Vasopressor Selection and Postoperative Delirium in Older Adults: A Propensity-Matched Database Analysis

Introduction

Postoperative delirium is a common neurological complication of anesthesia, characterized by transient alterations in consciousness, cognition, and awareness. It has been associated with adverse outcomes, including increased mortality and 1-year cognitive decline.^1,2 The incidence of postoperative delirium is significantly influenced by intraoperative factors such as cerebral deoxygenation and hypotension, both of which can compromise cerebral perfusion and contribute to acute cognitive dysfunction.^3,4 Vasopressor selection may contribute to this risk. Phenylephrine has previously been associated with increased postoperative delirium compared to ephedrine.^5,6

Although phenylephrine and ephedrine have been compared in this context, norepinephrine is another commonly used agent with unclear effects on postoperative delirium. Comparisons between norepinephrine and ephedrine have primarily focused on outcomes such as nausea and hypotension, as these agents are commonly used in anesthetic practices.^7,8 However, to our knowledge, the rates of postoperative delirium among all three agents have not been compared.

This multi-center, large database retrospective study aims to investigate the effects of norepinephrine compared to ephedrine and phenylephrine, administered on the day of surgery, on the incidence of postoperative delirium. We hypothesize that the incidence of postoperative delirium is similar following the administration of ephedrine vs norepinephrine on day of surgery due to similar mechanisms of action, and both vasopressors will have lower rates of postoperative delirium when compared to phenylephrine.

Methods

Results

Patients who met inclusion criteria for phenylephrine (n = 168,506), ephedrine (n = 107,930), and norepinephrine (n = 6390) were identified and propensity score matched individually. After matching, the analysis consisted of three separately matched cohorts: phenylephrine vs ephedrine (173,250 total patients), phenylephrine vs norepinephrine (12,116 total patients), and ephedrine vs norepinephrine (11,188 total patients). All covariates were adequately balanced after matching (SMD <0.1) except for BMI which remained moderately imbalanced between the phenylephrine vs norepinephrine (SMD = 0.1671) and ephedrine vs norepinephrine (SMD = 0.1276) comparisons.

Discussion

This multicenter retrospective database analysis provides an association between postoperative delirium and vasopressor selection (phenylephrine vs ephedrine vs norepinephrine) in a population of patients undergoing non-cardiac and non-neurosurgical procedures and receiving single vasopressor therapy. Key strengths include the large sample size and multi-center study design. When combined with propensity score matching and large sample size, the observed association is less likely to be influenced by confounding factors.

Our findings demonstrated significantly higher odds of developing postoperative delirium after phenylephrine administration compared with ephedrine. While the results are statistically significant, the incidence of postoperative delirium may not be clinically significant between these two cohorts, given the low absolute risk difference. This study estimates approximately 5 additional cases of postoperative delirium per 10,000 patients and an estimated number needed to harm of 2000. Therefore, the clinical relevance of this association may be insufficient to alter vasopressor selection in isolation. However, these findings support prior research, including a large multi-center retrospective cohort study by Ma et al which reported an odds ratio of 1.35 (95% CI: 1.06 to 1.71; P = 0.02) for phenylephrine compared to ephedrine with a postoperative delirium incidence of 0.9% (685 of 78,900) for patients receiving intraoperative phenylephrine as compared to 0.4% (85 of 24,112) for patients receiving intraoperative ephedrine. ⁶ A randomized controlled trial in 120 patients with total knee arthroplasty found a postoperative delirium incidence of 33% (20 of 60) on the first postoperative day in a group receiving continuous phenylephrine infusion vs 7% (4 of 60) in a group with continuous ephedrine infusion (P < 0.001). This study revealed a decreased rSO₂ in the phenylephrine group (−8.4% ± 0.6%) with a milder decrease in rSO₂ in the ephedrine group (−1.3% ± 0.6%) and a difference between the two cohorts of 7.2% (95% CI: 6.0 to 8.3; P < 0.001). ⁵

A potential explanation includes different mechanisms of action between these drugs. However, these mechanisms were not explicitly tested in this study and should therefore be interpreted as hypothesis-generating rather than definitive. Ephedrine exerts strong β₁ and β₂ effects, leading to increased cardiac output and therefore improved blood flow to the brain. Additionally, ephedrine causes a smaller increase in systemic vascular resistance due to its β₂-mediated vasodilation, which may enhance cerebral perfusion and oxygenation. ⁹ Another potential mechanism involves ephedrine’s central action on VMAT2 reuptake receptors, which induces amphetamine-like central stimulation of dopamine and serotonin. ¹⁰ For example, ephedrine exposure within monkey models was shown to alter neurotransmitter levels in the hippocampus. ¹¹ Administration of ephedrine thus indirectly releases stimulatory neurotransmitters and may improve cognition or awareness, thereby reducing the incidence of delirium. Despite the association between ephedrine and tachyphylaxis, ephedrine has recently been investigated as a slow infusion for delirium treatment, demonstrating a 26% absolute reduction in postoperative delirium rates on the first postoperative day, though no significant changes were observed over seven days. ⁵

When compared to ephedrine, phenylephrine is a pure α₁-adrenergic receptor agonist. ¹² The primary effects of phenylephrine are limited to peripheral vasoconstriction with minor increases in cardiac output or cerebral autoregulation. This potent vasoconstriction may reduce cerebral perfusion pressure and regional oxygen delivery, especially in vulnerable postoperative patients. Unlike ephedrine, phenylephrine lacks β₂ receptor activity, causing greater increases in systemic vascular resistance and further contributing to decreased cerebral perfusion. Additionally, phenylephrine does not cross the blood-brain barrier and therefore lacks the central neuromodulatory effects seen with ephedrine, which may help stabilize neurotransmitter balance and counteract neuroinflammation. ¹² As a result, phenylephrine is unlikely to stimulate dopamine and serotonin the way ephedrine does, potentially leading to reduced alertness.

Interestingly, norepinephrine did not exhibit any differences in postoperative delirium odds compared with phenylephrine or ephedrine. Since norepinephrine exhibits potent α₁-adrenergic receptor activity with modest β1 and minimal β2 receptor activity, ¹³ it is likely to affect cerebral perfusion and postoperative delirium similarly to phenylephrine. However, its cardiostimulatory properties may help maintain cerebral perfusion more effectively than phenylephrine, though still to a lesser extent than ephedrine. Given that norepinephrine is typically administered to more hemodynamically unstable or critically ill patients,^14-16 residual confounding by indication likely remains. Vasopressor selection is not random and often reflects the patient’s underlying physiology, severity of hemodynamic instability, and anesthetic management. Furthermore, norepinephrine is usually reserved for non-transient, continuous vasopressor support, and longer durations of intraoperative hypotension have been associated with increased delirium. ³ Consequently, the observed neutral association may underestimate a potentially protective effect of norepinephrine relative to phenylephrine with respect to postoperative delirium.

Given the association of postoperative delirium with increased mortality and complications rates in older patients, an understanding regarding the impact of vasopressor selection on postoperative cognitive function is essential. ¹⁷ Future studies are warranted to further assess the underlying mechanism by which vasopressors influence postoperative delirium. Prospective studies are also warranted to better assess the relationship between norepinephrine and postoperative delirium as retrospective analyses are likely confounded by underlying selection bias.

As with all retrospective studies, causality cannot be established. Electronic medical record data are subject to variations in billing codes and gaps in documentation which may affect the accuracy of the findings. Vasopressin, angiotensin II, and metaraminol were not considered in this analysis. Secondary outcomes, including downstream clinical consequences such as length of stay, discharge disposition, readmission, postoperative complications, and mortality, were not evaluated. Furthermore, due to the nature of TriNetX, vasopressor exposure was defined by medication administration codes and could not reliably account for dose, duration, timing, or mode of administration, such as bolus vs infusion. The authors also lacked access to the specific intraoperative clinical context surrounding vasopressor use, including hemodynamic instability, ASA physical status, and surgery duration. Vasopressor selection is often driven by intraoperative hemodynamic conditions or clinician preference, which may result in confounding by indication despite matching. For example, patients receiving norepinephrine may have had greater hemodynamic instability, contributing to reduced cerebral perfusion. Despite including somnolence and disorientation in our definition, the true incidence of postoperative delirium may have been underestimated due to under-documentation or transient episodes that resolve before being recorded. In contrast, inclusion of ICD-10 codes for somnolence and disorientation may have reduced the specificity of delirium ascertainment and introduced potential outcome misclassification. Additionally, the broad inclusion of surgical procedures likely captured many operations with a low baseline risk of postoperative delirium, which may have further contributed to the low observed incidence and should be considered when interpreting study power and generalizability. Although extensive cohort matching aimed to decrease confounders, propensity score matching was also limited to the variables available within the database.

Conclusion

In this large cohort of older adults undergoing non-cardiac, non-neurosurgical procedures, rates of postoperative delirium within seven days of surgery were found to be significantly lower with use of ephedrine compared to use of phenylephrine on day of surgery. Norepinephrine was not associated with increased risk of postoperative delirium when compared to either phenylephrine or ephedrine. These findings warrant further pharmacologic studies to individualize anesthetic choice, especially in the high-risk geriatric population.

Supplemental Material