Effect of Intravenous Acetaminophen vs Placebo Combined With Propofol or Dexmedetomidine on Postoperative Delirium Among Older Patients Following Cardiac Surgery: The DEXACET Randomized Clinical Trial

Key Points

Question  Does postoperative use of scheduled intravenous acetaminophen (paracetamol), combined with intravenous propofol or dexmedetomidine, reduce the risk of postoperative delirium among older patients undergoing cardiac surgery?

Findings  In this randomized clinical trial that included 120 patients undergoing cardiac surgery, scheduled intravenous acetaminophen administered postoperatively for 48 hours, combined with intravenous propofol or dexmedetomidine as a sedative, significantly reduced in-hospital delirium compared with placebo (10% vs 28%, respectively).

Meaning  In older patients undergoing cardiac surgery, postoperative use of scheduled intravenous acetaminophen, combined with intravenous propofol or dexmedetomidine, may be considered to reduce the incidence of postoperative delirium, although additional research is needed.

Importance  Postoperative delirium is common following cardiac surgery and may be affected by choice of analgesic and sedative.

Objective  To evaluate the effect of postoperative intravenous (IV) acetaminophen (paracetamol) vs placebo combined with IV propofol vs dexmedetomidine on postoperative delirium among older patients undergoing cardiac surgery.

Design, Setting, and Participants  Randomized, placebo-controlled, factorial clinical trial among 120 patients aged 60 years or older undergoing on-pump coronary artery bypass graft (CABG) surgery or combined CABG/valve surgeries at a US center. Enrollment was September 2015 to April 2018, with follow-up ending in April 2019.

Interventions  Patients were randomized to 1 of 4 groups receiving postoperative analgesia with IV acetaminophen or placebo every 6 hours for 48 hours and postoperative sedation with dexmedetomidine or propofol starting at chest closure and continued for up to 6 hours (acetaminophen and dexmedetomidine: n = 29; placebo and dexmedetomidine: n = 30; acetaminophen and propofol: n = 31; placebo and propofol: n = 30).

Main Outcomes and Measures  The primary outcome was incidence of postoperative in-hospital delirium by the Confusion Assessment Method. Secondary outcomes included delirium duration, cognitive decline, breakthrough analgesia within the first 48 hours, and ICU and hospital length of stay.

Results  Among 121 patients randomized (median age, 69 years; 19 women [15.8%]), 120 completed the trial. Patients treated with IV acetaminophen had a significant reduction in delirium (10% vs 28% placebo; difference, −18% [95% CI, −32% to −5%]; P = .01; HR, 2.8 [95% CI, 1.1-7.8]). Patients receiving dexmedetomidine vs propofol had no significant difference in delirium (17% vs 21%; difference, −4% [95% CI, −18% to 10%]; P = .54; HR, 0.8 [95% CI, 0.4-1.9]). There were significant differences favoring acetaminophen vs placebo for 3 prespecified secondary outcomes: delirium duration (median, 1 vs 2 days; difference, −1 [95% CI, −2 to 0]), ICU length of stay (median, 29.5 vs 46.7 hours; difference, −16.7 [95% CI, −20.3 to −0.8]), and breakthrough analgesia (median, 10 082.5 vs 12 609.0 µg morphine equivalents; difference, −2530 [95% CI, −5064 to −22]). For dexmedetomidine vs propofol, only breakthrough analgesia was significantly different (median, 10 110.0 vs 12 612.5 µg; difference, −2567 [95% CI, −5094 to −26]; P = .03). Fourteen patients in both the placebo-dexmedetomidine and acetaminophen-propofol groups (46% and 45%) and 7 in the acetaminophen-dexmedetomidine and placebo-propofol groups (24% and 23%) had hypotension.

Conclusions and Relevance  Among older patients undergoing cardiac surgery, postoperative scheduled IV acetaminophen, combined with IV propofol or dexmedetomidine, reduced in-hospital delirium vs placebo. Additional research, including comparison of IV vs oral acetaminophen and other potentially opioid-sparing analgesics, on the incidence of postoperative delirium is warranted.

Trial Registration  ClinicalTrials.gov Identifier: NCT02546765

Delirium, characterized by an acute onset and fluctuating course of inattention and either disorganized thinking or an altered level of consciousness, is reported in 50% of cardiac surgical patients¹ and is associated with increased length of intensive care unit (ICU) stay, morbidity, mortality, and long-term cognitive decline.^2,3Quiz Ref ID Risk factors such as preexisting cognitive dysfunction, postoperative pain, use of opioids and sedatives, and surgical inflammation have been associated with delirium.^2,4 Untreated pain from sternotomy, tube thoracostomy, and leg incision sites can induce delirium,⁵ but opioids may also increase this risk.⁶ Perioperative sedation and opioid-based analgesic regimens are potential risk factors for delirium that may be addressed with appropriate interventions.⁷

Nonsteroidal anti-inflammatory drugs and acetaminophen (paracetamol) are commonly used as adjuvant analgesics. Since nonsteroidal anti-inflammatory drugs are avoided after cardiac surgery because of bleeding and kidney injury risk, acetaminophen may be particularly useful in this setting. Intravenous (IV) acetaminophen administration may be advantageous compared with oral or rectal routes because of variability in bioavailability. In some settings, IV acetaminophen has been shown to reduce inflammation and may confer central analgesic properties⁸ and decrease opioid consumption in both opioid-naive and exposed patients.⁹ Despite these properties, IV acetaminophen has not been studied in the context of delirium prevention. Dexmedetomidine, a selective α₂ receptor agonist, may have opioid-sparing effects¹⁰ with anti-inflammatory properties¹¹; however, studies evaluating dexmedetomidine for prevention of delirium following cardiac surgery have shown mixed results.^12-14

This study aimed to investigate the effect of IV acetaminophen vs placebo combined with IV dexmedetomidine or propofol on the incidence of postoperative delirium. It was hypothesized that postoperative IV acetaminophen for 48 hours as an adjuvant analgesic and dexmedetomidine as a sedative would reduce the incidence of in-hospital delirium following cardiac surgery.

This randomized factorial trial was conducted at Beth Israel Deaconess Medical Center in Boston, Massachusetts, from September 2015 to April 2018. Institutional review board approval was obtained from the Committee on Clinical Investigations and all patients provided written informed consent. The study protocol has been previously published¹⁵ and is available in eAppendix 1 in Supplement 1.

Quiz Ref IDPatients aged 60 years or older undergoing coronary artery bypass graft surgery with or without aortic and/or mitral valve replacement requiring cardiopulmonary bypass were eligible for trial inclusion. Patients with a preoperative left ventricular ejection fraction of less than 30%, preexisting cognitive impairment, Alzheimer disease, Parkinson disease, medications for cognitive decline, history of recent seizures, serum creatinine levels above 2 mg/dL, liver dysfunction, recent history of alcohol misuse, English-language limitations, and hypersensitivity to study medications and patients undergoing emergent surgery were excluded. Participants who did not speak English were excluded because of their inability to complete the cognitive assessments, which have been extensively validated in English.

Patients were allocated using computer-generated block randomization of size 8 with the help of a research pharmacy to either acetaminophen or placebo as an adjunct analgesic and to either dexmedetomidine or propofol as a sedative. This resulted in 4 combinations (1:1:1:1) of analgesics and sedatives, namely, acetaminophen and dexmedetomidine, acetaminophen and propofol, dexmedetomidine and placebo, or propofol and placebo. To achieve blinding to analgesia group among patients, care teams, and research staff assessing outcomes, 1 g of IV acetaminophen and placebo (0.9% saline) were distributed in equal-volume IV bags. Given the distinct appearance of propofol, there was no blinding to sedation group.

Acetaminophen or placebo was administered within 1 hour of ICU admission and thereafter every 6 hours for 8 doses. Both groups received opioids and other analgesics according to standard care practices during these 48 hours; however, administering acetaminophen in any form outside the blinded study medication during this time was not allowed. Propofol or dexmedetomidine started during chest closure was continued for up to 6 hours postoperatively or until extubation, whichever occurred first. A fast-tracking pathway according to standard care practices was used in the ICU to enable early weaning, extubation, spirometry, early mobilization, and subsequent discharge from the ICU (eAppendix 2 in Supplement 1). The dexmedetomidine group received an IV bolus dose of 0.5 to 1 µg/kg during chest closure, followed by a maintenance infusion of 0.1 to 1.4 µg/kg per hour (eAppendix 2). The IV propofol group received a maintenance dose of 20 to 100 µg/kg per minute.

The primary outcome was the incidence of postoperative in-hospital delirium at any time during a patient’s hospital stay. Delirium was measured daily until discharge with the Confusion Assessment Method (CAM) or CAM-ICU for intubated patients. The CAM is a standardized tool used for identification of delirium through a diagnostic algorithm based on 4 cardinal features of delirium, namely acute onset and fluctuating course, inattention, disorganized thinking, and altered level of consciousness. Secondary outcomes included duration of delirium, postoperative cognition at discharge, 48-hour breakthrough analgesic requirements, and ICU and hospital lengths of stay. Additional postoperative analgesic requirements were measured as the amount of opioid (IV morphine or hydromorphone) and oral analgesics required for pain control. This was calculated as the sum of (fentanyl dose × 100) + (hydromorphone dose × 4) + morphine dose + (oxycodone dose × 1.5).¹⁶ For context, a morphine equivalent dose of 90 morphine equivalents a day would correspond to 90 mg of hydrocodone or roughly 60 mg of oxycodone. Hospital length of stay was defined as the number of days admitted in the hospital following surgery and ICU length of stay was defined as the number of hours admitted in the ICU prior to transfer to the general floor. Postoperative cognitive decline was assessed as a secondary outcome using the Montreal Cognitive Assessment (MoCA) (range, 0 [worst] to 30 [best]; minimally clinically important difference [MCID], 2 points)¹⁷ at baseline and discharge. Daily cognitive assessments were performed each day using a standard questionnaire comprising the Delirium Symptom Interview, a standard cognitive assessment consisting of tests of attention, orientation, and memory, and the CAM or CAM-ICU. Nonverbal (intubated) patients were administered the CAM-ICU, which includes both brief cognitive testing and the CAM algorithm to determine the presence or absence of delirium. Assessments were administered by study team members masked to treatment assignment and specifically trained in recognition and assessment of delirium and cognition. Post hoc analyses included delirium severity (CAM severity [CAM-S] score; range, 0 [best/no delirium] to 19 [worst]; MCID, 2 points),¹⁸ median time to ICU discharge among patients with and without delirium, and postoperative pain. All reported pain scores (11-point scale from 0 [no pain] to 10 [worst pain]; MCID, 2 points)¹⁹ were obtained from study participants by nursing staff blinded to the analgesic assignment and were entered into the medical record. The median and highest (worst) of these pain scores were recorded for analysis for each patient on each postoperative day.

The study protocol provides the schedules of enrollment, interventions, and assessments used in the trial.¹⁵ Additional details of the study protocol are documented in eAppendix 1 in Supplement 1. Results of the postdischarge follow-up and biomarker analyses are not included in this article.

Hypotension was predefined as a systolic blood pressure less than 90 mm Hg for 5 minutes or more. Bradycardia was defined as any heart rate less than 40/min. Adverse event data were collected from patients’ medical records daily for a maximum of 10 days. In a post hoc analysis, the difference between groups in the lowest documented blood pressures was assessed.

A sample size of 58 patients was needed for each group, acetaminophen and placebo, to detect a 50% reduction in delirium incidence from a baseline delirium incidence of 50% using 2-sided α = .05 and 80% power. The expected effect size in the primary outcome was based on a pilot study conducted at Beth Israel Deaconess Medical Center.²⁰ Previous studies have shown that reductions in the incidence of delirium exceeding at least one-third of the baseline rate are considered clinically meaningful; therefore, this study would be powered to detect a clinically meaningful difference.²¹ A sample size of 120 patients who received the intervention was targeted to account for potential patient withdrawals. At the time of conception, this study was also interested in the effect of sedation type on the incidence of in-hospital delirium. Hence, the inclusion of this second factor was preplanned. It was therefore included in the factorial design to allow for generation of hypotheses with sedation techniques; however, the trial was powered only to detect differences between those who did and did not receive acetaminophen.

Research Electronic Data Capture (REDCap) was used for data collection and included information on analgesics and sedatives, opioid dosage during the first 48 hours postoperatively, pain scores, patient delirium and cognition assessment data, adverse events, and demographic data. The primary analysis included all patients who were randomized and received at least 1 dose of the study medication. Patients who withdrew after receiving a study medication were still included in the analysis. Daily assessments of delirium were completed until the time of withdrawal. Data were abstracted from the medical record to assess secondary end points in patients who withdrew. Data are presented as means and standard deviations, medians and interquartile ranges (IQRs), or frequencies and proportions depending on variable type and distribution. Changes between groups are further presented as differences and 95% confidence intervals. The Shapiro-Wilk test was used to assess normality. Continuous data were assessed between groups using a parametric t test, Wilcoxon rank sum test, or Kruskal-Wallis test as appropriate. Differences in categorical variables, including the primary outcome (in-hospital delirium), were assessed with a χ² test (or a 2-tailed Fisher exact test when cell counts were small). Missing data were uncommon in the data set (<6%); therefore, no imputation was performed. Interrater reliability assessments of delirium, reported as percent agreement, κ, and weighted κ, were performed both for presence of delirium and for key features including acute change, inattention, disorganized thinking, and altered level of consciousness.

Quiz Ref IDThe primary analysis focused on the relationship between the primary outcome, in-hospital delirium, and acetaminophen use. Given the factorial design of the study, an analysis was performed to determine whether an interaction was observed between groups. In this analysis, a regression model was created with in-hospital delirium as the outcome variable, including an interaction term between the sedative and analgesic groups. Prespecified secondary analyses were performed for outcomes in the factorial design, with 4 groups representing each of the acetaminophen and sedative combinations as well as for the relationship between sedative groups comparing patients who received dexmedetomidine with those who received propofol. No adjustment was performed for multiple testing; therefore, all results for secondary analyses should be interpreted as exploratory.

In post hoc analyses, Kaplan-Meier curves and a Cox proportional hazards model were used to compare time to experiencing delirium after confirming that the proportional hazards assumption was met. To test whether the proportional hazards were met, a time-varying covariate with group was included in the model (P = .56). Given the factorial design, this approach was replicated in models for all 4 factorial groups as well as for the pairwise comparisons (dexmedetomidine vs propofol and acetaminophen vs placebo) to ensure that this assumption was valid. No violations of this assumption were detected; therefore, no time-varying covariates were retained in the final model. Repeated-measures analyses including generalized estimating equations with robust variance were used to assess for differences in the median and worst daily pain scores between groups using mixed-model regression. SAS version 9.4 (SAS Institute Inc) was used for all analyses, with 2-sided P<.05 considered statistically significant. Full details of the statistical analysis plan are available in eAppendix 1 in Supplement 1.

A total of 140 patients provided informed consent and were enrolled, of whom 121 were randomized and a total of 120 received the study intervention and were subsequently analyzed (Figure 1). Patients were excluded from the study if they had an active history of drug or alcohol use disorder (96/268 [36%]), serum creatinine higher than 2 mg/dL (58/268 [22%]), preoperative left ventricular ejection fraction less than 30% (50/268 [19%]), liver dysfunction (28/268 [10%]), lack of English-language proficiency (22/268 [8%]), preexisting cognitive impairment (17/268 [6%]), seizures in the last 3 months (9/268 [3%]), emergent procedure (7/268 [3%]), hypersensitivity to study drugs or procedures (8/268 [3%]), Alzheimer disease (3/268 [1%]), prophylactic medications for cognitive decline (2/268 [0.7%]), or isolated aortic surgery (1/268 [0.4%]). Overall, the majority of enrolled participants were male (101/120 [84%]), were white (111/120 [93%]), and had a median age of 69 (IQR, 63-76) years. Isolated coronary artery bypass grafting was the most frequent cardiac procedure (79/120 [66%]), followed by coronary artery bypass grafting with an aortic valve replacement (32/120 [27%]). The baseline characteristics of the 4 factorial groups were similar (Table 1). Baseline functional status was similar among groups, with the majority of participants living independently and having a high school degree or higher (110/120 and 113/120, respectively) (eTable 1 in Supplement 2).

Two patients in the dexmedetomidine group received propofol and 2 patients in the propofol group received dexmedetomidine in the first 6 hours. Five participants (2 in the placebo group and 3 in the acetaminophen group) missed 1 dose and 1 placebo participant received 1 additional dose of placebo. Four participants randomized to acetaminophen and 1 participant randomized to placebo received additional IV, oral, or rectal acetaminophen during the study period. Despite these results, all analyses were performed using data from all randomized patients who received at least 1 of the 4 study medications.

When conducting the analysis of the factorial groups, no significant interaction was observed between the sedation and analgesic groups (P = .62). Results of the main effects were therefore analyzed and are presented separately. Quiz Ref IDPatients assigned to acetaminophen had a significantly lower incidence of in-hospital delirium, the primary outcome, compared with patients assigned to placebo (10% vs 28%; difference, −18%; 95% CI, −32% to −5%; P = .01) (Table 2). Quiz Ref IDIn contrast, no significant difference was observed in the incidence of delirium (17% vs 21%; difference, −4%; 95% CI, −18% to 10%; hazard ratio, 0.8; 95% CI, 0.4-1.9; P = .54) between those who did and did not receive dexmedetomidine. Agreement was 100% and κ and weighted κ were 1.00 for all CAM variables collected.

Patients who received acetaminophen experienced a shorter duration of delirium compared with placebo (median, 1 vs 2 days; difference, −1 day; 95% CI, −2 to 0 days; P = .03) (Table 2). Postoperative cognition, as assessed via the median discharge MoCA score, was not significantly different between the 2 groups (median, 24.0 vs 24.0 points; difference, 0 points; 95% CI, −1.0 to 2.0 points; P = .29). Intensive care unit length of stay was significantly different in patients who received acetaminophen (29.5 vs 46.2 hours; difference, −16.7 hours; 95% CI, −20.3 to −0.8 hours; P = .02). There was no significant difference in the median hospital stay between those who did and did not receive acetaminophen (median, 8.0 vs 8.5 days; difference, −0.5 days; 95% CI, −2 to 0 days; P = .13). The acetaminophen groups received a lower opioid dose in the first 48 hours postoperatively compared with the placebo groups (median, 10 082.5 vs 12 609.0 µg; difference, −2530 µg; 95% CI, −5064 to −22 µg; P = .03) (eFigure 1 in Supplement 2). Total doses of each sedative administered are shown in eFigure 2 in Supplement 2. Although pain scores were reduced over time in post hoc analyses, no statistically significant difference was observed between those who did and did not receive acetaminophen for either worst (P = .44) or median (P = .21) pain scores (Figure 2). In post hoc analyses, the worst delirium severity (CAM-S) score was not significantly different between those who did and did not receive acetaminophen (median, 9.0 vs 8.0 points; difference, 1.0 point; 95% CI, −2.0 to 3.0 points; P = .81); however, time to delirium onset was significantly different between those in the acetaminophen and placebo groups (hazard ratio, 2.8; 95% CI, 1.1-7.8) (Figure 3). Furthermore, in a post hoc analysis of ICU length of stay, the median time to ICU discharge was significantly higher in patients who experienced delirium as opposed to those who did not (median, 52.4 [IQR, 29.7-123.6] hours vs 29.2 [IQR, 25.7-51.7] hours, respectively; P = .01).

When stratified by sedation group, duration of delirium, compared between those who did and did not receive dexmedetomidine, was not statistically significantly different (median, 1.0 vs 2.0 days; difference, −1 day; 95% CI, −2 to 0 days; P = .31) (Table 2). The median MoCA score at discharge was not statistically different between those who did and did not receive dexmedetomidine (median, 24.0 vs 24.0 points; difference, 0.9 points; 95% CI, −1.0 to 1.0 points; P = .55). Median lengths of stay in both the ICU (median, 32.0 vs 29.7 hours; difference, 2.3 hours; 95% CI, −3.2 to 5.0 hours; P = .66) and the hospital (median, 8.0 vs 8.0 days; difference, 0 days; 95% CI, −1 to 1 day; P = .71) were not significantly different between the dexmedetomidine and propofol groups. Patients who received propofol had larger total morphine equivalent doses given in the first 48 hours postoperatively compared with patients who received dexmedetomidine (median, 10 110.0 vs 12 612.5 µg; difference, −2567 µg; 95% CI, −5094 to −26 µg; P = .03). In post hoc analyses, worst and median pain scores were not statistically different between sedative groups (P = .39 and P = .13, respectively) (Figure 2). In post hoc analyses, the severity of delirium (median CAM-S score, 6.5 vs 9.0 points; difference, −2.5 points; 95% CI, −3.0 to 2.0 points; P = .39) was not statistically different between groups.

In a prespecified exploratory analysis of differences across all 4 groups, there was no statistically significant difference in the incidence of in-hospital delirium (P = .07) (Table 3). Specifically, in a logistic regression model including all 4 groups with placebo-propofol as the reference category, the acetaminophen-propofol group and the placebo-dexmedetomidine group showed no statistically significant difference in the odds of developing in-hospital delirium (odds ratios, 0.35 [95% CI, 0.09-1.28] and 0.85 [95% CI, 0.28-2.61], respectively). However, a statistically significant difference was observed in the acetaminophen-dexmedetomidine group (odds ratio, 0.17; 95% CI, 0.03-0.87). These and other results for secondary end points showed effect sizes consistent with those observed in the primary analysis.

Clinical outcomes and adverse events were not significantly different when stratified by randomization group (eTable 2 in Supplement 2). Fourteen patients each in the placebo-dexmedetomidine and acetaminophen-propofol groups (46% and 45%, respectively) and 7 patients each in the acetaminophen-dexmedetomidine and placebo-propofol groups (24% and 23%, respectively) experienced at least 1 episode of hypotension postoperatively. One patient (3%) in the placebo-dexmedetomidine group and 1 patient (3%) in the acetaminophen-propofol group experienced postoperative bradycardia. In a post hoc analysis, although roughly one-third of patients met the prespecified threshold for hypotension (systolic blood pressure <90 mm Hg for >5 minutes), the median of their lowest documented systolic blood pressure measurements was 83 (IQR, 79-86) mm Hg, which did not differ by group (P = .74).

In this randomized trial, patients who received scheduled IV acetaminophen analgesia in the first 48 hours postoperatively demonstrated a decreased in-hospital incidence of delirium compared with placebo. Additionally, delirium duration, ICU length of stay, and opioid and/or opioid equivalent use in the immediate postoperative period were lower in participants randomized to receive acetaminophen. In sedation analyses, although opioid use was significantly lower in the dexmedetomidine groups compared with the propofol groups, there was no statistically significant difference in delirium or length of stay. To our knowledge, this study is the first to show a significant and meaningful reduction of delirium with use of IV acetaminophen in patients undergoing cardiac surgery.

Mamoun et al²² previously assessed use of IV acetaminophen every 6 hours for 24 hours following cardiac surgery, in which reduced pain intensity was demonstrated without a reduction in opioid consumption. The key differences between the study by Mamoun et al and the present study are intervention and opioid assessment duration (48 hours in this study), use of patient-controlled analgesia by Mamoun et al vs nurse-driven postoperative analgesia in the present study, and inclusion of patients aged 18 years or older by Mamoun et al vs older patients in this study. It is possible that age-related²³ opioid-sparing effects could also partially explain the reduction in opioid consumption seen in this study. Several investigations have shown that IV acetaminophen may have better analgesic efficacy than oral acetaminophen and differing effects on opioid consumption.^23-26 Reduced opioid consumption⁶ may partially explain the mechanism behind the reduction in the incidence of delirium seen in the present study.

The finding of a decreased ICU length of stay in patients receiving IV acetaminophen was similar to the results from a meta-analysis in which critically ill patients with delirium had a 1-day longer mean length of ICU stay than those without delirium.²⁷ Furthermore, increased inspiratory lung capacity²⁴ and better pain relief with patient satisfaction²³ favor a shorter ICU stay, both of which could potentially be achieved more effectively with a successful opioid-sparing technique. These positive benefits need further exploration in larger clinical trials, both in the setting of enhanced recovery after cardiac surgery as well as in the face of the current opioid epidemic.

Although dexmedetomidine reduced opioid use in bariatric surgical patients,^10,28,29 its role in reducing postoperative delirium following cardiac surgery has yielded contradictory results.^12,13,30,31 In this study, although a decrease in opioid consumption in the first 48 hours postoperatively was observed, no statistically significant trend was observed in delirium incidence, severity, or duration. In addition, no interaction was found between dexmedetomidine and acetaminophen in reducing delirium incidence. Further studies in a larger cohort are needed to truly elucidate the effect of dexmedetomidine on aspects of delirium and cognition.

This study has several limitations. First, although powered to detect a difference in postoperative delirium between those who did and did not receive acetaminophen, the sample size was relatively small and the incidence of delirium was lower in this study compared with the pilot study used to perform the power calculation. It is therefore possible that the large effect size observed may not be replicated in larger trials. It is also possible that group interactions could be limited by the sample size, specifically the interaction between IV acetaminophen and dexmedetomidine. Given the small sample size, inferences should be interpreted with caution. Second, anesthetic management was left to the discretion of treating anesthesiologists. This may have affected the outcome given the potential differing effects of anesthetic drugs on the development of postoperative delirium. However, practice variability in perioperative care was minimized with a protocolized cardiac surgical environment and preservation of randomization would be expected to minimize confounding from any anesthetic practice differences. Third, it was not possible to blind the sedative (propofol or dexmedetomidine) to the study team. However, the primary outcome was based on blinded, placebo-controlled exposure to IV acetaminophen. Fourth, it is possible that during daily delirium assessments, some periods of acute-onset inattention, disorganized thinking, or altered level of consciousness may have been missed, potentially leading to misclassification of these outcomes. Despite this, routine testing once per day is an accepted method of diagnosing delirium.³² Fifth, the postoperative analgesic regimen was nursing driven (standard of care) even though patient-controlled analgesia would have been ideal, although randomization and the triple-blind design would have mitigated this. Sixth, the limited sample size from a single center and restrictive inclusion may place some limitations on generalizability. Specifically, non–English-speaking participants were excluded because of their inability to complete the cognitive assessments. The resulting cohort was predominantly white and male. It is possible that the findings observed may not be generalizable to other institutions or to people of other racial and ethnic backgrounds. As this was a single-center study, the results should be replicated in a larger multicenter trial.

Among older patients undergoing cardiac surgery, postoperative scheduled IV acetaminophen, combined with IV propofol or dexmedetomidine, significantly reduced the incidence of in-hospital delirium compared with placebo. Additional research, including comparison of IV vs oral acetaminophen and other potentially opioid-sparing analgesics on the incidence of postoperative delirium, is warranted.

Corresponding Author: Balachundhar Subramaniam, MD, MPH, Beth Israel Deaconess Medical Center, 375 Longwood Ave, W/MS-414, Boston, MA 02215 (bsubrama@bidmc.harvard.edu).

Accepted for Publication: January 16, 2019.

Correction: This article was corrected on July 16, 2019, for incorrect data in the Abstract, Results, Tables, and eFigures due to an incorrect conversion factor.

Author Contributions: Dr Subramaniam had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Subramaniam, Banner-Goodspeed, Gallagher, Talmor, Marcantonio.

Acquisition, analysis, or interpretation of data: Subramaniam, Shankar, Shaefi, Mueller, O’Gara, Banner-Goodspeed, Gasangwa, Patxot, Packiasabapathy, Mathur, Eikermann, Marcantonio.

Drafting of the manuscript: Subramaniam, Shankar, Shaefi, Mueller, O’Gara, Packiasabapathy, Mathur, Eikermann.

Critical revision of the manuscript for important intellectual content: Subramaniam, Shaefi, Mueller, O’Gara, Banner-Goodspeed, Gallagher, Gasangwa, Patxot, Packiasabapathy, Eikermann, Talmor, Marcantonio.

Statistical analysis: Mueller, Eikermann.

Obtained funding: Subramaniam, Banner-Goodspeed, Marcantonio.

Administrative, technical, or material support: Subramaniam, Shankar, Shaefi, Mueller, Banner-Goodspeed, Gasangwa, Patxot, Mathur, Talmor, Marcantonio.

Supervision: Subramaniam, Shaefi, O’Gara, Banner-Goodspeed, Gallagher, Talmor, Marcantonio.

Conflict of Interest Disclosures: Dr Subramaniam, Ms Banner-Goodspeed, and Dr Marcantonio reported receiving grant support from Mallinckrodt Pharmaceuticals for salary support during the conduct of the trial. Drs Subramaniam, Shaefi, Talmor, and Marcantonio reported receiving funds from the National Institutes of Health for unrelated work. Drs Shaefi and O’Gara reported receiving funding from the Foundation for Anesthesia and Education Research. No other disclosures were reported.

Funding/Support: The study was funded by Mallinckrodt Pharmaceuticals: funds were allotted to support time and effort of study personnel, sample acquisition, processing, storage, and regulatory compliance and to assist in paying for the study drugs and placebo.

Role of the Funder/Sponsor: Mallinckrodt Pharmaceuticals had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. The authors had sole authority for the data, analysis, write-up, and submission.

Data Sharing Statement: See Supplement 3.

Additional Contributions: We thank the cardiothoracic surgeons for their support to help complete this study and members of the Center for Anesthesia Research Excellence within the Department of Anesthesia at Beth Israel Deaconess Medical Center (BIDMC), who supported protocol implementation, compliance, and adherence throughout this clinical trial. We also thank the nurses from the cardiovascular ICU and cardiac surgical floors at BIDMC, as well as the BIDMC Society for Thoracic Surgeons Database Outcomes Manager.