Tonsillectomy Versus Watchful Waiting for Recurrent Throat Infection: A Systematic Review
CONTEXT: The effectiveness of tonsillectomy or adenotonsillectomy (“tonsillectomy”) for recurrent throat infection compared with watchful waiting is uncertain.
OBJECTIVE: To compare sleep, cognitive, behavioral, and health outcomes of tonsillectomy versus watchful waiting in children with recurrent throat infections.
DATA SOURCES: MEDLINE, Embase, and the Cochrane Library.
STUDY SELECTION: Two investigators independently screened studies against predetermined criteria.
DATA EXTRACTION: One investigator extracted data with review by a second. Investigators independently assessed risk of bias and strength of evidence (SOE) and confidence in the estimate of effects.
RESULTS: Seven studies including children with ≥3 infections in the previous 1 to 3 years addressed this question. In studies reporting baseline data, number of infections/sore throats decreased from baseline in both groups, with greater decreases in sore throat days, clinician contacts, diagnosed group A streptococcal infections, and school absences in tonsillectomized children in the short term (<12 months). Quality of life was not markedly different between groups at any time point.
LIMITATIONS: Few studies fully categorized infection/sore throat severity; attrition was high.
CONCLUSIONS: Throat infections, utilization, and school absences improved in the first postsurgical year in tonsillectomized children versus children not receiving surgery. Benefits did not persist over time; longer-term outcomes are limited. SOE is moderate for reduction in short-term throat infections and insufficient for longer-term reduction. SOE is low for no difference in longer-term streptococcal infection reduction. SOE is low for utilization and missed school reduction in the short term, low for no difference in longer-term missed school, and low for no differences in quality of life.
- CI —
- confidence interval
- GAS —
- group A streptococcus
- RCT —
- randomized controlled trial
- SOE —
- strength of evidence
Tonsillectomy in children has 2 primary indications: recurrent tonsillitis and obstructive sleep apnea. Recurrent or severe tonsillitis has been defined as ≥7 episodes of sore throat in the preceding year, or ≥5 episodes in the each of the preceding 2 years, or ≥3 episodes in each of the preceding 3 years.1 No gold standard diagnostic test exists, however, to predictably attribute symptoms to tonsillitis. Often “tonsillitis” is used synonymously with sore throat or pharyngitis without diagnostic testing. However, the degree to which either of these terms reflects true tonsillitis is not known. Bacterial pharyngitis with group A streptococcus (GAS) can be diagnosed via rapid testing or culture. It is not possible, however, to determine whether the tonsil represents the infectious nidus or if the suspected pathogen represents normal bacterial flora for a particular child’s pharynx (colonization without negative health impact). Sore throat or pharyngitis may or may not have a tonsillar origin, and it is possible that many cases have alternative explanations. Nonetheless, most diagnoses of “tonsillitis” are made without supportive documentation or bacterial testing.2 In addition, frequency of infections as a metric of severity used to determine tonsillectomy eligibility1,3,4 is fraught with complexity related to diagnostic variability and inconsistent medical documentation. Thus, heterogeneity in diagnostic accuracy, establishment of severity, and frequency of infections complicates treatment decisions regarding tonsillectomy and the performance of comparative effectiveness of its treatments.1,3–5
Previous systematic reviews of tonsillectomy for recurrent throat infections have combined studies of children and adults and reported moderate reductions in sore throats in the first postoperative year with greater benefit associated with more severe baseline infections.6,7 We conducted a new systematic review that included all currently available comparative studies to examine published evidence regarding the effectiveness of tonsillectomy compared with forms of watchful waiting (which could have included supportive treatment with medications such as antibiotics or nasal steroids) for children with recurrent throat infections. This review is a component of an Agency for Healthcare Research and Quality–commissioned comparative effectiveness review of tonsillectomy in children conducted by the Vanderbilt Evidence Based Practice Center. The full comparative effectiveness review and review protocol (PROSPERO registry number: CRD42015025600) are available at http://www.effectivehealthcare.ahrq.gov.
Search Strategy and Study Selection
We searched the MEDLINE database via PubMed, Embase, and the Cochrane Library from January 1980 to June 2016 using a combination of controlled vocabulary and key terms related to tonsillectomy and throat infections (eg, “tonsillectomy,” “adenotonsillectomy,” “streptococcal”). We also hand-searched the reference lists of included articles and recent reviews addressing tonsillectomy in children to identify potentially relevant articles.
We developed inclusion criteria in consultation with an expert panel of clinicians and researchers (Table 1). We included comparative study designs (eg, randomized controlled trials [RCTs], prospective or retrospective cohort studies) and studies published in English. We did not include studies with high risk of bias as part of the evidence base.
Data Extraction and Analysis
One investigator extracted data regarding study design, descriptions of study populations and intervention and comparison groups, and baseline and outcome data using a standardized form. A second investigator independently verified the accuracy of the extraction and revised as needed. Principal outcomes of interest included the number and severity of recurrent throat infections, quality of life, and health care utilization (number of clinician visits or contacts, number of courses of antibiotics). Significant heterogeneity in outcomes reported precluded meta-analysis; thus, we synthesized studies qualitatively and report descriptive statistics in summary tables.
Assessment of Study Risk of Bias and Strength of Evidence
Two investigators independently evaluated the methodologic quality of studies using prespecified questions appropriate to each study design to assess risk of bias of RCTs and observational studies.16 Senior reviewers resolved discrepancies in risk of bias assessment, and we used an approach described in the full review17 to determine low, moderate, or high risk of bias ratings. Assessment of the strength of the evidence (SOE) reflects the confidence that we have in the stability of treatment effects in the face of future research.18 The degree of confidence that the observed effect of an intervention is unlikely to change in additional research, the SOE, is presented as insufficient, low, moderate, or high. Assessments are based on consideration of 5 domains: study limitations, consistency in direction of the effect, directness in measuring intended outcomes, precision of effect, and reporting bias. We determined the strength of evidence separately for major intervention-outcome pairs using a prespecified approach described in detail in the full review.17
Our searches (conducted for the broader systematic review17) identified 9608 citations, of which 7 (reported in multiple publications) had low or moderate risk of bias and met inclusion criteria (Fig 1).3,8–15,19 Two unique RCTs were reported in 1 publication,3 and 1 set of investigators reported RCT and nonrandomized trial results together in multiple publications.8–10 Another RCT was reported in multiple publications.12–14 Table 1 outlines study characteristics and risk of bias assessments.
Number and Severity of Throat Infections and Utilization
All of the included studies evaluated children with a history of mild to moderate throat infection (classified by number of infections reported, Table 1). Four RCTs3,8–10,12–14 and 1 retrospective cohort study11 included children with at least 3 throat infections in the 1 to 3 years before surgery and reported on frequency of recurrent throat infections and clinician visits after surgery versus no surgery. In all studies reporting baseline data, number of infections decreased from baseline in both groups, with significantly greater decreases in sore throat days and diagnosed GAS infections in children who received tonsillectomy versus no surgery/watchful waiting with supportive care in the short term (<12 months).
Reduction in Sore Throat Episodes
One RCT and 1 nonrandomized trial8–10 reported that children who underwent tonsillectomy had fewer recorded days of sore throat in a symptom diary than children who had medical management with tonsillectomized children experiencing 0.50 ± 0.43 sore throat episodes per year after surgery compared with nonsurgical group who had 0.64 ± 0.49 episodes over the same time frame (95% confidence interval [CI]: 0.16 to 0.80). In the RCTs’ intention-to-treat analysis, number of sore throat episodes decreased by 3.5 (95% CI: 1.8 to 5.2) over the full 2-year study period for tonsillectomized children, but sore throats per month did not decrease significantly (Table 2). These studies are limited by a strong preference for surgery among parents of children with more severe symptoms, thus affecting the generalizability of the patients who were randomized. Children who were randomized were only moderately affected by their symptoms. The study also had significant attrition in return of the symptom diaries over time and difficulty obtaining provider records for review.
In 2 additional RCTs, children in the surgical groups had fewer visits for sore throat after tonsillectomy, but the number of visits for sore throat in the watchful waiting groups was also low.20 In the first postsurgical year, the tonsillectomy group had 1.74 (95% CI: 1.54 to 2.00) episodes of sore throat or throat infection compared with 2.93 episodes (95% CI: 2.69 to 3.22) in the control group. Although statistically significant, it is unclear whether this difference is clinically meaningful.
In another RCT including both children with mild symptoms of throat infection or tonsillar hypertrophy (≥7 throat infections in the previous year or ≥5 in the previous 2 years or ≥3 in the previous 3 years and Brouillette obstructive sleep apnea score of <3.5—ie, in no apnea or possible apnea range), children who received tonsillectomy had fewer throat infections (defined as throat pain and fever) compared with those who had no surgery (0.56/person-year vs 0.77/person-year, P = not reported).12–14 Of note, many children originally allocated to no surgery/watchful waiting (n = 50 of 149) crossed over to the surgery arm. One retrospective cohort study reported that children who did not undergo tonsillectomy were 3.1 times (95% CI: 2.1 to 4.6, P < .001) more likely to test positive for GAS infection that those who underwent surgery.11 Children who did not have tonsillectomy also experienced GAS infections at a shorter time interval than the tonsillectomized children.
Another retrospective cohort study included children who had <3 throat infections in the previous year and used Read-coded data to identify provider visits.15 The study reported a net reduction in 3-year mean sore throat visits (which included visits for sore throat symptoms, tonsillitis, pharyngitis, and nonspecific upper respiratory infection) for children who underwent tonsillectomy compared with those who did not. This reduction decreased over time with 2.46 fewer visits (95% CI: 2.29 to 2.63, P < .001) in years 1 to 3 and 1.21 fewer visits (95% CI: 1.04 to 1.38, P < .001) in years 4 to 6, or 0.61 sore throat visits per child per year (over the 6-year study period). This study captured provider visits and is subject to inaccuracies in medical coding. Episodes of sore throat that did not lead to a provider visit or visits with multiple concerns, which may have included throat infection, and were coded under another primary complaint were not captured. Table 2 outlines key outcomes.
Quality of Life
Quality of life was not markedly different between groups at any time point in 3 studies reporting these data.8–10,12 Fewer days of missed school or work were associated with tonsillectomy in the short term, with differences diminishing over time in 2 RCTs,3 whereas another RCT noted comparable school absences between groups.12–14 Overall, comparative effectiveness assessment of tonsillectomy versus no surgery to improve number of throat infections, associated health care utilization, days of work/school missed, and quality of life shows a benefit in the first postsurgical year, with diminishing benefit over time.
Compared with no surgery, tonsillectomy reduced utilization (clinician contacts) and missed school/work in the short term. We have low confidence in this conclusion (low SOE). We have greater confidence that compared with no surgery, tonsillectomy reduced sore throats/throat infections or streptococcal infections in the short term (<12 months; moderate SOE). In the longer term (>12 months), we found no difference between groups in reduction of streptococcal infections (low SOE). We also found no significant differences between groups in missed school/work or quality of life in the long term (>12 months) and have low confidence in these conclusions (low SOE).
We could not make a conclusion about longer-term effects on sore throats/throat infections (insufficient SOE) because few studies reported longer-term data, and those that did had high attrition rates. Only 1 study included children with <3 episodes of throat infection in the year before surgery15; we could not make conclusions about outcomes reported in this single study (insufficient SOE). Table 3 outlines SOE ratings.
Overall, children undergoing tonsillectomy to improve number of sore throats/throat infections, associated health care utilization (clinician visits), and days of work/school missed had improvements in these outcomes in the first postsurgical year compared with children not receiving surgery.3,8,11,12,15 These benefits did not persist over time. Most studies in this review included children with a history of mild to moderate throat infection (at least 3 episodes of sore throat in the year before surgery), and benefits of tonsillectomy were moderate in this population. Studies including children with a well-documented greater number of infections, such as the seminal Paradise et al study, may provide evidence for more marked benefits but were excluded from this review due to high risk of bias (due to potential selection bias and high attrition).1
As noted, most evidence addresses short-term effects. Although studies consistently reported reductions in sore throats and health care utilization in the first postsurgical year, evidence about long-term benefits of tonsillectomy for throat infection is limited. Thus, individual decision-making needs to balance the benefits of reducing illness-related outcomes (including missing school and work) with the risks associated with surgery. Caregivers and providers may wish to consider the potential benefits and drawbacks of attempting to manage children’s illnesses for a period of time to see if they outgrow the propensity for infection to avoid surgery.
Despite substantial research, the literature is largely silent on the natural history of throat infections that would provide a basis for the need for tonsillectomy in the long term. More data are needed to describe the potential to outgrow frequent infections and population factors that may predict resolution.21–23 Long-term data are needed to enable caregivers to weigh the benefits of surgery versus the reality of managing their child’s condition as they wait for it to resolve, although obtaining longer-term data is difficult, as evidenced by the high rate of attrition in most studies with >6 months’ follow-up included in this review.
Future studies should also attempt to characterize patient populations completely, including severity of throat infections, such that applicability can be much more specifically described and potential candidates for surgery or watchful waiting identified. Indeed, the literature lacks a consistent, consensus definition of infection; defining infection consistently is critical for promoting synthesis of research in the area. Tonsillitis or “sore throat” may also include cases of entities such as periodic fever, aphthous stomatitis, pharyngitis, cervical adenitis syndrome (PFAPA); clear characterization of children in studies is key for understanding effects on subpopulations. Relatively few data exist regarding predictable factors contributing to recurrence of symptoms of throat infections after tonsillectomy for primary management. A better understanding of these factors would also allow for more specific patient selection.
This review may be limited by our inclusion of studies published in English only; however, we identified few non-English studies of apparent relevance in a preliminary scan. Patient populations were generally poorly characterized, and little information was available on first-line treatment attempts before surgery. Particularly in studies intended to assess effects of tonsillectomy on throat infections, parents of more severely affected children were noted to refuse randomization and crossover to surgery at high rates. Among those studies focused on throat infection that did characterize patients more fully, most had low numbers of reported infections, and few reported culture-confirmed bacterial infections. Studies also did not describe the medical interventions children who did not undergo tonsillectomy may have received. Given heterogeneity and limited reporting in the literature, we were unable to identify potential subgroups that may respond more favorably to tonsillectomy or to supportive care.
Long-term effects are limited in the literature base, particularly regarding outcomes that include growth and development and quality-of-life outcomes. Exploration of demographics of patient populations more likely to be refractory to initial management strategies is also limited. It appears clear that throat infections decline in children over time regardless of treatment group, but with high loss to follow-up, the relative contribution of this decline on apparent effectiveness is unknown. Definitions of severity have been assigned somewhat arbitrarily, based on best evidence or consensus approach, and until there is a robust literature surrounding natural history of throat infections in childhood, it may be difficult to achieve consistency in this description.
Tonsillectomy can produce short-term reduction in throat infections compared with no surgery in children with ≥3 throat infections in the previous 1 to 3 years. Children undergoing tonsillectomy to improve number of throat infections, associated health care utilization (clinician visits), and days of work/school missed had improvements in these outcomes in the first postsurgical year compared with children not receiving surgery. These benefits did not persist over time, and data on longer-term results are lacking. Quality of life improved regardless of surgery. Additional research to assess longer-term benefits of tonsillectomy compared with no surgery, as well as subgroups of children who may experience greater benefit, is needed to inform decision-making for caregivers and clinicians.
Dr Shanthi Krishnaswami, Ms Jessica Kimber, and Ms Katherine Worley contributed to the data extraction. We gratefully acknowledge the input of the full research team and of our Agency for Healthcare Research and Quality Task Order Officers and Associate Editor.
- Accepted November 15, 2016.
- Address correspondence to Sivakumar Chinnadurai, MD, MPH, Doctors’ Office Tower, 7th Floor, 2200 Children’s Way, Nashville, TN 37232-8605. E-mail:
FINANCIAL DISCLOSURE: All authors received funding for this project under contract HHSA290201500003I from the Agency for Healthcare Research and Quality, US Department of Health and Human Services. Salary support for Dr Francis came from grants K23DC013559 and L30DC012687 from the National Institute for Deafness and Communication Disorders of the National Institutes of Health. The other authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This project was supported by the Agency for Healthcare Research and Quality (Contract HHSA290201500003I). Dr Francis' salary provided in part by the National Institutes of Health. Funded by the National Institutes of Health (NIH).
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2016-3491.
- Paradise JL,
- Bluestone CD,
- Colborn DK,
- Bernard BS,
- Rockette HE,
- Kurs-Lasky M
- Baugh RF,
- Archer SM,
- Mitchell RB, et al; American Academy of Otolaryngology-Head and Neck Surgery Foundation
- Georgalas CC,
- Tolley NS,
- Narula PA
- Lock C,
- Wilson J,
- Steen N, et al
- Wilson JA,
- Steen IN,
- Lock CA, et al
- Lock C,
- Wilson J,
- Steen N, et al
- van Staaij BK,
- van den Akker EH,
- Rovers MM,
- Hordijk GJ,
- Hoes AW,
- Schilder AG
- Buskens E,
- van Staaij B,
- van den Akker J,
- Hoes AW,
- Schilder AG
- Agency for Healthcare Research and Quality
- Francis DO,
- Chinnadurai S,
- Sathe NA, et al
- Copyright © 2017 by the American Academy of Pediatrics