Abstracts

Introduction

Alcohol use is a major public health problem ¹1. GBD 2016 Alcohol Collaborators. Alcohol use and burden for 195 countries and territories, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet 2018; 392:1015-35. which remains controversial among the general audience, stakeholders, and scientists. Individuals are strongly opinionated and the debate is often instigated by vested interests - either ideological, political, or financial ²2. Martino FP, Miller PG, Coomber K, Hancock L, Kypri K. Analysis of alcohol industry submissions against marketing regulation. PLoS One 2017; 12:e0170366.^,33. Casswell S. Vested interests in addiction research and policy. Why do we not see the corporate interests of the alcohol industry as clearly as we see those of the tobacco industry? Addiction 2013; 108:680-5.. Nevertheless, research on the consequences of alcohol consumption at individual and public health level is a huge scientific field comprising both natural and social sciences. As each area brings its specificities and evolves individually, several definitions and research methods are available. Its our job to create common ground to advance the field and the public policies.

Alcohol policies are the most effective prevention against alcohol-related harms ⁴4. Edwards G, Anderson P, Babor TF, Casswell S, Ferrence R, Giesbrecht N, et al. Alcohol policy and the public good. Cary: Oxford University Press; 1994.^,55. Babor TF, Caetano R, Casswell S, Edwards G, Giesbrecht N, Graham K, et al. Alcohol: no ordinary commodity. New York: Oxford University Press; 2003.^,66. Room R, Babor T, Rehm J. Alcohol and public health. Lancet 2005; 365:519-30.. Beyond acting in the health system, these policies encompass pricing and taxation, physical availability regulation, drinking and driving countermeasures, advertising and marketing restrictions, among others ⁵5. Babor TF, Caetano R, Casswell S, Edwards G, Giesbrecht N, Graham K, et al. Alcohol: no ordinary commodity. New York: Oxford University Press; 2003.^,66. Room R, Babor T, Rehm J. Alcohol and public health. Lancet 2005; 365:519-30.^,77. Monteiro M. The World Health Organization's SAFER initiative and the challenges in Brazil to reducing harmful consumption of alcoholic beverages. Epidemiol Serv Saúde 2020; 29:e202000.. Population-based surveys are milestones for planning, implementing, and monitoring these policies; it is therefore essential - though challenging - to measure alcohol consumption in a standardized, systematic, and reproducible way. The first challenge to remember when selecting questions to assess alcohol use is that survey results are likely to support population-level prevention strategies instead of “high-risk” (individual-level) strategies. For those of us who had clinical training, the seminal Sick Individual and Sick Populations by Geoffrey Rose is a must-read paper which explains this difference ⁸8. Rose G. Sick individuals and sick populations. Int J Epidemiol 1985; 14:32-8.. Although the professor only briefly mentioned heavy drinking, his theory was widely accepted in the field ⁹9. Skog O-J. The prevention paradox revisited. Addiction 1999; 94:751-7..

A second challenge are the many ways to investigate alcohol consumption. Choosing appropriate questions depends on the main objective of the survey (which is often pre-defined in public bids). Such questions (and their number/length) vary if the survey aims to specifically assess alcohol consumption/harms (such as the U.S. National Epidemiologic Survey on Alcohol and Related Conditions - NESARC) ¹⁰10. Grant BF, Dawson DA, Stinson FS, Chou SP, Dufour MC, Pickering RP. The 12-month prevalence and trends in DSM-IV alcohol abuse and dependence: United States, 1991-1992 and 2001-2002. Drug Alcohol Depend 2004; 74:223-34.^,1111. National Institute on Alcohol Abuse and Alcoholism. National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III). https://www.niaaa.nih.gov/research/nesarc-iii (accessed on 15/Aug/2021).
https://www.niaaa.nih.gov/research/nesar... , alcohol and substance use (i.e., U.S. National Survey on Drug Use and Health - NSDUH) ¹²12. Substance Abuse and Mental Health Services Administration. National Survey on Drug Use and Health (NSDUH). https://www.samhsa.gov/data/data-we-collect/nsduh-national-survey-drug-use-and-health (accessed on 15/Aug/2021).
https://www.samhsa.gov/data/data-we-coll... , or general health/behavior/risk factors (i.e., Behavioral Risk Factor Surveillance System - BRFSS) ¹³13. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System (BRFSS). https://www.cdc.gov/brfss/index.html (accessed on 15/Aug/2021).
https://www.cdc.gov/brfss/index.html... . In the United States, each of the above surveys is conducted by a different agency (the National Institute on Alcohol Abuse and Alcoholism - NIAAA, the Substance Use and Mental Health Services Administration - SAMHSA, and the Centers for Disease Control and Prevention - CDC, respectively).

Data from population-based surveys are also used to feed monitoring systems on alcohol and health. The World Health Organization (WHO), for example, has a Global Information System on Alcohol and Health (GISAH) ¹⁴14. Poznyak V, Fleischmann A, Rekve D, Rylett M, Rehm J, Gmell G. The world health organization's global monitoring system on alcohol and health. Alcohol Res 2013; 35:244-9.^,1515. World Health Organization. Global Information System on Alcohol and Health. https://www.who.int/data/gho/data/themes/global-information-system-on-alcohol-and-health (accessed on 15/Aug/2021).
https://www.who.int/data/gho/data/themes... . The GISAH provides a comprehensive overview of alcohol indicators worldwide, which are also compiled in the Global Status Report on Alcohol and Health¹⁶16. Poznyak V, Rekve D. Global status report on alcohol and health 2018. Geneva: World Health Organization; 2018..

In Brazil, nationally representative household surveys investigating alcohol consumption are quite recent. The pioneer study was conducted in 2001 ¹⁷17. Galduróz JCF, Carlini EA. Use of alcohol among the inhabitants of the 107 largest cities in Brazil - 2001. Braz J Med Biol Res 2007; 40:367-75. and the struggle to obtain optimal reproducibility continues until now ¹⁸18. De Boni RB, Vasconcellos MTL, Silva PN, Coutinho C, Mota J, Peixoto JNB, et al. Reproducibility on science: challenges and advances in Brazilian alcohol surveys. Int J Drug Policy 2019; 74:285-91.. Furthermore, the different methods used resulted in huge discrepancies across surveys and with international indicators. These discrepancies could confuse clinicians, researchers, and stakeholders who are unfamiliar with alcohol epidemiology, undermining the demand for stronger policies.

One possible step to improve communication and avoid misinterpretation of alcohol-related indicators is answering to the “Five W’s of Epidemiology”. The “Five W’s” is a mnemonic for the fundamental questions of descriptive Epidemiology: What (health event definition), Who (person/population), Where (place), When (time), and Why (causes, risk factors, modes of transmission) ¹⁹19. Centers for Disease Control and Prevention. Principles of epidemiology in public health practice. Atlanta: Centers for Disease Control and Prevention; 2012.. They help us to remember the key methodological information necessary to understand and compare data across different studies and countries as well as to assess the soundness of results. This paper aims to exemplify this approach by describing some of the GISAH indicators and the corresponding data available in the latest Brazilian household surveys.

Method

Nine alcohol-related indicators from the WHO GISAH ¹⁵15. World Health Organization. Global Information System on Alcohol and Health. https://www.who.int/data/gho/data/themes/global-information-system-on-alcohol-and-health (accessed on 15/Aug/2021).
https://www.who.int/data/gho/data/themes... were described to answer the “Five W’s”. The indicators were selected based on data from population-based surveys (although whenever a country has no survey or data is incomplete, indicators can be mathematically estimated using predefined algorithms). Only indicators referring to the adult population were included, following the three domains specified at GISAH: Level of Consumption, Patterns of Consumption, and Harms and Consequences - Morbidity (Table 1).

To show how the “Five W’s” would be answered by the information obtained from Brazilian surveys, key methodological information and results from the following nationwide household surveys conducted face-to-face are presented:

(1) 2nd Brazilian Household Survey About the Use of Psychotropic Drugs (II LNUDPB) conducted by the Brazilian Center of Information on Psychotropic Drugs (CEBRID) in 2005 ²⁰20. Centro Brasileiro de Informações sobre Drogas Psicotrópicas. I Levantamento domiciliar sobre o uso de drogas psicotrópicas no Brasil: estudo envolvendo as 108 maiores cidades do país, 2005. São Paulo: Centro Brasileiro de Informações sobre Drogas Psicotrópicas; 2006.;

(2) 2nd Brazilian National Alcohol and Drugs Survey (LENAD II) conducted by the National Institute of Science and Technology for Public Policies on Alcohol and Other Drugs, Federal University of São Paulo (INPAD/ UNIFESP) in 2011-2012 ²¹21. Unidade de Pesquisas em Álcool e Drogas; Instituto Nacional de Ciência e Tecnologia para Políticas Públicas do Álcool e Outras Drogas. Segundo Levantamento Nacional de Álcool e Drogas. Relatório 2012. São Paulo: Universidade Federal de São Paulo; 2012.;

(3) 3rd Brazilian National Survey on Drug Use by the Brazilian Population (3rd LNUD) conducted by the Institute of Communication and Scientific and Technological Information in Health, Oswaldo Cruz Foundation (ICICT/Fiocruz) in 2015 ²²22. Bastos FIPM, Vasconcellos MTL, De Boni RB, Reis NB, Coutinho CFS. 3rd National Survey on Drug Use by the Brazilian Population. https://www.arca.fiocruz.br/bitstream/icict/34614/2/iii%20lnud_english.pdf (accessed on 15/Aug/2021).
https://www.arca.fiocruz.br/bitstream/ic... ;

(4) 2nd Brazilian National Health Survey (PNS) conducted by the Brazilian Institute of Geography and Statistics (IBGE) in 2019 ²³23. Instituto Brasileiro de Geografia e Estatística. Pesquisa Nacional de Saúde: 2019. Percepção do estado de saúde, estilos de vida, doenças crônicas e saúde bucal: Brasil e grandes regiões. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2020.^,2424. Painel de indicadores de saúde - Pesquisa Nacional de Saúde. https://www.pns.icict.fiocruz.br/painel-de-indicadores-mobile-desktop/ (accessed on 15/Aug/2021).
https://www.pns.icict.fiocruz.br/painel-... ;

Only information from official/summary/executive reports publicly available online are presented. Data were selected from reports instead of scientific papers since those are a source of information for many stakeholders and lay citizens.

Results and discussion

What - definitions on alcohol consumption

Standardized definitions and questionnaires to assess alcohol use in population-based surveys are plenty available, as well as their main advantages/disadvantages, psychometric properties, and biases. Several studies ²⁵25. Bloomfield K, Stockwell T, Gmel G, Rehn N. International comparisons of alcohol consumption. Alcohol Res Health 2003; 27:95-109.^,2626. Gmel G, Rehm J. Measuring alcohol consumption. Contemp Drug Probl 2004; 31:467-540.^,2727. National Institute on Alcohol Abuse and Alcoholism. Recommended alcohol questions. https://www.niaaa.nih.gov/research/guidelines-and-resources/recommended-alcohol-questions (accessed on 15/Aug/2021).
https://www.niaaa.nih.gov/research/guide... ^,2828. Rehm J. How should prevalence of alcohol use disorders be assessed globally? Int J Methods Psychiatr Res 2016; 25:79-85. have been conducted to increase their accuracy and reduce measurement error. Young researchers on the field and stakeholders must understand the objective of the questions/measures and why they should be selected. Though these methods evolve, they are usually tested before their implementation in the highly expensive population surveys.

Levels of consumption

The main indicator to evaluate alcohol consumption at the population level is “Alcohol per capita consumption (in liters of pure alcohol) - APC”. It can be obtained using data from administrative public data on sales, considered the best estimate of alcohol consumption in any country ²⁹29. Hall WD, Chikritzhs TN, d'Abbs PHN, Room RGW. Alcohol sales data are essential for good public policies towards alcohol. Med J Aust 2008; 189:188-9.^,3030. Rehm J, Baliunas D, Borges GLG, Graham K, Irving H, Kehoe T, et al. The relation between different dimensions of alcohol consumption and burden of disease: an overview. Addiction 2010; 105:817-43.^,3131. Rehm J, Klotsche J, Patra J. Comparative quantification of alcohol exposure as risk factor for global burden of disease. Int J Methods Psychiatr Res 2007; 16:66-76.^,3232. Rehm J, Gmel GE, Gmel Sr. G, Hasan OSM, Imtiaz S, Popova S, et al. The relationship between different dimensions of alcohol use and the burden of disease - an update. Addiction 2017; 112:968-1001.. In fact, evidence shows that compared to sales data, all alcohol-related surveys underestimate alcohol consumption from 30% to 70%.

The WHO compiles total APC from almost all the signatory countries profiting from sales, agriculture, import/export, and other administrative records. Total APC, however, is an average for the entire population - including both individuals who drink and do not drink alcohol. For estimating APC among drinkers it is necessary to conduct a survey and obtain the number of individuals who have drank within a specific time frame. In Brazil, it was estimated that the total APC (including recorded and unrecorded alcohol consumption) was 7.8 liters of pure alcohol in 2016. Table 1 shows that, among drinkers, the APC was estimated at 19.8 liters.

Patterns of consumption

Quantity-frequency (QF) of alcohol consumption varies across population groups. These variations can be understood as “patterns of alcohol consumption”, which influence the frequency of both acute (such as drinking and driving, violence, sexual risk behaviors) and chronic alcohol-related problems (such as alcohol dependence and other clinical outcomes, including cirrhosis, cardiovascular diseases, and cancer) ⁵5. Babor TF, Caetano R, Casswell S, Edwards G, Giesbrecht N, Graham K, et al. Alcohol: no ordinary commodity. New York: Oxford University Press; 2003..

Five indicators are classified as patterns of alcohol consumption in the GISAH: lifetime abstainers, 12-month abstainers, former drinkers, 12-month drinkers, and heavy episodic drinking (past 30 days) (Table 1).

To estimate the proportion of lifetime abstainers, surveys must ask regarding lifetime use of alcohol. This can appear as a filter question such as “Have you ever drank an alcoholic beverage during your lifetime?” ¹²12. Substance Abuse and Mental Health Services Administration. National Survey on Drug Use and Health (NSDUH). https://www.samhsa.gov/data/data-we-collect/nsduh-national-survey-drug-use-and-health (accessed on 15/Aug/2021).
https://www.samhsa.gov/data/data-we-coll... , where if individuals answer they have not, they are considered lifetime abstainers and all the additional questions are imputed as “no”. The same logic is used to estimate the number of 12-month abstainers (and drinkers) whereas the estimated proportion of former drinkers combines these two measures (lifetime and 12-month drinking). Lifetime drinking/abstention may also be helpful, over time, to estimate the number of individuals who started drinking in a given year (incidence) ³³33. Lemmens P. Measuring lifetime drinking histories. Alcohol Clin Exp Res 1998; 22(2 Suppl):29S-36S.. Nevertheless, compared with 12-month and 30-day, lifetime drinking is most likely to present recall bias ³⁴34. Johnson T, Fendrich M. Modeling sources of self-report bias in a survey of drug use epidemiology. Ann Epidemiol 2005; 15:381-9.^,3535. Simon GE, VonKorff M. Recall of psychiatric history in cross-sectional surveys: implications for epidemiologic research. Epidemiol Rev 1995; 17:221-7..

Twelve-month drinking is a sine qua non question to evaluate alcohol use disorders (AUD) and the prevalence of alcohol consumers and abstainers per year. In turn, past 30 days-drinking is the most accurate timeframe to assess QF, being used to evaluate heavy episodic drinking and binge drinking.

Heavy episodic drinking (HED - i.e., drinking six or more alcoholic drinks in a single occasion) is a WHO definition whereas binge drinking (drinking five or more alcoholic drinks - or four drinks, for women - within a couple of hours) is a NIAAA definition ³⁶36. National Institute on Alcohol Abuse and Alcoholism. NIAAA council approves definition of binge drinking. NIAAA Newsletter 2004; 2004:3.. These are sometimes used as synonyms to estimate the ingestion of 60-80g of pure ethanol (which may lead to alcohol intoxication, i.e., “drunkenness”). One alcoholic drink is a standardized dose of ethanol (12-14g) which usually corresponds to a can of beer or a glass (120mL) of wine or 30mL of spirits (whisky, “cachaça”, gin, etc.). When asking about binge drinking, researchers must clarify what they mean by “one drink” since the percentage of pure ethanol varies among the types of beverages and cultures. These are approximate measures, which are not used for pharmacological studies and do not indicate the precise alcohol blood level for an individual.

Binge drinking/HED is the alcohol use pattern presenting the largest effect on the incidence of acute alcohol-related problems, morbidity and mortality at the population level. In addition, it is also a good predictor of alcohol dependence ³⁷37. Graff-Iversen S, Jansen MD, Hoff DA, Høiseth G, Knudsen GP, Magnus P, et al. Divergent associations of drinking frequency and binge consumption of alcohol with mortality within the same cohort. J Epidemiol Community Health 2013; 67:350-7.^,3838. Plunk AD, Syed-Mohammed H, Cavazos-Rehg P, Bierut LJ, Grucza RA. Alcohol consumption, heavy drinking, and mortality: rethinking the J-shaped curve. Alcohol Clin Exp Res 2014; 38:471-8.. Some authors argue it is the single most important question to screen for alcohol problems. In 2017, Rehm et al. ³²32. Rehm J, Gmel GE, Gmel Sr. G, Hasan OSM, Imtiaz S, Popova S, et al. The relationship between different dimensions of alcohol use and the burden of disease - an update. Addiction 2017; 112:968-1001. recommended adding HED in epidemiological surveys due to its overwhelming impact to the burden of disease. It was also suggested that HED could be used as a surrogate measure for estimating AUD - because it brings less stigma and is easily operationalized ²⁸28. Rehm J. How should prevalence of alcohol use disorders be assessed globally? Int J Methods Psychiatr Res 2016; 25:79-85..

Harms and consequences - morbidity

Alcohol consumption influences the risk of about 230 health problems, including infectious diseases, noncommunicable diseases (NCDs), cancer, and injuries. Three main mechanisms cause these harms: (i) toxic effects on the body (resulting mainly in chronic diseases such as liver disease, heart disease, and cancer); (ii) intoxication (see binge drinking above); and (iii) AUDs ³⁹39. World Health Organization. Global status report on alcohol and health 2014. Geneva: World Health Organization; 2014.. Much information related to harms from toxic effects on the body and from intoxication are/should be available on health information systems (such as Brazilian Mortality Information System - SIM and Brazilian Information System for Notificable Diseases - SINAN), traffic report systems (such as Brazilian National Traffic Department - DENATRAN), and violence register systems (which are not unified in Brazil). AUDs, however, are both mental health disorders and risk factors for other health/social conditions. Especially since AUDs are among the most prevalent and the least diagnosed/treated mental health disorders worldwide ⁴⁰40. Mitchell AJ, Meader N, Bird V, Rizzo M. Clinical recognition and recording of alcohol disorders by clinicians in primary and secondary care: meta-analysis. Br J Psychiatry 2012; 201:93-100.^,4141. Carvalho AF, Heilig M, Perez A, Probst C, Rehm J. Alcohol use disorders. Lancet 2019; 394:781-92., estimating its proportion in population-based surveys is essential.

The Lancet’s latest Seminar on Alcohol Use Disorders ⁴¹41. Carvalho AF, Heilig M, Perez A, Probst C, Rehm J. Alcohol use disorders. Lancet 2019; 394:781-92. provides a comprehensive review on how AUD is diagnosed. Briefly, AUD are defined differently by two diagnostic systems that evolved independently: the Diagnostic and Statistical Manual of Mental Disorders (DSM, version 5, since 2015) ⁴²42. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (DSM-5). Washington DC: American Psychiatric Association; 2013. and the International Classification of Disease (ICD, version 11, since 2018) ⁴³43. World Health Organization. ICD-11: International Classification of Disease. Geneva: World Health Organization; 2018.. The latest revision of the DSM remains controversial and many studies have compared it to the previous DSM-4 and to ICD-10 and ICD-11 ⁴¹41. Carvalho AF, Heilig M, Perez A, Probst C, Rehm J. Alcohol use disorders. Lancet 2019; 394:781-92.^,4444. Agrawal A, Heath AC, Lynskey MT. DSM-IV to DSM-5: the impact of proposed revisions on diagnosis of alcohol use disorders. Addiction 2011; 106:1935-43.^,4545. Grant BF, Chou SP, Saha TD, Pickering RP, Kerridge BT, Ruan WJ, et al. Prevalence of 12-month alcohol use, high-risk drinking, and DSM-IV alcohol use disorder in the United States, 2001-2002 to 2012-2013: results from the National Epidemiologic Survey on Alcohol and Related Conditions. JAMA Psychiatry 2017; 74:911-23.^,4646. Hasin DS, Auriacombe M, Borges G, Bucholz K, Budney A, Crowley T, et al. The DSM-5 field trials and reliability of alcohol use disorder. Am J Psychiatry 2013; 170:442-3.^,4747. Saunders JB, Degenhardt L, Reed GM, Poznyak V. Alcohol use disorders in ICD-11: past, present, and future. Alcohol Clin Exp Res 2019; 43:1617-31.. The revision of DSM-4, which agreed with ICD-10 ⁴⁷47. Saunders JB, Degenhardt L, Reed GM, Poznyak V. Alcohol use disorders in ICD-11: past, present, and future. Alcohol Clin Exp Res 2019; 43:1617-31., significantly changed AUD criteria. Overall, in DSM-5, AUD is diagnosed by 11 criteria, where its severity (light, moderate, and severe) is defined by the number of positive criteria. ICD-11, in turn, divides AUD into harmful use and alcohol dependence, the latter being the most severe. In short, alcohol dependence is characterized by compulsive heavy alcohol use and loss of control over intake, which continues despite adverse consequences and the availability of other rewarding activities ⁴¹41. Carvalho AF, Heilig M, Perez A, Probst C, Rehm J. Alcohol use disorders. Lancet 2019; 394:781-92.^,4343. World Health Organization. ICD-11: International Classification of Disease. Geneva: World Health Organization; 2018. (Figure 1). Standardized questions and questionnaires based either on DSM or ICD criteria can be used to screen for or diagnose AUD ⁴⁸48. Babor T, Higgins-Biddle J, Saunders J, Monteiro M. AUDIT - the Alcohol Use Disorders Identification Test: guidelines for use in primary care. Geneva: World Health Organization; 2001.^,4949. Amorim P. Mini International Neuropsychiatric Interview (MINI): validação de entrevista breve para diagnóstico de transtornos mentais. Rev Bras Psiquiatr 2000; 22:106-15.^,5050. Cottler LB, Robins LN, Helzer JE. The reliability of the CIDI-SAM: a comprehensive substance abuse interview. Br J Addict 1989; 84:801-14..

Table 2 shows the results from the Brazilian probability sample surveys. Probability sample surveys are designed to obtain the prevalence of something in the entire population without interviewing all individuals (the only study which evaluates all the population is the census). In these surveys, punctual prevalence is the population-weighted average of the proportions of subjects with the outcome being investigated in population strata defined in complex sampling.

Assessing the precision of the estimated prevalence value (“margin of error” ) requires estimating the 95% confidence interval (95%CI). The true value in the entire population, with 95% certainty, is therefore any value within the interval. For instance, the II LNUDPB ²⁰20. Centro Brasileiro de Informações sobre Drogas Psicotrópicas. I Levantamento domiciliar sobre o uso de drogas psicotrópicas no Brasil: estudo envolvendo as 108 maiores cidades do país, 2005. São Paulo: Centro Brasileiro de Informações sobre Drogas Psicotrópicas; 2006. estimated that 74.6% of the sample have drunk some alcoholic beverage during their lifetime, but the “true” estimate in the Brazilian population, with a 95% of certainty, is any value from 70.3% to 78.9%. To exemplify this, in the case of an election poll, the estimate would be “74.6% with a margin of error of 4.3 percentage points”.

The 95%CI is inversely related to sample size, i.e., the larger the sample, the lower the interval and the more precise is the estimate. The PNS ²³23. Instituto Brasileiro de Geografia e Estatística. Pesquisa Nacional de Saúde: 2019. Percepção do estado de saúde, estilos de vida, doenças crônicas e saúde bucal: Brasil e grandes regiões. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2020.^,2424. Painel de indicadores de saúde - Pesquisa Nacional de Saúde. https://www.pns.icict.fiocruz.br/painel-de-indicadores-mobile-desktop/ (accessed on 15/Aug/2021).
https://www.pns.icict.fiocruz.br/painel-... , which interviewed more than 90,000 persons, thus has a narrower 95%CI (Table 2).

Furthermore, whenever 95%CIs overlap, punctual prevalence cannot be considered different, assuming a 0.05 type I error. For instance, the prevalence of binge drinking was 16.5% in the 3rd LNUD ²²22. Bastos FIPM, Vasconcellos MTL, De Boni RB, Reis NB, Coutinho CFS. 3rd National Survey on Drug Use by the Brazilian Population. https://www.arca.fiocruz.br/bitstream/icict/34614/2/iii%20lnud_english.pdf (accessed on 15/Aug/2021).
https://www.arca.fiocruz.br/bitstream/ic... and 17.1% in the PNS ²³23. Instituto Brasileiro de Geografia e Estatística. Pesquisa Nacional de Saúde: 2019. Percepção do estado de saúde, estilos de vida, doenças crônicas e saúde bucal: Brasil e grandes regiões. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2020.^,2424. Painel de indicadores de saúde - Pesquisa Nacional de Saúde. https://www.pns.icict.fiocruz.br/painel-de-indicadores-mobile-desktop/ (accessed on 15/Aug/2021).
https://www.pns.icict.fiocruz.br/painel-... , but the prevalence in the 3rd LNUD cannot be considered lower because its 95%CIs overlap with that of PNS. Notably, only the 3rd LNUD and the PNS provide 95%CIs for all the estimates in the reports.

Who - study population

The second “W” is “Who”, i.e, the definition of the study population. Since the population-based surveys mentioned in this study aim to represent the entire population living in households, certain groups were not interviewed nor analyzed, such as the institutionalized, incarcerated individuals living in the streets, Indigenous people living in protected areas, etc.

Furthermore, studies may include different age ranges. While GISHA includes individuals older than 15 years, the II LNUDPB and 3rd LNUD include individuals 12-65 years, the LENAD II ²¹21. Unidade de Pesquisas em Álcool e Drogas; Instituto Nacional de Ciência e Tecnologia para Políticas Públicas do Álcool e Outras Drogas. Segundo Levantamento Nacional de Álcool e Drogas. Relatório 2012. São Paulo: Universidade Federal de São Paulo; 2012. includes those older than 14 years, and PNS includes those 18 years and older. Since alcohol consumption varies with age (e.g., adults usually drink more than adolescents, and young adults drink more than older adults), the mean prevalence of alcohol consumption and harms can also vary according to the age group evaluated.

To interpret HED, AUD, and alcohol dependence, analyzing the denominator is essential to understand if the prevalence refers to the entire population (drinkers and non-drinkers) or only to drinkers. Both measures are useful for different purposes, but prevalence will always be higher when the denominator is lower (i.e, among drinkers) (Table 1 - lines: Alcohol, HED *** (population) past 30 days and Alcohol, HED *** (drinkers only) past 30 days).

Finally, researchers must also consider who did not answer the survey (non-response) and how missing data was handled. Dealing with non-response/handling missing data in probability sampling can be done in several ways from the design to the statistical analysis. This is a highly specialized area and, although its discussion is beyond the scope of this paper, it is good practice to provide information on the methods used so that other scientists can scrutinize and reproduce them (or not).

Where - place

Surveys can have different coverages according to the study’s objective. GISAH indicators cover the entire country, though surveys which provide the data sometimes are not. The only Brazilian survey without a nationwide coverage is the II LNUDPB (which included only municipalities with ≥ 200,000 inhabitants).

An additional result obtained by probability samples and related to “Who” and “Where” is the “Total Estimated Population”. This value represents the approximate number of people presenting the characteristic studied. Though estimating 95%CI for it is also possible, most international reports do not - likely due to lack of space. The II LNUDPB, estimated that 74.6% of Brazilians, almost 38 million people, have drunk during their lifetime. The 3rd LNUD, on the other hand, estimated that 66.4% of Brazilians have drunk during their lifetime, but this smaller percentage represented around 101 million people. Before these discrepancies, one must look for additional methodological detail that can explain the differences in coverage, age group, denominators, and definitions. In this case, the difference was likely related to coverage.

When

The timeframe of data collection is the easiest aspect to evaluate. If all the above items are equal, successive surveys can be used to estimate trends, as long as statistical analysis is conducted. The simple comparison of two punctual prevalences (without appropriate statistical analysis) is not a scientifically valid time trend. Though estimating trends is possible even if some of the W’s differ, this requires additional analytical steps.

Why

The last “W”, which refers to causality, cannot be answered in population-based surveys. Because of this design limitation, it should be addressed by longitudinal or experimental designs ⁵¹51. Grimes DA, Schulz KF. Epidemiology 1: an overview of clinical research: the lay of the land. Lancet 2002; 359:57-61.. Most of the scientific papers profiting from population-based surveys aim to investigate associated factors ⁵²52. Bertoni N, Szklo A, De Boni RB, Coutinho C, Vasconcellos M, Silva PN, et al. Electronic cigarettes and narghile users in Brazil: do they differ from cigarettes smokers? Addict Behav 2019; 98:106007.^,5353. Krawczyk N, Mota JC, Coutinho C, Bertoni N, Vasconcellos MTL, Silva PLN, et al. Polysubstance use in a Brazilian national sample: correlates of co-use of alcohol and prescription drugs. Subst Abus 2022; 43:520-6., which requires performing a multiple regression model. Such models, however, must consider design effect, sample weights, and calibration.

Limitations

The “Five W’s” can be considered as a tool for describing and interpreting survey results despite not covering all the necessary knowledge to design and analyze a study. This manuscript does not aim to discuss specific survey results and clinical, epidemiological, sampling, and analytical methods. These are analyzed in the classical literature such as Edwards’ Treatment of Drinking Problems: A Guide for Helping Professions⁵⁴54. Humprehys K, Lingford-Hughes A. Edwards' treatment of drinking problems: a guide for helping professions. Cambridge: Cambrigde University Press; 2016., Modern Epidemiology⁵⁵55. Lash T, Vanderweele T, Haneuse S, Rothman K. Modern epidemiology. Alphen aan den Rijn: Wolters Kluwer; 2021., Survey Methodology⁵⁶56. Groves RM, Fowler Jr. FJ, Couper MP, Lepkowski JM, Singer E, Tourangeau R. Survey methodology. Hoboken: John Wiley & Sons; 2009., and Complex Surveys: A Guide to Analysis Using R⁵⁷57. Lumley T. Complex surveys: a guide to analysis using R. Hoboken: John Wiley & Sons; 2010., among others. Nevertheless, even if other approaches are used, assessing results in a systematic manner is essential to all the actors who wish to contribute to the field.

Furthermore, all aspects of the methods (questionnaire, sampling, data collection procedures, data systems, data entry, all the statistical analyses, and so on) can and should be improved. All survey procedures/methods must therefore be well documented, saved, and made available to those interested. In this regard, the “Five W’s” are insufficient to achieve successful results. Ideally, scientific studies should investigate the impact of every potential change. As an example, in 2014, the SAMHSA published a 230 pages-long report compiling the abstracts of scientific literature to evaluate the methodological changes in the NHSUD from 1971 to 2014 ⁵⁸58. Center for Behavioral Health Statistics and Quality. National Survey On Drug Use And Health: summary of methodological studies, 1971-2014. Rockville: Substance Abuse and Mental Health Services Administration; 2014..

Conclusions

This manuscript focused on the “Five W’s of Epidemiology” to present some of the fundamental concepts and key methodological aspects regarding alcohol-related indicators obtained in population-based surveys. The “Five W’s” are an easy framework to remember and both scientists and stakeholders can benefit from them. The hardest “W” to choose and evaluate is “What” since harmful alcohol use is a subjective and changeable measure based on different systems of classification. Nevertheless, all the other “W’s” and methodological elements of probability surveys must be clear. None of us should take results for granted, rather we should be able to evaluate them in a scientific, skeptical, and dispassionate way.

It is quite disappointing that none of the Brazilian reports presented all the GISAH indicators, and further analyzes would be necessary (provided the required data was collected). If planning did not foresee those indicators, comparisons with international data may be jeopardized. Some of the Brazilian surveys show significant differences between each other (almost 10 times higher prevalence of binge drinking and alcohol dependence, for example) compared to global GISAH indicators. Rather than dismissing or disqualifying these results, one could take a “detective approach” and look for possible methodological reasons which can explain these discrepancies. Well-documented methods will thus provide hints for the curious readers and the necessary information for skilled, independent scientists who can evaluate them and reproduce the survey.

Finally, there has been advance in the global agenda to prevent alcohol-related harms. For instance, to tackle the harmful consumption of alcohol became a target for the Sustainable Development Goals ⁵⁹59. United Nations. Goal 3: ensure healthy lives and promote well-being for all at all ages. https://www.un.org/sustainabledevelopment/health/ (accessed on 15/Aug/2021).
https://www.un.org/sustainabledevelopmen... , and the WHO launched the SAFER initiative ⁷7. Monteiro M. The World Health Organization's SAFER initiative and the challenges in Brazil to reducing harmful consumption of alcoholic beverages. Epidemiol Serv Saúde 2020; 29:e202000. (providing tools for the countries to implement cost-effective policies against alcohol). Population-based surveys are a crucial component for monitoring these strategies emphasizing the importance of a systematic collection and assessment of information on alcohol.

Acknowledgments

I would like to thank Dr. Marcelo Ribeiro-Alves and Dr. Flavio Kapczinski for reading an earlier version of this manuscript. I also acknowledge the long-term financial support received from Brazilian National Research Council (CNPq; #312543/2020-4) and Rio de Janeiro State Research Foundation (FAPERJ; E-26/203.154/2017).

References

Publication Dates

History