Incidence of adenoidectomy (A), tympanostomy (B), and tonsillectomy (C) operations in Finnish children by age and sex (Kaplan-Meier life-table analysis). The curves for female and male subjects are plotted separately. The number of individuals operated on and the number of all individuals are given in parentheses. The curves are drawn until age 25 years. The log rank tests for the difference between female and male subjects were χ2 = 0.409 (P= .32) for adenoidectomy, χ2 = 0.664 (P= .21) for tympanostomy, and χ2 = 0.928 (P = .17) for tonsillectomy.
Numbers of tonsillectomy and adenotonsillectomy operations at Helsinki University Central Hospital during 1997 and 1998. Each bar shows the number of operations performed on patients at a given age. The numbers of tonsillectomy operations (without concurrent adenoidectomy) performed on all patients are shown in Figure 2A. The numbers of tonsillectomy operations performed on female and male patients are shown in Figure 2B-C, respectively. The numbers of adenotonsillectomy operations (with concurrent adenoidectomy) performed on all patients are shown in Figure 2D, and the numbers of adenotonsillectomy operations performed on female and male patients are shown in Figure 2E-F, respectively. In each figure part the curve for the best fitting mixture of normal distributions is superimposed. Mixtures of normal distributions were fitted in sequence of 1, 2, and 3 normal distributions. A combination of 3 normal distributions adequately fitted the observed age distribution of all and female patients who had tonsillectomy or adenotonsillectomy (likelihood ratio test: Figure 2A, χ23= 28.7, P<.01; Figure 2B, χ23= 30.3, P<.01; Figure 2D, χ23= 19.7, P<.01; and Figure 2E, χ23= 19.6, P<.01. A combination of 2 normal distributions adequately fitted the observed age distribution of male patients who had tonsillectomy or adenotonsillectomy (likelihood ratio test: Figure 2C, χ23= 69.9, P<.01; and Figure 2F, χ23= 106.7, P<.01).
Numbers of tonsillectomy and adenotonsillectomy operations at Helsinki University Central Hospital during 1997 and 1998 as grouped by the indication for surgery. Each bar shows the number of operations performed on patients at a given age. The numbers of tonsillectomy and adenotonsillectomy operations performed because of hyperplasia are plotted in Figure 3A-D, respectively. The number of operations performed because of abscesses and other acute infections are plotted in Figure 3B-E, and the numbers of operations performed because of chronic tonsillitis are plotted in Figure 3C-F. In each figure part the curve for the best fitting mixture of normal distributions is superimposed. Mixtures of normal distributions were fitted in sequence of 1, 2, and 3 normal distributions. A combination of 3 normal distributions adequately fitted the observed age distribution of patients who had tonsillectomy because of abscesses and other acute infections likelihood ratio test: Figure 3B, χ23= 39.2, P<.01. A combination of 2 normal distributions adequately fitted the observed age distribution of the other patient groups (likelihood ratio test: Figure 3A, χ23= 64.0, P<.01; Figure 3C, χ23= 26.0, P<.01; Figure 3D, χ23= 107.8, P<.01; Figure 3E, χ23= 17.1, P<.01; and Figure 3F, χ23= 30.8, P<.01).
Mattila PS, Tahkokallio O, Tarkkanen J, Pitkäniemi J, Karvonen M, Tuomilehto J. Causes of Tonsillar Disease and Frequency of Tonsillectomy Operations. Arch Otolaryngol Head Neck Surg. 2001;127(1):37-44. doi:10.1001/archotol.127.1.37
To characterize the factors that influence the frequency of tonsillectomy and adenoidectomy operations.
Design and Setting
Nationwide questionnaire. Analysis of patients undergoing tonsillectomy or adenoidectomy at Helsinki University Central Hospital, Helsinki, Finland.
Four hundred eighty-three of 819 individuals randomly selected from the Finnish National Public Registry. Two thousand two hundred thirty-one individuals younger than 30 years who underwent tonsillectomy (888 patients), adenotonsillectomy (294 patients), or adenoidectomy (1049 patients) at Helsinki University Central Hospital from January 1, 1997, through December 31, 1998.
Main Outcome Measures
Age of the individual at the time of operation. Indication for the operation.
The frequency of adenoidectomies was 24% (116 persons) and that of tonsillectomies 8% (39 persons) among the 483 individuals who returned the questionnaire. The frequency of tonsillectomy operations by age was multimodal; the frequency of tonsillectomies increased in preschool-aged children, declined thereafter, and increased again in teenagers. Tonsillar hyperplasia was the most frequent among children younger than 10 years, peritonsillar abscesses among teenagers, and chronic tonsillitis among individuals older than 20 years. The proportion of females was higher than males among teenaged patients. However, the cause and sex distribution could not explain the multimodality in the age-specific frequency. The age-specific frequency of tonsillectomies performed because of peritonsillar abscesses still followed a multimodal distribution.
Factors relating to respiratory tract infections, maturation of the immune system, and the onset of puberty contribute to the cause of tonsillar disease. Distinct indications for tonsillectomy should be defined for preschool-aged children, teenagers, and individuals older than 20 years.
ADENOIDECTOMY AND tonsillectomy operations are frequently performed surgical procedures among children. The removal of the tonsils and the adenoids is considered to be safe with no known long-term immunological side effects. However, the plain presence of these organs in human beings suggests that they may have provided an evolutionary advantage to humans. In this regard knowledge of the epidemiology of tonsillar disease is important. First, such data may be used to evaluate the cause of adenotonsillar disease. Second, epidemiological data may be helpful in setting up clinical trials on the efficacy of adenoidectomy and tonsillectomy. Third, it may be used in planning further studies on the long-term effects and side-effects of adenoidectomy or tonsillectomy. We report frequent indications for adenoidectomy and tonsillectomy operations at different ages and evaluate the possible factors influencing the frequency of tonsillar disease.
A questionnaire was sent to 819 Finnish individuals in 1998. These individuals had previously served as a control population for children with diabetes mellitus and they had been selected on the basis of sex and date of birth from the Finnish National Population Registry matching 819 Finnish children with type 1 diabetes mellitus.1 The selection resulted in a population that was a random representative of Finnish children except for sex distribution (370 females, 449 males). The age of the individuals ranged from 10 to 27 years when the questionnaire was distributed. Altogether 483 (59%) of the 819 forms were returned.
The questionnaire included the following questions: (1) Have your adenoids been removed (adenoidectomy)? (2) If adenoidectomy was performed, at what age was it done? (3) What was the reason for adenoidectomy? (I) ear infections, (II) glue ear, (III) mouth breathing or snoring, (IV) sinusitis, or (V) other infections. (4) Have you had tympanostomy tubes inserted? (5) If tympanostomy was performed, at what age was it done? (6) Have your tonsils been removed (tonsillectomy?) (7) If tonsillectomy was performed, at what age was it done? The questions 1, 3 (I-V), 4, and 6 were followed by multiple-choice answer fields including yes/no, and do-not-know choices. The questionnaire included a picture with explanatory text showing the anatomical location of the adenoids and tonsils to help the respondent to better understand the questions. Data about the number of siblings, family income, location of the residence, and type of day care was obtained from results of another questionnaire sent to the same individuals during an earlier study.1
Data from the patient registry at the Helsinki University Central Hospital (HUCH) was obtained by searching the database for the International Classification of Diseases, 10th Revision (ICD-10) codes for tonsillectomy (EMB10, 888 operations), adenotonsillectomy (EMB20, 294 operations), and adenoidectomy (EMB30, 1049 operations) in the patient registry from January 1, 1997, through December 31, 1998. Patients who were younger than 30 years were included for the analysis. The indications for the operations were obtained from the ICD-10 code of the diagnosis of the patient when operated on.
Among the individuals who returned the questionnaire, the risk for adenoidectomy, tympanostomy, or tonsillectomy within a given period, the incidence proportion by age t, was estimated as a function of age:
where S(t) is the Kaplan-Meier estimate of the survival probability. Statistical significance of the difference between female and male subjects in the Kaplan-Meier estimate of the survival probability was tested by the log rank test.
The nature of the age distribution of subjects was studied by testing the goodness-of-fit and parameters for multimodal distribution models using maximum likelihood methods with the MIX software package.2 The statistical procedure involved stepwise testing of a model distribution of 1 standard distribution, subsequently a model distribution of 2 standard distributions, thereafter a model distribution of 3 standard distributions, and so on. Standard likelihood ratio test was performed comparing nested models to establish a model that fitted the data adequately with the smallest number of parameters. When there were too few observations the χ2 test was used to test the goodness-of-fit. A multimodal distribution that is a combination of 3 normal distributions has the following probability density function:
This represents a mixture of overlapping normal distributions of variable x (age at the operation), with means µ1, µ2, µ3, and SDs 𝛔1, 𝛔2, 𝛔3, and where α1 is the proportion of the subjects in the first component, α2 the proportion in the second, and (1 − a1 − α2) in the third. The statistical significances of the independences between variables in the tables of operation frequencies were tested using the χ2 test and the Fisher exact test.
The frequencies of adenoidectomy, tympanostomy, and tonsillectomy operations were estimated by a questionnaire sent to 819 individuals randomly selected from the Finnish National Population Registry. Among the individuals who returned the questionnaire, the frequency of adenoidectomy was 24% (116 persons); tympanostomy, 8% (39 persons); and tonsillectomy, 8% (39 persons) (Figure 1).
The frequency of tonsillectomies at distinct ages (age-specific frequency) appeared to depict a multimodal distribution (Figure 1C). The rate of operations increased between the ages of 4 and 8 years, decreased thereafter, and increased again between the ages of 13 and 23 years (Figure 1C). To further investigate the age-specific frequency of tonsillectomies, the patient registry at HUCH was analyzed for operations performed during 1997 and 1998. The distributions of the age-specific frequencies of tonsillectomies performed at HUCH also showed an apparent multimodality for both female and male subjects (Figure 2).
A statistical analysis was carried out to test multimodality of the distributions of the age-specific frequencies of operations performed at HUCH. The distributions of age-specific frequencies of both tonsillectomy and adenotonsillectomy operations fitted a 3-modal distribution in female subjects and a 2-modal distribution in male subjects (Figure 2).
The multimodal distribution of the age-specific frequencies of tonsillectomies suggested that the subjects could be grouped in 3 age groups corresponding to each of the 3 peaks of the multimodal distribution (<10 years, 10-19 years, 20-29 years). The proportions of female and male subjects undergoing operations in each of these age groups were then analyzed. Adenoidectomies were performed more frequently on male than on female subjects among children younger than 10 years (Table 1). Tonsillectomies and adenotonsillectomies were performed more frequently on female than male subjects among individuals aged between 10 and 20 years (Table 1).
The multimodal distribution of the age-specific frequencies of operations suggested that the age of the patient might be associated with a distinct cause of tonsillar disease. To facilitate the analysis, the indications for tonsillectomy and adenotonsillectomy were grouped in 3 categories. The first group of subjects had tonsillar or adenotonsillar hyperplasia (ICD-10 codes J35.1 and J35.3). This group of subjects included patients who were operated on because of obstructive symptoms such as mouth breathing and snoring but also patients with recurrent sinusitis or otitis media. The second group of subjects included patients who were operated on because of a severe acute tonsillar infection such as peritonsillar abscess (ICD-10 code J36), retropharyngeal abscess (ICD-10 code J39), neck lymphadenitis (ICD-10 code L04), acute tonsillitis (ICD-10 code J03.9), or mononucleosis (ICD-10 code B27). The third group of subjects had chronic tonsillitis (ICD-10 code J35.0) who suffered from recurrent episodes of tonsillitis and sore throat.
The indications for tonsillectomies (without concurrent adenoidectomy) were reviewed in each of the 3 age groups. The group of patients operated on because of abscesses and acute infections was the largest. The proportion of these patients was the highest in individuals aged 10 to 19 years (Table 2). The proportion of patients having tonsillar hyperplasia (ICD-10 code J35.1) decreased and the proportion of patients having chronic tonsillitis (ICD-10 code J35.0) increased with increasing age (Table 2). The indications for adenotonsillectomies (tonsillectomy with concurrent adenoidectomy) in each group resembled the indications of tonsillectomies, the proportion of patients having tonsillar or adenotonsillar hyperplasia (ICD-10 codes J35.1 or J35.3) decreased and the proportion of patients having chronic tonsillitis (ICD-10 code J35.0) increased with increasing age (Table 3).
When the age-specific proportions of females and males in each of the diagnosis group were analyzed, the results showed that female subjects were operated on for chronic tonsillitis more frequently than males in age groups 10 to 19 years (z = 2.17, P = .03) and 20 to 29 years (z = 2.858, P = .004, Table 4). Although tonsillectomies were apparently performed with similar frequencies on both female and male subjects aged 20 to 29 years (Table 2), females were operated on in this age group more frequently for chronic tonsillitis than for other reasons (χ2 = 12.37, P<.01, Table 4).
The distributions of the indications for tonsillectomies and adenotonsillectomies in each age group could only partly explain the multimodal distribution of the age-specific frequencies of tonsillectomies. The frequency of tonsillectomies performed because of peritonsillar abscesses and other acute infections still depicted an apparent multimodal distribution (Figure 3).
The major indications for adenoidectomy were otitis media, sinusitis, or obstructive symptoms related to adenoidal hyperplasia. Of the individuals returning the questionnaire, 109 gave information on the indication of adenoidectomy. Of these 83 (76%) had underlying otitis media or sinusitis. Otitis media was the indication for adenoidectomy, particularly in children younger than 4 years, for whom 79.2% otitis media was reported as the indication. The proportion of children undergoing adenoidectomy because of otitis media decreased with increasing age. In children aged 8 years or older this proportion was only 28.6% (χ2 = 16.09, P<.001, Table 5). A decreasing proportion of otitis media but an increasing proportion of sinusitis as the indication for adenoidectomy was also seen in patients operated on at HUCH.
The effect of the number of siblings, family income, site of residence, year of birth, and type of day care for children aged younger than 3 years on the frequency of adenoidectomy, tympanostomy, and tonsillectomy operations is given in Table 6. The frequency of tonsillectomies was low in children of families with a low annual family income (<15 200 Euros; P = .04, Fisher exact test). The frequency of the operations appeared to be higher in children who had attained day care at municipal day-care centers during the first 3 years of life. The number of children whose day-care status was known was, however, small and the observed difference did not reach statistical significance.
The frequencies of tonsillectomy operations at different ages followed a multimodal distribution, the first peak occurring around the age of 5 to 6 years, and the second around 14 to 16 years. Further, a third peak after the age of 20 years was observed in the age-specific frequencies of tonsillectomies performed on female subjects. In previous reports the frequencies of tonsillectomies have been reported on all individuals younger than 5 years, on all individuals aged from 5 to 9 years, and on all individuals aged from 10 to 14 years,3- 5 resulting in a possible failure to detect a relative decrease in the frequency of tonsillectomy operations performed on children aged between 9 to 11 years.
The cause of tonsillar disease was associated with the age of the patient in that tonsillar hyperplasia was the most frequent cause in children younger than 10 years, in teenagers the cause of tonsillar disease was most frequently related to abscesses and acute infections, whereas chronic tonsillitis was the most frequent in individuals older than 20 years. The temporal pattern of the cause of tonsillar disease starting from hyperplasia or abscesses and ending in chronic tonsillitis may represent a continuum of different manifestations of a single entity of a tonsillar disease. It may be that bacterial growth within a hyperplastic tonsil leads to the formation of abscesses in teenagers. Later on it is possible that previous formations of scar tissue prevent the development of large abscesses but the disease manifests as chronic tonsillitis. However, it is possible that small, clinically undetected abscesses may later lead to chronic tonsillitis.
An association of adult chronic tonsillitis and peritonsillar abscesses is also evident in that anaerobic bacteria have been implicated as pathogens in both diseases. Anaerobic bacteria such as Fusobacterium necrophorum, Fusobacterium nucleatum, Prevotella melaninogenica, Prevotella intermedia, Peptostreptococcus micros, and Actinomyces odontolyticus are frequently recovered from peritonsillar abscesses.6Fusobacterium, pigmented Prevotella, Porphyromonas, and Peptostreptococcus species can also be recovered in the tonsils of patients with chronic tonsillitis.7,8 In addition, patients with chronic tonsillitis have increased levels of antibodies against F nucleatum and P intermedia suggesting that these bacteria have a role in chronic tonsillitis.9
Although adenoidectomies performed on children younger than 10 years were more frequent among boys than girls, the frequencies of tonsillectomies and adenotonsillectomies were higher among female than male subjects aged between 10 and 19 years. The results are consistent with previous reports indicating that adenoidectomies are more frequent among boys than girls younger than 4 years,3- 5 and that the frequency of tonsillectomies among teenagers is higher in female than in male subjects.3- 5 The unequal sex distribution in the cause of tonsillar disease suggests that tonsillar disease is slightly different in the female than in the male patient.
The frequencies of adenoidectomy, tympanostomy, and tonsillectomy operations have usually been estimated by comparing hospital discharge records with the total child population.3,5,10- 13 For example, the frequency of adenotonsillectomy operations per 100 000 individuals younger than 15 years has been estimated to range from 303 to 787 in the United States.12 Although accurate comparisons between the cumulative frequencies of operations obtained from questionnaires and the frequencies of operations obtained from hospital discharge records cannot be made, it appears that the cumulative frequency of adenoidectomy operations is high in Finland.
Previous studies have shown large geographical variation in the frequencies of tonsillectomies.5,10,11,13- 18 Also temporal variations in the frequencies of adenoidectomies and tonsillectomies occur.3,5,12,19,20 The variations in the operation frequencies may reflect variations in the indications for the operations, in the availability of medical services, and in the incidence of upper respiratory tract infections.
The cumulative frequencies of adenoidectomy and tympanostomy operations in this article were close to those reported in a previous questionnaire study involving 1708 Finnish children in 119 randomly selected second grade classes.21 In our study the cumulative frequency of adenoidectomy among 9-year-old children was approximately 20% and that of tympanostomy 8%. The cumulative frequency of adenoidectomy operations in the previous study was 17% and that of tympanostomy operations 8.6% in a child population with a mean age of 8.7 years.21
Low family income was associated with a decreased rate of tonsillectomy operations but not with the rate of adenoidectomy or tympanostomy operations. A significant proportion of tonsillectomies were performed on children older than 10 years. It is, therefore, possible that older children from families with a low family income may not reach medical services as easily as other children. However, in this income class mothers are frequently not contributing to the family income and stay at home, which may, in turn, be associated with a decreased risk for tonsillar disease in children of these mothers.
The frequency of operations among children younger than 3 years who had day care at a municipal day-care center seemed to be higher than among those children who did not have day care at day-care centers. However, the number of children of which the type of day care was known was small and the difference in the rate of the operations in our study did not reach statistical significance. Day care at day care-centers has previously been associated with an increased frequency of adenoidectomy and tympanostomy tube operations in Finland.22 In another study day-care at day care centers was associated with an increased cumulative incidence of tympanostomy tube insertions (31%, 108/346) as compared with children who were taken care of at home (11%, 7/63).23 Increased frequency of operations performed on children receiving out-of-home day care may reflect increased exposure to pathogenic microbes at day-care centers.
The most frequent indication for adenoidectomy in children younger than 4 years was otitis media. In older children sinusitis was more prevalent. The development of nasal sinuses and the growth of the eustachian tube with increasing age of the child may result in the development of purulent infections into the nasal sinuses rather than into the middle ear cavity.
The grouping of individuals in 3 age groups according to sex and distinct distributions of the indications for the operations only partly explained the multimodal nature of the age-specific frequencies of tonsillectomy. The age-specific frequencies of tonsillectomies performed because of peritonsillar abscesses still followed a multimodal distribution. The observed multimodality may reflect potential differences in the causative factors leading to tonsillar disease in the 3 age groups. These factors may be environmental such as exposure to the causative agent of mononucleosis that in our material led to tonsillar pathology preferentially in teenagers. However, these factors may be intrinsic to normal development such as to endocrine phenomena associated with puberty that may contribute to tonsillar disease in teenagers and that may result to female preponderance among teenagers undergoing tonsillectomy.
Accepted for publication June 28, 2000.
This study received financial support from the Paulo Foundation, Helsinki, Finland.
Corresponding author and reprints: Petri S. Mattila, MD, Department of Otorhinolaryngology, Helsinki University Central Hospital, Haartmaninkatu 4 E, FIN-00290 Helsinki, Finland (e-mail: firstname.lastname@example.org).