Psychometric properties of the Pre-Sleep Arousal Scale in a large community sample.

Abstract

The purpose was to examine the psychometric properties of the Pre-Sleep Arousal Scale.
From a randomly selected sample of the general population (N=5000), 2327 participants completed a survey on nighttime symptoms, daytime symptoms, health outcomes, and psychological processes. The study sample consisted of 1890 participants who did not fulfill criteria for a sleep disorder other than insomnia.
Findings indicated that the PSAS did not produce an adequate factorial solution. When three problematic items were removed, the solution, accounting for 48.5% of the variance, improved (PSAS-13). One subscale, cognitive arousal (α=.88), consisted of five items (37.1%), and one subscale, somatic arousal (α=.72), of eight items (11.4%). The two factors were significantly inter-correlated (ρ=.51) and associated with the PSAS-13 (ρ=.91, ρ=.80). Among those with insomnia, a shortened PSAS (PSAS-14) was established, which consisted of a cognitive and a somatic subscale (48.6% of the variance). The PSAS-13 and the two subscales showed discriminant validity between three sleep groups (normal sleep, poor sleep, and insomnia disorder) (R(2)=.24-.34). The PSAS-13 and the subscales demonstrated convergent validity with measures on sleep-related worry, sleep-related beliefs, anxiety, and depression. The PSAS-13 and the two subscales were significantly correlated with sleep parameters and daytime impairment.
Though acceptable psychometric properties were established for the PSAS, the cognitive subscale's focus upon general pre-sleep arousal and the relatively low variance accounted for calls for further work on and a possible re-conceptualization of the PSAS.

Full text

Page 1

Psychometric properties of the Pre-Sleep Arousal Scale in a large community sample
Markus Jansson-Fröjmark⁎, Annika Norell-Clarke
School of Law, Psychology, and Social Work, Örebro University, Örebro, Sweden
a b s t r a c t a r t i c l ei n f o
Article history:
Received 11 April 2011
Received in revised form 5 October 2011
Accepted 6 October 2011
Keywords:
Insomnia
Sleep
Arousal
Scale
Objective: The purpose was to examine the psychometric properties of the Pre-Sleep Arousal Scale.
Methods: From a randomly selected sample of the general population (N=5000), 2327 participants com-
pleted a survey on nighttime symptoms, daytime symptoms, health outcomes, and psychological processes.
The study sample consisted of 1890 participants who did not fulfill criteria for a sleep disorder other than
insomnia.
Results: Findings indicated that the PSAS did not produce an adequate factorial solution. When three prob-
lematic items were removed, the solution, accounting for 48.5% of the variance, improved (PSAS-13). One
subscale, cognitive arousal (α=.88), consisted of five items (37.1%), and one subscale, somatic arousal
(α=.72), of eight items (11.4%). The two factors were significantly inter-correlated (ρ=.51) and associ-
ated with the PSAS-13 (ρ=.91, ρ=.80). Among those with insomnia, a shortened PSAS (PSAS-14) was
established, which consisted of a cognitive and a somatic subscale (48.6% of the variance). The PSAS-13
and the two subscales showed discriminant validity between three sleep groups (normal sleep, poor
sleep, and insomnia disorder) (R2=.24–.34). The PSAS-13 and the subscales demonstrated convergent va-
lidity with measures on sleep-related worry, sleep-related beliefs, anxiety, and depression. The PSAS-13
and the two subscales were significantly correlated with sleep parameters and daytime impairment.
Conclusion: Though acceptable psychometric properties were established for the PSAS, the cognitive sub-
scale's focus upon general pre-sleep arousal and the relatively low variance accounted for calls for further
work on and a possible re-conceptualization of the PSAS.
© 2011 Elsevier Inc. All rights reserved.
Introduction
Insomnia is a chronic condition that involves difficulties initiating
sleep, maintaining sleep, or waking in the morning not feeling
restored [1]. It is one of the most prevalent health problems being
reported by 10% of the population [2]. The consequences for the
sufferer are severe and include functional impairment, absenteeism,
impaired concentration and memory, increased use of medical ser-
vices, and an elevated risk of subsequently developing another psy-
chiatric disorder [3–5]. Insomnia is therefore viewed as a serious
public health problem.
In the insomnia literature, hyperarousal is often dichotomized
into two entities, cognitive and physiological/somatic arousal, and
these are usually seen as closely linked systems [6,7]. Both psycho-
logical and physiological models of insomnia underscore hyper-
arousal as a maintaining factor in insomnia [8–12]. For example, in
a neurocognitive model of psychophysiological insomnia, high fre-
quency EEG activity is considered to interfere with the normal estab-
lishment of sleep onset-related mesograde amnesia, influencing
information and/or memory processes [11]. In another model, the
psychobiological inhibition model, several mechanisms (i.e. atten-
tion, intention, and effort) are proposed that may underpin cognitive
arousal [12].
From a psychological viewpoint, cognitive arousal has been shown
to be associated with sleep difficulties and particularly “having an
overactive mind” [13–15]. Evidence from a population survey also in-
dicates that individuals dissatisfied with their sleep report pre-sleep
mental activity [16]. Though the results are mixed [17], evidence
suggests that there is a significant association between cognitive
arousal and sleep onset latency [18,19]. Experiments have shown
that an increase in cognitive arousal results in an increase in the sub-
jective estimation of time taken to fall asleep [20–22] and experi-
mental manipulations that decrease cognitive arousal result in a
decrease in sleep onset latency [23–25].
From a physiological perspective, early evidence indicated that
poor sleepers, relative to good sleepers, exhibit elevated autonomic
arousal, e.g. higher skin conductance, both prior and during sleep
[26]. When differences have been investigated using other physio-
logical parameters as well (e.g. EMG), later work on physiological
arousal in insomnia has produced conflicting evidence [6]. More re-
cent evidence suggests that individuals with insomnia, relative to
normal sleepers, show neurobiological differences, e.g. increased oxy-
gen use both day and night [27,28]. Recent empirical validation for
the neurocognitive model has been provided, mainly through the
Journal of Psychosomatic Research 72 (2012) 103–110
⁎ Correspondingauthorat:SchoolofLaw,Psychology,andSocialWork,ÖrebroUniversity,
SE-701 82 Örebro, Sweden. Tel.: +46 19 301042.
E-mail address: markus.jansson@oru.se (M. Jansson-Fröjmark).
0022-3999/$ – see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jpsychores.2011.10.005
Contents lists available at SciVerse ScienceDirect
Journal of Psychosomatic Research

Page 2

use of Power Spectral Analysis showing elevated beta power, or neu-
roimaging techniques [29–32]. Recent studies have also provided
empirical evidence for the psychobiological inhibition model
[33,34]. In a recent review, based on differing measurement tech-
niques (e.g. electrophysiological and neuroimaging), it is concluded
that primary insomnia is characterized by increased levels of arousal
in primary insomnia during both at night and daytime, thus indicat-
ing that physiological hyperarousal processes have an important
role in the pathophysiology of primary insomnia [7]. In another re-
cent review, the authors argue that insomnia is a conflict between
two systems; the sleep system and the central nervous system [35].
Given the potential role of hyper-arousal in chronic insomnia, an
important topic for research, which might ultimately be useful in
clinical practice, has been to develop and psychometrically validate
measures that assess arousal. Though several self-report instru-
ments have already been developed and validated (e.g. the Arousal
Predisposition Scale and the Hyperarousal Scale), the most frequent-
ly used subjective arousal measure is the Pre-Sleep Arousal Scale
(PSAS) [15]. The PSAS consists of sixteen items and comprises two
subscales, one of which is intended to assess cognitive arousal and
the otheronesomatic arousal.In theonly previous study psychomet-
rically evaluating the PSAS [15], a young, convenience sample, which
consisted of students, normal sleepers, and insomniacs, completed
the PSAS along with measures of anxiety, depression, and several
sleep parameters (e.g. sleep onset latency and sleep awakenings).
In the validation investigation, the PSAS was found to correlate
with anxiety, depression as well as with indices of sleep difficulties.
Also, the items and subscales of the PSAS were shownto discriminate
insomniacs from normal sleepers [15].
Though the PSAS has been shown to display psychometric prop-
erties of promising quality, there are several aspects of the PSAS
that need consideration. Firstly, the PSAS was evaluated in a young,
convenience sample, and questions remain whether the same prop-
erties can be demonstrated in a large, representative sample of
adults. A second aspect relates to the two PSAS subscales (cognitive
and somatic arousal). Though the scale constructors aimed to tap
two arousal domains, this was only verified by asking three clinical
psychologists to independently categorize the sixteen items [15].
Despite full agreement between the raters, it might be problematic
that no factor analysis has been performed on the scale since such
an analysis can more extensively determine whether the PSAS com-
prises one or several latent variables. Lastly, psychometric questions
regarding the PSAS pertain to the internal consistency of the scale,
the discriminant and convergent validity of the PSAS as well as its as-
sociation with sleep parameters and daytime impairment. This study
aimed to document the psychometric properties of the PSAS.
Method
Participants
This research is part of the Prospective Investigation on Psycho-
logical Processes for Insomnia (PIPPI) study, which was approved
by the Regional Ethics Board in Uppsala, Sweden. A random sample
of 5000 residents from two counties in Sweden (Örebro and
Värmland), 18 to 70 years old, was sent a survey in September
2008. The random sample was obtained from the national register
in which all residents in the two counties are listed. Of the total
sample, 58 (1.2%) were not eligible (incorrect address: N=38, par-
ticipation refusal: N=20). Of the 4942 eligible residents, 2327 par-
ticipants (47.1%) returned the survey.
An attritionanalysis intwo steps showed the following [36]. From
register data on age and gender, the analyses showed that the
respondents (mean: 47 years) were significantly older than the
non-respondents (mean: 41 years) but that there was no difference
on gender. The second step consisted of data from a one-page survey
that was sent out to 20% of the non-respondents (N=522), and
24.9% of the non-respondents returned the short survey. Analyses
between the respondents and the 130 short survey respondents
demonstrated that there were no differences on age (47 versus
48 years), gender (55% versus 48% women), sleep disturbance dur-
ing the past month (38% versus 40%), or mean Insomnia Severity
Index score (12.1 versus 11.7).
Among the 2327 study participants, the mean age was 47.1 years
and 54.9% were women. Regarding marital status, 14.9% reported
being single, 79.4% being cohabitant or married or having a partner,
4.2% being divorced, and 1.4% being widowed. On vocational status,
73.3% reported being full or part-time employed or students and
26.7% being unemployed, on sick leave, on pension or other status.
Concerning education, 25.7% reported compulsory school as their
highest level of education, 45.8% high school, and 28.5% college
or university. Regarding ethnicity, 92.1% reported being born in
Sweden. Comparisons between our sample and public register data
(e.g. Statistics Sweden and Swedish National Institute of Public
Health) reflecting the population in Sweden showed that our sample
was representative on several demographic parameters (i.e. age,
gender, civil status, occupational status, and education) as well as
the percentage of the participants reporting medical conditions,
medications, sleep disturbance, anxiety, and depression [36].
The only inclusion criterion that was used for this study required
that the participants did not fulfill criteria for another sleep disor-
der than insomnia (see under Sleep disorders below for criteria;
N=1890).
Procedure
The survey was mailed to the 5000 residents. It was accompanied
by an introductory letter and invitation to participate as well as a pre-
paid return envelope. If a response was not received within two
weeks a reminder was mailed. If an additional two weeks elapsed
without a response a new survey was sent. To increase the response
rate, a number of steps were taken in line with a Cochrane review
[37]. For example, we sent a pre-notification letter, a small incentive,
an information letter, we used closed questions, we placed relevant
and easy questions first, we included a pre-paid return envelope,
and we sent a reminder to non-responders. The odds of response
have been shown to increase in the range from 1.13 to 1.99 when
these steps are used in isolation [37].
Measures
The following demographic parameters were assessed: age, gen-
der, civil status, level of education, vocational status, and ethnicity.
For all the measures described in the section below, the participants
were asked to respond based on the past month.
Nighttime symptoms
To assess nighttime symptoms, the participants were asked to com-
plete the following categorical questions based on the previous month
[38]: sleep onset latency (SOL; b15 min, 16–30 min, 31–60 min,
>60 min), wake time after sleep onset (WASO; same alternatives as
for SOL), early morning awakening (EMA; same alternatives as for
SOL), total sleep time (TST; b4 h, 4–5 h, 5–6 h, 6–7 h, 7–8 h, 8–9 h,
9–10 h, >10 h), sleep restoration [completely (1), a lot (2), some-
what (3), a little (4), not at all (5)], and sleep quality [very good
(1), quite good (2), neither good nor poor (3), quite poor (4), very
poor (5)]. For the purpose of the study, four categorical sleep param-
eters were constructed [39]: SOL (30 min or less versus 31 min or
more), WASO (same as for SOL), EMA (same as for SOL), and TST
(more than 6 h versus 6 h or less). To determine sleep disturbance,
the participants were asked to complete the following questions:
sleep disturbance during the past month (yes or no; if no: continue
104
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110

Page 3

to daytime impairment section), frequency of sleep disturbance
during the past month (b1 night per week, 1–2 nights per week, 3–
5 nights per week, every night), and duration of sleep disturbance
(‘b1 year’, ‘1–5 years’, ‘6–10 years’, ‘>10 years’). In addition, the In-
somnia Severity Index (ISI; score range 0–28 points), a seven-item
scale, was also included [40].
Daytime symptoms
The participants were asked to report on the degree of sleep-
related impairment during the previous month: fatigue/malaise,
impairment in attention, concentration, or memory, social dysfunc-
tion, vocational dysfunction, mood disturbance, irritability, daytime
sleepiness, reduction in motivation, energy, or initiative, proneness
for errors or accidents at work or while driving, tension headaches,
gastrointestinal symptoms, and concerns or worries about sleep
[38]. The response alternatives for these indications of daytime im-
pairment were: not at all (1), somewhat (2), quite much (3), and a
lot (4). The response alternatives for the two functional domains
were: no negative consequences (1), small negative consequences
(2), marked negative consequences (3),large negativeconsequences
(4) and very large negative consequences (5). A composite score for
daytime impairment was computed, with all the impairment items
being merged.
Anxiety and depression
The anxiety and depression subscales of the Hospital Anxiety and
Depression Scale were used to assess anxiety and depression
(HADS-A and HADS-D) [41]. The two subscales are self-rating in-
struments with seven anxiety and seven depression questions in
which the severity is rated on 4-point scales (score range 0–21).
Neither of the subscales contains items determining sleep or insom-
nia. Based on the current sample, the internal consistency for the
subscales were high (anxiety: α=.85, depression: α=.88) accord-
ing to Kline's criteria [42].
Sleep-related beliefs
To assess sleep-related beliefs, the Dysfunctional Beliefs and
Attitudes about Sleep Scale were used (DBAS-10) [43]. The DBAS-
10 is employed to identify sleep-related dysfunctional beliefs and
contains 10 items. To make the response alternatives similar to
other psychological process measures, the alternatives were chan-
ged slightly (1–5; 1=strongly disagree, 5=strongly agree) and
the score range was thus 10–50. Based on the current sample, the
internal consistency was α=.85.
Sleep-related worry
The Anxiety and Preoccupation about Sleep Questionnaire was used
to assess sleep-related worry (APSQ) [44,45]. The response alternatives
for each of the 10 items were changed slightly in the current study to
make the alternatives similar to the other psychological process mea-
suresusedinthecurrentstudy[1–5;1=stronglydisagree,5=strongly
agree (original alternatives: 1–10; 1=strongly disagree, 10=strongly
agree)]. The score range was thus 10–50. A psychometric study of the
APSQ has demonstrated a two-factor solution, accounting for 70.7% of
the variance, determining (a) worries about the consequences of
poor sleep and (b) worries about the uncontrollability of sleep [44].
The APSQ and the two subscales also show discriminant validity,
convergent validity with measures on cognitive arousal, sleep-
related beliefs, anxiety, and depression, and significant correlations
with sleep parameters and daytime impairment.
Sleep-related arousal
The Pre-Sleep Arousal Scale consisting of 16 items was employed
to determine pre-sleep arousal (PSAS) [15]. The response alternatives
were 1 (‘not at all’) to 5 (‘extremely’) with a score range of 8–40. The
PSAS as well as all the other self-report instruments was first
translated into Swedish and then back-translated into English by
two persons with good skills in both Swedish and English.
Sleep disorders
The SLEEP-50 was used to assess seven DSM-IV-TR sleep disor-
ders: apnea, narcolepsy, restless legs/periodic limb movement disor-
der, circadian rhythm disorder, sleep walking, nightmares, and
hypersomnia [46]. The instrument has high internal consistency,
test–retest correlation ranging between .65 and .89, and a factor
structure that matches the DSM-IV-TR sleep disorders. The sensitivity
and specificity scores have been found to be reasonable for the sleep
disorders (sensitivity: 0.67–1.00; specificity: 0.69–1.00). The agree-
ment between clinical diagnoses and classification derived from the
SLEEP-50 is substantial (kappa=.77) [47]. The participants were
asked to rate to what extent the items have been applicable during
the past month (1=not at all, 4=very much). Three additional sub-
groups in the SLEEP-50 (insomnia, affective disorder and sleep state
misperception) were not used in the current study.
Sleep groups
The participants were classified in one of three groups according
to their sleep patterns, daytime impairment, and evidence of sleep
disorders other than insomnia. The classification used an algorithm
based on a combination of insomnia diagnostic criteria from Research
Diagnostic Criteria for insomnia [38], established quantitative criteria
for insomnia [39], and screening for sleep disorders other than insom-
nia [46]. Among the 2327 baseline responders, 407 (17.5%) fulfilled the
criteria for a sleep disorderother thaninsomnia.Of the 407 participants
with a sleep disorder than insomnia, restless legs/periodic limb move-
ment disorder was fulfilled by 46.3%, apnea by 44.3%, circadian rhythm
disorder by 23.2%, nightmares by 21.0%, hypersomnia by 15.4%, narco-
lepsy by 3%, and sleep walking by 0.2%. In addition to the 407 partici-
pants excluded for fulfilling the criteria for a sleep disorder other
than insomnia, 30 individuals were also excluded due to incomplete
data on nighttime symptoms, daytime symptoms, or the SLEEP-50.
Insomnia disorder
To belong to this group, four diagnostic criteria had to be met.
(1) The participant had to affirm a sleep disturbance during the past
month. (2) The individual had to report initial, middle, or late insom-
nia (31 min or more per night) or non-restorative sleep (‘a little’ or
less) or poor sleep quality (‘quite poor’ or less). This sleep pattern
had to be present for at least 3 nights per week. (3) The participant
had to report daytime impairment (for symptoms: ‘quite much’ or
more, for function: ‘marked negative consequences’ or more) [48].
(4) The individual must not meet the criteria for apnea, narcolepsy,
restless legs syndrome/periodic limb movement disorder, circa-
dian rhythm disorder, sleepwalking, nightmares or hypersomnia as
assessed with the SLEEP-50. Unpublished data from this project indi-
cates a high concordance between the current insomnia disorder def-
inition and validated Insomnia Severity Index cutoffs [49], which, in
turn, have shown nearly 98% correct classification rates differentiat-
ing clinical insomnia patients and normal sleepers. The concordance
betweenour insomnia disorder definition and the ISI cutoff at 8 points
was 99.4%. The concordance for the ISI cutoff at 11 points was also
high (89.4%). Also, the mean ISI score for the group was 15.4, indicat-
ing clinical insomnia [40]. This indicates that the current insomnia
definition captures the insomnia disorder construct to a high degree.
In the insomnia disorder group, 7.7% reported their duration of dis-
turbed sleep to have been less than a year, 43.8% 1–5 years, 23.1%
6–10 years, and 25.4% more than 10 years.
Poor sleep
To belong to this group, two diagnostic criteria had to be met.
(1) The participant had to affirm a sleep disturbance during the past
105
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110

Page 4

month. (2) The individual must not meet the criteria for apnea, nar-
colepsy, restless legs syndrome/periodic limb movement disorder,
circadian rhythm disorder, sleepwalking, nightmares, and hyper-
somnia as assessed with the SLEEP-50. In the poor sleep group,
19.3% reported their duration of disturbed sleep to have been less
than a year, 45.8% 1–5 years, 16.7% 6–10 years, and 18.3% more
than 10 years. This group was used to reflect the full spectrum from
normal sleep to insomnia disorder and can possibly be viewed as a
proxy for subclinical insomnia (mean Insomnia Severity Index
score for the group was 8.8, indicating subclinical insomnia [40]).
Normal sleep
To belong to this group, two criteria had to be fulfilled. (1) The
participant did not report a sleep disturbance during the past
month. (2) The participant must not meet the criteria for apnea, nar-
colepsy, restless legs syndrome/periodic limb movement disorder,
circadian rhythm disorder, sleepwalking, nightmares, and hypersom-
nia as assessed with the SLEEP-50.
Statistical analysis
To reveal any latent variables within the PSAS that cause the
manifest variables to co-vary, exploratory factor analysis was used
in line with recommendations [50]. To ensure that the characteris-
tics of the data set were suitable for the factor analysis to be con-
ducted on the study sample, the Kaiser-Meyer-Olkin measure of
sampling adequacy (KMO) and the Bartlett Test of Sphericity (BTS)
were conducted on the data. A maximum likelihood factor extrac-
tion procedure with oblique rotation (direct oblimin) was employed
since this approach is particularly useful in extracting meaningful
factors and because of the possibility that the extracted factors
may be correlated [50]. Oblique rotation was chosen since theory
posits that cognitive and somatic arousal are closely related [7] and
that the analyses on the current sample showed a high correlation
between the two factors (.62). The scree test was used to decide
the number of meaningful factors that might be in the data set. The
minimum loading of an item was determined at .32, a recommended
threshold for the minimum loading of an item [51]. To investigate
internal consistency, Cronbach's alpha was used, and 0.70 was con-
sidered as the minimum acceptable criterion of instrument internal
reliability [42]. The discriminant validity was examined with multi-
variate analysis of variance (MANOVA) in which sleep status
(insomnia disorder, poor sleep, and normal sleep) was used as the
fixed factor. In the MANOVA, the PSAS items as well as the total
PSAS scale were employed as the dependent variables. Assumptions
for MANOVA were checked before executing the analyses. One-way
analysis of variance (ANOVA) was employed as a follow-up test. The
analyses concerning convergent validity and association between
(a) the PSAS and its subscales and (b) sleep parameters and daytime
impairment employed a correlative approach (Spearman's Rho [52],
Eta, and Contingency Coefficient).
Results
Factorial validity, internal consistency and inter-relationship: entire sample
Preparatory analyses were first used to ensure that the data distribution satisfied
the psychometric criteria for a factor analysis of the PSAS. These analyses showed that
the KMO yielded an index of 0.92 and the BTS was significant (χ2
pb.001). Based on the psychometric results, a factor analysis was viewed
as appropriate. To check for stability of parameter estimates, the sample was first
divided into two random halves. In the two sub-samples and in the sample as a
whole, the scree plot indicated a two-factor solution for the PSAS. Given the estab-
lished stability of estimates, the sample of 1890 participants was used in the remain-
ing analyses. The factor solution is displayed at Table 1. The factor analysis revealed a
two-factor solution which explained 47.6% of the variance. The first factor, which was
labeled “cognitive arousal” as in the original paper (PSAS-C-8), explained 38.2% of the
variance. The second factor (previously termed “somatic arousal”, PSAS-S) consisted
of eight items and explained 9.4% of the variance. However, several indices indicated
(df = 120)=11752.2,
that the first factor was problematic, i.e. three items cross-loaded (items 9, 11, and
16) and there was an item-loading below .32 (item 16).
One and three-factor solutions were evaluated to examine if manually setting fac-
tors to retain would improve the factor solution. The one-factor solution (38.2% of the
variance) resulted in a problem, i.e. there was an item-loading below .32 (item 3: .29).
Based on the encountered low-loading item in the manually set one-factor solution as
well as the scale constructors' intention to assess both somatic and cognitive arousal,
the one-factor solution was discarded. The three-factor solution (53.6% of the variance)
displayed identical problems as the initial two-factor solution, i.e. cross- and low-
loading items, and was therefore also discarded.
To investigate whether dropping the three problematic items (items 9, 11, and 16)
would improve the factor solution, analyses excluding the three items, both individu-
ally and in different combinations, were executed. The initial analyses showed that
dropping the problematic items individually did not improve the factor solution. How-
ever, when all three problematic items were removed from the factor analysis, the so-
lution displayed several improved features (see Table 2); the solution explained 48.9%
of the variance [37.4% for the cognitive subscale, PSAS-C-5 (Eigenvalue 4.87) and 11.5%
for the somatic subscale, PSAS-S (Eigenvalue 1.49)], all the item-loadings were above
.32, there were no cross-loadings, and there were more than five items in each factor
(five in the cognitive factor and eight in the somatic factor) [50]. The KMO yielded
an index of 0.90 and the BTS was significant (χ2
PSAS version with 13 items, extracted in the entire sample, is hereafter termed as
PSAS-13 and the full version as PSAS-16.
The items in the cognitive factor showed moderate to high primary loadings (.63 to
.97) and the items in the somatic factor showed moderate loadings (.43 to .63). The
items showed minimal to low loadings on the other component, indicating two rela-
tively distinct arousal dimensions. The correlations (Spearman's Rho) between the
cognitive items ranged from .44 to .80, suggesting fair to excellent correlations, and be-
tween the somatic items from .14 to .37, indicating small to fair correlations. The inter-
nal consistency of the total PSAS-13 was α=.85, for the cognitive factor α=.88, and
for the somatic factor α=.72. The two factors were significantly inter-correlated
(ρ=.51) and significantly associated with the total PSAS-13 (ρ=.91, ρ=.80). As can
be seen in Table 3, the PSAS-13 and the PSAS-16 correlated very highly (ρ=.99) as
did the two versions of the cognitive subscale (PSAS-C-8 and PSAS-C-5: ρ=.97).
(df = 78)=8482.8, pb.001). The
Factorial validity, internal consistency and inter-relationship: insomnia sample
The preparatory analyses in the insomnia sample (n=170) showed that the KMO
yielded an index of 0.80 and the BTS was significant (χ2
and a factor analysis was thus viewed as appropriate. In the insomnia sample, the
scree plot indicated a two-factor solution for the PSAS which explained 44.8% of the
variance. The first factor (“cognitive arousal”) explained 30.8% of the variance and
the second factor (“somatic arousal”) 14.0% of the variance. However, item 2 cross-
loaded and loaded below .32 on both the factors. A one- and a three-factor solution
were also evaluated. The one-factor solution resulted in a problem, i.e. there were
items loading below .32 (items 1, 3, 4, 5, 5, 7, 8, and 16). Based on the low-loading
items in the one-factor solution and the original intention to assess both somatic and
cognitive arousal, the one-factor solution was discarded. The three-factor solution
(52.2% of the variance) also displayed problems, i.e. cross-loading items (2 and 12)
and low-loading items (2 and 16), and was therefore also discarded. When the
(df = 120)=1069.4, pb.001),
Table 1
Exploratory factor analysis of the pre-sleep arousal scale with 16 items in the entire
sample (N=1890).
Factor loadings
Cognitive
arousal
Somatic
arousal
1Heart racing, pounding or beating irregularly
2A jittery, nervous feeling in your body
3Shortness of breath or labored breathing
4A tight, tense feeling in your muscles
5Cold feeling in your hands, feet or your body in general
6Have stomach upset (knot or nervous feeling in stomach,
heartburn, nausea, gas, etc.)
7Perspiration in palms of your hands or other parts
of your body
8Dry feeling in mouth or throat
9Worry about falling asleep
10Review or ponder events of the day
11Depressing or anxious thoughts
12Worry about problems other than sleep
13Being mentally alert, active
14Can't shut off your thoughts
15Thoughts keep running through your head
16Being distracted by sounds, noise in the environment
(e.g. ticking of clock, house noises, traffic)
−.02
.21
−.09
.05
.06
.08
.47
.49
.49
.59
.38
.48
−.01
.45
−.01
.35
.70
.49
.64
.64
.89
.98
.31
.46
.32
.06
.38
.22
−.04
−.06
−.09
.25
Note. Bold numbers indicate the primary loading.
106
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110

Page 5

problematic item from the two-factor solution was dropped (item 2), one additional
item was identified as a low-loading item (16). When both items 2 and 16 were re-
moved in a further factor analysis, the solution displayed several improved
features (see Table 4); the solution explained 48.6% of the variance [32.8% for the cog-
nitive subscale (Eigenvalue 4.60) and 15.8% for the somatic subscale (Eigenvalue
2.20)], all the item-loadings were above .32, there were no cross-loadings, and there
were more than five items in each factor (seven in the cognitive factor and seven in
the somatic factor [50]). The KMO yielded an index of 0.79 and the BTS was significant
(χ2
nia sample, is hereafter termed as PSAS-14. The items in the cognitive factor displayed
moderate to high primary loadings (.42 to .98) and the items in the somatic factor
showed moderate loadings (.38 to .57). The items showed minimal to low loadings
on the other component, with the exception of one cognitive item (“Depressing or anx-
ious thoughts”). The correlations (Spearman's Rho) between the cognitive items ran-
ged from .28 to .89, suggesting fair to excellent correlations, and between the
somatic items from .04 to .37, indicating small to fair correlations. The internal consis-
tency of the total PSAS-14 was α=.82, for the cognitive factor α=.89, and for the so-
matic factor α=.66. The two factors were significantly inter-correlated (ρ=.21) and
significantly associated with the total PSAS-13 (ρ=.88, ρ=.62). The PSAS-14 and
the PSAS-16 correlated very highly (ρ=.99) as did the two versions of the cognitive
subscale (ρ=.98) and the two versions of the somatic subscale (ρ=.97).
(df = 91)=998.6, pb.001). The PSAS version with 14 items, validated in the insom-
Discriminant validity
To investigate the discriminant validity of the PSAS-13, the 170 participants with
insomnia disorder were compared with the 393 poor sleepers and the 1327 normal
sleepers. As can be seen at Table 5, multivariate analysis of variance showed that the
PSAS-13 items, the total PSAS-13 instrument, and the two retained factors discriminat-
ed the three sleep groups. For the five cognitive items, the insomnia disorder group had
higher scores than the other two groups (pb.01 in all instances), and the poor sleepers
had more elevated scores than the normal sleepers (pb.01 in all instances). For the
eight somatic items, an identical pattern emerged between the insomnia disorder
group and the other two groups (pb.05 in all instances) and between the poor and
the normal sleepers (pb.05 in all instances). Two exceptions emerged; for items 6
and 7, no significant difference was noted between the insomnia disorder group and
the poor sleepers. For the total PSAS-13 (pb.001) and the two factors (PSAS-S and
PSAS-C-5) (pb.001), the insomnia disorder group had more elevated scores than the
other two groups, and the poor sleepers scored higher than the normal sleepers. The
between-group effect sizes for the 13 items were between .03 and .25, for the total
PSAS-13 .34, for the cognitive factor .28, and for the somatic factor .24.
To explore the discriminative validity of the three excluded items and the full PSAS
(PSAS-16), additional analyses were executed. The three dropped items and the PSAS-
16 discriminated between the three sleep groups. For the three excluded items and the
PSAS-16, the insomnia disorder group had higher scores than the other two groups
(pb.01 in all instances), and the poor sleepers had more elevated scores than the nor-
mal sleepers (pb.01 in all instances).
Convergent validity
The correlations between the total PSAS-13 and its two factors and four related
constructs (APSQ, DBAS-10, HADS-A, and HADS-D) are displayed at Table 3. The
table also displays correlations for the PSAS-16 and for the original cognitive subscale
consisting of 8items.Asdepictedatthetable,thePSAS-13anditstwofactors(PSAS-Sand
PSAS-C-5) were significantly related to the APSQ at a fair to moderate level (ρ=.43–.52),
to the DBAS-10 at a fair level (ρ=.38–.45), to the HADS-A at a moderate to good level
(ρ=.49–.57), and to the HADS-D at a fair level (ρ=.38–.44).
Association between the PSAS-13 and sleep parameters and daytime impairment
The associations between the PSAS-13 and its two factors (PSAS-S and PSAS-C-5)
with sleep parameters and daytime impairment were investigated. First, the PSAS-13
and the two factors were correlated (Eta) with four categorical sleep parameters
[sleep onset latency, wake after sleep onset, and early morning awakening (30 min
or less versus more than 30 min); total sleep time (more than 6 h versus 6 h or less)]
and one continuous sleep parameter (sleep quality; Spearman's Rho). As is displayed
in Table 3, the PSAS-13 and its two subscales were moderately associated with sleep
onset latency (η=.26–.33), wake after sleep onset (η=.26–.31), early morning awak-
ening (η=.21–.23), and total sleep time (η=.20–.24). At the same table, it is also pos-
sible to verify that the PSAS-13 and the two subscales were correlated with sleep
quality at a fair level (ρ=.35–.43).
Table 2
Exploratory factor analysis of the pre-sleep arousal scale with 13 items in the entire
sample (N=1890).
Factor loadings
Cognitive
arousal
Somatic
arousal
1Heart racing, pounding or beating irregularly
2A jittery, nervous feeling in your body
3Shortness of breath or labored breathing
4A tight, tense feeling in your muscles
5Cold feeling in your hands, feet or your body in general
6Have stomach upset (knot or nervous feeling in stomach,
heartburn, nausea, gas etc.)
7Perspiration in palms of your hands or other parts
of your body
8Dry feeling in mouth or throat
10Review or ponder events of the day
12Worry about problems other than sleep
13Being mentally alert, active
14Can't shut off your thoughts
15Thoughts keep running through your head
−.01
.24
−.07
.05
.05
.07
.47
.43
.47
.60
.40
.52
−.02
.48
−.01
.70
.66
.61
.88
.98
.48
.05
.15
.01
−.04
−.07
Note. Bold numbers indicate the primary loading. Note that items 9, 11, and 16 have
been removed.
Table 3
Correlations between the Pre-Sleep Arousal Scale, the two PSAS Subscales, related constructs, sleep parameters, and daytime impairment.
PSAS-16 PSAS-C-5PSAS-C-8PSAS-S APSQDBAS HADS-AHADS-D SOLWASO EMATSTSQ DI
PSAS-13
PSAS-16
PSAS-C-5
PSAS-C-8
PSAS-S
APSQ
DBAS
HADS-A
HADS-D
SOLa
WASOa
EMAa
TSTb
SQ
DI
.99⁎⁎
.91⁎⁎
.91⁎⁎
.92⁎⁎
.94⁎⁎
.97⁎⁎
.80⁎⁎
.79⁎⁎
.51⁎⁎
.56⁎⁎
.52⁎⁎
.55⁎⁎
.47⁎⁎
.52⁎⁎
.43⁎⁎
.45⁎⁎
.47⁎⁎
.41⁎⁎
.44⁎⁎
.38⁎⁎
.61⁎⁎
.57⁎⁎
.59⁎⁎
.51⁎⁎
.55⁎⁎
.49⁎⁎
.49⁎⁎
.37⁎⁎
.44⁎⁎
.45⁎⁎
.39⁎⁎
.42⁎⁎
.38⁎⁎
.40⁎⁎
.31⁎⁎
.57⁎⁎
.33⁎⁎
.36⁎⁎
.32⁎⁎
.37⁎⁎
.26⁎⁎
.32⁎⁎
.23⁎⁎
.19⁎⁎
.17⁎⁎
.31⁎⁎
.33⁎⁎
.29⁎⁎
.32⁎⁎
.26⁎⁎
.36⁎⁎
.20⁎⁎
.22⁎⁎
.18⁎⁎
.35⁎⁎
.23⁎⁎
.25⁎⁎
.21⁎⁎
.23⁎⁎
.21⁎⁎
.30⁎⁎
.16⁎⁎
.20⁎⁎
.21⁎⁎
.17⁎⁎
.35⁎⁎
.24⁎⁎
.26⁎⁎
.22⁎⁎
.25⁎⁎
.20⁎⁎
.30⁎⁎
.07⁎⁎
.18⁎⁎
.22⁎⁎
.24⁎⁎
.34⁎⁎
.28⁎⁎
.43⁎⁎
.45⁎⁎
.38⁎⁎
.42⁎⁎
.35⁎⁎
.44⁎⁎
.30⁎⁎
.34⁎⁎
.33⁎⁎
.36⁎⁎
.46⁎⁎
.34⁎⁎
.43⁎⁎
.51⁎⁎
.53⁎⁎
.45⁎⁎
.49⁎⁎
.44⁎⁎
.56⁎⁎
.55⁎⁎
.53⁎⁎
.45⁎⁎
.19⁎⁎
.20⁎⁎
.19⁎⁎
.13⁎⁎
.36⁎⁎
Note. Spearman's Rho, Eta, and Contingency Coefficient were used as correlation statistics. PSAS = Pre-Sleep Arousal Scale (13 and 16 items), PSAS-C = Pre-Sleep Arousal Scale
cognitive subscale (5 and 8 items), PSAS-S = Pre-Sleep Arousal Scale somatic subscale, APSQ = Anxiety and Preoccupation about Sleep Questionnaire, DBAS = Dysfunctional Be-
liefs and Attitudes about Sleep, HADS-A = Hospital Anxiety and Depression Scale — anxiety, HADS-D=Hospital Anxiety and Depression Scale — depression, SOL = sleep onset
latency, WASO = wake after sleep onset, EMA = early morning awakening, TST = total sleep time, SQ = sleep quality, DI = daytime impairment.
⁎⁎ pb.01.
a30 min or less versus 31 min or more.
bMore than 6 h versus 6 h or less.
107
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110

Page 6

Second, the PSAS-13 and its two subscales were also correlated (Spearman's rho)
with daytime impairment (total score of the twelve daytime impairment items). At
the same table, it is possible to verify that the PSAS-13 and its two factors were signif-
icantly correlated with daytime impairment at a fair to moderate level (ρ=.44–.51).
Discussion
The first aim of this investigation was to explore the factorial va-
lidity and the internal consistency of the PSAS. The results suggested
that the original PSAS instrument with 16 items did not produce an
adequate factorial solution in the sample. More specifically, there
were several cross- and low-loading items in the cognitive subscale.
As a result, various methods to improve the factorial solution were
executed, and findings indicated that a shortened PSAS (PSAS-13)
resulted in two distinct factors, cognitive and somatic arousal, with
no cross- or low-loading items and with five items or more in each
subscale [50]. Additional, positive features of the PSAS-13 included
moderate to high primary loadings for the items in the cognitive sub-
scale and minimal to low loadings for the items on one of the sub-
scales on the other subscale. On a general level, the factor loadings
can be considered as statistically meaningful given the sample size
in the current study [53]. In addition, the loadings (all above 0.4)
imply that the PSAS-13 items are important in relation to the two
extracted factors [53]. The internal consistencies of the PSAS-13 and
of the two subscales were acceptable to good (α: .71–.88), a finding
very similar to the only previous study investigating the psychomet-
ric properties of the PSAS (α: .67–.88) [15].
When a factor analysis of the PSAS with 16 items was executed
in the insomnia sample, an adequate factorial solution could not
be established. When two problematic items were removed in the
analysis, a shortened PSAS (PSAS-14) displayed several improved
features, including two factors assessing cognitive and somatic
arousal, no cross- or low-loading items, and moderate to high pri-
mary loadings for the items in the cognitive subscale. The internal
consistency of the PSAS-14 and of the cognitive subscale was good
(α: .82–.89) and of the somatic subscale acceptable (α=.66).
A few of the PSAS-13 and PSAS-14 items warrant a comment. The
four dropped items consisted of domains tapping somatic nervous-
ness (“A jittery, nervous feeling in your body”), sleep-related worry
(“Worry about falling asleep”), distressing thoughts (“Depressing or
anxious thoughts”), and sound or noise distraction (“Being distracted
by sounds, noise in the environment”). One of the dropped items
(“Worry about falling asleep”) requires further discussion. Our inter-
pretation of the factorial solution is that though the item may be con-
sidered the most insomnia-specific cognitive one, it did not fit into
the remaining five cognitive items, which appear to measure general
cognitive arousal. This notion is also supported by our finding that the
item “Worry about falling asleep” was the cognitive item with the
lowest loading of the seven remaining items. It should be mentioned
that an additional dropped item (“Being distracted by sounds, noise
in the environment”) appear to be distinct from the other seven cog-
nitive items in terms of content and this notion is also confirmed by
displaying the lowest correlation with its subscale in the previous
PSAS study [15]. The dropped cognitive items seem thus to be rela-
tively distinct to the current study's extracted cognitive subscale,
comprising items assessing cognitive arousal in general. One somat-
ic item was dropped when the insomnia sample was examined
(“A jittery, nervous feeling in your body”). Our interpretation of
why the dropped somatic item did not fit within the somatic sub-
scale is that it taps somatic arousal in a more general way than the
remaining seven items, which all assess somatic arousal in a partic-
ular area of the body. This notion is also supported by our result that
the item “A jittery, nervous feeling in your body” was the somatic
item with the lowest loading when the factorial validity of the
PSAS-13 was explored.
The variance accounted for by the PSAS-16, the PSAS-13, and the
PSAS-14 warrants a few remarks since all three versions produced
relatively low percentages (47.1%, 48.5%, and 48.6%). It has been ar-
gued that the meaningfulness of factor analyses accounting for less
than the chance percentage may be questioned or that the usefulness
of the measure is doubtful [54]. Though the current study's findings
on the PSAS are unable to shed light on whether a factor analysis
accounting for less than the chance percentage is meaningful or
whether the PSAS is a useful measure, the results highlight a prob-
lematic notion with the PSAS. One interpretation of the modest vari-
ance accounted for by the PSAS is that the instrument taps arousal too
narrowly, only focusing upon cognitive and somatic arousal, and
reconsidering removing current or adding new items (e.g. behavioral
or emotional aspects) to the PSAS might possibly be an improvement.
Based on the current findings, this calls for a discussion on when
the PSAS should be used in clinical and research settings. First, based
on the finding that the extracted cognitive subscale with five items
(PSAS-C-5) seems not to assess sleep-related arousal, we see the
subscale's future potential in assessing general, cognitive arousal,
more so in research settings than in clinical practice since the
PSAS items are too vague to provide specific clinical information.
Table 4
Exploratory factor analysis of the pre-sleep arousal scale with 14 items in the insomnia
sample (N=170).
Factor loadings
Cognitive
arousal
Somatic
arousal
1Heart racing, pounding or beating irregularly
3Shortness of breath or labored breathing
4A tight, tense feeling in your muscles
5Cold feeling in your hands, feet or your body in general
6Have stomach upset (knot or nervous feeling in stomach,
heartburn, nausea, gas etc.)
7Perspiration in palms of your hands or other parts
of your body
8Dry feeling in mouth or throat
9Worry about falling asleep
10Review or ponder events of the day
11Depressing or anxious thoughts
12Worry about problems other than sleep
13Being mentally alert, active
14Can't shut off your thoughts
15Thoughts keep running through your head
−.02
.01
−.02
.06
.06
.49
.57
.48
.41
.46
−.03
.38
−.02
.42
.72
.64
.75
.65
.95
.98
.47
.09
-.06
.30
.12
.01
−.16
−.13
Note. Bold numbers indicate the primary loading. Note that items 2 and 16 have been
removed.
Table 5
Discriminant validity of the PSAS-13 at Item-, Scale-, and Subscale-level.
Insomnia disorder
(N=170)
Poor sleep
(N=393)
Normal sleep
(N=1327)
F
R2
M (SD)M (SD)M (SD)
Item 1
Item 2
Item 3
Item 4
Item 5
Item 6
Item 7
Item 8
Item 10
Item 12
Item 13
Item 14
Item 15
PSAS-13
Factor 1a
Factor 2b
1.7 (1.1)
1.8 (1.1)
1.4 (0.8)
2.4 (1.3)
2.2 (1.3)
1.9 (1.2)
2.0 (1.2)
1.9 (1.0)
3.0 (1.3)
2.9 (1.3)
2.7 (1.2)
3.1 (1.3)
3.2 (1.3)
30.2 (8.4)
14.9 (5.4)
15.3 (5.3)
1.4 (0.7)
1.6 (0.8)
1.1 (0.4)
1.8 (1.0)
1.8 (1.0)
1.7 (0.9)
1.7 (1.0)
1.6 (0.8)
2.6 (1.1)
2.3 (1.0)
2.2 (1.0)
2.5 (1.1)
2.7 (1.1)
24.9 (6.7)
12.3 (4.5)
12.6 (3.6)
1.1 (0.4)
1.2 (0.5)
1.1 (0.3)
1.3 (0.6)
1.6 (0.9)
1.2 (0.5)
1.2 (0.6)
1.2 (0.5)
1.8 (0.8)
1.6 (0.7)
1.6 (0.9)
1.6 (0.8)
1.7 (0.8)
18.3 (4.7)
8.3 (3.0)
10.0 (2.4)
73.00⁎⁎
89.12⁎⁎
33.15⁎⁎
126.75⁎⁎
22.92⁎⁎
74.00⁎⁎
76.93⁎⁎
67.21⁎⁎
130.81⁎⁎
185.10⁎⁎
86.68⁎⁎
219.77⁎⁎
220.01⁎⁎
332.30⁎⁎
259.79⁎⁎
204.88⁎⁎
.10
.12
.05
.16
.03
.10
.11
.09
.17
.22
.12
.25
.25
.34
.28
.24
⁎⁎ pb.01.
aCognitive arousal.
bSomatic arousal.
108
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110

Page 7

We further recommend that more recent instruments should be
used in research and in clinical settings to index insomnia-specific
mental arousal. For example, the Anxiety and Preoccupation about
Sleep Questionnaire has shown substantial promise in assessing
sleep-related worry (APSQ) [44,45]. Further, a new scale tapping
insomnia-specific rumination, the Daytime Symptom Response
Scale, may also be an important addition to the literature and to clin-
ical practice [55]. Second, based on the finding that shorter versions
of the PSAS displayed improved factorial validity compared with the
full PSAS, we recommend that when the aim is to index subjective
arousal that the shorter versions are used in clinical and research
settings, in combination with other insomnia-specific arousal mea-
surements (e.g. the APSQ). It should also be underscored that
when the purpose is to determine physiological/somatic arousal,
there are physiological measurements, e.g. autonomous parameters,
which are recommended [7].
The second purpose was to examine the discriminant validity of
the PSAS-13. The findings indicated that all the PSAS items, the total
scale, and the two subscales clearly discriminated the three sleep
groups. Across nearly all the analyses, the insomnia disorder group
had higher scores than the other two groups, and the poor sleepers
had more elevated scores than the normal sleepers. The current find-
ings on discriminant validity resemble a previous validation study
[15], which demonstrated discrimination between insomniacs and
normal sleepers. The exceptions in the current study were two so-
matic subscale items where no significant difference was noted be-
tween the insomnia disorder group and the poor sleepers. It should
also be noted that with the addition of the three dropped items, either
included in the PSAS-13 or in the cognitive subscale, the effect sizes
were only marginally increased (with the exception of the item
“Worry about falling asleep”), indicating that they add little in dis-
criminating the sleep groups.
The third aim was to explore the convergent validity of the PSAS-
13. The results indicated that the PSAS-13 and the two subscales
were significantly related to sleep-related worry, sleep-related be-
liefs, anxiety, and depression. The correlations found in this study
between the PSAS-13 and its subscales with HADS-A and HADS-D
resemble to a high degree what has been documented in a previous
study [15]. The inter-relationships between the PSAS-13 and sleep-
related worry and beliefs have previously not been examined and in-
dicate fair to moderate correlations. Despite the significant inter-
relationships, the strength of the associations indicates that the
PSAS-13 and its subscales are also relatively distinct from sleep-
related worry, sleep-related beliefs, anxiety, and depression, indicat-
ing divergence to a high degree.
The fourth purpose was to examine the association between the
PSAS-13 and its subscales with sleep parameters and daytime impair-
ment. The results suggested that the PSAS-13 and the two subscales
were moderately correlated with sleep parameters central to insom-
nia. The correlations between the PSAS, its subscales, and the sleep
parameters resemble to a high degree previous findings [15]. Cogni-
tive arousal was, relative to somatic arousal, in general more strongly
related to the five sleep parameters. The PSAS-13 and its subscales
were more highly correlated with the sleep parameters than sleep-
related beliefs, anxiety and depression. Though the PSAS-13 and the
two subscales were more highly correlated with the sleep parameters
than most of the study's additional measures, there was an exception:
sleep-related worry, as assessed by the APSQ, was, relative to the
PSAS-13 and its subscales, more highly correlated with the sleep pa-
rameters. The findings also indicated that the PSAS-13 and the two
subscales were correlated with daytime impairment at a fair to mod-
erate level. Though the PSAS-13 and its subscales were more strongly
associated with daytime impairment than depression, there was an
opposite observation for the remaining measures; sleep-related
worry, sleep-related beliefs, and anxiety were more strongly related
to daytime impairment than the PSAS-13 and its subscales.
A few methodological limitations should be mentioned. One ini-
tial limitation was the moderate response rate (47.1%), and the find-
ing that respondents were likely to be older than non-respondents. It
should also be emphasized that only age and gender were consid-
ered in the attrition analysis, which might be a limitation since
othersociodemographic factors are related to insomnia [56].Because
the short- and the long-survey respondents appeared to belong to
the same age group, it is possible that the short-survey respondents
were not representative, at least on age, for the group of initial non-
responders. Second, since this study was cross-sectional, causal
relationships cannot be deduced. Though a previous study has docu-
mented good test–retest correlations for the PSAS [15], a third limi-
tation was that no test–retest analysis was performed in the
current study. A fourth limitation was that all measures were based
on self-report, which might, for example, have resulted in a sleep
category misclassification, with, for example, non complaining poor
sleepers having been categorized as belonging to the normal sleep
group and vice versa. However, we note that we used a screening
instrument to assess sleep disorders as well as research diagnostic
criteria for insomnia that underscores the condition as a subjective
complaint.
Takenasawhole,thePSASappearstobeaself-reportinstrumentthat
needs further empirical work and possibly also a re-conceptualization
before it can be recommended as an assessment tool in research and
clinical settings. Though we see potential in the PSAS, we recommend
further studies on its psychometric properties.
Conflict of interest
The authors do not have any conflicts of interest to report.
Acknowledgments
We would like to express our gratitude to the Swedish Council for
Working Life and Social Research for financial support and to Steven
Linton, Allison Harvey, and Lars-Gunnar Lundh for study design.
References
[1] American Psychiatric Association. Diagnostic and statistical manual of mental dis-
orders: fourth edition. Washington, DC: American Psychiatric Association; 1994.
[2] Ancoli-Israel S, Roth T. Characteristics of insomnia in the United States: results of
the 1991 National Sleep Foundation Survey. Sleep 1999;22:347–53.
[3] Breslau N, Roth T, Rosenthal L, Andreski P. Sleep disturbance and psychiatric dis-
orders: a longitudinal epidemiological study of young adults. Biol Psychiatry
1996;39:411–8.
[4] Ford D, Kamerow DB. Epidemiologic study of sleep disturbances and psychiatric
disorders. An opportunity for prevention? JAMA 1989;262:1479–84.
[5] Roth T, Ancoli-Israel S. Daytime consequences and correlates of insomnia in the
United States: results of the 1991 National Sleep Foundation Survey. II. Sleep
1999;22:354–8.
[6] Harvey AG. A cognitive model of insomnia. Behav Res Ther 2002;40:869–93.
[7] Riemann D, Spiegelhalder K, Feige B, Voderholzer U, Berger M, Perlis M, et al. The
hyperarousal model of insomnia: a review of the concept and its evidence. Sleep
Med Rev 2010;14:19–31.
[8] Espie CA. Insomnia: conceptual issues in the development, persistence, and treat-
ment of sleep disorder in adults. Annu Rev Psychol 2002;53:215–43.
[9] Lundh LG, Broman JE. Insomnia as an interaction between sleep-interfering and
sleep-interpreting processes. J Psychosom Res 2000;49:299–310.
[10] Richardson GS, Roth T. Future directions in the management of insomnia. J Clin
Psychiatry 2001;62:39–45.
[11] Perlis ML, Giles DE, Mendelson WB, Bootzin RR, Wyatt JK. Psychophysiological
insomnia: the behavioural model and a neurocognitive perspective. J Sleep Res
1997;6:179–88.
[12] EspieCA.Understandinginsomniathroughcognitivemodeling.SleepMed2007;4:3–8.
[13] Broman JE, Hetta J. Perceived pre-sleep arousal in patients with persistent psy-
chophysiologic and psychiatric insomnia. Nord J Psychiatry 1994;48:203–7.
[14] Lichstein KL, Rosenthal TL. Insomniacs' perceptions of cognitive versus somatic
determinants of sleep disturbance. J Abnorm Psychol 1980;89:105–7.
[15] Nicassio PM, Mendlowitz DR, Fussell JJ, Petras L. The phenomenology of the pre-
sleep state: the development of the pre-sleep arousal scale. Behav Res Ther
1985;23:263–71.
[16] Ohayon MM, Caulet M, Guilleminault C. How a general population perceives its
sleep and how this relates to the complaint of insomnia. Sleep 1997;20:715–23.
109
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110

Page 8

[17] Van Egeren L, Haynes SN, Franzen M, Hamilton J. Presleep cognitions and attribu-
tions in sleep-onset insomnia. J Behav Med 1983;6:217–32.
[18] Borkovec TD, Grayson JB, O'Brien GT, Weerts TC. Relaxation treatment of pseu-
doinsomnia and idiopathic insomnia; an electroencephalographic evaluation. J
Appl Behav Anal 1979;12:37–54.
[19] Kuisk KA, Bertelson AD, Walsh JK. Presleep cognitive hyperarousal and affect as
factors in objective and subjective insomnia. Percept Mot Skills 1989;69:1219–25.
[20] Gross RT, Borkovec TD. The effects of a cognitive intrusion manipulation on the
sleep onset latency of good sleepers. Behav Ther 1982;13:112–6.
[21] Hall M, Buysse DJ, Reynolds CF, Kupfer DJ, Baum A. Stress-related intrusive
thoughts disrupt sleep-onset and continuity. Sleep Res 1996;25:163.
[22] Lichstein KL, Fanning J. Cognitive anxiety in insomnia: an analogue test. Stress
Med 1990;6:47–51.
[23] Harvey AG, Payne S. The management of unwanted pre-sleep thoughts in insom-
nia: distraction with imagery versus general distraction. Behav Res Ther 2002;40:
267–77.
[24] Haynes SN, Adams A, Franzen M. The effects of presleep stress on sleep-onset in-
somnia. J Abnorm Psychol 1981;90:601–6.
[25] Levey AB, Aldaz JA, Watts FN, Coyle K. Articulatory suppression and the treatment
of insomnia. Behav Res Ther 1991;29:8589.
[26] Monroe LJ. Psychological and physiological differences between good and poor
sleepers. J Abnorm Psychol 1967;72:255–64.
[27] Bonnet MH, Arand DL. 24 hour metabolic rate in insomniacs and matched normal
sleepers. Sleep 1995;18:581–8.
[28] Merica H, Blois R, Gaillard JM. Spectral characteristics of sleep EEG in chronic in-
somnia. Eur J Neurosci 1998;10:1826–34.
[29] Perlis MJ, Kehr EL, Smith MT, Andrews PJ, Orff H, Giles DE. Temporal and
stagewise distribution of high frequency EEG activity in patients with primary
and secondary insomnia and good sleeper controls. J Sleep Res 2001;10:93–104.
[30] Krystal AD, Edinger JD, Wohlgemuth WK, Marsh GR. NREM sleep EEG frequency
spectral correlates of sleep complaints in primary insomnia subtypes. Sleep
2002;25:630–40.
[31] Buysse DJ, Germain A, Hall ML, Moul DE, Nofzinger EA, Begley A, et al. EEG spectral
analysis in primary insomnia: NREM period effects and sex differences. Sleep
2008;31:1673–82.
[32] Nofzinger EA, Nowell PD, Buysee DJ, Vasko RC, Thase ME, Frank E, et al. Towards a
neurobiology of sleep disturbance in insomnia and depression: a comparison of
subjective, visually-scored, period amplitude, and power spectral density sleep
measures. Sleep 1999;22:99.
[33] Bastien CH, Guimond S, Lemelin S, St-Jean G, Turcotte I. Signs of insomnia in bor-
derline personality individuals. J Clin Sleep Med 2008;4:462–70.
[34] Bastien CH, St-Jean G, Turcotte I, Morin CM, Carrier J. Chronic psychophysiological
insomnia: hyperarousal and/or inhibition deficits? An ERPs investigation. Sleep
2008;31:887–98.
[35] Bonnet MH, Arand DL. Hyperarousal and insomnia: state of the science. Sleep Med
Rev 2010;14:9–15.
[36] Jansson-Fröjmark M, Harvey AG, Norell-Clarke A, Linton SJ. An epidemiologic
study of psychological processes in chronic insomnia. Submitted for publication.
[37] Edwards P, Roberts I, Clarke M, DiGuiseppi C, Pratap S, Wentz R, et al. Methods to
increase response rates to postal questionnaires (review). Cochrane Database
Syst Rev 2007;2.
[38] Edinger JD, Bonnet MH, Bootzin RR, Doghramji K, Dorsey SM, Espie CA, et al.
Derivation of research diagnostic criteria for insomnia: report of an American
Academy of Sleep Medicine Work Group. Sleep 2004;27:1567–96.
[39] Lichstein KL, Durrence HH, Taylor DJ, Bush AJ, Riedel BW. Quantitative criteria for
insomnia. Behav Res Ther 2003;41:427–45.
[40] Bastien CH, Vallieres A, Morin CM. Validation of the Insomnia Severity Index as an
outcome measure for insomnia research. Sleep Med 2001;2:297–307.
[41] Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr
Scand 1983;67:361–70.
[42] Kline P. The handbook of psychological testing. London: Routledge; 1993.
[43] Espie CA, Inglis SJ, Harvey L, Tessier S. Insomniacs' attributions: psychometric
properties of the Dysfunctional Beliefs and Attitudes about Sleep Scale and the
Sleep Disturbance Questionnaire. J Psychosom Res 2000;48:141–8.
[44] Jansson-Fröjmark M, Harvey AG, Lundh LG, Norell-Clarke A, Linton SJ. Psychometric
propertiesofaninsomnia-specificmeasureofworry:the AnxietyandPreoccupation
about Sleep Questionnaire. Cogn Behav Ther 2011;40:65–76.
[45] Tang NKY, Harvey AG. Correcting distorted perception of sleep in insomnia: a
novel behavioural experiment? Behav Res Ther 2004;42:27–39.
[46] Spoormaker VI, Verbeek I, van den Bout J, Klip C. Initial validation of the SLEEP-50
questionnaire. Behav Sleep Med 2005;3:227–46.
[47] Landis JR, Koch GG. The measurement of observer agreement for categorical data.
Biometrics 1977;33:159–74.
[48] Jansson-Fröjmark M, Linton SJ. The course of insomnia over one year: a longitudi-
nal study in the general population in Sweden. Sleep 2008;31:881–6.
[49] Morin CM, Belleville G, Bélanger L, Ivers H. The Insomnia Severity Index: psycho-
metric indicators to detect insomnia cases and evaluate treatment response. Sleep
2011;34:601–8.
[50] Osborne JW, Costello AB. Best practices in exploratory factor analysis: four rec-
ommendations for getting the most from your analysis. Pract Assess Res Eval
2005;10:1–9.
[51] Tabachnick BG, Fidell LS. Using multivariate statistics. Boston: Allyn and Bacon;
2001.
[52] Colton T. Statistics in medicine. Boston: Little, Brown; 1974.
[53] Stevens JP. Applied multivariate statistics for the social sciences (4th edition).
Hillsdale, NJ: Erlbaum; 2002.
[54] Peterson RA. A meta-analysis of variance accounted for and factor loadings in
exploratory factor analysis. Mark Lett 2000;11:261–75.
[55] Harris AL, Carney CE, Moss TG. Towards insomnia-specific models and assessment
of rumination. Sleep 2010;33:A210 (abstract supplement).
[56] Ohayon MM. Epidemiology of insomnia: what we know and what we still need to
learn. Sleep Med Rev 2002;6:97–111.
110
M. Jansson-Fröjmark, A. Norell-Clarke / Journal of Psychosomatic Research 72 (2012) 103–110