Significant complications, including protracted vomiting, dehydration, airway obstruction, bleeding, and unanticipated admission or readmission to hospital, can occur after tonsillectomy [1-8]. Because of associated somnolence, respiratory depression, and nausea, opioids may contribute to morbidity after tonsillectomy . Ketorolac, a nonsteroidal antiinflammatory drug, can provide analgesia comparable to morphine [9-15] and causes less drowsiness , respiratory depression , and vomiting [13,15]. However, ketorolac depresses platelet function and increases bleeding time by 1-3 min [16-19]. Ketorolac given before tonsillectomy increases intraoperative blood loss [14,19]; the hemostatic effects of ketorolac given after tonsillectomy have not been reported. Given the analgesic equivalence and side effects of ketorolac and morphine, we hypothesized that ketorolac given after tonsillectomy would result in faster recovery and discharge, decreased vomiting, and fewer unanticipated admissions without increasing the incidence of posttonsillectomy hemorrhage.
After institutional review board approval and informed, parental consent, children aged 1-12 yr presenting to Children's Hospital Medical Center (CHMC) or CHMC-Outpatient North (OPN) for tonsillectomy performed by, or under the direct supervision of, one of the two otolaryngologist investigators were randomized to receive either morphine 0.1 mg/kg or ketorolac 1 mg/kg intravenously (IV) after tonsillectomy. Children were excluded from recruitment for significant underlying medical conditions, airway abnormalities (except isolated obstructive sleep apnea), asthma, bleeding disorders or a family history thereof, or allergy to morphine, ketorolac, or aspirin. Randomization by computer-generated random numbers was stratified by location of surgery (CHMC or OPN) and outpatient/same day admit status. Study drugs were prepared by a pharmacist (CHMC) or an anesthesiologist not involved in the care of the subject (OPN). Subjects, families, anesthesiologists, nurses, and investigators were blinded to study drug assignment.
The unpremedicated subjects were taken with one parent to the induction room, where anesthesia was induced by inhalation of halothane in 66% nitrous oxide/33% oxygen; subjects were then transported to the adjacent operating room where an IV catheter was inserted, and the trachea was intubated without muscle relaxants at a deep level of anesthesia. Subjects then received metoclopramide 0.2 mg/kg and ampicillin 25 mg/kg IV. Anesthesia was maintained by spontaneous inhalation of halothane and nitrous oxide; the anesthetist was instructed to maintain the subject at a moderately deep level of anesthesia.
Tonsillectomy was performed by electrocautery dissection in all subjects; adenoidectomy, if performed, was by curettage or electrocoagulation. When surgery was complete and hemostasis was achieved, the study drug was administered as a bolus and the trachea extubated at a deep level of halothane anesthesia. Subjects then breathed 100% oxygen by face mask until able to maintain a patent airway without application of positive pressure, at which time they were transported to the postanesthesia care unit (PACU).
Subjects were given acetaminophen (15-20 mg/kg per rectum) on arrival to PACU. Supplemental oxygen was administered until subjects maintained an oxygen saturation >or=to95% on room air. Subjects who complained of pain, or who were thought to be in pain by their PACU nurse, received up to two doses of morphine 0.05 mg/kg IV at the discretion of their anesthesiologist. Subjects received total IV fluids equal to at least 8 h of maintenance and were discharged from PACU to outpatient surgery (OPS) or to the ward when they met standard CHMC PACU discharge criteria (awake, hemodynamically stable, able to maintain a patent airway without assistance, protective reflexes intact, comfortable, and free of persistent adenotonsillar bleeding). Outpatient subjects were required to maintain a room air saturation >or=to95% prior to discharge to OPS. Same day admit subjects were discharged to the ward on supplemental oxygen, if needed, to maintain an oxygen saturation >or=to95%. Subjects were discharged home when they met standard OPS discharge criteria (as above for PACU discharge, but also fully awake and alert and completely recovered from anesthesia); although subjects were encouraged to take fluids by mouth prior to discharge, they were not required to do so, provided they were adequately hydrated and free of persistent vomiting. Same day admit subjects were monitored with pulse oximetry the first night after surgery.
Age, weight, ASA physical status, diagnosis, surgical procedure, obstructive sleep apnea (OSA) score , and tonsil size [0+ to 4+, ] were recorded prior to surgery. During surgery, anesthesia induction time, surgeon's subjective assessment of intraoperative bleeding (minimal, little surgical bleeding; moderate, modest surgical bleeding, easily controlled; excessive, brisk surgical bleeding, difficult to control), and time of deep extubation were recorded. Blood loss was not measured, as the surgical technique was typically associated with <10-20 mL of blood loss. In the PACU, duration of oxygen requirement, supplemental analgesic requirements, emesis, bleeding, and time to readiness for discharge were recorded. In OPS, emesis, bleeding, and time to readiness for discharge home were recorded. Times were measured from extubation. On the ward, emesis, duration of oxygen requirement, and bleeding were recorded. The subjects' parents were contacted the day after surgery and again 2 wk later and emesis (24-h call only), bleeding, and any return to clinic or readmission to hospital were recorded. Bleeding episodes were scored for severity: 0 = none; 1 = self-limited; 2 = return to office; 3 = return to emergency room; 4 = admitted to hospital for observation; and 5 = return to operating room.
Because of the low incidence (1%-2% in CHMC experience) and critical nature of posttonsillectomy hemorrhage, bleeding was the primary determinant of sample size. Power calculations for bleeding were performed using one-tailed comparisons, as morphine, unlike ketorolac, has no effects on blood coagulation. With alpha = 0.05 and beta = 0.2, 300 subjects were required in each treatment group to detect an increased incidence of posttonsillectomy hemorrhage from 1%-2% to 4%-6%. With n = 300, power was adequate to detect clinically significant differences in other outcome variables. Data for bleeding was analyzed after the recruitment of each 100 subjects and the study was to be terminated if an increase in posttonsillectomy hemorrhage was seen with ketorolac.
Parametric data are presented as mean +/- SD. Nonparametric data are presented as median (25%-75% range). Rates are presented as rate (95% confidence interval) calculated from the Binomial distribution. Parametric data were analyzed with the unpaired Student's t-test or one-way analysis of variance with post hoc analysis by multiple unpaired Student's t-tests with Bonferonni correction for multiple comparisons. Ordinal nonparametric data were analyzed with the Mann-Whitney U-test. Nominal nonparametric data were analyzed with chi squared (with Yate's correction for continuity for 2 times 2 contingency tables) or Fisher's exact test; a one-tailed Fisher's exact test was used to analyze the incidence of bleeding, testing only the null hypothesis that bleeding after tonsillectomy with ketorolac is no worse than with morphine. Results were considered significant for corrected P < 0.05.
The first interval analysis of bleeding data was performed after 97 subjects had been recruited; because of an increased incidence of major bleeding and a greater number of bleeding episodes in the first 24 h after surgery in ketorolac subjects, the study was terminated (see below). Forty-nine subjects received ketorolac and 47 received morphine; one subject randomized to receive morphine who experienced severe intraoperative arterial bleeding was not treated with study drug and is not included in the analysis of the remaining 96 subjects. There were no differences between subjects receiving ketorolac or morphine in demographics Table 1 or surgical management Table 2.
Approximately one third of subjects in each group received supplemental morphine in the PACU (ketorolac, 18/49; morphine, 13/47). Three ketorolac subjects required more than one dose of supplemental morphine (one had a disconnected IV tube during the first dose). There were no differences in awakening, recovery, and discharge readiness times Table 3. One morphine and three ketorolac subjects required supplemental oxygen to maintain oxygen saturation >or=to95% after discharge from PACU.
There were no differences in unplanned admissions to hospital for outpatient subjects receiving ketorolac (4/29, three for bleeding and one for fever/poor oral intake/pneumonia) or morphine (4/28, two for arterial oxygen desaturation and two for vomiting/poor oral intake). Four subjects in each treatment group required readmission to hospital, two for bleeding and two for dehydration/poor oral intake in each group.
Ketorolac subjects had a lower median number of emetic episodes (1 [0-3] vs 3 [1-4]; P = 0.006). Although there was no difference in the proportion of subjects having at least one episode of emesis (32/49 vs 39/47), morphine subjects were more likely to have more than two episodes of emesis after PACU discharge (9/49 vs 22/47; P = 0.007).
The incidence and severity of bleeding episodes are shown in Table 4. Two ketorolac subjects required reoperation to control bleeding (one on the day of surgery, one on postoperative Day 5) compared with one morphine subject (postoperative Day 6). Ketorolac subjects were more likely to experience major bleeding (bleeding requiring intervention, bleeding severity score >or=to2) in the first 24 h after surgery than morphine subjects; no increase was seen in overall major bleeding or major bleeding after the first postoperative day. Ketorolac subjects had more bleeding episodes in the first 24 h after surgery (0.22 episodes/subject [0.12-0.35] vs 0.04 episodes/subject [0-0.11]); there was no increase in the overall number of bleeding episodes with ketorolac (0.31 episodes/subject [0.18-0.43] vs 0.21 episodes/subject [0.11-0.34]). Although most subjects in both treatment groups had a 24-h bleeding severity score of 0, among subjects who bled in the first 24 h, ketorolac subjects had a higher median bleeding severity score than morphine subjects (4 vs 1; P = 0.04 [one-tailed; from Mann-Whitney U-test for all subjects]); there was no increase in overall bleeding severity scores with ketorolac. Although not specifically solicited, there were reports in three ketorolac subjects of postoperative physical signs suggesting abnormal coagulation (two petechiae and one extensive bruising).
There were several minor protocol deviations. Four ketorolac and two morphine subjects received oral midazolam premedication. Two ketorolac and three morphine subjects received atropine. Three subjects, all in the morphine group, experienced laryngospasm treated with succinylcholine (two on induction [subsequently extubated awake] and one on extubation). One ketorolac and two morphine subjects received dexamethasone for airway swelling. Two subjects received inadvertent overdoses of ketorolac (2-3 mg/kg); neither subject experienced postoperative complications.
Our original hypothesis was supported only in that ketorolac subjects experienced fewer episodes of emesis after tonsillectomy. In contrast to our hypothesis, ketorolac subjects neither awakened sooner, achieved discharge readiness more quickly, nor had fewer unanticipated admissions after tonsillectomy than morphine subjects. Of most significance, ketorolac subjects were more likely than morphine subjects to experience major bleeding and had more bleeding episodes in the first 24 h after tonsillectomy.
The majority of subjects in both treatment groups experienced at least one episode of emesis; however, subjects who received morphine experienced more emesis after leaving the PACU. This result is consistent with another study which demonstrated an increased incidence of late postoperative vomiting after a single intraoperative dose of morphine . Children who received supplemental morphine in the PACU did not experience increased emesis compared to subjects in the same treatment group who did not require additional analgesia. Despite prophylactic treatment of emesis with metoclopramide, 75% of our subjects experienced at least one episode of emesis. This is in contrast to the 47% incidence of emesis in subjects receiving metoclopramide reported by Ferrari and Donlon ; we are unable to explain this discrepancy, as anesthetic management (with the exception of orogastric tube decompression of the stomach) and postoperative care appear comparable between the two studies.
We were unable to demonstrate any improvement in recovery in subjects receiving ketorolac. The fact that we tracheally extubated our subjects under deep anesthesia may have made awakening from the anesthetic the predominant effect in determining recovery. Our use of halothane, rather than an anesthetic associated with more rapid emergence, may have obscured any recovery advantage of ketorolac over morphine in the early postoperative period; however, given the short duration of surgery, any such effect would have been of minimal magnitude by PACU discharge. Administration of supplemental morphine in PACU did not affect awakening time, time to readiness for PACU discharge, or discharge home in either treatment group.
The early termination of our study does not allow us to draw any conclusions regarding unplanned admissions or readmissions. It appears that there is no difference between ketorolac and morphine in the rate of unplanned admission or readmission after tonsillectomy, although unexpected admissions after ketorolac were likely to be for bleeding whereas those after morphine were likely to be for protracted vomiting or desaturation. Given its side effect profile, ketorolac would be expected to be particularly appropriate for children with OSA. The small number of OSA subjects enrolled prevents us from testing this hypothesis. We considered continuing the study in a limited fashion, restricting recruitment to subjects with OSA; however, given the absence of any suggestion of faster recovery or decreased oxygen requirements with ketorolac and the potentially catastrophic consequences of posttonsillectomy hemorrhage, we concluded that continuation of the study would be ethically inappropriate.
Given the well documented adverse effects of ketorolac on coagulation [16-19] and the absence of such effects with morphine, our use of one-tailed statistical tests for bleeding, examining only the null hypothesis that bleeding with ketorolac is no worse than that with morphine, was appropriate. The use of one-tailed statistical tests does impose some ambiguity in the interpretation of negative results. When two-tailed statistical tests are used and the null hypothesis is not rejected (P > 0.05), it is possible to conclude that there is no difference between the two treatment groups. In contrast, when one-tailed statistical tests are used, acceptance of the null hypothesis implies only that there is no difference between the two treatment groups in the hypothesized direction; the possibility remains that there may be a difference between the two treatment groups in the direction opposite to that in the original null hypothesis. Thus, in instances where we did not demonstrate an increase in bleeding with ketorolac, we can only conclude that either bleeding does not differ between morphine and ketorolac or that bleeding is worse with morphine.
The surgical techniques used were associated with a dry surgical field and minimal blood loss, such that accurate measurement of blood loss and orogastric tube evacuation of stomach contents after tonsillectomy were unnecessary. A completely dry surgical field assured hemostasis before administration of ketorolac and provided an appropriate milieu for deep extubation of the trachea. We do not routinely obtain preoperative coagulation studies in patients having tonsillectomy; therefore, the differences seen in bleeding may have been due to an excess number of ketorolac subjects with undetected coagulation defects. The absence of intraoperative differences in bleeding and the fact that the differences in bleeding seen were limited to the first 24 h suggest that this is unlikely.
The overall incidence of bleeding was approximately 16% in each group. This number includes five subjects who experienced only self-limited bleeding and three subjects who presented to the clinic or emergency department with bleeding sufficiently mild to not require treatment or admission to hospital. The number of reports of mild bleeding episodes may have been increased in our study because of our decision to prospectively record all episodes of bleeding, irrespective of severity, and increased parental vigilance secondary to the informed consent process. The reoperation rate (3%) was consistent with our experience.
The apparently contradictory observation of increased major bleeding in the first 24 h with ketorolac but no overall increase in bleeding may be a Type II error, wherein the dramatic increase in major bleeding in the first 24 h produced only a small increment in the total incidence of bleeding, which was not detected because of insufficient numbers of subjects. Ketorolac subjects also tended to have multiple episodes of bleeding in the first 24 h; these multiple episodes are not reflected in the count of subjects with any episodes of bleeding. Finally, many of the subjects who received ketorolac bled both in the first 24 h and again later; perhaps ketorolac did not increase the number of subjects who bled, but rather altered when, how often, and how severely they bled.
Our demonstration of an increased incidence of major bleeding in the first 24 h after surgery is consistent with the pharmacodynamics of ketorolac, whose antiplatelet effects are reversible and limited to the time the drug is present in the body . As the half-life of ketorolac is 5-6 h , platelet function returns to normal within 24 h of a single dose of ketorolac.
In conclusion, we have demonstrated that, even when given after tonsillectomy, ketorolac increases the incidence of major bleeding and the number of bleeding episodes in the first 24 h after surgery. In addition, although ketorolac decreased the number of episodes of emesis after PACU discharge, it did not affect time to awakening or readiness for discharge or the rate of unplanned admission to hospital. Our demonstration of increased early posttonsillectomy hemorrhage, in concert with studies showing increased intraoperative bleeding when ketorolac was given before tonsillectomy [14,19], suggests that ketorolac is contraindicated in children having adenotonsillectomy.
The authors wish to acknowledge the nurses of the CHMC PACU, OPS, and OPN, without whose invaluable assistance this study would not have been possible. They also wish to thank Michelle Tate and Patricia Wilson for their patience and professionalism during the preparation of this manuscript.
1. Guida RA, Mattucci KF. Tonsillectomy and adenoidectomy: an inpatient or outpatient procedure? Laryngoscope 1990;100:491-3.
2. Carithers JS, Gebhart DE, Williams JA. Postoperative risks of pediatric tonsilloadenoidectomy. Laryngoscope 1987;97:422-9.
3. Segal C, Berger G, Basker M, Marshak G. Adenotonsillectomies on a surgical day-clinic basis. Laryngoscope 1983;93:1205-8.
4. Richmond KH, Wetmore RF, Baranak CC. Postoperative complications following tonsillectomy and adenoidectomy--who is at risk? Int J Pediatr Otorhinolaryngol 1987;13:117-24.
5. Shott SR, Myer CM, Cotton RT. Efficacy of tonsillectomy and adenoidectomy as an outpatient procedure: a preliminary report. Int J Pediatr Otorhinolaryngol 1987;13:157-63.
6. Maniglia AJ, Kushner H, Cozzi L. Adenotonsillectomy: a safe outpatient procedure. Arch Otolaryngol Head Neck Surg 1989;115:92-4.
7. Reiner SA, Sawyer WP, Clark KF, Wood MW. Safety of outpatient tonsillectomy and adenoidectomy. Otolaryngol Head Neck Surg 1990;102:161-8.
8. Haberman RS, Shattuck TG, Dion NM. Is outpatient suction cautery tonsillectomy safe in a community hospital setting? Laryngoscope 1990;100:511-5.
9. Buckley MMT, Brogden RN. Ketorolac: a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential. Drugs 1990;39:86-109.
10. Peirce RJ, Fragen RJ, Pemberton DM. Intravenous ketorolac tromethamine versus morphine sulfate in the treatment of immediate postoperative pain. Pharmacotherapy 1990;10:111S-5S.
11. Brown CR, Moodie JE, Wild VM, Bynum LJ. Comparison of intravenous ketorolac tromethamine and morphine sulfate in the treatment of postoperative pain. Pharmacotherapy 1990;10:116S-21S.
12. Kenny GNC, McArdle CS, Aitken HH. Parenteral ketorolac: opiate-sparing effect and lack of cardiorespiratory depression in the perioperative patient. Pharmacotherapy 1990;10:127S-31S.
13. Watcha MF, Jones MB, Lagueruela RG, et al. Comparison of ketorolac and morphine as adjuvants during pediatric surgery. Anesthesiology 1992;76:368-72.
14. Fitz-James I, Ho J, Pang LM, et al. Effect of ketorolac on perioperative bleeding and analgesia in pediatric tonsillectomy and adenoidectomy patients [abstract]. Anesth Analg 1995;80:S127.
15. Munro HM, Riegger LQ, Reynolds PI, et al. Comparison of the analgesic and emetic properties of ketorolac and morphine for paediatric outpatient strabismus surgery. Br J Anaesth 1994;72:624-8.
16. Conrad KA, Fagan TC, Mackie MJ, Mayshar PV. Effects of ketorolac tromethamine on hemostasis in volunteers. Clin Pharmacol Ther 1988;43:542-6.
17. Spowart K, Greer IA, McLaren M, et al. Haemostatic effects of ketorolac with and without concomitant heparin in normal volunteers. Thromb Haemost 1988;60:382-6.
18. Bean JD, Hunt R, Custer MD. Effects of ketorolac on postoperative analgesia and bleeding time in children [abstract]. Anesthesiology 1993;79:A1190.
19. Rusy LM, Houck CS, Sullivan LJ, et al. A double-blind evaluation of ketorolac tromethamine versus acetaminophen in pediatric tonsillectomy: analgesia and bleeding. Anesth Analg 1995;80:226-9.
20. Brouilette R, Hanson D, David R, et al. A diagnostic approach to suspected obstructive sleep apnea in children. J Pediatr 1984;105:10-4.
21. Brodsky L. Modern assessment of tonsils and adenoids. Pediatr Clin North Am 1989;36:1551-69.
22. Weinstein MS, Nicolson SC, Schreiner MS. A single dose of morphine sulfate increases the incidence of vomiting after outpatient inguinal surgery in children. Anesthesiology 1994;81:572-7.
23. Ferrari LR, Donlon JV. Metoclopramide reduces the incidence of vomiting after tonsillectomy in children. Anesth Analg 1992;75:351-4.
24. Mroszczak EJ, Jung D, Yee J, et al. Ketorolac tromethamine pharmacokinetics and metabolism after intravenous, intramuscular, and oral administration in humans and animals. Pharmacotherapy 1990;10:33S-9S.
© 1995 International Anesthesia Research Society
25. Olkkola KT, Maunuksela EL. The pharmacokinetics of postoperative intravenous ketorolac tromethamine in children. Br J Clin Pharm 1991;31:182-4.