aluminum–garnet capsulotomy, and vitreoretinal surgery). In addition, multivariate analysis involving inclusion of other preoperative and intraoperative risk factors identified in an earlier study were combined with these postoperative risk factors for a final overall analysis.

Statistical Analysis

Kaplan-Meier survival analysis was conducted to analyze graft survival times for postoperative factors. Postoperative follow-up factors were categorized, highlighting reference categories for comparisons. The Kaplan-Meier product limit method was applied to generate individual survival curves with corresponding survival rates. To avoid time-dependent bias, Cox proportional hazards regression with time-dependent covariates was used to analyze follow-up risk factors (eg, complications and second surgery) associated with risk of graft failure. Proportional hazards assumption was tested to identify variables over time. The identified variables that violated the proportional hazards were controlled in the model during the analyses. In the combined preoperative, intraoperative, and postoperative follow-up model, age was tested for the proportional hazard assumption. Because age violated the assumption, in the final model, age was no longer treated as a fixed variable and was controlled for. Univariate analyses were conducted for each variable to assess the significance. The effect of each individual risk factor was quantified and reported as a hazard ratio (HR) compared with a referent group. The final models were adjusted for age, sex, and time-dependent factors that violated proportional hazard assumptions. The final models also were checked for colinearity among variables by inspecting the standard errors associated with each variable. The first model was tested for postoperative factors only, and the second model was tested for combined preoperative and postoperative factors. In the preoperative and postoperative factors, multivariate analysis age and primary diagnosis interaction also were taken into account as identified in the first article. A P value (2-tailed) of less than .05 was considered to be statistically significant. All analyses were conducted using Statistical Package for Social Sciences version 15.0 (SPSS, Inc, Chicago, Illinois, USA).