MEDICINE
ORIGINAL ARTICLE
Long-Term Results From Tonsillectomy in Adults Götz Senska, Halil Atay, Carolin Pütter, Philipp Dost
SUMMARY Background: Tonsillectomy is performed more than 400 000 times in the European Union each year, making it one of the most common operations. Nonetheless, there have been only a few long-term studies of quality of life after tonsillectomy. Methods: In 2004, data on the quality of life after tonsillectomy were obtained from adult German-speaking tonsillectomy patients by means of the Glasgow Benefit Inventory and a questionnaire specifically designed for that study. The present study concerns the further followup of these patients, sometimes many years later. 114 patients with recurrent tonsillitis were included in this descriptive study. Results: Of the 114 patients, 97 (85%) provided further data at 14 months, and 71 (62%) at ca. 7 years. The Glasgow Benefit Inventory revealed postoperative improvement of quality of life at 14 months and at 7 years, with median values of 16.67 points (quartile 11.11/25) and 13.89 points (quartile 8.33/25) (p = 0.168). The mean number of annual episodes of sore throat fell from 10 preoperatively to 2 postoperatively (p = 0.0001). The number of visits to the doctor, the intake of analgesic drugs and antibiotics, and the number of medical absences from work also declined significantly over the period of observation. Conclusion: Tonsillectomy was associated with a longlasting improvement of health and quality of life, and with lower utilization of medical resources. The 62% response rate at 7 years leaves the question open whether patients with a favorable postoperative course may have been more likely than others to participate in the study. ►Cite this as: Senska G, Atay H, Pütter C, Dost P: Long-term results from tonsillectomy in adults. Dtsch Arztebl Int 2015; 112: 849–55. DOI: 10.3238/arztebl.2015.0849
Department of Otorhinolaryngology, Head and Neck Surgery, Marienhospital Gelsenkirchen GmbH: Dr. med. Senska, Dr. med. Atay, Prof. Dr. med. Dost Institute for Medical Informatics, Biometry and Epidemiology, University of Duisburg-Essen: Dr. rer. medic. Pütter
Deutsches Ärzteblatt International | Dtsch Arztebl Int 2015; 112: 849–55
T
onsillectomy is one of the most common surgical interventions overall (1). However, the number of studies on the subjective success for the patient is significantly lower than those on the economic and clinical data. As demonstrated by recent examples, both tonsillectomy itself as well as the frequency of its implementation—particularly in Germany—are sources of extensive discussions and criticism, which has also been expressed in the public media (2–4). The large increase in publications is indicative of this trend. For instance, a MEDLINE search we carried out in 2008 for the terms “life quality AND tonsillectomy,” “benefit AND tonsillectomy,” and “economic AND tonsillectomy” resulted in only 11 relevant published studies that addressed the quality of life after tonsillectomy. Our same searches in 2014 resulted in more than 30 relevant publications, although most of those studies addressed the clinical results, as measured by objective postoperative results (5–10). In the meantime, however, more studies are available in which the patient’s subjective assessment has also been taken into account (11–14). Another aspect that has also been discussed in the literature concerns the short follow-up period: so far, statements about the postoperative success can only be made in terms of months to a few years. Studies in which a longer follow-up period was used are uncommon (14). As we had previously published a study on this subject in 2010 (15), we decided to interview the same patient collective again to examine whether the newlyobtained results were consistent. Thus, the aim of this descriptive study was to evaluate both the perceived changes in the quality of life after tonsillectomy as well as objective parameters, such as the number of physician visits and the use of resources, over a long term.
Methods An exploratory analysis was performed. A total of 114 consecutive patients were interviewed preoperatively, who then underwent elective tonsillectomy in the Marienhospital Gelsenkirchen in 2004. Inclusion criteria were: ● Adult (≥ 18 years old) ● Good knowledge of German ● At least three episodes of acute tonsillitis in the previous 12 months.
849
850
n (%)
N
– Female
– Total
n (%)
N
– Female
– Total
Wilcoxon signed rank test
Unpaired Wilcoxon test for sex differences
Mean ± SD (total)
n (%)
– Male
Sex
Work absences
Wilcoxon signed rank test
26
22 (26%)
4 (13%) 15
8 (10%)
7 (23%)
32
20 (24%) 56
42 (51%)
p = 0.51
11.9 ± 13.2
9
7 (8%)
2 (6%)
p = 0.16
1
1 (1%)
14 (46%)
p = 0.32
Unpaired Wilcoxon test for sex differences
N
– Total
48
32 (39%)
16 (51%)
3 (10%)
>10
8 (33%)
1–3
13
37
41
4
3 (4%)
1 (0%)
4 (16%)
71
19
51 (69%) 15 (21%)
20 (84%)
14 (45%) 22 (92%) 49
82
2
1 (1%)
1 (4%)
p = 0.19
1.4 ± 4.0
2
2 (3%)
0 (0%)
7 (39%)
1–3
27
31
0
0 (0%)
0 (0%)
2 (13%)
50
16
35 (70%) 14 (28%)
15 (94%)
5
5 (7%)
0 (0%)
61
45 (85%)
16 (88%)
2
1 (2%)
1 (5%)
Postoperative at 14 months to 7 years: p = 0.8396
5
4 (5%)
1 (4%)
p = 0.62
1.8 ± 5.5
20
2 (4%)
0 (0%)
p = 0.22
0.7 ± 1.4
5
4 (8%)
1 (6%)
p = 0.16
1.8 ± 2.3
10
9 (17%)
1 (6%)
4–6
13
1 (2%)
0 (0%)
0
0 (0%)
0 (0%)
3
2(4%)
1(6%)
7–10
Postoperative at 7 years (n = 71)
18 (34%) 24 (45%)
9 (50%)
None
Postoperative at 14 months to 7 years: p < 0.6045
2
2 (3%)
0 (0%)
Pre- to postoperative at 7 years: p < 0.001
Pre- to postoperative at 14 months: p < 0.001
15
11 (13%) 35 (42%) 60 (84%)
4 (13%)
p = 0.68
0.7 ± 1.6
5
5 (7%)
0 (0%)
3
3 (4%)
0
>10
Postoperative at 14 months to 7 years: p < 0.56
6
3 (4%)
Pre- to postoperative at 7 years: p < 0.001
Pre- to postoperative at 14 months: p < 0.001
21
17 (20%)
4 (12%)
p = 0.10
2 ± 2.9
10
9 (12%)
3 (13%)
7–10
Pre- to postoperative at 7 years: p < 0.0001
Pre- to postoperative at 14 months: p < 0.0001
32
1 (4%)
4–6
Postoperative at 14 months (n = 97)
12 (50%)
None
27 (32%) 10 (12%) 25 (34%) 33 (45%)
5 (16%)
7–10
9.7 ± 33 (max. 365)
12 (42%)
21
14 (25%)
7 (23%)
4–6
Preoperative (N = 114) 1–3
5 ± 2.7
n (%)
– Female
0 (0%)
0
0
0
None
Mean ± SD
n (%)
– Male
Sex
Physician Visits
Wilcoxon signed rank test
Unpaired Wilcoxon test for sex differences
Mean ± SD (total)
n (%)
– Male
Sex
Sore throat episodes
Number of results (subgroup)
Sore throat episodes, physician visits, and work absences in the year previous to survey
TABLE 1
4
4 (8%)
1 (5%)
0
0 (0%)
0 (0%)
0
0
0
>10
MEDICINE
Deutsches Ärzteblatt International | Dtsch Arztebl Int 2015; 112: 849–55
MEDICINE
TABLE 2 Number of participants who took analgesics or antibiotics, per year and in comparison Preoperative (N=114)
Postoperative 14 months (n=97) Postoperative 7 years (n=71)
Number of participants who took analgesics Sex – Male
n (%)
14 (45%)
1 (4%)
4 (22%)
– Female
n (%)
44 (53%)
6 (8%)
12 (23%)
– Total
N
Fisher's exact test
58
7
16
p = 0.24
p = 0.67
p=1
Pre- to postoperative at 14 months: P< 0.001
Wilcoxon signed rank test
Postoperative at 14 months to 7 years: p = 0.0127
Pre- to postoperative 7 years: P< 0.001 Number of participants who took antibiotics Sex – Male
n (%)
30 (97%)
3 (13%)
2 (11%)
– Female
n (%)
80 (96%)
18 (25%)
15 (29%)
– Total
N
110
21
17
Fisher's exact test Wilcoxon signed rank test
p=1
p = 0.26 Pre- to postoperative at 14 months: p < 0.001
p = 0.20 Postoperative at 14 months to 7 years: p = 0.8318
Pre- to postoperative 7 years: p < 0.001
Patients were referred to us for surgery by a specialist or primary care physician. Exclusion criteria were: ● Abscess tonsillectomy ● Tonsillectomy with neoplasia (suspected or proven) ● Exclusively tonsillar hyperplasia ● Tonsillectomy à chaud. Patients were first informed about the procedures, and their written consent for both the operation and participation in the study was obtained. A day prior to the operation, each patient was evaluated with a structured interview developed by the authors, which—in contrast to the Glasgow Benefit Inventory (GBI)—allows a controlled before-and-after comparison. The study participants were asked the same questions 14 months later, and—taking the arithmetic mean—once again seven years after surgery (in 2004 or 2005), using a—likewise standardized—telephone interview. In each case, the evaluation was based on the previous 12 months. Up to three telephone attempts were initially made to reach each subject. If this was unsuccessful, a letter was first sent to the address given during their interviews and then, after contacting the Einwohnermeldeamt (Residents’ Registration Office), up to two further letters were sent to their current addresses. Only then were the patients who had not been reached (n = 43) classified as nonresponders. Patients Deutsches Ärzteblatt International | Dtsch Arztebl Int 2015; 112: 849–55
were also additionally interviewed with the GBI at 14 months and at the second time of contact. The GBI was developed in 1996 to measure the benefits after otorhinolaryngologic interventions and can be used for various interventions (16, 17). The GBI can measure changes in health and quality of life after a surgical intervention; the survey tool has been validated for this purpose (18). Surgeries were performed under general anesthesia by surgeons with varying degrees of qualification, using the so-called “cold steel” technique (19, 20). In this technique, a relieving mucosal incision is made with scissors, and then the tonsils are removed from the tonsillar fossa with a periosteal elevator. The main complications and side effects of tonsillectomy (bleeding and pain) (21) were the subject of a separate study that has been partially published (22). This study was approved by the Ethics Committee of the Medical Association of Westfalen-Lippe.
Statistical methods Patient characteristics were represented by descriptive measures (median, quartiles) and frequency tables and, where necessary, were compared using simple univariate tests (Wilcoxon rank sum test and Fisher’s exact test). All resulting p values were indicated descriptively (without further adjustment). To test the pre- and
851
MEDICINE
Box plots showing the comparison of the Glasgow Benefit Inventory (GBI) scores postoperatively and after 7 years, for the scales “General perception of wellbeing,” “Social support,” and “Overall physical health”
FIGURE GBI Score 100
50
+ +
+
+
+
+ +
0
+
–50
–100
p = 0.1682
p = 0.1227
p = 0.4994
p = 0.5913
post-OP post-OP_7 post-OP post-OP_7 post-OP post-OP_7 post-OP post-OP_7 Total
postoperative differences (analgesics, antibiotics, sore throat episodes, physician visits, and work absences), the Wilcoxon signed rank test was used. In a multivariate analysis using the linear mixed model for longitudinal data, the effects of time, age, and sex on the number of sore throat episodes per year was examined. To analyze the robustness, the linear mixed model was used with different covariance structures for random effects (diagonal, unstructured, and compound symmetry with heterogeneous variances). For the pre/post comparison, the effect sizes were ≥ 0.25 with a sample size of N = 114, and ≥ 0.33 for n = 71. Effect sizes of ≥ 0.5 are commonly assessed as a medium effect.
Results In 2004, 467 patients who had undergone tonsillectomy were excluded from this study based on the criteria given above. However, 114 patients met the inclusion criteria. Of these, n = 97 (85%) were interviewed again at 14 months postoperatively, and n = 71 (62%), at 7 years (2004–2011). Of the participants interviewed postoperatively, 73 or 53 (75%) were women, and 24 or 18 (25%) were men, for n = 97 or n = 71, respectively. In comparison, the gender distribution among all adult patients operated on in 2004 was 59% women to 41% men. The youngest study participant at the time of surgery was 18 years old, and the oldest participant was 62 years old.
852
General
Social
Physical
The median age of the study participants in 2004 was 26 years (mean 28 years), which did not change in the two interviewed groups once corrected for the time past (i.e., 14 months or seven years). The average age at baseline did not differ between men (28.2 years) and women (27.9 years). Before surgery, the median number of tonsillitis or sore throat episodes was 6 per year (mean: 10); postoperatively, it dropped to 1 (mean: 2), as shown in Table 1. The lower number of postoperative sore throat episodes differed significantly from that of preoperative episodes (p = 0.0001). The multivariate analyses with the mixed model confirmed the decrease observed for the sore throat episodes per year over time (p12 months) worldwide, published in English or German N Start
n End
Follow-up (months)
Age (mean)
Survey
Results
Bhattacharyya 2002 (7)
293
83
37.7
Mui 1998 (8)
147
60
24
27.3
GBI
Improvement in quality of life
ND
Own survey
Improvement in quality of life
Wireklint 2012 (14)
75
68
72
25
SF 36
Improvement in quality of life
Schwentner 2007 (24)
600
227
38
30.1
GBI
Improvement in quality of life
Ovesen 2013 (12)
614
386
36
ND
Own survey
Improvement in quality of life
Our present study
114
71
84
26
GBI
Improvement in quality of life
GBI, Glasgow Benefit Inventory; ND, not determined
The median number of physician visits preoperatively was 4 (mean: 5), but at 14 months and at 7 years postoperatively, this was reduced to 0 (mean: 0.7). At the 14-month or 7-year time point, 71 (73%) or 50 (70%) of the participants, respectively, did not consult a primary care or ENT physician due to a sore throat (Table 1). The median number of lost work days related to tonsillitis or sore throat decreased from 10 days (mean: 11.9) preoperatively to 0 days (mean: 1.4/1.8) postoperatively (p