Concerned significant others of persons with problem gambling and problem gaming: Prevalence and characteristics in the Norwegian population

Abstract

Background

Problem gambling (PG) and problem gaming (GAM) generate concern for close ones often denoted as concerned significant others (CSO). There is a dearth of evidence on the characteristics associated with being a CSO of PG and GAM, which is explored in the present study.

Methods

A total of 2,000 persons (aged 16–74 years), randomly drawn from the Norwegian population registry, were invited to respond to a questionnaire assessing current or past concerns about gambling and gaming of significant others, as well as questions about their own gambling, gaming, alcohol and illegal drug use, personality traits, and demographic characteristics. Multiple regression analyses were conducted to identify correlates of being a CSO of PG and GAM, split by close (i.e., first degree relatives and partners) and remote relationships.

Results

In all, 575 persons (46% men, age M = 44.05, SD = 17.31 years) completed the questionnaire (response rate = 30.4%). The prevalence of ever being a CSO was 23.2% for PG (6.8% for close, specifically) and 27.1% for GAM (17.4% for close, specifically). Close/PG was associated with higher Neuroticism. Remote/PG was associated with lower education and own PG. Close/GAM were younger. Remote/GAM correlates were lower income, own PG, and higher Extraversion scores.

Conclusions

PG and GAM have high CSO prevalence rates. In line with the problem behavior theory, the correlates of being a CSO suggest the presence of environments with problems and fewer resources connected to PG and GAM.

Introduction

Gambling is a recreational activity for most people engaging in it, but becomes problematic with severe psychological, social, and financial consequences for some persons (Livazović & Bojčić, 2019; Oliveira, Silveira, & Silva, 2008; Wardle et al., 2024). Globally, it is estimated that 46.2% of adults and 17.9% of adolescents have gambled in the past 12 months, and in adult samples, 8.7% engage in risky gambling and 1.4% in problem gambling (PG; Tran et al., 2024). The estimated prevalence in the Norwegian population (age 16–74 years) is 0.6% for PG and 2.3% for moderate risky gambling, where male sex, lower education, lower income, and birthplace outside Norway are risk factors (Pallesen et al., 2023).

Concerned significant others (CSO) of PG (CSO/PG) are persons who are worried about the gambling behavior of someone in their family or social network. Prevalence depends on definitions, i.e., if restricted to include partners and family only or also close friends. The number of affected-others for the typical PG is estimated to be six people, while persons with PG themselves typically estimate this number to be lower (Goodwin, Browne, Rockloff, & Rose, 2017). In a national survey in Australia, 5.7% of adults experienced harm due to another person's gambling and the affected other was most often a friend, thereafter a spouse/partner (Rockloff, Russell, Browne, & Hing, 2025). In a corresponding Australian study with data from Tasmania 5.1% of respondents reported being affected by other people's gambling the past-year (Dowling, Spence, Browne, Rockloff, & Merkouris, 2025). In Sweden, 18.2% reported that someone close to them, now or previously, suffered from PG (Svensson, Romild, & Shepherdson, 2013), while the corresponding number was 19.3% in Finland (Salonen et al., 2014, 2015). The only prevalence study conducted on CSO/PG in Norway found a prevalence of 2.0% (Wenzel, Øren, & Bakken, 2008).

It is conceivable that the prevalence of CSO/PG reflects the prevalence of PG in a given population, and since the prevalence of PG is not vastly higher in Sweden and Finland than in Norway (Abbott, Romild, & Volberg, 2018; Salonen, Hagfors, Lind, & Kontto, 2020), the above-mentioned discrepancy in prevalence of CSO/PG likely reflects methodical artifacts. The Norwegian prevalence study limited CSO/PG to “close relatives” whom responders both had noticed spent increasing amounts on gambling, and had lied about how much they gambled (Wenzel et al., 2008), which is a rather strict definition. Hence, we lack knowledge about the current prevalence of CSO/PG in Norway that is comparable to the prevalence in other Scandinavian countries and in line with recent accepted definitions and understanding of CSO.

CSO/PG report elevated emotional distress, higher rates of smoking and risky alcohol use, impaired physical health, hypervigilance, negative impact on partner relationship and increased conflict with the person in question, as well as financial distress, as compared to the general population (Castrén, Lind, Hagfors, & Salonen, 2021; Riley, Harvey, Crisp, Battersby, & Lawn, 2021). Knowledge about CSO/PG is thus important in order to estimate the total burden of PG, but also to tailor help to both the persons with PG themselves and to the CSO/PG. Treatment of PG is more effective when CSO are involved (Jimenez-Murcia et al., 2017), which is also attested to regarding treatment of other mental disorders (e.g., Claxton, Onwumere, & Fornells-Ambrojo, 2017; Meuleman, Sloover, & van Ee, 2023). Knowledge about characteristics of CSO/PG might inform about the relational resources the persons with PG can count on in their recovery, i.e., it may be hard to abstain from gambling if those close to oneself engage in gambling. Further, the health care system often provides support to CSO (Edgren, Pörtfors, Raisamo, & Castrén, 2022) and knowledge about CSO/PG would help tailoring these interventions.

While people suffering from PG have elevated scores on the personality trait neuroticism, and lower scores on conscientiousness (Bagby et al., 2007; Strømme et al., 2021), little is known about the traits of CSO/PG. Although personality traits are relatively stable individual differences in behavior, motivation, emotions, and thoughts, life events may lead to personality change (Bleidorn, Hopwood, & Lucas, 2018). People with mental disorders typically show elevated levels of neuroticism and low conscientiousness (Kotov, Gamez, Schmidt, & Watson, 2010), and the burden of being a CSO might increase scores on neuroticism. Knowledge about CSO’ personality might be useful when tailoring support given to CSO.

Playing video games is another recreational activity which for some become excessive and problematic. Gaming disorder (GD) refers to persons with impaired control over gaming, where gaming takes precedence over other activities, and continues despite negative consequences (World Health Organization, 2019). Prevalence of GD is estimated to be 1.4–3.3%, where sample size, mean age, and study quality are negatively associated with the estimates (Kim et al., 2022; Stevens, Dorstyn, Delfabbro, & King, 2021). In the Norwegian population (16–74 years), 45.7% participated in gaming during the last six months, while only 0.7% of the population were categorized as suffering from GD and 5% from problem gaming (GAM; Pallesen et al., 2023). GAM is associated with impaired mental health, and academic and social problems (Ferguson, Coulson, & Barnett, 2011; Männikkö, Ruotsalainen, Miettunen, Pontes, & Kääriäinen, 2020). Some video games have incorporated aspects of gambling, i.e., “loot boxes” (Spicer et al., 2021; Wardle et al., 2024). Pallesen et al. (2023) found that about 19% of person with PG reported GAM, and it has been proposed that gaming is a gateway to gambling, i.e. the “gateway hypothesis” (Delfabbro & King, 2020; Molde et al., 2019; Spicer et al., 2021).

So far, there is a dearth of research on CSO of people with GAM (CSO/GAM). There are some studies on the parents and partners of GAM, but seemingly no prevalence studies or studies on CSO/GAM outside the family. Szász-Janocha et al. (2023) found that CSO/GAMreported harms within the domains relationship, emotional well-being, physical or mental health, and work or study. The most frequent harms concerned the relationship domain, but half of parents and one third of partners reported a negative impact also on their financial situation.

We are not aware of any studies on the personality traits of CSO/GAM. People with GAM have lowered scores on conscientiousness (Akbari et al., 2021). As with CSO/PG, the burden of being a CSO/GAM might increase neuroticism.

Problem behavior theory (PBT; Jessor & Jessor, 1977) posits that problem behaviors are connected and that involvement in one problem behavior is connected to other problem behavior involvement. In line with PBT, apart from the gaming-gambling “gateway hypothesis” delineated previously, PG and GAM have similar risk factors and manifestations (Sanders & Williams, 2019). Additionally, there is a positive association between PG as well as GAM and various addictive behaviors such as exercise-, internet-, and work addiction and substance use (Burleigh, Griffiths, Sumich, Stavropoulos, & Kuss, 2019; Villella et al., 2011). As with personality, described above, it is plausible that the identified gateway and co-occurrence problems of PBT manifest similarly in CSO/PG and CSO/GAM.

Against this backdrop, the aim of the current study was to examine the prevalence rates of CSO/PG and CSO/GAM in Norway. Further, we aimed to explore characteristics of CSO in terms of demographic variables (age, sex, income, and education), their own gambling, gaming, and alcohol and illegal drug use, as well as their personality traits, as compared to the general population. In the current study, GAM comprises a broader concept than the specific diagnosis of GD (Kim et al., 2022; World Health Organization, 2019), reflecting concern about the gambling of close ones. Previous research indicates that close CSO, including partners and children, might be more affected than other CSO (Riley et al., 2021). Accordingly, in the current study, we delineate two CSO subgroups; “close” (representing first degree relatives and current partners), and “remote” (representing other CSO, e.g., friends, grandparents).

Methods

Procedure and participants

On January 2, 2024, 2000 Norwegian residents (995 men; age range 16–74, M = 43.56, SD = 16.26 years) were randomly drawn from the population register. The Directorate of Taxation provided information on their names, address, age, and sex. Each received a letter with an invitation to participate in a survey about personality and games, stating that completers of the survey would receive a gift card with a value of 250 NOK (approximately €20) for participation. They were asked to go to a web page and answer within three weeks. Later, a reminder including a paper-based version of the questionnaire, and a pre-paid return envelope was sent by postal mail. Each person was assigned a unique four-digit code. The project received ethical approval from details omitted for anonymous review.

Of the 2000 invitations sent, 109 invitations were returned due to invalid address, severe sickness, or death. This left us with 1891 potential participants of which 575 (266 men, age range 16–74, M = 44.05, SD = 17.31 years) participated, yielding a response rate of 30.4%. The demographic characteristics of participants are presented in Table 1.

Table 1.

Demographics and score on studied variables (N = 575)

Variable	Number/score
Age, years, M (SD)	44.05 (17.31)
Male, n (%)	266 (46.3)
Income1, M (SD)	4.90 (3.02)
Education2, M (SD)	4.89 (3.02)
Country of birth, n (%)
-Norway	492 (85.6)
-Other Europe	52 (9.0)
-Africa	8 (1.4)
-Asia	16 (2.8)
-America	7 (1.2)
PG, n (%)	18 (5.5)
GAM, n (%)	33 (5.7)
AUDIT-C risk, n (%)	216 (38.6)
DUDIT-C risk, n (%)	32 (5.7)
Mini-IPIP (Cronbach's α), M (SD)
Neuroticism (0.73)	10.99 (3.61)
Extraversion (0.80)	13.55 (3.52)
Openness (0.72)	13.92 (3.29)
Agreeableness (0.70)	16.54 (2.68)
Conscientiousness (0.72)	15.74 (3.11)

Note.

PG = Problematic Gambling. Score ≥3 on the Problem Gambling Severity Index (PGSI).

GAM = Problematic Gaming. Score ≥3 sometimes on at least four of seven items on Game Addiction Scale for Adolescents (GASA).

AUDIT-C risk = score ≥4 for women and score ≥5 for men on Alcohol Use Disorder Identification Test – Consumption.

DUDIT-C risk = Other answer than “never” on first question of Drug Use Disorders Identification Test – Concise.

Mini-IPIP = Mini – International Personality Item Pool.

1. Income: Gross annual income where 0 < 100,000 NOK; 1 = 100,000 – 199,000 NOK; 2 = 200,000 – 299,000 NOK; etc. up to 10 ≤ 1,000,000 NOK.
2. Education: Ordinal scale treated as continuous, where 1 = Not completed primary school, 2 = Primary school (9 or 10 years), 3 = High school/vocational, 4 = University, lower degree, 5 = University, higher degree, 6 = PhD.

Measures

Demographics

Age and sex were obtained directly from the population registry. The questionnaire included questions about income level, education level, and country of birth.

Concerned significant others (CSO)

Participants were asked (1) “Do any of the following persons who are close to you have problems with gambling now?”; (2) “Did any of the following persons who are close to you have problems with gambling earlier (not now)?” (3) “Do any of the following persons who are close to you have problems with computer games now?”; (4) “Did any of the following persons who are close to you have problems with computer games earlier (not now)?”. Here, participants themselves determined the distinction between “now” (current CSO) and “not now” (former CSO). For each question, the respondent answered yes or no to the following categories of persons: father, mother, sister/brother, grandparent, partner, ex-partner, child, other family member, friend, colleague, and other (who? Fill in).

Problem gambling

Participants who responded “yes” to a question about whether they had participated in gambling the last 12 months, were asked to complete the Problem Gambling Severity Index (PGSI), which is a subset of items from the Canadian Problem Gambling Index (CPGI; Ferris & Wynne, 2001; Holtgraves, 2009). PGSI consists of five items concerning amount of problematic gambling (e.g., “Have you bet more than you could really afford to lose?”) and four items about negative consequences of gambling (e.g., “Has gambling caused you any health problems including stress or anxiety?”). Items are scored on a four-point Likert scale from 0 = never to 3 = always, which provide a composite score ranging from 0 to 27. It has been translated to and used in several other Norwegian studies (e.g., Pallesen et al., 2023). Cronbach's alpha for PGSI in the current study was 0.84. As this was a population sample and not a clinical sample, cut-off for PG was set to a PGSI score ≥3 (corresponding to “moderate risk gambling” in Ferris & Wynne, 2001) to ensure a sufficient group size, and consequently acceptable statistical power.

Problem gaming

Participants who responded “yes” to a question about whether they had participated in gaming the last 6 months, were asked to complete the Game Addiction Scale for Adolescents (GASA; Lemmens, Valkenburg, & Peter, 2009). The scale is one of the most used scales for assessing gaming problems and has shown good psychometric properties in adult samples (Andreassen et al., 2016; Festl, Scharkow, & Quandt, 2013). GASA contains seven items about use of computer games (e.g., “How often during the last six months did you think about playing a game all day long?”) scored on a five-point Likert scale from 1 = never to 5 = very often. The composite score ranges from 7 to 35. GASA has been used in the Norwegian population study about prevalence of gambling and gaming problems (Pallesen et al., 2023). Cronbach's alpha of GASA in the current study was 0.84. In line with Lemmens et al. (2009), the cut-off for GAM was set to score ≥3 (i.e., sometimes or more often) on at least four of seven items.

Alcohol use

Alcohol use was assessed using the Alcohol Use Disorder Identification Test – Consumption (AUDIT-C; Khadjesari et al., 2017; Saunders, Aasland, Babor, De La Fuente, & Grant, 1993), which consists of three items about frequency and amount of alcohol intake (e.g., “How often do you have a drink containing alcohol?”). Responses are provided on a five-point scale. For this study, cut-off for risky drinking was set to an AUDIT-C score of ≥4 for women and ≥5 for men following Khadjesari et al. (2017)’ recommendation.

Illegal drug use

Illegal drug use was measured with the first item from the Drug Use Disorders Identification Test – Concise (DUDIT-C; Berman, Wennberg, & Sinadinovic, 2015; Hildebrand, 2015), where respondents state if they use other (non-prescribed or illegal) drugs than alcohol. Responses are provided on a five-point scale. Cut-off was set to score ≥1 (i.e., scores other than never).

Personality traits

Personality traits were assessed with the Mini International Personality Item Pool (Mini-IPIP; Donnellan, Oswald, Baird, & Lucas, 2006) which is a 20-item questionnaire assessing the Five-Factor Model of personality traits. The four items that reflect each trait are rated in terms of the degree they describe the respondent on a five-point scale from very inaccurate to very accurate. It has been used in several other Norwegian studies (e.g., Erevik, Pallesen, Vedaa, Andreassen, & Torsheim, 2017). Cronbach's alphas in the current study were 0.73 for Neuroticism, 0.80 for Extraversion, 0.72 for Openness, 0.70 for Agreeableness, and 0.72 for Conscientiousness.

Statistical analyses

Descriptive statistics comprising frequencies, and averages and deviations (M ± SD) were used to compute characteristics of the sample. Multiple regression analyses were conducted to identify the correlates of belonging to each CSO group. Dependent variables for the four regressions were: (1) close/PG, (2) remote/PG, (3) close/GAM, and (4) remote/GAM. Close CSO concerned, now and/or earlier, gambling/gaming of a parent, current partner, sibling or child. Remote CSO concerned gambling/gaming of a grandparent, ex-partner, other family member than mentioned above, friend, colleague or other (who?).

Independent variables were: The background variables age, sex, educational level, and income level (treated as continuous, except sex); participants' own gaming, gambling, drinking and drug use (dichotomous, following cut-offs for risk described above); and their personality traits (continuous). We used independent samples t-tests and chi-square tests for bivariate analyses to identify independent variables associated with being in each of the CSO groups. We investigated zero-order correlations between the dependent and independent variables and conducted multivariate logistic regressions for each CSO group. The analyses were conducted using SPSS statistics version 29 (IBM Corp.).

Ethics

Approval from the Regional Committee for Medical and Health Research Ethics, Health Region West (REK2023/621277).

Results

Prevalence of CSO/PG

The lifetime prevalence of CSO/PG across the two categories was 23.2%. While 14.7% were current CSO/PG, 17.8% were former CSO/PG. The concerned person was most frequently a friend (29 current, 38 former), sibling (15 current, 12 former), or ex-partner (14 previous, 13 former). For further analysis, 38 persons (6.8%) were categorized as close/PG, while 103 (19.1%) were categorized as remote/PG. Sixteen participants belonged to both categories, i.e., reporting concern of PG in both close and remote relationships. See Table 2 for details on which CSO relationship responders belonged to currently and previously. Almost half of the current CSO and a little more than half of the former CSO was somewhat, very or extremely concerned about the person with PG (see Fig. 1).

Table 2.

Number of Concerned Significant Others (CSO) per relationship in Norwegian population (N = 575)

	Gambling (PG)		Gaming (GAM)
Relationship	Current	Former	Current	Former
Father	5	4	1	3
Mother	2	4	0	1
Sibling	15	12	22	25
Child	5	4	32	25
Partner	3	6	9	12
Ex-partner	14	13	8	9
Grandparent	4	9	0	0
Other family	9	14	13	18
Friend	29	38	28	30
Collegue	11	17	10	8
Others	6	9	5	2
CSO1	79 (14.7%)	96 (17.8%)	96 (17.7%)	99 (18.5%)
CSO/Close2	25 (4.4%)	27 (4.8%)	61 (11.1%)	60 (10.9%)
CSO/Remote2	6,261 (11.5%)	79 (14.7%)	48 (8.9%)	52 (9.8%)
	PG, now and/or earlier3:		GAM, now and/or earlier3:
CSO	125 (23.2%)		144 (27.1%)
CSO/Close2	38 (6.8%)		96 (17.4%)
CSO/Remote2	103 (19.1%)		74 (14%)

Note.

Number of responders who have someone close to them who have problems with PG and GAM now and earlier.

1. Number (valid %) of responders who are CSO in at least one relationship.
   Current PG: Four responders are CSO in three relationships, 15 responders are CSO in two relationships, 60 responders are CSO in one relationship.
   Former PG: Four responders were CSO in three relationships, 22 responders were CSO in two relationships, 70 responders were CSO in one relationship.
   Current GAM: Three responders are CSO in four relationships, four responders are CSO in three relationships, 13 responders are CSO in two relationships, 76 responders are CSO in one relationship.
   Former GAM: One responder were CSO in four relationships, five responders were CSO in three relationships, 18 responders were CSO in two relationships, 75 responders were CSO in one relationship.
2. CSO/Close: Father, mother, sibling, partner, and child.
   CSO/Remote: Grandparent, ex-partner, friend, collegue, others.
3. Overlap current/former:
   Of the 125 CSO/PG, 29 were current, 46 was former, and 50 were both current and former CSO. Of the 38 close/PG, 11 were current, 13 were former, and 14 were both current and former CSO. Of the 103 remote/PG, 24 were current, 41 were former, and 38 were both current and former CSO.
   Of the 144 CSO/GAM, 45 were current, 48 were former, and 51 were both current and former CSO. Of the 96 close/GAM, 36 were current, 35 were former, and 25 were both current and former CSO.
   Of the 74 remote/GAM, 22 were current, 26 were former, and 26 were both current and former CSO.

Fig. 1.

Degree of concern per group of Concerned Significant Others (CSO)

Note.

PG = Problematic Gambling. GAM = Problematic Gaming.

Degree: Score on scale where 1 = not concerned, 2 = a little concerned, 3 = somewhat concerned, 4 = Very concerned, 5 = extremely concerned.

Numbers on bars are number of responders per answer.

Prevalence of CSO/GAM

The lifetime prevalence of CSO/GAM across the two categories was 27.1%. While 17.7% were current CSO/GAM, 18.5% were former CSO/GAM. The person of concern was most frequently a friend (28 current, 30 former), child (32 current, 26 former), or sibling (22 current, 24 former). For further analyses, 96 persons (17.4%) were categorized as close/GAM, while 74 persons (13.9%) were categorized as remote/GAM. A total of 25 participants belonged to both categories, i.e., reporting concern of GAM in both close and remote relationships. Most of the CSO/GAM were “not” or only “a little” concerned about the person with GAM. See Fig. 1.

Prevalence of being CSO of both PG and GAM

There were some overlaps between CSO/PG and CSO/GAM. Of the 144 responders who had ever been a CSO/GAM and the 125 responders who had ever been a CSO/PG, 59 belonged to both groups. This means that 47% of CSO/PG were also CSO/GAM and 41% of CSO/GAM were also CSO/PG.

Correlates of CSO/PG

Correlations between independent variables are shown in Table 3.

Table 3.

Correlations between independent variables (N = 575)

Variable	Age	Sex1	Income2	Education3	PG	GAM	AUDIT-Cr	DUDIT-Cr	Mini-IPIP
									N	E	O	A	C
Age
Sex1	.04
Income2	.35***	.15***
Education3	.20***	−.10*	.49***
PG	−.05	.09*	−.06	−.03
GAM	−.23***	.12**	−.18***	−.10*	.17***
AUDIT-Cr	−.10*	−0.11**	.12**	.08*	.04	.03
DUDIT-Cr	−.19***	.05	−.07	−.04	.09*	.07	.04
Mini-IPIP
N	−.21***	−.28***	−.23***	−.08	.06	.08	.05	.07
E	.01	−.03	.06	.13**	−.08	−.04	.18***	.06	−.09*
O	−.19***	.09*	−.03	.04	−.02	.12**	.04	.13**	.09*	.23***
A	−.02	−.36***	−.11**	.06	−.06	−.11*	.10*	−.06	.20***	.23***	.13**
C	.29***	−.15***	.21***	.20***	−.08	−.14***	−.03	−.09*	−.16***	.10*	−.16***	.13**
Mean (SD)	44.05 (17.31)	1.46 (0.50)	4.90 (3.02)	4.89 (3.02)	0.03 (0.18)	0.06 (0.24)	0.57 (0.50)	0.06 (0.23)	10.98 (3.60)	13.56 (3.52)	13.93 (3.29)	16.56 (2.67)	15.76 (3.01)
n	575	575	561	575	560	560	560	558	570	567	568	569	568
Skewness	0.01		0.05				0.23		0.13	−0.31	−0.12	−0.70	−0.61
Kurtosis	−1.27		−0.93				−0.27		−0.55	−0.33	−0.53	0.01	−0.22

Note.

1. Sex: Positive indicate correlation with being male, negative with being female. Mark: Stricter cut-off for women on AUDIT-Cr.
2. Income: Gross annual income where 0 < 100,000 NOK; 1 = 100,000 – 199,000 NOK; 2 = 200,000 – 299,000 NOK; etc. up to 10 ≤ 1,000,000 NOK.
3. Education: Ordinal scale treated as continuous variable, where 1 = Not completed primary school, 2 = Primary school (9 or 10 years), 3 = High school/vocational, 4 = University, lower degree, 5 = University, higher degree, 6 = PhD.

PG = score >2 on the Problem Gambling Severity Index (PGSI).

GAM = Problematic Gaming. Score ≥3 sometimes on at least four of seven items on Game Addiction Scale for Adolescents (GASA).

AUDIT-Cr = score ≥4 for women and score ≥5 for men on Alcohol Use Disorder Identification Test – Consumption.

DUDIT-Cr = Other answer than “never” on first question of Drug Use Disorders Identification Test – Concise.

Mini-IPIP = Mini – International Personality Item Pool.

N = Neuroticism, E = Extraversion, O = Openness, A = Agreeableness, C = Conscientiousness.

Pearson Correlation, (two-tailed) * p-value <.05. ** p-value <.01. *** p-value <.001.

Close/PG was associated with higher Neuroticism in the bivariate analyses, with a moderate effect size (see Table 4 for results from bivariate analysis). The final regression model including age, sex, and Neuroticism explained 2.6% of the variance (see Table 5 for the multiple regressions).

Table 4.

Bivariate analyses per CSO-group (N = 575)

Variable	Full sample (N = 575)	Close/PG (n = 38)	p/ES1	Remote/PG (n = 103)	p/ES1	Close/GAM (n = 96)	p/ES1	Remote/GAM (n = 74)	p/ES1
Age, M years (sd)	44.05 (17.31)	41.03 (14.81)	.26/0.19	41.46 (17.23)	.17/0.15	39.86 (15.04)	.007/0.28	34.73 (15.69)	<.001/0.59
Male, n (%)	266 (46.3)	13 (34.2)	.16/0.07	59 (57.3)	.02/0.10	47 (49.0)	.63/0.03	43 (58.1)	.04/0.09
Income2, M level (sd)	4.90 (3.02)	5.03 (2.68)	.79/0.05	4.47 (2.94)	.14/0.17	5.19 (3.16)	.30/0.12	3.34 (2.82)	<.001/0.59
Education3, M level (sd)	4.89 (3.02)	3.58 (0.98)	.44/0.13	3.47 (1.08)	.008/0.29	3.80 (1.07)	.25/0.13	3.62 (1.07)	.50/0.08
PG, n (%)	18 (5.5)	1 (4)	14/0.01	11 (57.9)	.002/0.15	6 (12.2)	.04/0.08	6 (15.4)	.02/0.10
GAM, n (%)	33 (5.7)	3 (7.9)	.494/0.02	11 (10.7)	.06/0.09	10(10.6)	.07/0.09	18 (24.3)	<.001/0.30
AUDIT-C risk, n (%)	216 (38.6)	17 (44.7)	.50/0.04	40 (38.8)	1/<0.01	36 (38.3)	1/<0.01	27 (36.5)	.81/0.02
DUDIT-C risk, n (%)	32 (5.7)	4 (10.5)	.304/0.06	9 (8.7)	.23/0.06	8 (8.5)	.27/0.18	10 (13.5)	.007/0.13
Mini-IPIP (Cronbach's α/skew/kurtosis for full sample), sum (sd)
Neuroticism (0.73/.13/−0.56)	10.99 (3.61)	13.11 (3.01)	<.001/0.64	10.98 (3.90)	.98/<0.01	11.36 (3.35)	.29/0.12	11.41 (3.58)	.33/0.12
Extraversion (0.80/−0.30/−0.33)	13.55 (3.52)	13.46 (3.55)	.89/0.02	13.56 (3.40)	.995/<0.01	13.46 (3.46)	.94/0.01	14.30 (3.28)	.04/0.26
Openness (0.72/−0.12/−0.53)	13.92 (3.29)	13.59 (3.74)	.50/0.12	13.85 (3.37)	.69/0.04	14.30 (3.06)	.26/0.13	14.85 (3.36)	.01/0.31
Agreeableness (0.70/−0.71/.04)	16.54 (2.68)	16.70 (2.53)	.75/0.06	16.38 (2.75)	.41/0.09	16.87 (2.42)	.24/0.14	16.49 (2.52)	.77/0.04
Conscientiousness (0.72/−0.61/−0.23)	15.74 (3.11)	15.11 (2.85)	.18/0.23	15.61 (2.92)	.71/0.04	15.20 (3.28)	.06/0.22	14.77 (3.28)	.005/0.35

Note.

CSO = Concerned Significant Others.

Close/PG = respond “yes” to concern about gambling of parent, current partner, sibling or child.

Remote/PG = respond “yes” to concern about gambling of grandparent, ex-partner, other family, friend, collegue or others.

Close/GAM = respond “yes” to concern about gaming of parent, current partner, sibling or child.

Remote/GAM = respond “yes” to concern about gaming of grandparent, ex-partner, other family, friend, collegue or others.

PG = Score ≥3 on the Problem Gambling Severity Index (PGSI).

GAM = Score ≥3 sometimes on at least four of seven items on Game Addiction Scale for Adolescents (GASA).

AUDIT-C risk = score ≥4 for women and score ≥5 for men on Alcohol Use Disorder Identification Test – Consumption.

DUDIT-C risk = Other answer than “never” on first question of Drug Use Disorders Identification Test – Concise.

Mini-IPIP = Mini – International Personality Item Pool.

1. ES = Effect size. For t-tests, Cohen's d with pooled standard deviation is reported. For chi-square tests, phi coefficient is reported.
2. Income: Gross annual income where 0 < 100,000 NOK; 1 = 100,000 – 199,000 NOK; 2 = 200,000 – 299,000 NOK; etc. up to 10 ≤ 1,000,000 NOK.
3. Education: Ordinal scale treated as continuous, where 1 = Not completed primary school, 2 = Primary school (9 or 10 years), 3 = High school/vocational, 4 = University, lower degree, 5 = University, higher degree, 6 = PhD.
4. Assumption regarding frequency in any cell above 4, is violated.

Bold indicate that score differed for those in the group compared to those not in this group on independent sample t-test for continuous and chi-square test for independence for categorical variables (p < .05), unadjusted.

Table 5.

Multiple regression per CSO-group with age, sex, and independent variables significant at p < 0.05 in bivariate analysis included

5.1) Close/PG (38 in CSO- group)
Predictor	B	SE(B)	β	t	p	95% CI (B)	VIF
Age	.000	.001	−.015	−0.344	.73	[−.144, .108]	1.049
Sex1	−.012	.022	−.024	−0.558	.58	[−.055, .031]	1.085
Neuroticism	.010	.003	.148	3.311	<.001	[−.004, .016]	1.135
Constant	-.018	.064		−0.285	.776
Model Summary: R = .16, R2 = .026, Adjusted R2 = .02, F (3, 555) = 4.854, p = .002
5.2) Remote/PG (103 in CSO-group)
Predictor	B	SE(B)	β	t	p	95% CI (B)	VIF
Age	−.001	.001	−.036	−0.827	.41	[−.003, .001]	1.047
Sex1	.066	.034	.083	1.920	.06	[−.002, .133]	1.022
Education2	−.038	.018	−.095	−2.165	.03	[−.073, −.004]	1.053
PG	.296	.096	.133	3.079	.002	[.107, .484]	1.012
Constant	.264	.091		2.904	.004	[.085, .442]
Model Summary: R = .20, R2 = .04, Adjusted R2 = .03, F (4, 524) = 5.57, p < .001
5.3) Close/GAM (96 in CSO group)
Predictor	B	SE(B)	β	t	p	95% CI (B)	VIF
Age	−.002	.001	−.102	−2.392	.02	[−.004, .000]	1.004
Sex1	.017	.033	.022	0.522	.60	[−.047, .081]	1.010
PG	.153	.092	.071	1.654	.10	[−.029, .334]	1.012
Constant	.243	.064		3.803	<.001	[.117, .368]
Model Summary: R = .13, R2 = .017, Adjusted R2 = .011, F (3, 540) = 3.098, p = .026
5.4) Remote/GAM (74 in CSO group)
Predictor	B	SE(B)	β	t	p	95% CI (B)	VIF
Age	−.002	.001	−.078	−1.690	.09	[−.003, .000]	1.289
Sex1	.056	.030	.081	1.888	.06	[−.002, .114]	1.091
Income3	−.016	.005	−.139	−3.083	.002	[−.026, −.006]	1.208
PG	.088	.082	.045	1.063	.29	[−.074, .249]	1.051
GAM	.347	.064	.236	5.445	<.001	[.222, .472]	1.121
DUDIT-Cr	.107	.063	.072	1.707	.09	[−.016, .230]	1.058
Extraversion	.011	.004	.107	2.514	.01	[.002, .019]	1.090
Openness	.002	.005	.017	0.389	.70	[−.007, .011]	1.147
Conscientiousness	−.003	.005	−.023	−0.517	.61	[−.012, .007]	1.179
Constant	.047	.121		0.391	.70	[−.191, .286]

Model Summary: R = .388, R² = .150, Adjusted R² = .135, F (9, 508) = 10.00, p < .001.

Note.

1. Sex: Positive indicate correlation with being male, negative with being female.
2. Education: Ordinal scale treated as continuous variable, where 1 = Not completed primary school, 2 = Primary school (9 or 10 years), 3 = High school/vocational, 4 = University, lower degree, 5 = University, higher degree, 6 = PhD.
3. Income: Gross annual income where 0 < 100,000 NOK; 1 = 100,000 – 199,000 NOK; 2 = 200,000 – 299,000 NOK; etc. up to 10 ≤ 1,000,000 NOK.

CSO = Concerned Significant Others.

Close/PG = respond “yes” to concern about gambling of parent, current partner, sibling or child.

Remote/PG = respond “yes” to concern about gambling of grandparent, ex-partner, other family, friend, colleague or others.

Close/GAM = respond “yes” to concern about gaming of parent, current partner, sibling or child.

Remote/GAM = respond “yes” to concern about gaming of grandparent, ex-partner, other family, friend, colleague or others.

PG = Problematic Gambling, score >2 on the Problem Gambling Severity Index.

GAM = Problematic Gaming, at least four items scored at least “sometimes” on Game Addiction Scale for Adolescents.

DUDIT-C r = Drug Use Disorders Identification Test – Concise, score>0 on first item.

Personality traits was measured by Mini – International Personality Item Pool.

p < .05 in bold.

Also, remote/PG was associated with male sex, lower education and PG in the bivariate analyses. In the multiple regression analysis including sex, age, education, and PG as independent variables, only education (p = 0.03) and PG (p = 0.002) remained significant predictors for remote/PG. The model explained 4% of the variance.

Correlates CSO/GAM

Close/GAM were associated with younger age and PG in bivariate analyses. In the multiple regression, only younger age (p = 0.017) remained a significant predictor for Close/GAM. The model only explained 1.7% of the variance.

Additionally, remote/GAM were associated with younger age, male sex, lower income, PG, GAM, and drug risk. Further, remote/GAM was positively associated with Extraversion and Openness and inversely related with Conscientiousness in bivariate analyses. These are independent variables which correlate (see Table 3). In the multiple regression, income (p = 0.002), problematic gaming (p < 0.001), and Extraversion (p = 0.01) remained significant predictors for remote/GAM. No problem with multicollinearity were detected. The model explained 15% of the variance.

Discussion

Our estimated lifetime prevalence of CSO/PG (23.2%) is higher than previously reported in Norway (Wenzel et al., 2008), Sweden (Svensson et al., 2013), and Finland (Salonen et al., 2015). This might be explained by a broader definition of being a CSO in the current study, including asking specific questions for each possible relationship, and not just “someone close”. One might also expect higher prevalence when asking if someone close has problems with PG (as is the case in the Nordic studies) rather than if the respondents are personally affected by someone's PG, as was examined in the prevalence studies from Australia (Dowling et al., 2025). Development of international consensus about how to operationalize CSO should be prioritized. In our study, responses to the questions about concern indicate that a several CSO-respondents are not very concerned (see Figure 1).

Only 14.7% reported being a current CSO/PG which is roughly five times the last-year prevalence of moderate risk gamblers (2.3%) and PG (0.6%) in Norway (Pallesen et al., 2023), which is consistent with the finding that PG affects approximately six other people (Goodwin et al., 2017). The CSO/PG most frequently reported having a sibling, ex-partner, friend, or colleague with PG. Notably, more reported having an ex-partner than a current partner with PG. In line with in-group bias or favoritism, this might reflect a tendency to trivialize problems of current partners or that such problems are stilled hidden (Holdsworth, Nuske, Tiyce, & Hing, 2013), while emphasizing ex-partners problems that might led to relationship break-up (Black, Shaw, McCormick, & Allen, 2012).

Close/PG did not differ from others on demographic variables or related to being at risk concerning their own gambling, gaming, alcohol or illegal drug use, while remote/PG had lower education and were more often risky gamblers themselves. Given that a third of the remote/PG were friends of the gambler, it is conceivable that they have more risky gambling and lower education, as friends often resemble each other (Laursen, 2017). The differences between sexes were not significant, which is surprising considering that males gamble more often than females (Tran et al., 2024).

Neuroticism was elevated among close/PG, while no differences in personality traits were found among remote/PG. High Neuroticism might reflect state-artifacts of the emotional burden of having a close family member suffering from PG (Riley et al., 2021), and/or common genetics with the family member with PG since persons with PG consistently seem to have elevated scores on this trait (Strømme et al., 2021). It is also possible that people high in Neuroticism are more easily concerned about the gambling of close ones, an explanation that cannot be ruled out as the current study is cross-sectional and hence proscribed from identifying causal direction.

Prevalence of CSO/GAM was 27.1%, which is lower than the 34.4% estimate reported for partners in an Australian study (Szász-Janocha et al., 2023). Gaming is a common activity in younger age groups, which might explain why the person of concern was most commonly a child, sibling, or friend. In addition, the blurry boundary between normal and problematic gaming might explain why it's more often close family members who report being a CSO/GAM, as one might need to be close to the person to realize the negative consequences of gaming (Pallesen, Lorvik, Bu, & Molde, 2015).

Close/GAM was associated with younger age. Bivariate analysis indicated that they were more likely to experience GAM themselves, but this was not a significant finding in the multiple regression analysis. The high prevalence combined with insignificant findings on characteristics of close/GAM, indicate that gaming of concern happens across many sociodemographic groups.

Remote/GAM were most often friends of the gamer and associated with several independent variables in the bivariate analysis, including younger age, male sex, and lower scores on Conscientiousness. In the final adjusted regression model, remote/GAM was associated with lower income, risk of GAM and illegal drug use, and elevated scores on Extraversion. It is reasonable that these are characteristics they share with the gamer, as friends often resemble each other (Russell, Langham, & Hing, 2018). However, the elevated score on Extraversion is somewhat surprising and contrasts with the stereotypical isolated gamer (Kowert, Festl, & Quandt, 2013). One explanation though is that people high on Extraversion have a larger social network than those with lower score, which by the share number of people in their network would increase the probability of knowing someone with a gaming problem. It was also surprising that we did not find robust sex differences, since men/boys game more often than women/girls, and CSO/GAM in previous studies are more often women/girls (Szász-Janocha et al., 2023).

Furthermore, as posited by the problem behavior theory (Jessor & Jessor, 1977) the overlap between CSO groups, the correlations between GAM, low income, low education and PG, and the several associations between these variables and remote/GAM suggest the presence of PG and GAM clusters within social networks.

Strengths, limitations, and implications

The use of a population-based sample is a major strength of the current study. However, only about 1/3 responded. Falling response rates is an increasing challenge in surveys (Stedman, Connelly, Heberlein, Decker, & Allred, 2019), and several strategies have been suggested to compensate for this (Sammut, Griscti, & Norman, 2021). All responders were given a gift card to increase the response rate, still, this did not increase the response rate as much as we hoped for. However, our evaluation of the background variables (age, sex, income and education level, see Table 1) of responders, suggest that they reflect the general Norwegian population on basic demographic variables (Statistics Norway, 2024a; 2024b). Still, selection bias cannot be ruled out. Although the sample size was reasonably large, it should be noted that low variability in some of the independent variables (e.g., PG) may have limited the statistical power.

The study investigated novel research questions. To the best of our knowledge, this is the first population-based survey on CSO/GAM and the first study of personality in CSO. As this is a cross-sectional study, it is not possible to conclude on causality or directionality between traits and CSO status. GAM is a relative new research field, and is often compared to studies on PG. An asset of this study is that respondents were given the same questions regarding CSO/PG and CSO/GAM.

As an exploratory study, no correction was made for multiple testing which may increase the risk of Type I error. Therefore, findings from the bivariate analyses should be interpreted with caution. As a cross-sectional study, causality and directionality cannot be inferred. The latter requires experimental and longitudinal research. Although statistically significant, explained variance in the multiple regressions was low and future studies should thus explore other potential predictors of CSO status, such as psychological distress and family dysfunction (Dowling et al., 2025; Szász-Janocha et al., 2023).

There are strengths and weaknesses with using questionnaires. Although anonymous, responders might still provide social desirable answers (Van de Mortel, 2008). “Yes-saying”/acquiescence bias (Bowling, 2005) might lead to a higher endorsement of the CSO items, since the same question was asked for each relationship, currently and previously, for both PG and GAM (fatigue bias; “ok, then, s/he was gambling a little too much”).

The rationale for merging “now” and “earlier” answers was that previous problems might influence the current relationship (Holdsworth et al., 2013). The merger of current and former concerns also compensated for the limitation that “now” and “earlier” was not defined in terms of specific time frames in the questionnaire. Still, it cannot be ruled out that concerns wane with time, thereby affecting CSO status. Further, we decided not to include ex-partners in the close CSO category, as the burden of being a CSO to ex-partners was judged to be less than being a CSO in a current relationship. On the other hand, CSO of ex-partners report a high level of problems (Castrén et al., 2021); and we regard it as likely that people underreport problems when it concerns their current partner, and overreport when it is about an ex-partner. Categorizing ex-partners in the remote category can be debated, as having common child(ren) with an ex-partner (i.e., still co-parent or shareing financial obligations) could explain why relative many were concerned. This should be addressed in future studies.

The prevalence of CSO/GAM should be investigated in other populations. The interaction between personality traits and CSO should be elucidated using longitudinal designs. Prevention and treatment of GAM should consider the social network of the person suffering from GAM.

Funding Statement

Funding sources: TThe Norwegian Competence Center for Gambling and Gaming Research (SPILLFORSK) at the University of Bergen founded the data collection.