One of the fastest ways to assess patients' level of cognitive functioning is by using
screening tools, which have a short administration time and can be applied to a broad group
of patients, in contrast to a full neuropsychological evaluation. Nevertheless, since the
screening tools consist of a range of subtests across cognitive domains, they do allow for
making relatively general assumptions for the patient group the investigators examine.
Studies with similar purpose have been performed previously, however, they have applied
cognitive screening tools developed originally to target other populations than stroke, such
as dementias. In contrast, in this study the investigators will use the Oxford Cognitive
Screen (OCS), a tool developed specifically to target stroke patients. The investigators have
chosen to focus on stroke population of working-age adults, since predicting future levels of
cognitive functioning bears a special relevance for this group due to future work life and/or
education, and since even a mild cognitive impairment can have far-reaching consequences.
Setting and recruitment:
This study will include approximately 90 stroke patients recruited from the subacute stroke
unit at Bispebjerg Hospital and the subacute stroke unit at Glostrup Hospital. The patients
will be assessed three times: during their hospitalization, after 3 months and again after 9
months. The project will not impact the rehabilitation of the patients in that the patients
will receive standard rehabilitation care during the course of the study.
A research assistant will look over the list of patients hospitalized at the stroke units
weekly and will recruit patients that comply with the inclusion and exclusion criteria. At
baseline the patients will be tested with OCS either by a research assistant,
neuropsychologist or a therapist working at the specific stroke unit, that has received
training by a neuropsychologist to perform the testing. The patient will also be asked to
fill out Hospital Anxiety and Depression Scale (HADS) and will be asked about the baseline
information that are not deductible from the medical journal (see section about baseline
measures).
The follow up sessions will take place either in the home address of the patient or at
Bispebjerg- or Glostrup Hospital where they will be invited in advance according to their
preferences. The subjects will receive a link and QR-code via e-mail a week in advance to the
follow-up session to fill out questionnaires described in the section "3- and 9-month
measures". At the follow-up sessions they will be interviewed to fill out MDS-HC, tested with
OCS and tested with supplementary cognitive tests by a research assistant. If they have not
filled out the questionnaires in advance, the research assistant will help them to do so
during the test session.
Sample size assessment based on power calculations:
To be able to prove a small effect size (d = 0.15) in the linear regressions the
investigators need to include at least 55 patients.
To be able to prove a small effect size (d = 0.15) in the Repeated Measures ANOVA the
investigators need to include at least 73 patients.
Because the Correlation analysis is not used to obtain results about the main objective,
which is to predict future cognitive level, a possible larger effect size (d = 0.3) is
accepted. To prove a minimum of d=0.3, the investigators need at least 82 patients in the
final analysis. The investigators choose therefore to include 90 patients, assuming possible
dropout.
Plan for missing data:
Missing/Unavailable/Non-reported/ Uninterpretable or considered missing because of data
inconsistency: The investigators will primarily use Maximum Likelihood Estimation to make
imputations when data is missing.
Out-of-range results: When the investigators identify an outlier, it will be investigated
whether there might be an obvious cause that is unrelated to what the investigators are
trying to measure. If this is the case, the datapoint will be ruled out, otherwise it will be
part of the final analysis and would be considered as natural variation.
Statistical analysis plan describing the analytical principles and statistical techniques to
be employed: The investigators plan to perform 6 separate multiple linear regressions.
Prognostic variables will be chosen for each multiple linear regression based on their
correlation with the specific outcome variable. Significance would be considered achieved at
p < .10.
The possible prognostic variables will be the same in all 6 analyses and are as follows:
Age, sex, years of education, type of stroke, location of stroke, NIHSS-score, additional
treatment, HADS-scores at baseline, OCS-score at baseline, pre-stroke working situation,
housing situation, cohabitant status, alcohol intake, and physical exercise level. Initially,
the investigators will perform a correlation analysis between the possible predictors and
outcome. Those that correlate significantly with outcome will be used for regression
analysis.
The investigators will perform a separate analysis for each outcome variable as follows:
MDS-HC performance-score after 3 months, MDS-HC difficulty-score after 3 months, MDS-HC
performance-score after 9 months, MDS-HC difficulty-score after 9 months, scores of the
supplementary cognitive tests after 3 months and scores of the supplementary cognitive tests
after 9 months.
Repeated measures ANOVA The investigators plan to perform a repeated measures ANOVA to test
if the OCS-score changes over time. This analysis will be corrected for influence of
depression and anxiety symptoms and fatigue by correcting for HADS-scores and MFI-20-score.
If this test proves significant, pairwise comparisons will be done (LSD-t-test). Significance
would be considered achieved at p < .05.
Correlational analyses The investigators plan to perform analyses of correlations between
MDS-HC- and OCS-scores. One will be done for the variables at 3 months, and one will be done
for the variables at 9 months.
The investigators also plan to perform correlational analyses of the OCS-scores and the
scores on the supplementary cognitive tests. One will be done for the variables at 3 months,
and one will be done for the variables at 9 months. Significance would be considered achieved
at p < .05.
To be able to explain results of the latter correlational analyses between OCS-scores and the
scores on supplementary tests the investigators want to execute McNemar's test and Kappa
measure of agreement. This is to further theorize how the results of the separate subtests in
OCS relates to the subtests of the supplementary testing. These analyses will be done for
each OCS-subtest matched to the corresponding subtest of the supplementary test battery.
Significance in McNemar's test would be considered achieved at p < .05.
