Effect of Intranasal Dexmedetomidine on the Night Before Surgery for Postoperative Delirium in Older Patients Undergoing Total Hip or Knee Arthroplasty: A Randomized Clinical Trial

Abstract

Importance

Postoperative delirium is a common and serious neurocognitive complication in elderly arthroplasty patients, with preoperative poor sleep as a key risk factor. Effective preoperative prevention strategies remain an urgent clinical need.

Objective

To assess preoperative intranasal dexmedetomidine for preventing postoperative delirium in older adults undergoing hip or knee arthroplasty.

Design

This trial was conducted at a single tertiary academic hospital in China from January 10, 2025 to June 10, 2025, with follow-up assessments performed from admission to 7 days after surgery.

Setting

This was a single-center, double-blind, placebo-controlled randomized trial.

Participants

232 elderly patients aged 60 years or older, with American Society of Anesthesiologists (ASA) status 1 to 3, scheduled for total hip or knee arthroplasty under general anesthesia were enrolled and randomized 1:1 after eligibility screening.

Interventions

Intranasal dexmedetomidine(100 μg total dose) or normal saline, administered at 21:00 on the night before surgery.

Main Outcomes and Measures

The primary outcome was the incidence of postoperative delirium within 3 postoperative days, evaluated by the Confusion Assessment Method. Secondary outcomes included postoperative delirium incidence within 7 days, duration, severity, subtype, perioperative sleep quality, pain, anxiety, quality of recovery, time to ambulation, and adverse events.

Results

In the intention-to-treat analysis, the postoperative delirium incidence within 3 days was 10.3% in the dexmedetomidine group and 20.7% in the placebo group [relative risk=0.50; 95% confidence interval (CI)=0.26–0.95; P=0.030]. Intranasal dexmedetomidine could improve sleep quality and reduce preoperative and postoperative anxiety and pain. No severe adverse events or nasal mucosal injuries were observed in either group.

Conclusion

Intranasal dexmedetomidine on the night before surgery could reduce the incidence of postoperative delirium and improve perioperative sleep quality in older patients undergoing joint arthroplasty. Sleep improvement may partly explain this association, suggesting a potential supportive strategy for delirium prevention.

Trial Registration

Chinese Clinical Trial Registry Identifier: ChiCTR2500095414.

Introduction

Postoperative delirium (POD), a common and serious neurocognitive complication that usually appears in the early postoperative period, is marked by an acute and fluctuating disruption in attention, awareness, and cognition.1–3 Clinical manifestations present as hyperactive, hypoactive, or mixed subtypes. Older adults undergoing non-cardiac surgery, particularly total hip or knee arthroplasty, exhibit a 10% to 30% incidence of POD.4,5 The condition has been associated with disability and distress for millions of patients every year, posing a public health burden.

Preoperative sleep disturbances have been identified as a modifiable and mechanistically important risk factor for POD.6–8 In older adults, preoperative nocturnal sleep deprivation serves as a marker for POD.9,10 Furthermore, after controlling for age, comorbidities, and baseline cognitive condition, two systematic reviews on major non-cardiac surgeries demonstrated that preoperative sleep fragmentation, wakefulness, and nocturnal awakenings are strongly associated with an increased risk of POD.11,12 Thus, disturbed sleep before surgery may increase susceptibility to delirium. Consequently, measures designed to enhance preoperative sleep quality may serve as a feasible option to lower the incidence of POD.13,14

Dexmedetomidine, a highly selective α2-adrenergic receptor agonist, is frequently used for sedation, analgesia, and anxiolysis. The activation of endogenous sleep regulatory pathways mediates its sedative action, resulting in a neurophysiological state that is similar to natural sleep.15–17 By inducing spindle activity and slow δ oscillation patterns that resemble N2 sleep stages, dexmedetomidine better maintains physiological sleep architecture than γ-aminobutyric acid drugs, such as benzodiazepines.15,18,19 Notably, in non-cardiac surgery populations, intravenous dexmedetomidine administration during or after surgery has been associated with a lower incidence of POD.20,21

Most existing studies focus on dexmedetomidine delivered intraoperatively, postoperatively, or in the intensive care unit.19,21,22, Evidence concerning preoperative intranasal dexmedetomidine for patients scheduled for joint arthroplasty is still inadequate, especially concerning its efficacy in neuroprotection and sleep regulation.

However, on the night before surgery, older adults undergoing joint arthroplasty tend to experience pain, anxiety, and stress, all of which worsen sympathetic activity and sleep. Intranasal dexmedetomidine confers anxiolytic and sedative effects comparable to intravenous administration while offering superior safety profiles.22–25 This necessitates determining whether intranasal dexmedetomidine administration on the night before surgery could provide neuroprotective benefits.

To this end, a randomized, double-blind trial was conducted to investigate whether intranasal dexmedetomidine administered the night before surgery reduces the incidence of POD in older adults undergoing total hip or knee arthroplasty. We hypothesized that enhancement of preoperative sleep quality could partially explain the protective effect against POD.

Methods

Study Design

This double-blind, randomized, placebo-controlled trial was conducted at the Affiliated Hospital of Xuzhou Medical University. The study was approved by the Medical Ethics Committee (approval number: XYFY2024-KL448-01) on September 23, 2024, and prospectively registered at Chinese Clinical Trial Registry (ChiCTR2500095414) on January 7, 2025. Patients were enrolled from January 10 to June 10, 2025 after written informed consent. The study complied with the Declaration of Helsinki and was reported according to the CONSORT guidelines.

Participants

Patients 60 years or older with American Society of Anesthesiologists (ASA) status 1 to 3 undergoing total hip or knee arthroplasty under general anesthesia were eligible for the trial.

The exclusion criteria were preoperative cognitive dysfunction (Mini-Mental State Examination score <15), history of schizophrenia, epilepsy, Parkinsonism, myasthenia gravis, preoperative communication inability (coma, severe dementia, or language barrier), left ventricular ejection fraction <30%, sick sinus syndrome, severe sinus bradycardia (<50 beats per minute), second-degree or higher atrioventricular block without a pacemaker, allergic rhinitis or nasal structural abnormalities, severe hepatic dysfunction (Child-Pugh class C), end-stage renal dysfunction requiring preoperative dialysis, dexmedetomidine allergy, severe pre-existing sleep disorders or a history of routine use of hypnotic medications and refusal of consent or inability to cooperate with the intervention.

Randomization and Blinding

Eligible patients were randomized 1:1 to intranasal dexmedetomidine or normal saline using permuted block randomization (block size 4). The sequence was generated by an independent statistician using R software. Group allocations were concealed in opaque envelopes. The study drug and normal saline were identical in appearance and volume.

All outcome assessors, patients, anesthesiologists, nurses, surgeons, and statisticians were blinded to group assignment. All procedures were performed by separate trained researchers, including intervention administration, safety parameter monitoring, primary outcome measurement, evaluation of secondary outcomes such as pain and anxiety, and sleep data collection. Unblinding was permitted only in medical emergencies.

Intervention

Drug intervention was performed by a trained researcher at 21:00 on the night before surgery.

Dosage was based on the package insert for dexmedetomidine hydrochloride nasal spray (Shanghai Hengrui Pharmaceutical Company Ltd., Shanghai, China). Each patient received a fixed cumulative dose of 100 μg, delivered as 2 sprays per nostril (25 μg per spray). After 30 seconds, a second spray was administered to each nostril. The entire administration was completed within 2 minutes. The placebo group received an equal volume of normal saline under the same conditions.

Following administration, patients were continuously monitored for 2 hours for adverse events (hypotension, bradycardia, hypoxemia). After 2 hours, patients were instructed to report any discomfort promptly for further physician intervention if symptoms occurred.

Anesthesia Procedure

Standard monitoring electrocardiography (ECG), peripheral oxygen saturation (SpO₂), and invasive blood pressure was applied before induction. Anesthesia was induced with propofol, etomidate, sufentanil, and rocuronium, and maintained with remifentanil and sevoflurane.

Regional blocks were performed: femoral nerve block (knee arthroplasty) or fascia iliaca compartment block (hip arthroplasty) with 0.375% ropivacaine. Postoperative analgesia was provided by intravenous patient-controlled analgesia. The postoperative patient-controlled intravenous analgesia solution contained sufentanil 2 μg/kg and tropisetron 6 mg diluted to a total volume of 100 mL with normal saline. The background infusion rate was 2 mL/h, with a bolus dose of 0.5 mL and a lockout interval of 15 min. Nurses not involved in the study conducted the NRS assessment on patients. For patients with a score >4, non-steroidal anti-inflammatory drugs were administered for rescue analgesia.

Outcomes

Primary Outcome

Baseline delirium was assessed preoperatively upon admission using the Confusion Assessment Method (CAM).26

The primary outcome was the incidence of POD within the first three postoperative days. POD was evaluated using CAM, a standardized tool that evaluates the four features of delirium, namely acute onset and fluctuating course, inattention, disorganized thinking, and an altered level of consciousness. POD was assessed twice daily by two fixed researchers who had completed systematic training. Given the fluctuating characteristics of delirium, researchers also inquired with the patient’s family and ward nurses for insights into their symptoms to minimize the rate of missed diagnosis. POD was defined as present if it occurred at either the morning or afternoon assessment.

Secondary Outcomes

Secondary outcomes included the incidence of POD within the 7 postoperative days, POD subtypes, POD severity and duration, pain, anxiety, the incidence of rescue analgesia, the score of 15-item Quality of Recovery Questionnaire (QOR-15) on the third postoperative day, time to ambulation, postoperative nausea and vomiting (PONV) within 24 h postoperatively, and duration of postoperative care (from the day of surgery to discharge).

The incidence of POD within the first 7 postoperative days was also evaluated by CAM, the POD subtypes, POD severity assessed by the Delirium Rating Scale Revised 98, and delirium duration (the days from delirium onset to the last day).

All patients were assessed for the objective and subjective quality of sleep. The sleep quality within three months before the surgery was evaluated using the Pittsburgh Sleep Quality Index (PSQI). Sleep quality from the night before surgery to postoperative days was evaluated by using the Richards-Campbell Sleep Questionnaire (RCSQ)27 and Fitbit Charge 2. The RCSQ was administered on the morning of surgery. Fitbit Charge 2 smart bracelets were used for perioperative sleep monitoring. The Fitbit Charge 2 was worn from 9 p.m. the night before surgery to 6 a.m. on the third day postoperatively. The parameters include total sleep duration, deep sleep duration, light sleep duration, and rapid eye movement (REM) sleep. Sleep efficiency was defined as the total sleep duration divided by the wearing duration.

Pain levels at rest and on movement were evaluated using a Numerical Rating Scale (NRS), and anxiety levels were assessed using the Self-Rating Anxiety Scale (SAS), both measured before anesthesia induction and each afternoon over the three postoperative days.

Frailty status was assessed at admission using the FRAIL scale (Fatigue, Resistance, Ambulation, Illnesses, Weight loss), which consists of 5 items. Patients with a total score ≥3 were defined as frail.

For safety reasons, during the intervention and surgery, other secondary outcomes included hypotension (systolic blood pressure <95 mm Hg or ≥30% decrease from baseline), hypertension (systolic blood pressure >180 mm Hg or ≥30% increase from baseline), bradycardia (heart rate <45 bpm), tachycardia (heart rate >100 bpm), and hypoxemia (peripheral oxygen saturation <90%) and nasal mucosal injuries were recorded.

Sample Size Calculation

Based on previous literature,20 POD incidence was 23% in the placebo group and 9% in the dexmedetomidine group. With 80% power and α=0.05, 104 patients were required. Considering a 10% dropout rate, 232 patients (116 per group) were enrolled. Calculations were performed using PASS 15.0.

Statistical Analysis

Continuous data were tested for normality and presented as mean (standard deviation, SD) or median (interquartile range, IQR). Categorical data were presented as n (%). Group comparisons used an independent t-test, Mann–Whitney U-test, χ²-test, or Fisher’s exact test.

The primary outcome was analyzed by intention-to-treat (ITT) and per-protocol (PP) sets. The ITT analysis included all randomized patients who received the study drug, adhering to standard trial reporting principles. To address the clinical reality that patients with canceled surgeries lacked postoperative observations, the PP analysis was performed as a pre-specified sensitivity analysis, excluding patients with post-randomization surgery cancellations. Time until POD onset within 3 days after the operation was evaluated with Kaplan–Meier survival curves.

Generalized estimating equations were applied to postoperative outcomes (anxiety, pain, and sleep data). Bonferroni correction was used for multiple comparisons.

To handle missing data, multiple imputation by chained equations was used under the missing-at-random assumption.

An exploratory causal mediation analysis was run in R 4.5.1 (mediation package), with treatment as the predictor, preoperative total sleep time as the continuous mediator, and POD as the binary outcome. Linear and logistic models were fitted respectively, with effects and 95% CIs derived from 1000 non-parametric bootstraps; no confounders or treatment-mediator interactions were adjusted for.

A Post hoc sensitivity analysis was performed using subgroup analysis and logistic regression. The logistic regression model was used to account for potential confounders by adjusting for variables related to POD, including age, sex, ASA, MMSE, PSQI, intraoperative hypotension, and anesthesia duration.

Statistical analyses were conducted with R 4.5.1; two-sided P<0.05 was considered significant.

The initial trial protocol was provided in the attachment (Supplementary Protocol, Supplementary Figure 1 and Supplementary Table 1).

Results

Patient Characteristics and Perioperative Variables

Initially, 336 patients were screened for eligibility. Of them, 104 were excluded (40 did not meet the inclusion criteria, and 64 met the predetermined exclusion criteria). Eventually, 232 patients were randomly assigned to either a dexmedetomidine group or a normal saline (placebo) group. Three patients, two in the placebo group and one in the dexmedetomidine group, cancelled their surgery, resulting in missing postoperative data. Therefore, 232 patients were included in the intention-to-treat (ITT) analysis, whereas 229 patients were included in the per-protocol (PP) analysis (Figure 1). The missing outcomes were handled using the multiple imputation model described above under the ITT analysis. Table 1 summarizes the baseline characteristics and preoperative variables.

Figure 1.

Study flow chart of participant screening, randomization and analysis.

Table 1.

Baseline Characteristics

Characteristics	Placebo Group (n=116)	Dexmedetomidine Group (n=116)	P Value
Age, mean (SD), year	69.9 (6.3)	69.8 (5.9)	0.872
Sex, n (%)			0.683
Male	41 (35.3)	44 (37.9)
Female	75 (64.7)	72 (62.1)
Body mass index, mean (SD), kg/m2	25.5 (3.6)	25.6 (3.0)	0.751
Current smoker, n (%)	24 (20.7)	20 (17.2)	0.503
Alcohol drinking, n (%)	16 (13.8)	16 (13.8)	>0.999
Educational level, n (%)			0.847
Illiteracy	38 (32.8)	36 (31.0)
Primary school	48 (41.4)	46 (39.7)
Junior high school or high	25 (21.6)	26 (22.4)
University or higher	5 (4.3)	8 (6.9)
Type of surgery, n (%)			0.658
Total knee arthroplasty	86 (74.1)	83 (71.6)
Total hip arthroplasty	30 (25.9)	33 (28.4)
ASA classification, n (%)			>0.999
I	1 (0.9)	1 (0.9)
II	101 (87.0)	101 (87.0)
III	14 (12.1)	14 (12.1)
Medical history, n (%)
Hypertension	47 (40.5)	49 (42.2)	0.790
Diabetes	12 (10.3)	15 (12.9)	0.539
Coronary artery disease	15 (12.9)	16 (13.8)	0.847
History of stroke	11 (9.5)	15 (12.9)	0.405
Hepatic dysfunction	1 (0.9)	0 (0.0)	>0.999
COPD	2 (1.7)	6 (5.2)	0.280
Age-adjusted Charlson Comorbidity Index, median (IQR), score	3 (2–3)	3 (2–3)	0.911
Mini-Mental State Examination, median (IQR), score	23 (21–25)	23 (21–25)	0.840
Self-Rating Anxiety Scale, median (IQR), score	40 (32–47)	40 (35–46)	0.531
Pittsburgh Sleep Quality Index, median (IQR), score	11 (7–13)	12 (7–15)	0.194
FRAIL, n (%)			0.677
Robust	53 (47.5)	51 (44.0)
Prefrail	46 (39.7)	43 (37.1)
Frail	17 (14.7)	22 (19.0)
Numeric Rating Scale for pain, median (IQR), score
At rest	2 (2–3)	2 (2–3)	0.661
On movement	6 (5–7)	6 (5–7)	0.837
NSAIDs use, n (%)	62 (53.4)	61 (52.6)	0.895
Preoperative systolic blood pressure, mean (SD), mmHg	139.9 (15.3)	140.6 (14.6)	0.729
Preoperative diastolic blood pressure, mean (SD), mmHg	81.8 (10.2)	80.3 (10.1)	0.241
Preoperative heart rate, mean (SD), beats/min	74.5 (9.1)	75.4 (8.4)	0.454
Preoperative laboratory values
White blood cell count, median (IQR), ×109/L	5.5 (3.3–6.5)	5.6 (3.2–6.0)	0.904
Neutrophil count, median (IQR), ×109/L	3.5 (2.8–4.6)	3.7 (3.2–4.1)	0.461
Hemoglobin, mean (SD), g/L	135.6 (13.0)	136.5 (12.1)	0.138
Platelet count, mean (SD), ×103/μL	250.0 (50.7)	237.0 (61.1)	0.584
Albumin, median (IQR), g/L	43.0 (40.2–45.0)	43.1 (40.9–45.8)	0.385
hs-CRP, median (IQR), mg/L	1.89 (0.5–3.0)	0.96 (0.5–2.8)	0.233
Erythrocyte sedimentation rate, median (IQR), mm/h	12.5 (8.0–17.5)	13.0 (9.0–18.0)	0.598
Creatinine, median (IQR), μmol/L	53.5 (47.0–61.0)	56.5 (47.0–64.5)	0.178

Notes: Continuous variables are presented as mean (SD) or median (IQR), and categorical variables as n (%), unless otherwise indicated.

Abbreviations: ASA, American Society of Anesthesiologists; COPD, chronic obstructive pulmonary disease; NSAIDs, non-steroidal anti-inflammatory drugs; hs-CRP, high-sensitivity C-reactive protein; SD, standard deviation; IQR, interquartile range.

Primary Outcome

All enrolled patients presented no preoperative delirium.

According to the ITT analysis, the dexmedetomidine group had a lower incidence of POD than the placebo group (10.3% [12/116] vs 20.7% [24/116]; relative risk [RR] =0.50; 95% confidence interval [CI] = 0.26–0.95; P=0.030). Consistent results were obtained from the PP analysis (10.4% [12/115] vs 21.1% [24/114]; RR = 0.49; 95% CI = 0.26–0.94; P=0.027; Table 2). Kaplan-Meier curves showed a difference in cumulative POD incidence between these groups (hazard ratio [HR] = 0.48; 95% CI = 0.26–0.95; P=0.031; Figure 2).

Table 2.

Primary and Secondary Outcomes

Outcomes	Placebo Group (n=116)	Dexmedetomidine Group (n=116)	Effect estimate (95% CI)	P Value
Postoperative delirium, n (%)
ITT analysis	24 (20.7)	12 (10.3)	0.50 (0.26–0.95)	0.030
PP analysis	24/114 (21.1)	12/115 (10.4)	0.49 (0.26–0.94)	0.027
POD subtypes, n(%)				0.930
Hypoactive	17 (70.8)	9 (75)	NA
Hyperactive	4 (16.7)	2 (16.7)	NA
Mixed	3 (12.5)	1 (8.3)	NA
POD within postoperative 7 days, n (%)	26/116 (22.4)	12/116 (10.3)	0.46 (0.25–0.87)	0.013
Days with POD, median (IQR), d	2 (1–2)	2 (1–2)	0 (−1 to 0)	0.691
POD severity, median (IQR), score	14 (12–16)	13 (11–15)	0 (−2 to 1)	0.659
Median pain at rest, mean (SD), score a
Before induction	2.3 (0.8)	2.1 (0.7)	0.23 (0.03–0.43)	0.023
24 h after surgery	2.4 (0.5)	2.3 (0.4)	0.19 (0.06–0.32)	0.003
48 h after surgery	1.9 (0.7)	1.8 (0.7)	0.1 (−0.09–0.30)	0.292
72 h after surgery	1.2 (0.6)	1.3 (0.6)	−0.07 (−0.23–0.09)	0.386
Median pain on movement, mean (SD), score b
Before induction	5.9 (1.1)	5.5 (0.9)	0.45 (0.19–0.71)	0.005
24 h after surgery	4.2 (0.5)	3.9 (0.7)	0.26 (0.10–0.42)	<0.001
48 h after surgery	3.0 (0.9)	3.1 (1.1)	−0.10 (−0.36–0.15)	0.424
72 h after surgery	2.8 (0.7)	2.8 (0.6)	0.01 (−0.15–0.17)	0.917
Rescue analgesia within 24 h, n (%)	21 (18.1)	24 (20.7)	1.14 (0.68–1.93)	0.618
Self-Rating Anxiety Scale, mean (SD), score c
Before induction	40.8 (8.4)	38.1 (7.0)	2.65 (0.52–4.77)	0.015
24 h after surgery	33.7 (7.1)	34.6 (6.5)	−0.82(−2.60–0.92)	0.357
48 h after surgery	31.4 (6.1)	29.9 (6.4)	1.53 (0.15–2.92)	0.230
72 h after surgery	26.7 (4.3)	26.2 (4.0)	0.47(−0.09–1.02)	0.102
RCSQ, median (IQR), score	51 (40–70)	70 (60–80)	15 (10–20)	<0.001
15-item quality of recovery, median (IQR), score	131.5 (120.5–140)	133.5 (125–140)	3(-1-6)	0.110
Time to first ambulation, mean (SD), h	45.5 (5.5)	44.1 (4.8)	−1.37 (−2.71 to −0.03)	0.045
PONV within 24 h, n (%)	12 (10.3)	13 (11.2)	0.93 (0.44–1.93)	0.832
Duration of postoperative care, median (IQR), day	4 (4–5)	4 (3–4)	0 (0–0)	0.032
Other outcomes (when intervention), n (%)
Hypotension	0	0	NA	NA
Hypertension	0	0	NA	NA
Bradycardia	0	2 (1.7)	NA	0.478
Tachycardia	1 (0.9)	1 (0.9)	1.00 (0.06–15.80)	>0.999
Hypoxemia	0	0	NA	NA
Nasal mucosal injury	0	0	NA	NA

Notes: Binary outcomes are reported as n (%) with relative risk (RR) and 95% CIs. Continuous data are expressed as mean (SD) or median (IQR). Categorical variables were compared using χ² or Fisher’s exact tests; between-group differences for non-repeated continuous outcomes were assessed via independent t-test or Mann–Whitney U-test. All effect estimates were calculated as dexmedetomidine versus placebo. Superscripts a, b, c indicate that NRS and SAS scores were approximately treated as continuous measurement data and presented as mean±SD. Generalized estimating equation (GEE) was applied for longitudinal data analysis. If the group-by-time interaction effect was statistically significant, pairwise between-group comparisons at each time point were performed, and all P values were adjusted by the Bonferroni method for multiple comparisons.

Abbreviations: ITT, intention-to-treat; PP, per-protocol; RCSQ, Richards-Campbell Sleep Questionnaire; PONV, postoperative nausea and vomiting; IQR, interquartile range; SD, standard deviation; CI, confidence interval; NA, not applicable; MD, mean difference.

Figure 2.

Kaplan-Meier curves for cumulative incidence of no postoperative delirium (POD) between groups. The Log rank test was applied for group comparison.

Abbreviations: POD, postoperative delirium; HR, hazard ratio; CI, confidence interval; Dex, dexmedetomidine.

Secondary Outcomes

POD Severity and Therapy

POD severity (14 [12–16] vs 13 [11–15]; P=0.659) and duration (IQR = 1.0–2.0 days; P=0.691) did not differ between the two groups. The hypoactive subtype accounted for 72.2% of all POD cases (26/36), followed by the hyperactive subtype (16.7%; 6/36) and the mixed subtype (11.1%; 4/36, Table 2).

Sleep Quality

On the night before surgery, patients in the dexmedetomidine group demonstrated considerably better subjective sleep quality, as evidenced by higher Richards-Campbell Sleep Questionnaire scores than patients in the placebo group (mean difference [MD] = 15; 95% CI = 10–20; P<0.001).

Objective sleep monitoring using the Fitbit Charge 2 demonstrated similar results. The Fitbit device was generally well tolerated. No participant reported significant physical discomfort, skin irritation, or notable sleep disruption attributed to the wristband. Three participants experienced partially missing sleep recordings caused by accidental Fitbit removal during overnight monitoring.

Patients in dexmedetomidine group exhibited an increase in total sleep time (405.4 vs 374.7 min; MD = 31.2; 95% CI = 19.8–41.6; P<0.001), deep sleep time (108.0 vs 98.3 min; MD = 10.2; 95% CI = 2.1–18.2; P=0.011), light sleep time (220.9 vs 207.6 min; MD = 13.3; 95% CI = 5.7–20.8; P<0.001), and rapid eye movement (REM) sleep time (76.1 vs 65.8 min; MD=11.2; 95% CI=0.4–14.1; P<0.001). Additionally, the dexmedetomidine group showed significantly higher sleep efficiency (Table 3).

Table 3.

Perioperative Sleep Outcomes Between the Groups

Outcomes	Placebo Group (n=116)	Dexmedetomidine Group (n=116)	GEE-Estimated Marginal Mean Difference (95% CI)	P Value
Total sleep time, mean (SD), min
D0	374.7 (23.9)	405.4 (24.9)	31.2 (19.8–41.6)	<0.001
D1	245.3 (27.5)	273.4 (28.1)	28.1 (17.2–38.9)	<0.001
D2	333.8 (25.3)	335.3 (25.2)	1.3 (−9.0–11.9)	0.784
D3	350.7 (29.2)	345.2 (30.8)	−5.5 (−9.9–16.9)	0.603
Deep sleep time, mean (SD), min
D0	98.3 (17.7)	108.0 (18.0)	10.2 (2.1–18.2)	0.011
D1	57.6 (15.2)	67.3 (16.9)	9.5 (4.4–15.1)	0.009
D2	98.4 (10.2)	98.3 (10.7)	0.3 (−4.8–4.8)	0.991
D3	96.5 (16.1)	99.5 (16.7)	3.1 (−4.2–10.4)	0.404
Light sleep time, mean (SD), min
D0	207.6 (16.8)	220.9 (16.5)	13.3 (5.7–20.8)	<0.001
D1	122.8 (17.8)	135.4 (17.1)	12.6 (4.7–20.5)	0.002
D2	168.3 (17.1)	168.7 (17.9)	0.3 (−7.5–8.5)	0.906
D3	177.5 (15.6)	178.1 (14.5)	0.6 (−6.1–7.3)	0.853
REM time, mean (SD), min
D0	65.8 (15.9)	76.1 (14.1)	11.2 (0.4–14.1)	<0.001
D1	62.9 (9.3)	70.7 (9.1)	7.7 (3.5–11.9)	0.006
D2	67.2 (9.7)	68.2 (9.7)	1.0 (−3.4–5.4)	0.655
D3	69.5 (9.6)	71.1 (9.9)	1.6 (−2.8–5.9)	0.480
Sleep efficiency, %
D0	69.4 (4.5)	75.1 (4.5)	5.9 (3.7–8.0)	<0.001
D1	45.4 (5.1)	50.6 (3.5)	5.1 (3.2–7.2)	<0.001
D2	61.8 (3.8)	62.1 (4.7)	0.3 (−1.7–2.2)	0.722
D3	64.7 (5.5)	65.4 (5.8)	0.6 (−1.9–3.2)	0.684
Sleep fragmentation index, median (IQR), times/h
D0	1 (1–2)	1 (1–1)	0.81 (0.58–1.13)	0.267
D1	2 (2–3)	2 (2–2)	0.76 (0.58–0.99)	0.051
D2	1 (1–2)	1 (1–2)	0.97 (0.71–1.33)	0.736
D3	1 (1–2)	1 (1–1)	0.79 (0.59–1.06)	0.111

Notes: Repeated sleep data were analyzed by generalized estimating equations (GEE), which handled partial missing records from 3 patients. If a significant group-by-time interaction was detected in the GEE model, simple main effects between the two groups at each time point were reported separately. Bonferroni correction was applied for multiple time-point comparisons. Continuous outcomes are shown as mean (SD) or median (IQR); MD = mean difference. Sleep-stage variables were derived from Fitbit Charge 2 estimates.

Abbreviations: D0, the night before surgery; D1, the first postoperative night; D2, the second postoperative night; D3, the third postoperative night; REM, rapid eye movement; SD, standard deviation; MD, mean difference; IQR, interquartile range; CI, confidence interval.

The sleep-enhancing effects of dexmedetomidine continued on the first postoperative night. Patients in the dexmedetomidine group demonstrated an increase in total sleep time (273.4 vs 245.3 min; MD = 28.1; 95% CI = 17.2–38.9; P<0.001), deep sleep time (MD = 9.5; 95% CI = 4.4–15.1; P=0.009), light sleep time (MD = 12.6; 95% CI = 4.7–20.5; P=0.002), and REM sleep time (MD = 7.7; 95% CI = 3.5–11.9; P=0.006). Similarly, the dexmedetomidine group showed significantly higher sleep efficiency than placebo group. However, there were no significant differences in the sleep parameters or sleep fragmentation index on postoperative nights 2 and 3 between the dexmedetomidine and placebo groups (Table 3).

Pain, Anxiety, and Recovery

Patients in the dexmedetomidine group reported significantly lower Numeric Rating Scale pain scores at rest and with movement than patients in the placebo group before anesthesia induction and on postoperative day 1. No differences were observed on postoperative days 2 and 3 (P>0.05). There was no difference in the incidence of PONV between the two groups (10.3% [12/116] vs 11.2% [13/116]; RR = 0.93; 95% CI = 0.44–1.93; P = 0.832; Table 2).

Before anesthesia induction, patients in the dexmedetomidine group reported lower Self-Rating Anxiety Scale (SAS) scores (MD = −2; 95% CI = −5–0; P=0.045); However, there were no differences in SAS scores on postoperative days 1 to 3 between the two groups (P>0.05).

Quality of Recovery-15 scale scores (MD = 39; 95% CI=−1 to 6, P = 0.110), and 24h postoperative nausea and vomiting incidence (10.3% vs 11.2%; RR = 1.09; 95% CI = 0.48–2.51; P=0.832) were comparable between the two groups. A statistically significant difference in the median duration of postoperative care was observed between the groups but was clinically insignificant (P=0.032, Table 2).

Mediation Analysis

In our exploratory mediation analysis, Dexmedetomidine reduced the incidence of POD (total effect = −0.112, 95% CI [−0.172, −0.049], P = 0.008), through prolonged sleep duration [Average Causal Mediation Effect (ACME)= −0.072, 95% CI [−0.125, −0.016], P = 0.016].

Exploratory causal mediation analysis suggested that sleep duration was associated with a proportion mediated of 64.0% of the total treatment effect, although the confidence interval was wide (95% CI: 11.2% to 110.8%, P=0.017) (Figure 3 and Table 4). This suggested that prolonged sleep duration may partly explain the association between dexmedetomidine and the reduction in postoperative delirium.

Figure 3.

Mediation analysis with total sleep time as the mediating variable for the association between intranasal dexmedetomidine and postoperative delirium (POD). Bootstrap resampling was used to verify the mediation effect.

Abbreviations: POD, postoperative delirium; CI, confidence interval.

Table 4.

Results of the Mediation Analysis

	Est.	95% CI LL	95% CI UL	P Value
ACME (Indirect)	−0.072	−0.125	−0.016	0.016
ADE (Direct)	−0.041	−0.134	0.049	0.385
Total Effect	−0.112	−0.172	−0.049	0.008
Prop. Mediated	0.640	0.112	1.108	0.017

Notes: In the mediation model, intranasal dexmedetomidine versus placebo was treated as the independent variable, total sleep time was treated as the mediator, and postoperative delirium within the first 3 postoperative days was treated as the dependent variable. ACME, ADE, total effect, and proportion mediated were estimated using causal mediation analysis with bootstrap resampling. The proportion mediated was calculated as ACME divided by the total effect.

Abbreviations: CI, confidence interval; LL, lower limit; UL, upper limit; ACME, average causal mediation effect; ADE, average direct effect; Prop. Mediated, Proportion Mediated (ACME/Total effect).

Intraoperative Data and Adverse Events

The length of operation, anesthesia duration, and intraoperative dosages of anesthetics and vasoactive drugs were similar between the dexmedetomidine and placebo groups (P>0.05). Adverse events, such as hypotension, hypertension, bradycardia, tachycardia, and hypoxemia, did not differ between the groups before or during surgery (P>0.05). After intranasal administration of dexmedetomidine or placebo, neither group experienced any incidence of nasal mucosal damage (Table 5).

Table 5.

Intraoperative Data Between the Groups

Characteristics	Placebo Group (n=116)	Dexmedetomidine Group (n=116)	P Value
Duration of operation, median (IQR), min	90 (80–98)	95 (86–100)	0.079
Duration of anesthesia, median (IQR), min	116.5 (111–120)	118 (110–121)	0.701
Fluid administered, median (IQR), mL	1250(1075–1500)	1250 (1250–1500)	0.728
Red blood cell transfusion, n (%)	0 (0.0)	0 (0.0)	NA
Propofol, median (IQR), mg	245 (200–295)	250 (210–280)	0.961
Sufentanil, median (IQR), μg	30 (30–30)	30 (30–30)	0.255
Remifentanil, median (IQR), mg	1.8 (1.5–2.0)	1.8 (1.5–2.0)	0.321
Phenylephrine, median (IQR), μg	200 (120–270)	200 (100–320)	0.470
Ephedrine, median (IQR), mg	0 (0–6)	0 (0–3)	0.238
Change in hemoglobin, median (IQR), g/L	23.0 (13.5–35.0)	22.0 (18.0–28.0)	0.628
Urapidil, n (%)	7 (6.0)	6 (5.2)	0.775
Hypotension, n (%)	24 (20.7)	21 (18.1)	0.618
Hypertension, n (%)	12 (10.3)	14 (12.1)	0.677
Bradycardia, n (%)	10 (8.6)	12 (10.3)	0.654
Tachycardia, n (%)	12 (10.3)	8 (6.9)	0.349

Notes: Continuous variables are presented as mean (SD) or median (IQR), and categorical variables as n (%), unless otherwise indicated. P values were calculated using the independent-samples t test for normally distributed continuous variables, the Mann–Whitney U-test for non-normally distributed continuous variables, and the χ²-test or Fisher’s exact test for categorical variables, as appropriate.

Abbreviations: SD, standard deviation; IQR, interquartile range; NA, not applicable.

Sensitivity Analysis

According to post-hoc subgroup analyses, dexmedetomidine significantly reduced the incidence of POD among patients older than 69 years, with Pittsburgh Sleep Quality Index scores >12, Mini-Mental State Examination scores >22, and classified as frail (Figure 4).

Figure 4.

Post hoc subgroup analysis for the effect of intranasal dexmedetomidine on postoperative delirium (POD). Odds ratios and 95% confidence intervals are presented for different clinical subgroups.

Abbreviations: DEX, dexmedetomidine; POD, postoperative delirium; OR, odds ratio; CI, confidence interval; PSQI, Pittsburgh Sleep Quality Index; MMSE, Mini-Mental State Examination; SAS, Self-Rating Anxiety Scale.

We identified variables that may influence POD, including age, sex, PSQI, MMSE, intraoperative hypotension, and anesthesia duration, and the dexmedetomidine still demonstrated the advantage of reducing risk of POD after adjusting for these potential confounding factors (OR, 0.28, 95% CI: 0.10–0.76, P = 0.012; Supplementary Figure 2).

Discussion

According to this randomized placebo-controlled trial, intranasal dexmedetomidine administration on the night before surgery reduced the incidence of POD in older adults undergoing total hip or knee arthroplasty. The intervention also improved perioperative sleep quality and alleviated preoperative and early postoperative pain and anxiety. Additionally, exploratory mediation analysis suggested that prolonged preoperative sleep duration may be associated with POD reduction.

Before total hip or knee arthroplasty, most patients, particularly older adults, experience anticipatory pain, anxiety, and sleep disturbance. Preoperative sleep disturbances have been identified as a major independent risk factor for POD.6,14,28,29 There is a strong correlation between sleep fragmentation, shorter sleep duration, and disrupted sleep architecture and the risk of POD.9,17 In line with our findings, a recent trial demonstrated that preoperative sleep improvement significantly reduced POD in elderly patients with pre-existing sleep disorders undergoing major noncardiac surgery,30 further supporting the potential value of preoperative sleep optimization for POD prevention. Disrupted sleep architecture is particularly detrimental in older adults, considering age-related changes in circadian rhythms and sleep homeostasis. This phenomenon might be attributed to the disrupted sleep homeostasis in the central nervous system by chronic or acute preoperative sleep deprivation.31–33 These findings reinforce the therapeutic importance of maximizing sleep before surgery. Thus, an approach to reduce the incidence of POD and optimize perioperative care is likely to improve preoperative sleep quality through safe and efficient therapies.

Dexmedetomidine is a highly selective α2-adrenergic agonist that promotes endogenous sleep pathways and inhibits locus coeruleus-mediated noradrenergic neuronal activity.15,34 In this trial, objective sleep monitoring data confirmed significant improvements in total sleep time, deep sleep duration, and REM sleep duration in the dexmedetomidine group. These improvements in sleep quality persisted into the first postoperative night. This sustained effect may be explained by dexmedetomidine’s capacity to interrupt the perioperative anxiety-pain-sleep disruption cycle. By restoring physiological sleep and suppressing sympathetic overactivity on the night before surgery, the intervention blunts the acute surgical neuroendocrine stress response, thereby preserving the integrity of central sleep regulatory centers on subsequent nights. Improvement of REM and deep sleep attributed to dexmedetomidine is essential for memory consolidation and synaptic plasticity.35,36 Furthermore, enhanced sleep quality reduces neuroinflammation and oxidative stress, protects hippocampal neurogenesis, and preserves cognitive function.37,38 When combined, these processes restore normal sleep architecture, strengthen synaptic connections, and maintain neural network integrity, thereby creating a neuroprotective environment that may lower POD incidence.

Patients who received intranasal dexmedetomidine the night before surgery demonstrated lower pain and anxiety scores within the first 24 postoperative hours, which is in line with previous trials.39–41 Dexmedetomidine has inherent anxiolytic and analgesic properties in addition to enhancing sleep quality, which may help reduce the incidence of POD. The hypothalamic-pituitary-adrenal axis is modulated by central α2-adrenoceptor activation, which reduces cortisol release and attenuates perioperative stress responses.18,33 This neuroendocrine stability prevents the downstream cascade of inflammatory mediators that contribute to POD. Additionally, dexmedetomidine directly reduces pain by inhibiting nociceptive transmission by activating the α2-adrenoceptor.42,43 Together, these mechanisms may produce a perioperative neuroprotective milieu that lessens susceptibility to POD when combined with sleep enhancement.42

Prior studies have exclusively focused on improving sleep during or after surgery. The present study expanded the intervention window to the preoperative phase in light of the discomfort, anticipatory anxiety, and circadian disturbance experienced by older adults before total hip or knee arthroplasty. Therefore, dexmedetomidine yielded more extensive perioperative advantages by focusing on the night before surgery, a crucial window of neurobiological susceptibility.

Although a statistically significant intergroup difference in duration of postoperative care was observed, this result should be interpreted prudently. First, the absolute intergroup difference was minor. Second, the study lacked adequate statistical power for this secondary exploratory outcome. Third, unmeasured confounding factors may interfere with the result. Furthermore, the duration of postoperative care was not the primary research focus of our trial. The current data only indicate a potential trend of shortened hospitalization associated with the intervention, and larger-sample prospective studies are needed to verify this finding in the future.

Several limitations of this study should be acknowledged. First, this was a single-center trial with a modest number of delirium events, which limits the generalizability of the findings and provides insufficient statistical power to detect rare adverse events. Second, although strict role separation was implemented to maintain blinding, the sedative and bradycardic effects of dexmedetomidine carry a theoretical risk of unblinding. Nevertheless, this risk had minimal impact on the primary outcome, as delirium assessments were performed after the sedative effects had fully resolved. We acknowledge that very brief or extremely subtle hypoactive delirium episodes may still have been missed, which is a common limitation of all clinical delirium studies. Third, both mediation and subgroup analyses were post-hoc exploratory analyses. The mediation estimate had a wide confidence interval extending above 100%, and the model did not adjust for potential mediator-outcome confounders or treatment-mediator interactions. Thus, it cannot establish a causal pathway. Fourth, sleep was monitored with a consumer wearable instead of polysomnography, which introduces measurement imprecision in sleep staging. Fifth, a fixed 100-μg dose was used for all patients without weight adjustment, which may not account for individual pharmacokinetic variability.

In summary, intranasal dexmedetomidine administered on the night before surgery was associated with a lower incidence of POD in older adults undergoing total hip or knee arthroplasty under general anesthesia. The intervention also improved perioperative sleep quality, and sleep improvement may partly explain the protective association against delirium, suggesting a potential supportive strategy for POD prevention. Larger multicenter trials are warranted to validate these findings.

Funding Statement

This study was supported by grant 22KJA320007 from the Natural Science Research Fund of Higher Education Institutions in Jiangsu Province, grant XWKYHT20230056 from the Science and Technology Project of Xuzhou Health Commission, grant GSPSJ20240803 and LCZX202503 from Construction Project of High-Level Hospital of Jiangsu Province, grant FXJDBF2024219 from Paired Assistance Scientific Research Project by The Affiliated Hospital of Xuzhou Medical University, grant 82301414 from the National Natural Science Foundation of China and grant BK20241766 from the Natural Science Foundation of Jiangsu Province.

Data Sharing Statement

Deidentified individual participant data, the full study protocol, statistical analysis code, and case report form template will be shared upon reasonable request. Interested researchers may contact the corresponding author (liusu_xzmu@163.com) to submit a request with a detailed research proposal. Data will be made available after manuscript publication and retained for at least 5 years.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.