The study design was that of a case-control study. Patients were diagnosed as having OSAS after laboratory-proven polysomnographic (PSG) evaluation and were recruited according to their date of birth. Healthy control subjects were not assessed by PSG.

The Institutional Ethics Committee of the Soroka University Medical Center (Beer-Sheva, Israel) approved this study.

The study took place at the Sleep-Wake Disorders Center at Soroka Medical Center (affiliated with Ben-Gurion University of the Negev). Children had been residents of the region for at least 4 yr before the PSG study.

All children (a total of 161 children, ages 3.11 to 5.09 yr), enrollees of Clalit Health Care Services (CHS), the largest health maintenance organization in the region, with clinically suspected OSAS were recruited consecutively during the period of January 2001 through December 2004. In our setting, these children are first referred by their family physician to otolaryngology (ear, nose, and throat [ENT]) surgeons or pulmonologists (OSAS-related consultants), who then refer the patients for a PSG study. The control group, selected randomly, was matched 1:1 to the OSAS group by age, sex, and primary care physician to adjust for socioeconomic factors and differences in physician daily practice. We explored “typical” patients with OSAS (14) by excluding children who exhibited extreme consumption of health care services, that is, more than 10 times the mean values of their group. Control subjects were otherwise healthy (i.e., were not included in the databases of severely ill children, based on a review of medical records during hospital visits and supplied drugs in the community clinics). It is possible that 1–2% of the control group might have had undiagnosed OSAS (14). Five children with OSAS and one control subject who had known comorbidity were excluded from the original participants. The final database included 156 otherwise healthy children with OSAS and their paired control subjects.

Data resources for the patients with OSAS included PSG data (15), sleep questionnaires, and hospital records. Diagnoses were classified (International Classification of Diseases, 9th revision) into three categories (13): upper respiratory diseases, lower respiratory diseases, and other. We were not permitted to contact either group to obtain additional information because of patient confidentiality (12–14). Data on health care use for both groups were obtained from the CHS Economics Department.

Health care use and costs were analyzed by going back from the date of OSAS diagnosis to the child's day of birth. Total annual cost (12–14) includes the following: number and duration of admissions, number of emergency department visits, and number of new and repeated visits for consultations. Number and type of prescriptions supplied were described according to anatomic and therapeutic criteria of the World Health Organization (16). Cost of the PSG study was not included.

Health care use was analyzed in two modes: for the 12-mo period before PSG diagnosis (n = 156) (12–14) and in full years from the first to the fourth year of life (Years 1–4). The latter was defined according to the information available in the CHS computerized databases, starting in January 1998.

A case-control data analysis (17) was performed with SPSS (version 12.0; SPSS, Chicago, IL) software. Cost elements were performed as previously described (12–14). Data were presented as means ± SD for all PSG parameters and as means ± SEM and median and range for costs. The null hypothesis was rejected at the 5% level.

One hundred and fifty-six typical children with OSAS with a mean age of 4.02 ± 0.7 yr (95 males and 61 females) at the time of PSG diagnosis were included in the study. The control group included 156 children matched 1:1 by age, sex, geographic location, and pediatrician or family physician. Among patients with OSAS, 43 (29.9%) of the parents reported smoking at least five cigarettes/d; of those, 32 (78%) were fathers.

The OSAS group (Table 1) had an average apnea–hypopnea index (AHI) of 8.1 ± 7.8 events/h. Relative to the average AHI, REM AHI was higher by 10 events/h (p < 0.0001), that is, compatible with OSAS of moderate severity (18) (i.e., in our settings we define mild, moderate, and severe OSAS as having an AHI of > 1 to ⩽ 4.9, 5 to ⩽ 9.9, and ⩾ 10 events/h, respectively). AHI severity was comparable in boys and girls. As a group, children with OSAS did not present evidence of considerable nocturnal hypoxemia, that is, as a group, mean T₉₀ = 2.1 ± 4.6%, and total desaturation index was 3.4 ± 4.9 events/h (Table 1). After PSG, 103 children (66%) underwent surgery.

Analysis of health care use was performed per complete 12-mo period to minimize seasonal effects. Annual health care use among the OSAS group was $275 ± 54 compared with $128 ± 22 in the control group—that is, 215% higher compared with the control group (p < 0.0001, Wilcoxon test). Cost elements per patient per year before the PSG study are summarized in Table 2. Patients with OSAS had an average of 0.14 new admissions per year compared with 0.07 new admissions in the control group (p = 0.8). Children with OSAS had a similar number of visits to the emergency room compared with the control group, 0.4 versus 0.3 visits (p = 0.2), respectively. In our CHS region, when ENT consultation is required, more than 85% of children are referred to the 10 ENT surgeons rotating among 11 clinics. Patients with OSAS and control subjects were referred to these surgeons in equal proportions (p = 0.6). Children with OSAS had more consultations (ENT consultation before PSG referral was not included in the analysis) compared with the control group, 1.2 versus 0.4 visits, respectively (p < 0.0001). The OSAS group needed more recurrent consultations (at least two visits) than the control group, 29.6 versus 7.7% (p < 0.0001, McNemar test). Common consultations (specialist referrals) in the OSAS group included ENT surgeons for 58 children (40%), pediatric pulmonologists for 16 children (10.3%), and ophthalmologists for 22 children (14.1%). Other consultations included neurologists, cardiologists, dermatologists, and orthopedic surgeons (less than 10% for each specialty). The control group had significantly lower visit rates to specialists: 20 children (12.8%) to ENT surgeons and 7 children (4.5%) to ophthalmologists; all other specialists had 1–5% referrals per specialty.

Costs for drugs for patients with OSAS were six times higher (p < 0.0001, Wilcoxon test) than for the control group (Table 2). The prevalence and number of supplied drugs 1 yr before PSG diagnosis are summarized in Table 3. More subjects with OSAS (odds were up to 6.1 times higher) were supplied with the medications presented in Table 3, which includes only the pharmacologic groups in which differences were found. The average cost of drugs supplied per patient per year to patients with OSAS was up to 140% more than to the control subjects. Specifically, more drugs in the general antiinfectives (J) category and respiratory system (R) category were supplied (p < 0.0001) to patients with OSAS, that is, 236 and 275%, respectively. Ninety percent of supplied system N group drugs were acetaminophen. Ninety-nine percent of the musculoskeletal drugs were ibuprofen. These drugs can also be purchased over the counter.

When the OSAS group is arbitrarily divided by cost, the upper 25% (n = 39) of patients, defined as the most costly, had a mean consumption per person per year of $864 ± 187, which was significantly higher than the lower 75% of patients ($79 ± 7; p < 0.0001, Mann-Whitney test). These upper 25% of costly patients consumed 78.4% of all annual OSAS group costs, that is, 11-fold more health care resources than the lower 75% of patients. The most costly subgroup had similar age, sex distribution, AHI, and sleep characteristics compared with the lower 75% of patients. For example, the AHI was 8.2 ± 6.9 versus 8.1 ± 8.2 events/h, respectively.

AHI correlated (Spearman correlation) only with cost for medical consultations (r = 0.16, p = 0.05) and J category drugs (r = 0.19, p < 0.02). Total annual costs did not correlate with AHI. Children with AHI ⩾ 5 events/h, versus AHI < 5 events/h, were supplied with 10% more drugs from group J (p < 0.0001), had 22% more repeated (two or more) consultations (p < 0.0001), and two times more hospital visits (p < 0.0001). Similar results regarding J group drugs and consultations were obtained in children with ⩾ 2 events/h versus AHI < 2. The percentage of time with Sa_O2 < 90% in our children with OSAS did not correlate with any of the health care use indices. Children exposed to passive smoking showed similar health care use compared with children who were not exposed to passive smoke ($209 ± 58 and $310 ± 79, respectively; p = 0.54, Mann-Whitney test).

During the fourth year of life, 44 children underwent adenotonsillectomy and did not complete the 12-mo period; they were excluded from that year's data. Compared with the remaining 112 (72%) children with OSAS, the 44 children (28%) who were excluded from data analysis because of adenotonsillectomy are younger (3.4 ± 0.3 vs. 4.3 ± 0.7 yr, p < 0.0001) and have more severe OSAS (AHI = 10.7 ± 8.3 vs. 7.1 ± 7.4 events/h, p < 0.0001). No significant differences were found regarding total annual cost, number of consultations per child per year, and percentage of children requiring two or more consultations ($258 ± 146, 1.2 ± 0.18, and 48% vs. $242 ± 49, 1.1 ± 0.15, and 42%, respectively) and sex distribution.

In comparison with the control group, total annual cost (Figure 1) was 160 to 190% higher (p < 0.0001, two-way analysis of variance [ANOVA]) for each year of life in the OSAS group. For each 1-yr increase in age, total annual cost decreased by 35% in the OSAS group and by 50% in the control group (p < 0.0001, two-way ANOVA). The total number of hospital visits throughout Years 1–4 was 40% higher in the OSAS group compared with the control subjects (3.2 ± 3.5 visits per child vs. 2.3 ± 2.6 visits per child, respectively; p = 0.048). Patients with OSAS required more (about 20%) repeated (two or more) hospital visits compared with control subjects (p < 0.0001, McNemar test). Compared with control subjects, the number of OSAS-related consultations (Table 4) was in the range of 50 to 600% higher (p < 0.0001). The rate of referral of patients with OSAS to general consultants (not including ENT or pediatric pulmonologists) was in the range of 30 to 50% higher (p < 0.0001), except for Year 1. Odds for referral of patients with OSAS to OSAS-related consultants (not including the last ENT surgeon visit before PSG study) is especially higher in Year 4 only: odds ratio, 9.4; 95% confidence interval, 4.2–21.2. Patients with OSAS required two to four times more (p < 0.0001, McNemar test) repeated (two or more) consultations with specialists.

Figure 1. Total annual costs from first year of life, including Year 4. Values were obtained for 156 children with obstructive sleep apnea syndrome (OSAS) and pair-matched control subjects (n = 112 for Year 4). Columns represent average annual costs (± SEM) per subject per year in U.S. dollars. ^#p < 0.0001, two-way analysis of variance.

Download Figure | Download Powerpoint

[More] [Minimize]

The annual cost of drugs for patients with OSAS was 70 to 200% higher through the first 4 yr of life (p < 0.0001). Significantly more (p < 0.0001, two-way ANOVA) patients with OSAS were supplied with J and R categories of drugs, beginning in Year 2 of life. The odds (95% CI) of supplying these drugs to patients with OSAS are presented in Table 5. The most supplied respiratory subcategory was nasal preparations, which was provided more (17 vs. 25%) to patients with OSAS (p < 0.0001). Interestingly, drugs for obstructive airway diseases were supplied more (15%) to subjects with OSAS in Years 3 and 4 (odds ratio, 2.0; 95% confidence interval, 1.1–3.9). Other pharmacologic categories were not consistently supplied more to patients with OSAS.

Medical diagnoses made during hospital visits in all 4 yr of life are summarized in Table 6. Compared with the control group, children with OSAS had higher rates of lower respiratory tract diseases (i.e., pneumonia, bronchiolitis, and asthma) and “other” diseases (e.g., gastrointestinal or orthopedic) (p < 0.0001). No differences were found in upper respiratory tract diseases (i.e., otitis media, tonsillitis, laryngitis, and croup).

The major new finding in the current report is that health care use and morbidity are increased for several years before OSAS diagnosis and treatment. Increased morbidity among children with OSAS was related to lower respiratory airway diseases. The total number of hospital visits from the time of diagnosis, starting from the first year of life, was 40% higher in the OSAS group, and these children required 20% more repeated hospital visits. Referral of children with OSAS to ENT and pediatric pulmonologists was significantly higher beginning in Year 1, and especially in Year 4. Antiinfective and respiratory categories of drugs were supplied to patients with OSAS significantly more from Year 2 and up.

We present results of “typical” otherwise healthy children with OSAS compared with their healthy control subjects (19, 20). The study group were less than 5 yr of age; children 5 yr of age or younger have more severe OSAS than do children older than 5 yr (14). In this age category, the occurrence of adenotonsillar hypertrophy peaks during the fifth year of life (7, 21) and health care use is maximal (14). To better understand the influence of OSAS per se on health care use in a physician's daily practice, we had to exclude children who either exhibited extreme consumption of health care services or had additional serious morbidities. Therefore, our result presents the minimum annual expenditures of children with OSAS because none of the children investigated had concomitant diseases. It may be postulated that control subjects should be “normal” subjects (regarding AHI and oxygen saturation) referred for PSG evaluation from the same doctor. However, most subjects referred to OSAS evaluation are symptomatic, mainly reporting snoring. It was suggested that snoring is not just an innocent noise during sleep in infants but may represent the lower end of the disease spectrum associated with sleep-disordered breathing (22). Therefore, laboratory-proven “normal” children are not an adequate control group.

We analyzed data in two modes: 1 yr before diagnosis and from the first year of life including Year 4. These analyses are complementary and minimize uncertainties regarding health care use. At the time of diagnosis, most children with OSAS had not completed the full last year of life. Forty-four children who underwent surgery were excluded from data analysis of Year 4, because adenotonsillectomy reduces health care use (13). These children were significantly younger and had more severe AHI, but their total annual costs 1 yr before diagnosis did not differ from those of the 112 children with OSAS included in the final analysis of this year.

Medical diagnoses made during hospital visits and the type of supplied drugs in the community settings revealed that children with OSAS had significantly higher rates of lower respiratory tract diseases. This finding does not corroborate our previous report (13) that increased morbidity among children with OSAS is related to upper respiratory tract infections. This may be partially explained by the possibility that in this study we explored morbidity in younger children from day of birth to date of diagnosis. In this age category, the upper respiratory sounds may be wrongly interpreted as small airway disease because clinical and pulmonary function testing of small airway diseases is difficult and not routinely performed in children who are younger than 1 yr (23). Both upper and lower airway diseases in young children may present similar symptoms preceding OSAS, signs that become more obvious after infancy (24). Inflammation occurs in lymphadenoid tissue in older children (25, 26), leading physicians to the diagnosis at a later age. This information may deviate from or mask a physician's decision to refer patients for PSG study.

There is a true link between lower respiratory tract illness and OSA that has not yet been well elucidated in children. This implies that children who have had lower respiratory tract infections should be watched more carefully for OSAS.

Elevation of health care consumption among children with OSAS has already been reported (14). The main factors in elevated total annual costs in children with OSAS and not exceeding 5 yr of age are in-patient admissions and visits to the emergency room, representing more than 50% of total annual costs. These cost elements reflect severe morbidity.

Objective variables such as AHI have little predictive value of health care use in patients with OSAS. Our findings support the evidence that in children (27) and adults (11–14) with OSAS, PSG parameters weakly correlate with outcome measures such as the Epworth Sleepiness Scale (27). The information that PSG findings did not predict health care consumption may result from the fact that AHI in these patients is probably above the threshold for elevated health care use. The increased number of referrals to specialists was the only significant cost element among children with OSAS during the first year of life. It is not clear why the odds for referral to OSAS-related consultants and diagnosis were considerably higher in Year 4, even after excluding ENT consultation before PSG referral. The answer is not straightforward. In fact, as in adults (28), awareness of sleep-disordered breathing symptoms in children by parents and physicians is low (29, 30). Among children with a significant history of snoring, only 8% of parents mentioned this symptom during the concurrent clinical evaluation, and only 15% had done so previously (29).

Health care use in adults with OSAS is related to obesity, alcohol use, caffeine and tobacco consumption, and cardiovascular comorbidity (8, 9, 11), among other risk factors, and probably to low socioeconomic status (31). It is possible that multiple modifiers, including passive smoking, socioeconomic status, and snoring (22, 31), together with predisposing genetic risk factors (32–37) and evolving airway inflammation (26), may contribute to the occurrence of sleep-disordered breathing starting in the first year of life.

Our data on health care use may be difficult to compare with those from other health care systems that have more than one payer, as in the United States. However, our data represent a health care system similar to that in Canada (8–11). The information presented reflects the “true” consumption of health care resources (12–14, 31) of children with OSAS: all PSGs and the relevant medical information regarding patients with OSAS are stored in the only sleep center in the region. CHS uses one billing system to include community and hospital services. According to the National Health Care Law, equal access to medical services is provided to all enrollees and there is no economic incentive to increase consumption of sleep laboratory services due to reimbursement policies. Physicians are paid a capitation fee once every 3 mo per patient and do not have any economic incentive to increase consumption of services (12–14).

The elevated health care costs attributed to lower respiratory tract diseases may represent missed diagnoses of OSAS, suggesting the need for early diagnosis and intervention. This would presumably be the case if an effective treatment is available, as in our system in which equal access to treatment is available and performed (13, 14, 31). Two-thirds of our children underwent surgery. This group of children probably benefited from treatment by improved sleep characteristics, behavior, and psychological and neurocognitive functions (15, 19, 23). In addition, total annual cost savings are maximal among children with AHI > 8 events/h (13). One-third of children were not treated surgically, probably because of a low level of awareness among patients and physicians to the potential benefits of surgery or to OSAS-associated morbidity (13).

The main challenge with these results is to derive a causal inference from the correlative data at hand. One cannot necessarily attribute increased health care use to OSAS per se. It could be that some parents assertively seek health care services for their children, including PSG, consultations, and medications, eventually being diagnosed as having OSAS. However, this possibility is unlikely because all enrollees have equal access to all medical services with no economic barrier (12–14). The ability of patients to pursue medical help may be influenced by socioeconomic status; however, we minimized this effect by selecting control subjects from the same geographic location (31). The ability of parents to pursue medical help may be related to the fact that children with OSAS have nonspecific symptoms including a deleterious effect on cognition, inattention, hyperactivity, and reduction of quality of life (27, 38–42) that can lead parents to engage health care services. Further studies are needed to explore this important issue. Moreover, passive cigarette smoke is a recognized environmental risk factor for snoring-related sleep fragmentation among infants (22). In our study, passive smoking did not increase health care use. This result should be interpreted with caution because the number of smoking parents was too small to enable us to reach a conclusion; therefore, this requires further study. It is unlikely that a practice pattern of individual ENT surgeons affects the “risk for sleep laboratory referral,” because in our region all children with OSAS and control subjects are referred to the same limited group of ENT consulting surgeons.

Children with OSAS present significantly greater morbidity and higher health care use, starting from the first year of life. Most consumed health care resources are related to respiratory diseases. Practitioners should be aware that an increase in airway diseases and health care costs, starting from the first year of life, may be associated with the presence of sleep-disordered breathing. This study suggests the need for early diagnosis and intervention in children with OSAS.