Correcting distorted perception of sleep in insomnia: a novel behavioural experiment?

Behaviour Research and Therapy 42 (2004) 27–39
www.elsevier.com/locate/brat
Correcting distorted perception of sleep in insomnia: a novel
behavioural experiment?
Nicole K.Y. Tang, Allison G. Harvey ?
Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford, OX1 3UD, UK
Received 15 October 2002; received in revised form 19 February 2003; accepted 27 February 2003
Abstract
Patients with primary insomnia overestimate their sleep onset latency (SOL) and underestimate their
total sleep time (TST). The present study aimed to test the utility of a novel behavioural experiment
designed to correct distorted perception of sleep among patients diagnosed with primary insomnia. Individ-
uals with primary insomnia were asked to wear an actigraph and keep a sleep diary for three nights. On
the following day, half were shown the discrepancy between the data recorded on the actigraph and their
sleep diary (Shown-Discrepancy Group), the other half were not shown the discrepancy (No-Demonstration
Group). Participants were then asked to wear the actigraph and keep a sleep diary for three further nights.
Following the behavioural experiment, the Shown-Discrepancy Group estimated their SOL more accurately
and reported less anxiety and preoccupation about sleep compared to the No-Demonstration Group. The
theoretical and clinical implications of these ?ndings are discussed.
? 2003 Published by Elsevier Ltd.
Keywords: Insomnia; Distorted perception; Sleep; Behavioural experiment; Anxiety; Preoccupation
The prevalence of insomnia in the United States is 33% with 9% reporting insomnia on a
regular nightly basis (Ancoli-Israel & Roth, 1999). As such, insomnia is among the most prevalent
psychological health problems. By de?nition, patients with chronic insomnia honestly and persist-
ently complain of not getting enough sleep (American Psychiatric Association, 1994). However,
the objective documentation of these has been somewhat illusory in that the majority of empirical
investigations have not been able to detect clinically signi?cant differences in polysomnograph-
ically measured total sleep time (TST) between patients with insomnia and good sleepers (e.g.,
Mendelson, 1993; Rosa & Bonnet, 2000; Spiegel, 1981). In studies where signi?cant statistical
? Corresponding author. Tel.: +44-1865-223912.
E-mail address: allison.harvey@psy.ox.ac.uk (A.G. Harvey).
0005-7967/$ - see front matter ? 2003 Published by Elsevier Ltd.
doi:10.1016/S0005-7967(03)00068-8

---

28
N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
differences were detected (e.g., Adam, Tomney, & Oswald, 1986; Gillin, Duncan, Pettigrew,
Frankel, & Snyder, 1979; Knab & Engel, 1988), the degree of sleep de?cit has often been regarded
as unremarkable (Adam et al., 1986; Coates & Thoresen, 1981). For example, in a review of 14
related studies, Chambers and Keller (1983) reported the average difference in objective total
sleep time between individuals with insomnia and normal sleepers to be only 35 min. This ?nding
suggests that there is a signi?cant proportion of patients for whom the severity of the complaint
is not commensurate with the amount of sleep obtained. Consistently, there is robust evidence
that a tendency toward overestimating sleep onset latency (SOL) and underestimating total sleep
time is ubiquitous among patients with insomnia (e.g., Adam et al., 1986; Carskadon et al., 1976;
Coates et al., 1982; Frankel, Coursey, Buchbinder, & Synder, 1976; Friedman et al., 2000; see
Perlis, Giles, Mendelson, Bootzin, & Wyatt, 1997 for review; Spielman, Sadkin, & Thorpy, 1987;
Wicklow & Espie, 2000).
In a recent cognitive model, we proposed that this characteristic distorted perception of sleep
is one of the core processes that serves to maintain insomnia (Harvey, 2002). We suggest that if
a patient with insomnia consistently perceives that he/she is not getting enough sleep, sleep related
anxiety and preoccupation will be fuelled (e.g., “I’m losing control”, “I’m not going to cope”, “I
will lose my job”). Escalating anxiety and preoccupation contribute to maintenance because they
are not conditions conducive to optimal sleep onset (Espie, 2002) nor to optimal daytime perform-
ance (Sarason, 1984).
To date, cognitive behavioural treatments for insomnia have not speci?cally targeted distorted
perception of sleep and the accompanying anxiety and preoccupation. The aim of the present
experiment was to pilot the ef?cacy of a novel behavioural experiment with a view to ?lling this
gap. Cognitive therapists employ two main methods to change distorted perceptions and unhelpful
beliefs (Clark, 1999; Salkovskis, 1999). The ?rst is a discussion technique. Through discussion,
therapists provide information to directly challenge distortion in perceptions and logic of the
patient’s beliefs (Beck, 1976; Ellis, 1962). However, as discussion is typically not suf?cient
(Teasdale, 1999), an alternative approach, behavioural experiment, has been developed. A behav-
ioural experiment involves arranging real life experiences to provide a crystal clear and memorable
demonstration to the patient that the perception or belief is not plausible, and from which the
patient can derive corrective feedback (Beck, 1995; Clark, 1999; Salkovskis, 1999).
Based on the cognitive model of insomnia, we predicted that a behavioural experiment designed
to correct distorted perception of sleep should be associated with (1) more accurate subjective
perception of sleep and (2) reduced sleep related anxiety and preoccupation. Based on the null
?ndings in comparisons of objectively measured sleep between good sleepers and individuals with
insomnia (e.g., Chambers & Keller, 1993), we did not predict a change to objective sleep esti-
mates. In an uncontrolled case series, an attempt to correct distorted perception of sleep among
individuals with insomnia showed promise (Downey & Bonnet, 1992) but is not amenable to
clinical practice as the procedure was intrusive (involving waking the patient 27 times on the
training night) and required access to technically trained staff and expensive equipment
(polysomnography).

---

N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
29
1. Method
1.1. Participants
Participants were staff and students at two local universities in Oxford. A total of 52 individuals,
aged 18–46 years, responded to an advertisement and attended a screening session. In the absence
of a psychometrically validated alternative, a structured clinical interview, the Insomnia Diagnos-
tic Interview (IDI), was used to identify the presence of insomnia. The IDI comprises ?ve sections
designed to carefully assess the presence of each DSM-IV (American Psychiatric Association,
1994) criteria for primary insomnia: (a) the predominant complaint is a dif?culty in initiating or
maintaining sleep or non-restorative sleep for at least 1 month (Cluster A); (b) the complaint
causes clinically signi?cant distress or impairment (Cluster B); (c) the insomnia does not occur
exclusively as a result of another sleep or mental disorder (Cluster C and D); and (d) the insomnia
is not due to the effects of a substance or illness (Cluster E). To be included in the present study,
participants needed to have met all the criteria above. Additionally, participants’ sleep disturbance
must have been present at least three nights per week for at least one month (Morin, 1993). Ten
participants were excluded for experiencing insomnia one or two nights a week (n = 4) or not
being clinically distressed or impaired (n = 6). Two further participants were excluded from the
?nal analyses due to unsuccessful experimental manipulation. Hence, the ?nal sample comprised
40 individuals diagnosed with primary insomnia. The structured clinical interview for the DSM-
IV (SCID; Spitzer, Williams, Gibbon, & First, 1996) was administered. Fifteen participants
(37.5%) met criteria for one or more DSM-IV Axis 1 diagnoses in the past (major depression
= 11, generalized anxiety disorder = 1, panic disorder = 2, post-traumatic stress disorder = 1).
Two participants (5%) met criteria for one current DSM-IV Axis 1 diagnosis (generalized anxiety
disorder, speci?c phobia). As insomnia is commonly comorbid with a range of psychological
disorders (Harvey, 2001b), individuals with comorbid problems were not excluded as (1) the
occurrence of comorbid psychological problems does not necessarily render insomnia as a second-
ary diagnosis (Harvey, 2001b), and (2) selecting ‘pure’ cases reduces the representativeness of
the sample (McCrae & Lichstein, 2001; Morin, Stone, McDonald, & Jones, 1994). Instead, we
made every effort to ensure that the sleep disturbance was currently the most distressing and
disabling problem (Di Nardo, Moras, Barlow, Rapee, & Brown, 1993), and that the sleep disturb-
ance did not occur exclusively as a result of the comorbid psychological disorder (American
Psychiatric Association, 1994).
1.2. Design and procedures
Participants were randomly allocated to either the Shown-Discrepancy Group (n = 20) or the
No-Demonstration Group (n = 20) and attended three sessions.
1.2.1. Session one
After obtaining informed consent, the IDI and the SCID (Spitzer et al., l996) were administered.
Participants then completed the Pittsburgh Sleep Quality Index (PSQI; Buysse, Reynolds, Monk,
Berman, & Kupfer, 1989), the Penn State Worry Questionnaire (PSWQ; Meyer, Miller,
Metzger, & Borkovec, 1990), the Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer,

---

30
N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
1988), and the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh,
1961). To index baseline distorted perception, participants were given an actigraph and a sleep
diary. They were instructed to wear the actigraph, on their non-dominant wrist, from two hours
prior to bedtime until rising in the morning for the next three consecutive nights and to complete
the sleep diary immediately on waking for the next three consecutive mornings (Days 1–3). Both
groups of participants were told that the purpose of the study was to observe the sleep/wake
schedule over a six-day period.
1.2.2. Session two
When participants returned for the second session, the actigraph and the sleep diaries for Days
1–3 were collected and the Anxiety and Preoccupation about Sleep Questionnaire (APSQ) was
administered. The Shown-Discrepancy Group then engaged in a behavioural experiment that
involved two parts. First, participants were shown and taught to read their actigraph recordings
from the last three days, after the experimenter had downloaded their sleep data from the actigraph
to a computer. The experimenter then gave each participant a short tutorial in which the participant
was informed that the data were generated from their wrist movement. The participant was then
orientated to the method of calculating their own SOL and TST. No further verbal interaction
ensued between the experimenter and the participants. Second, the participant completed a data
summary sheet on which the sleep diary and the actigraph recordings were juxtaposed. Participants
then calculated the discrepancy between the self-rated and objectively recorded sleep estimates.
The aim of this exercise was to provide a clear and memorable demonstration of the discrepancy
between how much participants thought they slept and how much the actigraph estimated they
had slept. The procedure was identical for the No-Demonstration Group except that they did not
do the behavioural experiment. To index the impact of the behavioural experiment, participants
were instructed to wear the actigraph for the next three nights and complete a sleep diary for the
next three mornings (Days 4–6).
1.2.3. Session three
During the ?nal session, the actigraph and the sleep diaries for Days 4–6 were collected. All
participants were asked to complete the APSQ with reference to the last three days. As a manipu-
lation check, the Shown-Discrepancy Group was asked to rate the extent to which they believed
the discrepancy between the sleep diary and actigraph recordings (1 “believe”, 10 “do not
believe”). Participants were debriefed as to the purpose of the study.
1.3. Materials
1.3.1. Anxiety and Preoccupation about Sleep Questionnaire
In the absence of an existing questionnaire to index our second hypothesis, we developed the
APSQ. The 10 items that comprise the APSQ were derived from statements made by patients
with primary insomnia as documented in previous research (Borkovec, 1982; Harvey, 2001a;
Watts, Coyle, & East, 1994). Participants were required to rate, on a 10-point scale, how true
each of the statements was for them over the past three days (1 “not true”, 10 “very true”). A total
score for the APSQ was obtained by summing across each rating. Example statements include: “I
worry about the amount of sleep I am going to get every night” and “I put great effort into

---

N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
31
rectifying my sleep problems”. The reliability and validity of the questionnaire was established on
a sample of 110 university students (41 of whom had PSQI global score
5, a cut off identifying a
clinically signi?cant sleep disorder with 90% speci?city; Buysse et al., 1989). The internal consist-
ency (Cronbach’s alpha) of the total scale was 0.92. The correlation with PSQI (r = 0.44, p
0.0001) and BAI (r = 0.37, p
0.0001) indicated that higher scorers on the APSQ were also
higher scorers on the PSQI and BAI.
1.3.2. Objective sleep estimates
An actigraph is a small, wrist-watch like device widely used in sleep research for the evaluation
of sleep–wake cycles. Embedded within is a miniaturized piezoelectric acceleration sensor that
detects and stores information about physical motion. Movement data can later be downloaded
into compatible software to generate an estimation of the sleep/wake cycle. In the present study,
the Mini Motionlogger Actigraph Basic (Ambulatory Monitoring Inc.) was used to provide an
objective estimate of SOL and TST. Data were collected in Zero-Crossing Mode at 60-s intervals.
Conversion of movement data in sleep parameters was performed by Action W using the Cole–
Kripke algorithm (Cole, Kripke, Gruen, Mullaney, & Gillin, 1992). To facilitate more accurate
scoring of sleep, participants were instructed to depress the event marker on the actigraph when
they turned the light off to go to sleep and to depress the event marker for the second time when
they wake up the following morning. The reliability and validity of the use of actigraphy has
been widely investigated. According to a comprehensive review by the American Sleep Disorders
Association (Sadeh, Hauri, Kripke, & Lavie, 1995), actigraphy is considered as a useful, accept-
able instrument to measure sleep–wake schedule and sleep quality. Compared to polysomnograph-
ically measured sleep, the ‘gold standard’, there was a strong agreement rate (above 90%; Jean-
Louis, Kripke, Mason, Elliott, & Youngstedt, 2001; Sadeh et al., 1995) in normal adults. However,
the agreement rate is likely to be lower in individuals who lie immobile for long periods (e.g.,
clinically depressed patients and insomnia patients) as the technology is less accurate in differen-
tiating quiet pre-sleep wakefulness from sleep (ASDA, 1995). For the use of actigraphy on insom-
nia patients, the epoch by epoch agreement rate ranges from 78.20% to 89.71% (Cole et al., 1992;
Jean-Louis et al., 1997; Mullaney, Kripke, & Messin, 1980; Sadeh, Alster, Urbach, & Lavie,
1989). Whilst there are concerns about the merits of actigraphy in providing a precise estimate
of sleep onset, the advantage of using actigraphy in the context of the present study was that (1)
it is non-intrusive, (2) it allows sleep to be monitored in the participants’ home, a natural sleeping
environment, (3) it is relatively inexpensive, and most importantly of all, (4) actigraphic data are
quickly downloaded and are presented in a clear and easily digestible manner such that participants
can easily calculate how long it took them to fall asleep and how long they slept in total.
1.3.3. Subjective sleep estimates
The daily sleep diary comprised four questions asking participants to record their bedtime,
SOL, TST, and wake time.

---

32
N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
2. Results
2.1. Participant characteristics
Table 1 presents the participant characteristics. There were no differences between the Shown-
Discrepancy Group and the No-Demonstration Group for sex composition (analysed with chi-
square), age, subjectively estimated SOL, subjectively estimated TST, PSQI, BAI, BDI, PSWQ
and insomnia duration (analysed with independent sample t-tests).
2.2. Baseline distorted perception of sleep
At baseline (Days 1–3), participants as a whole overestimated their SOL by 37 min and under-
estimated their TST by 46 min, indicating that this sample of participants did exhibit distorted
perception of sleep. Further, at baseline, the mean objective sleep data (SOL = 23.6 min, TST
= 7.2 h) were within the range recorded from good sleepers (SOL
30 min, TST = 7–8 h;
Morin, 1993), indicating that this sample of participants did not exhibit a ‘substantial de?cit in
nocturnal sleep time’ (Chambers & Keller, 1993, p. 649).
The analyses below involved a series of repeated measures ANOVAs with Group (Shown-
Discrepancy Group vs. No-Demonstration Group) as the between subjects factor and Session (pre-
experiment vs. post-experiment) as the within subjects factor. To explore signi?cant interactions,
t-tests were conducted with a Bonferroni adjustment to control for multiple comparisons (p
0.0125).
Table 1
Participant characteristics
Shown-Discrepancy Group
No-Demonstration Group
c2 (1)
Sex
Female
15
11
1.76
Male
5
9
t (38)
Age
22.8 (5.2)
24.6 (7.0)
0.92
sSOL
67.6 (33.9)
76.5 (55.3)
0.61
sTST
6.3 (0.9)
6.5 (1.4)
0.53
PSQI
11.0 (3.1)
10.1 (3.7)
0.84
BAI
14.8 (8.0)
11.2 (10.0)
1.28
BDI
12.5 (8.6)
12.2 (10.9)
0.10
PSWQ
55.3 (15.4)
48.9 (15.1)
1.33
Insomnia duration
5.2 (4.6)
4.9 (4.9)
0.19
Note: Mean values are reported with standard deviations in parentheses. Age is reported in years; sSOL =
subjectively estimated sleep onset latency in the last month (in minutes); sTST = subjectively estimated total sleep
time in the last month (in hours); PSQI = global score on the Pittsburgh Sleep Quality Index; BAI = total score on
the Beck Anxiety Inventory; BDI = total score on the Beck Depression Inventory; PSWQ = total score on the Penn
State Worry Questionnaire; Insomnia duration is reported in years

---

N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
33
2.3. The effect of the behavioural experiment on subjective sleep perception
2.3.1. Subjective SOL estimates
Fig. 1a depicts the change in subjective SOL for the two groups before and after the behavioural
experiment. There was no signi?cant main effect for Group. There was a signi?cant main effect
for Session, F (1, 38) = 13.78, p
0.001, such that the SOL estimates were shorter after the
behavioural experiment compared to before the behavioural experiment. There was a signi?cant
Group by Session interaction, F (1, 38) = 4.51, p
0.05. Follow up tests indicated that the
Shown-Discrepancy Group estimated their SOL to be shorter after the behavioural experiment
relative to before the behavioural experiment, t (19) = 3.52, p
0.002. This difference was not
observed for the No-Demonstration Group.
2.3.2. Subjective TST estimates
Fig. 1b depicts the change in subjective TST for the two groups before and after the behavioural
experiment. There was no signi?cant main effect for Group. There was a signi?cant main effect
for Session, F (l, 38) = 14.47, p
0.001, such that the TST estimate was longer after the behav-
ioural experiment compared to before the behavioural experiment. There was no interaction.
Fig. 1.
(a) Subjective sleep onset latency before and after the behavioural experiment. (b) Subjective total sleep time
before and after the behavioural experiment.

---

34
N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
2.4. The effect of the behavioural experiment on sleep related anxiety
The mean APSQ scores for the two groups before and after the behavioural experiment are
depicted in Fig. 2. There was no signi?cant main effect for Group. There was a signi?cant main
effect for Session, F (1, 38) = 4.09, p
0.05, such that the APSQ score was lower post-experiment
relative to pre-experiment. There was also a signi?cant Group by Session interaction, F (1,38)
= 13.15, p
0.001. Follow up tests indicated that the Shown-Discrepancy Group reported lower
sleep-related anxiety and preoccupation after the behavioural experiment compared to before the
behavioural experiment, t (19) = 3.83, p
0.001. This difference was not observed for the No-
Demonstration Group. Effect sizes were calculated so that a positive effect size denoted a
reduction in anxiety and preoccupation about sleep [d = (M
M ) / s
]. In order to give a
1
2
pooled
more precise estimate of the population variance, we used the pooled standard deviation, spooled
= ?(s12 + s22)/2 instead of the standard deviation of either group. Whilst the effect size for the
No-Demonstration Group was
0.13, that for the Shown-Discrepancy Group was 0.48, a medium
effect size according to Cohen’s (1988) de?nition.
2.5. The effect of the behavioural experiment on objective sleep
2.5.1. Objective SOL estimates
Before the behavioural experiment, the mean objective SOL for the Shown-Discrepancy Group
was 23.4 min (SD = 14.7) and for the No-Demonstration Group was 23.8 min (SD = 23.7). After
the behavioural experiment, the mean objective SOL for the Shown-Discrepancy Group was 17.9
min (SD = 12.3) and for the No-Demonstration Group was 26.5 min (SD = 21.3). No signi?cant
effects were observed on this variable.
2.5.2. Objective TST estimates
Before the behavioural experiment, the mean objective TST for the Shown-Discrepancy Group
was 7.1 h (SD = 1.0) and for the No-Demonstration Group was 7.2 h (SD = 1.7). After the
behavioural experiment, the mean objective TST for the Shown-Discrepancy Group was 7.5 h
Fig. 2.
Anxiety and Preoccupation about Sleep Questionnaire (APSQ) total score before and after the behavioural
experiment.

---

N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
35
(SD = 1.1) and for the No-Demonstration Group was 7.2 h (SD = 1.5). No signi?cant effects
were observed on this variable.
2.5.3. Manipulation check
The mean belief rating for the Shown-Discrepancy Group was 3.4 (SD = 1.9), indicating that
participants readily accepted the information conveyed during the behavioural experiment and
that, from the participants’ point of view, the behavioural experiment had credibility. Pearson
correlation analyses (2-tailed) were performed to examine the relationship between the credibility
of the behavioural experiment, as re?ected in the belief rating, and reduction in anxiety and
preoccupation about sleep, improvement in SOL and TST estimation. No signi?cant correlations
were detected. This is not surprising as the variability of the belief rating was limited by the
exclusion of individuals (n = 2) who scored more than 7.
3. Discussion
The ?rst prediction tested was that the behavioural experiment would be associated with
improved subjective sleep estimates. In support, the Shown-Discrepancy Group estimated their
SOL to be shorter after, compared to before, the behavioural experiment. This difference was not
signi?cant for the No-Demonstration Group. It is worth noting that after the behavioural experi-
ment the Shown-Discrepancy Group’s subjective SOL estimate (M = 27.9 min) improved to within
30 min, the cut-off for ‘normal’ SOL (Morin, 1993; Morin et al., 1999).
Interestingly, whilst the behavioural experiment was successful in correcting perception of SOL,
it was not as potent for TST. One possible explanation for this result was that perhaps the discrep-
ancy for SOL was demonstrated with more clarity: SOL was easily estimated at a glance, whereas
TST involved subtracting periods of wakefulness after sleep onset. It should be noted that for
TST there was a signi?cant main effect for Session, suggesting that both groups slept better across
time. This result is likely to be attributable to a diary keeping effect, such that an “enhanced
awareness” of sleep patterns may reduce anxiety over sleep loss and thus contribute to better
sleep (Morin, 1993, p.71).
The second prediction was that the behavioural experiment should be associated with reduced
sleep related anxiety and preoccupation. Consistent with this prediction, after the behavioural
experiment, the Shown-Discrepancy Group reported signi?cantly lower levels of anxiety and pre-
occupation about sleep compared to before the behavioural experiment, an improvement that was
not observed for the No-Demonstration Group. This ?nding is consistent with the view that dis-
torted perception of sleep serves to fuel anxiety and preoccupation with sleep (Harvey, 2002).
Several alternative explanations of the results need consideration. First, perhaps the therapeutic
effect seen in the Shown-Discrepancy Group was a response to demand characteristics inherent
to the experimental procedure. However, as great care was taken throughout the study to avoid
giving participants any indication that they were expected to show any improvement in their sleep,
this account is not entirely convincing. In future research, consideration should be given to
employing a proactive counterdemand procedure. Second, perhaps the therapeutic effect seen in
the Shown-Discrepancy Group was an artefact of the interaction between the participant and
experimenter during the behavioural experiment. Since interpreting actigraphic data was novel to

---

36
N.K.Y. Tang, A.G. Harvey / Behaviour Research and Therapy 42 (2004) 27–39
all participants, some verbal exchange was inevitable. However, this is an unlikely account of
the results as the verbal exchange was minimal. Third, perhaps the therapeutic effect seen in the
Shown-Discrepancy Group was a placebo effect. That is, perhaps participants would have shown
a reduction in sleep-related anxiety and preoccupation if they were simply shown any feedback.
Whilst the current design does not allow us to eliminate this possibility, it would be interesting
to test in future research whether or not participants given false feedback would show comparable
therapeutic effect. Nevertheless, it should be emphasised that during the session it was common
for participants to check that they were viewing their own data typi?ed by comments such as “I
remember I was woken up by my dog at 5 a.m. this morning and from then onward I couldn’t
go back to sleep. Wow, I can see that here (pointing to the screen)” and “Oh it’s interesting... I
did go to the toilet twice last night... I can see where I got up here (pointing to the screen) and
here (pointing to the screen)”.
There are several limitations to this study. First, the sample was drawn from the university
population. However, it should be noted that all participants met strict DSM-IV criteria for primary
insomnia, the mean PSQI global score for the sample was 10.5, and the average duration of the
insomnia for the sample was 5 years. Nonetheless, replication with a clinical sample is a crucial
“next step”. In particular, it would be interesting to check whether the behavioural experiment
has potential to improve objective sleep in more severe samples. Second, as participants were
followed up for three days, the long term durability of the behavioural experiment is not known.
Third, whilst the obvious advantage of actigraphy is that it is non-intrusive and inexpensive, the
disadvantage is that it provides an indirect measure of sleep based on movement and as such,
some error is inevitably introduced. For instance, actigraphy might have over scored sleep in
individuals who experienced long periods of quiet wakefulness before sleep onset. Nonetheless,
it should be noted that the epoch by epoch agreement rate between polysomnography and actigra-
phy for patients with insomnia is overall reasonably high (78.20–89.71%; e.g., Sadeh et al., 1995)
and the design of the experiment has contributed to keeping the error rate constant across the
two groups tested. Importantly, when our participants examined their own sleep data during the
behavioural experiment, they rated the actigraphic data to be a plausible objective representation
of their sleep. Fourth, the sample was not objectively assessed for comorbid sleep disorders such
as sleep apnea, periodic limb movement syndrome and narcolepsy. However, the IDI includes a
section designed to carefully screen for the key symptoms of comorbid sleep disorders. No partici-
pant endorsed any screening question. According to the Standards of Practice report by the Amer-
ican Academy of Sleep Medicine, full overnight polysomnographic assessment is not mandatory
for the routine evaluation of chronic insomnia, unless there is a valid rationale to support the use
of it (Chesson et al., 2000).
After treatment, patients with insomnia often demonstrate improvements on subjective indices
of sleep despite a lack of change in objectively measured sleep (e.g., Engle-Friedman, Bootzin,
Hazlewood, & Tsao, 1992 used polysomnography; Friedman et al., 2000 used actigraphy). This
pattern of ?ndings is consistent with those of the present study and adds weight to the argument
that subjective perception of sleep plays a central role in insomnia. As is re?ected in the current
diagnostic criteria, insomnia is essentially a subjective complaint of insuf?cient sleep and is ident-
i?ed primarily on the basis of the patient’s self-report. Future research may bene?t from greater
consideration of the clinical signi?cance of the role of subjective perception of sleep in the context
of insomnia.

---

Document Outline

• Correcting distorted perception of sleep in insomnia: a novel behavioural experiment?
  • Method
    • Participants
    • Design and procedures
      • Session one
      • Session two
      • Session three
    • Materials
      • Anxiety and Preoccupation about Sleep Questionnaire
      • Objective sleep estimates
      • Subjective sleep estimates
  • Results
    • Participant characteristics
    • Baseline distorted perception of sleep
    • The effect of the behavioural experiment on subjective sleep perception
      • Subjective SOL estimates
      • Subjective TST estimates
    • The effect of the behavioural experiment on sleep related anxiety
    • The effect of the behavioural experiment on objective sleep
      • Objective SOL estimates
      • Objective TST estimates
      • Manipulation check
  • Discussion
  • Acknowledgment
• References

---