Prevalence, distribution, and social determinants of tobacco use in 30 sub-Saharan African countries

Tobacco use has been long known to be a major cause of premature mortality [1] and has been attributed to cause 9% of all deaths worldwide [2]. Each year, an estimated 5.7 million deaths, 6.9% of years of life lost, and 5.5% of disability adjusted life years are caused by tobacco-related diseases [3]. The prevailing pattern of the tobacco epidemic could cause one billion deaths during the 21^st century and 80% of them could occur in low- and middle-income countries (LMICs) [4]. Though recent global estimates have shown a decreasing trend of smoking among both men and women, in 2012 there remained an estimated 967 million smokers living in 187 countries, with the number being expected to increase as the population grows [5].

The Framework Convention on Tobacco Control (FCTC) adopted in 2003 has been ratified by 177 countries worldwide [6]. Under the FCTC, monitoring of tobacco use worldwide by population-based surveys has been prioritized to understand disease patterns, assess the impact of tobacco control measures, and to assist tobacco control policy changes [7]. Major steps in this direction are the Global Tobacco Surveillance system [8], World Health Organization’s STEPS program [9], World Health Surveys (WHS) [10], and the International Tobacco Control (ITC) policy evaluation project [11] carried out in a number of countries spanning all continents. Nevertheless, data from these surveys do not comprehensively reflect tobacco use estimates, patterns, and types of tobacco products consumed in sub-Saharan Africa (SSA). For example, smokeless tobacco (SLT) products commonly consumed in South and South-East Asia [12] have health effects that are different from those of smoking [13],[14] and are usually not emphasized much in tobacco control policies. This assumes great importance since the type of tobacco products consumed not only varies across countries [15] and regions [5], but also by age, gender, education, and economic status [16]-[19].

Among the multi-country surveys, tobacco use data from 14 SSA countries is available from the WHS [10] and Nigeria and Uganda in GATS [20], but none from ITC projects [11]. Further, the WHS and ITC projects focus on cigarette smoking only [11],[17]. Information about tobacco use gathered from Demographic and Health Surveys (DHSs) performed on nationally representative samples of men and women can provide national-level estimates for each country and study the social distribution of tobacco use and type of tobacco products consumed in a particular region [15]. Pampel has provided estimates of cigarette smoking and its social determinants from DHSs performed in 14 SSA countries during the year 2006 or prior [21]. However, Pampel’s study does not provide estimates of SLT use and distribution of tobacco use by economic (wealth) status [21]. A systematic review of studies on adult tobacco use prior to the year 2005 in 14 SSA countries showed that the information varied due to the heterogeneity of the included studies [22]. The scale and pattern of the tobacco epidemic currently prevailing in the SSA countries is not clearly known except for some reports based on DHSs from Ghana [23] and Madagascar [24] and a national survey from Mozambique [25]. Inclusion of tobacco use questions in 30 countries’ DHSs during recent years provides a clear picture of the tobacco epidemic in the SSA region [26]. We aim to provide country-level prevalence estimates for smoking and SLT use and assess their social distribution (determinants) in 30 SSA countries.

Our analyses of DHS data provided national-level estimates for tobacco use in 30 out of 47 SSA countries by sex and type of tobacco consumed. The data obtained highlights the scale of the tobacco epidemic and describes the pattern of smoking and SLT use according to social groups. Among men, the prevalence of smoking (mainly cigarettes) was very high relative to SLT use in all countries except Madagascar and Mozambique. Among all the SSA countries, smoking prevalence among men was high in Sierra Leone, Lesotho, and Madagascar, where nearly a third of adult men were current smokers. In most countries, the prevalence of both smoking and SLT use among women was very low compared to men and the highest prevalence of smoking and SLT use was found in Burundi and Madagascar, respectively. Compared to men, women were using more diverse tobacco products such as cigarettes, pipe, snuff, chewing tobacco, and other types.

Since most SAA countries are poor and have lower literacy rates, it is commonly thought that the prevalence of tobacco use is lower, resulting in a low priority for tobacco control. However, following the economic growth currently experienced in many SSA countries, it is estimated that smoking prevalence will increase [31]. The lower prevalence estimates for most SSA countries presented herein were comparable to those in Pampel’s study from 14 SSA countries [21]. A systematic review has also concluded that adult tobacco use prevalence in many SSA countries is lower than in developed and other developing countries [22]. The very small differences in prevalence estimates between our study and those of Pampel’s indicate that the prevalence has changed minimally or else the differences may have been solely due to sampling errors. Nevertheless, another DHS-based study reported that prevalence of smoking among Ghanaian men had decreased by 1.7% between 2003 and 2008 [23]. Further, the current smoking and SLT use prevalence in most of the SSA countries, except Madagascar, Sierra Leone, and Lesotho, was much lower than in South and South-East Asian countries [15]. However, the current smoking prevalence estimates presented herein cannot be compared with those in Nigeria and Uganda assessed by GATS [20] or with the WHS [17], which included 14 SSA countries, since these surveys defined current smoking as smoking any form of tobacco either daily or occasionally [17],[19]. Moreover, our estimates are also different from those by Ng et al. [5], since the authors adopted a different definition of daily smoking and used comprehensive data sources and robust statistical analyses.

Overall, prevalence rates of smoking and SLT use among both men and women in 30 SSA countries were much lower than in South and South-East Asian countries [15] and other regions of the world [5],[32]. A higher prevalence of SLT use among men was found in some SSA countries only, for example, chewing tobacco in Madagascar [24] and Mozambique [25], and snuff inhalation in Rwanda and Senegal. Prevalence of SLT use among women was very low in most SSA countries except in Madagascar (mainly chewing tobacco) [24] and Lesotho (mainly inhaling snuff), unlike the pattern in South Asian countries where both men and women used more diverse types of SLT products [15]. A higher prevalence of SLT use among men and women was reported in India, Pakistan, Nepal, and Bangladesh, but not in other South-East Asian countries [15]. Thus, the high prevalence of SLT use among men and women in Madagascar may be explained by a high proportion of Madagascans having South Asian descent where SLT use is very high [23]. Sierra Leone had the highest prevalence of smoking among men, consistent with a previous study [23]. However, of all SSA countries, Burundi had the highest prevalence (9.8%) of smoking among women. The lower prevalence of smoking and SLT use among women in most SSA countries is in accordance with an earlier study [19]. A positive gradient by age among both men and women for smoking has been previously reported [15],[19],[21],[22]. This pattern may be explained by cohort effects, i.e., smoking was less likely to be initiated in more recent decades, or by age effects, i.e., respondents continued to initiate as they grew older. However, we could not assess whether the positive gradient by age was caused by cohort or age effects since we analyzed single cross-sectional survey data from each country. Previous studies have reported the existence of wealth-related inequalities in smoking [17] and the social determinants of tobacco use [19] in LMICs. These studies indicated that, in most LMICs, the poorest men and women were more likely to smoke than the richest, which is similar to our results. In SSA countries, SLT use was strongly associated with wealth, i.e., poorer men and women were more likely to use SLT, similar to the findings from South and South-East Asian countries [15]. It is thought that poorer people may consume tobacco to suppress their hunger [33] since many smokers believe that smoking has an appetite-suppressing effect; many tobacco companies have exploited this by introducing appetite suppressant additives to the cigarettes [34]. Compared to men and women with a higher education, uneducated men and women were more likely to smoke and use SLT, consistent with the results of previous studies from Africa [21] and other regions [19]. Less educated (illiterate) people may be more vulnerable to tobacco use as they lack knowledge about their adverse health effects [35] or else this pattern may be due to parental influence, peer pressure [36], and cultural acceptance [37].

Cigarette smoking has been reported to be higher among urban residents [21],[38]; however, in our study, rural men and women in SSA countries were more likely to smoke. Our findings suggest that perhaps smokers who are usually poor and uneducated may be living in rural areas. In our study, Catholic, traditional religions, or no religious affiliation were associated with smoking and SLT use, which may likely be a residual confounder. Although none of the religions promote smoking or other unhealthy behaviors, there is no conclusive evidence for this negative relationship between religion and tobacco use [39]. As compared to those who were never in union, single men (separated, divorced, or widowed) were more likely to use tobacco; this was also the case for married women. These findings are not consistent with our previous studies or with others from Africa [15],[21],[23],[40]. Our analyses show that agriculturists and unskilled or manual workers had a higher risk of smoking or SLT use, which is in accordance to a previous study in 14 SSA countries [21] and another in the United States of America [41]. Nevertheless, we agree that occupation, which reflects an individual’s social standing, is also related to an individual’s education and income [42] and therefore social standing affects health and health behaviors [43].

In general, the prevalence of smoking and SLT use was very low among women, in agreement with previous reports from SSA countries [21],[22]. However, the higher prevalence of smoking among women in Burundi, Sierra Leone, and Namibia, and SLT use in Madagascar and Lesotho warrants gender-specific tobacco control interventions in these countries. Our analysis identified that poor, uneducated or less educated (up to primary school), agriculturists, and manual or unskilled workers as the most vulnerable groups. Research has shown that in lower socio-economic groups, who already have a scarce income, smoking may divert their resources from education, health care, housing, and quality food to purchase cigarettes [44]. Even though tobacco use is generally lower in SSA countries, they have a higher burden of nutrition and communicable disease [45] and may face an additional burden from non-communicable diseases [46] if timely action is not taken to curb this early-stage tobacco epidemic. To reduce the burden of tobacco-related ill health, interventions should be directed at eliminating its root causes such as illiteracy and poverty itself [17]. There is a need for the relevant authorities to act by addressing the disparities in tobacco use, failing which inequalities in health may widen further [16]. As of July 2014, 42 of the 47 SSA countries had ratified the FCTC, whose provisions include a ban on tobacco advertising, promotions, and partnerships, warning labels on tobacco product packages, measures to prevent exposure to second hand smoke, and increased taxation [7]. It is necessary that all SSA countries ratify the FCTC, especially Malawi and Mozambique, which have a higher smoking prevalence.

Analyses of DHSs provided a regional overview of the tobacco epidemic and such data may be utilized for monitoring the tobacco epidemic at country-level and assess prevalence by population subgroups. If DHSs in more countries collect tobacco use data, an updated analysis can provide a complete scenario of tobacco use in the SSA region. In a vast continent such as Africa, prevalence of tobacco use is lower than in other regions, but varies much across the 30 countries included herein. For example, smoking among men was 5.4% in Sao Tome & Principe but 37.7% in Sierra Leone. What can explain the between-country differences in prevalence of smoking and SLT use? Such inter-country variations may be explained by country characteristics such as the economy, i.e., gross domestic product, cultural factors, access of tobacco companies to sales, tobacco control policies, and pricing of tobacco products. Nevertheless, we did not study these factors as it was beyond the scope of this paper. Further research using country-level aggregate data about these factors and multi-level modeling may provide a better understanding about the reasons for inter-country variations in tobacco use.

Prevalence estimates retrieved from DHS data have some limitations due to the survey design and questions asked to assess tobacco use. DHSs have limited the age of men and women respondents from 15–64 and 15–49 years, respectively. Therefore, true population prevalence rates may be underestimated if the prevalence rates among older men (>64 years) and women (>49 years) were higher. The association between social factors and tobacco use lacks a temporal relationship due to the cross-sectional design of the DHSs. We could only estimate current smoking and current SLT use since limited information was collected about tobacco use. Tobacco use based on self-reports may have been underreported due to stigma, especially among the young and women, leading to misclassification bias and underestimation of prevalence rates. However, there was no means to verify self-reported tobacco use by estimating biomarkers such as urinary cotinine levels.

The prevalence of smoking among women was much lower than among men, but showed similar social patterns. Tobacco control strategies should target the poor, not (least) educated, and agricultural and unskilled workers, who are the most vulnerable social groups in the SSA region. DHSs can provide reliable estimates for surveillance of tobacco use at country-level and by social groups. As most SSA countries are at the early stages of the tobacco epidemic, tobacco control efforts in Africa should focus on health promotion to stop the initiation of tobacco use in addition to cessation.