The study by Dr O'Brien et al1 was read with great interest. These authors described their multicenter comparison of the clinical efficacy and safety of ofloxacin vs cefazolin and tobramycin as a double-masked, prospective, randomized clinical trial. However, the protocol actually implemented in this study does not support the claim of random allocation for the investigation of efficacy.
Restriction of the analysis to the subset of 140 patients (56%) who had positive cultures, rather than to the 248 patients who were actually randomized, results in a nonrandomized comparison. The bias introduced by this analysis strategy is well documented.2 Although baseline characteristics of the comparison groups were not significantly different, this does not imply random allocation. The valid randomized comparison in this investigation would be between two treatment strategies and would compare the results of all patients who were randomized to a treatment group, independent of culture status as
Baker RS, Flowers CW, Casey R, Fong DS, Wilson MR. Efficacy of Ofloxacin vs Cefazolin and Tobramycin in the Therapy for Bacterial Keratitis. Arch Ophthalmol. 1996;114(5):632. doi:10.1001/archopht.1996.01100130624031