Research Report

Addiction Eur Addict Res Received: April 4, 2018

Research DOI: 10.1159/000492160 Accepted: July 16, 2018
Published online: August 28, 2018

Neurocognitive Functioning in Alcohol

Use Disorder: Cognitive Test Results
Do not Tell the Whole Story
Steffen Moritz a Schaimaa Irshaid a Thies Lüdtke a, b Ingo Schäfer a
Marit Hauschildt a Michael Lipp a
a Department of Psychiatry and Psychotherapy, University Medical Center Hamburg-Eppendorf, Hamburg,
Germany; b Department of Psychology, UiT – The Arctic University of Norway, Tromsø, Norway

Keywords viduals underwent a comprehensive neuropsychological

Alcohol use disorder · Neuropsychological test battery. The neurocognitive assessment was flanked by
functioning · Motivation the Momentary Influences, Attitudes and Motivation Impact
on Cognitive Performance Scale (MIAMI), which captures
momentary influences affecting performance. Results: Pa-
Abstract tients with AUD performed worse than nonclinical controls
Objectives: It is textbook knowledge that individuals with on most test parameters. Group differences achieved a very
alcohol use disorder (AUD) show large neurocognitive defi- large effect size for parameters tapping speed and accuracy.
cits. However, these patients display a number of addition- Patients with AUD showed deviant scores, particularly on
al impairments (e.g., lack of drive and motivation) that may the post version of the MIAMI (retrospective assessment of
contribute to poor test results. The impact of these second- symptoms and influences during testing) and the total
ary mediators has not been explored systematically. Based scores. For accuracy, the MIAMI scores represented a partial
on prior findings that low performance motivation, a nega- mediator. For speed, significant group effect sizes were ren-
tive attitude toward cognitive assessment, and momentary dered nonsignificant when the MIAMI was taken into ac-
symptoms compromise neuropsychological test results in count. Conclusion: Like other psychiatric patients, patients
depression, schizophrenia, and obsessive-compulsive disor- with AUD show marked neurocognitive impairments that
der, we examined the possibility that impaired test results
in AUD partially represent an epiphenomenon. Methods:
Fifteen patients with AUD and 20 matched nonclinical indi- S.M. and S.I. contributed equally to this work.
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

© 2018 S. Karger AG, Basel Dr. Steffen Moritz

Department of Psychiatry and Psychotherapy
University Medical Center Hamburg-Eppendorf
E-Mail karger@karger.com
Martinistrasse 52, DE–20246 Hamburg (Germany)
www.karger.com/ear
Downloaded by:

E-Mail moritz @ uke.de

seem to be aggravated by, for example, distraction and lack ever, as the authors detected no significant interaction for
of effort. This tentatively suggests that performance only the 2 groups and the 2 experimental conditions, they re-
partly reflects cortical impairments in areas hosting neuro- frained from concluding that neuropsychological deficits
cognitive faculties. Contextual factors deserve greater at- are caused by motivational limitations. A study on 1,306
tention in patients with addiction. The cross-sectional de- individuals from the general population [25] indicated
sign of our study limits conclusions relating to causality. that neurocognitive deficits were not correlated with the
© 2018 S. Karger AG, Basel amount of alcohol consumption but with the desire/crav-
ing for alcohol [17], which suggests the possibility that
craving distracts individuals and lowers their perfor-
Introduction mance.
To achieve our purpose, we compared patients with
Meta-analyses suggest neuropsychological impair- AUD with healthy controls on a large neurocognitive bat-
ment in alcohol-dependent patients across many do- tery. Assessment was flanked by the Momentary Influ-
mains [1, 2], including executive functioning [3]. Cogni- ences, Attitudes and Motivation Impact on Cognitive
tive deficits in severe alcoholism are usually attributed to Performance Scale (MIAMI) [19, 21, 23], which allowed
brain damage in the frontal and limbic regions [4, 5], par- us to test whether momentary influences such as poor
ticularly due to thiamine deficiency/malnutrition and the motivation and current symptoms compromise neuro-
consequences of liver damage [6–8]. Yet, a direct neuro- psychological test results. We hypothesized that individ-
toxic effect of alcohol consumption for neurocognitive uals with AUD would show worse results than controls
deficits in humans is not fully established, and there is an but that a substantial part of the variance could be ex-
ongoing debate regarding whether mild to moderate al- plained by contextual factors.
cohol consumption is harmful [9] or even neuroprotec-
tive [10, 11]. Age may also play a role; mild to moderate
alcohol consumption has been implicated in slower cog- Methods
nitive decline in older individuals [12], while moderate to
heavy drinking may decrease neuropsychological perfor- Fifteen inpatients who met the criteria for AUD (moderate or
mance in adolescents across time [13]. Longitudinal stud- severe) according to the DSM-5 were recruited from the addiction
ward of the Department of Psychiatry and Psychotherapy of the
ies suggest that with long-term abstinence, neurocogni- University Medical Center Hamburg-Eppendorf (Germany). Di-
tive results improve and may at times achieve (almost) agnoses were verified by the physician in charge. Overall symptom
premorbid levels [1, 14–16]. severity was assessed with the Brief Psychiatric Rating Scale [26].
The present study examined whether at least some of One patient was medicated with an antidepressant and one with
the neurocognitive deficits seen in alcohol-dependent pa- an antipsychotic drug; one patient was receiving an anti-craving
substance. Bipolar disorder, schizophrenia psychosis, or any other
tients stem from poor motivation [4] and other secondary primary psychiatric diagnoses represented exclusion criteria. The
influences. To illustrate, if the individual is not motivated performance of the patients was compared to 20 individuals with-
to perform a cognitive task or if he or she is distracted by out any psychiatric disorder (nonclinical controls). Most of these
craving [17] or rumination, tests results may not mirror participants had served as clinical controls in prior studies on OCD
the full potential of an individual and thus may underes- [19], depression [19], or schizophrenia [19]. Blind to results, the
initial control group was iteratively reduced to match patients on
timate true cognitive functioning (secondary malperfor- background characteristics.
mance). For a number of disorders, such as obsessive- General exclusion criteria were major neurological disorders
compulsive disorder (OCD) [18, 19], depression [20, 21], (e.g., multiple sclerosis) and age below 18 or above 65 years. None
and schizophrenia [22, 23], it has been already deter- of the non-clinical participants fulfilled criteria for psychiatric
mined that impairments in test performance in part rep- disorders, as confirmed through the Mini International Neuro-
psychiatric Interview [27]. The study was approved by the ethics
resent epiphenomena of other psychological influences. committee of the German Psychological Association (DGPS, EK
Impairments in these groups were sometimes even elim- 122016). All participants provided written informed consent pri-
inated when these influences were taken into account. or to participating in the trial. The samples are characterized in
Few studies have addressed this possibility in alcohol use Table 1.
disorder (AUD). Scheurich et al. [24] found that motiva-
Neurocognitive Tasks
tional (goal-setting) instructions could significantly en- Trail-Making Test (TMT) [28]
hance cognitive performance in patients with AUD, The TMT Part A (adult version) assesses psychomotor speed;
whereas no such increment was found in controls. How- the individual has to connect numbers in ascending order as
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

2 Eur Addict Res Moritz/Irshaid/Lüdtke/Schäfer/

DOI: 10.1159/000492160 Hauschildt/Lipp
Downloaded by:
Table 1. Differences between individuals with AUD and healthy individuals on demographic characteristics and scores on the MIAMI
subscales

Variables AUD (n = 15) Healthy (n = 20) Statistics

n % n % χ2 df p value

Background characteristics
Gender, female/male 4/11 27/73 10/10 50/50 1.94 1 0.163 –

M SD M SD |t|/χ2 df p value |d|

Age, years 52.33 11.11 47.55 10.27 1.32 33 0.197 0.45

Years of formal education 9.54 4.45 11.15 1.57 1.26 13.94 0.229 0.51
Brief Psychiatric Rating Scale total 48.13 12.77 – – – – – –
MIAMI pre
Low motivation 2.03 0.38 1.94 0.35 0.69 33 0.493 0.25
Concerns about assessment 2.16 0.42 2.15 0.48 0.04 33 0.972 0.02
Fear of a poor outcome 2.02 0.64 1.78 0.69 1.05 33 0.301 0.36
Unfavorable momentary influences 2.17 0.58 1.65 0.43 3.04 33 0.005 1.43
Total score 2.09 0.30 1.88 0.30 2.08 33 0.045 0.70
MIAMI post
Low motivation 1.75 0.45 1.39 0.41 2.48 33 0.019 0.84
Concerns about assessment 1.83 0.50 1.55 0.44 1.73 33 0.094 0.60
Fear of a poor outcome 2.27 0.57 1.75 0.53 2.78 33 0.009 0.95
Unfavorable momentary influences 2.11 0.38 1.61 0.29 4.49 33 <0.001 1.51
Total score 1.99 0.35 1.57 0.25 4.09 33 <0.001 1.42
Total score (pre and post) 2.04 0.28 1.73 0.23 3.67 33 0.001 1.23

MIAMI, Momentary Influences, Attitudes and Motivation Impact on Cognitive Performance Scale (higher scores designate greater
impairment).

quickly as possible. Part B assesses set-shifting ability; the partici- as fast as possible. Norm values derived from a large sample of par-
pant has to alternate between numbers and letters in ascending ticipants were applied for median reaction times and number of
order (i.e., 1-A-2-B, etc.). Age-adapted norm scores were applied omissions [32].
[29].
Selective Attention Subtest from the Test for
Wisconsin Card Sorting Test [30] Attentional Performance [32]
The Wisconsin Card Sorting Test [31] was used in a com- For this selective attention test, participants were consecutively
puterized version to assess executive functioning. Partici- shown either a “+” (distractor) or “×” (target). If the target stimu-
pants were presented a maximum of 128 cards with colored (i.e., lus appeared, the participant had to press a button as fast as pos-
red, blue, yellow, or green) symbols (i.e., stars, rectangles, tri- sible. Norm scores for median reaction times and errors from a
angles, or crosses) in different frequencies (i.e., 1–4). Cards had large German population were applied [32].
to be matched according to 3 principles that–unbeknownst to
the participant–changed during the course of the assessment Block Design from the Wechsler Adult Intelligence
(i.e., color, symbols/shape, number of items). A tone (high vs. Scale–4th Edition [33, 34]
low) and a corresponding verbal cue indicated whether a selec- Individuals were asked to match colored cubes to a two-dimen-
tion was correct or incorrect. The dependent variables were the sional pattern as quickly as possible. Scaled scores from a large
number of categories completed (i.e., 0–6) and perseverative er- German population were applied [34].
rors.
Auditory Verbal Learning Test (AVLT) [35]
Divided Attention Subtest from the Test for We measured learning capacity and retention with the Ger-
Attentional Performance [32] man version of the AVLT [36]. A list of 15 words was read to the
For this test, participants needed to perform 2 tasks concur- participants 5 times. At the end of each trial, the individuals had
rently. Whenever asterisks formed a rectangle on a 4 × 4 dot matrix to repeat as many words as they could recall. Then, a new (inter-
(optical target) or 2 tones of the same frequency (high or low) were ference) list was read and had to be remembered. Afterward,
heard (acoustical target), the participant had to press a space bar items from the first list had to be recalled, immediately afterwards
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

Neurocognitive Functioning in AUD: Eur Addict Res 3

Cognitive Test Results DOI: 10.1159/000492160
Downloaded by:
and after a delay of 30 min. Learning was indexed by the sum of Assessment of Subjective Influences on Cognitive Performance
correctly recalled items from trials 1–5. Long-term memory was The neuropsychological assessment was flanked by the 2 parts
indexed by the number of correctly reproduced items after 30 of the Momentary Influences, Attitudes and Motivation Impact on
min. To avoid interference, no other verbal memory test was per- Cognitive Performance Scale (MIAMI), with each part consisting
formed during the retention interval. Normative scores for the of 18 items. The pre version was handed out before the evaluation
German version of the task are available for different age groups and the post version immediately afterwards. The post version asks
[36]. participants to provide a retrospective assessment of motivational
and contextual factors during the assessment. Each version assess-
Selective Attention Test d2 [37] es the following 4 domains: (1) poor motivation (e.g., “I am not
Test d2 is a letter cancellation test that captures selection atten- motivated at all to take part in the assessment” [pre]; “I was not
tion. The participants had to cross out a target item (letter d with motivated and therefore did not achieve my best performance”
2 small lines) while ignoring distractor items. Participants had to [post]); (2) concerns about the assessment (e.g., “I am experiencing
complete as many items as possible in a fixed time. Normative the situation as very unpleasant and would like to leave” [pre]; “I
scores for the parameter concentration performance (German: felt pressured by the test situation” [post]); (3) fear of a poor out-
Konzentrationsleistung) served as the outcome measure. Age-ad- come (e.g., “I worry that the tasks will be too difficult for me” [pre];
justed normative scores from a large population sample were ap- “I am fearful about the results” [post]); and (4) negative momentary
plied [38]. influences (e.g., “Right now, I feel very tired and exhausted” [pre];
“I had a headache during the assessment” [post]). Items have to be
Wechsler Memory Scale–Revised Story Recall [39] endorsed on a 4-point Likert scale (1 = completely agree, 2 = rather
Participants had to repeat as much information as possible agree, 3 = rather disagree, 4 = completely disagree). Subscale scores
from 2 short stories. Recall was requested immediately after ad- were aligned to facilitate interpretation; higher scores designate
ministration and after 30 min. To avoid interference (see also more negative influences and attitudes. Cronbach’s α of the MIAMI
above for the AVLT), no other verbal memory test was presented is 0.75. The baseline version of the MIAMI was modestly correlated
during the retention interval. Scoring followed instructions in the (r = 0.42) with an expert rating scale in patients with OCD.
manual; German norm values were applied [40].
Statistical Analyses
Similarities from the Wechsler Adult Intelligence Group comparisons were analyzed using independent t tests
Scale–4th Edition [33, 34] for single and aggregated (speed and accuracy) neuropsychologi-
In this verbal reasoning task, the individual was asked to de- cal parameters. Then, 2 mediation analyses were run using Hayes’
duce the superordinate category of 2 items. Scaled scores from a process procedure (model 4; 5,000 bootstrap samples to correct for
large German population were applied [34]. potential biases of non-normality in the sample) with group (0 =
healthy, 1 = AUD) as the independent variable (x), neuropsycho-
Matrix Reasoning from the Wechsler Adult Intelligence logical impairment (i.e., aggregated scores for speed and accuracy)
Scale–4th Edition [33, 34] as the dependent variable (y), and the MIAMI total score (pre and
The individual was presented with a pattern sequence and had post scores pooled) as the mediator (M). The null hypothesis can
to select the item that completed the sequence from 5 alternatives be rejected when the range of the 95% CI (LL = lower limit to UL =
(nonverbal reasoning). Scaled scores were applied from a large upper limit) does not include zero. We also report the Sobel test
German population. for the indirect effect. In addition to regular assumptions of linear
models, mediation analyses rely on the assumption that indepen-
Zoo Test of the Behavioral Assessment of the dent variable, mediator, and outcome are causally related. How-
Dysexecutive Syndrome [41] ever, it is debated whether this causality has to be unequivocally
Planning, a core executive function, was measured with the established empirically by experimental and longitudinal designs
Zoo Test. The participants had to plan a route encompassing 6 out ([43], p.89). Hayes ([43], p.89) claims that there is “no problem” to
of 12 possible locations in a fictive zoo by adhering to several cri- conduct mediation analyses even if the design does not allow clear
teria. The test consists of 2 subtests. First, a demanding situation causal inferences, as long as theoretical considerations support
with little external structure was presented. The individual had to causality. Therefore, it should be mentioned that although theory
plan in advance the order in which they would visit designated lo- supports a presumed causal relationship, our results should be in-
cations in a zoo. Then, a concrete, externally imposed strategy had terpreted cautiously.
to be followed to reach locations in the zoo. In a second step, we calculated analyses of covariance by enter-
ing the MIAMI total score (pre and post assessments pooled) as
Aggregated Scores covariate. This was done to determine whether the magnitude of
From the neuropsychological parameters, we composed 2 new the group effect would be reduced. The latter analyses were sub-
indices that aggregated either speed or accuracy parameters. Com- sidiary to the mediation analyses to determine the magnitude of
posite scores are well accepted in neuropsychology and help to the effect of the mediator/covariate on speed and performance.
condense information from different domains (e.g., IQ tests; 37); Effect sizes are expressed as Cohen’s d (the formula was adjusted
they are useful to make studies with different tests tapping similar for unequal sample sizes), with d = 0.2 representing a small effect
functions more comparable [42]. For the present analysis, we cal- size, d = 0.5 a medium effect size, and d = 0.8 a large effect size. We
culated z-scores for each parameter and calculated the mean z- confirmed t tests on group differences as well as parametric cor-
value. Higher scores designate worse performance. Hence, some relations using non-parametric statistics (U-test; Spearman’s rho)
scores had to be reversed (e.g., memory scores). and found compatible results.
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

4 Eur Addict Res Moritz/Irshaid/Lüdtke/Schäfer/

DOI: 10.1159/000492160 Hauschildt/Lipp
Downloaded by:
 Results see above for the uncorrected average effect size, d = 1.10).
When the MIAMI total score was accounted for, 9 of
The 2 samples did not differ on any demographic char- 18 comparisons remained significant; only 2 yielded very
acteristics (Table 1). Patients differed significantly from large effect sizes (d ≥1.2).
controls on the MIAMI total scores for pre, post, and the
mean of pre and post. Regarding subscales, samples dif-
fered on “unfavorable momentary influences” on pre, Discussion
while at post all group differences were significant except
for “concerns about assessment”. In line with a plethora of studies in the field, our results
Patients performed worse on 14 out of 18 cognitive test confirm large neurocognitive test impairments in pa-
parameters relative to controls (Table 2); all of these dif- tients with AUD. In the current study, despite low power
ferences were large (d ≥0.8), and 6 were even very large (d to detect significant differences owing to rather small
≥1.2). On average, between-group effect sizes were large sample sizes, group differences emerged for 14 of 18 pa-
(Md = 1.10). Group differences for the aggregated scores rameters. All of these differences achieved a large effect
achieved very large effect sizes, particularly for accuracy. size. As expected, patients also performed worse on ag-
When looking at norm scores (Table 2), 47% of patients gregated scores for speed and accuracy at large effect siz-
showed impairment (median) compared to only 10% in es. However, mean differences from controls should not
healthy individuals (mean: 46 vs. 12%). overshadow the fact that not all patients showed deficits;
Significant associations were found between aggregat- diagnostically relevant deficits (1 SD below the mean)
ed neuropsychological and MIAMI scores in 4 out of 6 were displayed by approximately half of the sample per
correlations rating from a medium to large effect size (Ta- test (median of all norm scores).
ble 3). As shown in other population samples [18–23], pa-
Figure 1 shows that the relationship between group tients displayed more concerns about the assessment,
and neurocognition was mediated by MIAMI scores, as more fear of a poor outcome, and more unfavorable mo-
evidenced by a significant Sobel test; in addition, CIs of mentary influences than the controls, as indexed by the
the indirect effects did not cross zero (5,000 bootstrap MIAMI questionnaire. Seven of the comparisons were
samples). For speed, the direct effect became nonsignifi- significant. When the pooled total score was entered as a
cant when MIAMI scores were considered; for perfor- covariate in the group comparisons, neurocognitive im-
mance, the direct effect was reduced but remained sig- pairment remained significant in 9 of the 18 comparisons.
nificant. This was corroborated by mediation analyses. The MI-
We reran the analyses by removing the subscale Unfa- AMI scores significantly mediated the relationship be-
vorable Momentary Influences from the total score. This tween group membership and outcome (accuracy, speed).
was done as this subscale may be regarded as a subjective For speed, the direct effect was not significant once the
reflection of true neurocognitive impairments. The vari- MIAMI was taken into account. For accuracy, the direct
able may be considered an index of the outcome rather effect was reduced but remained significant indicating
than a mediator and might inflate the mediation effects. that MIAMI scores acted as a partial mediator. When we
However, results remained essentially unchanged when repeated the analyses using the MIAMI pre score only,
the subscale was removed (the lower and upper limit did the significant indirect effect remained.
not cross zero, suggesting a significant indirect effect). While a number of studies already show that patients
Following the suggestion of a reviewer we reran the anal- with AUD show poor motivation and impairment of oth-
yses for the pre score of the MIAMI as a mediator only. er factors relevant to volition and performance, to the best
Results were less strong than for the main analysis but the of our knowledge no study has looked systematically at
indirect effects were also significant: again, CIs of the in- the impact of these variables on performance. Unlike rou-
direct effects did not cross zero (speed: 0.0044–0.0960; tine tests in medicine (e.g., blood samples), we can only
accuracy: 0.0003–0.0519). derive meaningful inferences from neuropsychological
In line with this, the effect sizes of group differences test results if participants perform to the best of their po-
for single scores were significantly decreased (paired t tential and have fair and similar conditions for the assess-
test, t(17) = 9.56, p < 0.001) when MIAMI scores were ac- ment.
counted for (for analyses of covariances using the MIAMI Our study does not question the results of prior studies
baseline score, the average effect size dropped to d = 0.69; indicating neuropsychological impairment in patients
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

Neurocognitive Functioning in AUD: Eur Addict Res 5

Cognitive Test Results DOI: 10.1159/000492160
Downloaded by:
 6
Table 2. Group differences between individuals with AUD and nonclinical controls

Variables/domains AUD (n = 15) Healthy (n = 20) 1 SD ≤ Statistics ANCOVA (MIAMI)

norm, %
M SD norm M SD norm |t|/χ2 df p value |d| p value |d|

Speed
Trail-making test A, s 44.00 18.04 31.33P 28.95 8.98 53.95P 47/10 3.00 19.20 0.008 1.11 0.149 0.52
Divided attention RT auditory, ms 726.40 292.71 39.47T 618.00 90.65 41.60T 60/40 1.38 16.03 0.185 0.53 0.469 0.26
Divided attention RT visual, ms 944.53 193.03 43.00T 808.25 116.14 51.10T 40/15 2.69 33 0.011 0.89 0.531 0.22
Go/no go RT, ms 484.07 103.02 42.60T 427.85 99.95 49.00T 53/25 1.65 33 0.108 0.56 0.579 0.20
Attention
Go/no go errors 2.33 2.32 44.20T 1.32 2.14 49.32T 40/19 1.33 33 0.194 0.46 0.211 0.46

Eur Addict Res

d2 concentration index 108.69 38.87 25.62P 150.10 38.19 59.55P 54/5 3.02 31 0.005 1.08 0.042 0.78
Divided attention omissions 4.33 4.81 40.53T 1.70 1.78 47.50T 33/25 2.02 16.90 0.059 0.77 0.673 0.16

DOI: 10.1159/000492160
Memory
AVLT learning 43.13 12.28 45.40T 56.95 9.10 57.00T 33/0 3.83 33 0.001 1.31 0.035 0.78
AVLT retention 8.73 3.88 44.13T 12.10 2.90 53.45T 32/10 2.94 33 0.006 1.01 0.106 0.59
Logical memory immediate 22.93 8.63 34.80P 30.05 5.62 62.85P 33/0 2.95 33 0.006 1.01 0.224 0.44
Logical memory delayed 17.93 10.30 34.13P 27.00 5.76 64.90P 40/10 3.07 20.50 0.006 1.13 0.134 0.54
Reasoning
Similarities 19.87 8.55 7.67S 26.70 2.72 10.15S 47/5 2.98 16.13 0.009 1.15 0.029 0.81
Matrix reasoning 13.47 6.02 8.60S 20.45 2.91 12.85S 47/0 4.14 18.91 0.001 1.55 0.010 0.97
Spatial Performance
Block design 30.87 12.45 7.07S 50.85 9.10 12.20S 60/0 5.25 24.55 <0.001 1.88 <0.001 1.51
Executive functioning
Trail-making test B, s 170.53 98.82 19.93P 61.45 15.41 52.80P 73/20 4.27 14.51 0.001 1.67 0.002 1.18
WCST categories 2.43 2.10 – 5.00 1.75 – – 3.88 33 <0.001 1.35 0.001 1.27
WCST perseveration 24.79 13.41 – 13.27 7.98 – – 3.14 32 0.004 1.08 0.034 0.79
Zoo test total 12.43 4.07 – 15.90 0.45 – – 3.18 13.22 0.007 1.30 0.022 0.87
Aggregated scores (z-transformed)
Speed (slowing) 0.48 0.88 – −0.36 0.45 – – 3.39 19.52 0.001 1.26 0.076 0.65
Accuracy (malperformance) 0.55 0.62 – −0.39 0.31 – – 5.37 19.29 <0.001 2.01 <0.001 1.39

T T-scores (M = 50), P percentile (M = 50), S scaled score (M = 10).

Hauschildt/Lipp
Moritz/Irshaid/Lüdtke/Schäfer/
Downloaded by:
Karolinska Institutet, University Library
130.237.122.245 - 8/29/2018 1:19:10 PM
Fig. 1. Results from the mediation analyses
using Hayes’ process macro. The indirect MIAMI MIAMI
effects for speed (a) and accuracy (b) were
significant, indicating that momentary in-
fluences, concerns, and fears as well as mo- a = 0.052 b = 1.265
tivation (Momentary Influences, Attitudes (0.014)**** (0.412)*** a = 0.052 b = 0.821
(0.014)**** (0.300)**
and Motivation Impact on Cognitive Per-
formance Scale [MIAMI] total for pre and
post) largely contributed to group differ- c = 0.140 (0.038)**** c = 0.156 (0.267)****
ences (0 = healthy, 1 = alcohol use disorder) Group Speed Group Accuracy
on test results. For speed, the direct effect c’ = 0.073 (0.040)+ c’ = 0.113 (0.029)****
was reduced to a statistical trend. + p < 0.1; (0.020–0.133; sobel p = 0.021) (0.017–0.128; sobel p = 0.031)
* p < 0.05; ** p < 0.01; *** p < 0.005; **** p < a b
0.001.

with addiction. Yet, we question the validity of bold Table 3. Correlations between MIAMI scores and aggregated neu-
claims that monocausally attribute these deficits to brain ropsychological parameters for speed (slowness) and accuracy
(malperformance)
regions hosting neurocognitive faculties.
We found a very large group difference for memory. Slowness Malperformance
However, the patients as a group achieved better scores
than might be expected from this result (T-score > 40; Total score pre 0.594* 0.410
percentile > 30), and no more than one third showed Total score post 0.314 0.569*
malperformance on one of the 4 parameters as defined Total score pre and post 0.529* 0.591*
by 1 standard deviation below the mean. This result is * p < 0.05.
partly owing to above-normal performance in controls,
so we advise researchers to calculate norm scores in their
samples; some control groups might not be representa-
tive because they include, for example, students or vol- of potential metacognitive deficits in patients with psy-
unteers from data bases who are assessed repeatedly for chological problems, interview scales are needed to docu-
different studies and thus have an advantage due to pri- ment signs of fatigue, test anxiety, and distractibility (e.g.,
or experience with such assessment situations (even if number of requests to terminate the assessment; signs of
they are not familiar with the exact tests). As stated be- poor motivation). Recently, we found evidence of an ef-
fore [21], poor neurocognitive functioning may not be a fect of symptoms and motivation on neurocognitive per-
consequence but is perhaps a risk factor for addiction formance by both objective and subjective assessments
since poor IQ and neuropsychological functioning [19]. Fourth, no alcoholism scale was administered; such
clearly compromise academic achievement [44]. Indi- a scale might elucidate which factors are especially tied to
viduals with lower cognitive performance are known to primary cognitive or secondary malperformance. For ex-
have worse vocational functioning [45] and are less like- ample, craving may lead to distraction from the assess-
ly to have well-paid, permanent jobs, which may foster ment and likely represents an important mediator (see
psychological problems, including depression and ad- also our second point); tentative evidence in this direc-
diction [46, 47]. tion has already been obtained [17, 25]. Finally, and as
We would like to acknowledge a number of limita- stated in the preceding paragraph, there is a chicken-or-
tions. First, the sample size was small, so independent the-egg problem; due to the cross-sectional design, we
replication in other institutions is warranted. Second, cannot make clear causal inferences as to whether deficits
other contextual factors or mediators need to be exam- already existed at a premorbid stage or even whether pa-
ined for their role in functioning. For example, it was re- tients with normal scores showed a neurodegenerative
cently found that stereotype threat may also lead to sec- decline from a higher-than-normal baseline level. We
ondary impairment [48]. In AUD, motor impairment cannot rule out that unmeasured confounding variables
(e.g., ataxia) may also play a role in timed tests with a have further attenuated the observed indirect effect. It
strong motor component (e.g., the TMTs). Third, in view also deserves to be tested whether some of the contextual
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

Neurocognitive Functioning in AUD: Eur Addict Res 7

Cognitive Test Results DOI: 10.1159/000492160
Downloaded by:
factors that we examined, particularly motivation and er short sessions with clear instructions, preferably also in
distraction by symptoms, are caused by AUD (or corre- written form so that treatment effects are sustained in
lates) and whether some neurocognitive impairments those with memory problems. In addition, treatment
(e.g., lower inhibitory control) are risk factors for AUD. should also consider barriers to treatment engagement
Finally, better-powered trials should examine the dy- such as anxiety, ambivalence and poor motivation.
namic interrelations between neurocognitive deficits and The cross-sectional design of our study does not rule
negative test-taking attitudes, particularly the possibility out neurotoxic effects of alcohol although evidence is
of a vicious circle (e.g., cognitive impairments lead to now accumulating that impairment abates following a
poor motivation and test anxiety, among other conse- longer period of abstinence. As with other disorders, we
quences, which in turn compromise test results). should be cautious not to infer macroscopic dysfunction
in brain areas governing cognitive functioning based on
deviant neuropsychological scores when poor scores may
Conclusions also reflect other factors. Biogenetic explanations may
harm the psychological well-being of our patients and
Patients with an AUD were much slower and less ac- contribute to (self-)stigma [49].
curate than controls on neurocognitive tests. Group dif-
ferences on accuracy were maintained at a very large ef-
fect size when the effects of contextual and motivation Acknowledgments
factors were accounted for but shrank substantially in
None.
size; the mediation was significant. As the list of examined
mediators was clearly not exhaustive, it cannot be ruled
out that the deficits pertaining to accuracy are even lower
Disclosure Statement
and, as for speed, may be nonsignificant when additional
factors are taken into account. However, while the ques- The authors declare that there are no conflicts of interest.
tion if deficits are primary or secondary is important for
basic research and the conceptualization of illness mod-
els, it is perhaps of less relevance for psychological treat- Clinical Trial Registration Details
ment. Irrespective of its cause, treatment should take into
account cognitive deficits, for example by providing rath- Not applicable.

References
1 Stavro K, Pelletier J, Potvin S: Widespread and 7 Vetreno RP, Hall JM, Savage LM: Alcohol- A, Solfrizzi V: Alcohol consumption in mild
sustained cognitive deficits in alcoholism: a related amnesia and dementia: Animal mod- cognitive impairment and dementia: harmful
meta-analysis. Addict Biol 2013;18:203–213. els have revealed the contributions of differ- or neuroprotective? Int J Geriatr Psychiatry
2 Bates ME, Bowden SC, Barry D: Neurocogni- ent etiological factors on neuropathology, 2012;27:1218–1238.
tive impairment associated with alcohol use neurochemical dysfunction and cognitive im- 12 Ganguli M, Vander Bilt J, Saxton JA, Shen C,
disorders: Implications for treatment. Exp pairment. Neurobiol Learn Mem 2011; 96: Dodge HH: Alcohol consumption and cogni-
Clin Psychopharmacol 2002;10:193–212. 596–608. tive function in late life: a longitudinal com-
3 Stephan RA, Alhassoon OM, Allen KE, et al: 8 Fama R, Le Berre AP, Hardcastle C, Sassoon munity study. Neurology 2005; 65: 1210–
Meta-analyses of clinical neuropsychological SA, Pfefferbaum A, Sullivan EV, Zahr NM: 1217.
tests of executive dysfunction and impulsivity Neurological, nutritional and alcohol con- 13 Squeglia LM, Spadoni AD, Infante MA, My-
in alcohol use disorder. Am J Drug Alcohol sumption factors underlie cognitive and mo- ers MG, Tapert SF: Initiating moderate to
Abuse 2017;43:24–43. tor deficits in chronic alcoholism. Addict Biol heavy alcohol use predicts changes in neuro-
4 Scheurich A: Neuropsychological function- 2017. Epub ahead of print. psychological functioning for adolescent girls
ing and alcohol dependence. Curr Opin Psy- 9 Anderson ML, Nokia MS, Govindaraju KP, and boys. Psychol Addict Behav 2009;23:715–
chiatry 2005;18:319–323. Shors TJ: Moderate drinking? Alcohol con- 722.
5 Xiao P, Dai Z, Zhong J, Zhu Y, Shi H, Pan P: sumption significantly decreases neurogene- 14 Rourke SB, Grant I: The interactive effects of
Regional gray matter deficits in alcohol sis in the adult hippocampus. Neuroscience age and length of abstinence on the recovery
­dependence: a meta-analysis of voxel-based 2012;224:202–209. of neuropsychological functioning in chronic
morphometry studies. Drug Alcohol Depend 10 Neafsey EJ, Collins MA: Moderate alcohol male alcoholics: a 2-year follow-up study. J Int
2015;153:22–22. consumption and cognitive risk. Neuropsy- Neuropsychol Soc 1999;5:234–246.
6 Zakhari S: Overview: how is alcohol metabo- chiatr Dis Treat 2011;7:465–484. 15 Bernardin F, Maheut-Bosser A, Paille F:
lized by the body? Alcohol Res Health 2006; 11 Panza F, Frisardi V, Seripa D, Logroscino G, ­Cognitive impairments in alcohol-dependent
29:245–254. Santamato A, Imbimbo BP, Scafato E, Pilotto subjects. Front Psychiatry 2014;5:78.
130.237.122.245 - 8/29/2018 1:19:10 PM
Karolinska Institutet, University Library

8 Eur Addict Res Moritz/Irshaid/Lüdtke/Schäfer/

DOI: 10.1159/000492160 Hauschildt/Lipp
Downloaded by:
16 Oscar-Berman M, Valmas MM, Sawyer KS, 25 Werkle N: Prädiktoren mnestischer Prob- 39 Woodard JL, Axelrod BN: Wechsler Memory
Ruiz SM, Luhar RB, Gravitz ZR: Profiles of leme in der Allgemeinbevölkerung unter be- Scale – Revised. Psychol Assess 1987; 7: 445–
impaired, spared, and recovered neuropsy- sonderer Berücksichtigung von Stimmung, 449.
chologic processes in alcoholism. Handb Clin Alkoholgenuss, Schlaf und Geschlecht, 2016. 40 Härting C, Markowitsch HJ, Neufeld H, Cal-
Neurol 2014;125:183–210. University of Hagen, Hagen. abrese P, Deisinger K, Kessler J: Wechsler
17 Naim-Feil J, Fitzgerald PB, Bradshaw JL, Lub- 26 Overall JE, Gorham DR: The brief psychiatric Gedächtnistest – Revidierte Fassung (WMS-R)
man DI, Sheppard D: Neurocognitive deficits, rating scale. Psychol Rep 1962;10:799–812. [Wechsler Memory Scale – revised version
craving, and abstinence among alcohol-de- 27 Lecrubier Y, Sheehan DV, Weiller E, Amorim (WMS-R)]. Hans Huber, Bern, 2000.
pendent individuals following detoxification. P, Bonora I, Sheehan KH, Janavs J, Dunbar 41 Wilson B, Alderman N, Burgess P, Emslie H,
Arch Clin Neuropsychol 2014;29:26–37. GC: The Mini International Neuropsychiatric Evans J: Behavioural Assessment of the Dys-
18 Moritz S, Hottenrott B, Jelinek L, Brooks AM, Interview (MINI). A short diagnostic struc- executive Syndrome (BADS). J Occup Psy-
Scheurich A: Effects of obsessive-compulsive tured interview: reliability and validity ac- chol Employ Dis 2003;5:33–37.
symptoms on neuropsychological test perfor- cording to the CIDI. Eur Psychiatry 1997;12: 42 Miskowiak KW, Ott CV, Petersen JZ, Kes-
mance: complicating an already complicated 224–231. sing LV: Systematic review of randomized
story. Clin Neuropsychol 2012;26:31–44. 28 Reitan RM: Trail Making Test. Manual of ad- controlled trials of candidate treatments
19 Moritz S, Hauschildt M, Saathoff K, Jelinek L: ministration and scoring. Reitan Neuropsy- for cognitive impairment in depression
Does impairment in neuropsychological tests chology Laboratory, Tucson, AZ, USA, 1992. and methodological challenges in the field.
equal neuropsychological impairment in ob- 29 Tombaugh TN: Trail Making Test A and B: Eur Neuropsychopharmacol 2016; 26: 1845–
sessive-compulsive disorder (OCD)? Mo- normative data stratified by age and educa- 1867.
mentary influences, testing attitude, and mo- tion. Arch Clin Neuropsychol 2004; 19: 203– 43 Hayes AF: Introduction to mediation, mod-
tivation are related to neuropsychological 214. eration, and conditional process analysis: a
performance in OCD. J Obsessive-Compul- 30 Heaton RK: Wisconsin Card Sorting Test. regression-based approach. Guilford Press,
sive and Related Disorders 2017;14:99–105. Manual. Psychological Assessment Resourc- 2013.
20 Scheurich A, Fellgiebel A, Schermuly I, Bauer es, Odessa, FL, 1981. 44 Mayes SD, Calhoun SL, Bixler EO, Zimmer-
S, Wölfges R, Müller MJ: Experimental evi- 31 Loong JW: The Wisconsin Card Sorting Test man DN: IQ and neuropsychological predic-
dence for a motivational origin of cognitive (IBM version). Wang Neuropsychological tors of academic achievement. Learn Individ-
impairment in major depression. Psychol Laboratory, San Luis Obispo, CA, 1990. ual Diff 2009;19:238–241.
Med 2008;38:237–246. 32 Zimmermann P, Fimm B: Test for attentional 45 Sadek JR, van Gorp WG: The prediction of
21 Moritz S, Stöckert K, Hauschildt M, Lill H, performance (TAP), 1995. doi: 10.1017/ vocational functioning from neuropsycho-
Jelinek L, Beblo T, Diedrich S, Arlt S: Are we CBO9781107415324.004. logical performance; in: Marcotte TD, Grant
exaggerating neuropsychological impairment 33 Wechsler D: Wechsler adult intelligence scale I (eds): Neuropsychology of Everyday Func-
in depression? Reopening a closed chapter. – Fourth edition (WAIS-IV). Pearson, San tioning. Guilford Press, New York, 2010, pp
Expert Rev Neurother 2017;17:839–846. Antonio, TX, 2008. 113–135.
22 Fervaha G, Zakzanis KK, Foussias G, Graff- 34 Jacobs C, Petermann F: Wechsler Intelligenz- 46 Lorant V: Socioeconomic inequalities in de-
Guerrero A, Agid O, Remington G: Motiva- test für Erwachsene (WIE) [Wechsler Intelli- pression: a meta-analysis. Am J Epidemiol
tional deficits and cognitive test performance gence Scale for Adults]. Zeitschrift für Psychi- 2003;157:98–112.
in schizophrenia. JAMA Psychiatry 2014; 71: atrie, Psychologie und Psychotherapie 2007; 47 Heflin CM, Iceland J: Poverty, material hard-
1058–1065. 55:205–208. ship, and depression. Soc Sci Quart 2009; 90:
23 Moritz S, Klein JP, Desler T, Lill H, Gallinat J, 35 Lezak MD: Neuropsychological assessment, 1051–1071.
Schneider BC: Neurocognitive deficits in 1995. 48 Moritz S, Spirandelli K, Happach I, Lion D,
schizophrenia. Are we making mountains out 36 Helmstaedter C, Lendt M, Lux S: VLMT Ver- Berna F: Dysfunction by disclosure? Stereo-
of molehills? Psychol Med 2017; 47: 2602– baler Lern- und Merkfähigkeitstest [VLMT type threat as a source of secondary neurocog-
2612. Verbal Learning and Memory Test]. Belz nitive malperformance in obsessive-compul-
24 Scheurich A, Müller MJ, Szegedi A, Anghe- Test, Göttingen, 2001. sive disorder. J Int Neuropsychol Soc (in
lescu I, Klawe C, Lörch B, Kappis B, Bialonski 37 Brickenkamp R: Aufmerksamkeits-Belas- press).
HG, Haas S, Hautzinger M: Neuropsycholog- tungs-Test Handanweisung d-2 [Attention- 49 Kvaale EP, Haslam N, Gottdiener WH: The
ical status of alcohol-dependent patients: In- Pressure-Test Manual d2], 1962. “side effects” of medicalization: a meta-ana-
creased performance through goal-setting in- 38 Schmidt-Atzert L: Test d2: Aufmerksamkeits- lytic review of how biogenetic explanations
structions. Alcohol Alcoholism 2004;39:119– Belastungs-Test [Test d2: attention-strain- affect stigma. Clin Psychol Rev 2013;33: 782–
125. test], (ed 6). Hogrefe, Göttingen, 2004. 794.

130.237.122.245 - 8/29/2018 1:19:10 PM

Karolinska Institutet, University Library

Neurocognitive Functioning in AUD: Eur Addict Res 9

Cognitive Test Results DOI: 10.1159/000492160
Downloaded by: