Adherence to and Effectiveness of Positive Airway Pressure Therapy in Children With Obstructive Sleep Apnea

Study Group

Patients who were aged 2 to 16 years, had newly diagnosed OSAS, and were candidates for PAP were eligible for the study. Children who were younger than 2 years and those with Down syndrome were excluded from the study because the standard of care at 1 of the participating centers was to provide formal behavioral therapy for these patients to enhance adherence.⁹ Patients with pathologic central apnea or chronic respiratory failure were excluded because they would require BPAP. Patients who had received previous PAP treatment were excluded. All patients received standard medical care from a pediatric, board-certified sleep specialist, including therapy for treatable risk factors for OSAS, before study enrollment.

OSAS was diagnosed according to pediatric standards.¹⁰ Patients had to meet at least 1 of the following criteria: (1) obstructive apnea hypopnea index (AHI) ≥5/hour (hypopnea defined as a ≥50% decrease in airflow of any duration associated with 4% desaturation and/or arousal; obstructive apneas scored if ≥2 respiratory cycles in duration); (2) arterial oxygen saturation (Sao₂) nadir ≤85%, (3) peak end-tidal Pco₂ ≥55 mm Hg, or (4) Sao₂ <92% for ≥10% of total sleep time (TST) and end-tidal Pco₂ ≥50 mm Hg for ≥10% TST. In addition, patients had to have at least 1 of the following symptoms or signs: (1) snoring and difficulty breathing during sleep, (2) failure to thrive (weight <5th percentile for age), (3) pulmonary hypertension, (4) excessive daytime sleepiness (school-aged children), or (5) symptoms of attention-deficit/hyperactivity disorder or other behavioral problems (school-aged children).

Baseline Assessment

Patients underwent baseline polysomnography (PSG) as part of their clinical assessment, according to standard pediatric recommendations.^10–12 Height, weight, and awake BP were measured. Care was taken to follow pediatric guidelines and use appropriately sized BP cuffs when measuring the BP.¹³ A standardized questionnaire regarding symptoms was administered to the caregivers.

Subjective Assessment

During a standardized interview, parents were asked to assess subjectively their child's degree of sleepiness, temperament, and school performance. Some questions had a yes/no response, and others rated frequency of events (eg, never, once a week, 2–3 times a week, >3 times a week). Questions regarding sleepiness included the general question of whether the child was sleepy, as well as more specific questions to assess the frequency of the child's falling asleep while watching television, at school (for children older than kindergarten level), and when active. The frequency of snoring, labored breathing at night, and enuresis (for children ≥6 years of age) was assessed. Yes/no questions included questions as to whether the child was irritable, was hyperactive, had age-appropriate behavior, and had delayed development. For school-aged children, parents were asked to rate the child's school performance as “above average (mainly A's and B's),” “average (mainly C's),” or “below average (mainly D's and F's).”

Institution of PAP Therapy

Children were given a mask and headgear without the PAP unit to practice wearing for 2 weeks while awake to help them habituate to the system. Each family also received a standardized behavioral instruction sheet.⁹ After 2 weeks, the patient underwent an overnight laboratory titration study to determine the optimal pressure required. A standardized protocol that was based on previous clinical pediatric experience was used. The goal of the titration polysomnogram was to eliminate all obstructive apneas, desaturation, and hypercapnia at a level of pressure tolerated by the patient without excessive awakenings. The minimal CPAP/expiratory pressure (EPAP) that was provided by the study equipment was 3 cm H₂O. For patients who were randomly assigned to CPAP, PAP was started at 3 cm H₂O and was increased to 4 cm H₂O and then increased further in 2-cm H₂O increments as needed. For patients who were randomly assigned to BPAP, the aim was to keep a 6-cm H₂O difference between inspiratory and expiratory pressures. The patient was started on 4/3 (minimum) cm H₂O; pressure then was increased by 2-cm H₂O increments to 6/3, 8/3, 10/4, 12/6, 14/8, 16/10 cm H₂O, etc. Supplemental oxygen was added when the patient desaturated persistently to <92% in the absence of apnea, paradoxical breathing, or snoring.

After the titration study, the patient was provided with a VPAP II ST-A (ResMed, San Diego, CA) PAP machine, mask, headgear, and humidifier, free of charge, and the family was instructed on its use. This machine can provide either CPAP or BPAP. The patient, the patient's family, and all study physicians and coordinators were blinded as to the mode of therapy. Because patients and their families were naïve to the use of PAP and had not experienced either CPAP or BPAP treatment in the past, it was not anticipated that they would guess which modality they were receiving. By necessity, the sleep technician was unblinded; however, the technician followed a strict protocol. Randomization was performed using a randomization table generated by the study sponsor, and numbered slips in envelopes were provided for each patient. The sleep technician set the equipment to the randomized mode during the titration polysomnogram. For preservation of blinding, the machine control panel and display were covered. Because few pediatric masks are available in the United States, a variety of masks were used to provide the best fit. For maximization of comfort and, hence, adherence, all patients were provided with a heated humidifier, and a 20-minute pressure ramp was used at sleep onset and after awakenings. After the sleep study, patients took the equipment home and were instructed to use it during all hours of sleep.

Follow-up

Patients received a follow-up telephone call after 48 hours and again after 1 week of PAP use. They then were seen in clinic every other month and received a telephone call on alternate months when they were not being seen. During clinic visits, the patient was assessed clinically; a standardized questionnaire regarding symptoms of OSAS, as well as side effects of positive pressure use, was administered; weight, height, and BP were measured; the equipment was inspected; and the usage meter was downloaded. During telephone calls, a standardized verbal questionnaire was administered regarding subjective symptoms as well as potential PAP problems and side effects. Side effects were assessed and treated at the discretion of the sleep specialist as per standard clinical practice (eg, prescribing nasal steroids for nasal congestion). After 6 months, a repeat polysomnogram was performed on current PAP settings; and height, weight, BP, and subjective complaints were reevaluated.

Patients who did not return for follow-up received telephone calls followed by letters. Patients who did not return their equipment also received registered letters.

Assessment of Adherence

Objective adherence was obtained by downloading the equipment at each clinic visit. Poor adherence was defined arbitrarily as use of PAP for an average of <3 hours/night. Patients with poor adherence received appropriate counseling and troubleshooting as per standard clinical practice. The protocol called for patients who were nonadherent at the 3-month evaluation to be switched to the alternative mode of therapy, with a repeat titration sleep study to ensure adequate ventilation. In practice, only 1 patient was switched to the other arm of the study, because most nonadherent patients were also not adherent with repeat visits. Patients who were lost to follow-up were assumed to be nonadherent.

Statistical Analysis

The primary objectives in this study were to determine adherence to and effectiveness of PAP, both CPAP and BPAP, in children with OSAS. The main outcome measurements were mean nightly hours of use and AHI. For continuous variables, such as age, values are given as means ± SDs.

For categorical variables, such as gender, the proportion of patients in each category is presented. Differences between the 2 modes (BPAP vs CPAP) for demographics (gender, race, and age) and baseline characteristics (height, weight, BP, and reason for OSA) were compared using the independent t test or χ² test where appropriate. These same types of analyses were done to determine differences in the adherent versus nonadherent groups. Differences between initial visit and end visit (5 month or 6 month) for outcome variables were assessed using paired t tests, χ² test, or Pearson correlation coefficients. Statistical significance required P < .05.

During the interim analyses, comparisons between modes were performed by 2-way analysis of variance with daily and monthly usage as a within-subjects factor and with mode as a between-subjects factor. There did not seem to be any differences between the BPAP and CPAP modes, so future analyses used combined data (CPAP + BPAP).

Data that were collected in this study were documented on case report forms. A Code of Federal Regulations Part 11 compliant database system was created using Microsoft Access 97 software (Redmond, WA) for computer entry and verification of the data collected on the case report forms. All computations of results were generated using SAS software (Version 8.2; SAS Institute, Cary, NC).

Study Group

Thirty children were enrolled. One patient who was randomly assigned to BPAP was excluded later because it was unclear whether his equipment was used or not, because there was a technical problem with the usage meter as a result of electricity surges in the home. Thus, the results from 29 patients are included in this article. Sixteen patient were randomly assigned to BPAP, and 13 were assigned to CPAP. Seven children weighed <30 kg, the FDA cutoff for PAP use. Study group details are shown in Table 1. OSAS was attributed to obesity in two thirds of the patients; 11 children weighed >100 kg (range: 12–171 kg).

For the CPAP group, the mean pressure used was 8 ± 3 cm H₂O (range: 4–12). For the BPAP group, the mean inspiratory pressure was 11 ± 4 cm H₂O (range: 4–16), and the mean EPAP was 5 ± 3 cm H₂O (range: 3–10).

Patient follow-up is shown in Fig 1. Nineteen patients completed all aspects of the protocol. Seven patients dropped out between 1 and 5 months of the study; therefore, 6-month data were unavailable. Two patients had 6-month downloads but no PSG, and 1 had 6-month PSG but no download. There were no significant differences in baseline polysomnographic parameters between patients who completed the study and those who dropped out.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Flow diagram of patient enrollment and follow-up.

Adherence

Patients were classified as nonadherent when mean use was <3 hours/night, despite counseling and troubleshooting. In addition, patients who did not return for follow-up visits (despite frequent telephone calls and reminder letters) were classified as nonadherent; it was assumed that these patients were not using the equipment. Of the 8 patients who did not have 6-month downloads, 2 had 0 follow-up after initially receiving their equipment, despite concerted efforts to contact them. Of the remaining 6 patients, 4 used their machine for a mean of <1 hour/night in the month before dropping out, and 1 patient used it for 0, 2, and 5 hours, respectively, for the 3 months before he dropped out. Thus, it seems as though all of these dropouts were nonadherent. The last child was an exception, using the equipment for 4 to 7 hours/night for 5 months. After month 5, the child was placed out of the family and could not be located for follow-up. For the purpose of analysis, this patient was classified as nonadherent.

Mean nightly use is shown in Fig 2. Figure 2A shows the group mean hour use per month for all patients for whom hour meter readings were obtained; Fig 2B shows the hour use for all patients, assuming 0 hours for those for whom a download could not be obtained. At month 6, the mean nightly use for patients for whom downloads could be obtained was 5.3 ± 2.5 hours. The mean nightly use for all 29 patients, assuming that those who did not return to have their equipment downloaded were using the machines for 0 hours/night, was 3.8 ± 3.3 hours. Interindividual use was variable, with 3 children using their equipment for 9 to 10 hours/night. Among those who used the equipment for >3 hours a night, PAP use increased from month 1 to month 6 (P = .008).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2

The mean (SD) nightly use is shown for patients who received CPAP (○) and BPAP (•). A, Data are shown only for patients for whom downloads could be obtained. The numbers above each data point reflect the number of patients who were using BPAP and for whom downloads could be obtained at that time point, over the number of patients who initially were enrolled. The numbers below each data point are the corresponding values for patients who were using CPAP. B, Data for all patients, assuming 0 usage for those for whom downloads could not be obtained. The tables show the t values, P values, and degrees of freedom for CPAP versus BPAP.

There was no difference in use between patients who were using CPAP and BPAP (Fig 2), although the study was not powered to detect a difference between the 2 groups. It was assumed that a difference of 1 hour of use per night between the 2 treatment groups would be clinically significant. On the basis of an acceptable difference of 1 hour and an SD of 2.45 (the pooled SD), a sample size of 63 patients per group would have been required to provide a power of 80% at the 95% confidence level; this sample size was not practical for the current study.

There was no significant difference in age, gender, race, severity of OSAS (based on AHI, arousal index, and Sao₂ nadir), subjective symptoms (eg, sleepiness, enuresis, hyperactivity, irritability, school performance), mode of PAP, pressure level, or indication for PAP (obesity vs all other indications) between those who were nonadherent and those who were adherent.

Subjectively, parents were asked to estimate how many hours per night their child used the PAP. At the 5-month telephone call, 18 parents gave an estimated duration of use and had their equipment downloaded. Parents stated that their child used the PAP a mean of 7.6 ± 2.6 hours/night, whereas actual adherence, as determined by the usage meter, was 5.8 ± 2.4 hours (P < .001). However, 78% of parents admitted that their child did not use the machine every night. It was frequently volunteered that the machine was not used during weekends or vacations.

Effectiveness

PSG on PAP after 6 months of treatment was compared with baseline PSG for the 20 patients who returned for 6-month PSG. Of those who did not return for 6-month PSG, 2 were totally lost to follow-up, with no downloads available; 3 were adherent with their equipment before dropout; and 4 were nonadherent.

Polysomnographic parameters at baseline (while breathing spontaneously) and after 6 months (while on PAP) are shown in Table 2 for the 20 patients who returned for the 6-month study. Because there were no differences in any parameter between CPAP and BPAP, combined data are shown. There was no significant change in sleep architecture (sleep efficiency and sleep stage distribution) on treatment. However, major improvements were shown in respiratory parameters, with highly significant improvements in both the AHI (P = .003; Fig 3) and the Sao₂ nadir (P = .001). At baseline, the AHI ranged from 3 to 115/hour, with 8 patients having an AHI >15/hour. On PAP, the AHI ranged from 0 to 17/hour, with only 1 patient having an AHI >15/hour. Two patients had a moderate increase in AHI; 1 from 8/hour to 12/hour, and the other from 12/hour to 17/hour. The reasons for this were unclear, but possibilities include reduced TST during PSG for the first patient (216 minutes at baseline and 311 minutes at 6 months), which may have resulted in an inaccurate representation of the true AHI, and weight gain in the second patient (from 106 kg to 112 kg).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3

The AHI at baseline and on PAP after 6 months of therapy is shown for the 20 adherent patients. There was a significant decrease in AHI on PAP (P = .003, t = 3.44, degrees of freedom = 19).

Growth

At baseline, most patients were of normal height (Table 1) but were overweight. The patients as a whole had a mean BMI z score of 2.1 ± 1.1. No patient had a weight or BMI z score <−1. Two patients had a height z score <−2, but both were obese, and the short stature was most likely related to their underlying disease (achondroplasia; renal transplant) rather than their OSAS. Six-month growth data were available for 20 patients. Growth parameters did not change significantly after 6 months of PAP. Only 1 obese patient lost weight (a 1.6-kg reduction from a baseline weight of 159 kg).

BP

Baseline BP data were available for 28 patients. At baseline, 36% of patients had a systolic BP >95th percentile for age, gender, and height; and 43% had a systolic BP >90th percentile. Seven percent had a diastolic BP >95th percentile, and 11% had a diastolic BP >90th percentile. Some but not all of these patient were obese, and 1 patient with systolic hypertension had a renal transplant and a longstanding history of hypertension on treatment. Six-month follow-up BP values were available for 18 patients. There was no significant difference in any of the BP parameters at 6 months.

Subjective Outcomes

After 6 months, there was a significant improvement in subjective parental assessment of sleepiness. This included an improvement in the number of children who were reported to be sleepy (P = .03), falling asleep while watching television (P = .004), and falling asleep at school (P = .007). Of note, at study entry, 58% of school-aged children were reported to fall asleep at school at least once a week, compared with only 6% after 6 months of PAP. However, there was no significant subjective improvement in age-appropriate behavior, delayed development, irritability, hyperactivity, enuresis, or reported school performance.

At baseline, 100% of parents reported that their children snored most nights (>3 nights a week). After 6 months of PAP, there was a significant improvement in reported snoring on PAP (P < .0001): 11% reported snoring most nights, 26% reported snoring some nights (<3 nights a week), and 63% reported no snoring. Similarly, there was an improvement in those who were reported to have difficulty breathing during sleep. At baseline, 62% of patients were reported to have frequent difficulty breathing during sleep, 34% had occasional difficulty breathing, and 3% were not observed to have difficulty breathing; at 6 months, the corresponding values were 5%, 16%, and 79% (P < .0001).

Side Effects and Technical Problems

Side effects and equipment problems were assessed by telephone interview soon after PAP was initiated (ie, at 48 hours) and after long-term use (5 months). After 48 hours of use, 14% of families complained of problems with the equipment (eg, equipment kept shutting off; confusion regarding setup) and 21% complained of problems with the mask (eg, “bubble” buckling, mask leaks). After 5 months, 10% of families complained of equipment problems (eg, tubing worn out) and 10% complained of occasional mask problems, primarily leaks. One of these families stated that the child had “pink eye,” which may have been secondary to the mask leak.

Despite the use of heated humidification, nasal symptoms were common, occurring in 17% of patients at 48 hours and 38% of patients at 5 months. Nasal symptoms included congestion, rhinorrhea, and 2 cases of recurrent epistaxis. Of note, nasal symptoms at 5 months were significantly more common in those who were using CPAP than in those who were using BPAP (P = .02), although there was no difference between the 2 groups at 48 hours (P = .31).

Other side effects were rare. At 48 hours, 3% of families noted skin erythema related to the mask; and at 5 months, 10% noted erythema. However, in all cases, the erythema was only occasional and was transient in nature. One patient complained of acne in the mask distribution and 1 of irritation beneath the nose.

Adherence

The most important finding in this study was the relatively poor adherence. First, 8 patients dropped out of the study without providing final downloads. Two dropped out immediately, and no follow-up was obtained; it is unlikely that these patients used the equipment. Of the remaining dropouts, all but 1 had very low nightly use of their equipment before dropping out. The remaining child was placed out of the home and lost to follow-up. Thus, it is reasonable to assume that all but 1 of the dropouts stopped using their equipment. This is despite the fact that all patients received intensive support at a level greater than typically provided in the clinical situation. This support included the use of free equipment, including a ramp option and heated humidification, and frequent telephonic and clinic follow-up, where problems and side effects were addressed by pediatric sleep specialists. Second, those who continued to use the equipment used it only for a mean of 5.3 hours/night. Although this is very similar to adherence rates in adults,^18–21 children sleep longer than adults and thus would require longer PAP use. For the age range of 2 to 16 years, children sleep an average of 8 hours/night.²²

In this study, we arbitrarily chose a cutoff of 3 hours of use a night to define adherence to evaluate the potential predictors of adherence. Clearly, 3 hours of use is suboptimal. However, as no previous studies have documented objective PAP use in children, we chose a low number that would clearly separate those who were using the equipment little if at all from those who were using it more consistently. There were no trends suggesting that a different cutoff point would yield different results.

Previous studies in children have reported relatively good rates of subjective adherence on the basis of parental report.^5,15 However, in this study, there was a marked discrepancy between parents' report of PAP use and objective adherence. Thus, it is important to monitor adherence objectively.

We found no difference in adherence or effectiveness between the two modes of pressure delivery. It should be noted that this study was not powered to detect a statistically significant difference. Nevertheless, there was no indication of a trend toward a difference in any of the parameters, although the mean CPAP pressures (8 cm H₂O) were higher than the mean EPAP pressures (5 cm H₂O). In adults, adherence has been shown to be similar between CPAP and BPAP use.²¹ Nasal side effects after chronic use were more common in those who were using CPAP than BPAP. Future, larger studies to evaluate the difference in adherence between the 2 modes are required.

Two thirds of the patients in this study required PAP because of obesity. The prevalence of obesity is increasing rapidly in children and adolescents and is currently 15%.²³ Whereas in previous years OSAS in most children was related to adenotonsillar hypertrophy, it is now frequently related to obesity.²⁴ A significant number of children were extremely obese. Although not evaluated in this study, it is possible that these children had low self-esteem and other psychosocial issues related to their obesity that had an impact on their adherence with PAP.²⁵

Effectiveness

Both CPAP and BPAP were highly effective at treating OSAS, as documented by the polysomnographic results. This is in agreement with several other studies that were performed in infants^16,17 and older children.^7,26,27

No changes were seen in growth. This is in contrast to a large number of studies that demonstrated increased weight and height after treatment of OSAS by tonsillectomy and adenoidectomy.^28–30 The difference between the current studies and the surgical studies is probably related to differences in the population. Most patients in our study had OSAS related to obesity rather than adenotonsillar hypertrophy, and the only 2 short patients (height z score ≥2 SD) had underlying diseases (achondroplasia, renal transplant) in which height would not be expected to improve. Unlike a previous report that evaluated the response to surgical treatment of OSAS,³¹ the current study did not find an increase in obesity after PAP treatment.

A surprisingly large proportion of patients had systolic hypertension before institution of PAP. BP did not improve during this study. It is likely that the hypertension was attributable to obesity rather than OSAS, especially because diastolic hypertension may be more common than systolic hypertension in children with OSAS.³²

Subjectively, parents noted an improvement in snoring and nocturnal breathing and an improvement in all parameters of daytime sleepiness, although they did not note any changes in temperament, hyperactivity, or school performance. Side effects were minor, although nasal symptoms were frequent.

Midfacial hypoplasia has been reported as a side effect of PAP in children, although it is probably rare.^33,34 We did not see any instances of this in the current study, but this may have been because infants were excluded from the study and because the study duration (6 months) may have been too short for this development to occur.

Limitations

The main limitation of this study was the high dropout rate. However, we posit that the dropouts, with the exception of 1 child who was placed out of the home, actually reflect nonadherence and thus are part of the outcome data for this study.

This study did not include an untreated control group. Theoretically, therefore, the study results (especially the subjective results) may have been attributable to a placebo effect. However, we consider it unlikely that all of the observed changes were attributable to a placebo effect because of the highly significant improvement in objective measurements such as the polysomnographic parameters. Previous studies that evaluated the effects of placebo (sham CPAP) in adults showed that placebo may cause a slight improvement in subjective measures such as quality-of-life parameters³⁵ but does not result in an improvement in measures of sleep-disordered breathing.^35,36