What gambling screening can tell you
The Problem Gambling Severity Index (PGSI) and South Oaks Gambling Screen (SOGS) are standardized questionnaires used to identify patterns of potentially harmful gambling and to decide when a fuller evaluation is warranted. A PGSI score describes reported gambling-related risk and problems over a specified period. A SOGS result indicates how closely answers resemble patterns associated with serious gambling problems in the populations where that instrument has been studied. Neither score, on its own, is a clinical diagnosis of gambling disorder. Diagnostic assessment examines behavior over time, clinically significant distress or impairment, psychiatric and medical explanations, financial and interpersonal consequences, and immediate safety needs. These distinctions are reflected in the 2025 NICE gambling-related harms guideline and in research comparing screening scales with diagnostic interviews.
The distinction matters in both directions. Someone with a high questionnaire score may require urgent support for debt, family conflict, or psychological distress even before a clinician establishes a formal disorder. Someone with a lower score may still face serious consequences that the questionnaire has not captured. A valid assessment therefore treats the score as one source of information within a wider understanding of the person's gambling and circumstances. The World Health Organization explicitly recognizes gambling harms below the diagnostic threshold.
Screening, case identification, and diagnosis
Screening is a systematic way of flagging people who may benefit from further questions, support, or clinical assessment. Case identification can begin with one direct, respectful question about gambling or its consequences. Severity assessment characterizes the extent and pattern of reported difficulties, and a formal diagnostic interview tests whether established criteria are met. These processes may occur together in a specialist service, but they answer different questions. A public-health survey estimates patterns in a population; a clinical encounter determines needs and care for an individual.
The PGSI is especially familiar from community surveys. The SOGS originated as a screen for pathological gambling, the earlier diagnostic terminology. Both have also appeared in clinical research, but the fact that a scale discriminates between groups does not establish that it can accurately classify every individual in primary care, treatment settings, adolescents, or culturally distinct populations. The systematic review by Otto and colleagues (2020) found that far fewer screening studies met stringent diagnostic-reference and methodological requirements than the number of questionnaires in circulation might suggest.
A clinical diagnosis of gambling disorder is governed by the relevant diagnostic system and professional judgment. The American Psychiatric Association places gambling disorder in the category of substance-related and addictive disorders as the recognized nonsubstance condition. ICD-11 also identifies gambling disorder under disorders due to addictive behaviors. Neither system uses the PGSI category labels or a SOGS threshold as a substitute for diagnostic requirements. See the American Psychiatric Association's clinical explanation and the WHO ICD-11 clinical descriptions.
The Problem Gambling Severity Index
What the PGSI measures
The PGSI is a nine-item component of the Canadian Problem Gambling Index. Its content addresses several complementary dimensions: needing to wager more money for the same excitement, staking more than one can afford, chasing losses, borrowing or selling something to gamble, perceiving gambling as a problem, health effects including stress or anxiety, criticism by others, financial difficulties, and guilt. This combination intentionally covers gambling behavior and experienced consequences. It is designed for measuring a continuum of problems in general-population contexts, and it is often used to characterize risk groups in epidemiological studies. Holtgraves (2009) examined its psychometric structure and explained its role relative to the clinically derived SOGS.
Standard PGSI administration asks about the preceding 12 months. Each item is scored 0 for never, 1 for sometimes, 2 for most of the time, or 3 for almost always. Responses are summed, producing a total from 0 to 27. The conventional interpretive bands are 0 (non-problem gambling according to the items), 1–2 (low-risk gambling), 3–7 (moderate-risk gambling), and 8–27 (problem gambling). These labels follow a survey classification convention; they are not DSM-5-TR severity grades. The UK Gambling Commission provides the question content, scoring, and cautions regarding interpretation.
Interpreting a PGSI score of zero
Zero means that the respondent did not endorse any of the nine experiences at the relevant response frequency during the reference period. It does not prove the complete absence of gambling harm. The person might have gambled without recognizing a consequence, misunderstood an item, avoided disclosure, experienced a consequence outside the nine topics, or encountered harm after the questionnaire period. Population surveys may also assign a score of zero to people who did not gamble during the preceding year; studies should state clearly how they handle nongamblers. A clinician should still attend to specific disclosed financial, psychological, or relationship difficulties. This is a measurement boundary, rather than a contradiction in the person's experience.
Interpreting scores from one to seven
A score of 1 or 2 suggests that at least one concerning indicator has been reported, while scores of 3 to 7 indicate more or more frequent reported difficulties. The descriptive terms low risk and moderate risk should not be read as predictions that an individual will inevitably progress to gambling disorder. Nor should they imply that harm is only hypothetical: some people in these categories already experience consequences. The Gambling Commission's methodological analysis discusses why category labels can obscure ongoing harm and why trajectories vary across longitudinal studies. Currie, Hodgins, and Casey (2013) specifically assessed the validity of PGSI interpretive categories.
A person reporting a single intense consequence may need more help than another person with a somewhat larger total. For example, gambling-related loss of rent money, fear of a violent response to debt, or acute suicidal thoughts requires attention based on the consequence itself. Adding questionnaire points cannot express urgency reliably; a total score loses information about which questions were endorsed, their context, and who else was affected.
What a PGSI score of eight or more means
Eight or more is the conventional threshold for the PGSI problem-gambling category. It signals a pattern warranting careful follow-up, discussion of harms, and consideration of specialized support. It does not establish that the person meets four DSM diagnostic criteria, satisfies ICD-11 requirements, or has a particular treatment prognosis. A screening category cannot be translated into a medical diagnosis by a simple numerical equivalence. The distinction is explicit in national gambling statistics and in NICE's clinical guidance.
The same number may arise from different answer patterns. Elevated gambling-related guilt and criticism can coexist with financial disruption; repeated loss chasing can occur alongside relatively limited recognition of a problem. The answer profile, narrative description, and recent changes in risk matter more clinically than an isolated total. Investigators may use alternative PGSI thresholds for particular research questions, but such cutoffs require context-specific validation and must be labeled rather than silently substituted for conventional categories.
The South Oaks Gambling Screen
Purpose, historical origin, and content
The SOGS was introduced by Lesieur and Blume in 1987 as a structured questionnaire associated with then-current concepts of pathological gambling. It contains 20 scored items addressing gambling-related experiences such as perceived loss of control, concealment, chasing, criticism, borrowing, and consequences. It was designed before the contemporary DSM-5-TR formulation of gambling disorder; its historical role is central to understanding why its item content and threshold cannot be interpreted as a present-day diagnostic checklist. The original development study included participants with substance-use and gambling diagnoses, mutual-help participants, students, and hospital employees (Lesieur and Blume, 1987).
SOGS scoring conventionally yields a total from 0 to 20. A score of 5 or higher has commonly served as a marker of probable pathological gambling in research and screening. This is an instrument-specific historical cutoff, not a current DSM threshold. Variants differ in whether questions refer to lifetime history or the previous 12 months, and the wording, sampling design, and score interpretation must be checked before comparing results. A lifetime endorsement can remain positive long after behavior has changed, so assessment of current needs requires an explicit time frame.
Strengths and limitations of the SOGS
The SOGS has a substantial research history and helps identify clinically important gambling difficulties. Its diagnostic classification performance nevertheless depends on the setting, population, reference standard, and cutoff. In Stinchfield's 2002 validation study, SOGS distinguished many people in a gambling-treatment sample appropriately, while its performance was weaker in the general-population sample and included substantial false-positive classification relative to DSM-IV criteria. Quoting an impressive sensitivity or specificity from a specialist sample without stating the denominator and reference standard would mislead a reader.
SOGS also reflects the diagnostic concepts of its era and devotes marked attention to financial sources and borrowing. Contemporary gambling disorder includes important features such as attempts to stop and continued engagement despite harm that need to be established through clinical discussion. An old threshold may be useful for research comparability; it cannot replace updated criteria. NICE therefore names SOGS and PGSI as examples of tools clinicians may consider while separately recommending a broad assessment of history, context, co-occurring conditions, and risk (NICE, 2025).
PGSI versus SOGS: choosing a screening instrument
The central comparison concerns purpose. PGSI was developed for population-level measurement of gambling problems on a graded scale and is widely used for surveillance. SOGS grew out of clinical case-finding and older pathological-gambling constructs and often uses a binary threshold to flag possible severe difficulties. Both collect self-report data, both can identify problems deserving attention, and both are influenced by disclosure, recall, wording, and which consequences are measured. Their numeric totals have different ranges, so a PGSI 8 and a SOGS 8 do not have equivalent meanings.
A clinician choosing between them should establish the question to be answered: Is a service identifying possible gambling harms? Estimating population prevalence? Characterizing change over repeated visits? Comparing findings with a historical cohort? Selecting a tool based only on brevity or a published accuracy percentage can import an unsuitable cutoff into a new setting. The chosen version should be validated in the target language, age range, and clinical setting and administered using its intended reference period. The primary outcome should also be specified: gambling harm, a PGSI-defined category, or a formal gambling disorder diagnosis.
In measurement research, consistency of scores is called reliability; correspondence with the intended construct is validity; and ability to classify cases against a credible diagnostic reference is diagnostic accuracy. High internal consistency does not guarantee that a screen correctly identifies a clinical condition. Correlation between two questionnaires can demonstrate related constructs while leaving the accuracy of both unresolved. The stringent review by Otto and colleagues (2020) underscores the need for independent diagnostic interviews, representative samples, and transparent performance reporting.
How diagnostic accuracy should be evaluated
Sensitivity, specificity, and predictive value
Sensitivity is the proportion of people with the reference-standard disorder who screen positive; specificity is the proportion without that disorder who screen negative. Positive predictive value is the proportion of screen-positive people who truly meet the diagnostic reference standard; negative predictive value is the corresponding proportion of screen-negative people without the disorder. Predictive values depend on how common the condition is among those tested, so a positive screen may be less informative in a low-prevalence community than in a high-risk treatment service, even when sensitivity and specificity are unchanged.
A thought experiment illustrates the base-rate problem without claiming that any particular gambling questionnaire has the example's accuracy. Imagine 1,000 adults, 20 of whom meet independently established diagnostic criteria, and an imaginary test with 90% sensitivity and 90% specificity. It flags 18 true cases and 98 people who do not meet the criteria, for 116 positive results. Only about 16% of positive results would be true cases. The numerical example is hypothetical. Its lesson is that a screen designed for triage should lead to assessment and support rather than automatic labeling.
Researchers must also consider verification bias if only screen-positive participants receive the diagnostic interview; spectrum bias if testing focuses on treatment seekers rather than community members; incorporation bias if the reference diagnosis uses the same questionnaire items; language and cultural translation differences; and changes in DSM criteria across decades. Reports should give actual denominators and uncertainty intervals rather than present performance as a universal property of a test. NICE's 2025 evidence review of assessment tools identified gaps in evidence for accuracy across care settings.
False positives and false negatives have human consequences
A false-positive label can provoke shame, reduce trust, create insurance or employment concerns in some settings, and distort prevalence estimates. A false negative can delay recognition of accumulating debt, suicidal distress, or harm to family members. The response to these errors is better clinical process: explain what a questionnaire assesses, discuss positive responses individually, offer confidential opportunities to disclose, repeat measurements when clinically appropriate, and pursue diagnostic assessment when warranted. A clinical decision should never rest on a total score alone.
DSM-5-TR diagnosis and the diagnostic interview
In DSM-5-TR, gambling disorder is defined by persistent and recurrent problematic gambling resulting in clinically significant impairment or distress. At least four of nine specified criteria must occur within a 12-month period, and the pattern must be distinguished from gambling behavior better accounted for by a manic episode. The nine themes are increasing wager size for excitement, restlessness or irritability when stopping, repeated unsuccessful control efforts, preoccupation, gambling when distressed, returning to recover losses, concealment, jeopardized important relationships or opportunities, and reliance on others to relieve gambling-caused financial crises. These are diagnostic concepts summarized by the American Psychiatric Association.
A clinician clarifies the frequency, timing, persistence, subjective experience, and functional effects of each criterion. Borrowing for gambling may indicate a serious concern yet requires the context of financial resources and behavior over time. Conversely, a person could display marked preoccupation, failed attempts to cut back, or persistent loss chasing without framing those experiences as harm on a questionnaire. Clinical judgment also addresses severity, course, remission, environmental pressures, and the possibility of incomplete or conflicting information.
ICD-11 approaches gambling disorder through impaired control over gambling, increasing priority assigned to gambling, and continuation or escalation despite negative consequences, together with significant distress or impairment. The pattern is generally evident for at least 12 months, with a shorter period possible when requirements are fully met and symptoms are severe. ICD-11 explicitly includes predominantly online and predominantly offline presentations. For classification and country-specific practice, clinicians refer to the WHO's ICD-11 clinical descriptions and applicable national guidance.
What a comprehensive gambling assessment examines
Gambling history and exposure
Assessment begins by establishing the first gambling experiences and the evolution of spending, frequency, duration, venues, and products. It explores sports betting, casino games, slots, lotteries, online and mobile gambling, live betting, and gambling-like activities where monetary risk is involved. Product accessibility, rapid repeated bets, promotions, advertising, payment methods, and the ability to gamble privately can contribute to risk exposure. Frequency and expenditure require interpretation against income, obligations, and actual consequences. Someone making infrequent but catastrophic wagers may need urgent care even if total gambling time is low.
Questions about loss chasing help identify a cycle in which a person returns to gambling to recover losses, often increasing exposure to further losses. Other relevant experiences include feeling compelled to continue after near misses, beliefs about luck or control over random outcomes, seeking relief from distress, and difficulty resisting cues. Such processes can help explain behavior and guide intervention, but identifying a cognitive distortion or strong emotional motivation does not itself establish a disorder. Each is considered within the broader pattern of behavior, harm, and functioning.
Financial, family, and occupational consequences
Gambling impacts should be recorded concretely: missed housing payments, high-interest loans, depletion of savings, payday borrowing, concealment of debt, loss of work or study opportunities, sleep disruption, conflict, caregiving effects, and dependence on relatives. A clinical history may include attempts to access money or support after a loss and whether coercion, financial abuse, or domestic violence is present. Family members may suffer serious harm regardless of whether the person who gambles satisfies diagnostic criteria. WHO's public-health account places these consequences within the wider definition of gambling-related harm.
The assessment should distinguish financial pressure caused by gambling from financial precarity that precedes it, because both pathways can occur and reinforce one another. The person may also experience shame that limits disclosure. Respectful, specific questions about day-to-day realities usually reveal more than an accusation or generalized description such as irresponsible spending. Professional financial and social support may be warranted alongside psychological care.
Mental health, physical health, and differential diagnosis
Co-occurring depression, anxiety, substance use problems, ADHD, trauma-related symptoms, sleep disturbance, and other psychiatric or neurodevelopmental conditions can influence gambling behavior and treatment needs. Assessment should establish the timing and relationship of symptoms instead of assuming that gambling caused every additional condition. A history of episodic markedly elevated mood, decreased need for sleep, and broader risk taking warrants evaluation for mania or hypomania; medication effects, including impulse-control changes associated with certain dopamine agonists, can also be clinically relevant. NICE specifically recommends attention to these contexts (NICE, 2025).
Functional impairment is assessed in context: the person's responsibilities, income and financial dependents, relationships, health, work or education, and ability to control participation despite adverse outcomes. Some heavy or habitual gambling may be associated with no currently identifiable diagnostic syndrome, while severe personal or family harms can exist at a subdiagnostic level. A careful interview allows both possibilities to be described accurately.
Suicide and self-harm risk assessment
Gambling-related debt, shame, relationship crisis, and psychological distress can accompany elevated suicide risk. The scale total is not a suicide-risk assessment and must never be used to exclude acute risk. NICE recommends directly asking people experiencing gambling harm about suicidal ideation and intent, assessing past attempts, and arranging urgent specialist or emergency help when risk is considerable or immediate (NICE NG248, recommendations 1.1.11–1.1.13). This assessment is relevant even when a person does not disclose other established psychiatric risk factors.
A clinician should assess current thoughts, intent, plans, access to means, recent losses, intoxication, protective supports, and capacity to remain safe, then determine an appropriate safety response and follow-up. If someone may act on suicidal thoughts or cannot stay safe, emergency or crisis services should be contacted immediately. Screening for gambling harm and responding to an urgent crisis are different clinical tasks, and the safety response takes priority.
When brief screens and alternatives are appropriate
Short case-finding instruments may be useful when time or setting makes a nine- or twenty-item questionnaire impractical. The three-item Brief Biosocial Gambling Screen (BBGS) was developed using population data and historical DSM-IV-TR definitions, with promising accuracy in its development analysis (Gebauer, LaBrie, and Shaffer, 2010). Its original performance estimates should not be assumed to transfer unchanged to contemporary DSM-5-TR criteria, all cultures, or every clinical environment.
The absence of any questionnaire may also be appropriate at the very first conversation. NICE advises practitioners to ask directly about gambling in settings including mental health care, substance-use services, financial distress, homelessness, and some medication-related impulse-control concerns. The person can then be offered a validated measure and a broader assessment if concerns arise (NICE NG248). A supportive first question may make disclosure more likely than a premature diagnostic label.
Assessment among adolescents and different populations
Adult PGSI and SOGS norms should not be treated as automatic diagnostic standards for adolescents. Some youth research uses age-adapted measures, including the SOGS-RA, but adapting wording, reference periods, and scoring changes the instrument and requires its own validation. Gambling opportunities and financial dependence also differ by age. Family involvement, safeguarding, and applicable confidentiality requirements should be considered in age-appropriate care. A teenager's gaming activity needs a separate gaming-related assessment; video game engagement is not equivalent to wagering money on uncertain outcomes.
Language proficiency, cultural interpretations of gambling, local products and regulations, disability, socioeconomic pressures, and patterns of online participation can affect what respondents recognize and report. Translated instruments need validation beyond literal word substitution. Older adults may face different financial vulnerabilities and social circumstances. Clinical practice also needs accessible communication for people with intellectual disability, acquired brain injury, and neurodivergent traits, while distinguishing those traits from the diagnostic construct of gambling disorder.
Using screening in treatment and follow-up
Repeated PGSI or other validated scores can help track selected self-reported experiences during treatment, provided the same version and time frame are used and the changes are interpreted cautiously. A reduction in score may reflect meaningful improvement, but a 12-month recall period can remain influenced by events before treatment began. Recovery is better evaluated through multiple outcomes: gambling participation and control, safety, money management, family stability, functioning, distress, and goals identified by the person receiving care. A lower score without safer circumstances is an incomplete outcome.
Screening is often the beginning of collaborative treatment planning. Depending on assessment, care may include gambling-specific cognitive behavioral therapy, motivational approaches, financial and practical support, self-exclusion measures, management of co-occurring disorders, and family or peer support. The strength of evidence differs across interventions and outcomes. NICE recommends gambling-specific CBT and an individualized pathway for people experiencing harms, with particular attention to complex co-occurring needs (NICE NG248). This article addresses measurement and clinical assessment; detailed treatment effectiveness requires its own distinct evidence review.
How to interpret an online gambling addiction test
An online PGSI-style result can be a prompt for reflection or help seeking, especially if the site transparently identifies the instrument and scoring. It becomes misleading when it presents a category as a confirmed medical diagnosis, claims that a particular score proves brain disease, or asserts a precise personal prognosis. Some websites alter the item wording, shorten the scale, change the recall period, or combine items from different measures; scores from such tools should not automatically be treated as equivalent to the validated original.
A responsible result page identifies the tool and reference period, explains whether the assessment has been clinically validated, states what the score can and cannot tell the individual, and provides access to appropriate support. A reader can also consider the concrete experiences behind the answers: gambling beyond affordable limits, concealment, repeated unsuccessful attempts to stop, debt, distress, and harm to relationships. Asking for a professional discussion is reasonable at any score when such concerns arise.
Questions to bring to a clinical appointment
Helpful information includes the types of gambling used, approximate frequency and recent expenditure, money lost relative to essential needs, whether losses are chased, attempts to reduce gambling, borrowing and debts, changes in sleep or mood, current medications, substance use, family consequences, and any immediate safety concerns. Previous questionnaire results can be brought to an appointment with the date and version recorded. People can also describe what they hope will change and which financial, relationship, or mental-health problems are most pressing.
Evidence quality, uncertainty, and research priorities
The established findings are that PGSI and SOGS are recognized gambling screening measures, that gambling disorder has independent diagnostic criteria, and that harm can occur below diagnostic thresholds. Evidence also supports asking about gambling when risk factors or relevant clinical presentations are present. Less certain is the portability of one numeric cutoff across populations, languages, age groups, diagnostic reference standards, and care environments. The 2020 systematic review by Otto and colleagues and 2025 NICE evidence review B both identify substantial limitations in the comparative diagnostic-accuracy literature.
Differences between studies may arise from recruitment, instrument version, weighting, recall period, self-report concealment, prevalence, and changing diagnostic manuals. Many studies evaluate correlations or internal consistency without an independent blinded diagnostic reference. Strong clinical validation requires prespecified thresholds, representative recruitment, inclusion of screen-negative participants in reference interviews, clear handling of missing responses, reporting of false positives and false negatives, and calibration in external samples. Research should also measure harms experienced by relatives and the socioeconomic context beyond what a nine-item score captures.
Frequently asked questions
Is PGSI 8 or higher a diagnosis of gambling disorder?
No. Eight or higher is the conventional PGSI problem-gambling classification. Formal diagnosis requires an assessment against DSM-5-TR or ICD-11 requirements, with attention to impairment, context, and alternative explanations. A person in the PGSI 8+ category should be offered informed follow-up and support without being reduced to the label.
Can a PGSI score of zero rule out harm?
No. It means none of the PGSI's nine assessed indicators was reported in the measured period. A particular severe consequence, a family member's harm, or a changing risk situation can still justify intervention. A zero score should be interpreted alongside direct discussion and practical circumstances.
What SOGS score suggests serious gambling problems?
Five or more is a commonly used historical threshold for probable pathological gambling. The meaning of that threshold depends on the instrument version, time frame, population, and reference standard; it is a prompt for fuller assessment rather than a confirmed diagnosis. Clinical services may use other validated thresholds for specified purposes, which should be stated explicitly.
Is PGSI or SOGS the more accurate test?
There is no universal winner. Accuracy depends on what outcome is being assessed, how participants were recruited, which versions were used, and how diagnosis was independently determined. PGSI is particularly common in population research; SOGS has a long history in clinical screening. Choosing a test requires matching the measure to the intended use and local validation evidence.
Should someone seek help before reaching a threshold?
Yes. Financial danger, psychological distress, family harm, repeated failed control efforts, and suicidal thoughts deserve attention regardless of questionnaire category. A qualified health professional or gambling treatment service can discuss the situation, establish what is happening, and organize support. Immediate danger to life requires emergency or crisis assistance.
Related Articles
Cluster Gambling, Gaming and Behavioral Addictions — the overarching research chapter on recognized disorders, harms, assessment, and treatment.
Habit Formation vs Addiction: Learning, Motivation, and Compulsion — conceptual context for distinguishing repeated behavior from clinically significant impairment.
Stress and Addiction: Coping, Negative Reinforcement, and Relapse — related mechanisms of distress-driven repetition, interpreted according to the condition under study.
Gambling Reward Psychology: Near Misses, Loss Chasing, and Illusion of Control — experimental and cognitive research on gambling feedback, learning, loss chasing, and perceived control, complementary to screening and clinical assessment.
References
American Psychiatric Association. (2024). Gambling disorder. Psychiatry.org.
Currie, S. R., Hodgins, D. C., and Casey, D. M. (2013). Validity of the Problem Gambling Severity Index interpretive categories. Journal of Gambling Studies, 29, 311–327. https://doi.org/10.1007/s10899-012-9300-6.
Gambling Commission. (2021, updated). Problem gambling screens. Official methodology.
Gebauer, L., LaBrie, R., and Shaffer, H. J. (2010). Optimizing DSM-IV-TR classification accuracy: A brief biosocial screen for detecting current gambling disorders among gamblers in the general household population. Canadian Journal of Psychiatry, 55(2), 82–90. https://doi.org/10.1177/070674371005500204.
Holtgraves, T. (2009). Evaluating the Problem Gambling Severity Index. Journal of Gambling Studies, 25, 105–120. https://doi.org/10.1007/s10899-008-9107-7.
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(9), 1184–1188. https://doi.org/10.1176/ajp.144.9.1184.
National Institute for Health and Care Excellence. (2025a). Gambling-related harms: Identification, assessment and management (NG248). Clinical recommendations.
National Institute for Health and Care Excellence. (2025b). Tools for identification and assessment of harmful gambling: Evidence review B. NCBI Bookshelf.
Otto, J. L., and colleagues. (2020). A systematic review evaluating screening instruments for gambling disorder finds lack of adequate evidence. Journal of Clinical Epidemiology. https://doi.org/10.1016/j.jclinepi.2019.12.022.
Stinchfield, R. (2002). Reliability, validity, and classification accuracy of the South Oaks Gambling Screen (SOGS). Addictive Behaviors, 27(1), 1–19. https://doi.org/10.1016/S0306-4603(00)00158-1.
World Health Organization. (2024a). Clinical descriptions and diagnostic requirements for ICD-11 mental, behavioural and neurodevelopmental disorders. Official publication.
World Health Organization. (2024b). Gambling. Fact sheet.