Helping Professionals

by Jackie Ferris

When comparing prevalence rates, it is important to pay attention to the instrument used (typically SOGS or DSM-IV), the period the survey covered (e.g., was it gambling and gambling-related problems in the last year, or lifetime?), and whether or not the survey included youth (under 18 years of age). These factors are important because they can bias the data one way or another. For instance, the SOGS typically categorizes twice as many people as (probably) pathological as the DSM-IV. With the SOGS, it is also important to note which cutoff scores were used. The standard is that a score of 5 or more falls into the “probably pathological” category. Those with scores of 3-4 are labelled “problem” gamblers. Measures based on the last year are more conservative than those that ask about behaviour over a lifetime (for obvious reasons). Studies that include youth will typically have a higher prevalence of problem gambling, and these studies often use a revised version of the SOGS, the SOGS-RA that scores 4 or more as “problem” and 2-3 as “at-risk.”

Table 2, provides a sample of some of the results of problem gambling prevalence surveys. This list is by no means comprehensive, and is intended to provide an idea of the range of prevalence rates across Canadian, North American and other international jurisdictions. The table includes only current rates (based on the last 12 months) unless otherwise specified. table 2 shows only studies including adults 18 and over, with the exception of the Swedish study, which included adults 15-74. The study in Washington State separated out several age groups, and the results are included in this table because they show how much higher gambling problem prevalence rates are for young adults (18-24).

Table 2. Comparison of prevalence rates across jurisdictions between people affected by problem gambling (PG) (scores of 3-4 on SOGS) (%) and Probable Pathological Gamblers (PPG) (%)

Table 3, shows the prevalence of gambling problems for youth, typically 13 to 17 years old. Although the “problem” category used in the adolescent surveys uses a score of 4 or more on the SOGS-RA as a cut-off, rates of gambling problems for youth are clearly higher than for adults.

The regional differences in results can be explained in part by differences in the availability of gambling between jurisdictions. In general, greater availability in terms of number of outlets and types of gambling means higher rates of problem gambling. This is particularly true with continuous play games (VLT's or slot machines, for example). In the United States, for instance, prevalence rates tend to be highest in states where the availability of legal gambling is high or has increased rapidly, and also within a 50-mile drive of a casino. Some evidence suggests that gambling problem prevalence rates increase sharply for the first few years after an increase in availability, and then level off, although the evidence is not clear cut (Wynne Resources, 1998; Productivity Commission, 1999).

The Australian study (Productivity Commission, 1999) noted that lotteries and scratch-and-win tickets represented fairly low risk gambling in Australia, while continuous play gaming machines and casino table games, in particular, were associated with higher prevalence rates. The Australian study also noted that rates of presenting for treatment were more closely linked to the accessibility of gaming machines than other types of gambling, and suggested that presenting for treatment may be a better indicator of problems with gambling than prevalence rates as measured in surveys.

Table 3. Comparison of prevalence rates for youth between At-risk (a-r) (scores of 2-3 on SOGS-RA) (%) and Problem (P) (scores of 4+ on SOGS-RA) (%)

Variations in problem gambling prevalence by demographic characteristics

It is difficult to paint a portrait of an “average” problem gambler. The best predictor of problem status (or potential problem status) according to the data from general population surveys is the playing of gaming machines. Those who play these machines are much more likely to experience problems than those who do not. While there are some socio-demographic factors that differentiate between problem and non-problematic gamblers, it is important to note that there are often bigger differences between gamblers and non-gamblers than there are between people affected by problem gambling and gamblers in general (Productivity Commission, 1999).

Some groups do have higher prevalence rates than others, and there are some consistencies across jurisdictions. For instance, people affected by problem gambling are more likely to be found among ethnic minorities, young adults (under 25), and those with lower levels of formal education (NORC report). Prevalence rates have also been shown to be higher for prison inmates, and those who have or have had mental health, alcohol or other drug problems, as well (NORC report; Productivity Commission Report). table 4, below, demonstrates the differences in some of these demographic groups. Even with the error margins taken into account, the differences are striking — adolescent prevalence rates tend to be about twice the rate for adults. College students’ rates are also about twice the adult average, and the rates for institutionalized populations studied are five to seven times higher than the rates for other adults. The data in this table are based on 120 studies conducted in the 1977 to 1997 period, all of which met Shaffer et al.’s criteria for research standards.

Table 4. Mean Prevalence Rates and Confidence Intervals of Gambling Problems: A Meta-Analysis of North American Surveys, 1977-1997* 

The Australian study suggested that the higher rates among groups who might be considered minorities are a function of a learned greater resistance to social sanctions. In support of this argument, those with fewer significant others are more likely to be people affected by problem gambling. For instance, those who are separated, divorced, or living in single-person households, as well as those who are unemployed, have higher prevalence rates. However, the Australian study pointed out that the causality here is complex, as unemployment study pointed out that the causality here is complex, as unemployment and/or a return to singlehood may be consequences, rather than predictors, of gambling problems.

The report from the 1999 US national study suggested that middle-aged women made up the new growth area for problem gambling. This group had the highest lifetime prevalence rates. The US researchers speculated that this might be due to this group’s growing involvement with slot machines and video lottery terminals, as well as Internet gambling. Women seem to be more interested in casino games without the trappings of the casino, in playing the games in their own homes in relative anonymity.

Prevalence rates as they are currently reported don’t appropriately capture gender differences in problem gambling. They tend to obscure the differences between male and female gambling behaviours. Most male people affected by problem gambling report seeking “action,” or excitement, while many female gamblers say that they gamble to “escape,” to get away from personal problems. A score of 5 on a DSM-IV or SOGS screen may actually represent two totally different gambling behaviour patterns with entirely different sets of motivators.

The increasing accessibility of continuous play gambling machines, in particular, will likely have a huge impact on the numbers of female people affected by problem gambling in the future. The Australian report already notes that problem gambling, which used to be predominantly a male issue, has become increasingly feminized. In Australia, the numbers of men and women in treatment are now roughly equal (The Productivity Commission, 1999). In order to capture the many types of problem gambling, and to provide a prevalence estimate that is meaningful, the way prevalence is measured, and even conceptualized, has changed in the last few years.

Trends in the measurement of prevalence

Dissatisfaction with the South Oaks Gambling Screen as a tool for the measurement of prevalence in general population surveys, as well as the model of gambling that labels some gamblers as “pathological” has led to a trend away from labelling, and toward less value-laden typologies of gambling. Shaffer and Hall (1996) suggest a scheme that standardizes the language and takes into account the scope, nature and severity of the gambling behaviour involved (p. 8), based on various levels of gambling problems. They use the term “disordered” gambling, and offer a classification system that allows for more precise communication about specific levels of gambling problems. They suggest that the concept of levels of gambling better represents gambling behaviour and gambling problems as a continuum, rather than simply dividing gambling into problematic and non-problematic categories of behaviours. Shaffer and Hall’s schema has four levels. Level 0 is reserved for non-gamblers, level 1 for non-problematic gambling, level 2 for gamblers with sub-clinical levels of gambling problems (this would include “problem,” “at-risk,” “in transition” and “potential pathological” gamblers), level 3 for those who have the most severe gambling problems (those who would in most cases meet the established clinical diagnostic criteria for pathological gambling), and level 4 gamblers, who are a sub-set of level 3 gamblers, those who present themselves for treatment (Shaffer et al., 1997).

This development is attractive for several reasons. First, it recognizes severe gambling problems in two ways: those who meet the diagnostic criteria, and those who actually present for treatment. Second, it separates those who do not gamble from those who experience no problems gambling. These categories are often lumped together in prevalence research, thus losing an opportunity to track movement between the categories. Finally, it allows for two-way movement along a continuum, from more severe to less severe levels of gambling problems, as well as vice versa. It is also important to recognize that not only are those in treatment a very small subset of level 3 gamblers, but this group may not be those who have the most serious problems.

When the U.S. team developed their new measure, the nods (based on the DSM-IV criteria), they used a classification scheme very similar to Shaffer and Hall’s. The primary difference between the nods schema and Shaffer and Hall’s typology is that those who experience no harm are separated from those who experience one or two adverse consequences. The other notable difference is the use of a threshold for losses. The nods typology ranges from Type A (those who report losing less than $100 in a single day or year) to Type E (those who report losing $100 or more in a single day or year, as well as five or more adverse effects — this group is equivalent to the “pathological” gamblers in DSM-IV) (NORC Report, 1999).

More recently, other researchers have begun to examine ways to position problem gambling within its social context and to develop profiles of different types of people affected by problem gambling. The Canadian Problem Gambling Index (the CPGI), tries to integrate DSM-IV criteria with a broader understanding of problem gambling in its community context. This instrument includes a number of correlates of problem gambling addressing family history and social context in particular (Ferris, Wynne & Single, 1999). A similar effort under way in Australia, the Victorian Authority Gambling Screen (VAGS), entails “a multi-disciplinary reconceptualization of the impacts of gambling on the individual and family” (VCGA, 1998, as cited in the Productivity Commission Report, section 6.43, 1999). These attempts to reconceptualize gambling as more of a social issue than a clinical one represent a positive step toward producing prevalence rates that more accurately reflect the reality of their communities.