The Effects of Smoking on the Rate of Postoperative Hemorrhage After Tonsillectomy and Uvulopalatopharyngoplasty

Objective  To evaluate the rate of posttonsillectomy hemorrhage and any change that smoking causes in patients who undergo tonsillectomy.

Design  Retrospective chart review.

Setting  Military tertiary referral center.

Patients  The study included 1036 adult patients who underwent tonsillectomy either alone or in conjunction with another procedure.

Main Outcome Measures  The rate of postoperative hemorrhage was reviewed in all patients who underwent tonsillectomy at our medical center, and an investigation was conducted to determine whether smoking caused any alteration in this rate.

Results  A total of 1010 patients were included in the study, with a total bleeding rate of 6.7%. There was a significant increase in the rate of bleeding in all patients when they were divided into smokers and nonsmokers (10.2% and 5.4%, respectively; P = .01). The large difference was found by subset analysis to be attributable to a marked increase in postoperative hemorrhage in the patients who underwent uvulopalatopharyngoplasty (10.9% in smokers vs 3.3% in nonsmokers; P = .006) and remained significant when they were further subdivided by sex. Men who underwent tonsillectomy alone also bled significantly more than women (11.2% and 5.4%, respectively; P = .02). All other subsets analyzed did not reach statistical significance.

Conclusions  Smoking does appear to increase the rate of posttonsillectomy hemorrhage in patients who undergo uvulopalatopharyngoplasty with tonsillectomy, but not in those who undergo tonsillectomy alone. This modifiable risk factor may help clinicians further counsel their patients before surgery, but further study is needed to ascertain that these findings apply to a broader patient base.

Tonsillectomy, despite many changes in the indications and fervor for the operation over the years, remains one of the most commonly performed surgical procedures in otolaryngology. After extensive research into reducing the perioperative morbidity of tonsillectomy and uvulopalatopharyngoplasty (UPPP) with tonsillectomy, there remains an appreciable risk of complications, the greatest of which is postoperative hemorrhage. Numerous changes in operative techniques, perioperative medications administered, postoperative pharmacotherapy, and precautions have been unable to remove postoperative hemorrhage as the most common complication of both procedures. Numerous risk factors, including age, sex, instruments used, phases of the moon, and even hair color, have been studied.^1-5

The detrimental influence of smoking on wound healing is well known.⁶ Yet to date, no studies have looked specifically at the effects of smoking on posttonsillectomy or post-UPPP hemorrhage. In military medicine, we perform a large number of tonsillectomies and UPPPs in the young adult population. It has been our anecdotal experience that smokers have a higher rate of postoperative hemorrhage than do nonsmokers. To evaluate whether smoking is truly a modifiable risk factor, we undertook a retrospective review of all patients who underwent tonsillectomy or UPPP at our medical center in the last 5 years.

After approval was obtained from the institutional review board, the surgical database of our tertiary care training hospital was searched from June 2000 to September 2005 for all patients with the Current Procedural Terminology codes corresponding to tonsillectomy, adenotonsillectomy, or UPPP. All patients younger than 18 years were excluded from the study. A separate database was created that included patients with Current Procedural Terminology codes corresponding to control of oropharyngeal hemorrhage and any patients with visits to either the otolaryngology clinic or the emergency department within 15 days of surgery. The 2 collected databases were merged, and the medical records of the patients were reviewed. We excluded from the combined database all patients who underwent a tonsillectomy as part of the treatment of a known carcinoma of the head and neck, yielding a total sample size of 1036 patients. An additional 25 patients with incomplete medical records and 1 patient who was diagnosed as having lupus anticoagulant were also removed, for a final sample size of 1010 patients.

The following information was collected on the cohort: age, sex, preoperative diagnosis, surgical procedure(s) performed, tobacco use (specifically cigarette smoking), number of visits within 15 days of the surgery, and reason for the visits. The patients were then grouped according to whether they were smokers or nonsmokers and whether they underwent tonsillectomy alone or in conjunction with UPPP. Posttonsillectomy hemorrhage was considered significant only if it required intervention for control of the bleeding. Intervention was defined as either removal of clot and electrocautery with the patient under local anesthesia or the return of a patient to the operating room for control of the hemorrhage. Cases in which no clot or bleeding was visualized were considered insignificant and were not included in the total number of hemorrhages. The χ² method was used for statistical analysis except in cases with expected sample sizes of fewer than 5, when the Fisher exact test was used.

The final sample size comprised 1010 patients (Table 1). A total of 569 patients underwent tonsillectomy alone, and 441 patients underwent tonsillectomy with UPPP. There were significantly fewer women in the smoking group (P = .007) and the UPPP group (P < .0001) and significantly fewer men in the tonsillectomy group (P < .001). Two hundred seventy-four of the patients (27.1%) were smokers.

There were a total of 68 bleeding episodes that required intervention in 56 patients, for a total posttonsillectomy hemorrhage rate of 6.7%. Twenty of the bleeding episodes (in 10 smokers and 10 nonsmokers) required a return visit to the operating room, and 48 were managed in the clinic or emergency department with the patient under local anesthesia. Four bleeding episodes occurred within the first 24 hours after surgery, for a primary hemorrhage rate of 0.4%. Among the smokers, there were 28 bleeding episodes, for a rate of 10.2%. This rate was significantly higher (P = .01) than that among nonsmokers (5.4%). Subset analysis between the patients who underwent UPPP and those who underwent tonsillectomy (Table 2) showed that the difference between smokers and nonsmokers was attributable to the hemorrhage rate in the patients who underwent UPPP. The rate of bleeding episodes among smokers who underwent UPPP was more than 3 times the rate among nonsmokers (10.9 vs 3.3%; P = .006). In the tonsillectomy group, the difference in the incidence was not statistically different between the smokers (9.8%) and the nonsmokers (7.2%; P = .41).

To control for any influence of sex on the hemorrhage rates, the groups were then divided once more (Table 3 and Figure). The overall rate of postoperative hemorrhage was higher in men (7.9%) than in women (4.9%); however, this difference did not reach statistical significance (P = .10). Also, the postoperative hemorrhage rate among male smokers was higher than that among female smokers (12.0% vs 6.0%) but also failed to reach statistical significance (P = .22). Furthermore, the rate of bleeding among male nonsmokers (6.0%) was higher than that among female nonsmokers (4.6%), but, again, this did not reach statistical significance (P = .51). However, the rate of postoperative hemorrhage was significantly greater in both the men (P = .05) and the women (P = .03) who smoked and underwent UPPP. Also, there was an increased rate of bleeding among men who underwent tonsillectomy (11.2%) compared with women (5.3%; P = .02). The rate of bleeding among patients who underwent UPPP, however, was similar among men and women (P = .51).

Wide variation exists in the rate of posttonsillectomy hemorrhage reported in the literature to date. A key contributing factor to this variability is the lack of consensus on what constitutes a significant postoperative hemorrhage. A majority of the literature uses the number of patients who require a return to the operating room.^7-10 At least 2 classification schemes have been proposed in an attempt to stratify the severity of the hemorrhage at presentation.^11,12 Several other studies have included any bleeding episode that requires intervention.^13,14 Obviously, those studies that rely on a return trip to the operative theater will have lower hemorrhage rates. In our military medical center, control of bleeding with the patient under local anesthesia is often attempted before a return to the operating room. This practice of hemorrhage control led to the inclusion of all patients who required medical attention for control of bleeding. The hemorrhage rate was increased from 2.0% to 6.7%, but we believe that is a more accurate measurement based on the current practice at our medical center. Many studies have looked at the rate of postoperative hemorrhage after tonsillectomy, but fewer studies have focused on adults only. Posttonsillectomy hemorrhage rates are generally thought to be higher in adults than in children, with a range of 1.5% to 18.0% reported in the literature, but the majority of rates reported are between 3.0% and 5.0%.^2,8,13,15-18

Studies have taken differing approaches to UPPP and tonsillectomy vs tonsillectomy alone. Most exclude all patients who have undergone concomitant procedures, while some lump both groups together. We looked at UPPP and tonsillectomy patients both as a group and as subsets. Rates of hemorrhage after UPPP have not been studied as extensively as those after tonsillectomy. The range encountered in the literature varies from 0.6% to 14.0%.^19-22 In the current study, the same classification that was previously discussed for inclusion as a posttonsillectomy hemorrhage was used in the patients who underwent UPPP. The rate of patients with multiple hemorrhages in the present study (14.0%) is similar to the published rates of Wei et al² (12%) and Liu et al¹⁴ (14%). The rate of multiple bleeding episodes was similar across all groups evaluated in our study. The increased rate of hemorrhage found in men is also consistent with several previous studies.^4,10,16 Other studies, however, have not found sex significant.^2,3,15,18

When smokers and nonsmokers were compared, a significant difference in the rate of hemorrhage was apparent. To our knowledge, no previous study has noted this correlation. One study in Spain included smoking among a variety of factors; however, in that study, smoking was found to be insignificant.²³ One limitation of the Spanish study was a small sample size. As previously mentioned, in our study the patients who underwent UPPP were the cause of the divergent posttonsillectomy hemorrhage rates when the smokers and nonsmokers were compared. The reasons for the variation between the tonsillectomy and UPPP subsets are not known. The tonsillectomy portion of the procedures is performed in an identical manner. We considered the sex disparity between the groups a possible contributing factor. However, when the bleeding rates among male and female UPPP patients were considered, they remained significantly higher among both male and female smokers. One study that looked at the influence of smoking on the pharyngeal mucosa found that male patients had a higher degree of alterations on histologic examination when compared with dose-matched female patients.²⁴ No attempt was made to quantify the tobacco use in any patient, which might also be an influencing factor.

Smoking has previously been shown to impair random flap survival and to increase wound complications.⁶ The possible etiology of the influence of tobacco use on healing includes impaired epithelialization, decreased oxygen delivery, microvascular injury, and effects on leukocytes, macrophages, and platelets.⁶ The smoke may also have a direct drying and irritant effect on the pharynx that becomes a factor in pharyngeal surgery. Further division of patients by amount of tobacco use could also be informative because several studies have shown a dose-dependent response to the change in the pharyngeal mucosa.^24,25 The timing and duration of any perioperative smoking cessation may also be important.

Our study has several strengths. First, in a restricted environment such as the military, all patients receive their entire care at our hospital. Patients who live outside the immediate vicinity of the hospital receive their care at one of the civilian or military facilities in their area and were not included in this study. Therefore, because all of our patients were treated at 1 institution, we strongly believe that we have an accurate representation of the postoperative hemorrhage rates. Second, we have had a computerized system at our medical center for more than 10 years that records the date and place of every visit, including surgical procedures. Consequently, we have been able to record with great accuracy which patients were treated for postoperative hemorrhage. Finally, because of our electronic medical record, very few charts (n = 25) were excluded as a result of incomplete documentation. This small number of patients is not likely to have affected our results, given the size of the overall cohort.

There are some weaknesses to our study. First, it is a retrospective chart review. Therefore, causation cannot be demonstrated. The patients also were not randomly distributed among the groups studied. Furthermore, we did not make an attempt to determine which patients stopped smoking just before and after undergoing surgery, nor was there an attempt to stratify patients by the amount of cigarettes smoked. As mentioned above, the number of cigarettes may have a dose-dependent effect on oropharyngeal healing.

In conclusion, smoking appears to increase the rate of hemorrhage in patients who undergo UPPP but not in those who undergo tonsillectomy alone. Further investigation of this relationship is needed, with stratification of patients by the number of cigarettes smoked and attention to the length of time before and/or after surgery that patients refrain from smoking. Better understanding of this connection could be helpful in altering one of the few potentially modifiable risk factors for posttonsillectomy hemorrhage.

Correspondence: Sean M. Demars, MD, Department of Otolaryngology, Madigan Army Medical Center, 9040A Fitzsimmons Dr, MCHJ-SET, Tacoma, WA 98431 (sean.demars@us.army.mil).

Submitted for Publication: October 11, 2007; final revision received January 4, 2008; accepted January 6, 2008.

Author Contributions: Dr Demars had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Demars and Harsha. Acquisition of data: Demars. Analysis and interpretation of data: Demars, Harsha, and Crawford. Drafting of the manuscript: Demars. Critical revision of the manuscript for important intellectual content: Demars, Harsha, and Crawford. Study supervision: Harsha and Crawford.

Financial Disclosure: None reported.

Previous Presentation: This study was presented as a poster at the American Academy of Otolaryngology Annual Meeting; September 17-20, 2006; Toronto, Ontario, Canada.