Sociodemographic Disparities and Healthcare Utilization in Pediatric Obstructive Sleep Apnea Management

Abstract

Objectives:

We examined (1) disparities in obstructive sleep apnea (OSA) care by insurance coverage, and by child race and ethnicity among Medicaid-insured children (MIC), and (2) healthcare utilization changes after OSA care.

Methods:

IBM MarketScan insurance claims were used to index OSA care 1-year before and after initial OSA diagnosis in 2017 among 2–17-year-old children (n=31,787, MIC: 59%). OSA care and healthcare utilization analyses adjusted for child age, gender, obesity, and complex chronic conditions.

Results:

We identified 8 OSA care pathways, including no care, which occurred in 34.4% of the overall sample. MIC had 13% higher odds of no OSA care compared to commercially-insured children (CIC). MIC had 32–48% lower odds of any treatment pathway involving specialty care, but a 13–46% higher likelihood of receiving surgical care without polysomnogram (PSG) and PSG only. In MIC, non-Latinx Black/African American (Black) and Hispanic/Latinx children were 1.3–2.2 times more likely than White children to receive treatment involving specialty care and/or PSG, while Black children were 31% less likely than White youth to undergo surgery. In the full sample, surgical care was associated with less outpatient and emergency healthcare utilization compared to those untreated or not surgically treated.

Conclusions:

Varied OSA management by insurance coverage suggests disparities in access to and engagement in care and potentially greater disease burden among MIC. Surgical care is associated with reduced healthcare utilization. The lower odds of surgery in Black MIC should be further evaluated in the context of OSA severity, healthcare biases, and family preferences.

1. Introduction

Up to 5.7% of children are diagnosed with obstructive sleep apnea (OSA),^1,2 making it one of the most common childhood medical conditions.³ Untreated OSA is a public health concern given its associations with infection, cardiometabolic issues, and asthma exacerbations.^1,4,5 OSA is also associated with impairments in neurobehavioral skills that are crucial for academic success.^6–8 Untreated OSA has been linked to a 215% elevation in child healthcare utilization, with 40% more hospital visits than controls.⁹

There are racial, ethnic, and socioeconomic disparities in the incidence, severity, and management of pediatric OSA.^10,11 Much of this research has focused on comparing non-Hispanic/Latinx Black/African American (‘Black’) to non-Hispanic/Latinx White (‘White’) youth, or comparing OSA management by family and/or neighborhood-level socioeconomic status (SES).¹⁰ In addition to research showing an increased prevalence of OSA among Black compared to White youth, the Childhood Adenotonsillectomy Trial (CHAT) identified more severe OSA and less resolution following treatment in Black versus White youth, even when covarying for obesity and other OSA predictors.^7,12 Other studies, including CHAT, have found increased prevalence and severity of OSA in youth living in lower SES homes and neighborhoods, adjusting for child race and ethnicity.^10,13,14 Public insurance as a proxy for lower family SES has also been associated with surgical treatment delays for children with OSA.¹⁵

Mechanisms of these disparities are understudied, but are likely due to multi-level factors.¹⁶ Such factors include exposure to neighborhood toxins¹³ and systemic racism across contexts, including in healthcare settings.^17–19 To this end, there may be barriers to accessing specialty care services based on family resources and/or insurance coverage.^20–22 Prior research has found that surgical treatment of OSA is linked to lower healthcare utilization,^23,24 underscoring the importance of early identification and timely, equitable OSA treatment. Identifying insurance-related and racial and ethnic disparities along the continuum of OSA management, from initial diagnosis to follow-up care, can guide interventions at multiple levels of the healthcare system (e.g., provider-level education, policies to improve healthcare access). However, few studies have examined OSA management by insurance status and in samples large enough to compare across child racial and ethnic groups, including youth of Hispanic/Latinx backgrounds.

This study leveraged large-scale insurance claims data to examine disparities in OSA clinical care in the one year before and after diagnosis among youth ages 2 to 17. Based on previous research, we used Medicaid insurance as a proxy for lower family SES.^10,15 Among Medicaid-insured children (MIC), race and ethnicity were examined as socio-political (rather than genetic or biological) constructs reflecting differential access to care and experiences of ongoing and historical racism and bias.^25,26 In line with previous research, we hypothesized that MIC and Black children would have lower odds of OSA follow-up care relative to commercially insured children (CIC) and to White youth, respectively. We also hypothesized that, as in prior research, surgical care of OSA would be linked to less utilization of outpatient as well as emergency department (ED)/inpatient hospitalization healthcare.

2. Methods

2.1. Data

We used IBM MarketScan Research insurance claims data (Truvan Health Analytics, New York, NY)^27,28 to identify the children with OSA diagnosis (ICD-10 diagnosis codes G47.33, G47.3), OSA unspecified, or other sleep apnea (G47.30, G47.39). Out of 12,394,902 enrolled patients in 2017, 0.7% had OSA diagnosis. OSA prevalence varied by insurance type, with 0.5% of CIC and 0.8% of MIC diagnosed with OSA (p< 0.001). A total of 31,787 children ages 2 to 17 years with a first OSA diagnosis in 2017 and continuously enrolled one-year pre- and post-diagnosis were included in analyses. We opted to use 2017 data to maintain consistency in applying ICD-10 diagnostic codes.

2.2. Variables

Sociodemographic variables included child age in years, gender, and insurance provider (Medicaid or commercial). Race and ethnicity data were available for MIC only; these data were self-reported by patients/families and grouped into categories by IBM. US region was available for CIC only. Obesity and complex chronic conditions (CCCs) during 1-year prior to and at the time of OSA diagnosis were included, given the potential impact of obesity and other medical conditions on OSA diagnosis, follow-up care, and healthcare utilziations.^27,28 CCCs include cardiovascular, neuromuscular, congenital or genetic, technology dependence, gastrointestinal, hematology or immunology, malignancy, metabolic, renal, respiratory, neonatal and transplant categories. We included any other sleep disordered breathing diagnoses that occurred in the year pre- and post-OSA diagnosis, as follows: snoring, central sleep apnea, apnea not otherwise specified, sleep-related hypoxia, sleep-related hypoventilation, and hypoxemia.²⁸

Pediatric guidelines for OSA management include consultation with appropriate specialty care services (i.e., otolaryngology, sleep, or pulmonary medicine), confirmation of suspected OSA via polysomnography (PSG), and, depending on PSG-confirmed OSA severity, treatment via surgical care (adenotonsillectomy, primary/secondary adenoidectomy, or tonsillectomy), positive airway pressure (PAP), or watchful waiting.¹ As MarketScan data do not include OSA severity, we characterized OSA management via 8 distinct treatment pathways, some of which could reflect appropriate watchful waiting or no care following a non-severe or reassuring PSG result. Of note, the MarketScan databases contain 135 provider codes, including otolaryngology and pulmonary medicine, but codes for subspecialties such as sleep medicine are not identified within the provider codes. Thus, our coding of specialty care reflects otolaryngology and pulmonary medicine, where pediatric OSA management is typically provided.

As shown in Figure 1, the pathways included: (1) Surgical care + PSG, reflecting patients who received surgical care in the 1-year post-OSA diagnosis and any PSG in the 1-year prior to, at the time of, or 1-year post-OSA diagnosis; and (2) Surgical care only, reflecting surgical care and no PSG history. Additional pathways reflected the possibility that PSG indicated no OSA or a non-severe/reassuring result: (3) PSG + specialty care, reflecting any PSG in the 1-year prior to, at the time of, or 1-year post-OSA diagnosis as well as pulmonary and/or otolaryngology specialty care visits in the 1-year post-OSA diagnosis; (4) PSG only, reflecting any PSG in the 1-year prior to, at the time of, or 1-year post-OSA diagnosis without any other OSA-related surgical care, specialty care, or PAP provision. Other non-surgical treatment pathways included: (5) Specialty care only, reflecting only pulmonary and/or otolaryngology specialty care visits in the 1-year post-OSA diagnosis, with no PSG history; and (6) PAP care, which included a PAP prescription and/or PSG with PAP titration, with no surgical care. Two additional care pathways were: (7) No care, reflecting no PSG history and no surgical care, specialty care, or PAP care; and (8) Comprehensive care, reflecting patients who received a combination of PSG in the 1-year prior to, at the time of, or 1-year post-OSA diagnosis as well as surgical care, PAP, and/or specialty care. We examined the distinct OSA treatment pathways in analyses of disparities by insurance and, within MIC, by race and ethnicity, except for the Comprehensive care pathway, due to the small sample size (n=202 patients, 0.6% of the overall sample).

Figure 1.

Summary of sample inclusion, exclusion, treatment pathways, and care groups

Note. PSG = polysomnogram; PAP = Positive airway pressure, OSA= obstructive sleep apnea.

(1)PSG + surgical care: Any PSG prior to, at the time of, or post-OSA diagnosis plus surgical care (adenotonsillectomy, tonsillectomy, adenoidectomy).

(2)Surgical care only: Surgical care without any PSG history.

(3)PSG + specialty care: Any PSG prior to, at the time of, or post-OSA diagnosis plus pulmonary and/or otolaryngology visits.

(4)PSG only: Any PSG prior to, at the time of, or post-OSA diagnosis with no surgical care, no PAP, and no specialty care.

(5)Specialty care only: Pulmonary and/or otolaryngology visits only with no PSG, no surgical care, and no PAP.

(6)PAP care: PAP prescription and/or PSG with PAP titration.

(7)No treatment: No PSG history, surgical care, specialty care, or PAP care.

(8)Comprehensive care: Excluded due to small sample size (n=202) and high utilization of OSA-related care (i.e., surgical care, PAP, and/or PSG, and/or specialty care).

To examine change in healthcare utilization after OSA treatment, we first categorized 5 of the distinct pathways into 3 care groups (Figure 1). The (A) Surgical care group reflected patients that followed the (1) Surgical care + PSG pathway or the (2) Surgical care only pathway. The (B) Non-surgical care group was comprised of patients that followed the (5) Specialty care only or (6) PAP care pathways. The (C) No care group included all patients following the (7) No care pathway. Patients on the (3) PSG + specialty care and (4) PSG only pathways were excluded from healthcare utilization analyses, given ambiguity about their follow-up care. Patients on these pathways have had no care following a PSG that showed evidence of OSA, or they could have had PSG results that showed no evidence of OSA or a non-severe/reassuring result that did not require further care.

We then compared the changes in outpatient and ED visits/inpatient hospitalizations in the 1-year before OSA diagnosis and 1-year after OSA treatment among children in the Surgical care, Non-surgical care, and No care groups, as in prior research.²³ Outpatient visits were defined as encounters with primary care/family medicine or with subspecialty providers other than pulmonary and otolaryngology. For both outpatient visits and ED visits/inpatient hospitalizations, the number of encounters were used to reflect healthcare utilization outcomes. Across groups, we excluded those who did not have 1-year data after the first OSA care receipt and those having multiple OSA-related surgeries with more than a month gap in between each surgery (21.0%; see Figure 1). For all those in the Surgical care group, we excluded healthcare utilization data in the 3 months after surgery to avoid confounding medical utilization with perisurgical morbidity.²³ Thus, healthcare utilization was examined within 12 months pre-OSA diagnosis for all groups, while the timing of the 12-month post period differed by group, with the No care group timeframe reflecting 1-year post OSA-diagnosis, the Non-surgical care group timeframe reflecting 1-year post first non-surgical OSA treatment, and the Surgical care group reflecting 1-year post-surgical period (Supplementary Table 1).

2.3. Statistical analysis

We used descriptive statistics to summarize the sociodemographic characteristics of patients with OSA. To examine whether there were disparities in likelihood of following distinct OSA treatment pathways 1-year post-OSA diagnosis in MIC versus CIC, and by race and ethnicity for MIC, we used Chi-square tests and binomial logistic regression models with adjustments for child gender, age, and the presence of obesity and CCCs during the 1-year prior to and at the time of OSA diagnosis. As noted above, the Comprehensive care pathway was excluded from analyses given its small sample size. In analyses comparing race and ethnicity among MIC, White children were selected as the reference group to reflect racial privilege and potentially increased access to unbiased healthcare.^26,29 Data for the racial and ethnic group coded as “Other” by IBM are presented but not interpreted given that we are unable to disaggregate these data according to specific racial and ethnic groups.

We also examined interactions between child age and race and ethnicity to better understand disparities across developmental phases, including early childhood (2–5 years), middle childhood (6–11 years) and adolescence (12–17 years). Mixed-effects negative binomial regression models were used to examine the changes in outpatient and in ED/inpatient hospitalization healthcare utilization before and after OSA treatments after adjustments for covariates. All statistical analyses were conducted using STATA 16 (StataCorp LLC, College Station, TX) and SAS 9.4 (SAS Institute Inc, Cary, NC). This research was deemed exempt by the Institutional Review Board of Children’s Hospital of Philadelphia.

3. Results

3.1. Patient characteristics

Patient sociodemographics are shown in Table 1. Patients were 53.3% male and 59.2% MIC, with a mean age of 7.8 years (SD = 4.2). Among MIC, 53.8% were White, followed by Black (35.2%), Hispanic/Latinx (8.1%), and Other racial and ethnic backgrounds (2.9%). Overall, 93.5% of patients did not have any CCCs. Neuromuscular disorders were the most common CCC (2.5%).

Table 1.

Patient characteristics at the time of first OSA diagnosis in 2017 by insurance provider

	All		Commercial	Medicaid	p-value+
	n=31,787		n=12,965	n=18,822
	n	%	%	%
Age group					<0.001
2–5 years	11860	37.3	40.2	35.4
6–11 years	13264	41.7	38.0	44.3
12–17 years	6663	21.0	21.8	20.4
Age, Mean (SD)	7.8	(4.2)	7.7 (4.4)	7.8 (4.1)	0.003
Gender					<0.001
Male	16935	53.3	54.7	52.3
Female	14852	46.7	45.3	47.7
Race and ethnicitya					--
Hispanic/Latinx	--	--	--	8.1
Non-Hispanic/Latinx Black	--	--	--	35.2
Non-Hispanic/Latinx White	--	--	--	53.8
Other	--	--	--	2.9
Obesity	1600	5.0	3.4	6.2	<0.001
Complex chronic conditions
Cardiovascular	239	0. 8	0.7	0.8	0.17
Neuromuscular	782	2.5	2.2	2.7	0.003
Congenital or genetic	571	1.8	1.7	1.8	0.50
Technology dependence	262	0.8	0.6	1.0	<0.001
Gastrointestinal	212	0.7	0.5	0.8	<0.001
Hematology or immunology	186	0.6	0.4	0.7	<0.001
Malignancy	74	0.2	0.2	0.2	0.97
Metabolic	227	0.7	0.6	0.8	0.008
Renal	51	0.2	0.1	0.2	0.28
Respiratory	222	0.7	0.5	0.9	<0.001
Neonatal	56	0.2	0.1	0.2	0.007
Transplant	4	0.01	0.01	0.02	0.65
Any conditions above	2075	6.5	5.7	7.1	<0.001
Total number (Mean, SD)	0.09	(0.40)	0.07 (0.36)	0.10 (0.43)	<0.001
Regiona					--
Northeast	--	--	21.1	--
North Central	--	--	22.4	--
South	--	--	39.6	--
West	--	--	16.4	--
Unknown	--	--	0.4	--
Visit with OSA diagnosis in 2017b					<0.001
At polysomnography	2121	6.7	7.6	6.1
Pulmonary visit	651	2.0	2.8	1.5
Otolaryngology visit	7655	24.1	33.9	17.3
Primary care visit	3528	11.1	14.1	9.0
Other outpatient specialty or non-specialty care visit	17061	53.7	39.8	63.3
Other inpatient/hospital visit	648	2.0	1.7	2.3
Provider or visit type unknown	128	0.4	0.2	0.5

^aRace and ethnicity were available for Medicaid only, and US region was available for commercial only.

^bVisit with OSA diagnosis was defined by the order of categories listed above, as these categories were not mutually exclusive. For example, 271 patients in the polysomnography category had primary care visits on the same day.

⁺Chi-squared tests and t-test.

Nearly one-quarter of patients (24.1%) were diagnosed with OSA at an otolaryngology visit, but most patients’ OSA diagnoses occurred at other outpatient specialty or non-specialty care (excluding Pulmonary) visits (64.8%) (Table 1). Only 6.7% of OSA diagnoses occurred in the context of a PSG, and 2.0% occurred during a Pulmonary visit. In addition to OSA, snoring was the most commonly diagnosed sleep disordered breathing disorder both in the year prior to the OSA diagnosis (27.7%) and at the same time as the OSA diagnosis (18.8%), followed by sleep-related hypoventilation (pre-OSA: 2.2%) and hypoxemia (pre-OSA: 1.5%). The prevalence of central sleep apnea, sleep-related hypoxia, and apnea-not otherwise specified were all < 1.5% in the 1-year prior to and at the same time as the OSA diagnosis (Supplementary Table 2).

3.2. Variation in OSA treatment pathways by insurance coverage

Of the 8 identified OSA treatment pathways (Figure 1), No care was the most prevalent, with 34.4% of the full sample receiving no OSA care of any kind in the 1-year post-OSA diagnosis, and no history of PSG in the 1-year pre, at the time of, or 1-year post-diagnosis. (Table 2). Surgical care only was the next most prevalent pathway (31.8%), followed by PSG + surgical care (11.4%), and PSG only (6.4%). Compared to MIC, a consistently higher percentage of CIC received Specialty care only, PSG + specialty care, and PAP care, whereas a higher percentage of MIC received Surgical care only and PSG only (Table 2).

Table 2.

Differences in the 1-year OSA treatment pathways by insurance provider (n=31,585)

	Percentage of patients on each treatment pathway			Odds ratio for likelihood of each treatment pathway in Medicaid patients (ref: Commercial)+
	All	Commercial	Medicaid	OR	95% CI
(1) Surgical care + PSG	11.4%	11.3%	11.7%	1.02	.95, 1.09
(2) Surgical care only	31.8%	30.6%	32.9%	1.21	1.15, 1.28
(3) PSG + specialty care	5.9%	8.0%	4.5%	.52	.47, 57
(4) PSG only	6.4%	5.0%	7.3%	1.46	1.32, 1.61
(5) Specialty care only	6.1%	8.1%	4.7%	.57	.52, .63
(6) PAP care	3.4%	4.0%	3.1%	.68	.59, .77
(7) No care	34.4%	33.0%	35.8%	1.13	1.07, 1.18
(8) Comprehensive care	0.6%	--	--	--	--

PAP=positive airway pressure; PSG=polysomnography.

⁺Binomial logistic regression models after adjusting for child gender, age, any complex chronic conditions, and obesity during 1-year prior to and at time of OSA diagnosis. Bold indicates significant associations (p<0.05).

⁽¹⁾PSG + surgical care: Any PSG prior to, at the time of, or post-OSA diagnosis plus surgical care (adenotonsillectomy, tonsillectomy, adenoidectomy).

⁽²⁾Surgical care only: Surgical care without any PSG history.

⁽³⁾PSG + specialty care: Any PSG prior to, at the time of, or post-OSA diagnosis plus pulmonary and/or otolaryngology visits.

⁽⁴⁾PSG only: Any PSG prior to, at the time of, or post-OSA diagnosis with no surgical care, no PAP, and no specialty care.

⁽⁵⁾Specialty care only: Pulmonary and/or otolaryngology visits only with no PSG, no surgical care, and no PAP.

⁽⁶⁾PAP care: PAP prescription and/or PSG with PAP titration.

⁽⁷⁾No treatment: No PSG history, surgical care, specialty care, or PAP care.

⁽⁸⁾Comprehensive care: Excluded due to small sample size (n=202) and high utilization of OSA-related care (i.e., surgical care, PAP, and/or PSG, and/or specialty care).

After adjusting for child age, gender, CCCs, and obesity, MIC were 13% more likely to receive No care compared to CIC (MIC=35.8% versus CIC=33.0%; OR=1.13, 95% CI=1.07–1.18). Although MIC and CIC were equally likely to follow the PSG + surgical care pathway MIC had a higher odds than CIC of receiving Surgical care only (OR [95% CI]=1.21 [1.15, 1.28]), without any PSG, as well as PSG only (OR [95% CI]=1.46 [1.32, 1.61]), without any other OSA care receipt. Compared to CIC, MIC showed 32–48% lower odds of receiving PSG + specialty care (OR [95% CI]=0.52 [0.47, 0.57]), Specialty care only (OR [95% CI]=0.57 [0.52, 0.63]), and PAP care (OR [95% CI]=0.68 [0.59, 0.77]).

3.3. Variation in OSA treatment pathways by child race and ethnicity in MIC

Among MIC, Surgical care only was the most common treatment pathway in the 1-year period after OSA diagnosis, followed by No care (Table 3). Compared to White youth, Black youth had lower odds of undergoing Surgical care only (OR [95% CI]=0.69 [0.64, 0.74]), but higher odds of receiving Surgical care + PSG (OR [95% CI]=1.15 [1.05, 1.27]). No differences in likelihood of following these two pathways emerged when comparing Hispanic/Latinx youth to White youth. However, Hispanic/Latinx youth were less likely to receive PAP care than White youth (OR [95% CI]=0.65 [0.45, 0.93]), as well as less likely to receive No care (OR [95% CI]=0.77 [0.68, 0.97]). However, compared to White youth, Black and Hispanic/Latinx youth were 1.3–2.2 times more likely to follow the PSG only, PSG + specialty care, and Specialty care only pathways.

Table 3.

Racial and ethnic differences in the 1-year OSA treatment pathways among Medicaid patients (n=18,699)

	Percentage of patients on each treatment pathway				Odds ratio (95% CI) for likelihood of each treatment pathway in (ref: White)
	White	Black	Hispanic/Latinx	Other	Black	Hispanic/Latinx	Other
(1) Surgical care + PSG	11.0%	12.6%	11.9%	11.9%	1.15 (1.05, 1.27)	1.05 (.89, 1.25)	1.08 (.82, 1.4)
(2) Surgical care only	36.0%	28.1%	33.6%	32.5%	.69 (.64, .74)	.94 (.84, 1.06)	0.83 (.68, 1.00)
(3) PSG + specialty care	3.6%	5.5%	5.5%	5.6%	1.60 (1.38, 1.86)	1.55 (1.21, 1.98)	1.64 (1.12, 2.41)
(4) PSG only	6.5%	8.6%	8.4%	5.2%	1.36 (1.21, 1.53)	1.30 (1.07, 1.59)	.82 (.55, 1.21)
(5) Specialty care only	3.9%	5.1%	8.0%	5.2%	1.33 (1.15, 1.55)	2.20 (1.78, 2.72)	1.34 (.90, 1.99)
(6) PAP care	3.1%	3.2%	2.3%	2.6%	.98 (.81, 1.18)	.65 (.45, .93)	.88 (.50, 1.54)
(7) No care	35.9%	36.9%	30.3%	37.0%	1.05 (.98, 1.12)	.77 (.68, .87)	1.07 (.89, 1.28)
(8) Comprehensive care	--	--	--	--	--	--	--

Note. PAP=positive airway pressure; PSG=polysomnography. +Seven binomial logistic regression models after adjusting for child gender, age, any CCCs and obesity during 1-year prior to and at time of OSA diagnosis. Bold indicates significant associations (p<0.05).

⁽¹⁾PSG + surgical care: Any PSG prior to, at the time of, or post-OSA diagnosis plus surgical care (adenotonsillectomy, tonsillectomy, adenoidectomy).

⁽²⁾Surgical care only: Surgical care without any PSG history.

⁽³⁾PSG + specialty care: Any PSG prior to, at the time of, or post-OSA diagnosis plus pulmonary and/or otolaryngology visits.

⁽⁴⁾PSG only: Any PSG prior to, at the time of, or post-OSA diagnosis with no surgical care, no PAP, and no specialty care.

⁽⁵⁾Specialty care only: Pulmonary and/or otolaryngology visits only with no PSG, no surgical care, and no PAP.

⁽⁶⁾PAP care: PAP prescription and/or PSG with PAP titration.

⁽⁷⁾No treatment: No PSG history, surgical care, specialty care, or PAP care.

⁽⁸⁾Comprehensive care: Excluded due to small sample size (n=202) and high utilization of OSA-related care (i.e., surgical care, PAP, and/or PSG, and/or specialty care).

3.4. Comparing changes in healthcare utilization across OSA care groups

Across the CIC and MIC youth included in healthcare utilization analyses (Figure 1), 38.4% were in the Surgical care group, 10.8% were in the Non-surgical care group, and 50.8% received No care. Youth in the Surgical care group had the largest drops in the number of outpatient visits from pre to post-OSA care (adjusted pre vs. post difference [SE]= −0.21 [0.01]) compared to youth in the Non-surgical care group (−0.04 [0.02]) and to those in the No care group (−0.14 [0.01]; Table 4; both p values for comparing the Non-surgical care group and No care group vs. the Surgical care group <0.01). Similarly, youth in the Surgical care group had the largest drops in ER visits and inpatient hospitalization (adjusted pre vs. post difference [SE]= −0.29 [0.02]), compared to those in the Non-surgical care group (−0.09 [0.04]) and those in the No care group (−0.20 [0.02}; Table 4; both p values for comparing the Non-surgical care group and No care group vs. the Surgical care group <0.01).

Table 4.

Comparing the change in healthcare utilization 1-year pre-OSA diagnosis and post-OSA treatment vs. 1-year pre and post-OSA diagnosis among no care, non-surgical care, and surgical care groups

	(A) Surgical care group			(B) Non-surgical care group			(C) No care group
	Pre mean (SD)	Post mean (SD)	Adjusted difference, beta (SE)a	Pre mean (SD)	Post mean (SD)	Adjusted difference, beta (SE)a	Pre mean (SD)	Post mean (SD)	Adjusted difference, beta (SE)a
Outpatient visits	3.12 (4.82)	2.55 (4.30)	−.21 (.01) ***	4.77 (7.03)	4.65 (6.37)	−.04 (.02)	2.64 (4.46)	2.33 (4.26)	−.14 (.01) ***
ER visits and inpatient hospitalizations	0.89 (1.57)	0.66 (1.21)	−.29 (.02) ***	0.78 (1.51)	0.71 (1.40)	−.09 (.04) *	0.88 (1.62)	0.70 (1.44)	−.20 (.02) ***

^***p<0.001,

^*p<0.05.

^aMultilevel mixed-effects negative binomial regression model to compare the number of healthcare utilization by OSA treatment after adjusting for child gender, age, insurance type, any CCCs and obesity during 1-year prior to and at time of OSA diagnosis. Bold indicates significant associations. ER= Emergency room. The drops in both healthcare utilization outcomes among the Surgical care group were significantly larger than those in the No care group (outpatient p<0.001, ER and inpatient p=0.003) and Non-surgical care group (outpatient p=0.003, ER and inpatient p<0.001).

^(A)Surgical care group: Patient that followed the Surgical care + PSG or Surgical care only treatment pathways.

^(B)Non-surgical care group: Patients that followed the Specialty care only or PAP care treatment pathways.

^(C)No care group: Patients that followed the No care pathway.

4. Discussion

Using a large insurance claims dataset, we observed that overall, over one-third of pediatric patients diagnosed with OSA in 2017 had no history of PSG evaluation within one year of OSA diagnosis and no management of OSA via surgery, specialty care, or PAP. We also found sociodemographic disparities in the likelihood of following distinct OSA treatment pathways. Although OSA diagnosis was more prevalent in MIC than CIC, MIC were less likely to receive care following diagnosis, with no PSG history. However, among those that received any OSA care, MIC were more likely than CIC to follow the Surgical care only and PSG only treatment pathways. We also found racial and ethnic disparities in OSA management among the MIC subsample, particularly regarding the likelihood of following surgical versus non-surgical pathways, including PSG + specialty care, PSG only, and Specialty care only. Importantly, we also found that overall, surgical management of OSA was associated with greater drops in subsequent healthcare utilization, underscoring the need for equitable OSA management.

4.1. Variation by insurance coverage

Our findings of increased OSA diagnosis in MIC versus CIC align with a systematic review showing greater risk of sleep disordered breathing (SDB) among children of lower-SES backgrounds¹⁰ across a range of indicators, including public insurance,³⁰ living in a lower-SES household,^31,32 lower caregiver educational attainment,^33–35 and living in a US Census-defined lower-SES neighborhood.^13,33,36,37 In our study, the nature of OSA management also varied by insurance coverage across most of the identified treatment pathways. It is heartening that there were no insurance-related differences in the likelihood of having Surgical care + PSG, as this pathways reflects evidence-based care, with gold standard OSA evaluation via PSG.¹ However, even after adjusting for covariates, MIC had a greater likelihood than CIC of receiving no care or receiving surgical care without any PSG history. MIC also had a lower likelihood of receiving any treatment pathway involving specialty care, which corresponds with income-related inequities in access to care across pediatric specialties.³⁸ On the one hand, MIC could have received medical management of OSA via their primary care provider, which was not captured in this study. On the other hand, variation in physician practices could explain these insurance-related disparities.³⁰ In a survey of otolaryngologists in California, only 27% of surgeons indicated they would offer an office appointment to children with government-funded insurance due to administrative burdens, including excessive paperwork and poor reimbursement.³⁰ Different documentation and regulatory requirements could also have contributed to variation in OSA care by insurance type, although many previous studies have used MarketScan data to validly reflect receipt of billable services.³⁹

Variation in the timing and severity of OSA diagnosis could also contribute to differences in care by insurance status. Severity information was not available, but some pathways could reflect PSG evidence of mild OSA that did not require surgical treatment. However, our finding that MIC were more likely than CIC to receive Surgical care only, without specialty care, could reflect increased OSA severity based on clinical (i.e., non-PSG) evaluation among MIC. It could also be that PSG evaluation for MIC had occurred more than 1 year before OSA diagnosis, which our data did not capture. Especially given research showing delays in PSG receipt and surgical treatment for publicly-insured children with SDB,¹⁵ our findings that MIC were more likely to receive PSG only and less likely to receive specialty care treatment could reflect delays in follow-up care rather than MIC have no or mild OSA on PSG.

4.2. Variation by race and ethnicity in Medicaid-insured children

In MIC, we expected lower odds of OSA follow-up care in Black and Hispanic/Latinx children compared to White children.^10,11 Instead, we found that Black and Hispanic/Latinx children with Medicaid were more likely than White children with Medicaid to follow pathways involving specialty care and PSG, but less likely to receive Surgical care only. Prior research similarly shows that Black children in the US are less likely than their White counterparts to undergo adenotonsillectomy and experience longer wait times for this procedure, but are more likely to undergo PSG.^11,40,41 In previous research, Black children have also had more severe OSA on PSG compared to White children,^7,12 with more OSA-related oxygen desaturation and nasal breathing difficulties.⁴² However, Black children had a lower odds of undergoing Surgical care only compared to White children. This finding could suggest clinicians ordered more PSGs in Black children due to the well-documented higher prevalence of SDB in Black compared to White youth,^2,10 or could reflect delays in surgical care. Hispanic/Latinx children were also less likely than White youth to receive No care, which is counter to findings in prior work¹⁰ and should be explored further given underrepresentation of Hispanic/Latinx youth in OSA research.

Socio-cultural variation in family attitudes, beliefs, and decision-making about OSA care could also contribute to these findings. Several recent qualitative studies have examined caregiver perspectives, decision-making, and predispositions around surgical intervention in families of children with SDB.^43–45 In one study, parents predisposed to choose surgical intervention were more likely to be White and to use medical jargon when interacting with the otolaryngologist.⁴⁵ Another study of clinician-family communication in SDB care found that otolaryngologists were less likely to explore the surgery-related emotions of racially and ethnically minoritized families compared to when speaking with White families, suggesting bias in clinician-family interactions.⁴⁴ These findings highlight the importance of assessing family perspectives about their child’s treatment options and medical care team as well as addressing implicit biases in clinician-family treatment planning.^44,45 Indeed, in another study, caregivers tended to view surgical intervention as a “last resort” and referenced trust and confidence in their medical providers as key determinants of treatment decisions.⁴³ Medical mistrust among Black versus White families, stemming from historical and ongoing racism in medicine at multiple levels (i.e., interpersonal, institutional, systemic) could also drive our findings that Black patients were less likely to undergo surgical care, particularly in the absence of PSG evaluation. Future research should examine caregiver perceptions about OSA management and medical mistrust both across and within racial and ethnic groups, as no group is a monolith, and many other individual and family factors could contribute to medical decision-making. It is also necessary to further investigate racism and bias in healthcare practices and clinician-family communication in relation to disparities in surgical and other forms of OSA care.^43–45

A number of environmental factors could additionally contribute to our findings. Such factors include neighborhood-level toxins, irritants, and allergens that trigger nasopharyngeal inflammation found in lower-SES neighborhoods.¹³ For instance, in one study, about 50% of the observed racial disparity in OSA severity was accounted for by increased neighborhood-level socioeconomic disadvantage.¹² Additional studies, especially those with Hispanic/Latinx children and those of other racial and ethnic backgrounds, are necessary to better understand modifiable determinants of disparities in OSA severity, follow-up care, and surgical outcomes in both publicly and privately insured youth.

4.3. Healthcare utilization

Sociodemographic disparities in OSA management, and surgical care in particular, are especially concerning given the adverse impacts of untreated OSA on child health and wellbeing. Surgery is still the first-line treatment for pediatric OSA,¹and we found that outpatient healthcare utilization and ED visits/inpatient hospitalizations dropped more significantly among youth who received surgery for OSA compared to youth with non-surgical care or to those with no care. Those with severe OSA may be more likely to obtain surgery, resulting in improvement of related health concerns post-surgery and, in turn, less healthcare utilization. By contrast, youth who did not receive any OSA care may have more contact with the healthcare system due to the impact of OSA on overall child health. Our findings align with previous research showing reductions in ED visits, hospitalizations, and specialty care visits following adenotonsillectomy for OSA.²³ A quality improvement initiative also found that initiating PAP to treat pediatric OSA was associated with reductions in the number of ED visits, inpatient hospitalizations, and the length of any hospital stay.²⁴ The nature of non-surgical OSA care, which in this study included subspecialty care visits, PAP prescription, and PSG with or without PAP titration, could reflect children engaged in continued OSA evaluation and/or treatment, with subsequent healthcare utilization reductions occurring more than 1-year after initiating this care. Future research should examine healthcare utilization longitudinally, as well as among patients who received non-specialty care management of OSA, including through family medicine or primary care visits.

Future studies on the impact of surgical OSA care and healthcare utilization should also assess family beliefs and any hesitancy around pursuing surgical versus non-surgical care, as these important family factors could inform efforts to increase patient/family-centered educational resources about OSA treatment options.

4.4. Limitations

As our study is limited to care in patients with an initial OSA diagnosis, results are not generalizable to children who have OSA symptoms but have not yet received a diagnosis. OSA was defined according to diagnosis codes rather than PSG results. Our OSA treatment pathways did not capture all possible forms of care, including watchful waiting, intranasal corticosteroids or saline, via family medicine or primary care, which could be appropriate for mild OSA. Misclassification may also have occurred due to shared ICD-10 codes for primary snoring without OSA and snoring as an OSA symptom. Specialty care codes in the MarketScan database are limited by lack of information about subspecialists (i.e., sleep medicine), and some of the follow-up care with otolaryngology and pulmonary medicine could have been related to other concerns, such as asthma. In addition, our data were limited to 1-year pre- and post-OSA diagnosis, and as such children could have received an OSA diagnosis and/or PSG evaluation prior to this timeframe. As noted above, OSA severity data were unavailable and could also explain variation in care.

Race and ethnicity data were only available for MIC, so we could not investigate these potential disparities in CIC. We lacked specific information for the children coded as being of “other” racial and ethnic backgrounds, which precludes any interpretation of results for this likely heterogeneous group. The regional distribution of racial and ethnic groups, including proximity to cities with more specialty care services, could explain some of our findings, but MarketScan data are limited to US region, which is only available for CIC. Additional unmeasured patient/family factors, such as non-English language preference, could also contribute to the likelihood of following different OSA treatment pathways, and should be examined in future research. Nonetheless, our study overcame the limitations of the small samples in previous research by using large-scale claims data to examine child sociodemographic disparities and healthcare utilization in OSA management 1-year postdiagnosis. These results provide important insights into healthcare patterns among pediatric OSA patients, as well as the potential benefits of surgical intervention on healthcare utilization, which can inform future research on interventions to promote equitable OSA care.

4.5. Conclusions

Overall, over one-third of youth in our study received no care following OSA diagnosis, and there were notable disparities in care by insurance coverage and, within Medicaid-insured youth, by race and ethnicity. Equitable access to critical specialists and both surgical and non-surgical treatment options is necessary to ensure comprehensive OSA management. Given that surgical treatment was associated with more reductions in healthcare utilization in the 1-year post-diagnosis, efforts to address OSA care disparities and support evidence-based practices are needed. Collectively, these findings suggest that an in-depth examination of family healthcare preferences, as well as physician and insurance practices, related to OSA decision-making, management, and outcomes is warranted.

Highlights.

• OSA diagnosis is more prevalent in Medicaid-insured vs. commercially-insured children.

• One-third of youth had no OSA care, including no PSG history, post-OSA diagnosis.

• Surgical care is linked to less healthcare utilization 1-year post-diagnosis.

Abbreviations:

Black

non-Hispanic/Latinx Black/African American

CCC

complex chronic conditions

CIC

commercially insured children

CPT

current procedural terminology

MIC

Medicaid-insured children

OSA

obstructive sleep apnea

PAP

positive airway pressure

PSG

polysomnography

SES

socioeconomic status

White

non-Hispanic/Latinx White

Data availability:

The data underlying this article were provided by IBM MarketScan under license/by permission. Data will be shared on request to the corresponding author with permission of IBM MarketScan.