Discussion
The unique characteristic of our study is that it focuses entirely on neonates (aged <1 month). In this retrospective study, we found that SDB can be a serious issue in high-risk neonatal population, occurring in almost 91% (73/80) of the study population. The majority had moderate to severe SDB. OSA was the most predominant SDB seen in our study cohort. This study demonstrates how medically complex these neonates are given the presence of spectrum of risk factors and comorbidities (table 2). However, with involvement of multidisciplinary specialist teams, these neonates were able to be managed appropriately.
The most common presenting symptoms were increased respiratory work of breathing and unexplained apnoea or desaturation highlighting that neonates do not present with the typical symptoms of OSA in older infants and children with snoring and restless sleep. Some infants were referred after a Brief Resolved Unexplained Episode (BRUE). Literature suggests that an apparent life-threatening event (ALTE) or BRUE may be the first clinical sign of SDB and these infants should be further evaluated .18 19 Guilleminault et al observed that ALTE/BRUE were more prevalent in children with micrognathia, retrognathia and anatomic anomalies that increase the risk of SDB.20 About 9% of the neonates in our study were evaluated due to family history of sudden infant death syndrome (SIDS). In a large cohort study from Denmark,21 a fourfold higher risk of SIDS was observed among siblings of children who died of SIDS. Other studies also support investigating siblings.22 Almost a third of our study cohort was born preterm, which is in line with the literature suggesting that premature infants may be at higher risk for SDB secondary to anatomical and physiologic variances.23 24
While tonsil and adenoid hypertrophy are the main cause of OSA in older children, literature suggests anomalies of upper airway anatomy, and craniofacial malformations are the main underlying factors in neonates,6 11 12 which is in keeping with the comorbidities seen in our study. Several genetic syndromes were found to be associated with SDB in our study, similar to what is described in the literature.6 11 12 These syndromes most likely either lead to increased upper airway obstruction (large tongue, small midface and/or small jaw) or cause hypotonia predisposing neonates to hypoventilation and SDB. Our data emphasise the need to be proactive in neonates with these conditions to the possible diagnosis of SDB. Overall, very high incidence of SDB in our high-risk cohort is supported by the literature showing that children with craniofacial anomalies had an estimated prevalence of SDB in up to 67% patients.25
This study underlines the need for multidisciplinary specialist teams in the management of SDB in neonates. In our study, multiple specialists were involved in the care of each patient with sleep physicians, otolaryngologists and geneticists the most frequently involved. Kaditis et al underlined the importance of multidisciplinary approach for management of SDB in children aged 1–23 months.26
PSG was done in majority (84%) of our study cohort, which is the gold standard to diagnose SDB. Diagnosing SDB based on neonatal PSG remains challenging as normative data are not well established and cut-off levels for various parameters in PSG to categorise pathologically significant SDB in neonates are still being defined.14 A recent study from Daftary et al3 showed that healthy newborns have much higher mean AHI (14.9±9.1), OAI (2.3±2.5) and CAI (5.4±6.2) compared with older children. Our study cohort had significantly elevated AHI and OAI, confirming the severity of SDB in our cohort. AHI is calculated by adding central and obstructive apnoea and hypopnoea. The high mean AHI (62.5) of our study cohort was mainly contributed by the OAI, which was significantly higher (38.7±34) than the expected range described in the literature.2 3 27 Although mean CAI of our study cohort was slightly higher than Daftary et al’s study,3 it was within the expected range for healthy neonates as per the review by Ng and Chan.2 It is well known that SDB can be a mixture of obstructive or central apnoea, especially in high-risk neonates with certain comorbidities. However, based on the PSG results, it is clear that the most predominant type of SDB in our study cohort was OSA and in majority it was moderate–severe as reflected by very high mean OAI. We decided to choose OAI as a marker of severity of OSA because its scoring criteria is clearly defined by AASM while scoring hypopnoea is difficult and categorising central versus obstructive hypopnoea is even more difficult. Therefore, OAHI can be difficult to replicate with possibility of significant interobserver variability. Despite the lack of universally accepted guideline for diagnosis and severity criteria for SDB in neonates,12 our data support the usefulness of PSG in medically complex neonates at an early age not only to diagnose SDB but also to guide management.
Thirteen (16%) neonates had compelling clinical signs/symptoms of OSA like increased work of breathing, high CO2 levels and oxygen desaturation. Therefore, treatment was initiated, and diagnosis was based on obvious clinical features and oxycapnography study. The international classification of sleep disorders also supports clinical diagnosis of OSA in paediatric age based on one or more symptoms.28 Previous studies in children have shown that oxycapnography alone is a reasonable screening tool in a diverse patient cohort with neuromuscular, congenital cardiac and genetic problems.29 30 However, its sensitivity depends on the severity of SDB; a normal oximetry does not rule out SDB and therefore does not replace PSG. Another limitation is its inability to differentiate between obstructive and central apnoea. A high prevalence of movement artefacts can make data interpretation challenging. A recent study showed good correlation between the desaturation index obtained from an overnight oximetry study to the OAHI obtained from PSG in young infants (<6 months) at risk of SDB.31 In our study, oxycapnography was only used in very severe cases who had obvious clinical symptoms and physician decided to initiate treatment over waiting for PSG date. We believe that oxycapnography can play an important role in screening for moderate to severe SDB in high-risk infants and help in treatment monitoring once diagnosis is established by PSG.
The clinical intervention was triggered for all but seven neonates (73/80, 91%) in our study based on very high mean AHI, OAI and obvious clinical features of SDB. Clinical factors, such as comorbidities, current symptoms and physical examination findings played equally important role in determining SDB diagnosis and management decisions.
Adeno-tonsillectomy is the first-line treatment in children with OSA, however establishing the specific cause of SDB is critical to selecting the optimal therapy in newborns. Ten per cent neonates in our study underwent surgery to relieve OSA. CPAP was prescribed to neonates with no indication of surgical treatment. All neonates underwent a pressure titration study during PSG to determine the optimal CPAP setting. Efficacy of CPAP was judged by improvement in clinical signs and normalisation of respiratory parameters, which was confirmed by follow-up oxycapnography and blood gases. Majority of our study cohort (70%) were successfully stablished, maintained and discharged on home CPAP therapy. CPAP is reported to be an effective treatment for upper airway obstruction in Pierre Robin sequence and can be used as a first-line treatment in selected patients.32 33 In a study of sleep apnoea in infants by Robinson et al,34 CPAP showed the highest decrease in AHI for all nonsurgical interventions used. Retrospective studies support the use of CPAP to treat SBD in infants with genetic and muscular problems.35 36 Midface hypoplasia has been described with the prolonged use of CPAP, but long-term outcome and potential reversibility of this finding are unknown.37 Long-term compliance and a high dropout rate is a major issue with CPAP use.38
Neonates with severe OSA, who did not respond to medical management by CPAP, were evaluated for surgical interventions, including lip–tongue adhesion, MDO and tracheostomy. MDO has been reported to relieve OSA in micrognathic infants and improve feeding and growth. Most of the studies have reported both short-term and long-term near-normalisation of OSA after MDO.39 40 In our study, only one patient underwent tracheostomy. Literature supports doing tracheostomy only as a last resort when other interventions fail.26
Other therapies included anti gastro-oesophageal reflux disease (GORD) treatment was used in 30% of our study population. Literature shows that GORD is commonly seen in infants with SDB, however, it remains unclear whether a causal relationship exists between the two.41
Limitations
Our study had several limitations including being a single-centre retrospective study. Being the major referral centre of the state, our cohort represents a highly biased population who were at high risk of SDB. This is a likely explanation for the very high rate of polysomnographic abnormalities among our study cohort. It is possible that infants with severe disease were selectively referred compared with those with milder SDB. Another limitation was that in number of cases diagnosis was based on oxycapnography study and clinical features. The major reason for this choice of diagnostic test in some cases were obvious signs of SDB and inability to undertake a full sleep study due to the medical fragility of the child while they were in a neonatal intensive care unit. We acknowledge that while oxycapnography has a high diagnostic positive predictive value, a reliable negative predictive value is lacking. The criteria for SDB severity based on AHI cutoffs vary in the literature for newborns and it is possible that we might have under or overestimated the burden of SDB. Another limitation was lack of repeat PSG data prior to discharge to confirm resolution of SDB postintervention. In our cohort, treatment was monitored by serial follow-up oxycapnography, blood gas parameters and clinical improvement of signs and symptoms. Our usual protocol is to do a follow-up PSG in 3–4 months postdischarge.