OBJECTIVES: To estimate differences in adjusted hospital length of stay (LOS) between linezolid and vancomycin using multivariate survival analysis. Linezolid's bioequivalent IV and oral formulations may enable earlier hospital discharge compared to vancomycin treatment. METHODS: 460 hospitalized patients with suspected/confirmed methicillin‐resistant Staphylococcus infections were treated with either linezolid (LZD) or vancomycin (VAN) in a randomized controlled trial. Covariate imbalances between treatment groups were tested using t‐tests and chi‐square tests. Multivariate Cox proportional hazards and several parametric models for LOS were tested for best fit using the Akaike Information Criteria and log‐likelihood ratio statistics. The Cox proportional hazards assumption was rejected (p < .05); log‐logistic survival models fit best. The log‐logistic estimates are used to create two alternative adjusted survivorship functions, one based on individual corrections to LOS (Individual correction), and the other based on means of the predicted survivorship function for each individual (Mean survivorship). Adjusted LOS at quartiles of % discharged for each function was compared with unadjusted LOS using Kaplan‐Meier method. RESULTS: Hospital unit type at randomization and number of comorbidities both differed significantly between groups and significantly affected LOS. With covariate controls in the log‐logistic model, linezolid treatment significantly reduced LOS (p = .04). Adjusted/unadjusted LOS at quartiles of % discharged were:
CONCLUSIONS: When adjusted for covariate differences, median LOS for linezolid patients was at least 2 days shorter than for vancomycin patients. Other differences in the LOS distribution are evident and may be important to decision‐makers but off‐median estimates may be sensitive to the adjustment method used. Methodologic considerations are explored further.
%
discharged
Adjusted LOS (days) Unadjusted LOS
Individual correction Mean survivorship Kaplan‐Meier
LZDVANLZDVANLZDVAN25610696850 (median)14161316141575262828292929
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.