Intensive vs. conventional insulin therapy in neurosurgical patients: a meta-analysis of postoperative infections and hypoglycemic risks.

Background Hyperglycemia significantly increases postoperative infection risk in neurosurgical patients. While intensive insulin therapy (IIT) has shown promise in general critical care populations, its specific benefits and risks in neurosurgical patients remain debated. Objective This meta-analysis evaluated the impact of intensive versus conventional insulin therapy on postoperative infection rates and hypoglycemic events in neurosurgical patients, examining diabetes as a potential effect modifier. Methods A systematic review was conducted following PRISMA guidelines, searching PubMed, Google Scholar, and Scopus databases through October 2024. We included randomized controlled trials and prospective observational studies comparing IIT versus conventional insulin therapy (CIT) in neurosurgical patients. Primary outcomes were postoperative infection rates; secondary outcomes included hypoglycemic events. Data were analyzed using random- and fixed-effects models, with subgroup analyses by publication year and diabetes prevalence. Results Seven studies encompassing 1,146 neurosurgical patients (578 IIT, 568 CIT) were included. Overall, 458 patients (40.0%) developed postoperative infections. The random-effects model yielded non-significant results (RR: 0.782, 95% CI: 0.610-1.003, P = 0.053), though the fixed-effects model demonstrated significant infection reduction with IIT (RR: 0.746, 95% CI: 0.649-0.859, P < 0.001). Studies published before 2010 showed significant infection reduction (RR: 0.66, 95% CI: 0.64-0.69, I² = 0%), whereas post-2010 studies demonstrated no significant benefit (RR: 0.96, 95% CI: 0.40-2.29, I² = 63.7%). Hypoglycemic events were more frequent with IIT. Diabetes prevalence negatively correlated with IIT efficacy (B = 0.13, p = 0.000). Conclusions While IIT shows potential benefits in reducing postoperative infections in neurosurgical patients, this effect was not statistically significant in the random-effects model and increases hypoglycemic risk. Efficacy appears diminished in diabetic patients and has declined over time. These findings highlight the need for individualized glycemic management strategies balancing infection prevention with hypoglycemic risks.