Improving the diagnosis of obstructive sleep apnea in children with nocturnal oximetry-based evaluations

KEYWORDS:

• Obstructive sleep apnea
• children
• oximetry
• sleep-disordered breathing
• diagnosis
• adolescents

1. Introduction

Sleep-disordered breathing (SDB) includes a spectrum of clinical entities that can be defined as a clinically relevant disturbance of nocturnal breathing patterns, and includes primary snoring, upper airway resistance syndrome, and obstructive sleep apnea (OSA). OSA is considered the end of the spectrum of SDB [1], and is characterized by intermittent cycles of upper airway collapse usually associated with intermittent desaturations and arousal during sleep. It is a prevalent disorder in childhood, affecting 2–3% of children [2]. Several factors can predispose children to OSA, including adenotonsillar hypertrophy, neuromuscular disorders, craniofacial abnormalities, and obesity. Consequently, certain subgroups of children for example children with obesity and Down syndrome exhibit a higher prevalence of OSA [1].

Several complications are associated with OSA in childhood, including cardiovascular, metabolic, and neurocognitive complications. Ultimately, these complications can diminish the general health and quality of life of these children. Therefore, a timely diagnosis of OSA is important for appropriate treatment. Furthermore, it is also of clinical importance to determine the severity of the disease since moderate-to severe OSA is less likely to resolve on its own, and treatment is essential [3].

Polysomnography (PSG) is the current gold standard for the diagnosis of OSA in children. During this investigation, sleep architecture, gas exchange, respiratory events, arousals, heart rate, and body position are measured. Although PSG is an indispensable diagnostic tool in sleep medicine, it is associated with several disadvantages. First, it is an expensive investigation that is time consuming and labor intensive. Furthermore, it is not universally available, especially in resource-constrained environments. The limited availability of sleep centers may also result in long waiting lists. As a result, a timely diagnosis of OSA is not always possible.

Nocturnal pulse oximetry has been proposed as an abbreviated and low-cost screening method for SDB both in adults and children. Brouillette et al. [4] were the first to investigate the feasibility of nocturnal pulse oximetry as a diagnostic tool for OSA. Since then, several studies around this topic have been initiated. In a recent systematic review by Kaditis et al., these studies on the use of nocturnal pulse oximetry for diagnosing OSA in children were discussed [5]. They concluded that overnight oximetry is useful for the diagnosis of OSA, as most studies in otherwise healthy children referred to a pediatric sleep clinic for suspected OSA had a good positive predictive value (range: 84–98%) and specificity (range: 60–99%). Moreover, it was concluded that oximetry can facilitate treatment decisions and help predict post-adenotonsillectomy complications in children. Overall, it was concluded that a desaturation index of more than two episodes per hour can predict both mild and moderate-to-severe OSA, while criteria based on clusters of desaturations as described by Brouillette et al. [4], and the McGill oximetry score can predict moderate-to-severe OSA. However, not all reports are in agreement and it seems that overnight oximetry is less reliable in certain subgroups of children. It remains essential to study the role of pulse oximetry in subgroups of children, as different abnormalities in nocturnal oximetry could occur in different subgroups. For example, a study in obese children showed that nocturnal oximetry alone is insufficient for a diagnosis of OSA [6]. Oximetry seems less reliable in predicting the presence of OSA in obese children compared to normal-weight children. This could be attributed to the fact that overweight and obesity can be associated with the presence of sleep-related hypoxia even in the absence of OSA [7]. Also in patients with Down syndrome, oximetry seems to be less reliable as it has a higher night-to-night variability compared to other children [8]. In the paper by Pavone and colleagues, a good night-to-night consistency of nocturnal oximetry was described in otherwise normal children referred for suspected SDB [9]. The discrepancy between these papers again demonstrates that nocturnal oximetry can have a different outcome in specific subgroups of children.

There have been attempts to determine normative data for oximetry across the pediatric age spectrum. Scholle et al. [10] described that the 90th percentile for the oxygen desaturation index ≥3% was 2.2 events/hour during the second year of life; less than 1.2 events/hour for ages 2–10 years; and less than 0.5 events/hour for children ages 11–18 years. An important problem identified from this study was that the higher ranges of the oxygen desaturation index among asymptomatic children overlap with those of children with OSA, thus distinctly reducing the accuracy of oximetry in the mild OSA range.

2. How can we improve a diagnosis of OSA by oximetry?

Even though oximetry has shown promise as a diagnostic tool for OSA, it seems that it is insufficient as an entity on its own [6,8]. Therefore, oximetry should best be combined with other techniques to improve diagnostic accuracy for OSA.

A recent study by Hornero et al. investigated the use of automated signal processing algorithms on oximetry recordings in children [11]. The automated signal analysis consists out of four phases. The first phase involves a preprocessing step to standardize signals to the same frequency of 25 Hz and remove artifacts. The second phase is the feature extraction phase in which the presence of pediatric OSA of each recording is characterized. Because the second phase gives a very comprehensive characterization of the oximetry recordings, a third phase was implemented for feature selection. During this phase the relevance and redundancy of the OSA-related data is evaluated. Finally, the last phase consists of a multilayer-perceptron model that is able to automatically estimate the apnea-hypopnea index (AHI). By means of this technique, they were able to reliably identify children with OSA using clinical cutoff values for the AHI. The technique was most reliable when the AHI ≥5, which is a clinically relevant cutoff associated with higher morbidity and need for treatment in children with OSA. The neural network algorithm reduced the need for PSG by 53%, while only treating 6% of snoring children with an actual AHI <1. Furthermore, only 5.5% of children with an AHI ≥5 would have missed treatment. It seems that this technique is as reliable as respiratory polygraphy. However, the technique is still in the beginning phase of development and not yet universally available. But for the future, it could make nocturnal pulse oximetry more reliable as a screening tool for OSA.

Garde et al. combined oximetry and pulse rate variability by using a smartphone for diagnosing OSA in children [12]. They found that the combination of oximetry and pulse rate variability improved the diagnostic performance of this Phone Oximeter compared to only an oximetry signal, with in an increase in sensitivity from 80% to 88%.

The ApneaLink is an outpatient screening device that combines pulse oximetry with oronasal airflow. This device has shown good concordance with PSG in adults [13]. However, to date there are only a limited number of studies that focused on the ApneaLink in children. Stehling et al. concluded in recent study that the ApneaLink would be a good screening tool for diagnosing OSA in children older than 10 years [14]. In contrast to the oximetry, the ApneaLink has also demonstrated good results in obese children [15]. Alonso-Alvarez el al. studied another respiratory polygraphy device that on top of pulse oximetry and airflow, also measured thoracoabdominal movement, electroencephalography (EEG), and body position [16]. They found a sensitivity of 91% and a specificity of 94% for a cutoff value of AHI ≥5 events/hour. It seems that the combination of pulse oximetry with at least oronasal airflow gives a more reliable result compared to pulse oximetry alone in a pediatric cohort.

Pattern analysis of mandibular motion by means of a midsagittal mandibular movement magnetic sensor is another method that shows promise. A recent study in children indicated that several patterns of mandibular movement during sleep in children with upper airway obstruction correlated with increased respiratory effort during upper airway obstruction [17]. Only one study in adults has investigated the combination of mandibular motion and oximetry, with promising results [18].

3. Conclusion

Nocturnal pulse oximetry is an easy low-cost screening tool for OSA in children when PSG is not available. However, oximetry does not perform perfectly inherent to a screening tool and the clinical interpretation should be based on a good knowledge of its pros and cons. We therefore conclude that it should be combined with other techniques for a more reliable result. Automated signal processing algorithms of oximetry signals and mandibular motion analysis show clear promise for the future. Home respiratory polygraphy is an existing good alternative to single channel pulse oximetry, but is more expensive and also requires manual scoring. Nevertheless, screening methods seem to be most reliable in the case of moderate-to-severe OSA. A dependable screening option for mild OSA is still missing, and future studies should explore this option.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This manuscript has not received funding.