In 2011, the Finnish past-year problem gambling (SOGS ≥ 3) rate was 2.7% (population estimate 110 000). Based on Williams and colleagues, the standardised problem-gambling rate for Finland would be 1.5% in 2011, which would be considered average and which would be comparable also to other standardised rates seen in Sweden, Australia, Canada and the United States [27]. However, when looking at the impacts of one’s gambling problem from the perspective of the CSOs, the percentage is considerably higher. The large proportion of CSOs reflects the fact that an individual’s gambling problem has extensive impacts at the community level.
Almost a fifth (19.3%) of Finnish respondents were defined as CSOs (population estimate 786 000). The Finnish CSO prevalence is almost identical to that of the Swedish study [9]. However, the Swedish study used a more open approach in defining CSOs: they were not able to identify the relationship with the problem gambler in spite of their interest in doing so [9, 33]. In our study, relationships to a problem gambler varied from close family members to close friends, excluding more distant friends, colleagues or relatives. Our results showed that the problem gambler was most typically not a family member but a close friend. In the Norwegian study, the approach was restricted to the family context and they identified only 2.0% of the population as CSOs [3]. Despite this, the proportion of CSOs was greater in Finland than in Norway, since 8.6% of Finnish respondents had a problem gambler in the family. The differences between our results and the Norwegian results may be because the instrument used in the Norwegian study required respondents to know that they had been lied to by their gambling relative and to have noticed that their relative had spent more and more money on gambling [3].
Overall, the proportion of the CSOs did not differ between men and women. However, females (10.4%) had at least one family member who had had a gambling problem statistically significantly more often than male CSOs (6.8%). This finding is consistent with the Norwegian population study, which found female gender associated with having a problem gambler in the family [3]. However, the Swedish population study’s more open approach indicated that males were somewhat more likely to be CSOs than females [9]. On the other hand, our results indicate that men had close friends who were problem gamblers more often than women. Therefore, the differences between these three population studies may be explained by different definitions of CSOs. In order to assure better comparability, the use of a coherent definition for CSO should be pursued in further research.
The proportion of female CSOs that were concerned about partners’ or sisters’ or brothers’ gambling was larger than the proportion of the male CSOs, whereas male CSOs were concerned about a close friend’s gambling more often than females. These gender differences may reflect that men are more likely to have gambling problems compared with women. Therefore, one may assume that women are more likely to be married or be the sibling of a problem gambler. Similarly, if men are more likely to be close friends with other men rather than other women, thus perhaps men are more likely to be friends with a problem gambler. Findings from two previous help-seeking CSO samples also indicate that CSOs were the intimate partner of or were in a relationship with a problem gambler [34, 35]. In addition, help-seeking CSOs were mainly female [34]. Overall, the CSOs of problem gamblers encounter a great deal of general relationship and personal distress. However, it is probable that distress caused by family members’ gambling is different from distress caused by friend’s gambling [2, 7, 8, 36, 37], and they should be taken into account when planning and developing support for CSOs. Further studies evaluating these differences are needed.
Our bivariate analyses revealed that the gambling behaviour of the CSOs also paralleled the gambler’s gambling behaviour. First, the number of game types gambled, the past-year gambling problems (SOGS ≥ 3) and the onset age of gambling less than 18 years were associated with being a CSO for both genders. Furthermore, multivariate analyses revealed that female CSOs had two statistically significant gambling-related variables as underlying factors that may cause distress in their lives: their own gambling problem and past-year gambling. In all analyses, any past-year gambling involvement was statistically significantly associated with being a female CSO, but not being a male CSO. Further studies are needed to confirm and explore this finding.
CSOs reported impairments in several aspects of perceived health and well-being correlates: for both genders, mental health problems were significantly associated with being a CSO. This finding was in line with the Swedish study, which also found substance abuse clearly associated with being a CSO [9]. In our multivariate analyses, risky alcohol consumption was statistically significantly associated with being a male CSO, while daily smoking was statistically significantly associated with being a female CSO. In fact, CSO’s daily smoking was the only correlate where the 95% CI did not overlap for males and females (Table 3). Therefore, the clearest gender difference that might have practical importance [38] was seen in smoking. This is an intriguing finding, since smoking is also strongly associated with female gamblers: women who smoke are 14 times more likely to be pathological gamblers than non-smoking women, whereas for males the corresponding figure is five [39].
Overall, comorbid mental health problems and substance abuse, including alcohol and nicotine, are clearly associated with the severity of gambling problems [40, 41]. In the clinical context, it has been proposed that smoking may enhance the gambling experience or serve as a cue for gambling among problem gamblers [39, 42]. Our results indicate that similar correlates are also associated with being a CSO. Further research on CSOs should seek to establish whether it is because they live in the same environment and most likely share similar lifestyles, as has been suggested by Etcheverry and colleagues [43], or whether it is a coping method for different types of life stress.
Since there are obvious personal and familial costs associated with problem gambling, it is important to acknowledge the specific needs of CSOs whilst planning support and treatments for problem gamblers. Recovering gamblers have indeed reported that family members play an assistive role in their recovery process and that CSOs are naturally motivated to work on their own situation as well as helping the gambler to recover [2, 11, 42, 44, 45]. Reports from gambling helplines also show that CSOs request assistance and support [46, 47]. The Finnish gambling helpline Peluuri provides support, information and consultation services via telephone, short message service and internet for problem gamblers and their CSOs. In 2012, 34% of the Peluuri helpline calls came from CSOs, and during 2013, CSOs reported experiencing increasingly more social- and health-related harms due to the gambling of their significant other [48, 49].
The representativeness of this population study is limited by attrition. However, the data were weighted based on age, gender and residency in order to enhance sample representativeness [22]. A power analysis was not calculated when determining the sample size. The sampling frame was drawn from both landline phones and mobile phones, but the relevance of the use of the dual-frame was not analysed. However, Jackson and colleagues (2013) have recommended the use of a dual-frame sampling methodology, since a traditional landline sampling has proven to impair the sample representativeness [50].
Due to the cross-sectional design, conclusions about the causality of the correlates and CSOs are not possible. In addition, our study did not look comprehensively at the different types of gambling harms that CSOs experienced. For example, Svensson and colleagues (2013) noticed that both male and female CSOs had more financial difficulties than other non-CSOs. The CSOs had lent money to someone who they thought or knew would use it to gamble or pay bills and they were more often exposed to violence during the past year than other people [9]. Previous results from clinical contexts have also indicated that emotional and physical abuse is associated with problem gambling [5–7, 35]. Thus, it would be important to include these topics in further research. In particular, it would be useful to clarify the consequences of having a problem gambler as a significant other, as well as to study the help-seeking of CSOs in more depth.
Although the CSO’s gambling behaviour was assessed with a validated instrument, the gambling problems of their significant other were based on the CSOs own perceptions and were not assessed with a validated instrument or diagnostic interview. In this study, the concern mainly reflects the potential existence of the gambling problems of the significant others without any evaluation of the amount or type of concern. Additionally, the verb tense used to assess this was ‘has had problems’. With this type of wording, the prevalence of problem gambling as assessed by CSOs should be considered as ‘lifetime prevalence’ rather than ‘past-year prevalence’.
The past-year gambling problems, the alcohol consumption and mental health problems of respondents were evaluated using previously validated instruments [23–27]. However, with the SOGS, excessive weight is given to items concerned with borrowing money, with nearly half of the 20 equally weighted items dealing with sources for funding gambling. The SOGS has also been criticized for not being sufficiently sensitive to slot machine-related problems or to gambling problems in women [26, 51]. This is notable in Finland where slot machines are among the three most popular game types and women’s gambling prevalence has increased in 2007 and 2011 [22, 52].
Finally, previous population-based research on CSOs is rare and the existing theoretical or empirical knowledge on CSOs are limited. Due to the exploratory aims of this study, a large number of statistical tests were undertaken without corrections being made for multiple comparisons. Therefore, it is expected that some of the findings of significant relationships in this manuscript may be incorrect and further research is needed to test the specific hypotheses arising here so as to confirm the existence of these specific relationships. This study mainly offers valuable suggestive information and recommendations for further research.