The OSA patient journey: pathways for diagnosis and treatment among commercially insured individuals in the United States

Abstract

Study Objectives:

The aims of this study were to characterize obstructive sleep apnea (OSA) care pathways among commercially insured individuals in the United States and to investigate between-groups differences in population, care delivery, and economic aspects.

Methods:

We identified adults with OSA using a large, national administrative claims database (January 1, 2016–February 28, 2020). Inclusion criteria included a diagnostic sleep test on or within ≤ 12 months of OSA diagnosis (index date) and 12 months of continuous enrollment before and after the index date. Exclusion criteria included prior OSA treatment or central sleep apnea. OSA care pathways were identified using sleep testing health care procedural health care common procedure coding system/current procedural terminology codes then selected for analysis if they were experienced by ≥ 3% of the population and assessed for baseline demographic/clinical characteristics that were also used for model adjustment. Primary outcome was positive airway pressure initiation rate; secondary outcomes were time from first sleep test to initiation of positive airway pressure, sleep test costs, and health care resource utilization. Associations between pathway type and time to treatment initiation were assessed using generalized linear models.

Results:

Of 86,827 adults with OSA, 92.1% received care in 1 of 5 care pathways that met criteria: home sleep apnea testing (HSAT; 30.8%), polysomnography (PSG; 23.6%), PSG-Titration (19.8%), Split-night (14.8%), and HSAT-Titration (3.2%). Pathways had significantly different demographic and clinical characteristics. HSAT-Titration had the highest positive airway pressure initiation rate (84.6%) and PSG the lowest (34.4%). After adjustments, time to treatment initiation was significantly associated with pathway (P < .0001); Split-night had shortest duration (median, 28 days), followed by HSAT (36), PSG (37), PSG-Titration (58), and HSAT-Titration (75). HSAT had the lowest sleep test costs and health care resource utilization.

Conclusions:

Distinct OSA care pathways exist and are associated with differences in population, care delivery, and economic aspects.

Citation:

Wickwire EM, Zhang X, Munson SH, et al. The OSA patient journey: pathways for diagnosis and treatment among commercially insured individuals in the United States. J Clin Sleep Med. 2024;20(4):505–514.

BRIEF SUMMARY

Current Knowledge/Study Rationale: The typical journey toward obstructive sleep apnea (OSA) diagnosis and management requires an individual to complete multiple steps, which can affect health outcomes, costs, and equitable access to care. Large-scale, real-world studies are needed to characterize the OSA patient journey from diagnosis to treatment in the United States.

Study Impact: This study represents the largest study of the OSA patient journey to date and the first to use a national database to characterize common OSA care pathways in the United States. Five distinct care pathways accounted for 92% of OSA care, marked by important differences in patient populations, care delivery, and economic aspects. Care pathways were associated with time to treatment initiation. These results add an important real-world perspective to understanding the OSA journey and provide future directions for research and clinical care.

INTRODUCTION

Obstructive sleep apnea (OSA) is a common and costly chronic medical condition that affects approximately one-fourth of adults in the United States and nearly 936 million individuals worldwide.^1,2 Untreated OSA is strongly associated with adverse cardiovascular, metabolic, neurocognitive, cerebrovascular, and psychiatric outcomes, as well as diminished quality of life.^3–6 These sequelae result in increased economic burden, estimated at $149.6 billion in 2015 in the United States.^7–9 Despite these well-documented consequences of OSA, the vast majority of individuals with OSA remain undiagnosed and untreated.^10,11

In the broader health care landscape, the importance of the patient journey has been increasingly recognized as a vital component of high-quality, cost-effective, patient-centered health care. Understanding the journey is particularly important when considering OSA, because people living with OSA frequently experience multiple barriers to both diagnosis and treatment.^1,10,12–14 For example, a typical journey in OSA diagnosis and management requires a person to complete multiple steps including a physician consultation (generalist or sleep specialist), diagnostic sleep test (in-laboratory polysomnogram [PSG] or home sleep apnea test [HSAT]), provision of results, provider and patient decision-making around therapy, referral to additional providers and/or coordination of third parties such as durable medical equipment suppliers, and initiation of therapy (eg, positive airway pressure [PAP] therapy or oral appliance therapy or other approaches). After initiation of therapy, ongoing clinical follow-up is required to ensure adequacy of therapy and then maintenance of treatment gains.^15–19

Each of these multiple steps in the journey to OSA diagnosis and treatment can affect health outcomes, costs, and equitable access to care. However, little has still been reported about how OSA care is delivered in the real world, with respect to patient journeys and care pathways. The majority of published studies are single-center, interventional studies with small sample sizes that lack diversity in the patient populations.^20–22 Estimates suggest that the total waiting time for diagnosis and treatment with continuous PAP ranges from 2–10 months,^12,23 but very few published reports about duration of time to advance through each step in the pathway exist.²³ Claims data from a large database are uniquely able to provide insight into these questions and to provide practical understanding to help identify gaps in ensuring timely and equitable access to OSA diagnosis and treatment.

To gain insight into the journey from OSA diagnosis to treatment, the objectives of this study were to (1) identify and characterize OSA care pathways in the United States and (2) investigate between-groups differences in populations, care delivery (rates of PAP initiation and time to treatment initiation), and economic aspects. For these analyses, we hypothesized that (1) distinct OSA care pathways exist and (2) between-groups differences are observed in rates of PAP initiation as well as time from diagnostic sleep testing to treatment with PAP.

METHODS

Data source and study population

The data source was the MarketScan Commercial Claims and Encounters-Medicare Supplemental and Coordination of Benefits (CCAE-MDCR) database (Merative, Ann Arbor, Michigan), a large, deidentified, Health Insurance Portability and Accountability Act (HIPAA)-compliant, national administrative database. This is a very large convenience sample (> 45 million current enrollees, > 250 million total enrollees in database) drawn from throughout the United States. Retired individuals with Medicare fee-for-service or Medicare Advantage plans are not included. Using the study period of January 1, 2016–February 28, 2020, we identified adults (age ≥ 18 years) who received their first OSA diagnosis between January 1, 2017, and February 28, 2019 (range for index date). Inclusion criteria were at least 12 months of continuous enrollment in either the CCAE or MDCR database (typically with the same carrier except for certain life-changing events) both before and after the date of first OSA diagnosis (index date) and at least 1 sleep test on or within 12 months before the index date. Exclusion criteria were evidence of PAP treatment within the 12 months prior to the index date, diagnosis of central sleep apnea at any time during the study, and Medicare coverage prior to the age of 65 years. Because this was a retrospective, observational study that utilized deidentified data for individuals who met eligibility criteria, this study was exempted from full Institutional Review Board approval by the Case Western Reserve University, University Hospitals Institutional Review Board. All aspects of this study were conducted in compliance with HIPAA regulations and the HIPAA Omnibus Rule of 2013.

Demographic characteristics

Baseline demographic characteristics including age, sex, and geographic region were derived from MarketScan files. Comorbid conditions present at the index date were considered chronic.

Clinical characteristics

A diagnosis of OSA was defined as the presence of 1 or more physician-assigned International Classification of Diseases, 10th Revision OSA diagnoses (G47.33) in any position on the claim and at least 1 diagnostic sleep test. The sleep test was required to be on the same date as, or within 12 months before, the day of OSA diagnosis. Diagnostic sleep tests were identified using current procedural terminology (CPT) and health care common procedure coding system (HCPCS) codes (Table S1 ^{(145.4KB, pdf)} in the supplemental material). For visits in which a PAP titration code was the only testing code present, this was assumed to be a split-night study and referred to as such. PAP initiation was identified using HCPCS codes for PAP devices and supplies; treatment with oral appliance therapy was identified using the HCPCS code for an oral sleep appliance (Table S2 ^{(145.4KB, pdf)} ). Relevant medical history and comorbidities were identified using International Classification of Diseases, 10th Revision codes in the 12 months prior to and including the index date (Table S3 ^{(145.4KB, pdf)} ).

OSA care pathways

In this study, OSA care pathways were identified by examining all claims for diagnostic sleep tests from 12 months before the date of first OSA diagnosis to PAP initiation or 12 months after diagnosis, whichever came first, for a total of up to 24 months per individual. The type, number, and costs of sleep tests were determined using both professional and facility charges on all dates of service. Raw claims data and summary statistics were reviewed in detail by the investigative team. A multitude of OSA care pathways were identified, but only pathways experienced by at least 3% of the total study cohort were analyzed. Of note, this study examined OSA care pathways only for individuals who had at least 1 sleep test and an OSA diagnosis. Individuals who underwent a sleep test but had no OSA diagnosis were excluded from the study cohort.

Study outcomes

The primary outcome was PAP initiation rate, defined as the proportion of individuals who initiated PAP therapy within 12 months after the first OSA diagnosis. PAP initiation rate and differences in initiation according to demographic characteristics were determined for the total study population; PAP initiation rates by pathway were investigated.

The secondary outcomes were assessed by pathway only in the subset of individuals who initiated PAP treatment within 12 months of OSA diagnosis. Secondary outcomes included care delivery (time to treatment initiation) as well as economic aspects from the payer perspective (payer cost of sleep tests, and health care resource utilization [HCRU] in the year following PAP initiation). Time to treatment initiation was defined as the time from the date of completion of the first sleep test to the date of PAP initiation. The cost of sleep tests was determined from claims and adjusted based on the Consumer Price Index to 2022 US dollars to account for inflation.²⁴ To determine HCRU, individuals who initiated PAP by February 28, 2019, and had at least 12 months of continuous enrollment after PAP initiation were evaluated. For these individuals, the median number of medical encounters (inpatient, outpatient, and emergency visits) and medication claims were determined within 12 months after PAP initiation using the In-patient Admissions/Out-patient Services and Drug Claims tables of the CCAE-MDCR database, respectively. We also analyzed the subgroup of individuals with at least 1 visit or medication claim in the 12 months after PAP initiation.

Statistical analysis

Descriptive statistics were calculated using mean with standard deviation or median with interquartile range for continuous variables and frequencies with percentages for categorical variables. Differences between groups were evaluated using Kruskal–Wallis tests for continuous variables and chi-squared tests for categorical variables. Generalized linear models were adjusted for baseline demographic characteristics and the Charlson Comorbidity Index.²⁵ Diagnoses used in the Charlson Comorbidity Index were defined using International Classification of Diseases, 10th Revision codes, available in Table S3 ^{(145.4KB, pdf)} . Characteristics used for model adjustment are shown in Table 1. Generalized linear models were used to examine potential associations between pathways and time to treatment initiation, expressed by duration ratios, defined as the ratio of time to treatment initiation for each pathway compared to the PSG pathway. Statistical significance was considered at P < .05. All analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, North Carolina).

Table 1.

Patient characteristics.

Characteristica	Overall (n = 86,827)	HSAT (n = 26,736)	PSG (n = 20,508)	PSG-Titration (n = 17,197)	Split-Night (n = 12,812)	HSAT-Titration (n = 2,746)
Sex, n (%)
Male	47,680 (54.9)	15,115 (56.5)	9,549 (46.6)	9,188 (53.4)	8,584 (67.0)	1,637 (59.6)
Female	39,147 (45.1)	11,621 (43.5)	10,959 (53.4)	8,009 (46.6)	4,228 (33.0)	1,109 (40.4)
Age, yearsb
Mean ± SD	49.0 ± 11.8	48.7 ± 11.0	47.7 ± 12.8	50.3 ± 11.8	49.8 ± 11.2	50.9 ± 11.2
Age groups, n (%)b
18–34 years	10,217 (11.8)	2,904 (10.9)	3,309 (16.1)	1,668 (9.7)	1,187 (9.3)	215 (7.8)
35–44 years	18,536 (21.4)	5,991 (22.4)	4,350 (21.2)	3,436 (20.0)	2,703 (21.1)	528 (19.2)
45–54 years	28,102 (32.4)	9,137 (34.2)	6,172 (30.1)	5,542 (32.2)	4,325 (33.8)	944 (34.4)
55–64 years	25,076 (28.9)	7,695 (28.8)	5,518 (26.9)	5,307 (30.9)	3,902 (30.5)	862 (31.4)
65+ years	4,896 (5.6)	1,009 (3.8)	1,159 (5.7)	1,244 (7.2)	695 (5.4)	197 (7.2)
Region, n (%)
Northeast	15,272 (17.6)	6,633 (24.8)	3,098 (15.1)	2,359 (13.7)	1,211 (9.5)	540 (19.7)
North Central	19,125 (22.0)	4,841 (18.1)	4,786 (23.3)	3,878 (22.6)	3,444 (26.9)	927 (33.8)
South	40,859 (47.1)	10,921 (40.9)	10,120 (49.4)	9,949 (57.9)	5,802 (45.3)	1,007 (36.7)
West	11,271 (13.0)	4,315 (16.1)	2,435 (11.9)	964 (5.6)	2,217 (17.3)	270 (9.8)
Unknown	300 (0.4)	26 (0.1)	69 (0.3)	47 (0.3)	138 (1.1)	2 (0.1)
Primary payer, n (%)
Medicare	4,896 (5.6)	1,009 (3.8)	1,159 (5.7)	1,244 (7.2)	695 (5.4)	197 (7.2)
Commercial	81,931 (94.4)	25,727 (96.2)	19,349 (94.4)	15,953 (92.8)	12,117 (94.6)	2,549 (92.8)
Charlson Comorbidity Index
Mean ± SD	0.9 ± 1.5	0.7 ± 1.3	0.9 ± 1.5	1.0 ± 1.6	0.9 ± 1.6	0.9 ± 1.5
Median [25th, 75th]	0.0 [0.0, 1.0]	0.0 [0.0, 1.0]	0.0 [0.0, 1.0]	0.0 [0.0, 1.0]	0.0 [0.0, 1.0]	0.0 [0.0, 1.0]
PAP initiation, n (%)c
Yes	52,809 (60.8)	15,259 (57.1)	7,053 (34.4)	14,207 (82.6)	10,480 (81.8)	2,323 (84.6)

Patient characteristics for the overall cohort and within the top care pathways at baseline. ^aYear (2017–2019), payer plan type, and patient setting (> 99.7% outpatient) were included for adjustment in models. ^bCharacteristic not used for generalized linear modeling due to the overlap of individuals over 65 years in the Medicare payer cohort. ^cPAP initiation rate was defined as the proportion of individuals who initiated PAP therapy within 12 months after the first OSA diagnosis; it was not used for adjustment in models. HSAT = home sleep apnea test, PAP = positive airway pressure, PSG = polysomnography, SD = standard deviation.

RESULTS

The final cohort included 86,827 individuals (Figure 1), of which 54.9% were men and the mean age was 49.0 years (standard deviation: ±11.8). The majority of individuals had commercial insurance (94.4%) (Table 1).

Figure 1. Study population.

OSA = obstructive sleep apnea, PAP = positive airway pressure.

OSA care pathways

This study identified 103 unique OSA care pathways, each comprising from 1 to 14 sleep tests. There were 5 pathways that accounted for 92.1% of the total study population and represented at least 3% of the study cohort: HSAT (30.8%), PSG (23.6%), PSG-Titration (diagnostic sleep test followed by a titration study; 19.8%), Split-night (14.8%), and HSAT-Titration (HSAT followed by a titration sleep test; 3.2%) (Table 1). All had a higher proportion of males than females, except for PSG (46.6% male). All pathways had the greatest proportion of individuals residing in the South (36–58%), followed by the North Central (18–34%) and Northeast (10–25%) regions. Individuals using the PSG-Titration pathway had the highest mean Charlson Comorbidity Index, whereas individuals in the HSAT pathway had the lowest Charlson Comorbidity Index. Hypertension, hyperlipidemia, and obesity were the most common comorbidities in all pathways (Table 2).

Table 2.

Patient comorbidities.

Characteristic	Overall (n = 86,827)	HSAT (n = 26,736)	PSG (n = 20,508)	PSG-Titration (n = 17,197)	Split-Night (n = 12,812)	HSAT-Titration (n = 2,746)
Charlson comorbidities, n (%)
Cancer	3,630 (4.2)	985 (3.7)	911 (4.4)	780 (4.5)	515 (4.0)	118 (4.3)
Cerebrovascular disease	3,765 (4.3)	851 (3.2)	1,016 (5.0)	937 (5.5)	490 (3.8)	137 (5.0)
COPD	14,101 (16.2)	3,846 (14.4)	3,656 (17.8)	3,071 (17.9)	1,991 (15.5)	440 (16.0)
Congestive heart failure	3,500 (4.0)	659 (2.5)	873 (4.3)	913 (5.3)	671 (5.2)	114 (4.2)
Dementia	312 (0.4)	48 (0.2)	108 (0.5)	73 (0.4)	42 (0.3)	10 (0.4)
Diabetes without complications	14,306 (16.5)	3,711 (13.9)	3,005 (14.7)	3,423 (19.9)	2,669 (20.8)	437 (15.9)
Diabetes with complications	3,976 (4.6)	961 (3.6)	829 (4.0)	982 (5.7)	778 (6.1)	138 (5.0)
HIV/AIDS	227 (0.3)	69 (0.3)	65 (0.3)	34 (0.2)	32 (0.2)	7 (0.3)
Mild liver disease	4,815 (5.6)	1,324 (5.0)	1,118 (5.5)	1,070 (6.2)	744 (5.8)	165 (6.0)
Metastatic solid tumor	403 (0.5)	118 (0.4)	89 (0.4)	86 (0.5)	61 (0.5)	13 (0.5)
Moderate or severe liver disease	119 (0.1)	30 (0.1)	28 (0.1)	27 (0.2)	21 (0.2)	2 (0.1)
Myocardial infarction	1,418 (1.6)	272 (1.0)	368 (1.8)	351 (2.0)	266 (2.1)	52 (1.9)
Paraplegia and hemiplegia	440 (0.5)	78 (0.3)	130 (0.6)	101 (0.6)	82 (0.6)	16 (0.6)
Peptic ulcer	589 (0.7)	147 (0.6)	184 (0.9)	118 (0.7)	80 (0.6)	21 (0.8)
Peripheral vascular disease	3,353 (3.9)	780 (2.9)	818 (4.0)	830 (4.8)	493 (3.8)	120 (4.4)
Renal disease	2,724 (3.1)	622 (2.3)	602 (2.9)	687 (4.0)	501 (3.9)	101 (3.7)
Rheumatic disease	1,998 (2.3)	524 (2.0)	571 (2.8)	450 (2.6)	247 (1.9)	57 (2.1)
Other comorbidities, n (%)
Anxiety	18,350 (21.1)	5,537 (20.7)	5,130 (25.0)	3,516 (20.4)	2,149 (16.8)	498 (18.1)
Asthma	9,177 (10.6)	2,634 (9.9)	2,434 (11.9)	1,829 (10.6)	1,256 (9.8)	276 (10.1)
Depression	15,423 (17.8)	4,452 (16.7)	4,318 (21.1)	3,007 (17.5)	1,927 (15.0)	420 (15.3)
GERD	17,070 (19.7)	4,851 (18.1)	4,221 (20.6)	3,833 (22.3)	2,292 (17.9)	461 (16.8)
Hypertension	42,207 (48.6)	11,986 (44.8)	8,900 (43.4)	9,636 (56.0)	7,170 (56.0)	1,457 (53.1)
Hyperlipidemia	32,452 (37.4)	9,468 (35.4)	7,013 (34.2)	7,326 (42.6)	5,104 (39.8)	1,060 (38.6)
Insomnia	12,008 (13.8)	3,117 (11.7)	3,784 (18.5)	2,414 (14.0)	1,423 (11.1)	256 (9.3)
Mild cognitive impairment	441 (0.5)	80 (0.3)	145 (0.7)	105 (0.6)	52 (0.4)	14 (0.5)
Obesity	31,383 (36.1)	8,911 (33.3)	6,872 (33.5)	6,736 (39.2)	5,498 (42.9)	1,068 (38.9)

Patient comorbidities for the overall cohort and within the top care pathways at baseline. AIDS = acquired immunodeficiency syndrome, COPD = chronic obstructive pulmonary disease, GERD = gastroesophageal reflux disease, HSAT = home sleep apnea test, HIV = human immunodeficiency virus, PSG = polysomnography.

Primary outcome: PAP initiation

In the overall sample, 60.8% of participants initiated PAP within the year following OSA diagnosis. Of the individuals who did not initiate PAP, 4.0% initiated oral appliance therapy. PAP initiation rates were significantly higher for individuals with Medicare vs commercial insurance (63.5% vs 60.7%, P < .0001) and for males vs females (64.8% vs 56.0%, P < .0001). The most common initial PAP modality was continuous PAP/automatic PAP (95.1% of individuals), followed by bilevel PAP without back-up rate (4.7%) and bilevel PAP with back-up rate (0.2%).

PAP initiation rate was significantly associated with region (P < .0001): Individuals in the North-Central region had the highest rate of PAP initiation (67.5%), followed by individuals in the West region (60.5%), South region (60.4%), and Northeast region (53.7%). PAP initiation rate was significantly associated with OSA care pathway (P < .0001), with the highest rate for HSAT-Titration (84.6%), followed by PSG-Titration (82.6%), Split-night (81.8%), HSAT (57.1%), and PSG (34.4%) pathways (Figure 2).

Figure 2. PAP initiation rates among the 5 most common OSA care pathways in the United States.

HSAT = home sleep apnea test, OSA = obstructive sleep apnea, PAP = positive airway pressure, PSG = polysomnography.

Secondary outcomes

Secondary outcomes were evaluated only among individuals who initiated PAP.

Time to treatment initiation

The time to treatment initiation (median [interquartile range]) was shortest for Split-night (28 [32] days), followed by HSAT (36 [40] days), PSG (37 [43] days), PSG-Titration (58 [57] days), and HSAT-Titration (75 [61] days) (Figure 3A). After adjustment for baseline demographic and clinical characteristics to account for potential confounding, generalized linear models showed a significant association between pathways and time to treatment initiation. (Figure 3B). Duration ratios indicated that Split-night and HSAT pathways had shorter time to treatment initiation compared with the PSG reference pathway, whereas PSG-Titration and HSAT-Titration had longer time to treatment initiation compared with PSG (all comparisons P < .001).

Figure 3. Time to treatment initiation.

(A) Total time to treatment initiation for individuals who initiated PAP (n = 52,809). Time to treatment initiation was defined as the time from first sleep test to PAP initiation. The boxed regions indicate the interquartile range and the middle horizontal line represents the median (labeled). The whiskers show the range and the diamond indicates the mean. All pairwise comparisons were statistically significant (P < .05). (B) Duration ratios for time to treatment initiation by each care pathway, compared with the PSG care pathway. Generalized linear modeling was used to adjust for patient characteristics. Overall association was significant (P < .0001) and each pathway was significantly different compared with PSG (P < .0001). HSAT = home sleep apnea test, PAP = positive airway pressure, PSG = polysomnography.

Economic aspects

The median sleep testing cost was the lowest for HSAT ($362), followed by Split-night ($1,937), PSG ($1,991), HSAT-Titration ($2,600), and PSG-Titration ($2,644) (Figure 4A). Individuals from the PSG pathway had the highest number of outpatient visits during the 12 months after PAP initiation (median [interquartile range], 23 [19]) while those from the HSAT pathway had the lowest number of outpatient visits (19 [16]). Individuals from the PSG and PSG-Titration pathways had the highest number of medication claims (20 [30] and 20 [31], respectively) whereas those from the HSAT and HSAT-Titration pathways had the lowest number of medication claims (17 [26] and 17 [26], respectively) (Table 3). In the subgroup of individuals with at least 1 visit or medication claim during the 12 months after PAP initiation, individuals from the PSG pathway had the greatest median number of outpatient visits; those from the PSG-Titration pathway had the greatest median number of medication claims (Figure 4B).

Figure 4. Economic impact by OSA care pathway.

Relative to other pathways, the HSAT pathway was associated with significantly fewer outpatient visits (P < .0001). Relative to other pathways (except vs HSAT-titration), the HSAT pathway was associated with significantly fewer medication claims (P < .0001). (A) Cost of sleep tests in each care pathway for individuals who initiated PAP. All pairwise comparisons for individuals who initiated PAP (n = 52,809) were statistically significant except for PSG vs Split-night. The boxed regions indicate the interquartile range and the middle horizontal line represents the median (labeled). The whiskers show the range and the diamond indicates the mean. (B) Health care resource utilization outcomes within 12 months after PAP initiation for individuals who initiated PAP by February 28, 2019, and had at least 12 months of continuous enrollment after PAP initiation with ≥ 1 visit or ≥ 1 medication claim in the 12 months after PAP initiation. HSAT = home sleep apnea test, OSA = obstructive sleep apnea, PAP = positive airway pressure, PSG = polysomnography, USD = US dollars.

Table 3.

Health care resource utilization (HCRU).

Type of HCRU, median (IQR)	HSAT (n = 26,736)	PSG (n = 20,508)	PSG-Titration (n = 17,197)	Split-Night (n = 12,812)	HSAT-Titration (n = 2,746)
Inpatient visits	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
Outpatient visits	19 (16)	23 (19)	22 (19)	21 (17)	21 (16)
Emergency visits	0 (0 (0)0)	0 (0)	0 (0)	0 (0)	0 (0)
Medication claims	17 (26)	20 (30)	20 (31)	19 (30)	17 (26)

Health care resource utilization within 12 months after PAP initiation for 47,481 individuals who initiated PAP by February 28, 2019, and had at least 12 months of continuous enrollment after PAP initiation. Relative to other pathways, the HSAT pathway was associated with significantly fewer outpatient visits (P < .0001). Relative to other pathways (except vs HSAT-Titration), the HSAT pathway was associated with significantly fewer medication claims (P < .0001). HSAT = home sleep apnea test, IQR = interquartile range, PAP = positive airway pressure, PSG = polysomnography.

DISCUSSION

This national analysis is the largest study of the journey from OSA diagnosis to treatment to date and the first to characterize common OSA care pathways in the United States. Five distinct care pathways accounted for the vast majority of OSA care, with important differences in patient populations, as well as clinical characteristics, care delivery, and economic aspects between groups. These results add an important real-world, population health perspective to understanding the OSA journey in the United States and provide meaningful future directions for both research and clinical care.

The most common OSA care pathways were HSAT and PSG, which together accounted for OSA care delivered to over half of the individuals in our national sample (54%). These results are consistent with a recent abstract describing a US claims-based analysis of the journey from OSA testing to treatment.²⁶ In terms of clinical characteristics, individuals in the HSAT pathway had the lowest comorbidity burden, which is consistent with clinical guideline recommendations that HSAT is appropriate for people with a high pretest probability of OSA and low comorbidity burden.^27,28

Care delivery differed across OSA care pathway groups. Interestingly, individuals in the HSAT-Titration pathway (3.2% of sample) demonstrated the highest rates of PAP initiation (84.6%) and also the longest time to PAP initiation (75 days). Given the relatively high comorbidity burden observed among these individuals, these results might indicate that these people have relatively complex OSA that takes substantial time to diagnose, but once diagnosed they are more likely to initiate treatment. By contrast, individuals in the PSG pathway (23.6% of sample) demonstrated the lowest rates of PAP initiation (34.4%) and a moderate time to PAP initiation (37 days). One possible explanation for this finding is that perhaps individuals in the PSG pathway were diagnosed with mild OSA, for which PAP treatment is less strongly recommended than for moderate/severe OSA.²⁹ Further supporting this hypothesis, individuals with mild OSA are also less likely to accept and adhere to PAP treatment.³⁰ Similarly, relative to individuals in other OSA care pathways, those in PSG were more likely to be younger and female and to experience higher prevalence of insomnia and depression, both of which are known to impact PAP initiation.³⁸ Finally, it is worth noting that time to treatment initiation is a potential marker of access to care, which can be influenced by sex, proximity to sleep medicine centers, insurance payer policies, and other factors, not to mention wait times for in-laboratory sleep tests including titrations. These influences should be examined in future studies.

In addition to care delivery, OSA care pathways also differed substantially in terms of economic aspects, including both sleep testing costs and subsequent HCRU over 12 months. We observed significant differences in the cost of sleep tests by pathway, with the HSAT pathway demonstrating the lowest costs. These results align with a recent economic analysis that showed that home-based sleep tests are less expensive than in-laboratory sleep tests from the payer’s perspective.³¹ Previous analyses have also demonstrated the cost-effectiveness of using home-based diagnostic testing as compared to in-laboratory testing from a patient’s (ie, quality of life) perspective.^32,33 Interestingly, we found that individuals in care pathways with more than 1 sleep test demonstrated higher HCRU over the year following PAP initiation compared with individuals in pathways with a single study. One possible explanation for this trend is that more frequent sleep tests are a marker of increased medical complexity that might result in greater HCRU as well as delays in care.

Strengths of the present study include the important and timely research question regarding population-level trends in OSA diagnosis and treatment in the United States. Second, our data source was large and nationally representative, derived from all regions in the United States. This resulted in high generalizability for descriptive analyses as well as adequate statistical power for statistical tests. Finally, we employed a novel, data-driven analytic approach to identifying common OSA care pathways in the real world.

At the same time, results of this study must be interpreted in light of several limitations. Most important, the present analysis captures only a portion of the OSA journey, beginning with the initial diagnostic test. Future work should examine patterns and characteristics of earlier stages in the OSA journey, including how individuals travel from initial suspicion of OSA to formal evaluation and management. Second, a number of limitations to our MarketScan administrative claims data source must be acknowledged. Although MarketScan includes individuals from all regions of the country, all individuals possess commercial insurance and the database does not include Medicaid, Veteran’s Affairs, Indian Health Service, or uninsured patients; generalizability of our findings to these populations is unknown. Most employers who contribute to the database are large employers, so medium and small employers are underrepresented. Likewise, administrative claims databases do not include key patient-level clinical characteristics such as sleep, daytime symptoms, or patient preferences that can affect all aspects of OSA care, including OSA care pathway. Similarly, MarketScan does not include direct measures of race, income, employment, or educational attainment, or other social determinants of health, all of which are known to influence OSA care. Certainly, more nuanced assessment of social determinants of health including area-level influences is an important direction for future research. Third, the commercial insurance plans included in the MarketScan database are heterogeneous, and we were unable to determine the impact of individual payers’ policies (including patient out-of-pockets costs) on OSA testing or treatment. More nuanced assessment of insurance type and payer policies is an important area for future research. Fourth, we were unable to assess facility type (eg, American Academy of Sleep Medicine–accredited sleep center), provider-level characteristics (such as board certification in sleep medicine or midlevel provider), or other system-level variables that are likely to influence OSA care.^34,35 Fifth, we were unable to assess the impact of COVID-19 on OSA care pathway assignment or outcomes because this study did not cover the COVID-19 era; broadly speaking, COVID-19 has increased adoption of telehealth approaches, including sleep medicine clinical care and research.^36,37 Sixth, the reference group for the generalized linear model on time to PAP initiation was chosen to be the PSG care pathway, which had the lowest PAP initiation rate (34.4%). Individuals in the OSA care pathway might have experienced barriers or other influences that could have affected assessment of time to treatment initiation. Seventh, given our large sample sizes, some statistically significant findings would be expected, so we have strived to focus our discussion on clinical meaningful findings. Finally, our retrospective observational design precludes determination of causality.

In conclusion, this national analysis is the largest study to date of the OSA diagnosis and treatment journey and the first to characterize OSA care pathways in the real world. Important between-group differences were observed in terms of patient population, care delivery, and economic aspects. Future research should seek to replicate these pathways in different populations and data sources; identify patient-, provider-, and systems-level influences of pathway assignment and outcomes; and optimize the OSA care pathway experience to maximize clinical outcomes and population health to the benefit of payers, clinicians and, most importantly, people with OSA.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. This study was funded by Primasun, Inc. E.M.W.’s institution has received research support from the American Academy of Sleep Medicine Foundation, Department of Defense, Merck, ResMed, and the ResMed Foundation. E.M.W. has served as a scientific consultant to DayZz, Eisai, EnsoData, Idorsia, Merck, Nox Health, Purdue, Primasun, and ResMed and is an equity shareholder in WellTap. Xuan Zhang and Sibyl H. Munson disclose employment at Boston Strategic Partners, Inc. Adam V. Benjafield discloses employment at ResMed. Shannon S. Sullivan discloses consultancy for Verily Life Sciences. Mojgan Payombar discloses employment at Primasun, Inc. Susheel P. Patil discloses consultancy for Primasun, Inc.

ABBREVIATIONS

CCAE

Commercial Claims and Encounters

CPT

current procedural terminology

HCPCS

health care common procedure coding system

HCRU

health care resource utilization

HIPAA

health insurance portability and accountability act

HSAT

home sleep apnea test

MDCR

Medicare Supplemental and Coordination of Benefits

OSA

obstructive sleep apnea

PAP

positive airway pressure

PSG

polysomnogram