The perioperative use of non-steroidal anti-inflammatory drugs (NSAIDs) is well established as an integral part of pain management after surgery [1-4]. Acting at both the peripheral as well as at the central level, NSAIDs are very effective analgesics and also offer a morphine sparing effect as part of a multimodal approach [5,6]. Tonsillectomy is increasingly managed as short-stay surgery with patients leaving hospital within 24 h. In order to facilitate such an approach, effective management of postoperative pain and nausea becomes crucial not to mention for the sake of patient comfort. The combination of paracetamol and a NSAID offers effective routine analgesia after surgery but opioids are still needed as rescue medication. The emetogenic potential of opioids is however considered as one causative factor in keeping short-stay patients in hospital. Furthermore postoperative vomiting increases the risk of re-bleeding after tonsillectomy due to increased venous pressure at the tonsillar bed. Haemorrhage in the immediate postoperative period is a serious complication following tonsillectomy. Thus the perioperative use of NSAIDs in tonsillectomy is considered as relatively contraindicated due to their platelet inhibiting effects, although this has recently been the focus of debate [7-11]. The newly introduced cyclo-oxygenase-2 (COX-2) selective NSAIDs have no clinically relevant platelet inhibiting effect and have been shown to be equianalgesic with the classical NSAIDs for postoperative pain . As much of the literature has reported on post-tonsillectomy pain in paediatrics, we found it of interest to study the effect on postoperative pain and opioid usage, of premedication with a COX-2 selective NSAID in adult patients scheduled for tonsillectomy.
After obtaining approval from the Regional Ethics Committee and written informed consent, 40 otherwise healthy, adult (>15 yr old) patients scheduled for tonsillectomy were consecutively randomized using a random numbers table  into one of two groups to receive a premedication with either paracetamol and placebo or paracetamol and rofecoxib. Patients were enrolled by one of the anaesthetists at the preassesement clinic. Identical placebo capsules and rofecoxib 50 mg capsules were prepared and packed individually by the hospital pharmacist. Patients were excluded if they had any allergy or intolerance to the trial medications or if they had been taking NSAIDs within a week prior to the study. Further exclusion criteria were chronic pain, a history of liver, kidney or cardiovascular disease, established pregnancy, breast-feeding, re-operation within the first 24 h or inability to understand the protocol as explained.
In the preoperative waiting area, 1.5 h prior to anaesthesia, the patients received either 1.5 g of paracetamol and one placebo capsule (Group P) or 1.5 g of paracetamol and one capsule of rofecoxib 50 mg (Group R) orally. The medications were given by a nurse blinded as to the premedication administered and so were all the involved physicians. Anaesthesia was conducted as per routine of the department. Induction with propofol 1.5-2 mg kg−1, fentanyl 1-2 μg kg−1 and rocuronium 0.4-0.6 mg kg−1 for tracheal intubation, and was maintained with O2 : N2O (1 : 3) and sevoflurane (end tidal concentration 1.6 -2.8%). All patients had ondansetron 4 mg and droperidol 0.5 mg intravenous (i.v.) on induction. Neuromuscular blockade was reversed with neostigmine 2.5 mg and glycopyrrolate 0.4 mg if surgical time was <35 min or if deemed necessary by the attending anaesthesiologist. Thereafter the patient was extubated. The swabs used for haemostasis were weighed before and after surgery and the difference in grams was considered equal to the blood loss in milliliters. Patients were transferred to the recovery area and were given morphine i.v. if the nursing staff deemed it necessary. Repeat doses of 2.5 mg were given until the pain score was <3 or the patient was comfortable. After 30-45 min patients were transferred to the ward. The total dose of morphine at 15 min and the time to first rescue dose of morphine was recorded. Pain scores (visual analog scale (VAS) 1-10) were obtained at 4, 6, 8, 12 and 20 h and if not discharged, at 24 h. The percentage of patients with a pain score >5 within the first 8 h postoperatively (at 4, 6 and 8 h) was calculated. Morphine (2.5 mg increments as according to the hospital protocol) was administered i.v. during the first 24 h when pain (VAS) was above 3-4 and the cumulative dose was recorded at all time points. Paracetamol was administered orally every 6 h. The postoperative need for antiemetics was recorded and ondansetron 1-2 mg or cyclizine 25 mg was administered i.v. at the nursing staff's discretion.
Surgery was performed by one of three surgeons according to a uniform procedure. In the anaesthetized, intubated patient the Boyle-Davis gag was introduced. Tonsillectomy was performed after infiltration of the tonsillar mucosa with local anaesthetic (bupivacaine 0.25% with epinephrine 1 : 400 000 at 0.3 mL kg−1). The mucosa was incised and blunt dissection performed. Haemostasis was achieved by silk ties and/or monopolar diathermy. Dry gauze swabs were used for compression of the tonsillar bed until any visible blood loss had subsided.
Our hypothesis was that premedication with a COX-2 selective NSAID in addition to our routine paracetamol premedication would provide better pain relief in the early postoperative phase and possibly decrease opioid consumption.
In accordance with the objective of the study the primary endpoint was pain and the secondary endpoint was the cumulative postoperative opioid dose. A pre-study power analysis, based on the primary endpoint estimated a number of 21 patients in each group to be required based on the assumption of pain scores in the range of 4-6 and a minimal relevant difference in pain of 2 and a standard deviation (SD) of 2 (obtained from the literature) with two-tailed value of α = 0.05 and a power of 90%. With 21 patients in each group the study would have a power of 80% to detect a difference of 10 mg in opioid consumption. The U-test was used for the comparison of categorical data and Fisher's exact test for comparison of proportions. Non-parametric analysis of variance (ANOVA) for repeated measures (Kruskal-Wallis) was used for comparison of the two groups for treatment and time effects and non-parametric ANOVA (Friedman) supplemented with Dunnett's test for within group differences. Values are given as median and extreme values (minimum-maximum). A value of P < 0.05 was considered statistically significant. The StatView+SE and GraphPad InStat software programs were used for the calculations.
Forty-two patients were allocated and randomized to enter the study. Due to reoperation for haemostasis within the study period one patient (P-group) was excluded from the calculations, and due to the withdrawal of rofecoxib by the manufacturer the very last patient randomized could not enter the study. When the randomization code was broken this patient happened to have been randomized to the R-group leaving the final number of patients in each group at 20. A post hoc power analysis for n = 20 showed that the power of the study for the primary endpoint would be 85% for the actual differences to be detectable. All surgery and anaesthesia was considered as uncomplicated. There were no significant differences between the two groups with respect to age, weight, gender, surgery and anaesthesia time or intraoperative fentanyl dosage (Table 1).
There were no overall significant differences in pain scores between the two groups (Kruskal-Wallis) (Table 2). However 60% of patients in the P-group experienced a pain score >5 within the first 8 h postoperatively as opposed to 25% of patients in the R-group (P = 0.02). There was no difference in the opioid consumption at 15 min. The time to first rescue dose was 7.5 (1-60) min (P-group) and 5.0 (16-90) min (R-group), respectively (P = 0.35). The cumulative opioid consumption increased significantly in both groups within the study period. Four hours postoperatively the P-group had consumed significantly more opioid whereas in the R-group the steady increase in opioid consumption did not reach statistical significance until 12 h postoperatively. This was reflected in significant inter group differences at 6 and 8 h postoperatively. The total cumulative opioid consumption at 24 h was 24.5 (4.0-82.5) mg (P-group) and 12.0 (0-75) mg (R-group), respectively (P = 0.15) (Table 2).
Five patients in the P-group and three patients in the R-group needed a rescue antiemetic within the first 8 h. The numbers were too small for further analysis. A significant (P < 0.02) difference was found for intraoperative blood loss with a median of 36.0 mL in the P-group as opposed to 23.6 mL in the R-group (Table 1).
The present study points to an early (i.e. within the first 8 h) but relatively minor postoperative analgesic benefit of premedication with a COX-2 selective NSAID as compared with paracetamol in adult tonsillectomy. This finding was reflected in a lower opioid consumption within the first 12 h postoperatively. A significantly decreased intraoperative bleeding in the COX-2 group (R-group) was an unexpected finding, albeit of possible theoretical interest.
A reduction in the postoperative use of morphine in order to facilitate fast tracking of the perioperative process is a recent focus of debate [3,14,15]. The use of a multimodal regimen for the control of pain and postoperative nausea and vomiting (PONV) has become the standard of treatment in many centres. The intraoperative use of a local anaesthetic peritonsillar infiltration is favoured by many surgeons often including epinephrine for haemostasis. Only a few studies have addressed this issue in adults and any prolonged pain relieving effect has been questioned [16-18]. Different surgical techniques using blunt or sharp dissection have been discussed in the context of postoperative pain but the results are equivocal .
We found a small but statistically significant early postoperative analgesic effect of rofecoxib, administered 1.5 h preoperatively with 1.5 g of paracetamol. We chose this timing for premedication in accordance with the pharmacokinetic profiles of rofecoxib and oral paracetamol. Our concurrent use of paracetamol may have blunted the sole effect of the NSAID but we found it unethical not to administer our routine baseline analgesic (paracetamol). Whilst we saw no differences in pain scores or opioid consumption in the recovery unit (15 min) an effect of rofecoxib became evident during the first 12 h postoperatively. We ascribe the lower pain scores in the R-group within the first 8 h to an effect of the study drug in agreement with its pharmacokinetic profile. It could be speculated that an early postoperative (4-8 h) analgesic effect of rofecoxib would have implications for the development of late postoperative pain as the COX-2 enzyme is up-regulated as part of the inflammatory response to surgery. This COX-2 induction is thought to be in part responsible for sensitization in the periphery as well as in the central nervous system [1,5,11,20]. It could be criticized that the overall relatively low pain scores (Table 2) in the study may have obscured any major impact of the study drug, as it was recently pointed out by Kalso and colleagues . A relatively small number of patients may have further increased the risk of committing a Type II error.
With respect to any opioid sparing effect the study was possibly too small to show any significant difference in total cumulative opioid consumption between the two groups. However the R-group tended to consume less morphine within the first 12 h. It has been suggested that the opioid sparing effect of NSAIDs is specifically related to morphine and its metabolites and that the use of alternative opioids (e.g. fentanyl) in future studies might disclose a more realistic picture of NSAIDs effects in this respect .
Frequencies (3.3-8.6%) for PONV in adult tonsillectomy have recently been published . Preoperative risk stratification, use of morphine and prophylactic intervention in patients at increased risk is advocated. However, studies in pediatric populations have shown considerably higher figures of about 30% . The present study protocol dictated the intraoperative use of anti-emetics in all patients. Overall 22% needed an anti-emetic within the study period.
Within the first 8 h postoperatively five patients in the P-group and three patients in the R-group were administered an antiemetic. The numbers were too small for further analysis. Our patients were allowed morphine as rescue medication which might explain the somewhat higher figures for PONV in our study as opposed to the incidences reported by Church .
Due to an inhibitory effect on platelet aggregation  and because of a potential for serious postoperative bleeding, the perioperative as well as the postoperative use of NSAIDs in tonsillectomies has traditionally been discouraged [9,26,27]. The figures reported for spontaneous early post-tonsillectomy haemorrhage are relatively small (2-10% in most studies). A strict contraindication to the use of NSAIDs in tonsillectomy has recently been questioned and there seems to be little evidence for an increased intraoperative or early postoperative bleeding with the preoperative administration of a classical NSAID [8,11]. The new generation of NSAIDs with selectivity for the COX-2 enzyme do not affect platelet function [28,29]. As they have been shown to possess equianalgesic potency as compared to the classical NSAIDs as well as conferring an opioid sparing effect in the perioperative setting, they have been suggested to be of potential value in surgery carrying a high bleeding risk [30,31].
We have no explanation for our observation of a significantly lower blood loss in the rofecoxib group. The method of weighing the swabs used for direct haemostasis at the tonsillar bed was recently used and published by others . A change of the balance between thromboxane and prostacycline in the surgical field in favour of thromboxane and thus favouring haemostasis could possibly explain our finding, but the present study was not powered for the observation . A recent meta-analysis of outcomes with the use of preoperative coxibs found no impact on blood loss in seven studies .
The differences in pain scores between our two groups were relatively small and the addition of an NSAID to the analgesic regimen seems, although statistically significant, to confer only limited extra pain relief. In this respect our observations are in agreement with recent studies in children  and in adults . With the present cost of COX-2 selective NSAIDs and the efficacy of paracetamol alone or possibly in combination with a classic NSAID, the implications for clinical practice continues to be a matter of debate. Study endpoints others than pain relief and opioid consumption have been recently been advocated [31,33-36].
In conclusion we have found a relatively small but statistically significant early postoperative analgesic benefit from premedication with rofecoxib in adult tonsillectomy. There was a trend towards a higher consumption of opioids during the first 12 h in the control group but we found no effect on total postoperative morphine consumption. A possible haemostatic effect of the COX-2 selective NSAID was an unexpected finding and a larger scale study focusing on this issue will be required in order to evaluate this observation. In order to optimize further the analgesic approach in adult tonsillectomy patients, a COX-2 selective NSAID is safe and effective.
The staff of the operating theatre, recovery room and the surgical wards at Timaru and Bidwill Trust Hospitals is thanked for their friendly assistance with the study. Dr F. Beswick is thanked for her kind review and comments.
The authors have no financial obligations to the manufacturer of the study drug. The study has not been supported financially by any company or foundation.
1. McCrory CR, Lindahl SGE. Cyclooxygenase inhibition for postoperative analgesia. Anesth Analg
2. Kehlet H. Multimodal approach to control postoperative pathophysiology and rehabilitation. Br J Anaesth
3. Wilmore DW, Kehlet H. Management of patients in fast track surgery. BMJ
4. Dahl V, Raeder JC. Non-opioid postoperative analgesia. Acta Anaesthesiol Scand
5. Cashman JN. The mechanism of action of NSAIDs in analgesia. Drugs
1996; 52(Suppl 5)
6. Kehlet H, Dahl JB. The value of ‘Multimodal’ or ‘Balanced Analgesia’ in postoperative pain
treatment. Anesth Analg
7. Romsing J, Walther-Larsen S. Peri-operative use of nonsteroidal
anti-inflammatory drugs in children: analgesic efficacy and bleeding. Anaesthesia
8. Moiniche S, Romsing J, Dahl JB, Tramèr M. Nonsteroidal
anti-inflammatory drugs and the risk of operative site bleeding after tonsillectomy
: a quantitative systematic review. Anesth Analg
9. Marret E, Flahault A, Samama C-M, Bonnet F. Effects of postoperative, non-steroidal, anti-inflammatory drugs on bleeding risk after tonsillectomy
10. Smith I, Wilde A. Secondary tonsillectomy
haemorrhage and non-steroidal anti-inflammatory drugs. J Laryngol Otol
11. Hiller A, Silvanto M, Savolainen S et al
. Propacetamol and diclofenac alone and in combination for analgesia after elective tonsillectomy
. Acta Anaesthesiol Scand
12. Gajraj NM. Cyclooxygenase-2 inhibitors. Anesth Analg
13. Dawson-Saunders B, Trapp RG. Basic and Clinical Biostatistics
, 2nd edn. New York: Appleton and Lange, 1994.
14. Kehlet H, Rung GW, Callesen T. Postoperative opioid analgesia: time for a reconsideration. J Clin Anesth
15. Goudas LC, Carr DB. Postoperative analgesia - reconsider, don't reject. J Clin Anesth
16. Molliex S, Haond P, Baylot D et al
. Effect of pre- vs postoperative tonsillar infiltration with local anesthetics on postoperative pain
. Acta Anaesthesiol Scand
17. Schoem SR, Watkins GL, Kuhn JJ et al
. Control of early postoperative pain
with bupivacaine in adult local tonsillectomy
. Arch Otolaryngol Head Neck Surg
18. Warnock FF, Lander J. Pain progression, intensity and outcomes following tonsillectomy
19. Akural EI, Koivunen PT, Teppo H et al
pain: a prospective, randomized and double-blind study to compare an ultrasonically activated scalpel technique with the blunt dissection technique. Anaesthesia
20. Hinz B, Brune K. New insights into the physiology and pathophysiological functions of cyclo-oxygenase-2. Curr Opin Anesthesiol
21. Kalso E, Smith L, McQuay HJ, Moore A. No pain, no gain: clinical excellence and scientific rigor - lessons learned from IA morphine. Pain
22. Hyllested M, Jones S, Pedersen JL, Kehlet H. Comparative effect of paracetamol
, NSAIDs or their combination in postoperative pain
management: a qualitative review. Br J Anaesth
23. Church JJ. Is postoperative nausea and vomiting following tonsillectomy
really a problem? Anaesthesia
24. Kotiniemi LH, Ryhänen PT, Valanne J et al
. Postoperative symptoms at home following day-case surgery in children: a multicentre survey of 551 children. Anaesthesia
25. Schafer IA. Effects of nonsteroidal
anti-inflammatory therapy on platelets. Am J Med
26. Gallagher JE, Blauth J, Fornadley JA. Perioperative ketorolac tromethamine and postoperative haemorrhage in cases of tonsillectomy
and adenoidectomy. Laryngoscope
27. Salonen A, Kokki H, Tuovinen K. I.V. ketoprofen for analgesia after tonsillectomy
: a comparison of pre- and post-operative administration. Br J Anaesth
28. Silverman DG, Halaszynski T, Sinatra R et al
does not compromise platelet aggregation during anaesthesia and surgery. Can J Anesth
29. Blaicher AM, Landsteiner HT, Al-Falaki O et al
. Acetylsalicylic acid, diclofenac and lornoxicam, but not rofecoxib
, affect platelet CD 62 expression. Anesth Analg
30. Gilron I, Milne B, Hong M. Cyclooxygenase-2 inhibitors in postoperative pain
31. Reuben SS, Bhopatkar S, Maciolek H et al
. The preemptive analgesic effect of rofecoxib
after ambulatory arthroscopic knee surgery. Anesth Analg
32. FitzGerald GA, Patrono C. The coxibs, selective inhibitors of cyclooxygenase-2. New Engl J Med
33. Straube S, Derry HJ, McQuay H et al
. Effect of preoperative COX-II selective NSAIDs (coxibs) on postoperative outcomes: a systematic review of randomized studies. Acta Anaesthesiol Scand
34. Pickering AE, Bridge HS, Nolan J et al
. Double-blind, placebo-controlled analgesic study of ibuprofen or rofecoxib
in combination with paracetamol
in children. Br J Anaesth
35. Watcha MF, Issioui T, Klein KW, White PF. Costs and effectiveness of rofecoxib
, celecoxib, and acetaminophen for preventing pain after ambulatory otolaryngologic surgery. Anesth Analg
36. White PF. Changing Role of COX-2 Inhibitors in the perioperative period: is parecoxib really the answer? Anesth Analg