Shivering is an unpleasant and frequent complication in the postoperative period (1). The origin of postoperative shivering is unclear; various mechanisms have been proposed. Shivering may happen as a thermoregulatory response to hypothermia or muscle hyperactivity with clonic or tonic patterns; and different frequencies have been reported (1). However, in the postoperative period, muscle activity may be increased even with normothermia (2), suggesting that other mechanisms than heat loss and subsequent decrease in core temperature may contribute to the development of shivering. These include uninhibited spinal reflexes, postoperative pain, decreased sympathetic activity, pyrogen release, adrenal suppression, and respiratory alkalosis (1).
In a survey on 33 clinical problems, anesthesiologists ranked postoperative shivering 8th when its frequency was considered and 21st when asked about the importance of preventing this complication (3). This suggests that most anesthesiologists do not consider shivering to be a true medical problem. However, in a shivering patient, oxygen consumption may increase by 200% to 500%(4,5). Also, hypothermia may trigger vasoconstriction and thus increase vascular resistance. Thus, in a patient with already limited myocardial oxygen supply because of arteriosclerosis, shivering may further compromise myocardial function. Shivering may also increase intraocular and intracranial pressure, and it may contribute to increased wound pain (1).
Numerous pharmacological interventions have been proposed for the treatment of postoperative shivering. Textbooks suggest that apart from applying radiant heat to the body surface, shivering may be treated with meperidine, clonidine, or ketanserin (1). The relative efficacy of these different molecules, however, remains unclear. The aim of this systematic review was to evaluate efficacy and harm of pharmacological interventions that are currently used for the treatment of postoperative shivering.
Inclusion and Exclusion Criteria
Relevant studies were full reports of randomized comparisons of any pharmacological antishivering intervention (active) compared with placebo or no treatment (control) in postoperative, nonventilated patients. Studies in any language were considered. Data on the efficacy of prevention of shivering or on nonpharmacological interventions were not included. We did not consider data from retrospective analyses, from studies without randomization, or from abstracts. Studies with group sizes <10 patients were excluded.
Two authors (PK and MRT) independently searched the MEDLINE (http://www.nlm.nih.gov), EMBASE (http://stneasy.fiz-karlsruhe.de and DataStar®) and Cochrane Controlled Trials Register (http://www. cochrane.de) databases using different search strategies. Free text words used were “postoperative OR postanesthetic OR postanaesthetic” AND “shivering OR shaking OR tremor” AND “randomized OR randomised.”
The date of the last electronic search was August 11, 2000. Reference lists of retrieved reports and of relevant review articles (6,7) were screened. Locally available anesthesia journals were hand searched. The German manufacturers of pethidine (meperidine) (Aventis Pharma) and clonidine (Boehringer Ingelheim) were contacted by letter and asked for additional information, including unpublished data.
Critical Appraisal and Data Extraction
All retrieved reports were screened by one author (PK). Irrelevant data were excluded at that stage. Each potentially relevant report was then read by at least two other authors independently to assess adequacy of randomization and blinding and description of withdrawals according to the validated three-item, five-point Oxford scale (8). We assigned one point to studies described as being “randomized,” and an additional point if the method of randomization was described and adequate (for instance a table of random numbers). One point was assigned when the trial was described as “double-blind,” and an additional point was assigned if the method of double blinding was described and adequate (for instance, identical ampoules). Finally, reports that described the number of and reasons for withdrawals were given one point. Because we included only randomized trials, the minimum score of a valid trial was one and the maximum score was five.
Relevant efficacy data were extracted by one author (PK) and checked by another author (LHE). Numerous different efficacy endpoints were reported in these trials. To avoid unnecessary heterogeneity of the data and to reduce the risk of interpretational bias, there was a pre hoc decision to extract only efficacy data on complete absence of shivering after application of the study drugs. When there were repeated drug administrations at different time intervals we calculated the respective cumulative drug doses for each time interval. Variable doses (for instance, milligrams per kilogram of body weight) were extrapolated to fixed doses using the average body weight of the study populations as reported in the original trials. Consensus on both quality scores and extracted data was reached by discussion.
As an estimate of the statistical significance of a difference between active and control, we calculated relative risks (RR) with 95% confidence intervals (CI) (9). Because all combined data were clinically and statistically homogenous (P > 0.1) we used a fixed effect model throughout. A statistically significant difference between active and control was assumed when the 95% CI of the RR did not include 1. As an estimate of the clinical relevance of any difference between active and control we calculated the number-needed-to-treat (NNT) (10) with 95% CI (11). Because control event rates were similar across different trials, we graphically plotted NNTs and 95% CI for individual treatments to test for relative efficacy (12). Data on adverse drug reactions were extracted when they were reported in dichotomous form, and they were analyzed as for efficacy data. Analyses were done using Excel on a Power Macintosh G3 and checked in RevMan 4.0 provided as a freeware by the Cochrane Collaboration (http://www.cochrane.org/) (13).
Excluded and Included Reports
We retrieved 41 potentially relevant reports (Fig. 1). Twenty-one were later excluded; 3 evaluated the effect of a muscle relaxant in ventilated patients (14–16), 11 had no placebo or “no treatment” control (5,17–26), 5 were not randomized or the randomization process was unclear (27–31), in 1, group size was <10 patients (32), and 1 did not report on dichotomous efficacy data (33).
We eventually analyzed data from 20 randomized controlled trials published between 1984 and 2000 (34–53) (Table 1). Pharmaceutical companies did not provide relevant additional data. In those trials, 944 adults received an active intervention, and 413 controls received a placebo. The median size of active and control groups was 18 patients (range, 9–37) and 20 patients (range, 10–32), respectively. The median Oxford scale was 2.5 (range, 1–4). One report scored 1, 9 scored 2, 6 scored 3, and 4 scored 4. In five trials, efficacy data had to be extrapolated from graphs (34,39,42,49,50). There were two dose-finding studies (37,53). In one trial, patients’ average body temperature was between 34°C and 35°C (41); in all other trials that reported on body temperature, it was above 35°C.
The active interventions were opioids (morphine, fentanyl, alfentanil, pethidine [meperidine], nalbuphine, pentazocine), other centrally acting analgesics (tramadol, metamizol, nefopam), sodium channel blocker (lidocaine), α2-agonist (clonidine), methylphenidate, doxapram, ketanserin, and magnesium.
The observation times were between 1 min and 45 min. There was a direct relationship between the length of the observation period and the percentage of control patients who shivered after application of a placebo (i.e., the control event rate) (Fig. 2). Because we have to assume that there is some relationship between the control event rate and the true underlying risk, we tested the impact of the observation period on the antishivering efficacy of active interventions. For three drugs, there was an adequate number of data for one dose to test for this potential relationship; this was with ketanserin 10 mg (36,39,45), meperidine 25 mg (42,44,46,49,51–53), and clonidine 150 μg (39,41,43,47). With all three regimens, the RR for further shivering compared with placebo decreased over time, i.e., the antishivering efficacy decreased with increasing length of the observation period (Fig. 3).
Many regimens were tested in one small trial only. Because undue weight could be given to these interventions, an arbitrary decision was made to estimate antishivering efficacy for only those interventions that were tested in at least two trials. These were meperidine 25 mg (42,44,49,51–53), clonidine 150 μg (39,41,43,47), doxapram 100 mg (48,49,53), ketanserin 10 mg (36,39,45), and alfentanil 250 μg (42,52). Because antishivering efficacy obviously depended on the length of the observation time, data pooling, and estimation of the relative efficacy of different interventions was performed only within similar observation periods (Table 2, Fig. 4). Within time periods, all combined efficacy data were statistically homogenous (P > 0.1).
All these interventions were statistically significantly more effective than placebo at all time points, except for ketanserin 10 mg after 30 min. Meperidine 25 mg was most often tested and performed consistently best at 1, 5, 10, and 15 min. After one min, the NNT to prevent any further shivering with meperidine 25 mg compared with placebo was approximately 3, suggesting that it acted faster than alfentanil 250 μg or doxapram 100 mg. After 5, 10, and 15 min, the NNTs for meperidine 25 mg were less than 2. Some of the other interventions came close to this; clonidine 150 μg and doxapram 100 mg, for instance, had NNTs less than 2 at 5 min but not thereafter.
Regimens of fentanyl (34,46), lidocaine (34), magnesium (40), metamizol (44), methylphenidate (41), morphine (46), nalbuphine (51), nefopam (38), pentazocine (50), and tramadol (37) were tested in one trial each with a limited number of patients. No further conclusions could be drawn in addition those provided by the individual papers.
Three trials reported on nausea or vomiting with meperidine 25 mg (46), 50 mg (43) or 0.4 mg/kg (44). In those, 6 of 82 patients (7.3%) had nausea or vomited with meperidine compared with 2 of 76 (2.6%) with placebo, a difference that was not statistically significant (RR, 2.84; 95% CI, 0.60–13.5). Two trials reported on respiratory depression with meperidine (defined as “bradypnoe and pulse oximetry <85%”(34) or “bradypnoe not needing a specific treatment”(44)). Using these criteria, 2 of 48 patients (4.2%) had bradypnoe with meperidine compared with 0 of 45 with placebo, a difference that was not statistically significant (RR, 2.87; 95% CI, 0.31–26.4). Other adverse effects, reported in no more than one trial, were nausea or vomiting with fentanyl (34), and bradycardia with clonidine or ketanserin (39).
Shivering, as nausea or vomiting, never becomes chronic and it is unlikely to kill a patient. However, in shivering postoperative patients, left ventricular systolic work index and oxygen consumption index may be increased (54). It is, therefore, encouraging that some simple and inexpensive interventions are effective in the treatment of this adverse effect of anesthesia and surgery. Two shivering patients need to be treated with meperidine 25 mg, clonidine 150 μg, or doxapram 100 mg for one to stop shivering within five minutes who would have continued to shiver had they all received a placebo. This degree of efficacy relates to the “average” shivering adult patient in the postoperative period. Most often, shivering was diagnosed clinically; the role of body temperature remained unclear. As these trials used randomized treatment allocation, differences in body temperature (core or peripheral) were unlikely to have an effect on the overall efficacy. In none of the trials, a stratified (factorial) design with efficacy evaluation in normothermic versus hypothermic patients was applied. Average body temperatures were comparable across the studies, and no profound hypothermia was reported. We do not know, if these antishivering interventions are more (or less) effective in specific subgroups of patients, for instance in those who are hypothermic.
There were several methodological problems with this analysis. First, we had to rely on indirect comparisons (i.e., comparisons from placebo-controlled trials). This was necessary because, for shivering, as for many other clinical settings, a “gold standard” treatment is unknown. As a consequence, trialists do not know what standard comparator they should choose to test the efficacy of a new, experimental intervention. We retrieved many potentially valid randomized trials that compared an experimental intervention with a randomly chosen comparator. Combining these heterogeneous data was impossible. Active comparisons, however, may be used to test the validity of the league table of relative antishivering efficacy (Fig. 4). For instance, in a recent direct comparison, there was equivalence between meperidine 25 mg and clonidine 150 μg (55), and the league table suggested the same.
Second, as in previous similar analyses (56,57), we had to choose a dichotomous efficacy end point to overcome the large variability of the different reported end points. The only way to reduce both the risk of interpretation bias and heterogeneity of the data was to define a clearly dichotomous hurdle of efficacy, for instance, absence of any shivering after treatment. Such a hurdle may be unnecessarily high; a treatment that does not stop shivering but that alleviates symptoms to a great extent may be clinically useful. However, a dichotomous end point allows combination of data from independent trials, and thus increases power.
Third, when analyzing the control event rates (i.e., the incidence of shivering in placebo patients) it became obvious that the length of the observation period was a factor (Fig. 3). In such circumstances it was invalid to combine efficacy data that originated from different observation times. The relative efficacy of the different experimental interventions was, therefore, compared within similar observation periods only, and indeed the data within these periods were statistically homogenous. Because observation times were inconsistently reported and because many interventions were tested in no more than 1 trial each, only 5 regimens could be compared: meperidine 25 mg, doxapram 100 mg, clonidine 150 μg, ketanserin 10 mg, and alfentanil 250 μg after 5 minutes. One drug only was tested up to 30 minutes; this was ketanserin 10 mg, and it was not significantly different from placebo. Thus, the information on the relative efficacy of these drugs remains limited.
Reporting of adverse drug reactions was sparse. The problem then is that we do not know if adverse drug reactions did not occur, or if they were not reported. Typically opioid-related side effects, such as nausea, vomiting or respiratory depression, seemed to be rare with relatively small and single doses of morphine, fentanyl, alfentanil, meperidine, nalbuphine, or pentazocine. The absence of bradycardia and hypotension with clonidine up to 150 μg is less obvious; it may be cautious to titrate the clonidine.
The fact that simple regimens such as meperidine 25 mg are very effective in treating shivering patients begs the question as to whether more trials to investigate the efficacy of yet other drugs that may have some antishivering efficacy are actually needed. However, there were some weaknesses in almost all these trials; this may justify the definition of a research agenda. For instance, there was a lack of data on long-term outcome. Many trials stopped observations after only 5 or 10 minutes. Such short observation periods may not be long enough to identify the true efficacy of these interventions and their long-term effect on patients. Also, data on the efficacy of these drugs in children are lacking. Finally, because the time course plays a crucial role in this setting, it seems reasonable to report the time to the first treatment application to allow for meaningful comparisons between studies.
We thank Daniel Haake from the Medical libraries of the Center Medical Universitaire, Geneva University, Geneva, for his help in searching electronic databases. We are grateful to Keinosuke Nara, MD, PhD, Alexander Stetsenko, MD, and Greta Poglia, MD, who translated Japanese, Russian, and Italian papers.
1. Sessler DI. Temperature monitoring. In: Miller RD, ed. Anesthesia. New York, Edinburgh, London, Madrid, Melbourne, Milan, Tokyo: Churchill Livingstone, 1994: 1363–82.
2. Horn EP, Sessler DI, Standl T, et al. Non-thermoregulatory shivering in patients recovering from isoflurane or desflurane anesthesia. Anesthesiology 1998; 89: 878–86.
3. Macario A, Weinger M, Truong P, Lee M. Which clinical anesthesia outcomes are both common and important to avoid? The perspective of a panel of expert anesthesiologists. Anesth Analg 1999; 88: 1085–91.
4. Bay J, Nunn JF, Prys-Roberts C. Factors influencing arterial PO2 during recovery from anaesthesia. Br J Anaesth 1968; 40: 398–407.
5. Macintyre PE, Pavlin EG, Dwersteg JF. Effect of meperidine on oxygen consumption, carbon dioxide production, and respiratory gas exchange in postanesthesia shivering. Anesth Analg 1987; 66: 751–5.
6. Prys-Roberts C. Postanesthetic shivering. Clin Anesth 1968; 3: 358–69.
7. Buggy DJ, Crossley AW. Thermoregulation, mild perioperative hypothermia and postanaesthetic shivering. Br J Anaesth 2000; 84: 615–28.
8. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17: 1–12.
9. Morris JA, Gardner MJ. Calculating confidence intervals for relative risk, odds ratio, and standardised ratios and rates. In: Altman DG, Gardner MJ, eds. Statistics with confidence: confidence intervals and statistical guidelines. London: BMJ Books, 1995: 50–63.
10. Laupacis A, Sackett DL, Roberts RS. An assessment of clinically useful measures of the consequences of treatment. N Engl J Med 1988; 318: 1728–33.
11. Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment effect. BMJ 1995; 310: 452–4.
12. Altman DG. Confidence intervals for the number needed to treat. BMJ 1998; 317: 1309–12.
13. Review Manager (RevMan), Version 4.0 for Windows. Oxford: The Cochrane Collaboration, 1999.
14. Cruise C, MacKinnon J, Tough J, Houston P. Comparison of meperidine and pancuronium for the treatment of shivering after cardiac surgery. Can J Anaesth 1992; 39: 563–8.
15. Dupuis JY, Nathan HJ, DeLima L, et al. Pancuronium or vecuronium for treatment of shivering after cardiac surgery. Anesth Analg 1994; 79: 472–81.
16. Sladen RN, Berend JZ, Fassero JS, Zehnder EB. Comparison of vecuronium and meperidine on the clinical and metabolic effects of shivering after hypothermic cardiopulmonary bypass. J Cardiothorac Vasc Anesth 1995; 9: 147–53.
17. Brichard G, Johnstone M. The effect of methylphenidate (Ritalin) on post-halothane muscular spasticity. Br J Anaesth 1970; 42: 718–22.
18. Götz E, Bogosyan S, Müller E, Litz R. Treatment of postoperative shivering with nalbuphine. Anasthesiol Intensivmed Notfallmed Schmerzther 1995; 30: 28–31.
19. Heffline MS. A comparative study of pharmacological versus nursing interventions in the treatment of postanesthesia shivering. J Post Anesth Nurs 1991; 6: 311–20.
20. Izzo V, Mariconti P, Tiengo M. Action and effectiveness of nefopam chloride in the control of postoperative shivering. Minerva Anestesiol 1991; 57: 760–2.
21. Pagani I, Ramaioli F, Marra M, Benegiamo M. The use of doxapram chlorohydrate (t-doxapril) as an anti-shivering drug. Minerva Anestesiol 1980; 46: 653–8.
22. Pausawasdi S, Jirasirithum S, Phanarai C. The use of tramadol hydrochloride in the treatment of post-anesthetic shivering. J Med Assoc Thai 1990; 73: 16–20.
23. Rosa G, Pinto G, Orsi P, et al. Control of post anaesthetic shivering with nefopam hydrochloride in mildly hypothermic patients after neurosurgery. Acta Anaesthesiol Scand 1995; 39: 90–5.
24. Saltanov AI, Boshkoev ZB, Kadyrova EG. Analgesia and treatment of muscular shivering during the period of postanesthetic adaptation in oncologic patients. Anesteziol Reanimatol 1998; 32–5.
25. Salvi L, Susini G, Ceriani R, et al. Effectiveness and hemodynamic effects of clonidine, meperidine, taurine, ketanserin in the therapy of post-anesthesia shivering in patients undergoing ECC in moderate hypothermia. Minerva Anestesiol 1991; 57: 756–7.
26. Miyakawa H, Matsumoto K, Matsumoto S, et al. A comparison of three drugs (pethidine, magnesium sulfate and droperidol) in patients with post-anesthesia shivering. Masui 1991; 40: 1503–6.
27. Beliaev AV, Ryzhy SM, Gubysh VP. The efficacy of the use of calcium chloride for preventing postoperative shivering. Klin Khir 1990; 6–7.
28. Beliaev AV, Ryzhy SM, Dubov AM. Use of magnesium sulfate for controlling postoperative shivering. Klin Khir 1991; 42–4.
29. Tempia A, Livigni S, Castino PM, et al. The use of nefopam in the prophylaxis and treatment of postoperative shivering. Minerva Anestesiol 1992; 58: 547–51.
30. Vogelsang J, Hayes SR. Butorphanol tartrate (Stadol) relieves postanesthesia shaking more effectively than meperidine (Demerol) or morphine. J Post Anesth Nurs 1992; 7: 94–100.
31. Vogelsang J. The treatment of postanesthesia shaking. AORN J 1993; 57: 1449–56.
32. Güzeldemir ME, Türkan H, Bayhan N, et al. Intravenous clonidine for postanaesthetic shivering complication. Anest ve Rean Cem Mecmuasi 1993; 21: 357–60.
33. Pizzirani E, Clemente E, Rizzoli G, et al. The therapy of postoperative shivering: comparison of drugs. Minerva Anestesiol 1991; 57: 758–9.
34. Alfonsi P, Hongnat JM, Lebrault C, Chauvin M. The effects of pethidine, fentanyl and lignocaine on postanaesthetic shivering. Anaesthesia 1995; 50: 214–7.
35. Capogna G, Celleno D. IV clonidine for post-extradural shivering in parturients: a preliminary study. Br J Anaesth 1993; 71: 294–5.
36. Crisinel D, Bissonnette B, Feihl F, Gardaz JP. Efficacy of ketanserin on postanesthetic shivering. Ann Fr Anesth Reanim 1997; 16: 120–5.
37. De Witte J, Deloof T, De Veylder J, Housmans PR. Tramadol in the treatment of postanesthetic shivering. Acta Anaesthesiol Scand 1997; 41: 506–10.
38. Ferri L, Bertelli G, Innocenti S, et al. The treatment of postoperative shivering with nefopam hydrochloride. Minerva Anestesiol 1993; 59: 317–20.
39. Joris J, Banache M, Bonnet F, et al. Clonidine and ketanserin both are effective treatment for postanesthetic shivering. Anesthesiology 1993; 79: 532–9.
40. Kizilirmak S, Karakas SE, Akca O, et al. Magnesium sulfate stops postanesthetic shivering. Ann N Y Acad Sci 1997; 813: 799–806.
41. Launo C, Palermo S, Germi MR, et al. Clonidine and postoperative shivering. Minerva Anestesiol 1991; 57: 427–31.
42. Lyons B, Carroll M, McDonald NJ. The treatment of postanaesthetic shivering: a double blind comparison between alfentanil and pethidine. Acta Anaesthesiol Scand 1995; 39: 979–82.
43. Mercadante S, De Michele P, Letterio G, et al. Effect of clonidine on postpartum shivering after epidural analgesia: a randomized, controlled, double-blind study. J Pain Symptom Manage 1994; 9: 294–7.
44. Monso A, Riudeubas J, Barbal F, et al. A randomized, double-blind, placebo-controlled trial comparing pethidine to metamizol for treatment of post-anaesthetic shivering. Br J Clin Pharmacol 1996; 42: 307–11.
45. Nalda MA, Gomar C, Luis M. The effect of ketanserin on post-anaesthetic vasoconstriction and shivering. Eur J Anaesthesiol 1985; 2: 265–77.
46. Pauca AL, Savage RT, Simpson S, Roy RC. Effect of pethidine, fentanyl and morphine on post-operative shivering in man. Acta Anaesthesiol Scand 1984; 28: 138–43.
47. Pern U, Tryba M, Zenz M, Tegenthoff H. Neue möglichkeiten in der therapie des shivering. In: Klose R, Büttner J, eds. Clonidin in anästhesie und intensivmedizin. Nürnberg: PERIMED-spitta, 1993: 53–60.
48. Sarma V, Fry EN. Doxapram after general anaesthesia: its role in stopping shivering during recovery. Anaesthesia 1991; 46: 460–1.
49. Singh P, Dimitriou V, Mahajan RP, Crossley AW. Double-blind comparison between doxapram and pethidine in the treatment of postanaesthetic shivering. Br J Anaesth 1993; 71: 685–8.
50. Terasako K, Yamamoto M. Comparison between pentazocine, pethidine and placebo in the treatment of post-anesthetic shivering. Acta Anaesthesiol Scand 2000; 44: 311–2.
51. Wang JJ, Ho ST, Lee SC, Liu YC. A comparison among nalbuphine, meperidine, and placebo for treating postanesthetic shivering. Anesth Analg 1999; 88: 686–9.
52. Wrench IJ, Cavill G, Ward JE, Crossley AW. Comparison between alfentanil, pethidine and placebo in the treatment of post-anaesthetic shivering. Br J Anaesth 1997; 79: 541–2.
53. Wrench IJ, Singh P, Dennis AR, et al. The minimum effective doses of pethidine and doxapram in the treatment of post-anaesthetic shivering. Anaesthesia 1997; 52: 32–6.
54. Bilotta F, Pietropaoli P, La Rosa I, et al. Effects of shivering prevention on haemodynamic and metabolic demands in hypothermic postoperative neurosurgical patients. Anaesthesia 2001; 56: 514–9.
55. Schwarzkopf KR, Hoff H, Hartmann M, Fritz HG. A comparison between meperidine, clonidine and urapidil in the treatment of postanesthetic shivering. Anesth Analg 2001; 92: 257–60.
56. Picard P, Tramèr MR. Prevention of pain on injection with propofol: a quantitative systematic review. Anesth Analg 2000; 90: 963–9.
57. Eberhart LH, Morin AM, Bothner U, Georgieff M. Droperidol and 5-HT3-receptor antagonists, alone or in combination, for prophylaxis of postoperative nausea and vomiting. A meta-analysis of randomised controlled trials. Acta Anaesthesiol Scand 2000; 44: 1252–7.