Content uploaded by Cheryl Ritenbaugh
Author content
All content in this area was uploaded by Cheryl Ritenbaugh on Jan 29, 2018
Content may be subject to copyright.
Development and Validation of the EXPECT Questionnaire:
Assessing Patient Expectations of Outcomes
of Complementary and Alternative Medicine
Treatments for Chronic Pain
Salene M. W. Jones, PhD,
1
Jane Lange, PhD,
2
Judith Turner, PhD,
3
Dan Cherkin, PhD,
1
Cheryl Ritenbaugh, PhD, MPH,
4
Clarissa Hsu, PhD,
1
Heidi Berthoud,
1
and Karen Sherman, PhD
1
Abstract
Background: Patient expectations may be associated with outcomes of complementary and alternative medicine
(CAM) treatments for chronic pain. However, a psychometrically sound measure of such expectations is needed.
Objectives: The purpose of this study was to develop and evaluate a questionnaire to assess individuals’
expectations regarding outcomes of CAM treatments for chronic low back pain (CLBP), as well as a short form
of the questionnaire.
Methods: An 18-item draft questionnaire was developed through literature review, cognitive interviews with
individuals with CLBP, CAM practitioners, and expert consultation. Two samples completed the questionnaire:
(1) a community sample (n=141) completed it via an online survey before or soon after starting a CAM
treatment for CLBP, and (2) participants (n=181) in randomized clinical trials evaluating CAM treatments for
CLBP or fibromyalgia completed it prior to or shortly after starting treatment. Factor structure, internal con-
sistency, test–retest reliability, and criterion validity were examined.
Results: Based on factor analyses, 10 items reflecting expectations (used to create a total score) and three
items reflecting hopes (not scored) were selected for the questionnaire. The questionnaire had high internal
consistency, moderate test–retest reliability, and moderate correlations with other measures of expectations. A
three-item short form also had adequate reliability and validity.
Conclusions: The Expectations for Complementary and Alternative Medicine Treatments (EXPECT) ques-
tionnaire can be usedin research to assess individuals’ expectations oftreatments for chronic pain. It is recommended
that the three hope questions are included (but not scored) to help respondents distinguish between hopes and
expectations. The short form may be appropriate for clinical settings and when expectation measurement is not a
primary focus.
Keywords: back pain, yoga, acupuncture, complementary and alternative medicine, massage, chiropractic
Introduction
Greater knowledge about the effects of individuals’
expectations regarding outcomes from complementary
and alternative medicine (CAM) treatments for chronic pain
is needed. Progress in this area has been hampered by the lack
of valid and reliable measures, possibly explaining the
inconsistent research.
1–6
Most measures of individuals’ ex-
pectations regarding CAM treatments were not developed
systematically and were developed for one specific treat-
ment.
7,8
No questionnaires have been developed to assess
individuals’ expectations across CAM therapies for chronic
1
Group Health Research Institute, Seattle, WA.
2
Fred Hutchinson Cancer Research Center, Seattle, WA.
3
University of Washington, Seattle, WA.
4
University of Arizona, Tucson, AZ.
THE JOURNAL OF ALTERNATIVE AND COMPLEMENTARY MEDICINE
Volume 22, Number 11, 2016, pp. 936–946
ªMary Ann Liebert, Inc.
DOI: 10.1089/acm.2016.0242
936
pain. Many studies used measures developed to assess treat-
ment credibility.
7,8
However, an individual’s views regarding
the credibility of a treatment may differ from their expecta-
tions about the likely effects of the treatment on their pain.
Prior work also supports the importance of distinguishing
expectations from other perceptions, such as credibility and
hope.
9–11
Hopes reflect individuals’ views of the best-case
scenario, whereas expectations are their appraisals of the most
likely outcomes, and asking about hope may affect how re-
spondents interpret questions about expectations.
9–11
To ad-
dress these gaps in the literature, a measure of expectations for
different CAM treatments for chronic low back pain (CLBP)
was developed and evaluated. In recognition of the frequent
need in research studies to minimize respondent burden, a
short form of this questionnaire was also developed.
Methods
Questionnaire development: overview
The draft Expectations for CAM Treatments (EXPECT)
questionnaire items were developed through an iterative pro-
cess, including a review of the literature and interviews with
individuals with CLBP and CAM practitioners. Questions were
drafted to capture hopes and expectations for CAM treatment
effects on three domains identified in previous work
9,10
(pain
relief, improved function, and improved mood and overall
well-being). Questions were drafted about specific domains
(sleep, mood) and how much change individuals hoped for and
realistically expected in their back pain, and the impact ofback
pain on their life in one year. The draft questions were evalu-
ated using cognitive interviews with individuals with CLBP
11
and were reviewed by seven CAM and pain experts. Using this
input, the items were revised, and they were evaluated in ad-
ditional cognitive interviews with individuals with CLBP.
The draft questionnaire included 18 items regarding indi-
viduals’ hopes and expectations about outcomes of CAM care
for their pain. Each item had a 0–10 response scale, and some
had an option for selecting ‘‘not applicable.’’ Anchors dif-
fered across items, but 0 always represented no change or
worse and 10 represented maximal improvement (complete
relief or no impact of back pain on the domain). All study
procedures and materials were approved by the Group Health
Institutional Review Board, and all participants provided in-
formed consent.
Development of EXPECT questionnaire
using community volunteers
Participants, setting, and procedures. Between July 2013
and November 2014, individuals were recruited who were
planning to begin or who had recently started acupuncture,
chiropractic, massage, or yoga for CLBP. They were recruited
from several sources, including practitioners, online adver-
tisements, print advertisements, and panelists assembled by a
survey research organization. Advertisements directed par-
ticipants to the study Web site, which contained a description
of the study and screening questions. Eligibility requirements
included age between 20 and 70 years old; the ability to read
English; Internet access; back pain for at least 3 months; back
pain intensity in the previous week rated ‡3 on a 0–10 scale;
plan to begin acupuncture, chiropractic, massage, or yoga for
CLBP in the next month or started of one of those therapies in
the past week; and no use of those therapies in the previous 3
years (except for the prior week). Eligible individuals then
completed the draft questionnaire and other questionnaires.
Participants received $20 for completed questionnaires. One
day later, participants who had completed questionnaires
prior to their first treatment were invited to complete the
questionnaire again if they still had not started treatment. One
day later, they were e-mailed a link to the online questionnaire
and asked to complete it again within 7 days. Participants
received $10 for completing the second questionnaire.
Measures. To assess criterion validity, the six-item
treatment Credibility/Expectancy Questionnaire (CEQ),
7
the
four-item Acupuncture Expectancy Scale,
8
and a single
question asking how helpful the participant believed the
treatment would be for their CLBP were administered.
4,8,12–17
The CEQ includes two subscales (treatment credibility and
treatment expectations). The CEQ has demonstrated construct
validity and relationships with clinical outcomes.
4
The four-
item Acupuncture Expectancy Scale, which assesses patients’
expectations for the efficacy of acupuncture, has demonstrated
validity and reliability.
8
It was adapted by replacing ‘‘illness’’
with ‘‘pain’’ and, for non-acupuncture treatment, by replacing
‘‘acupuncture’’ with the treatment the participant was begin-
ning. The single CLBP treatment helpfulness question
12–17
asks respondents to rate their expectations of a treatment being
effective for their backpain on a 0 (no change) to 10 (complete
change) scale.
Evaluation of EXPECT questionnaire
in randomized controlled trials
Participants, setting, and procedures. After the draft
questionnaire was developed, it was evaluated in three ran-
domized controlled trials (RCTs) of CAM therapies for chronic
pain: Back to Health (BTH),
18
Mind–Body Approaches to Pain
(MAP),
19
and Pain Outcomes Comparing Yoga versus Struc-
tured Exercise (POYSE; http://clinicaltrials.gov/ct2/show/
NCT01797263). The BTH trial randomized participants with
CLBP to yoga, physical therapy, and self-care education.
Questionnaire data, completed before starting treatment, were
obtained from 72 people randomly assigned to either physical
therapy or yoga. The MAP trial
19
randomized participants with
CLBP to group cognitive–behavioral therapy or mindfulness-
based stress reduction. Participants (n=72) were invited to
complete the draft questionnaire within 1 week of their first
treatment session, and were paid $20 for completing the
questionnaire. The POYSE trial randomized veterans with fi-
bromyalgia to yoga or structured exercise classes. Participants
(n=100) completed the paper version of the draft questionnaire
after randomization and before treatment began. In all three
trials, participants also provided demographic data and pain
intensity ratings. Participants in BTH and MAP completed the
23-item modified RolandModified Disability Questionnaire,
20
and POYSE participants completed the Fibromyalgia Impact
Questionnaire Revised.
21
Participants were required to have
had pain for 12 weeks and to rate their pain as ‡3 on a 0–10
scale.
Statistical analyses
Initial psychometric analyses: community sample. The
psychometric properties of the draft questionnaire were
EXPECTATIONS AND CAM TREATMENT FOR PAIN 937
examined in the community sample to develop the final
questionnaire. Items with >5% ‘‘not applicable’’ responses
were excluded. Next, an exploratory factor analysis (EFA)
was conducted to determine the number of factors and to
select items for the final questionnaire. The factor solutions
were obtained via a minimum residual approach using R
v3.2.0
22
and the psych package.
23
The two-factor solution
was obtained using Promax rotation, allowing factors to
correlate. One- and two-factor solutions were compared
based on the Bayesian information criterion (BIC),
24
pro-
portion of total variance explained, factor loadings, and fit
statistics. Models with a lower BIC are favored. The fit
statistics included those that do not assume multivariate
normality (standardized root mean square of the residual
[SRMSR]; <0.08 suggests a good fit)
25
and other statistics
that assume a multivariate normal structure. These included
the chi-square test of perfect fit (non-significant p-values
>0.05 are desirable), root mean square error of approxima-
tion (RMSEA; <0.08 suggests a good fit), and the Tucker–
Lewis index (TLI; >0.95 suggests a good fit).
26
Table 1. Community Sample and Randomized Clinical Trial Sample Characteristics
Community sample Clinical trials
Characteristic
All
(n=163)
Retest subsample
(n=71)
BTH
(n=72)
MAP
(n=72)
POYSE
(n=100)
Age
20–30 years, % (n) 27.0 (44) 22.5 (16) 15.5 (11) 13.9 (10) 2.0 (2)
31–40 years, % (n) 20.9 (34) 21.1 (15) 21.1 (15) 9.7 (7) 7.0 (7)
41–50 years, % (n) 20.2 (33) 25.4 (18) 25.4 (18) 18.1 (13) 18.0 (18)
51–60 years, % (n) 22.7 (37) 22.5 (16) 33.8 (24) 29.2 (21) 34.0 (34)
61–70 years, % (n) 9.2 (15) 8.5 (6) 4.2 (3) 29.2 (21) 33.0 (33)
‡71 years, % (n) 0 0 0 0 6.0 (6)
Female, % (n) 57.7 (94) 52.1 (37) 62.5 (45) 62.5 (45) 31.0 (31)
Race
Asian/Pacific islander, % (n) 1.9 (3) 1.4 (1) 4.2 (3) 4.2 (3) 0
Black, % (n) 9.9 (16) 16.9 (12) 55.6 (40) 1.4 (1) 35.7 (35)
More than one, % (n) 6.8 (11) 7.0 (5) 6.9 (5) 0 7.1 (7)
Other, % (n) 2.5 (4) 4.2 (3) 9.7 (7) 1.4 (1) 3.1 (3)
White, % (n) 78.9 (127) 70.4 (50) 23.6 (17) 93.0 (66) 54.1 (53)
Hispanic, % (n) 9.3 (15) 7.0 (5) 12.5 (9) 2.8 (2) 4.2 (4)
PHQ-2 (0–6), M(SD) 2.85 (2.14) 2.53 (2.02)
Average pain intensity past week
(0–10 scale), M(SD)
6.72 (1.75) 6.68 (1.73) 4.9 (2.3) 5.8 (1.4) 6.4 (1.9)
Days with back pain, last 3 months, M(SD) 75.05 (16.76) 74.18 (16.1)
Receiving disability compensation
for pain, % (n)
8.8 (14) 11.6 (8)
RMDQ, M(SD) 16.47 (5.58) 17.21 (5.36) 15.3 (5.3) 11.1 (4.8)
FIQR, M(SD) 56.5 (16.6)
Ever had a back injection, % (n) 30.1 (49) 32.4 (23)
Ever had back surgery, % (n) 12.9 (21) 8.5 (6)
Pain below the knee in last month, % (n) 44.8 (73) 40.8 (29)
Use of prescription medicine for
back pain in last week, % (n)
74.8 (122) 80.3 (57)
Treatment initiating
Acupuncture, % (n) 35.6 (58) 42.3 (30)
Chiropractic, % (n) 24.5 (40) 19.7 (14)
Massage, % (n) 20.2 (33) 18.3 (13)
Yoga, % (n) 19.6 (32) 19.7 (14) 50.0 (36) 0 48.0 (48)
MBSR, % (n) 0 61.1 (44) 0
CBT, % (n) 0 38.9 (28) 0
PT, % (n) 50.0 (36) 0 0
SEP, % (n) 0 0 52.0 (52)
Credibility/Expectancy Questionnaire (7)
Credibility subscale (range 3–27), M(SD) 19.52 (5.13) 19.27 (5.57)
Expectancy subscale (range 3–27), M(SD) 11.13 (4.27) 10.65 (4.37)
Total score (range 6–54), M(SD) 30.64 (8.78) 29.92 (9.29)
Acupuncture Expectancy Scale
(range 4–20, (8)), M(SD)
12.53 (4.15) 12.13 (4.05)
Helpfulness of treatment (range 0–10), M(SD) 6.72 (2.45) 6.56 (2.63)
BTH, Back to Health trial; CBT, cognitive–behavioral therapy; FIQR, Fibromyalgia Impact Questionnaire Revised; MAP, Mind–Body
Approaches to Pain trial; MBSR, mindfulness-based stress reduction; PHQ, Patient Health Questionnaire; POYSE, Pain Outcomes
Comparing Yoga versus Structure Exercise; PT, physical therapy; RMDQ, Roland Morris Disability Questionnaire; SEP, Structured
Exercise Program; SD, standard deviation.
938 JONES ET AL.
As detailed in the Results, a two-factor model was selected,
and then items for the final questionnaire were selected based
on factor loadings. Items were retained that loaded >0.50 on
the primary factor and that loaded <0.30 on the secondary
factor. The final questionnaire was scored as the mean of the
answered expectations questions. Scores could range from
0 to 10, with higher scores indicating more favorable ex-
pectations. For the final questionnaire, floor and ceiling ef-
fects (defined as >15% having the highest or lowest possible
score
27
), Cronbach’s alpha to assess internal consistency
(‡0.70 is considered adequate
28
), and corrected item–total
correlations were evaluated. Test retest reliability of the
questionnaire was assessed in the test–retest subgroup of the
community sample using the intra-class correlation coefficient
(ICC), with ‡0.70 considered adequate.
29
Criterion validity of
the questionnaire was assessed by calculating Pearson corre-
lations with the other measures of expectations.
Psychometric evaluation in clinical trials samples. Although
the RCT samples completed all 18 draft questions, only the
final 10-item questionnaire was scored. Confirmatory factor
analyses (CFA) was conducted to assess the factor structure of
the final questionnaire in the RCT samples using the lavaan
package.
30
The CFA was fit via maximum likelihood. Fit was
examined using the chi-square test of perfect fit, RMSEA,
TLI, and SRMSR. To assess fit without the assumption of
multivariate normality, robust test statistics using sandwich-
based covariance estimates were computed in addition to the
standard minimum function test statistics. Floor and ceiling
effects, item–total correlations, and Cronbach’s alpha were
also examined.
Short-form item selection. From the final 10-item ques-
tionnaire and the questions about hopes from the draft ques-
tionnaire (see Results and Appendix A), three items were
selected that were summary or general measures of expecta-
tions, rather than expectations specific to one outcome domain
(e.g., mood,sleep), based on the rationale that these items were
likely to be relevant for all participants and could provide ad-
equate coverage of the construct of expectations with reduced
items. One hope item was also selected to be administered but
not scored. The three selected items were ‘‘how much change
do you realistically expect’’ in: (1) your back pain, (2) the
impact of your backpain on your life, and (3) the impact of your
back pain on your daily activities. The short form was evalu-
ated by examining its internal consistency (all four samples),
test–retest validity (community sample only), correlation with
the long form (all four samples), and correlations with the
criterion measures (community sample only).
Results
Table 1 provides descriptive statistics for all samples.
Across the samples, diversity in age, race/ethnicity, and sex
was represented, although few adults >70 years of age were
included by design.
Initial psychometric analyses in community sample
Figure 1 shows the study flow in the community sample
(n=163, n=71 for test–retest subsample). Table 2 shows, for
each item, the number of missing/not applicable responses,
the mean (SD) rating, corrected item–total correlations, and
factor loadings in the one- and two-factor solutions in the
EFA. The corrected item–total correlations were acceptable
(0.60–0.88).
31
Before factor analyses, one question (about
impact of back pain on family/friends) was excluded because
9.8% (n=10) of the sample reported that the question was not
applicable. All other items had <5% not applicable or missing
responses. This left complete data from 141 participants for
the EFA. The EFA factor loadings indicated that the two-
factor solution consisted of one factor conceptually related to
expectations (items 2, 4–8, and 10–13) and the other factor
conceptually related to hopes (items 1, 3, 9, 15, and 17). Two
items (16 and 18) concerning expectations of outcomes in 1
year loaded on both factors.
The two-factor solution had better fit statistics and ex-
plained more variance compared with the one-factor solu-
tion (Table 3). In addition to eliminating the question about
impact of back pain on family/friends, four other questions
FIG. 1. Study flow in the commu-
nity sample that completed the online
draft questionnaire.
EXPECTATIONS AND CAM TREATMENT FOR PAIN 939
were eliminated, resulting in a 10-item final questionnaire.
Items 16 and 18 were eliminated because they loaded on
both factors. Items 15 and 17 were eliminated because al-
though they loaded on the hope factor, they measured hope
for a time frame that differed from that for the other three
hope questions. Based on these results, 10 items about ex-
pectations were chosen for the final questionnaire to be
scored, and three questions about hopes (1, 3, and 9) were
chosen to be administered but not scored.
Reliability and validity statistics for the final 10-item
EXPECT questionnaire (scored from the 18 draft question-
naire items administered) are shown in Table 4. Cronbach’s
Table 3. Fit Indexes for the Exploratory and Confirmatory Factor Analyses in the Community
Sample and the Clinical Trials Samples
Exploratory analyses:
community sample (17 items)
Confirmatory analysis:
clinical trials, one-factor solution (10 items)
One-factor
solution
(n=141)
Two-factor
solution
(n=141)
BTH yoga
study
(n=55)
MAP study
(n=41)
POYSE study
(n=85)
Total % variance
explained
0.62 0.71 Total % variance
explained
0.61 0.65 0.74
Chi-square 926.7 527.8 SRMSR 0.07 0.05 0.05
Chi-square df 119 103 ML chi-square 124.9 57.0 141.2
p-Value <0.001 <0.001 Chi-square df 35 35 35
RMSEA (90% CI) 0.23
(0.21–0.23)
0.18
(0.16–0.19)
ML chi-square
p-value
<0.001 0.01 <0.001
RMSR 0.10 0.05 ML RMSEA
(90% CI)
0.22 (0.18–0.26) 0.12 (0.06–0.18) 0.19 (0.16–0.22)
TLI 0.65 0.79 ML TLI 0.75 0.92 0.86
BIC 337.8 18.1 ML CFI 0.81 0.94 0.89
Robust chi-square 54.36 49.90 80.18
Robust chi-square
p-value
0.02 0.049 <0.001
Robust RMSEA
(90% CI)
0.10 (0.06–0.13) 0.10 (0.02–0.16) 0.12 (0.10–0.15)
Robust TLI 0.93 0.87 0.92
Robust CFI 0.94 0.90 0.94
BIC, Bayesian information criterion; CFI, comparative fit index; CI, confidence interval; df, degrees of freedom; ML, maximum
likelihood; RMSEA, root mean square error of approximation; SRMSR, standardized root mean square residual; TLI, Tucker–Lewis index.
Table 2. Draft Questionnaire Performance Characteristics and Factor Loadings
in One- and Two-Factor Exploratory Factor Analyses, Community Sample (N=141)
Question
number
NA/missing,
%(n)M(SD)
Corrected item–total
correlation
One factor Two factor
Loadings 1 Loading 1 Loading 2
Q1 0 (0) 7.4 (2.2) 0.60 0.56 0.08 0.57
Q2 0 (0) 5.8 (2.2) 0.75 0.77 0.61 0.18
Q3 0 (0) 7.6 (2.4) 0.74 0.72 -0.04 0.91
Q4 0 (0) 6.0 (2.3) 0.80 0.83 0.58 0.29
Q5 3.7 (6) 6.5 (2.6) 0.80 0.83 0.88 -0.04
Q6 3.1 (5) 6.5 (2.5) 0.79 0.82 0.92 -0.09
Q7 2.5 (4) 6.2 (2.6) 0.79 0.83 0.94 -0.11
Q8 0 (0) 6.0 (2.3) 0.78 0.81 0.77 0.07
Q9 2.5 (4) 7.2 (2.5) 0.72 0.70 0.04 0.80
Q10 2.5 (4) 5.9 (2.4) 0.85 0.87 0.73 0.17
Q11 2.5 (4) 6.0 (2.6) 0.86 0.89 0.98 -0.08
Q12 2.5 (4) 6.1 (2.5) 0.83 0.86 0.94 -0.06
Q13 0.6 (1) 6.2 (2.4) 0.88 0.90 0.98 -0.06
Q14 9.8 (16) 6.1 (2.5) 0.78
Q15 0 (0) 7.6 (2.5) 0.78 0.76 -0.05 0.98
Q16 0 (0) 6.0 (2.5) 0.77 0.78 0.39 0.46
Q17 0 (0) 7.5 (2.6) 0.65 0.64 -0.26 1.08
Q18 0 (0) 6.0 (2.5) 0.79 0.79 0.34 0.53
Note: Q14 was not included in the factor analyses.
NA, not applicable.
940 JONES ET AL.
alpha (internal consistency) was very high (0.96). Floor
(<2% had a score of 0) and ceiling (3% had a score of 10)
effects were low in the community sample. In the commu-
nity sample subgroup that completed the expectations
questions on two separate occasions, the test–retest ICC was
0.75 (95% confidence interval [CI] 0.63–0.84), consistent
with expectations being a moderately stable construct over a
short time period before treatment. The Pearson correlations
of the 10-item EXPECT questionnaire with other measures
of expectations ranged between 0.54 and 0.65 (Table 4).
Psychometric evaluation in clinical trials samples
Tables 3 and 5 display psychometric statistics, factor
loadings, and fit statistics for the 10 expectation items of the
final EXPECT questionnaire completed in the three clinical
trial samples. CFA were limited to individuals in the clinical
trials with non-missing values (55/72 for the BTH trial, 41/72
for MAP, and 85/100 for POYSE). In all three trials, the
maximum likelihood fit statistics demonstrated some lack of
fit. In contrast, the robust fit statistics indicated good fit.
Using the combined data from the three RCT samples,
Cronbach’s alpha was 0.96, indicating high internal consis-
tency. Corrected item–total correlations across the three trials
ranged from 0.61 to 0.90, further indicating sufficient reli-
ability. Floor and ceiling effects were minimal; no individuals
had a score of 0 and only 7 (2.9%) had a score of 10.
Short-form evaluation
In the four samples combined, the three-item short form
showed adequate internal consistency (Cronbach’s a=0.89;
95% CI 0.81–0.96). In the subsample that completed the
questionnaire twice, test–retest stability was 0.79 (95% CI
0.69–0.89). Pearson correlations between the short form and
the parent questionnaire ranged from 0.90 to 0.95 across the
four samples. In the community sample, the three-item short
form correlated moderately with the CEQ total score (0.64),
credibility subscale (0.54), expectancy subscale (0.67),
Acupuncture Expectancy Scale (0.51), and the single ex-
pectations item (0.56).
Discussion
The EXPECT questionnaire assesses individuals’ expec-
tations of outcomes of CAM treatments for chronic pain.
The questionnaire showed good reliability in community
and clinical trials samples of individuals with chronic pain.
The factor analyses supported the conceptualization of
hopes and expectations as two distinct, but related, con-
structs. The questionnaire was moderately correlated with
other treatment expectation measures likely due to differ-
ences across measures in aspects of expectations assessed by
the measures or due to one measure consisting of a single
item and the restricted range of another. The short-form
version also showed good reliability and validity, and was
highly correlated with the parent questionnaire.
The EXPECT questionnaire includes 10 items on ex-
pectations for various outcomes relevant for adults with
CLBP. It is recommended that three questions about hopes
for treatment outcomes also be administered with the ex-
pectation items but not scored. The intent of these questions
is to increase respondents’ focus on answering the 10 ex-
pectations questions in terms of their realistic expectations
as opposed to their hopes for treatment outcomes. Data to
Table 5. Final 10-Item Expectations Questionnaire: Factor Loadings for Clinical Trials Samples
Question
number
BTH trial MAP trial POYSE trial
Factor
loadings
Corrected
item–total
correlation
Factor
loadings
Corrected
item–total
correlation
Factor
loadings
Corrected
item–total
correlation
Q2 0.76 0.72 0.72 0.71 0.78 0.77
Q4 0.84 0.8 0.86 0.84 0.82 0.80
Q5 0.85 0.85 0.59 0.56 0.77 0.76
Q6 0.66 0.64 0.76 0.72 0.85 0.84
Q7 0.79 0.78 0.82 0.82 0.86 0.85
Q8 0.82 0.77 0.54 0.56 0.87 0.86
Q10 0.84 0.79 0.91 0.89 0.9 0.88
Q11 0.64 0.61 0.89 0.85 0.9 0.87
Q12 0.81 0.79 0.94 0.90 0.91 0.90
Q13 0.79 0.76 0.91 0.88 0.88 0.86
Table 4. Final 10-Item EXPECT Questionnaire:
Reliability and Validity Statistics
Community
sample,
n=163
Clinical
trials,
n=244
Cronbach’s alpha 0.96 0.96
Test–retest reliability, ICC (95% CI) 0.75
(0.63–0.84)
Floor effect (score of 0), n(%) 2 (1.2) 0
Ceiling effect (score of 10), n(%) 3 (1.8) 7 (2.9)
Pearson correlations of the
Expectations Questionnaire
with other expectations measures
Credibility/Expectancy Questionnaire
Treatment credibility subscale 0.54
Treatment expectations subscale 0.65
Total score 0.65
Acupuncture Expectancy Scale 0.54
Treatment helpfulness rating 0.60
ICC, intraclass correlation.
EXPECTATIONS AND CAM TREATMENT FOR PAIN 941
support the use of these three hope questions for studying
hopes about treatment outcomes are not available. For the
short-form questionnaire, the administration of the first hope
item (‘‘How much change do you hope for in your back
pain?’’), followed by the three short-form expectation items,
is recommended, giving a total of four items administered.
As with the parent questionnaire, only the expectation items
in the short form are included in the scoring.
The EXPECT questionnaire has advantages compared with
previous measures of treatment expectations. The Acu-
puncture Expectancy Questionnaire was found to have floor
and ceiling effects
8
; the EXPECT questionnaire did not have
floor or ceiling effects. Unlike CAM-treatment specific ex-
pectations measures,
8
this questionnaire was developed for
use with diverse CAM treatments. In contrast to measures
that assess treatment credibility
32
(i.e., a treatment’s potential
ability to improve pain in general), the EXPECT question-
naire asks about expectations in terms of the effects of the
treatment on the respondent’s own pain outcomes. Finally,
the EXPECT questionnaire asks about multiple outcomes
related to chronic pain, whereas other measures (e.g., mood
and sleep) assess expectations for only a few outcomes.
8
This study had a number of limitations. In the EFA, many of
the fit statistics based on multivariate normality exceeded
cutoff values. However, one of the main purposes of the factor
analysis was to compare one- and two-factor models to deter-
mine whether hope and expectations are different rather than to
compare models to cutoffs or to test the multivariate normality.
The CFA, despite small sample sizes, showed adequate fit with
the robust test statistics. The samples included few individuals
‡70 years of age. The study did not evaluate empirically
whether including the hope items affected responses to the
expectation questions. Finally, the full or short-form ques-
tionnaires, once finalized, were not evaluated in new samples.
The strengths of the study include development of the
questionnaire using a sequential process of qualitative in-
terviews with practitioners and individuals with chronic
pain, cognitive interviews, and psychometric evaluation.
Other strengths are questionnaire testing in samples diverse
in sociodemographic characteristics and CAM therapies,
and confirmatory testing in samples of individuals beginning
clinical trials. The POYSE sample consisted of adults with
fibromyalgia but showed results consistent with those in the
CLBP samples, supporting the use of the questionnaire in
diverse pain populations. A strength of the short-form de-
velopment was selection of items based on item content
rather than empirical selection. Selection of items for a short
form solely through empirical methods can lead to omission
of important aspects of the construct.
33
Conclusions
The EXPECT questionnaire can be used in research to
assess individuals’ expectations of treatments for chronic
pain. Several directions for future research are indicated.
Further research is needed to assess the psychometric
characteristics of the EXPECT questionnaire and short form
in samples with different sociodemographic and clinical
characteristics. Examination of the association of EXPECT
scores with outcomes after CAM treatments may help in-
crease knowledge about the role of individuals’ expectations
in their outcomes after CAM treatments. Finally, although
the questionnaire can be used with individuals beginning
CAM treatments for CLBP, the questionnaire might be
adapted for use with individuals with other pain problems
and for use with non-CAM treatments.
Acknowledgments
This study was funded by the National Center for Com-
plementary and Integrative Healthcare (grant number R01
AT005809).
Author Disclosure Statement
The authors do not have any conflicts of interest to declare.
References
1. Thomas E, Croft PR, Paterson SM, et al. What influences
participants’ treatment preference and can it influence
outcome? Results from a primary care-based randomised
trial for shoulder pain. Br J Gen Pract 2004;54:93–96.
2. Thomas KJ, MacPherson H, Thorpe L, et al. Randomised
controlled trial of a short course of traditional acupuncture
compared with usual care for persistent non-specific low
back pain. BMJ 2006;333:623.
3. Galer BS, Schwartz L, Turner JA. Do patient and physician
expectations predict response to pain-relieving procedures?
Clin J Pain 1997;13:348–351.
4. Smeets RJ, Beelen S, Goossens ME, et al. Treatment ex-
pectancy and credibility are associated with the outcome of
both physical and cognitive–behavioral treatment in
chronic low back pain. Clin J Pain 2008;24:305–315.
5. Goldstein MS, Morgenstern H, Hurwitz EL, et al. The
impact of treatment confidence on pain and related dis-
ability among patients with low-back pain: results from the
University of California, Los Angeles, low-back pain study.
Spine J 2002;2:391–399; discussion 399–401.
6. Kalauokalani D, Cherkin DC, Sherman KJ, et al. Lessons
from a trial of acupuncture and massage for low back pain:
patient expectations and treatment effects. Spine (Phila Pa
1976) 2001;26:1418–1424.
7. Devilly GJ, Borkovec TD. Psychometric properties of the
credibility/expectancy questionnaire. J Behav Ther Exp
Psychiatry 2000;31:73–86.
8. Mao JJ, Armstrong K, Farrar JT, et al. Acupuncture ex-
pectancy scale: development and preliminary validation in
China. Explore (NY) 2007;3:372–377.
9. Schafer LM, Hsu C, Eaves ER, et al. Complementary and
alternative medicine (CAM) providers’ views of chronic low
back pain patients’ expectations of CAM therapies: a qual-
itative study. BMC Complement Altern Med 2012;12:234.
10. Hsu C, Sherman KJ, Eaves ER, et al. New perspectives on
patient expectations of treatment outcomes: results from
qualitative interviews with patients seeking complementary
and alternative medicine treatments for chronic low back
pain. BMC Complement Altern Med 2014;14:276.
11. Sherman KJ, Eaves ER, Ritenbaugh C, et al. Cognitive
interviews guide design of a new CAM patient expectations
questionnaire. BMC Complement Altern Med 2014;14:39.
12. Cherkin DC, Eisenberg D, Sherman KJ, et al. Randomized
trial comparing traditional Chinese medical acupuncture,
therapeutic massage, and self-care education for chronic
low back pain. Arch Intern Med 2001;161:1081–1088.
13. Sherman KJ, Cherkin DC, Erro J, et al. Comparing yoga,
exercise, and a self-care book for chronic low back pain: a
942 JONES ET AL.
randomized, controlled trial. Ann Intern Med 2005;143:
849–856.
14. Sherman KJ, Cherkin DC, Ichikawa L, et al. Treatment
expectations and preferences as predictors of outcome of
acupuncture for chronic back pain. Spine (Phila Pa 1976)
2010;35:1471–1477.
15. Beckmann AM, Kiviat NB, Daling JR, et al. Human pap-
illomavirus type 16 in multifocal neoplasia of the female
genital tract. Int J Gynecol Pathol 1988;7:39–47.
16. Cherkin DC, Sherman KJ, Kahn J, et al. A comparison of
the effects of 2 types of massage and usual care on chronic
low back pain: a randomized, controlled trial. Ann Intern
Med 2011;155:1–9.
17. Sherman KJ, Cherkin DC, Wellman RD, et al. A random-
ized trial comparing yoga, stretching, and a self-care book
for chronic low back pain. Arch Intern Med 2011;171:
2019–2026.
18. Saper RB, Sherman KJ, Delitto A, et al. Yoga vs. physical
therapy vs. education for chronic low back pain in pre-
dominantly minority populations: study protocol for a
randomized controlled trial. Trials 2014;15:67.
19. Cherkin DC, Sherman KJ, Balderson BH, et al. Comparison
of complementary and alternative medicine with conven-
tional mind–body therapies for chronic back pain: protocol
for the Mind–body Approaches to Pain (MAP) randomized
controlled trial. Trials 2014;15:211.
20. Roland M, Fairbank J. The Roland–Morris Disability
Questionnaire and the Oswestry Disability Questionnaire.
Spine 2000;25:3115–3124.
21. Bennett RM, Friend R, Jones KD, et al. The Revised Fi-
bromyalgia Impact Questionnaire (FIQR): validation and
psychometric properties. Arthritis Res Ther 2009;11:R120.
22. R Core Team. R: A Language and Environment for Statistical
Computing. Vienna, Austria: R Foundation for Statistical
Computing, 2014.
23. Revelle W. pysch: Procedures for Personality and Psy-
chological Research. 1.5.4 ed. Evanston, IL: Northwestern
University, 2015.
24. Schwarz G. Estimating the dimension of a model. Ann Stat
1978;6:461–464.
25. Browne M, Cudeck R. Alternative Ways of Assessing
Model Fit. London: Sage, 1993.
26. Hu LT, Bentler PM. Cutoff criteria for fit indexes in co-
variance structure analysis: conventional criteria versus
new alternatives. Struct Equ Modeling 1999;6:1–55.
27. Terwee CB, Bot SD, de Boer MR, et al. Quality criteria
were proposed for measurement properties of health status
questionnaires. J Clin Epidemiol 2007;60:34–42.
28. Nunnally JC, Bernstein IH. Psychometric Theory. 3rd ed.
New York: McGraw-Hill, 1994.
29. Fayers PM, Machin D. Quality of Life. Assessment, Ana-
lysis and Interpretation. Chichester, England: Wiley, 2000.
30. Yves R. Iavaan: an R package for strucutral equation
modeling. J Stat Softw 2012;48:1–36.
31. Everitt BS. The Cambridge Dictionary of Statistics. 2nd ed.
New York: Cambridge University Press, 2002.
32. Prady SL, Burch J, Vanderbloemen L, et al. Measuring ex-
pectations of benefit from treatment in acupuncture trials: a
systematic review. Complement Ther Med 2015;23:185–199.
33. Smith GT, McCarthy DM, Anderson KG. On the sins of
short-form development. Psychol Assess 2000;12:102–111.
Address correspondence to:
Salene M.W. Jones, PhD
Group Health Research Institute
Seattle, WA 98101
E-mail: salenewu@gmail.com
I am now going to ask you a series of questions about the effects that [INSERT TREATMENT MODALITY] may
have on your back pain and on how back pain impacts your life. In each case, the question is asking about the
results at the end of the treatment period.
Back Pain Questions:
1. How much change do you hope for in your back pain? Please answer on a scale of 0 to 10, where 0 is ‘‘no
change or worse’’ and 10 is ‘‘complete relief.’’
012345678910
No Change/Worse Complete Relief
2. How much change do you realistically expect in your back pain? Please answer on a scale of 0 to 10, where 0 is
‘‘no change or worse’’ and 10 is ‘‘complete relief.’’
012345678910
No Change/Worse Complete Relief
Appendix A: Expectations for Complementary and Integrative
Treatments (EXPECT) Questionnaire
EXPECTATIONS AND CAM TREATMENT FOR PAIN 943
Impact of back pain on life questions:
3. How much change do you hope for in the impact of back pain on your life? Please answer on a scale of 0 to 10,
where 0 is ‘‘no change or worse’’ and 10 is ‘‘pain no longer impacts my life.’’
012345678910
No Change/Worse Back Pain No Longer Impacts My Life
4. How much change do you realistically expect in the impact of back pain on your life? Please answer on a scale
of 0 to 10, where 0 is ‘‘no change or worse’’, and 10 is ‘‘pain no longer impacts my life.’’
012345678910
No Change/Worse Back Pain No Longer Impacts My Life
Sleep/Mood/Energy:
The next set of questions asks about areas of your life such as sleep, mood, and energy. If any of these questions are
not relevant for you because back pain does not impact that area of your life, please answer Not Applicable.
5. How much change do you realistically expect in your back-related sleep problems? Please answer on a scale
of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘back pain no longer affects my sleep.’’ If back pain
does not impact your sleep, please choose ‘‘not applicable.’’
012345678910
No Change/Worse Back Pain No Longer Affects My Sleep
,Not Applicable
6. How much change do you realistically expect in your mood or irritability? Please answer on a scale of 0 to 10,
where 0 is ‘‘no change or worse’’ and 10 is ‘‘back pain no longer affects my mood.’’ Or you may choose
‘‘not applicable.’’
012345678910
No Change/Worse Back Pain No Longer Affects My Mood or
Irritability
,Not Applicable
7. How much change do you realistically expect in your energy? Please answer on a scale of 0 to 10, where 0 is
‘‘no change or worse’’ and 10 is ‘‘back pain no longer affects my energy.’’ Or you may choose
‘‘not applicable.’’
012345678910
No Change/Worse Back Pain No Longer Affects My Energy
,Not Applicable
Coping Question:
The next question is about your expectations about coping with back pain.
8. How much improvement in your ability to cope with back pain do you realistically expect as a result of [INSERT
TREATMENT MODALITY]? Please answer on a scale of 0 to 10, where 0 is ‘‘no improvement’’
and 10 is ‘‘extreme improvement.’’
012345678910
No Improvement Extreme Improvement
944 JONES ET AL.
Limitations due to back pain questions:
The following questions are about effects that [INSERT TREATMENT MODALITY] may have on your physical
limitations due to back pain. In each case, the question is asking about the results at the end of the treatment period.
If these questions are not relevant for you because you do not have any physical limitations due to back pain,
please choose ‘‘not applicable.’’
9. How much change do you hope you will have in your back pain–related physical limitations? Please answer on
a scale of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘limitations completely resolved.’’ Or you may
choose ‘‘not applicable.’’
012345678910
No Change/Worse Limitations Completely Resolved
,Not Applicable
10. How much change do you realistically expect in your back pain-related physical limitations? Please answer on a
scale of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘limitations completely resolved.’’ Or you may
choose ‘‘not applicable.’’
012345678910
No Change/Worse Limitations Completely Resolved
,Not Applicable
Impact of back pain on specific areas of life questions:
The next questions to ask about the effects that [INSERT TREATMENT MODALITY] may have on the impact of
back pain on specific areas of your life. In each case, the question is asking about the results at the end of the
treatment period. If any of these questions are not relevant for you because back pain does not impact that area
of your life, please choose ‘‘not applicable.’’
11. How much change do you realistically expect in the impact of back pain on your work, including housework?Please
answer on a scale of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘back pain no longer impacts my work,’’
including housework. Or you may choose ‘‘not applicable’’ if back pain does not impact your work/housework now.
012345678910
No Change/Worse Back Pain No Longer Impacts my Work
,Not Applicable
12. How much change do you realistically expect in the impact of back pain on your social and recreational
activities? Please answer on a scale of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘back pain
no longer impacts my social and recreational activities.’’ Or you may choose ‘‘not applicable’’ if back
pain does not impact your social and recreational activities now.
012345678910
No Change/Worse Back Pain No Longer Impacts my Social
and Recreational Activities
,Not Applicable
13. How much change do you realistically expect in the impact of back pain on your daily activities? Please answer
on a scale of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘back pain no longer impacts my
daily activities.’’ Or you may choose ‘‘not applicable’’ if back pain does not impact your daily activities now.
012345678910
No Change/Worse Back Pain No Longer Impacts My Daily Activities
,Not Applicable
EXPECTATIONS AND CAM TREATMENT FOR PAIN 945
Appendix B: Short Form of the EXPECT Questionnaire
I am now going to ask you a series of questions about the effects that [INSERT TREATMENT MODALITY] may
have on your back pain and how back pain impacts your life. In each case, the question is asking about the results
at the end of the treatment period.
1. How much change do you hope for in your back pain? Please answer on a scale of 0 to 10, where 0 is ‘‘no change
or worse’’ and 10 is ‘‘complete relief.’’
012345678910
No Change/Worse Complete Relief
2. How much change do you realistically expect in your back pain? Please answer on a scale of 0 to 10, where 0 is
‘‘no change or worse’’ and 10 is ‘‘complete relief.’’
012345678910
No Change/Worse Complete Relief
3. How much change do you realistically expect in the impact of back pain on your life? Please answer on a scale
of 0 to 10, where 0 is ‘‘no change or worse’’, and 10 is ‘‘pain no longer impacts my life.’’
012345678910
No Change/Worse Back Pain No Longer Impacts My Life
4. How much change do you realistically expect in the impact of back pain on your daily activities? Please answer
on a scale of 0 to 10, where 0 is ‘‘no change or worse’’ and 10 is ‘‘back pain no longer impacts my daily activities.’’
Or you may choose ‘‘not applicable’’ if back pain does not impact your daily activities now.
012345678910
No Change /Worse Back Pain No Longer Impacts My Daily Activities
,Not Applicable
946 JONES ET AL.