Transient Insomnia Versus Chronic Insomnia: A Comparison Study of Sleep-Related Psychological/Behavioral Characteristics Chien-Ming Yang,1 Shih-Chun Lin,1 and Chung-Ping Cheng2 1 2
National Chengchi University, Taiwan National Cheng Kung University, Taiwan
Objectives:
Vulnerability to transient insomnia is regarded as a predisposing factor for chronic insomnia. However, most individuals with transient insomnia do not develop chronic insomnia. The current study investigated the differential contributing factors for these two conditions to further the understanding of this phenomenon. Method: Chronic insomnia patients and normal sleepers with high and low vulnerability to transient insomnia completed measures of pre-sleep arousal, dysfunctional sleep beliefs, and sleep-related safety behaviors. Results: Both cognitive and somatic pre-sleep arousals were identified as significant predictors for transient insomnia. Dysfunctional beliefs regarding worry about insomnia and cognitive arousal were predictors for chronic insomnia. Sleep-related safety behavior, although correlated with insomnia severity, was not a significant predictor for both conditions. Conclusions: Dysfunctional beliefs associated with worry and losing control over sleep are the most critical factors in differentiating chronic insomnia from transient insomnia. These factors should be addressed to help prevent individuals with high sleep vulnerability from developing chronic sleep C 2013 Wiley Periodicals, Inc. J. Clin. Psychol. 00:1–14, 2013. disturbance. ! Keywords: insomnia; acute insomnia; transient insomnia; chronic insomnia; dysfunctional sleep belief; safety behavior; pre-sleep arousal
Insomnia is among the most common health-related complaints. A recent review of data from different countries reported that approximately one third of the adult population has reported insomnia symptoms, 9%-15% has shown sleep difficulties with daytime consequences, and approximately 6% has shown symptoms of diagnosable insomnia (Ohayon, 2002). Insomnia can significantly affect not only nighttime sleep but also daytime functioning (e.g., Godet-Cayr´e et al., 2006; Roth & Ancoli-Israel, 1999; Ustinov et al., 2009; see Riedel & Lichstein, 2000, for a review). Insomnia can have adverse repercussions for both the individual and society. These include the high-direct costs for its evaluation and treatment, and the indirect costs related to increased medical and psychiatric comorbidity, increased rates of absenteeism and work-related accidents, and reduced productivity (Chilcott & Shapiro, 1996; L´eger & Bayon, 2010; Martin, Aikens, & Chervin, 2004; Stoller, 1994; Walsh & Engelhart, 1999). Despite the negative effects of insomnia, sleeplessness itself is not necessarily a disorder. Acute insomnia is a common experience for most people who experience stress. It can even be an adaptive reaction to stress in life-threatening situations, because sleep can make one vulnerable to threats in ecological situations. After the stressor has been dissipated, the individual should then default back to normal sleep due to the plastic and automatic nature of normal sleep regulation. However, in patients with chronic insomnia (CI), sleep does not normalize because of the activation of other processes that interfere with sleep regulation (Espie, 2002; Spielman, 1986). Empirical data have also supported the view that a significant proportion of cases of transient and/or milder insomnia might not develop into chronic and/or severe insomnia (Morin et al., This study is partially supported by the National Science Council of Taiwan (Grant No. NSC98-2410-H004-023-MY3). Please address correspondence to: Chien-Ming Yang, Department of Psychology and Research Center for Mind, Brain and Learning, National Chengchi University, 64 Sec. 2 Chih-Nan Rd., Taipei 116, Taiwan. E-mail:
[email protected] JOURNAL OF CLINICAL PSYCHOLOGY, Vol. 00(0), 1–14 (2013) Published online in Wiley Online Library (wileyonlinelibrary.com/journal/jclp).
! C 2013 Wiley Periodicals, Inc. DOI: 10.1002/jclp.22000
2
Journal of Clinical Psychology, xxx 2013
2009). A 3-year longitudinal study followed 119 participants with insomnia syndrome that fulfilled the Diagnostic and Statistical manual of Mental Disorders Fourth Edition (DSM-IV; APA, 2000) diagnostic criteria, and 269 with subsyndromal insomnia symptoms that did not fulfill all diagnostic criteria (i.e., reported insomnia symptoms but were satisfied with their sleep, were free of distress or daytime consequences, or suffered symptoms lasting less than 1 month). Individuals with insomnia symptoms at baseline had a remission rate of 63%, with 33% of the remitters relapsing. More importantly, only about 10%-14% of the participants with subsyndromal insomnia at baseline reported insomnia syndrome at one of the follow-up assessments. On the other hand, participants with insomnia syndrome at baseline had a remission rate of 34%, with 11% of the remitters eventually relapsing. The most common trajectory in both groups was to have a consistent status at all subsequent follow-up assessments. While insomnia syndrome tends to be chronic in course in most patients, milder insomnia symptoms tend to be transient in nature. Only a small proportion of cases of milder insomnia develop into CI. Therefore, it is of interest to explore the differential contributing factors for transient and CI to understand why only some transient sleep disturbances develop into chronic problems. Prophylactic strategies and intervention programs targeting the different groups can then be developed accordingly to reduce the costs of insomnia to both individuals and society. Another longitudinal study also conducted by Morin’s group addressed this issue by examining which variables predicted the incidence of insomnia symptoms and insomnia syndrome within a 1-year follow-up period. Among the 464 good sleepers followed, 7.3% had developed insomnia syndrome and 30.7% had developed subsyndromal symptoms of insomnia at the end of the follow-up period (LeBlanc et al., 2009). The number of life events and negative events experienced and their intensity within the last 6 months were higher in both groups of participants with insomnia symptoms and insomnia syndrome than in good sleepers. In addition, the participants who developed insomnia syndrome were characterized by premorbid elevations in anxiety, depression, and arousability and lowered extraversion; these trends were not found in those with insomnia symptoms. The results can be interpreted under the framework of the 3-P model proposed by Spielman (1986). Although stress may play a major role in precipitating the onset of sleep disturbance in both groups, cases tended to develop into diagnosable insomnia in those who had the predisposing characteristics of emotional disturbances and hyperarousal. According to the 3-P model, for individuals with these predisposing factors, stress-related transient insomnia may further elicit maladaptive coping behaviors and emotional arousal that perpetuate transient sleep disturbances into chronic sleep problems. Although this study identified possible predisposing characteristics for insomnia syndrome, the perpetuating factors were not addressed in this study and require further investigation. Several etiological models of insomnia have been proposed to illustrate the interactions among psychological and behavioral factors that lead to the perpetuation of insomnia (e.g., Espie, 2006; Harvey, 2002a; Morin, 1993 ; Perlis, Giles, Mendelson, Bootzin, & Wyattett, 1997). For example, Morin (1993) raised the importance of dysfunctional beliefs about sleep in the maintenance of CI. Such beliefs are thought to form a vicious cycle with maladaptive sleep habits, pre-sleep arousal, and the consequences of sleeplessness that exacerbates and perpetuates sleep disturbances. The association between dysfunctional sleep beliefs and insomnia was supported by the finding that more dysfunctional beliefs about sleep were reported in insomnia patients than in good sleepers (Carney & Edinger, 2006; Edinger, Fins, & Glenn, 2000; Morin, Stone, Trinkle, Mercer, & Remsberg, 1993). Further, reductions in dysfunctional sleep beliefs are associated with greater improvements in sleep after cognitive behavioral therapy for insomnia (Carney & Edinger, 2006; Morin, Blais, & Savard, 2002). Harvey (2002a) further theorized that dysfunctional sleep beliefs may lead to “sleep-related safety behaviors”—those meant to avoid the consequences of sleeplessness but in contrast prevent individuals from disconfirming their dysfunctional beliefs and thereby increase the likelihood of sleep disruptions (e.g., clock watching during the night, monitoring of the expected consequences of sleeplessness after a night of poor sleep). Subsequent studies supported this view by showing higher prevalences of sleep-related safety behaviors and self-monitoring in
Transient Insomnia Versus Chronic Insomnia
3
individuals with insomnia than in good sleepers (Harvey, 2002b; Ree & Harvey, 2004; Semler & Harvey, 2004; Woodley & Smith, 2006). Another major etiological factor suggested to perpetuate insomnia is hyperarousal. The threat of sleeplessness in insomnia patients was thought to change their cognitive processes and increase their pre-sleep cognitive activity levels; this conditioned arousal may interfere with the process of falling asleep (Espie, Broomfield, MacMahon, Macphee, & Taylor, 2006; Harvey, 2002a; Perlis et al., 1997). Research has shown evidence of attentional bias toward sleep-related stimuli (Jones, Macphee, Broomfield, Jones, & Espie, 2005; MacMahon, Broomfield, & Espie, 2006; Marchetti, Biello, Broomfield, Macmahon, & Espie, 2006; Spiegelhalder, Espie, Nissen, & Riemann, 2008; Woods, Marchetti, Biello, & Espie, 2009) as well as increased pre-sleep arousal and cognitive activities prior to sleep (Harvey, 2000; Nelson & Harvey, 2003; Nicassio, Mendlowitz, Fussell, & Petras, 1985; Perlis, Smith, Orff, Andrews, & Giles, 2001; Wicklow & Espie, 2000) in insomnia patients. Although the above-mentioned research findings did support that these sleep-related psychological and behavioral factors are associated with insomnia, it is not clear whether these variables are general factors associated with sleep disturbances or are particularly important in the perpetuation of insomnia. One way to further clarify this issue is to compare these factors between individuals who are vulnerable to transient insomnia and those who have developed CI. Levels of the perpetuating factors are expected to be higher in CI patients than in good sleepers and individuals vulnerable to transient sleep disturbances. The Ford Insomnia Response to Stress Test (FIRST) was recently developed to assess vulnerability to stress-related transient sleep disturbance; it asks subjects to rate their likelihood of experiencing sleep disturbances in reaction to various daily life stressors (Drake, Richardson, Roehrs, Scofield, & Roth, 2004 ). Scores on the FIRST were shown to predict higher levels of sleep disturbance during the first night of sleeping in a laboratory and in reaction to a caffeine challenge (Drake et al., 2004; Drake, Jefferson, Roehrs, & Roth, 2006). Recent sibling and twin studies also demonstrated moderate genetic influence in the FIRST score (Drake, Scofield, & Roth, 2008; Drake, Friedman, Wright, & Roth, 2011). These results suggest that FIRST scores may reflect the individual trait of sleep reactivity to stress. In one study, higher sleep vulnerability was found to be associated with higher arousability, perceived stress, rumination, extraversion, and suppression among middle-aged sleepers and higher arousability, pre-sleep arousal, emotion-oriented coping, and rumination among youngadult good sleepers (Fernandez-Mendoza et al., 2010); in another study, it was associated with dysfunctional sleep beliefs in young adults who reported good sleep (Yang, Chou, & Hsiao, 2011). These studies suggested that cognitive–emotional hyperarousal, emotion-oriented coping, perceived stress, and dysfunctional sleep beliefs might all modulate vulnerability to stress-related insomnia. However, these studies did not include comparison groups of patients with CI; therefore, we are unable to evaluate the relative contributions of these factors to transient and chronic sleep disturbances. For the present study, we took a further step by comparing sleep-related psychological and behavioral factors among good sleepers, individuals vulnerable to transient insomnia, and CI patients. On the bases of the foregoing literature review and etiological models described previously, we selected dysfunctional sleep beliefs, pre-sleep arousal, and sleep-related safety behaviors as the variables of interest. Furthermore, levels of anxiety and depression were assessed to control for possible confounding due to emotional disturbances. Because the trait of sleep vulnerability may reflect higher arousability in response to stress (as it was associated with arousability in previous studies; Fernandez-Mendoza et al., 2010), we hypothesized that pre-sleep arousal is higher in individuals with high sleep vulnerability than those with low vulnerability. Pre-sleep arousal is expected to increase further as the insomnia becomes chronic in course; therefore, pre-sleep arousal was also hypothesized to be highest in insomnia patients. Similarly, dysfunctional sleep beliefs and sleep-related safety behaviors are thought to play a major role in perpetuating insomnia; therefore, they were hypothesized to be more prevalent in CI patients than in the noninsomnia groups. Dysfunctional sleep beliefs are also hypothesized to be higher—but to a lesser degree—in the high-sleep-vulnerability group than in the low-sleep vulnerability group because a significant correlation between sleep vulnerability and
4
Journal of Clinical Psychology, xxx 2013
dysfunctional beliefs was reported in a previous study (Yang et al., 2011). On the other hand, sleep-related safety behaviors were hypothesized to show no difference in prevalence between individuals low and high in sleep vulnerability, as those behaviors are considered as maladaptive coping strategies to the threat of sleep problems. Because transient insomnia is less likely to be considered as a threat, we expected no significant difference between the high-vulnerability (HV) and low-vulnerability (LV) groups in terms of this measure.
Methods Subjects The subjects included 50 patients with CI (16 men and 34 women; mean age = 44.3 [10.6] years) and 102 normal sleepers (29 men and 73 women; mean age = 45.92 [8.4] years). CI patients were recruited from outpatient clinics of a general hospital and were evaluated and referred by a psychiatrist who was certified in both psychiatry and sleep medicine. Insomnia diagnoses were based on the criteria for primary insomnia in the DSM-IV-TR (APA, 2000), except that we altered the duration criterion to longer than 6 months to include chronic patients only. The inclusion criteria for the subjects with CI were as follows: (a) predominant complaint of difficulty initiating or maintaining sleep or nonrestorative sleep for at least 6 months; (b) reported significant distress or impairment in daytime functioning associated with sleep disturbance; (c) no current or past history of other sleep disorders, as evaluated by clinical interview; (d) no current or past history of other mental disorders, as evaluated by clinical interview; (e) no current or past history of substance abuse reported; and (f) no current medical conditions and/or medication use that might affect sleep. The normal sleepers were recruited from the community by flyers and word of mouth and were evaluated by a clinical psychology graduate student under the supervision of a licensed clinical psychologist, who was certified in behavioral sleep medicine. The inclusion criteria for normal sleepers were (a) no reported symptoms of insomnia, (b) no current or past history of any sleep disorders including insomnia, (b) no current or past history of psychiatric disorders, (c) no current major medical conditions, and (d) no current use of substances/medications that may affect sleep. In addition, potential subjects who scored >19 on the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977) and/or >13 on the Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988) were excluded from participation. The normal sleepers were further divided into a LV group (FIRST ≤ 16; n = 51; 31 women and 20 men) and a HV group (HV; FIRST > 16; n = 51; 36 women and 15 men) by a median-split according to their FIRST scores. The median-split was applied to categorize the normal sleepers, because a validated cutoff score has not yet been established for the FIRST; this method was used in previous studies and was shown to track differences in sleep-related characteristics between the split groups (Drake et al., 2004; Drake et al., 2006).
Procedure A clinical interview was conducted with each potential subject by a psychiatrist and/or a trained graduate student to confirm the diagnosis and rule out possible comorbid disorders. The CESD, BAI, and Pittsburgh Sleep Quality Inventory (PSQI; Buysse, Reynolds, Monk, Berman, & Kupfer, 1989) were also administered to assist screening for depression, anxiety disorders, and possible sleep disorders and to serve as control variables for emotional disturbances. Those who passed the screening were asked to complete a battery of self-rating scales, including the FIRST (Drake et al, 2004), the Insomnia Severity Index (ISI; Morin, 1993; Bastien, Vallieres, & Morin, 2001), a brief version of the Dysfunctional Beliefs and Attitudes about Sleep Scale (DBAS-16; Morin, 1993), the Sleep-Related Behaviors Questionnaire (SRBQ; Ree & Harvey, 2004), and the Pre-Sleep Arousal Scale (PSAS; Nicassio et al., 1985). The COPE scale –which assesses general styles of stress coping–was also administered, but the data are not reported in the present article. Informed consent was obtained from all subjects before they completed the questionnaires. The
Transient Insomnia Versus Chronic Insomnia
5
study procedure was conducted according to the ethical standards of the Taiwan Psychological Association and was approved by the departmental Ethics Committee for Human Study.
Measurement FIRST (Drake et al., 2004). The FIRST is a 9-item questionnaire designed to quantify the degree of vulnerability to stress-related sleep disturbance as an individual trait. For each item, subjects were asked to rate the likelihood of experiencing sleep disturbances in response to a common stressful situation in daily life (e.g., having an important meeting the next day, watching a frightening movie or TV show, or an argument) on a 4-point Likert-scale ranging from 1 (not likely) to 4 (very likely). The test-retest reliability coefficient was r = .92 with a 2-week interval between test administrations. Cronbach’s α for the 9-item scale was .83, indicating a high level of internal consistency (Drake et al., 2004). ISI. The ISI is a 7-item Likert-type self-rating scale designed to assess subjective perception of the severity of insomnia (Morin, 1993). The scale contains items that measure the symptoms, associated features, and impacts of insomnia, including difficulty falling asleep, difficulty maintaining sleep, early morning awakening, dissatisfaction with sleep, concerns regarding insomnia, and functional impacts of insomnia. The total score can be used to differentiate patients with and without insomnia and categorize patients with different levels of insomnia severity. The scale has adequate internal consistency (Cronbach’s α = .74; Bastien et al., 2001). It also has small, moderate correlations with polysomnographic indices of sleep quality (r = .32–.55). DBAS-16. The original DBAS comprises 30 items concerning beliefs, attitudes, expectations, and attributions regarding sleep and insomnia (Morin, 1993). Subjects rated their level of agreement with each item on a 10-point scale. A brief version of the DBAS has recently been developed and validated by Morin and colleagues (Morin, Vallieres, & Ivers, 2007); it includes subfactors representing four domains: (a) perceived consequences of insomnia, (b) worry and helplessness about insomnia, (c) sleep expectations, and (d) beliefs about medication. The scale has good internal consistency (Cronbach’s α = .79) and test-retest reliability (r = .83). It was further validated among sufferers of primary insomnia and comorbid insomnia and among long-term hypnotics users in a multisite study employing a larger sample size (Carney et al., 2010). SRBQ. The SRBQ was designed to measure sleep-related safety behaviors in which insomnia patients may engage to cope with their sleep problems (Ree & Harvey, 2004). It comprises 32 items rated on a 5-point Likert scale and investigates the presence of sleep-interfering behaviors intended to compensate for sleep loss, such as clock watching, trying to stop all thinking while in bed, or catching up on sleep by napping. The SRBQ has good internal consistency (Cronbach’s α = .92) and can discriminate normal sleepers from patients with insomnia. PSAS. The PSAS comprises 16 items designed to measure the level of arousal prior to sleep (Nicassio et al., 1985). The items are further divided into two subscales: somatic arousal and cognitive arousal. Previous study has shown that the PSAS has good internal consistency (Cronbach’s α = .79-.84 for the somatic subscale; α = .67–.88 for the cognitive subscale) and test-rest reliability with a 3-month interval (r = .76 for somatic arousal; r = .72 for the cognitive subscale). PSQI. The PSQI contains 19 items designed to measure various aspects of sleep quality and sleep disturbances during a 1-month period (Buysse et al., 1989). A global sleep quality score ranging 0–21 can be derived from this inventory, with higher scores indicating poorer sleep quality. The PSQI global score has good internal consistency (Cronbach’s α = .83) and test-retest reliability (r = .85). A cutoff score of greater than 5 was found to correctly identify 88.5% of patients with sleep disturbances.
6
Journal of Clinical Psychology, xxx 2013
CES-D (Radloff, 1977). The CES-D is a 20-item self-rating scale designed to measure self-reported symptoms associated with depression experienced within the past week. It comprises six subscales reflecting major dimensions of depression: depressed mood, feelings of guilt and worthlessness, feelings of helplessness and hopelessness, psychomotor retardation, loss of appetite, and sleep disturbance. Higher item and total scores indicate more depressive symptoms. High internal consistency has been reported across studies (Cronbach’s α = .85–.90). A score of 16 or higher was originally suggested as this scale’s cutoff point for high depressive symptoms. However, that cutoff was shown to have low specificity, particularly for participants in Asian countries. Therefore, in accordance with previous studies on Asian populations, a cutoff score of 19 was used for the present study (Wada et al., 2007; Kim, Decoster, Huang, & Chiriboga, 2011). BAI (Beck et al., 1988; Beck & Steer, 1993). The BAI is a 21-item self-rating scale that measures the severity of subjective anxiety. The items comprise descriptive statements of anxiety symptoms, and subjects rated how applicable these symptoms were to them on a 4-point Likert scale. The scale has good internal consistency for adults (Cronbach’s α = .92-.94) and good test-retest reliability across a 1-week interval (r = .75). Concurrent validity was also documented through the demonstration of significant correlations with other scales for anxiety, including the Hamilton Anxiety Rating Scale, the anxiety subscale of the Cognition Check List, and the State-Trait Anxiety Inventory. The total score on the BAI is the sum of the ratings of the 21 symptoms. A total score of 13 has been recommended as the cutoff score for anxiety problems. Data Analysis One-way ANOVAs (analyses of variance) were first conducted to compare background information and BAI and CES-D scores among the HV, LV, and CI groups. Tukey’s honestly significant difference method was applied for post hoc comparisons when significant main effects were found. Multivariate analyses of covariance (MANCOVAs) and analyses of covariance (ANCOVAs) were then conducted to compare scores on the rating scales of sleep-related factors among the three groups, with BAI and CES-D used as covariates to control for the influence of emotional disturbances. In addition, Pearson’s partial correlations between ISI scores and the various measures of sleep-related psychological/behavioral factors were calculated separately for patients with insomnia and normal sleepers, with BAI and CES-D scores used as control variables. Hierarchical multiple regression analyses were then conducted in two stages. In the first stage, BAI and CES-D scores were entered into the regression model to control for the effects of emotional disturbances; in the second stage, the sleep-related psychological and behavioral variables that correlated significantly with ISI scores were entered into the model stepwise to identify which variables best predict the severity of insomnia symptoms. All statistics were conducted using the SPSS version 10 software package. A value of α = .05 was used as the criterion for statistical significance.
Results Demographic Data, Mood, and Sleep Ratings Demographic data, mood ratings on the CES-D and BAI, and sleep ratings on the ISI, PSQI, and FIRST for the three groups are presented in Table 1. There were no significant differences in age, F(2, 149) = 1.223, n.s., or years of education, F(2, 149) = 1.255, n.s., among the three groups. Significant main effects of scores on the FIRST, ISI, PSQI, CES-D, and BAI were found, Fs(2, 149) = 75.035, 79.534, 138.841, 25.497, 29.411, respectively; ps < .001. Post hoc comparisons revealed that FIRST scores were higher in the HV group than the LV group, with no difference between the HV group and the CI group. The scores on the rest of the scales were higher in the CI group than the other two groups and showed no differences between the HV and LV groups.
Transient Insomnia Versus Chronic Insomnia
7
Table 1
Demographic and Clinical Characteristics and Comparison of Variables Among Low-Vulnerability (LV), High-Vulnerability (HV), and Chronic Insomnia (CI) Groups Variables/group Age (y) Education (y) FIRST CES-D BAI PSQI ISI
LV group 46.08 13.43 12.62 8.14 2.64 7.45 4.08
± ± ± ± ± ± ±
8.59 2.50 2.29 5.26 4.68 2.07 3.64
HV group 43.12 13.59 22.46 11.35 6.08 8.46 6.12
± ± ± ± ± ± ±
9.66 2.59 4.03 6.35 7.18 2.10 4.78
CI group 44.28 12.81 22.33 18.82 14.19 14.82 15.04
± ± ± ± ± ± ±
10.57 2.60 6.57 10.31 10.16 2.71 6.39
F
Effect sizea
1.223 1.255 75.035** 25.497** 29.411** 138.841** 66.433**
.016 .017 .512 .260 .289 .670 .473
Post hoc comparisons
LV < HV = CI LV = HV < CI LV = HV < CI LV = HV < CI LV = HV < CI
Note. FIRST = Ford Insomnia Response to Stress Test; CES-D = Center for Epidemiologic Studies Depression Scale; BAI = Beck Anxiety Inventory; PSQI = Pittsburgh Sleep Quality Index; ISI = Insomnia Severity Index. a Eta squared. **p < .001.
Comparisons of Psychological and Behavioral Factors Among Groups A one-way MANCOVA with CES-D and BAI applied as covariates showed significant differences among the three groups on the DBAS-16, F(8, 276) = 9.324, p < .001. One-way ANCOVAs were then conducted to compare the DBAS-16 subscale scores among the groups, controlling for the influences of CES-D and BAI scores. Significant differences were found for the “worry/helplessness about insomnia” and “beliefs about medication” subscales, Fs(2, 142) = 9.931 and 28.737, respectively, ps < .001, but not for the “perceived consequences of insomnia” and “sleep expectations” subscales, Fs(2, 142) = 1.165 and 1.140, n.s. Post hoc comparisons showed that for “worry/helplessness about insomnia,” the CI group scored significantly higher than the other two groups. For the “beliefs about medication” subscale, post hoc comparisons revealed a significantly higher rating for the CI group than the other two groups and no significant difference between the LV and HV groups (see Table 2). For the SRBQ, a one-way ANCOVA with BAI and CES-D used as control variables showed significant between-group differences, F(2, 149) = 7.362, p = .001. Post hoc comparisons indicated that the CI group scored significantly higher than the other two groups; further, the HV group scored higher than the LV group (see Table 2). In terms of pre-sleep arousal levels, a one-way MANOVA that controlled for the CES-D and BAI showed significant differences among the three groups on the PSAS, F(4, 282) = 15.137, p < .001. One-way ANCOVAs were then conducted for each subscale; these showed significant between-group differences for the cognitive arousal subscale, F(2, 143) = 26.903, p < .001, but not for the somatic arousal subscale, F(2, 143) = .296, n.s. For the cognitive arousal subscale, post hoc comparisons revealed that the CI group scored higher than the other two groups and that the HV group scored higher than the LV group (see Table 2).
Partial Correlations and Hierarchical Regressions Among the noninsomnia participants, after controlling for BAI and CES-D scores, ISI scores were correlated significantly with scores on the SRBQ, r(96) = .325, p = .001, and both the somatic, r(96) = .454, p < .001, and cognitive, r(96) = .496, p < .001, subscales of the PSAS. ISI scores were also correlated significantly with the “worry/helplessness about insomnia” subscale of the DBAS-16, r(96) = .294, p < .005, but not with the other subscales (see Table 3). The first step of multiple hierarchical regression analysis showed that CES-D and BAI scores account for a significant proportion of the variance in ISI, R2 = .233, F(2, 99) = 15.033, p < .001; the second step subsequently identified the cognitive and somatic subscales of the PSAS, cognitive
8
Journal of Clinical Psychology, xxx 2013
Table 2
Psychological/Behavioral Measures and Corrected Comparison of Variables Among LowVulnerability (LV), High-Vulnerability (HV), and Chronic Insomnia (CI) Groups CI group
F
Effect sizea
22.64 ± 11.02 25.49 ± 9.13
28.08 ± 10.34
1.165
.016
23.22 ± 10.30 27.84 ± 8.03
37.76 ± 11.41
9.931**
.123
1.140 28.737**
.016 .288
15.26 ± 5.42 .294 23.54 ± 6.86 26.903** 53.46 ± 14.79 7.362*
.004 .273 .093
Variable/group DBAS-16 Perceived consequences Worry about insomnia Sleep expectations Medication PSAS Somatic arousal Cognitive arousal SRBQ
LV group
13.28 ± 5.06 4.04 ± 4.70
HV group
13.62 ± 4.67 4.000 ± 3.93
12.40 ± 4.96 12.640 ± 5.15
10.10 ± 2.68 11.30 ± 3.69 11.40 ± 3.64 15.42 ± 4.60 28.76 ± 20.34 37.96 ± 11.83
Post hoc comparisons
LV = HV < CI LV = HV < CI LV < HV < CI LV < HV < CI
Note. Scores on the Beck Anxiety Inventory and the Center for Epidemiologic Studies Depression Scale were used as covariates to control for the influences of anxiety and depression symptoms, respectively.PSAS = Pre-sleep arousal scale; DBAS-16 = Dysfunctional Beliefs and Attitudes about Sleep, 16-item version; SRBQ = Sleep-Related Behavior Questionnaire. a Partial eta squared. *p < .01. **p < .001.
Table 3
Pearson’s Partial Correlations Between Scores on the Insomnia Severity Index and Other SleepRelatedPsychological and Behavioral Variables Groups Insomnia Variables DBAS-16 Perceived consequences Worry about insomnia Sleep expectations Medication PSAS Somatic arousal Cognitive arousal SRBQ
Normal sleepers
r
p
r
p
.229 .595 .206 –.137
.127
