Neuroprotection by anaesthetics in rodent models of traumatic brain injury. A systematic review and network meta-analysis

Background: Anaesthetic neuroprotection in the setting of traumatic brain injury (TBI) remains unproven and is based upon results in the preclinical experiments. Here we sought to synthesize the results in rodent models of TBI and to evaluate the effects of publication bias, experimental manipulation and poor study quality on the effect estimates. Methods: Following a systematic review, we use pairwise meta-analysis to estimate the effect of anaesthetics, opioids, and sedative-hypnotics on neurological outcome and network meta-analysis to compare their relative efficacy. We sought evidence of bias related to selective publication, experimental manipulation, and study quality. Results: Sixteen studies, involving 32 comparisons were included (546 animals). Treatment improved neurological outcomes by 35%, 95% confidence interval (C.I.) 26 to 44%, P<0.001. Statistical heterogeneity was small (12%) but the 95% prediction interval for the estimate was wide (15 to 56%). Statistical power was low: 61% (90% C.I. 22-86%). Small sample size in the studies was a serious shortcoming reducing statistical heterogeneity and obscuring differences in outcome between drugs and between experimental conditions. Conclusions: Anaesthetics do provide neuroprotection in rodent models of TBI. The effect size estimates do not appear to be exaggerated by selective publication, experimental manipulation, or study design. The main shortcoming of the included studies was small sample sizes leading to low power and imprecision, which precluded the network meta-analysis from providing a meaningful ranking for efficacy among the drugs. Reliable preclinical investigations of neuroprotection by anaesthetics will require larger sample.