- Email: [email protected]
Effect of tonsillar fossa closure on postoperative pain and bleeding risk after tonsillectomy
International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
Contents lists available at SciVerse ScienceDirect
International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl
Effect of tonsillar fossa closure on postoperative pain and bleeding risk after tonsillectomy Bruce H. Matt a,*, Bryan J. Krol a,b, Yan Ding c, Beth E. Juliar c a
Indiana University School of Medicine, Department of Otolaryngology – Head & Neck Surgery, School of Medicine, Indiana University, Indianapolis, IN, USA Head & Neck Surgery Associates PSC, Ft. Thomas, KY, USA c Division of Biostatistics, Department of Medicine, School of Medicine, Indiana University, Indianapolis, IN, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 5 August 2012 Received in revised form 27 August 2012 Accepted 3 September 2012 Available online 25 September 2012
Objective: To determine if closing the tonsil fossa after tonsillectomy leads to less pain and bleeding. Study design: Interventional, Randomized, Single Blind, Active Control, Single Group Assignment, Safety/ Efficacy Study. Follow-up: 2 months. Methods: Generalized Estimating Equations (GEE) analyzed effects of tonsillar pillar closure, surgeon experience and interaction on outcomes. Setting: Primarily academic tertiary care referral center, institutional practice, primarily children, both ambulatory and hospitalized care. Subjects: 763 subjects (age 8–264 months) undergoing tonsillectomy. Exclusions: suspected malignancy or active peritonsillar abscess. At the discretion of the attending surgeon, patients undergoing tonsillectomy during the 4 year study period were offered participation. A computer selected the side closed. 131 subjects withdrew (complete lack of follow-up information) after the first 72 h. Intervention: 3-0 chromic sutures on tapered needles to close one tonsillar fossa. The subject was not told which side was closed. Main outcome measures: postoperative bleeding (at any time) and pain reported was sought on days 1, 7, 14, 21, and 28. Results: Closure of the tonsillar fossa did not change the risk of bleeding. Closing the tonsillar fossa had a 40% increase in the odds ratio of postoperative pain. In the tonsillar fossa sides left open, greater surgeon experience decreased the risk of bleeding. In closed sides, enriched surgeon experience increased the risk of bleeding (p < .0.05). Conclusions: Suture closure of the tonsillar fossa after tonsillectomy does not reduce the risk of bleeding. Additionally, closing the tonsillar fossa increased postoperative pain. Level of evidence: 1b (individual randomized controlled trial). ß 2012 Elsevier Ireland Ltd. All rights reserved.
Keywords: Tonsillectomy Postoperative pain Hemostasis Pillar sutures Tonsillar fossa closure
1. Introduction 1.1. Tonsillectomy incidence Tonsillectomy is one of the most common operations performed by otolaryngologists [1,2] 287,000 tonsillectomies (with or without adenoidectomy) were performed in 1996 on children under 15 years of age, almost doubling to 530,000 in 2006 [1–3]. This represents roughly $701 million/year for tonsillectomy in the United States alone [4]. Techniques to improve the experience for the patient by having better outcomes and lowering
* Corresponding author at: Suite 0860, Riley Hospital, 705 Riley Hospital Drive, Indianapolis, IN 46202-5230, USA. Tel.: +1 317 944 6670; fax: +1 317 948 4720. E-mail address: [email protected] (B.H. Matt). 0165-5876/$ – see front matter ß 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijporl.2012.09.004
risks and discomfort, as well as making the operation and postoperative care easier for the surgeon have been sought for decades, if not centuries [5]. 1.2. What can be done to reduce postoperative issues and complications after tonsillectomy? This paper is not focused on the differences between techniques of removal, but on a method to reduce postoperative complications after complete tonsillectomy. The specific method chosen here to test is suturing together of the tonsillar pillars to close the tonsillar fossa. Immediate wound closure allows for healing by primary intention, which is faster than allowing the wound to close spontaneously by secondary intention healing. Mechanical integrity of a healed wound allows for quicker return to normal function [6].
1800
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
The feasibility and value of tonsillar fossa closure has been studied by several investigators. Weighill et al. [7] evaluated 60 consecutive adult (age greater than 16 years) subjects, and by randomization, sutured one side closed after tonsillectomy with 30 plain catgut. Weighill placed 3 interrupted sutures to approximate the anterior and posterior tonsillar pillars. No statistically significant difference was noted between the sewn and unsewn sides at postoperative days 1, 2, and 3. More than half the subjects were lost to follow up at 6 weeks postoperatively. 3 Subjects developed a palatal hematoma on the sutured side, while one subject had a postoperative hemorrhage from the unsutured side. Nandapalan and McIlwain [8] compared 50 consecutive adult patients (age greater than 15 years) undergoing tonsillectomy, and by randomization, sutured one side closed with 2-0 polydioxanone (PDS), with 3 interrupted sutures, sewing the pillars to the tonsillar fossa bed to eliminate dead space. The sutured side was more painful at postoperative days 1, 2, and 3, but thereafter the unsutured side was more painful at postoperative days 4–10. Eighteen of 19 patients complained of otalgia on the unsutured side. None had a palatal hematoma and 1 had mild postoperative hemorrhage from the unsutured side. Ramjettan and Singh [9] compared two non-randomized, consecutive groups: an initial control group of 20 whose tonsillar pillars were not sewn, and a second experimental group of 20 whose tonsillar pillars were sewn with 3-0 chromic catgut in a running, continuous fashion, bilaterally. The ages ranged from 3 to 35 years. The adults had more pain than the children in both groups at the immediate postoperative measurement and at the 24-h point. The difference in pain scores for the two groups (control vs. treatment) was not statistically significant. In the sutured group, 2 subjects had palatal hematomas, and one subject had temporary nasal regurgitation. No subjects had postoperative bleeding. Genc et al. [10] in 39 patients, ages 3–15 years, at the preference of the attending surgeon, selected one tonsillar fossa to be sutured closed with 4-0 plain catgut on a sharp needle in 4–5 interrupted sutures. Genc et al. found no difference in pain on the first postoperative day. However, in contrast to the above studies, pain was less on the sutured side from postoperative days 3–10. Edema was worse on postoperative day 10 on the sutured side, but healing was better. None of the subjects had palatal hematomas, or postoperative bleeding reported, however the subjects were not followed beyond postoperative day 10. The sutured side was statistically significantly more likely to have edema. The above studies suffer from small sample sizes, and included primarily adults. This study was undertaken to prospectively and systematically investigate whether a simple technique could be used to reduce complications in a large group of children undergoing tonsillectomy in a randomized trial of closed vs. open tonsillar fossa.
2.1.1. Specific aim 1: determine if suturing the tonsillar fossa has an effect on postoperative bleeding 2.1.1.1. Working hypothesis 1. The working hypothesis of this aim is that suturing closed the tonsillar fossa reduces the risk of postoperative bleeding after adenotonsillectomy. Alternatively, the null hypothesis is that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not statistically change the risk of postoperative bleeding after adenotonsillectomy. 2.1.2. Specific aim 2: determine if suturing the tonsillar fossa has an effect on postoperative pain 2.1.2.1. Working hypothesis 2. The working hypothesis of this aim is that suturing closed the tonsillar fossa would reduce the postoperative pain after adenotonsillectomy. Alternatively, the null hypothesis is that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not statistically change the postoperative pain after adenotonsillectomy. 2.1.3. Specific aim 3: determine if experience of the surgeon has an effect on postoperative bleeding 2.1.3.1. Working hypothesis 3. The working hypothesis of this aim is the more experienced the surgeon, the less postoperative bleeding for the subject. Alternatively, the null hypothesis is that, compared to a less experienced surgeon, the more experienced surgeon does not statistically change the rate of postoperative bleeding after adenotonsillectomy. 2.1.4. Specific aim 4: determine if experience of the surgeon has an effect on postoperative pain 2.1.4.1. Working hypothesis 4. The working hypothesis of this aim is the more experienced the surgeon, the less pain for the subject. Alternatively, the null hypothesis is that, compared to a less experienced surgeon, the more experienced the surgeon does not statistically change the postoperative pain after adenotonsillectomy. 2.1.5. Specific aim 5: determine if suturing the tonsillar fossa has an effect on other postoperative complications 2.1.5.1. Working hypothesis 5. The working hypothesis of this aim is suturing closed the tonsillar fossa reduces other postoperative complications after adenotonsillectomy. Alternatively, the null hypothesis is that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not statistically change the risk of other postoperative complications after adenotonsillectomy. 3. Materials and methods
2. Aims and objectives
3.1. Institutional review and subject protection
The long-range goal is to implement safe and effective surgical strategies when tonsillectomy is performed. The paper investigates a specific surgical technique in an effort to reduce postoperative complications by suture closure of the tonsillar pillars. Closed wounds require ‘‘less healing’’, as the body does not have to bring the wound edges together, and thus the raw surfaces are not exposed to abrasive disruptions.
This study was approved by the Institutional Review Board (IRB) of our institution (Indiana University Office of Research Administration, Human Subjects Office), and monitored annually by the IRB. This clinical trial was registered with http:// www.clinicaltrials.gov.
2.1. Central hypothesis
3.2.1. Inclusion criteria Our inclusion criteria were: any patient, male or female, of any ethnicity, or any age, for whom tonsillectomy was recommended for recurrent pharyngitis, obstructive sleep disorder, snoring, halitosis, or feeding difficulty associated with adenotonsillar
Our null hypothesis is: no difference in bleeding or pain or other complications exists between the sutured and unsutured tonsillar fossa.
3.2. Subject selection
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
hypertrophy, and who in the investigator’s opinion, was capable of providing reliable responses to post-operative follow-up questions. 3.2.2. Exclusion criteria Any patient for whom tonsillectomy was recommended for suspected malignancy or active peritonsillar abscess was excluded from the study. Other exclusion criteria included: inability to read or understand the post-operative questionnaire; language barriers or cognitive impairment that would prevent the patient from freely giving informed consent and failure to agree to follow-up. Other procedures in the subject were permitted. 3.2.3. Subject recruitment The option of participating in this study was offered to all eligible patients who were scheduled to undergo tonsillectomy with one of the 7 attending otolaryngologists. No financial advantage was offered to the patients. No written advertisements were used to publicize the study. 3.2.4. Study intervention 3.2.4.1. Control group. One tonsillar fossa remained open (unsutured) and would serve as a control for the closed tonsillar fossa. Because the subject acts as his own control, i.e. comparing the pain of one side to the other, differences in tonsillectomy technique are controlled (internally by the subject). This approach has the advantages of avoiding the use of a separate control group that does not receive the experimental intervention and does not require the use of a pain scale for comparing one subject to another. This experimental design also internally controls for any other methods or techniques used in each individual subject either perioperatively or postoperatively. This design is not perfect however, as younger children may not necessarily have the intellectual ability to be able to clearly indicate which, if any, side hurt. Also, the design does not specifically differentiate between pain at a particular site and referred pain. 3.2.5. Surgical procedure and closure After induction of appropriate general anesthesia, a tonsillectomy was performed. The specific method of tonsillectomy was per surgeon’s preference. Three specific methods of tonsillectomy were used: (1) electrocautery dissection, (2) cold dissection and snare, and (3) harmonic scalpel, though the vast majority was done with electrocautery dissection (refer to Table 2). Intraoperative control of bleeding was achieved as was routine for each individual otolaryngologist. After the tonsillectomy was completed, closure of a tonsillar fossa was performed by the surgeon. The side-to-close chosen was determined by a computer-generated randomization schedule. The subject and the family were not informed which side was closed. This study intervention used 3-0 chromic (absorbable) sutures on tapered needles to close only one tonsillar fossa. The other (contralateral) tonsillar fossa remained open. The anterior and posterior pillar mucosa was sewn together with 2–3 interrupted stitches of 3-0 chromic. No attempt was made to obliterate dead space, or to completely seal the fossa: a small gap inferiorly was often present with the mouth-gag in place. 3.2.6. Postoperative care Surgeon-specific routines were followed for postoperative care. Because the patient acted as his or her own control, postoperative choices left at the discretion of the surgeon included: antibiotic, analgesia, diet, and outpatient vs. inpatient care.
1801
3.2.7. Data collection and extraction Information collected included the following independent variables: age at operation, gender, indication for tonsillectomy, date of surgery, experience of surgeon, method of tonsillectomy, method of hemostasis, side closed, and known clotting abnormalities. Dependent variables were: which side was more painful on or about postoperative day 1 (the first 72 h), days 7, 14, 21, and overall assessment at the postoperative clinic visit (on or about day 28). Overall assessment included the ‘‘gestalt’’ assessment of the subject (and/or legal guardian), the ‘‘gestalt’’ assessment of the investigator and any concerns which the subject (and/or legal guardian) or investigator raised. The pain scale was ordinal. We asked, ‘‘As of today, which side hurts more?’’ and categorized the responses as [1 = left, 2 = right 3 = equal, 4 = unable to determine, 5 = no pain]. Information was obtained from the caregiver or subject, either in person, or by phone. A physician or other departmental personnel obtained the information using a scripted questionnaire. Information about the subject’s pain was sought on or about postoperative day 1 (the first 72 h), days 7, 14, 21, and at the postoperative clinic visit (on or about day 28). Additional information was obtained if postoperative bleeding occurred, including which side bled, when the bleeding occurred, what evaluation and treatment were undertaken and what the outcome was. 3.3. Statistical methods 3.3.1. Prospective sample size calculations Pre-trial estimates were for 720 subjects enrolled to obtain a sample size of 36 subjects who had postoperative bleeding (‘‘bleeders’’) (5% of total enrollment). With 36 bleeders, there is 80% power to detect a statistically significant difference of 0.33 in proportion of bleedings from closed fossae vs. proportion of bleedings from open fossae, at the 0.05 level of significance in a two-sided McNemar’s test for correlated data [11]. If 720 patients were enrolled and 95% of them reported pain, there would be 99% power to detect a difference in proportions of 0.10 at any of 5 post surgery data collection times with a McNemar’s test at alpha of 0.01 using Bonferroni’s method to adjust for multiple comparisons. 3.3.2. Statistical analysis Descriptive statistics were obtained for demographical characteristics (race and age), surgery related information (method of tonsillectomy, method of hemostasis, clotting abnormalities, reasons for surgery, hospital information, experience of surgeon), pain, bleeding and adverse events. Generalized Estimating Equations (GEE) were used to analyze the effects of tonsillar pillar closure, time trend, experience of surgeon and its interaction with tonsillar pillar closure on postoperative pain, risk of bleeding, and risk of complications. Analysis was performed using GEE model due to correlation among repeated measurements on the same patient (both tonsillar fossa sides and survey over time for pain and bleeding) [11]. All tests were conducted as two-sided at alpha level of 0.05 for statistical significance. The statistical software was SAS version 9.1.3 (SAS Institute, Inc., Cary, NC, USA). 4. Results The CONSORT 2010 Flow Diagram is show in Fig. 1.The demographic description of the subjects who met the criteria for this study and enrolled is presented in Table 1 with the age distribution in Fig. 2. The categorization of the operations is listed in Table 2. Surgeon data are listed in Table 3. Seventeen
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
1802
Fig. 1. CONSORT 2010 Flow Diagram. Notes: *This study was conducted prior to 2010, when the CONSORT flow diagram was published. The total number of patients who underwent tonsillectomy by the surgeons involved is not known, nor retrievable. The number listed, as ‘‘assessed for eligibility’’ is the number of tonsillectomies performed by the first author during the time period of the study.
interviewers collected postoperative information about the subjects. 4.1. Withdrawal from study One hundred thirty one (131) subjects had no follow-up information after the first 72 h. 340 did not have follow-up information at or about postoperative day 7. 346 did not have follow-up information at or about postoperative day 14. 357 did not have follow-up information at or about postoperative day 21.
Table 1 Preoperative subject demographics. Age (months)
Mean = 75.7 m
Male Female Indications Obstructive sleep disorder Pharyngitis Velopharyngeal insufficiency in preparation for pharyngeal flap Feeding problems Other Missing value Ethnicity Hispanic Race Caucasian African American Asian Mixed Other Preoperative clotting abnormality Gender
Range 8–264 m 446 317 684 60 11
(58.5%) (4.15%) (89.7%) (7.9%) (1.4%)
1 5 2 7 644 114 3 1 1 34
(0.1%) (0.7%) (0.3%) (1.7%) (84.4%) (14.9%) (0.4%) (0.1%) (0.1%) (4.5%)
326 did not have follow-up information at or about postoperative day 28. These categories are not mutually exclusive; that is, some subjects did not provide data at one point in time, but then provided data at a later collection time. Altogether, 123 had complete follow-up; i.e. data was collected at each data request time. Of these 123 with complete follow-up, 61 had the left side sutured closed, 62 subjects had the right side sutured. 4.2. Specific aims 4.2.1. Specific aim 1: determine if suturing the tonsillar fossa has an effect on postoperative bleeding We sought to compare postoperative bleeding rates between the sutured (closed) tonsillar fossa and the non-sutured (open) tonsillar fossa. When a patient was unable to determine which side bled, bleeding information was considered as missing for GEE analysis. 69% (529/743) of subjects did not have postoperative bleeding, while 16% (121 subjects) were known to have some type of postoperative bleeding. This includes all bleeding, including minor bleeding (from any source, e.g. nose) that did not require intervention. Of those 121 subjects that had bleeding, only 43 subjects had bleeding attributable to the tonsil fossae (43/ 763 = 5.6%). Overall, only 27 subjects (27/763 = 3.5%) required control of bleeding. When bleeding was known to occur from the tonsillar fossae, it was from only the open side in 18/43 patients (42%), from only the closed side in 17/43 (40%) (p = 0.86) and from both sides in 4/43 (9%). In 4/43 (9%) subjects who had bleeding from the tonsil fossae, the side of bleeding could not be
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
1803
Table 3 Surgeon data. Number of cases Faculty (n = 7 individuals) Resident (n = 18 individuals) Resident Subcategory at time of operationa PGY5 (n = 4) PGY4 (n = 5) PGY3 (n = 15) PGY2 (n = 9) PGY1 (n = 1)
61 702 13 19 633 36 1
a Because the study extended over 58 months, some of the resident surgeons had advanced in their training, and so performed the operation in more than one postgraduate year of training. For the data analysis, years of experience after residency were used to stratify the faculty surgeons.
Fig. 2. Graphical representation of age distribution.
determined. We accept the null hypothesis that suturing the tonsil fossa does not change the risk of postoperative bleeding. 4.2.2. Specific aim 2: determine if suturing the tonsillar fossa has an effect on postoperative pain We assessed pain after tonsillectomy to discover if suturing the tonsillar fossa would help our patients by reducing their postoperative pain. The results are indicated in the following figure (Fig. 3). At all times, subjects reported more pain on the closed side. After 14 days, the majority of patients did not have pain on either side. In analysis of the likelihood of pain, the effects of treatment and postoperative days are both statistically significant. Closing the tonsillar pillar is associated with a 40% increase in the odds ratio of postoperative pain on the closed side compared to leaving the tonsillar pillar open (95% confidence interval = 1.21–164, p < 0.001). Elapse of time has a healing effect on pain and the odds ratio of having pain decrease is 10% per day during the study period (95% confidence interval = 0.89–0.91, p < 0.001). This includes all pain reported by the subjects, whether or not one side was more painful than the other.
Table 2 Surgical methods, procedures and type of facility. Tonsillectomy method
Hemostasis method
Side closed Concurrent procedures (subject may have more than one concurrent procedure)
Hospital type
Cautery Harmonic scalpel Cold dissection and snare Cautery Cautery and tannic acid Sutures Oxymetazoline, topical Tannic acid, topical Other Right Left Adenoidectomy Laryngoscopy and bronchoscopy Supraglottoplasty Myringotomy and tube(s) Cerumen removal Tracheostomal revision Lingual frenotomy Tracheostomy Mastoid bowl debridement Excision preauricular cyst Uvulectomy Children’s Hospital County Hospital Private Community Hospital
757 4 2 751 5 4 1 1 1 379 384 756 595
We reject the null hypothesis, that there is no difference in postoperative pain after adenotonsillectomy comparing leaving the fossa open and suturing closed the tonsillar fossa. Our results show that, instead of reducing pain, suturing increased pain. This finding is noteworthy since primary wound closure is a long accepted surgical approach, since at least the time of the ancient Egyptians (2900-1000 BCE) [12]. 4.2.3. Specific aim 3: determine if experience of the surgeon has an effect on postoperative bleeding Significant interaction was found between the experience of surgeon and the treatment (p < 0.05). In the tonsillar fossa left open, a surgeon with more experience was associated with decreased risk of bleeding. However in the closed tonsillar fossa, an increased risk of bleeding occurred with the more experienced surgeons. This is an unexpected result, since in general, the more experienced surgeon is thought to have better outcomes. Certainly pediatric heart transplant surgeons with more experience have better outcomes [13]. One would expect the experienced surgeon to be able to handle the tissues more gently and delicately, and thus have better outcomes. At least in this study, that expectation is not supported. Hence we reject the hypothesis that the more experienced the surgeon, the less postoperative bleeding for the subject in closed tonsillar fossa. This is not initially comforting for experienced surgeons, though it can lead to re-evaluation of technique as part of practice-based learning and improvement and systems-based practice, two important competencies of a physician [14].
(99.2%) (0.5%) (0.3%) (98.4%) (0.7%) (0.5%) (0.1%) (0.1%) (0.1%) (49.7%) (50.3%)
377 150 35 3 2 1 1 1 1 746 (97.8%) 15 (2.0%) 2 (0.3%)
Fig. 3. Responses about pain during the study period.
1804
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
4.2.4. Specific aim 4: determine if experience of the surgeon has an effect on postoperative pain Interaction between treatment (i.e. whether or not a tonsillar fossa was closed or remained open) and experience of surgeon is not significant (p = 0.30), interpreted as no difference between the open side and the closed side in the risk of pain associated with the experience of surgeon. Experience of the surgeon is not significantly related to the duration of postoperative pain for the subject on either side (p = 0.08). In an alternative analysis using experience of surgeon as number of years after residency, surgeon experience shows no statistically significant difference. Therefore, we accept the null hypothesis that, compared to a less experienced surgeon, the more experienced the surgeon does not change the duration of postoperative pain after adenotonsillectomy. This is good news for otolaryngology trainees, as it implies that their skills are statistically no different in the pain domain from those of their attending surgeons for tonsillectomy. 4.2.5. Specific aim 5: determine if suturing the tonsillar fossa has an effect on other postoperative complications The majority of subjects (550/763 or 72%) did not report any adverse events. Specifically, no deaths or palatal hematoma occurred. The most common adverse event was ‘‘requiring medical or surgical intervention’’ (n = 112). Upon review of the individual records of the involved subjects, we were unable to determine if the experimental treatment of suturing was responsible for these adverse events. In fact, these adverse events are the same seen in standard tonsillectomy techniques. No unusual complications (not usually associated with tonsillectomy) such as palatal hematoma occurred. We did not have a control group of subjects with no suturing of either tonsillar fossa, so we cannot directly compare subjects with suturing vs. subjects with no suturing. Thus, we accept the null hypothesis that that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not change the risk of other postoperative complications after adenotonsillectomy. 5. Discussion This study was designed to evaluate the effect of suture closure of one tonsillar fossa after tonsillectomy on postoperative pain and bleeding. 5.1. Paradoxical results with bleeding We found that suturing the tonsillar fossa did not reduce the risk of bleeding, which is counter-intuitive. In general, closure of wounds is thought to reduce the risk of bleeding. We cannot explain why suturing did not help in this experiment. One supposition is that the wound was not tightly closed at all levels, as a small opening was left inferiorly in the fossa. However, the unsutured side remained completely open, but did not show an increased risk. Another supposition is that not all the dead space was obliterated during suturing. However, even if bleeding occurred in the closed side, one could argue that increased pressure in a closed wound would help control bleeding. Again, we have no evidence to support that idea. In contrast, both Weighill et al. [7] and Nandapalan and McIlwain [8] reported bleeding only on the unsewn side and the only other two papers dealing with suturing the tonsillar fossa had no postoperative bleeding [9,10]. 5.2. Paradoxical effects of surgical experience The more experienced surgeon had a decreased risk of bleeding from the open (unsutured) side. That is an intuitive finding confirmed by our study. Paradoxically, the closed (sutured) side
had an increased risk of bleeding in the hands of the more experienced surgeon. Perhaps the more experienced surgeon engaged more tissue, and passed the suture needle more deeply. If so, more structures would be expected to be at risk for bleeding. If a less experienced surgeon incorporated less tissue in suturing, then only vessels closer to the surface would be at risk. We did not measure the needle depth or amount of tissue engaged during suturing, so we are unable to evaluate that idea in this study. The literature, in only one article, specifically addresses, in tonsillectomy, the role of experience for the surgeon. Roberts et al. [15] found that the experience of the surgeon had no effect on the complication rate. 5.3. Paradoxical results of pain Subjects had more pain with the closed tonsillar fossa. Again this is an initially unexpected result, though Nandapalan and McIlwain [8] had similar findings for postoperative days 1–3, while Genc et al. noted less pain on the sutured side at postoperative days 3–10 [10] We found that the sutured side continued to have more pain than the unsutured side for the month of follow-up. Perhaps, suturing led to tugging and pulling of the tissues, whereas the open side was less constrained during swallowing. Additionally, perhaps merely doing more (suturing) induced more trauma, and hence inflammation, which might explain the increased pain. A build-up of fluid under pressure in the tissues, as is seen in a hematoma, often causes pain. In this study, the tonsillar fossae were not closed water-tight, so any fluids that might have accumulated in the fossa were allowed to drain. Our data does not reveal a provable explanation for the finding of increased pain in the sutured fossa. 5.4. Clinical relevance Clinically, this study supports the idea that the common practice during tonsillectomy of not suturing the tonsillar fossa is preferable suturing. The patients will have less risk of bleeding and less pain with an open (unsutured) tonsillar fossa. Hence the surgical desire to close a wound is not advantageous, in the case of tonsillectomy. However, our study actually supports, with data, the common practice of leaving the fossae unsutured during tonsillectomy. We found that a non-sutured fossa hurt less with a more experienced surgeon, but hurt more in a sutured fossa. This is not consistent with the findings of Weighill et al. [7] and Ramjettan and Singh [9], who did not find a difference between sewn and unsewn sides. The factors that contribute to a good outcome in surgery are many, and not always intuitive or easy to discern. 5.5. Design considerations One unaccounted for factor may be differences in actually suturing the tonsillar pillars, such as not sewing the muscle and mucosa, but just sewing the mucosa, or conversely, just sewing the muscle without including mucosa. We had some concern that if each surgeon sutured the pillar slightly differently, the results might be affected and we did not control for this variation. At follow-up examinations, often the sutured side had not stayed closed when the pharynx was viewed in the first 2 weeks postoperatively. Perhaps the suture type was the wrong choice, as chromic does not last long (10–14 days), though longer than gut sutures (7–10 days), but not as long as PDS (up to 6 weeks). The precise number of subjects whose sutured side stayed closed was not collected and is retrospectively unobtainable. We cannot speak to whether subjects who had a sutured side open up had more or less pain. Another problem is the incomplete data. Information was obtained from the subject, or caregiver(s). Both of these sources of information could introduce distortion of the true pain conditions
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
of the subject. Caregivers may misinterpret or confuse their left and right (or the subject’s left or right), and subjects may not be precise in declaring exactly where the throat hurt. Additionally, most of the subjects received concurrent operations besides tonsillectomy. We believe that bilateral myringotomy and tubes, or bilateral cerumen removal would not be expected to increase pain unilaterally, or even if it did, the pain from these procedures is almost always gone in 24 h. Adding adenoidectomy to tonsillectomy is common in treating pharyngitis and obstructive sleep apnea in children. As the adenoid is a central structure, the pain is expected to be midline, and thus not bias the patient’s perception to one side. In any event, the average duration of pain after adenoidectomy alone is 3 days (range 0– 8 days) [16]. The literature is silent on whether tonsillectomy pain is additive to adenoidectomy pain, though personal observations of the authors imply that the more the surgeon operates (more tissue disruption, more area involved), the more it hurts the patient. A number of children also received a supraglottoplasty. Fortunately, that pain is often minor and probably gone in less than 1 day [17], so it would not be expected to influence the sided of any pain, or cause any bleeding in the tonsillar fossa. Overall, the subjects were relatively homogenous in the indication for tonsillectomy, as almost 90% underwent the operation for obstructive sleep disorder symptoms. 377 of 763 patients undergoing tonsillectomies had supraglottoplasties. Evaluation and treatment of the entire airway in obstructive sleep apnea is not considered routine for many pediatric otolaryngology practices. Groblewski et al. [17] support the idea that supraglottoplasty does not hurt much, nor for very long, and therefore would not be expected to influence the sidedness of the pain scores. Walton et al. [18] in a meta-analysis find the mean resolution of pain for complete tonsillectomy to be (mean SD) 7.6 ( 2.14) days. This supports the idea that tonsillectomy hurts much more than supraglottoplasty, and hence tonsillectomy would be expected to account for the vast majority of the pain and for longer time periods, compared to supraglottoplasty. Another criticism that can be raised with this study is the inclusion of younger children who would not necessarily have the developmental capacity to identify unilateral pain. Certainly pain in general can be deduced by evaluation of a subject’s vital signs and other measures like crying or irritability. Excluding children younger than 3 would reduce the sample size to 561. However, many of those under age 3 either had no pain by postoperative day 7 (as assessed by the caregivers), or did not provide data on which if any side hurt. The statistical analysis we used has accounted for missing data points and thus our conclusions on pain remain statistically valid. 6. Conclusions The data show that the use of suture closure of the tonsillar pillars after tonsillectomy does not reduce the risk of bleeding. In fact, in tonsillar fossae left open, the more experienced the surgeon, the lower the risk of bleeding. In closed tonsillar fossae, the more experienced the surgeon, the higher the risk of bleeding. This finding is counter-intuitive, as one would expect more experience to lead to less bleeding, and better suturing. Additionally, closing the tonsillar pillar increased the odds of postoperative pain by 40% compared to leaving the tonsillar pillar open. This is also counterintuitive, as closed wounds elsewhere on the body (e.g. skin) tend to hurt less. Though the subjects were not completely homogenous (e.g. differences exist in indications for operation, comorbidities,
1805
age, surgeon, and hospital), they represent a real-world cohort of children undergoing tonsillectomy. Based on this study, we cannot recommend suturing of the tonsillar fossae after tonsillectomy. Funding Funding for biostatistical services was provided by the General Otorhinolaryngological Foundation, Indianapolis, IN. Conflict of interest statement The authors have no conflicts of interest to report. Acknowledgments The authors gratefully acknowledge the contributions of the other attending surgeons: M.S. Chand, S.R. Cordes, M.H. Fritsch, R.A. Hamaker, R.C. Miyamoto, A.O. Yekinni, the School of Medicine, Department of Otolaryngology – Head and Neck Surgery residents, nurses, secretaries and phone-interviewers, for this study. References [1] J.L. Paradise, C.D. Bluestone, D.K. Colborn, B.S. Bernard, H.E. Rockette, M. KursLasky, Tonsillectomy and adenotonsillectomy for recurrent throat infection in moderately affected children, Pediatrics 110 (2002) 7–15. [2] M.F. Owings, L.J. Kozak, Ambulatory and inpatient procedures in the United States, 1996, in: Vital & Health Statistics – Series 13: Data from the National Health Survey, 1998, pp. 1–119. [3] K. Cullen, M. Hall, A. Golosinskiy, Centers for disease control, ambulatory surgery in the United States, 2006, in: National Health Statistics Reports, 2009, p. 28, Available at: http://www.cdc.gov/nchs/data/nhsr/nhsr011.pdf (accessed 01.09.11). [4] N. Bhattacharyya, L.J. Kepnes, Economic benefit of tonsillectomy in adults with chronic tonsillitis, Ann. Otol. Rhinol. Laryngol. 111 (2002) 983–988. [5] J.M. Curtin, The history of tonsil and adenoid surgery, Orthop. Clin. North Am. 20 (1987) 415–419. [6] L. Neumayer, D. Vargo, Principles of preoperative and operative surgery, in: C.M. Townsend, Jr., R.D. Beauchamp, B.M. Evers, K.L. Mattox (Eds.), Sabiston Textbook of Surgery: The Biological Basis of Modern Surgical Practice, 18th ed., Saunders Elsevier, Philadelphia, 2008, pp. 251–279. [7] J.S. Weighill, D.W. Proops, D. Jeffries, J. Brandrick, Pain relief following tonsillectomy (does sewing the faucial pillars together help?), J. Laryngol. Otol. 100 (1986) 307–310. [8] V. Nandapalan, J.C. McIlwain, Tonsillar fossa obliteration and post-operative pain, Clin. Otolaryngol. Allied Sci. 20 (1995) 127–129. [9] S. Ramjettan, B. Singh, Are sutured faucial pillars really an advantage in tonsillectomy? S. Afr. J. Surg. 34 (1996) 189–191. [10] E. Genc, D. Hanci, N.T. Ergin, T. Dal, Can mucosal sealing reduce tonsillectomy pain? Int. J. Pediatr. Otorhinolaryngol. 70 (2006) 725–730. [11] B. Dawson, R.G. Trapp, Basic and Clinical Biostatistics, 4th ed., Lange Medical Books/McGraw-Hill, New York, 2004. [12] S.J. Doshi, A Complete Review of Wound Closure, Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, 1992. [13] R.R. Davies, M.J. Russo, K.N. Hong, S. Mital, R.S. Mosca, J.M. Quaegebeur, et al., Increased short- and long-term mortality at low-volume pediatric heart transplant centers: should minimum standards be set? Retrospective data analysis, Ann. Surg. 253 (2011) 393–401. [14] ACGME, Common Program Requirements: General Competencies, 2007 http:// www.acgme.org/outcome/comp/GeneralCompetenciesStandards21307.pdf (accessed 30.07.11). [15] C. Roberts, S. Jayaramachandran, C.H. Raine, A prospective study of factors which may predispose to post-operative tonsillar fossa haemorrhage, Clin. Otolaryngol. Allied Sci. 17 (1992) 13–17. [16] H.M. Tuomilehto, H.M. Kokki, R.P. Ahonen, J.M. Nuutinen, Postoperative behavioral changes in children after adenoidectomy, Arch. Otolaryngol. Head Neck Surg. 128 (2002) 1159–1164. [17] J.C. Groblewski, R.K. Shah, G.H. Zalzal, Microdebrider-assisted supraglottoplasty for laryngomalacia, Ann. Otol. Rhinol. Laryngol. 118 (2009) 592–597. [18] J. Walton, Y. Ebner, M.G. Stewart, M.M. April, Systematic review of randomized controlled trials comparing intracapsular tonsillectomy with total tonsillectomy in a pediatric population, Arch. Otolaryngol. Head Neck Surg. 138 (2012) 243– 249.
Contents lists available at SciVerse ScienceDirect
International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl
Effect of tonsillar fossa closure on postoperative pain and bleeding risk after tonsillectomy Bruce H. Matt a,*, Bryan J. Krol a,b, Yan Ding c, Beth E. Juliar c a
Indiana University School of Medicine, Department of Otolaryngology – Head & Neck Surgery, School of Medicine, Indiana University, Indianapolis, IN, USA Head & Neck Surgery Associates PSC, Ft. Thomas, KY, USA c Division of Biostatistics, Department of Medicine, School of Medicine, Indiana University, Indianapolis, IN, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 5 August 2012 Received in revised form 27 August 2012 Accepted 3 September 2012 Available online 25 September 2012
Objective: To determine if closing the tonsil fossa after tonsillectomy leads to less pain and bleeding. Study design: Interventional, Randomized, Single Blind, Active Control, Single Group Assignment, Safety/ Efficacy Study. Follow-up: 2 months. Methods: Generalized Estimating Equations (GEE) analyzed effects of tonsillar pillar closure, surgeon experience and interaction on outcomes. Setting: Primarily academic tertiary care referral center, institutional practice, primarily children, both ambulatory and hospitalized care. Subjects: 763 subjects (age 8–264 months) undergoing tonsillectomy. Exclusions: suspected malignancy or active peritonsillar abscess. At the discretion of the attending surgeon, patients undergoing tonsillectomy during the 4 year study period were offered participation. A computer selected the side closed. 131 subjects withdrew (complete lack of follow-up information) after the first 72 h. Intervention: 3-0 chromic sutures on tapered needles to close one tonsillar fossa. The subject was not told which side was closed. Main outcome measures: postoperative bleeding (at any time) and pain reported was sought on days 1, 7, 14, 21, and 28. Results: Closure of the tonsillar fossa did not change the risk of bleeding. Closing the tonsillar fossa had a 40% increase in the odds ratio of postoperative pain. In the tonsillar fossa sides left open, greater surgeon experience decreased the risk of bleeding. In closed sides, enriched surgeon experience increased the risk of bleeding (p < .0.05). Conclusions: Suture closure of the tonsillar fossa after tonsillectomy does not reduce the risk of bleeding. Additionally, closing the tonsillar fossa increased postoperative pain. Level of evidence: 1b (individual randomized controlled trial). ß 2012 Elsevier Ireland Ltd. All rights reserved.
Keywords: Tonsillectomy Postoperative pain Hemostasis Pillar sutures Tonsillar fossa closure
1. Introduction 1.1. Tonsillectomy incidence Tonsillectomy is one of the most common operations performed by otolaryngologists [1,2] 287,000 tonsillectomies (with or without adenoidectomy) were performed in 1996 on children under 15 years of age, almost doubling to 530,000 in 2006 [1–3]. This represents roughly $701 million/year for tonsillectomy in the United States alone [4]. Techniques to improve the experience for the patient by having better outcomes and lowering
* Corresponding author at: Suite 0860, Riley Hospital, 705 Riley Hospital Drive, Indianapolis, IN 46202-5230, USA. Tel.: +1 317 944 6670; fax: +1 317 948 4720. E-mail address: [email protected] (B.H. Matt). 0165-5876/$ – see front matter ß 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijporl.2012.09.004
risks and discomfort, as well as making the operation and postoperative care easier for the surgeon have been sought for decades, if not centuries [5]. 1.2. What can be done to reduce postoperative issues and complications after tonsillectomy? This paper is not focused on the differences between techniques of removal, but on a method to reduce postoperative complications after complete tonsillectomy. The specific method chosen here to test is suturing together of the tonsillar pillars to close the tonsillar fossa. Immediate wound closure allows for healing by primary intention, which is faster than allowing the wound to close spontaneously by secondary intention healing. Mechanical integrity of a healed wound allows for quicker return to normal function [6].
1800
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
The feasibility and value of tonsillar fossa closure has been studied by several investigators. Weighill et al. [7] evaluated 60 consecutive adult (age greater than 16 years) subjects, and by randomization, sutured one side closed after tonsillectomy with 30 plain catgut. Weighill placed 3 interrupted sutures to approximate the anterior and posterior tonsillar pillars. No statistically significant difference was noted between the sewn and unsewn sides at postoperative days 1, 2, and 3. More than half the subjects were lost to follow up at 6 weeks postoperatively. 3 Subjects developed a palatal hematoma on the sutured side, while one subject had a postoperative hemorrhage from the unsutured side. Nandapalan and McIlwain [8] compared 50 consecutive adult patients (age greater than 15 years) undergoing tonsillectomy, and by randomization, sutured one side closed with 2-0 polydioxanone (PDS), with 3 interrupted sutures, sewing the pillars to the tonsillar fossa bed to eliminate dead space. The sutured side was more painful at postoperative days 1, 2, and 3, but thereafter the unsutured side was more painful at postoperative days 4–10. Eighteen of 19 patients complained of otalgia on the unsutured side. None had a palatal hematoma and 1 had mild postoperative hemorrhage from the unsutured side. Ramjettan and Singh [9] compared two non-randomized, consecutive groups: an initial control group of 20 whose tonsillar pillars were not sewn, and a second experimental group of 20 whose tonsillar pillars were sewn with 3-0 chromic catgut in a running, continuous fashion, bilaterally. The ages ranged from 3 to 35 years. The adults had more pain than the children in both groups at the immediate postoperative measurement and at the 24-h point. The difference in pain scores for the two groups (control vs. treatment) was not statistically significant. In the sutured group, 2 subjects had palatal hematomas, and one subject had temporary nasal regurgitation. No subjects had postoperative bleeding. Genc et al. [10] in 39 patients, ages 3–15 years, at the preference of the attending surgeon, selected one tonsillar fossa to be sutured closed with 4-0 plain catgut on a sharp needle in 4–5 interrupted sutures. Genc et al. found no difference in pain on the first postoperative day. However, in contrast to the above studies, pain was less on the sutured side from postoperative days 3–10. Edema was worse on postoperative day 10 on the sutured side, but healing was better. None of the subjects had palatal hematomas, or postoperative bleeding reported, however the subjects were not followed beyond postoperative day 10. The sutured side was statistically significantly more likely to have edema. The above studies suffer from small sample sizes, and included primarily adults. This study was undertaken to prospectively and systematically investigate whether a simple technique could be used to reduce complications in a large group of children undergoing tonsillectomy in a randomized trial of closed vs. open tonsillar fossa.
2.1.1. Specific aim 1: determine if suturing the tonsillar fossa has an effect on postoperative bleeding 2.1.1.1. Working hypothesis 1. The working hypothesis of this aim is that suturing closed the tonsillar fossa reduces the risk of postoperative bleeding after adenotonsillectomy. Alternatively, the null hypothesis is that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not statistically change the risk of postoperative bleeding after adenotonsillectomy. 2.1.2. Specific aim 2: determine if suturing the tonsillar fossa has an effect on postoperative pain 2.1.2.1. Working hypothesis 2. The working hypothesis of this aim is that suturing closed the tonsillar fossa would reduce the postoperative pain after adenotonsillectomy. Alternatively, the null hypothesis is that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not statistically change the postoperative pain after adenotonsillectomy. 2.1.3. Specific aim 3: determine if experience of the surgeon has an effect on postoperative bleeding 2.1.3.1. Working hypothesis 3. The working hypothesis of this aim is the more experienced the surgeon, the less postoperative bleeding for the subject. Alternatively, the null hypothesis is that, compared to a less experienced surgeon, the more experienced surgeon does not statistically change the rate of postoperative bleeding after adenotonsillectomy. 2.1.4. Specific aim 4: determine if experience of the surgeon has an effect on postoperative pain 2.1.4.1. Working hypothesis 4. The working hypothesis of this aim is the more experienced the surgeon, the less pain for the subject. Alternatively, the null hypothesis is that, compared to a less experienced surgeon, the more experienced the surgeon does not statistically change the postoperative pain after adenotonsillectomy. 2.1.5. Specific aim 5: determine if suturing the tonsillar fossa has an effect on other postoperative complications 2.1.5.1. Working hypothesis 5. The working hypothesis of this aim is suturing closed the tonsillar fossa reduces other postoperative complications after adenotonsillectomy. Alternatively, the null hypothesis is that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not statistically change the risk of other postoperative complications after adenotonsillectomy. 3. Materials and methods
2. Aims and objectives
3.1. Institutional review and subject protection
The long-range goal is to implement safe and effective surgical strategies when tonsillectomy is performed. The paper investigates a specific surgical technique in an effort to reduce postoperative complications by suture closure of the tonsillar pillars. Closed wounds require ‘‘less healing’’, as the body does not have to bring the wound edges together, and thus the raw surfaces are not exposed to abrasive disruptions.
This study was approved by the Institutional Review Board (IRB) of our institution (Indiana University Office of Research Administration, Human Subjects Office), and monitored annually by the IRB. This clinical trial was registered with http:// www.clinicaltrials.gov.
2.1. Central hypothesis
3.2.1. Inclusion criteria Our inclusion criteria were: any patient, male or female, of any ethnicity, or any age, for whom tonsillectomy was recommended for recurrent pharyngitis, obstructive sleep disorder, snoring, halitosis, or feeding difficulty associated with adenotonsillar
Our null hypothesis is: no difference in bleeding or pain or other complications exists between the sutured and unsutured tonsillar fossa.
3.2. Subject selection
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
hypertrophy, and who in the investigator’s opinion, was capable of providing reliable responses to post-operative follow-up questions. 3.2.2. Exclusion criteria Any patient for whom tonsillectomy was recommended for suspected malignancy or active peritonsillar abscess was excluded from the study. Other exclusion criteria included: inability to read or understand the post-operative questionnaire; language barriers or cognitive impairment that would prevent the patient from freely giving informed consent and failure to agree to follow-up. Other procedures in the subject were permitted. 3.2.3. Subject recruitment The option of participating in this study was offered to all eligible patients who were scheduled to undergo tonsillectomy with one of the 7 attending otolaryngologists. No financial advantage was offered to the patients. No written advertisements were used to publicize the study. 3.2.4. Study intervention 3.2.4.1. Control group. One tonsillar fossa remained open (unsutured) and would serve as a control for the closed tonsillar fossa. Because the subject acts as his own control, i.e. comparing the pain of one side to the other, differences in tonsillectomy technique are controlled (internally by the subject). This approach has the advantages of avoiding the use of a separate control group that does not receive the experimental intervention and does not require the use of a pain scale for comparing one subject to another. This experimental design also internally controls for any other methods or techniques used in each individual subject either perioperatively or postoperatively. This design is not perfect however, as younger children may not necessarily have the intellectual ability to be able to clearly indicate which, if any, side hurt. Also, the design does not specifically differentiate between pain at a particular site and referred pain. 3.2.5. Surgical procedure and closure After induction of appropriate general anesthesia, a tonsillectomy was performed. The specific method of tonsillectomy was per surgeon’s preference. Three specific methods of tonsillectomy were used: (1) electrocautery dissection, (2) cold dissection and snare, and (3) harmonic scalpel, though the vast majority was done with electrocautery dissection (refer to Table 2). Intraoperative control of bleeding was achieved as was routine for each individual otolaryngologist. After the tonsillectomy was completed, closure of a tonsillar fossa was performed by the surgeon. The side-to-close chosen was determined by a computer-generated randomization schedule. The subject and the family were not informed which side was closed. This study intervention used 3-0 chromic (absorbable) sutures on tapered needles to close only one tonsillar fossa. The other (contralateral) tonsillar fossa remained open. The anterior and posterior pillar mucosa was sewn together with 2–3 interrupted stitches of 3-0 chromic. No attempt was made to obliterate dead space, or to completely seal the fossa: a small gap inferiorly was often present with the mouth-gag in place. 3.2.6. Postoperative care Surgeon-specific routines were followed for postoperative care. Because the patient acted as his or her own control, postoperative choices left at the discretion of the surgeon included: antibiotic, analgesia, diet, and outpatient vs. inpatient care.
1801
3.2.7. Data collection and extraction Information collected included the following independent variables: age at operation, gender, indication for tonsillectomy, date of surgery, experience of surgeon, method of tonsillectomy, method of hemostasis, side closed, and known clotting abnormalities. Dependent variables were: which side was more painful on or about postoperative day 1 (the first 72 h), days 7, 14, 21, and overall assessment at the postoperative clinic visit (on or about day 28). Overall assessment included the ‘‘gestalt’’ assessment of the subject (and/or legal guardian), the ‘‘gestalt’’ assessment of the investigator and any concerns which the subject (and/or legal guardian) or investigator raised. The pain scale was ordinal. We asked, ‘‘As of today, which side hurts more?’’ and categorized the responses as [1 = left, 2 = right 3 = equal, 4 = unable to determine, 5 = no pain]. Information was obtained from the caregiver or subject, either in person, or by phone. A physician or other departmental personnel obtained the information using a scripted questionnaire. Information about the subject’s pain was sought on or about postoperative day 1 (the first 72 h), days 7, 14, 21, and at the postoperative clinic visit (on or about day 28). Additional information was obtained if postoperative bleeding occurred, including which side bled, when the bleeding occurred, what evaluation and treatment were undertaken and what the outcome was. 3.3. Statistical methods 3.3.1. Prospective sample size calculations Pre-trial estimates were for 720 subjects enrolled to obtain a sample size of 36 subjects who had postoperative bleeding (‘‘bleeders’’) (5% of total enrollment). With 36 bleeders, there is 80% power to detect a statistically significant difference of 0.33 in proportion of bleedings from closed fossae vs. proportion of bleedings from open fossae, at the 0.05 level of significance in a two-sided McNemar’s test for correlated data [11]. If 720 patients were enrolled and 95% of them reported pain, there would be 99% power to detect a difference in proportions of 0.10 at any of 5 post surgery data collection times with a McNemar’s test at alpha of 0.01 using Bonferroni’s method to adjust for multiple comparisons. 3.3.2. Statistical analysis Descriptive statistics were obtained for demographical characteristics (race and age), surgery related information (method of tonsillectomy, method of hemostasis, clotting abnormalities, reasons for surgery, hospital information, experience of surgeon), pain, bleeding and adverse events. Generalized Estimating Equations (GEE) were used to analyze the effects of tonsillar pillar closure, time trend, experience of surgeon and its interaction with tonsillar pillar closure on postoperative pain, risk of bleeding, and risk of complications. Analysis was performed using GEE model due to correlation among repeated measurements on the same patient (both tonsillar fossa sides and survey over time for pain and bleeding) [11]. All tests were conducted as two-sided at alpha level of 0.05 for statistical significance. The statistical software was SAS version 9.1.3 (SAS Institute, Inc., Cary, NC, USA). 4. Results The CONSORT 2010 Flow Diagram is show in Fig. 1.The demographic description of the subjects who met the criteria for this study and enrolled is presented in Table 1 with the age distribution in Fig. 2. The categorization of the operations is listed in Table 2. Surgeon data are listed in Table 3. Seventeen
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
1802
Fig. 1. CONSORT 2010 Flow Diagram. Notes: *This study was conducted prior to 2010, when the CONSORT flow diagram was published. The total number of patients who underwent tonsillectomy by the surgeons involved is not known, nor retrievable. The number listed, as ‘‘assessed for eligibility’’ is the number of tonsillectomies performed by the first author during the time period of the study.
interviewers collected postoperative information about the subjects. 4.1. Withdrawal from study One hundred thirty one (131) subjects had no follow-up information after the first 72 h. 340 did not have follow-up information at or about postoperative day 7. 346 did not have follow-up information at or about postoperative day 14. 357 did not have follow-up information at or about postoperative day 21.
Table 1 Preoperative subject demographics. Age (months)
Mean = 75.7 m
Male Female Indications Obstructive sleep disorder Pharyngitis Velopharyngeal insufficiency in preparation for pharyngeal flap Feeding problems Other Missing value Ethnicity Hispanic Race Caucasian African American Asian Mixed Other Preoperative clotting abnormality Gender
Range 8–264 m 446 317 684 60 11
(58.5%) (4.15%) (89.7%) (7.9%) (1.4%)
1 5 2 7 644 114 3 1 1 34
(0.1%) (0.7%) (0.3%) (1.7%) (84.4%) (14.9%) (0.4%) (0.1%) (0.1%) (4.5%)
326 did not have follow-up information at or about postoperative day 28. These categories are not mutually exclusive; that is, some subjects did not provide data at one point in time, but then provided data at a later collection time. Altogether, 123 had complete follow-up; i.e. data was collected at each data request time. Of these 123 with complete follow-up, 61 had the left side sutured closed, 62 subjects had the right side sutured. 4.2. Specific aims 4.2.1. Specific aim 1: determine if suturing the tonsillar fossa has an effect on postoperative bleeding We sought to compare postoperative bleeding rates between the sutured (closed) tonsillar fossa and the non-sutured (open) tonsillar fossa. When a patient was unable to determine which side bled, bleeding information was considered as missing for GEE analysis. 69% (529/743) of subjects did not have postoperative bleeding, while 16% (121 subjects) were known to have some type of postoperative bleeding. This includes all bleeding, including minor bleeding (from any source, e.g. nose) that did not require intervention. Of those 121 subjects that had bleeding, only 43 subjects had bleeding attributable to the tonsil fossae (43/ 763 = 5.6%). Overall, only 27 subjects (27/763 = 3.5%) required control of bleeding. When bleeding was known to occur from the tonsillar fossae, it was from only the open side in 18/43 patients (42%), from only the closed side in 17/43 (40%) (p = 0.86) and from both sides in 4/43 (9%). In 4/43 (9%) subjects who had bleeding from the tonsil fossae, the side of bleeding could not be
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
1803
Table 3 Surgeon data. Number of cases Faculty (n = 7 individuals) Resident (n = 18 individuals) Resident Subcategory at time of operationa PGY5 (n = 4) PGY4 (n = 5) PGY3 (n = 15) PGY2 (n = 9) PGY1 (n = 1)
61 702 13 19 633 36 1
a Because the study extended over 58 months, some of the resident surgeons had advanced in their training, and so performed the operation in more than one postgraduate year of training. For the data analysis, years of experience after residency were used to stratify the faculty surgeons.
Fig. 2. Graphical representation of age distribution.
determined. We accept the null hypothesis that suturing the tonsil fossa does not change the risk of postoperative bleeding. 4.2.2. Specific aim 2: determine if suturing the tonsillar fossa has an effect on postoperative pain We assessed pain after tonsillectomy to discover if suturing the tonsillar fossa would help our patients by reducing their postoperative pain. The results are indicated in the following figure (Fig. 3). At all times, subjects reported more pain on the closed side. After 14 days, the majority of patients did not have pain on either side. In analysis of the likelihood of pain, the effects of treatment and postoperative days are both statistically significant. Closing the tonsillar pillar is associated with a 40% increase in the odds ratio of postoperative pain on the closed side compared to leaving the tonsillar pillar open (95% confidence interval = 1.21–164, p < 0.001). Elapse of time has a healing effect on pain and the odds ratio of having pain decrease is 10% per day during the study period (95% confidence interval = 0.89–0.91, p < 0.001). This includes all pain reported by the subjects, whether or not one side was more painful than the other.
Table 2 Surgical methods, procedures and type of facility. Tonsillectomy method
Hemostasis method
Side closed Concurrent procedures (subject may have more than one concurrent procedure)
Hospital type
Cautery Harmonic scalpel Cold dissection and snare Cautery Cautery and tannic acid Sutures Oxymetazoline, topical Tannic acid, topical Other Right Left Adenoidectomy Laryngoscopy and bronchoscopy Supraglottoplasty Myringotomy and tube(s) Cerumen removal Tracheostomal revision Lingual frenotomy Tracheostomy Mastoid bowl debridement Excision preauricular cyst Uvulectomy Children’s Hospital County Hospital Private Community Hospital
757 4 2 751 5 4 1 1 1 379 384 756 595
We reject the null hypothesis, that there is no difference in postoperative pain after adenotonsillectomy comparing leaving the fossa open and suturing closed the tonsillar fossa. Our results show that, instead of reducing pain, suturing increased pain. This finding is noteworthy since primary wound closure is a long accepted surgical approach, since at least the time of the ancient Egyptians (2900-1000 BCE) [12]. 4.2.3. Specific aim 3: determine if experience of the surgeon has an effect on postoperative bleeding Significant interaction was found between the experience of surgeon and the treatment (p < 0.05). In the tonsillar fossa left open, a surgeon with more experience was associated with decreased risk of bleeding. However in the closed tonsillar fossa, an increased risk of bleeding occurred with the more experienced surgeons. This is an unexpected result, since in general, the more experienced surgeon is thought to have better outcomes. Certainly pediatric heart transplant surgeons with more experience have better outcomes [13]. One would expect the experienced surgeon to be able to handle the tissues more gently and delicately, and thus have better outcomes. At least in this study, that expectation is not supported. Hence we reject the hypothesis that the more experienced the surgeon, the less postoperative bleeding for the subject in closed tonsillar fossa. This is not initially comforting for experienced surgeons, though it can lead to re-evaluation of technique as part of practice-based learning and improvement and systems-based practice, two important competencies of a physician [14].
(99.2%) (0.5%) (0.3%) (98.4%) (0.7%) (0.5%) (0.1%) (0.1%) (0.1%) (49.7%) (50.3%)
377 150 35 3 2 1 1 1 1 746 (97.8%) 15 (2.0%) 2 (0.3%)
Fig. 3. Responses about pain during the study period.
1804
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
4.2.4. Specific aim 4: determine if experience of the surgeon has an effect on postoperative pain Interaction between treatment (i.e. whether or not a tonsillar fossa was closed or remained open) and experience of surgeon is not significant (p = 0.30), interpreted as no difference between the open side and the closed side in the risk of pain associated with the experience of surgeon. Experience of the surgeon is not significantly related to the duration of postoperative pain for the subject on either side (p = 0.08). In an alternative analysis using experience of surgeon as number of years after residency, surgeon experience shows no statistically significant difference. Therefore, we accept the null hypothesis that, compared to a less experienced surgeon, the more experienced the surgeon does not change the duration of postoperative pain after adenotonsillectomy. This is good news for otolaryngology trainees, as it implies that their skills are statistically no different in the pain domain from those of their attending surgeons for tonsillectomy. 4.2.5. Specific aim 5: determine if suturing the tonsillar fossa has an effect on other postoperative complications The majority of subjects (550/763 or 72%) did not report any adverse events. Specifically, no deaths or palatal hematoma occurred. The most common adverse event was ‘‘requiring medical or surgical intervention’’ (n = 112). Upon review of the individual records of the involved subjects, we were unable to determine if the experimental treatment of suturing was responsible for these adverse events. In fact, these adverse events are the same seen in standard tonsillectomy techniques. No unusual complications (not usually associated with tonsillectomy) such as palatal hematoma occurred. We did not have a control group of subjects with no suturing of either tonsillar fossa, so we cannot directly compare subjects with suturing vs. subjects with no suturing. Thus, we accept the null hypothesis that that, compared to leaving the fossa open, suturing closed the tonsillar fossa does not change the risk of other postoperative complications after adenotonsillectomy. 5. Discussion This study was designed to evaluate the effect of suture closure of one tonsillar fossa after tonsillectomy on postoperative pain and bleeding. 5.1. Paradoxical results with bleeding We found that suturing the tonsillar fossa did not reduce the risk of bleeding, which is counter-intuitive. In general, closure of wounds is thought to reduce the risk of bleeding. We cannot explain why suturing did not help in this experiment. One supposition is that the wound was not tightly closed at all levels, as a small opening was left inferiorly in the fossa. However, the unsutured side remained completely open, but did not show an increased risk. Another supposition is that not all the dead space was obliterated during suturing. However, even if bleeding occurred in the closed side, one could argue that increased pressure in a closed wound would help control bleeding. Again, we have no evidence to support that idea. In contrast, both Weighill et al. [7] and Nandapalan and McIlwain [8] reported bleeding only on the unsewn side and the only other two papers dealing with suturing the tonsillar fossa had no postoperative bleeding [9,10]. 5.2. Paradoxical effects of surgical experience The more experienced surgeon had a decreased risk of bleeding from the open (unsutured) side. That is an intuitive finding confirmed by our study. Paradoxically, the closed (sutured) side
had an increased risk of bleeding in the hands of the more experienced surgeon. Perhaps the more experienced surgeon engaged more tissue, and passed the suture needle more deeply. If so, more structures would be expected to be at risk for bleeding. If a less experienced surgeon incorporated less tissue in suturing, then only vessels closer to the surface would be at risk. We did not measure the needle depth or amount of tissue engaged during suturing, so we are unable to evaluate that idea in this study. The literature, in only one article, specifically addresses, in tonsillectomy, the role of experience for the surgeon. Roberts et al. [15] found that the experience of the surgeon had no effect on the complication rate. 5.3. Paradoxical results of pain Subjects had more pain with the closed tonsillar fossa. Again this is an initially unexpected result, though Nandapalan and McIlwain [8] had similar findings for postoperative days 1–3, while Genc et al. noted less pain on the sutured side at postoperative days 3–10 [10] We found that the sutured side continued to have more pain than the unsutured side for the month of follow-up. Perhaps, suturing led to tugging and pulling of the tissues, whereas the open side was less constrained during swallowing. Additionally, perhaps merely doing more (suturing) induced more trauma, and hence inflammation, which might explain the increased pain. A build-up of fluid under pressure in the tissues, as is seen in a hematoma, often causes pain. In this study, the tonsillar fossae were not closed water-tight, so any fluids that might have accumulated in the fossa were allowed to drain. Our data does not reveal a provable explanation for the finding of increased pain in the sutured fossa. 5.4. Clinical relevance Clinically, this study supports the idea that the common practice during tonsillectomy of not suturing the tonsillar fossa is preferable suturing. The patients will have less risk of bleeding and less pain with an open (unsutured) tonsillar fossa. Hence the surgical desire to close a wound is not advantageous, in the case of tonsillectomy. However, our study actually supports, with data, the common practice of leaving the fossae unsutured during tonsillectomy. We found that a non-sutured fossa hurt less with a more experienced surgeon, but hurt more in a sutured fossa. This is not consistent with the findings of Weighill et al. [7] and Ramjettan and Singh [9], who did not find a difference between sewn and unsewn sides. The factors that contribute to a good outcome in surgery are many, and not always intuitive or easy to discern. 5.5. Design considerations One unaccounted for factor may be differences in actually suturing the tonsillar pillars, such as not sewing the muscle and mucosa, but just sewing the mucosa, or conversely, just sewing the muscle without including mucosa. We had some concern that if each surgeon sutured the pillar slightly differently, the results might be affected and we did not control for this variation. At follow-up examinations, often the sutured side had not stayed closed when the pharynx was viewed in the first 2 weeks postoperatively. Perhaps the suture type was the wrong choice, as chromic does not last long (10–14 days), though longer than gut sutures (7–10 days), but not as long as PDS (up to 6 weeks). The precise number of subjects whose sutured side stayed closed was not collected and is retrospectively unobtainable. We cannot speak to whether subjects who had a sutured side open up had more or less pain. Another problem is the incomplete data. Information was obtained from the subject, or caregiver(s). Both of these sources of information could introduce distortion of the true pain conditions
B.H. Matt et al. / International Journal of Pediatric Otorhinolaryngology 76 (2012) 1799–1805
of the subject. Caregivers may misinterpret or confuse their left and right (or the subject’s left or right), and subjects may not be precise in declaring exactly where the throat hurt. Additionally, most of the subjects received concurrent operations besides tonsillectomy. We believe that bilateral myringotomy and tubes, or bilateral cerumen removal would not be expected to increase pain unilaterally, or even if it did, the pain from these procedures is almost always gone in 24 h. Adding adenoidectomy to tonsillectomy is common in treating pharyngitis and obstructive sleep apnea in children. As the adenoid is a central structure, the pain is expected to be midline, and thus not bias the patient’s perception to one side. In any event, the average duration of pain after adenoidectomy alone is 3 days (range 0– 8 days) [16]. The literature is silent on whether tonsillectomy pain is additive to adenoidectomy pain, though personal observations of the authors imply that the more the surgeon operates (more tissue disruption, more area involved), the more it hurts the patient. A number of children also received a supraglottoplasty. Fortunately, that pain is often minor and probably gone in less than 1 day [17], so it would not be expected to influence the sided of any pain, or cause any bleeding in the tonsillar fossa. Overall, the subjects were relatively homogenous in the indication for tonsillectomy, as almost 90% underwent the operation for obstructive sleep disorder symptoms. 377 of 763 patients undergoing tonsillectomies had supraglottoplasties. Evaluation and treatment of the entire airway in obstructive sleep apnea is not considered routine for many pediatric otolaryngology practices. Groblewski et al. [17] support the idea that supraglottoplasty does not hurt much, nor for very long, and therefore would not be expected to influence the sidedness of the pain scores. Walton et al. [18] in a meta-analysis find the mean resolution of pain for complete tonsillectomy to be (mean SD) 7.6 ( 2.14) days. This supports the idea that tonsillectomy hurts much more than supraglottoplasty, and hence tonsillectomy would be expected to account for the vast majority of the pain and for longer time periods, compared to supraglottoplasty. Another criticism that can be raised with this study is the inclusion of younger children who would not necessarily have the developmental capacity to identify unilateral pain. Certainly pain in general can be deduced by evaluation of a subject’s vital signs and other measures like crying or irritability. Excluding children younger than 3 would reduce the sample size to 561. However, many of those under age 3 either had no pain by postoperative day 7 (as assessed by the caregivers), or did not provide data on which if any side hurt. The statistical analysis we used has accounted for missing data points and thus our conclusions on pain remain statistically valid. 6. Conclusions The data show that the use of suture closure of the tonsillar pillars after tonsillectomy does not reduce the risk of bleeding. In fact, in tonsillar fossae left open, the more experienced the surgeon, the lower the risk of bleeding. In closed tonsillar fossae, the more experienced the surgeon, the higher the risk of bleeding. This finding is counter-intuitive, as one would expect more experience to lead to less bleeding, and better suturing. Additionally, closing the tonsillar pillar increased the odds of postoperative pain by 40% compared to leaving the tonsillar pillar open. This is also counterintuitive, as closed wounds elsewhere on the body (e.g. skin) tend to hurt less. Though the subjects were not completely homogenous (e.g. differences exist in indications for operation, comorbidities,
1805
age, surgeon, and hospital), they represent a real-world cohort of children undergoing tonsillectomy. Based on this study, we cannot recommend suturing of the tonsillar fossae after tonsillectomy. Funding Funding for biostatistical services was provided by the General Otorhinolaryngological Foundation, Indianapolis, IN. Conflict of interest statement The authors have no conflicts of interest to report. Acknowledgments The authors gratefully acknowledge the contributions of the other attending surgeons: M.S. Chand, S.R. Cordes, M.H. Fritsch, R.A. Hamaker, R.C. Miyamoto, A.O. Yekinni, the School of Medicine, Department of Otolaryngology – Head and Neck Surgery residents, nurses, secretaries and phone-interviewers, for this study. References [1] J.L. Paradise, C.D. Bluestone, D.K. Colborn, B.S. Bernard, H.E. Rockette, M. KursLasky, Tonsillectomy and adenotonsillectomy for recurrent throat infection in moderately affected children, Pediatrics 110 (2002) 7–15. [2] M.F. Owings, L.J. Kozak, Ambulatory and inpatient procedures in the United States, 1996, in: Vital & Health Statistics – Series 13: Data from the National Health Survey, 1998, pp. 1–119. [3] K. Cullen, M. Hall, A. Golosinskiy, Centers for disease control, ambulatory surgery in the United States, 2006, in: National Health Statistics Reports, 2009, p. 28, Available at: http://www.cdc.gov/nchs/data/nhsr/nhsr011.pdf (accessed 01.09.11). [4] N. Bhattacharyya, L.J. Kepnes, Economic benefit of tonsillectomy in adults with chronic tonsillitis, Ann. Otol. Rhinol. Laryngol. 111 (2002) 983–988. [5] J.M. Curtin, The history of tonsil and adenoid surgery, Orthop. Clin. North Am. 20 (1987) 415–419. [6] L. Neumayer, D. Vargo, Principles of preoperative and operative surgery, in: C.M. Townsend, Jr., R.D. Beauchamp, B.M. Evers, K.L. Mattox (Eds.), Sabiston Textbook of Surgery: The Biological Basis of Modern Surgical Practice, 18th ed., Saunders Elsevier, Philadelphia, 2008, pp. 251–279. [7] J.S. Weighill, D.W. Proops, D. Jeffries, J. Brandrick, Pain relief following tonsillectomy (does sewing the faucial pillars together help?), J. Laryngol. Otol. 100 (1986) 307–310. [8] V. Nandapalan, J.C. McIlwain, Tonsillar fossa obliteration and post-operative pain, Clin. Otolaryngol. Allied Sci. 20 (1995) 127–129. [9] S. Ramjettan, B. Singh, Are sutured faucial pillars really an advantage in tonsillectomy? S. Afr. J. Surg. 34 (1996) 189–191. [10] E. Genc, D. Hanci, N.T. Ergin, T. Dal, Can mucosal sealing reduce tonsillectomy pain? Int. J. Pediatr. Otorhinolaryngol. 70 (2006) 725–730. [11] B. Dawson, R.G. Trapp, Basic and Clinical Biostatistics, 4th ed., Lange Medical Books/McGraw-Hill, New York, 2004. [12] S.J. Doshi, A Complete Review of Wound Closure, Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, 1992. [13] R.R. Davies, M.J. Russo, K.N. Hong, S. Mital, R.S. Mosca, J.M. Quaegebeur, et al., Increased short- and long-term mortality at low-volume pediatric heart transplant centers: should minimum standards be set? Retrospective data analysis, Ann. Surg. 253 (2011) 393–401. [14] ACGME, Common Program Requirements: General Competencies, 2007 http:// www.acgme.org/outcome/comp/GeneralCompetenciesStandards21307.pdf (accessed 30.07.11). [15] C. Roberts, S. Jayaramachandran, C.H. Raine, A prospective study of factors which may predispose to post-operative tonsillar fossa haemorrhage, Clin. Otolaryngol. Allied Sci. 17 (1992) 13–17. [16] H.M. Tuomilehto, H.M. Kokki, R.P. Ahonen, J.M. Nuutinen, Postoperative behavioral changes in children after adenoidectomy, Arch. Otolaryngol. Head Neck Surg. 128 (2002) 1159–1164. [17] J.C. Groblewski, R.K. Shah, G.H. Zalzal, Microdebrider-assisted supraglottoplasty for laryngomalacia, Ann. Otol. Rhinol. Laryngol. 118 (2009) 592–597. [18] J. Walton, Y. Ebner, M.G. Stewart, M.M. April, Systematic review of randomized controlled trials comparing intracapsular tonsillectomy with total tonsillectomy in a pediatric population, Arch. Otolaryngol. Head Neck Surg. 138 (2012) 243– 249.