Psychometric measurement of PEI in chronic musculoskeletal pain
589
Table I. Description of the background and outcome variables in the study groups
Baseline characteristics
Age, years
Sex, men/women
Body mass index (BMI): normal weight, BMI< 25; overweight, BMI 25 ≥ BMI< 30; obese, BMI ≥ 30.
Country of origin: Sweden; other Scandinavian; non-Scandinavian
Living conditions: living alone; living with others (e.g. with wife/husband, children, parents, or other adults)
Education: compulsory; high school; university/college; other
Working status: (self-)employed/student, part- or full time; Unemployed; Other
Sick leave, No/Yes
Worries about finances (4-point scale): often/quite often; seldom/not at all worried
Expectations for treatment (4-point scale): full recovery/some improvement; some relief/no expectations at all
Probability of working within 6 months (7-point scale), dichotomized into: very large/large vs moderate to very small.
Outcome measures
Disability
Neck Disability Index (NDI) (34). 10 items, each item was scored from 0 (no activity limitations) to 5 (major activity
limitations). Transformed into 0–100%; 0%=no disability.
Functional Rating Index (FRI) (35). 10 items, each item was scored from 0 (no activity limitations) to 4 (major activity
limitations). Transformed into 0–100%; 0%=no disability.
WAD CR MixCP
Mental health
Pain Catastrophizing Scale (PCS) (36). Contains 13 items on 5-point scale (0 (not at all) to 4 (always); higher scores indicate
higher catastrophizing.
Self-Efficacy Scale (SES) (37). Contains 20 items on 0–10 point scale (0=not confident at all, to 10=very confident to perform
different activities) and generates a total score from 0-200.
Coping Strategies Questionnaire (38). The catastrophizing subscale comprises 6 items that describe catastrophic thoughts.
Total score 0-36, higher score signifies higher level of catastrophizing.
Hospital Anxiety and Depression Scale (HADS) [29, 30, 31]. 7 items for anxiety (HAD-A), 7 items for depression (HAD-D).
Subscale scores range from 0 to 21: 0–7: non-case; 8–10 indicates a possible case; 11 or more indicates a definite case.
Health-Related Quality of Life (HRQoL)
European Quality of Life instrument (EQ-5D) [32, 33], consists of 2 parts:
(baseline)
(baseline)
1) The EQ-5D (3-L) contains 5 dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression.
Each dimension has 3 levels: no problems, some problems, and extreme problems. The answers on the 5 dimensions are
converted into a single EQ-5D index ranging from −0.594 to 1, where 1 indicates optimal health.
2) The EQ VAS records the respondent’s self-rated health on a vertical visual analogue scale that ranges from 0 (”worst
possible health state”) to 100 (”best possible health state”).
Self-reported current work ability was measured using the wording,
”Current work ability compared with best,” from the Work Ability Index (WAI) (39, 40). Score: 0 (completely unable to work)
to 10 (best work ability).
WAD: whiplash-associated disorders; CR: cervical radiculopathy; MixCP: mixed chronic pain.
Ethics
The ethics committee at the Faculty of Health Sciences at Lin-
köping University, Linköping in Sweden approved the WAD
study (Dnr 2010/1888-31 and 2011/262-32) (15) and the CR
study (Dnr-M126-08 and M126-08 T99-09) (16). For the MixCP
study the patients received treatment as usual and completed
routine questionnaires for MMR. The questionnaires formed the
basis for the Swedish Quality Registry for Pain rehabilitation
in primary care (SQRP), which today is part of routine care in
39 rehabilitation/primary care clinics in Sweden. SQRP data is
stored with the approval of the National Swedish Data Inspec-
tion Agency (permission number 1580-97). The study followed
the ethical principles of the Declaration of Helsinki and Swedish
law regarding the use of personal information (20, 21). Local
health authorities approved the study design and protocol.
In all studies the participants were free to leave the study wit-
hout explanation with no negative consequences on future tre-
atment. Personal participant details were rendered anonymous
before data-entry. There were no commercial interests connected
to the studies. Informed written consent was obtained from all
participants included in the studies.
Statistical analysis
Power calculation in the WAD (15) and CR (16) studies was
performed for the original RCT studies to detect significant
effects between treatments. In the current study the number
of participants was higher than the proposed minimum requi-
rements of 5 participants per included item to perform factor
analysis (22) and sufficient to perform all comparisons
The Student’s t-test, χ 2 test, and Fisher’s exact test were
used for within-group and between-group comparisons, as
appropriate. Some important psychometric properties of PEI
were investigated.
Content validity was investigated using the proportion of
participants that had missing responses (23), and that gave
“not applicable” responses based on the assumption that they
perceived these items/questions not to be relevant (19). Differ-
ences were analysed with the χ 2 test.
Using MPlus version 8 (MPlus, Muthén & Muthén, Los
Angeles, CA, USA), a confirmatory factor analysis (CFA) was
conducted to test the one-factor measurement model of the
6-item PEI. Because of the Likert-type scale, where response
categories range from “much better/much more” to “no change”
the data were treated as ordered categorical when the model
parameters were estimated in MPlus. The CFA was assessed for
exact fit with the means and variance-adjusted weighted least
squares (WLSMV) χ 2 and approximate fit with standardized
root mean square residual (SRMR) according to the guidelines
set out in 2018 by Asparouhov & Muthén (24). Specifically,
exact fit was concluded if the χ 2 was not significant (p > 0.05).
Otherwise approximate fit was concluded if the χ 2 test rejects the
model (p < 0.05), but SRMR ≤ 0.08 and standardized residuals
were small (|r res |<0.10 (25)).
J Rehabil Med 51, 2019