Impact of Dexmedetomidine Supplemented Analgesia on Delirium in Patients Recovering From Orthopedic Surgery

A Randomized Controlled Trial

Hong Hong; Da-Zhi Zhang; Mo Li; Geng Wang; Sai-Nan Zhu; Yue Zhang; Dong-Xin Wang; Daniel I. Sessler


BMC Anesthesiol. 2021;21(223) 

In This Article


This randomized, double-blinded, placebo-controlled trial with two parallel groups was performed at the Peking University First Hospital and Beijing Jishuitan Hospital, both in Beijing, China. The study protocol was approved by the Biomedical Research Ethics Committee of Peking University First Hospital (2018–131 on July 18, 2018; No.6 Da-Hong-Luo-Chang Street, Beijing 100,034, China; Chairperson Prof. Xiao-Hui Guo) and the Ethics Committee of Beijing Jishuitan Hospital (201,808–06 on August 28, 2018; No.31 Xin-Jie-Kou East Street, Beijing 100,035, China; Chairperson Prof. Xiao-Lan Zhao). Written informed consents were obtained from all patients or their legal representatives. The trial was registered prior to patient enrolment at (ChiCTR1800017182; principal investigator: Dong-Xin Wang; date of registration: July 17, 2018) and (NCT03629262; principal investigator: Dong-Xin Wang; date of registration: August 14, 2018). This manuscript adheres to the applicable Consolidated Standards of Reporting Trials (CONSORT) guidelines.

Potential participants were screened and consented pre-operatively. We included patients aged 65–90 years who were scheduled for elective hip or knee arthroplasties, hip fracture repair, or spinal surgery and who agreed to use patient-controlled intravenous analgesia postoperatively. We excluded patients who: (1) were scheduled for cancer surgery; (2) had a pre-operative history of schizophrenia, epilepsy, parkinsonism, or myasthenia gravis; (3) inability to communicate due to coma, profound dementia, or language barrier; (4) sick sinus syndrome, severe sinus bradycardia (< 50 beats per min), or second- or third-degree atrioventricular block without a pacemaker; (5) diagnosed sleep apnea syndrome or a STOP-Bang score ≥ 3 combined with a serum bicarbonate ≥ 28 mmol.l−1; or, (6) severe hepatic dysfunction (Child–Pugh class C), renal failure (requiring dialysis before surgery), American Society of Anesthesiologists physical status > IV, or estimated survival ≤ 24 h.


Hip and knee arthroplasties were performed with neuraxial anesthesia or a peripheral nerve block, either combined with general anesthesia. Neuraxial anesthesia included epidural and combined spinal-epidural anesthesia. Peripheral nerve blocks included lumbar plexus, sciatic nerve, femoral nerve, and iliac fascial space. All were performed with ultrasound guidance. Patients were sedated during block insertion with dexmedetomidine and/or midazolam. The target was to maintain Richmond Agitation-Sedation Scale (RASS) scores between -2 and 0. RASS scores range from –5 (unarousable) to + 4 (combative), with 0 indicating an alert and calm subject.[16]

Spinal anesthesia was performed with bupivacaine; epidural anesthesia and peripheral nerve blocks used ropivacaine. Per routine, epidural catheters, if used, were withdrawn at end of surgery because patients were given prophylactic antithrombotic therapy after surgery. Regional analgesia was not used in spine surgery patients.

About a third of participating patients had general anesthesia alone, and a small fraction had general anesthesia combined with epidural anesthesia or peripheral nerve block. General anesthesia was induced with midazolam (1–3 mg), propofol or etomidate and sufentanil or remifentanil, and maintained with propofol infusion, sevoflurane and/or nitrous oxide inhalation, and sufentanil or remifentanil. Anesthetic drugs were adjusted to maintain Bispectral Index between 40 and 60. The Bispectral Index is an electroencephalographic measure of hypnotic depth, ranging from 0 to 100, with values between 40 and 60 considered optimal.

Random numbers were computer-generated in a 1:1 ratio with a block size of 4 using SAS 9.2 software (SAS Institute, Cary, NC, USA). Randomization was stratified by trial site and surgical location (hip or knee versus spine). Trial drugs, either dexmedetomidine 200 μg/2 ml or a comparable volume of 0.9% saline, were provided as clear aqueous solutions in identical appearing 3-ml ampules (Yangtze River Pharmaceutical Group Co., Ltd., Jiangsu, China). Sequential randomization numbers were assigned to vials by a pharmacist who was otherwise not involved in the trial. Allocation was concealed in sequentially numbered sealed opaque envelopes until the end of the trial. All investigators, clinicians, and patients were therefore completely blinded to treatment allocation. But in case of emergency (such as unexpected, rapid deterioration in a participant's clinical status), clinicians could adjust or stop drug administration if deemed clinically necessary. Unmasking was allowed only if clearly needed for clinical purposes.

Postoperative analgesia was primarily provided by patient-controlled intravenous administration of the trial drug (either dexmedetomidine 200 μg or 0.9% saline) and 200 μg sufentanil, diluted with 0.9% saline to 160 ml. The patient-controlled pump was programmed to deliver 2-ml boluses with a lockout interval of 8 min and a background infusion of 1 ml.h−1. We adopt this dosing regimen because it has been safely used in our clinical practice and our previous studies.[13] Patient-controlled analgesia was continued for at least 24 h, but not longer than 72 h after surgery. Other analgesics including non-steroidal anti-inflammatory drugs, acetaminophen, and opioids were administered when the Numeric Rating Scale (NRS, an 11-point scale where 0 indicates no pain and 10 the worst pain) of pain remained > 3 despite self-controlled analgesia. As a routine practice, patient-controlled analgesia was stopped after 48 h by anesthesia or ward nurses when the NRS pain score with movement was ≤ 3, analgesics could be taken orally, and/or hospital discharge was planned. Open-label dexmedetomidine was not allowed except for treatment of delirium.

Patients were transferred to the intensive care unit (ICU) when clinically indicated; otherwise, they remained in the post-anesthesia care unit for at least 30 min, and were then sent to a surgical ward. Electrocardiogram, invasive or non-invasive blood pressure, and pulse oxygen saturation were monitored continually in critical care and recovery units. Non-invasive blood pressure and pulse oxygen saturation were monitored intermittently until next morning. Non-invasive blood pressure and heart rate were then monitored once or twice daily until hospital discharge. Those with unstable hemodynamic were monitored frequently and transferred to an intensive care unit if necessary.

Non-pharmacological strategies to reduce delirium, including restoring hearing and vision aids, reorientation, cognitive stimulation, early mobilization, sleep-promotion and timely correction of dehydration were all used per clinical routine.[17] Patients with delirium were initially managed with non-pharmacological measures and treatment of primary diseases. Severe agitation (RASS score of + 3 or more) was treated with haloperidol and/or dexmedetomidine.[18]


Baseline data included demographic characteristics, surgical diagnosis, pre-operative comorbidities, surgical history, smoking and alcohol consumption, and pre-operative medications and laboratory test results. The Charlson comorbidity index was calculated.[19] During the pre-operative interview, cognitive function was evaluated with the Mini-Mental State Examination score (MMSE; scores range from 0 to 30, with higher scores indicating better cognitive function).[20]

Routine intra-operative monitoring included electrocardiogram, non-invasive blood pressure, pulse oxygen saturation and urine output. We also recorded Bispectral Index, end-tidal carbon dioxide partial pressure, and volatile anesthetic concentration. Intra-arterial and central venous pressures were monitored when clinically indicated. Other intra-operative data included the type and duration of anesthesia, types and doses of medication during anesthesia, type and duration of surgery, estimated blood loss, administered fluid volumes, and blood transfusions. Postoperative data included intensive care unit admission after surgery, study drug and sufentanil consumption during patient-controlled analgesia, supplemental analgesics and hypnotics within 5 days, and other medications.

Postoperative pain severity was assessed twice daily, between 8–10 AM and between 6–8 PM, with the NRS, both at rest and with movement. "Movement" was defined as turning over on/getting off the bed for patients after spinal surgery and flexion–extension/rehabilitation exercise for those after joint surgery. The most severe pain score during movement was recorded. Subjective sleep quality was assessed once daily, between 8–10 AM, with the NRS. Patients were asked to give a comprehensive score that best evaluate their overall sleep quality last night, i.e., a good night's sleep or a bad night's sleep. The scale ranged from 0 to 10, with 0 representing the best possible sleep and 10 the worst possible sleep. A minimum difference of 1 point was considered clinically meaningful.[21]

Our primary outcome was delirium which was assessed twice daily, between 8–10 AM and between 6–8 PM, with the Confusion Assessment Method (CAM) in patients who were not intubated or the CAM for the Intensive Care Unit (CAM-ICU) in patients who were intubated.[22,23] Delirium assessments continued until the 5th postoperative day or hospital discharge, whichever occurred first. Immediately before assessing delirium, patients' sedation or agitation status was assessed with the RASS. When patients were deeply sedated or unarousable (RASS score –4 or –5), they were considered to be comatose and delirium was not assessed. In patients with positive CAM assessments, delirium was classified into three motoric subtypes: (1) hyperactive (RASS score was consistently positive, + 1 to + 4); (2) hypoactive (RASS score was consistently neutral or negative, −3 to 0); and, (3) mixed.[24]

Secondary outcomes included pain, subjective sleep quality, and RASS score during the first 5 days; postoperative opioid consumption within 5 days; postoperative duration of hospitalization; postoperative complications within 30 days; 30-day mortality; and cognitive function and quality-of-life in 30-day survivors. Sufentanil equivalent dose was calculated in order to compare opioid consumption.[25] Postoperative complications were defined as newly occurred adverse conditions that required therapeutic intervention; that is, class 2 or higher by Clavien-Dindo classification.

Thirty days after surgery, cognitive function was assessed with the Chinese version Telephone Interview for Cognitive Status-modified (TICS-m; scores range from 0 to 48, with higher scores indicating better function).[26] Quality-of-life was assessed with the World Health Organization Quality of Life-brief version, WHOQOL-BREF; a 24-item questionnaire that provides assessments of the quality of life in physical, psychological, and social relationship, and environmental domains. For each domain, the score ranges from 0 to 100, with higher score indicating better function; minimal important difference 0.5 SD.[27]

Adverse events were monitored from the beginning of patient-controlled analgesia until 72 h after surgery. Among anticipated abnormalities, we defined bradycardia as heart rate < 45 beats per minute, hypotension as systolic blood pressure < 90 mmHg or a decreased of more than 30% from baseline, tachycardia as heart rate > 100 beats per minute, hypertension as systolic blood pressure > 180 mmHg or an increase of more than 30% from baseline, and hypoxemia as pulse oxygen saturation < 90%.

Pre-operative interview and postoperative follow-up were performed by two qualified investigators (HH and ML) who did not participate in anesthesia or perioperative care. Both investigators were trained to follow the study protocol, and in use of the CAM and the CAM-ICU by a psychiatrist. During the training process, the symptoms, diagnosis and treatment of delirium were presented, the uses of the CAM and the CAM-ICU were explained, and simulation training courses on patient-actors were performed and continued until the diagnosis of delirium reached 100% agreement between the investigators and the psychiatrist. The training process was repeated every 4–6 months throughout the trial.

Sample Size Estimation

Based on previous results,[28,29] we expected that delirium would occur in 12.5% of elderly patients after orthopedic surgery in the placebo group. In a recent trial, low-dose dexmedetomidine reduced postoperative delirium by about 60%.[30] We assumed that delirium would be reduced by 50% in the dexmedetomidine group. With significance set at 0.05 and power set at 80%, the sample size was 676 patients. Anticipating about 5% loss-to-follow-up, we planned to enroll 712 patients. Sample size was calculated with the PASS 11.0 software (Stata Corp. LP, College Station, TX, USA).

Statistical Analysis

The balance of baseline data between groups was assessed using absolute standardized difference, calculated as the absolute difference in means, medians, or proportions divided by the pooled standard deviation.[31] Baseline variables with an absolute standardized difference ≥ 0.147 (i.e.,1.96× were considered imbalanced and would be adjusted for in all analyses when considered necessary.

The primary outcome, i.e., the incidence of delirium within 5 days after surgery, was compared with chi square tests, with differences between groups expressed as relative risk (95% CI). For patients who were discharged or died within 5 days, the last delirium assessment results were used to replace the missing data when calculating incidence within 5 days; missing data were not replaced when calculating daily prevalence of delirium. The interactions between treatment effect and predefined factors were assessed separately with logistic regression models.

Other numeric variables were analyzed using independent-sample t or Mann–Whitney U tests. Differences (and 95% CIs for the differences) between medians were calculated with Hodges-Lehmann estimators. Categorical variables were analyzed using chi square, continuity-corrected chi square, or Fisher exact tests. Ordinal data were assessed by Mann Whitney U tests. Time-to-event variables were evaluated with Kaplan–Meier estimators, with differences between groups assessed with log-rank tests. Patients who died within 30 days were censored at the time of death; and those who stayed in hospital for longer than 30 days were censored at 30 days after surgery. Missing data were not replaced.

Outcome analyses were performed in the intention-to-treat population. For the primary outcome, a per-protocol analysis was also performed. Differences were calculated as dexmedetomidine group vs. or minus placebo group. No interim analysis was planned. The trial stopped when planned sample size was reached. For all hypotheses, two-tailed P values < 0.05 were considered statistically significant. For the interactions between treatment effect and predefined factors, P values < 0.10 were considered statistically significant. Statistical analyses were performed on SPSS 25.0 software package (IBM SPSS, Chicago, IL).