pii: jc- 00515 -15http://dx.doi.org/10.5664/jcsm.

6130

S CI E NT IF IC IN VES TIGATIONS

Development of the Pain-Related Beliefs and Attitudes about Sleep (PBAS)

Scale for the Assessment and Treatment of Insomnia Comorbid with
Chronic Pain
Esther F. Afolalu, MSc1; Corran Moore, BSc1; Fatanah Ramlee, MHSc1; Claire E. Goodchild, PhD2; Nicole K.Y. Tang, DPhil1
Department of Psychology, University of Warwick, Coventry, UK; 2Department of Psychology, Institute of Psychiatry, London, UK
1

Study Objectives: Dysfunctional beliefs and attitudes about sleep is a cognitive-behavioral factor central to the development and perpetuation of insomnia.
Previous works to unravel the complex interrelationship between pain and insomnia have not explored the role of inflexible beliefs about the sleep-pain
interaction, possibly due to a lack of a valid instrument for doing so. The current study evaluated the psychometric and functional properties of a 10-item Pain-
Related Beliefs and Attitudes about Sleep (PBAS) scale.
Methods: The PBAS scale was administered to four clinical samples of chronic pain patients with comorbid insomnia: to examine the scale’s psychometric
properties (n = 137), test-retest reliability (n = 26), sensitivity to treatment (n = 20), and generalizability (n = 62). All participants completed the PBAS together
with validated measures of pain interference, insomnia severity, and cognitive-behavioral processes hypothesized to underpin insomnia.
Results: The PBAS scale was found to be reliable, with adequate internal consistency and temporal stability. Factor analysis suggested a 2-factor solution
representing beliefs about “pain as the primary cause of insomnia” and the “inevitable consequences of insomnia on pain and coping.” The PBAS total
score was positively correlated with scores from the Insomnia Severity Index (ISI) scale, Dysfunctional Beliefs and Attitudes about Sleep (DBAS) scale, and
the Anxiety and Preoccupation about Sleep Questionnaire (APSQ). It was a significant predictor of insomnia severity and pain interference. A significant
reduction in PBAS was also observed in patients after receiving a hybrid cognitive-behavioral intervention for both pain and insomnia.
Conclusions: Pain-related sleep beliefs appear to be an integral part of chronic pain patients’ insomnia experience. The PBAS is a valid and reliable
instrument for evaluating the role of these beliefs in chronic pain patients.
Keywords: pain, insomnia, beliefs, assessment, scale
Citation: Afolalu EF, Moore C, Ramlee F, Goodchild CE, Tang NK. Development of the pain-related beliefs and attitudes about sleep (PBAS) scale for the
assessment and treatment of insomnia comorbid with chronic pain. J Clin Sleep Med 2016;12(9):1269–1277.

I N T RO D U C T I O N
BRIEF SUMMARY
Current Knowledge/Study Rationale: Maladaptive beliefs about
Sleep beliefs are featured in cognitive-behavioral theories of the sleep-pain interaction are possible factors perpetuating pain-
insomnia as a factor central to the development and perpetu- related insomnia. This study examined the psychometric properties
ation of sleep disturbances.1–8 A key hypothesis across these of the Pain-Related Beliefs and Attitudes about Sleep (PBAS) scale,
theories of insomnia is that certain inflexible beliefs about specifically designed for the assessment and treatment of insomnia
sleep may exacerbate emotional responses, heighten cognitive comorbid with chronic pain.
Study Impact: Thinking about the interaction between pain and
and physiological arousal, and promote sleep practices that are sleep is an integral part of chronic pain patients’ insomnia experience.
paradoxically sleep interfering.1,2,5–9 Findings from the current study suggest that the PBAS is a valid
The Dysfunctional Beliefs and Attitudes about Sleep and reliable tool for detecting and assessing these unhelpful beliefs,
(DBAS) scale was developed as a research and clinical tool opening up new avenues for research and interventions.
to measure these maladaptive sleep beliefs.6 Both the origi-
nal 30-item version6,10 and the abbreviated 16-item version11
have shown satisfactory psychometric properties. Compared Dysfunctional beliefs may also mediate therapeutic change - in
with the original version, the abbreviated version has a similar treatment studies, a reduction in DBAS has been associated
but more compact 4-factor structure, which reflects themes of with an improvement in sleep as assessed with both subjective
thoughts concerning (1) inevitable consequences of insomnia, and objective measures of sleep efficiency at posttreatment.12,13
(2) worry/helplessness about insomnia, (3) unrealistic sleep ex- The improvement in sleep efficiency was well maintained at
pectations, and (4) assumed effectiveness of sleep medications. up to one-year follow-up.12,13 Together, these findings point to
Functionally, previous research has indicated that these dys- these dysfunctional beliefs as a logical treatment target.
functional beliefs are a clinical correlate of insomnia. Holding The growing interest in the relationship between pain and
these beliefs may be maladaptive; a high DBAS score is associ- sleep has driven the investigation of possible cognitive-be-
ated with greater insomnia severity, anxiety, and depression.11 havioral mechanisms underpinning pain-related insomnia.14–17

1269 Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016

EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

This has also led to the identification of elevated DBAS scores for insomnia and chronic pain.25 Sensitivity to change asso-
across a range of painful conditions, such as cancer, fibromy- ciated with treatment was examined by comparing responses
algia, and different types of chronic musculoskeletal, neu- to the PBAS collected at pretreatment and posttreatment as-
ropathic, and inflammatory pain.18–21 Not all patients with sessments. Participants in Sample 4 were 60 patients who, like
chronic pain have problems sleeping, but among those who do, participants in the first sample, completed the PBAS alongside
the vast majority report an onset of sleep disturbance during a range of validated questionnaires. Sample 4 was consisted
or following the onset of pain,22 and many believe that sleep of chronic pain patients with and without insomnia recruited
disruption is a secondary symptom of pain.23 Clinical expe- from a pain clinic in Gloucestershire, UK, whereas Samples
rience and qualitative research further suggest that some pa- 1, 2, and 3 were chronic pain patients with insomnia recruited
tients hold firm beliefs about how their pain affects sleep and from pain clinics based within the city of London, UK. The
how they would struggle to manage their pain following a poor difference in insomnia presentation and geographical setting
night’s sleep.23,24 Example thoughts are “I can never get com- provided a test of the generalizability of the findings across
fortable in bed because of the pain,” “The pain will wake me different demographic profiles.
up predictably,” and “I won’t be able to cope with my pain if I The 4 samples were recruited using similar inclusion and ex-
don’t sleep well.” clusion criteria. Essentially, all participants were adults (aged
These thoughts and beliefs reflect chronic pain patients’ 18 years or older) with chronic pain and recruited from hospi-
perceived cause(s) and consequence(s) of their insomnia and tal-based pain clinics. To be included in the study, participants
their assumed relationship between pain and sleep. These needed to be literate in English, have experienced non-malig-
thoughts and beliefs have an idiopathic focus on pain and nant pain for at least 6 months, and for Samples 1, 2, and 3,
its interaction with sleep that is not covered by items of the scored ≥ 15 on the Insomnia Severity Index.26 Sample 4 did not
DBAS, e.g., “I believe that insomnia is essentially a result of require participants to have a minimum ISI score, thus further
a chemical imbalance,” “I can’t ever predict whether I will testing the generalizability of the PBAS scale to chronic pain
have a good or poor night’s sleep,” and “Without an adequate patients that may present with subclinical threshold insomnia.
night’s sleep, I can hardly function the next day.” Smith et al. Exclusion criteria were: (1) receiving an injection or operation
found that chronic pain patients frequently reported pain-re- for their pain in the last month, (2) comorbid major psychiatric
lated thoughts during the pre-sleep period and that pre-sleep disorders, (3) hospitalized or with a life-threatening medical
thoughts pertaining to pain were significantly associated condition, and (4) visual/cognitive impairments that prevent
with poorer sleep continuity.15 However, the effect of these questionnaire completion. The participants’ demographic,
thoughts on subsequent sleep and pain management has never pain and sleep characteristics are presented in Table 1.
been empirically examined, as there is currently no instru-
ment specifically assessing these beliefs about sleep in the Measures
context of chronic pain.
Pain-Related Beliefs and Attitudes about Sleep (PBAS)
The current study evaluated the psychometric and functional
properties of a 10-item Pain-Related Beliefs and Attitudes The PBAS scale is a 10-item self-completed questionnaire
about Sleep (PBAS) scale, designed to assess pain-related dys- designed to assess patients’ beliefs about the interaction be-
functional beliefs and attitudes about sleep among people with tween pain and sleep (see Table 2 for individual items). The
chronic pain. Specifically, the factor structure, internal consis- items were generated based on previous research evidence and
tency, temporal stability, concurrent validity, predictive valid- clinical interviews with patients with concurrent chronic pain
ity, sensitivity to treatment and generalizability of the scale and insomnia.23,24 They reflect the pain-related sleep beliefs
were examined. and statements commonly endorsed by this population. Ad-
ministered as an extension to the DBAS, efforts were made to
avoid repetition of content. Only 10 items were developed and
METHODS included to keep the questionnaire short, user-friendly, and
easy to administer and score in clinical settings. Instruction
Participants for completion is identical to the DBAS, requiring participants
Secondary analysis of data drawn from 4 clinical study sam- to rate their level of agreement with each statement between 0
ples informed the evaluation of the psychometric and func- “strongly disagree” and 10 “strongly agree.” The total score is
tional properties of the PBAS scale. Each of the 4 studies based on the average score of all items, with a higher average
received ethical approval from the relevant research ethics score indicating stronger or more inflexible beliefs that pain
committee. and sleeplessness are inextricably linked. All samples com-
Participants in Sample 1 were 137 chronic pain patients who pleted the PBAS scale, alongside the following measures.
completed the PBAS scale and a collection of validated mea-
Dysfunctional Beliefs and Attitudes about Sleep
sures assessing sleep, pain and psychological characteristics.
Scale (DBAS-16)
Participants in Sample 2 were 26 patients who completed the
PBAS scale on 2 occasions, one week apart, for the assessment The brief version of the DBAS11 contains 16 items for the as-
of test-retest reliability. Participants in Sample 3 were 20 pa- sessment of general negative beliefs and attitudes about sleep-
tients who completed a pilot study investigating the effective- lessness. DBAS-16 was used because it has proven to be as
ness of a 4-week hybrid cognitive behavioral therapy (CBT) reliable and valid as the original 30-item version, but shorter,

Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016 1270

 EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

Table 1—Participant characteristics by sample.

Sample 1 Sample 2 Sample 3 Sample 4 Between-Samples
(n = 137) (n = 26) (n = 20) (n = 62) Test Statistics # Differences
Age (y) 46.0 (11.3) 50.1 (10.5) 48.5 (8.9) 52.3 (11.1) F (3,241) = 4.95** 1–4
Body mass index 27.8 (6.1) 27.7 (5.8) 28.8 (5.6) 27.6 (7.1) F (3,232) = 0.17 n/a
Sex (female %) 75.9 53.8 90.0 67.7 χ2 (3, n = 245) = 9.14* 2–3
Ethnicity (Caucasian %) 72.3 80.8 65.0 100.0 χ2 (3, n = 245) = 23.07*** 1–4, 2–4, 3–4
Education (degree %) 21.2 8.0 30.0 22.6 χ2 (3, n = 245) = 3.63 n/a
Marital status (married %) 48.2 50.0 55.0 66.1 χ2 (3, n = 245) = 5.68 n/a
Employment status (unemployed or 40.1 48.0 65.0 66.1 χ2 (3, n = 245) = 13.60** 1–4
on sick leave %)
Benefit status (receiving benefit %) 53.3 50.0 70.0 58.1 χ2 (3, n = 245) = 2.48 n/a
Pain duration (y) 8.0 4.3 6.1 8.0 χ2 (3, n = 243) = 5.59 n/a
Pain intensity (0–10 rating) 5.7 (2.6) 6.1 (2.3) 6.1 (1.8) 5.1 (2.3) F (3,240) = 1.43 n/a
Insomnia duration (y) 5.0 4.0 6.0 4.5 χ (3, n = 227) = 6.38
2
n/a
Insomnia severity (ISI) 20.4 (3.9) 20.4 (3.6) 20.3 (3.3) 14.4 (8.0) F (3,236) = 2.01*** 1–4, 2–4, 3–4

Means are presented with standard deviations in parentheses, except for pain and insomnia duration where medians are presented instead. Cases with
missing data were excluded from the analysis on a test-by-test basis, and hence the different sample sizes. 0–10 rating: The 0–10 pain rating scale of the
Brief Pain Inventory (Samples 1–3) and Short Form – McGill Pain Questionnaire (Sample 4) was used to assess pain intensity. ISI is Insomnia Severity
Index total score. # F values for one-way ANOVAs are reported for significant between sample-differences for continuous data categories. Bonferroni tests
post hoc analyses were used to follow up significant results of One-way ANOVA, except when the assumption of homogeneity was not assumed Dunnett
T3 post hoc tests were used. χ2 for chi square tests are reported for between-samples differences for categorical data with the critical p value set to 0.01 to
control for multiple comparisons. *p < 0.05, **p < 0.01, ***p < 0.001.

Table 2—The 10 items of the Pain-Related Beliefs and Attitudes about Sleep (PBAS) and their psychometric properties based
on Sample 1 (n = 137).
Factor Loadings (Rotated Solution)
Factor 1: Factor 2:
Standard Item-Total Pain as the primary Inevitable consequences of
Items Mean Deviation Correlations cause of insomnia insomnia on pain and coping
1. My insomnia is largely a result of the pain and 6.47 3.05 0.55 0.79 0.17
there is nothing I can do about it.
2. With the pain, I can never get myself comfortable 7.50 2.59 0.53 0.82 0.12
in bed.
3. The pain is always there when you try to have a 7.31 2.63 0.46 0.78 0.05
good night’s sleep.
4. When I am in pain, I simply can’t get to sleep no 6.94 2.65 0.55 0.64 0.31
matter how hard I try.
5. I know I can’t sleep through the night because 6.42 3.11 0.57 0.64 0.32
the pain will wake me up.
6. I get very annoyed when the pain wakes me up. 6.39 3.19 0.49 0.18 0.64
7. Not sleeping well is going to make my pain 5.01 3.37 0.55 0.06 0.83
worse the next day.
8. I won’t be able to cope with the pain if I don’t 4.68 3.40 0.63 0.11 0.88
sleep well.
9. Unless I get rid of the pain, I won’t sleep well. 5.98 3.38 0.55 0.26 0.65
10. The insomnia is taking away one of my few 5.56 3.31 0.53 0.24 0.65
respites from pain.
Eigenvalue – – – 2.92 2.97
Variance accounted for (R2) – – – 29.15 29.66
Internal consistency (α) of items in bold type 0.84 – – 0.82 0.81
Mean score (standard deviation) of items in bold type 6.23 (2.00) – – 6.93 (2.14) 5.53 (2.51)

and thus less burdensome to complete. Participants were asked average score is indicative of more strongly held negative be-
to rate their level of agreement with each statement between liefs about sleep. The DBAS-16 has demonstrated acceptable
0 “strongly disagree” and 10 “strongly agree,” thus a higher internal consistency (Cronbach α ≥ 0.77), temporal stability

1271 Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016

EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

(r = 0.83) over a 2-week interval and concurrent validity (cor- of the McGill Pain Questionnaire and sensitive to the effect of
relation with Insomnia Severity Index: r = 0.45). clinical pain treatments.30
Anxiety and Preoccupation about Sleep Questionnaire (APSQ) Analysis
Sleep-related anxiety was assessed using the 10-item APSQ.27 Statistical analyses were conducted using SPSS version 22.
Participants were asked to reference the previous month and Comparisons of demographic details (Table 1) across all 4
rate their agreement to each item between 1 (not true) and 10 samples were performed using ANOVA for continuous vari-
(very true). Higher total scores indicate greater sleep related ables (age, body mass index, pain intensity, and insomnia sever-
anxiety. The APSQ has shown good internal consistency ity index score) and χ2 for categorical variables (sex, ethnicity,
(Cronbach α = 0.92) and concurrent validity (Pittsburgh Sleep education level, marital status, employment status, benefits
Quality Index: r = 0.44; Beck Anxiety Inventory: r = 0.37). status, pain duration, and insomnia duration). All variables
were visually and statistically checked for normal distribution.
Insomnia Severity Index (ISI) Factor structure, internal consistency (Cronbach α), concurrent
The ISI26 is a measure of insomnia severity. Participants were in- validity (Pearson r), temporal stability (test-retest), sensitivity
structed to reference their sleep during the previous month and to treatment (change associated with treatment), predictive va-
rate the 7 items between 0 (not at all) and 4 (extremely). A higher lidity (stepwise linear regression), and generalizability of the
total score is indicative of more severe insomnia, with a clini- scale are reported below.
cal cutoff ≥ 15 that has optimal sensitivity (94%) and specificity
(94%). The ISI has demonstrated acceptable internal consistency
(Cronbach α = 0.76–0.78; item-total r = 0.36–0.67), concurrent R ES U LT S
validity (correlations with other sleep assessments ranging from
0.07 to 0.91) and high sensitivity to the effect of treatment. Demographics
For indexing pain severity and pain interference, Samples There were no significant differences on all demographic and
1–3 completed the Brief Pain Inventory (BPI) and Sample clinical characteristics (reported in Table 1) measured be-
4 completed the Short Form – McGill Pain Questionnaire tween the three London samples, except that Sample 3 had a
(SF-MPQ). Both scales are introduced below: significantly greater % of female participants than Sample 2
(χ2 (3, n = 245) = 9.14, p < 0.05).
Brief Pain Inventory (BPI) Overall, participants in Sample 4 were older, more ethni-
The BPI 28 is a self-report questionnaire administered to as- cally homogenous and had a lower level of insomnia severity,
sess the severity of pain and a measure of pain-related inter- compared with Samples 1–3. A significant difference in age
ference. For the pain interference subscale, the interference was found between Samples 1 and 4 (F3,241 = 4.95, p < 0.001).
sub-scale assesses the extent to which pain interferes with Additionally, a significant difference in insomnia severity
(1) general activity; (2) mood; (3) walking ability; (4) work was found between Samples 1–4, 2–4 and 3–4 (F3,236 = 2.01,
both inside and outside the home; (5) relations with people; p < 0.001). A significant difference was also found for ethnic-
(6) sleep; and (7) enjoyment of life. Participants were asked to ity between Samples 1–4, 2–4 and 3–4 (χ2 (3, n = 245) = 23.07,
rate the 7 items between 0 (does not interfere) and 10 (inter- p < 0.001) and between Samples 1 and 4 for employment sta-
feres completely) during the past week. A total pain interfer- tus (χ2 (3, n = 245) = 13.60, p < 0.01). No significant between
ence score is calculated as the average of the 7 items. A higher sample differences were found for the remaining categories.
average interference subscale score indicates greater interfer- See Table 1 for ANOVA and χ2 results.
ence in daily life due to pain. The BPI has been shown to have
good internal consistency (Cronbach α = 0.88), concurrent Distribution of PBAS Scores
validity with the Roland-Morris Disability Questionnaire The mean PBAS score was 6.23 (standard deviation = 1.99).
(r = 0.57) and high sensitivity to the effect of treatment.29 Visual examination of the histogram and Q-Q Plot of PBAS
Of the 4 items assessing pain severity, the numerical rating scores (Figure 1) did not indicate any significant deviations
scale of current pain rating was utilized to index present pain from a normal distribution or notable outliers (Skewness = −0.19
intensity in Samples 1–3 (0 [no pain at all] and 10 [pain as bad [standard error (SE) = 0.21]; Kurtosis = 0.34 [SE = 0.41]).
as you can imagine]).
Content Validity
Short Form - McGill Pain Questionnaire (SF-MPQ) A principal component analysis (PCA) was conducted with
SF-MPQ30 consists of 15 descriptors of pain (11 sensory; 4 af- orthogonal rotation (varimax), following satisfactory results
fective) and participants rate these on a scale of 0 (none) to from the Kaiser-Meyer-Olkin (KMO) test (0.81) and Bartlett
3 (severe). Pain scores are calculated from the sum of the in- test of sphericity (p = 0.001) verifying sampling adequacy
tensity rank values of the words chosen for sensory, affective, (i.e., suitability of data for structure detection). The KMO sta-
and total descriptors. The scale also includes a visual analogue tistic indicates the proportion of variance in the data that are
scale (VAS) and present pain index (PPI). The VAS (0–10) is attributable to underlying factors. KMO values greater than
similar to the current pain rating scale of the BPI and was used 0.80 are considered good and suggest that a factor analysis is
to index present pain intensity in Sample 4. The scale has been potentially useful for understanding the data structure. The
shown to correlate very strongly with the long-form version Bartlett test evaluates the data’s appropriateness for PCA by

Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016 1272

 EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

Figure 1—Histogram (left) and Q-Q Plot (right) of the distribution of PBAS scores in Sample 1.

Table 3—Intercorrelations of scores on PBAS, DBAS-16, APSQ, and ISI in Samples 1 and 4.
Sample 1 (n = 137) Sample 4 (n = 57#)
PBAS DBAS-16 APSQ ISI PBAS DBAS-16 APSQ ISI
PBAS – –
DBAS-16 0.65*** – 0.57*** –
APSQ 0.57*** 0.66*** – 0.45*** 0.71*** –
ISI 0.37*** 0.37*** 0.35*** – 0.64*** 0.46** 0.53*** –
#
Missing data due to incomplete response from chronic pain patients without insomnia. *p < 0.05, **p < 0.01, ***p < 0.001. PBAS, Pain-Related Beliefs and
Attitudes about Sleep; DBAS-16, Dysfunctional Beliefs and Attitude about Sleep Scale 16-item version; APSQ, Anxiety and Preoccupation about Sleep
Questionnaire; ISI, Insomnia Severity Index.

checking whether or not the correlation matrix concerned is high, demonstrating good internal consistency. Item-total cor-
an identity matrix, in which variables are noncollinear and relations were moderate to strong, ranging from 0.46 to 0.63.
unsuitable for structure detection. A rejection of the null
hypothesis (i.e., identity matrix) signifies that there is a re- Concurrent Validity
lationship between the variables.31 The 10 items loaded on Intercorrelations of the PBAS score with the scores of the
2 factors, both of which had eigenvalues over Kaiser’s cri- DBAS-16, APSQ and ISI were examined to establish concur-
terion of 1. These two factors combined explained 58.8% of rent validity (Table 3). In Sample 1 (n = 137), there were positive
the total variance. correlations between PBAS and DBAS-16 (r = 0.65, p < 0.001),
Table 2 shows items 1–5 loading on the first factor, account- APSQ (r = 0.57, p < 0.001), and ISI (r = 0.37, p < 0.001). A simi-
ing for 29.15% of the item variance. Factor 1 was labeled “pain lar pattern of relationships was observed in Sample 4 (n = 57),
as the primary cause of insomnia.” Items 6–10 loaded most where positive correlations were again found for PBAS with
strongly on the second factor, accounting for 29.66% of the DBAS-16 (r = 0.57, p < 0.001), APSQ (r = 0.45, p < 0.001), and
item variance. Factor 2 was labeled “inevitable consequences ISI (r = 0.64, p < 0.001).
of insomnia on pain and coping.”
Temporal Stability of the PBAS
Internal Consistency Twenty-six participants completed the PBAS twice, with a one-
The internal consistency of the PBAS was measured with week interval between administrations. A Pearson correlation
Cronbach α and item-total correlation. As evident in Table 2, coefficient calculated between the two PBAS scores showed a
the Cronbach α coefficients for both the full scale (α = 0.84) significant correlation, r = 0.91, p < 0.0001, indicating a high
and the subscales (Factor 1 α = 0.82 and Factor 2 α = 0.81) were level of test-retest reliability and temporal stability.

1273 Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016

EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

Table 4—Summary of stepwise multiple regression analysis predicting insomnia severity and pain interference in Sample 1
(n = 137).
Predicted Variable Predictors F R R2 Adjusted R2 ΔR2 B SE B β t
ISI Model 1 Predictor:
PBAS 21.49** 0.37 0.14 0.13 0.14 0.72 0.16 0.37 4.64**
Model 2 Predictors: 16.40** 0.44 0.20 0.19 0.06
PBAS + 0.60 0.16 0.31 3.82**
BPI Pain Intensity 0.38 0.12 0.25 3.14*
Model 3 Predictors: 13.12** 0.48 0.23 0.21 0.03
PBAS + 0.36 0.18 0.18 1.93*
BPI Pain Intensity 0.39 0.12 0.26 3.26**
APSQ 0.04 0.02 0.22 2.33*
BPI Pain Interference Model 1 Predictor:
BPI Pain intensity 38.62** 0.42 0.22 0.22 0.22 2.46 0.40 0.47 6.21**
Model 2 Predictors: 27.47** 0.29 0.28 0.07 0.07
BPI Pain Intensity + 2.10 0.39 0.40 5.35**
PBAS 1.82 0.51 0.27 3.59**

*p < 0.05, **p < 0.001.

PBAS scores decreased slightly from the first (mean = 6.87, pretreatment-to-posttreatment reductions in DBAS-16 (r = 0.85,
SE = 0.39) to the second (mean = 6.79, SE = 0.41) adminis- p < 0.0001), BPI pain interference (r = 0.58, p < 0.01) and ISI
tration, however, a paired t-test revealed that this was not a (r = 0.57, p < 0.01).
significant decrease, t25 = 0.44, p = 0.66. The temporal stabil-
ity of PBAS scale compared well with that of the DBAS-16, PBAS was the Strongest Predictor of Insomnia
which also showed a significant correlation between scores, Severity (ISI) among Chronic Pain Patients
r = 0.94, p < 0.0001. However, a paired t-test revealed that Linear regression analyses were performed to determine
DBAS-16 mean scores dropped from the first (mean = 6.0, whether PBAS was a predictor of insomnia severity. The step-
SE = 0.42) to the second (mean = 5.63, SE = 0.42) admin- wise method was used to enter all potential predictors (PBAS,
istration. The drop in DBAS-16 was small but significant, BPI/SF-MPQ pain intensity, DBAS-16, and APSQ) to the re-
t25 = 2.45, p < 0.05. gression models, to identify a combination of predictors that
account for the most variance in the predicted variable. Since
Sensitivity to Treatment the correlations among the potential predictors were all less
The responses of 20 participants receiving a 4-week course of than 0.80, multicollinearity was not assumed to be of major
hybrid CBT for sleep and pain management as part of a pilot concern in the multiple regression analyses. Further tests to
study were analyzed, in order to determine the sensitivity of explore if the data met the assumption of collinearity indicated
the PBAS for detecting reduction in dysfunctional pain-related that multicollinearity was not a concern, Variance Inflation
sleep beliefs after treatment. Factor (VIF) values were 1.72 or less and Tolerance statistics
The hybrid CBT was effective in reducing ISI scores of in- all greater than 0.58.31 As it is often recommended to carry out
somnia severity from pretreatment (mean = 20.3, SE = 0.73) some form of validation analysis when stepwise regression is
to below the clinical threshold at posttreatment (mean = 7.75, used,31 we have two samples with which we can cross-validate
SE = 1.48), t19 = 9.39, p < 0.05, r = 0.90. The hybrid CBT was the regression model predicting ISI scores.
also associated with a significant reduction in BPI pain interfer- For Sample 1 (Table 4), PBAS scores was selected first into
ence score from pretreatment (mean = 7.06, SE = 0.22) to post- the model, significantly predicting insomnia severity individu-
treatment (mean = 4.09, SE = 0.38), t19 = 8.99, p < 0.05, r = 0.90. ally (F1,135 = 21.49, β = 0.37, p < 0.001), accounting for 14%
Pretreatment endorsement of dysfunctional beliefs and at- of the variance. PBAS scores also jointly predicted insomnia
titudes from the treatment sample was similar to that given severity with pain intensity in the second model and with pain
by Sample 1 (PBAS mean = 6.23, DBAS-16 mean = 4.67). intensity and APSQ in the third model. The addition of pain in-
PBAS scores showed a significant reduction in this sample, tensity and APSQ in the model increased the percentage vari-
from pretreatment (mean = 6.29, SE = 0.31) to posttreatment ance explained to 19% and 21% respectively.
(mean = 3.33, SE = 0.39), t19 = 6.94, p < 0.0001, r = 0.85. A simi- A similar regression analysis (Table 5) was carried out in
lar significant reduction following treatment was also observed Sample 4 to cross-validate the predictive model. In this sec-
for DBAS-16 scores from 4.73 (SE = 0.33) to 2.26 (SE = 0.33), ond analysis, PBAS was again a significant predictor of insom-
t19 = 6.8, p < 0.0001, r = 0.84. nia severity individually (F1,54 = 35.55, p < 0.001), accounting
Of relevance, pretreatment-to-posttreatment reduc- for 40% of the variance, and jointly with pain intensity in the
tions in the PBAS scores were significantly correlated with second model and with pain intensity and APSQ in the third

Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016 1274

 EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

Table 5—Summary of stepwise multiple regression analysis predicting insomnia severity in Sample 4 (n = 56).
Predicted Variable Predictors F R R2 Adjusted R2 ΔR2 B SE B β t
ISI Model 1 Predictor:
PBAS 35.55** 0.63 0.40 0.39 0.40 2.09 0.35 0.63 5.96**
Model 2 Predictors: 30.10** 0.73 0.53 0.51 0.14
PBAS + 1.56 0.33 0.47 4.61**
SF-MPQ Pain Intensity 0.13 0.04 0.40 3.91**
Model 3 Predictors: 25.81** 0.77 0.60 0.58 0.07
PBAS + 1.14 0.35 0.35 3.29*
SF-MPQ Pain Intensity 0.13 0.03 0.40 4.15**
APSQ 0.09 0.03 0.29 2.93*

*p < 0.05, **p < 0.001.

model. The addition of pain intensity and APSQ in the model Modeling on the DBAS, the PBAS contains items that capture
increased the percentage of variance explained to 53% and unhelpful beliefs about sleep that are prevalent among chronic
60%, respectively. In both samples, DBAS-16 did not emerge pain patients. The two emerging factors from the PBAS scale
as a significant predictor of insomnia severity. are “pain as the primary cause of insomnia” and the “inevi-
table consequences of insomnia on pain and coping,” reflecting
PBAS Strengthened the Prediction of Pain Interference a negative view of the bidirectional association between pain
(BPI) among Chronic Pain Patients and sleep. The development of such view is not ungrounded,
Using Sample 1 data (Table 4), a further stepwise regression considering that poor sleep is usually a marker for worsened
was conducted with BPI pain intensity, PBAS, DBAS-16, and physiological and psychological pain-related outcomes.32 Poor
APSQ entered as potential predictors of pain interference (BPI sleep also contributes to the exacerbation of pain processes in
pain interference score). Pain Intensity scores was selected first both healthy and chronic pain individuals.33,34 The concern here
into the model as a significant predictor of pain interference is that inflexible thinking about the sleep-pain interaction may
(F1,135 = 38.62, β = 0.47, p < 0.001), accounting for 22% of the exacerbate emotional responses to sleep disturbance, accentu-
variance. The addition of PBAS (F1,134 = 27.47, p < 0.001) to the ate pre-sleep cognitive and physiological arousal, and promote
predictive model increased the percentage variance explained maladaptive sleep practices and pain coping strategies.
to 28%. DBAS-16 (t = −0.96, p = 0.34) and APSQ (t = −0.30, In the context of primary insomnia, it has been known for
p = 0.74) were not significant predictors of pain interference. some time that dysfunctional beliefs about sleep—as measured
with DBAS—contribute to the development and maintenance
of insomnia symptoms.3 Specifically, cognitive processes of
D I SCUS S I O N worry and holding negative dysfunctional beliefs about sleep
are significant predictors of persistent insomnia symptoms
Maladaptive beliefs about the sleep-pain interaction are pos- over a period of 18 months.35 The development of PBAS offers
sible factors perpetuating pain-related insomnia. In this paper, a tool to examine the extent to which rigid thinking about the
we examined the psychometric properties of the PBAS for the pain-sleep relationship contributes to the manifestation of in-
assessment of these beliefs among chronic pain patients. The somnia in clinical groups for which pain is a constant or recur-
scale showed good reliability with adequate internal consis- rent feature (e.g., fibromyalgia, arthritis, temporal mandibular
tency and temporal stability, and the total score of the PBAS joint disorder, cancer).
correlated with established measures of insomnia severity. The PBAS scale showed moderate to strong correla-
With scale items specifically designed to tap into the patients’ tions with DBAS and APSQ, suggesting that there are likely
perceived interaction between sleep and pain, the PBAS out- overlaps in terms of the psychological factors at play in the
performed the DBAS and emerged as the best predictor of in- maintenance of primary insomnia and pain-related insom-
somnia severity. It also independently added to the prediction nia. However, functionally speaking, the PBAS was a better
of pain interference, above and beyond the expected effect of predictor of insomnia severity and pain interference among
pain intensity. Importantly, the scale was sensitive to treat- chronic pain patients with comorbid insomnia. This provides
ment; significant reduction in PBAS was observed in chronic evidence for the specificity of the scale and highlights pain-
pain patients following a course of hybrid CBT for pain and in- focused sleep beliefs, as assessed by PBAS, as potential cogni-
somnia. Reduction in PBAS was also correlated with improve- tive treatment targets when addressing pain-related insomnia.
ments in insomnia severity and pain interference. Psychological treatments for chronic pain have predominantly
The development of the PBAS was motivated by the find- focused on pain management. Although sleep hygiene advice
ings of Smith et al.,15,16 which demonstrated that the focus and is routinely given as part of the standard treatment, emphasis
content of sleep-related cognitions were different between on sleep is light, and not enough time and effort are devoted
“primary” insomnia and insomnia comorbid with chronic pain. to helping patients understand the pain-sleep interaction and

1275 Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016

EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

addressing the cognitive-behavioral factors perpetuating their and allows the effect of these beliefs on sleep and pain man-
insomnia. Not surprisingly, outcome data have shown that agement to be empirically examined. Clinically, it provides a
completion of pain management programs does not always re- tool to identify dysfunctional thoughts that require cognitive
sult in improved sleep.36 In developing a new hybrid CBT that therapy and can be used as a measure of treatment progress
aims to tackle pain and sleep simultaneously, special attention when treating chronic pain patients with comorbid insomnia.
has been given to addressing patients’ unhelpful beliefs about
pain, sleep and the interaction of the two.25 The PBAS was
successful at detecting changes in these pain-related sleep be- A B B R E V I AT I O N S
liefs following a course of this hybrid CBT. Of particular clini-
cal relevance, those patients who showed a reduction in PBAS APSQ, anxiety and preoccupation about sleep questionnaire
scores were also those who demonstrated greater reduction in BPI, brief pain inventory
insomnia symptoms and pain interference. The PBAS is poten- CBT, cognitive behavioral therapy
tially a useful clinical tool for guiding and assessing progress DBAS, dysfunctional beliefs and attitude about sleep
of insomnia treatment among pain patients. ISI, insomnia severity index
The development of the PBAS followed the recommended KMO, kaiser meyer olkin
procedure of scale development.37,38 Data from four indepen- PBAS, pain-related beliefs and attitudes about sleep
dent samples of chronic pain patients were collected to ex- PCA, principal component analysis
amine its score distribution, structural dimension, internal PPI, present pain index
consistency, temporal stability, criterion validity, sensitivity SE, standard error
to treatment, and generalizability. The four samples were of SF-MPQ, short form-mcgill pain questionnaire
different sizes recruited from different clinics. Samples 1, 2, VAS, visual analogue scale
and 3 were chronic pain patients with clinical levels of comor- VIF, variance inflation factor
bid insomnia, whereas Sample 4 consisted of a mix of chronic
pain patients with and without clinical insomnia. Such differ-
ence in participant composition and insomnia severity might R E FE R E N CES
also explain why DBAS was not found to be a significant pre-
1. Espie CA, Broomfield NM, MacMahon KM, Macphee LM, Taylor LM. The
dictor of ISI. In other words, when compared with the PBAS, attention-intention-effort pathway in the development of psychophysiologic
DBAS may not address the prominent cognitive feature among insomnia: a theoretical review. Sleep Med Rev 2006;10:215–45.
pain patients with subclinical and clinical levels of insomnia. 2. Harvey AG. A cognitive model of insomnia. Behav Res Ther 2002;40:869–93.
Further research administering the PBAS in heterogeneous 3. Jansson M, Linton SJ. Psychological mechanisms in the maintenance of
community samples of chronic pain individuals would help insomnia: arousal, distress, and sleep-related beliefs. Behav Res Ther
determine to what extent the scale distinguishes good-sleeping 2007;45:511–21.
from poor-sleeping pain individuals. There is also potential for 4. Jansson-Fröjmark M, Linton SJ The role of psychological mechanisms to
insomnia in its early phase: a focus on arousal, distress, and sleep-related
reducing the number of scale items in order to cut administra- beliefs. Psychol Health 2008;23:691–705.
tion times short but still maintain the necessary psychometric 5. Lundh L-G, Broman J-E. Insomnia as an interaction between sleep-interfering
properties of the full scale. Additional treatment studies with and sleep-interpreting processes. J Psychosom Res 2000;49:299–310.
bigger sample sizes and longer follow-ups are also required to 6. Morin CM. Insomnia: psychological assessment and management. New York,
further establish the scale’s sensitivity to treatment-associated NY: Guildford Press, 1993.
changes in pain-related sleep beliefs. Apart from assessing the 7. Perlis ML, Giles DE, Mendelson WB, Bootzin RR, Wyatt JK.
Psychophysiological insomnia: the behavioural model and a neurocognitive
association of changes in PBAS with self-reported improve-
perspective. J Sleep Res 1997;6:179–88.
ment in sleep, it would also be interesting to clarify whether 8. Riemann D, Spiegelhalder K, Feige B, et al. The hyperarousal model
changes in PBAS are associated with changes in objective of insomnia: a review of the concept and its evidence. Sleep Med Rev
sleep and pain outcomes. This would lend support to the hy- 2010;14:19–31.
pothesized role of dysfunctional beliefs about sleep and pain 9. Edinger JD, Means MK. Cognitive-behavioral therapy for primary insomnia.
interaction in the development and maintenance of pain-related Clin Psychol Rev 2005;25:539–58.
insomnia and its contribution to subsequent pain interference. 10. Morin CM, Stone J, Trinkle D, Mercer J, Remsberg S. Dysfunctional beliefs
and attitudes about sleep among older adults with and without insomnia
complaints. Psychol Aging 1993;8:463–67.
11. Morin CM, Vallières A, Ivers H. Dysfunctional Beliefs and Attitudes about
CO N CLUS I O N S Sleep (DBAS): validation of a brief version (DBAS-16). Sleep 2007;30:1547–54.
12. Edinger JD, Wohlgemuth WK, Rodney AR, Marsh GR, Quillian RE. Does
Excessive cognitive arousal is a cardinal feature of both pri- cognitive-behavioral insomnia therapy alter dysfunctional beliefs about sleep?
mary and pain-related insomnia.39 Thinking about the interac- Sleep 2001;24:591–99.
tion between pain and sleep is an integral part of chronic pain 13. Morin CM, Blais F, Savard J. Are changes in beliefs and attitudes about sleep
related to sleep improvements in the treatment of insomnia? Behav Res Ther
patients’ insomnia experience, and the findings of the present 2002;40:741–52.
study suggest that the PBAS is a valid and reliable instru- 14. Ashworth PC, Davidson KM, Espie CA. Cognitive-behavioral factors
ment to detect and assess unhelpful beliefs about the sleep- associated with sleep quality in chronic pain patients. Behav Sleep Med
pain interaction. Theoretically, it elaborates the concept of 2010;8:28–39.
dysfunctional sleep beliefs within the context of chronic pain

Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016 1276

 EF Afolalu, C Moore, F Ramlee et al. Pain-Related Beliefs and Attitudes about Sleep

15. Smith MT, Perlis ML, Carmody TP, Smith MS, Giles DE. Presleep cognitions 31. Field A. Discovering statistics using SPSS. London: SAGE Publications
in patients with insomnia secondary to chronic pain. J Behav Med Ltd, 2009.
2001;24:93–114. 32. Finan PH, Goodin BR, Smith MT. The association of sleep and pain: an update
16. Smith MT, Perlis ML, Smith MS, Giles DE, Carmody TP. Sleep quality and and a path forward. J Pain 2013;14:1539–52.
presleep arousal in chronic pain. J Behav Med 2000;23:1–13. 33. Irwin MR, Olmstead R, Carrillo C, et al. Sleep loss exacerbates fatigue,
17. Tang NKY, Goodchild CE, Hester J, Salkovskis PM. Pain-related insomnia depression, and pain in rheumatoid arthritis. Sleep 2012;35:537–43.
versus primary insomnia. A Comparison study of sleep pattern, psychological 34. Smith MT, Edwards RR, McCann UD, Haythornthwaite JA. The effects of
characteristics, and cognitive-behavioral processes. Clin J Pain sleep deprivation on pain inhibition and spontaneous pain in women. Sleep
2012;28:428–36. 2007;30:494–505.
18. Carney CE, Edinger JD, Manber R, Garson C, Segal ZV. Beliefs about sleep 35. Norell-Clarke A, Jansson-Frojmark M, Tillfors M, Harvey AG, Linton SJ.
in disorders characterized by sleep and mood disturbance. J Psychosom Res Cognitive processes and their association with persistence and remission of
2007;62:179–88. insomnia: findings from a longitudinal study in the general population. Behav
19. Crönlein T, Wagner S, Langguth B, Geisler P, Eichhammer P, Wetter TC. Are Res Ther 2014;54:38–48.
dysfunctional attitudes and beliefs about sleep unique to primary insomnia? 36. Ashworth PCH, Burke BL, McCracken LM. Does a pain management
Sleep Med 2014;15:1463–7. programme help you sleep better? Clin Psychol Forum 2008;184:35–40.
20. Theadom A, Cropley M. Dysfunctional beliefs, stress and sleep disturbance in 37. Clark LA, Watson D. Constructing validity: basic issues in objective scale
fibromyalgia. Sleep Med 2008;9:376–81. development. Psychol Assessment 1995;7:309–19.
21. Tremblay V, Savard J, Ivers H. Predictors of the effect of cognitive behavioral 38. Hinkin TR, Tracey JB, Enz CA. Scale construction: developing reliable and
therapy for chronic insomnia comorbid with breast cancer. J Consult Clin valid measurement instruments. J Hosp Tour Res 1997;21:100–20.
Psychol 2009;77:742–50.
39. Byers HD, Lichstein KL, Thorn BE. Cognitive processes in comorbid poor
22. Morin CM, Gibson D, Wade J. Self-reported sleep and mood disturbance in sleep and chronic pain. J Behav Med 2016;39:233–40.
chronic pain patients. Clin J Pain 1998;14:311–14.
23. Kleinman L, Mannix S, Arnold LM, et al. Assessment of sleep in patients with
fibromyalgia: qualitative development of the fibromyalgia sleep diary. Health ACK N O W L E D G M E N T S
Qual Life Outcomes 2014;12:1–11.
The authors thank Rhian Tait, Philippa Beckwith, and Polly Ashworth for their help
24. Theadom A, Cropley M. ‘This constant being woken up is the worst with data collection.
thing’ – experiences of sleep in fibromyalgia syndrome. Disabil Rehabil
2010;32:1939–47.
25. Tang NKY, Goodchild CE, Salkovskis PM. Hybrid cognitive-behaviour therapy SUBM I SSI O N & CO R R ESPO NDENCE I NFO R M ATI O N
for individuals with insomnia and chronic pain: a pilot randomised controlled
trial. Behav Res Ther 2012;50:814–21. Submitted for publication December, 2015
Submitted in final revised form May, 2016
26. Bastien CH, Vallières A, Morin CM. Validation of the Insomnia Severity Index
Accepted for publication June, 2016
as an outcome measure for insomnia research. Sleep Med 2001;2:297–307.
Address correspondence to: Esther F. Afolalu, Department of Psychology, University
27. Tang NKY, Harvey AG. Correcting distorted perception of sleep in insomnia: a of Warwick, Coventry, CV4 7AL, UK; Tel: +44 (0) 24 765 28294; Email: e.f.afolalu@
novel behavioural experiment? Behav Res Ther 2004;42:27–39. warwick.ac.uk
28. Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain
Inventory. Ann Acad Med Singapore 1994;23:129–38.
29. Tan G, Jensen MP, Thornby JI, Shanti BF. Validation of the Brief Pain Inventory D I SCLO S U R E S TAT E M E N T
for chronic nonmalignant pain. J Pain 2004;5:133–7.
This was not an industry supported study. The authors have indicated no financial
30. Melzack R. The short-form McGill Pain Questionnaire. Pain 1987;30:191–97. conflicts of interest.

1277 Journal of Clinical Sleep Medicine, Vol. 12, No. 9, 2016