Perceptual and motor Skilk, 2008, 107,691-706.

C Perceptual and Motor Skills 2008

A NEW SCALE FOR MEASURING INSOMNIA:

THE BERGEN INSOMNIA SCALE '

STALE PALLESEN B J B m l BJORT7ATN

fa cult?^ of Psychology Department o f Pz~blicHealth atzd
Universzty of Bergen P r t ~ n a qHealth Care
lilzic'ersity of Bergen
Nortoe iaw Competence Center
$r Sleep Disorders Norwe zaiz Competence Center
j8r Sleep Disorders

INGER HILDE NORDHUS BBRGE SIVERTSEN, hlARI HJBKNEVIK

Facultj, of Psychology
University of Bergelz
Norwe hn Competence Center
j r .Sleep ~ i . s o r d c ~ s

CHARLES hl. MORTN

Summaty-A new scale for the measurement of insomnia, the Bergen Insomnia
Scale. was constructed on the basis of current formal and clinical diagnostic criteria
for insomnia. There are six items, of which the first three pertain to sleep onset,
maintenance, and early morning wakening insomnia, respectively. The last three items
refer to not feeling adequately rested. experiencing daytime impairment, and being
dissatisfied with current sleep. This scale was validated in three samples, 320 students.
2,645 community persons, and 225 patients. Cronbach alphas in the three samples
were .79,2 7 , and 30,respectively. The 2-urk. rest-retest reliability for students was
--
. I I . In the student and the patient samples, a two-factor solution was found, noctur-
nal symptoms and daytime symptoms, but in the community sample, a one-factor solu-
tion was found. The Bergen Insomnia Scale discriminated well between the patient
sample and the other two. In all three. values of convergent and discriminative valid-
ity in relation to other self-report measures were good, as well as in relation to poly-
somnographic data for patients. It is concluded that the Bergen Insomnia Scale has
good psychometric properties. It is one of very few insomnia scales which provide nor-
mative data for comparisons and which has been validated against subjective as well
as polysomnographic data.

According to the Diagnostic and Statistical Manual of Mental Disorders

(DSM-IV-TR; American Psychiatric Association, 2000), insomnia is defined
as a complaint consisting of difficulty in initiating or maintaining sleep or of
nonrestorative sleep, which lasts for at least 1 mo. To qualify for an insom-
nia diagnosis, it is also required that the sleep difficulty causes clinically

'Address correspondence to Stale Pallesen, Ph.D., Facult of Pvycholog University of Bergen,

Christiesgt. 12, 5015 Bergen, Norway or e-mail ~ s t a a l e . p a ~ ~ ~ s e n ~ ~ s y s p . u i ~ ~ n o ~ .

DO1 10.2466/PMS.107.1.691-706
692 S. PALLESEN. ET AL.

significant distress or impairment in social, occupational or other important

areas of functioning. Similar definitions can be found in other diagnostic sys-
tems, such as the International Classification of Disorders (ICD-10; World
Health Organization, 1992), and the International Classification of Sleep Dis-
orders (ICSD-2; American Academy of Sleep Medicine, 2005). Epidemiolog-
ical studies of insomnia have produced highly divergent estimates of its prev-
alence, ranging from about 2 % (Linjenberg, Almqvist, Hetta, Roos, & Ag-
ren, 1989) to 48% (Quera-Salva,Orluc, Golderberg, & Guilleminault , 1991).
This variability seems to be related to differences in the definition of insom-
nia, data-collection procedures, and the construction of items. Newer epide-
miological studies typically have emphasized formal diagnostic systems (Oha-
yon, Caulet, Priest, & Guilleminault, 1997; Pallesen, Nordhus, Nielsen,
Havik, Kvale, Johnsen, & Skjmskift, 2001) whereas older ones have normal-
ly not been based upon such systems (Karacan, Thornby, Anch, Holzer,
Warheit, Schwab, & XJilliams, 1976; Bixler, Kales, Soldatos, Kales, & Hea-
ley, 1979). Many of the self-report questionnaires in use today for the mea-
surement of insomnia are based upon criteria of formal diagnostic systems
(Soldatos, Dikos, & Paparrigopoulos, 2000; Levine, Lewis, Bowen, Kripke,
Kaplan, Naughton, rt al., 2003; Violani, Lucidi, Lombardo, & Russo, 2004).
In addition, insomnia scales, especially those used in epidemiological con-
texts, should be brief, have a short time frame and be easy to administer
(Moul, Hall, Pilkonis, & Buysse, 2004). Moreover, scales for the measure-
ment of insomnia should distinguish among sleep onset, sleep maintenance,
and mixed insomnia (Martin & Ancoli-Israel, 2002). Finally, it has also been
suggested that insomnia scales which provide normative data would be ex-
tremely useful in assessing the extent of sleep impairment at baseline and
the end-state status of patients who have completed insomnia treatment
(Morin, 2003). Although there are scales, such as the Insomnia Severity In-
dex, which adhere to many of the above-mentioned ideals (Bastien, Val-
lii.res, & Morin, 2001), there still seems to be room for improvement in the
measurement of insomnia. One potential limitation of the scales in use is
that they do not adhere to any specific cutoff value for defining waking time
as insomnia, such as the 30-min. cutoff value which has been suggested as a
clinical marker for insomnia (Lacks & Morin, 1992; Pallesen, Nordhus,
Havik, & Nielsen, 2001). In addition, many scales use rather unspecific
response categories such as "never," "seldom," etc. in spite of the fact that
people may interpret the meaning of such adjectives very differently
(Schwarz, 1999).
Against this background, a scale was constructed for the measurement
of insomnia which ( I ) was based upon the inclusion criteria for insomnia in
the DSM-IV-TR (American Psychiatric Association, 2000), (2) used the 30-
min. cutoff value to define waking time as insomnia, (3) specified the fre-
 BERGEN INSOMNIA SCALE 693

quency of the various insomnia symptoms in terms of days per week, (4)
was brief and easy to administer, (5) distinguished among subtypes of in-
somnia, (6) had a short time frame, and (7) provided normative data.
METHOD
Description of Scale
The Bergen Insomnia Scale was constructed on the basis of the inclu-
sion criteria for insomnia in the DSM-IV-TR (American Psychiatric Associa-
tion, 2000). There are six items (Appendix A, p. 706). The first four items
(assessing difficulties with sleep initiation, sleep maintenance, early morning
awakening, and nonrestorative sleep) correspond to the DSM-IV-TR crite-
rion A for insomnia. The last two items (assessing daytime impairment and
satisfaction with sleep) adhere to the DSM-IV-TR criterion B for insomnia.
All items cover the last month and are in accord with the time criterion for
insomnia in the DSM-IV-TR. Each item is rated on an 8-point scale, rang-
ing from O to 7 days per week. A total composite score is calculated by add-
ing together the scores for each item, yielding a total score with a possible
range of 0 to 42.
Samples
Student sample.-This sample comprised 320 undergraduate psychology
students at the University of Bergen, 248 women and 72 men. Their mean
age was 21.4 yr. (SD=3.7). Respondents were recruited at a lecture at which
they completed a set of questionnaires comprising the Athens Insomnia Scale
(Soldatos, et al., 2OOO), the Pittsburgh Sleep Quality Index (Buysse, Rey-
nolds, Monk, Berman, & Kupfer, 19891, the Beck Depression Inventory I1
(Beck, Brown, & Steer, 1996), the Beck Anxiety Inventory (Beck, Epstein,
Brown, & Steer, 1988), and the Bergen Insomnia Scale. Each set of ques-
tionnaires had a unique number, which the students were instructed to write
down and to keep for later. After 2-u.k., the Bergen Insomnia Scale was re-
administered to a subsample of about 200 students. They were then asked to
write their numbers on the questionnaires. A total of 182 students provided
their correct numbers. Thus, for these students data from the first set of
questionnaires could be matched with data from the re-administration of the
Bergen Insomnia Scale.
Communzty sample.-This sample comprised 5,000 subjects (18 to 80
years) randomly drawn from the Norwegian population register. They re-
ceived one set of questionnaires by post and were invited to participate in a
survey about sleep, which included the Bergen Insomnia Scale. Up to two
postal reminders were sent to those who did not return the questionnaires.
A total of 2645 subjects participated, a response rate of 52.9%. Of these,
1292 were men and 1353 were uTomen. The mean age was 48.1 yr. (SD=
15.3).
694 S. PALLESEN. ET AL

Patient sample.-This sample comprised all patients who were sched-

uled for an all-night clinical polysomnography assessment at the Sleep Clinic
at Voss County Hospital during the period March to December 2006 due to
suspicion of serious, but still undiagnosed sleep disorders (such as sleep ap-
nea, periodic limb movement disorder or hypersomnias such as narcolepsy
or idiopathic hypersomnia). A total of 225 patients, 115 men and 108 wom-
en (2 with missing sex information), constituted the patient sample. The
mean age was 42.1 yr. (SD= 14.4). In addition to undergoing clinical poly-
somnography, all patients completed both the Epworth Sleepiness Scale
(Johns, 1991) and the Hospital Anxiety and Depression Scale (Zigmond &
Snaith, 1983) in addition to the Bergen Insomnia Scale. Sleep stages, respira-
tory disturbances, and limb movement were scored according to standard cri-
teria (Rechtschaffen & Kales, 1968). Respiration (air flow, tidal volume, and
oxygen saturation) and anterior tibialis electromyographic readings were re-
corded to detect sleep apnea or periodic limb movements. The polysomnog-
raphy alas scored by a trained and certified polys~mnograph~ technician.
Measures
Athens Insovznia Scale.-This scale of eight items was constructed in ac-
cord with the ICD-10 criteria for insomnia. The first five items comprise
difficulties initiating sleep, maintaining sleep, early morning awakening, sleep
duration, and perceived sleep quality. The last three items measure aspects
of daytime impairment (well-being, physical and mental functioning, and
sleepiness). Each item is scored on a 4-point scale and a higher score indi-
cates more severe problems. The total score can range from O to 24. The
Athens Insomnia Scale has shown good reliability and validity (Soldatos, et
al., 2000).
Pittsburgh Sleep Quality Index.-The Index comprises 19 items which
measure different sleep problems during the last month. There are seven sub-
scales: subjective sleep quality, sleep latency, sleep duration, habitual sleep
efficiency, sleep disturbances, use of sleeping medication, and daytime dys-
function. A total score, ranging from 0 to 21, can be calculated by summing
the score for each subscale, a high score indicating more severe problems.
The index has been validated in several samples (Buysse, et al., 1989).
Beck Depression Inventory-II.-On this self-report measure of depres-
sion the respondent is instructed to describe how he has felt during the last
two weeks. The inventory has 21 items, each scored on a 4-point scale, yield-
ing a total score with possible range from 0 to 63. A high score indicates
problems of greater severity. The inventory has demonstrated good psycho-
metric properties (Beck, et al., 1996).
Beck Anxiety Inventory.--This self-report inventory measures anxiety
symptoms experienced by the respondent during the last week. It contains
21 items, each rated on a 4-point scale. The total score ranges from 0 to 63,
 BERGEN INSOMNIA SCALE 695

and a higher score is indicative of Inore severe symptoms. For the scale,
good reliability and validity have been reported (Beck, et dl., 1988).
Epworth Sleepiness Scale.-The scale describes eight situations, and re-
spondents estimate the likelihood of dozing off in each of these situations
using a +point scale, ranging from O = Would never doze to 3 = A high
chance of dozing. The total score can range from O to 24, and the scale has
!good psychometric properties (Johns, 1991).
Hospital Anxiety and Depressiorz Scale.-This scale is a self-report mea-
sure with 14 items, seven of which measure nonvegetative symptoms of anxi-
ety and seven of which measure nonvegetative symptoms of depression.
Each item is scored on a 4-point scale (0-3); a high score indicates more se-
vere problems (Zigmond & Snaith, 1983). The scale has good psychometric
properties (Bjelland, Dahl, Haug, & Neckelmann, 2002).
Clinical po1ysomnography.-Polysomnography comprises standard physi-
ological measures of brain activity (electroencephalogram), muscle tension
(electromyogram), and eye movements (electrooculogram). Based on these
measures one can distinguish between a waking state and sleep, and divide
the latter into five different stages (Stages 1-4 and rapid eye moveinent
sleep). In clinical contexts, the standard polysomnography setup is supple-
mented by measures of heart rate (electrocardiogram), respiration, oxygen
saturation (oximetry), and leg movements. Among other things, clinical poly-
somnography allows for the calculation of the number of apneas (breathing
pauses lasting for at least 10 sec.) or hypopneas (reduction of airflow lasting
for at least 10 sec.), and the number of leg moveinents per hour of sleep.
These measures are the basis for the Apnea-Hypopnea Index and the Peri-
odic Limb Movement Index, respectively (Butkov, 2002).
Stuttstzn
The data were analyzed using SPSS, Version 14.0. The internal consis-
tency was calculated as Cronbach alpha. The test-retest reliability was calcu-
lated using Pearson product-moment correlation coefficients, and the correct-
ed item-total correlation was calculated. Discriminant validity in terms of
group discrimination was investigated through an analysis of variance in
which the mean score of each item and the total score of the Bergen Insom-
nia Scale were compared across the three samples. Given many comparisons,
the significance level u7as adjusted by a Bonferroni adjustment. Based upon
the polysomnographic data, the patient sample was divided into four groups:
(1) sleep apnea patients (having an Apnea-Hypopnea Index above 51, (2) pa-
tients suffering from hypersomnia (having a mean sleep-onset latency of less
than 8 min. on the multiple sleep latency test), ( 3 ) patients with periodic
limb movement disorder (scoring above 15 on the Periodic Limb Movement
Index), and (4) patients with other sleep pathology (e.g., suffering from in-
somnia, circadian rhythm sleep disorders, depression). As some patients
696 S. PALLESEN. ET AL

could suffer from more than one sleep disorder, the patients in Groups 1 to
3 were compared with the patients in Group 4 on each of the single items
of the Bergen Insomnia Scale as well as on the total score using a one-way
analysis of variance.
To investigate the factor structure of the Bergen Insomnia Scale, a prin-
cipal component analysis was conducted for the three samples separately.
Parallel analysis was used to assess the number of factors to be retained
(Horn, 1965). When more than one factor was found, direct oblimin was
used as the rotation method. To investigate convergent and discriminant va-
lidity compared to other measures, Pearson product-moment correlations
were calculated between the summary scores of the Bergen Insomnia Scale
and the summary scores of the other measures administered in the student
and patient samples. In the patient sample, Pearson product-moment corre-
lations were also calculated for the relation of scores on the individual items
of the Bergen Insomnia Scale with relevant polysomnographic parameters.
To provide normative data, the data from the community sample were di-
vided into four age groups (18-29, 30-44, 45-59, and 60-80 years) and by
sex. A two-way analysis of variance was conducted to investigate whether
mean scores differed between the groups, based upon age and sex.

Reliability
Cronbach alphas were .79. 3 7 , and .80 in the student, community, and
patient samples, respectively. The 2-wk, test-retest correlation coefficient
based on a subsample of the students was . i i . The corrected item-total cor-
relation coefficients for the three samples are shown in Table 1.
Group Discrivzination
The analysis showed that there were significant differences in the mean
scores between the samples for Item 1 (F, ,,,, = 60.1, p < .01), Item 2 (F, ,,,,,=
31.1, p<.01), Item 3 (F,,,,,=57.1, p<.01), Item 4 (F,,,,,=179.2, p<.01),
Item 5 ( F,,,,:, =280.4, p< ,011, Item 6 (F,, ,,,= 180.81, p < ,011, and the total
score (FZ,,,,,= 151.1, p < .01). The post hoc test indicated the mean scores of
the patient sample arere significantly higher than the mean scores of the stu-
dent sample for all items except for Item 1, and for the total score. The
mean scores of the patient sample were significantly higher than the mean
scores of the community sample for all items and the total score. The mean
scores for Items 1 , 4,5 , and 6 were significantly higher for the student sam-
ple than for the community sample, whereas the mean scores for Items 2
and 3 were significantly higher for the community than for the student sam-
ple. There were no significant differences in the mean total composite scores
for the Bergen Insomnia Scale between the student and the community sam-
ples (see Table 1).
 BERGEN IKSOMNIA SCALE 697

MEANS,STANDARD
DEVIATIONS,
AND CORRECTED
ITEM-TOTAL
CORRELATION
COEFFICIENTS
FORBERGENINSOMNIA
SCALE
-. ~ -

Item Student Sample Community Sample Patient Sample Significant

I\, .SD r,, ,bl SD r;, SD 5, Differencesq
1 3.03 2.14 .47 1.90 2.17 .61 3.07 2.59 .45 b, c
2 0.91 1.51 .52 1.63 2.10 .68 2.31 2.38 .62 a,b,d
3 0.68 1.24 .35 1.54 2.04 .57 2.57 2.58 .53 a, b, d
1 2.90 1.89 .63 2.46 2.21 .68 5.41 1.94 .58 a, b, c
5 1.87 1.59 .57 1.20 1.81 .65 1.24 2.29 .46 a, b, c
6 2.49 2.09 .74 1.94 2.20 .80 4.88 2.16 .72 a,b,c
Total 11.88 7.43 10.67 9.73 22.49 9.86
--
a, b
Note.-a = Patient sample mean higher than student salm le h Patient sample mean higher
than community sample. c = Student sample mean higher tgan cArnunitT sample. d = Commu-
nity sample mean higher than student sample. i;,=Irem-total correlations. 'Bonferroni adjust-
ment of p < .05.

The results from the comparisons of patients with different sleep disor-
ders are presented in Table 2. Patients with sleep apnea scored lo\ver on
Items 1 and 4 than patients with other sleep pathology. Patients diagnosed
with hypersomnia scored lower on Items 1, 2, and 3 and on the total score
than patients with other sleep pathology. There was no difference on the
mean score on any items or the total score between patients suffering from
periodic limb movement disorder and patients with other sleep pathology.
TABLE 2
STANDARD
I ~ ~ E A KAKD
S L~E\'I.~T~oNs
FORBERGEN INSOMNIA
SCALE
FORPATIENTS
WITHDIFFERENT SLEEPDISORDERS
- - ~ ~-

Item Sleep hpnea Hlpersomn~a Periodlc Limb O t h e ~Sleep

in=6l) ( n= 2 0 ) hloxement in =49) Pnthology ( n = 119)
ZI SD X1 ~
SD
- ~ -
,LI SD hf SD
1 2.50" 2.43 1.08'' 1.61 3.06 2.44 3.41 2.67
2 2.40 2.42 0 1.83 2.52 2.25 2.33 2.48
3 2.34 2.57 1.30" 2.20 2.60 2.73 2.58 2.57
4 5.00'; 2.02 5.45 1.67 5.15 2.16 5.71 1.80
5 3.96 2.30 1.20 2.38 1.07 2.16 4.48 2.31
6 4.64 2.25 4.65 2.28 4.98 2.21 4.95 2.13
Total 20.52 9.67 17.78" 7.59 22.12 9.59 23.46 9.83
"'Mean is significantly different from the Other Sleep Pathology group.

Factor Structurr
In the student sample, a two-factor solution was found. The first factor
explained 49.6% of the variance while the second factor explained 19.0% of
the variance. Items 1,5 , and 6 loaded (2.40) on the first factor, and Items
1, 2, and 3 loaded ( 2 .40) on the second factor. The Pearson correlation co-
efficient between the two factors was .36 (see Table 3). In the community
sample, a one-factor solution was most appropriate. In the patient sample, a
698 S. PALLESEN. ET AL.

two-factor solution upas found. The first factor explained 51.0% of the vari-
ance, whereas the second factor explained a total of 19.7% of the variance.
Items 1, 2, 3, and 6 loaded ( 2 .40) on the first factor, and Items 4, 5 , and 6
loaded (2.40) on the second factor (see Table 3). The Pearson correlation
coefficient between the two factors was 3 8 .
TABLE 3
COMPONEKT
MATRIX(COMMUNITY
SAMPLE)4 K D PATTERN
M.ATRIX(STUDENT
AND P.411~~1
SAMPLES)
SHOWING FACTOR (2.40) FOR ITEMSOF BERGENINSOMNIA
LOADINGS SCALE
Tcem Community Sarnple Student Sample Patient Sample --

Factor 1 Factor 1 Factor 2 Factor 1 Facror

.
2 --

BISl .73 .43 .56

BIS2 .78 .81 .90
BISS .69 .88 .L)0
BIS4 .80 .88 .85
BIS5 .76 .87 .92
BIS6 .88 .83 .50 .52
% Variance 60.2 49.6 19.0 51.0 19.7

Convergent and Discriminant VaZidz'ty

The correlation coefficients for scores on the Bergen Insomnia Scale
and the other questionnaires in the student sample are shown in Table 4.
Scores on the Bergen Insomnia Scale correlated higher with scores on the
Athens Insomnia Scale than with ratings on the Beck Anxiety Inventory
(t,,, = 12.1, p < .01) and on the Beck Depression Inventory-I1 (t,,, =7.7,p <
,131). Scores on the Bergen Insomnia Scale also correlated higher with those
on the Pittsburgh Sleep Quality Index than with ratings on the Beck Anxi-
ety Inventory Jt2,,= 9.2, p < .01) and on the Beck Depression lnventory (t,,, =

PEARSON
CORREWTIOKS
FORTOTALSCORESON BERGENINSOMNI.ISCALE
AND OTHERQUESTIONKAIRES
I N TWO S.WPLLS
- --

2 3 4 Beck Anxiety
Inoentorv
Student Sample
1. Bergen Insomnia Scale .79t .73t .55+ .32+
2. Athens Insomnia Scale .76f .6l+ .39t
3. Pittsburgh Sleep Quality Index .50+ .36t
4. Beck Depression Inventory11 - -
.hot
Epworth Sleepiness Scale
Patient Sample
1. Bergen Insomnia Scale .32t
IIospital Anxiety and Depression Scale
2. Anxiety
3. Depression
 BERGEN INSOhlNIA SCALE 699

4.7, p < .01). Table 4 also shows the correlation coefficients between the
scores on the Bergen Insomnia Scale and the other questionnaires adminis-
tered to the student sample. In Table 5 , the correlation coefficients between
the individual items and total score of the Bergen Insomnia Scale and rele-
vant polysomnographic parameters are presented.
TABLE 5
PEARSON
CORRELATIOUS
FORITFM AND TOTALSCORESON BERGEN
INSOMKIA
SCALE
AKD POLYSOMNOGRAPHICVARIABLES
IK PATIENT
SAMPLE

Polysomnograpl~ic'l'ariable Item Total

1 2 3 4 5 6 Score
Time to sleep onset .31 .23 .12 .13 .05 .15 .26
'Waking time -.01 .28 .23 .O1 .01 .10 .15
Time Stage 1 -.02 .08 -.04 -.I1 -.01 -.07 -.06
Time Stage 2 -.I1 -.20 -.l6 -.06 .02 -.I1 -.I2
Time slow \\aye sleep .13 -.I8 -.19 -.02 -.03 -.06 -.08
Time RELI sleep -.05 -.42 -.40 .O1 .01 -.I8 -.25
Sleep efficiency -.I5 -.37 -.31 -.OX -.02 -.I6 -.26
No, of anakcnings .08 .23 .19 -.04 -.0$ .02 .10
Total sleep time -.06 -33 -33 -.06 .02 -.I7 -.20
Apnea-Hypopnea Index -.I9 .03 -.03 -.I3 -.06 -.03 -.lo
Periodic Limb Llovement Index .07 .15 .13 -.08 -.OS .06 .04
-- ~ --- -
- -.
Xote.-Item are: 1 =Sleep onset > 3 0 min.; 2 =Wake time > 30 min.; 3 =Early awakening; 4 =
Nonresrorative sleep; 5 =Negative daytime consequences: 6 =Dissatisfied with sleep. .00 < Y <
.14, ns: . 1 5 < r < . 1 7 , p < . 0 5 ; . 1 8 < r < l . 0 0 . p < . 0 1 .

The normative data for the Bergen Insomnia Scale are shown in Table
6 . No main effect was found for age group (F,,,-, = 0.5, p > .05). The main
effect of sex was, however, significant (F,,,,,=38.1, p<.05), indicating a
higher rating on insomnia among women than men. The interaction of age
group x sex was not significant ( F ,,, = 2.5, p > ,051.

In all three samples, the Cronbach alpha of the Bergen Insomnia Scale
indicated acceptable internal consistency. In addition, the corrected item-to-
tal correlation coefficients for all items in all samples were positive and great-
er than .40,except for Item 3 in the student sample, where the corrected
item-total correlation coefficient was .35. The latter finding may be explain-
ed by the fact that the range of Item 3 scores in the student sample was
quite small. Thus, the data show that there is high conceptual consistency
among the items (Kirshner & Guyatt, 1985). The 2-wk. test-retest reliability
a-as .77. As symptoms of insomnia may vary somewhat over time (Hohagen,
Kappler, Schramm, Rtemann, K'eyerer, & Berger, 19941, the magnitude of
the test-retest correlation coefficient seems adequate. In sum, these data sug-
gest that the reliability of the Bergen Insomnia Scale is good.
S. PALLESEN, ET AL.

n N rn
C C r n
3\

N C N

---
.1
- - N
c CC

m rn m
N r n m
N N N

2r2
N N N

---
M N % -
C N C

r.01-
N m m
"I N N 1
M S 0
C Q -
m N -1

I
CZCC
rn m

-- c.l

n u? a
a - r n
N N N
2 C N
M CC lA
3 - N

-
30r.C
-m

I,"
N N N '

m r . n

V V m
CCOC
W - t N

L L L i
X h X
7 - co
In 2

m % - c
 BERGEK INSOMNIA SCALE 701

The scores on the individual items and the total composite score for the
Bergen Insomnia Scale were expected to be higher in the patient sample
than in the two other samples. Compared to the students, the patients scored
higher on five of the six items and the total composite score. Only on Item
1 (problems initiating sleep) did the students score as high as the patient
sample. However, this finding is not very surprising, as previous surveys
have shown that difficulty falling asleep is among the most frequently report-
ed insomnia symptoms among young adults (Pallesen, Nordhus, Nielsen,
Havik, Kvale, Johnsen, et al., 2001). Compared to the comnlunity sample,
the data clearly showed that the patients scored significantly higher on all
the individual items and on the composite score for the Bergen Insomnia
Scale. Thus, the Bergen Insomnia Scale has good discriminant validity for
groups expected to differ with respect to insomnia symptoms (Kirshner &
Guyatt, 1985). Patients with sleep apnea scored lower on Item 1 (difficulties
falling asleep) than patients with other sleep pathology. This is in line with
findings that sleep apnea seldom is associated with severe problems falling
asleep (Chung, 2005). Sleep apnea patients scored lower than patients with
other sleep pathology on Item 4 (nonrestorative sleep). This may at first
sight seem illogical, but it should be kept in mind that the patients with oth-
er sleep pathology initially were referred to the polysomnographic investiga-
tion given severe subjective sleep problems that could represent serious sleep
pathology. The patients with hypersomnia scored significantly lower on Items
1, 2, and 3 and on the total score of the Bergen Insomnia Scale. This should
be expected as patients with hypersomnia seldom experience problems sleep-
ing but typically complain of excessive daytime sleepiness (Damiilliers, 2006).
The patients suffering from periodic limb movement disorder did not differ
from the patients with other sleep pathology on the Bergen Insomnia Scale.
Taken together, the findings from the comparisons of groups with different
sleep disorders were more or less as expected. It should be noted that the
Bergen Insomnia Scale is not aimed at differentiating between patients with
different sleep disorders.
The factor analyses showed a two-factor solution for both the student
and patient samples, whereas a one-factor solution seemed to be most appro-
priate for the community sample. In the student sample, Items 1, 2 , and 3
all loaded on one factor explaining 19.0% of the variance. whereas Items 4,
5, and 6 loaded on another factor which explained nearly half of the vari-
ance (49.6%). Given the content of the items, the first factor seems to rep-
resent nocturnal symptoms, and the second factor could be interpreted as
daytime symptoms of insomnia. A similar pattern of loadings was found in
the patient sample, but here, the nocturnal symptoms factor explained more
variance (51.0%) than the daytime symptoms factor (19.7%). In the patient
sample it should be noted that Item 6 (being dissatisfied with your sleep)
702 S. PALLESEN, ET AL

loaded on both factors, whereas this item only loaded on the daytime symp-
toms factor in the student sample. We suggest that Item 6 should be in-
cluded in the daytime symptoms factor. The two-factor solutions reported
here correspond well with current diagnostic criteria for insomnia that em-
phasize both nocturnal difficulties in sleeping and negative daytime conse-
quences (American Psychiatric Association, 2000) as well as with newer cog-
nitive models of insomnia in which nocturnal and daytime processes are as-
sumed to interact in vicious circles (Harvey, 2007).
On the other hand, a one-factor solution for the Bergen Insomnia Scale
was found in the community sample. The fact that the two factors found
both in the student and the patient sample correlated positively and moder-
ately in magnitude, taken together with the reasonably high Cronbach alphas
in all three samples, suggests that a one-factor solution may also be a mean-
ingful interpretation of the factor structure of the Bergen Insomnia Scale.
When it comes to the convergent and discriminative validity, investi-
gated by analyzing correlation coefficients in relation to other questionnaires,
the Bergen Insomnia Scale showed high correlations with two other sleep
scales, indicating that convergent validity is adequate. The Bergen Insomnia
Scale showed lower correlations with scales measuring anxiety and depres-
sion. This finding suggests that the Bergen Insomnia Scale also has good dis-
criminant validity (Kirshner & Guyatt, 1985). In both the student and pa-
tient samples, the Bergen Insomnia Scale correlated moderately and positive-
ly with measures of anxiety and depression. This finding is expected, as in-
somnia is often associated with these two psychopathological states (Benca,
Obermeyer, Thisted, & Gillin, 1992). The Bergen Insomnia Scale correlated
negatively with the Epworth Sleepiness Scale. Although sleepiness may be a
daytime consequence of insomnia (American Psychiatric Association, 2000),
many studies have shown that insomniacs are often hyperaroused and thus
have difficulty falling asleep both at night and during the daytime (Bonnet
& Arand, 2000). The correlation coefficients for polysomnographic parame-
ters with scores on the Bergen Insomnia Scale were in general small to mod-
erate. This would be expected, as self-report and objective measures of sleep
usually do not show very high concordance (Coates, Killen, George, Mar-
chini, Hamilton. & Thoresen, 1982) and given that the Bergen Insomnia
Scale covers a time period of one month, whereas the polysomnograpl~ic
data were collected from one night only. Item 1 (difficulties falling asleep)
correlated significantly with time to sleep onset and negatively with sleep ef-
ficiency and the Apnea-Hypopnea Index. This is in line with what would be
expected and the findings showing that sleep-apnea patients seldom report
problems initiating sleep (Chung, 20051. Item 2 (awakening from sleep) cor-
related positively with time to sleep onset, waking time, number of awaken-
ings, and the Periodic Limb Movement Index, and negatively with time in
 BERGEN INSOMNIA SCALE 7 03

Stage 2, Slow Wave Sleep and Rapid Eye Movement sleep, as well as with
sleep efficiency and total sleep time. All these correlations are in the ex-
pected direction. Item 3 (early morning awakening) correlated positively
with waking time and the number of awakenings and negatively with time in
Stage 2, Slow Wave Sleep, and Rapid Eye Movement sleep, in addition to
total sleep time and sleep efficiency. These findings are also in the expected
direction. Neither Items 4 nor 5 correlated significantly with any of the poly-
somnographic parameters. This is probably because Item 4 (not feeling ade-
quately rested) and Item S (experiencing daytime impairment) concern day-
time aspects of insomnia and not nocturnal events such as those measured
by the nocturnal polysomnography. This is also in agreement with studies
showing that insomniacs do not differ from healthy controls on several mea-
sures of daytime functioning and that insomniacs, in spite of improvement
on sleep parameters following treatment, do not show corresponding im-
provement on measures of daytime functioning (Omvik, Sivertsen, Pallesen,
Bjorvatn, Havik, & Nordhus, 2008). Item 6 (being dissatisfied with sleep)
correlated positively with sleep onset and negatively with time in Rapid Eye
Movement Sleep, sleep efficiency, and total sleep time. This finding suggests
that some objective indicators of poor sleep correlate with being dissatisfied
with sleep. The fact that Item 6 loaded both on the nocturnal and the day-
time symptom factors of insomnia strengthens this conclusion. The total
scores on the Bergen Insomnia Scale correlated positively with time to sleep
onset and waking time and negatively with time in Rapid Eye Movement
sleep, sleep efficiency, and total sleep time. In sum, all the correlation coeffi-
cients between both the individual Bergen Insomnia Scale items and the to-
tal score and the poly~omno~raphic parameters were in the expected direc-
tion, further strengthening the evidence for convergent validity of the Ber-
gen Insomnia Scale.
As has been confirmed in many previous studies (e.g., Palleson, Nord-
hus, Nielsen, Havik, Kvale, Johnsen, et al., 20011, the normative data for the
Bergen Insomnia Scale showed that women reported more insomnia than
men. However, there were no significant mean differences between age
groups on the Bergen Insomnia Scale. As medication status was not partial-
led out, this may explain why insomnia did not reliably progress with age.
As the scale has norms, comparisons with new individual samples are possi-
ble, an advantage many other insomnia scales do not provide. However, one
should take into account that the normative data were based on a random
sample from the general population, including poor as well as good sleepers.
Limitations
Although the Bergen Insomnia Scale was based on the inclusion criteria
for insomnia found in the Diagnostic and Statistical Manual of Mental Dis-
704 S. PALLESEN, ET A L .

orders (DSM-IV-TR; American Psychiatric Association, 20001, it does not

differentiate between insomnia symptoms stemming from primary insomnia
and those stemming from other sleep disorders. other mental disorders, sub-
stance use, or a general medical condition. LLke most sleep questionnaires,
the Bergen Insomnia Scale does not specifically target the night-to-night
variability that is so typical of persons with insomnia (PaUeson, Nordhus,
Havik, & Nielsen, 2001). Given wording of items, a few respondents may
have misunderstood and answered the items according to the last werk and
not the last month. The 30-min. cutoff was emphasized to establish a com-
mon reference when completing the scale, and this cutoff is in line with
suggestions by Lacks and Morin (1992). It should, however, be noted that
for some persons an unbroken wake time of above40 min. may not repre-
sent a subjective sleep problem, whereas others may react negatively to an
unbroken wake time of less than 30 min. (Lichstein, Durrence, Taylor,
Bush, & Riedel, 2003).
Recommendations and Future Dzrrctzons
The Bergen Insomnia Scale is free for use by clinicians and researchers.
It can probably fulfill the requirements for a unidimensional measure of in-
somnia, but it should not be substituted for broader and multidimensional
sleep questionnaires such as the Pittsburgh Sleep Quality Index (Buysse, et
al., 1989). The scale can provide a continuous score (0-42) of insomnia prob-
lems. It can also be scored according to the definitions of insomnia by the
American Psychiatric Association (2000) and the clinical criteria suggested
by Lack5 and Morin (1992) so that scoring 3 or above on at least one of the
first four items and scoring 3 or above on at least one of the last two items
would indicate the presence of insomnia. Severity may be judged from the
percentile scores given in Table 5 . Future studies with the Bergen Insomnia
Scale should concern its sensitivity and specificity in differentiating benveen
insomniacs and healthy controls. Items 1 and 3 may also be validated against
measures of circadian preference, such as the Horne-Ostberg Morningness-
Eveningness Questionnaire (Horne & Ostberg, 1976).
REFERENCES
AMERICAN ACADEMY OF SLEEPM ~ D I C I N12005)
E. The iizteynational classificatzoiz of sleep dzsor-
d m : diagnostic and coding maizual. (2nd ed.) Y'esrchester, IL: Author.
AMERICAN PSYCHIATRIC ASSOCIATION. (2000) Dzagno,tic and statistical rznnual of metital disor-
ders. (4th ed., Rev.) Washington, DC: Author.
BASTIEN,C. H., I ~ L L I ~ RA,,
E S&MORIN,
, C. M. 12001) Validation of the Insomnia Severity In-
dex as an outcome tor insomnia research. Sleep lMeledicine, 2 , 297-308.
BECK,A. T., BROWX,G. K., &STEER,R. A. (1996) BDI-11. Beck Depression Iricentory, nzanual.
San Antonio, TX: The Psychological Corp.
BECK,A. T., EPSTEIN, N., BROWN,G., &STEER,R. A. (1988) An inventory for measuring clinical
anxiety: psychometric properties. lozcrnul of Coizszilting and Clzntcal Psychology. 56, 893-
897.
BEXCA,R. M., OBERMEYER, W, H., T H I S I ~ DR.. A,, &GILLIN,J. C. (19921 Sleep and psychiatric
disorders: a meta-analysis.Archit'es oJGenera1 P.rychiatry. 4 9 , 651-686.
 BERGEN INSOMNIA SCALE 705

BIXLER,E. O., KALES,A., SOLDATOS, J. , D., &HEALEY,S. (1979) Prevalence of sleep

C. R., K ~ L E S
disorders in the Los Angeles metropolitan area. The America12 Journal of Psychlbtu?'. 136.
,9<7 ,9Lq
ILII-ILUL.
BJELLAND,I., DAHL,A. A., HAUG,T. T., &NECKELMANN, D. (2002) The validity of the Hospital
Anxiety and Depression Scale: an updated literature review. journi71 of Psychorotnatic Re-
.search. 52. 69-77.
BONNET, M. H., &ARAND,D. L. 12000) Activity, arousal. and the MSLT in patients with in-
somnia. Sleep, 23, 205-212.
BUTKOV, N. 12002) Polysomno ra hy In T. L Lee Chiong, M.J. Sateia, & M.A. Carskadon
iEds.1, Sleep medicine. ~h!a&l~hia, PA: H a n ~ & Belfus. Pp. 605-638.
BUYSSE,D. I . , REYNOLDS, K , R., &KUPFER,D. J. 11989) The Pitts-
C. F.. MONK,T. H., B ~ R M A S.
burgh Sleep Quality Index: a new instrument for ps~chiatricpractice and research. Pry-
chia~tyResearch, 28, 193-213.
CHUKG,K. F. (2005) Insomnia subtypes and their relationships to daytime sleepiness in pa-
tients with obstructive sleep apnea. Respzratzon, 72, 460-165.
COATES,T. J., KILLEN,J. D., GEORGE, J., MARCHINI, E., HAMILTON, S., &THOFSSEN, C. F.. (1982)
Estimating sleep parameters: a muititrait-multimethod analysis. Journal of Const~ltzngatzd
Clrnrcal Psycholog,,. 50, 345-352.
DAWILLIERS, Y.120061 Differential diagnosis in hypersomnia. Current Neurolog?; and "druro~ci-
ence Reports, 6, 156-162.
HARVEY. A. G. (2007) A cognitive model of insomnia. Behauioz~rResearch ~7ndTherapy. 40,
Xh9-X7I
-.
HOHAGEN, t-, GPPLER, C., SCHR~MM, E., RIEMANN, D.. WEYLRER, S., &BERGER, M. (1994) Sleep
onset insomnia, sleep maintaining insomnia and insomnia with early-morning awakening:
temporal stability oi subtypes in a longitudinal study on general-practice attetlders. Sleep,
17. 551-554.
HORK,J: L. (1965) A rationale and test for the number of factors in factor analysis. Psj~cho-
vzetrtka. 30.,179-185.
HORNE,J. A,, & OSTBERG, 0 . (1976) A self-assessment questionnaire to determine morning-
ness-eveningness in human circadian rhythms. Itzternatiotzal Jourtzal of Chrotzobzolo~:, 1 ,
97.1
,, 1 f l
JOHNS,M. W (1991) 4 new method for measuring daytime sleepiness: the Epworth Sleepiness
Scale. Sleep, 11, 540-545.
IC\R~CAN, I., THORNBY, J. I., ~ C H A/I., ! HOLZER,C. E., ~ ~ R H E G.I TI.,
, SCHWNB, 1. J., &WII~LI.AMS,
R. L. (1976) Prevalence of sleep disturbance in a primarily ~ ~ r b aFlorida n county. Sonnl
Sczences and I V l e d t ~ i t 10,
~ ~ .239-244.
KIRSHNER, B., & GUYATT, G. (1985) A methodological framework for assessing health indices.
joz~rnal of Chrorzic Diseases, 8, 27-36.
L C K S ,l?, &MORIN,C. M.(1992) Recent advances in the assessment and treatment of insom-
nia. ]oz~rnalof Consulting and Clinical P.s?;cholog, 60, 586-594.
L E ~ N ED., W., LEWIS,M. A,, BOWEK,D. J., KRIPKE,D. F., KAPLAN, R. hI.,NAUGHTON, M. J., &
SHL~MAKER, S.A. (2003) Reliability and validity of the Women's Health Initiative Insom-
nia Rating Scale. l'sycholoyical Assessvzent, 15, 137-118.
LICHSTEIN, K. L., DURRENCE, H. H., TAYLOR, D. J., BUSH,A. J., &&EDEL,B. K'. (2003) Quanti-
tative criteria for insomnia. Behavzozrr Resrarch and Therap?;,11, 427-445.
LINJENBERG, B., ALMQVIST, M.. HETT.~, J., ROOS,B. E., &AGREN,H. (1989) Age and the preva-
lence of insomnia in adulthood. European I'~?~chzatry, 3, 5-12.
MARTIN,J. L.. &ANCOLI-ISRAEL, S. 12002) Assessment and diagnosis of insomnia in non-phar-
nlacological intervention studies. Sleep Medicine Reuiclcs, 6, 379-406.
MORIN,C. M. 12003) Measuring outcomes in randomized clinical trials of insomnia treat-
ments. Sleep Medicine Reuiews. 7. 263.279.
MOUL,D. E., HALL,M., PILKONIS, P. A,, &BWSSE,D. J. 12001) Self-report measures of insom-
nia in adults: rationales, choices. and needs. Sleep "dediczne Reviez~>r, 8, 177-198.
OHAYON, M. IM., CAULET.M., PRIEST,R. G., & GUILLEMINAIJLT, C. (1997) DShl-IV and ICSD-
90 insomnia syniptonls and sleep dissatisfaction. BrititA journal 0.f Psychzatr?;, 171, 382-
388
,
OMYIK,S., SIVERTSEN, B., PALLESEN, S.. RJORVATN, B., HAVIK,0. E., & NORDHUS, I. H. (2008)
Daytime functioning in older patients suffering from chronic insomnia: treatment out-
come in a randomized contro led tr~alcomparing CBT with Zopiclone. Behau~oztr Re-
~earchand the rap^, 16, 623-611.
P.+LLESEN,S., NORDHUS,'~. H., HAVIK,0.F.., ~ N I E L S EG.K , H. (2001) Clinical assessment and
treatment of insomnia. Profes.szot?al Prychology~Research and Practice, 32. 115-124.
P.~LLESEN,S.,NORDHL~S, I. H., NIELSEN, .c; H., HAWK,O. E., KVALE,G., JOIIKSEK, B. H., &
706 S. PALLESEN, ET AL.

SKJOTSKIFT, S. 120011 Prevalence of insomnia in the adult Norwegian population. Sleep.

-71, , 771
, , - -779
,,*.
QUERA-SALVA, M. A., ORLUC,A., GOLDERBERG. F., & GCILLEMINAULT, C. 11991) Insomnia and
use of hypnotics: study of a French population. Sleep, 14, 386-391.
RECHTSCHAFFEN, A,, &KALES,A. (1968) A manual of standardzsrd terrn?nology and scoring sys-
tewz for. sleep stages of hurzau sleep. Los Anflea, CA: Univer. of California at Los Acge-
les, Brain Information ServiceIBrain Kesearc Institute.
SCHWARZ, N. (19991 Self-report: how the questions shape the answers. Amcricarz Psychologist.
54, 93-10>,
SOLDATOS,C. R.,DIKOS,D. G., &PAPARRIGOPOULOS, T. J. 12000) Athens Insomnia Scale: valida-
tion of an instrument based on ICD-10 criteria. Jozrrnal of Psychosomatic Research, 48,
555-560.
VIOLANI, C., LUCIDI,D. F,, LOMBARDO, C., &RUSSO,P. M. 12004) 17alidityof a Short Insomnia
Questionnaire: the SDQ. Brain Research Bulletzn. 63, 415-421.
WORLDHEALTHORGANIZATION. 11992) The ICD-10 clasrz~cationof mental atzd behavioral disor-
d e r ~ clrntcal
: descrrpt~onsand diagnostic guidelines. Geneva: Author.
ZIGMOND, A. S., &SKAITH,R. l? (1983) The Hospital Anxiety and Depression Scale. Acta Psy-
chtatrzca Scandinnz'zca, 67. 361-370.

Accepted September 16, 2008.

APPENDIX A
BERGENINSOMNIA
SCALE
Instructions. The questionnaire below contains six questions relating to sleep and tiredness.
Please circle the alternative (number of days per week) that suits you best. 0 means n o days
during the course of a week, 7 means every day during the course of a week.

Example. If, on three days during the course of a week, it has taken you more than 30 minutes
to fall asleep after you have switched the light off, circle alternative 3.

During the past month, how many days

a week has it taken you more than 30
minutes to fall asleep after the light was 0 1 2 0 4 5 6 7
switched off?

Number of Days Per Week

1. During the past month. how many days a
week has it taken you more than 30 minutes
to fall asleep after the light was switched off? 0 1 2 3 4 5 6 7

2. During the past month, how many davs a

week have you been awake for more than
30 minutes between per~odsof sleep? 0 1 2 3 4 5 6 7

3. During the past month, how many days a

week have you awakened more than 30
minutes earlier than you uished without
managing to fall asleep again? 0 1 2 3 4 5 6 7

4. During the past month, how many days a

week have vou felt that vou have not had
enough resf after wakinb up? 0 1 2 3 4 5 6 7

5 . During the past month, how many days a

week have you been so sleepp/tired that it
has affected you at school/work or in your
private life? 0 1 2 3 4 5 6 7

6. During the past month, how many days a

week have you been dissatisfied with your
sleep? 0 1 2 3 4 5 6 7