A systematic review of peri-operative melatonin
You can respond to this article at http://www.anaesthesiacorrespondence.com
Summary
We systematically reviewed randomised controlled trials of peri-operative melatonin. We included 24 studies of 1794 participants that reported eight peri-operative outcomes: anxiety; analgesia; sleep quality; oxidative stress; emergence behaviour; anaesthetic requirements; steal induction; and safety. Compared with placebo, melatonin reduced the standardised mean difference (95% CI) pre-operative anxiety score by 0.88 (0.44–1.33) and postoperative pain score by 1.06 (0.23–1.88). The magnitude of effect was unreliable due to substantial statistical heterogeneity, with I2 87% and 94%, respectively. Qualitative reviews suggested the melatonin improved sleep quality and emergence behaviour, and might be capable of reducing oxidative stress and anaesthetic requirements.
Introduction
Evidence-based, multi-modal, surgical and anaesthetic approaches have reduced morbidity and mortality following surgical procedures 1. New peri-operative interventions may further improve recovery. Melatonin is a pineal hormone regulating circadian rhythms in mammals 2. Melatonin has documented effects on sleep disturbances, anxiety and pain 3-7. Melatonin also has anti-inflammatory and anti-oxidative effects 8. Melatonin may therefore be a useful peri-operative drug, particularly as it does not have any known serious adverse effects 9. The analgesic and anxiolytic effects of melatonin in adult surgical patients have been qualitatively reviewed by Yousaf et al. 7. We conducted a systematic review and meta-analysis the efficacy and safety of melatonin for peri-operative patients.
Methods
We searched the PubMed, Embase, CENTRAL, Web of Science and LILACS databases (September 2013) for randomised controlled trials (RCTs) with the terms, ‘melatonin AND (peri-operative period OR postoperative period OR surgery OR surgical OR operation OR surgical procedures OR operative procedures)'. The review was conducted in accordance to the PRISMA guidelines 10. We also searched the reference lists of identified studies. Two authors (LPHA, IG) independently assessed RCTs for inclusion, categorising risk of bias domains and extracting participant and procedure variables, melatonin regimens and outcomes 11. Disagreements were solved by consensus.
We intended to synthesise quantitatively all outcomes, consigning qualitative review to outcomes we were unable to pool through meta-analysis. We did not analyse outcomes reported as side-effects. We applied a random-effects model, combining continuous outcomes as standardised mean differences (SMD) with their associated 95% confidence intervals (95% CI). We transformed median (IQR or range) to mean (SD), excluding transformed data in a sensitivity analysis 11. We quantified heterogeneity of the studies with I2 statistics. All analyses were performed with RevMan (version 5.2; The Nordic Cochrane Centre, Copenhagen, Denmark).
Results
We included 24 RCTs of 1794 participants (Fig. 1) 4-6, 12-32. Table 1 documents the risk of bias, assessed by the Cochrane Risk of Bias assessment tool. Studies reported nine outcomes: pain; anxiety; sleep quality; oxidative stress; emergence behaviour; anaesthetic doses; steal anaesthetic induction (in a sleeping patient); and safety (Table 2).
Bias domain | |||||||||
---|---|---|---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | |
Acil et al. 2004 17 | ? | ? | ? | + | + | ? | + | − | − |
Almenrader et al. 2013 31 | + | + | + | + | + | + | + | + | + |
Borazan et al. 2010 13 | + | + | + | + | ? | ? | + | + | + |
Capuzzo et al. 2006 20 | + | + | + | ? | + | ? | + | + | − |
Caumo et al. 2007 6 | + | + | + | + | ? | ? | + | + | + |
Caumo et al. 2009 5 | + | + | + | + | + | ? | + | + | + |
Evagelidis et al. 2009 30 | + | + | + | + | ? | ? | + | − | − |
Gögenur et al. 2009 16 | + | + | + | + | + | ? | + | + | + |
Haghjooy et al. 2013 28 | + | + | + | + | + | ? | + | − | − |
Ismail & Mowafi 2009 14 | + | + | + | + | ? | ? | + | + | − |
Isik et al. 2008 27 | ? | ? | + | + | ? | ? | + | + | − |
Ionescu et al. 2008 15 | ? | ? | + | + | ? | ? | + | + | − |
Kain et al. 2009 21 | + | + | + | + | ? | ? | + | + | + |
Khezri & Merate 2013 4 | + | + | + | + | + | ? | + | − | + |
Kücükakin et al. 2010 25 | + | + | ? | ? | + | + | + | + | + |
Kücükakin et al. 2010 26 | + | ? | ? | ? | + | − | + | + | + |
Naguib & Samarkandi 1999 18 | ? | ? | + | + | ? | ? | + | + | − |
Naguib & Samarkandi 2000 19 | ? | ? | + | + | ? | ? | + | + | − |
Naguib et al. 2006 24 | + | + | + | + | ? | ? | + | − | − |
Nickkholgh et al. 2011 32 | + | + | + | + | + | + | + | − | + |
Mowafi & Ismail 2008 12 | + | + | ? | + | ? | ? | + | + | − |
Özcengiz et al. 2011 29 | + | + | + | + | ? | ? | + | + | − |
Samarkandi et al. 2005 22 | + | + | + | + | ? | ? | + | + | − |
Turkistani et al. 2007 23 | ? | ? | ? | + | ? | ? | + | − | − |
- 1, random sequence generation; 2, allocation concealment; 3, blinding of participants and personnel; 4, blinding of outcome assessors; 5, incomplete outcome data; 6, selective reporting; 7, other bias; 8, sample size calculation; 9, defined primary outcome; ‘+’, ‘?’, ‘−’, low, unclear and high risks of bias, respectively.
n | Surgery | Anaesthetic | Control | Route | Dose | Timing | Outcomes | |
---|---|---|---|---|---|---|---|---|
Acil et al. 2004 17 | 66 | Laparoscopic cholecystectomy | GA | Midazolam, placebo | SL | 5 mg | 90 min < | Analgesia, anxiety |
Almenrader et al. 2013 31 | 87 | Various | GA | Clonidine | Oral | 5 mg | 60 min < | Steal induction |
Borazan et al. 2010 13 | 52 | Open prostatectomy | GA | Placebo | Oral | 6 mg × 2 | Night < and 60 min < | Analgesia, sleep |
Capuzzo et al. 2006 20 | 138 | Various | GA/Spinal | Placebo | Oral | 10 mg | 90 min < | Analgesia, anxiety |
Caumo et al. 2007 6 | 33 | Open hysterectomy | Epidural | Placebo | Oral | 5 mg × 2 | Night < and 60 min < | Analgesia, anxiety, sleep |
Caumo et al. 2009 5 | 59 | Open hysterectomy | Epidural | Clonidine, placebo | Oral | 5 mg × 2 | Night < and 60 min < | Analgesia, anxiety |
Evagelidis et al. 2009 30 | 71 | Hysteroscopy | GA | Placebo | SL | 0.1, 0.25 or 0.5 mg.kg−1 | 30 min < | Anaesthetic requirement |
Gögenur et al. 2009 16 | 121 | Laparoscopic cholecystectomy | GA | Placebo | Oral | 5 mg | > for three nights | Analgesia, sleep |
Haghjooy et al. 2013 28 | 30 | CABG | GA | Placebo | Oral | 10 mg | For one month daily and 60 min < | Oxidative stress |
Ismail & Mowafi 2009 14 | 40 | Ophthalmological | Topical | Placebo | Oral | 10 mg | 90 min < | Analgesia, anxiety |
Isik et al. 2008 27 | 60 | Dental | Sedation | Midazolam, placebo | Oral | 3 mg or 0.5 mg.kg−1 | 60 min < | Sedation |
Ionescu et al. 2008 15 | 53 | Laparoscopic cholecystectomy | GA | Midazolam, placebo | SL | 3 mg × 2 | Night < and 90 min < | Analgesia, anxiety |
Kain et al. 2009 21 | 148 | Various | GA | Midazolam | Oral | 0.05, 0.2 or 0.4 mg.kg−1 | 45 min < | Anxiety, emergence |
Khezri & Merate 2013 4 | 60 | Ophthalmological | Topical | Placebo | SL | 3 mg | 60 min < | Analgesia, anxiety |
Kücükakin et al. 2010 25 | 41 | Laparoscopic cholecystectomy | GA | Placebo | IV then oral | 10 mg | At incision | Oxidative stress |
Kücükakin et al. 2010 26 | 50 | Major vascular | GA/ Epidural | Placebo | IV then oral | 50 mg then 10 mg | At incision and > for 3 nights | Oxidative stress |
Naguib & Samarkandi 1999 18 | 75 | Laparoscopic gynaecological | GA | Midazolam, placebo | SL | 5 mg | 100 min < | Analgesia, anxiety |
Naguib & Samarkandi 2000 19 | 84 | Laparoscopic gynaecological | GA | Midazolam, placebo | SL | 0.05, 0.1 or 0.2 mg.kg−1 | 100 min < | Analgesia, anxiety |
Naguib et al. 2006 24 | 200 | Various | GA | Placebo | Oral | 0.2 mg.kg−1 | 50 min < | Anaesthetic requirement, anxiety |
Nickkholgh et al. 2011 32 | 36 | Hepatic | GA | Placebo | Oral | 50 mg.kg−1 | After intubation | Safety |
MowafI & Ismail 2008 12 | 40 | Hand | IVRA | Placebo | Oral | 10 mg | 90 min < | Analgesia, anxiety |
Özcengiz et al. 2011 29 | 100 | Oesophageal dilatation | GA | Dex’, midazolam, placebo | Oral | 0.1 mg.kg−1 | 45 min < | Emergence |
Samarkandi et al. 2005 22 | 105 | Various | GA | Midazolam, placebo | Oral | 0.1, 0.25 or 0.5 mg.kg−1 | 60 min < | Anxiety, emergence, sleep |
Turkistani et al. 2007 23 | 45 | Various | GA | Placebo | Oral | 3 or 5 mg | 100 min < | Anaesthetic requirements, anxiety |
- CABG, coronary artery bypass grafting; GA, general anaesthesia; IVRA, intravenous regional analgesia; Dex', dexmedetomidine; SL, sublingual; IV, intravenous; <, before surgery; >, after surgery.
Most studies gave 3–10 mg (0.05–0.5 mg.kg−1) oral melatonin 4-6, 12-16, 20-24, 27-29, 31 or sublingual melatonin 4, 15, 17-19, 25, 30. One study gave 3500 mg orally 32 and one study gave 50 mg intravenously 26. Most studies gave melatonin 30–100 min before surgery 4, 12, 14, 17-24, 27, 29-31. One study gave melatonin daily for one month pre-operatively 28, whereas four studies gave melatonin the night before surgery, in addition to the immediate pre-operative dose 5, 6, 13, 15. Three studies gave melatonin at surgical incision 25, 26 or after intubation 32. One study gave three separate melatonin doses on the first three postoperative nights 16. The controls were inactive placebo 4-6, 12-20, 22, 24-30, 32, clonidine 5, dexmedetomidine 29 and midazolam 15, 17-19, 21, 22, 27, 29. One study included a control group in which patients did not receive a placebo 23. Participants had various operations: abdominal 15-17, 25, 32; dental 27; gynaecological 5, 6, 18, 19, 30; ophthalmological 4, 14; orthopaedic 12; thoracic 28, 29; urological 13; vascular 26; and mixed 20-24, 31. The anaesthetic techniques included: general anaesthesia 13, 15-26, 28-32; intravenous regional anaesthesia 12; neuraxial block 5, 6, 20, 26, 31; sedation 27; and topical analgesia 4, 14.
Pain was investigated by 12 RCTs (n = 821) as scores 16, 17, 20, analgesic requirements 15, or both 4-6, 12-14, 18, 19. Pain scores were documented with a visual analogue scale (VAS) 5, 6, 13, 16-19, verbal score 4, 12, 14 or numerical rating scale (NRS) 20 (Table 3). Melatonin reduced postoperative pain (Fig. 2); the extreme heterogeneity was unaffected by exclusion of RCTs for which we had converted median (IQR or range) to mean (SD) values. In individual RCTs, melatonin either reduced analgesic doses or there was no difference to control (Table 4).
Effect of melatonin on pain score | |
---|---|
Capuzzo et al. 2006 20 | → |
Ismail & Mowafi 2009 14 | ↓ |
Khezri & Merate 2013 4 | → |
Mowafi & Ismail 2008 12 | ↓ |
- →, no statistical difference; ↓, statistically less pain after melatonin vs placebo.
Analgesic dose | ||
---|---|---|
Intra-operative | Postoperative | |
Borazan et al. 2010 13 | ↓ | ↓ |
Caumo et al. 2007 6 | → | ↓ |
Caumo et al. 2009 5 | → | ↓ |
Ionescu et al. 2008 15 | ↓ | |
Ismail & Mowafi 2009 14 | ↓ | |
Khezri & Merate 2013 4 | → | |
Mowafi & Ismail 2008 12 | ↓ | ↓ |
Naguib & Samarkandi 1999 18 | → | → |
Naguib & Samarkandi 2000 19 | → | → |
- →, no statistical difference; ↓, statistically less analgesic after melatonin vs placebo.
Fourteen studies (n = 1146) investigated anxiety before, during and after surgery using four different scales: VAS 4, 12, 14, 17-19, 23, 24; NRS 20; state-trait anxiety inventory 5, 6, 15; Yale pre-operative anxiety scale 21, 22. Melatonin reduced pre-operative anxiety, but the meta-analysis was very heterogeneous, which was partly resolved by exclusion of RCTs for which we had converted median (IQR or range) to mean (SD) values (Fig. 3). In individual RCTs, melatonin either reduced intra-operative and postoperative anxiety or was no different to control (Table 5).
Anxiety score | ||
---|---|---|
Intra-operative | Postoperative | |
Acil et al. 2004 17 | ↓ | |
Capuzzo et al. 2006 20 | → | |
Caumo et al. 2007 6 | ↓ | |
Caumo et al. 2009 5 | ↓ | |
Ionescu et al. 2008 15 | ↓ | |
Ismail & Mowafi 2009 14 | ↓ | |
Khezri & Merate 2013 4 | ↓ | ↓ |
Naguib & Samarkandi 1999 18 | → | |
Naguib & Samarkandi 2000 19 | → |
- →, no statistical difference; ↓, statistically less anxiety after melatonin vs placebo.
Only one study in children intended to record sedation as an outcome (n = 60), and found no difference between melatonin and placebo 27. However, sedation has been registered as an adverse effect in several studies using multiple measuring scales 13, 15, 17-19, 25, 26.
Four RCTs (n = 311) recorded the effect of melatonin on postoperative sleep quality in adults 6, 13, 16 and children 22. Sleep was assessed using sleep questionnaires 6, 13, 16 or accelerography 6. Melatonin improved subjective sleep quality in the early postoperative period 13, 16 and circadian rhythm during the first postoperative week 6. Melatonin reduced sleep disturbance for two postoperative weeks in children 22.
Three RCTs (n = 121) measured concentrations of molecules in adults associated with oxidative stress. Melatonin had no effect on malondialdehyde, ascorbic acid, dehydroascorbic acid or C-reactive protein 25, 26. Melatonin increased intra-operative levels of nuclear factor erythroid 2-related factor 2, suggesting potentiation of the anti-oxidative response 28.
Three RCTs (n = 353) studied the effect of melatonin on behaviour in children emerging from anaesthesia compared with placebo 22, 29, dexmedetomidine 29 and midazolam 21, 22, 29. Behaviour was assessed using the Keegan score 21, a discomfort scale 22 and an agitation scale 29. Melatonin improved behaviour compared with placebo 21, 22, but was no different to dexmedetomidine or midazolam 29.
Three RCTs (n = 316) investigated the effect of melatonin on anaesthetic induction dose 23, 24, 30. The studies used bispectral index 23, 30 or clinical assessment 24. Melatonin reduced the intravenous anaesthetic dose 23, 24 but not the dose of sevoflurane 30.
Melatonin was less effective than clonidine in facilitating successful ‘steal’ induction of anaesthesia in 87 children 31. Melatonin had no effect on complication rates of 36 liver resection patients 32.
Discussion
Compared with placebo, melatonin reduced postoperative pain and pre-operative anxiety. However, we consider these effects to be unreliable due to the profound heterogeneity of the meta-analyses. We were unable to conduct meta-analyses of other outcomes owing to the variability in study design. Qualitative reviews of these outcomes suggested that melatonin improved sleep quality and emergence behaviour and might reduce oxidative stress and anaesthetic requirements.
Most RCTs were vulnerable to bias (Table 1). Only one out of 24 RCTs reported adequate limitation of risk of bias in all methodological domains 32. Sample size calculations were inadequately documented in 7/24 RCTs, increasing the risk of false negative findings (type-2 errors) 4, 17, 23, 24, 28, 30, 32. A primary outcome was not documented in 14/24 RCTs 12, 14, 15, 17-20, 22-24, 27-30. Finally, some RCTs failed to adjust when testing the significance of multiple outcomes, increasing the risk of false positive findings (type-1 errors).
The analgesic effect of melatonin remains uncertain. The reduction in SMD of 1.06 equates to 20 mm in the VAS: as previously commented, the extreme heterogeneity within this analysis makes this pooled effect unreliable. Therefore, additional RCTs would be needed to establish melatonin's clinical effect. Melatonin affects nociception 33, 34. Analgesia may be mediated through peripheral and central effects 35-37. The central action is attributed to specific melatonin receptors (MT1 and MT2) and interaction with opioid, ?-aminobutyric acid (GABA) or N-methyl-D-aspartate receptor systems 35, 36. Peripherally, melatonin has anti-inflammatory actions 37. Any effect of peri-operative melatonin would vary with dosage, administration route and timing. A robust dose-response relationship for melatonin has yet to be demonstrated in humans, which might be provided in non-clinical settings 38, 39.
Peri-operative anxiety is unpleasant and may increase postoperative pain 40. Benzodiazepines are associated with psychomotor impairment, paradoxical psychological reactions, amnesia and respiratory depression, particularly when combined with the residual effects of hypnotics and opioids. These adverse effects may impede early mobilisation and recovery of the patient 41, 42. Melatonin is not associated with these adverse effects. Melatonin reduced pre-operative anxiety by the equivalent of 19 mm on a VAS. However, as with postoperative analgesia, the considerable heterogeneity makes this pooled effect unreliable.
Surgery initially abolishes ‘rapid eye movement’ (REM) sleep, followed by a REM rebound, increased duration of light sleep and reduced slow wave sleep 43. Melatonin induces sleep and shifts the circadian phase 44. However, there is only limited evidence for the sleep-regulating effect of melatonin in surgical patients that might be documented by the use of peri-operative polysomnography.
An excess of oxidising free radicals results in oxidative stress 25, 26. Free radicals can damage cells and is associated with morbidity, e.g. pulmonary oedema 45, 46. Haghjooy et al. demonstrated that prolonged melatonin treatment induces anti-oxidant enzyme activity in patients undergoing coronary artery bypass grafts 28. In infants, anti-oxidative effects of high-dose melatonin administration (up to 10 repeated administrations of 10 mg.kg−1 bodyweight over 72 h) have been correlated to clinical outcomes 8, 47-49. As two RCTs in this meta-analysis did not show any effect of melatonin, these results in infants should not be generalised to adults 25, 26. These findings need to be confirmed in future series.
Confusion on emergence from anaesthesia may last 15–30 min 50-54 and is associated with an increased rate of peri-operative complications and increased staffing costs 52, 54, 55. The aetiology, diagnosis and management of emergence behaviour are still uncertain, making the conclusions concerning the effect of melatonin inconclusive.
The possible reduction in dose of intravenous anaesthetic induction agent by melatonin may be through GABA-A-receptors, comparable with similar properties of benzodiazepines 56-58. ‘Steal induction’ is the induction of anaesthesia in a sleeping child. The rationale is that this might reduce peri-operative stress and anxiety, including subsequent anaesthetics. Melatonin has not been compared with placebo, but appears to be inferior to clonidine 31.
Several studies have demonstrated that melatonin is without the adverse effects associated with opioids, non-steroidal anti-inflammatory drugs1 and benzodiazepines, in both animals and in humans 9, 59, 63. Melatonin was not associated with any serious side-effect in the RCTs included in this systematic review.
In conclusion, melatonin significantly reduced postoperative pain scores and pre-operative anxiety scores. The profound heterogeneity of the meta-analyses, especially with regard to analgesic effect, limits our conclusions, which can only be regarded as preliminary.
Acknowledgements
We thank Janne Wendt, Librarian at Herlev Hospital for her work in relation to the literature search. We thank Tobias Wirenfeldt Klausen, Statistician at Herlev Hospital for meta-analyses. LPHA receives a salary from the University of Copenhagen.
Competing interests
No additional external funding and competing interests declared.