Search and Selection Results
The database searches returned 329 papers for review and 7 of these were excluded as duplicates. Analysis of titles and abstracts led to a further 59 exclusions for not discussing gambling, 67 exclusions for not discussing harms, or for discussing harms from other related sources such as gambling adverts, and 3 for investigating non-human test subjects. The remaining 193 studies were reviewed in full, resulting in 37 studies being excluded because they were articles, book chapters, letters or editorial pieces that discussed the topic without providing new information. Studies discussing only harms to others resulted in 13 exclusions, 4 studies were unobtainable, and 11 studies were not available in English. During the full-text review studies previously thought to discuss harms were removed for not discussing specifics (32), or discussing harm minimisation (40) rather than harm analysis, leaving 56 studies remaining.
Main Results
Description of Included Studies
Of the 56 studies included in this review, 14 used interviews, 2 used observation, 4 used focus groups, 26 involved surveys and 10 conducted reviews. Of the 10 reviews, 7 were systematic, and the remainder were narrative. Secondary data analysis was conducted in 14 of the studies, 2 studies used case notes, 1 study used online forum analysis, and 1 used health state valuation vignettes.
The most common funding bodies for this selection of studies were the Ministry of Social Affairs and Health Helsinki (5) and the Victorian Responsible Gambling Foundation (6). In total, the government funded 11 studies, medical institutes and foundations funded 3, gambling focused institutes or trusts funded 17, Colleges and Universities funded 5, research councils and institutes funded 3, and Star City Casino in Sydney funded 1. Of the remaining 16 studies, 11 received no funding and 5 did not declare their funding status.
Risk Severity
Fourteen studies include data on risk severity, which is the measure of behaviour that puts someone at risk of developing a problem with gambling or developing Gambling Disorder. Rintoul, Deblaquiere [34] found that venues looked for signs of problem gambling behaviour such as ‘shouting at the machine or other people in the venue’, appearing depressed, being withdrawn or emotional, excessive sweating, extended play and continuing to gamble with the winnings. Although the venue considered these items to be gambling harms it could be argued that continuing to gamble with the winnings is a predictor of financial harm rather than a harm itself. In 34 hours they observed these signs in all venues and concluded that risk factors for the behaviour included multiple machine use at one time and withdrawing money several times at the venue.
In academic research the risk of developing Gambling Disorder is most frequently measured using the PGSI, a component of the Canadian Problem Gambling Index [35], which categorises an individual as low risk or non-problem (0 score), moderate risk (1-2 score), high risk (3-7 score), or a problem gambler (8+ score). These scores increase the confidence of accurate diagnosis but do not provide any clarity on the nature or severity of harms experienced by the individual. It is therefore important to state again that risk severity cannot be considered a gambling harm, as high-risk scores do not always lead to actual experienced harms or diagnosis. However, Shannon, Anjoul [36] found that the PGSI reliably correlated with 48 indicators of harm. These harm indicators were compiled using data from a scoping literature review, clinical case notes in a treatment setting, and a cohort of seven specialist clinicians.
Atherton and Beynon [37] found that people who self-reported as a problem gambler were twice as likely to see their general practitioner for support with mental health concerns, five time as likely to be hospital inpatients, and eight times as likely to seek counselling. Angus, Anjoul [28] also found that the number of harms experienced increased with PGSI classification, and significantly less non-problem, low risk, or moderate risk gamblers reported harms compared with problem gamblers. Problem gamblers were also more likely to come from the clinical sample, who had significantly greater severity of harms in all domains, and of which 100% reported psychological harms, compared to only 54.69% in the community.
Many of the studies looking into risk severity focused on gathering evidence of the Prevention Paradox. This describes a situation where the majority of cases of a disease come from a population at low or moderate risk of that disease. In the case of gambling this means, the majority of harms are found within the low to moderate risk gamblers, and the minority is found within high risk or problem gamblers.
In one study the prevalence of harms across non-problem (PGSI), or non-pathological (SOGS), gamblers were twice that of problem gamblers in their participant sample [22 ]. Browne and Rockloff [19] found that only 10% of financial harms across the study population were in the problem gambling or pathological gambling groups and that more than 50% of cases where someone sold their belongings to fund their gambling were in recreational or low risk gamblers. In contrast to this, they found that more than 50% of social deviance harms are found within problem gamblers, and the remaining categories of harm were evenly distributed across the severity groups.
Evidence of the prevention paradox was found by several other studies [20, 38, 39], though it was also found that the highest severity harms were generally only present in the problem gambling groups, as defined by the PGSI. Skaal, Sinclair [40] found psychological distress was only associated with problem gambling and Lloyd, Hawton [41] found that self-harm thoughts were associated with problem gambling. Rawat, Browne [20] found that problem gamblers show similar disability weights to those of Bipolar Disorder or alcohol dependence, whereas the low risk group show disability weights equal to moderate anxiety. Disability weights are a health-related measure of quality of life using a ratio scale between 0 and 1, representing ideal health and death. These estimates further suggest that although the majority of harms may be found in the lower risk groups, the more severe harms are only prevalent within problem gambling groups.
Li, Browne [42] found that selling personal items, absence from work or study, reduced performance, poor sleep and extreme distress had the highest correlation with PGSI categories. They also found that reduced spending on essential expenses, absence from work or study, feelings of worthlessness, increased relationship conflict, and feeling like an outcast were the most effective discriminators between the low and high-risk groups.
Shannon, Anjoul [36] found that across both a clinical and non-clinical sample the lowest reported harms were substance use, suicide, bankruptcy, and education problems. They found that the distribution of averaged harm scores was consistent across both samples, excluding reduced savings and decreased happiness, suggesting that psychological and financial harms are the most significant. The most common gambling harms found represented negative impacts on general quality of life and psychological well-being; however, the mean level of harm for the 15 most common indicators in the community sample was less than one empirically defined unit.
Despite these results, there has been some criticism of the PGSI as a measure of gambling behaviour. Victorian Responsible Gambling Foundation [43] categorises the PGSI groups as non-problem gambling, low risk, moderate risk and problem gambling, which is distinctly different to the categories used in some of these studies. Delfabbro and King [31] argued that harms cannot be confidently attributed to gambling in low risk individuals, as some studies consider those who score a 0 on the PGSI as low risk, rather than no risk. They also argue that harms are not appropriately scaled, citing that “suicide is not equal to shame,” and that when controlling for severity low risk categories show very few harms. However, they did find that in financial harm categories the items they considered as more severe, such as selling belongings, were present even in the low risk groups. This may be due to the less affluent socioeconomic status of the low risk category, and it is hard to attribute these financial harms to gambling.
Age
Twenty-one studies include data on age, and several of these found that being younger was associated with a higher risk of experiencing gambling harms [37, 44-50]. In particular one study found that younger age groups (16-34) were at risk of dependence and social harms Canale, Vieno [51], and Ferrara, Franceschini [48] found that as well as higher rates of what they label “problematic gambling”, younger age groups showed a higher comorbidity with other addictions. Desai, Maciejewski [47] found that younger gamblers were more likely to abuse alcohol and other substances, be incarcerated, and be bankrupt and depressed, and older gamblers were more likely to report good subjective general health.
In contrast, Melendez-Torres, Anthony [52] found that harms experienced increased with age; however, they only researched participants attending school who would be categorised in the younger age groups of other studies. Some studies found that younger gamblers were less at risk of financial harms [49, 53], despite one suggesting that they spent more [54]. And Larsen, Curtis [55] found that alcohol use increased with age in lifetime problem gamblers, as defined by the DSM-IV criteria for ‘pathological gambling’, in opposition to the trend seen in a general population.
Two studies found that age had no impact on harm profiles [22, 39], and Lloyd, Hawton [41] found no association between age and gambling-induced thoughts of self-harm. Pitt, Thomas [53] found that children aged 8-16 showed little or no current harms as they were gambling at home with their families, spending small amounts of pocket money, or betting with activities such as push-ups against family members.
Despite the apparent absence of harms in the youngest age groups children developed false beliefs around gambling, such as that skill can be used to win, or that it is necessary for everyone to try gambling at least once. Children were also found to understand how gambling could gain them money. In Breen [56] it was found that youth who were exposed to gambling at a young age were more likely to gamble later in life to increase their income, and that youth who missed school had reduced lifelong aspirations and reduced opportunities.
Further research is needed to understand the distribution of harms across age groups as it was found by Estevez, Herrero-Fernandez [57] that sensation seeking and impulsivity were high in young gamblers. Anxiety, depression and psychoticism were partially mediated by impulsivity, and somatisation, obsessive-compulsive behaviour, interpersonal sensitivity, paranoid ideation and hostility were perfectly mediated. Hubert and Griffiths [58] found that online gamblers are more likely to be younger, and Bergh and Kuhlhorn [59] found that individuals aged 20-34 gambled for almost twice as long per session compared to participants over 35. Breen, Hing [54] discovered that younger gamblers preferred to play poker or other card games, while Ferrara, Franceschini [48] found that younger gamblers were more likely to be sports bettors. These additional variables could explain some of the variance in harm presentation and help to target appropriate interventions, as Ferrara, Franceschini [48] found that 0.2-12.3% of adolescents in their study met the criteria for a diagnosis of Gambling Disorder depending upon country of residence.
Gender
Twenty studies examined gender, and only a few of these found evidence that harms affected men and women differently [22 , 44 , 45, 59]. Where researchers found differences between genders, these were explainable by other factors. Though some studies show men have a higher prevalence of harms than women, [46, 48, 60, 61], we know that on average men gamble more frequently and spend more money when gambling [38, 51, 52, 62]. Hing, Breen [63]found that females from small villages and men from towns were more likely to be heavy commercial gamblers, however the harms suffered were the same and so this was likely due to usage level rather than gender.
Livazovic and Bojcic [64] found that males in Croatia scored significantly higher on psychological, social, and financial consequences than females. However, they also scored significantly higher on risk behaviour and were more likely to score as a problem gambler on the Canadian Adolescent Gambling Inventory. Raisamo, Makela [38] found that although the prevalence of harms was higher in males, when controlling for frequency of play and amount spent gender was no longer a significant factor. Moreover, Splevins, Mireskandari [65] found that men started gambling earlier than women did and found it more exciting. This led to increased spending and therefore an increased risk of harms such as substance use and interpersonal conflicts.
Despite this some studies have suggested key differences in how gambling harms present between genders. In Singapore, Goh, Ng [66] reported that “tentative evidence… points to the risk of child neglect when the problem gambler is the mother.” They also found that verbal abuse was most commonly males towards their mother, but found no difference in cases of physical abuse between genders. McCarthy, Thomas [67] found that women were more likely to report mental health comorbidity than males, however causality is not discussed, and Raisamo, Kinnunen [68] found that although the most common harm was guilt for both genders, the second most common was disrupted schoolwork for females and conflict with friends for males.
One key study that found an opposing result was Salonen, Alho [69] who found that while gambling was more common in young males, women displayed an increase in specific harms between 2011 and 2015 where men did not. This particular study also looked at attitudes towards gambling and found that while female attitudes were generally negative over the age of 25, male attitudes were generally positive for all but the 15-17 age range. As the study used self-report data these differences in attitude could have affected how individuals reported harms.
Culture
Eleven studies include data on culture and the majority of these discuss Australia and New Zealand. The included studies largely focus on single groups or comparing indigenous people and migrants to a society, so there are currently significant gaps that future studies may address. Although direct comparison is therefore not possible, some estimates and inferences can be made.
Kolandai-Matchett, Langham [70] found that Pacific New Zealand people experienced gambling through collectivist cultural values, meaning that additional harm dimensions were present. Some of the listed cultural harms include a loss of belonging or isolation, shame; loss of the community’s respect; disruption of trusting relationships; transference of communal responsibilities; and an overall loss of social cohesion. In a quotation from one of the interviewed participants, the researchers noted that the wider collective might exclude non-present or non-contributing members of the society.
Breen [56] studied indigenous Australians and found that one key harm was the neglect of children when a parent gambles, and the eldest daughter would become their main caregiver. They also found that many people would gamble within a group, increasing their behaviour, but they would feel shame from losses and from the potential gossip within their close community, and Breen, Hing [54] also found that a large concern for Indigenous Australians was the spread of a harms impact throughout the community. The range of harms shown within indigenous Australian communities included gambling away their pensions, relationship issues [54], betting above their means, guilt and shame, chasing losses [54], and to a lesser extent financial harms like borrowing money or selling items, and health problems. However, it is important to consider whether betting above their means, chasing losses, and borrowing or selling can be considered harms or simply predictors of harm.
McCarthy, Thomas [67] suggested that women from ethnic minorities, indigenous communities and specifically Maori and Pacific women in New Zealand were more vulnerable to gambling harms than European women were. Melendez-Torres, Anthony [52] also found that participants from white ethnicities were less likely to feel guilt from gambling, and a non-white British background was associated with more harms. Ferrara, Franceschini [48] found that non-white males were most at risk of developing a gambling problem and, in the UK, Wardle, Bramley [71] found that although migrants were less likely to gamble they were more likely to experience harms than individuals born in the country. By contrast, Currie, Hodgins [46] found that white men were more likely to report harms, and are more likely to gamble overall, with a higher frequency and higher spends.
The final study that investigated culture is that of Goh, Ng [66] who found that families in Singapore were at risk of acute financial harms when the problem gambler was a parent. Most households suffered double financial harms through loss of income and large debt. When the gambler was a mother without income, they found that the father would leave employment to care for the children, resulting in an income reduction for the entire household. Goh, Ng [66] also found that many people in Singapore viewed gamblers as self-centred, and siblings would often give up on them, rather than accepting problem gambling as an illness.
Gambling Behaviour
One of the most predictable influences on gambling harms is the specific gambling habits and behaviours of the individual. Fourteen studies include data on gambling behaviours and several studies have agreed that a higher frequency of play, and higher amount of spending per session, leads to more harms [38, 45, 46, 49, 51, 72]. In particular, Nigro, D'Olimpio [61] reported that higher involvement in gambling was associated with higher levels of depression and self-reported memory impairment. Hing, Breen [73] also reported that heavy gambling led to participants spending their entire pay or pension, borrowing money, and playing all night and day in both commercial and card games. Once again we need to consider whether borrowing money can be considered a harm, however in partnership with spending their entire pension it is clear that some financial mismanagement is present.
Game choice also affected harms, as heavy commercial gamblers reported more harms than card players did. These included debts, relationship issues, loss of home, missed bills, no food and low nutrition, child neglect and abuse from stress, lying, domestic violence, depression, suicidality, criminality, and selling their belongings.
Some other studies have found links between the quantity of harms and the choice of game type. Castrén, Perhoniemi [45] found that six out of twelve game type predictors were associated with more harmful consequences, including scratch games, betting, slot machines, non-poker online games, online poker, and non-monopoly games. They found that lottery play caused the lowest number of harms, and this finding is consistent with findings reported by Currie, Hodgins [46] who found that frequency of play on lottery games did not increase the harms experienced, whereas electronic gambling machines, ticket gambling, bingo and casino games did. Ronzitti, Soldini [50] also concluded that Fixed Odds Betting Terminals and casino tables were associated with the highest scores on the PGSI, and gambling machines showed high PGSI scores compared to other methods of play, despite similar play times.
Two studies within the search looked at motivations for gambling, and although Browne, Hing [44] found no link between motivation of play and harms, Lee, Chung [62] found that excitement, escape and challenge motives were linked with positive outcomes, but financial motivation led to harms. Lloyd, Hawton [41] also found that self-harm thoughts were associated with money as a motivator but was negatively associated with enjoyment motivations. Within this category it is also important to discuss factors such as player skill level and whether gambling is taking place in a group or individually, however none of the studies included in this review examined these specific areas.
Online vs. Offline Gambling
As well as specific game type five studies look at the broader categories of online or offline gambling. Castrén, Perhoniemi [45] found only a weak link between online gambling and an increase in harms, however Gainsbury, Abarbanel [16] found that online gamblers tended to have higher PGSI scores. Moreover, online gamblers showed more variation in their choice of game and gambling behaviours, and in comparison to offline gamblers, they showed no preference for skill-based games.
Yani-de-Soriano, Javed [74] found that online gambling was associated with binge drinking, cigarette smoking and an increased risk of developing problem gambling. However, they did not find a link between what they labelled as internet addiction and gambling. Hubert and Griffiths [58] also found a link between online gambling and alcohol dependence, and they discovered that online gamblers were less likely to have jobs, children and a stable relationship, leading to unemployment and less money later in life. They further found that online gamblers were less able to control impulsivity and frustration, but despite this, they had fewer suicidal thoughts than offline gamblers, although actual suicide attempts were comparable in both groups.
Feelings of shame appeared to be lower in online gamblers relative to offline gamblers [75], and it was suggested that online gamblers feel less judged since their behaviours could be more secretive and private. Despite these reduced feelings of shame, Fulton [75] observed that secretive gambling increased financial harms due to the likelihood of concealed debt; and by living a double life secretive gamblers experienced increased stress, relationship conflicts, and emotional deterioration.
Socioeconomic Status
There were eleven studies examining the socioeconomic factors that influenced the presentation of gambling harms, and all but one study concluded that less affluent socioeconomic groups are more at risk of experiencing harms than more affluent groups. Melendez-Torres, Anthony [52] found that households that are more affluent were associated with more gambling behaviour and subsequently more harms. In comparison Ferrara, Franceschini [48] found that individuals in routine socioeconomic groups were at the highest risk of developing an addiction to gambling, and Currie, Hodgins [46] concluded that participants who reported harms were more likely to be in a lower income bracket, and to have received no further education than high school. Similarly, Atherton and Beynon [37] found that lower income households spent a higher proportion of their income on gambling, and were more likely to bet more than they could afford to lose, and Angus, Anjoul [28] found that clinical participants had significantly lower incomes than a community sample and a higher proportion of them reported harms.
Skaal, Sinclair [40] reported that urban residents were more likely to report psychological distress and be at a high risk of problem gambling on the PGSI. In comparison, harms were associated with employment status in peri-urban areas, with unemployment doubling the risk of problem gambling. Browne, Hing [44] found that income and education level are indirect risk factors of harm, and Lloyd, Hawton [41] found that gambling related thoughts of self-harm were more prevalent in the unemployed.
Lee, Chung [62] reported that married people with a low income are more likely to be financially motivated to gamble, and this motivation could be associated with greater harms. Lloyd, Hawton [41], however, found no link between problem gambling and marital status.
In an apparent contrast with other results, Tu, Gray [76] found that people in managerial or professional occupations appeared more likely to participate in gambling than people in routine occupations. However, although gambling rates in the most affluent groups dropped during times of recession, the rates within deprived communities did not, suggesting that less wealthy people may be more likely to gamble in times of economic stress. When controlling for confounding variables the most deprived groups were 4.5 times as likely to experience a gambling-related argument or money issues than people living in the least deprived areas were.
Other Factors
Studies that examined unique factors affecting harms include Jeffrey, Browne [77] who compared gamblers with their partners and found that gamblers were more likely to report individual harms that affect themselves. They also found that gamblers identified a wider range of harms and were better at identifying harms than their partners. This suggests that as well as experiencing more harms than affected others, gamblers are more aware of those harms and therefore may be more consciously impacted. Li, Browne [42] also found that harms in all areas accumulated more quickly in gamblers than in affected others.
Langham, Russell [78] found that an individuals’ sense of coherence correlated strongly with gambling harms. Sense of coherence is the extent to which someone feels confident in the predictability of his or her environment, and that things will generally turn out as expected.
Binde [79] looked specifically at harms in the context of an individual’s place of work and found that signs of a problem gambler at work included excessive talk on gambling, gambling during breaks or during work time, borrowing money from colleagues, poor work performance, lateness or absence and more. Despite this, they did not compare different levels of employment or types of job role that could have provided a more thorough view of gambling harms and employment. Also, borrowing from colleagues should only be considered a gambling harm if the money is not paid back or the transaction causes a strain on the working relationship.
May-Chahal, Humphreys [80] investigated harms within the British prison population and found that although the prevalence of problem gambling in terms of the PGSI was higher in prisons, the prevalence of gambling behaviour prior to incarceration was significantly lower. They found that there was no link between PGSI score and criminal career, and no statistical link between gambling and drug use, however high rate offenders in their mid-20s were 5.3 times more likely to be frequent loss chasers than other categories. May-Chahal, Humphreys [80] also found that occasional gamblers were less likely to use alcohol or drugs in prison, with nearly 2/3 of the problem-gambling group abstaining completely from substance use. The researchers suggest that this may be because the individuals’ ‘addiction needs’ are being met by their gambling behaviour.
Three studies reported on how student status affected gambling harms, with Livazovic and Bojcic [64] reporting that high-achieving students reported less psychological harms, and vocational students were significantly more at risk of harms than other students. However, there was only a weak correlation between success in school and harms from problem gambling. Melendez-Torres, Anthony [52] also found that a reduced feeling of belonging at school was associated with more harms, but also higher rates of gambling, and Apinuntavech, Viwatwongkasem [81] found that average GPA was significantly lower in those who gambled, though only by a small margin. They also reported that students who had gambled were more likely to feel guilt, lie, experience depression, perceived poor health, and insomnia. With some students reporting substance use to manage stress, school absences and considering suicide.
Four studies also looked at how gambling harms present within a family unit, or how an individuals’ home life may influence their gambling. Anderson, Rempusheski [82] examined senior family members and found that co-dependency within the family, where each person was expected to bail the other out, caused relationship breakdowns, stress and tension. Some participants expressed shame at spending their children’s trust funds or savings; however, participants also reported using the addiction model to neutralize shame and guilt. Ferrara, Franceschini [48] and Larsen, Curtis [55] both found that an individuals’ home life affected their gambling habits. Ferrara, Franceschini [48] stated that having separated parents increased the risk of gambling later in life, and Larsen, Curtis [55] found that the odds of showing one or more addictive behaviours increased in households without children. Perhaps suggesting that the presence of a traditional nuclear family unit is a protective factor against gambling harms. Alternatively, Livazovic and Bojcic [64] found that family life and the parents’ level of education both had no significant effect on harms experienced, though they did affect the likelihood of taking part in risky behaviour.