Our current review has managed to identify more randomized trials than Cristea et al. (2016)’s [13] prior meta-analytical review for opioid and cocaine use disorder. Six randomised trials were identified to be included in the current systeatic review, with three trials involving participants with opioid use disorder and three trials involving participants with cocaine use disorders. The evidence synthesized from these trials have provided strong evidence that attention biases are present in opioid and stimulant use disorders. There was similarity in the demographics of the participants of the three trials with opioid use disorder, in that the sample were treatment-seeking individuals, predominantly male, and the diagnosis was ascertained based on psychiatric assessments. Across all six trials, attention bias was reported to be present, despite the different methods of assessment with either the use of Stroop test or Visual probe task. Across the six trials, there remained no consensus that attention bias modification was effective. Evidence synthesis for other secondary outcome measures could not be performed given the heterogeneity in the measures reported and the limited number of trials. The risk of bias assessment for the included trials revealed a high risk of selection and attrition bias.

Our current review, by synthesizing the evidence from randomized trials, has provided evidence that attention biases are present in both opiate-use disorders and stimulant (cocaine) use disorders. The current findings complement and expand on the findings that Cristea et al. (2016) [13] previously reported. As aforementioned, their meta-analytical review included 24 studies that involved alcohol use and tobacco use participants. Even though the prior review made use of a search strategy that included keywords such as “cannabis, marijuana, cocaine, heroin, opiates, amphetamines”, randomized trials were not identified and hence evidence synthesis was limited to alcohol and tobacco addiction only. We acknowledge that in Christiansen et al.’s (2015) [14] review, they have identified more studies than in our reiew. They have identified three studies involving heroin-dependent participants and five studies involving cocaine-dependent participants. However, as aforementioned, the authors did not specify their inclusion and exclusion criteria and failed to utilize a systematic review approach and this limits the quality of the evidence synthesized. Nevertheless, their findings help to demonstrate that attentional bias are present in these disorders and are evident across a variety of study designs. In summary, our results fulfills the gaps in knowledge of Cristea’s prior meta-analysis and add on to the evidence synthesized in Christiansen et al.’s (2015)’s review [14]. Attention biases are present in opioid and cocaine use use disorders, and this finding is of clinical importance, as if attentional biases are present, specialized interventions could be devised to help retrain these automatic biases.

The results arising from the opioid trials demonstrate the presence of attention biases in individuals who were maintained on opiate substitution therapy (Lubman, D.I. et al., 2000) [18]. The presence of attention biases amongst individuals maintained on opiate substitution therapy implies that, whilst pharmacological interventions might help in the stabilization of lifestyle and minimization of harms associated with illicit usage, psychological interventions such as attention bias modification needs to be considered as augmentative treatment, to help address the automatic processes that might result in a relapse. Franken, I.H. et al. [17] proposed in his review that drug-related stimuli is known as “cognitive intermediates” prior to a lapse or relapse, given that such salient stimuli activates unconscious processes leading to one having increased attention, but also leaving the individual with fewer resources available to apply learned coping strategies. Hence, there is clearly a need for attention bias modification to be considered for individuals with addictive disorders. There has also been increased recognition of this need, as evidenced by Heitman’s et al. (2017) [24] proposed protocol, in which they attempted to investigate both the effectiveness and cost-effectiveness of an online, Internet-based attention bias modification delivered in complement to usual intervention for individuals with alcohol or cannabis use disorders. Whilst there was a prior study (Hullu et al., 2017) [25] that evaluated combined cognitive bias modification and cognitive behavioural therapy, there remains, to our knowledge, no similar published studies or protocol for substance use disorders. A consideration of the integration of both modalities of therapy is crucial, given that cognitive behavioural therapies typically target the top down or reflective conscious decision-making processes, whilst bias modification targets the bottom-up or unconscious processes that could be responsible for lapse and relapse.

Amongst the studies identified for the current synthesis, only one study evaluated attention bias modification. The remaining studies evaluated other forms of intervention, including that of cue-exposure therapy and computerized cognitive behavioral intervention. Given this, this limits the evidence synthesis for the overall effectiveness of bias modification. Notably, most of the included studies were studies that merely evaluated for the presence of attention biases using mainly the Stroop task (in four studies). The remaining studies evaluated for attention biases using the visual probe task. Since different interventions were used, it is hard to compare whether the different tasks used (Stroop versus the visual probe) did have an effect on the overall results.

There are many strengths of this current review. We have comprehensively searched through the literature for randomized trials involving highly prevalent addictive disorders. Our current review identified more trials than Cristea et al.’s (2016) [13] review for opioid and stimulant use disorders. This is because we have searched through additional databases, namely that of MEDLINE, ScienceDirect, and Scopus, and the search was conducted up until 2017. Through our extensive search strategy, we managed to include trials that have been conducted among participants with highly prevalent substance use disorders, specifically that of cocaine use disorder. The findings arising from the current review help to provide evidence for the presence of attentional biases in other addictive disorders, and this is important clinically. It is important for clinicians to recognize that automatic processes such as attention biases could predispose individuals with opioid and cocaine use disorders towards a lapse or relapse. Clinicians need to recognize that for the chronically addicted individuals, psychological interventions such as cognitive behavioural therapy for relapse might not be adequate and there is a potential role for augmenting that treatment with attention bias modification. We managed to perform a risk of bias assessment to assess the quality of the trials that were included. We have adhered to our planned protocol and used a predefined search strategy and have adhered to the specified inclusion and exclusion criteria.

Despite the strengths, there are also several limitations. In our current review, we only included six trials. Our main limitations pertain to the fact that there might be other studies that have been published and were evaluated for the presence of attentional bias and the effectiveness of attentional bias modification but were excluded due to their study designs. We have not in our current review included the articles that Christiansen et al. (2015) [14] included in their prior review, as the prior review included studies with different study designs. We have originally considered in our review only randomised trials for inclusion in order to enhance the quality of the evidence derived by the synthesis [26]. There are intrinsic advantages with regard to considering only randomised trials. Trials involving randomisation remove the inherent chance of confounding, and the double-blinded process also helps in the minimisation of biases (Shrier, I. et al., 2007) [26]. However, Shrier, I. et al. (2007) [26] previously articulated the need to consider the inclusion of other trial designs as the additional information derived could help in clinical reasoning and provide a solid foundation for causal inferences. Similarly, Peinemann, F. et al. (2013) [27] also opinionated that the consideration of the types of studies to be included is dependent on the clinical question that needs to be answered. Thus, for our current review, the main clinical question relates to whether attentional biases are present in highly prevalent substance use disorders, and whether these biases could be modified, and to look at its effect on other secondary outcome measures. As such, restricting the inclusion criteria to include only randomised trials would result in a significant number of observational studies being left out, as demonstrated by Christiansen et al. (2015)’s [14] review. Moreover, whilst attention biases are widely studied for alcohol and tobacco use disorders, attention bias and bias modification for stimulant use, cannabis use, and opiate use disorders are less well characterised. Also, given that many of these substances are considered illicit globally, it is potentially difficult for researchers to conduct randomised trials, given the ethical and legality concerns associated with a randomised trial. Given the high risk of bias in some of these randomised trials, especially for attrition and selection bias, randomisation has potentially not minimized the risk of confounders. Hence, the quality of the evidence synthesised from these limited randomised trials might not be superior in comparison to a review that considered the inclusion of a variety of study designs. Hence, in retrospect, to better answer the research and clinical question, a consideration of different study designs might yield more evidence for attentional bias modification in this emerging field, instead of limiting just to randomised trials. In addition, our review is limited as it did not include any studies involving participants with cannabis use disorder, which is a highly prevalent disorder globally. Lastly, we were limited to a qualitative synthesis of the data extracted from each of the studies, and a meta-analytical synthesis was not appropriate due to the heterogeneity in the outcomes reported.