Journal of Rehabilitation Medicine 51-1CompleteIssue | Page 23

20
M. L. M. Wijenberg et al.
to the left. All analyses were performed using an alpha level
of 0.05.
Baseline differences. Group differences on baseline characteris-
tics between the included and excluded participants were ana-
lysed using Pearson χ 2 tests for nominal variables including all
available levels, independent t-tests (T) for normally distributed
continuous variables, and non-parametric tests (Mann-Whitney
U test) for continuous variables in case of non-normality.
Changes over time. To reveal changes over time, paired samples
t-tests were performed. For further inspection, changes over
time were categorized into 3 categories: “worse”, “equal” and
“better”. Hereby, we used a change of 0.5 SD as pragmatic in-
dicator of change of the psychological variable score over time,
as suggested by Norman et al. (26). For the variables proactive
coping, self-efficacy, extraversion and optimism, an increase
of more than 0.5 SD over time was considered an adaptive
change (“better”) (see Introduction). For the variables passive
coping, neuroticism and pessimism an increase of more than 0.5
SD over time was considered a maladaptive change (“worse”)
(see Introduction). A change of less than 0.5 SD over time was
considered as no change over time (“equal”). Effect sizes were
calculated using Glass’s delta (27).
Correlation and factor analyses. To explore associations
among the psychological variables, Pearson correlations were
calculated and interpreted as suggested by Evans (28) (≤ 0.19
very weak; 0.20–0.39 weak; 0.40–0.59 moderate; 0.60–0.79
strong; ≥ 0.80 very strong). To explore dimensions among the
psychological variables, exploratory factor analysis (EFA)
(including the extraction methods principal axis factoring and
maximum likelihood estimation) was performed using data
from participants who completed all questionnaires at both
time-points (n = 282). Factor extraction was based on multiple
criteria, as suggested by Osborne & Costello (29).The K1 crite-
rion (eigenvalue ≥1.0), the Scree plot, small residual correlations
([–0.1; 0.1]) and a non-significant goodness-of-fit test (based
on maximum likelihood estimation) were used as 4 indicators
for extraction. If the criteria suggested multiple solutions for
factor extraction, e.g. a 1-factor structure on the basis of the
Scree plot and a 2-factor structure on the basis of the K1 crite-
rion, all suggestions of factor extraction were executed and the
best solution, indicated by smaller residual correlations and a
non-significant goodness-of-fit test, was chosen. Only factor
loadings greater than 0.30 were displayed. Because we expected
the factors to be correlated, Oblimin rotation was preferred over
Varimax rotation.
RESULTS
Sample
Of the 395 participants in the Restore4Stroke cohort,
324 (82.0%) provided data on psychological factors at
2 years post-stroke and were included in the current
study. Furthermore, depending on the psychological
factor, imputation was performed for 0.0–9.6% of
participants. Baseline characteristics of both the in-
cluded group (n = 324) and excluded group (n = 71)
and significant differences between them are shown in
Table I. The participants in the included group were sig-
nificantly younger, less often highly educated and more
often involved in a relationship in comparison with
the excluded group. Regarding stroke-related factors,
the included group had a significantly higher level of
cognitive functioning 2 months post-stroke and a sig-
nificantly higher level of independence in ADL 4 days
post-stroke in comparison with the excluded group.
Change of psychological factors over time
Table II provides the results of the temporal stability
analysis of psychological factors. All psychological
factors except passive coping showed significant
changes over time. At 2 years post-stroke, participants
reported less proactive coping, lower self-efficacy,
less extraversion, less optimism, more neuroticism
Table I. Patients’ characteristics at baseline
Demographic factors
Male, %
Age, years, mean (SD)
Marital status: in a relationship, %
Higher educational level, %
Stroke-related factors
Ischaemic, %
Left hemisphere, %
Severity of stroke (NIHSS), mean (SD)
No stroke symptoms (NIHSS 0), %
Minor stroke symptoms (NIHSS 1–4), %
Moderate stroke symptoms (NIHSS 5–12), %
Moderate to severe stroke symptoms (NIHSS 13), %
ADL 4 days post-stroke (BI), mean (SD)
Cognitive functioning 2 months post-stroke (MoCA)
Cognitively impaired (MoCA ≤ 25), %
Destination after discharge from hospital
Home, %
(Geriatric) rehabilitation, %
a
Included group
(n  = 324) Excluded group
(n  = 71) Test statistic
χ 2 /T/U p-value
64.8
65.9 (12.1)
71.3 25.4 a 64.8
70.4 (14.0)
56.3 32.8 b 0.00
2.79
6.05
19.68 1.00
< 0.01
0.01
< 0.01
92.3 97.2
45.1 2.23
8.05 0.33
0.09
3.3 (3.3)
21.1
50.7
25.4
2.8
16.1 (4.9) 9,975.00 0.08
9,575.50
5,072.50 0.02
< 0.05
39.0 a
2.7 (3.2)
24.4
57.1
16.4
2.1
17.0 (4.8)
23.8 (3.7) c
66.3
21.9 (5.4) d
75.6
4.52
72.2
27.8
0.10
62.0
38.0
n  = 323, b n  = 61, c n  = 306, d n  = 41. T: independent t-test; U: Mann-Withney U test; NIHSS: National Institutes of Health Stroke Scale; ADL: activities of daily living;
BI: Barthel Index; MoCA: Montreal Cognitive Assessment; SD: standard deviation.
www.medicaljournals.se/jrm