Rates of Problematic Gambling in a British Homeless
Sample: A Preliminary Study
Steve Sharman •Jenny Dreyer •Mike Aitken •Luke Clark •
Published online: 23 January 2014
Springer Science+Business Media New York 2014
Abstract Homelessness and problem gambling are two public health concerns in the UK
that are rarely considered concurrently, and little is known about the extent of gambling
involvement and problematic gambling in the homeless. We recruited 456 individuals
attending homelessness services in London, UK. All participants completed a screen for
gambling involvement, and where gambling involvement was endorsed, the Problem
Gambling Severity Index (PGSI) was administered. The PGSI risk categories were com-
pared against data from the 2010 British Gambling Prevalence Survey (BGPS). PGSI
problem gambling was indicated in 11.6 % of the homeless population, compared to 0.7 %
in the BGPS. Of participants endorsing any PGSI symptoms, a higher proportion of
homeless participants were problem gamblers relative to the low and moderate risk groups,
compared to the BGPS data. These results conﬁrm that the homeless constitute a vul-
nerable population for problem gambling, and that diagnostic tools for gambling
involvement should be integrated into homelessness services in the UK.
S. Sharman (&)M. Aitken L. Clark
Department of Psychology, University of Cambridge, Downing Street, Cambridge CB2 3EB, UK
Connection @ St Martins, 12 Adelaide Street, London WC2N 4HW, UK
Kings College, London, UK
Department of Psychology, Institute of Psychiatry, Kings College London, De Crespigny Park, London
SE5 8AF, UK
National Problem Gambling Clinic, Soho Centre for Health & Care, 1 Frith Street, London W1D 3HZ,
Department of Medicine, Imperial College London, South Kensington Campus, London SW7 2AZ,
J Gambl Stud (2015) 31:525–532
Keywords Problem gambling Homelessness Prevalence UK London
Problem gambling and homelessness are public health concerns that have a signiﬁcant
impact on affected individuals, their families, and the wider society. Until recently,
problem gambling was categorized as an Impulse Control Disorder in the Diagnostic and
Statistical Manual Fourth Edition (DSM-IV, APA 1994). Based on shared characteristics
with substance use disorders, including symptom hallmarks like tolerance, craving and
withdrawal, the recent DSM-5 has reclassiﬁed gambling disorder into the ‘Addiction and
Related Disorders’ category (APA 2013). Existing research in the homeless has revealed
elevated levels of mental health problems (Scott 1993), including drug and alcohol use
disorders (Wincup et al. 2003), as well as depression and loneliness (Sumerlin 1995).
Despite this evident vulnerability of the homeless population to addictive disorders, little
research has been done to investigate the relationship between homelessness and gambling,
and indeed these issues are predominantly researched independently.
Within the general population, it is well-recognized that gambling is a popular leisure
activity. In the UK, the 2010 British Gambling Prevalence Survey (BGPS, Wardle et al.
2010) found that 73 % of adults reported gambling in some form over the past 12 months.
Using the Problem Gambling Severity Index (PGSI, Ferris and Wynne 2001) to quantify
disordered gambling, the BGPS indicated a prevalence rate of 0.7 %, which is comparable
to rates in other countries in Europe (e.g. Switzerland, 0.8 %, Bondolﬁ et al. 2008) and
slightly lower than prevalence rates suggested for North America (US, 1.6 %, Shaffer et al.
1999; Canada, 2.0 %, Cox et al. 2005).
Ascertaining levels of homelessness is a difﬁcult task due to the transient nature of the
homeless population, combined with the fact that many homeless people strive to stay
anonymous. Homelessness can constitute either living in temporary accommodation, or
‘sleeping rough’ on the streets (the government deﬁnition of rough sleepers is ‘people
sleeping…or bedded down in the open air…or other places not designed for habitation’,
(Department for Communities and Local Government 2012). Council ﬁgures for Central
London in the year starting April 2011 recorded 5,678 rough sleepers (12 % female) in
contact with outreach workers (Street to Home Annual Report 2012), with around half of
these in the Westminster authority in Central London where the current study was con-
ducted. These ﬁgures represented a 43 % increase from the previous year (2010–2011),
indicating that this is not only a signiﬁcant social problem in London, but one that appears
to be increasing.
A small number of previous international studies have examined levels of gambling and
problematic gambling in the homeless. Using the Massachusetts Gambling Screen (based
on the DSM-IV criteria), Shaffer et al. (2002) studied 171 homeless people who were
seeking treatment for substance use disorders. ‘At-risk’ gambling was present in 18.3 %,
and pathological gambling in 5.5 %. Two studies using the South Oaks Gambling Screen
(SOGS, Lesieur and Bloom 1987) reported a prevalence rate of probable pathological
gambling of 17.2 % in 87 individuals relying on community assistance in Canada (LePage
et al. 2000), and 23 % in 275 homeless people in the USA (Nower et al. 2014). Qualitative
studies have also provided some insight on the causal connection here, that gambling
represents a contributory factor for some individuals becoming homeless (Holdsworth and
526 J Gambl Stud (2015) 31:525–532
Tiyce 2012; van Laere et al. 2009). For example, gambling was listed in the top ten
contributing factors for homelessness in older adults surveyed across the US, Australia and
UK (Crane et al. 2005).
The present study represented the ﬁrst attempt to measure levels of gambling
involvement and problem gambling in homeless individuals accessing services in the UK.
We investigated the distribution of gambling involvement as indicated by PGSI risk cat-
egories. Secondary aims were to characterize any gender differences, associations with
current housing circumstances (i.e. temporary accommodation versus rough sleeping), and
the different preferred forms of gambling engaged in by the homeless.
Participants were recruited from 16 homeless centres across Westminster, London in
January 2012 (n=456). The centres from which participants were recruited included
shelters, hostels and day centres. From the overall sample, 160 participants provided age
(m=41.9, SD 11.9, range 18–78), 264 participants provided gender information (246
male, 18 female) and 291 provided current housing circumstance (206 Hostel Residents, 83
Rough sleepers, 2 squatters). The recruitment and study protocol was given ethical
approval by Kings College London. The study was also approved by the commissioning
manager of Westminster City Council’s Rough Sleeping team. Participants were informed
of the nature of the study, and provided verbal consent.
Problem Gambling Severity Index (PGSI, Ferris and Wynne 2001)
The PGSI is a 9 item questionnaire measuring gambling severity, derived from the longer
Canadian Problem Gambling Index. Each item is scored from 0 to 3 (never =0, some-
times =1, most of the time =2, almost always =3), resulting in a total score range of
0–27. Cronbach’s areliability coefﬁcient was 0.95 in the present study, indicating that this
is a reliable scale to use on this population. The gambling risk categories were based upon
Currie et al. (2010): a score of 0 indicates a non-problem gambler, scores of 1–4 indicate a
‘low risk’ gambler, 5–7 indicate a ‘moderate risk’ gambler and a score of [7 indicates a
Data Collection and Analysis
To allow comparison of prevalence rates with the BGPS, a full PGSI was administered to
participants who scored C1 on item one of the PGSI (‘‘In the last 12 months, have you bet
more than you could afford to lose?’’). All participants (n=457) completed this item as a
screening question, with 363 (79.6 %) scoring zero by answering ‘‘0—never’’ to this item.
At some study sites, participants who answered ‘never’ to item one were nonetheless
administered the full PGSI (n=147). Of these, 135 participants (91.8 %) scored zero. In
the remaining 12 individuals who answered never to item one, but scored [0 on the full
scale, 6 scored in the low risk category, six in the moderate risk category, and none scored
in the problem gambler category.
J Gambl Stud (2015) 31:525–532 527
To establish any difference between prevalence rates in the homeless compared to the
existing BGPS dataset, a Chi square (v
) analysis was conducted. The distribution of
gamblers scoring C1 amongst risk proﬁles within the homeless and the BGPS dataset was
also analysed using a Chi square analysis. A pvalue of \.05 was considered signiﬁcant.
Participant’s scores on the PGSI were classiﬁed as no risk (n=350), low risk (n=38),
moderate risk (n=15) or problem gamblers (n=53) based on the published thresholds.
In the overall homelessness sample (n=456), the rate of problem gambling on the
PGSI was 11.6 %, with moderate risk gambling in 3.3 %, low risk gambling in 8.3, and
76.8 % registering no risk (i.e. PGSI =0) (see Fig. 1a).
In subjects who conﬁrmed their gender, males (n=246) had a problem gambling rate
of 20.8 %, a moderate risk rate of 5.6 %, a low risk rate of 14.2 % and no risk rate of
59.4 %. In the female participants (n=18), the problem gambling rate was 5.5 %, the
moderate risk rate was also 5.5 %, the low risk rate was 11.1 % and the no risk rate was
77.8 %. Due to the very low number of conﬁrmed female participants, gender differences
are not discussed further.
Comparing the proportions of the overall sample falling in the different gambling risk
groups against the general UK population from the BGPS, there was a reliable difference
between two datasets (v
(3) =11.1, p\.011), with the largest differences in the no risk
and problem gambler groups. Removing the no risk group, and analysing only participants
with PGSI scores C1, there was a reliable difference in the proportions of participants in
the ‘at risk’ categories between the homeless dataset and the BGPS data (v
p\.001). The BGPS data indicates a stepwise decline in prevalence as gambling severity
increases (i.e. moving from low risk to problem gambling). In contrast, in the homeless
sample, there was a signiﬁcantly greater proportion of problem gamblers relative to the low
risk and moderate risk categories (see Fig. 1b).
Further analysis looked at the gambling risk categories as a function of current housing
circumstance (see Fig. 2a). There was a signiﬁcant difference in risk proﬁles between
Hostel Residents and Rough Sleepers, (v
(2) =9.9, p=.007): hostel residents displayed
a larger proportion of low risk gambling, whereas the rough sleepers displayed a higher
rate of problem gambling (Fig. 2a). In 106 participants who indicated game preferences,
electronic roulette machines and horse racing were the most popular gambling activities;
online and casino gambling were the least common (Fig. 2b).
In a convenience sample of service-accessing homeless individuals attending outreach
centres in Central London, UK, the rate of problem gambling detected using the PGSI was
11.6 %, which is substantially higher than the general population ﬁgures for the UK
indicated by the BGPS. A second ﬁnding is that the distribution of PGSI risk categories
differed markedly in our homeless sample relative to the BGPS data. While the BGPS data
show the expected proﬁle of decreasing prevalence with greater gambling severity, of those
individuals who scored C1 on the PGSI, the proportion of problem gamblers was sub-
stantially higher in the homeless sample, and the proportion of low risk gamblers was
substantially lower. This high rate of problem gambling was evident in our homeless
528 J Gambl Stud (2015) 31:525–532
participants who self-reported male gender, and it was particularly evident in the rough
sleepers. Analysis of gambling type demonstrated that shop-based gambling activities
including electronic roulette (on a ‘Fixed Odds Betting Terminal’), slot machines and
sports/horse betting, were the most common forms of betting among the homeless problem
These observed rates of problem gambling in the homeless are similar to past studies
from North America (LePage et al. 2000, 17.2 %; Nower 2014, 23 %; Shaffer et al. 2002,
5.5 %). It is possible that our detected rate is a conservative estimate due to the imple-
mentation of our screening question, which assumed an overall PGSI score of zero for
participants who did not endorse the ﬁrst item on the PGSI. This assumption was supported
by an analysis in a subset of our sample (n=147) who did not endorse the screening item
but nevertheless completed the full PGSI. While 92 % of these participants did score zero
on the full scale, 12 participants manifested some level of problematic gambling, and thus
our screening question may have slightly under-estimated the overall prevalence rate. The
instrument for assessing problematic gambling also differs across studies; the two studies
demonstrating the highest prevalence rates (LePage et al., Nower et al.) used the SOGS,
Fig. 1 PGSI risk proﬁle including and excluding ‘No Risk’ category
Fig. 2 Risk proﬁle per housing status and gambling activity
J Gambl Stud (2015) 31:525–532 529
Shaffer et al. used the MAGS, and our study used the PGSI. It is possible that these
different screens may capture distinct facets of problematic gambling amongst the
homeless; cross-screen tool validation in this population has yet to be conducted.
Despite the possibility of a conservative estimate, the prevalence ﬁgure for problem
gambling in the homeless of 11.6 % is dramatically higher than the population prevalence
estimate from the BGPS. It should be acknowledged that the BGPS data were collected by
post, resulting in the exclusion of a number of vulnerable populations, including the
homeless (also the prison population and student halls of residence). Thus, the BPGS
prevalence ﬁgure is itself likely to represent an overly conservative estimate of problem
gambling in the general population. One caveat to this comparison is that the BGPS data
were collected prior to 2010, and our data were collected in 2012, and therefore it is
possible that national gambling involvement may have ﬂuctuated over this time; however
the increase in problem gambling rates observed between previous versions of the BPGS
(from 1999 to 2007, and from 2007 to 2010) are minimal, and unlikely to inﬂuence these
The elevated prevalence of problem gambling was particularly apparent in males.
However, the sampling strategy did not attain equal representation of both genders, and a
very small number of homeless women participated, precluding statistical analysis by
gender. The heavily male-weighted sample could also bias the analysis of preferred
gambling forms, as the most popular activities were those found in bookmakers’ shops,
which are traditionally male-dominated environments. Within the subgroup of the home-
less who reported at least some level of gambling, the rate of problem gambling was
particularly elevated in the rough sleepers compared to hostel residents. One possible link
between sleeping status, gambling and gambling type may be the shelter offered by high
street gambling venues in the UK. We estimated that there were 61 such venues in the
immediate vicinity of our outreach centres. Amusement arcades and bookmaker’s shops
can have lengthy opening hours, with typical opening times from 8 a.m. to 11 p.m., and
some open 24 h. High street amusement arcades also offer very low stake gambling, from
as little as 5 p per play, and some offer free hot drinks and snacks. Extended exposure to
such an environment may increase risk of problem gambling in the homeless. Consistent
with this notion, the most common forms of gambling among our cohort were those offered
by bookmakers (roulette machines, sports betting).
The observed increased rate of problem gambling amongst the homeless population
highlights the relationship between poverty and ﬁnancial risk taking. When faced with
poverty, an individual may display risky behaviour in an effort to exit poverty (Sadler
2000). In the case of homelessness, the experienced level of poverty is extremely severe.
Nevertheless, our data do not allow any conclusions to be drawn regarding the directional
causality, as to whether problem gambling is a cause or a consequence of homelessness.
We also note that our sample was self-selecting, in that we were only able to recruit
individuals who accessed services provided by Westminster Local Authority.
This is the ﬁrst study to use a clinically recognized diagnostic tool to show a signiﬁcantly
higher rate of problem gambling in a service-accessing homeless population compared to
the general population in the UK. We observed a markedly higher proportion of problem
gamblers compared to low-risk gamblers in the homeless. Our ﬁndings conﬁrm that
homeless people constitute a vulnerable population for excessive gambling, and imply that
530 J Gambl Stud (2015) 31:525–532
the problems of homelessness and problem gambling may beneﬁt from being addressed
concurrently rather than independently. Homelessness services should consider including
questions about gambling behavior in their support pathways, to enable homeless indi-
viduals to better access treatment.
Conﬂict of interest The authors declare that there is no conﬂict of interest in this study.
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.).
Washington, DC: American Psychiatric Association.
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.).
Arlington, VA: American Psychiatric Publishing.
Bondolﬁ, G., Osiek, C., & Ferrero, F. (2008). Prevalence estimates of pathological gambling in Switzerland.
Acta Psychiatrica Scandinavica, 117, 236–239.
Cox, B., Yu, N., Aﬁfﬁ, T., & Ladouceur, R. A. (2005). National survey of gambling problems in Canada.
Canadian Journal of Psychiatry, 50, 213–217.
Crane, M., Byrne, K., Fu, R., Lipmann, B., Mirabelli, F., Rota-Bartelink, A., et al. (2005). The causes of
homelessness in later life: Findings from a 3-nation study. Journal of Gerontology Social Sciences,
Currie, S. R., Casey, D. M., & Hodgins, D. C. (2010). Improving the psychometric properties of the problem
gambling severity index. Report submitted to the Interprovincial Problem Gambling Research Consor-
tium. Retrieved from http://www.ccgr.ca/wp-content/uploads/2013/03/Improving-the-Psychometric-
Ferris, J., & Wynne, H. (2001). The Canadian Problem Gambling Index user manual. Report to the
Canadian inter-provincial task force on problem gambling. Canada: The Canadian Centre on Sub-
Holdsworth, L., & Tiyce, M. (2012). Exploring the hidden nature of gambling problems among people who
are homeless. Australian Social Work, 65(4), 474–489.
Homelessness Statistics. Department for Communities and Local Government. Retrieved Aug 7th, 2013 from
Lepage, C., Ladouceur, R., & Jacques, C. (2000). Prevalence of problem gambling among community
service users. Community Mental Health Journal, 36(6), 597–601.
Lesieur, H., & Blume, S. (1987). The South oaks gambling screen (SOGS): A new instrument for the
identiﬁcation of pathological gamblers. American Journal of Psychiatry, 144, 1184–1188.
Nower, L., Eyrich-Garg, K. M., Pollio, D. E., & North, C. S. (2014). Problem gambling and homelessness:
Results from an epidemiologic study. Journal of Gambling Studies, 1–13.
Petry, N. M., Stinson, F. S., & Grant, B. F. (2005). Comorbidity of DSM-IV pathological gambling and
other psychiatric disorders: Results from the national epidemiologic survey on alcohol and related
conditions. Journal of Clinical Psychiatry, 66(5), 564–574.
Sadler, M. A. (2000). Escaping poverty: Risk-taking and endogenous inequality in a model of equilibrium
growth. Review of Economic Dynamics, 3(4), 704–725.
Scott, J. (1993). Homelessness and mental illness. The British Journal of Psychiatry, 162(3), 314–324.
Shaffer, H. J., Freed, C. R., & Healea, D. (2002). Gambling disorders among homeless persons with
substance use disorders seeking treatment at a community centre. Psychiatric Services, 53(9),
Shaffer, H. J., Hall, M. N., & Vander Bilt, J. (1999). Estimating the prevalence of disordered gambling
behaviour in the United States and Canada: A research synthesis. American Journal of Public Health,
Shaffer, H. J., LaBrie, R., Scanlan, K. M., et al. (1994). Pathological gambling among adolescents: Mas-
sachusetts gambling screen (MAGS). Journal of Gambling Studies, 10, 339–362.
Street to Home Annual Report, 1st April 2011–31st March 2012. (2012). Retrieved Aug 7th, 2013, from
Sumerlin, J. R. (1995). Adaptation to homelessness: Self-actualization, loneliness, and depression in street
homeless men. Psychological Reports, 77(1), 295–314.
J Gambl Stud (2015) 31:525–532 531
van Laere, I. R., de Wit, M. A., & Klazinga, N. S. (2009). Pathways into homelessness: Recently homeless
adults problems and service use before and after becoming homeless in Amsterdam. BMC Public
Wardle, H., Moody, A., Spence, S., Orford, J., Volberg, R., Jotangia, D., et al. (2010). British Gambling
Prevalence Survey 2010. London (UK): National Centre for Social Research.
Welte, J., Barnes, G., Wieczorek, W., Tidewell, M., & Parker, J. (2001). Alcohol and gambling pathology
among US adults: Prevalence, demographics patterns and comorbidity. Journal of Studies on Alcohol,
Wincup, E., Buckland, G., & Bayliss, R. (2003). Youth homelessness and substance use: Report to the drugs
and alcohol research unit. London: Home Ofﬁce.
532 J Gambl Stud (2015) 31:525–532