Combining approach bias modification with working memory training during inpatient alcohol withdrawal: an open-label pilot trial of feasibility and acceptability

Contemporary neurocognitive models suggest that in alcohol use disorder (AUD), information-processing and decision-making are biased towards relatively automatic, implicit processes which drive initial approach tendencies [1, 2]. According to these models, normative information processing involves a series of iterations which progressively allow more associations and contextual information to be activated, enabling evaluation and response selection to move from more automatic to more reflective styles. In AUD, however, rapidly-activated, automatic impulses become sensitised to alcohol-related stimuli, resulting in strong activation of alcohol-approach tendencies, (i.e. approach bias [3, 4]), before more reflective processing styles are engaged. The ability to engage reflective processing may be further undermined by the neurocognitive deficits that are common among people with AUD [5, 6]. Increased approach bias has been found to have cross-sectional associations with increased hazardous drinking [7, 8], and to prospectively predict increased drinking after treatment in problem drinkers seeking to reduce their drinking [9], although one study found the opposite effect (lower approach bias predicting relapse) [10]. Importantly, however, approach bias can be reduced in alcohol dependent individuals through computerised approach bias modification (ABM) training, which has been shown to lead to reduced relapse rates [11‐14]. Indeed, a recent systematic review suggests that ABM is effective in reducing a range of addictive behaviours [15], including alcohol use when applied as an adjunctive approach for individuals with AUDs [16].

Impairment in working memory, which refers to the ability to maintain and manipulate goal-relevant information, and which is thus implicated in reflective control [17], is one of the neurocognitive deficits most commonly observed among people with AUD [5, 6]. Such WM deficits make it harder to implement strategies to circumvent habitual behaviours, such as impulsive alcohol-approach tendencies, and can therefore hamper treatment outcomes [18]. Research has shown poorer WM among individuals with higher alcohol approach bias [19]. Cognitive training designed to improve WM capacity has shown promising results, including improved performance on WM tasks [20], reductions in delay discounting (the tendency to prioritise immediate small rewards over larger, delayed rewards) [21], and reductions in heavy drinking, particularly among individuals with high approach bias [22]. Therefore, strategies that both dampen approach bias and strengthen WM could improve treatment outcomes, yet to date this has not been explored, despite the fact that both ABM and WM training are low-cost, non-invasive, easy-to-administer interventions.

Treatment for AUD often commences with inpatient detoxification. Research has shown that the initial weeks of abstinence are a period of neuroplasticity and neural reorganisation [23‐25], making this a potentially opportune time to modify neurocognitive processes. However, if detoxification is not followed by post-withdrawal treatment, around 85% of patients relapse [26]. Early relapse prevents subsequent engagement in post-withdrawal treatment. This highlights the importance of examining adjunctive interventions during residential withdrawal to prevent rapid relapse and increase the chance of ongoing post-withdrawal treatment engagement.

Our pilot study of a 4-session course of ABM during detoxification significantly reduced rates of early relapse relative to a control condition [12]. We expected that adding WM training to ABM (“dual-training”) would increase abstinence rates even further. Nevertheless, before executing costly randomised controlled trials, it is important to establish whether dual-training is acceptable and feasible (e.g., tolerated and completed) during detoxification treatment. The aim of the current pilot study was to examine the feasibility and acceptability of a novel computerised ‘dual-training’ program combining ABM and WM training during inpatient withdrawal, and examine rates of abstinence at 2-week follow-up relative to those observed in our earlier study with ABM-alone or sham-training.

Participants’ baseline characteristics are shown in Table 1.

The study findings suggest that engaging patients in computerised WM and ABM training during residential alcohol withdrawal treatment is feasible. Most participants found dual training to be acceptable, rating it as interesting and helpful in terms of perceived improved attention. The two-week abstinence rate was similar to what we previously observed following no (‘sham’) training in our earlier study [12] and lower than we observed with ABM alone (absent WM training). However, it is important to note that this was a feasibility study which was not powered to examine effectiveness, and the estimated abstinence rate may lack precision due to the small sample size (note the 95% confidence interval extends from 36 to 70%).

We expected that adding WM training to ABM (dual training) could enhance abstinence rates, given that previous studies supported the effectiveness of WM training for substance use disorders [22, 32]. However, these previous studies adopted multi-session training protocols (e.g., 25 sessions delivered over one month). Thus, our 4 sessions were likely insufficient to have a positive impact on relapse rates, and future studies should consider testing this approach in settings where longer training is possible (e.g. post-withdrawal rehabilitation), or use post-discharge “booster” training sessions. Our findings regarding feasibility and acceptability suggest that a large proportion of patients would engage with such neurocognitive interventions, making them a worthwhile investment for future research. Limitations of this study include the lack of acceptability questionnaire assessments among the 7 participants who withdrew (meaning acceptability may be over-estimated). Moreover, the acceptability questionnaire was just a 3-item instrument that provides very limited feedback, and did not distinguish between the ABM and WM training. Self-reported alcohol outcomes were also not objectively-verified.

Our findings indicate that dual-training is acceptable and feasible during alcohol withdrawal. We cannot draw conclusions regarding the effectiveness of this intervention from a small single-group, open-label study. However, considering the likelihood that WM training requires more than 4 sessions to be effective, we would suggest that larger trials designed to test the efficacy of this dual-training approach either be in treatment settings where longer courses of WM training are possible (e.g., post-withdrawal rehabilitation programs), or incorporate post-discharge “booster” sessions into their protocol.