In adults, the treatment of choice for obstructive sleep apnea (OSA) is continuous positive airway pressure (CPAP) device, which functions by blowing air into the airway to overcome the critical closing pressure of the pharynx and maintain a patent pharyngeal airway during sleep [1]. Unlike adults, CPAP is considered in children only when OSA persists despite adenotonsillectomy or such surgery is contraindicated. The pressure level that each patient requires needs to be titrated individually under attended polysomnography, during which CPAP pressure is started at a low level and progressively increased to find the level at which the following obstructive respiratory events, including apneas, hypopneas, respiratory effort-related arousals, and snoring, are eliminated. The pressure required for resolution of events is often higher if the patient is supine, and may be higher in REM sleep when muscle tone is the lowest. Therefore, the final prescription should reflect the single fixed pressure required to treat the worst apnea in supine REM sleep [2]. To standardize CPAP titration technique, the American Academy of Sleep Medicine (AASM) have published a clinical guideline for the manual titration of CPAP pressure in OSA patients in 2008 [3].

Over the last two decades, auto-CPAP machines have developed and focused mainly for adult patients. The CPAP pressure was initially set up as a range: Lowest and highest pressure levels. The machines are programmed to adjust CPAP pressure automatically based on detection of apnea, hypopnea, flow limitation, snoring, pressure fluctuations or increased airway resistance. With the implement of auto-CPAP, the need for CPAP titration under polysomnography is supposedly decreased. A systematic review with meta-analyses on 24 randomized controlled trials comparing auto-CPAP with single fixed pressure CPAP in adults with OSA found that auto-CPAP improved compliance by 11 minutes per night and reduced the Epworth Sleepiness Scale by 0.5 points. Although statistically significant differences were found but clinical importance was unclear. Therefore the choice of therapy still depends on patient’s preference or cost [4].

There have been relatively fewer studies of auto-CPAP use in children as compared to adults. In 2004, Polombini et al. demonstrated that auto-CPAP was safe and effective for children with OSA in an attended setting. It was also found to be useful for pressure titration [5]. In 2009, Marshall et al. demonstrated that auto-CPAP was feasible and safe to be used at home in children with sickle cell anemia, which resulted in improved sleep-related breathing disorders and at least one aspect of cognition [6].

All auto-CPAP devices have a slot in the machine allowing for a smart card to capture data. The data can be downloaded and viewed or printed out. Some machines report not only the patients’ adherence but also the amount pressure the patients have received and express as an average pressure and as a percentile pressure. In 2017, Mihai et al. conducted a retrospective study in children comparing these downloaded pressure data with the CPAP pressure determined by PSG. They found that downloaded pressure from auto-CPAP report were usually below treatment pressure determined under PSG [7]. Since this study was a retrospective review and recruited CPAP-naïve children whose parents reported CPAP use of at least 1 to 2 hours a night, 39% of the children were found to be non-adherence as using auto-CPAP less than 90% of the time. Therefore downloaded pressure data may be confounded with the data from non-adherence children. We hypothesized that if the children were adherence to auto-CPAP, downloaded pressure data might be closed to the CPAP pressure determined by attended PSG titration. If so, instead of performing PSG titration, auto-CPAP could be started in a new patient right after diagnosis of OSA and the optimum CPAP pressure level could be obtained easily from downloaded data. For any cost reasons, the patient could transition to fixed pressure CPAP afterwards. So the objective of this prospective study was to compare the pressure level derived from auto-CPAP downloaded data with the CPAP pressure determined under attended PSG in children with OSA.

Subjects

The study was conducted prospectively from February 2016 to January 2017. We enrolled children aged 2 to 18 years old, diagnosed with OSA by PSG. They needed CPAP therapy and were willing to use auto-CPAP at home with acceptance from parents. Children with neuromuscular weakness and central hypoventilation were excluded. This study was approved by the Institutional Review Board (IRB) of Ramathibodi Hospital. All children and parents provided informed consent.

Methods

Polysomnographic studies

A standard attended PSG for titration of CPAP, was performed at Ramathibodi Hospital Sleep Disorder Center. We used PSG equipment (Grael system, Compumedics, Melbourne, Australia). The measured parameters included the recording of frontal, central and occipital electroencephalograms, left and right electrooculograms, submental and bilateral tibial electromyogram, electrocardiogram, nasal thermistor, nasal pressure, pulse oximetry, end tidal CO₂, thoracic and abdominal belt for respiratory effort monitoring, and infra-red camera. Signals were recorded with Compumedics® (Compumedics Ltd, Abbotsford, Australia) and scored following the American Academy of Sleep Medicine (AASM) criteria [8]. Apneas were defined as a drop in peak signal excursion by ≥ 90% of pre-event baseline using oronasal thermal sensor. Obstructive apnea defined by criteria of apnea for at least duration of 2 breaths during baseline breathing and associated with presence of respiratory effort. Hypopneas were defined as a peak signal excursions drop by ≥ 30% of pre-event baseline using nasal pressure and the duration lasts for ≥ 2 breaths and ≥ 3% oxygen desaturation from pre-event baseline of the event associated with arousal. Children were accompanied by parents throughout the night. The studies were conducted by two skillful clinical sleep technicians experienced with children. One night PSG titration was performed following AASM clinical guidelines for the manual titration of positive airway pressure [3], with the goal to eliminate all obstructive apneas and hypopneas and minimize arousals and snoring.

Auto-CPAP studies

All patients were instructed to use auto-CPAP devices (System One REMstar Auto, Philips Respironics Inc, Murrysville, Pennsylvania, United States). This device detects upper airway resistance by flow limitation. It measures changes in airflow by an internal pneumotachograph to identify respiratory events. If flow decreases by 40% to 80% for at least 10 sec, the event is labeled a hypopnea; a decrease in flow by more than 80% for at least 10 sec is labeled an apnea [9,10]. The device starts off at the minimum pressure set point for five minutes. At this minimum pressure setting, the patient is awake and the upper airway is patent and upper airway resistance is low. As the patient falls asleep the upper airway begins to collapse thus increasing the resistance. The device then delivers proper pressure until there is no flow limitation [11].

Auto-CPAP was assigned to be used at home for 2 weeks either before or after PSG titration. The period in between the use of auto- CPAP at home and PSG titration was less than 8 weeks. The study flow diagram is shown in Figure 1. Auto-CPAP pressure were adjusted to 4-15 cm H₂O in children aged <12 years, and 4-20 cm H₂O in children aged >12 years. Mask fitting was completed by our clinical sleep technicians. We let the patients select their own masks according to their preference to optimize acceptability and comfort. Another important aim of mask fitting was to minimize leaks. The patient came back to our sleep clinic after two weeks of home auto-CPAP. The data from auto-CPAP were downloaded using EncorePro Basic (Philips Respironics Inc, Murrysville, Pennsylvania, United States). The pressure parameter taken from the standard report was “average device pressure ≤ 90% of time”, at which the device spent 90% of the session time at or below this pressure. Adherence data were also downloaded. Data from only those who used auto-CPAP ≥ 4 hours per day, for >consecutive 7 days and percent days with device usage ≥ 80% were selected for analysis. Average device pressure ≤ 90% of time from auto- CPAP was compared with PSG titrating pressure.

Figure 1: Study flow diagram.

Statistical analysis

Data were analyzed using SPSS version 20. Descriptive statistics include mean, standard deviation, and median values. The data are expressed as the mean+SD for continuous variables or numbers and percentages for categorical variables. Continuous variables were compared using paired Student t test or the Mann-Whitney test. Categorical variables were compared using the λ2 test or the Fisher’s exact test. Relationship between two continuous variables was assessed using Pearson’s correlation coefficient. A p value of <0.05 was considered statistically significant.

Seventeen OSA patients who needed CPAP therapy were invited to participate to the study. Three patients refused to use auto-CPAP at home. Another three patients who failed to meet the criteria of CPAP adherence were excluded. Informed consent was obtained from all individual participant included in the study. Data from 11 patients were analyzed in this study. Of the remaining 11 patients, 9 (82%) were male. Demographic data are shown in Table 1. Six (55%) were obese. Mean body mass index z-score was above 2.0. Five (45%) presented with enuresis. Nine (82%) had severe OSA as apnea hypopnea index >10 events/h. All of our patients had been treated with intranasal corticosteroids and leukotriene inhibitor. Six (55%) had undergone adenotonsillectomy.

Table 1: Demographic data of participants (n = 11).

Average device pressure ≤ 90% of time derived from auto-CPAP was 11.05+3.68 cmH₂O, and PSG titrating pressure was 11.18+3.34 cm H₂O. There was no statistically significant difference between these two pressure variables (p value=0.84). However, when considering on actual values of the pressure as shown in Figure 2, 6 (55%) patients had pressure differences >2 cm H₂O as shown in Figure 3, 3 (27%) had pressure differences >4 cm H₂O. PSG pressure was higher than auto- CPAP pressure in 5 patients and lower in 6 patients. The largest difference was 9.3 cm H₂O found in the last patient as shown in Figure 3.

Figure 2: Comparison of pressure levels obtained from polysomnographic titration and auto-CPAP.

Figure 3: Pressure difference (PSG pressure – auto-CPAP pressure) of each patient in cm H₂O.

Of 6 patients with history of adenotonsillectomy, 5 (83%) had pressure differences >2 cmH₂O. In contrast to 5 patients without history of adenotonsillectomy, there was only 1 (20%) patient having pressure differences >2 cmH₂O. Fisher’s exact test showed a statistically significant difference between these two percentages (p value=0.036). No other conditions of the patients (including age, sex, body mass index, OSA severity, allergic rhinitis, Prader Willi Syndrome) were found to be associated with the pressure differences >2 cmH₂O.

Figure 4a shows no correlation between PSG pressure and auto- CPAP pressure in patients with history of adenotonsillectomy (r=0.03, p value=0.518) in contrast to the patients without history of adenotonsillectomy, in whom PSG pressure correlates well with auto- CPAP pressure (r=0.91, p value=0.035) as shown in Figure 4b.

Figure 4a: Poor correlation between PSG and auto-CPAP pressure in patients with history of adenotonsillectomy (n=6).

Figure 4b: Strong correlation between PSG and auto-CPAP pressure in patients without history of adenotonsillectomy (n=5).

No patients reported adverse events from auto-CPAP.

This pilot study indicates that although auto-CPAP devices are safe to be used in children, they cannot be used to determine the optimal CPAP pressure requirement or replace CPAP titration under PSG. The actual pressure levels derived from auto-CPAP differed from PSG pressure greater than 2 cmH₂O in more than half of the patients. Auto- CPAP pressure levels were found to be either higher or lower than PSG pressure. And the pressure difference was as high as 9 cm H₂O in one patient. The findings of this study suggest that a manual CPAP titration under attended PSG is still essential to determine the most effective pressure for fixed-pressure CPAP in children.

Most of the studies with similar objectives comparing the pressure levels derived from auto-CPAP with PSG titrating pressure have been conducted in adults. The results were not consistent. Luo et al. found that automatic titration pressure derived from REM star Auto was significantly higher than that derived from manual titration under PSG [12]. Choi et al. reported no significant differences in pressure level between full-night manual titration and auto-adjusting titration [13]. Lloberes et al. found that daytime auto-CPAP titration yielded a higher treatment pressure than PSG titration but the clinical outcomes were similar [14].

So far very few studies on auto-CPAP have been conducted in children. Mihai et al. conducted a retrospective study comparing treatment pressure derived from auto-CPAP and those from overnight PSG titration. They found that average device pressure < 90% of time downloaded from auto-CPAP did not statistically differ from PSG pressure. However the pressure difference falling within 2 cm H₂O were noted only in 69% of cases. The magnitude of difference ranged from 8 cm H₂O under PSG pressure to 3 cm H₂O over PSG pressure [7]. The results of Mihai’s study were compatible with our current study. In addition to the results and recruiting children as subjects, the similarity between ours and Mihai’s study was the use of REMstar Auto as a single auto-CPAP device at home, not in the sleep laboratory. We also selected average device pressure <90% of time as a representative of auto-CPAP pressure. Dissimilar factors that strengthened of our study were the prospective study design and non-adherence subjects were excluded. We selected patients only those who used auto-CPAP ≥ 4 hours per day, for >consecutive 7 days and percent days with device usage ≥ 80% in comparison with Mihai’s study that included patients who used at least 1 to 2 hours a night. One limitation of our study was a smaller sample size that may explain why the differences in pressure levels derived from the two techniques could not be demonstrated.

Our study confirmed that auto-CPAP devices were well tolerated and safe to be applied in children with no adverse effects reported. Only 3 out of 14 recruited children failed to meet the criteria of CPAP adherence. The youngest age of our adherence patients was found to be 4 years with the body weight of 12.6 Kg. This indirect evidence confirmed the usefulness of auto-CPAP in children with OSA as reported by Palombini et al, Marshall et al. and Mihai et al. Unfortunately we did not collect data of the polysomnographic parameters, oxygen saturation or the quality of life that might be improved after auto-CPAP therapy [5-7].

One interesting finding was the effect of adenotonsillectomy to the pressure difference. In children with no history of adenotonsillectomy the pressure determined by auto-CPAP correlated well with PSG titration pressure. However, in children who had undergone adenotonsillectomy there was no significant correlation. The effect of adenotonsillectomy on the function of auto-CPAP devices have never been described before. In adults, upper airway surgical treatment was found to have some benefits by reducing nasal CPAP pressure levels [15,16]. In children, adenotonsillectomy also reduces CPAP requirement most likely due to alleviation of upper airway obstruction [17]. However it is unknown why auto-CPAP devices function differently in patients with and without history of adenotonsillectomy. It is likely that there may be some changes in the pattern of air flow through the upper airway surrounded by surgical scar resulting in the changing pattern of auto-CPAP response. Further studies on this issue are needed to confirm and explain this finding.

It should be noted that we used the same auto-CPAP device in all children. Our results cannot be applied to other brands of auto-CPAP since different devices used different proprietary algorithms to adjust pressure and the performance of one device should not be assumed to be equivalent to that of another [18]. Isetta et al. conducted a bench test to assess how 7 available auto-CPAP devices respond to a simulated OSA patient. They found wide variations in mean and maximum pressure, and also the time to reach maximum pressure [19]. Therefore the pressure derived from each brand of auto-CPAP may not be the same.

In conclusion, although auto-CPAP is safe to be used at home in children, it cannot replace the role of PSG titration in determining the most effective fixed pressure level of long-term CPAP therapy in children with OSA.

Although the number of subjects is only 11, we still see the difference of pressure levels greater than 2 cm H₂O in more than half of the patients. We use 2 cm H₂O as a cutoff of pressure difference because this is important in clinical practice. The magnitude of pressure difference in one patient is as high as 9 cm H₂O which may cause harmful effect clinically. Therefore the pressure level obtained from auto-CPAP cannot be used to replace the pressure derived from polysomnography. Auto-CPAP pressure levels were either higher or lower than PSG pressure. So it would be too difficult to predict optimal pressure from auto-CPAP.

We thank Teeradej Kuptanon, Wanaporn Anantaseree, Pravit Jetanachai, Anchalee Leejakpai, Vijittra Hongtong and Ramathibodi Hospital Sleep Disorder Center for all their support. Moreover, our special thanks go to children and parents who kindly participated in this project.

The authors declare that they have no conflict of interest.

No funding was received for this research.