Safety and Efficacy of Antibacterial Prophylaxis After Craniotomy: A Decision Model Analysis

Antibiotic prophylaxis has revolutionized the safety of neurosurgical procedures in the last century. Today, the clinician's drug of choice before surgery often is based on the antibiotic's resistance profile and drug-induced complications.A decision tree model was developed to compare cefazolin (cephalosporin), vancomycin, or their combination on 90-day mortality postcraniotomy. We modeled the infection type (methicillin-sensitive, methicillin-resistant, or other organisms), antibiotic-related complications that could affect mortality (e.g., renal injury), and Clostridium difficile infections. Parameters' values were extracted from published sources. One-way sensitivity analysis was used to examine results' robustness to plausible variations in input parameter values.The expected value (EV) of 90-day survival was the greatest among patients on cefazolin (EV = 0.9145), followed by patients on vancomycin (EV = 0.8898), and patients on the combination (EV = 0.8886). Cefazolin was the preferred strategy in most one-way sensitivity analyses, except for a few cases in which other options could be preferred based on expected survival. Vancomycin was preferred if kidney injury risk was ≤0.056 conditional on vancomycin intake or ≥12% conditional on cefazolin intake. The combination was preferred if kidney injury conditional risk was ≤0.083 or that for kidney injury-mortality was ≤4.7%. Varying other risks (e.g., postsurgical-site infections (methicillin-resistant Staphylococcus aureus, methicillin-sensitive Staphylococcus aureus, or other); Clostridium difficile infections' risks conditional on each antibiotic) did not change the preferred strategy.According to this decision analysis, patients undergoing a craniotomy who had cefazolin as prophylaxis had a slightly greater expected survival compared with other strategies. These results were sensitive to changes in kidney injury development risk and kidney failure-associated death.