Association of propofol vs. sevoflurane maintenance anesthesia with postoperative delirium, cognitive decline, and mortality in older adults: a global comparative-effectiveness study.

BACKGROUND

Postoperative delirium and longer-term neurocognitive disorders are major complications of surgery in older adults. As surgical volumes rise in ageing populations, identifying modifiable intraoperative factors—particularly anesthetic choice—has become increasingly important. Propofol is often favoured for its recovery profile, whereas sevoflurane offers organ-protective preconditioning that may benefit vulnerable patients. Comparative evidence in high-risk real-world populations remains limited.

METHODS

We performed a multicentre retrospective comparative-effectiveness study using a global federated health network. Adults aged 60–80 year undergoing surgery with general anesthesia were included if they received propofol or sevoflurane for maintenance. Propensity-score matching (1:1) and doubly robust Cox models estimated adjusted hazard ratios (aHRs). The primary outcome was postoperative delirium; secondary outcomes were 30-day mortality, long-term postoperative cognitive dysfunction (180–365 days), and long-term mortality. Sensitivity analyses incorporated antipsychotic-based delirium ascertainment and time-stratified hazards.

RESULTS

After matching, 91,046 patients were included (propofol 45,545; sevoflurane 45,545). Propofol was associated with higher risks of postoperative delirium (3.36% vs. 2.70%; aHR 1.37, 95% CI 1.25–1.48) and 30-day mortality (0.66% vs. 0.53%; aHR 1.25, 95% CI 1.05–1.48), with the greatest excess on postoperative day 1 (aHR 2.58). Long-term cognitive dysfunction (aHR 1.16, 95% CI 1.09–1.24) and long-term mortality (aHR 1.23, 95% CI 1.17–1.29) were also increased.

CONCLUSIONS

In this large real-world cohort, propofol maintenance anesthesia was associated with higher risks of postoperative delirium, long-term cognitive decline, and mortality compared with sevoflurane, particularly in physiologically vulnerable patients. These findings generate hypotheses regarding anesthetic choice in high-risk older adults and support the need for adequately powered randomized trials. Limitations include potential outcome under-ascertainment, lack of intraoperative physiologic data, and residual confounding inherent to electronic health records.