Smoking-attributable mortality (SAM) is a serious public health concern in the 21st century which now outnumbers that from human immunodeficiency virus (HIV), tuberculosis, and malaria combined [1
]. According to the 2013 World Health Organization (WHO) report on the global tobacco epidemic, tobacco use accounts for millions of premature deaths and billions of dollars of economic damage every year [2
]. Apart from the health issues, tobacco use has significant bearings on socioeconomic status, quality of life and general well-being. Some of the common impacts of tobacco use include direct expenses to purchase tobacco products, medical expenditure for treating tobacco-related illnesses and lower workplace productivity, which can contribute substantially to household poverty, poor nutrition and low education, particularly in low- and middle-income countries (LMICs) [3
]. Many LMICs are undergoing early stages of the tobacco epidemic with a lower burden of tobacco use and SAM compared with the high-income countries (HICs): 18% in the HICs vs. 11% in middle-income countries and 4% in LMICs [4
]. However, the prevalence of smoking in the LMICs is increasing at a fast pace which is projected to account for a higher proportion of SAM during the coming decades. Between 1990 and 2009, the prevalence of tobacco use in the Western Europe declined by a quarter (26%), while that in Africa and some Middle Eastern countries rose by more than half during the same period (57%) [5
The slow decline of tobacco use in developed countries has resulted in a focus-shift of the tobacco industry from the West to emerging economies in Africa and Asia [6
]. During 2003–2012, the total area harvested for tobacco in Africa increased by two-thirds, while in the USA and Europe that decreased by 18% and 40.4%, respectively [7
]. In the context of African countries, this is particularly worrisome as the underdeveloped healthcare systems struggle to cope with the rising prevalence of non-communicable chronic diseases (NCDs) [6
]. In the wake of this situation, there has been growing concern regarding tobacco control in African countries in recent years as more and more countries are willing to implement the WHO Framework Convention on Tobacco Control (FCTC) [8
]. Kenya, as the largest economy in eastern Africa, has shown concrete commitments for tobacco control policies and has become a signatory of the FCTC framework since 2004. Ethiopia, which is projected to surpass Kenya to become the largest economy in the region, has ratified the framework convention in 2014.
As two fast-growing economies, both Ethiopia and Kenya have been experiencing an expanding middle class with ensuing alternations in demographic, dietary and epidemiological structures marked by the rising prevalence of lifestyle diseases including tobacco-induced ones [10
]. Hence, reducing the use of tobacco at the population level remains a key imperative from both healthcare and national development aspects. A major requirement for developing and achieving tobacco control targets is the availability of workable information on the magnitude of the burden and underlying causes. However, research evidence on the nationwide prevalence of tobacco use is scarce in Ethiopia and Kenya. Previous research on smoking in these two countries was either based on non-representative samples at sub-national level, or failed to show the trends in prevalence rates [12
]. As the case with most low-income settings, resources for comprehensive surveys are limited in Ethiopia and Kenya. For this purpose, in the present study we used datasets from the Demographic and Health Survey (DHS) which provide high-quality data on various indicators of population health and risk factors such as tobacco smoking and alcohol drinking. Based on these datasets, the aim of this study was to estimate the trends in prevalence of smoking among adult men aged 15–59 years. In addition, we also investigated the association of smoking with several sociodemographic and economic variables.
Effective policy-making for controlling tobacco use requires routine monitoring of prevalence rates. With that purpose in mind, in the present study we have analyzed the Demographic and Health Survey datasets conducted in last 15 years in Ethiopia and Kenya and have shown the trends and current prevalence of smoking among adult men. Findings indicated that in Ethiopia the prevalence of smoking has increased since the 2005 survey, while in Kenya the rate has been decreasing at a slow pace since the 2003 survey. The declining trend in Kenya might be due to the implementation of the Framework Convention on Tobacco Control of the WHO (which has been in effect since 2005) to which the country was one of the earliest signatories [17
]. Ethiopia, on the other hand, has joined the treaty only recently (2014); however the overall prevalence in Ethiopia was still lower than that of Kenya (17% in 2014 in Kenya vs. 11.7% in 2011 in Ethiopia), and of the African Region average of 15.8% (2010 estimate) [18
]. Data were not available for subjects under 15 year of age, however the Global Youth Tobacco Survey of 2007 reported an equally high prevalence of tobacco consumption (18.5%) among youth aged 13 to 15 years. Analysis of the data on 23 African countries from the Global Progress Report on implementation of the treaty reveals that the rate of implementation varied substantially, ranging from 9% in Sierra Leone to 78% in Kenya [19
]. Thus, the high rate of smoking in Kenya comes as a surprise given the highest rate of implementation of the framework.
Findings also indicated the existence of regional disparities in the prevalence of smoking as urban men were significantly more likely to report smoking, except for the 2005 DHS survey in Ethiopia. Although the population in Eastern African region is predominantly rural, both countries are urbanizing at a fast pace led by the recent boom in employment in urban areas, which is usually associated with higher adoption of smoking behavior [20
]. As urban and rural people share varying degrees of exposure to environmental and lifestyle-related risk factors, understanding the geographical variation in smoking patterns is essential for regional priority setting and policy actions. The findings of this study support the fact that urban men should be regarded as a priority concern for smoking interventions, especially the age group of 25 to 34 years.
Religious affiliation also appeared to be significantly associated with smoking behavior among both urban and rural men in Kenya. In the context of the present study, it is hard to discern the causality of this relationship; however, previous studies have reported that religious commitment is an important predictor of lifestyle and health-related behavior [21
]. Given the considerably high percentage of religious attachment in this region, this can be regarded as an opportunity for the promotion of healthier lifestyles through various religious teachings and relevant norm-setting that discourages smoking and other unhealthy behavior which may lead to greater health outcomes and social well-being [23
With regard to education, the association was significant in the urban areas only and was relatively stronger for Kenya than in Ethiopia. Surprisingly, compared to men who had no formal education, the likelihood of being a smoker was higher among those with a higher educational profile. While the finding appears to be counterintuitive, this is in line some previous studies [24
] and inconsistent with others [25
]. In general, higher educational attainment is associated with better self-efficacy and adherence to healthy behavior. However, the relationship may not be monotonic and subject to variation depending on sociocultural and environmental factors such as the level of exposure to smoking advertisements, stages of urbanization, and degrees to which anti-smoking policies are applied. As countries tend to urbanize, an increasing proportion of the labor force shifts from agriculture/manual to service/managerial jobs indicating a higher socioeconomic status and disposable income, which can serve as an enabling factor for healthier lifestyles. Findings from Kenya also support these facts as men employed in service and managerial jobs in rural areas and with higher wealth status in the urban areas had significantly lower odds of smoking. However, interpretation of these findings should be considered in the context of the relevant factors (e.g., aggressive marketing, social smoking, occupational prestige) deterring abstinence from or cessation of smoking.
Previous studies from developed countries have shown that higher income and educational status can influence increased consumption of alcohol [27
], which can trigger higher consumption of tobacco [28
]. Our findings suggest that alcohol consumption was associated with significantly higher odds of smoking among both urban and rural men in Kenya. This finding suggests that anti-smoking policies and campaigns should also focus on addressing alcohol use/abuse to effectively reduce the prevalence of smoking.
Although our findings do not indicate any causal relationship between smoking and demographic and socioeconomic variables, some interesting contrasts emerged from the analysis. As the associations tended to be more significant for Kenya than in Ethiopia, this might indicate dissimilarities in the underlying determinants of smoking in these countries. Further comparative studies will be required by including more context-specific variables associated with the adoption of smoking, success in cessation, and effectiveness of the current tobacco-control programs. Clearly, both of the counties are far from achieving their full potential in terms of anti-smoking policy development and implementation of the Framework Convention on Tobacco Control program. One important policy implication of the findings of the present study is that tobacco-control programs need to take into account the age and regional differentials along with the socioeconomic factors relevant to the standard of living among smokers (e.g., education, economic status). Long-term success in reducing the burden of smoking will require appropriate social policy instrumentation to address social factors, as well as political commitment against tobacco production and marketing that can undermine the efficacy of the programmatic efforts.
As far as we are concerned, this is the first study to estimate the trend in the prevalence of smoking among adult men in eastern Africa. Sample size was relatively large and representative of men aged between 15 and 59 years. The findings also provide important insights for policy action and for more comprehensive studies in both countries. However, besides the important contributions our study makes, it has several limitations that need attention when interpreting the results. Firstly, as the data were secondary, this limited the choice and measurement of the variables. As such, some key variables relevant to tobacco-smoking behaviour were not possible to include in the study. The prevalence of smoking and other variables was self-reported, which is subject to reporting bias. The results represent the prevalence of smoking only, and not the overall prevalence, as other forms of tobacco use were not considered. Also, the associations derived from the cross-sectional analysis cannot confirm any causality.