Delirium, characterized as an acute decline in attention and global cognitive dysfunction, occurs in up to 50% of patients following cardiac surgery (Rudolph and Marcantonio, 2011). It is associated with increased morbidity and mortality, prolonged length of hospital stay, postoperative cognitive dysfunction [POCD], and decline in functional status (Rudolph and Marcantonio, 2011). POCD, which is defined as a clinically significant decline in cognitive function following recovery from surgery, is a serious concern for those undergoing any form of surgery but is especially prevalent among cardiac surgery patients (Hovens et al., 2016). While the prevalence of POCD in those undergoing non-cardiac surgery is approximately 12%, the reported incidence of POCD among those undergoing cardiac surgery ranges from 20–50% at 3 months postoperatively (Hovens et al., 2016; Evered and Silbert, 2018). Recent studies have highlighted the increased risk for an accelerated course of cognitive decline following delirium at 12 and 36 months postoperatively (Saczynski et al., 2012; Inouye et al., 2016).

While the mechanisms for both delirium and POCD following cardiac surgery are likely multifactorial, well-documented risk factors include baseline cognitive impairment, psychoactive medications, pain, and inflammation related to surgery (Brown, 2014; Inouye et al., 2014; Indja et al., 2017). Although the pathophysiological mechanism of delirium is still largely unknown, it is believed that cholinergic deficiency is the final pathway. In fact, a heightened systemic inflammatory response to surgery or postoperative infection may lead to increased permeability of the blood-brain barrier with resultant neuro-inflammation, triggering both delirium and POCD. This direct injury to the neurons can lead to long term downstream adverse effects (Marcantonio, 2012). Acetaminophen has analgesic, antipyretic, antioxidant and anti-inflammatory action. Neuronal cell survival is improved in oxidatively challenged cell cultures exposed to acetaminophen (Tripathy and Grammas, 2009). Current literature comparing the formulations of acetaminophen show that IV acetaminophen offers higher CSF concentration, greater area under the curve (AUC), shorter time to reach maximum plasma concentration, improved analgesic coverage and decreased requirement for narcotic agents when used as a part of multimodal analgesia in the postoperative period (Singla et al., 2012; Politi et al., 2017). Because of its analgesic properties, as well as possible protection against neuro-inflammation, intravenous (IV) acetaminophen was considered as a potential protective factor against postoperative delirium, as well as associated longer-term sequelae (Subramaniam et al., 2019).

In the recent DEXACET trial, we demonstrated a significant reduction in delirium in older cardiac surgery patients treated with IV acetaminophen, combined with IV propofol or dexmedetomidine, compared with placebo (Subramaniam et al., 2019). We did not find differences in the rates of delirium in the IV propofol vs. dexmedetomidine arms, therefore the current analysis focuses only on the effects of IV acetaminophen. Using 12-month follow up data from that trial, this analysis investigates the relationship between IV acetaminophen and longer-term functioning following cardiac surgery. Given the significant reduction in delirium observed in the primary analysis, we hypothesized that in-hospital use of IV acetaminophen six-hourly for eight doses after cardiac surgery is associated with an improved cognitive functioning and functional status at 1 month and 12 months following surgery as compared to placebo.

The DEXACET trial was a prospective, randomized, placebo-controlled, triple-blinded, factorial design trial, carried out from September 2015 through April 2018, with long-term follow up completed in April 2019. The purpose of the study was to assess the impact of postoperative scheduled acetaminophen and sedatives (propofol or dexmedetomidine) to decrease the incidence of postoperative delirium in elderly cardiac surgical patients. Long term follow up was an apriori aim for this study as well. This study was approved by the Committee on Clinical Investigations Institutional Review Board (Protocol 2014P000413) at Beth Israel Deaconess Medical Center (BIDMC) and was registered at clinicaltrials.gov (Protocol NCT02546765).

Full details of the study protocol and results were published previously (Shankar et al., 2018; Subramaniam et al., 2019). In brief, 120 patients age 60 years and older undergoing either coronary artery bypass graft (CABG) or CABG plus valve replacement surgery were included. Patients were excluded if they were undergoing either an emergent procedure or isolated aortic surgery, had preoperative ejection fraction less than 30%, preexisting cognitive impairment, Alzheimer’s disease, Parkinson’s 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, or hypersensitivity to any study medication. After confirming subject eligibility, written informed consent was obtained by a study physician. Using computer generated block randomization, patients were assigned into groups (1:1:1:1) to receive either acetaminophen and propofol, acetaminophen and dexmedetomidine, placebo and propofol, or placebo and dexmedetomidine. The dexmedetomidine group received an IV bolus dose of 0.5–1 μg/kg during chest closure, followed by a maintenance infusion of 0.1–1.4 μg/kg per hour. The IV propofol group received a maintenance dose of 20–100 μg/kg per minute. Eight Doses of acetaminophen (1 g IV) or placebo (0.9% normal saline) were administered upon immediate post-operative admission to the intensive care unit (ICU) and, repeated every six-hourly for the first 48 h after surgery. Blinding could only be achieved between the analgesic treatment arms (acetaminophen vs. placebo), not the sedation arms (propofol and dexmedetomidine).

Preoperative baseline and discharge cognitive assessments were performed in all subjects using the Montreal Cognitive Assessment (MoCA) (Nasreddine et al., 2005). MoCA consists of a 30-point questionnaire used to test the various domains of cognitive function. A briefer cognitive assessment was administered daily in the hospital to assess for delirium. This assessment included questions evaluating language, attention, concentration, working memory, short term memory, visuo-spatial abilities, abstract reasoning and orientation to time place and person (Nasreddine et al., 2005). Additionally, the Delirium Symptom Interview (DSI) was used to elicit specific symptoms of delirium, such as perceptual disturbances and delusions (Albert et al., 1992). These daily assessments in the hospital were used to score the Confusion Assessment Method (CAM), a standardized approach to identify delirium, with high sensitivity (94–100%) and high specificity (90–95%) (Inouye et al., 1990; Inouye et al., 2016). All study assessors were blinded to treatment group.

Following discharge, patients received follow-up assessments to measure cognition, mood, and functional status, predefined secondary outcomes, at 1 month and 12 months following surgery. Cognitive function was assessed using the Telephone Montreal Cognitive Assessment (T-MoCA), a 22-point validated telephone-based cognitive assessment tool that can be used remotely to follow-up patients after discharge. The T-MoCA contains a subset of MoCA items (22 out of 30) that can be administered by phone, facilitating comparison with baseline MoCA scores, from which the T-MoCA items were used for comparison. Sustained attention tasks, days of week backwards (DOWB) and months of year backwards (MOYB), were added to the T-MOCA to increase sensitivity for detection of inattention, a feature of delirium (Zietemann et al., 2017). Patients were also assessed with the short-form Geriatric Depression Scale (GDS), scored 0-15 (15 = worst); scores ≥5 indicated depressive symptomatology (Yesavage and Sheikh, 1986; Albert et al., 1992).

Functional status was evaluated during the same interview using the activities of daily living (ADL) and instrumental activities daily living (IADL) scales. The ADL scale consists of a set of questions assessing the patient’s requirement for assistance to carry out six basic daily activities: feeding, walking, toileting, bathing, dressing and grooming (Katz et al., 1970). the ADL scale was scored 0-6 (6 = independent), with a score <6 indicating any ADL impairment. IADL, an instrument that assesses higher order activities needed to maintain independent living in the community, cooking, housework, finances, medications, transportation, and telephone, was administered using a separate scale (Marcantonio, 2012); the IADL scale was scored 0-6 (6 = independent), with a score <6 indicating any IADL impairment. Declines in cognitive and functional status from preoperative baseline to follow-up time points were evaluated since they may reflect important changes in quality of life for the participants.

Following the standard practice in many epidemiologic studies to limit losses to follow up, the protocol permitted a time window for completion of the follow-up interviews of 14 days before and after the 1-month scheduled time-point, as well as 3 months before and after the 12-month scheduled time-point.

In this analysis of the 1-month and 12-month outcomes of our clinical trial, no significant differences between acetaminophen and placebo groups were observed in cognitive or physical functioning at either time point for T-MoCA, GDS scores, ADLs, or IADLs. While we observed significant improvements between median T-MoCA scores at baseline as compared to 12 months follow-up, this could largely be attributable to the learning effects commonly associated with repeated cognitive testing (Heilbronner et al., 2010). For all measures, except GDS we observed significant time effects at 1 and 12 months relative to baseline, but there were no significant treatment effects, or time × treatment interaction effects, for cognitive, mood or functional outcomes. Although in our initial study we observed a significant association between treatment with acetaminophen and a lower rate of in-hospital delirium following cardiac surgery, this did not translate into improved cognitive or functional outcomes at 1-month or 12-months follow-up (Marcantonio, 2012).

Previous studies have documented increased rates of depression following cardiac surgery (Pourafkari et al., 2016). Importantly, both baseline and postoperative depression are strongly associated with adverse long-term outcomes and mortality following cardiac surgery (Takagi et al., 2017; Drudi et al., 2018). While we did not find any significant trends within our own cohort, it is important to continue to monitor postoperative depression closely in similar studies, particularly among the elderly.

Experimental animal models have shown improved cognition when administered acetaminophen after orthopedic surgery or post LPS–induced cognitive impairment potentially due to its central modulatory effects (Zhao et al., 2017; Garrone et al., 2021). This was demonstrated by alteration in hippocampal cytokines levels and markers of microtube dynamics. The literature for long term impact of acetaminophen, on cognition, in both experimental animals and humans, remains limited. Our study did not demonstrate a significant difference in cognitive function between the groups at 1 month and 12 months. This may in part be because of the losses to follow up that were incurred at 1 month and 12 months. Our findings regarding cognitive outcomes are similar to a prior small cohort study by Neufeld et al. examining the long-term effects of delirium following surgery with general anesthesia (Neufeld et al., 2015). In that study, no significant differences between those with and without postoperative delirium were found in cognitive ability or functional status at 18 months postoperatively (Neufeld et al., 2015). While 45% of their study population (41 people) had diagnosed delirium in the post anesthesia care unit and 96% of survivors followed-up, they did not find any significant long-term differences (Neufeld et al., 2015). However, a number of other studies have demonstrated a significant decline in cognitive ability up to 3 months following postoperative delirium, which resolved by 12 months postoperatively (Brown et al., 2018; Austin et al., 2019). We did not perform any assessments between the 1 month and 12 months period, and thus, we are unable to comment on whether a nadir occurred for these patients between our scheduled follow up assessments.

While significant declines in cognitive ability and functional ability between one and 3 months following postoperative delirium have been frequently described (Brown et al., 2018; Austin et al., 2019), only a few studies have shown persistent long-term cognitive dysfunction (Saczynski et al., 2012; Inouye et al., 2016). Saczynski et al. demonstrated an association between postoperative delirium and significantly lower postoperative cognition at both 1 month and 12 months following surgery (Saczynski et al., 2012). Inouye et al. demonstrated a significant decline in cognitive ability 36 months following postoperative delirium in a large cohort of 560 elective surgery patients without dementia at baseline (Inouye et al., 2016). Maclullich et al. reviewed the literature from 2003 to 2008 for long term cognitive decline in patients with acute in-hospital delirium. They concluded that stress responses, chronic neurodegeneration triggered by inflammation and infarct in critical brain areas can precipitate delirium acutely and long-term changes in cognition (MacLullich et al., 2009). Most of the studies reviewed diagnosed delirium using CAM and cognition using the mini mental state examination (MMSE). Considering that MoCA has higher sensitivity for diagnosing mild cognitive impairment and NCD in elderly patients, we chose to use the MoCA scoring in our study (Ciesielska et al., 2016; Siqueira et al., 2019).. Similar to our study, postoperative long term cognition using T-MoCA in non-emergent surgery was studied by Austin et al., concluding no difference between delirious and non-delirious patients at 90 days (Austin et al., 2019). In addition, while delirium is described as an independent risk factor for decline in functional ability, a significant decline has typically only been demonstrated at 1 month, but not at 12 months (Rudolph et al., 2010). Furthermore, these studies had larger samples and higher long-term retention rates than our study.

Our study has many important strengths. We conducted a rigorous randomized controlled trial with a factorial design and used state of the art delirium assessments to measure the primary outcome. With little missing data (<6% during the hospital stay) we were able to robustly quantify the effect of IV acetaminophen vs placebo combined with propofol or dexmedetomidine on postoperative delirium among older patients following cardiac surgery. We demonstrated a strong positive effect for prevention of delirium in the original study. For this follow-up study, we used a standardized measure (T-MoCA) to track cognitive and functional outcomes over 1-year by trained assessors, who were carefully blinded to the treatment group assignment.

However, several limitations are worthy of comment. First, the sample size is relatively small and was not explicitly powered to examine longer term cognitive or functional decline. The relatively small sample size and lack of power leads to an increased risk of a Type II error (Button et al., 2013). Moreover, a 1-year timeframe may have been too short to detect a significant long-term trajectory of cognitive decline. Additionally, there were losses to follow-up raising the possibility of informative censoring bias, such that patients with worse status failed to participate in the follow-up interviews, which may have affected our ability to elucidate the true relationship. It is also possible that additional measurements at three and/or 6 months would have enhanced retention of patients in the study and may have enabled us to better characterize cognitive changes over time. It is important to note that this was a single-site study, and the population enrolled were white and predominantly males, potentially limiting the generalizability. Finally, the interview used to collect data on the IADL and ADL status of patients during follow-up was adapted and did not include the multiple response categories typically used in the Lawton-Brody IADL test, which may have limited our ability to detect more subtle changes over time.

In the present analyses no statistically, significant improvement was demonstrated in the cognitive, depressive or functional outcomes at one and 12 months with the use of acetaminophen when compared to placebo. Given the limited sample size and participant retention in the study, further investigation is warranted. It will be important to further investigate both the long-term relationship between delirium with cognitive, emotional, and physical function, and also the long-term cognitive effects of acetaminophen in combination with dexmedetomidine or propofol following cardiac surgery. An interesting area of future research is to understand the molecular target of acetaminophen for delirium prevention. This would potentially provide a better understanding of acetaminophen’s ability to impact cognition and function over the long term. It would also be interesting to investigate other pharmacological and non-pharmacological interventions in reducing long term functional decline after delirium. Long-term outcome studies in vulnerable populations are challenging, and our study highlights the need for robust efforts to meet these challenges.