Effects of Ramelteon on the Prevention of Postoperative Delirium in Older Patients Undergoing Orthopedic Surgery: The RECOVER Randomized Controlled Trial

Abstract

OBJECTIVES:

Postoperative delirium, associated with negative consequences including longer hospital stays and worse cognitive and physical outcomes, is frequently accompanied by sleep-wake disturbance. Our objective was to evaluate the efficacy and short-term safety of ramelteon, a melatonin receptor agonist, for the prevention of postoperative delirium in older patients undergoing orthopedic surgery.

DESIGN:

A quadruple-masked randomized placebo-controlled trial (ClinicalTrials.gov NCT02324153) conducted from March 2017 to June 2019.

SETTING:

Tertiary academic medical center.

PARTICIPANTS:

Patients aged >65 years, undergoing elective primary or revision hip or knee replacement.

INTERVENTION:

Ramelteon (8 mg) or placebo

MEASUREMENTS:

Eighty participants were randomized to an oral gel cap of ramelteon or placebo for 3 consecutive nights starting the night before surgery. Trained research staff conducted delirium assessments for 3 consecutive days starting on postoperative day (POD) 0, after recovery from anesthesia, and on to POD2. A delirium diagnosis was based upon DSM-5 criteria determined by expert panel consensus.

RESULTS:

Of 80 participants, 5 withdrew consent (1 placebo, 4 ramelteon) and 4 were excluded (4 ramelteon) after randomization. Delirium incidence during the two days following surgery was 7% (5/71) with no difference between the ramelteon versus placebo: 9% (3/33) and 5% (2/38), respectively. The adjusted odds ratio for postoperative delirium as a function of assignment to the ramelteon treatment arm was 1.28 (95% CI 0.21–7.93;z-value 0.27;p-value= 0.79). Adverse events were similar between the two groups.

CONCLUSION:

In older patients undergoing elective primary or revision hip or knee replacement, ramelteon was not efficacious in preventing postoperative delirium.

INTRODUCTION

Postoperative delirium, an acute confusional state arising after surgery, is caused by an underlying physiological disturbance and characterized by sleep/wake cycle disturbances.(1) It is a serious public health problem that disproportionately affects older people.(2) Delirium is associated with many negative outcomes including: increased mortality, longer hospital stays, increased institutionalization after acute care hospitalization, worse cognitive and physical outcomes (both short- and long-term), and significant distress for patients and their families.(3–5) Estimated costs to the healthcare system due to all episodes of delirium among the elderly range from $38 to $152 billion/year.(6) With the aging population, increasing numbers of older adults are undergoing surgery - making postoperative delirium prevention an urgent priority.(7)

While the mechanisms leading to postoperative delirium are unclear, it is almost always accompanied by symptoms of sleep-wake disturbance, such as daytime somnolence and nighttime sleeplessness.(8) Diurnal dysregulation may be causally important in the genesis of delirium, and thus melatonin agonists may prevent delirium by maintaining or normalizing the circadian rhythm.(9) The rationale for using a melatonin agonist is further strengthened by multiple biological effects on neuro-inflammatory mechanisms implicated in delirium,(10) via antioxidant and possibly anti-amyloid effects.(11) There are two published double-blind, randomized controlled trials (RCTs) examining the efficacy of melatonin on delirium prevention in orthopedic surgery patients with conflicting results.(12, 13) In one study involving patients over 65 years of age undergoing hip arthroplasty, patients who received 5 mg of melatonin preoperatively had significantly lower proportion of postoperative delirium (5/53; 9.4%) compared to those who did not receive melatonin (16/49; 32.7 %).(13) The other study included patients aged 65 years or older undergoing emergent hip fracture surgery. There was no difference in postoperative delirium incidence between those who received 3 mg of melatonin for 5 consecutive evenings after admission (55/186;29.6%) compared to those who received placebo (49/192; 25.5%).(12)

We conducted a quadruple-masked (participant, care provider, investigator, outcomes assessor) randomized placebo-controlled trial to evaluate the efficacy and short-term safety of a melatonin receptor agonist, ramelteon, for the prevention of postoperative delirium in an orthopedic surgery population. We chose ramelteon as it is a highly selective agonist of MT₁ (Mel_1a) and MT₂ (Mel_1b) receptor subtypes that are thought to be associated with the melatonin’s effect on circadian rhythms, and with greater affinity for the binding sites compared to melatonin.(14) The preoperative administration of ramelteon was hypothesized to correct melatonergic deficiencies observed immediately following surgery and to enhance circadian rhythm and sleep effects.(15)

Previous studies demonstrated the presence of delirium immediately upon recovery from general anesthesia as an important predictor of subsequent delirium and cognitive outcomes at hospital discharge.(3, 16) In addition to measuring delirium incidence on postoperative day (POD) 1 and/or 2, the proposed trial incorporated assessment of delirium on the day of surgery (POD 0) upon recovery from anesthesia as a means of early delirium detection in helping to determine the efficacy of the ramelteon intervention.

METHODS

Study Design Overview

This randomized placebo-controlled trial was conducted from March 2017 to June 2019 at a single clinical center. The research protocol was approved by the Johns Hopkins Institutional Review Board (IRB00097232) and registered at ClinicalTrials.gov (NCT02324153). All participants and their legally authorized representatives provided their written informed consent to participate.

The trial was undertaken at the Johns Hopkins Bayview Medical Center (JHBMC). Eligible patients were identified from the orthopedic surgical scheduling roster (after obtaining HIPAA waiver) and/or from referral by the patient’s surgeon. Upon receiving informed consent from the patient or their legally authorized representative (LAR), cognitive screening, using the Mini-mental State Exam (MMSE),(17) and delirium assessment determined eligibility. Both elective and emergency surgeries were eligible.

Participants

Patients ≥65 years of age without preoperative delirium undergoing elective primary or revision joint (hip or knee) replacement were randomized 1:1 to receive either 8 mg of ramelteon orally or placebo. The inclusion criteria were (1) planned orthopedic surgery and inpatient stay following surgery, (2) age > 65 years, (3) MMSE > 15 before surgery, (4) ability to understand, speak, read and write English. The exclusion criteria were (1) delirium diagnosis by Confusion Assessment Method (CAM)(18) prior to surgery, (2) inability to give informed consent due to cognitive impairment and a suitable Legally Authorized Representative (LAR) cannot be identified, (3) declines participation, (4) currently taking ramelteon, melatonin and medications that will alter systemic ramelteon levels including fluvoxamine, rifampin, ketoconazole, or fluconazole, (5) history of ramelteon or riboflavin intolerance, (6) heavy daily alcohol intake by medical record or history, (7) current moderate to severe liver failure (as defined by Charlson criteria (19)), (8) evidence of Systemic Inflammatory Response Syndrome (SIRS) as measured by > 2 criteria, or (9) presence of a condition that in the opinion of the Principal Investigator (PI) might compromise patient safety if enrolled in the study. The PI (KJN) personally obtained informed consent from all participants at the pre-surgical study visit.

Study Procedures

Randomization and Intervention

Since preoperative cognitive impairment is strongly associated with postoperative delirium, the randomization process was stratified by preoperative MMSE exam score to ensure balanced allocation (MMSE ≥24 vs. <24). Under supervision of the statistician, research pharmacy staff who had no contact with the patient or clinical staff randomized participants using a computer program, utilizing dynamic allocation probabilities to balance the two groups on the dichotomous MMSE variable.

The research pharmacy staff prepared the active treatment (ramelteon 8 mg or placebo) using identical gel caps containing riboflavin for doses to be self-administered outside of the hospital. A high-dose of riboflavin (100 mg) was added to the first dose of the assigned intervention (both ramelteon and placebo) to act as a marker of adherence. Since riboflavin fluoresces under UV light, a urine sample obtained prior to entering the operating room was sent for quantitative fluorometer using a cutoff of > 900 ng/ml as evidence of adherence.(20)

The participants received active treatment or placebo on three consecutive evenings: (1) the night immediately prior to surgery, (2) the night of surgery postoperative day (POD) 0 and (3) the following day POD 1. All participants who completed the study received the assigned intervention (ramelteon 8 mg/placebo) at the study visit prior to surgery and received a reminder call by research staff the night prior to surgery.

For all inpatient doses, clinical nursing staff administered medication each evening (8–10 PM). Administration was documented in the clinical electronic medication record with adherence closely monitored by research staff. Participants who were admitted to hospital the night before surgery had study assigned medication administered by clinical staff, similar to the procedure for POD 0 and POD 1 doses.

Clinical Assessment

We included two time periods of delirium incidence as our primary outcome: first during POD 1 and/or 2 and second on POD 0 after recovery from anesthesia as measured by Aldrete score > 8.(21) This tool, used by clinical nurses in the post anesthesia care unit, measures arousal, hemodynamic, and respiratory function. This allowed us to report traditionally described delirium incidence in the days following surgery as well as incidence immediately following recovery from anesthesia on the day of surgery. For participants who were discharged early from the hospital (prior to POD 2), delirium assessments were conducted via telephone. Secondary outcomes included the number of participants with adverse events, based upon daily clinical interview and medical record review, as a measure of safety and tolerability during their postoperative course.

Trained research staff conducted delirium assessments at baseline, and at the three later time points (POD 0, 1, 2). This assessment included: (1) face-to-face (or telephone if discharged) standardized interview of the patient, (2) standardized cognitive testing including the MMSE and the abbreviated Digit Span,(22) (3) history from family/friends familiar with the patient’s cognitive baseline, (4) interview of nursing and other medical staff caring for the patient, and (5) review of medical records for evidence of behavioral symptoms in previous 24 hours. Information on sleep disturbance was sought from all sources. Assessors rated the CAM and the Delirium Rating Scale-Revised 98 (DRS-R98)(23) for the prior 24 hours and described the patient’s behavior and mental state in a written narrative. Final adjudication of delirium diagnosis was derived after all data were presented to a panel of expert delirium diagnosticians (KJN, PBR and ESO) who rated each of the DSM-5 criteria at each assessment timepoint throughout the patient’s course. If a patient met some, but not all of the DSM-5 criteria, they were considered to be subsyndromal in presentation. The primary diagnosis of delirium was based upon meeting all DSM-5 criteria, and determined by consensus of the expert panel as previously described.(24) Secondary outcome measures included the analysis of those patients assessed as subsyndromal by the consensus panel in addition to those meeting all of the DSM-5 criteria, and the continuous variable of delirium severity derived from the DRS-98R severity measure.

Measures of covariates included the Clinical Dementia Rating (CDR), which was adjudicated as previously described(25); baseline sleep was measured using the Pittsburgh Sleep Quality Index (PSQI) to evaluate subjective nighttime sleep quality over the past month prior to surgery. To measure the previous night’s sleep, the Richards-Campbell Sleep Questionnaire (RCSQ) was administered on POD1 and POD2.(26) Originally developed for critically ill patients, RCSQ is a five-item visual analog scale for the patients to rate their sleep. All patient assessment tools were purchased or used with appropriate permission from copyright holders.

Statistical Analysis

Descriptive analyses (means, standard deviations (SD); counts and percentages) of demographics and other baseline characteristics were calculated separately for each study arm. Analyses of primary outcomes were conducted in accordance with the a priori modified intention-to-treat analytic plan, and included those individuals who were adherent to the first (pre-surgical) dose of study medication. Results from regression models are reported as odds ratio/coefficient, 95% CI, and p-value, to convey the magnitude and direction of the effect, and the degree of uncertainty around the estimate. We also report the z-test value (calculated as coefficient divided by its standard error).

Odds of delirium on POD’s 1 and 2 were modeled via a Markov logistic regression model(27), adjusting for baseline MMSE, and mental status on the previous day, where two transitions were possible: from normal mental status to delirium and/or coma (abnormal mental states) and the reverse. Odds of postoperative delirium upon immediate recovery from anesthesia on POD 0 as a function of treatment assignment were modeled via logistic regression, adjusting for baseline MMSE, surgical duration, and PSQI. Correlation of mental status on consecutive days for the same individual were addressed via the method generalized estimating equations (GEE).(27) The continuous DRS-R-98 severity measure of delirium was modeled via longitudinal linear regression with correlated observations within participants addressed via the method of GEE with treatment assignment as the covariate of interest. The power to detect a difference in the development of postoperative delirium at the previously reported rates(28) (30% placebo-treated vs. 3% ramelteon-treated) was greater than 90%. With 40 participants in each treatment arm and assuming 30% of the placebo-treated group will develop delirium on POD 1 or 2, we would have had 84% power to detect a ramelteon-related reduction in delirium from 30% (placebo arm) to 6% (ramelteon arm).

Adverse events were tabulated by body system, preferred term, and treatment assignment. Risk of a severe adverse event by treatment arm was compared via a Fisher’s exact test due to the small numbers of severe adverse events which occurred. Counts of adverse events were modeled via ordinal logistic regression.

RESULTS

The Consolidated Standard of Reporting Trials (CONSORT) diagram is outlined in Figure 1. Of patients undergoing planned orthopedic procedures during March, 2017 to October 2018, 412 patients were screened and 19% (80/412) were randomized to receive either ramelteon 8 mg or placebo. The most common reason for ineligibility was age (< 65 years old) 16% (66/412). Of the 240 patients eligible for the study, 26% (62/240) declined participation. Total of 5 patients withdrew consent after randomization (placebo group: 1, ramelteon group: 4). Four patients were excluded from the ramelteon group: one became delirious prior to surgery, and 3 had their surgeries cancelled on the day of surgery. All of these patients took the preoperative dose of the study drug. Overall, a total of 71 patients completed the study protocol, with 38 in the placebo group and 33 in the ramelteon group. All participants tested positive for riboflavin (Vitamin B₂) in the preoperative urine samples, confirming adherence to the first dose of the assigned intervention.

Figure 1. Consolidated Standard of Reporting Diagram for RECOVER study.

Patients undergoing elective orthopedic surgeries during March 2017 to October 2018 were recruited.

Baseline Characteristics

Baseline characteristics were similar between the ramelteon and placebo groups (Table 1). More participants underwent knee surgery compared to hip surgery in both groups. All patients stayed at least one night in the hospital following surgery, and none required intensive care unit care after recovery from anesthesia. None were comatose during their hospital stay.

Table 1.

Baseline Characteristics

Demographics	Placebo (n = 39)	Ramelteon (n = 41)
Age, mean (SD), years	75.4 (5.0)	74.3 (5.5)
Sex, No. (%)
Male	14 (37)	14 (42)
Female	24 (63)	19 (58)
Years of Education, No. (%)
1–8 (Elementary)	0 (0)	1 (3)
9–11 (Some HS)	3 (8)	4 (12)
12 (HS or GED)	14 (37)	3 (9)
13–15 (Some College)	7 (18)	4 (12)
16 (College Graduate)	14 (37)	21 (64)
Race, No. (%)
African American	6 (16)	6 (18)
White	32 (84)	27 (82)
Married or Living with Partner, No. (%)	23 (59)	21 (51)
Living at Home, No. (%)	35 (90)	29 (71)
PSMS (ADL), mean (SD)	4.8 (1.2)	3.9 (2.3)
IADL, mean (SD)	6.9 (1.9)	6.0 (3.1)
Charlson Comorbidity Index score, mean (SD)	1.1 (1.1)	1.3 (1.5)
Cognition and Sleep
MMSE, mean (SD)	28.2 (1.9)	28.6 (1.5)
CDR Sum of Boxes, mean (SD)	0.4 (0.9)	0.3 (0.7)
CDR Global, No. (%)
0	31 (82)	27 (82)
0.5	6 (16)	6 (18)
1	1 (3)	0 (0)
DRS-R98 Total, mean (SD)	2.6 (3.0)	2.2 (2.6)
Pittsburg Sleep Quality Index, mean (SD)	6.2 (3.3)	5.7 (3.4)
STOP-Bang, Total, mean (SD)	3.5 (1.5)	3.5 (1.6)
Surgery and Anesthesia a
Joint Site
Knee, No. (%)	25 (66)	19 (58)
Hip, No. (%)	13 (34)	14 (42)
Anesthesia Type
General Anesthesia, No. (%)	7 (18)	8(24)
Spinal Anesthesia, No. (%)	31 (82)	26 (79)
Anesthesia Time, hours, mean (SD)	2.8 (1.3)	3.1 (1.5)
Surgical time, hours, mean (SD)	2.3 (1.2)	2.7 (1.4)

Abbreviations: CDR, Clinical Dementia Rating (modified(25)); DRS-R-98, Delirium Rating Scale-Revised-98(23) (16 items each rated 0 to 3, max. score 46 points; higher scores = greater delirium severity); IADL, Instrumental Activity of Daily Living (Lawton) (47) (8 items assessing of 8 domains of function, max. score = 8; higher scores = greater function); MMSE, Mini-Mental State Examination (17) (10 items, max score = 30 points; No, Number; PSMS, Physical Self-Maintenance Scale(47) (6 items assessing 6 domains of function, max. score = 6; higher score = greater function); PSQI, Pittsburg Sleep Quality Index (48) (19 items combined into 7 component scores 0 to 3, max. score = 21; higher scores = worse sleep quality); STOP-Bang Sleep Apnea Screening Tool, Snoring, Tiredness, Observed apnea and high blood Pressure and Bang: BMI, age, neck circumference, gender (49) (8 items assessing sleep apnea risk, max. score = 8; cut-off ≥3= increased risk for sleep apnea)

^aFor surgical data, the denominators consist of individuals who remained in the study and had surgery/anesthesia – placebo (n=38), ramelteon (n=33)

The overall incidence of delirium during POD 1 and 2 was 7% (5/71). There was no difference in the delirium incidence between the two groups - 5% (2/38) and 9% (3/33) in placebo and ramelteon, respectively. The adjusted odds ratio for post-operative delirium as a function of assignment to the ramelteon treatment arm was 1.28 (95% CI 0.21–7.93; z-value 0.27; p-value=0.79). On the day of surgery following recovery from anesthesia (POD 0), the overall delirium incidence was 14% (10/71), with 8% (3/38) in the placebo group and 21% (7/33) in the ramelteon group (Table 2). The adjusted odds ratio for POD 0 delirium as a function of assignment to the ramelteon treatment arm was 3.91 (95% CI 0.73–20.92; z-value 1.59; p-value=0.11).

Table 2.

Clinical Outcomes

	Placebo (n = 38)	Ramelteon (n = 33)	p-value
Delirium, No. (%)
Postoperative day (POD) 1 and 2	2 (5.3)	3 (9.1)	0.66 a
POD 0: After recovery from anesthesia	3 (7.9)	7 (21.2)	0.17 a
Maximum DRS-R-98 Severity Score:
POD 1 and 2, mean (SD)
Without delirium	3.1 (2.4)	3.5 (2.4)	0.54 b
With delirium	19 (11.3)	19.7 (8.0)	0.56 b
Maximum DRS-R-98 Severity Score:
POD 0 after surgery, POD 1 and 2 mean (SD)
Without delirium	4.3 (2.7)	3.7 (2.1)	0.93 b
With delirium	18.4 (7.8)	20.1 (8.5)	0.56 b
Delirium duration (days) c	1.2 (0.5)	1.3 (0.5)	0.27 b
Discharged before POD 2, No. (%)	26 (68.4)	15 (45.5)	0.06 a
Study participants experiencing AE, No. (%)	13 (33.3)	13 (31.7)	0.53 a
Total number of AE e, No.	18	21	0.95 d
Richards-Campbell Sleep Questionnaire, mean (SD)
POD 0 (night of surgery)	55.5 (25.7)	58.0 (28.4)	6.4 (71) 0.70f
POD 1	37.2 (32.1)	33.9 (24.4)	6.8 (71) 0.63 f

AE, Adverse Event; DRS-R-98, Delirium Rating Scale-Received-98 (23); No, Number; POD, postoperative day; Richards-Campbell Sleep Questionnaire, RCSQ, (26) (5-item sleep questionnaire, evaluating i) depth, ii) latency, iii) number of awakening, iv) efficiency (percent of time awake); v) quality scored on a 100 millimeter visual-analogue scale (VAS), higher score = better sleep; SD, Standard Deviation.

^aFisher’s exact tests.

^bCalculated via Mann-Whitney test due to small counts of those with delirium.

^cFrom POD 0 to POD2.

^dCalculated from ordinal logistic regression of number of AEs on treatment assignment (log OR: 0.03, 95% CI 0.89–0.95)

^eSome patients experienced AE on more than one occasion.

^fstudent’s t-test [t-value (df), pvalue]. Table 2 only contains data from patients who completed the study (placebo n = 38, ramelteon n = 33) as no information could be collected from patients who withdrew consent after randomization (but before surgery) as well as from patients who did not undergo surgery.

The overall incidence of subsyndromal delirium during POD 1 and 2 was 28% (20/71), with 26 % (10/38) in the placebo group and 30% (10/33) in the ramelteon group. The adjusted odds ratio for post-operative delirium including subsyndromal delirium as a function of assignment to the ramelteon treatment arm was 1.20 (95% CI 0.43–3.36; z-value 0.35; p-value= 0.73). On POD 0, following recovery from anesthesia, the overall incidence of subsyndromal delirium was 28% (20/71), with 29% (11/38) in the placebo group and 27% (9/33) in the ramelteon group. The adjusted odds ratio for PACU delirium including subsyndrdomal delirium as a function of assignment to the ramelteon treatment arm was 1.81 (95% CI 0.67–4.93; z-value 1.16; p-value=0.25).

Delirium severity was also assessed using DRS-R-98 severity score. There were no differences in the mean maximum DRS-R-98 scores between the placebo and ramelteon groups. Via linear regression models, assignment to the ramelteon arm was not associated with postoperative delirium severity (β=0.70, 95% CI −0.93–3.44; z-value; p-value=0.40). There were also no differences in delirium duration between the two groups (Table 2). Perceived sleep quality during hospitalization, measured by the Richards-Campbell Sleep Questionnaire (RCSQ), was not significantly different between the two groups (Table 2).

A total of 39 adverse events were reported (Appendix Table 1): 13 participants randomized to placebo reported 18 events, and 13 participants randomized to ramelteon reported 21 events. The most common were nausea (2 in placebo, 5 in ramelteon), hypotension (1 in placebo, 2 in ramelteon), and dizziness (2 in placebo, 1 in ramelteon). Via adjusted ordinal logistic regression the odds ratio for an adverse event as a function of assignment to ramelteon was 1.03 (95% CI 0.31–2.47; z-value 0.06; p=0.95). A total of 3 severe adverse events were reported in 3 different participants, all randomized to ramelteon: bigeminy, hypotension, and a fall. Via Fisher’s exact test, this difference in SAE rate was not statistically significant (p=0.24) nor were any of these events thought to be clinically related to the study intervention. (Appendix Table 1)

DISCUSSION

In our study of an older (≥65) planned orthopedic surgery population, ramelteon was not efficacious in preventing postoperative delirium compared to placebo. We did not demonstrate differences in the delirium incidence between the ramelteon and the placebo groups during POD 1 to 2 as well as including delirium measured on POD 0 after recovery from anesthesia. There were also no differences in the incidence of subsyndromal delirium or delirium severity between the ramelteon and the placebo groups during the same study time period. There were few adverse events in both ramelteon and placebo groups with the most common side effect being nausea.

Our finding is similar to a triple-blind randomized placebo-controlled trial using ramelteon for prevention of intensive care unit (ICU) delirium in adults who underwent elective pulmonary thromboendarterectomy. While higher delirium rates were detected in both placebo (36%) and ramelteon (32.2%) groups with a longer duration of follow-up compared to our study, ramelteon was not efficacious in preventing delirium compared to placebo (relative risk RR, 0.8; 95% CI, 0.5–1.4; p=0.52). A smaller subgroup analysis of individuals ≥65 years old was also not significant.(29) While another triple-blind randomized placebo-controlled trial using ramelteon for prevention of ICU delirium in adults demonstrated reduced incidence in the ramelteon group (24.4%) compared to placebo (46.5%), this trial did not meet the primary outcome of reduced ICU length of stay(30). Another single-blind randomized placebo-controlled study of ramelteon for prevention of delirium in older adults (≥65) admitted to medical wards showed similar findings, with delirium incidence of 32% in placebo group compared to 3% in the ramelteon group.(28)

The low overall incidence of delirium in our study may be due in part to the characteristics of our participants, most of whom had high preoperative cognitive and physical functional levels. However, delirium incidence rates are comparable to other elective orthopedic populations.(31, 32) In addition, there have been many efforts to improve perioperative care through enhanced recovery pathways in orthopedic surgery at our institution over the past 5 years. These include a change in anesthesia administration type, with a minority of orthopedic patients receiving general anesthesia, multimodal pain management with the use of nonsteroidal anti-inflammatory drugs (NSAIDS), acetaminophen, reduced opioid consumption and improved physical rehabilitation which currently routinely begins in the post anesthesia care unit or PACU once the patient has recovered from anesthesia. These interventions contribute to reduced delirium incidence and hospital length of stay.(33)

Several recent systematic reviews and meta-analyses have summarized the effect of melatonin-receptor agonists on delirium.(34–38) Although some of the meta-analyses show beneficial effects of melatonergic drugs on delirium prevention, there are conflicting study outcomes and substantial study heterogeneity.(36) Reasons for variability in study outcomes include heterogeneous precipitating factors and underlying pathophysiology of delirium across different study populations.(39,40) Additionally, factors that influence pharmacokinetics of melatonergic agents may be variable in different populations. A recent systematic review and meta-analysis suggested that in addition to the administration of melatonin one night prior to surgery, additional administration of melatonin closer to the surgery time may be more effective in preventing postoperative delirium.(37) A study involving younger adults reported alterations in melatonin pharmacokinetics after surgery including delayed absorption and clearance as well as reduced peak blood concentration.(41) Some proposed mechanisms include changes in absorption due to restarting food after fasting, neurohormonal and inflammatory changes associated with surgery, factors that affect gastrointestinal motility (e.g. sympathetic, inflammatory factors and opioids), and reduced hepatic clearance.(41) In contrast, a pharmacokinetics study involving two cohorts of ventilated older adults in the medical ICU reported faster absorption, increased peak blood concentration and slightly delayed clearance of melatonin compared to pharmacokinetics reported for healthy volunteers. Exposure of melatonin to greater area of intestinal mucosa due to nasogastric administration as well as lipid content and prokinetic effects of enteral nutrition were proposed as potential mechanisms for altered pharmacokinetics.(42) Therefore, pharmacokinetic variations across different populations may partly explain some of the differences in the outcomes seen in previous studies.(34, 35) Whether it is possible to improve the efficacy of melatonergic drugs preventing delirium in the surgical population by timing the drug administration closer to the surgery and augmenting the postoperative dose remains to be seen.(41)

Pharmacogenomic heterogeneity may also contribute to variable outcomes. A recent study involving cardiac surgery demonstrated that the melatonin receptor 1B gene polymorphism may cause pathological dysfunction of the receptor, with over five-fold increase in delirium incidence in those with the risk genotype.(43) However, we did not examine this variable in our population.

Currently, multicomponent non-pharmacological interventions are recommended for prevention of postoperative delirium, as the evidence for pharmacological treatments is lacking. (44) Additional strategies targeting different mechanisms of postoperative delirium have included controlling sedation depth during anesthesia administration and opioid-sparing strategies.(24, 45, 46) In the future, it would also be important to examine the effect of melatonergic drugs in conjunction with other interventions that target different pathways in delirium pathogenesis.

The limitations of the study include a small sample size at a single-site and a low incidence of delirium which may have limited power to detect a treatment effect. However we evaluated both full and subsyndromal delirium two days after surgery as well as immediately following recovery on the day of surgery, when early symptoms are very prevalent and predictive of subsequent delirium, and consistently found no evidence of a treatment benefit in any of these comparisons. The strengths of the study include conduct of a rigorous, quadruple-masked RCT with a carefully standardized delirium assessment methodology and team successfully deployed in a previous study.(24) The design included a test of participant adherence to self-administered study medication with the addition of riboflavin to the first doses of active agent and placebo and allowing for confirmatory preoperative urine testing. Our findings suggest that contacting the participants the night prior to surgery to take their medication resulted in excellent adherence to treatment intervention.

CONCLUSION

Melatonin receptor-agonist ramelteon was not efficacious in preventing the incidence of postoperative delirium in a population of patients undergoing orthopedic surgery compared to placebo.

Highlights.

• What is the primary question addressed by this study?

  The primary question addressed by the study is whether the melatonin receptor agonist ramelteon is efficacious and safe for prevention of postoperative delirium in older patients undergoing elective orthopedic surgery.

• What is the main finding of this study?

  There was no difference in the delirium incidence between ramelteon versus placebo groups: 9% (3/33) and 5% (2/38), respectively. Adverse events were similar between the two groups.

• What is the meaning of the finding?

  Ramelteon was not more efficacious for prevention of postoperative delirium in older patients undergoing elective orthopedic surgery compared to placebo.