All studies suffer from severe methodological limitations which constrain possible conclusions. Only two studies included control groups [38, 41], whereas two studies performed post hoc classifications between improvers and nonimprovers [37, 40], and four studies [39, 42�C44] did not employ control conditions at all. Sample sizes were very small [39, 40, 42�C44] or contained unequal group ZD6474 sizes [38] in six studies, and most studies did not perform intention-to-treat analyses to correct for the high number of dropouts. The study participants were mostly in the postacute phase after detoxification, except for the study by Li et al. [41]. Three studies did not specify the patients’ diagnoses [39, 40] or included subclinical participants [37].

Furthermore, two studies included culture-specific interventions that were no pure EX interventions [38, 41] which hampers generalizability. In three studies [37, 39, 41], group differences concerning specific outcome variables were found already at the beginning of the study, partly explaining group differences at the end of treatment. Finally, self-reported substance use was not chemically validated in three studies [37, 42, 44].In summary, evidence is very weak concerning the efficacy of EX as an adjunct therapy in the treatment of illicit drug abuse/dependence. The studies published so far are not methodologically sound and generalizable. Therefore, only a few preliminary conclusions can be drawn pointing towards unspecific benefits of EX, given a certain duration and training intensity of the EX intervention.

Well-designed studies using adequate sample sizes and control groups are needed to determine if EX programs are effective for treating SUD and, if so, under which conditions.4. Potential MechanismsA couple of mechanisms are discussed when arguing for a beneficial role of EX intervention in psychiatric disorders and especially in SUD. However, most of these mechanisms have not directly been investigated in human SUD population. Therefore, conclusions for mechanisms can cautiously be drawn from animal studies and studies in other human populations, but their translation to SUDs remains speculative at present. Two reviews listed some of these potential mechanisms for alcohol use disorders [54, 55].4.1. Neurochemical Alterations by EXIn alcohol-dependent patients, dysfunctions of dopaminergic, glutamatergic, and opioidergic neurotransmission have been linked to craving and relapse [56]. A large number of animal studies reported acute and chronic EX-induced alterations in GSK-3 different transmitter systems, by modifying transmitter release, reuptake, turnover, or receptor density and sensitivity, respectively.