- Email: [email protected]
Diathermy power settings as a risk factor for hemorrhage after tonsillectomy
Otolaryngology–Head and Neck Surgery (2009) 140, 23-28
ORIGINAL RESEARCH–GENERAL OTOLARYNGOLOGY
Diathermy power settings as a risk factor for hemorrhage after tonsillectomy David Lowe, FRCS, David A. Cromwell, PhD, James D. Lewsey, PhD, Lynn P. Copley, MSc, Peter Brown, FRCS, Matthew Yung, FRCS, and Jan H. van der Meulen, PhD, London, Glasgow, Buckinghamshire, and Suffolk, UK OBJECTIVE: To investigate bipolar diathermy power settings as a risk factor for postoperative hemorrhage following tonsillectomy. STUDY DESIGN AND SETTING: A prospective cohort study was undertaken between July 2003 and September 2004 in National Health Service (NHS) and independent hospitals in England and Northern Ireland. Data were collected on patient characteristics, tonsillectomy technique, and postoperative hemorrhage within 28 days of surgery. RESULTS: Among the 9572 patients who had a tonsillectomy performed with bipolar diathermy dissection and hemostasis, the overall rate of hemorrhage was 4.6 percent and the risk of hemorrhage was not associated with the diathermy power setting. Among the 8465 patients who had tonsillectomy with cold steel dissection and bipolar diathermy hemostasis, the rate of hemorrhage increased from 1.8% in patients with the lowest power settings (6 to 8 watts) to 3.7% in those with settings above 18 watts (P value for trend ⫽ 0.005). CONCLUSION: In tonsillectomies using cold steel dissection and bipolar diathermy for hemostasis, the risk of postoperative hemorrhage becomes greater as diathermy power increases. © 2009 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
O
ver the past 20 years, an increasing number of tonsillectomies have been performed using “hot” electrosurgical techniques, with diathermy being among the most widely used.1,2 On March 24, 2004, the National Institute for Health and Clinical Excellence (NICE), together with the British Association of Otorhinolaryngologists–Head and Neck Surgeons (BAO-HNS), published interim guidance on the use of diathermy in tonsillectomy, recommending that surgeons use as little diathermy as possible especially when it was used for both dissection and hemostasis.3 The guidance was developed from the interim results of the National Prospective Tonsillectomy Audit (NPTA), which had begun in July 2003 to monitor the occurrence of postoperative hemorrhage and other complications. The interim results showed that the rate of hemorrhage was three times higher
with the use of diathermy throughout an operation when compared to the traditional approach of “cold” dissection, with subsequent bleeding controlled with ties or gauze packs.4 In addition, hemorrhage rates were found to be greater when diathermy was used for both tonsillar dissection and hemostasis compared to when it was used for hemostasis alone. The NPTA continued until September 2004 and these results were confirmed in the final analysis,5 with both monopolar and bipolar diathermy showing similar levels of increased risk. As well as collecting information on surgical technique, the NPTA asked surgeons to record the diathermy power setting used during a tonsillectomy. In this paper, we provide a detailed analysis of the relationship between bipolar diathermy power setting for tonsillectomy and postoperative hemorrhage.
MATERIALS AND METHODS The design and method of data collection used by the NPTA has been described previously.4-6 In brief, patients were eligible for inclusion if they had a tonsillectomy between July 7, 2003 and September 30, 2004 in a National Health Service (NHS) or independent hospital in England and Northern Ireland. Patients were excluded if the procedure was a tonsillectomy revision or a tonsillar biopsy, if it was for known cancer, or if it was undertaken with palatal surgery. Data were prospectively collected on patient characteristics, the surgical operation, and subsequent complications. A primary postoperative hemorrhage was defined as any bleeding that led to delayed hospital discharge, blood transfusion, or return to theatre during the initial stay. A secondary hemorrhage was any postoperative bleed that occurred after discharge from hospital and led to readmission to hospital within 28 days of surgery. The audit was approved by the Northern and Yorkshire multicenter research ethics committee.
Received March 18, 2008; revised August 21, 2008; accepted August 21, 2008.
0194-5998/$36.00 © 2009 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2008.08.025
24
Otolaryngology–Head and Neck Surgery, Vol 140, No 1, January 2009
Tonsillectomies were grouped according to the techniques used for dissection and hemostasis. In this analysis, we focus on two of these categories: the first contained procedures in which bipolar diathermy was used for both dissection and hemostasis, while the second contained procedures in which bipolar diathermy was used for hemostasis after cold steel dissection (ie, the use of scissors and/or blunt dissector and snare). Surgeons recorded maximum power setting for both dissection and hemostasis, indicating both the power setting and the unit. During the analysis, it was found that 97 percent of all power settings were reported in watts, and so the analysis was restricted to these data. The hemorrhage rate and relative risks were calculated for specified ranges of power settings. The relative risks were calculated as ratios of the hemorrhage rates for each group, with the group with the lowest power setting being used as the reference. The groups were defined to reflect the distribution of the power setting values, which tended to take particular discrete values (8, 10, 12, 15, and 20 watts). Data were also grouped because there was no reason to suppose beforehand that the risk of hemorrhage increased linearly with every unit increase in power. Values higher than 50 watts and lower than 6 watts were judged to be unrealistic because they would have had either an excessive or insufficient dissection or coagulative effect. The data of these patients were therefore excluded. Procedures performed using diathermy throughout were grouped using the power setting for dissection only as there was a very high correlation between the values recorded for dissection and hemostasis (Pearson’s r ⫽ 0.97). Evidence of a relationship between the power categories and the unadjusted hemorrhage rates was assessed using the 2 test for trend.7 Multiple multilevel logistic regression8 was then used to adjust the risk estimates for potential confounding factors, namely, patient age and sex, indication for surgery, grade of operating surgeon, type of hospital (NHS or independent), and whether the procedure was performed before or after the guidance was published. A twostage analysis was performed. In the first stage, the regression model contained the power setting categories used in the unadjusted analysis. In the second stage, if justified, the regression model would include power setting as a continuous variable because this would more precisely estimate any dose-response relationship if the trend was approximately linear. Multilevel logistic regression was used to account for any lack of independence in the data due to patients being clustered within hospitals. All P values were two-sided, and those lower than 0.05 were judged to be statistically significant. STATA software (version 9.2) was used for all statistical calculations.
RESULTS The NPTA database contained 12,562 patients who had a tonsillectomy performed with bipolar diathermy dissection
and hemostasis and 11,956 patients who had cold steel dissection with bipolar diathermy hemostasis. Excluding patients with missing data or invalid power settings (see Table 1) meant that the analysis included 9572 (76%) of the diathermy dissection patients and 8465 (71%) of the cold steel dissection patients. There were slightly more female than male patients, and around 60 percent were under 16 years of age, with recurrent acute tonsillitis being the most common indication for surgery (Table 1). Over 85 percent of procedures were performed in NHS hospitals, and most patients were admitted as inpatients rather than day cases. Among the 9572 patients who underwent diathermy dissection and hemostasis, a total of 438 patients (4.6%) experienced a postoperative hemorrhage within 28 days. Specifically, these patients had 43 (0.4%) primary hemorrhages and 404 (4.2%) secondary hemorrhages (some patients had both a primary and secondary hemorrhage). The rate of hemorrhage increased with age and was noticeably lower for patients whose indication for surgery was an obstructive rather than infectious condition (Table 1). The rate of hemorrhage did not depend upon the power setting used for dissection (2 test for trend, P ⫽ 0.8), with the relative risks of the various groups not differing from the reference group by more than 20 percent (Table 2). The lack of association between hemorrhage rates and power setting remained after using logistic regression to adjust for possible confounders. Among the 8465 patients that had cold steel dissection and diathermy hemostasis, a hemorrhage was reported for 229 patients (2.8%). In total, there were 39 (0.5%) primary hemorrhages and 196 (2.3%) secondary hemorrhages. As before, the rate of hemorrhage increased with age and was lower for patients with an obstructive rather than infectious condition (Table 1). The rate of hemorrhage rose steadily as the power setting increased, from 1.8% in patients with the lowest power settings (6 to 8 watts) to 3.7% in those with settings above 18 watts (Table 2), which represents a doubling in risk (2 test for trend, P ⫽ 0.005). The relationship between diathermy power setting and the overall risk of hemorrhage reflects the risk of a patient experiencing a secondary hemorrhage (Table 3). The trend in secondary hemorrhage risk across the groups was statistically significant (2 test for trend, P ⬍ 0.001). The study did not have the statistical power to draw conclusions about the relationship between diathermy power setting and primary hemorrhage (2 test for trend, P ⫽ 0.1). Using multiple multilevel logistic regression, we observed that adjustment for confounding did not affect the relationship between the diathermy power setting and the overall risk of hemorrhage (Table 4). Consequently, the second model was used to estimate the change in risk associated with an increase of 1 watt of power. Because the evidence that the trend was approximately linear extended over the range of the categories, it was decided to winsor the power setting data.9 Winsorization prevented the highest values from having undue influence on the estimated slope and involved replacing the highest power setting values with a maximum
Lowe et al
Diathermy power settings as a risk factor for . . .
25
Table 1 Characteristics of patients undergoing tonsillectomy involving diathermy Bipolar diathermy for dissection and hemostasis Patients (%) Patients All patients in NPTA database Patients with no power settings or power settings not in watts Patients with invalid power values Patients included in analysis Sex Male Female Age group Up to 5 years 5-15 years 16 years and over Indication for surgery Recurrent acute tonsillitis Chronic tonsillitis Previous quinsy Pharyngeal obstruction Other Grade of operating surgeon Consultant Nontraining staff grade Specialty registrar Senior house officer
Hemorrhage rate
Cold steel dissection and bipolar diathermy hemostasis Patients (%)
12,562
11,956
2106 884 9572
2849 642 8465
Hemorrhage rate
3928 (41) 5581 (59)
4.8% 4.4%
3342 (40) 5073 (60)
2.9% 2.6%
1627 (17) 4534 (47) 3395 (36)
2.2% 3.7% 7.0%
1207 (14) 3969 (47) 3284 (39)
1.2% 2.3% 3.8%
7159 650 204 1147 163
(77) (7) (2) (12) (2)
4.9% 6.5% 4.9% 1.8% 3.1%
6601 620 182 695 178
(80) (7) (2) (8) (2)
2.9% 2.1% 5.5% 1.2% 1.7%
3925 1874 2532 1232
(41) (20) (26) (13)
3.7% 5.7% 4.6% 5.5%
2716 2561 1758 1418
(32) (30) (21) (17)
2.5% 2.6% 2.5% 3.5%
value (set at 23 watts). A total of 369 values (4.4%) were winsored. Using this model, it was estimated that the risk of a hemorrhage increased by a factor of 1.05 (95% CI for odds ratio ⫽ 1.01 to 1.10; P ⫽ 0.017) for an increase of 1 watt
in power, or equivalently by a factor of 1.67 (95% CI for odds ratio ⫽ 1.09 to 2.54) for an increase of 10 watts in power. This estimate produced similar odds ratios for the original power setting categories (Table 4).
Table 2 Rates of tonsillar hemorrhage by surgical technique for various diathermy power settings Power setting Diathermy for dissection and hemostasis* 6 to 8 watts 9 to 11 watts 12 to 14 watts 15 to 17 watts 18 watts plus Total Cold steel dissection and diathermy only for hemostatis 6 to 8 watts 9 to 11 watts 12 to 14 watts 15 to 17 watts 18 watts plus Total
Number of operations (%)
Events
Rate†
Relative risk† (95% CI)
1098 (11%) 2719 (28%) 1863 (19%) 2552 (27%) 1340 (14%) 9572
48 126 91 103 70 438
4.37% 4.63% 4.88% 4.04% 5.22% 4.58%
1.06 1.12 0.92 1.19
1 (0.77 (0.79 (0.66 (0.83
to to to to
1.47) 1.57) 1.29) 1.71)
735 (8%) 2025 (21%) 1516 (16%) 3037 (32%) 1152 (12%) 8465
13 48 37 88 43 229
1.77% 2.37% 2.44% 2.90% 3.73% 2.71%
1.34 1.38 1.64 2.11
1 (0.73 (0.74 (0.92 (1.24
to to to to
2.46) 2.56) 2.92) 3.90)
*Based on power setting specified for dissection. †2 test for trend P ⫽ 0.8 (diathermy for dissection and hemostasis) and P ⫽ 0.005 (diathermy only for hemostasis).
26
Otolaryngology–Head and Neck Surgery, Vol 140, No 1, January 2009
Table 3 Rates of primary and secondary hemorrhage among patients undergoing cold steel dissection and diathermy hemostasis* Primary hemorrhage Power setting (watts)
Secondary hemorrhage
Number of operations
Events
Rate†
Relative risk
Events
Rate†
Relative risk
735 2025 1516 3037 1209 8522
6 13 5 9 6 39
0.82% 0.64% 0.33% 0.30% 0.50% 0.46%
1 0.79 0.40 0.36 0.61
9 36 33 80 44 202
1.22% 1.78% 2.18% 2.63% 3.64% 2.37%
1 1.45 1.78 2.15 2.97
6 to 8 9 to 11 12 to 14 15 to 17 18 and over Total
*The sum of the primary and secondary hemorrhages for the various groups may exceed the total number of hemorrhages in Table 2 because some patients had both a primary and secondary hemorrhage. †2 test for trend P ⫽ 0.1 (primary hemorrhage) and P ⬍ 0.001 (secondary hemorrhage).
DISCUSSION The NPTA has previously demonstrated that the risk of postoperative hemorrhage after tonsillectomy was greater when bipolar diathermy was used for both tonsillar dissection and hemostasis compared to diathermy hemostasis alone.4-6 In this study, we found that there is also a “doseresponse” relationship between the risk of hemorrhage and the diathermy power setting when bipolar diathermy is used for hemostasis after cold steel dissection. Specifically, the rate of hemorrhage for patients with diathermy power settings above 18 watts was 3.7 percent, approximately twice the rate for patients with power settings between 6 and 8 watts (1.8%). Indeed, the risk of hemorrhage when power was above 18 watts approached the rate of hemorrhage observed when bipolar diathermy was used for dissection and hemostasis (4.6% overall). The latter could represent a threshold above which increasing amounts of diathermy power do not affect the risk of hemorrhage. This would explain why we did not find a relationship between diathermy power setting and hemorrhage when diathermy was used for both dissection and hemostasis.
The NPTA is the largest prospective cohort study to date on complications after tonsillectomy, collecting data on over 24,000 operations involving bipolar diathermy in England and Northern Ireland. The size of the study and the prospective data collection are particular strengths. But, being an observational study, it is susceptible to a number of biases. First, the results are based on procedures and outcomes reported by the participating hospitals. Incomplete data submission could have influenced the results if the probability of a patient not being captured by the study was related to the choice of technique, the power setting, or the other risk factors. We compared the audit data with data from an administrative database (called HES) that collects information on all acute episodes in English NHS hospitals. This showed that the audit included about 60 percent of the eligible patients and the case mix of patients in the audit was representative. The observed hemorrhage rates in the preand postguidance periods in the NPTA data were 4.2 percent and 2.9 percent, respectively, compared to the HES-derived pre- and postguidance hemorrhage rates of 6.2 percent and 5.9 percent, respectively.10 As this analysis was concerned primarily with
Table 4 Unadjusted and adjusted odds ratios of postoperative hemorrhage for tonsillectomy using diathermy for hemostasis
Power setting
Number of operations
Hemorrhage rate
Unadjusted OR
6 to 8 watts 9 to 11 watts 12 to 14 watts 15 to 17 watts 18 watts plus
735 2,025 1516 3037 1152
1.77% 2.37% 2.44% 2.90% 3.73%
1 1.35 1.39 1.66 2.15
Adjusted OR* (95% CI)
1.19 1.47 1.68 2.33
1 (0.6 (0.7 (0.8 (1.1
to to to to
2.4) 3.1) 3.3) 4.7)
OR estimated with power setting treated as continuous† 1 1.17 1.36 1.58 1.84
*Odds ratio (OR) adjusted for sex, age, type of hospital (NHS or independent), grade of operating surgeon, indication for surgery, and whether the procedure was performed before or after the NICE guidance was published. †Odds ratio for each power setting category based on the adjusted odds ratio for a unit increase in power (1 watt). The odds ratio was estimated as 1.0524 (95% CI 1.01 to 1.10; P ⫽ 0.017) for each 1-watt increase.
Lowe et al
Diathermy power settings as a risk factor for . . .
the differences in the relative risks and not in the risks themselves, this underreporting would influence the results only if hospitals submitted data selectively. There was no evidence to suggest selective reporting of hemorrhage among patients of similar characteristics. Another weakness of this study concerns the quality of the power settings data. Bias may result from the exclusion of patients with missing or invalid values. However, the characteristics of excluded patients, and their outcomes, were similar to those with power setting values. For the diathermy dissection group, the hemorrhage rates in patients with and without power values were 5.1 and 4.6, respectively (2 test P ⫽ 0.17). For the cold steel dissection group, the values were 2.72 percent and 2.71 percent, respectively (P ⫽ 0.96). The reported power setting values may have also differed from the actual power used. Some diathermy generators give only limited information on the power being delivered, and this may partially explain the grouping of the settings around particular values. In addition, the study did not collect data on how long diathermy was applied to a patient. However, any such errors are likely to cause the strength of the relationship between hemorrhage risk and diathermy power setting to be underestimated rather than overestimated due to an effect known as “regression dilution.”11 It might be argued that the use of diathermy for hemostasis after cold steel dissection was a response to excessive intraoperative bleeding when ties have already been used, and so might identify patients who were predisposed to intraoperative as well as postoperative hemorrhage. This seems an unlikely explanation because the audit received information on 4285 tonsillectomies using cold steel dissection and ties for hemostasis. Moreover, the grouping of power settings around specific values suggests that these were the routine settings of surgeons with a preference for diathermy12 rather than settings used in problematic situations. It might also be argued that the use of higher diathermy settings in hemostasis was caused by excessive intraoperative bleeding among patients predisposed to postoperative hemorrhage. We feel that this is unlikely because the adjusted analysis included indications for surgery and patient age, factors that are generally considered to be the best predictors of a particularly difficult procedure. The reason why higher power settings increase the risk of secondary hemorrhage is not clear. We suspect that higher diathermy power settings increase the area of tissue damage due to thermal injury, which results in a larger area of tissue necrosis within the tonsillar bed. It has also been suggested that the pathology of tissue injury differs between cold steel and diathermy dissection, with diathermy resulting in more damage.13 Greater tissue damage would be expected to increase pain in the first week after the operation, and this was observed in the only other study that we found to have examined how diathermy power may influence the outcomes after tonsillectomy. Cardozo et al14 enrolled 101 patients age over 13 years who had cold steel dissection followed by diathermy hemostasis, and measured
27
the amount of diathermy energy used by timing how long diathermy was applied at a preset power setting of 15 W. They reported a positive correlation between diathermy power and postoperative pain at five and 10 days after tonsillectomy (r ⫽ 0.55 and r ⫽ 0.44, respectively). They also reported that patients with any secondary hemorrhage had been exposed to higher median energy levels than those with no bleeding (1449 vs 1036 joules, respectively, P ⫽ 0.032). More generally, a systematic review of studies comparing hot vs cold techniques reported that all studies found higher pain scores among those patients undergoing hot techniques between four and 10 days postoperatively compared to patients undergoing cold tonsillectomy.15 The overall hemorrhage rates for the two techniques observed in the NPTA are similar to rates described in other recent studies on the relationship between postoperative hemorrhage and tonsillectomy technique. Data from the Welsh National Registry16 gave a hemorrhage rate of 2.1 percent for 141 patients undergoing cold steel dissection and bipolar diathermy hemostasis, and 9.8 percent for 325 patients undergoing bipolar diathermy dissection and hemostasis (relative risk ⫽ 4.6; 95 percent CI: 1.4 to 14.9). Several other large studies have reported a slight increase in the risk of hemorrhage for diathermy dissection compared to cold steel dissection, but the differences have not been statistically significant.13,17 A systematic review published in 2006 combined data from 24 studies16 and reported that a hemorrhage occurred in 5.3 percent of the 1676 patients undergoing bipolar diathermy dissection and hemostasis. However, among 983 patients undergoing cold steel dissection and bipolar diathermy hemostasis, the rate was 7.3 percent, giving a relative risk for diathermy dissection of 0.72 (95% CI: 0.54 to 0.98). The reason for the higher rate among the “cold” tonsillectomies was not clear and the authors of the review acknowledge that the results reflect differences in study design as well as the characteristics of the enrolled patients. The original guidance from NICE focused on the combination of cold and hot surgical techniques in terms of their use for dissection and hemostasis.3 It recommended that surgeons use as little diathermy as possible but it did not provide information to help determine what level of diathermy is acceptable. A survey of UK ENT surgeons conducted after the guidance was published reported that some surgeons responded by using cold steel rather than diathermy for dissection while continuing to use diathermy for hemostasis.18 Other surgeons began performing tonsillectomy with cold steel and ties, the technique found in the NPTA to have the lowest overall hemorrhage rate (1.7%). However, this carries some increased risk of primary/reactionary bleeding, probably related to the slipping of ligatures in the early postoperative period. This study is of importance because it provides more precise information on how diathermy use may be limited. The results show that the hemorrhage rate for cold steel dissection with diathermy hemostasis when performed at 6 to 8 watts is comparable to the hemorrhage rate observed for cold steel and ties (1.8%
28
Otolaryngology–Head and Neck Surgery, Vol 140, No 1, January 2009
vs 1.7%, respectively). The results therefore suggest that cold steel tonsillectomy with careful bipolar diathermy hemostasis at low settings is an equally effective method for minimizing postoperative hemorrhage.
and Public Safety (Northern Ireland). Jan van der Meulen is supported by a National Public Health Career Scientist Award, Department of Health– NHS R&D Programme UK.
REFERENCES ACKNOWLEDGEMENTS We thank the staff of all the hospitals in England and Northern Ireland who contributed data to the National Prospective Tonsillectomy Audit. This audit was carried out by the Comparative Audit Group of the British Association of Otorhinolaryngologists–Head and Neck Surgeons and the Clinical Effectiveness Unit of The Royal College of Surgeons of England– London School of Hygiene and Tropical Medicine.
AUTHOR INFORMATION From the Clinical Effectiveness Unit (Drs Lowe, Cromwell, Copley, and van der Meulen), The Royal College of Surgeons of England, London, England, UK; Public Health and Policy (Drs Cromwell and van der Meulen), London School of Hygiene and Tropical Medicine, London, England, UK; the Division of Community Based Sciences (Dr Lewsey), University of Glasgow, Glasgow, Scotland, UK; the ENT Department (Dr Brown), Milton Keynes General NHS Trust, Milton Keynes, Buckinghamshire, England, UK; and the Department of Otolaryngology (Dr Yung), Ipswich Hospital NHS Trust, Ipswich, Suffolk, UK. Corresponding author: Dr David Cromwell, Clinical Effectiveness Unit, The Royal College of Surgeons of England, 35-43 Lincoln’s Inn Fields, London WC2A 3PE United Kingdom. E-mail address: [email protected].
AUTHOR CONTRIBUTIONS David Lowe, coordinated the data collection; supported data analysis, and co-wrote first draft of paper; David A. Cromwell, performed data analysis and co-wrote first draft of paper; James D. Lewsey, supported data analysis, contributed to revision of article; Lynn P. Copley, coordinated data collection and supported data analysis, contributed to revision of article; Peter Brown, designed the audit and supervised its execution, contributed to revision of article; Matthew Yung, designed the audit and supervised its execution, contributed to revision of article; Jan H. van der Meulen, designed the audit and supervised its execution, contributed to revision of article.
FINANCIAL DISCLOSURES The National Prospective Tonsillectomy Audit was funded by the Department of Health (England) and the Department of Health, Social Services
1. Murty GE, Watson MG. Diathermy haemostasis at tonsillectomy: current practice—a survey of UK otolaryngologists. J Laryngol Otol 1990;104:549 –52. 2. Walner DL, Parker NP, Miller RP. Past and present instrument use in pediatric adenotonsillectomy. Otolaryngol Head Neck Surg 2007;137: 49 –53. 3. NICE. Interim guidance on the use of diathermy in tonsillectomy. Available at: http://www.nice.org.uk. Accessed: April 20, 2005. 4. Lowe D, van der Meulen J, National Prospective Tonsillectomy Audit. Tonsillectomy technique as a risk factor for postoperative haemorrhage. Lancet 2004;364:697–702. 5. Lowe D, van der Meulen J, Cromwell D, et al. Key messages from the National Prospective Tonsillectomy Audit. Laryngoscope 2007;117: 717–24. 6. National Prospective Tonsillectomy Audit: Final report. Available at: http://www.entuk.org/members/audits/tonsil/Tonsillectomyauditreport_ pdf. Accessed: December 10, 2007. 7. Altman DG. Practical statistics for medical research. London: Chapman and Hall; 1991. 8. Leyland AH, Goldstein H. Multilevel modelling of health statistics. Chichester: Wiley & Sons; 2001. 9. Hoaglin DC, Mosteller F, Tukey JW. Understanding robust and exploratory data analysis. Chichester: Wiley & Sons; 2000. 10. National Prospective Tonsillectomy Audit. Impact of NICE guidance on rates of haemorrhage after tonsillectomy: an evaluation of guidance issued during an ongoing national tonsillectomy audit. Qual Saf Health Care 2008;17:264 – 8. 11. MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart disease. Part 1, Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet 1990;335:765–74. 12. Myssiorek D, Alvi A. Post-tonsillectomy haemorrhage: an assessment of risk factors. Int J Pediatr Otorhinolaryngol 1996;37:35– 43. 13. O’Leary S, Vorrath J. Postoperative bleeding after diathermy and dissection tonsillectomy. Laryngoscope 2005;115:591– 4. 14. Cardozo AA, Hallikeri C, Lawrence H, et al. Teenage and adult tonsillectomy: dose-response relationship between diathermy energy used and morbidity. Clin Otolaryngol 2007;32:366 –71. 15. Leinbach RF, Markwell SJ, Colliver JA, et al. Hot versus cold tonsillectomy: a systematic review of the literature. Otolaryngol Head Neck Surg 2003;129:360 – 4. 16. Mowatt G, Cook JA, Fraser C, et al. Systematic review of the safety of electrosurgery for tonsillectomy. Clin Otolaryngol 2006;31:95–102. 17. Walker P, Gillies D. Post-tonsillectomy hemorrhage rates: Are they technique-dependent? Otolaryngol Head Neck Surg 2007;136(4 Suppl): S27–31. 18. Hopkins C, Ajulo P, Krywawych M, et al. The impact of the interim report of the National Prospective Tonsillectomy Audit. Clin Otolaryngol 2005;30:475– 6.
ORIGINAL RESEARCH–GENERAL OTOLARYNGOLOGY
Diathermy power settings as a risk factor for hemorrhage after tonsillectomy David Lowe, FRCS, David A. Cromwell, PhD, James D. Lewsey, PhD, Lynn P. Copley, MSc, Peter Brown, FRCS, Matthew Yung, FRCS, and Jan H. van der Meulen, PhD, London, Glasgow, Buckinghamshire, and Suffolk, UK OBJECTIVE: To investigate bipolar diathermy power settings as a risk factor for postoperative hemorrhage following tonsillectomy. STUDY DESIGN AND SETTING: A prospective cohort study was undertaken between July 2003 and September 2004 in National Health Service (NHS) and independent hospitals in England and Northern Ireland. Data were collected on patient characteristics, tonsillectomy technique, and postoperative hemorrhage within 28 days of surgery. RESULTS: Among the 9572 patients who had a tonsillectomy performed with bipolar diathermy dissection and hemostasis, the overall rate of hemorrhage was 4.6 percent and the risk of hemorrhage was not associated with the diathermy power setting. Among the 8465 patients who had tonsillectomy with cold steel dissection and bipolar diathermy hemostasis, the rate of hemorrhage increased from 1.8% in patients with the lowest power settings (6 to 8 watts) to 3.7% in those with settings above 18 watts (P value for trend ⫽ 0.005). CONCLUSION: In tonsillectomies using cold steel dissection and bipolar diathermy for hemostasis, the risk of postoperative hemorrhage becomes greater as diathermy power increases. © 2009 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved.
O
ver the past 20 years, an increasing number of tonsillectomies have been performed using “hot” electrosurgical techniques, with diathermy being among the most widely used.1,2 On March 24, 2004, the National Institute for Health and Clinical Excellence (NICE), together with the British Association of Otorhinolaryngologists–Head and Neck Surgeons (BAO-HNS), published interim guidance on the use of diathermy in tonsillectomy, recommending that surgeons use as little diathermy as possible especially when it was used for both dissection and hemostasis.3 The guidance was developed from the interim results of the National Prospective Tonsillectomy Audit (NPTA), which had begun in July 2003 to monitor the occurrence of postoperative hemorrhage and other complications. The interim results showed that the rate of hemorrhage was three times higher
with the use of diathermy throughout an operation when compared to the traditional approach of “cold” dissection, with subsequent bleeding controlled with ties or gauze packs.4 In addition, hemorrhage rates were found to be greater when diathermy was used for both tonsillar dissection and hemostasis compared to when it was used for hemostasis alone. The NPTA continued until September 2004 and these results were confirmed in the final analysis,5 with both monopolar and bipolar diathermy showing similar levels of increased risk. As well as collecting information on surgical technique, the NPTA asked surgeons to record the diathermy power setting used during a tonsillectomy. In this paper, we provide a detailed analysis of the relationship between bipolar diathermy power setting for tonsillectomy and postoperative hemorrhage.
MATERIALS AND METHODS The design and method of data collection used by the NPTA has been described previously.4-6 In brief, patients were eligible for inclusion if they had a tonsillectomy between July 7, 2003 and September 30, 2004 in a National Health Service (NHS) or independent hospital in England and Northern Ireland. Patients were excluded if the procedure was a tonsillectomy revision or a tonsillar biopsy, if it was for known cancer, or if it was undertaken with palatal surgery. Data were prospectively collected on patient characteristics, the surgical operation, and subsequent complications. A primary postoperative hemorrhage was defined as any bleeding that led to delayed hospital discharge, blood transfusion, or return to theatre during the initial stay. A secondary hemorrhage was any postoperative bleed that occurred after discharge from hospital and led to readmission to hospital within 28 days of surgery. The audit was approved by the Northern and Yorkshire multicenter research ethics committee.
Received March 18, 2008; revised August 21, 2008; accepted August 21, 2008.
0194-5998/$36.00 © 2009 American Academy of Otolaryngology–Head and Neck Surgery Foundation. All rights reserved. doi:10.1016/j.otohns.2008.08.025
24
Otolaryngology–Head and Neck Surgery, Vol 140, No 1, January 2009
Tonsillectomies were grouped according to the techniques used for dissection and hemostasis. In this analysis, we focus on two of these categories: the first contained procedures in which bipolar diathermy was used for both dissection and hemostasis, while the second contained procedures in which bipolar diathermy was used for hemostasis after cold steel dissection (ie, the use of scissors and/or blunt dissector and snare). Surgeons recorded maximum power setting for both dissection and hemostasis, indicating both the power setting and the unit. During the analysis, it was found that 97 percent of all power settings were reported in watts, and so the analysis was restricted to these data. The hemorrhage rate and relative risks were calculated for specified ranges of power settings. The relative risks were calculated as ratios of the hemorrhage rates for each group, with the group with the lowest power setting being used as the reference. The groups were defined to reflect the distribution of the power setting values, which tended to take particular discrete values (8, 10, 12, 15, and 20 watts). Data were also grouped because there was no reason to suppose beforehand that the risk of hemorrhage increased linearly with every unit increase in power. Values higher than 50 watts and lower than 6 watts were judged to be unrealistic because they would have had either an excessive or insufficient dissection or coagulative effect. The data of these patients were therefore excluded. Procedures performed using diathermy throughout were grouped using the power setting for dissection only as there was a very high correlation between the values recorded for dissection and hemostasis (Pearson’s r ⫽ 0.97). Evidence of a relationship between the power categories and the unadjusted hemorrhage rates was assessed using the 2 test for trend.7 Multiple multilevel logistic regression8 was then used to adjust the risk estimates for potential confounding factors, namely, patient age and sex, indication for surgery, grade of operating surgeon, type of hospital (NHS or independent), and whether the procedure was performed before or after the guidance was published. A twostage analysis was performed. In the first stage, the regression model contained the power setting categories used in the unadjusted analysis. In the second stage, if justified, the regression model would include power setting as a continuous variable because this would more precisely estimate any dose-response relationship if the trend was approximately linear. Multilevel logistic regression was used to account for any lack of independence in the data due to patients being clustered within hospitals. All P values were two-sided, and those lower than 0.05 were judged to be statistically significant. STATA software (version 9.2) was used for all statistical calculations.
RESULTS The NPTA database contained 12,562 patients who had a tonsillectomy performed with bipolar diathermy dissection
and hemostasis and 11,956 patients who had cold steel dissection with bipolar diathermy hemostasis. Excluding patients with missing data or invalid power settings (see Table 1) meant that the analysis included 9572 (76%) of the diathermy dissection patients and 8465 (71%) of the cold steel dissection patients. There were slightly more female than male patients, and around 60 percent were under 16 years of age, with recurrent acute tonsillitis being the most common indication for surgery (Table 1). Over 85 percent of procedures were performed in NHS hospitals, and most patients were admitted as inpatients rather than day cases. Among the 9572 patients who underwent diathermy dissection and hemostasis, a total of 438 patients (4.6%) experienced a postoperative hemorrhage within 28 days. Specifically, these patients had 43 (0.4%) primary hemorrhages and 404 (4.2%) secondary hemorrhages (some patients had both a primary and secondary hemorrhage). The rate of hemorrhage increased with age and was noticeably lower for patients whose indication for surgery was an obstructive rather than infectious condition (Table 1). The rate of hemorrhage did not depend upon the power setting used for dissection (2 test for trend, P ⫽ 0.8), with the relative risks of the various groups not differing from the reference group by more than 20 percent (Table 2). The lack of association between hemorrhage rates and power setting remained after using logistic regression to adjust for possible confounders. Among the 8465 patients that had cold steel dissection and diathermy hemostasis, a hemorrhage was reported for 229 patients (2.8%). In total, there were 39 (0.5%) primary hemorrhages and 196 (2.3%) secondary hemorrhages. As before, the rate of hemorrhage increased with age and was lower for patients with an obstructive rather than infectious condition (Table 1). The rate of hemorrhage rose steadily as the power setting increased, from 1.8% in patients with the lowest power settings (6 to 8 watts) to 3.7% in those with settings above 18 watts (Table 2), which represents a doubling in risk (2 test for trend, P ⫽ 0.005). The relationship between diathermy power setting and the overall risk of hemorrhage reflects the risk of a patient experiencing a secondary hemorrhage (Table 3). The trend in secondary hemorrhage risk across the groups was statistically significant (2 test for trend, P ⬍ 0.001). The study did not have the statistical power to draw conclusions about the relationship between diathermy power setting and primary hemorrhage (2 test for trend, P ⫽ 0.1). Using multiple multilevel logistic regression, we observed that adjustment for confounding did not affect the relationship between the diathermy power setting and the overall risk of hemorrhage (Table 4). Consequently, the second model was used to estimate the change in risk associated with an increase of 1 watt of power. Because the evidence that the trend was approximately linear extended over the range of the categories, it was decided to winsor the power setting data.9 Winsorization prevented the highest values from having undue influence on the estimated slope and involved replacing the highest power setting values with a maximum
Lowe et al
Diathermy power settings as a risk factor for . . .
25
Table 1 Characteristics of patients undergoing tonsillectomy involving diathermy Bipolar diathermy for dissection and hemostasis Patients (%) Patients All patients in NPTA database Patients with no power settings or power settings not in watts Patients with invalid power values Patients included in analysis Sex Male Female Age group Up to 5 years 5-15 years 16 years and over Indication for surgery Recurrent acute tonsillitis Chronic tonsillitis Previous quinsy Pharyngeal obstruction Other Grade of operating surgeon Consultant Nontraining staff grade Specialty registrar Senior house officer
Hemorrhage rate
Cold steel dissection and bipolar diathermy hemostasis Patients (%)
12,562
11,956
2106 884 9572
2849 642 8465
Hemorrhage rate
3928 (41) 5581 (59)
4.8% 4.4%
3342 (40) 5073 (60)
2.9% 2.6%
1627 (17) 4534 (47) 3395 (36)
2.2% 3.7% 7.0%
1207 (14) 3969 (47) 3284 (39)
1.2% 2.3% 3.8%
7159 650 204 1147 163
(77) (7) (2) (12) (2)
4.9% 6.5% 4.9% 1.8% 3.1%
6601 620 182 695 178
(80) (7) (2) (8) (2)
2.9% 2.1% 5.5% 1.2% 1.7%
3925 1874 2532 1232
(41) (20) (26) (13)
3.7% 5.7% 4.6% 5.5%
2716 2561 1758 1418
(32) (30) (21) (17)
2.5% 2.6% 2.5% 3.5%
value (set at 23 watts). A total of 369 values (4.4%) were winsored. Using this model, it was estimated that the risk of a hemorrhage increased by a factor of 1.05 (95% CI for odds ratio ⫽ 1.01 to 1.10; P ⫽ 0.017) for an increase of 1 watt
in power, or equivalently by a factor of 1.67 (95% CI for odds ratio ⫽ 1.09 to 2.54) for an increase of 10 watts in power. This estimate produced similar odds ratios for the original power setting categories (Table 4).
Table 2 Rates of tonsillar hemorrhage by surgical technique for various diathermy power settings Power setting Diathermy for dissection and hemostasis* 6 to 8 watts 9 to 11 watts 12 to 14 watts 15 to 17 watts 18 watts plus Total Cold steel dissection and diathermy only for hemostatis 6 to 8 watts 9 to 11 watts 12 to 14 watts 15 to 17 watts 18 watts plus Total
Number of operations (%)
Events
Rate†
Relative risk† (95% CI)
1098 (11%) 2719 (28%) 1863 (19%) 2552 (27%) 1340 (14%) 9572
48 126 91 103 70 438
4.37% 4.63% 4.88% 4.04% 5.22% 4.58%
1.06 1.12 0.92 1.19
1 (0.77 (0.79 (0.66 (0.83
to to to to
1.47) 1.57) 1.29) 1.71)
735 (8%) 2025 (21%) 1516 (16%) 3037 (32%) 1152 (12%) 8465
13 48 37 88 43 229
1.77% 2.37% 2.44% 2.90% 3.73% 2.71%
1.34 1.38 1.64 2.11
1 (0.73 (0.74 (0.92 (1.24
to to to to
2.46) 2.56) 2.92) 3.90)
*Based on power setting specified for dissection. †2 test for trend P ⫽ 0.8 (diathermy for dissection and hemostasis) and P ⫽ 0.005 (diathermy only for hemostasis).
26
Otolaryngology–Head and Neck Surgery, Vol 140, No 1, January 2009
Table 3 Rates of primary and secondary hemorrhage among patients undergoing cold steel dissection and diathermy hemostasis* Primary hemorrhage Power setting (watts)
Secondary hemorrhage
Number of operations
Events
Rate†
Relative risk
Events
Rate†
Relative risk
735 2025 1516 3037 1209 8522
6 13 5 9 6 39
0.82% 0.64% 0.33% 0.30% 0.50% 0.46%
1 0.79 0.40 0.36 0.61
9 36 33 80 44 202
1.22% 1.78% 2.18% 2.63% 3.64% 2.37%
1 1.45 1.78 2.15 2.97
6 to 8 9 to 11 12 to 14 15 to 17 18 and over Total
*The sum of the primary and secondary hemorrhages for the various groups may exceed the total number of hemorrhages in Table 2 because some patients had both a primary and secondary hemorrhage. †2 test for trend P ⫽ 0.1 (primary hemorrhage) and P ⬍ 0.001 (secondary hemorrhage).
DISCUSSION The NPTA has previously demonstrated that the risk of postoperative hemorrhage after tonsillectomy was greater when bipolar diathermy was used for both tonsillar dissection and hemostasis compared to diathermy hemostasis alone.4-6 In this study, we found that there is also a “doseresponse” relationship between the risk of hemorrhage and the diathermy power setting when bipolar diathermy is used for hemostasis after cold steel dissection. Specifically, the rate of hemorrhage for patients with diathermy power settings above 18 watts was 3.7 percent, approximately twice the rate for patients with power settings between 6 and 8 watts (1.8%). Indeed, the risk of hemorrhage when power was above 18 watts approached the rate of hemorrhage observed when bipolar diathermy was used for dissection and hemostasis (4.6% overall). The latter could represent a threshold above which increasing amounts of diathermy power do not affect the risk of hemorrhage. This would explain why we did not find a relationship between diathermy power setting and hemorrhage when diathermy was used for both dissection and hemostasis.
The NPTA is the largest prospective cohort study to date on complications after tonsillectomy, collecting data on over 24,000 operations involving bipolar diathermy in England and Northern Ireland. The size of the study and the prospective data collection are particular strengths. But, being an observational study, it is susceptible to a number of biases. First, the results are based on procedures and outcomes reported by the participating hospitals. Incomplete data submission could have influenced the results if the probability of a patient not being captured by the study was related to the choice of technique, the power setting, or the other risk factors. We compared the audit data with data from an administrative database (called HES) that collects information on all acute episodes in English NHS hospitals. This showed that the audit included about 60 percent of the eligible patients and the case mix of patients in the audit was representative. The observed hemorrhage rates in the preand postguidance periods in the NPTA data were 4.2 percent and 2.9 percent, respectively, compared to the HES-derived pre- and postguidance hemorrhage rates of 6.2 percent and 5.9 percent, respectively.10 As this analysis was concerned primarily with
Table 4 Unadjusted and adjusted odds ratios of postoperative hemorrhage for tonsillectomy using diathermy for hemostasis
Power setting
Number of operations
Hemorrhage rate
Unadjusted OR
6 to 8 watts 9 to 11 watts 12 to 14 watts 15 to 17 watts 18 watts plus
735 2,025 1516 3037 1152
1.77% 2.37% 2.44% 2.90% 3.73%
1 1.35 1.39 1.66 2.15
Adjusted OR* (95% CI)
1.19 1.47 1.68 2.33
1 (0.6 (0.7 (0.8 (1.1
to to to to
2.4) 3.1) 3.3) 4.7)
OR estimated with power setting treated as continuous† 1 1.17 1.36 1.58 1.84
*Odds ratio (OR) adjusted for sex, age, type of hospital (NHS or independent), grade of operating surgeon, indication for surgery, and whether the procedure was performed before or after the NICE guidance was published. †Odds ratio for each power setting category based on the adjusted odds ratio for a unit increase in power (1 watt). The odds ratio was estimated as 1.0524 (95% CI 1.01 to 1.10; P ⫽ 0.017) for each 1-watt increase.
Lowe et al
Diathermy power settings as a risk factor for . . .
the differences in the relative risks and not in the risks themselves, this underreporting would influence the results only if hospitals submitted data selectively. There was no evidence to suggest selective reporting of hemorrhage among patients of similar characteristics. Another weakness of this study concerns the quality of the power settings data. Bias may result from the exclusion of patients with missing or invalid values. However, the characteristics of excluded patients, and their outcomes, were similar to those with power setting values. For the diathermy dissection group, the hemorrhage rates in patients with and without power values were 5.1 and 4.6, respectively (2 test P ⫽ 0.17). For the cold steel dissection group, the values were 2.72 percent and 2.71 percent, respectively (P ⫽ 0.96). The reported power setting values may have also differed from the actual power used. Some diathermy generators give only limited information on the power being delivered, and this may partially explain the grouping of the settings around particular values. In addition, the study did not collect data on how long diathermy was applied to a patient. However, any such errors are likely to cause the strength of the relationship between hemorrhage risk and diathermy power setting to be underestimated rather than overestimated due to an effect known as “regression dilution.”11 It might be argued that the use of diathermy for hemostasis after cold steel dissection was a response to excessive intraoperative bleeding when ties have already been used, and so might identify patients who were predisposed to intraoperative as well as postoperative hemorrhage. This seems an unlikely explanation because the audit received information on 4285 tonsillectomies using cold steel dissection and ties for hemostasis. Moreover, the grouping of power settings around specific values suggests that these were the routine settings of surgeons with a preference for diathermy12 rather than settings used in problematic situations. It might also be argued that the use of higher diathermy settings in hemostasis was caused by excessive intraoperative bleeding among patients predisposed to postoperative hemorrhage. We feel that this is unlikely because the adjusted analysis included indications for surgery and patient age, factors that are generally considered to be the best predictors of a particularly difficult procedure. The reason why higher power settings increase the risk of secondary hemorrhage is not clear. We suspect that higher diathermy power settings increase the area of tissue damage due to thermal injury, which results in a larger area of tissue necrosis within the tonsillar bed. It has also been suggested that the pathology of tissue injury differs between cold steel and diathermy dissection, with diathermy resulting in more damage.13 Greater tissue damage would be expected to increase pain in the first week after the operation, and this was observed in the only other study that we found to have examined how diathermy power may influence the outcomes after tonsillectomy. Cardozo et al14 enrolled 101 patients age over 13 years who had cold steel dissection followed by diathermy hemostasis, and measured
27
the amount of diathermy energy used by timing how long diathermy was applied at a preset power setting of 15 W. They reported a positive correlation between diathermy power and postoperative pain at five and 10 days after tonsillectomy (r ⫽ 0.55 and r ⫽ 0.44, respectively). They also reported that patients with any secondary hemorrhage had been exposed to higher median energy levels than those with no bleeding (1449 vs 1036 joules, respectively, P ⫽ 0.032). More generally, a systematic review of studies comparing hot vs cold techniques reported that all studies found higher pain scores among those patients undergoing hot techniques between four and 10 days postoperatively compared to patients undergoing cold tonsillectomy.15 The overall hemorrhage rates for the two techniques observed in the NPTA are similar to rates described in other recent studies on the relationship between postoperative hemorrhage and tonsillectomy technique. Data from the Welsh National Registry16 gave a hemorrhage rate of 2.1 percent for 141 patients undergoing cold steel dissection and bipolar diathermy hemostasis, and 9.8 percent for 325 patients undergoing bipolar diathermy dissection and hemostasis (relative risk ⫽ 4.6; 95 percent CI: 1.4 to 14.9). Several other large studies have reported a slight increase in the risk of hemorrhage for diathermy dissection compared to cold steel dissection, but the differences have not been statistically significant.13,17 A systematic review published in 2006 combined data from 24 studies16 and reported that a hemorrhage occurred in 5.3 percent of the 1676 patients undergoing bipolar diathermy dissection and hemostasis. However, among 983 patients undergoing cold steel dissection and bipolar diathermy hemostasis, the rate was 7.3 percent, giving a relative risk for diathermy dissection of 0.72 (95% CI: 0.54 to 0.98). The reason for the higher rate among the “cold” tonsillectomies was not clear and the authors of the review acknowledge that the results reflect differences in study design as well as the characteristics of the enrolled patients. The original guidance from NICE focused on the combination of cold and hot surgical techniques in terms of their use for dissection and hemostasis.3 It recommended that surgeons use as little diathermy as possible but it did not provide information to help determine what level of diathermy is acceptable. A survey of UK ENT surgeons conducted after the guidance was published reported that some surgeons responded by using cold steel rather than diathermy for dissection while continuing to use diathermy for hemostasis.18 Other surgeons began performing tonsillectomy with cold steel and ties, the technique found in the NPTA to have the lowest overall hemorrhage rate (1.7%). However, this carries some increased risk of primary/reactionary bleeding, probably related to the slipping of ligatures in the early postoperative period. This study is of importance because it provides more precise information on how diathermy use may be limited. The results show that the hemorrhage rate for cold steel dissection with diathermy hemostasis when performed at 6 to 8 watts is comparable to the hemorrhage rate observed for cold steel and ties (1.8%
28
Otolaryngology–Head and Neck Surgery, Vol 140, No 1, January 2009
vs 1.7%, respectively). The results therefore suggest that cold steel tonsillectomy with careful bipolar diathermy hemostasis at low settings is an equally effective method for minimizing postoperative hemorrhage.
and Public Safety (Northern Ireland). Jan van der Meulen is supported by a National Public Health Career Scientist Award, Department of Health– NHS R&D Programme UK.
REFERENCES ACKNOWLEDGEMENTS We thank the staff of all the hospitals in England and Northern Ireland who contributed data to the National Prospective Tonsillectomy Audit. This audit was carried out by the Comparative Audit Group of the British Association of Otorhinolaryngologists–Head and Neck Surgeons and the Clinical Effectiveness Unit of The Royal College of Surgeons of England– London School of Hygiene and Tropical Medicine.
AUTHOR INFORMATION From the Clinical Effectiveness Unit (Drs Lowe, Cromwell, Copley, and van der Meulen), The Royal College of Surgeons of England, London, England, UK; Public Health and Policy (Drs Cromwell and van der Meulen), London School of Hygiene and Tropical Medicine, London, England, UK; the Division of Community Based Sciences (Dr Lewsey), University of Glasgow, Glasgow, Scotland, UK; the ENT Department (Dr Brown), Milton Keynes General NHS Trust, Milton Keynes, Buckinghamshire, England, UK; and the Department of Otolaryngology (Dr Yung), Ipswich Hospital NHS Trust, Ipswich, Suffolk, UK. Corresponding author: Dr David Cromwell, Clinical Effectiveness Unit, The Royal College of Surgeons of England, 35-43 Lincoln’s Inn Fields, London WC2A 3PE United Kingdom. E-mail address: [email protected].
AUTHOR CONTRIBUTIONS David Lowe, coordinated the data collection; supported data analysis, and co-wrote first draft of paper; David A. Cromwell, performed data analysis and co-wrote first draft of paper; James D. Lewsey, supported data analysis, contributed to revision of article; Lynn P. Copley, coordinated data collection and supported data analysis, contributed to revision of article; Peter Brown, designed the audit and supervised its execution, contributed to revision of article; Matthew Yung, designed the audit and supervised its execution, contributed to revision of article; Jan H. van der Meulen, designed the audit and supervised its execution, contributed to revision of article.
FINANCIAL DISCLOSURES The National Prospective Tonsillectomy Audit was funded by the Department of Health (England) and the Department of Health, Social Services
1. Murty GE, Watson MG. Diathermy haemostasis at tonsillectomy: current practice—a survey of UK otolaryngologists. J Laryngol Otol 1990;104:549 –52. 2. Walner DL, Parker NP, Miller RP. Past and present instrument use in pediatric adenotonsillectomy. Otolaryngol Head Neck Surg 2007;137: 49 –53. 3. NICE. Interim guidance on the use of diathermy in tonsillectomy. Available at: http://www.nice.org.uk. Accessed: April 20, 2005. 4. Lowe D, van der Meulen J, National Prospective Tonsillectomy Audit. Tonsillectomy technique as a risk factor for postoperative haemorrhage. Lancet 2004;364:697–702. 5. Lowe D, van der Meulen J, Cromwell D, et al. Key messages from the National Prospective Tonsillectomy Audit. Laryngoscope 2007;117: 717–24. 6. National Prospective Tonsillectomy Audit: Final report. Available at: http://www.entuk.org/members/audits/tonsil/Tonsillectomyauditreport_ pdf. Accessed: December 10, 2007. 7. Altman DG. Practical statistics for medical research. London: Chapman and Hall; 1991. 8. Leyland AH, Goldstein H. Multilevel modelling of health statistics. Chichester: Wiley & Sons; 2001. 9. Hoaglin DC, Mosteller F, Tukey JW. Understanding robust and exploratory data analysis. Chichester: Wiley & Sons; 2000. 10. National Prospective Tonsillectomy Audit. Impact of NICE guidance on rates of haemorrhage after tonsillectomy: an evaluation of guidance issued during an ongoing national tonsillectomy audit. Qual Saf Health Care 2008;17:264 – 8. 11. MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart disease. Part 1, Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet 1990;335:765–74. 12. Myssiorek D, Alvi A. Post-tonsillectomy haemorrhage: an assessment of risk factors. Int J Pediatr Otorhinolaryngol 1996;37:35– 43. 13. O’Leary S, Vorrath J. Postoperative bleeding after diathermy and dissection tonsillectomy. Laryngoscope 2005;115:591– 4. 14. Cardozo AA, Hallikeri C, Lawrence H, et al. Teenage and adult tonsillectomy: dose-response relationship between diathermy energy used and morbidity. Clin Otolaryngol 2007;32:366 –71. 15. Leinbach RF, Markwell SJ, Colliver JA, et al. Hot versus cold tonsillectomy: a systematic review of the literature. Otolaryngol Head Neck Surg 2003;129:360 – 4. 16. Mowatt G, Cook JA, Fraser C, et al. Systematic review of the safety of electrosurgery for tonsillectomy. Clin Otolaryngol 2006;31:95–102. 17. Walker P, Gillies D. Post-tonsillectomy hemorrhage rates: Are they technique-dependent? Otolaryngol Head Neck Surg 2007;136(4 Suppl): S27–31. 18. Hopkins C, Ajulo P, Krywawych M, et al. The impact of the interim report of the National Prospective Tonsillectomy Audit. Clin Otolaryngol 2005;30:475– 6.