Abstract
Study Objectives:
There is a complex interplay between obstructive sleep apnea (OSA) and type 2 diabetes. There are minimal data regarding the effects of treating OSA with positive airway pressure (PAP) therapy on outcomes and health care resource utilization (HCRU) in patients with OSA and type 2 diabetes. We investigated the impact of PAP adherence on HCRU and costs in this population.
Methods:
A retrospective analysis was conducted with a cohort of OSA patient from a US administrative claims dataset linked to objective device data (AirView, ResMed Corp., San Diego, California). Propensity score matching was used to control for potential imbalance in baseline covariates between PAP-adherent and -nonadherent patients. Newly diagnosed patients with OSA aged ≥ 18 years with type 2 diabetes were included. PAP adherence was defined as meeting Centers for Medicare and Medicaid Services compliance criteria in all 8 90-day periods over 2 years. HCRU was based on the number of all-cause doctor visits, emergency room visits, inpatient hospitalizations, and PAP equipment and supplies.
Results:
In years 1 and 2 of PAP therapy, HCRU was significantly lower in adherent vs nonadherent patients (number/patient for emergency room visits 0.68 ± 1.47 vs 0.99 ± 1.91 [year 1], 0.69 ± 1.43 vs 0.95 ± 1.89 [year 2]; for hospitalizations 0.16 ± 0.58 vs 0.22 ± 0.62 [year 1], 0.15 ± 0.51 vs 0.21 ± 0.74 [year 2]; all P < .001). Changes in estimated total 24-month payments were higher for nonadherent patients ($2,282, 95% confidence interval: $1,368, $3,205).
Conclusions:
Consistent use of PAP therapy over 2 years was associated with decreased HCRU in patients with OSA and type 2 diabetes, strongly suggesting a role for screening and treating OSA in type 2 diabetes.
Citation:
Sterling KL, Cistulli PA, Linde-Zwirble W, et al. Association between positive airway pressure therapy adherence and health care resource utilization in patients with obstructive sleep apnea and type 2 diabetes in the United States. J Clin Sleep Med. 2023;19(3):563–571.
Keywords: obstructive sleep apnea, positive airway pressure, diabetes, adherence, resource utilization, hospitalization
BRIEF SUMMARY
Current Knowledge/Study Rationale: There is a complex interplay between obstructive sleep apnea (OSA) and type 2 diabetes. Minimal data are available regarding the effects of treating OSA with positive airway pressure therapy on outcomes and health care resource utilization in patients with OSA and type 2 diabetes.
Study Impact: This real-world study identified a positive association between adherence to positive airway pressure therapy and reductions in health care resource utilization and associated costs in OSA patients with type 2 diabetes. Patients with type 2 diabetes should be screened for OSA because of the potential for positive airway pressure therapy to decrease health care resource use and improve outcomes.
INTRODUCTION
Obstructive sleep apnea (OSA) is one of the most common chronic diseases and is estimated to affect up to 1 billion people worldwide.1,2 Perturbations such as intermittent hypoxia,3 hypercapnia, frequent arousals, and insufficient sleep cause sympathetic activation and are thought to contribute to cardiometabolic comorbidities, which occur in > 50% of patients with OSA.4–6
There is a complex interplay between OSA and type 2 diabetes.7 Studies consistently demonstrate that the prevalence of OSA is up to 60% in patients with type 2 diabetes and even higher in those who are also obese.4,8 The relative risk of incident diabetes is increased in individuals with vs without OSA.4,9 Furthermore, untreated OSA is associated with poor glycemic control in type 2 diabetes10 and increases the risk of microvascular complications.11,12
Positive airway pressure (PAP) therapy is the first-line treatment for moderate-to-severe OSA.13–15 Randomized controlled trials investigating the effect of PAP in individuals with type 2 diabetes are scarce, have small sample sizes, and report inconsistent results. Several meta-analyses report an inconsistent effect of PAP on glycemic control, possibly due to variable adherence to therapy, as well as differences in baseline glycemic control, study duration, and diabetes medication use.16–18 In a substudy of the Sleep Apnea cardioVascular Endpoints (SAVE) trial19 that randomly assigned patients with OSA and stable cardiovascular disease to PAP plus usual care or usual care alone, PAP treatment of OSA for up to 4 years did not improve glycemic control in patients with preexisting diabetes, although mean nightly PAP usage was only 3.6 hours.20,21 However, PAP has been shown to reduce blood pressure significantly and to improve quality of life in OSA patients with type 2 diabetes.22
Randomized controlled trials evaluating the effects of treating OSA in patients with type 2 diabetes have several limitations, including low adherence to PAP, which is common in patients with diabetes.23 In addition, randomized controlled trials may not be representative of what happens in real life and may be biasing populations agreeing to PAP treatment.24 Both OSA and type 2 diabetes are associated with high health care resource utilization and costs.25–28 Therefore, another way to address whether PAP effectively improves diabetes health trajectories would be to determine whether the economic burden of type 2 diabetes is influenced by PAP therapy adherence. A recent systematic review identified a gap in current understanding of the economic impact of OSA in the presence of diabetes.29
This study investigated whether PAP adherence, based on the Centers for Medicare and Medicaid Services (CMS) compliance, was associated with improved health outcomes and reduced costs in OSA patients with comorbid type 2 diabetes over a 2-year period.
METHODS
Data sources
This study was conducted using a linked dataset that combined administrative claims data and objective individual patient usage data collected from cloud-connected PAP devices. The deidentified payer-sourced (“closed”) adjudicated administrative claims data contain more than 100 geographically dispersed health plans in the United States (licensed from Inovalon Insights LLC, Bowing, Maryland). PAP therapy usage data were obtained from AirView (ResMed Corp., San Diego, California), which collects compliance and therapy data when patients use their PAP devices. Objective PAP data collected in AirView include treatment usage, clinical therapy metrics (such as leak and pressure), and residual respiratory events.30–32 The AirView and Inovalon datasets were linked through a tokenization process via a third-party health data connectivity company. Irreversible encrypted tokens for each dataset were derived from personally identifiable information (eg, name and date of birth). Prior to linkage, each dataset was deidentified, and then the deidentified datasets were linked using the encrypted tokens. The resulting linked database underwent a Health Insurance Portability and Accountability Act (HIPAA) expert evaluation to ensure HIPAA compliance. The study design was reviewed by an Institutional Review Board (Advarra, reference number Pro0004005) and deemed exempt from Institutional Review Board oversight.
Study population
Eligible patients were adults (age ≥ 18 years) with a new OSA diagnosis between June 2014 and April 2018. OSA (International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM]: 327.23 and International Classification of Diseases, 10th Revision, Clinical Modification [ICD-10-CM]: G47.33) was diagnosed within 60 days of a sleep test (Healthcare Common Procedure Coding System 95808, 95810, 95811, G0398–G0400). All patients received PAP therapy (continuous or automatic PAP) using an AirSense 10 device (ResMed Corp., San Diego, California) with ≥ 1 year of claims data prior to the first sleep test and 2 years of claims data after device setup. Type 2 diabetes was identified by the presence of ≥ 2 claim dates with a diagnosis of type 2 diabetes (ICD-9-CM: 250.*0, 250.*2, 357.2, 362.0*, 366.41; ICD-10-CM E11.*) in the year prior to PAP device setup. Patients were excluded if any of the following occurred in the year before PAP device setup: evidence of PAP resupply; presence of diagnosis codes for type 1 diabetes (ICD-9-CM: 250.*1, 250.*3; ICD-10-CM: E10.*); central sleep apnea (ICD-9-CM: 327.21, 327.27; ICD-10-CM: G47.31, G47.37); nocturnal hypoventilation (ICD-9-CM: 327.26; ICD-10-CM: G47.36); pregnancy (ICD-9-CM: 630.*-679.*, 792.3, 796.5, V22.*, V23.*, V27.*, V28.*, V72.42, V91.*; ICD-10-CM: O00.*-O9A.*); end-stage renal disease (ICD-9-CM: 585.6; ICD-10-CM: N18.6); or dialysis use (ICD-9-CM: V45.11; ICD-10-CM: Z99.2).
Adherence to PAP therapy was defined based on CMS criteria, which classifies a patient as compliant if there is device usage for ≥ 4 hours/night on ≥ 70% of nights in a consecutive 30-day period within a 90-day timeframe. For our study, patients were defined as being adherent if they met CMS criteria for eight consecutive 90-day timeframes from device setup. Patients were considered nonadherent if CMS criteria were met in none of the 8 consecutive 90-day timeframes. Patients who met the CMS criteria in some, but not all, timeframes were classified as having “intermediate” adherence (Table S1 (104.1KB, pdf) in the supplemental material) and excluded from analyses so that comparisons of health care usage were made between two distinct adherence pattern groups (ie, nonadherent and adherent). PAP usage of the “intermediate” adherence group widely varied (number of quarters in which patients met CMS criteria ranged from 1–7).
Covariates
Baseline covariates were derived from claims data. These included demographic factors (age, payer, sex, and obesity), comorbidities in the year before PAP setup (coronary artery disease, heart failure, atrial fibrillation, chronic obstructive pulmonary disease, asthma, cancer, anxiety, depression, psychotic mood disorders, other mood disorders, hyperlipidemia, gastroesophageal reflux disease, and hypertension), type 2 diabetes complications in the prior year (retinopathy, nephropathy, neuropathy, peripheral vascular disease, foot ulcers, amputations, circulatory, and other), prior-year health care resource use (all-cause inpatient hospitalization and all-cause emergency room [ER] visits), and diabetes medication utilization. Diabetes-related medication use was evaluated in the 90 days prior to initiation of PAP therapy.
Outcomes
Health care resource utilization was determined based on the number of all-cause doctor (physician and outpatient hospital) visits, all-cause ER visits, all-cause inpatient hospitalizations, and PAP equipment and supplies (from claims data). Health care resource costs used for analyses do not represent charged or paid amounts but rather are calculated costs based on a proprietary Proxy Financials algorithm developed by the claims database provider Inovalon Insights LLC. The Inovalon Insights LLC algorithm is based on CMS Medicare prospective payment system fee schedules. This approach enriches the dataset with relative amounts that reflect Medicare-allowed payments across provider services and treatments. Risk-adjusted models were used to determine the impact of 2-year PAP adherence on the number of all-cause inpatient hospitalizations and all-cause ER visits.
Statistical analysis
Baseline characteristics are presented using descriptive statistics. Models were constructed on the unmatched population to determine the expected mean number of 2-year all-cause hospitalizations and all-cause ER visits using a generalized linear model with a logit link. Goodness of fit was assessed by McFadden’s grouped log-likelihood R squared and explanatory potential by 90th percentile predicted range. The linear predictor from the model was then scaled to create a risk score to show the effects of predictors and quantify the risk group for participants. The number needed to treat to avoid an event (ER visit or hospitalization) was calculated as 1/absolute risk reduction.
A hybrid approach was used to ensure appropriate comparison between adherence groups when assessing the impact of adherence on health care resource use and costs. First, a logistic regression model based on risk of not adhering to PAP therapy was developed using baseline demographics, comorbidities, and prior-year health care resource utilization. All other covariates were not significant and therefore not included in the final model. A propensity score was calculated based on model coefficients and used in greedy matching. Additionally, exact matching of the following baseline variables was performed to ensure balanced groups: age group, sex, payer type, prior all-cause hospitalizations, and prior all-cause ER visits. Finally, to ensure that the matched cohorts resembled the original nonadherent cohort, adherent patients were matched to a randomly down-sampled subgroup of nonadherent patients without replacement. The balance between the matched cohorts was assessed using standardized mean differences of baseline covariates; a standardized mean difference of less than 0.1 indicates negligible difference.33 Differences in health care resource use between matched samples before and after PAP setup were assessed using paired Wilcoxon tests. Difference-in-differences analysis was used to compare changes in mean health care resource use and estimated total payments in the first year immediately after PAP initiation, as well as the estimated total payments in the 2 years after PAP initiation. This comparison was used to identify changes before and after PAP use in the nonadherent group above or below the changes in the adherent group. The analysis was performed using 10,000 bootstrap samples of the difference between post and pre periods for paired differences (nonadherent – adherent) in the matched cohort.34 Statistical analyses were performed using the R statistical package version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria).35
RESULTS
Study population
A total of 32,318 OSA patients with comorbid type 2 diabetes who were treated with PAP therapy were identified (mean age 57.4 years, 46.5% female, 26.7% nonadherent, 61.0% with commercial insurance) (Table 1). Forty percent of the population had at least 1 diabetes-related complication; the mean number of diabetes-related complications was 0.63 and the most common complication was neuropathy (21%). Covariates identified as being associated with greater risk of nonadherence included younger age (18–54 years), Medicaid insurance coverage, diabetes-related complications, being less compliant with diabetes medication (medication fill rates < 9/year), comorbid mood disorders (of which psychotic disorders were associated with the highest nonadherence risk), and comorbid chronic obstructive pulmonary disease. After matching, adherent and nonadherent patient cohorts were well-balanced (all standardized mean differences < 0.1) on all baseline characteristics (Table 1). Diabetes-related medication use in the 90 days prior to starting PAP was similar in the matched adherent and nonadherent cohorts, including the use of sodium-glucose cotransporter 2 inhibitors and glucagon-like peptide-1 receptor agonists (Table S2 (104.1KB, pdf) ). Medication use remained similar during follow-up years (Table S3 (104.1KB, pdf) ).
Table 1.
Baseline characteristics of the study population, overall and for propensity score–matched adherence subgroups.
| Characteristics | Overall (n = 32,318) | Matched Cohort* | ||
|---|---|---|---|---|
| Adherent (n = 3,203) | Nonadherent (n = 3,203) | SMD | ||
| Female sex, n (%) | 15,034 (46.5) | 1,546 (48.3) | 1,546 (48.3) | 0 |
| Age, years | 57.4 ± 11 (57) | 56.7 ± 10.6 (57) | 56.3 ± 11.2 (56) | |
| Age group, n (%) | ||||
| 18–54 years | 12,455 (38.5) | 1,388 (43.3) | 1,388 (43.3) | 0 |
| 55–69 years | 15,300 (47.3) | 1,421 (44.4) | 1,421 (44.4) | 0 |
| ≥70 years | 4,563 (14.1) | 394 (12.3) | 394 (12.3) | 0 |
| Payer, n (%) | ||||
| Commercial | 19,716 (61.0) | 1,752 (54.7) | 1,752 (54.7) | 0 |
| Medicare Advantage | 6,459 (20.0) | 891 (27.8) | 891 (27.8) | 0 |
| Medicaid | 6,143 (19.0) | 560 (17.5) | 560 (17.5) | 0 |
| Obesity, n (%) | ||||
| Morbidly obese | 12,731 (39.4) | 1,319 (41.2) | 1,230 (38.4) | 0.06 |
| Obese | 9,529 (29.5) | 908 (28.3) | 912 (28.5) | <0.01 |
| No listed obesity | 10,058 (31.1) | 976 (30.5) | 1,061 (33.1) | 0.06 |
| Comorbidities, n (%) | ||||
| Mean number† | 1.9 | 2.0 | 2.0 | 0.03 |
| Coronary artery disease | 7,954 (24.6) | 795 (24.8) | 801 (25.0) | <0.01 |
| Heart failure | 4,368 (13.5) | 441 (13.8) | 487 (15.2) | 0.04 |
| Cerebrovascular disease | 2,995 (9.3) | 274 (8.6) | 341 (10.6) | 0.07 |
| Asthma | 6,243 (19.3) | 649 (20.3) | 688 (21.5) | 0.03 |
| COPD | 5,475 (16.9) | 606 (18.9) | 640 (20.0) | 0.03 |
| Pneumonia | 2,162 (6.7) | 243 (7.6) | 210 (6.6) | 0.04 |
| Depression | 8,153 (25.2) | 884 (27.6) | 911 (28.4) | 0.02 |
| Anxiety | 6,667 (20.6) | 690 (21.5) | 732 (22.9) | 0.03 |
| Psychotic disorders | 1,656 (5.1) | 188 (5.9) | 189 (5.9) | <0.01 |
| Other mood disorders | 2,401 (7.4) | 244 (7.6) | 273 (8.5) | 0.03 |
| Hypertension | 28,164 (87.1) | 2,768 (86.4) | 2,794 (87.2) | 0.02 |
| Hyperlipidemia | 25,794 (79.8) | 2,526 (78.9) | 2,513 (78.5) | <0.01 |
| Gastroesophageal reflux disease | 10,335 (32.0) | 1,081 (33.7) | 1,115 (34.8) | 0.02 |
| Diabetes complications, n (%) | ||||
| Mean number | 0.63 | 0.68 | 0.70 | 0.03 |
| Any complication | 12,925 (40.0) | 1,370 (42.8) | 1,383 (43.2) | <0.01 |
| Retinopathy | 3,131 (9.7) | 339 (10.6) | 323 (10.1) | 0.02 |
| Nephropathy | 3,725 (11.5) | 440 (13.7) | 350 (10.9) | 0.09 |
| Neuropathy | 6,800 (21.0) | 734 (22.9) | 793 (24.8) | 0.04 |
| Circulatory | 1,865 (5.8) | 184 (5.7) | 227 (7.1) | 0.05 |
| Peripheral vascular disease | 2,150 (6.7) | 202 (6.3) | 256 (8.0) | 0.07 |
| Foot ulcers | 570 (1.8) | 55 (1.7) | 75 (2.3) | 0.04 |
| Diabetes medications‡, n (%) | 26,915 (83.3) | |||
| ≥ 1 oral or noninsulin medication | 18,646 (69.3) | 1,924 (70.7) | 1,903 (69.6) | 0.01 |
| ≥ 1 insulin medication | 5,352 (19.9) | 545 (20.0) | 606 (22.1) | 0.05 |
| None | 7,166 (26.6) | 680 (25.0) | 714 (26.1) | 0.03 |
Values are mean ± SD (median) or number of patients (%). *Matched cohort created based on propensity model (risk of not adhering to the positive airway pressure therapy model) and exact matching on age group, sex, payer type, prior all-cause hospitalizations, and prior all-cause emergency room visits. †Excluding hyperlipidemia, hypertension, and obesity. ‡Denominators for percentages represent the number of patients with available medication information. COPD = chronic obstructive pulmonary disease, SMD = standardized mean difference.
PAP usage
Mean PAP usage during the first year of therapy for the overall population was 4.1 days/week, 3.7 hours/day, and 5.3 hours per use day. Patients who adhered to PAP used the device almost every day of the week (mean 6.7 days/week) and consistently for 7 hours each day based on both hours per day or hours per use day (6.8 and 7.1, respectively). In contrast, matched nonadherent patients used their PAP devices infrequently (0.9 days/week) and for a short period of time on days when the device was used (0.4 hours/day and 3.0 hours/use day).
Diabetes complications
Compared with baseline, the number of diabetes-related complications increased in the year after PAP therapy initiation in both the adherent and nonadherent matched cohorts, with no statistically significant between-group difference (mean 0.76 vs 0.80, P = .20) (Table S4 (104.1KB, pdf) ). Approximately 20% of patients in both cohorts who did not have complications at baseline had an emergent complication in the year after PAP therapy initiation; the most common incident complications in the adherent and nonadherent cohorts were neuropathy (8.4% and 9.3%) and nephropathy (4.4% and 4.6%) (Table S4 (104.1KB, pdf) ). For patients with complications at baseline, 29.6% of the adherent cohort and 33.0% of the nonadherent cohort had an emergent complication in the year after PAP initiation (Table S4 (104.1KB, pdf) ). Statistical comparisons were not performed on subgroups due to covariate imbalance when limiting to subgroup comparisons.
Risk-adjusted outcomes
The results of models predicting the number of 2-year all-cause hospitalizations and all-cause ER visits based on PAP adherence are shown in Figure 1. Being adherent to PAP was significantly associated with reduced number of all-cause hospitalizations (P < .001), with a number needed to treat of 4. The magnitude of benefit varied by patient health state, from a number needed to treat of 23.2 in those with a score of 0–3 (low hospitalization risk; 30% of the population) to a number needed to treat of 2.7 for those with a score > 8 (high hospitalization risk; 21% of the population).
Figure 1. Impact of positive airway pressure therapy adherence on the number of all-cause hospitalizations (A) and all-cause emergency room (ER) visits (B).
Red and green solid lines represent values predicted by the respective health care resource risk model. Blue vertical bars represent number of patients in each risk group. Dots represent the actual mean values from the observed data, with error bars representing the 90% confidence interval. Q1 = first quartile, Q2 = second quartile, Q3 = third quartile, Q4 = fourth quartile.
Being adherent to PAP also significantly reduced the number of ER visits (P < .001), with an average of 0.7 ER visits avoided per adherent patient. Again, the magnitude of benefit varied by patient health state, from 1 ER visit avoided for every 4.7 treated in low-risk patients (score of 0–2; 34% of the population) to 2 ER visits avoided for every adherent patient in the highest risk group (score ≥ 8; 23% of the population).
Two-year health care resource utilization
The number of ER visits and inpatient hospitalizations in year 1 and year 2 of PAP usage was significantly lower in adherent vs nonadherent patients from the matched cohort (Table 2 and Figure 2). Reductions in health care resource utilization seen in year 1 were maintained in year 2. Similarly, the difference-in-difference analyses demonstrated that in the first year of PAP use nonadherent patients had a significantly larger increase in ER visits per person (mean 0.30, 95% confidence interval [CI]: 0.23, 0.37) and inpatient hospitalizations (mean 0.06, 95% CI: 0.04, 0.09) and fewer doctor visits (mean −1.21, 95% CI: −1.60, −0.81). There was a 0.7 difference in doctor visits per patient between adherent and nonadherent patients in PAP year 1 when analyzing doctor visits associated with an OSA diagnosis, accounting for more than half of the difference-in-difference.
Table 2.
Health care resource use in the unmatched and matched cohorts.
| Resource Use (Mean Number per Patient) | Unmatched | Matched* | |||
|---|---|---|---|---|---|
| Adherent (n = 10,235) | Nonadherent (n = 8,622) | Adherent (n = 3,203) | Nonadherent (n = 3,203) | P† | |
| Year before PAP | |||||
| Doctor visits | 10.24 ± 8.00 | 13.70 ± 11.15 | 11.93 ± 9.44 | 13.05 ± 10.90 | <.001 |
| ER visits | 0.57 ± 1.13 | 1.30 ± 2.47 | 0.91 ± 1.49 | 0.92 ± 1.62 | .21 |
| Inpatient hospitalizations | 0.17 ± 0.52 | 0.27 ± 0.75 | 0.22 ± 0.59 | 0.22 ± 0.59 | .32 |
| PAP year 1 | |||||
| Doctor visits | 11.0 ± 9.0 | 14.0 ± 12.0 | 13.13 ± 10.99 | 13.05 ± 11.36 | .76 |
| ER visits | 0.46 ± 1.09 | 1.26 ± 2.57 | 0.68 ± 1.47 | 0.99 ± 1.91 | <.001 |
| Inpatient hospitalizations | 0.13 ± 0.51 | 0.27 ± 0.79 | 0.16 ± 0.58 | 0.22 ± 0.62 | <.001 |
| PAP year 2 | |||||
| Doctor visits | 10.46 ± 9.25 | 12.31 ± 11.42 | 12.31 ± 11.15 | 11.69 ± 10.59 | .99 |
| ER visits | 0.47 ± 1.08 | 1.20 ± 2.49 | 0.69 ± 1.43 | 0.95 ± 1.89 | <.001 |
| Inpatient hospitalizations | 0.12 ± 0.48 | 0.26 ± 0.90 | 0.15 ± 0.51 | 0.21 ± 0.74 | <.001 |
Values are mean ± SD. *Matched cohort created based on propensity model (risk of not adhering to PAP therapy model) and exact matching on age group, sex, payer type, prior all-cause hospitalizations, and prior all-cause ER visits. †P values are for the adherent vs nonadherent comparison in the matched population. ER = emergency room, PAP = positive airway pressure.
Figure 2. Mean number of hospitalizations (A) and emergency room (ER) visits (B) in the different patient cohorts.
Error bars represent 90% confidence intervals. Blue, orange, and gray vertical bars represent the mean number of health care resource uses per patient in cohort. PAP = positive airway pressure.
Baseline health care resource costs were similar between matched adherent and nonadherent patients (Table S5 (104.1KB, pdf) ). In the matched cohort, adherent patients had lower ER visit costs in PAP year 1 (P < .001) and PAP year 2 (P < .001) compared with nonadherent patients; there was a similar trend for inpatient hospitalization costs (Table S5 (104.1KB, pdf) ). OSA-related test costs prior to PAP therapy initiation were similar in the adherent and nonadherent groups (P = .62) but, as expected, equipment-related costs in PAP years 1 and 2 were significantly higher for adherent vs nonadherent patients (year 1: $1,246 vs $794, P < .001; year 2: $509 vs $97, P < .001). The difference-in-difference analysis of estimated total payments for the first year after PAP therapy initiation demonstrated statistically higher estimated total payments for the nonadherent cohort: $675 (95% CI: $107, $1,249) and higher payments when OSA-related expenses are not included ($1,135, 95% CI: $565, $1,702) (Table 3). Greater differences in estimated total payments were observed when assessing the 24-month period after PAP therapy initiation with or without OSA-related expenses included ($1,410, 95% CI: $500, $2,335; without OSA-related expenses: $2,282, 95% CI: $1,368, $3,205).
Table 3.
Difference-in-differences for estimated total payments in matched cohort.
| Estimated Total Payments ($) | Year Before PAP (Nonadherent – Adherent) | Year(s) After PAP (Nonadherent – Adherent) | Difference-in-Differences* |
|---|---|---|---|
| 12 months after PAP | |||
| Estimated total payments | 47 (−340, 431) | 722 (216, 1,226) | 675 (107, 1,249) |
| Estimated total payments – no OSA† | 39 (−342, 419) | 1,174 (671, 1,677) | 1,135 (565, 1,702) |
| 24 months after PAP | |||
| Estimated total payments | 47 (−340, 431) | 1,457 (518, 2,364) | 1,410 (500, 2,335) |
| Estimated total payments – no OSA† | 39 (−342, 419) | 2,321 (1,386, 3,223) | 2,282 (1,368, 3,205) |
Values are mean (95% confidence interval). *Difference-in-differences estimates of mean estimated total payments using 10,000 bootstrap samples of the difference between post and pre year for paired differences (nonadherent – adherent) in the matched cohort. The 95% bootstrap confidence interval for the estimate is also shown. †Does not include sleep test and apnea equipment estimated payments. OSA = obstructive sleep apnea, PAP = positive airway pressure.
DISCUSSION
Our study identified a positive association between adherence to PAP therapy and reductions in health care resource utilization in patients with OSA and comorbid type 2 diabetes. These findings highlight an opportunity that may exist for improved health outcomes for patients treated with PAP therapy, particularly since health care utilization by patients with diabetes has been estimated to be high, $327 billion in 2017.36–39 Adherence to PAP therapy was associated with the avoidance of 4 all-cause hospitalizations and 0.7 all-cause ER visits per adherent patient on average. These values were favorable and suggest the value of encouraging nonadherent patients to adhere to therapy, especially in high-risk type 2 diabetes patients, where benefits appeared to be even greater.
To our knowledge, this is the largest health care economic analysis of OSA patients with comorbid type 2 diabetes. Previous analyses have identified a positive association between PAP adherence and both glycemic and blood pressure control.17,40–44 In a proof-of-concept mechanistic study, PAP treatment in a laboratory setting to ensure full compliance for 1 week improved 24-hour mean glucose levels, especially overnight and in those with poor glycemic control at baseline.40 The duration of PAP usage might play a significant role in glucose metabolism because OSA during rapid eye movement sleep was more strongly related to glycemic control than OSA during non-rapid eye movement sleep.41 Babu et al analyzed sleep clinic patients and found that PAP usage of > 4 hours/day was associated with glycemic control, but no such association was found when PAP usage was < 4 hours/day.42 A retrospective analysis of 1,295 sleep clinic medical records in type 2 diabetes with OSA showed that adherent PAP users had significantly lower blood pressure than nonadherent PAP users and demonstrated a dose response with greater hours of PAP usage.43 A UK case-control study also reported better blood pressure and glycemic control during PAP therapy and determined that PAP treatment in OSA patients with type 2 diabetes was a cost-effective use of National Health Services resources.44
The findings of our study help to address a gap identified in a 2019 systematic review29 regarding the economic impact of OSA in the presence of comorbid diseases, including diabetes. Our results are in alignment with previous studies suggesting that PAP adherence was associated with a reduction in health care cost in acute care and inpatient settings, and in overall health care utilization in patients with OSA, with and without preexisting cardiovascular disease or associated chronic obstructive pulmonary disease (overlap syndrome).36 We further extend previous findings by including a longer 2-year duration of PAP use, incorporating for the first time in the literature objective PAP usage metrics to define adherence and focusing on patients with high documented health care utilization.36–39 The use of a linkage of data from a national claims database and objective PAP usage data is a key strength of this study. In contrast, previous studies have often estimated PAP adherence using surrogate measures rather than objectively recorded device usage. Based on its strengths and taken together with previous studies, our data support the benefits of OSA treatment in patients with type 2 diabetes.
There are also some limitations of our study worth noting. While we were able to use statistical methods to control for many potential differences in baseline characteristics of the patient population, some minor imbalances remained after matching, notably for obesity, fills of insulin medications, and some diabetes-related complications. Additionally, due to the imbalance of subgroups, we were unable to conduct statistical comparisons to assess the association of PAP adherence with diabetic complications. Further research is needed to understand this relationship. It is also important to note that medication costs were not available and therefore were not included in total cost estimates. However, given the similarities in diabetes medication use for adherent and nonadherent patients throughout the study, we believe this is unlikely to have substantially affected results around the differences in total costs. Other potentially important factors in a patient’s overall health, including lifestyle factors (eg, smoking, alcohol intake, nutrition, physical activity, and overall sleep duration), laboratory tests, and patient-reported outcomes are not captured in our data.
Our dataset did not include Medicare fee-for-service patients, potentially limiting the generalizability of our findings. There was also heterogeneity in adherence across payer types observed in the data, with Medicaid patients being less likely to adhere. This may reflect differences in the quality of follow-up pathways. Given this, further research is needed to better understand the association between payer and adherence.
Finally, use of the CMS definition of compliance for adherence in our study is conservative as patients had to achieve CMS compliance in all 8 quarters over a 2-year period. Thus, our adherent cohort represents a phenotype of highly adherent patients and were compared with those who had very limited PAP usage. Further research is needed to understand whether there is a dose–response relationship between PAP usage and health outcomes. This would allow for better understanding of patients with intermediate adherence, who were excluded from the current analysis, as their usage varied widely enough that we were unable to identify an appropriate cut-off point for classification.
CONCLUSIONS
This real-world analysis demonstrated a positive association between adherence to PAP therapy and reductions in all-cause hospitalizations, all-cause ER visits, and corresponding health care costs in patients with OSA and comorbid type 2 diabetes compared to patients who are not adherent to PAP therapy. This finding provides additional evidence to support clinical screening and treating type 2 diabetes patients for OSA and the importance of encouraging patients to be adherent to prescribed PAP therapy.
DISCLOSURE STATEMENT
All authors have reviewed and approved the manuscript. This study was funded by ResMed. Drs. Sterling, Malik, Benjafield, Armitstead, and Nunez; Ms. Cole; and Mr. More, Emami, and Woodford are employees of ResMed. Mr. Linde-Zwirble has received fees from ResMed for consultancy services. Prof. Cistulli has an appointment to an endowed academic Chair at the University of Sydney that was established from ResMed funding; has received research support from ResMed, SomnoMed, and Zephyr Sleep Technologies; is a consultant to ResMed, SomnoMed, Signifier Medical Technologies, and Bayer; and has a pecuniary interest in SomnoMed related to a 2004 role in research and development. Prof. Malhotra is funded by the National Institutes of Health. Prof. Malhotra received funding from Equillium, Corvus, Jazz, and Livanova for medical education. ResMed, Inc. provided a philanthropic donation to the University of California San Diego in support of a sleep center. Prof. Reutrakul reports no conflicts of interest but does receive funding from the National Institutes of Health (R01EY029782). Prof. Pépin is partially supported by MIAI @ Grenoble Alpes (ANR-19-P3IA-0003), and his department has received research support from ResMed, Philips Respironics, and Fisher and Paykel.
ACKNOWLEDGMENTS
The authors thank Nicola Ryan, Ann Cameron, and Naomi Alpert for editing support, formatting, and presentation of the manuscript. MedXcloud is an academic–industry collaboration involving global academic thought leaders in the fields of sleep and respiratory medicine and employees and consultants of ResMed.
ABBREVIATIONS
- CI
confidence interval
- CM
clinical modification
- CMS
Centers for Medicare and Medicaid Services
- ER
emergency room
- HIPAA
Health Insurance Portability and Accountability Act
- ICD
International Classification of Diseases
- OSA
obstructive sleep apnea
- PAP
positive airway pressure
REFERENCES
- 1. Lévy P , Kohler M , McNicholas WT , et al . Obstructive sleep apnoea syndrome . Nat Rev Dis Primers. 2015. ; 1 ( 1 ): 15015 . [DOI] [PubMed] [Google Scholar]
- 2. Benjafield AV , Ayas NT , Eastwood PR , et al . Estimation of the global prevalence and burden of obstructive sleep apnoea: a literature-based analysis . Lancet Respir Med. 2019. ; 7 ( 8 ): 687 – 698 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Thanaviratananich S , Cheng H , Chirakalwasan N , Reutrakul S . Association between nocturnal hypoxemic burden and glucose metabolism [published online ahead of print, 2021. Aug 14]. Sleep Breath 10.1093/sleep/zsab072.476 . [DOI] [PubMed]
- 4. Reutrakul S , Mokhlesi B . Obstructive sleep apnea and diabetes: a state-of-the-art review . Chest. 2017. ; 152 ( 5 ): 1070 – 1086 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ryan S , Arnaud C , Fitzpatrick SF , Gaucher J , Tamisier R , Pépin JL . Adipose tissue as a key player in obstructive sleep apnoea . Eur Respir Rev. 2019. ; 28 ( 152 ): 190006 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ryan S , Cummins EP , Farre R , et al . Understanding the pathophysiological mechanisms of cardiometabolic complications in obstructive sleep apnoea: towards personalised treatment approaches . Eur Respir J. 2020. ; 56 ( 2 ): 1902295 . [DOI] [PubMed] [Google Scholar]
- 7. Subramanian A , Adderley NJ , Tracy A , et al . Risk of incident obstructive sleep apnea among patients with type 2 diabetes . Diabetes Care. 2019. ; 42 ( 5 ): 954 – 963 . [DOI] [PubMed] [Google Scholar]
- 8. Foster GD , Sanders MH , Millman R , et al. Sleep AHEAD Research Group . Obstructive sleep apnea among obese patients with type 2 diabetes . Diabetes Care. 2009. ; 32 ( 6 ): 1017 – 1019 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Kendzerska T , Gershon AS , Hawker G , Tomlinson G , Leung RS . Obstructive sleep apnea and incident diabetes. A historical cohort study . Am J Respir Crit Care Med. 2014. ; 190 ( 2 ): 218 – 225 . [DOI] [PubMed] [Google Scholar]
- 10. Kent BD , Grote L , Ryan S , et al . Diabetes mellitus prevalence and control in sleep-disordered breathing: the European Sleep Apnea Cohort (ESADA) study . Chest. 2014. ; 146 ( 4 ): 982 – 990 . [DOI] [PubMed] [Google Scholar]
- 11. Leong WB , Jadhakhan F , Taheri S , Thomas GN , Adab P . The association between obstructive sleep apnea on diabetic kidney disease: a systematic review and meta-analysis . Sleep. 2016. ; 39 ( 2 ): 301 – 308 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Zhu Z , Zhang F , Liu Y , Yang S , Li C , Niu Q , Niu J . Relationship of obstructive sleep apnoea with diabetic retinopathy: a meta-analysis . BioMed Res Int. 2017. ; 2017 : 4737064 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Morgenthaler TI , Aurora RN , Brown T , et al. American Academy of Sleep Medicine . Practice parameters for the use of autotitrating continuous positive airway pressure devices for titrating pressures and treating adult patients with obstructive sleep apnea syndrome: an update for 2007. An American Academy of Sleep Medicine report . Sleep. 2008. ; 31 ( 1 ): 141 – 147 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Gay P , Weaver T , Loube D , Iber C ; Positive Airway Pressure Task Force Standards of Practice Committee American Academy of Sleep Medicine . Evaluation of positive airway pressure treatment for sleep related breathing disorders in adults . Sleep. 2006. ; 29 ( 3 ): 381 – 401 . [DOI] [PubMed] [Google Scholar]
- 15. Rotenberg BW , Murariu D , Pang KP . Trends in CPAP adherence over twenty years of data collection: a flattened curve . J Otolaryngol Head Neck Surg. 2016. ; 45 ( 1 ): 43 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Shang W , Zhang Y , Wang G , Han D . Benefits of continuous positive airway pressure on glycaemic control and insulin resistance in patients with type 2 diabetes and obstructive sleep apnoea: a meta-analysis . Diabetes Obes Metab. 2021. ; 23 ( 2 ): 540 – 548 . [DOI] [PubMed] [Google Scholar]
- 17. Labarca G , Reyes T , Jorquera J , Dreyse J , Drake L . CPAP in patients with obstructive sleep apnea and type 2 diabetes mellitus: systematic review and meta-analysis . Clin Respir J. 2018. ; 12 ( 8 ): 2361 – 2368 . [DOI] [PubMed] [Google Scholar]
- 18. Zhu B , Ma C , Chaiard J , Shi C . Effect of continuous positive airway pressure on glucose metabolism in adults with type 2 diabetes: a systematic review and meta-analysis of randomized controlled trials . Sleep Breath. 2018. ; 22 ( 2 ): 287 – 295 . [DOI] [PubMed] [Google Scholar]
- 19. McEvoy RD , Antic NA , Heeley E , et al. SAVE Investigators and Coordinators . CPAP for prevention of cardiovascular events in obstructive sleep apnea . N Engl J Med. 2016. ; 375 ( 10 ): 919 – 931 . [DOI] [PubMed] [Google Scholar]
- 20. Loffler KA , Heeley E , Freed R , et al. SAVE Substudy Investigators . Continuous positive airway pressure treatment, glycemia, and diabetes risk in obstructive sleep apnea and comorbid cardiovascular disease . Diabetes Care. 2020. ; 43 ( 8 ): 1859 – 1867 . [DOI] [PubMed] [Google Scholar]
- 21. Reutrakul S , Mokhlesi B . Can long-term treatment of obstructive sleep apnea with CPAP improve glycemia and prevent type 2 diabetes? Diabetes Care. 2020. ; 43 ( 8 ): 1681 – 1683 . [DOI] [PubMed] [Google Scholar]
- 22. Shaw JE , Punjabi NM , Naughton MT , et al . The effect of treatment of obstructive sleep apnea on glycemic control in type 2 diabetes . Am J Respir Crit Care Med. 2016. ; 194 ( 4 ): 486 – 492 . [DOI] [PubMed] [Google Scholar]
- 23. Pépin JL , Bailly S , Rinder P , et al. on behalf of the medXcloud Group . CPAP therapy termination rates by OSA phenotype: a French nationwide database analysis . J Clin Med. 2021. ; 10 ( 5 ): 936 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Pépin JL , Bailly S , Rinder P , et al. medXcloud Group . Relationship between CPAP termination and all-cause mortality: a French nationwide database analysis . Chest. 2022. ; 161 ( 6 ): 1657 – 1665 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wickwire EM , Tom SE , Vadlamani A , et al . Older adult US Medicare beneficiaries with untreated obstructive sleep apnea are heavier users of health care than matched control patients . J Clin Sleep Med. 2020. ; 16 ( 1 ): 81 – 89 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Tarasiuk A , Reuveni H . The economic impact of obstructive sleep apnea . Curr Opin Pulm Med. 2013. ; 19 ( 6 ): 639 – 644 . [DOI] [PubMed] [Google Scholar]
- 27. Leger D , Bayon V , Laaban JP , Philip P . Impact of sleep apnea on economics . Sleep Med Rev. 2012. ; 16 ( 5 ): 455 – 462 . [DOI] [PubMed] [Google Scholar]
- 28. Tinajero MG , Malik VS . An update on the epidemiology of type 2 diabetes: a global perspective . Endocrinol Metab Clin North Am. 2021. ; 50 ( 3 ): 337 – 355 . [DOI] [PubMed] [Google Scholar]
- 29. Wickwire EM , Albrecht JS , Towe MM , et al . The impact of treatments for OSA on monetized health economic outcomes: a systematic review . Chest. 2019. ; 155 ( 5 ): 947 – 961 . [DOI] [PubMed] [Google Scholar]
- 30. Cistulli PA , Armitstead J , Pepin JL , et al . Short-term CPAP adherence in obstructive sleep apnea: a big data analysis using real world data . Sleep Med. 2019. ; 59 : 114 – 116 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Drager LF , Malhotra A , Yan Y , et al. medXcloud Group . Adherence with positive airway pressure therapy for obstructive sleep apnea in developing vs. developed countries: a big data study . J Clin Sleep Med. 2021. ; 17 ( 4 ): 703 – 709 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Liu D , Armitstead J , Benjafield A , et al . Trajectories of emergent central sleep apnea during CPAP therapy . Chest. 2017. ; 152 ( 4 ): 751 – 760 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Austin PC . An introduction to propensity score methods for reducing the effects of confounding in observational studies . Multivariate Behav Res. 2011. ; 46 ( 3 ): 399 – 424 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Burks SV , Anderson JE , Panda B , et al . Employer-mandated obstructive sleep apnea treatment and healthcare cost savings among truckers . Sleep. 2020. ; 43 ( 4 ): zsz262 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Matching: multivariate and propensity score matching with balance optimization . https://cran.r-project.org/web/packages/Matching/ . Accessed 11 Aug 2021. .
- 36. Kirsch D , Yang H , Maslow AL , Stolzenbach M , McCall A . Association of positive airway pressure use with acute care utilisation and costs . J Clin Sleep Med. 2019. ; 15 ( 9 ): 1243 – 1250 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Wickwire EM , Bailey MD , Somers VK , et al . CPAP adherence is associated with reduced inpatient utilization among older adult Medicare beneficiaries with pre-existing cardiovascular disease . J Clin Sleep Med. 2022. ; 18 ( 1 ): 39 – 45 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Walter RJ , Hagedorn SI , Lettieri CJ . Impact of diagnosing and treating obstructive sleep apnea on healthcare utilization . Sleep Med. 2017. ; 38 : 73 – 77 . [DOI] [PubMed] [Google Scholar]
- 39. Sterling KL , Pépin JL , Linde-Zwirble W , et al . Impact of positive airway pressure therapy adherence on outcomes in patients with OSA and COPD . Am J Respir Crit Care Med. 2022. ; 206 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Mokhlesi B , Grimaldi D , Beccuti G , et al . Effect of one week of 8-hour nightly continuous positive airway pressure treatment of obstructive sleep apnea on glycemic control in type 2 diabetes: a proof-of-concept study . Am J Respir Crit Care Med. 2016. ; 194 ( 4 ): 516 – 519 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Grimaldi D , Beccuti G , Touma C , Van Cauter E , Mokhlesi B . Association of obstructive sleep apnea in rapid eye movement sleep with reduced glycemic control in type 2 diabetes: therapeutic implications . Diabetes Care. 2014. ; 37 ( 2 ): 355 – 363 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Babu AR , Herdegen J , Fogelfeld L , Shott S , Mazzone T . Type 2 diabetes, glycemic control, and continuous positive airway pressure in obstructive sleep apnea . Arch Intern Med. 2005. ; 165 ( 4 ): 447 – 452 . [DOI] [PubMed] [Google Scholar]
- 43. Sheth U , Monson RS , Prasad B , et al . Association of continuous positive airway pressure compliance with complications in patients with type 2 diabetes and obstructive sleep apnea [published online ahead of print, 2021. Mar 17]. J Clin Sleep Med 10.5664/jcsm.9248 . [DOI] [PMC free article] [PubMed]
- 44. Guest JF , Panca M , Sladkevicius E , Taheri S , Stradling J . Clinical outcomes and cost-effectiveness of continuous positive airway pressure to manage obstructive sleep apnea in patients with type 2 diabetes in the U.K . Diabetes Care. 2014. ; 37 ( 5 ): 1263 – 1271 . [DOI] [PubMed] [Google Scholar]