In this study we reveal complex regional patterns of tobacco and alcohol consumption in six communities across four SSA countries. First, in the total cohort, almost half (49%) of all participants, men and women, either consumed tobacco products, alcohol, or both. Secondly, men were more likely to be current smokers than women (34.5% vs 2.1%), in line with global studies, but more women (14.4%) used smokeless tobacco (snuff and chewing tobacco) than men (5.3%), with some regional variation [2, 3, 5, 6]. Third, overall, being a current smoker was associated with alcohol consumption, lower education levels, and was less common among married individuals or those living with partners. Fourth, alcohol consumption was also more common among men, but had higher prevalence than smoking in both sexes (60.8% of men and 29.3% of women). And lastly, problematic alcohol use was associated with being male, widowed or divorced, and using tobacco products (smoking and smokeless use).
Prevalence estimates of current smoking among men from almost all the AWI-Gen sites were much higher when compared to the age-adjusted prevalence for daily smoking among men in their respective countries in a study conducted in 2015 . A reason for this difference might be that the previous study to which we compare our findings was based on the prevalence of daily smoking, whereas our smoking variable was ‘current smoking’ with some men reporting that they were occasional current smokers. However, the direct comparison of prevalence of daily smokers is still higher in AWI-Gen and these differences could be due to several other factors including age distribution, and selection of specific communities in AWI-Gen, whereas Reitsma et al. (2017) may have used publications with small participant numbers, or specific co-morbidities from different communities to develop country-specific prevalence estimates. Patterns of tobacco use were highly sex-specific, with smoking tobacco most common among men, and women preferring snuff or chewing tobacco supporting previous findings in SA, Ethiopia, the Gambia, Kenya and Angola [3, 9, 10, 13, 15, 21, 25, 39]. In the combined sample, more than half of the men ever smoked, and 34.5% were current smokers. A similarly high prevalence of current smoking was previously reported in men from other African countries [3, 6, 9, 10, 13, 15, 21, 25, 39]. In our study, age was associated with current smoking in men (aOR = 0.98 (95%CI 0.97–0.99)) and smokeless tobacco consumption across both sexes (aOR = 1.06 (1.05–1.08)), and while statistically significant in the full dataset, the effects were small (per year of age) and in opposite directions. Adults in the highest wealth categories and those who had attained higher education levels were less likely to use tobacco products, in concordance with the findings of previous studies in Africa [3, 6, 9, 10, 13, 15, 21, 25, 39]. Alcohol consumption was strongly associated with tobacco use. Furthermore, former smokers appeared to be more likely to consume some form of smokeless tobacco, which suggests that there may be a substitution process at play. Smoking cessation was significantly associated with the highest SES quintiles, being married/living with a partner, and ceasing to consume alcohol. Within the AWI-Gen study, smoking was less common among adults with tertiary education and if they did smoke, they were more likely to stop smoking than those with lower levels of education.
Overall current alcohol consumption (both problematic and non-problematic) was present in 40.5% of the AWI-Gen cohort. The lower prevalence of alcohol consumption among women, as well as the decreased likelihood to engage in harmful alcohol consumption has been attributed to socio-cultural stigmas surrounding women who consume alcohol. This finding is supported by several studies in different African contexts [7,8,9, 17, 18, 20, 24]. Lifetime alcohol abstainers were more prevalent in East and South (Kenya and South Africa) Africa, compared to West Africa (Burkina Faso and Ghana). This difference was found to be more defined among women, with women in Nairobi (Kenya), Agincourt, Dikgale, and Soweto (South Africa) more likely to abstain from alcohol than participants in Navrongo (Ghana) and Nanoro (Burkina Faso). The frequency of daily alcohol consumption was highest among men in Nanoro (27.2%) and 10.3% of women were daily consumers. This trend of high daily alcohol consumption in both sexes was also observed during the national WHO-STEPS survey in 2013 in Nanoro where 26.2% of men and 16.7% of women were identified as daily alcohol consumers .
Differences in alcohol use were observed between the West African sites, Nanoro and Navrongo, despite having similar rates of alcohol consumption. Navrongo reported a much higher prevalence of problematic alcohol consumption (31.2%) than Nanoro (12.5%). In Navrongo, spirits were also more popular than in Nanoro. Binge drinking was found to be highest among men in Dikgale and Agincourt, corroborating previous data suggesting that SA has one of the highest rates of alcohol consumption per capita in the world [7, 17, 18, 40]. Binge drinking was the main feature of problematic alcohol consumption together with the feeling of needing to “cut down”. It also appears that social pressure through criticism (“people criticising you”) was most prevalent in Navrongo, followed by Dikgale. In Agincourt however, fewer surveyed drinkers reported facing criticism and feeling guilty related to their drinking patterns. In this study of adults aged 40 to 60 years, age was significantly associated with problematic alcohol consumption, where older people in this age range were less likely to drink, but the effect was small. Age effects related to drinking behaviour were observed in other African studies [12, 13, 16, 41]. Across the combined dataset, problematic drinking was not significantly correlated with SES, but SES associations were complex and varied at site level. A recent cross-sectional study examining correlates of alcohol use in the slums in Kenya suggested that alcohol use was associated with higher income, whereas the reverse relationship was found for problematic drinking in the Nairobi sample of this study (with lower odds of problematic drinking in association with higher SES) . In our study, alcohol consumption was highly correlated with tobacco use, likely reflecting addictive behavior, rather than a preference for one substance above the other. This correlation has been observed in many cross-sectional studies that were also not specifically designed to reveal the likely sequence of substance use behaviours [9, 10, 13, 14, 16, 21, 39]. Current (problematic and non-problematic) and former drinkers were more likely to use tobacco products than those who had never consumed alcohol.
To our knowledge, this is one of the larger cross-sectional African population studies of multi-site comparisons of the prevalence of tobacco and alcohol consumption with related socio-demographic correlates. Patterns of tobacco and alcohol consumption show sex specific, regional (East, West, South Africa), as well as within-region differences. These differences may be due to differences in socio-economic transition across regions, but may also be affected by different national policies related to the regulation and taxation of tobacco and alcohol products, in turn affecting accessibility to those substances [2, 42, 43]. In addition, there are differences across religions and cultural beliefs which are known to influence patterns of tobacco and/or alcohol consumption [9, 10, 15, 17, 19, 24, 36].
This cross-sectional study was not designed to infer causality and there is no data to assess the temporal sequence of substance use. Our study was limited to persons aged 40 to 60 years, which is not representative of the general population, and data generated from a single community is not necessarily generalisable to be representative of the geographic region or an entire country. The missing data from Soweto limited inference for this site. Under-reporting of substance use may have occurred due to cultural differences within regions, as women would be less likely to report substance use due to potential stigmatisation, and because the CAGE questionnaire asks sensitive questions which participants may have found difficult to answer honestly and objectively.