Initial low-dose gentamicin for Staphylococcus aureus bacteremia and endocarditis is nephrotoxic.

BACKGROUND The safety of adding initial low-dose gentamicin to antistaphylococcal penicillins or vancomycin for treatment of suspected Staphylococcus aureus native valve endocarditis is unknown. This study evaluated the association between this practice and nephrotoxicity. METHODS We performed a prospective cohort study of safety data from a randomized, controlled trial of therapy for S. aureus bacteremia and native valve infective endocarditis involving 236 patients from 44 hospitals in 4 countries. Patients either received standard therapy (antistaphylococcal penicillin or vancomycin) plus initial low-dose gentamicin (n=116) or received daptomycin monotherapy (n = 120). We measured renal adverse events and clinically significant decreased creatinine clearance in patients (1) in the original randomized study arms and (2) who received any initial low-dose gentamicin either, as a study medication or <or= days before enrollment. RESULTS Renal adverse events occurred in 8 (7%) of 120 daptomycin recipients, 10 (19%) of 53 vancomycin recipients, and 11 (17%) of 63 antistaphylococcal penicillin recipients. Decreased creatinine clearance occurred in 9 (8%) of 113 of evaluable daptomycin recipients, 10 (22%) of 46 vancomycin recipients, and 16 (25%) of 63 antistaphylococcal penicillin recipients. An additional 21 patients received initial low-dose gentamicin <or=2 days before study enrollment. A total of 22% of patients who received initial low-dose gentamicin versus 8% of patients who did not receive initial low-dose gentamicin experienced decreased creatinine clearance (P = 005 ). Independent predictors of a clinically significant decrease in creatinine clearance were age >or=65 years and receipt of any initial low-dose gentamicin. CONCLUSIONS Initial low-dose gentamicin as part of therapy for S. aureus bacteremia and native valve infective endocarditis is nephrotoxic and should not be used routinely, given the minimal existing data supporting its benefit.