Reduced usage of upper airway stimulation therapy in patients with comorbid insomnia and obstructive sleep apnea

Abstract

Study Objectives:

Upper airway stimulation (UAS) is a hybrid surgical-medical device used to treat moderate-to-severe obstructive sleep apnea (OSA). Comorbid insomnia and OSA (COMISA) is present in ∼50% of these patients. Our aim was to study UAS outcomes and adherence in patients with COMISA.

Methods:

A retrospective review of 379 patients with OSA who underwent UAS implantation at a single institution between 2014 and 2021. Demographics, OSA severity metrics, and insomnia data were collected. Patients were categorized into OSA alone (OSAa) or COMISA. Objective adherence data were collected from device downloads during follow-up. Data were analyzed with using R Studio (R Foundation for Statistical Computing, Vienna, Austria) and Prism (Boston, MA, USA).

Results:

Of the 274 patients included, 148 had COMISA (54.0%) and 126 OSAa (46.0%). Average follow-up time was 2.5 years and OSAa had more males than COMISA (P < .001). Patients with COMISA had higher insomnia severity index scores than OSAa preoperatively (16 vs 8.7; P = .003). All groups showed significant decreases in objective and self-reported OSA outcomes postoperatively, but there was no difference between COMISA and OSAa. Patient with COMISA had decreased device usage (4.9 vs 5.8 h/night; P = .015) and paused therapy more often than patients with OSAa (1.4 vs 0.4 pauses/night; P < .001). Multivariate linear regression, when controlling for sex as a covariate, showed insomnia to be an independent predictor of lower UAS hours/night and more pauses/night (P < .01).

Conclusions:

Patients with COMISA use UAS therapy for shorter durations and require more breaks from therapy when compared with those with OSAa. Future research is needed to explore the underlying mechanism and improve UAS treatment adherence in patients with COMISA.

Citation:

Kaffenberger TM, Chandna M, Kaki P, et al. Reduced usage of upper airway stimulation therapy in patients with comorbid insomnia and obstructive sleep apnea. J Clin Sleep Med. 2023;19(12):1997–2004.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Obstructive sleep apnea (OSA) is a common sleep disorder that is frequently comorbid with insomnia (COMISA). In patients with moderate-to-severe OSA who are intolerant to continuous positive pressure, upper airway stimulation is an effective hybrid medical-surgical treatment. This retrospective review of a large sample of upper airway stimulation patients at a single institution compares OSA outcomes among patients with and without COMISA.

Study Impact: Both COMISA and OSA-alone groups showed significant improvement in OSA-related metrics, but when comparing adherence outcomes between these groups, there was a significant decrease in mean device usage in patients with COMISA (4.9 vs 5.8 h/night; P = .015) and an increase in therapy pauses (1.4 vs 0.4 pauses/night; P < .001). This pattern persisted even when controlling for sex differences between groups.

INTRODUCTION

Upper airway stimulation (UAS) is a common and effective treatment modality for patients with moderate-to-severe obstructive sleep apnea (OSA) who are unable to tolerate continuous positive airway pressure (CPAP) therapy.^1,2 UAS is a hybrid medical-surgical therapy that requires long-term maintenance and good adherence to appropriately treat OSA akin to CPAP.³ Results from the Adherence and Outcome of Upper Airway Stimulation for OSA International Registry (ADHERE) of 1,017 patients demonstrated an average of 5.6 ± 2.1 hours of nightly usage at 12 months postimplantation.² The recognition of comorbid insomnia and OSA (COMISA) has increased over the past decade and estimates of its prevalence range widely from 7% to 84% in the general population.⁴ The combination of insomnia and OSA in patients provides unique treatment challenges as insomnia symptoms can impact tolerance to OSA therapies like CPAP.⁵ Within the UAS literature, COMISA has also been a recent topic of interest and, while the early data show stable improvement in OSA objective treatment metrics, there have been conflicting data if insomnia affects UAS adherence or usage. One of these recent studies observed no difference in device usage between patients with UAS COMISA and those without.⁶ But another study reported that low users of UAS therapy (defined as using the device < 4 h/night) are significantly more likely to have COMISA.⁷

This study aims to further explore the relationship of COMISA on UAS outcomes, specifically on patient adherence to UAS therapy. It represents the largest cohort to date examining the prevalence of COMISA in UAS patients and we hypothesize that COMISA has a negative impact on adherence to UAS therapy.

METHODS

Institutional review board approval was obtained for this retrospective analysis. Federal Drug Administration requirements to qualify for UAS implantation with the Inspire Medical System (Minneapolis, MI, USA) include CPAP intolerance, age ≥ 22 years, an apnea-hypopnea index (AHI) between 15 and 65 events/h, central or mixed respiratory events must be < 25% of the total events, a body mass index (BMI) < 32 or 35 kg/m² depending on insurance, and an additional procedure called a sleep endoscopy to rule out a complete concentric collapse of the soft palate.⁸ We reviewed records from all patients who underwent UAS implantation at a large tertiary care hospital system between 2014 and 2021. Included patients had UAS adherence data available for review within the electronic medical record. Any patient who had the UAS procedure but had no follow-up after the device activation was excluded.

We collected demographic data including age, sex, BMI at the time of surgery, pre- and postoperative polysomnogram (PSG) data, and follow-up time. Objective OSA outcomes were collected from PSGs and included the AHI, oxygen desaturation index (ODI), and time spent below 88% oxygen saturation (T88%). The type of PSG was also recorded and included attended type 1 diagnostic PSGs, type 3 home sleep apnea tests (HSATs; Clevemed, Cleveland, OH, USA), and in-laboratory UAS titration PSGs (PSGt). For HSATs, the respiratory event index was recorded and then labeled as AHI for simplicity. Epworth Sleepiness Scale (ESS) and Insomnia Severity Index (ISI) scores were collected at pre- and postoperative time points. UAS adherence data were recorded from data device downloads from the Inspire remotes in clinic and included hours/night of therapy use, number of pauses/night, and percentage of nights with ≥ 4 hours of use. Furthermore, we collected the sensation and functional thresholds of the UAS device at the activation appointment. The sensation threshold is the amplitude of UAS at which point the patient feels the stimulation, and the functional threshold is the amplitude at which the tongue protrudes beyond the incisors.

Patients were categorized into the insomnia group (COMISA) if they met 1 of the following inclusion criteria: treated pre- or postoperatively for insomnia symptoms, were diagnosed with an insomnia disorder based on the International Classification of Sleep Disorders, third edition (ICSD-3),⁹ or if they had a preoperative ISI score ≥ 15.¹⁰ We also included those patients who had received a cognitive behavior therapy for insomnia referral/treatment or hypnotic agents for their symptoms, which included the drugs zolpidem, eszopiclone, zaleplon, trazodone, doxepin, ramelteon, suvorexant, lemborexant, daridorexant, or temazepam at any time point in the electronic medical record. Based on the information within the electronic medical record, we also categorized patients with insomnia into sleep initiation, sleep maintenance, or combined subtypes, and the timing of insomnia diagnosis or symptoms in relation to surgery was collected. Patients with a diagnosis of restless legs syndrome (RLS) were excluded from the analysis.

Statistical analysis was completed using R Studio (R Foundation for Statistical Computing, Vienna, Austria) and Prism 9 (Boston, MA, USA). The nonparametric Mann-Whitney U test and Kruskal-Wallis test were used to analyze continuous variables, and Pearson’s χ² test was used for categorical variables. Multivariate linear regression was performed to control for sex as a covariate in predicting UAS device adherence. Spearman correlation was used to measure the relationship between UAS adherence and threshold variables. Statistical significance was defined as P < .05.

RESULTS

Of 379 patients with UAS, 274 met the inclusion criteria and their demographic characteristics are summarized in Table 1. Average follow-up after surgery was 2.5 years (standard deviation = 1.7 years). One hundred forty-eight patients were diagnosed with COMISA (54%) and 126 patients were categorized as OSA alone (OSAa; 46%). The OSAa group had significantly more men (75% vs 55%; P < .001) and lower ISI scores (8.6 vs 17.5; P < .001) preoperatively; otherwise, there were no other significant differences between groups.

Table 1.

Demographic characteristics of included patients and breakout groups.

Characteristic	n	Overall (n = 274)	OSAa (n = 126)	COMISA (n = 148)	P a
Age, y	274	62.0 (11.2)	62.2 (11.9)	61.8 (10.5)	.6
Male	274	176 (64%)	94 (75%)	82 (55%)	<.001
BMI at surgery, kg/m2	274	28.6 (3.6)	28.5 (3.4)	28.6 (3.8)	.8
Years to last follow-up	274	2.5 (1.7)	2.3 (1.6)	2.6 (1.7)	.2
BMI at last follow-up, kg/m2	273	28.4 (3.9)	28.4 (3.6)	28.4 (4.2)	.8
Insomnia Severity Index preop	57	14.7 (7.3)	8.6 (4.8)	17.5 (6.5)	<.001
No. of postop sleep studies	274	2.0 (1.3)	2.0 (1.3)	2.0 (1.3)	.8
No. of postop HSAT	274	1.2 (1.1)	1.3 (1.1)	1.2 (1.1)	.5
No. of postop titration PSGs	274	0.8 (0.6)	0.7 (0.5)	0.8 (0.6)	.7

Data are presented as mean (standard deviation) or n (%) unless otherwise indicated. ^aWilcoxon rank-sum test or Pearson’s chi-square test. BMI = body mass index, COMISA = comorbid insomnia and obstructive sleep apnea, HSAT = type 3 home sleep apnea test, OSAa = obstructive sleep apnea alone, postop = postoperatively, preop = preoperatively, PSG = type 1 polysomnogram.

Table 2 shows the characteristics of the COMISA group based on insomnia symptom timing as well as symptom subtypes, including sleep initiation insomnia, sleep maintenance insomnia, or both. Eighty-four (57%) of the COMISA cohort were diagnosed preoperatively. Seventy-four (50.6%) patients with COMISA reported sleep maintenance disturbance and 25 (17%) reported both sleep initiation and sleep maintenance disturbances.

Table 2.

Insomnia group characteristics based on sleep initiation vs maintenance symptoms and timing of first symptom or diagnosis in relation to upper airway surgery.

Characteristic	n	Overall (n = 133)	Sleep Initiation (n = 33)	Sleep Maintenance (n = 73)	Both (n = 27)
Timing of insomnia diagnosis	133
Postop		64 (48%)	15 (45%)	41 (56%)	8 (30%)
Preop		69 (52%)	18 (55%)	32 (44%)	19 (70%)

Values are presented as n (%). postop = postoperatively, preop = preoperatively.

Table 3 summarizes self-reported and objective OSA therapy metrics. There were significant improvements in ESS (P < .001), AHI (P < .001), ODI (P < .001), and T88% (P = .01) for the cohort overall and within each group. There were no significant differences in these changes between patients with OSAa and those with COMISA according to Mann-Whitney U tests.

Table 3.

Upper airway stimulation therapy efficacy for obstructive sleep apnea metrics.

Characteristic	n	Overall (n = 274)	OSAa (n = 126)	COMISA (n = 148)	P a
ESS preop	180	9.9 (5.2)	10.5 (5.6)	9.4 (4.9)	.2
Postop change in ESS	78	−3.0 (4.8)	−3.3 (5.1)	−2.8 (4.6)	.5
AHI preop	274	33.5 (13.7)	34.0 (14.0)	33.1 (13.5)	.6
AHI change	260	−17.8 (18.1)	−18.0 (17.4)	−17.6 (18.7)	>.9
ODI preop	100	27.3 (13.6)	28.8 (13.3)	25.9 (13.9)	.3
Postop change in ODI	73	−12.5 (18.0)	−10.8 (19.1)	−14.4 (16.8)	.8
T88%	213	27.5 (61.9)	39.3 (84.6)	17.0 (26.2)	.02
T88% change	163	−7.6 (74.8)	−18.8 (93.1)	4.5 (45.0)	.11

Data are presented as mean (standard deviation) unless otherwise indicated. ^aWilcoxon rank-sum test. All changes in ESS, AHI, ODI and T88% are significant (P < .01) across the entire cohort and OSAa and COMISA subcohorts. AHI = apnea-hypopnea index, COMISA = comorbid insomnia and obstructive sleep apnea, ESS = Epworth Sleepiness Scale, ODI = oxygen desaturation index, OSAa = obstructive sleep apnea alone, postop = postoperatively, preop = preoperatively, SaO₂ = oxygen saturation, T88% = minutes below oxygen saturation of 88%.

Device usage and adherence characteristics are reported in Table 4. Hours/night usage was significantly lower among patients with COMISA compared with patients with OSAa (4.9 vs 5.8 h/night; P = .015; Figure 1A). The percentage of nights with ≥ 4 hours of use was not significantly different between patients with OSAa and those with COMISA (77.1% vs 73.9%; P = .5). The recorded number of pauses/night was significantly higher in the COMISA group compared with the OSAa group (1.5 vs 0.4 pauses/night; P < .001; Figure 1B). A multivariate linear regression analysis was used to control for the differences in sex inherent to our 2 cohorts. COMISA predicted a 1.04-hour reduction in device usage (P = .001) and 0.97 more pauses/night (P = .0002). Being male was associated with a 0.73-hour reduction in device usage (P = .03) but was not a significant predictor of pauses/night.

Table 4.

Device usage between patients with and without a comorbid sleep disorder.

Characteristic	n	Overall (n = 274)	OSAa (n = 126)	COMISA (n = 148)	P a
Hours/night of use	274	5.3 (2.7)	5.8 (2.4)	4.9 (2.8)	.015
Percent of nights ≥ 4 h of use	103	75.4 (25.7)	77.1 (24.7)	73.9 (26.6)	.5
Pauses/night	105	0.9 (1.3)	0.4 (0.7)	1.4 (1.6)	<.001

Data are presented as mean (standard deviation). ^aWilcoxon rank-sum test. COMISA = comorbid insomnia and obstructive sleep apnea, OSAa = obstructive sleep apnea alone.

Figure 1. Upper airway stimulation (UAS) adherence data visualized with Tukey box-and-whisker plots.

(A) Hours per night of use overall and between the patients with obstructive sleep apnea alone (OSAa) and comorbid insomnia and OSA (COMISA). Patients with COMISA had a significantly decreased nightly usage; P = .015. * indicates P < 0.05. (B) The average number of UAS pauses per night was significantly elevated in the COMISA group compared with the OSAa group; P = .001. *** indicates P < 0.001. Testing was completed with Mann-Whitney unpaired, nonparametric 2-tailed t tests.

The relationship between insomnia subtype (sleep initiation, sleep maintenance, or both) and adherence metrics was also explored, but no statistically or clinically significant results were found (Figure 2 and Table 5). The timing of original insomnia diagnosis (pre- or postoperatively in relation to UAS implantation) was also assessed but showed no statistical or clinical differences (Figure 3 and Table 6). Hours/night and pauses/night were further analyzed and compared with patients’ sensation and functional UAS device thresholds at the time of activation, but there were no clinically or statistically significant differences identified (Table S1 ^{(54.1KB, pdf)} and Table S2 ^{(54.1KB, pdf)} in the supplemental material).

Figure 2. Upper airway stimulation adherence data in patients with comorbid insomnia and obstructive sleep apnea (COMISA) separated by insomnia symptom subtypes using Tukey box-and-whisker plots.

Statistical analysis showed no significant difference in hours/night of usage (A) (P = .5) or pauses/night (B) (P = .8). Testing was completed with Mann-Whitney unpaired, nonparametric 2-tailed t tests.

Table 5.

Device usage between insomnia subtypes in patients with COMISA.

Characteristic	n	Overall (n = 133)	Sleep Initiation (n = 33)	Sleep Maintenance (n = 73)	Both (n = 27)	P a
Hours/night of use	133	4.8 (2.8)	4.9 (3.1)	4.6 (2.7)	5.2 (2.6)	.4
Percent of nights ≥ 4 h of use	49	74.7 (26.1)	83.1 (25.2)	72.8 (24.5)	71.1 (31.2)	.2
Pauses/night	50	1.5 (1.6)	1.4 (1.1)	1.4 (1.6)	1.6 (2.2)	.8

Data are presented as mean (standard deviation). ^aKruskal-Wallis rank-sum test. COMISA = comorbid insomnia and obstructive sleep apnea.

Figure 3. Upper airway stimulation (UAS) adherence data in patients with comorbid insomnia and obstructive sleep apnea (COMISA) separated by timing of insomnia diagnosis in relation to UAS surgery using Tukey box-and-whisker plots.

Statistical analysis showed no significant difference in hours/night of usage (A) (P = .6) or pauses/night (B) (P = .3). Testing was completed with Mann-Whitney unpaired, nonparametric 2-tailed t tests.

Table 6.

Device usage in patients with COMISA based on timing of insomnia diagnosis.

Characteristic	n	Overall (n = 147)	Postop (n = 64)	Preop (n = 83)	P a
Hours/night of use	147	4.9 (2.8)	4.8 (2.8)	5.0 (2.9)	.6
Percent of nights ≥ 4 h of use	54	73.9 (26.6)	72.0 (29.6)	75.1 (24.8)	>.9
Pauses/night	55	1.4 (1.6)	1.6 (1.8)	1.2 (1.4)	.3

Data are presented as mean (standard deviation). Timing of diagnosis is in relation to the timing of upper airway stimulator implantation. ^aWilcoxon rank-sum test. COMISA = comorbid insomnia and obstructive sleep apnea, Postop = postoperatively, Preop = preoperatively.

DISCUSSION

In this study, we demonstrated UAS therapy to be an effective treatment in patients with OSAa as well as in those with COMISA, but those patients with COMISA have significantly more difficulty tolerating UAS therapy. Our data indicate that they use UAS, on average, almost 1-hour/night less and are much more likely to rely on pausing therapy during the night than a patients with OSAa. To our knowledge, this is the first study to show that patients with COMISA have decreased UAS adherence compared with those with OSAa.

Our work builds on 4 prior studies looking at COMISA in patients with UAS. In Luyster et al’s study⁷ of 24 patients, the authors approached the question of how COMISA affects UAS adherence from a different direction. They first identified low users of UAS therapy, defined as those using the device < 4 hours/night, and then explored qualitative factors that could be responsible. Notably, ISI scores and anxiety were significantly higher in the low-user group when compared with the high-user group. In Jomha et al’s study,⁶ 64 patients with UAS were analyzed, and a similar prevalence of COMISA was described in our study (47% vs 54%, respectively). However, in contrast to the current study, they did not find a difference in UAS usage in patients with COMISA. This conflicting result may be due to differences in the length of follow-up between the 2 studies. Jomha et al assessed adherence at the 3-month postoperative UAS PSGt while ours assessed adherence at the most recent follow-up appointment, which was, on average, 2.5 years from surgery. This explanation is supported by Luyster et al,⁷ as their low-usage patients had an increased length of time from surgery compared with high users (26.8 vs 22.8 mo). Another study by Pordzik et al¹¹ noted an improvement in the ISI at 3 months after implantation in 27 patients, but no adherence data were published. Last, Dhanda Patil et al¹² studied a cohort of 53 veterans implanted with a UAS device and found similar results to this study. Namely, the prevalence of COMISA was 56.6% and, despite the adherence rate not reaching statistical significance, there was a trend toward patients with COMISA having a lower usage when compared with patients with OSAa (5.6 vs 6.4 h/night; P = .17). We suspect that if Dhanda Patil et al had more statistical power that this result would have reached statistical significance. Another possible explanation for the differences in adherence seen in our sample is the differences in sex between the OSAa and COMISA groups. Women represented only 25% of our OSAa group, while women made up 45% of the COMISA group (P < .001). This is consistent with several other studies that have demonstrated a female predominance of insomnia and COMISA.⁵ When sex was controlled for in a multivariate regression analysis, the prevalence of COMISA remained strongly associated with poor UAS adherence in the form of lower device usage/night and more pauses/night. On that multivariate analysis we also found that male sex was associated with reduced usage. This appears to be in line with prior reports from the ADHERE registry, which demonstrated that women have improved UAS outcomes and that women may also have improved UAS adherence.^2,13 An additional finding from our study was that patients with COMISA had significantly less hypoxia (as measured by time spent < 88% oxygen saturation) on their preoperative sleep study (17 vs 39 minutes; P = .02), despite similar ODIs. A possible explanation for this finding (and the minimal improvement in the hypoxic burden postoperatively) in the COMISA group is that these patients have a lower arousal threshold compared with those with OSAa. The patients with COMISA would then be more sensitive to changes in hypoxia and have shorter respiratory events.¹⁴ Interest in using the arousal threshold to predict UAS success has also been recently tested and demonstrated that those patients with high arousal thresholds have higher rates of treatment success.¹⁵

Compared with prior COMISA reports, a novel feature of our study is the reporting of the average number of UAS therapy pauses. This pause feature in the Inspire UAS system allows the patient to suspend therapy for a short period during nocturnal awakenings. In our results, the COMISA cohort used the pause feature much more frequently (1.5 vs 0.4 pauses/night; P = .001). The most likely explanation behind this observation is that those patients with insomnia, having frequent awakenings at night or needing longer to fall asleep initially, will rely on the pause button to keep the therapy off as they try to fall asleep. A cluster analysis of UAS adherence data from the first 3 months of UAS therapy in 2,098 patients demonstrated that the cluster of patients with the lowest UAS usage also had the highest number of pauses/night.¹⁶ Factors such as insomnia and RLS were not assessed, but the authors did speculate that this cluster may have higher rates of comorbid sleep disorders.

Outside of insomnia, additional sleep disorders like RLS may play an important role in UAS adherence. In this study cohort, we did identify 20 UAS patients who had been diagnosed with RLS. Given the low number of patients in this cohort, these patients were ultimately excluded, but preliminary data did not show a difference in therapy adherence or pauses. Jomha et al⁶ found a high percentage of RLS within their cohort (28%) and showed a significant decrease in the nightly usage of UAS in the RLS group. Our small sample of patients with RLS may be partially explained by the demographic differences between the 2 studies. Women are known to have higher rates of RLS, and our study was composed of only 36% women compared to 47% in the Jomha et al study.⁶

In our study, we did expand our COMISA analysis to assess the impact of insomnia symptom subtypes and the timing of insomnia diagnosis on adherence (Table 2). We suspected that patients who were diagnosed with insomnia before UAS implantation would have worse adherence, but no clinically or statistically significant differences were uncovered (Figure 3; Table 6). We also suspected that patients with sleep maintenance insomnia would be most susceptible to the effects of UAS. There was a predominance of patients with COMISA with sleep maintenance symptoms (67.6%), but no difference in adherence was identified between sleep maintenance, initiation, or patients exhibiting both (4.6 vs 4.8 vs 5.2 h/night, respectively; Figure 2; Table 5; P = .5). Plausible explanations for this lack of differences include inadequate statistical power or that there may not be a difference to begin with. In the CPAP literature, there are mixed data linking sleep maintenance insomnia and decreased CPAP adherence, with 2 studies supporting a link^17,18 and an additional study refuting it.¹⁹

Our study has several limitations, with the foremost being its retrospective design and the limitations that carries with it. Additionally, other common comorbidities of insomnia, such as anxiety, depression, and chronic pain diagnosis, were not recorded. Much of the objective sleep study data came from HSATs, which are well known to underestimate objective disease burden. This study also encompasses over 7 years of clinical practice at a single institution and captures several changes to UAS protocols, including more detailed data download reports, integration of the ISI survey preoperatively, and the transition to using HSATs as the primary modality to assess objective OSA improvement. While these factors do contribute to an overall increase in the heterogeneity of data, we believe that these downsides were balanced by analyzing COMISA characteristics in the largest UAS sample published from a single institution. Furthermore, our study assessed the nuances of objective UAS adherence data from device downloads, insomnia subtypes at pre- and postoperative time points, as well as the sensation and functional thresholds of the UAS devices. While only a few of these metrics showed clinical or statistical significance, the underlying theme of patients with COMISA having more difficulty adapting to UAS therapy suggests multiple areas of prospective intervention.

One of those areas is improving the identification of patients with COMISA or those who are susceptible to insomnia prior to UAS surgery. Preoperative screening of UAS candidates with tools like the ISI may aid in the identification of future low utilizers of therapy. Our study showed higher ISI in the COMISA population than in the OSAa group (17.5 vs 8.6, respectively; P < .001). Based on our data, providers could potentially use an ISI score > 15 (correlating to clinical insomnia—moderate)¹⁰ as a preoperative threshold to institute cognitive behavior therapy for insomnia or hypnotic therapy to control insomnia symptoms prior to UAS implantation. An additional area would be further study on the Inspire UAS device’s “comfort settings,” which include additional electrode configurations, pulse width, and pulse frequency. A recent study by Steffen et al²⁰ demonstrated that modification of these parameters allows for more efficient tongue stimulation and could improve patient comfort with stimulation. Anecdotally, utilization of “comfort settings” to improve tolerance to therapy is often discussed but there are no data to support or refute their efficacy at this time. Last, the use of hypnotic agents and cognitive behavior therapy for insomnia management is well established, but whether they can also be used safely and efficaciously within the UAS COMISA population remains to be seen. These therapies could be used prior to UAS surgery, during an accommodation phase immediately after UAS activation, or as a salvage therapy for patients who develop insomnia after UAS activation. Further studies will be needed to identify effective preventative and therapeutic strategies for UAS patients with COMISA.

CONCLUSIONS

Approximately 50% of patients with moderate-to-severe OSA treated with UAS therapy carry a comorbid diagnosis of insomnia. This study demonstrated that these patients show relatively poor UAS adherence in terms of nightly usage and pauses/night, indicating that patients with COMISA face unique challenges in adherence to UAS therapy that could hinder its overall efficacy. Further research is needed to understand the factors underlying this observation as well as better preoperative screening protocols and postoperative salvage therapies.

DISCLOSURE STATEMENT

M.B.: Consultant for Inspire Medical Systems and research support from Nyxoah and Inspire. K.D.: Consultant for Axsome, Harmony, Jazz, Janssen, Idorsia, and Harmony. Research grants from Inspire Medical Systems, Zoll, Nyxoah, and Kem Pharma. Merck Stock. C.H.: Research support from Nyxoah, Inspire Medical Systems, and Sommetrics. The other authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

BMI

body mass index

COMISA

comorbid insomnia and obstructive sleep apnea

CPAP

continuous positive airway pressure

ESS

Epworth Sleepiness Scale

HSAT

type 3 home sleep apnea test

ISI

Insomnia Severity Index

ODI

oxygen desaturation index

OSA

obstructive sleep apnea

OSAa

obstructive sleep apnea alone

PSG

polysomnogram

RLS

restless legs syndrome

T88%

time spent below 88% oxygen saturation

UAS

upper airway stimulation