Potential Maladaptive Sleep-Related Cognitions in Depression with Comorbid Hypersomnolence: An exploratory investigation

Abstract

Objective/Background:

Dysfunctional sleep-related cognitions (SRCs) have been demonstrated in both insomnia and depression, but have not been evaluated in patients experiencing depression with co-occurring hypersomnolence. Given the prominence of maladaptive thinking in depression with comorbid insomnia, dysfunctional SRCs may also exist in depressed persons experiencing hypersomnolence. Identifying potentially maladaptive SRCs may assist development of cognitive-behavioral strategies to alleviate hypersomnolence and its related impairment, particularly when comorbid with depression.

Participants:

Twenty-two unmedicated persons with major depressive disorder (MDD) with comorbid hypersomnolence (MDD+/HYP+), as well as age- and sex-matched persons with MDD without hypersomnolence (MDD+/HYP−) and healthy controls (HC).

Methods:

Participants completed the Dysfunctional Beliefs and Attitudes About Sleep–16-item (DBAS-16) and underwent overnight polysomnography. Groups were compared across clinical and sleep domains, as well as DBAS-16 global, subscale, and individual item scores. Additional analyses evaluated DBAS-16 components while controlling for depression severity.

Results:

Groups significantly differed across all collected sleep and mood metrics consistent with diagnostic classification. MDD+/HYP+ DBAS-16 global score was significantly elevated, relative to HC, and was comparable to MDD+/HYP−. A DBAS-16 global score significant group effect was maintained while controlling for depression symptom severity, however only individual DBAS-16 items related to quantity and quality of sleep demonstrated particular relevance to MDD+/HYP+ compared to other groups.

Conclusions:

Results suggest potentially maladaptive SRCs in MDD+/HYP+. Further efforts are needed to clarify whether these beliefs and attitudes about sleep in persons with hypersomnolence are in fact dysfunctional, as well as identify relevant content for development of a novel hypersomnolence-related SRC metric.

1. Introduction

Depression and sleep complaints, particularly insomnia, have a well-established bidirectional relationship (Franzen & Buysse, 2008; Murphy & Peterson, 2015; Soehner, Kaplan & Harvey, 2014; Tsuno, Besset, & Ritchie, 2005). It has been estimated that upwards of 90% of persons with depression also experience a co-occurring sleep condition, most commonly insomnia (Franzen & Buysse, 2008). Dysfunctional, inaccurate thoughts and beliefs (e.g. all-or-nothing thinking, negative predictions, mind reading, etc.) are common in depression, and are often targeted as a point of intervention during cognitive therapy (Beck, Rush, Shaw, & Emery, 1979; Cristea et al., 2015). These maladaptive thought patterns have been shown to also extend to sleep-related content (e.g. erroneous expectations about sleep requirements, exaggerated beliefs concerning the daytime consequences of disturbed sleep, worry and helplessness related to sleep, etc.) (Levenson, Benca, & Rumble, 2015; Morin, Vallières, & Ivers, 2007). Relationships between depression, maladaptive thought processes, and sleep have been primarily evaluated within the framework of insomnia as the co-occurring sleep complaint (Carney & Edinger, 2006; Carney et al., 2010; Crönlein et al., 2014). These efforts have helped shape components of cognitive-behavioral therapy for insomnia, which is a first-line treatment option for insomnia occurring both with-and-without comorbid depression (Cunningham & Shapiro, 2018; Manber et al., 2016; Morgenthaler et al., 2006; Siebern & Manber, 2011).

Despite the primary emphasis placed on the relationship between insomnia and depression, a sizeable minority of patients with depression will experience hypersomnolence, broadly characterized by the presence of excessive daytime sleepiness (EDS), with normal to prolonged sleep duration, rather than insomnia (Kaplan and Harvey, 2009). As a symptom, hypersomnolence can also occur across a range of neurological, medical, and other psychiatric disorders. Similar to patients with insomnia, persons with hypersomnolence are extremely burdened by their symptoms, which often translates into significantly impaired functionality, poor socio-economic outcomes, negative health implications, and a deteriorated quality of life (Bayon, Léger, & Philip, 2009; Billiard & Dauvilliers, 2001; Ingravallo et al., 2012; Khan & Trotti, 2015; Ozaki et al., 2012; Sowa, 2016). Although it is likely that hypersomnolence and depression share a similar, bidirectional relationship to that of insomnia and depression (Kaplan & Harvey, 2009), no previous investigation has been conducted to assess whether dysfunctional sleep-related cognitions (SRCs) may exist in persons with depression and comorbid hypersomnolence (MDD+/HYP+).

At present, pharmacotherapy is generally considered the primary treatment for hypersomnolence. However, a recent survey study demonstrated that patients with hypersomnolence have a sizeable interest in receiving behavioral sleep medicine services, such as cognitive-behavioral therapy (CBT) (Neikrug, Crawford, & Ong, 2017). The absence of a well-established CBT derivative for hypersomnolence is thus an important area of research for behavioral sleep medicine. Previous research has highlighted certain behavioral techniques that may be applied in hypersomnolence management (Conroy, Novick, & Swanson, 2012), but the utility and foundations of cognitive therapy for hypersomnolence are unclear. As such, clarifying whether potentially dysfunctional SRCs occur in persons with hypersomnolence, particularly among those with co-occurring depression, is an important step in advancing this area of inquiry.

Presently, limited empirical evidence exists demonstrating that persons with hypersomnolence as a primary sleep complaint experience dysfunctional SRCs. One previously conducted study analyzed the Dysfunctional Beliefs and Attitudes about Sleep (DBAS-16) scale in a number of sleep disorders, including a group with central nervous system hypersomnias (e.g. narcolepsy and idiopathic hypersomnia) (Crönlein et al., 2014). This study demonstrated a non-significant elevation in DBAS-16 global scale score for the hypersomnolence group, relative to HC (Crönlein et al., 2014). However, wide variation in age, sex, and sample size across study groups, as well as methodological complexities raised by combining patients with narcolepsy and idiopathic hypersomnia, who often have divergent reports of sleep continuity, reduce the generalizability and applicability of these findings (Plante, 2018; Roth et al., 2013). Perhaps more importantly, this study required the absence of depression for inclusion, which may be a critical factor when considering the existence of dysfunctional SRCs in the context of hypersomnolence as the primary sleep complaint.

Thus, given the paucity of data in this important research area, and the methodological limitations of the scant available evidence, this study sought to expand upon the existing literature, by applying the widely used DBAS-16 to examine potentially maladaptive SRCs in persons with depression and comorbid hypersomnolence, relative to age- and sex-matched comparison groups of healthy sleepers and persons with depression but without hypersomnolence.

2. Methods

2.1. Participants and Study Design

The data utilized in this investigation were obtained from a parent study investigating the neurophysiologic correlates of hypersomnolence in depression. Relevant to these analyses, the study employed a multi-level screening process whereby prospective participants’ general health, psychological history, and sleep patterns were initially assessed via telephone before being considered for an in-person, comprehensive evaluation. In-person assessment included a Structured Clinical Interview for the DSM-IV (SCID) Axis 1 disorders (First, Spitzer, Gibbon, & Williams, 2002), semi-structured medical and sleep history, and physical examination, all performed by a physician board certified in sleep medicine and psychiatry (DTP). Depression diagnosis was determined from the SCID and hypersomnolence was established based on operationalized criteria proposed by Ohayon and colleagues (Ohayon, Dauvilliers, & Reynolds, 2012), which have since been adopted with only minor modifications as the diagnostic criteria for hypersomnolence disorder in the Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition (American Psychiatric Association, 2013). Additionally, MDD+/HYP+ hypersomnolence could not be attributable to another sleep or medical disorder (e.g. obstructive sleep apnea, restless legs syndrome, narcolepsy with cataplexy, etc.). By definition, participants meeting criteria for hypersomnolence disorder also did not meet diagnostic criteria for insomnia disorder (i.e. their sleepiness was not a result of or related to difficulty initiating and/or maintaining sleep).

Comparison groups of healthy controls (HC) and depression without hypersomnolence (MDD+/HYP−) were established through an age-and-sex matched recruitment process. These groups underwent a multi-level screening and assessment process identical to that of MDD+/HYP+. For HC, inclusion required the absence of any psychiatric or severe medical condition, no report of sleep complaint or difficulty staying awake during the day, and a self-reported habitual sleep duration between 7 and 9 hours. For MDD+/HYP−, depression diagnosis was derived from SCID interview and these persons did not meet hypersomnolence criteria outlined by Ohayon and colleagues. While many in this group experienced insomnia, neither the presence of insomnia nor a specific insomnia severity was required for inclusion, only the absence of hypersomnolence and presence of MDD was necessary.

Additionally, the following exclusionary parameters were applied across all persons enrolled in the parent study: participants could not be younger than 18 years old, have a history of head trauma, loss of consciousness > 30 minutes, or significant neurologic illness, be habitually using psychotropic medication within two weeks of in-person visit, smoke more than 15 cigarettes per day or be of imminent risk for self-harm/suicide. Furthermore, women were excluded if they were pregnant, breastfeeding, or < 6 months post-partum. Lastly, participants were excluded if they met DSM-IV criteria for alcohol or substance abuse within the preceding 6 months.

Participants deemed eligible for further participation at the initial, in-person evaluation were scheduled for a polysomnographic (PSG) evaluation at Wisconsin Sleep, the sleep clinic and laboratory affiliated with the University of Wisconsin-Madison.

All participants provided informed consent and all parent study procedures were approved by the University of Wisconsin-Madison Health Sciences Institutional Review Board.

2.2. Collected Measures: Demographics and Characteristics

Participants completed a questionnaire packet at the initial, in-person evaluation. The primary measure used to evaluate for potential SRCs was the widely used Dysfunctional Beliefs and Attitudes about Sleep – 16 item (DBAS-16) (Morin et al., 2007). Additional measures included the Beck Depression Inventory-II (BDI-II) (Wang & Gorenstein, 2013), Epworth Sleepiness Scale (ESS) (Johns, 1991; Kendzerska, Smith, Brignardello-Peterson, Leung, & Tomlinson, 2014), Pittsburgh Sleep Quality Index (PSQI) (Buysse, Reynolds, Monk, Berman, & Kupfer, 1989), Hypersomnia Severity Index (HSI) (Kaplan, Plante, Cook, & Harvey, 2019), and Insomnia Severity Index (ISI) (Morin, Belleville, Bélanger, & Ivers, 2011).

The DBAS-16 is a validated and reliable metric for assessing SRCs (Chung, Ho, & Yeung, 2016; Morin et al., 2007). This commonly utilized tool assesses a respondent’s level of agreement/disagreement to statements focused on sleep/insomnia-related cognitions (Morin et al., 2007). Individuals evaluate statements across a Likert-scale that ranges from 0 (strongly disagree) to 10 (strongly agree) (Morin et al., 2007). Previous research has established four thematic subscales within the DBAS-16: (1) Consequences of Insomnia (Consequences); (2) Worry or Helplessness About Sleep (Worry); (3) Expectations for Sleep (Expectations); and (4) Causal Attributions for Insomnia (Attributions) (Morin et al., 2007). Although this metric was designed to assess SRCs in patients with insomnia complaints (Carney et al., 2010; Morin et al., 2007), it was selected here because 1) it has been utilized by other investigations to examine potentially maladaptive beliefs and attitudes across a range of sleep disorders (Crönlein et al., 2014), and 2) no better alternative metric existed at the time of data collection (or at present) that specifically assesses SRCs in patients with complaints of hypersomnolence.

2.3. Collected Measures: Polysomnography

PSG data were collected using an integrated platform, with electroencephalography (EEG) recorded using a 256-channel high-density array (Electrical Geodesics, Eugene, OR) and additional physiological measurements captured for sleep staging and scoring (Alice® Sleepware; Philips Respironics, Murrysville, PA). PSG staging was performed by a registered technologist across 30-second epochs according to AASM criteria, using a standard six-channel EEG montage at approximate 10–20 locations (F3, F4, C3, C4, O1, and O2) referenced to the mastoids, submental electromyogram and electrooculogram criteria (Berry et al., 2015).

An ad libitum sleep study protocol was utilized. Participants were minimally disturbed throughout the night and not awakened at a prescribed time the following morning. PSG start-time was tailored to align as closely as logistically possible with the participant’s habitual sleep-wake pattern and end-time was determined by the participant’s stated desire to terminate the nocturnal sleep period upon awakening. Additionally, if a participant demonstrated obstructive sleep apnea (Apnea-Hypopnea Index ≥ 10 hr) or periodic limb movement disorder (PLM arousal index ≥ 10/hr) during their PSG assessment, then they were excluded from the analyses and a replacement was recruited.

2.4. Data Analyses

Groups were compared across collected demographics, characteristics, and PSG results, as well as DBAS-16 global, subscale, and individual item scores. Additionally, supplemental analyses across all DBAS-16 components were performed while controlling for depression symptom severity (Levenson et al., 2015). These analyses included a modified BDI-II score (BDI_NoSleep), absent of sleep-related questions (Levenson et al., 2015), as a covariate.

Analysis of variance (ANOVA) determined the main effect of group. F-Statistic and corresponding p value are provided from ANOVA. Initial significance level was set at 0.05. Post-hoc pairwise comparisons were performed for all comparisons that resulted in a significant main group effect. Bonferroni correction was applied across all post-hoc pairwise comparisons to correct for multiple comparisons.

Data analyses were performed using R Studio (The R Foundation, Boston, MA).

3. Results

3.1. Group Comparisons: Demographics, Characteristics, and Polysomnography Results

A full summary of group demographics and characteristics is presented in Table 1. Each group was comprised of 22 participants that were non-obese, young- to middle-age, and predominantly female. Among the depressed participants, 2 met diagnostic criteria for atypical depression (MDD+/HYP+ = 2; MDD+/HYP− = 0), 2 met criteria for a seasonal MDD pattern (MDD+/HYP+ = 2; MDD+/HYP− = 0), and 11 provided responses indicative of a singular current depressive episode (MDD+/HYP+ = 3; MDD+/HYP− = 8), while the majority met criteria for recurrent MDD (MDD+/HYP+ = 19; MDD+/HYP− = 14).

Table 1.

Group Demographics and Characteristics

Demographics
Variable	MDD+/HYP+	MDD+/HYP−	HC	F(2,63) (p value)	MDD+/HYP+ vs. MDD+/HYP−	MDD+/HYP+ vs HC	MDD+/HYP− vs HC
Group Size (N)	22	22	22
Percent Female (%)	81.8	81.8	81.8
Age	28.1 ± 5.80	26.8 ± 4.83	28.5 ± 5.70	0.59 (−)	N/A	N/A	N/A
BMI	24.6 ± 4.91	22.4 ± 2.91	25.2 ± 3.93	2.88 (−)	N/A	N/A	N/A
Characteristics
BDI	22.3 ± 7.40	24.0 ± 9.00	1.14 ± 1.70	77.2 (<0.0001)	-	<0.0001	<0.0001
BDI_NoSleep	21.4 ± 7.04	22.5 ± 8.67	1.05 ± 1.65	75.7 (<0.0001)	-	<0.0001	<0.0001
ESS	12.4 ± 2.63	6.00 ± 3.61	4.64 ± 2.28	45.2 (<0.0001)	<0.0001	<0.0001	-
HSI	21.0 ± 4.22	11.2 ± 4.56	3.55 ± 2.18	117 (<0.0001)	<0.0001	<0.0001	<0.0001
HSI-Symptoms	10.8 ± 1.97	2.73 ± 1.98	2.32 ± 1.36	157 (<0.0001)	<0.0001	<0.0001	-
ISI	11.6 ± 3.39	14.9 ± 5.54	1.14 ± 1.42	76.7 (<0.0001)	0.02	<0.0001	<0.0001
ISI-Symptoms	1.95 ± 1.84	6.14 ± 2.61	0.27 ± 0.55	57.5 (<0.0001)	<0.0001	0.01	<0.0001
PSQI	5.68 ± 2.03	8.77 ± 2.35	1.77 ± 1.45	69.2 (<0.0001)	<0.0001	<0.0001	<0.0001

Results across collected demographics and characteristics are presented for depression with comorbid hypersomnolence (MDD+HYP+), depression without hypersomnolence (MDD+/HYP−), and healthy controls (HC). Values are provided as mean ± standard deviation where appropriate. Demographics include group size, percentage of females within the group (Percent Female; %), age, and body mass index (BMI). Characteristics include Beck Depression Inventory (BDI), Beck Depression Inventory without sleep items (BDI_NoSleep), Epworth Sleepiness Scale (ESS), Hypersomnia Severity Index (HSI), Hypersomnia Severity Index with only items specific to hypersomnolence symptoms (HSI-Symptoms; Items 1a – 1d & 6), Insomnia Severity Index (ISI), Insomnia Severity Index with only items specific to insomnia symptoms (ISI-Symptoms; Items 1–3), and Pittsburgh Sleep Quality Index (PSQI). F-Statistic and corresponding p value are documented for ANOVA results. Bonferroni corrected p values are provided for pairwise comparisons, when applicable. A hyphen (−) denotes a nonsignificant result. N/A indicates that pairwise comparisons were not performed due to a nonsignificant group effect.

Groups were comparable across demographics. A main group effect was determined for all collected sleep and depression characteristics. Since HSI and ISI have similar question items related to daytime dysfunction and impact of sleep-related symptoms, hypersomnolence and insomnia specific values were also calculated omitting these questions. This, along with other data demonstrated that 1) MDD+/HYP+ had minimal to no insomnia complaints and 2) the MDD+/HYP− group reported the highest level of insomnia symptoms. Also consistent with diagnoses and group classification, MDD+/HYP+ endorsed the greatest daytime sleepiness and hypersomnolence symptom severity relative to other groups.

In terms of PSG, a main effect of group was only observed for Time in Bed (TIB), Total Sleep Time (TST), Non-rapid eye movement sleep (NREM) stage 1 (N1) duration, N1 percentage, and NREM stage 2 (N2) duration. Aligning with diagnoses, MDD+/HYP+ demonstrated the longest TIB and TST. A full summary of group PSG results is presented in Table 2.

Table 2.

Polysomnography (PSG) Variables and Staging

PSG Variables and Staging
Variable	MDD+/HYP+	MDD+/HYP−	HC	F(2,63) (p value)	MDD+/HYP+ vs. MDD+/HYP− p value	MDD+/HYP+ vs HC p value	MDD+/HYP− vs HC p value
TIB (min)	602 ± 100	545 ± 96.4	512 ± 62.6	6.00 (0.004)	-	0.003	-
TST (min)	527 ± 98.4	469 ± 91.6	448 ± 63.0	5.07 (0.009)	-	0.01	-
SE (%)	87.3 ± 6.47	86.0 ± 7.05	87.3 ± 7.01	0.18 (−)	N/A	N/A	N/A
SOL (min)	14.8 ± 13.3	16.5 ± 20.2	13.8 ± 9.62	0.25 (−)	N/A	N/A	N/A
WASO (min)	61.6 ± 36.1	59.8 ± 37.8	51.8 ± 29.8	0.49 (−)	N/A	N/A	N/A
Awakenings (N)	26.0 ± 9.12	23.9 ± 10.0	20.9 ± 7.45	1.86 (−)	N/A	N/A	N/A
REML (min)	126 ± 79.1	102 ± 40.3	125 ± 49.6	1.12 (−)	N/A	N/A	N/A
N1 (min)	36.9 ± 19.6	22.3 ± 12.4	24.1 ± 12.3	6.08 (0.004)	0.007	0.02	-
N1 (%)	6.98 ± 3.49	4.81 ± 2.49	5.33 ± 2.54	3.43 (0.04)	0.04	-	-
N2 (min)	308 ± 65.4	266 ± 64.7	260 ± 56.9	3.92 (0.03)	0.007	0.02	-
N2 (%)	58.6 ± 6.95	56.7 ± 8.60	57.4 ± 8.51	0.30 (−)	N/A	N/A	N/A
N3 (min)	78.5 ± 31.9	78.1 ± 41.4	71.1 ± 31.5	0.30 (−)	N/A	N/A	N/A
N3 (%)	15.2 ± 6.36	16.3 ± 7.13	16.3 ± 7.65	0.17 (−)	N/A	N/A	N/A
REM (min)	105 ± 41.3	103 ± 26.0	95.1 ± 33.5	0.47 (−)	N/A	N/A	N/A
REM (%)	19.2 ± 5.84	22.2 ± 5.53	21.0 ± 5.87	1.52 (−)	N/A	N/A	N/A

Results from in-laboratory, ad libitum PSG are presented for depression with comorbid hypersomnolence (MDD+HYP+), depression without hypersomnolence (MDD+/HYP−), and healthy controls (HC). Values are provided as mean ± standard deviation where appropriate. PSG variables included time in bed (TIB; minutes), total sleep time (TST; minutes), sleep efficiency (SE; %), sleep onset latency (SOL; minutes), wake after sleep onset (WASO; minutes), number of awakenings (Awakenings; N), and rapid eye movement sleep latency (REML; minutes). Additionally, duration (minutes) and percentage (%) of non-rapid eye movement sleep (NREM) stage 1 duration (N1), NREM stage 2 (N2), NREM stage 3 (N3), and REM (REM) are provided. F-Statistic and corresponding p value are documented for ANOVA results. Bonferroni corrected p values are provided for pairwise comparisons, when applicable. A hyphen (−) denotes a nonsignificant result. N/A indicates that pairwise comparisons were not performed due to a nonsignificant group effect.

3.2. Group Comparisons: DBAS-16

A main group effect was observed for DBAS-16 global score. MDD+/HYP+ and MDD+/HYP− scores were both significantly elevated, relative to HC. Notably, MDD+/HYP+ and MDD+/HYP− global scores were comparable. Additionally, significant variance among groups emerged across all DBAS-16 subscales. The presence of depression was uniquely important to subscale 2 (Worry or Helplessness About Sleep).

In terms of individual DBAS-16 items, a significant main effect was determined for 11 of the 16 items. MDD+/HYP+ was uniquely elevated, relative to both MDD+/HYP− and HC, for item 2 (When I don’t get the proper amount of sleep on a given night, I need to catch up on the next day by napping or on the next night by sleeping longer), item 5 (After a poor night’s sleep, I know that it will interfere with my daily activities on the next day), and item 9 (Without an adequate night’s sleep, I can hardly function the next day). Depression appeared to uniquely contribute to elevated ratings on item 11 (I have little ability to manage the negative consequences of disturbed sleep) and item 16 (I avoid or cancel obligations (social, family) after a poor night’s sleep).

A full summary of the analyses across DBAS-16 components is presented in Table 3.

Table 3.

DBAS-16: Global Score, Subscales, and Individual Items

DBAS-16: Group Effect and Pairwise Comparisons
DBAS-16	MDD+/HYP+	MDD+/HYP−	HC	F(2,63) (p value)	MDD+/HYP+ vs. MDD+/HYP−	MDD+/HYP+ vs HC	MDD+/HYP− vs HC
Global Score	79.3 ± 19.3	78.8 ± 27.8	45.4 ± 16.0	17.8 (<0.0001)	-	<0.0001	<0.0001
Subscale 1: Consequences	5.88 ± 1.51	5.02 ± 2.24	3.95 ± 1.72	6.02 (0.004)	-	0.003	-
Subscale 2: Worry	4.21 ± 2.02	5.58 ± 2.20	1.20 ± 1.14	32.5 (<0.0001)	0.05	<0.0001	<0.0001
Subscale 3: Expectations	7.57 ± 2.08	5.16 ± 2.61	6.09 ± 1.94	6.53 (0.003)	0.002	-	-
Subscale 4: Attributions	3.32 ± 1.52	3.29 ± 1.68	2.18 ± 1.39	3.83 (0.03)	-	-	-
Item 1	7.36 ± 2.68	5.77 ± 3.13	6.82 ± 3.54	1.84 (−)	N/A	N/A	N/A
Item 2	7.77 ± 1.90	4.55 ± 2.94	5.36 ± 2.36	10.4 (<0.0001)	0.0001	0.005	-
Item 3	3.64 ± 3.58	6.64 ± 2.71	0.96 ± 2.19	21.4 (0.003)	0.003	0.009	<0.0001
Item 4	3.27 ± 2.78	5.59 ± 2.96	0.46 ± 1.11	24.1 (<0.0001)	0.007	0.0008	<0.0001
Item 5	8.68 ± 1.36	6.82 ± 2.82	5.91 ± 2.65	8.07 (0.0008)	0.03	0.0006	-
Item 6	2.96 ± 3.37	3.64 ± 3.29	2.68 ± 2.46	2.3 (−)	N/A	N/A	N/A
Item 7	5.46 ± 2.69	4.96 ± 2.46	4.19 ± 2.63	1.35 (−)	N/A	N/A	N/A
Item 8	4.55 ± 2.87	3.86 ± 2.53	1.32 ± 1.78	10.7 (0.0001)	N/A	N/A	N/A
Item 9	6.36 ± 2.22	3.73 ± 2.73	2.96 ± 2.24	12.0 (<0.0001)	0.002	<0.0001	-
Item 10	5.23 ± 3.18	6.64 ± 2.84	2.32 ± 2.50	13.1 (<0.0001)	0.32	0.004	<0.0001
Item 11	5.91 ± 2.29	6.32 ± 2.94	2.00 ± 2.09	20.6 (<0.0001)	-	<0.0001	<0.0001
Item 12	5.14 ± 2.59	5.82 ± 2.11	5.23 ± 2.39	0.54 (−)	N/A	N/A	N/A
Item 13	5.19 ± 1.72	4.41 ± 1.79	4.10 ± 2.43	1.67 (−)	N/A	N/A	N/A
Item 14	2.64 ± 2.65	4.46 ± 2.76	0.14 ± 0.35	21.1 (<0.0001)	0.03	0.001	<0.0001
Item 15	1.59 ± 1.92	1.82 ± 1.68	0.64 ± 1.18	3.29 (0.04)	-	-	0.06
Item 16	3.77 ± 3.19	3.77 ± 3.32	1.50 ± 1.86	4.61 (0.01)	-	0.03	0.03

DBAS-16 global, subscale, and individual items scores are presented for comorbid depression and hypersomnolence (MDD+/HYP+), depression without hypersomnolence (MDD+/HYP−), and healthy controls (HC). Values are provided as means ± standard deviation. F- Statistic and corresponding p value are documented for the main effect of group. Bonferroni corrected p values are provided for pairwise comparisons, when applicable. A hyphen (−) denotes a nonsignificant result. N/A indicates that pairwise comparisons were not performed due to a nonsignificant group effect.

3.3. Group Comparisons: DBAS-16 - Controlling for depression symptom severity

The significant group effect for DBAS-16 global score was maintained in analyses that controlled for depression symptom severity. However, groups were no longer distinguishably different in post-hoc, pairwise comparisons using global DBAS-16 scores. Significance was also preserved for group effect across all DBAS-16 subscales.

A main significant group effect was identified for the same 11 DBAS individual items observed in analyses unadjusted for depression. MDD+/HYP+ maintained its uniquely high rating, relative to both MDD+/HYP− and HC, for item 2, and was still significantly greater than MDD+/HYP− for items 5 and 9.

A full summary of the analyses across DBAS-16 components is presented in Table 4.

Table 4.

DBAS-16 Group Effect and Pairwise Comparisons: Controlling for depression symptom severity

DBAS-16: Group Effect and Pairwise Comparisons Controlling for Depression Symptom Severity
DBAS-16	Group Effect F(2,63)	Group Effect p value	MDD+/HYP+ vs. MDD+/HYP− p value	MDD+/HYP+ vs HC p value	MDD+/HYP− vs HC p value
Global Score	18.0	<0.0001	-	-	-
Subscale 1: Consequences	6.05	0.004	-	-	-
Subscale 2: Worry	34.1	<0.0001	-	-	0.008
Subscale 3: Expectations	6.48	0.003	0.003	-	-
Subscale 4: Attributions	3.81	0.028	-	-	-
Item 1	1.81	-	N/A	N/A	N/A
Item 2	10.4	0.0001	0.0002	0.02	-
Item 3	21.2	<0.0001	0.004	-	0.004
Item 4	27.1	<0.0001	0.008	-	-
Item 5	8.07	0.0008	0.03	-	-
Item 6	2.29	-	N/A	N/A	N/A
Item 7	1.35	-	N/A	N/A	N/A
Item 8	10.8	<0.0001	-	-	-
Item 9	12.0	<0.0001	0.002	-	-
Item 10	12.3	<0.0001	-	-	-
Item 11	21.0	<0.0001	-	-	-
Item 12	1.02	-	N/A	N/A	N/A
Item 13	1.65	-	N/A	N/A	N/A
Item 14	22.1	<0.0001	0.03	-	-
Item 15	3.27	0.05	-	-	-
Item 16	4.60	0.01	-	-	-

Results from analyses controlling for depression symptom severity are presented for group and pairwise DBAS-16 comparisons. F-Statistic and corresponding p value for the main group effect are documented. Bonferroni corrected p values are provided for pairwise comparisons, when applicable. A hyphen (−) denotes a nonsignificant result. N/A indicates that pairwise comparisons were not performed due to a nonsignificant group effect.

4. Discussion

The primary results of this investigation suggest the presence of potentially maladaptive sleep-related cognitions (SRCs) in persons with depression and comorbid hypersomnolence (MDD+/HYP+). Global DBAS-16 scores were elevated in the MDD+/HYP+ sample relative to the healthy control (HC) group, and were comparable to depression without hypersomnolence (MDD+/HYP−). When adjusted for depression severity, a significant main group effect on DBAS-16 global score was still observed, however, global DBAS-16 score for MDD+/HYP+ was no longer significantly different compared to HC in post hoc comparisons. However, individual items on the DBAS-16 were elevated for MDD+/HYP+ compared to HC and/or MDD+/HYP− groups, even when adjusted for depression severity. These results suggest that discrete components of the DBAS-16 may be more relevant to persons experiencing hypersomnolence and co-occurring depressive illness.

In many ways, it is not surprising that components of the DBAS-16, rather than the overall scale, may be more relevant to persons with hypersomnolence. The DBAS-16 was initially designed to assess dysfunctional beliefs and attitudes about sleep in persons with insomnia (Carney et al., 2010; Morin et al., 2007). Therefore, while the metric is currently the best available measure for evaluating SRCs, its applicability to those experiencing hypersomnolence is likely quite limited. While these findings suggest there may be potentially maladaptive SRCs in persons with hypersomnolence, translating these results into treatment strategies that impact the care of persons with hypersomnolence will require several levels of clarification and further investigation.

It is noteworthy that elevated responses (strong agreeance) on some of the particular DBAS-16 items may not reflect dysfunctional SRCs, but rather valid reports of hypersomnolence symptomatology. Items 2, 5, and 9 assess beliefs about the quantity and quality of sleep and their resulting impact of functioning, which may be valid reflections of hypersomnolence symptomatology relating to sleep propensity and impaired functionality, rather than dysfunctional cognitions, per se. In this context, the field would benefit from a novel metric designed to assess SRCs that may be more appropriate for and specific to persons with hypersomnolence.

The optimal content of focus for such a novel metric has not been delineated. While many perceive hypersomnolence to be chronic and uncurable, relatively high remission rates have been described for many hypersomnolence conditions outside of type 1 narcolepsy (Kim, Lee, Lee, & Yoon, 2016). In this context, inventory items that assess feelings of hopelessness related to improvement of hypersomnolence symptoms may be particularly salient. Also, given the nonrestorative aspects of sleep for many persons with hypersomnolence, items that quantify the perceived futility of sleep may be useful addressing perceptions that might inhibit effective treatment. Going forward, qualitative studies would provide increased clarity regarding relevant content that might be included in a hypersomnolence-specific SRC measure, including patient-focused interviews to gain insight into relevant SRC themes among these patients. Another approach could utilize practitioners specializing in hypersomnolence who develop an initial inventory based on their clinical experience. After developing an initial inventory, quantitative research and statistical refinement would be utilized to construct a deliverable metric (Boateng, Neilands, Frongillo, Melgar-Quiñonez, & Young, 2018). This metric would then undergo scale evaluation to determine reliability and validity as a tool to measure SRCs in persons with hypersomnolence. Regardless of the specific approaches used, it is clear that the DBAS-16 is not sufficient for individuals with hypersomnolence complaints, and that behavioral sleep medicine would benefit from a measure tailored to evaluate potentially maladaptive cognitions in these patients.

There are limitations to this investigation that are worthy of discussion. First, participant recruitment resulted in an unequal sex distribution (81.6% female) within each of the samples. Although female predominance is expected within depression (Kessler et al., 2003) and hypersomnolence (Leu-Semenescu, Louis, & Arnulf, 2016), the dearth of male participants limits the generalizability of findings among men. Second, the relatively small sample size utilized for each group limits statistical power to detect differences between groups. However, the fact that participants were thoroughly assessed clinically and with ad libitum polysomnography, suggests groups were well characterized and appropriately categorized, which strengthens the overall study results. In addition, recent data suggest good inter-rater reliability for DSM-5 defined hypersomnolence disorder, and excellent interrater reliability for insomnia disorder, strengthening the presupposition that these sleep problems can be distinguished from one another by clinical history (Taylor et al., 2018). However, the lack of multiple sleep latency testing among participants prevents extension of these results to other nosological systems such as the International Classification of Sleep Disorders. Lastly, this investigation did not have access to DBAS-16 data from a sample of persons experiencing hypersomnolence absent of depression or a chronic insomnia sample for comparison. Future research that includes such groups, particularly in the development of hypersomnolence specific SRC metrics, is clearly warranted.

In summary, this investigation is the first to empirically assess sleep-related cognitions (SRCs) in persons with depression and comorbid hypersomnolence. After adjusting for depression severity, some DBAS-16 items related to sleep quality and quantity appeared particularly salient to persons with depression and co-occurring hypersomnolence. Future research that expands upon these findings to develop a more appropriate measure of potentially maladaptive sleep related cognitions in persons with hypersomnolence, both with and without mood disorders, is indicated.