Abstract
Background
We switched our antibiotic prophylaxis for elective hip and knee surgery from cefuroxime to flucloxacillin with single dose gentamicin in order to reduce the incidence of C. Diff diarrhoea. More patients subsequently appeared to develop acute kidney injury (AKI).
Methods
During a twelve month period we examined the incidence of AKI sequentially in 198 patients undergoing elective hip or knee surgery: cefuroxime (n = 48); high dose flucloxacillin (median 8g) (n = 52); low dose flucloxacillin (median 4g) (n = 46); and cefuroxime again (n = 52).
Results
There were no statistically significant differences between the four groups by chi-square tests for age, gender, nature of operation (hip or knee surgery), American Society of Anaesthesia (ASA) grade, mode of anaesthesia (spinal ± general anaesthetic v GA), baseline serum creatinine, pre-operative co-morbidity (hypertension, diabetes), pre-operative medication (NSAIDs, ACEI/ARBs or betablockers) and post-operative hypotension. Patients receiving high dose flucloxacillin required more vasopressors during surgery (p = 0.02 by Kruskal-Wallis test). The proportion of patients in each antibiotic group with any form of AKI by RIFLE criteria was: first cefuroxime group (8%), high dose flucloxacillin (52%), low dose flucloxacillin (22%), second cefuroxime (14%) (p < 0.0001). Odds ratios (OR) for AKI derived from a multivariate logistic regression model and arbitrarily assigning an OR of 1 to first cefuroxime group, were: high dose flucloxacillin 14.5 (95% CI, 4.2–49.7); low dose flucloxacillin 3.0(0.8–10.8); cefuroxime again 2.0(0.5–7.7). Three patients required temporary haemodialysis. Biopsies in two of these showed acute tubulo-interstitial nephritis. All three patients belonged to the high dose flucloxacillin group. None of the patients developed C Diff diarrhoea.
Summary
We have shown a strong association between high dose flucloxacillin with single dose gentamicin prophylaxis and subsequent development of AKI which was not confounded by any of the co-variates we measured
