Secondary Logo

Journal Logo

Perioperative Steroid Use for Tonsillectomy and Its Association With Reoperation for Posttonsillectomy Hemorrhage: A Retrospective Cohort Study

Miyamoto, Yoshihisa MD, MPH; Shinzawa, Maki MD, PhD; Tanaka, Shiro PhD; Tanaka-Mizuno, Sachiko PhD; Kawakami, Koji MD, PhD

doi: 10.1213/ANE.0000000000002681
Anesthetic Clinical Pharmacology: Original Clinical Research Report
Free
SDC

BACKGROUND: Steroids reduce postoperative complications after tonsillectomy such as nausea and vomiting, pain, and delayed recovery. However, steroids may also increase the risk of severe posttonsillectomy bleeding requiring reoperation.

METHODS: To evaluate the risk of postoperative bleeding requiring reoperation related to perioperative steroid use, we conducted a retrospective cohort study of 6149 patients treated at 68 hospitals using a hospital-based claims database. The primary outcome was reoperation for bleeding within 14 postoperative days. We estimated odds ratios (ORs) between perioperative steroid use and reoperation by multivariable logistic regression analysis adjusted for confounders. We also estimated differences in the adjusted risk. Subgroup analyses after dividing patients into adults and children were also performed.

RESULTS: The incidence of reoperation did not differ significantly between patients who received steroids on the day of tonsillectomy and those who did not (1.8%, n = 15 vs 1.5%, n = 79; adjusted OR 0.81, 95% confidence interval [CI], 0.45–1.43; P = .46). We also found nonsignificant associations in both adults (OR, 0.73; 95% CI, 0.38–1.38; P = .33) and children (OR, 1.18; 95% CI, 0.34–4.11; P = .80). The adjusted risk differences estimated by the logistic regression model were −0.30% (95% CI, −1.05 to 0.45) in all patients, −0.64% (95% CI, −1.82 to 0.54) in adults, and 0.13% (95% CI, −0.93 to 1.19) in children.

CONCLUSIONS: Steroid use on the day of tonsillectomy was not associated with an increased risk of reoperation for bleeding. Although the wide range of CIs for the ORs could not eliminate the possibility of increased risk, especially in children, the incremental risks of reoperation for steroid use were within an acceptable range for both adults and children. Our results support the safety of perioperative steroid use for tonsillectomy, considering the magnitude of risk of reoperation because of bleeding.

From the Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Kyoto, Japan.

Published ahead of print November 22, 2017.

Accepted for publication October 20, 2017.

Funding: Institutional and departmental.

Conflicts of Interest: See Disclosures at the end of the article.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Yoshihisa Miyamoto, MD, MPH, is currently affiliated with the Department of Anesthesia, Kanagawa Children’s Medical Center, Yokohama, Japan. Maki Shinzawa, MD, PhD, is currently affiliated with the Department of Nephrology, Graduate School of Medicine, Osaka University, Suita, Japan. Shiro Tanaka, PhD, is currently affiliated with the Department of Clinical Biostatistics, Graduate School of Medicine, Kyoto University, Kyoto, Japan. Sachiko Tanaka-Mizuno, PhD, is currently affiliated with the Department of Medical Statistics, Shiga University of Medical Science, Otsu, Japan.

Reprints will not be available from the authors.

Address correspondence to Koji Kawakami, MD, PhD, Department of Pharmacoepidemiology, Graduate School of Medicine and Public Health, Kyoto University, Yoshidakonoecho, Sakyo-ku, Kyoto 606–8501, Japan. Address e-mail to kawakami.koji.4e@kyoto-u.ac.jp.

There is evidence that dexamethasone reduces posttonsillectomy complications, such as nausea and vomiting, pain, and delayed recovery.1,2 Guidelines recommend intraoperative administration of dexamethasone for tonsillectomy in both children and adults to reduce postoperative complications3,4; however, there are ongoing concerns that intraoperative dexamethasone may increase the risk of reoperation for posttonsillectomy hemorrhage, and conflicting results have been reported regarding this issue even after the publication of the guidelines. One randomized controlled trial indicated that dexamethasone use increases the risk of posttonsillectomy bleeding.5 Conversely, a noninferiority trial concluded that dexamethasone use has no association with an increased risk of reoperation for bleeding.6 However, this trial was criticized for setting the same noninferiority margin at 5% for all kinds of bleeding, including any bleeding, bleeding requiring rehospitalization, and bleeding requiring reoperation.7 Several meta-analyses have also concluded that there is no significant increase in posttonsillectomy bleeding incidence in patients administered dexamethasone,8–11 but one of these did report an increased incidence of reintervention for bleeding in children administered systemic steroids.10 Moreover, a recent large observational study reported a significant association between systemic steroids and a higher incidence of posttonsillectomy reoperation in children.12

Severe bleeding, ie, bleeding requiring reoperation, is the most clinically relevant bleeding and can be life-threatening. However, thousands of participants are necessary to confirm a significant difference between treatment groups, given that the incidence of bleeding requiring reoperation is approximately 3%.13 Under this circumstance, a retrospective observational study using a multihospital database may provide more definitive evidence by allowing for the inclusion of large patient groups.

In addition to the controversy regarding the influence of steroids on posttonsillectomy bleeding, little is known about differences in the absolute risk of severe bleeding related to steroid use because most studies so far have reported only relative effect measures such as odds ratios (ORs) for posttonsillectomy bleeding with steroid use. Although an OR can approximate a risk ratio under the rare disease assumption such as reoperation for bleeding, a risk difference presents more directly pertinent information than relative effect measures and is particularly useful when considering the balance between benefits and harms.14 Therefore, reporting both relative and absolute effect measures is recommended even in observational studies.15 Furthermore, postoperative steroid use can benefit the recovery after tonsillectomy16,17; however, its influence on posttonsillectomy bleeding has not ever been reported. Hence, we conducted a retrospective cohort study using a claims database shared among many hospitals to investigate the association between perioperative steroid use, including postoperative steroids, and postoperative bleeding requiring reoperation and the difference in the absolute risk of reoperation related to perioperative steroid use.

Back to Top | Article Outline

METHODS

This study was approved by the Ethics Committee of Kyoto University Graduate School and Faculty of Medicine (E2224). This study was exempt from obtaining individual informed consent on the basis of the Ethical Guidelines for Epidemiologic Research by the Ministry of Health, Labor and Welfare because the data had been completely deidentified before being provided to us. This manuscript adheres to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.18

Back to Top | Article Outline

Design, Setting, and Data Sources

We performed a retrospective cohort study using a multihospital claims database provided by Medical Data Vision Co, Ltd (Tokyo, Japan). The database included records of approximately 6 million patients as of March 2014 from nonacademic 130 diagnosis procedure combination (DPC) hospitals, roughly 10% of all DPC hospitals in Japan. These DPC hospitals are acute-phase hospitals administered under the Diagnosis Procedure Combination/Per-Diem Payment System. The database contains not only administrative claims data but also detailed patient data, including an anonymized patient identifier, age, sex, diagnosis code, prescriptions, medical procedures including operations, and the results of laboratory tests.

Back to Top | Article Outline

Study Population

Patients who underwent tonsillectomy (Japanese original procedure code: K377) with or without adenoidectomy (K370) or tympanostomy tube insertion (K309) from April 1, 2008, to March 17, 2014, were eligible for inclusion. Patients were excluded if they received steroid or anticoagulant/antiplatelet therapy within 14 days preoperatively, underwent tonsillectomy because of malignancy, were administered anticoagulants/antiplatelets other than nonsteroidal anti-inflammatory drugs (NSAIDs) within 7 days postoperatively, were without data on anesthesia time, or had concomitant operations other than adenoidectomy or tympanostomy tube insertion.

Back to Top | Article Outline

Data Extraction

Independent Variables.

The main exposure was the administration of intravenous steroids on the day of tonsillectomy. We defined postoperative steroids as oral or intravenous steroids that were administered within 1–14 days postoperatively. For those who underwent reoperation, we confined postoperative steroids to those administered between the day after the index operation and the day before the reoperation. Also, we obtained the following patient data from the database: sex, age, indications for tonsillectomy, comorbidities, type of surgery, perioperative medications other than steroids, and anesthesia time. We defined the indication for surgery and diagnosis of comorbidities according to the International Statistical Classification of Diseases and Related Health Problems 10th Revision. Indications for tonsillectomy were classified as either infection (recurrent tonsillitis [International Classification of Diseases, 10th Revision code: J03.9], chronic tonsillitis [J35.0], and peritonsillar abscess [J36]) or noninfection (sleep apnea syndrome [G47.3], hypertrophic tonsil [J35.1], IgA nephropathy [N02.8], and palmoplantar pustulosis [L40.3]). Comorbidities considered for analysis were hypertension (I10 and I15), diabetes mellitus (E10–E14), and asthma (J45). Perioperative medicines identified with Anatomical Therapeutic Chemical classification system codes included steroids (Anatomical Therapeutic Chemical code: H02), NSAIDs (M01A and N02B), anticoagulants/antiplatelet agents (B01), and antibiotics (J01).

Back to Top | Article Outline

Outcomes.

The primary outcome was reoperation attributable to bleeding 1–14 days after tonsillectomy. Reoperation on the same day as the initial surgery was not a primary outcome measure because bleeding within 24 hours of surgery is generally attributable to surgical technique.3,19 We defined reoperation for bleeding as hemostasis for posttonsillectomy bleeding (K377). We also included the incision of retropharyngeal abscess (K367) as “reoperation for bleeding” because this procedure is considered as such by the health insurance claim system in Japan if performed on or after postoperative day 1 in tonsillectomy patients. Patients who were discharged within 13 postoperative days and of whom there were no claims data after 14 postoperative days were presumed not to have undergone reoperation because reoperation is considered to be usually performed at the same hospital as the index tonsillectomy.

Back to Top | Article Outline

Statistical Analysis

We compared relevant variables between tonsillectomy patients receiving steroids on the day of tonsillectomy (steroid group) and tonsillectomy patients receiving no steroids on the day of tonsillectomy (no-steroid group [who may, however, have received postoperative steroids]) by the Mann-Whitney U test or Fisher exact test. The variables compared included patient characteristics (sex, age, and comorbidities), procedural data (indication for tonsillectomy, type of surgery, and anesthesia time), perioperative medications (steroids, NSAIDs, and antibiotics on the day of tonsillectomy and postoperatively), and postoperative outcomes (reoperation for bleeding, postoperative day of reoperation, and postoperative length of stay).

We investigated the association of reoperation for bleeding with steroid use, sex, age, comorbidities, NSAIDs, and anesthesia time by univariate logistic regression analyses. We performed multivariable logistic regression analyses to evaluate the association between steroid use on the day of tonsillectomy and reoperation for bleeding adjusted for potential confounders (sex, age, age squared, indication for surgery, baseline hypertension, NSAIDs use, and anesthesia time). We used age and age squared as a priori confounders because an observational study reported a nonlinear relationship between posttonsillectomy hemorrhage and age.20 We used anesthesia time as a proxy for surgical time because our database did not include data on surgical time. We performed a stepwise selection of covariates to develop the final regression model using Akaike information criterion for parsimony.21 Furthermore, we estimated the adjusted risk difference of steroid use on reoperation for bleeding using Stata’s postestimation margins command. We estimated 2 probabilities of reoperation under each exposure status for each patient regardless of the actual exposure: one assuming the patient did receive steroid on the day of surgery and the other assuming the patient did not. Using the inverse logit function, we calculated these probabilities with covariates other than steroid use held at their observed values. These probabilities were then averaged over the entire study population. The adjusted risk difference here refers to the difference between the means of calculated probabilities.22 We compared the adjusted risk difference with the previously suggested acceptable margin of increased risk (1.5%).7

To test the robustness of our results and address an issue of the small number of events (ie, reoperation) per probable confounder,23 we estimated the risk differences by doubly robust estimation methods using propensity scores, ie, augmented inverse probability weighting (AIPW) estimator24 and inverse probability weighting with regression adjustment (IPWRA) estimator (Wooldridge double-robust estimator).25 These 2 methods combine an outcome regression model and a propensity score weighting method such that the effect estimator is unbiased if either outcome model or propensity score model is correctly specified. We calculated the propensity scores for receiving steroid on the day of tonsillectomy by multivariable logistic regression including sex, age, age squared, indication for tonsillectomy (infection or noninfection), whether adenoidectomy or tympanostomy tube insertion was performed simultaneously with tonsillectomy, hypertension, diabetes, asthma, preoperative inhaled steroid use, use of NSAIDs and antibiotics on the day of tonsillectomy, and the year of surgery. We used the same outcome model as the primary logistic regression model for doubly robust estimation methods.

We conducted sensitivity analyses by categorizing patients into 3 groups to exclude those who received steroids postoperatively from the reference group and to assess the effect of postoperative steroid use on bleeding risk. This was because steroids were used postoperatively in a substantial number of patients (30% of all patients) in this study. We defined the 3 groups according to the pattern of steroid use: never administered perioperatively; administered on the day of tonsillectomy (including those who were administered postoperatively); administered only postoperatively. We assessed the association between steroid use and reoperation for bleeding by multivariable stepwise logistic regression analyses adjusted for the same variables as in the primary analysis using Akaike information criterion. We also evaluated differences in the adjusted risk as described above except for estimating propensity scores for receiving steroids by multinomial logistic regression.

Posttonsillectomy hemorrhage is classified as primary (within 24 hours of surgery) and secondary (more than 24 hours after surgery).3 Primary hemorrhage is generally attributed to surgical technique. Secondary hemorrhage is thought to be caused by premature separation of the eschar and less likely related to surgical technique. We also performed sensitivity analyses by defining the main outcome as reoperation 2–14 days after the index tonsillectomy with the aim of complete exclusion of reoperation for primary hemorrhage because reoperation on postoperative day 1 could be due to primary hemorrhage. Furthermore, we conducted subgroup analyses as planned a priori to investigate whether the effects of steroids differed between children (age ≤18 years) and adults (>18 years).

We assumed the incidence of reoperation to be 1.5% in the no-steroid group and 3% in the steroid group according to a recent meta-analysis.10 When we set the α level at .05, the power was 0.82 considering the number of patients in the steroid group and the no-steroid group. All statistical analyses were conducted using Stata Version 14 (Stata Corp, College Station, TX). All statistical tests were 2 sided with P < .05 considered statistically significant. Continuous and categorical variables are expressed as medians (interquartile range) and number of individuals (percentage), respectively.

Back to Top | Article Outline

RESULTS

From a multihospital claims database, we identified 6513 patients from 68 institutions who underwent tonsillectomy between April 1, 2008, and March 17, 2014. Of these, we excluded 84 who received anticoagulants/antiplatelet agents (except NSAIDs) from 14 days before to 7 days after the surgery, 75 who received steroids within 14 days before tonsillectomy, 51 whose indication for surgery was presumed to be malignancy, 5 with missing anesthesia time records, and 149 with concomitant operations other than adenoidectomy or tympanostomy tube insertion. Consequently, 6149 tonsillectomy patients (3110 adults and 3039 children) were included in the study cohort.

Back to Top | Article Outline

Perioperative Steroid Use

Table 1.

Table 1.

Table 1 shows details of the steroids administered on the day of tonsillectomy and postoperatively. Steroids were used in 13.7% of the patients on the day of tonsillectomy (steroid group), with dexamethasone the most commonly used in 39.1% of patients in the steroid group. Steroids were used postoperatively (on the day following tonsillectomy or later) in 30.2% of the patients, with betamethasone the most frequently administered in 51.9% of patients receiving postoperative steroids. Steroids were started on postoperative day 1 in 78.1% of those administered steroids postoperatively. The median duration of postoperative administration was 6 days (interquartile range 3–7) in patients administered postoperative steroids.

Back to Top | Article Outline

Univariate Comparison of Patient Characteristics

Table 2.

Table 2.

Table 2 shows the demographic and baseline clinical characteristics of the tonsillectomy patients. Patients receiving steroids on the day of tonsillectomy (steroid group) were significantly older than those in the no-steroid group (25 [16–34] years vs 17 [6–30] years; P < .001). Infection was significantly more frequent in the steroid group as the indication for surgery (71.9% vs 63.5%; P < .001).

Back to Top | Article Outline

Univariate Comparison of Operative and Postoperative Characteristics

Table 3.

Table 3.

NSAIDs were more frequently administered to the steroid group than to the no-steroid group on the day of tonsillectomy (48.9% vs 42.3%; P < .001) (Table 3). The proportions of patients administered steroids and NSAIDs postoperatively (on the day after tonsillectomy or later) were significantly higher in the steroid group than in the no-steroid group (50.9% vs 27.0%, P < .001; 60.0% vs 52.1%, P < .001, respectively). The incidences of reoperation did not differ significantly between the steroid and no-steroid groups (1.8% vs 1.5%; P = .54).

Back to Top | Article Outline

Multivariable Analyses

Table 4.

Table 4.

Table 5.

Table 5.

Our primary analysis showed that steroid use on the day of tonsillectomy was not associated with reoperation for bleeding according to the multivariable logistic regression analysis (OR, 0.81; 95% confidence interval [CI], 0.45–1.43; P = .46) (Table 4). In contrast, age, age squared, and anesthesia time were significantly associated with reoperation (OR, 1.19; 95% CI, 1.12–1.27; P < .001; OR, 0.997; 95% CI, 0.996–0.998; P < .001; OR, 1.11 per 10 minutes; 95% CI, 1.06–1.17; P < .001, respectively). The adjusted risk difference in reoperation for steroid use on the day of tonsillectomy was −0.30% (95% CI, −1.05 to 0.45) (Table 5). Risk differences estimated by doubly robust estimation methods were −0.19% (95% CI, −1.03 to 0.65) for AIPW and −0.22% (95% CI, −1.02 to 0.59) for IPWRA. None of the upper limits of 95% CIs of the estimated risk differences exceeded 1.5%.

Back to Top | Article Outline

Sensitivity Analyses

We divided the patients into 3 groups according to the pattern of steroid use (never administered perioperatively; administered on the day of tonsillectomy [including those who were administered postoperatively]; and administered only postoperatively). The incidence of reoperation was 1.5% (60 of 3876) in the never-administered group, 1.8% (15 of 843) in patients administered steroids on the day of tonsillectomy, and 1.3% (19 of 1430) in patients administered steroids only postoperatively. The multivariable logistic regression analysis revealed that steroid use (both administered on the day of surgery and administered only postoperatively) was not significantly associated with reoperation for bleeding (OR, 0.76; 95% CI, 0.42–1.37; P = .37; OR, 0.81; 95% CI, 0.48–1.37; P = .43, respectively) (Table 6). Adjusted risk differences for reoperation compared to never-administered patients were −0.39% (95% CI, −1.17 to 0.40) for steroid administered on the day of tonsillectomy and −0.31% (95% CI, −1.04 to 0.42) for steroid administered only postoperatively (Table 5). Risk differences for steroid administered on the day of tonsillectomy and steroid administered only postoperatively estimated by doubly robust estimation methods were −0.28% (95% CI, −1.15 to 0.59) and −0.34% (95% CI, −1.07 to 0.39) for AIPW, and −0.32% (95% CI, −1.15 to 0.52) and −0.34% (95% CI, −1.07 to 0.40) for IPWRA, respectively (Table 5).

Table 6.

Table 6.

We found no significant association between steroid use and reoperation when we confined reoperation to that which occurred 2–14 days after the index tonsillectomy (Supplemental Digital Content, Table 1, http://links.lww.com/AA/C147). Neither did this confinement lead to substantial changes in risk differences from the primary analyses.

Back to Top | Article Outline

Subgroup Analyses of Adults and Children

Subgroup analyses revealed that steroid use was not significantly associated with an increase in reoperation incidence in either adults or children (OR, 0.73; 95% CI, 0.38 to 1.38; P = .33; OR, 1.18; 95% CI, 0.34 to 4.11; P = .80, respectively) (Table 4). Moreover, the upper limits of 95% CIs of the estimated risk differences were below 1.5% in both adults and children (Table 5). We observed similar results when we categorized steroid use into 3 and defined the outcome as reoperation within 2–14 days after the index tonsillectomy in sensitivity analyses for both adults and children (Table 6, Supplemental Digital Content, Table 1, http://links.lww.com/AA/C147).

Back to Top | Article Outline

DISCUSSION

We investigated the association between perioperative steroid use, including postoperative steroids, and reoperation for posttonsillectomy bleeding in this observational cohort study using a multihospital claims database. We observed no significant association between perioperative steroid use and reoperation 1–14 days after the procedure. The subgroup analyses showed similar results in both adults and children. The adjusted risk differences estimated by the logistic regression model were all within the suggested threshold of 1.5%, and these results were also confirmed by doubly robust estimation methods.

Steroid use on the day of tonsillectomy had no significant association with reoperation for bleeding in the primary analysis of all patients and subgroup analyses of adults and children. That there was no significant association between steroid use and reoperation does not necessarily mean that there is no increased risk of steroid use considering the wide CIs of the ORs, especially in children. However, the CIs of risk differences between patients administered and not administered steroids were all within the suggested acceptable threshold of 1.5% in both adults and children in subgroup and sensitivity analyses. These results support the safety of perioperative steroid use for tonsillectomy.

Our subgroup analyses showed no significant association between steroid use on the day of tonsillectomy and reoperation in both adults and children. This finding was consistent in adults, but not in children, with a recent meta-analysis reporting an increased frequency of reintervention for posttonsillectomy bleeding in children receiving systemic steroids on the operative day (OR, 3.43; 95% CI, 1.29–9.13).10 However, that meta-analysis could have been affected by including 1 trial of pediatric patients that was terminated prematurely due to a relatively high incidence of reoperation (5.2%, 8 of 154 participants).5

Excluding patients administered postoperative steroids from the reference group did not affect the result in our subgroup analysis of children (OR, 1.06; 95% CI, 0.30–3.81). This result differs from that of a larger observational study reporting a significantly higher reoperation rate in the steroid group in children (OR, 2.50; 95% CI, 1.47–4.23).12 This discrepancy may be partly attributable to different eligibility criteria because the aforementioned observational study excluded patients who had received intravenous steroids on any day except the day of tonsillectomy. In our study, a substantial number of patients (almost 30% in both adults and children) received postoperative intravenous steroids. We also found a significantly longer anesthesia time in children administered steroid only on the day of tonsillectomy than in children never administered steroid perioperatively (107 minutes [75–140] vs 97 minutes [80–116]; P = .02). In other words, it is possible that steroids were administered more frequently to patients with perioperative complications indicated by a long anesthesia time than to patients with an uneventful perioperative course, especially in children. Thus, if these conditions also had been true in the cohort of the above-described study, the exclusion of patients administered intravenous steroids postoperatively could have caused selection bias.

Postoperative steroids were not associated with an increased risk of reoperation. If steroids truly increase the risk of posttonsillectomy bleeding in a dose-dependent manner, we should have observed an increased risk in patients administered steroids postoperatively because most of those patients received multiple doses of steroids. Moreover, a recent study reported that a 1-week administration of oral prednisolone after tonsillectomy resulted in faster reepithelialization of the tonsillectomy bed.17 Thus, we speculate that steroids may benefit recovery rather than do harm.

This study revealed that steroid use for tonsillectomy appears much less frequent in Japan, especially for children than in the United States, where according to 1 study, approximately 70% of children received steroids.26 One possible reason for the lower proportion is that the use of steroids on the day of tonsillectomy is not approved for the prevention of postoperative nausea and vomiting by the health insurance system in Japan. Another possible reason is that tonsillectomy in the United States is usually performed as an outpatient surgery, whereas in Japan, tonsillectomy is usually an inpatient procedure. On the other hand, steroid administration after otolaryngological surgery is approved in Japan, explaining the more frequent use of postoperative steroids.

The number of patients included in this study was disproportionately small for the number of institutions. This disproportion is partly because the number of institutions registered to the database gradually increased during the study period. The number of institutions where tonsillectomies were undertaken reached 68 as of March 2014, and the monthly number of patients per hospital was 3–4 on average. Notwithstanding, the average patient number is much smaller than that in the United States.20 We presume that this gap is attributable to the difference in the health care system: patients undergoing surgery are not aggregated in Japan as much as in the United States.

While we found no significant relationship between steroid use on the day of tonsillectomy and reoperation for bleeding, a prolonged anesthesia time had a significant association with reoperation. We could not identify patients who underwent reoperation on the same day as the initial tonsillectomy because reoperation on the same day cannot be charged to the health insurance system in Japan. Also, anesthesia time for multiple procedures including reoperation on the same day is merged in our database. It has been shown that anesthesia time and surgical time are well correlated.27,28 Thus, patients with prolonged anesthesia time may include not only those with prolonged surgery due to intraoperative complications (eg, difficulty in achieving hemostasis) but also those with reoperation on the same day of tonsillectomy. Although the reason for the association between prolonged anesthesia and reoperation remains unclear, we conjecture that perioperative complications, as reflected by longer anesthesia time, may have affected the incidence of reoperation for bleeding.

Several limitations of this study should be considered. First, the database used does not include information on surgical techniques and institutions. There is little evidence that one method of tonsillectomy has a lower risk of bleeding than another,29 although the surgical technique may affect the incidence of posttonsillectomy bleeding.30,31 Furthermore, we observed no substantial change in results by confining reoperation to that within 2–14 days after the initial tonsillectomy to exclude reoperation for primary hemorrhage, which was thought to be affected by surgical technique19 or institutional practice. Second, we could not estimate the dose effect of steroids on bleeding risk because the database records only the number of ampules administered and not exact dosages. However, dose-dependent increases in the risk of bleeding have not been established.9,32 Also, we found no significant association between steroid use and reoperation among patients administered steroids postoperatively although most of these patients received multiple doses of steroids. Third, we could not discriminate patients administered steroids intraoperatively from those administered steroids postoperatively on the day of surgery because our database only included the data on the date when drugs were prescribed. Considering the delayed onset of action of steroids (6–8 hours),33 however, we believe that the difference between intraoperative and postoperative steroids on the day of surgery hardly affected the incidence of reoperation after the first postoperative day. Fourth, residual confounding by indications for steroid use can remain. Although we included anesthesia time as an adjusting factor, this is only an indirect index of perioperative complications. Last, our results may not apply to patients undergoing tonsillectomy in other settings (eg, at an academic hospital) because the database used in this study included records of only nonacademic acute-phase hospitals. Besides, anesthesia time was quite long for tonsillectomy in this study compared to that in other studies34,35; therefore, the unit OR of reoperation for anesthesia time may not be applicable in other countries. Our results must be interpreted based on the specific clinical situation because the acceptable incremental risk depends on the risk–benefit balance and clinical context.

Back to Top | Article Outline

CONCLUSIONS

In this observational study, steroid use on the day of tonsillectomy and/or postoperatively was not associated with an increased risk of reoperation for bleeding. Although the wide range of CIs for the OR could not eliminate the possibility of increased risk, especially in children, the incremental risks of reoperation for steroid use were within an acceptable range in both adults and children. We conclude that our results support the safety of perioperative steroid use for tonsillectomy, considering the magnitude of risk of reoperation for bleeding.

Back to Top | Article Outline

ACKNOWLEDGMENTS

The authors thank Masaki Nakamura (Medical Data Vision Co, Ltd, Tokyo, Japan) for supplying the dataset.

Back to Top | Article Outline

DISCLOSURES

Name: Yoshihisa Miyamoto, MD, MPH.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflicts of Interest: None.

Name: Maki Shinzawa, MD, PhD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflicts of Interest: None.

Name: Shiro Tanaka, PhD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflicts of Interest: None.

Name: Sachiko Tanaka-Mizuno, PhD.

Contribution: This author helped design the study, conduct the study, and write the manuscript.

Conflicts of Interest: None.

Name: Koji Kawakami, MD, PhD.

Contribution: This author helped design the study, conduct the study, and write the manuscript.

Conflicts of Interest: K. Kawakami received honoraria from Astellas, AbbVie, Eisai, Mitsubishi Tanabe Pharma, Takeda Pharmaceutical Company Limited, and Sanofi K.K., as well as consulting fees from Olympus, Kyowa Hakko Kirin, Kaken Pharmaceutical, and Otsuka Pharmaceuticals.

This manuscript was handled by: Ken B. Johnson, MD.

Back to Top | Article Outline

REFERENCES

1. Steward DL, Grisel J, Meinzen-Derr J. Steroids for improving recovery following tonsillectomy in children. Cochrane Database Syst Rev. 2011:CD003997. doi: 10.1002/14651858.CD003997.pub2.
2. Diakos EA, Gallos ID, El-Shunnar S, Clarke M, Kazi R, Mehanna H. Dexamethasone reduces pain, vomiting and overall complications following tonsillectomy in adults: a systematic review and meta-analysis of randomised controlled trials. Clin Otolaryngol. 2011;36:531–542.
3. Baugh RF, Archer SM, Mitchell RB, et al.; American Academy of Otolaryngology-Head and Neck Surgery Foundation. Clinical practice guideline: tonsillectomy in children. Otolaryngol Head Neck Surg. 2011;144:S1–S30.
4. Scottish Intercollegiate Guidelines Network (SIGN), Healthcare Improvement Scotland. Management of sore throat and indications for tonsillectomy: SIGN Guideline No 117. Available at: http://www.sign.ac.uk/pdf/sign117.pdf. Accessed January 3, 2017.
5. Czarnetzki C, Elia N, Lysakowski C, et al. Dexamethasone and risk of nausea and vomiting and postoperative bleeding after tonsillectomy in children: a randomized trial. JAMA. 2008;300:2621–2630.
6. Gallagher TQ, Hill C, Ojha S, et al. Perioperative dexamethasone administration and risk of bleeding following tonsillectomy in children: a randomized controlled trial. JAMA. 2012;308:1221–1226.
7. Czarnetzki C, Elia N, Tramèr MR. Bleeding risk and dexamethasone use in children undergoing tonsillectomy. JAMA. 2013;309:437.
8. Geva A, Brigger MT. Dexamethasone and tonsillectomy bleeding: a meta-analysis. Otolaryngol Head Neck Surg. 2011;144:838–843.
9. Shargorodsky J, Hartnick CJ, Lee GS. Dexamethasone and postoperative bleeding after tonsillectomy and adenotonsillectomy in children: a meta-analysis of prospective studies. Laryngoscope. 2012;122:1158–1164.
10. Plante J, Turgeon AF, Zarychanski R, et al. Effect of systemic steroids on post-tonsillectomy bleeding and reinterventions: systematic review and meta-analysis of randomised controlled trials. BMJ. 2012;345:e5389.
11. Bellis JR, Pirmohamed M, Nunn AJ, et al. Dexamethasone and haemorrhage risk in paediatric tonsillectomy: a systematic review and meta-analysis. Br J Anaesth. 2014;113:23–42.
12. Suzuki S, Yasunaga H, Matsui H, Horiguchi H, Fushimi K, Yamasoba T. Impact of systemic steroids on posttonsillectomy bleeding: analysis of 61 430 patients using a national inpatient database in Japan. JAMA Otolaryngol Head Neck Surg. 2014;140:906–910.
13. Windfuhr JP, Chen YS, Remmert S. Hemorrhage following tonsillectomy and adenoidectomy in 15,218 patients. Otolaryngol Head Neck Surg. 2005;132:281–286.
14. Deeks JJ, Higgins JPT, Altman DG. Higgins JPT, Green S. Analysing data and undertaking meta-analyses. Cochrane Handbook for Systematic Reviews of Interventions. 2008: 243–Chichester, England: Wiley-Blackwell296.
15. Vandenbroucke JP, von Elm E, Altman DG, et al.; STROBE Initiative. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. Epidemiology. 2007;18:805–835.
16. Palme CE, Tomasevic P, Pohl DV. Evaluating the effects of oral prednisolone on recovery after tonsillectomy: a prospective, double-blind, randomized trial. Laryngoscope. 2000;110:2000–2004.
17. Park SK, Kim J, Kim JM, Yeon JY, Shim WS, Lee DW. Effects of oral prednisolone on recovery after tonsillectomy. Laryngoscope. 2015;125:111–117.
18. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147:573–577.
19. Williams JD, Pope TH Jr. Prevention of primary tonsillectomy bleeding. An argument for electrocautery. Arch Otolaryngol. 1973;98:306–309.
20. Wei JL, Beatty CW, Gustafson RO. Evaluation of posttonsillectomy hemorrhage and risk factors. Otolaryngol Head Neck Surg. 2000;123:229–235.
21. Akaike H. A new look at the statistical model identification. IEEE Trans Automat Contr. 1974;19:716–723.
22. Kleinman LC, Norton EC. What’s the risk? A simple approach for estimating adjusted risk measures from nonlinear models including logistic regression. Health Serv Res. 2009;44:288–302.
23. Cepeda MS, Boston R, Farrar JT, Strom BL. Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders. Am J Epidemiol. 2003;158:280–287.
24. Funk MJ, Westreich D, Wiesen C, Stürmer T, Brookhart MA, Davidian M. Doubly robust estimation of causal effects. Am J Epidemiol. 2011;173:761–767.
25. Wooldridge JM. Inverse probability weighted estimation for general missing data problems. J Econom. 2007;141:1281–1301.
26. Mahant S, Keren R, Localio R, et al.; Pediatric Research in Inpatient Settings (PRIS) Network. Variation in quality of tonsillectomy perioperative care and revisit rates in children’s hospitals. Pediatrics. 2014;133:280–288.
27. Silber JH, Rosenbaum PR, Even-Shoshan O, et al. Estimating anesthesia time using the Medicare claim: a validation study. Anesthesiology. 2011;115:322–333.
28. Burgette LF, Mulcahy AW, Mehrotra A, Ruder T, Wynn BO. Estimating surgical procedure times using anesthesia billing data and operating room records. Health Serv Res. 2017;52:74–92.
29. Pinder DK, Wilson H, Hilton MP. Dissection versus diathermy for tonsillectomy. Cochrane Database Syst Rev. 2011:CD002211. doi: 10.1002/14651858.CD002211.pub2.
30. Lowe D, van der Meulen J; National Prospective Tonsillectomy Audit. Tonsillectomy technique as a risk factor for postoperative haemorrhage. Lancet. 2004;364:697–702.
31. Tomkinson A, Harrison W, Owens D, Harris S, McClure V, Temple M. Risk factors for postoperative hemorrhage following tonsillectomy. Laryngoscope. 2011;121:279–288.
32. Brigger MT, Cunningham MJ, Hartnick CJ. Dexamethasone administration and postoperative bleeding risk in children undergoing tonsillectomy. Arch Otolaryngol Head Neck Surg. 2010;136:766–772.
33. Becker DE. Basic and clinical pharmacology of glucocorticosteroids. Anesth Prog. 2013;60:25–31.
34. Ved SA, Walden TL, Montana J, et al. Vomiting and recovery after outpatient tonsillectomy and adenoidectomy in children. Comparison of four anesthetic techniques using nitrous oxide with halothane or propofol. Anesthesiology. 1996;85:4–10.
35. Kako H, Corridore M, Seo S, Elmaraghy C, Lind M, Tobias JD. Tracheal extubation practices following adenotonsillectomy in children: effects on operating room efficiency between two institutions. Paediatr Anaesth. 2017;27:591–595.

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2017 International Anesthesia Research Society