Survival analysis for evaluation of corneal ulcer healing times in calves with naturally acquired infectious bovine keratoconjunctivitis

A clinical trial examining the efficacy of 2 drugs for treatment of a natural epizootic of infectious bovine keratoconjunctivitis was performed. The study was conducted in 103 grazing Hereford calves during the summer of 1985. The calves were prospectively and randomly assigned to 1 of 3 groups at the beginning of the study on June 17, and were examined 3 times weekly thereafter until the final observation on August 6. Calves in group 1 (n = 34) were not treated and were used as controls. Calves of group 2 (n = 34) with corneal ulcers were treated with a long-acting oxytetracycline formulation (OTC group). The parenteral treatment was repeated in 72 hours. Affected calves of group 3 (n = 35) were treated topically with furazolidone spray when they developed new corneal ulcers, or when existing lesions worsened during subsequent examination periods (NFZ group). Healing times of the corneal ulcers were reported in 3 ways: the combined times for ulcers present in both eyes of a calf simultaneously (method A), independent times of each ulcer on a calf (method B), and time of the first ulcer for each calf (method C). Censored healing times were examined as left censored (ulcer present at the beginning of the study), right censored (ulcer not healed at the end of the study), or uncensored (true) healing times. The effect that the treatments had on healing times were investigated by use of notched box and whisker plots, life tables, and Cox regression models. The analysis indicated that treatment of calves with either antimicrobial reduced the healing time of corneal ulcers, compared with untreated controls. Calves treated with OTC had shorter periods with ulcers present on both eyes than did NFZ-treated calves. The healing time of the first ulcer on a calf was faster when treated with either antimicrobial than when not treated, but no significant difference between periods for OTC and NFZ treatments was found. Censored healing times were consistently longer than uncensored healing times. Box and whisker plots indicated that both treatments shortened healing times more than those for controls, and OTC shortened healing times more than did NFZ for responses A and B (but not C). Life tables showed that OTC healing times were shorter than those for controls, and NFZ shorter than controls for response B and C (but not A). Cox regression model (for response A) showed a borderline significant difference between times for OTC group and controls, and no significant difference between times for NFZ group and controls.