This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Alcohol use disorder (AUD) has been associated with diverse physical and mental morbidities. Among the main consequences of chronic and excessive alcohol use are cognitive and executive deficits. Some of these deficits may be reversed in specific cognitive and executive domains with behavioral approaches consisting of cognitive training. The advent of computer-based interventions may leverage these improvements, but randomized controlled trials (RCTs) of digital interactive-based interventions are still scarce.
The aim of this study is to explore whether a cognitive training approach using VR exercises based on activities of daily living is feasible for improving the cognitive function of patients with AUD undergoing residential treatment, as well as to estimate the effect size for this intervention to power future definitive RCTs.
This study consisted of a two-arm pilot RCT with a sample of 36 individuals recovering from AUD in a therapeutic community; experimental group participants received a therapist-guided, VR-based cognitive training intervention combined with treatment as usual, and control group participants received treatment as usual without cognitive training. A comprehensive neuropsychological battery of tests was used both at pre- and postassessments, including measurement of global cognition, executive functions, attention, visual memory, and cognitive flexibility.
In order to control for potential effects of global cognition and executive functions at baseline, these domains were controlled for in the statistical analysis for each individual outcome. Results indicate intervention effects on attention in two out of five outcomes and on cognitive flexibility in two out of six outcomes, with effect sizes in significant comparisons being larger for attention than for cognitive flexibility. Patient retention in cognitive training was high, in line with previous studies.
Overall, the data suggest that VR-based cognitive training results in specific contributions to improving attention ability and cognitive flexibility of patients recovering from AUD.
ClinicalTrials.gov NCT04505345; https://clinicaltrials.gov/show/NCT04505345
Alcohol is a psychoactive substance that acts on the central nervous system, leading to dependence while causing severe physical, mental, and social problems [
These impairments at the brain level may disrupt behavior in such a way that individuals become overreactive to external cues related to the substance; this causes them to be unable to control substance-seeking behaviors and to make long-term decisions [
However, cognitive deficits resulting from chronic alcohol consumption are usually at least partially reversed during inpatient recovery periods, mostly as a result of abstinence [
Meanwhile, recent technological progress has allowed new solutions for cognitive training based on the use of computerized systems to be developed. This trend is known as computerized cognitive training (CCT) and includes a growing number of systems available for cognitive training in different clinical contexts. An early study involving CCT did not show significant cognitive improvements from this treatment among individuals in residential treatment for AUD in comparison to controls [
A recent systematic review of cognitive training in AUD recovery suggests that these approaches may be useful to promote cognitive functioning on top of improvements due to mere abstinence. However, the available data with AUD individuals only provide evidence for near-transfer effects to very similar tasks, with no evidence regarding far-transfer effects to dissimilar tasks or to everyday functionality [
Some authors argue that virtual reality (VR)–based cognitive training is an especially ecologically valid form of CCT because it includes exercises that mirror everyday life activities and those that involve similar demands to those of everyday living [
Thus, previous research suggests that there are positive effects of computer-based cognitive training, but the specific contributions of VR-based cognitive training reproducing everyday life activities have not yet been demonstrated. In particular, randomized controlled trials (RCTs) have been lacking. This study builds on previous research but proposes a pilot RCT to estimate the effects of a VR-based cognitive intervention on patients with AUD at the level of memory, attention functions, and executive functions; this approach will help in reducing biases associated with previous noncontrolled studies [
The study design consisted of a two-arm pre-post RCT in which participants were assigned either to an experimental group that underwent VR-based cognitive training combined with treatment as usual in residential community rehabilitation or to a passive control group comprising treatment as usual but without VR-based cognitive training. This trial consisted of an open-label RCT in which patients, researchers, assessors, and therapists were not blinded to group allocation. The allocation ratio was 1:1. Both groups underwent the same treatment program for alcohol recovery that is administered to all inpatients in residential treatment. Random allocation was concealed prior to the start of the study and was based on simple randomization after baseline assessment with random number generation in Microsoft Excel, Office 365. The intervention model consisted of a parallel design. The trial was retrospectively registered at ClinicalTrials.gov (NCT04505345).
The sample was recruited from September 2017 to May 2018 at a clinic for recovery from AUD,
The outcomes of this study were selected from well-established neuropsychological tests. Primary outcomes were based on general cognitive functioning and executive functions, while secondary outcomes were based on specific cognitive tests for memory, attention, and cognitive flexibility.
The Montreal Cognitive Assessment (MoCA) [
The Frontal Assessment Battery (FAB) [
The Rey Complex Figure test (RCF) [
The Toulouse Pierón test (TP) [
The Wisconsin Card Sorting Test (WCST), developed by Grant and Berg [
The ethics committee of the School of Psychology and Life Sciences of the host institution approved the human subjects protocol used in this study. It was conducted according to best practices on human research and the principles of the Declaration of Helsinki.
After reading and signing the offline informed consent document, participants from both groups first completed a sociodemographic and clinical data questionnaire and then a flexible neuropsychological assessment battery consisting of tests of global cognition (ie, MoCA), executive functions (ie, FAB), attention (ie, TP), memory (ie, RCF), and cognitive flexibility (ie, WCST). Participants took between 1 and 1.5 hours to complete this session. Participants were given a code to ensure anonymity, and the questionnaires and battery of tests were identified with this code to pair them with posttreatment assessments.
Participants in the experimental group underwent the VR intervention plus treatment as usual during their second and third weeks of hospitalization. This intervention consisted of 10 sessions, each lasting 30 to 40 minutes, which ran twice a week over a period of 5 weeks; these sessions consisted of the performance of VR exercises based on activities of everyday life guided by one therapist in individual sessions, increasing in difficulty from session to session. Participants in the control group underwent treatment as usual for AUD.
The training intensity can be considered low in terms of the number of sessions, session duration, and total treatment dose, which fell below the common range of 10 to 14 hours [
In the last week of hospitalization, both the control group and the experimental group participants completed the same battery of neuropsychological evaluation tests that were applied in the first session. This session was led by the same evaluator of the first assessment. None of the therapists or evaluators involved in these sessions were the owners of the software used.
Cognitive training was conducted with the Systemic Lisbon Battery (SLB), release 2016 [
The SLB was developed as an alternative to the conventional methods of neurocognitive rehabilitation, but in this study it was used only for cognitive training. The platform consists of a virtual city with several built-up areas, a mini market, a pharmacy, an art gallery, and an interactive home; it also includes nonplayer characters walking around the city. The user is free to walk around the city and is given tasks to pick up certain objects in order to achieve a number of preset goals (eg, buy ingredients from a list at a grocery store). Elements from serious games, such as amount of money used or saved being used as a performance score along with visual and auditory feedback following completion of the tasks, are included in this platform to increase patients’ motivation and retention rates with therapy.
Examples of the virtual reality tasks used for cognitive training.
The data were analyzed using SPSS Statistics for Windows, version 21.0 (IBM Corp). Normality was assessed by analyzing the distributions for each outcome according to skewness and kurtosis and was tested with the Shapiro-Wilk test. Skewness and kurtosis were within limits (±2) for the study variables, except for the DI of the TP. Shapiro-Wilk tests revealed that only the DI of the TP at both assessments and the copy trial from the RCF at postassessment violated the normality assumption. Therefore, those variables were assessed with nonparametric tests.
The baseline characteristics of the groups were compared using Student
To test the effects of the treatment, we used repeated-measures analyses of covariance (ANCOVAs) with one within-subjects factor (ie, pretreatment vs posttreatment assessment) and one between-subjects factor (ie, experimental vs control group) while controlling for potential confounders. Effect sizes in the ANCOVAs are reported as η2 and are given by the following equation:
where SSB is the sum of squares for between-subjects factors and SST is the sum of squares for the total model. Following Cohen [
The final sample consisted of 36 patients; 30 (83%) were male, they were aged between 24 and 65 years (mean 44.83, SD 12.04), and the mean number of years they consumed alcohol was 14.31 (SD 4.34). From those 36 patients, 19 (53%) were assigned to the experimental group and 17 (47%) were assigned to the control group. From the total sample of 36 participants, 5 (14%) reported having used other substances in addition to alcohol in the past. Out of 36 participants, 6 (17%) had completed only 4 years of school (ie, elementary school), 22 (61%) had completed 6 years of school (ie, middle school), 7 (17%) had completed 9 years of school (ie, high school), and 1 (3%) had completed more than 12 years of school (ie, a higher degree). No differences were found between the experimental and control groups regarding gender distribution (
CONSORT (Consolidated Standards of Reporting Trials) 2010 flow diagram. VR: virtual reality.
The comparisons between groups at the baseline assessment for each of the outcomes showed statistically significant differences in the total scores on the MoCA test (
However, because these differences could be due to the higher average age of controls, further analyses accounted for this by using the MoCA and FAB scores at baseline as covariates. Age did not correlate with MoCA (
Pre-post comparisons were performed using ANCOVAs with treatment assessment point (ie, time 1 vs time 2) as a within-subjects factor and treatment (ie, experimental vs control group) as a between-subjects factor controlling for baseline MoCA and FAB total scores on outcomes that were normally distributed. For outcomes that violated normal distribution, separate Wilcoxon tests were conducted for the experimental and control groups controlling these confounders. These analyses revealed a significant effect of the intervention on attention in two out of five of the TP outcomes and on cognitive flexibility in two out of six of the WCST outcomes. Improvements between pre- and posttreatment assessments in the experimental group were found for attention, concentration, cognitive flexibility, visual perception, and memory; these are discussed in more detail in the following three sections.
The ANCOVA revealed a significant interaction effect between factors for correct responses (
Attention outcomes through the Toulouse Pierón test (TP).
Attention outcome | Experimental group score, mean (SE) | Control group score, mean (SE) |
|
||
|
Pretreatment | Posttreatment | Pretreatment | Posttreatment |
|
TP no correct responses | 128.47 (7.95) | 192.76 (10.12) | 140.17 (8.48) | 155.85 (10.79) |
|
TP working efficiency | 106.54 (8.40) | 171.34 (17.57) | 112.79 (8.96) | 108.61 (18.73) |
|
Wilcoxon tests were conducted separately for the experimental and control groups, controlling for MoCA and FAB scores that were divided into two groups by median split. Therefore, eight Wilcoxon tests were conducted comparing pre-post assessments for the DI: experimental group MoCA score of 23 and below (ie, median) versus experimental group MoCA score above 23, and experimental group FAB score of 14 and below (ie, median) versus experimental group FAB score above 14; the same design was used for the control group. These comparisons showed significant differences between pre-post assessments only for the group below the median score for the FAB, revealing a decrease in the DI for both the experimental group (
The ANCOVA for each of the WCST outcomes indicated a significant interaction effect for the total number of errors (
Cognitive flexibility outcomes through the Wisconsin Card Sorting Test (WCST).
Cognitive flexibility outcome | Experimental group score, mean (SE) | Control group score, mean (SE) | ||
|
Pretreatment | Posttreatment | Pretreatment | Posttreatment |
WCST no errors | 74.72 (3.14) | 52.43 (2.70) | 71.79 (3.62) | 72.45 (3.12) |
WCST no trials to complete first category | 40.31 (3.92) | 15.86 (1.62) | 23.95 (4.53) | 12.89 (1.87) |
Despite the fact that the ANCOVA did not reveal a significant effect in the total score from the RCF for the memory trial, the analysis of simple effects (ie, Bonferroni corrected) showed a significant improvement between pre- and posttreatment assessments in memory ability for the experimental group (
Memory outcomes through the Rey Complex Figure test (RCF).
Memory outcome | Experimental group score, mean (SE) | Control group score, mean (SE) | |||
|
Pretreatment | Posttreatment | Pretreatment | Posttreatment | |
RCF memory trial | 21.54 (4.92) | 37.80 (5.68) | 22.16 (5.24) | 24.99 (6.06) |
There is evidence that cognition improves during recovery treatment for AUD, while such improvements may be enhanced by the specific effects of cognitive training [
The results found in this study converge with previous research [
At the initial assessment, the two groups had different results in the MoCA and FAB measures, suggesting different levels of general cognitive ability and executive functions. In order to control for the effect of this difference on the outcomes, these variables were controlled in the pre-post analyses. In those analyses, we found significant differences in cognitive performance between pre- and posttreatment for both groups in attention, as assessed with the TP, and in cognitive flexibility, a component of executive functions, as assessed with the WCST; this shows evidence of the positive effects of the residential treatment plan and alcohol abstinence, confirming a robust finding in the literature [
However, in-depth analyses show that there were significant differences between the groups at the final assessment point for the correct responses and DI of the TP and total number of errors in the WCST; however, in other indicators (ie, number of trials to complete the first category in the WCST) that showed improvements between pre- and postassessments, the experimental and control groups did not significantly differ. This pattern of results suggests that there were effects by the VR training sessions in the experimental group beyond simple abstinence-related effects; this also suggests that the task-specific learning in cognitive training promoted neuroplasticity more clearly than abstinence by itself, thus promoting more consistent improvements in attention and cognitive flexibility, as specific components of executive functioning. In fact, executive dysfunctions typically associated with the prefrontal cortex are among the most pronounced deficits due to alcohol abuse [
Patient retention was another positive feature of this intervention that speaks to its feasibility, as only 14% (3/22) of patients discontinued the cognitive training intervention, which is near the lower bound of the range of 8% to 41% found in a recent review of previous studies [
Overall, this study suggests that the use of VR scenarios is a feasible option to enhance cognitive recovery in patients with AUD, specifically at the level of attention, and that it may also support improvements in cognitive flexibility; these are important cognitive abilities also underlying decision making and retention in recovery programs for AUD [
One main limitation of this study was the small dose of the intervention. Given the time of stay for rehabilitation in residential treatment at the partner institution, it was not possible to extend rehabilitation over a longer intervention period. Thus, training intensity was inferior to most studies with cognitive rehabilitation [
In this study, we found a positive impact of the VR training on the cognitive rehabilitation, particularly on attention and executive functions, of individuals with AUD. Although the residential treatment according to the Minnesota Model has as its main objective the promotion and maintenance of abstinence behavior in relation to alcohol, it can also promote a recovery from alcohol dependence. Such cognitive improvements may not only contribute to a better quality of life among patients but also to their social and family functioning and, therefore, their ability to maintain abstinence.
Future studies should focus on more general outcomes related to functionality, well-being, or quality of life to help understand whether such cognitive-focused approaches also contribute to overall psychological adjustment or whether there are far-transfer effects of skills to dissimilar tasks than those trained in the program. It is also worth studying whether these effects remain stable with time when assessed at longer follow-ups.
analysis of covariance
alcohol use disorder
computerized cognitive training
dispersion index
Frontal Assessment Battery
Montreal Cognitive Assessment
Rey Complex Figure test
randomized controlled trial
Systemic Lisbon Battery
sum of squares for between-subjects factors
sum of squares for the total model
substance use disorder
Toulouse Pierón test
virtual reality
Wisconsin Card Sorting Test
PG was responsible for study conceptualization, writing the original draft, methodology, and software development and is the owner of the software used in this study. JO was also responsible for study conceptualization, conducted the formal analysis, and wrote, reviewed, and edited the manuscript. MM and EC, under the supervision of AD and PL, prepared the evaluation protocol and conducted the data collection. PL was also involved in study conceptualization. RB was responsible for revising and editing the final manuscript. All authors contributed to, and have approved, the final manuscript.
None declared.