Early Improvement of Neuropsychological Impairments During Detoxification in Patients with Alcohol Use Disorder

Abstract

Aims

To assess recovery of alcohol-related neuropsychological deficits in a group of patients with pure severe alcohol use disorder (AUD) during a detoxification program using the Brief Evaluation of Alcohol-Related Neuropsychological Impairment (BEARNI) test.

Methods

Thirty-two patients with severe AUD using DSM-IV criteria (24 men, mean age = 45.5 ± 6.8 years old) were assessed using the BEARNI 8 ± 2 days after alcohol cessation (T1) and then were reassessed within 18 ± 2 days after alcohol cessation (T2). The primary study endpoint was the number of patients initially impaired at T1 who recovered cognitive functions at T2 assessment.

Results

At T1, 59% (n = 19) patients with pure severe AUD had at least one impaired cognitive function assessed by the BEARNI. At T2, 63% of the patients with AUD with deficits at T1 had normal BEARNI cognitive scores (χ² = 7.7, P = 0.005); specifically, the percentages of participants with normal subtest scores were 63% on memory (χ² = 12.4, P = 0.0004), 100% on verbal fluency (χ² = 16; P = <0.0001), 60% on alphabetical span (χ² = 12.8; P = 0.0003) and 67% on visuospatial (χ² = 15, P = 0.0001).

Conclusions

The cognitive impairments of two-thirds of patients with pure AUD included in the present study recovered within 18 days of abstinence, earlier than reported in previous studies.

INTRODUCTION

Roughly 30–80% of patients hospitalized in substance abuse departments present with cognitive impairments (Alarcon et al., 2015; Hayes et al., 2016) such as deficits in attention, executive functioning, processing speed, visuospatial abilities and both episodic and working memory (Stavro et al., 2013). Deficits in impulsivity, long-term memory and higher order executive functions have been shown to predict relapse (Rolland et al., 2019), worsen the course of the disease and hinder rehabilitative efforts (Domínguez-Salas et al., 2016). These deficits can also alter motivation-to-change behavior and the decision-making abilities involved in modifying alcohol habits and maintaining abstinence (Le Berre et al., 2012).

The dynamics of neuropsychological recovery after alcohol cessation are still unclear, as studies investigating the course of cognitive improvement have yielded mixed results (Schulte et al., 2014). Return to the normal range of cognitive functioning varies with time (see Table 1) and type of cognitive function (Pitel et al., 2012; Ridley et al., 2013). Each cognitive function recovers independently depending on the extent of damage to the corresponding brain network (Galandra et al., 2019). Some studies have found that short-term abstinence (i.e. before 30 days of abstinence) is sufficient to recover from alcohol-related cognitive deficits (Manning et al., 2008; Mon et al., 2013; Pelletier et al., 2016). For example, some investigations have shown that recovery of flexibility impairments can be observed within 4 weeks of abstinence (Schulte et al., 2014; Mulhauser et al., 2018). Other studies found that working memory recovered after 3 weeks of abstinence (Manning et al., 2008; Petit et al., 2017), yet still others found that 35 days after alcohol cessation, episodic memory performances had not significantly improved (Mon et al., 2013; Pennington et al., 2013). Finally, another study showed that visuospatial abilities in recent detoxified patients with alcohol use disorder (AUD) were significantly improved 18 days after alcohol cessation (Mulhauser et al., 2018). Taken together, these findings reveal that there is no clear consensus on the speed of cognitive recovery during early withdrawal.

Table 1

Clinical characteristics of longitudinal studies assessing cognitive performance in patients with AUD at early stages

Study	n	Current BDZ treatments at baseline	Current Cannabis opioid or stimulant use	Psychiatric comorbidities	Neurological comorbidities	Cirrhosis	Length Of alcohol abstinence at baseline (T1)	Healthy control group	Number of days between T1 and T2 assessment	Mean Daily alcohol consumption (g/day)	Cognitive domains (course
Maillard et al. (2020)	84	Excluded	n = 10	n = 9	n = 15	Not reported	Not reported	No	33.1 ± 11.90	170.76 ± 100.35	F ↑, WM↑, SP↑, EF↑, I↓,
Luquiens et al. (2019)	395	Not reported	Excluded	Excluded	Excluded	N = 65	Not reported	No	42	210.15 ± 110.64	T ↑ (45%), EF↑, A↑, STM↑, O↑
Mulhauser et al. (2018)	28	Not reported	Cannabis (n = 4)	Not reported	Not reported	Not reported	Mean: 7.75 days (SD = 4.13)	No	17.9	140.10 ± 50.68	T↑(26%), STM↑(32%), VS↑(18%), A↓(14%),
Petit et al. (2017)	41	28.7 ± 14.1 mg/day at T1 10.3 ± 15.2 mg/day at T2	Excluded	Mainly depression BDI =  20.34 ± 11.92)	Symptoms of dementia excluded	Not reported	1 day	Yes	18	Not reported	WM, F, VF, I ↓, STM ↑
Pelletier et al. (2016)	236	31.5% of the patients at T1 29.8% were treated chronically	Cannabis (n = 50)	Excluded	Excluded	N = 43	Mean: 21 days (14–56 days)	No	33.5 ± 7	200.4 ± 130.2	T↑ (45.2%), EF↑, A↑, STM↑, O↑
Pennington et al. (2013)	35	Not reported	Excluded	N = 11	Excluded	excluded	Mean: 6 ± 3 days	No	35 ± 9	140.1 ± 60.8	WM ↑, SP↑, STM↓
Cordovil et al. (2010)	35	T1 = patients were starting benzodiazepine- pines. At T2 most subjects had stopped BDZs	Excluded	Not reported	Not reported	Not reported	<24 h	Yes	14–18	Not reported	A↑, F↑, I↑
Manning et al. (2008)	30	Patients received BDZs over the first 9 day	Cannabis (N = 11)	Excluded	Excluded	Not reported	2–4 days	No	27.6 ± 1.7	300.08 ± 140.64	WM↑, VF↑, I↑, STM↑

Cognitive domains assessed (% of patients who recovered); T = total score EM = episodic memory, STM = short term memory, WM = working memory, VS = visuospatial, F = flexibility, VF = verbal fluency, EF = executive functions, I=inhibition, A = attention, SP = speed processing, O=orientation; ↑ = significant improvement, ↓ = no significant improvement, ‘—’= not impaired (vs. HCs)

The course of recovery of these impairments may be influenced by multiple factors, including level of education, IQ, baseline addiction severity, duration of abstinence, repeated previous withdrawals, current benzodiazepine (BDZ) treatments, psychiatric comorbidities and substance use disorder comorbidities, especially cafnnabis use disorders and opiate use disorders (Loeber et al., 2010). To our knowledge, no previous longitudinal studies investigating the course of recovery of neuropsychological impairments have simultaneously excluded these confounding factors (see Table 1).

Because the exact time of onset of cognitive recovery is unclear, the objective of this study was to investigate the early changes in neuropsychological impairments and their sociodemographic and clinical correlates in inpatient participants with AUD using the Brief Evaluation of Alcohol-Related Neuropsychological Impairment (BEARNI), a test specifically designed to easily assess neuropsychological impairments in clinical practice (Ritz et al., 2015) (Ritz et al., 2021). We excluded patients with previous or current BDZ use, other substance use disorders, psychiatric or neurological comorbidities and cirrhosis to address methodological concerns in prior studies (see Table 1).

METHODS

Participants

Patients with AUD

All patients consecutively admitted to an alcohol detoxification program in the substance abuse department of a psychiatric hospital (Amiens, France) between April 2018 and January 2019 were invited to participate to the study. All patients met at least six DSM-IV criteria for alcohol dependence using the Mini-International Neuropsychiatric Interview (MINI) (a structured interview for DSM-IV and ICD-10 diagnoses) (Sheehan et al., 1998), which is a proxy for severe AUD according to the DSM-5. Inclusion criteria were agreement to participate in the study, age between 18 and 65 years and ability to understand and speak French. The following situations that could have had any effect on cognitive functioning constituted exclusion criteria: history or current use of cannabis, opioids, cocaine, amphetamines or opiates (assessed with urinary toxicology tests and declarative data). Other exclusion criteria were a history of or current comorbid psychiatric disorders assessed using the MINI, including psychotic disorders, depressive disorders and anxiety disorders (Sheehan et al., 1998); prior or current use of psychotropic medications, including BDZs, selective serotonin reuptake inhibitors, serotonin and norepinephrine reuptake inhibitors, tricyclic antidepressants, antipsychotic medications, anxiolytic medications, mood stabilizers and hypnotic medications; history of liver fibrosis; history of fetal alcohol syndrome; stroke; coma; head trauma; epilepsy; and any encephalopathy related to alcohol consumption. All participants were informed about the study and provided their written informed consent before their inclusion, according to the Helsinki declaration, and all study procedures were approved by the local ethical committee.

All patients included in this study participated in a detoxification program in the substance abuse department of a psychiatric hospital (Amiens, France). The residential treatment program used in the present study is similar to those offered in United States and Europe, according to international guidelines (Mayo-Smith, 1997). Treatment workshops included content on communication and social skills, popular beliefs about alcohol, identification of high-risk situations, coping skills training and motivational interviews (Rolland et al., 2016). All participants received oral thiamine for 8 days before T1 (range: 250–1000 mg/day) to help prevent Wernicke-Korsakoff encephalopathy and other thiamine-related deficiencies (Dervaux and Laqueille, 2016).

To date, BDZs represent the gold-standard in the treatment of alcohol withdrawal syndrome. All of the patients included in this study received diazepam or oxazepam, which may impair cognition (Crowe and Stranks, 2017). BDZs were prescribed starting on the day of the admission and treatment lasted for 5–9 days. The BDZ treatment was progressively tapered until the cessation of withdrawal symptoms (i.e. when the Clinical Institute Withdrawal Assessment for Alcohol scores were lower than 7) (Cushman et al., 1985). The dosage of BDZs was assessed using equivalent doses of diazepam (mg/day), according to the UK guidelines (Ashton, 2019). Prior to cognitive assessment in the current study, all patients had received diazepam (n = 31) or oxazepam (n = 1). The mean cumulative dose of BDZ medication was 77.5 ± 62.0 mg of diazepam or equivalent, for a mean time period of 7.0 ± 1.8 days before the cessation of BDZ medication. In this study, none of the patients was taking BDZ medication at T1.

Healthy Controls

A control group of healthy controls (HCs) was constituted from the database used for the validation study of the BEARNI. In this validation study, 58 healthy participants underwent the extensive neuropsychological battery (Ritz et al., 2015). None of the controls had been treated for mental disorders, neurological disorders or severe somatic diseases. They did not meet substance use disorder criteria for alcohol nor any other psychoactive substance except for nicotine (Ritz et al., 2015). None of the HCs had neuropsychological impairments assessed with an extensive battery of neuropsychological tests (data not shown; Ritz et al., 2015).

Sociodemographic and clinical assessments

A questionnaire designed for the current study assessed sociodemographic variables, substance use and several factors previously identified as associated with alcohol-related cognitive impairments: age (Angerville et al., 2019), gender, education level (12 years or greater), mean daily alcohol consumption before detoxification (g/day), duration of AUD (in years), age of onset of first alcohol consumption and last alcohol consumption (in days). The Alcohol Use Disorders Identification Test (AUDIT) (WHO, French version; Gache et al., 2005) also was used to assess alcohol consumption before detoxification in the AUD sample and in the HCs to ensure that they did not meet the criteria for AUD over the lifetime (AUDIT <7 for men and < 6 for women) (see Table 2). Biological liver function tests including aspartate amino transferase (ASAT) and gamma-glutamyl-transferase (GGT), and ultrasonographic data, were collected in the patients with AUD. ASAT and GGT mean values were 66.81 ± 64.82 UI/L and 261.83 ± 473.01UI/L, respectively.

Table 2

Socio-demographic and clinical characteristics of the current samples

	Patients with AUD (n = 32)	HCs (n = 32)	P-values
	Mean scores ± SD	Mean scores ± SD
Age (years)	45.5 ± 6.84	46.56 ± 5.7	0.50
Educational level (years of school)	11 ± 2.77	11.56 ± 2.5	0.4
AUDIT	29.62 ± 4.07	2.9 ± 1.7	< 0.0001

Neuropsychological assessment

The BEARNI developed by Ritz et al. (2015) was used to screen for cognitive impairments in patients with AUD. The BEARNI includes five subtests: (a) Delayed free recall (inspired by the California Verbal Learning Test (Delis et al., 1988), which assesses verbal episodic memory and includes two learning trials of a 12-word list (4 words × 3 semantic categories), followed by a delayed free-recall assessment (/6 points); an alternative 12-word list was available to avoid a potential learning effect and was used in the retest condition; (b) Alphabetical span subtest adapted from the alpha-span task (Belleville et al., 1998), which assesses verbal working memory (/5 points); (c) Alternating verbal fluency subtest (/6 points), which assesses executive functions, specifically flexibility abilities; (d) Hidden figure test, an adapted version of the Hidden Figures Test (Corkin, 1979), which assesses visuospatial abilities, and includes five complex figures each containing two separate hidden figures that the patient has to find within 1-min per figure (/5 points); and (e) Ataxia, drawn from the Walk-a-Line Ataxia Battery (Fregly et al., 1972), assessed by asking the patient to stand on each foot in turn for 30 s, first with eyes open, then with eyes closed (/8 points).

The BEARNI uses two overall total scores: the total of the five subscores (1 for each of the subtests) (maximum score: 30 points), and a total cognitive score that excludes the ataxia subtest (maximum score: 22 points). To classify patients’ neuropsychological functioning as impaired, the proposed cut-offs for the total score are ≤19 or ≤ 21 for patients with a low (<12 years) or high (≥12 years) education level, respectively, and ≤ 16 or ≤ 17 for the cognitive score. Each subscore also has its own normal cut-off score. The cut-offs validated for delayed free recall, ataxia, alternating verbal fluency, alphabetical span and hidden figure test subtests were ≤ 2, ≤3, ≤2, ≤2.5 and ≤ 1, respectively. An exploratory factor analysis validated the BEARNI’s underlying structure, highlighting five factors that reflected visuospatial abilities, executive functions, ataxia, verbal episodic memory and verbal working memory (Ritz et al., 2015). The standardization of each BEARNI subtest and the two total scores revealed that this test has sufficient diagnostic accuracy for the detection of cognitive impairments related to alcohol use (Ritz et al., 2015; Ritz et al., 2021). Compared with the Montreal Cognitive Assessment (MoCA) and Mini-Mental State Examination, the BEARNI was specifically developed to screen for cognitive impairments related to alcohol use, with a larger range for each subtest.

Procedure

At baseline (T1), patients with AUD were assessed using the BEARNI after alcohol cessation and between 24 and 48 h after BDZ cessation. Patients with AUD were reassessed (T2) using the BEARNI with the alternative 12-word list for the delayed memory subtest, 10 days after BDZ cessation (i.e. 5 plasma elimination half-lives) (Friedman et al., 1992). At each timepoint, we used the BEARNI Total Score and the BEARNI Cognitive score for each patient. The HCs were assessed only once using the BEARNI because we used the control population from the BEARNI validation study (Ritz et al., 2015), and thus they could not be reassessed.

Statistical analysis

Quantitative variables were described using their means and standard deviations, and pairwise comparisons were performed using Student’s t-tests or the Mann–Whitney test when appropriate. Proportion data were described using their frequencies and were compared using the Chi-square test. Significant changes from T1 to T2 in AUD sample were assessed statistically, with analysis of covariance (ANCOVA) by examining BEARNI total scores, BEARNI cognitive scores and all cognitive subtest scores, after adjustment for age, gender and educational level. To highlight BEARNI scores improvement, we compared BEARNI scores of patients with AUD at T2 to BEARNI scores of HCs using ANCOVA, after adjustment for age, gender and educational and level partial square eta (ηp²) calculated to measure the effect size.

Thereafter, we examined whether each clinical variable was linked to neuropsychological recovery in patients with AUD using a logistic regression model, as we dichotomized the AUD sample according to the recovery status (0: no recovery at BEARNI cognitive score, 1: recovery at BEARNI cognitive score) with age, gender and educational level as covariates. Each patient was considered recovered if they had an impaired score at the T1 assessment, but a normal score at the T2 assessment. P-values <0.05 were considered to be significant. SPSS software V24.0 was used for all statistical analyses (IBM SPSS Inc., Armonk, NY).

RESULTS

Sociodemographic and clinical characteristics

Thirty-two patients with AUD and 32 HCs were included in the study (Table 2). There were no significant differences between the group of patients with AUD (75% male) and the control group (65% male) with respect to gender, age or educational level (Table 2). No patient had any clinical, biological or ultrasonographic signs of cirrhosis.

In the AUD sample, the mean quantity of alcohol consumed before the detoxification program was 142.3 ± 87.3 g/day and the mean duration of AUD was 10.5 ± 7.0 years, with a mean age at first alcohol consumption of 16.4 ± 5.2 years. At T1, breath samples were negative for recent use of alcohol in all patients, and urinalyses were negative in all patients for recent use of cocaine, cannabis, opiates, methamphetamines and amphetamines.

Neuropsychological assessment

In patients with AUD, the T1 assessment using the BEARNI was performed 8.0 ± 1.1 days on average after alcohol cessation and 1.34 ± 0.08 days after BDZ cessation. The T2 assessment was performed 18 ± 1.13 days on average after alcohol cessation and 10.1 ± 0.04 days after the last dose of BDZs.

At T1, we found that 97% (n = 31) of patients with AUD had impaired BEARNI total score and 59% (n = 19) were impaired on the BEARNI cognitive score, according to published norms (Ritz et al., 2015). In addition, 25% (n = 8) of the patients with AUD had a deficit on the delayed memory subtest, 25% (n = 8) had impaired alternating verbal fluency scores, 47% (n = 15) had impaired alphabetical span subtest scores, 47% (n = 15) had impaired visuospatial subtest scores and (Fig. 1) 47% (n = 15) of patients with AUD had more than two subtest scores impaired. Alphabetical span and visuospatial abilities subtest appeared to be more frequently associated (n = 11 (34.4%).

Fig. 1

Percentage of patients with AUD showing impairment at T1 and T2 as assessed by BEARNI subtests.

At T2, 65% of the patients with AUD with deficits at T1 (n = 20) had a normal BEARNI total score (χ² = 27.7, P < 0.0001); 63% (n = 12) had a normal BEARNI cognitive score (χ² = 7.7, P = 0.005); 63% (n = 5) had normal memory subtest scores (χ² = 12.4; P = 0.0004); 100% (n = 8) had normal alternating verbal fluency subtest scores (χ² = 16; P = <0.0001); 60% (n = 9) had normal alphabetical span subtest score (χ² = 12.8; P = 0.0003); and 67% (n = 10) had normal visuospatial subtest score (χ² = 15, P = 0.0001) (Fig. 1).

Of the patients with deficits at T1, 25% did not have normal BEARNI cognitive scores at the T2 assessment. All patients had cumulative deficits. At the T2 assessment, three patients had scores impaired at delayed memory subscale, eight patients had scores impaired at the alphabetical span and five patients had scores impaired at the visuospatial abilities’ subscale. Scores on the alphabetical ordination and visuospatial abilities subscales appeared to be more frequently associated (n = 4 (50%). Patients who had deficits at T2 had a lower age of onset of first alcohol consumption than patients without deficits at T2 (13.8 ± 2.8 year old vs. 17.4 ± 5.7 year old; t = −3.6, P = 0.002).

When we examined the changes between T1 and T2 in the AUD patients, the mean BEARNI total score significantly increased (F(1, 62) = 13.3, P = 0.001, ηp² = 0.12), as did the cognitive BEARNI score (F(1, 62) = 20.6, P < 0.0001, ηp² = 0.17). Among the BEARNI subtests, mean scores on the delayed free recall, alternating verbal fluency, visuospatial abilities and alphabetical span significantly increased (respectively F(1, 62) = 8.2, P = 0.006 ηp² = 0.10, F(1, 62) = 9.6, P = 0.002, ηp² = 0.12; F(1, 62) = 10.0, P = 0.002, ηp² = 0.12; F(1, 62) = 14.7, P = 0.0002, ηp² = 0.03; Fig. 2).

Fig. 2

(A and B) Neuropsychological performance at Baseline (AUD T1) and Follow-up (AUD T2) in patients with AUD and in HCs. #P < 0.05, ## P < 0.01, ### P < 0.001: P-values of comparisons between T1 and T2 in the AUD patients using ANCOVA ^*P < 0.05, ^**P < 0.01, ^***P < 0.0001: P-values of comparisons of T2 and HCs using ANCOVA.

No significant differences were observed between the AUD at follow-up (T2) and the HC group regarding the mean BEARNI (17.9 ± 4.1 vs. 19.5 ± 4.2; t = 1.5; P = 0.2) or in mean cognitive BEARNI score (14.9 ± 3.6 vs. 13.8 ± 3.2; t = −1.1; P = 0.2), mean scores on visuospatial ability (2.9 ± 1.4 vs. 2.9 ± 0.7; t = −0.03; P = 0.9), delayed memory (4.0 ± 4.1 vs. 3.5 ± 1.2; t = −0.4; P = 0.1), alphabetical span (3.4 ± 1.2 vs. 2.9 ± 0.7; t = −0.5; P = 0.06) and alternating verbal fluency (4.6 ± 1.0 vs. 4.7 ± 0.8; t = 0.2; P = 0.5).

Clinical variables associated to cognitive recovery

After adjusting for age, gender and educational level, AUDIT scores and the age of onset of alcohol consumption were significantly associated with recovery of BEARNI cognitive scores in patients with AUD (Table 3).

Table 3

Logistic regression analysis of clinical variables related to Recovery (BEARNI cognitive scores) in patients with AUD

	Coefficient	Standard error	P-values
Age of onset alcohol consumption	-0.5	0.3	0.02
Duration of AUD	0.05	0.07	0.5
BDZs dosage	-0.002	0.01	0.9
AUDIT scores	-0.4	0.2	0.01
Quantity of alcohol consumed before detoxification	-0.04	0.02	0.08

Result from the logistic regression analysis adjusted on educational level, age and gender

DISCUSSION

Main results

This study investigated the early changes in neuropsychological impairments in inpatient participants with severe AUD using the BEARNI 8 days after alcohol cessation (T1) and 18 days after alcohol cessation (T2). We found that at T2, 63% of the patients with impaired BEARNI cognitive score at T1 had normal scores (i.e. within 18 days of abstinence). To the best of our knowledge, this study is the first to observe recovery in cognitive impairments in a majority of patients with severe AUD within 18 days after alcohol cessation, and to have excluded possible cognitive impairment induced by BDZs, other psychotropic medications, cannabis or other drugs, or cirrhosis. This study provides to the literature the expected rates of cognitive recovery during short-term abstinence from alcohol using a simple tool in clinical practice and could disentangle the cognitive impairments induced by other causes.

Considering cognitive performance overall, a previous study found that after 18 days on average, 33% of patients initially impaired had recovered (Mulhauser et al., 2018). Another study used the MoCA test and observed that 45% of patients had recovered 6 weeks after alcohol cessation (Luquiens et al., 2019). Finally, an additional study found that 45% of patients had recovered 7 weeks after alcohol cessation (Pelletier et al., 2016). In contrast to our findings, the results of these prior studies may be due to several confounding factors, such as psychiatric comorbidities (Mulhauser et al., 2018), the use of BDZ medications (Pelletier et al., 2016; Mulhauser et al., 2018; Luquiens et al., 2019), comorbid cirrhosis (Luquiens et al., 2019) or cannabis use (Pelletier et al., 2016) (see Table 1).

Although we observed an overall improvement in BEARNI cognitive scores, 37% of this pure AUD sample still scored in the ‘deficit range,’ suggesting no recovery within the first 3 weeks of abstinence. All of the patients who had not recovered at T2 had scores impaired at the working memory task. Half of the patients had impaired scores at both the working memory and visuospatial tasks. This association raises the question of a possible common neural mechanism linking these cognitive domains.

Episodic and working memory

In the current study, the recovery rates of patients initially impaired on working memory and episodic memory (63 and 60%, respectively) were not consistent with those of Mulhauser et al., who did not find any recovery in episodic memory or working memory (assessed with the Repeatable Battery for the Assessment of Neuropsychological Status, or RBANS) after 25 days of abstinence. However, they did not report BDZs, psychiatric, neurological or cirrhosis comorbidities, which can impair performance on cognitive tasks and may have biased their results. Maillard et al. (2020) found that 33% of the patients who initially showed episodic memory impairments and 27% of the patients who initially showed working memory impairments subsequently recovered; however, the exact abstinence duration at baseline assessment was not reported in their study (Maillard et al., 2020).

Regarding the improvements in working and episodic memory over time, the results of this study were consistent with the improvement found in previous studies: 25 days after alcohol cessation using the (RBANS) reported by Mulhauser et al. (2018) and 30 days after alcohol cessation using the Wechsler Memory Scale and Verbal Fluency Test (Manning et al., 2008; Maillard et al., 2020).

Flexibility

The rate of recovery in flexibility performance of patients initially impaired observed here (100%) was not consistent with that of a previous study which found that only 12% of patients had recovered after 18 days of abstinence using the Trail Making Test (TMT) A and B (Mulhauser et al., 2018). Another study found retrospectively that after 3 weeks only 12% of those initially impaired on flexibility performance in the Modified Card Sorting Test returned to a normal level (z-scores> − 1.65) (Maillard et al., 2020). This difference may be due to a higher sensitivity of the Modified Card Sorting Test, as stated by Henry (2006) and the presence of other confounding factors (Table 1). The comparison of the results of this study for the flexibility domains with previous studies should be nuanced. Even though the TMT, the Modified Card Sorting Test and verbal fluency test are assessing flexibility, they are different from a sensory-motor perspective.

The significant improvement in verbal fluency subscores between T1 and T2 found in this study was in line with some previous studies that observed improvement in flexibility after 30 days of abstinence using the Verbal Fluency Test (Manning et al., 2008; Maillard et al., 2020) (Table 1).

Visuo-spatial abilities

We found that 67% of patients with AUD impaired at T1 had recovered at T2. To date, no previous study has investigated the rate at which patients with severe AUD recover in a longitudinal design study (Table 1). The significant improvement in mean score on visuospatial abilities that we found between T1 and T2 is in line with a previous study (Mulhauser et al., 2018), which found a significant improvement in visuospatial abilities using the RBANS after 18 days of abstinence.

Clinical variables related to cognitive recovery

This study found that total AUDIT scores were significantly associated with cognitive recovery. However, we were unable to examine whether AUDIT subscales (i.e. consumption and problems) were associated with recovery because only total AUDIT scores were available for analysis.

The age of onset of alcohol consumption was significantly associated with the absence of recovery in patients with AUD. A previous study suggests that early onset of drinking increases the risk for alcohol-related neurocognitive vulnerabilities and that initiation of any or weekly alcohol use at younger ages appears to be a risk factor for poorer subsequent neuropsychological functioning (Nguyen-Louie et al., 2017).

Strengths

The population of our study was homogenous: patients were included in this study only after the washout of BDZs and if they had been diagnosed with ‘pure’ AUD (i.e. without the influence of other psychotropic medications; cannabis, stimulant, or opioid use; psychiatric comorbidities; neurologic comorbidities or cirrhosis). No previous studies assessing cognitive impairments (such as memory, executive functions and visuospatial abilities) in patients with AUD systematically excluded all of these characteristics (Table 1).

Limitations

The results of this study should be viewed within the context of some limitations. First, our sample size was small, but necessarily so due to the pureness of the sample with numerous exclusion criteria and the objective of recruiting a homogeneous sample. Nevertheless, the present findings may not generalize to the clinical population of AUD patients commonly seen in AUD treatment.

Second, this study did not distinguish between the effects of abstinence and those of residential treatment-as-usual; however, our patients did not benefit from any cognitive training programs as described in some studies (Luquiens et al., 2019; Maillard et al., 2020). The residential treatment program used in this study is representative of similar programs offered in United States and Europe, according to international guidelines (Mayo-Smith, 1997). It is possible that the improvements in the BEARNI scores were linked in part with retesting (i.e. a learning effect from T1 to T2) (Jones, 2015). Regarding the episodic memory subtest, the use of an alternating form of the BEARNI could have mitigated this bias; however, with respect to visuospatial abilities and working memory, the test–retest reliability of the Complex Figure Test is high in healthy subjects (Woodrome and Fastenau, 2005), and the alphabetical span and verbal fluency tests have shown satisfactory reliability over time (Waters and Caplan, 2003).

A third limitation may be that tobacco smokers were not excluded in this study. Tobacco could have a potential impact on cognitive recovery in patients with AUD (Schmidt et al., 2017). Weinberger et al. found that more lifetime years of smoking was related to poorer memory recovery over 8 months of alcohol abstinence; however, excluding patients with AUD who also use tobacco could have markedly limited the number of subjects included in this study (Weinberger et al., 2019).

Implications

This study found that 63% of severe AUD patients demonstrating a significant cognitive improvement within 18 days of abstinence. Caregivers should take into account the neuropsychological impairments before 18 days of abstinence, considering that cognitive impairments are linked to the treatment addiction outcomes (Berre et al., 2012). Eighteen days after alcohol cessation could represent a critical timepoint to begin psychotherapies, such as cognitive behavioral therapy, which requires intact cognitive functioning to be effective (Verdejo-Garcia, 2016).

CONCLUSIONS

The current study found that cognitive impairments recovered within 18 days of abstinence in approximately two-thirds of patients with AUD. The BEARNI could be used more systematically in patients with AUD in substance abuse settings. Additional studies assessing cognitive improvements during abstinence, and especially earlier in abstinence, are needed. Further studies should also assess the early course of social cognition, attentional bias and inhibition deficits in patient with AUD early in abstinence, given their clinical impact (Rupp et al., 2021).

AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: a.d., B.A., a.d., M.N.

Performed the experiments: B.A., a.d.

Analyzed the data: B.A., H.H., M.N., a.d.

Contributed reagents/materials/analysis tools: M.N., A.L.P., H.B., L.R.

Editorial comments: M.N., M.P.M., A.L.P., H.B., L.R.

Wrote the paper: B.A., a.d.

ACKNOWLEDGMENTS

The authors thank Dr Wolfgang Persyn and Dr Guillaumont Cyril for their support in accessing the clinical setting.

FUNDING

None.

CONFLICT OF INTEREST

Alain Dervaux has received honoraria for lectures and non-financial support from Camurus, Indivior, Janssen, Otsuka and Lundbeck, outside the submitted work. Mickaël Naassila has received honoraria for lectures from Indivior, Merck-Serono and Lundbeck. All other authors declare that they have no conflicts of interest.

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author, AB, upon reasonable request.