Our review provides the first systematic description of individuals who have self-excluded from terrestrial and online gambling. The description included basic sociodemographic features, gambling behavior characteristics, the use of additional help, goals, and motives as well as barriers for self-exclusion. 
With regard to sociodemographic characteristics, individuals who self-excluded from terrestrial gambling were generally men in their early to mid-40s. Gamblers who had excluded themselves from online gambling were on average 10 years younger than self-excluders from terrestrial gambling. These results are in line with studies showing a higher prevalence of problem gambling among men compared to women as well as a younger average age of online gamblers compared to terrestrial gamblers (Dowling et al., 2017; Johansson, Grant, Kim, Odlaug, & Götestam, 2009; Hing, Russell, Gainsbury, & Blaszczynski, 2015; Wardle, Moody, Griffiths, Orford, & Volberg, 2011; Wood & Williams, 2011). Moreover, the results show an uneven gender ratio between self-excluders from terrestrial and online gambling with a higher proportion of men among self-excluders from online gambling than from terrestrial gambling. This observation is consistent with the results of previous studies (Blaszczynski, Russell, Gainsbury, & Hing, 2016; Gainsbury et al., 2012; Hing, Russell, Gainsbury, et al., 2015). 
Gambling on EGMs, including slot machines, was identified as the form of terrestrial gambling that most frequently led to the development of gambling-related problems among self-excluded gamblers. This confirms observations of studies reporting EGMs and slot machines as high-risk games more strongly associated with gambling problems than other types of gambling (Breen & Zimmerman, 2002; Hodgins et al., 2012; Lund, 2006; Meyer, Häfeli, Mörsen, & Fiebig, 2010; Petry, 2003). Therefore, self-exclusion programs should be tailored to the high-risk group of EGM and slot machine gamblers. 
The major motives for initiating self-exclusion were financial issues and significant others. While for gamblers self-excluded from terrestrial gambling, significant others played an important role in decision-making; they had only little impact on gamblers self-excluded from online gambling. This might be related to findings that online gamblers were less often married or living with a partner than gamblers of terrestrial gambling (Griffiths, Wardle, Orford, Sproston, & Erens, 2009; Wood &Williams, 2009).
Despite potentially problematic features specific for online gambling, for example, high accessibility and anonymity as well as the use of immature payments (Griffiths, 2003; Siemens & Kopp, 2011), gamblers excluded from online gambling reported financial reasons less often than gamblers excluded from terrestrial gambling. This may be explained by sociodemographic differences between online and terrestrial gamblers. Results suggest that online gamblers have a higher sociodemographic status, a higher household income and are more likely to work full-time than terrestrial gamblers (Gainsbury et al., 2012; Griffiths et al., 2009; Wood & Williams, 2011). It may alternatively relate to the younger age of self-excluded online gamblers and the fewer financial obligations related to this, that is, self-excluded online gamblers are presumably less likely to have started a family. Moreover, online gamblers may have made use of self-exclusion earlier than terrestrial gamblers due to online self-exclusion being more anonymous and faster to implement. Early self-exclusion might prevent (at least in part) financial problems, and hence, they may have been less important for self-excluders from online gambling. These findings illustrate that advertisement and media campaigns for self-exclusion should be tailored to the specific motives and needs of online and terrestrial gamblers.
The primary aim of self-exclusion was to prevent further gambling and to facilitate abstinence (Hayer & Meyer, 2011a, 2011b; Hing & Nuske, 2012; Hing et al., 2014). This is not surprising considering the fact that the majority of self-excluders from terrestrial or online gambling were identified to gamble problematically or pathologically. This illustrates that the concept of self-exclusion meets the needs of problem and pathological gamblers to regain control over their gambling behavior. However, these results also underline that self-exclusion seems to be used more often as a measure to regulate gambling behavior and to reduce further negative consequences at a later stage of the problem gambling career rather than a measure of early intervention. This supports previous notions of self-exclusion as a measure of harm minimization (Gainsbury, 2014; Hayer & Meyer, 2011b; Hing & Nuske, 2012; Hing, Russell, Gainsbury, et al., 2015; Hing et al., 2014). Nevertheless, if initiated earlier, self-exclusion may also be used to prevent intensive forms of gambling and consequently the development of problem and pathological gambling. To increase the effectiveness of self-exclusion and to promote early utilization, self-exclusion needs to be promoted and advertised as an effective means to regulate gambling behavior. In addition, self-exclusion programs for terrestrial gambling as well as self-exclusion options for online gambling need major structural changes to meet the needs of gamblers who wish to take a break from gambling as early intervention measure.
The necessity to improve exclusion programs for terrestrial gambling was underlined by the evaluation of frequently named structural barriers to self-exclusion that delay or prevent gamblers from using self-exclusion, such as impossibility of multiple venue exclusion and problems with venue staff due to false or incomplete information or lack of sensitivity and privacy (Gainsbury, 2014; Hing, Nuske, et al., 2015; Hing et al., 2014; Parke et al., 2014). Availability and accessibility are the key factors that should be considered in the (re)organization of self-exclusion. For instance, (a) gamblers should be able to exclude themselves from multiple venues within a single enrollment preferably outside of gambling venues. In addition, self-exclusion programs should be organized state-wide. Multiple venue exclusion reduces shame and the exposure to gambling within enrollment procedures. Moreover, restricting access to multiple venues will increase the effectiveness of self-exclusion by reducing the likelihood of circumvention and prevention of further gambling; (b) gamblers should be able to determine the length of selfexclusion themselves (Gainsbury, 2014). This might also encourage gamblers at an early stage of their problem gambling career to initiate self-exclusion and to experience a time out from gambling; (c) venue staff members need to be trained in all aspects of self-exclusion, for example, giving adequate information, treating applicants respectfully, ensuring adherence to self-exclusion by strict identity checks, etc. Moreover, staff members need to be trained in early recognition of problem gambling behavior and they should be encouraged to motivate possible problem gamblers to take a break from gambling. By improving staffs’ competences and adherence to rules, gamblers will establish trust in self-exclusion programs.
Although research on barriers to self-exclusion from online gambling is lacking, the results indicate rather diverse abstinence goals. To address different needs, gambling providers should enable self-excluders to make individual decisions on the duration as well as on the extent of the exclusion, for example, exclusion from specific types of gambling or blocking the account. While some gambling providers already offer different self-exclusion options, this should be mandatory for all gambling websites. However, the exclusion is then limited to a single provider and therefore legal regulations and technical solutions are necessary to prevent further gambling using another gambling website or creating a new account.
The results indicate that the utilization of additional counseling increased during the self-exclusion period. However, additional forms of support were utilized rather rarely. In general, the use of further health care offers strongly depends on the interconnection between selfexclusion programs and further intervention measures. For example, accepting a direct offer of further professional help within the process of self-exclusion needs less self-initiative and is therefore more likely than active help seeking by the gambler. Besides offers for additional professional help at enrollment, follow-ups should be a tool accompanying selfexclusion. By this, further professional help can be offered through the whole self-exclusion period and help to increase their utilization. The self-exclusion program in Montreal (Canada) can be considered as a good example of the interconnection of self-exclusion and other forms of support. Here, gamblers are offered a session with an independent psychological adviser during the implementation of self-exclusion, who provides information on options for professional help (Tremblay et al., 2008). In contrast, the links between the German self-exclusion program and other offers of support have been evaluated as not sufficiently developed (Fiedler, 2014).
When interpreting the results of this review, some limitations need to be considered. First, the included studies used different methodological approaches, for example, quantitative and qualitative. Moreover, quantitative data are based on self-reports, self-exclusion files, or behavioral tracking data. With regard to the aims of this review, the use of this broad range of methodological approaches was beneficial, especially concerning the analyses of aspects that were not constantly focused in research to date, for example, motives or perceived barriers related to self-exclusion. Second, self-exclusion of gamblers is a rather new research field, where randomized controlled trials or other more rigorous methods are practically impossible or difficult to perform. Our results were thus extracted from mostly descriptive studies, which ranged in their quality. To ensure the highest possible data quality, we only included peer-reviewed publications and assessed their data quality. Publications below the recommended thresholds were excluded. Third, the results may be biased by the different self-exclusion programs in countries included in the study. Due to the scarce information given in the studies on self-exclusion jurisdictions and regulations, generalizability of results and conclusions may be limited or only applicable to certain programs. Finally, due to lack of data, cultural factors that may influence gamblers’ attitudes toward the use of self-exclusion could not be considered.
