Hanasono MM, Lalakea ML, Mikulec AA, Shepard KG, Wellis V, Messner AH. Perioperative Steroids in Tonsillectomy Using Electrocautery and Sharp Dissection Techniques. Arch Otolaryngol Head Neck Surg. 2004;130(8):917–921. doi:10.1001/archotol.130.8.917
To determine the effect of preoperative dexamethasone sodium phosphate administration on posttonsillectomy morbidity for electrocautery ("hot") and sharp ("cold") dissection techniques.
Prospective, randomized, double-blind study.
University pediatric hospital and county teaching hospital.
A total of 219 children, aged 9 months to 12 years, undergoing tonsillectomy.
Participants who underwent tonsillectomy were randomly assigned to receive either intravenous dexamethasone sodium phosphate (1 mg/kg) or placebo.
Pain scores, oral intake, and emesis on postoperative day (POD) 1.
A total of 106 subjects (62 undergoing hot and 44 cold tonsillectomies) received preoperative steroids, and 113 (56 hot and 57 cold tonsillectomies) received placebo. On POD 1, pain scores reported by patients (P = .02), parents (P = .002), and physicians (P<.001) were significantly lower in subjects receiving steroids than in those receiving placebo. Emesis was reduced from a mean of 2.1 (placebo group) to 1.2 episodes (steroid group) (P = .02). Oral intake improved from 24.5% of normal diet (placebo) to 31.7% (steroid group) (P = .004). When all 4 groups were compared (cold placebo, cold steroid, hot placebo, and hot steroid), pain scores reported by physicians and parents were significantly lower in the cold steroid group than in the other groups.
Perioperative dexamethasone use reduces posttonsillectomy morbidity in pediatric patients in the early postoperative period after hot or cold tonsillectomy. The combination of steroid and cold dissection technique provided the greatest advantage in reducing posttonsillectomy subjective pain levels.
Tonsillectomy with or without adenoidectomy is one of the most commonly performed pediatric procedures in this country. However, this procedure is accompanied by significant postoperative morbidity, which may include postoperative bleeding, pain, nausea, vomiting, poor oral intake, and dehydration.1 Fever and foul mouth odor are also common.2,3 In an effort to reduce posttonsillectomy morbidity, numerous modifications and adjuncts to standard surgical technique have been proposed.
In recent years, many have advocated steroids for the reduction of postoperative morbidity after tonsillectomy. 4,5 Steroids are believed to act to reduce tissue damage and postoperative pain by suppressing fibrin deposition, capillary dilation, edema formation, and leukocyte migration.6,7 Steroids are also believed to increase thirst and appetite, resulting in a more rapid return to regular diet and adequate fluid intake. A prolonged antiemetic effect of intravenous (IV) dexamethasone has been well documented in patients undergoing chemotherapy.8
The effect of a single preoperative intravenous dose of steroid on tonsillectomy morbidity has been the subject of several studies with conflicting results.4,5 Several studies have shown reduced postoperative pain, less frequent emesis, and earlier return to normal diet associated with preoperative administration of a single dose of intravenous steroid.1,3,9- 12 Other studies have found no significant differences.13,14
The present study is the largest placebo-controlled, prospective, randomized, double-blind study to examine postoperative morbidity in pediatric patients given perioperative intravenous dexamethasone to date. The goal of this investigation was to elucidate the effect of steroids with regard to surgical technique, a factor that, to our knowledge, has not previously been evaluated. Specifically, we hypothesized that the surgical technique chosen for tonsillectomy (electrocautery [hereinafter, "hot"] vs sharp [hereinafter, "cold"] dissection) might influence the potential effects of steroids on postoperative morbidity.
Candidates for this study included all patients 12 years or younger scheduled to undergo elective tonsillectomy, tonsillectomy and adenoidectomy, or tonsillectomy with another procedure. Patients were excluded if they had significant comorbidities requiring perioperative steroids (eg, severe reactive airway disease) or conditions for which steroids are contraindicated (eg, diabetes mellitus). Informed consent to participate in the study was obtained from the subjects' parents and from all subjects older than 7 years, per the protocols of the Stanford University human subjects committee and the research and human subjects committee at Santa Clara Valley Medical Center. The procedures were performed at Stanford University Medical Center/Lucile Salter Packard Children's Hospital, a university children's hospital, or Santa Clara Valley Medical Center, a university-affiliated county teaching hospital, under the direction of 1 of 3 faculty surgeons (M.L.L., K.G.S., or A.H.M).
Subjects underwent tonsillectomy by either hot (on odd-numbered days of the month) or cold (on even-numbered days) surgical technique; wire snare transection with directed cautery for hemostasis was used with the cold technique. In all patients who underwent adenoidectomy, the technique consisted of curette and/or punch excision with directed suction cautery for hemostasis. Administration of steroid vs isotonic sodium chloride (normal saline) placebo was determined by a random number list; patients and physicians were blinded to the assignment. This randomization scheme resulted in 4 study arms: (1) cold tonsillectomy with steroid (CS); (2) cold tonsillectomy with placebo (CP); (3) hot tonsillectomy with steroid (HS); and (4) hot tonsillectomy with placebo (HP).
General anesthesia and postanesthesia care were standardized for all patients. Oral midazolam hydrochloride was administered as a premedication at a dose of 0.5 to 0.6 mg/kg up to a maximum of 20 mg. Induction was performed with sevoflurane and 60% nitrous oxide. Propofol was not used during induction because it has antiemetic effects that could confound the results of the study. Rocuronium bromide (0.6 mg/kg IV) was given prior to intubation. Anesthesia was maintained with nitrous oxide and isoflurane. The reversal regimen consisted of neostigmine methylsulfate (70 µg/kg IV) and glycopyrolate (10 µg/kg IV) and was mandatory for all patients. Fentanyl citrate (2 µg/kg IV) was given for pain during surgery, and morphine sulfate (0.05-0.1 mg/kg IV) was given in the recovery room for pain as needed. Metoclopramide (0.15 mg/kg IV) was given in the recovery room if there were 2 episodes of emesis or retching. Ondansetron hydrochloride (0.15 mg/kg IV) was given if metoclopramide was ineffective after 20 minutes.
Steroids or placebo was administered at the start of the procedure. Study medication (dexamethasone or normal saline placebo) was supplied to the investigators by pharmacy staff in a blinded manner, and medication records were maintained in the pharmacy until the conclusion of the study. Steroid treatment consisted of intravenous dexamethasone at a dose of 1 mg/kg up to a maximum dose of 50 mg. An equal volume of saline was administered intravenously to patients randomized to the control group.
Postoperatively, patients were prescribed a 5-day course of amoxicillin at 40 mg/kg per day or trimethoprim/sulfamethoxazole at 4 mg/kg per day and 20 mg/kg per day, respectively, for penicillin-allergic patients. For postoperative analgesia, patients were instructed to use acetaminophen at 12 to 15 mg/kg orally or rectally or acetaminophen with codeine elixir (120 mg of acetaminophen with 12 mg of codeine phosphate per 5 mL) at a codeine phosphate dose of 0.5 mg/kg every 4 hours according to patient/parental preference. Parents were strongly encouraged to give 1 of these pain medications every 4 hours around the clock for at least the first 3 days.
Data collected at the time of the surgical procedure included demographic and clinical information (age, sex, weight), indication for the procedure, associated medical conditions, surgical technique used, and any operative complications. The main outcome measures included differences in oral intake, pain scores, and vomiting. These measures were assessed at the time of patient follow-up in the clinic on postoperative day (POD) 1. On the POD 1 visit, an examination was performed, and a standard study encounter form was completed. Parents were queried regarding the quantity of food ingested (expressed as a percentage of normal intake rounded to the nearest quartile: 0%, 25%, 50%, 75%, and 100%) and occurrence of emesis. Subjective pain was assessed by the parents, the physician, and, in children older than 4 years, by the patients themselves. Pain was measured using the Wong-Baker visual analog scale,15 where 0 represents no pain and 10 represents the worst pain imaginable. On POD 3, patients were contacted by phone and queried regarding complications such as bleeding or any other postoperative problem that required an unplanned office visit or emergency department visit.
Outcome data were compared using the Kruskal-Wallis rank sum test. Nonparametric data (complication rates) were analyzed with the Mann-Whitney U test. A P value <.05 was considered statistically significant. Ninety-five percent confidence intervals were calculated for mean outcome measures where appropriate.
A total of 219 subjects aged 9 months to 12 years participated in the study. Seventeen patients who received dexamethasone (including 10 who underwent hot tonsillectomy and 7 who underwent cold tonsillectomy) and 16 patients (7 hot and 9 cold) who received placebo were kept overnight for reasons that included age younger than 2 years, history of severe obstructive sleep apnea, and history of significant comorbidity (eg, Down syndrome). These patients (15% of total) were not excluded from the study. The demographic data for the study populations are detailed in Table 1. There were no statistically significant differences in any of the demographic measures.
Seven patients had complications, 5 of which required readmission. These complications are listed in Table 2. There were no statistically significant differences in complication rates between the steroid and placebo groups, or when comparing HS, HP, CS, and CP groups. However, the nature of complications experienced by patients undergoing hot tonsillectomy and cold tonsillectomy were markedly different. Persistent emesis or dehydration occurred only in the hot group, while posttonsillectomy hemorrhage occurred only in the cold-group. Note that complications were not recorded after POD 3.
Postoperative outcome measures are listed in Table 3 and Table 4. On POD 1, significantly lower pain scores, based on the Wong-Baker visual analog scale15, were recorded in steroid-receiving subjects as rated by the child, parent, and physician compared with those recorded for patients receiving placebo (P<.05) (Table 3). Likewise, significantly fewer episodes of emesis (1.2 vs 2.1 episodes) and greater amounts of oral intake relative to the patient's normal intake (31.7% vs 24.5%) were noted in the steroid group. When hot tonsillectomy and cold tonsillectomy groups were compared without regard to steroid administration, no statistically significant differences in outcome measures were found.
Results were further analyzed by comparing each of the 4 treatment groups (CS, CP, HS, and HP) in an effort to determine whether results were more favorable for any group(s) (Table 4). Parent- and physician-rated pain scores on POD 1 were significantly lower for the steroid groups than for the placebo groups. Moreover, among the steroid-receiving groups, a significant advantage was found for the CS group over the HS group with respect to parent- and physician-rated pain scores. Oral intake on POD 1 was significantly greater in the CS and HS groups than in the placebo groups, but the CS and HS groups did not differ significantly from each other for this outcome measure. No significant differences were found in the 4-group analysis for subject-rated pain scores or number of emesis episodes.
In the present study, the administration of preoperative dexamethasone sodium phosphate at a dose of 1 mg/kg was associated with a reduction in child-, parent- and physician-rated posttonsillectomy pain scores on POD 1 compared with the administration of saline placebo. In addition, a significant improvement in oral intake and a reduction in the number of emesis episodes were noted for patients receiving steroids. When we compared CS, CP, HS, and HP groups, the CS group had significantly lower parent- and physician-rated pain scores than all other groups, which suggests that the combination of cold dissection and steroid may offer the greatest advantage in terms of posttonsillectomy morbidity reduction. In the 4-group comparison, subject-rated pain scores and episodes of emesis failed to reach statistical significance. This is likely owing to the relatively small sample size: a benefit was demonstrated for both of these outcome measures with larger sample size in the 2-group comparison between steroid and placebo groups.
Eight previous studies have examined the effects of preoperative intravenous dexamethasone in pediatric patients undergoing adenotonsillectomy.1,3,9- 14 These studies are briefly summarized in Table 5. Pappas et al,1 Tom et al,3 and April et al9 reported that preoperative dexamethasone administration at a dose of 1 mg/kg was associated with an improvement in oral intake and reduced emesis postoperatively. Vosdoganis and Baines10 reported similar findings using a dexamethasone dose of 0.4 mg/kg. Splinter and Roberts,11 using an even lower dose of dexamethasone (0.15 mg/kg; maximum dose, 8 mg), demonstrated an improvement in emesis, although oral intake was not measured as an end point. One of these studies3 found that pain was reduced as well in steroid-treated patients compared with controls.
In contrast, Volk et al13 and Ohlms et al14 found no statistically significant differences in oral intake, pain, level of activity, and analgesic use in patients treated with dexamethasone compared with placebo prior to tonsillectomy. Ohlms et al14 also failed to demonstrate any improvement in nausea and emesis in patients receiving steroid. Similarly, a small study by Catlin and Grimes12 concluded that a single preoperative dose of 8 mg/m2 of dexamethasone resulted in no differences in pain, nausea, emesis, or need for pain medication, although patients receiving steroid did experience a more rapid return to normal diet.
One outstanding difference between the several studies reporting a benefit from steroids and those that reported no benefit was the surgical technique used by the investigators. In studies by Pappas et al,1 Tom et al,3 and April et al,9 all of which showed an advantage to steroid administration, tonsillectomy was performed using the hot technique. In studies by Volk et al,13 Ohlms et al,14 and Catlin and Grimes,12 which failed to find a benefit for steroid use, tonsils were excised by the cold technique.
Hot and cold techniques have previously been compared. The advantage of the hot technique is that it reduces blood loss and dissection time. However, some studies have shown that there is less pain associated with the cold technique, presumably owing to reduced thermal injury to remaining tissue.16,17 Ohlms et al14 suggested that steroids might have a greater effect if given to patients undergoing tonsillectomy by the hot technique in whom more inflammation and edema is expected to develop. Further studies of steroid use during tonsillectomy using hot vs cold techniques were recommended by Ohlms et al,14 and that recommendation prompted the present investigation.
Apart from dissection technique, another factor that may explain the conflicting results of previous studies is the difference in dexamethasone dose used. Three of the 5 prior studies that showed a positive effect of IV dexamethasone on postoperative morbidity used a dose of 1 mg/kg, as used in the present study, with a maximum dose ranging from 10 to 25 mg.1,3,9 These 3 studies were also the studies in which tonsillectomy was performed strictly using a hot technique. While the study by Splinter and Roberts,11 in which a dexamethasone dose of 0.15 mg/kg was used, demonstrated a positive effect of IV steroid use, only postoperative emesis was examined. Ohlms et al,14 who used a dexamethasone dose of 0.5 mg/kg, and Volk et al,13 who gave 10 mg of dexamethasone to all children (approximately 2% of the children in our study received 10 mg or less), failed to demonstrate an advantage for lower-dose IV steroids.
Differences in anesthetic technique and lack of standardization may also explain in part the conflicting results reported in prior studies.1,3,9- 12 The anesthetic regimen used may greatly affect the incidence of postoperative nausea as well as early pain control. Moreover, in 2 studies, different anesthetic regimens were used within the studies themselves.10,11 In the study by Splinter and Roberts,11 some patients were given propofol, and others were not. Patients who received propofol, an agent known to have antiemetic effects, had a reduced incidence of postoperative emesis. Propofol was also used in the study by Vosdoganis and Baines,10 although its effects were not examined separately. In the present study, the anesthetic regimen was standardized in an effort to avoid any confounding effect on postoperative outcome measures.
In a recent meta-analysis of the above-mentioned 8 published studies of dexamethasone use in pediatric patients undergoing tonsillectomy or adenotonsillectomy,1,3,9- 14 Steward et al18 found that children given dexamethasone were 2 times less likely to vomit in the first 24 hours than children given placebo. In addition, children given dexamethasone were more likely to advance to a soft or solid diet on POD 1 than those given placebo. By POD 3, no differences in oral intake were found. Because of varied outcome measures and missing data, pain could not meaningfully be analyzed as an end point. The regression analysis of the antiemetic effect as a function of dexamethasone dose also suggested that the benefit was enhanced by increasing the dose to 1 mg/kg. This dose is somewhat higher than that used in several prior studies, particularly in studies that failed to find an advantage in steroid use. The early benefit noted for steroid use in the study by Steward et al18 appears consistent with the known 36- to 48-hour half-life of dexamethasone.
Long-term corticosteroid therapy may have significant adverse effects. Brief corticosteroid use (24-48 hours) may cause burning and itching at mucocutaneous junctions, precipitation of glucose intolerance in patients with latent diabetes mellitus, superficial ulcerations of the stomach mucosa, and very rare episodes of multifocal premature ventricular contractions.6 Complications from use of corticosteroids for less than 24 hours, however, are virtually nonexistent, even in the presence of viral or bacterial infection.6 Short courses of corticosteroids can be stopped abruptly and do not require a tapering regimen. No complications related to steroid use were reported in any of the studies examining the effect of preoperative steroids in tonsillectomy, including the present study.1,3,9- 14
In conclusion, a single, 1-mg/kg IV dose of dexamethasone results in a statistically significant reduction in patient-, parent-, and physician-rated pain scores; fewer episodes of emesis; and greater quantities of oral intake in the first 24 hours following tonsillectomy surgery compared with saline placebo. Dexamethasone appears to positively influence postoperative pain and oral intake regardless of whether hot or cold dissection techniques are used. Moreover, our data indicate that the combination of cold technique and steroid use may offer a further reduction in perioperative morbidity, as a statistically significant improvement in parent- and physician-rated pain scores and oral intake was noted in these patients on POD 1. The results of our study support the use of IV dexamethasone at a dose of 1 mg/kg at the time of tonsillectomy.
Correspondence: Anna H. Messner, MD, Otolaryngology–Head and Neck Surgery, Lucile Salter Packard Children's Hospital, 725 Welch Rd, Palo Alto, CA 94304 (email@example.com).
Submitted for publication May 14, 2002; final revision received December 1, 2003; accepted January 29, 2004.
This article was presented at the American Society of Pediatric Otolaryngology Spring Meeting; May 13-14, 2002; Boca Raton, Fla.
We thank Charles R. Mathis for his statistical expertise in the preparation of this article.