By Lennart E. (Len) Henriksson, Ph.D. ¹

Before 1969, most types of gambling were illegal in Canada. Today, it is ubiquitous, and growth continues to outstrip most other industries (Marshall, 1998). Every province has a state-owned lottery and charity gaming activities. Casinos can be found in every region of the country (Canada West Foundation, 1998).

For governing politicians, the benefits of expansion are evident. "New" revenues can be mustered without unpopular increases in taxes. At the societal level, the picture is less clear. When a casino opens, some proportion of the revenues and jobs may represent new, incremental gains. But another proportion is diverted from activities that might ultimately be more productive. There is also a risk of ills such as increased crime and problem gambling, or the encouragement of undesirable attitudes and behaviours (Henriksson and Lipsey, in press).

The expansion of gambling has not gone unchallenged. But the balance of power is overwhelmingly in favour of promoters, who seldom "give up" because the potential for private profit and increased gross revenue to government is immense. The purpose of this paper is to review issues of interest to the health research community, and present suggestions for further study.

Three themes are key to understanding the recent history and the Canadian policy environment:

During the early 1970s, politicians tended to view lotteries as a way to fund "little extras", such as the Montreal Olympics or charitable pursuits. Today, gambling accounts for small but significant proportions of government revenue in every province, ranging from 1.4% ($290 million) in British Columbia to 4.0% ($698 million) in Alberta (Canada West Foundation, 1998). It has become an integral component of government revenue-raising mechanisms. "Downloading" of responsibilities, voter dislike of high taxes (or user charges), and an aging society have also been identified as factors behind gambling’s explosive growth (Foot, 1996; Henriksson, 1996).

Historically, Canadian governments were expected to enforce criminal restrictions against illegal gambling, and provide oversight for small-scale games of chance that occurred within the context of agricultural fairs or charitable fundraising. Today, the "protective" regulatory role continues, but government has also become an active provider and promoter. The potential for real or perceived conflicts-of-interest is evident (Auditor-General of British Columbia, 1996/97; Campbell and Smith, 1997: 34).

For governments, experience has shown that expansion is a "slippery slope". Protective measures have often faded with the advent of intensified competition or waning revenue growth (Goodman, 1994; Lipsey, 1997). As the "product life-cycle" of various forms of gambling progresses–and as competition increases–the pressure on protective measures can be expected to continue.²

On the whole, the research is still "catching up" to what is available on phenomena such as alcohol and other drug use. This is only partly due to the recent increases in gambling activity. The small size of the circle of investigators and reviewers who conduct research in some areas (such as the prevalence of problem and pathological gambling; Shaffer, Hall and Vander Bilt, 1997: 61) has not been helpful. Too, departmental "silos" at many universities and colleges have discouraged the interdisciplinary approach that is so important to conducting meaningful studies (Henriksson, 1996).

The overwhelming domination of research funding by gambling promoters (including governments) is worrisome. At a time when stable employment and funding is often hard to come by, few researchers (or institutions) can afford to risk alienating these sources. More than one Canadian provincial government has funded "studies" that have grossly inflated the benefits of more gambling, while trivializing cost issues (Henriksson and Lipsey, in press). This is at least a mildly onerous signal for those contemplating research in the area.³

Gambling is hardly new to modern times. In various forms, it occurs universally in all cultures, all ages, and is participated in widely by those of all societies and social strata (Bolen and Boyd, 1968). The vast majority of the population participates in games of chance without any noticeable ill effect. But as is well known, some spend more than they can afford.

What is new is the medicalization of the problem.⁴ In 1980, the American Psychiatric Association officially recognized "compulsive gambling" in the third edition of the Diagnostic and Statistical Manual of Mental Disorders. In the 1987 revision and again in the fourth edition, the diagnosis was called "pathological gambling". As shown in Figure 1, the criteria are similar to those of psychoactive substance dependence: preoccupation with and need to engage in the activity, a buildup of tolerance, loss of control, and withdrawal symptoms. Associated personality characteristics include difficulties with problem solving and a predisposition to anger, depression and anxiety (Setness, 1997)

There is evidence that expanding legalized gambling increases the prevalence of problem and pathological gambling generally (Room, Turner and Ialomiteanu, 1998; Volberg, 1994). However, in a very recent Alberta study, the authors found that while current "problem gambling" prevalence rates decreased from 4.0% to 2.8% between 1994 to 1998, the current "probable pathological" rate increased from 1.4% to 2% (Wynne Resources, Ltd., 1998). While the relationship between gambling expansion and increased prevalence rates may require further study, the sheer number of cases can logically be expected to rise.

Measures for the incidence and prevalence of disorders such as problem and pathological gambling are evolving, and fraught with uncertainty (Ferris, Stirpe and Ialomiteanu, 1996: 6; Shaffer et al., 1997). Most extant prevalence studies have been done using the South Oaks Gambling Screen (SOGS) (Lesieur and Blume, 1987; Rönnberg, 1999). Originally, it consisted of 20 items derived from the definition of pathological gambling in DSM III (American Psychiatric Association, 1980. As newer versions of DSM appeared, SOGS was revised. These made measures incomparable with earlier versions; therefore, it is not unusual that the older version of SOGS is used even though it no longer corresponds to the current DSM (Rönnberg, 1999).

Recently, a study conducted for the National Gambling Impact Study Commission in the United States proposed a new scale out of a conviction that SOGS had been "eclipsed" (National Opinion Research Centre, 1999: 14). The new "NODS" scale consists of 17 items, each of which is carefully matched with DSM-IV criteria. In Canada, some pioneering work on conceptualizing, defining and measuring problem gambling behaviours is now in progress, and a final report is expected shortly.

As most studies rely on self-report (survey) data, a review of the often-overlooked limitations may be helpful. Measurement error is a serious problem. In particular, when a sample of people is asked questions about sensitive topics, it is likely that many will consciously or unconsciously bias their responses toward the socially acceptable answer (Organ and Podsakoff, 1986; Smart and Ferris, 1996; Streiner, Norman and Monroe-Blum, 1989). The result will be misleading estimate of true prevalence. As a recent study noted, some under-reporting of substance use occurs in alcohol and other drug addiction research; under-reporting appears to be higher among very light and very heavy users (Wynne Resources, Ltd., 1998: I-16).

Several techniques have been developed to detect and remedy "social desirability bias" and other sources of measurement error (Streiner et al.,1989: 97). For example, in their empirical study, Embree and Whitehead (1993) provide several helpful ideas on how the reliability and validity of self-reported drinking behaviour can be enhanced. The studies by Sobell and her colleagues (e.g., Sobell, Brown, Leo and Sobell, 1996) are also a useful resource. Enhancing the validity of problem gambling self-reports is a challenge for future research.

Non-response error occurs because data provided by subjects who respond to a survey do not necessarily generalize to non-responders. This is a particular concern when the response rate is very low. Today, this issue is acute. The use of technological aids such as voicemail, answering machine and caller ID to screen out strangers who call to collect survey data or make sales pitches is "old hat". Many would-be respondents refuse to participate, particularly if the survey is lengthy.⁵

In their recent analysis of prevalence studies, Shaffer and his colleagues (1997) found that "sample response rates of less than 50% are unscientific and offer little to our understanding of disordered gambling" (p. 118). This casts doubt on the usefulness of the studies that have been done in some provinces. The response rate for a 1994 British Columbia study was only 25% (Gemini Research, 1994: 15) – the lowest of the studies. A replication completed two years later (Angus Reid Group, 1996) failed to report the response rate.⁶

Sampling error results from heterogeneity on the incidence measures among members of the population. It is attributable to the fact that certain members of the population are deliberately excluded by selection of the subset of members for which responses were obtained (Dillman, 1991: 227).

Non-coverage error arises because some members of the population are not covered by the sampling frame and therefore have no chance of being selected into the sample (Dillman, 1991:227). In the case of the telephone survey, not all gamblers have access to a telephone. Other groups often subject to non-coverage error include non-English speakers,⁷ prisoners and hospital patients.

In sum, these sources of error all have the potential to cause understated prevalence measures.⁸ When the public is repeatedly presented with very low figures, they are encouraged to trivialize the disorder because prevalence by itself conveys no cost information. This may work to the benefit of vested interests.^9,10

The "Harvard study": Meta-analysis is a research procedure for aggregating the findings of empirical studies. It has been used to estimate the prevalence of mental health disorders (e.g., Ritchie, Kildea and Robine, 1992). Shaffer and his colleagues (1997) conducted a meta-analysis of some 152 problem and pathological gambling prevalence studies. One hundred and twenty studies were included in their final work.¹¹ The researchers found that the proportion of adults who qualify as "Level 3" disordered gamblers rose from 0.84% before 1994 to 1.29% since (p. 43).¹²

Sacks and his colleagues (1987) offer six criteria that are useful in evaluating the Harvard study¹³: study design, combinability, control and measurement of bias, statistical analysis, sensitivity analysis and application of the results. On the whole, the criteria are met well. Much of the content and suggestions will serve as a useful guide to future researchers. Particularly interesting are the authors’ suggestions for assessing the quality of prevalence surveys (p. 28-29).

One criterion that deserves additional discussion, however, is combinability: "the extent to which the results of the separate trials can be meaningfully combined" (Sacks et al., 1987: 451). The authors of the Harvard study demonstrate an awareness of the combinability issue (p. 52-3). They carefully describe the decision process used to select eligible studies (pp. 16-19). Their discussion of measurement differences and how these were treated is thorough. However, illnesses and addictions are at least in part culturally-defined phenomena (Helman, 1994: 194-223; Payer, 1996). More could have been done to caution readers about the pitfalls of combining prevalence studies from different cultures.¹⁴

An additional concern lies in the authors’ application of the results. As Sacks and his colleagues write, "once the results of the pooling process are available, the meta-analysis should attempt to put them into perspective…" (p. 453). While meta-analysis is a helpful way to "round out" the defects that are inevitable in any one study, it does not attenuate defects that are pervasive throughout the set. In this case, a very heavy proportion of the studies used self-report data collected by telephone interview. Shaffer and his colleagues are certainly aware of the various sources of error, but again, could have been more cautionary as they interpreted the results.

Problem and pathological gambling bring a variety of costs for individuals, families and communities. Many of these are "social costs"; that is, they are not relevant to any individual decision-maker, but are incurred by society as a whole. Estimates of total social costs¹⁵ (or, costs incurred by each individual problem gambler¹⁶ cover a wide range because of the tremendous variation in the definitions and assumptions researchers use to generate figures.¹⁷ What follows is a discussion of some principal costs.

Crime: Some problem gamblers commit crimes in order to support themselves (e.g., Crockford and el-Guebaly, 1998: 47). The comparative newness of large-scale casinos in Canada makes it inappropriate to dismiss the crime issue in light of unfavourable experiences in other countries (Burke, 1996; Friedman, Hakim and Weinblatt, 1989; Illinois State Police, 1992). Some authors have suggested that the "incubation period" for pathological gambling is roughly 10 years (Miller, 1996: 623): a good deal of time may pass before criminal behaviour associated with the disorder occurs. Further, a recent literature review notes that accurate rates of the prevalence and extent of gambling-related criminal behaviours are difficult to obtain, and may tend to understate the problem:

Unfortunately, promoters often understate or ignore the crime issue (Nicol and Nolen, 1998: 45). For example, the authors of the British Columbia Gaming Review claim that "fears of increased crime have not materialized in other locations in Canada where the availability of gambling has increased" (British Columbia Ministry of Employment and Investment, 1997: 3), and cite the Montreal casino as an example. They ignore problems that have arisen at that location because of loansharking and related violence ("Loan sharks," 1997). As in many promotional documents, the costs of crime itself, incarceration, justice administration, higher insurance rates, and preventive measures are ignored.

Suicide: When the consequences come home to roost, suicide may appear to be an easy out (Lesieur, 1998; Phillips, Welty and Smith, 1997). Some researchers have found that up to 90% of pathological gamblers have considered suicide and that about 20% of those in treatment have actually attempted it (National Council of Welfare, 1996: 34; Setness, 1997).

Family and employment difficulties: That problem and pathological gambling often inhibit effective job performance is well known (National Council of Welfare, 1996). Outcomes such as reduced productivity, dismissal, and bankruptcy all have consequences for spouses, children, and employers (Overman, 1990). One Canadian study revealed that spousal and child abuse was more frequent among pathological gamblers (Bland, Newman, Orn & Stebelsky, 1993). Jacobs (1989) found that children of problem gamblers experienced almost twice the incidence of broken homes caused by separation, divorce, or death of a parent before they had reached the age of 15 (p. 227).

Health costs: There is evidence that problem gambling is associated with increased health care costs (Henriksson, 1996). However, extant studies appear to be inconsistent in their treatment of comorbidity and the secondary role problem gambling may play to other disorders. Depending on the true nature of the causal linkages, economic and social costs (notably health-related ones) may be understated or overstated as a result.¹⁸

In sum, a great deal remains to be learned about the economic and social costs. Existing studies use a wide variety of methods to compute estimates. Often, these are very sensitive to assumptions and time horizon. Some costs may lend themselves to measurement relatively easily, such as incremental health care, treatment and justice administration. Other costs are far more difficult to measure. Reduced productivity and quality of life, or increased use of non-market resources (such as friends and family) are examples of these (e.g. Rice, Kelman, Miller and Dunmeyer, 1990: 37-9).¹⁹

An excellent review is contained in Crockford and el-Guebaly’s recent article in the Canadian Journal of Psychiatry (1998). These researchers conducted an extensive literature search using MEDLINE and other databases. Spunt and his colleagues (1998) have also published a review article that focuses on gambling and substance misuse. The discussion here is an overview of the broad findings, supplemented by a small number of studies that were not included in the reviews.

Alcohol or other drug abuse: Generally, both reviews confirm the summary of a senior practitioner: "addictions often come in bunches". The association between pathological gambling and alcohol or other drug abuse was found in several studies to be strong and significant. This finding is consistent with recent Canadian studies (National Council of Welfare, 1996: 33; Rupcich, Frisch and Govoni, 1997).

Crockford and el-Guebaly note that "the extent of the co-morbidity…varies depending on the sample characteristics and the instruments used" (p. 45). They note a possible role for ethnicity and culture in influencing the prevalence of pathological gambling, a point we emphasize again later in this paper. The results of two Alberta studies suggest that there is a linkage between problem drinking and pathological gambling in Aboriginal populations (Nechi Training, Research and Health Promotions Institute, 1994, 1995).

Crockford and el-Guebaly found some evidence of an association between pathological gambling and high rates of tobacco use (p. 45). The association is one that may vary with population groups. In a pioneering study of over 21,000 high-school students, alcohol and cigarette use proved to be non-significant predictors of "gambling problems" after adjustment for other risk behaviours. However, regular cocaine use and anabolic steroids were significantly associated with a report of problems attributable to gambling. There was an increase in the absolute number of risk behaviours reported between those who had not gambled, those who had gambled, and those for whom gambling had created problems (Proimos, DuRant, Pierce and Goodman, 1998).

Mood disorders: Both review articles cite studies showing a strong association between pathological gambling and mood disorders, notably depression. However, Crockford and el-Guebaly are somewhat guarded in their interpretation. While they note that in all likelihood, "there is at least a subpopulation of pathological gamblers who have a comorbid mood disorder", they emphasize the need for research to determine the nature of the association (p. 47).

Anxiety disorders: Three studies were found that reported an increased prevalence of anxiety disorders among pathological gamblers. However, because of limitations in the studies (notably small sample sizes and inconsistent measures), Crockford and el-Guebaly concluded that there "would appear to be insufficient data to support the theory that anxiety disorders are comorbid with pathological gambling" (p. 47).

An interesting question is the extent of comorbidity between Obsessive-Compulsive Disorder (OCD) and pathological gambling.²⁰ Crockford and el-Guebaly concluded that the literature revealed "little support" for comorbidity with OCD (p. 47). However, Hollander (1998) argued that significant comorbidity does exist between pathological gambling, OCD and depression. Consequently, he initiated a controlled trial with fluvoxamine, and found encouraging results (Hollander, DeCaria et al., 1998).²¹

Population health may be defined as a conceptual framework for the thinking about why some people are healthier than others (Thompson, 1999). It increases our understanding of the determinants of health and reaffirms the need to examine their antecedents in the population as a whole. Determinants have been classified in a variety of ways (Evans and Stoddart, 1994; Green and Ottoson, 1999; Kindig, 1997).

At times, identifying the relationship between gambling and population health will pose a challenge for researchers because it is now so widely available.²² With this caveat in mind, we suggest some key determinants and associations that may exist:

Income distribution: In their analysis of heterogeneities in health status, Hertzman and his colleagues note that "(t)he way in which income is distributed and used in a society may be more important than its average level.…(D)eveloped countries show rather strong correlations between the degree to which national income is equitably distributed and average health status" (Wilkinson, 1992; Hertzman, Frank and Evans, 1994:70). Rose (1990) was more forceful: he identified the health gap between rich and poor as the greatest of today’s public health challenges (p. 687).

Other things being equal, the effect of gambling expansion on population health may operate through an association with income distribution. Large prizes from lotteries, slot machines and other games are concentrated in the hands of a few. Moreover, a recent Statistics Canada study confirmed that gambling-derived taxes are regressive (Marshall, 1998: 8; see also Borg, Mason and Shapiro, 1991); that is, they constitute a smaller proportion of income as income rises.

Socio-economic status: The most compelling evidence for the role of non-medical factors in population health is the persistent finding that higher levels of socio-economic status (SES) are directly associated with lower levels of mortality and morbidity (Kindig, 1997: 75). To the extent that persons of lower SES demonstrate a greater propensity to divert their resources and savings towards gambling, the effect on population health may be adverse. Moreover, the literature suggests that persons of higher SES have enhanced self-efficacy perceptions, which could render them less vulnerable to problem gambling behaviour (Evans & Stoddart, 1994: 50).

Culture: Culture can be defined as the constellation of values and beliefs in a population or community. Its relevance to population health has been noted in the literature. The role of culture as a determinant of alcohol abuse has also been stressed (Abrams and Niaura, 1998: 135-7; Helman, 1994). Assessing whether a similar argument holds for the gambling context remains a project for future research. Rose provides some encouragement for this notion with his suggestion that cultures influence the incidence of deviant behaviour (and hence, adverse health consequences) by the extent to which it is defined and institutionalized:

Recognizing the downsides of gambling expansion, some provinces have tried "remedial measures" in an effort to mitigate these. For example, casino operators have sometimes been required to display "know-when-to-quit" signs, conduct training programs to help staff spot problem gamblers, or implement procedures whereby customers may voluntarily bar themselves from access to the facility. While these measures may be commendable, there is a dearth of peer-reviewed efficacy studies. In situations where a substantial proportion of revenue comes from problem gamblers (e.g., Miller, 1996), it may be unwise to expect that providers will eagerly fulfil the spirit of requirements such as these.

Additional measures (many of which appear to be based on "harm reduction" principles) have been implemented by addiction foundations and public authorities in most provinces (for an overview, see Canada West Foundation, 1998). The Canadian Medical Association and some of its provincial counterparts are also promoting awareness among their memberships.

The promotion of gambling has received little attention in Canada. Concurrently, advertising’s role in increasing tobacco and alcohol consumption is well known, and a clear public health concern (Mosher, 1994). Specific knowledge exists on sub-groups in the population, notably the young (O’Keefe and Pollay, 1996; Rachlis and Kushner, 1995). Examples of standards, regulations (or more broadly, "health promotion programs") are not difficult to find (Shah, 1994: 159-60) at both Provincial and Federal levels.

Meanwhile, there is some evidence that advertising is also a factor in causing problem and pathological gambling, notably among the young (Moran, 1994). But unlike other "legal", potentially addictive products, government is both the producer and regulator of state-sponsored gambling, and mandated standards for gambling promotions are virtually nowhere to be found.²³ Meanwhile, agencies such as the British Columbia Lottery Corporation rank among the highest-volume advertising accounts in their provinces. As a former director of marketing put it, "We believe any promotion that can alter the regular purchasing habits of the consumer is viewed as significantly benefiting our long-term success" (Goodman, 1994: 137).

Experience from the alcohol and tobacco cases leads to the suggestion that the boom in gambling advertising and the lack of standards contribute to health difficulties. As in most businesses, declining or disappointing revenues are often followed by more aggressive promotional campaigns. Sometimes, the targeting of gambling promotions on the addicted, the poor or youth market segments has been apparent (Goodman, 1994; Padavan, 1994).²⁴

All 10 provinces now offer secondary or tertiary treatment for problem gambling. New Brunswick was the first province to begin funding, in December, 1993 (Canada West Foundation, 1998: 2-3), while British Columbia was the last (in 1998).

The services offered range from a "telephone hot-line" service to more extensive counselling programs. A few peer-reviewed studies are available. For example, Sylvain, Ladouceur and Boisvert (1997) evaluated the efficacy of a cognitive-behavioural treatment package and found encouraging results. Analysis of data from six- and 12-month follow-ups revealed maintenance of therapeutic gains. Similar results were found by Ladouceur and his colleagues in their evaluation of a cognitive treatment program (1998).

Some countries outside Canada have a longer experience with treatment. In a recent review article, López Viets and Miller (1997) concluded that "empirical outcome data … provide an encouraging picture of treatment outcome for pathological gamblers" (p. 697). However, the authors subsequently pointed out several serious limitations, including small sample sizes and lack of post-treatment follow-up in many studies. They also found that improved outcome measures were needed, a conclusion also reached by Canada’s National Council of Welfare (1996: 60). Another very real concern is the "file-drawer problem"–that is, the extent to which studies which reveal nonsignificant or contrary findings remain unavailable or unpublished (e.g., Sacks et al., 1987; Shaffer et al., 1997:28).

Although drug addictions, alcoholism and pathological gambling are distinct phenomena (e.g., Briggs, Goodin and Nelson, 1996), experience suggests that the recidivism rate in problem gambling treatment programs will be significant. Many who are afflicted will never find their way to the treatment process (e.g., National Opinion Research Center, 1999: 35; Roman and Blum, 1991:757; Spunt et al., 1998: 2546).

The increase in treatment programs for problem and pathological gamblers raises the issue of resource allocation, an important issue in the health care field. After all, funds dedicated to problem-gambling related programs are unavailable for other purposes, such as treating drug addiction or depression. Another limited resource that is easy to overlook is time. If a busy family practitioner allocates her continuing education time to a "problem gambling workshop", she forgoes an opportunity to learn more about alcoholism or obsessive-compulsive disorder, even if the latter might ultimately be more beneficial to the patients in her practice.

Is pathological gambling a "useful" concept?²⁵ Consider the flowchart shown in Figure 2. We assemble a group of 100 patients with unknown mental health difficulties. Some proportion present to the physician, who makes an initial diagnosis. At times, the physician will correctly identify the primary disorder; in other cases, he or she will be "wrong". Either way, the physician recommends a course of treatment. Some (but not all) of the patients who are correctly diagnosed improve as a result. But what is also true is that some patients who are incorrectly diagnosed will improve, particularly where there is significant comorbidity between that diagnosis and the "true" primary disorder.

The scenario described is simplistic,²⁶ but does suggest several questions for thought and study:

(i) How can patients be encouraged to present, in the first place, and in a timely fashion? As Figure 2 shows, a patient who has a disorder but does not make an affirmative decision to seek help is left untreated. Experience shows that often, "early intervention" can be very effective.

(ii) What is the prior probability that a given disorder is present? A very low prior probability (that is, prevalence) brings with it a high probability of "false-positive" diagnoses, no matter how sophisticated the diagnostic tools may be (Barnum and Gleason, in press; Rönnberg, 1999a). Indeed, the formal or informal use of Bayesian revision could help physicians identify criteria that help pinpoint an initial diagnosis (Lilford and Braunholtz, 1996: 607)²⁷

(iii) What are the economic and human costs of an incorrect diagnosis? Sometimes, the costs of an incorrect diagnosis will be low. If the "true" problem gambler is misidentified as a depressive but improves following treatment for the latter disorder, for example, the cost may be negligible. In other situations it may be devastating.

(iv) What is the effectiveness (and cost-effectiveness) of treatment options? In order to make sound decisions, physicians need to know what treatments are most likely to produce sustainable improvements in the patient’s well-being. This again raises the need for relevant outcome measures and follow-up studies.

Understandably, many treatment decisions made by physicians will not reflect cost-related concerns. After all, their primary mission is to maximize the well-being of the patient. But at the societal (or government) level, cost issues do become relevant for it is here that the global allocation decisions are made. These decisions should optimize the use of the existing technologies and knowledge base in the health care community.

(v) Who should provide diagnostic and treatment services? There are many professionals capable of providing some or all of these services, such as physicians, registered psychologists and problem gambling counsellors. Specific perspectives, capabilities and cost implications are associated with each group.

Causality: Pathological gambling may well be a "useful" concept. But as Shaffer and his colleagues (1997) so eloquently suggest, more research is needed to determine whether it is best conceptualized as a primary or a secondary disorder (p. 72-3). If it is indeed a discrete entity, the nature of its causal relationships (or associations) with other disorders such as depression or alcoholism deserves careful study (Spunt et al., 1998:2549).²⁸

Prevention and treatment programs: More efficacy studies are needed, using adequate sample sizes that are representative of the population. Outcome measures (including follow-up) should be selected carefully. Given the relationships that exist between problem gambling and other disorders, researchers should find ways to help physicians unravel the complicated interactions efficiently and effectively (e.g., Walker, Unützer & Katon, 1998). The education community should be included in efforts to design additions to the school curriculum aimed at preventing problem gambling among youth.

Special populations: The growing role of the state in supplying and promoting gambling may have had the effect of altering cultural beliefs such that it has become a "valued" activity. This may be of particular concern for the young. As one reviewer pointed out, young people who have emergent gambling problems often show difficulty and dysfunction in relation to family, school performance and substance use (and abuse). Research is needed to further explore gambling’s association with these outcomes. Additional studies could address whether gambling behaviour is associated with lifestyle norms, support mechanisms, and coping skills–all of which are relevant to the population health paradigm.

Consequences of particular forms of gambling: There are many varieties of gambling available in Canada today. There is evidence that certain games are more appealing to at-risk segments of the population than others, but more studies (and replications) will afford a greater level of confidence to health experts and policymakers. Researchers should also study the outcomes of specific operating practices at casinos (notably hours of service and the availability of alcohol, credit, ATMs and cheque cashing services).

Cost-benefit analyses: The state of cost-benefit analysis could be significantly improved (Henriksson and Lipsey, in press). One area of particular interest is the allocation of health care resources vis-à-vis other activities that are equally (or more) important in promoting population health. Resources used in health care represent a net claim on wealth of the community, and overexpansion of the health care system can have negative impacts (Evans and Stoddart, 1994: 55). The issue is an important one because of the tendency of some provincial governments to use health care needs as a justification for gambling expansion.

The growth of state-sponsored gambling has given rise to a host of complicated (but important) questions. At a time when some have argued that "conceptual chaos" pervades the addictions field (Shaffer, 1997), the paradigm of population health may prove to be an extremely useful tool to refine and study these questions (e.g., Kuhn, 1962). At the same time, it is likely that many projects would benefit from an interdisciplinary approach because they touch on a broad array of fields. Building bridges between these diverse groups is a difficult but probably necessary means to achieve the goal of healthy communities.

¹ Sessional Lecturer, University of British Columbia Faculty of Commerce and Business Administration. (E-mail: Len_Henriksson@bc.sympatico.ca) The author gratefully acknowledges the comments and suggestions provided by Jason Azmier (Canada West Foundation), Virginia Carver, Lise Mattar (both of Health Canada), Dr. Sten Rönnberg (University of Stockholm), Edward Sawka (Alberta Alcohol and Drug Abuse Commission), Dr. Eric Single (University of Toronto), Herb Thompson (Addictions Foundation of Manitoba) and Dr. Harold Wynne (University of Alberta). The author is solely responsible for views expressed, and for any errors of omission or commission.

² "Product life-cycle" is a term from the marketing field. It is an effort to recognize distinct stages in the sales history of a product. It is normally portrayed as an "S"-shaped sales curve with four phases: introduction, growth, maturity and decline (Kotler & Turner, 1979: 244). Many lottery products have a very short life-cycle (Goodman, 1994).

³ This is not to suggest that Canadian researchers have been pressured to suppress or change findings. But those whose studies produce findings that are very unfavourable to gambling or expansion may be genuinely concerned about where their next grant will come from.

⁴ Medicalization can be defined as "the way in which the jurisdiction of modern medicine has expanded … and now encompasses many problems that formerly were not defined as medical entities" (Gabe & Calnan, 1989; Helman, 1994: 156).

⁵ Recent home invasions in some of Canada’s larger cities have produced a flood of campaigns urging residents to be cautious in divulging information to strangers. This could aggravate the non-response problem further.

⁶ As Shaffer and his colleagues note, the response rates of the studies in their meta-analysis ranged from 25% to 100%. Thirty-six studies failed to report a response rate. Of those that did report the response rate, it was incorrectly calculated in over half of the studies (p. 28).

⁷ There appear to be disparate views on the issue of gender balance. Spunt and his colleagues (1998) argue that "the vast majority of research studies on pathological gambling have focused only on male subjects" (p. 2539). The Harvard study is far gentler in its assessment (Shaffer et al., 1997: 25-7, 39-42).

⁸ Rönnberg (1999a) writes, "we do not have good methods to generalize data to a country’s or state’s populations from samples of household members from telephone directories. If you have good information about the population, you may be able to account for non-responders, weight them, and decrease the influence of low response. But if you have a low response rate, a small sample, and little knowledge about the population, your research is very much guesswork."

⁹ As Ferrel and Gold (1998) observed, "(t)he industry seized upon that 1% figure in arguing that the problem’s breadth has been vastly overblown" (p. A24).

¹⁰ In reviewing a meta-analysis, "it is useful for the reader to know who supported a study when deciding how much credence to give to its conclusions. Potential conflicts of interest do not necessarily disqualify a study, but they should be clearly acknowledged" (Sacks et al., 1987: 452). While the Harvard study identified the National Center for Responsible Gaming as the funding agency, the close linkage between that agency and gambling interests was not clearly disclosed.

¹¹ About half of the studies used some form of the South Oaks Gambling Screen (Shaffer et al., 1997: 55).

¹² The authors’ classification of disordered gambling consists of three levels: no problems (Level 1), sub-clinical problems (Level 2) and severe clinical problems (Level 3).

¹³ The criteria by Sacks et al. pertain to randomized controlled trials (not prevalence estimates) but remain relevant.

¹⁴ For example, the high reported incidence of problem gambling in some First Nations should concern health advocates. But the distinct cultural meaning of gambling in many Aboriginal communities (e.g., McMillan, 1995: 204-5) may limit the comparability of incidence rates with those of other cultures.

¹⁵ Example studies include Politzer, Yesalis and Hudak (1992) and National Opinion Research Center (1999).

¹⁶ Example reviews and studies include Goodman (1994) and National Council of Welfare (1996).

¹⁷ The problem of varying definitions of "social cost" (in the context of tobacco, alcohol and other drug abuse) has been reviewed by Choi, Robson and Single (1997, p. 153).

¹⁸ For a review of this issue, see Rice et al., 1990.

¹⁹ Some costs appear easier to estimate than they truly are. For example, one study uses a simple and intuitively appealing method to calculate the employer costs associated with termination. Unfortunately, the calculation does not consider the benefits of turnover for the employer, which are well documented in the human resources literature.

²⁰ For a review of OCD, see Jenike, Baer and Minichiello (1998). Comorbidity is discussed at page 689.

²¹ The researchers cautioned that "additional randomized placebo-controlled and maintenance trials are required to confirm the findings and determine whether improvement persists" (p. 1781).

²² As Rose (1985) writes, "the hardest cause to identify is the one that is universally present, for it then has no influence on the distribution of disease" (p. 32).

²³ The Province of Quebec is a notable exception.

²⁴ For example, one lottery organization posted billboards of a man holding a lottery ticket in a poverty-stricken neighbourhood, above the caption, "This could be your ticket out".

²⁵ Shaffer and his colleagues (1997) pose a similar question: "is pathological gambling a primary disorder?" (p. 71).

²⁶ For example, it ignores the possibility that patients will "cure" themselves through self-help or non-medical means.

²⁷ The Lilford and Braunholtz article includes a concise summary of Bayesian principles.

²⁸ If pathological gambling is only a secondary disorder, subordinate to other dysfunctional behaviour, then pathological gambling "will only exist as a consequent of another condition (e.g. manic episode, antisocial personality, alcohol abuse, obsessive-compulsive disorder or adolescence)" (Jessor and Jessor, 1977; Shaffer et al., 1997:72).

Abrams, D.B. and Niaura, R.S. Social Learning Theory. In Blane, H.T. and Leonard, K.E. (eds.), Psychological Theories of Drinking and Alcoholism (pp. 131-178). New York: Guilford Press.

American Psychiatric Association. (1980). Diagnostic and Statistical Manual of Mental Disorders (3^rd ed.). Washington, DC: Author.

American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders (4^th ed.). Washington, DC: Author.

Angus Reid Group. (1996). Social Gaming and Problem Gambling in British Columbia [Available from British Columbia Lottery Corporation, Kamloops, BC.].

Armstrong, J.S. (1985). Long-Range Forecasting. New York: Wiley.

Auditor-General of British Columbia. (1996/97). A review of government revenue and expenditure programs relating to alcohol, tobacco and gaming. Victoria, BC: Author.

Barnum, D. and Gleason, J. (in press). Analyzing the Proficiency of Technology System Proficiency Studies: The Case of Drug Testing. IEEE Transactions on Engineering Management.

Bland, R., Newman, S., Orn, H. and Stebelski, G. (1993). Epidemiology of pathological gambling in Edmonton. Canadian Journal of Psychiatry/Revue canadienne de psychiatrie, 38, 108-112.

Blaszcynski, A. and Silove, D. (1996). Pathological gambling: forensic issues. Australian and New Zealand Journal of Psychiatry, 30, 358-369.

Bolen, D.W. and Boyd, W.H. (1968). Gambling and the Gambler. Archives of General Psychiatry, 18, 617-630.

Borg, M.O., Mason, P. and Shapiro, S.L. (1991). The Incidence of Taxes on Casino Gambling: Exploiting the Tired and the Poor. American Journal of Economics and Sociology, 50(3), 323-332.

British Columbia Ministry of Employment and Investment. (1997). Gaming Review: Expansion Options and Implications. Victoria, BC: Author.

Burke, J.D. (1996, September). Problem Gambling Hits Home. Wisconsin Medical Journal , 95(9), 611-614.

Campbell, C.S. and Smith, G.J. (1998, March). Canadian Gambling: Trends and Public Policy Issues. Annals of the American Academy of Political and Social Science, 556, 22-35.

Canada West Foundation. (1998, October). The State of Gambling in Canada. Calgary, AB: Author.

Choi, B. C.K., Robson, L. and Single, E. (1997). Estimating the Economic Costs of the Abuse of Tobacco, Alcohol and Illicit Drugs: A Review of Methodologies and Canadian Data Sources. Chronic Diseases in Canada, 18(4), 149-165.

Crockford, D.N. and el-Guebaly, N. (1998, February). Psychiatric Comorbidity in Pathological Gambling: A Critical Review. Canadian Journal of Psychiatry/Revue canadienne de psychiatrie, 43, 43-50.

Dillman, D.A. (1991). The Design and Administration of Mail Surveys. Annual Review of Sociology, 17, 225-249.

Embree, B.G. and Whitehead, P.C. (1993). Validity and Reliability of Self-Reported Drinking Behavior: Dealing with the Problem of Response Bias. Journal of Studies on Alcohol, 54, 334-344.

Evans, R.G. and Stoddart, G.L. (1994). Producing Health, Consuming Health Care. In Evans, R.G., Barer, M.L. and Marmor, T.R. (eds.), Why are some people healthy and others not? (pp. 67-92). New York: Aldine de Gruyter.

Ferrel, D. and Gold, M. (1998, December 14). Casino Industry Fights an Emerging Backlash. Los Angeles Times, A1, A24-A25.

Ferris, J., Stirpe, T. and Ialomiteanu, A. (1996). Gambling in Ontario: A report from a general population. Toronto, ON: Addiction Research Foundation.

Foot, M. (1996). Boom, bust, and echo: how to profit from the coming demographic shift. Toronto, ON: Macfarlane, Walter & Ross.

Friedmann, J., Hakim, S. and Weinblatt, J. (1989). Casino Gambling as a "Growth Pole" Strategy and its Effect on Crime. Journal of Regional Science, 29(4), 615-623.

Gabe, J. and Calnan, M. (1989). The limits of medicine: women’s perception of medical technology. Social Science and Medicine, 28, 223-231.

Gemini Research. (1994). Social Gaming and Problem Gambling in British Columbia. [Available from British Columbia Lottery Corporation, Kamloops, BC].

Goodman, R. (1994). Legalized Gambling as a Strategy for Economic Development. Amherst, MA: Center for Economic Development.

Green, L.W. and Ottoson, J.M. (1999). Community and population health (8^th ed.). New York, NY: McGraw Hill.

Helman, C.G. (1994). Culture, Health and Illness. Oxford, UK: Butterworth-Heinemann.

Henriksson, L.E. (1996). Hardly a Quick Fix: Casino Gambling in Canada. Canadian Public Policy/Analyse de Politiques, XXII(2), 116-128.

Henriksson, L.E. and Lipsey, R.G. (in press). Should Provinces Expand Gambling? Canadian Public Policy/Analyse de Politiques.

Hertzman, C., Frank, J. and Evans, G. (1994). Heterogeneities in Health Status. In Evans, R.G., Barer, M.L. and Marmor, T.R. (eds.), Why are some people healthy and others not? (pp. 67-92). New York: Aldine de Gruyter.

Hollander, E. (1998). Treatment of obsessive-compulsive spectrum disorders with SSRIs. British Journal of Psychiatry, 173 (suppl. 35), 7-12.

Hollander, E., DeCaria, C.M. et al. (1998). Short-Term Single-Blind Fluvoxamine Treatment of Pathological Gambling. American Journal of Psychiatry, 155(12), 1781-1783.

Illinois State Police. (1992). How Casino Gambling Affects Law Enforcement. Springfield, IL: Division of Criminal Investigation Intelligence Bureau.

Jacobs, D.F. (1989). Illegal and undocumented: A Review of Teenage Gambling and the Plight of Children of Problem Gamblers in America. In Shaffer, H.J. (ed.), Compulsive Gambling: Theory, Research and Practice (pp. 249-292). Lexington, MA: Lexington Books.

Jenike, M.A., Baer, L. and Minichiello, W.E. (1998). Obsessive-compulsive disorders: practical management (3^rd ed.). St. Louis, MO: Mosby.

Jessor, R. and Jessor, S.L. (1977). Problem behavior and psychosocial development: A longitudinal study of youth. New York: Academic Press.

Kindig, D.A. (1997). Purchasing Population Health: paying for results. Ann Arbor, MI: University of Michigan Press.

Kotler, P. and Turner, R.E. (1979). Marketing management (3^rd Canadian ed.) Scarborough, ON: Prentice-Hall.

Kuhn, T.S. (1962). The Structure of Scientific Revolutions. Chicago, IL: University of Chicago Press.

Ladouceur, R. et al. (1998). Cognitive treatment of pathological gamblers. Behaviour Research and Therapy, 36, 1111-1119.

Lesieur, H.R. (1998). Costs and Treatment of Pathological Gambling. Annals of the American Academy of Political and Social Science, 556, 153-171.

Lesieur, H.R. and Blume, S.B. (1987). The South Oaks Gambling Screen (SOGS): A New Instrument for the Identification of Pathological Gamblers. American Journal of Psychiatry, 144, 1184-1188.

Lilford, R.H. and Braunholtz, D. (1996, September 7). The statistical basis of public policy: a paradigm shift is overdue. BMJ, 313, 603-7.

Lipsey, R.G. (1997). The Proposed Increase in Legalized Gambling in B.C.: Evidence and Appropriate Procedures. Unpublished manuscript.

Loan Sharks hunt for prey in Quebec casinos. (1997, January 20). Vancouver Sun, B10.

López Viets, V.C. and Miller, W.R. (1997). Treatment Approaches for Problem Gamblers. Clinical Psychology Review, 17(7), 689-702.

McMillan, A.D. (1995). Native peoples and cultures of Canada. Vancouver, B.C.: Douglas & McIntyre, Ltd.

Marshall, K. (1998, Winter). The gambling industry: raising the stakes. Perspectives on Labour and Income, 10(4), 7-11.

Miller, M.M. (1996). Medical Approaches to Gambling Issues I: The Medical Condition. Wisconsin Medical Journal, 95(9), 623-634.

Moran, E. (1995). Majority of secondary school children buy tickets, BMJ, 311, 1225-1226.

Mosher, J.F. (1994). Alcohol Advertising and Public Health: An Urgent Call for Action. American Journal of Public Health, 84, 180-181.

National Council of Welfare. (1996, Winter). Gambling in Canada. Ottawa, ON: Author.

National Opinion Research Center. (1999). Overview of National Survey and Community Database Research on Gambling Behavior [preliminary report to the National Gambling Impact Study Commission.] Chicago, IL: Author.

Nechi Training, Research and Health Promotions Institute. (1994). Spirit of Bingoland: A Study of Problem Gambling Among Alberta Native People. Edmonton, AB: Author.

Nechi Training, Research and Health Promotions Institute. (1995). Firewatch on Aboriginal Adolescent Gambling. Edmonton, AB: Author.

Nicol, J. and Nolen, S. (1998, May 11). The Curse of Casinos. Maclean’s, 44-47.

O’Keefe, A. and Pollay, R.W. (1996). Deadly Targeting of Women in Promoting Cigarettes. JAMWA, 51(1/2), 67-69.

Organ, D. and Podsakoff, P. (1986). Self-Reports in Organizational Research: Problems and Prospects. Journal of Management, 12(4), 531-544.

Overman, S. (1990). Addiction: Odds Are, Gamblers Cost Companies. HR Magazine, 35(4), 50-54.

Padavan, F. Sen. (1994). Rolling the Dice: Why Casino Gambling is a Bad Bet for New York State (Legislative Report, New York State Senate). Albany, NY.

Payer, L. (1996). Medicine and culture: varieties of treatment in the United States, England, West Germany and France. New York, NY: Henry Holt.

Phillips, D.P., Welty, W. and Smith, M. (1997). Elevated Suicide Levels Associated with Legalized Gambling. Suicide & Life-Threatening Behavior, 27(4), 373-378.

Politzer, R.M., Yesalis, C.E. and Hudak, C.J. (1992). The Epidemiologic Model and the Risk of Legalized Gambling: Where Are We Headed? Health Values, 16(2), 20-27.

Proimos, J., DuRant, R.H., Pierce, J.D. and Goodman, E. (1998). Gambling and Other Risk Behaviors Among 8^th- to 12^th-Grade Students. Pediatrics, 102(2), 1-6.

Rachlis, M. and Kushner, C. (1995). Strong Medicine: How to Save Canada’s Health Care System. Toronto, ON: Harper Collins Publishers.

Rice, D.P., Kelman, S., Miller, L.S. and Dunmeyer, S. (1990). The Economic Costs of Alcohol and Drug Abuse and Mental Illness: 1985. San Francisco, CA: Institute for Health and Aging, University of California, 1990.

Ritchie,K., Kildea, D. and Robine, J.M. (1992). The relationship between age and the prevalence of senile dementia: A meta-analysis of recent data. International Journal of Epidemiology, 21(4), 763-769.

Roman, P.M. and Blum, T.C. (1991) The Medicalized Conception of Alcohol-Related Problems: Some Social Sources and Some Social Consequences of Murkiness and Confusion. In Pittman, D.J. and White, H.R. (eds.), Society, Culture and Drinking Patterns Reexamined (pp. 753-774). New Brunswick, NJ: Rutgers Center of Alcohol Studies.

Rönnberg, S. (1999). Några funderingar att diskutera i anslutning till den del av den pågående nationella studien om spel och spelberoende. Unpublished manuscript, University of Stockholm.

Rönnberg, S. (1999a). Personal communication.

Room, R., Turner, N. and Ialomiteanu, A. (1998). Community Effects of the Opening of the Niagara Casino: A First Report [Available from Addiction Research Foundation Division, Centre for Addiction and Mental Health, Toronto].

Rose, G. (1985). Sick Individuals and Sick Populations. International Journal of Epidemiology, 14(1), 32-38.

Rose, G. (1990). Reflections on the changing times. BMJ, 301, 683-687.

Rupcich, N., Frisch, G.R. and Govoni, G. (1997). Comorbidity of Pathological Gambling in Addiction Treatment Facilities. Journal of Substance Abuse Treatment, 14(6), 573-574.

Sacks, H.S., Berrier, J., Reitman, D., Ancona-Berk, V.A. and Chalmers, T.C. (1987). Meta-Analysis of Randomized Controlled Trials. New England Journal of Medicine, 316, 450-455.

Setness, P.A. (1997). Pathological gambling. Postgraduate Medicine, 102(4), 13-18.

Shah, C.P. (1994). Public Health and Preventive Medicine in Canada (3^rd ed.). Toronto, ON: University of Toronto Press.

Shaffer, H.J. (1997). The Most Important Unresolved Issue in the Addictions: Conceptual Chaos. Substance Use and Misuse, 32(11), 1573-1580.

Shaffer, H.J., Hall, M.N. and Vander Bilt, J. (1997). Estimating the Prevalence of Disordered Gambling Behavior in the United States and Canada: A Meta-analysis. Boston, MA: Harvard Medical School Division on Addictions.

Smart, R.G. and Ferris, J. (1996, February). Alcohol, Drugs and Gambling in the Ontario Adult Population. Canadian Journal of Psychiatry/Revue canadienne de psychiatrie, 41, 37.

Sobell, L.C., Brown, J., Leo, G.I. and Sobell, M.B. (1996). The reliability of the Alcohol Timeline Followback when administered by telephone and by computer. Drug and Alcohol Dependence, 42, 49-54.

Spunt, B., Dupont, I., Liberty, H.J. and Hunt, D. (1998). Pathological Gambling and Substance Misuse: A Review of the Literature. Substance Use and Misuse, 33(13), 2535-2560.

Streiner, D.L., Norman, G.R. and Munroe Blum, H. (1989). Epidemiology PDQ. Toronto, ON: BC Decker.

Sylvain, C., Ladouceur, R. and Boisvert, J-M. (1997). Cognitive and Behavioral Treatment of Pathological Gambling: A Controlled Study. Journal of Consulting and Clinical Psychology, 65(5), 727-732.

Thompson, H. (1999). Personal communication.

Volberg, R.A. (1994). The Prevalence and Demographics of Pathological Gamblers: Implications for Public Health. American Journal of Public Health, 84(2),237-241.

Walker, E.A., Unützer, J. and Katon, W. (1998). Understanding and Caring for the Distressed Patient With Multiple Medically Unexplained Symptoms. Journal of the American Board of Family Practice, 11, 347-356.

Wilkinson, R.G. (1992). Income Distribution and Life Expectancy. BMJ, 304, 165-168.

Wynne Resources, Ltd. (1998, June). Adult Gambling and Problem Gambling in Alberta, 1998. Edmonton, AB: Alberta Alcohol and Drug Abuse Commission.