Ketamine and melatonin for the prevention of postoperative delirium in patients with colorectal cancer: a feasibility study

Abstract

Purpose

This study aimed to examine the feasibility and effects of preoperative 5 mg of melatonin and intraoperative 50 mg of ketamine on postoperative delirium (POD) prevention in candidates for colorectal cancer surgery.

Methods

In this randomized controlled trial, adults (> 18 years) who were candidates for elective colorectal cancer surgery were included in the study. Patients were randomized into four groups: placebo/saline (PS), melatonin/saline (MS), placebo/ketamine (PK), and melatonin/ketamine (MK). The groups received either 5 mg of oral melatonin or a placebo the night before surgery and 50 mg of ketamine or normal saline after anesthesia induction. The occurrence and severity of POD and pain severity were assessed via the confusion assessment method for the intensive care unit (CAM-ICU) and visual analogue scale (VAS), respectively (twice daily), until postoperative day 4.

Results

One-hundred and four patients (51% male, mean age: 56.29 ± 12.65) with a rate of 4.7 patients per week were recruited, with an attrition rate of 13.3%. The prevalence of POD was 17.3%, 22.23%, 16.67%, and 16.67% in the PS group, the MS group, the MK group, and the PK group, respectively. Compared with the control, none of the interventions significantly reduced the likelihood of POD occurrence.

Conclusion

This randomized controlled trial demonstrated the feasibility of recruiting and retaining surgical patients for a multi-arm perioperative intervention study. Although the interventions did not significantly reduce the incidence of POD, the study design and procedures were feasible, with acceptable recruitment and attrition rates. Compared to placebo, none of the interventions significantly reduced the incidence of POD; however, time was a significant factor, with POD incidence, severity, and pain decreasing longitudinally.

Trial regsitration

IR.TUMS.IKHC.REC.1401.374, registration date: 14 February 2023 and IRCT code: IRCT20120527009886N2, registration date: 07/03/2023.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13741-025-00571-3.

Introduction

Delirium is a state of mind that presents with disturbances in the patient’s wakefulness, awareness, orientation, thought process, and attention, with a fluctuating pattern over a short period (Diagnostic and Statistical Manual of Mental Disorders : DSM-5.5th ed.Arlington, VA :American Psychiatric Association, 2013). Postoperative delirium (POD) occurs in the acute phase, mostly following complex (major) surgeries, e.g., cardiac and abdominal operations, with prevalences of 32% and 30%, respectively (Igwe et al. 2023; Varpaei et al. 2024a). Colorectal cancer is the third most commonly diagnosed malignancy, with the second-highest attributed mortality rates globally, and is often accompanied by POD in the elderly (Baidoun et al. 2021; Bray et al. 2024). Moreover, POD increases hospitalization duration, costs, postoperative cognitive dysfunction, mortality, and morbidity (Boone et al. 2020).

The risk factors for POD are divided into modifiable (e.g., surgery duration, postoperative pain, intraoperative hypotension, and blood loss) and non-modifiable (e.g., age, cognitive function, and comorbidities) categories (Lai et al. 2023; Varpaei et al. 2024b). As pain and neuroinflammation are modifiable risk factors for POD, ketamine, a noncompetitive N-methyl-D-aspartate receptor antagonist and a phencyclidine-derived analgesic with neuroprotective, pain-reducing, and opioid-sparing characteristics, has been widely researched in the field of POD prevention, with conflicting results (Xiao et al. 2023). With respect to spinal surgery, Elsamadicy et al. reported that intraoperative ketamine administration increases the risk of POD, whereas Plyler et al. reported the opposite findings (Plyler et al. 2019; Elsamadicy et al. 2019). Similar contradictory results were reported for cardiopulmonary surgeries (Siripoonyothai and Sindhvananda 2021; Hudetz et al. 2009). In a meta-analysis, Fellous et al. reported no significant preventive effect of intraoperative ketamine against POD (Fellous et al. 2023). This study was later challenged by a trial sequential analysis showing insufficient power, thus requiring further randomized clinical trials (RCTs) for a decisive result (Hung et al. 2024).

Melatonin, a serotonin-derived neurohormone, has been studied for POD prevention in surgery candidates because of its sleep‒wake cycle regulatory effects (Jiang et al. 2023). In a systematic review and meta-analysis, melatonin and ramelteon, a melatonin agonist, were found to reduce POD incidence (Barnes et al. 2023). A recent multicenter feasibility study by Khaled et al. assessed the use of 3 mg of melatonin, administered once preoperatively and for 7 days postoperatively, in 85 patients. They observed no significant difference in POD incidence and postoperative cognitive function while concluding that for a definitive trial assessing the effects of melatonin on POD, a sample size of at least 1000 patients is needed (Khaled et al. 2025a). However, further studies are still required to determine the optimum regimen in terms of effectiveness and feasibility (Arbabi et al. 2018).

As mentioned by Scicutella et al., the etiology of delirium is not fully understood, and any pharmacological intervention to prevent delirium requires a multicomponent approach to prevent neuroinflammation, sleep‒wake cycle disruption, pain management, and perioperative risk factor management (Scicutella 2020). This study aimed to assess the feasibility of an RCT with a preoperative intervention and an intraoperative intervention to examine the effects of the administration of melatonin and ketamine on POD prevention in candidates for colorectal cancer surgery. By proposing this combination, we aim to achieve the sleep cycle-adjusting effects of melatonin in conjunction with pain reduction and the anti-inflammatory effects of ketamine, as evidenced through the inhibition of the postoperative IL-6 inflammatory response (Dale et al. 2012). This study aims to assess the administration of 5 mg of oral melatonin the night of surgery, administered as a single dose, as it is more feasible and applicable than long-term perioperative administration, combined with 50 mg of intraoperative ketamine, compared to placebo.

Methods

Study design

After approval by the Institutional Review Board (IRB, IRCT Code: IRCT20120527009886N2), the study started as a parallel, double-blind, placebo-controlled, single-center, feasibility randomized clinical trial conducted at the IKHC, a tertiary teaching referral hospital in Tehran, Iran. This study was conducted in compliance with the Declaration of Helsinki, ensuring ethical principles and participant protection. No prior publication of data in this manuscript was performed.

Patients

The inclusion criteria included patients who were older than 18 years, were diagnosed with colorectal malignancies, and had an expected surgery duration of more than 60 min. The exclusion criteria included a diagnosis of delirium before surgery, emergent or multiple surgeries, severe kidney or liver disease, previous reaction to ketamine or melatonin, history of myocardial ischemia, history of high intracranial pressure, and inability to complete preoperative cognitive and delirium assessments owing to language, hearing, or visual barriers.

Recruitment, randomization, and blinding

After screening for inclusion criteria using the patient’s medical record, written informed consent was obtained from patients the night before surgery. Baseline preoperative data—including demographics (age, sex, occupation, marital status, weight, and height), laboratory values, and cognitive function (assessed using the Montreal Cognitive Assessment [MoCA])—were then collected. Patients were randomized using computer-generated random numbers at an allocation ratio of 1:1:1:1, which was performed by an independent biostatistician who was not involved in data collection. Eligible participants were enrolled sequentially, and their group assignment was performed the night before the surgery. To ensure double-blinding, melatonin and the oral placebo were manufactured with identical appearances, and ketamine and normal saline were packaged in similar packages. Patients received their assigned drugs with numbers corresponding to their randomization codes.

Intervention and routine anesthetic care

The night before surgery, a registered nurse (not involved in the study design) administered an oral tablet containing either 5 mg of melatonin or a placebo to the patient. During surgery, after anesthetic induction, a vial containing 50 mg of ketamine or normal saline was handed to the anesthetist, thus shaping the four arms of the study (placebo/saline [PS], melatonin/saline [MS], placebo/ketamine [PK], and melatonin/ketamine [MK]). All the research team members, care providers (except the anesthetist), outcome assessor, and the patients were blinded to the interventions.

Anesthesia induction was performed with 0.03 mg/kg midazolam and 2–3 µg/kg fentanyl as premedication; then, 2 mg/kg propofol and 0.5 mg/kg atracurium were administered for induction. Maintenance was achieved via isoflurane, and the dose was adjusted on the basis of the patient’s hemodynamic conditions and depth of anesthesia. BIS electrodes were attached to the forehead, and the depth of anesthesia was maintained between 40 and 60. Every 30 min, 10 mg of atracurium and 50 mg of fentanyl were injected. At the conclusion of the surgery, muscle relaxant reversal was provided with neostigmine and atropine. Ondansetron (4 mg, as an antiemetic) was administered to prevent postoperative nausea and vomiting.

Outcomes

As the primary outcome, feasibility was assessed with regard to the following:

1. Recruitment rate (number of patients recruited per week) and attrition rate (number of patients who were lost to follow-up after allocation)

2. Assessment completion rates: Percentage of assessments (preoperative, intraoperative, and postoperative) that were completed for each patient.

3. Anesthetic drug variability: Given the feasibility constraints and the non-exploratory design of the study, and to adhere to ethical guidelines, the anesthetic management of patients was not restricted to a fixed protocol. Instead, the intraoperative medications administered were recorded and analyzed to account for potential variability.

4. Delirium incidence rates: POD assessments were done twice daily from the night after surgery until discharge. To ensure POD was not mistaken for emergence agitation, assessments were done at least 6 h after the end of the surgery. To evaluate the number of assessments needed to reach at least 80% POD detection, we examined different assessment modalities (once daily vs. twice daily, night-time assessment vs. day-time assessment).

The secondary outcomes included POD incidence, POD severity, and postoperative pain severity. The occurrence of POD was assessed via the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). The CAM-ICU is a reliable and validated tool for delirium assessment (Ely et al. 2001). Patients’ alertness was assessed primarily via the Richmond Agitation and Sedation Scale (RASS); if a score higher than − 3 was attained, delirium was then assessed via the CAM-ICU. To assess POD severity, we used the CAM-ICU-7, a valid and reliable tool with an 8-point rating scale (0 = no delirium, 7 = highest severity) (Khan et al. 2017). Postoperative pain was assessed via the visual analogue scale (VAS). Assessments were performed every 12 h (7–9 AM and 7–9 PM) every day, including weekends, from the day of surgery until patient discharge, performed by researchers educated in the RASS and the CAM-ICU.

During surgery, intraoperative data, including administered drugs, blood pressure, blood loss, duration of surgery, and urine output, were measured. Postoperative data, including ICU stay, ward stay, total hospitalization days, start of oral feeding, and ambulation, were collected after surgery. The decision not to exclude patients with missing data was taken before gathering the data to ensure no attrition bias was introduced.

Sample size

To our knowledge, no prior study has measured the combined effect of melatonin and ketamine for POD prevention, nor has any study assessed the effects of 5 mg of melatonin or 50 mg of ketamine on POD prevention in colorectal cancer surgeries. In this study, we aimed to assess the feasibility of a study on POD prevention; thus, no formal sample size calculations were needed. As a rule of thumb, we included 30 patients in each group.

Statistical analysis

SAS software (version 9, SAS Institute Inc., Cary, NC, USA) and SPSS software (version 27, SPSS Inc., Chicago, IL, USA) were used to analyze the data. Continuous data are presented as the means ± standard deviations, and categorical/nominal variables are reported as counts and percentages. All exposures and covariates were compared among groups via ANOVA for continuous variables and the chi-square test or Fisher’s exact test for categorical data. We used PROC MIXED to analyze both primary and secondary continuous outcomes (e.g., POD severity and postoperative pain), with fixed effects for time, treatment group, and surgery duration. For the binary outcome of postoperative delirium incidence, generalized estimating equations (GEE (Hardin and Hilbe 2012)) with an unstructured covariance matrix were used. Based on the guidelines by Hardin and Hilbe (2013), a χ²/df ratio close to or below 1 is considered indicative of acceptable model fit. Patients with a MoCA score ≤ 10 were excluded from secondary outcome analyses to reduce heterogeneity due to severe cognitive impairment. However, for sensitivity analyses and all primary outcome models, all patients were included, as excluding those with MoCA ≤ 10 led to model convergence issues. All analyses were restricted to postoperative days 0–4 due to minimal outcome variation beyond day 4. Significance was set at p < 0.05.

Results

Data collection began on April 4, 2023 (after the IRCT registration dated March 7, 2023). A total of 135 patients were assessed for eligibility, and 120 were included in the study. Of the 15 patients who did not enter the study, 7 declined to provide consent, 5 had language or hearing impairments, and 3 had a prior history of ischemic heart disease. Among the 120 patients randomized into 4 study arms, 11 did not receive their allocated treatment due to surgery cancellations (n = 10) or induction with ketamine because of blood pressure instability (n = 1). Five patients were excluded after receiving the intervention due to deviations from the planned procedure, two had surgeries lasting less than 60 min, and three underwent multiple surgeries (Fig. 1). Ultimately, 104 patients remained in our study, and no medication delivery was missed; thus, 23 patients were included in the PS group, 27 in the MS group, 24 in the MK group, and 30 in the PK group.

Fig. 1.

Eligibility assessment and randomization of patients in each group (CONSORT flow diagram)

Descriptives

While we included patients aged 18 years or older, according to the baseline clinical variable comparison, the mean age of patients in our study was 56.29 ± 12 years, with 83 (80%) patients being 45 years or older. Patients in the PK group had a significantly greater duration of surgery than those in the MS group (mean difference = 57.19 min, Bonferroni adjustment p = 0.023, Table 1). Preoperative laboratory data were also compared between groups (see Additional file 1: Complementary patient data comparison). On the postoperative night of surgery, patients in group PK reported significantly more pain compared to group PS (mean difference = 3.22, p = 0.011). Although ANOVA results showed a significant difference in the level of pain on night 4 postoperatively, the post hoc Bonferroni adjustment did not reveal any significant differences (patients in the PK group reported the lowest pain level compared to all three groups).

Table 1.

Between-group differences in preoperative, intraoperative, and postoperative data

Characteristics		PS (N = 23)	MS (N = 27)	MK (N = 24)	PK (N = 30)	p-value
Sex	Male	16 (69.57%)	12 (44.44%)	12 (50.0%)	13 (43.33%)	0.226
	Female	7 (30.43%)	15 (55.56%)	12 (50.0%)	17 (56.67%)
Employment	Not working	8 (34.78%)	13 (48.14%)	13 (54.17%)	13 (43.33%)	0.446
	Active worker	4 (17.39%)	4 (14.81%)	5 (20.83%)	10 (33.33%)
	Retired	10 (43.48%)	7 (25.93%)	6 (25.0%)	5 (16.67%)
Chemotherapy		9 (39.13%)	12 (44.44%)	10 (41.67%)	14 (46.67%)	0.930
Radiotherapy		10 (43.48%)	11 (25%)	10 (41.67%)	14 (46.67%)	0.978
Married		4 (17.39%)	1 (3.7%)	2 (8.33%)	4 (13.33%)	0.408

PS, placebo-saline; MS, melatonin-saline; MK, melatonin-ketamine; PK, placebo-ketamine

We found no differences in cases of sex, employment, marital status, history of chemotherapy or radiotherapy, other intraoperative data, postoperative ICU and ward hospitalization, ambulation, and start of feeding day, as well as other baseline characteristics and laboratory results (Table 2).

Table 2.

Group-level differences in sociodemographic and treatment characteristics

Characteristics		PS (N = 23) Mean (SD)	MS (N = 27) Mean (SD)	MK (N = 24) Mean (SD)	PK (N = 30) Mean (SD)	p-value
Preoperative	Age (year)	60.78 (11.33)	54.63 (12.75)	58.04 (13.54)	52.83 (12.12)	0.11
	Height (cm)	169.3 (9.22)	167.07 (10.46)	166.0 (9.38)	167.73 (8.18)	0.698
	Weight (kg)	75.27 (13.02)	69.77 (14.07)	68.5 (10.76)	71.96 (11.86)	0.309
Postoperative	ICU stay (day)	1.74 (1.10)	1.67 (2.06)	1.71 (1.12)	2.30 (2.73)	0.572
	Ward stay (day)	6.65 (2.89)	6.04 (3.07)	6.62 (5.66)	5.6 (2.03)	0.661
	Total stay (day)	8.39 (2.93)	7.70 (2.32)	8.33 (5.36)	7.90 (3.38)	0.886
	Start of oral feeding (day)	3.14 (1.15)	3.41 (1.40)	3.00 (0.89)	2.70 (1.10)	0.199
	Ambulation (day)	2.15 (0.81)	2.52 (1.12)	2.16 (0.50)	2.46 (2.71)	0.826
Intraoperative	Duration (min)	220.24 (77.14)	177.63 (53.68)	202.88 (51.48)	234.83 (77.56)	0.033
	Mean MAP (mmHg)	98.90 (11.30)	100.74 (11.11)	93.57 (14.28)	94.60 (14.28)	0.224
	Maximum MAP (mmHg)	122.85 (16.83)	122.47 (12.88)	117.43 (20.24)	116.79 (18.01)	0.515
	Minimum MAP (mmHg)	75.90 (11.54)	81.32 (12.53)	74.76 (13.83)	78.50 (17.08)	0.477
	Surgery BIS	49.86 (3.69)	48.84 (4.19)	50.50 (5.12)	49.60 (4.41)	0.675
	Blood loss (cc)	171.43 (116.80)	124.74 (88.90)	147.50 (58.48)	174.67 (91.04)	0.231
	Urine output (cc)	466.71 (314.30)	436.84 (229.64)	600.00 (558.20)	436.28 (228.56)	0.351
Preoperative MOCA	Total (out of 30)	18.57 (7.55)	19.59 (6.68)	19.00 (7.44)	22.9 (5.75)	0.085
	Visuospatial (out of 5)	2.39 (1.78)	2.52 (1.81)	2.21 (1.91)	3.34 (1.78)	0.106
	Naming (out of 3)	2.26 (0.81)	2.48 (0.89)	2.25 (0.94)	2.52 (0.63)	0.517
	Attention (out of 6)	3.91 (2.45)	4.00 (1.80)	4.12 (1.96)	4.79 (1.37)	0.307
	Language (out of 3)	1.30 (1.18)	1.41 (1.15)	1.21 (1.25)	1.86 (1.16)	0.645
	Abstraction (out of 2)	0.87 (0.87)	0.96 (0.94)	0.96 (1.00)	1.45 (0.83)	0.083
	Memory (out of 5)	2.35 (1.58)	2.81 (1.62)	2.83 (1.55)	3.07 (1.67)	0.456
	Orientation (out of 6)	5.22 (1.31)	5.40 (1.12)	5.46 (1.06)	5.86 (0.58)	0.144
Surgery night	CAM-ICU-7	1.74 (2.36)	1.89 (1.87)	1.42 (1.84)	1.57 (1.98)	0.849
	RASS	− 0.7 (1.15)	− 0.89 (0.93)	− 0.54 (0.83)	− 0.73 (0.91)	0.638
	VAS	2.77 (3.10)	5.52 (3.98)	4.62 (3.28)	6.00 (3.79)	0.012
1 st day	CAM-ICU-7	0.43 (1.34)	0.81 (1.49)	0.67 (1.24)	0.97 (1.27)	0.534
	RASS	− 0.04 (0.47)	− 0.26 (0.59)	− 0.12 (0.34)	− 0.37 (0.67)	0.151
	VAS	5.83 (3.24)	5.46 (3.05)	4.21 (3.6)	5.13 (3.10)	0.358
1 st night	CAM-ICU-7	0.91 (1.86)	0.63 (1.45)	0.58 (0.97)	0.4 (0.93)	0.584
	RASS	− 0.39 (0.84)	− 0.26 (0.66)	− 0.08 (0.50)	− 0.17 (0.59)	0.407
	VAS	3.78 (2.95)	5.33 (3.22)	4.29 (3.38)	4.77 (3.01)	0.348
2nd day	CAM-ICU-7	0.52 (1.41)	0.3 (0.91)	0.5 (1.22)	0.47 (0.90)	0.881
	RASS	− 0.08 (0.51)	− 0.14 (0.36)	− 0.08 (0.28)	− 0.20 (0.40)	0.668
	VAS	3.57 (2.90)	4.37 (2.99)	3.5 (2.83)	4.07 (3.13)	0.685
2nd night	CAM-ICU-7	0.43 (1.34)	0.26 (0.94)	0.17 (0.56)	0.2 (0.61)	0.739
	RASS	0.39 (1.90)	− 0.07 (0.38)	− 0.04 (0.20)	− 0.1 (0.31)	0.218
	VAS	2.61 (2.08)	3.26 (3.15)	3.67 (2.97)	3.43 (2.67)	0.592
3rd day	CAM-ICU-7	0.22 (1.04)	0.48 (0.98)	0.21 (0.72)	0.23 (0.97)	0.671
	RASS	− 0.09 (0.42)	− 0.22 (0.51)	− 0.04 (0.20)	− 0.07 (0.25)	0.279
	VAS	3.17 (2.76)	3.85 (3.03)	3.08 (3.20)	3.20 (2.71)	0.764
3rd night	CAM-ICU-7	0.17 (0.83)	0.33 (0.83)	0.17 (0.64)	0.24 (0.74)	0.853
	RASS	− 0.04 (0.21)	− 0.07 (0.38)	− 0.08 (0.41)	− 0.1 (0.31)	0.937
	VAS	3.17 (2.62)	3.48 (2.61)	3.58 (3.11)	3.59 (3.21)	0.956
4th day	CAM-ICU-7	0.39 (1.49)	0.03 (0.19)	0.16 (0.48)	0.27 (1.16)	0.609
	RASS	0.17 (1.15)	0.0 (0.0)	− 0.04 (0.2)	− 0.07 (0.26)	0.450
	VAS	2.47 (2.74)	2.62 (2.80)	2.83 (2.46)	2.55 (2.55)	0.970
4th night	CAM-ICU-7	0.17 (0.83)	0.0 (0.0)	0.08 (0.41)	0.28 (1.16)	0.569
	RASS	− 0.09 (0.42)	0.0 (0.0)	− 0.04 (0.2)	− 0.07 (0.26)	0.649
	VAS	3.35 (3.13)	3.11 (2.42)	3.04 (2.48)	1.59 (1.92)	0.041

PS, placebo-saline; MS, melatonin-saline; MK, melatonin-ketamine; PK, placebo-saline; SD, standard deviation; ICU, intensive care unit; MAP, mean arterial pressure; BIS, bispectral index score; MoCA, Montreal Cognitive Assessment; CAM, Confusion Assessment Method; RASS, Richmond Agitation Sedation Scale; VAS, visual analogue scale

The incidence rates were similar across groups and time points, with prevalence ranging from 16.67% to 22.23% (Table 3). Statistical analysis revealed no significant differences in the incidence of delirium overall or at any time point between the groups.

Table 3.

Postoperative POD prevalence and daily incidence

Delirium incidence	PS (N = 23)	MS (N = 27)	MK (N = 24)	PK (N = 30)	p-value
Surgery night	4 (17.39%)	3 (11.11%)	3 (12.5%)	4 (13.33%)	0.941
1 st day	1 (4.35%)	3 (11.11%)	2 (8.33%)	0 (0.0%)	0.261
1 st night	1 (4.35%)	2 (7.41%)	1 (4.17%)	0 (0.0%)	0.530
2nd day	2 (8.7%)	1 (3.7%)	1 (4.17%)	0 (0.0%)	0.369
2nd night	1 (4.35%)	1 (3.7%)	0 (0.0%)	0 (0.0%)	0.465
3rd day	1 (4.35%)	1 (3.7%)	0 (0.0%)	1 (3.33%)	0.893
3rd night	1 (4.35%)	2 (7.41%)	0 (0.0%)	1 (3.33%)	0.730
4th day	1 (4.35%)	0 (0.0%)	0 (0.0%)	1 (3.33%)	0.718
4th night	1 (4.35%)	0 (0.0%)	0 (0.0%)	1 (3.33%)	0.718
Prevalence (95% exact confidence interval)	4 (17.39%) (4.95–38.78%)	6 (22.23%) (8.62–42.26%)	4 (16.67%) (4.74–37.38%)	5 (16.67%) (5.64–34.72%)	0.957

PS, placebo-saline; MS, melatonin-saline; MK, melatonin-ketamine; PK, placebo-saline

Feasibility analysis

Regarding the feasibility of the study, four indicators were evaluated.

Recruitment and attrition rates

A total of 120 patients were recruited, and 104 patients were analyzed from April 4, 2023, until September 30, 2023; as a result, the recruitment rate was 4.7 patients per week, with an attrition rate of 13.3% (16 out of 120 participants). The mean age of the study population was 52.3 years (± 17.6), with 43.8% of the participants being male.

Missed patient assessments

A physician-researcher collected baseline demographic and laboratory data from all patients. Postoperative assessments, including CAM-ICU, CAM-ICU-7, and VAS, were gathered from all patients, and no missing data were found. Among the laboratory data, the inflammatory markers were not measured for every patient; 64.4% (67/104) of the patients had a pre-operative C-reactive protein (CRP) level, and only 35.6% (37/104) had an erythrocyte sedimentation rate (ESR) documented in their medical records prior to surgery. Patients with missing ESR and CRP showed no difference in case of age, sex, duration of surgery, and group (Additional file 2). Additionally, the intraoperative assessments were missing in 11.5% (12/104) of patients. This proportionally large amount of missing data was attributed to having only one researcher in the operating room and having no previously established system of data registration in the operating room; acquiring the data after the surgery was not possible.

Variability in anesthetic management

Table 4 shows that no significant differences were found between the groups in terms of intraoperative medications, except for midazolam, which had a significant mean difference of 0.39 between the PK and PS groups after post hoc Bonferroni adjustment (p = 0.033).

Table 4.

Intraoperative anesthetic medication

Drugs	PS (N = 23)	MS (N = 27)	MK (N = 24)	PK (N = 30)	p-value
Ketorolac	18 (78.26%)	15 (55.56%)	17 (70.83%)	23 (76.67%)	0.831
Acetaminophen	18 (78.26%)	15 (55.56%)	18 (75.0%)	26 (86.67%)	0.634
Fentanyl	21 (91.3%)	19 (70.37%)	22 (91.67%)	29 (96.67%)	0.685
Morphine sulfate	18 (78.26%)	17 (62.96%)	23 (95.83%)	28 (93.33%)	0.188
Midazolam	7 (30.43%)	13 (48.15%)	13 (54.17%)	21 (70.0%)	0.035
Cisatracurium	5 (21.74%)	2 (7.41%)	4 (16.67%)	3 (10.0%)	0.581
Atracurium	15 (65.22%)	17 (62.96%)	18 (75.0%)	26 (86.67%)	0.316
Isoflurane	18 (78.26%)	17 (62.96%)	20 (83.33%)	27 (90.0%)	0.859
Sodium thiopental	14 (60.87%)	16 (59.26%)	21 (87.5%)	27 (90.0%)	0.079
Ondansetron	19 (82.61%)	19 (70.37%)	23 (95.83%)	29 (96.67%)	0.091

PS, placebo-saline; MS, melatonin-saline; MK, melatonin-ketamine; PK, placebo-saline

Delirium assessment

Among the 22 patients who developed POD during their hospitalization, 14 (63.6%) became delirious on the night of the surgery, 4 (18.2%) on the day after surgery, and 1 (4.6%) on the third day after surgery. Thus, 81.8% of patients who developed POD were recognizable until the first day after surgery. The daytime assessments detected only 10 (45.4%) PODs, whereas the nighttime assessments detected 19 (86.4%) PODs.

Analytic analysis

The GEE model fit statistics indicated an acceptable fit (χ² = 185.37, χ²/df ratio = 0.23). The tests of fixed effects for the main outcome revealed that treatment group (p = 0.8769) and surgery duration (p = 0.6398) were not statistically significant predictors of the incidence of POD. However, time (p = 0.0016) was a significant predictor, suggesting that the incidence of POD varies significantly across different time points.

The factor of time was a significant predictor for POD severity (F = 5.81, p < 0.0001). However, the treatment group and duration of surgery were not significantly related to the POD severity (p = 0.5738 and p = 0.4366, respectively). Specifically, the effects of time were statistically highest the night after the surgery (F = 6.29, β = 1.30, p < 0.0001), followed by the first day after the surgery (F = 4.13, β = 0.538, p < 0.0001), and the first night after the surgery (F = 2.49, β = 0.375 p = 0.0151). As shown in Fig. 2A, the change in delirium severity over time was similar between groups. However, the POD severity difference at later time points, such as the second day after surgery (estimate = 0.1875, p = 0.0911) and the third day after surgery (β = 0.07498, p = 0.3080), was not statistically significant.

Fig. 2.

A POD severity in different groups. B Pain severity (VAS) in different groups

The time variable showed a significant relation with postoperative pain, as measured by the VAS. The greatest differences between the groups were observed at specific time points when compared to the fourth night (treatment group, p = 0.153; surgery duration, p = 0.194). Notably, VAS was significantly higher when measured the night after the surgery (β = 2.3720, p < 0.0001), the first day after the surgery (β = 2.8285, p < 0.0001), and the first night after the surgery (β = 1.9660, p = 0.0001). Additionally, VAS measured on the second day after surgery also had a significant effect (β = 1.0535, p = 0.0110). As shown in Fig. 2B, the change in pain severity was similar between the groups. However, it did not reach statistical significance when measured on the second night after the surgery (β = 0.7660, p = 0.0609), the third day after the surgery (β = 0.5535, p = 0.1047), and the third night after the surgery (β = 0.5694, p = 0.1346). Similar to the POD severity outcome, the later time points showed no significant differences in pain severity (Table 5).

Table 5.

POD incidence, POD severity, and pain severity

Effect	Estimate	SE	p-value
POD incidence (CAM-ICU)
Group	0.23	-	0.8769
Time	3.16	-	0.0016
Surgery duration	0.22	-	0.6398
POD severity (CAM-ICU-7)
Group PS	− 0.1245	0.1194	0.3002
Group MS	− 0.1471	0.1202	0.2249
Group MK	− 0.1193	0.1154	0.3045
Group PK	0
Surgery night	1.3000	0.2067	<.0001
1 st day	0.5375	0.1302	<.0001
1 st night	0.3750	0.1507	0.0151
2nd day	0.1875	0.1095	0.0911
2nd night	0.02498	0.09727	0.7980
3rd day	0.07498	0.07305	0.3080
3rd night	0.05192	0.09597	0.5901
4th day	0.01256	0.03786	0.7409
4th night	0
Pain severity (VAS)
Group PS	− 0.2702	0.4799	0.5751
Group MS	0.8843	0.4832	0.0712
Group MK	0.2680	0.4634	0.5648
Group PK	0
Surgery night	2.3720	0.4890	<.0001
1 st day	2.8285	0.3905	<.0001
1 st night	1.9660	0.4911	0.0001
2nd day	1.0535	0.4041	0.0110
2nd night	0.7660	0.4025	0.0609
3rd day	0.5535	0.3369	0.1047
3rd night	0.5694	0.3765	0.1346
4th day	0.02052	0.3021	0.9460
4th night	0

PS, placebo-saline; MS, melatonin-saline; MK, melatonin-ketamine; PK, placebo-saline; CAM, Confusion Assessment Method; VAS, visual analogue scale

Sensitivity analysis

Sensitivity analyses including all patients, regardless of MoCA score, showed no attenuation of effect estimates and did not materially alter the significance levels of the primary or secondary outcomes.

Discussion

This study evaluated the feasibility of an interventional approach involving two components, preoperative and intraoperative interventions. Specifically, we examined the effects of 5 mg of oral melatonin and 50 mg of IV ketamine on preventing POD in candidates undergoing colorectal cancer surgery.

In this single-center study, we successfully recruited 120 patients, achieving a recruitment rate of 4.7 patients per week. In a multicenter clinical trial assessing the effect of melatonin on POD, Khaled et al. assessed the recruitment rate of patients at three sites and were able to recruit 0.96 patients per week, which is far lower than our 4.7 rate (Khaled et al. 2025a). Although they included candidates for vascular, thoracic, gynecological, otolaryngological, general, and gastrointestinal surgeries, we included only candidates for colorectal cancer surgeries. Their lower recruitment rate could be justified by limiting their population to elderly patients (age > 65), while we included the general population. Recruitment of patients from various surgical populations can introduce confounders, which are important factors in study design; therefore, limiting the study population to a single surgical group, despite lowering recruitment rates, can benefit statistical reasoning (Hulley et al. 2013).

Older age is a significant risk factor for delirium incidence, and many interventional studies aimed at POD prevention have been conducted on the elderly population (Janssen et al. 2019); however, we included adults over 18 years of age not only to enhance the generalizability of our study but also in accordance with a recent recommendation from the Network for Investigation of Delirium: Unifying Scientists (NIDUS), which justified that patients with moderate baseline vulnerability will probably benefit more from interventions aimed at preventing POD (Devlin et al. 2025).

Attrition is most common in studies with long-term patient follow-ups, including Siddiqi et al.’s study, which was a 16-month interventional study with 38.8% attrition (Siddiqi et al. 2016). Nevertheless, we measured attrition (13.3%) to provide a sensible oversampling rate in future studies and believe that an oversampling of 10–15% will be reasonable. This attrition is primarily due to unforeseen events during surgery, such as cancellations and deviations from the planned procedure.

A retrospective study indicated that the median day of POD incidence was day 1, aligning with our findings (Siddiqi et al. 2016), which demonstrates that POD predominantly occurs on the first night following surgery, when pain levels are at their peak. This dose-dependent relation between postoperative pain and POD was also found in a meta-analysis by Khaled et al. (Khaled et al. 2025b).

As for our secondary outcomes, we hypothesized that ketamine and melatonin could reduce the incidence and severity of POD, especially when used in combination. Contrary to our expectations, melatonin and ketamine had no significant effect on POD prevention. We also showed that POD severity was greater at earlier time points. However, the treatment group and surgery duration did not significantly contribute to the severity of POD in our analysis. Previous studies have shown that melatonin can prevent the incidence of POD. Different doses and durations of melatonin administration were tested, ranging from 3 to 10 mg and from a single dose up to 7 days of duration (Jiang et al. 2023). Our focus was to assess the effects of a single dose of 5 mg of melatonin on POD, as it is more feasible. Although we found no significant effect, a study by Thakur et al. revealed that administering 3 mg of melatonin in addition to routine preoperative medication (alprazolam and ranitidine) reduces POD severity in patients with MoCA scores greater than 25 (Thakur et al. 2024). This contradictory result in our study could be attributed to the inclusion of patients with MoCA scores below 25, who had a higher risk of developing delirium, thereby nullifying the effects of melatonin (Varpaei et al. 2024b).

Another explanation for our nonsignificant results could be the use of prespecified non-weight-adjusted doses of melatonin and ketamine. Although in the case of melatonin previous studies have also used fixed doses as mentioned earlier (e.g., 3, 5, or 10 mg), studies assessing ketamine have used weight-based administration (Viderman, et al., 2023).

As mentioned previously, the effects of ketamine administration on POD may range from having no effect to being protective or even increasing the risk of POD (Fellous et al. 2023). Our study was unable to determine the significant effects of ketamine on POD, primarily due to the small sample size and diverse patient characteristics, including age, MoCA score, duration of surgery, and the use of intraoperative medication. Previously, studies with a focus on senior patients reported significant results either in favor of or against the use of ketamine for POD prevention (Tekletsadik et al. 2024; Abd Ellatif et al. 2024; Ghazaly et al. 2023). Ketamine could be a double-edged sword for use in seniors, as while it provides neuroprotective, pain-reducing, and anti-inflammatory activities, it also induces hallucinations (Johnston et al. 2023; Murphy et al. 2021). In particular, senior patients are more prone to delirium; thus, with minor changes in cognition, such as ketamine-induced hallucination, they can easily develop delirium (Kerguelen Murcia et al. 2024). Recently, esketamine, an enantiomer of ketamine, has been studied for POD prevention. In a meta-analysis, unlike ketamine, it has shown significant preventive effects, possibly due to its higher affinity for NMDA receptors, lower side effects, and stronger anti-inflammatory power (Fellous et al. 2023; Zhang et al. 2024). Additionally, unlike previous studies reporting on the pain-reducing properties of ketamine, our study did not yield significant results, possibly because of the small sample size, as this study was conducted as part of a research dissertation’s work, facing various limitations, including resource and time limitations and institutional guidelines, and various confounders (e.g., postoperative opioid consumption), limiting the statistical power and precision of secondary outcome analyses (Kerguelen Murcia et al. 2024).

Previous studies have assessed the effects of postoperative inflammation on POD and found detrimental properties of surgical inflammation on the blood–brain barrier integrity, causing neuroinflammation through microglia activation, and thus neuronal dysfunction and cognitive disturbance (Pang et al. 2022; Brattinga et al. 2022). In a meta-analysis by Wan et al., markers of systemic inflammation, CRP and interleukin-6, were found to be significantly related to POD, supporting the role of inflammation in the incidence of delirium (Wang et al. 2022). Ketamine is known to act as an anti-inflammatory agent by suppressing pro-inflammatory cytokines, modulating immune cell activity, and reducing systemic inflammatory markers (Niu et al. 2022). Although our study was unable to gather systemic inflammatory biomarkers sufficiently, which affects the conclusion of our research, we showed that this missing data was random.

In this study, we found no difference in intraoperative medication between the groups (except for midazolam). Previous studies have shown that the effect of midazolam on POD is insignificant (Yoshimura et al. 2023; Li et al. 2025). While Kowark et al. in a RCT showed no difference in POD incidence in patients receiving preoperative midazolam (Kowark et al. 2024), and with similar results, Azeem et al. showed no difference in POD between intraoperative dexmedetomidine compared to midazolam and morphine (Azeem et al. 2018). Still, intraoperative fentanyl has been found to increase POD (Zhou et al. 2021), thereby requiring standardization.

Limitations

As this study was part of a dissertation’s research work, it was limited by the small number of patients in each group, and due to resource constraints and time limitations causing the low sample size of the study, a powerful conclusion on the effects of ketamine and melatonin on POD cannot be obtained. Additionally, our interventions were not stratified by weight; therefore, the doses of 50-mg ketamine and 5-mg melatonin may have had different effects on patients. We propose that future studies provide interventions based on the weight of patients and include a larger sample size. Although the majority of assessments in our study were nearly complete, inflammatory markers (ESR and CRP) were not adequately collected. This highlights the need for future studies to fully evaluate inflammatory markers, as they represent a significant risk factor for POD incidence. Especially in studies assessing the effects of ketamine, due to its anti-inflammatory properties, it is necessary to assess postoperative inflammatory markers (e.g., ESR, CRP, and interleukin-6). We also recommend prespecifying ways to handle the missing data prior to the study to reduce the risk of bias. We aimed not to limit the intraoperative management of patients, and we acknowledge this as a significant limitation in our study. However, we found no difference in intraoperative management (except for midazolam), but not all patients received the same medications, which introduced bias into our study. We propose that the standardization of anesthesia may be appropriate in early exploratory studies, particularly when the causal effect of an intervention is still uncertain and controlling for confounding is critical. Conversely, once efficacy is established, pragmatic trial designs are preferable to evaluate effectiveness in routine clinical practice. We also acknowledge that the inclusion of younger patients could introduce low-risk patients into our study, thereby diluting the intervention’s effect and also increase the sample size requirements. We recommend further studies to use age-stratified interventions.

Implications for research

Any study assessing an intervention’s effect on POD must determine a balance between the recruitment rate, generalizability, and population vulnerability to POD. Broadening the inclusion criteria will lower costs and increase generalizability and the recruitment rate while introducing bias through confounders. While narrowing the study population to the elderly population can increase power and reduce the sample size needed, it lowers generalizability and possibly diminishes the effect of the intervention due to the high vulnerability of the patients.

Conclusion

The study demonstrated feasibility in implementing a multi-arm perioperative intervention trial among colorectal surgery patients. While no significant differences in POD incidence were observed between groups, recruitment, retention, and protocol adherence were acceptable. We showed that POD incidence, POD severity, and pain severity decreased over time. The results of our feasibility study are beneficial for the development of future studies.

Authors’ contributions

KF and MM were responsible for the conceptualization of the study. KF developed the methodology. HAV and KF conducted the formal analysis. Project administration was managed by KF and MR. The study was supervised by KF, MR, and MM. KF, ER, PM, and SMMTT gathered and curated the data. Data validation and visualization were performed by KF and HAV. KF and ER wrote the original draft. The draft was revised by KF, MM, and MR.

Funding

This research received no specific grant from any funding agency in the government, public, commercial, or not-for-profit sectors.

Data availability

The corresponding author can provide all de-identified data used in this study upon reasonable request via email.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board (IRB) of Imam Khomeini Hospital Complex (IKHC), Tehran, Iran, 14 February 2023 (ethics code: IR.TUMS.IKHC.REC.1401.374 and IRCT code: IRCT20120527009886N2, registration date: 07/03/2023, prior to the commencement of the trial and enrollment of the first patient). Written informed consent was taken from all patients. The purpose of the study was explained to the participants, and they were assured that their participation would be voluntary and that their data would be anonymized per Iranian regulations.

Consent for publication

Written informed consent was taken from all patients.

Competing interests

The authors declare no competing interests.