 |
 |
|
|
|
Legalized gambling has expanded in many parts of Canada and the US in recent years (1–3). Increases in the types of gambling activities available have been associated with increases in rates of problem gambling in the general population (4,5). Policy-makers recognize there is a social cost for the financial revenues from gambling expansion. For this reason, government-commissioned general population surveys are used in part to monitor the nature and extent of gambling problems in the community. It is important to ensure that assessment strategies have high validity. There is also a need to identify factors that differentiate problem gambling from recreational gambling. From a public health perspective, accurate data from community surveys are required to plan education, prevention, and treatment services (2,3). Canadian academic research and government-sponsored surveys on the prevalence of gambling problems in the general population have often adopted the approach used in previous US surveys, employing the South Oaks Gambling Screen (SOGS) (6), a DSM-IV–based instrument (7). The SOGS was developed in Long Island, New York, using criterion groups of substance abuse inpatients and members of Gamblers Anonymous (GA). Scores of 5 or more on this measure are used to indicate “probable pathological gambling.” Epidemiological researchers typically employ a lower score of 3 or 4 to define “problem gambling”(8). The use of the SOGS to determine the prevalence of problem and probable pathological gambling in community-based surveys has attracted some controversy (9–11). One major criticism is that this “screen” instrument produces a de facto diagnosis because it is not generally followed by any further diagnostic interviewing. A second criticism is that the SOGS scoring attaches too much weight to the number of different types of debt accumulated (for example, 9/20 scored items refer to sources of borrowing money). The SOGS also relies on the outdated DSM-III (12) criteria in its development. The DSM has undergone several important changes in the content of pathological gambling symptoms since that time. The current DSM-IV criteria (13) now include addiction characteristics to reflect tolerance, withdrawal, and use of gambling as a way to relieve dysphoric mood. These domains are not directly assessed in the SOGS. At the same time, the content of the DSM-IV and SOGS significantly overlap. Recent US research suggests these assessment strategies may have good agreement, but they contain different thresholds or cut-off scores for defining caseness in community surveys. Volberg found that the SOGS correlated with a brief DSM-IV screen at 0.52 (14), while Stinchfeld used a more extensive DSM-IV checklist and obtained a correlation of 0.77 with the SOGS (15). The DSM-IV cut-off of 5 symptoms produced lower prevalence figures for pathological gambling in both of these studies, compared with the SOGS. Volberg concluded from her analysis that the DSM-IV cut-off of 5 for pathological gambling was too severe for general population studies (14). Not surprisingly, there has also been some concern about the validity of the SOGS for use in Canadian epidemiological research. In 1997, a network of senior addictions administrators and representatives of government regulatory bodies recommended replacing the SOGS with a new made-in-Canada assessment instrument that did not yet exist (16). An inter- provincial funding envelope was established, and efforts have been made to develop new content for a Canadian problem gambling measure (17). However, given the fledgling knowledge base on problem gambling in Canada, retaining the SOGS in epidemiological research has some advantages. It may be beneficial to continue to use the SOGS, but to attach more importance to some SOGS items or to set different cutoff scores for defining gambling-related difficulties. Retaining the SOGS in Canadian problem gambling research would allow direct comparisons with SOGS-based research findings that have accumulated from many different centres, including the US. These advantages do not negate the need to compare the SOGS with other indices of gambling problems in community surveys. While there is no universally accepted definition of gambling disorder in this field, the DSM-IV is widely recognized in many mental health settings. An early advocate of SOGS use in epidemiological research on problem gambling recently noted that the DSM-IV criteria set has now been widely adopted as the gold standard in the field and is being used in conjunction with the SOGS in several different general population surveys (14). Volberg stressed the need to calibrate DSM-IV–based assessment strategies with the results of more than a decade of SOGS-based research (18). Direct comparisons between DSM-IV ratings of pathological gambling symptoms and the SOGS have not yet been conducted in a Canadian community-based adult sample. This is the primary purpose of the present study. We used an annual household community survey with a high response rate, similar to our previous community-based gambling research using the SOGS (19). This study closely examines the distribution of scores on the DSM-IV assessment system to determine whether a continuum of gambling problems exists. We determined characteristics of recreational gamblers (that is, adults who gamble but do not endorse any DSM-IV symptoms) and gamblers with varying levels of DSM-IV symptoms. We note that the DSM-IV Impulse Control Disorders work group actually recommended a lower symptom threshold of 4/9 possible symptoms (20), but the DSM-IV Task Force imposed a cut-off of 5/10 symptoms. The DSM-IV Impulse Control Disorders committee originally recognized that “with greater public education and awareness of gambling as a disorder, many individuals who have gambling problems but who are not pathological gamblers will be requesting consultation. Pathological gamblers will also be seeking help in earlier stages of the disorder” (20; p 1009). These issues may have direct relevance for the clinically significant gambling difficulties seen in Canada, where casinos and video gambling machines have only recently been introduced to society. In this study, we hypothesized that subthreshold pathological gamblers (that is, individuals with a DSM-IV symptom score of 1 to 4) would more closely resemble threshold pathological gamblers (that is, individuals with a DSM-IV symptom score of > 5) than recreational gamblers (that is, individuals with no DSM-IV symptoms) on types and frequency of gambling behaviours. MethodParticipants DesignWe generated a random sample of telephone numbers from a complete listing of all active telephone numbers in Winnipeg. The household was the primary sampling unit, and we contacted 750 respondents in each year (see note 1). For selection criteria we used sex, age, and household residency. An eligible respondent was someone aged 18 years or over who lived at that address and who matched a random predesignation of sex. We provided a description of the survey, and all respondents gave informed verbal consent to participate. Consistent with previous years, the response rate was high, with 71% of eligible households in 2000 responding and 70% responding in 2001. Professional interviewers took part in the study, most of whom had previous experience conducting telephone interviews for the WAS. Interviewers attended 2 training sessions on WAS protocol and received several hours of training. They also conducted several pretest interviews accompanied by debriefing sessions, and each received an interviewer’s handbook written specifically for the 2000 and 2001 WAS. We did not include the pretest interviews in the final dataset. Interviewers asked respondents whether they were involved in 1 or more types of gambling activities in their lifetime. To respondents who answered yes to at least 1 of 10 different types of gambling queried, interviewers asked a series of questions relating to gambling behaviour (for example, motivating factors and family history of gambling problems) similar to those used in previous gambling surveys (19). Interviewers also administered the 20-item SOGS (1) and a DSM-IV interview previously developed in our research program (2). The interview assesses the 10 DSM-IV symptoms for pathological gambling. We administered both lifetime and past-year versions of the gambling questions. We analyzed data using SPSS statistical software (23). To compute the single degree of freedom chi-square contrasts among the 3 groups of gamblers in Table 1, we used the PROC GENMOD procedure in SAS statistical software (24) with the specification that the distribution be binary.
ResultsOf the total sample (n = 1489), 1320 individuals (88.7%) reported having participated in at least 1 gambling activity in their lifetime. According to SOGS-based criteria, the lifetime prevalence of “probable pathological gambling” (that is, SOGS score of > 5) in the total sample was 2.6%, and a further 3.5% met the criteria for “problem gambling” (that is, SOGS score of 3 or 4), for a combined lifetime prevalence figure of 6.1% problem or probable pathological gambling. The 1-year prevalence figure was 3.2% (1.3% probable pathological and 1.9% problem gambling). According to DSM-IV–based criteria of 5 or more symptoms for pathological gambling, lifetime prevalence in the total sample was 1.3%. A further 6.6% had symptomatic or subthreshold pathological gambling (DSM-IV symptom scores of 1 to 4). The 1-year prevalence was 0.7% for pathological gambling and 4.5% for subthreshold level. The correlation between the SOGS items and DSM-IV symptoms was high, both for lifetime (r = 0.81) and past-year (r = 0.79) versions. At an individual level, most of the SOGS items were at least moderately correlated with the DSM-IV total score. The exception was 3 items concerning debt (that is, selling personal or family property, borrowing from loan sharks, and borrowing from someone without paying them back), which produced correlations of approximately 0.20 or less. Less severe items, such as borrowing from household money or from relatives, were more strongly correlated with DSM-IV scores (r = 0.56 and r = 0.46, respectively). Of the 1320 respondents who reported participation in at least 1 gambling activity in their lifetime, 1202 (91.1%) obtained a lifetime DSM-IV score of 0. These individuals were classified as recreational gamblers. Table 1 presents comparisons between recreational gamblers (DSM-IV score of 0), subthreshold pathological gamblers (DSM-IV score of 1 to 4), and pathological gamblers (DSM-IV score of > 5) on various gambling-related variables. These variables include a breakdown of low gambling activity frequency (1 or 2 different types in the past year) vs high gambling activity (> 5 different activities in the past year). We also assessed preference for different types of gambling activity, specific motivations to gamble, family history of gambling problems, and median monthly gambling expenditure. Recreational gamblers significantly differed from the subthreshold pathological gamblers and pathological gamblers. The latter 2 groups generally did not significantly differ from each other; in some cases, the subthreshold pathological gamblers fell on a continuum between recreational and pathological gamblers. DiscussionThis community survey of almost 1500 adults had a good response rate (70.5%) and produced a sample that was representative of federal census figures for the region on several sociodemographic indicators (21,22). The lifetime prevalence of probable pathological gambling in this study is 2.6%, according to the frequently used method of employing the SOGS and applying a minimum cut-off score of 5 (8). A further 3.5% met traditional criteria for problem gambling levels (that is, SOGS scores of 3 or 4). These figures are nearly identical to the results of a separate survey conducted in the same region 3 years earlier (19) and are among the highest prevalence figures reported in Canada using the SOGS (3). This consistently high rate is likely associated with the widespread availability of legalized gambling in Manitoba; for example, video lottery terminals (VLTs) in neighbourhood bars and restaurant lounges. This is the first Canadian study to directly compare the popular SOGS-based assessment with the DSM-IV symptom criteria for pathological gambling. The 2 forms of assessment were highly correlated (0.80), suggesting they measure the same phenomenon. Similar results have recently been found in US studies (14,15). Some SOGS items concerning amount of debt accumulated should be considered for deletion from the assessment, and 2 items not scored on the SOGS might also be deleted in community surveys: having a credit line with a casino or bookie. Despite the considerable measurement overlap, however, the DSM-IV threshold score of 5 or more symptoms for a diagnosis of pathological gambling produced a prevalence figure (1.3%) that is one-half the prevalence figure produced by the SOGS cut-off method (2.6%). The issue is not whether the SOGS items closely resemble the official clinical symptoms used in the DSM-IV to diagnose pathological gambling but rather, where exactly to set the cut-off for defining a gambling disorder in Canada. This context is characterized by the expansion of the number and types of legalized gambling activities that has occurred over the past decade in many regions of the country. Some may argue that a gambling disorder prevalence figure of 1.3% is too conservative, while others may argue that 2.6% is too liberal. Even the official DSM-IV cut-off of 5 symptoms is arbitrary and is based in part on characteristics of mostly male US members of GA(20). Further, the Impulse Control Disorders subcommittee had originally recommended a lower cutoff of 4 symptoms (of a possible total of 9). To empirically examine the threshold issue in a Canadian community sample, we compared the anchor point defined by recreational or “normal” gamblers with those gamblers who reported some DSM-IV gambling symptoms but who did not reach the symptom cut-off for pathological gambling (that is, subthreshold). At the other anchor point we examined respondents who formally met the criterion of at least 5 symptoms for pathological gambling. We compared the 3 groups on several important gambling-related variables, and the pattern that emerged suggests a continuum. The group that most often distinguished itself was the recreational gamblers. The subthreshold gamblers either did not differ from the pathological gamblers or fell somewhere between the recreational gamblers and the pathological gamblers. For example, on frequencies of high vs low levels of gambling activities, the subthreshold gamblers did not significantly differ from pathological gamblers, and both groups significantly differed from recreational gamblers. The largest chi-square difference across the 3 groups was on frequency of use of slot or poker machines (VLTs). This form of gambling particularly has been shown to be associated with the likelihood of developing gambling problems (25). VLT use is the most common type of gambling favoured by individuals seeking treatment for gambling problems (26), and it was common among the subthreshold pathological gamblers in this study (43%). Similarly, motivations to gamble that are thought to be common in pathological gambling (that is, to distract oneself from everyday problems) were rare in recreational gamblers (5%) and were significantly more common in subthreshold gamblers. Together, the empirical findings from this study echo Volberg’s recent caution that a DSM-IV cut-off score of 5 may be too high to accurately identify individuals with significant gambling problems, especially in community-based settings (14). From a clinical perspective, many Canadian adults may be seeking consultation or early intervention and might not yet have DSM-IV scores as high as 5 (the latter may better characterize end-stage or “rock bottom” pathological gambling, such as that seen in GA groups). While a score of 1 or 2 DSM-IV symptoms may be too liberal, scores of 3 or 4 on the DSM-IV symptom list should be considered indicative of a clinically significant gambling disorder. This cut-off effectively captures the higher end of subthreshold cases. It is also consistent with opinions stated in an influential published report (27) submitted to the US National Gambling Impact Study Commission that recommended DSM-IV scores of 3 or 4 be considered indicative of problem gambling behaviour (see note 2). As in previous studies (19), we found no differences between problem and nonproblem gamblers concerning forms of gambling that have been common to society for some time (that is, problem and nonproblem gamblers were similar regarding use of lottery tickets and charity gambling, with the motivation to support worthy causes). Notably, even among those gamblers with higher DSM-IV scores, most respondents did not report having a parent with a gambling problem. This finding is consistent with the likely increase in gambling-related problems in recent years that accompanied gambling expansion (that is, VLTs and casinos). It is possible that associations among first-degree relatives with gambling problems will increase in future years. To our knowledge this study is also the first to examine SOGS and DSM-IV scores for both lifetime and 1-year prevalence of gambling problems. Consistently, the prevalence figures for 1-year gambling problems were considerably lower than lifetime prevalence figures. Further research is needed to examine those respondents with lifetime but not past-year gambling problems to determine whether many individuals eventually “recover” from problems (that is, individuals no longer meet SOGS or DSM-IV criteria). Such lifetime recovered cases would still represent a significant social cost because of personal suffering and family disruption. However, accurate barometers of gambling problems in the community will need to differentiate between current (that is, 1-year) and lifetime prevalence estimates. For example, a population spike in gambling problems may have occurred in Manitoba in the mid- to late-1990s after VLTs and casinos had been available in the community for a few years. During that time, nearly 1400 adults were seeking treatment for gambling problems (26). Lifetime prevalence estimates may not reflect such changes and instead, may even show minor increases. One-year prevalence figures may be more sensitive to changes within society across time. Our research has several limitations. The study is cross- sectional in design, and we used only an urban sample. A clinical reappraisal investigation using mental health specialists would have been useful but was too prohibitive in cost and response rate to undertake in this study. To ensure a large sample with a high response rate, there were several additional questions warranting examination that we could not include in the survey (for example, cooccurring mental health problems). However, this study contributes to our emerging understanding of gambling problems in Canada. The SOGS, commonly used in community surveys, shows a high degree of association with the official DSM-IV assessment more commonly employed in clinical settings. Efforts should be directed toward establishing a consensus on appropriate cutoff scores for defining gambling disorder using existing assessment methods. Our results suggest that a lower DSM-IV symptom score cutoff of 3 or 4 may help to better capture clinically significant caseness in Canada. Funding and SupportThis project was supported in part by the Ruth Hurd Memorial Fund for substance abuse research (Department of Psychiatry, University of Manitoba) and by an infrastructure award from the Canada Foundation for Innovation. Dr Cox is supported by the Canada Research Chairs program. AcknowledgementsWe thank Ian Clara for assistance with data analysis. Notes1. Consistent with our previous research (19), we adopted a conservative approach and removed 11 participants from the original sample of 1500 adults because of missing data. Examination of these cases reveals that missing data typically arose because participants reported gambling at some point in their lifetime (for example, purchasing a lottery ticket) and were therefore classified as a “gambler”; however, participants then felt that the SOGS questions regarding problem or pathological gambling were not applicable, and the gambling section of the interview was stopped. Three individuals did not wish to respond to any questions. Seven completed the SOGS, obtained a score of 0, and did not wish to answer more gambling questions. Only 1 of 11 individuals had a SOGS score in the problem gambling range and did not wish to complete the remainder of the interview. 2. To further examine the proposition that DSM-IV scores of 3 or 4 may be important from a clinical and public health perspective, we compared Table 1 data using DSM-IV scores of 3 or 4 as the middle comparison group. The characteristics of respondents with 3 or 4 symptoms were more similar to the pathological gamblers than to recreational gamblers on several variables, such as the frequency of 5 or more gambling activities (36% for 3 or 4 DSM-IV symptoms group and 40% for the $ 5 group) and VLT use (55% for the 3 or 4 symptoms group and 70% for the $ 5 symptoms group). A total of 22 individuals (1.5% of the sample) scored 3 or 4. References1. National Gambling Impact Study Commission. Final report. Washington (DC): US Government; 1999. 2. Volberg RA. The prevalence and demographics of pathological gamblers: implications for public health. Am J Pub Health 1994;84:237–41. 3. Ladouceur R. The prevalence of pathological gambling in Canada. J Gambl Stud 1996;12:129–42. 4. Ladouceur R, Jacques C, Ferland F, Giroux I. Prevalence of problem gambling: a replication study 7 years later. Can J Psychiatry 1999;44:802–4. 5. Shaffer HJ, Hall MS, Vander Bilt J. Estimating the prevalence of disordered gambling behavior in the United States and Canada: a research synthesis. Am J Pub Health 1999;89:1369–76. 6. Lesieur HR, Blume SB. The South Oaks Gambling Screen (SOGS): a new instrument for the identification of pathological gamblers. Am J Psychiatry 1987;144:1184–8. 7. Beaudoin CM, Cox BJ. Characteristics of problem gambling in a Canadian context: a preliminary study using a DSM-IV based questionnaire. Can J Psychiatry 1999;44:483–7. 8. Volberg RA, Steadman HJ. Refining prevalence estimates of pathological gambling. Am J Psychiatry 1988;145:502–5. 9. Lesieur HR, Blume SB. Revising the South Oaks Gambling Screen in different settings. J Gambl Stud 1993;9:213–19. 10. Walker MB, Dickerson MG. The prevalence of problem and pathological gambling: a critical analysis. J Gambl Stud 1996;12:233–49. 11. Culleton RP. The prevalence rates of pathological gambling: a look at methods. J Gambl Stud 1989;5:22–41. 12. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 3rd ed. Washington (DC): American Psychiatric Association; 1980. 13. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Washington (DC): American Psychiatric Association; 1994. 14. Volberg RA. Research methods in the epidemiology of pathological gambling: development of the field and directions for the future. Annuario de Psicologia 1999;30:33– 46. 15. Stinchfield R. Reliability, validity and classification accuracy of the South Oaks Gambling Screen (SOGS). Addict Behav 2002;27:1–19. 16. Measuring problem gambling in Canada: a request for proposals. Inter-provincial task force on problem gambling. Ottawa; 1997. Unpublished report. 17. Ferris J, Wynne H. The Canadian problem gambling index final report. Phase II final report to the Canadian inter-provincial task force on problem gambling. Ottawa: Canadian Centre on Substance Abuse; 2001. Unpublished report. 18. Volberg RA. The epidemiology of pathological gambling. Psychiatr Ann 2002;32:171–8. 19. Cox BJ, Kwong J, Michaud V, Enns MW. Problem and probable pathological gambling: considerations from a community survey. Can J Psychiatry 2000;45:548–53. 20. Bradford J, Geller J, Lesieur HR, Rosenthal R, Wise M. Impulse control disorders. In: Widiger TA, Frances AJ, Pincus HA, Ross R, First MB, Wakefield-Davis W, editors. DSM-IV Sourcebook. Volume 2.Washington (DC): American Psychiatric Association; 1996. p 1007–32. 21. Michaud V. Lewis T.K. Selected findings from the 2000 Winnipeg Area Study. Winnipeg (MB): Department of Sociology, University of Manitoba; 2001. 22. Michaud V, Lewis TK. Selected findings from the 2001 Winnipeg Area Study. Winnipeg (MB): Department of Sociology, University of Manitoba; 2002. 23. Stastical package for the social sciences. Version 11.5. Chicago: SPSS Inc, 2002. 24. Statistical analysis systems (SAS). Version 8e. Cary (NC): SAS Institute Inc, 2002. 25. Breen RB, Zimmerman M. Rapid onset of pathological gambling in machine gamblers. J Gamb Stud 2002;18:31–43. 26. Wiebe JMD, Cox BJ. A profile of Canadian adults seeking treatment for gambling problems and comparisons with adults entering an alcohol treatment program. Can J Psychiatry 2001;46:418–21. 27. Gerstein D, Hoffman J, Larison C, Engelman L, Murphy S, Palmer A, and others. Gambling impact and behavior study: report to the National Gambling Impact Study Commission. Chicago (IL): National Opinion Research Center at the University of Chicago; 1999. Author(s)Manuscript received February 2003, revised, and accepted January 2004. 1. Canada Research Chair in Mood and Anxiety Disorders, Associate Professor, Department of Psychiatry, University of Manitoba, Winnipeg, Manitoba. 2. Professor, Department of Psychiatry, University of Manitoba, Winnipeg, Manitoba. 3. Director, Winnipeg Area Study, Department of Sociology, University of Manitoba, Winnipeg, Manitoba. Address for correspondence: Dr B Cox, PZ-430 PsycHealth Centre, 771 Bannatyne Avenue, Winnipeg, MB R3E 3N4 e-mail: coxbj@cc.umanitoba.ca
1 | 2
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||