Invited Commentary
January 29, 2025
Benzodiazepines Not Main Suspect in Cardiac Surgery Delirium
Amanda M. Kleiman, Michael P. Calgi, John S. McNeil
JAMA Surg. Published online January 29, 2025
doi:10.1001/jamasurg.2024.6603
Delirium after cardiac surgery continues to confound. Hypotension, emboli, hemorrhage, nonpulsatile flow, medications, preexisting vulnerabilities—the list of likely contributors is robust enough that 1 prime suspect is unlikely to emerge. However, with the recent study by Spence et al,1 we can probably cross benzodiazepines off this list. The Benzodiazapine-Free Cardiac Anesthesia (B-Free) trial, a large (approximately 20 000 patients), rigorous, multicenter, cluster crossover trial, randomized entire hospitals to restrictive (essentially no) or liberal benzodiazepine administration for cardiac surgery. The primary outcome, delirium within 72 hours of surgery, had a similar incidence in both groups (14.0% in the restrictive group and 14.9% in the liberal group).
If performed 25 years ago during the golden age of opioid- and benzodiazepine-heavy cardiac anesthetics, we suspect that B-Free would not have failed to reject its null hypothesis. But in our current enhanced recovery era with multimodal anesthesia and regional blocks, the low dose (approximately 4 mg) of midazolam given in the liberal group is likely primarily for anxiolysis preoperatively. The cardiopulmonary bypass (CPB) machine further lowers serum midazolam concentrations.2 Investigators have recently tried to tease out the delirium effect from other enhanced recovery drugs, with one recent study finding that only acetaminophen was protective.3 Dexmedetomidine may be of benefit, although study findings are conflicting.3,4
There is a bit of circumstantial evidence incriminating benzodiazepines in B-Free. Post hoc analyses demonstrated a statistically significant difference in delirium between the 2 groups when participants in the restrictive group were removed from analysis if they had received benzodiazepines up to 24 hours preoperatively. However, this population remained in the liberal group and thus may have confounded results through home use of benzodiazepines. Long-term benzodiazepine use is associated with increased risk of postoperative delirium, especially in older patients.5
The absence of any cases of intraoperative awareness in either group is reassuring. Surgery requiring CPB has historically been considered to have an elevated risk of awareness (frequently quoted as up to 1%), as CPB alters the pharmacokinetics and pharmacodynamics of anesthesia agents. Widespread adoption of processed electroencephalographic monitoring may have contributed to the lack of reported awareness, as it is superior to monitoring clinical signs for awareness prevention.6 Cases of intraoperative awareness may also have been missed, as the study protocol relied on self-reporting and limited postoperative follow-up to 72 hours rather than the 30 days recommended by anesthesia guidelines.
If benzodiazepines are not the culprit, then what should be interrogated next? A recent study of asymptomatic cardiac surgery patients found that central nervous system biomarkers of injury (tau, neurofilament light chain, etc) were elevated postoperatively.7 The accumulation of proteins appears to be central to brain injury, so shining a light on the connections between known delirium risk factors and biomarker levels may yield new evidence. Until then, patients with coronary artery bypass grafting can have a little midazolam in the preoperative period.
