Introduction
Obstructive sleep apnoea (OSA) is a common childhood condition caused by a spectrum of factors which lead to upper airway obstruction during sleep.1 Adenotonsillar hypertrophy is a common cause and adenotonsillectomy is one of the most commonly performed surgical interventions for treatment of OSA in children.Reference Brouillette, Fernbach and Hunt2 Previous studies have reported growth failure to be more common in children with OSA.Reference Bar, Tarasiuk, Segev, Phillip and Tal3, Reference Vontetsianos, Davris, Christopoulos and Dacou-Voutetakis4 In addition, recent literature has indicated an association between adenotonsillectomy and post-operative weight gain, but so far no review articles have specifically investigated this phenomenon in children with OSA.Reference Jeyakumar, Fettman, Armbrecht and Mitchell5
Childhood obesity has increased on a worldwide scale for the past three decades, and current figures show that 28 per cent of children aged 2–15 years in Britain are either overweight or obese.Reference Wang and Lobstein6, 7 Established health complications as a result of childhood obesity are significantly detrimental, and include cardiovascular risks and type II diabetes. It is therefore important to ascertain whether adenotonsillectomy in this cohort puts patients at risk of further obesity.
This review aimed to evaluate the current evidence for weight gain in a paediatric population following adenotonsillectomy performed for OSA.
Materials and methods
A systematic literature search was conducted by two independent researchers using PubMed, Embase and Ovid databases. Articles published between 1995 and December 2014 that included the following key search terms were screened and selected: ‘growth’ or ‘weight’, ‘adenoidectomy’ or ‘adenotonsillectomy’, and ‘OSA’. In total, 784 articles were retrieved; the abstracts were reviewed after removal of duplicate articles. The full text of suitable papers was analysed and the references were screened for further articles missed in the primary search.
The inclusion criteria were: children with OSA, aged 0–14 years, who underwent adenotonsillectomy. The exclusion criteria were: studies with less than 30 patients; articles in a language other than English; animal studies; and investigations of children with craniofacial, cardiac, pulmonary or neurological disorders. The final articles selected were divided up and analysed in separate groups depending on the reported weight outcomes. No ethical approval was required for this study.
Results
An initial search yielded 475 unique articles; after the primary screening, this was narrowed down to 18 papers that fitted the inclusion criteria (Figure 1). Further screening of the full text articles led to the exclusion of 12 further articles as a result of: the inability to extract details regarding the effect of the intervention on weight (n = 1), fewer than 30 participants in the intervention arm (n = 1), the inclusion of children older than 14 years (n = 2), and the inclusion of patients who had either undergone tonsillectomy only or who had undergone surgery for indications other than OSA (mostly for recurrent tonsillitis) (n = 8).Reference Vontetsianos, Davris, Christopoulos and Dacou-Voutetakis4, Reference Amin, Anthony, Somers, Fenchel, McConnell and Jefferies8–Reference Wijga, Scholtens, Wieringa, Kerkhof, Gerritsen and Brunekreef18 The 6 remaining studies included a total of 729 children.Reference Katz, Moore, Rosen, Mitchell, Amin and Arens19–Reference Roemmich, Barkley, D'Andrea, Nikova, Rogol and Carskadon24 The demographics of the articles are summarised in Table I.
Preferred Reporting Items for Systematic Reviews and Meta-Analyses (‘PRISMA’) flow diagram for reviewing papers.
Research articles and study demographics
OSA = obstructive sleep apnoea; PSG = polysomnography; BMI = body mass index
Given the variation in weight outcome measures and follow-up times recorded in the various papers, a meta-analysis was not conducted. Five papers utilised pre- and post-operative body mass index (BMI) z scores as the outcome measure, and one paper recorded standardised percentile weight.
Five papers involving 675 patients measured the pre- and post-operative change in BMI z scores. The BMI z score allows for the correction of height and gender according to age. A z score of 0 denotes the mean BMI of the population and the z values either side of this represent the standard deviation from the mean. In these five studies, the patients were followed up for seven months to five years. Four of them were prospective studies and one was a randomised, controlled trial.Reference Katz, Moore, Rosen, Mitchell, Amin and Arens19–Reference Kang, Auo, Yoo, Cho and Kim23 The prospective studies had relatively few participants, ranging from 40 to 88 patients. The randomised, controlled trial, conducted by Katz et al., comprised 464 patients, who all had OSA and were eligible for adenotonsillectomy.Reference Katz, Moore, Rosen, Mitchell, Amin and Arens19 The patients were randomly assigned to either the intervention arm, which consisted of early adenotonsillectomy (performed within four weeks), or the control arm, where the patients were managed by watchful waiting. In that study, post-operative BMI z scores were significantly greater than the pre-operative z scores, even those for children who were already overweight prior to surgery compared to the control group.
Although all five studies saw an increase in BMI z scores, the difference was only statistically significant for the studies performed by Kang et al.,Reference Kang, Auo, Yoo, Cho and Kim23 Jabbari Moghaddam et al.Reference Jabbari Moghaddam, Golzari, Saboktakin, Seyedashrafi, Sabermarouf and Gavgani22 and Katz et al.Reference Katz, Moore, Rosen, Mitchell, Amin and Arens19 (Figure 2). The populations varied between the studies. For instance, Jabbari Moghaddam et al. studied an underweight cohort with an average BMI z score of less than −1 pre-operatively,Reference Jabbari Moghaddam, Golzari, Saboktakin, Seyedashrafi, Sabermarouf and Gavgani22 whereas the other studies were all within 1 standard deviation pre-operatively. After surgical intervention, the mean BMI of the children had increased in four of the studies to fall into the normal range.Reference Karalok, Akdag, Turhan, Uzun, Ozdem and Dinc20–Reference Kang, Auo, Yoo, Cho and Kim23 The study performed by Katz et al. saw an increase in the BMI z score that exceeded 1 after surgery, shifting the mean BMI to the overweight category post-operatively.Reference Katz, Moore, Rosen, Mitchell, Amin and Arens19
Pre- and post-operative body mass index (BMI) z scores (*p < 0.001). Pre-op = pre-operative; post-op = post-operative
Roemmich et al. recorded weight outcomes in terms of standardised weight percentiles and showed a statistically significant increase in weight on follow up at 12 months (Figure 3).Reference Roemmich, Barkley, D'Andrea, Nikova, Rogol and Carskadon24
Pre- and post-operative standardised weight percentiles (*p < 0.001). Pre-op = pre-operative; post-op = post-operative
Three studies confirmed OSA with the help of polysomnographyReference Katz, Moore, Rosen, Mitchell, Amin and Arens19, Reference Huang, Guilleminault, Lee, Lin and Hwang21, Reference Roemmich, Barkley, D'Andrea, Nikova, Rogol and Carskadon24 and the other three studies established the diagnosis on clinical grounds (Table I).Reference Karalok, Akdag, Turhan, Uzun, Ozdem and Dinc20, Reference Jabbari Moghaddam, Golzari, Saboktakin, Seyedashrafi, Sabermarouf and Gavgani22, Reference Kang, Auo, Yoo, Cho and Kim23
Discussion
Our review paper shows evidence for weight gain in the short term following adenotonsillectomy in children with OSA, regardless of their initial weight status. Most of the studies included in the review were observational studies, with relatively few patients. However, although a meta-analysis could not be performed, the study with the highest level of evidence – the randomised, controlled trial performed by Katz et al. – confirms weight gain, as reflected by an increase in BMI z score (of 0.31) post-operatively at seven months follow up.Reference Katz, Moore, Rosen, Mitchell, Amin and Arens19 There was also an increase in BMI z score in the control group (0.13), but the increase was significantly greater in the intervention group. In addition, subgroup analyses showed a statistically significant BMI z score increase in the intervention group children who were underweight and overweight at baseline compared to the control group. In contrast, the studies conducted by Karalok et al.Reference Karalok, Akdag, Turhan, Uzun, Ozdem and Dinc20 and Huang et al.,Reference Huang, Guilleminault, Lee, Lin and Hwang21 which had longer follow-up times, of 12 months and 3 years respectively, did not show a significant weight gain; therefore, we do not know if this effect is sustained in the long term.
Czechowicz et al. conducted a retrospective study of 815 patients who underwent adenotonsillectomy for OSA.Reference Czechowicz and Chang11 They found a significant increase in weight percentile at 18 months’ follow up compared to baseline (6.3 weight percentiles, p < 0.001). However, they did not detect an increase in obesity rate among their cohort. Instead, they found that mainly underweight and younger children gained weight, which accounted for the overall increase in weight percentiles. The study by Czechowicz et al. was retrospective, with no randomisation. In addition, the follow-up times varied, from 12 to 27 months, making direct comparison less reliable as the time factor was not consistent. This paper was excluded from the review as they included patients over the age of 14 years.
Further randomised, controlled trials with longer follow-up times are needed to ascertain whether the weight gain is only a transient phenomenon. Another limitation was the lack of standardised outcome measures. Many of the studies used their own national growth charts to calculate the BMI z scores, which could affect direct comparisons of results depending on when they were updated for their respective populations. None of the studies reported using the most recent World Health Organization growth charts from 2007, which represent projected healthy growth potential in children.25 Body mass index z score measurement enables reliable comparisons across studies; however, this measurement is subject to a ceiling effect. This means that children with higher BMIs who gain weight will show a smaller increase in BMI z score because of the smaller population distribution at the extremes of the bell curve.
There are several proposed explanations as to why adenotonsillectomy results in post-operative weight gain. The intermittent upper airway obstruction during sleep has been thought to increase the work of breathing in children with OSA, and therefore increase energy expenditure at night. However, a study by Bland et al. did not find a significant difference in the pre- and post-operative energy expenditure of children with OSA who were treated with adenotonsillectomy.Reference Bland, Bulgarelli, Ventham, Jackson, Reilly and Paton26 Enlarged adenoids and/or tonsils could lead to more frequent infections, swallowing problems and pain, which could reduce appetite and in turn reduce overall calorific intake.Reference De Serres, Derkay, Sie, Biavati, Jones and Tunkel27, Reference Mitchell, Kelly, Call and Yao28 Several studies have also shown a hormonal dysregulation in children with OSA, particularly those investigating insulin-like growth factor 1 (which reduces growth potential), and an increase in growth hormone secretion after adenotonsillectomy.Reference Kiris, Muderris, Celebi, Cankaya and Bercin14, Reference Nieminen, Lopponen, Tolonen, Lanning, Knip and Lopponen15, Reference Jabbari Moghaddam, Golzari, Saboktakin, Seyedashrafi, Sabermarouf and Gavgani22, Reference Kang, Auo, Yoo, Cho and Kim23 This can be explained by the fact that OSA causes sleep fragmentation and reduces the amount of slow-wave sleep, which is the phase when the growth hormone is primarily released.Reference Saaresranta and Polo29, Reference Issa and Sullivan30
The reasons for post-operative weight gain seem to be multifactorial. Nevertheless, regardless of the aetiology, current evidence shows that there is a significant quantifiable increase in weight gain for children who undergo adenotonsillectomy to treat OSA. When considering children who are underweight at the outset, this may be a positive result, allowing them to regain their normal weight potential. However, in the case of normal weight, overweight or obese children, we might consider weight monitoring and referral to appropriate pathways for nutritional counselling and obesity management. We might want to factor this into our pre- and post-operative consultations in order to manage these patients holistically.
Conclusion
The current body of evidence points towards a significant weight gain in the short term following adenotonsillectomy in children with OSA. Although this effect might be beneficial in children who are underweight at baseline, there may be a need to raise awareness of the likely effect of surgery in the pre-operative consultation, and to consider weight monitoring and nutritional counselling post-operatively for children who are at risk of further obesity.
