Sleep and Behavior 24 Months After Early Tonsillectomy for Mild OSA: An RCT

This phase of the randomized study was undertaken at 2 of the 3 original Australian Tertiary Children’s Hospitals: Children’s Hospital at Westmead (CHW) and Queensland Children’s Hospital. The third center ended their involvement at the 12-month follow-up, according to the original protocol design. The protocol and 12-month outcomes are previously published.^4,5  The trial was registered with the Australian and New Zealand Clinical Trials Registry (registration number ACTRN12611000021976).

Centralized ethical approval was obtained from the Human Research Ethics Committee at CHW and at each site (Human Research Ethics Committee /14/Sydney Children's Hospital Network/332). Written informed consent was obtained from the parents and/or caregivers of the participating children. Regular teleconferences and/or face-to-face meetings were held among the investigators to ensure adherence to the protocol. The study was funded by government and philanthropic agencies, with no commercial support.

Details of eligibility, random assignment, and initial assessments are published.^4,5  Briefly, we recruited children aged 3 to 5 years at the time of referral to ear, nose, and throat (ENT) or sleep services. They were deemed eligible if they had a positive pediatric sleep questionnaire (PSQ) score (more than one-third of questions answered in the affirmative). Baseline testing included ENT review to ensure suitability for T/A, overnight polysomnography to exclude severe OSA, and a hearing test to ensure suitability for neurocognitive testing. Children with severe OSA (>10 obstructive events per hour) were excluded from the study and referred immediately for clinical management.

The Brief Intellectual Ability (BIA) score of the Woodcock-Johnson III Tests of Cognitive Abilities (WJ III)⁶  was our primary outcome of cognitive function because it can be applied throughout life, beyond 2 years of age, and is suitable for use across the age range of our study group when many children could not complete all the tests required to generate a general intellectual ability (GIA) score.

The BIA combines the comprehension-knowledge (verbal ability), fluid reasoning (thinking ability), and processing speed (efficiency in performing cognitive tasks) domains in a short, reliable measure of intelligence. The GIA includes more domains and is a composite of oral vocabulary, number series, verbal attention, letter-pattern matching, phonological processing, story recall, and visualization. The GIA combines tests that account for the largest variance in intellectual ability and does not vary much with age. However, the BIA score correlates well with the GIA and other measures of intellectual ability.

The psychologist performing the testing was blinded to the random assignment of the subjects, and parents were coached to avoid disclosure. We used the Australian adaptation Woodcock-Johnson III Normative Update WJ III software program (WJ III NU Compuscore and Profiles Program; Houghton, Mifflin and Harcourt) to generate test and functional values with output that included mean (SD), percentiles, z scores, and W-scores (task proficiency). The W-scale is a Rasch-derived equal-interval scale that is useful for reporting an individual’s growth over time in a skill, ability, or area of knowledge. An increase in a person’s W ability represents actual growth in the trait measured, as opposed to a scaled score that measures how a person compares to a reference group.

Parents completed questionnaires to evaluate their child’s executive function and behavior at baseline and at each follow-up evaluation. A parental stress index was also completed.⁷ 

To assess behavior, we used the Behavior Assessment Scale for Children 2 (BASC-2) and its parent rating scale (BASC-PRS).⁸  In the preschool age group (2–5 years), this includes 139 items in a 4-choice response format to provide clinical and adaptive measures of behavior.

To assess executive functioning, we used the Behavior Rating Inventory of Executive Function (BRIEF), including the preschool version option for younger children. This questionnaire has 63 items in 5 nonoverlapping scales, and all items were evaluated.⁹ 

When analyzing the BASC-2 and BRIEF questionnaires we used the T-scores (mean 50, SD 10) after excluding tests with high inconsistency scores.

Full overnight sleep studies (polysomnography) were performed in clinical pediatric sleep laboratories. Sleep staging used 4 electroencephalogram leads (central and occipital), bilateral electro-oculographic leads, and submental electromyogram. Respiratory channels included inductance plethysmography for chest wall and abdominal movement, along with surface electromyogram measures of diaphragm and abdominal muscle activity, and a pressure signal at the nose, using nasal prongs to monitor airflow. Oxyhemoglobin saturation was measured by pulse oximetry (arterial saturated oxygen) and carbon dioxide levels by using transcutaneous CO₂. Cardiac rhythm was monitored with standard electrocardiogram leads. A maximum of 2 people analyzed polysomnography data at each site, and they were blinded to the random assignment of the subjects. Analysis criteria were derived from the American Academy of Sleep Medicine 2007 guidelines and Australasian Sleep Association commentary, which were current when the study commenced.^10,11 

Obstructive apnea was defined as a cessation of airflow for at least 2 respiratory cycles. Hypopnea was defined as a reduction in airflow resulting in either an arousal or oxyhemoglobin desaturation of at least 3%. Central apnea was defined as cessation of airflow and respiratory effort for at least 20 seconds or if <20 seconds in duration but associated with an arousal or oxyhemoglobin desaturation of at least 3%. Total sleep time was evaluated in minutes and apnea indices as episodes per hour of sleep. The obstructive apnea-hypopnea index (OAHI) is the number of obstructive apneas, mixed apneas, and obstructive hypopneas per hour of sleep, whereas the apnea-hypopnea index (AHI) includes central events.

At 2-month intervals after random assignment, caregivers were contacted by telephone and asked 7 standard questions, which are detailed in Supplemental Table 3. Answers to these questions were recorded and analyzed.

All children in this study (both the eT/A and T/A₁₂ groups) underwent complete extracapsular tonsillectomy and adenoidectomy, regardless of the size of the adenoids, with the option of using diathermy. The adenoidal bed was visualized at the end of the procedure to ensure that removal was satisfactory. Adverse events are listed in the Supplemental Information. Assessments at the time of surgery included tonsil size (graded 1–4 out of 4), weight of the tonsils, assessment of the palate, and of the degree of choanal obstruction (normal = 1, partial = 2, or complete = 3).

Two-sample t tests and χ² tests were used to compare characteristics between participants who were lost to follow-up before 12 months and those who remained in the study until at least 12 months (see Supplemental Information).

Continuous and binary outcomes at 12 and 24 months were analyzed by using generalized estimating equations assuming normal and binomial distributions, respectively. Predictors in these models were randomized treatment, time (12 and 24 months) and their interaction, and baseline value of the outcome variable. If there was no evidence of interaction, then a model with main effects of treatment, time, and baseline value was fitted. These models account for the correlation between repeated observations in the same study participant at the 2 times. If there was evidence of interaction, then separate models at each time were fitted, with randomized treatment and baseline values as predictors.

Participant demographics are shown in Table 1, with the average age of children being 75.8 ± 10.3 months (6.3 years) at 24-month follow-up and no difference between groups. We evaluated 250 children with polysomnography and 220 children with WJ III before recruiting and randomly assigning 190 children into the study. At the 12-month follow-up, we evaluated 157 children, and at this 24-month follow-up, we evaluated 131 (83.4% of those at 12-month evaluation) (Fig 1). No differences were found between children included compared with those who were not followed-up at 24 months (Supplemental Table 4).

A summary of the assessment outcomes is shown in Table 2. Details of the parameters measured and outcomes according to group and evaluation point are shown in Table 2.

In the 24-month analysis, 3 patterns of change were seen:

1. factors that improved in the eT/A group at 12 months and showed sustained improvement at 24 months compared with the T/A₁₂ group;

2. factors that improved with surgery, regardless of the timing of surgery; and

3. factors that improved temporarily or showed an enhanced rate of improvement after surgery in the eT/A group.

Sustained improvement was seen in the eT/A group for the BASC-2 Somatization scale and the arousal indices on polysomnography. For BASC-2 Somatization, P < .003 at 12 months and P = .02 at 24 months. For polysomnography total arousal indices, P < .001 at 12 months and P = .06 at 24 months. For polysomnography arousal indices in nonrapid eye movement (NREM) sleep, P = .06 at 12 and at 24 months.

Improvement after surgery for both groups was seen for the majority of the parameters we measured, regardless of the timing of surgery (Table 2). These included the PSQ total score and behavior and sleep study parameters. For behavior, the BASC-2 behavioral symptoms score was better at 12 months in the eT/A group (P = .03), and although the T/A₁₂ group improved at 24 months, the interaction with time was not significant (P = .11). Both groups’ BASC-2 Internalization (including sex specific) and Somatization subscale scores improved after surgery, although this remained slightly higher in the T/A₁₂ group at 24 months. Indices that improved on polysomnography included the arousal index in rapid eye movement (REM), the central apnea index, the obstructive apnea index (total, REM, and NREM indices), and the AHIs (total, REM, and NREM indices).

Parents reported improvements after surgery in all rating questions asked, including the overall child rating (out of 10). These included the history of snoring, reports of trouble sleeping, sleeping during the day (Fig 2), having trouble breathing when asleep, and the score for eating well. Parents were asked if there was “a change in sleep,” and both groups reported an initial change after T/A.

There was a temporary enhancement in the rate of improvement after surgery in the eT/A group for the WJ III subscores of sound blending (P = .05) and incomplete words (P = .03), with a similar pattern for the BASC-2 withdrawal score (P = .02) (Fig 3).

Twelve-month outcomes (previously published)⁵  revealed improvement in the WJ III subscore of long-term retrieval and BASC-2 anxiety scores for the eT/A group. In our current analysis, long-term retrieval (memory) on WJ III was not different, with no interaction between study track (eT/A versus T/A₁₂) and time (P = .22) (Fig 4). At 12 months, the eT/A group revealed some improvement in the BASC-2 anxiety score (P = .07 compared with T/A₁₂), but no difference was seen between groups at 24 months.

Additional details of the analyses of individual BASC-2 and BRIEF responses, and WHII GIA and BIA scores are shown in Supplemental Table 3.

This study adds several findings to the literature regarding improvements in preschool-aged children after T/A: It is the only randomized study to measure cognitive and behavioral outcomes of T/A exclusively in the preschool age group. Also unique to this study is that all participants underwent T/A regardless of their OAHI and subsequent follow-up. This review allowed early (12-month) and late (24-month) comparison of outcomes for the intervention group (eT/A) against the T/A₁₂ group. To our knowledge, this is also the longest follow-up data period for these outcomes in preschool-aged children undergoing T/A.

Total intellectual ability scores were not different between groups at baseline or at follow-up. Consistent with the fact that we studied children without comorbidities, mean intellectual ability scores for both groups were higher than average. However, the WJ III⁶  cognitive subscore for long-term memory showed improvement in both groups after surgery; that is, we demonstrated improvement for the eT/A group at 12-month follow-up, but at 24-month follow-up, the groups’ scores were equivalent. Thus, all children had improved long-term memory scores 12 months after T/A, regardless of the timing of that surgery. Different statistical analyses for the primary analysis at 12 months led to subscale differences between reports. Studies now support both structural and functional changes relating to hippocampal function in association with OSA in children^12,13  as well as adults, in which pathologic studies reveal hippocampal abnormalities¹⁴  and some reversal after 3 months of treatment.¹⁵  Cerebral blood flow has also been demonstrated to be increased in children with sleep disordered breathing, with a subsequent reduction seen after treatment.^16,17  Further studies in this area are required to confirm that recovery of brain injury does occur after treatment of SDB.

After surgery, improvements appeared to be briefly enhanced for cognitive tests that include auditory elements (sound blending and incomplete words). At each follow-up, hearing was assessed in these children to ensure that changes seen were not attributable to hearing deficits.¹⁸ 

In the T/A₁₂ group, followed without intervention until their 12-month follow-up, a proportion had lower apnea indices (obstructive and total) on polysomnography. Nonetheless, for the group as a whole, the polysomnography indices did not improve (Table 2). Of those with an OAHI of >1.0/hour, 22 of 40 (55%) fell to <1.0/hour on their 12-month polysomnography. A similar pattern was seen in the control group of the CHAT study.¹⁹  However, 4 children (10% of those with an AHI of >1.0 at baseline) had worsening of disease over the same time frame. Regardless of this pattern of spontaneous improvement, once surgery had been undertaken, there were further improvements in polysomnography parameters. Despite the polysomnography results, parents reported persisting symptoms before surgery in this group and also reported clear improvement after T/A. This supports the concept that symptomatic children, even with an AHI of <1, may have periods of continuous airway obstruction that contribute to symptoms not reflected in the OAHI.²⁰ 

We conclude that improved sleep quality overnight permits the cessation of daytime napping after surgery. The improvement in sleep and breathing parameters seen on polysomnography postsurgery likely explains the finding of cessation in daytime napping seen in both groups. Although they were a year older at the time of surgery, a greater proportion of the T/A₁₂ group than the eT/A group continued to nap before surgery, and these children stopped daytime naps after T/A: a response to surgery that was immediate and sustained. Although napping has been associated with improved daytime performance in preschoolers,²¹  our results suggest that after T/A, the night sleep improved sufficiently to end the need for additional day napping because total sleep time tends not to differ among groups who do, or do not, nap at this age.²² 

Regardless of the timing of surgery, parents report improvements in their children’s behavior after surgery. From the earliest reports about T/A for SDB in children, changes in behavior have been consistently reported.^23,24  This study adds to existing evidence by demonstrating that preschool-aged children with mild OSA on polysomnography experience behavioral difficulties, which improve posttreatment, and that this improvement is sustained 24 months after surgery. The association of these symptoms with even mild but symptomatic OSA raises concern that, especially in children, polysomnography metrics do not accurately reflect the underlying cause of daytime symptoms because they do not capture periods of partial airway obstruction that occur through the night without discrete respiratory events.^25,26 

The major limitation of this study is the reduction in participants over the 24-month follow-up period. This difficulty with follow-up of participants is consistent with any longitudinal randomized controlled study, especially one in which parents are likely to have a treatment preference that affects recruitment and retention in the trial.²⁷  Responses in this study suggested that parental preference was for children to have eT/A. The groups available for analysis at 24 months were also reduced by having only 2 of the original 3 centers continuing to participate in the study. However, the center that discontinued had the least number of participants (10% of those originally randomly assigned). Additional comparisons of participants from the 3 centers are included in the supplement. Participation in this study required considerable commitment, and for parents who had other young children, this contributed to difficulties experienced when attending the follow-up appointments.

Regarding the primary study outcome of intellectual ability, we found no differences in overall scores between the eT/A and T/A₁₂ groups, including no difference in long-term memory, regardless of the timing of their surgery. OSA severity on polysomnography revealed some spontaneous improvement in the T/A₁₂ group before their T/A, but after surgery, there were further improvements in polysomnography parameters. Sleep quality improved after surgery, with supportive evidence given by cessation of daytime napping, even in the T/A₁₂ group after surgery. Parents reported significant additional behavioral improvements after surgery in the T/A₁₂ group.

The study provides support for T/A producing improvements in neurocognition, behavior, and sleep of otherwise normal children who are symptomatic for OSA, with polysomnography-proven mild to moderate OSA. Timing of surgery did not affect responses to treatment in preschool-aged children.

We thank the families of the participants in this study and the Australian Sleep Research Network for their support toward the protocol development.

AHI

apnea-hypopnea index

BASC-2

Behavior Assessment Scale for Children 2

BIA

Brief Intellectual Ability

BRIEF

Behavior Rating Inventory of Executive Function

CHW

Children’s Hospital at Westmead

eT/A

early adenotonsillectomy

ENT

ear, nose, and throat

GIA

general intellectual ability

NREM

nonrapid eye movement

OAHI

obstructive apnea-hypopnea index

OSA

obstructive sleep apnea

POSTA

Preschool Obstructive Sleep Apnea Tonsillectomy and Adenoidectomy

PSQ

pediatric sleep questionnaire

REM

rapid eye movement

T/A

adenotonsillectomy

T/A₁₂

adenotonsillectomy after the 12-month follow-up assessment

WJ III

Woodcock-Johnson III Tests of Cognitive Abilities