Effect of Linezolid versus Vancomycin on Length of Hospital Stay in Patients with Complicated Skin and Soft Tissue Infections Caused by Known or Suspected Methicillin-Resistant Staphylococci: Results from a Randomized Clinical Trial

Background: Complicated skin and soft tissue infections are common surgical indications usually requiring patients to be hospitalized, and are often caused by gram-positive bacteria, including methicillin-resistant staphylococci such as MRSA. Vancomycin has been the standard treatment for methicillin-resistant staphylococcal infections in many countries, but its intravenous-only formulation for systemic infections often confines patients to the hospital for the treatment. Linezolid, a novel oxazolidinone antibiotic available in intravenous and 100% bioavailable oral forms, was shown in a randomized trial to be as efficacious as vancomycin for suspected or proven methicillin-resistant staphylococcal infections. To determine if oral linezolid can reduce length of hospital stay (LOS) when compared to vancomycin, we compared the LOS for the 230 complicated skin and soft tissue infection patients enrolled in this trial. Materials and Methods: Patients received up to four weeks of linezolid (intravenous followed by optional oral) or vancomycin (intravenous only), followed by up to four weeks of observation. Unadjusted LOS was estimated using Kaplan-Meier survival functions, whereas the log-logistic survival analysis model was used to estimate the multivariate-adjusted LOS controlling for patient demographics and selected baseline clinical variables. Analysis was done on the intent-to-treat ( n = 230) sample as well as on two subsamples of the clinically evaluable ( n = 144) and surgical site infection ( n = 114) patients. Results: The unadjusted Kaplan-Meier median LOS was five days shorter for the linezolid group than the vancomycin group in the intent-to-treat sample (9 vs. 14 days, p = 0.052). It was eight days shorter (8 vs. 16 days, p = 0.0025) in the clinically evaluable sample, but the difference in the surgical site infection sample was not significant (10 vs. 14 days; p = 0.29). The linezolid group's unadjusted mean LOS was 1.7, 5.3 and 0.8 days shorter in the intentto-treat, clinically evaluable, and surgical site infection samples, respectively. After adjusting for age, gender, race, geographic region, bacteremia, type of inpatient location, and number of concurrent medical conditions using the log-logistic model, between-treatment differences in the multivariate-adjusted median LOS decreased to 3, 6, and 3 days, whereas the differences in mean LOS increased to 3.1, 6.5 and 2.5 days for the intent-to-treat, clinically evaluable, and surgical site infe