Association of adult tobacco use with household food access insecurity: results from Nepal demographic and health survey, 2011

Food insecurity defined as “‘limited or uncertain availability of nutritionally adequate and safe foods or limited or uncertain ability to acquire acceptable foods in socially acceptable ways” [1] is a well-known determinant of human development [2]. Millions of people in low-and-middle-income countries (LMIC) suffer from extreme hunger and malnutrition [3]; therefore, Sustainable Development Goals (SDGs) have underscored access to food as a basic human right needing urgent measures to end extreme hunger and food insecurity [4]. Food insecurity is linked with poverty [5] and it results in poorer quantity and quality of diet consumed [6]. Nepal is a low-income country where about 40% of its population lives on less than a dollar ($) per day [7]. Household food insecurity(HFI) is widely prevalent in Nepal where 45 of the 75 districts were considered to be food deficient [8] and two-thirds of the households in Kailali district had some degree of food insecurity [9].

Globally, tobacco smoking accounted for about 6.1 million deaths and 143.5 million DALYs in 2013 [10]. Overall smoking rates among men and women are decreasing in high income countries; yet an estimated 967 million smokers live worldwide mainly in LMICs [11]. Both smoking and smokeless tobacco (SLT) use are prevalent in South Asia [12, 13]. Tobacco use not only affects individuals’ health [10, 14] but also affects the household nutrition and budgets especially in low-income houses [15‐17]. Families with at least one tobacco user spends up to 20% of the income on buying tobacco products [17, 18] diverting the household income from essential needs such as food, health, education etc. A few studies have demonstrated that adult smoking is associated with food insecurity among the adults and children in the United States of America [19] and with the household food insecurity in Indonesia [20]. Tobacco smoking as well as SLT use are prevalent in Nepal [12]. Since food insecurity, poverty, and tobacco use are all prevalent in Nepal, we further tested, if there was an association between adult tobacco use and household food access insecurity using nationally representative data from Nepal Demographic and Health Survey (NDHS), 2011.

The households included for analyses were mainly from rural areas (86.4%); central (24.3%) and eastern (23.7%) regions. Men’s mean age was 29 years (Standard Deviation [SD] 10.2), and women’s mean age was 27.1 years (6.1). Women were mostly not educated (49.8%), whereas men were mostly (52%) educated up to secondary level only, and were either agriculturists (42.9%) or manual workers (31.6%) (Table 1). Among 4121 households selected for men’s survey, HFIAS score was skewed in distribution; mean score was 3.5 (SD, 4.6); median was 0 (interquartile range, 0-6) (Table 2) as HFIAS score in 51% of the households was zero. Three common conditions of HFI reported as occurring ‘often’ during 12 months prior to the survey date were ‘worried about not having enough food’ (18.2%), ‘not able to eat preferred foods because of lack of resources’, (12.8%) and ‘ate limited variety due to lack of resources’ (9.9%) (Table 2). Among men, weighted prevalence rates (%) of smoking, SLT use and any tobacco use were 30.2 (95% CIs 28.1, 32.3), 35.1 (95% CIs 32.7, 37.9), and 50.2 (95% CIs 47.9, 52.6), respectively while the corresponding rates for women were 13.6 (95% CIs 12.7, 14.9), 5.4 (95% CIs 4.5, 6.3), and 17.3 (95% CIs 15.7, 18.9) (data not shown).

By univariate comparisons of HFIAS scores with dependent variables, HFIAS scores were significantly higher among the households in which men smoked and as well as used SLT(4.96), and men either smoked or used SLT (3.82) as compared to households in which men do not use any form of tobacco (2.79). HFIAS score was also higher among the households where men were agriculturists (4.36) and manual workers (3.62) than those households where men were professionals or in service or business (1.92). HFIAS score showed a negative gradient across men’s educational attainment i.e. highest (6.44) among the households where men were not educated to lowest (1.13) among those with higher education and household wealth index (7.16 in poorest households versus 0.94 in richest households). HFIAS score was also significantly higher among the households in urban areas, Mid-Western and mountainous regions of Nepal. HFIAS scores significantly correlated with number of household members (Table 3). By zero-inflated negative binomial regression analyses, men’s tobacco use was associated with HFIAS score in all three models; however, the effect size slightly decreased in final model when spatial variables were included in the model (adjusted OR 1.18 to 1.11). Women’s tobacco use was not associated with HFIAS score. Development and ecological regions, household wealth index were rather strongly associated with HFIAS score (adjusted OR ratios ranging from 0.57 to 0.87, p ≤ 0.001) (Table 4). HFIAS score was not associated with occupation after the inclusion of household and spatial variables into the regression models. However, HFIAS score was not associated with men’s and women’s age, tobacco use by women, sex of household head, and total household members in all models (Table 4).

About half the men surveyed were using some type of tobacco product [12] and half the households did not report any food insecurity while the remaining had varying degrees of HFI confirming its existence in Nepal [8, 9]. Our study based on a nationally representative sample has further strengthened association between food insecurity and smoking reported in previous studies [19, 20, 27]. Households in which men used any form of tobacco (smoking and SLT) had highest HFIAS score but association was weaker than association observed with wealth index. Cutler-Triggs et al. studied a large nationally representative sample of 8817 households in the United States of America (USA) reported that living in a house with an adult smoker increased the risk of food insecurity (also its severity) among both children and adults [19]. Another study from the USA reported that smoking was associated with an increased food insecurity in low-income households [27]. Semba et al. studied 26,380 rural households in eight of the 12 provinces in Indonesia found that father’s smoking was associated with increased HFI score [20]. Consistent with previous studies [19, 20], our study also showed that smoking as well as SLT use were associated with higher HFIAS score even after controlling for socio-demographic factors and spatial variables. A study from Indonesia [20] did not asses association between women’s tobacco use and HFI. However, we studied women’s tobacco use which was not associated with HFIAS score. Lack of association may be due to lower tobacco use rates among women, lower quantity of tobacco consumed by women or role of women as mainly housewives in a conservative Nepalese society.

Our findings support the theory that tobacco use diverts household finances away from buying food in developing countries where expenditures on tobacco takes the place of expenditures on food, healthcare etc. particularly in low-income households [15, 17]. Data on household expenditures were not available to explore this association further; nevertheless this seems to be a plausible explanation for association observed between tobacco use and HFI at least in low-income households [15, 17]. Nepalese households spend an estimated 5% of their annual expenditure on tobacco products [28]. However, proportion of all household income spent on tobacco products is more important and this was 22% in poor urban households in Indonesia [29]. Efroymson et al. reported that if limited finances of poor households were not spent on tobacco up to 50% surplus money would be available to buy food or other consumer items [17]. In our study, even after controlling for covariates, wealth index was strongly associated with HFIAS score suggesting that adult tobacco use in the poor(er) households may have deprived them of buying consumer items. Association of tobacco use with socio-economic status is well known in both developing and developed countries [12, 27]. Ironically, in developing countries, tobacco use is significantly higher among economically weaker sections [12]. Kim and Tsoh argue that adults in poorer households may take up smoking to cope with financial stress resulting in a reciprocal effect on food insecurity i.e. poverty increasing smoking which in turn worsens the food insecurity which is a bidirectional association between food insecurity and smoking [30]. It has been hypothesised that smoking may be taken up to overcome psychological distress caused by food insecurity [31] or even to suppress their appetite when food was unavailable [32]. Even after controlling for covariates, HFIAS score was associated with spatial (urban or rural, development regions and ecological zones) factors probably due to residual confounding we could not adjust for in the regression models. Regional disparities in health indicators [22] and food insecurity [33] are well known in Nepal to support our findings.

Our study has following limitations arising from secondary data analyses, design and data collection methods of DHS. Cross-sectional data limits our ability to draw causal inferences of the observed associations in our study. Our results may have unrecognised biases arising from unmeasured explanatory factors not adjusted for in the analyses. Although, effect of poverty on HFI is well known [5], extraneous factors such as ownership of cultivable land, cattle, availability of workforce, rainfall etc. may also affect HFI in rural Nepal, an agrarian country [8]; data on all these factors were not collected in NDHS 2011. Information on daily consumption was not collected for other types of smoking and SLT; hence, we could not include quantity of tobacco consumed into the regression models. Non-validation of modified HFIAS (7-items and 12 months recall period) and extending the period of recall from 30 days to adjust for seasonal variations in food supply in Nepal may have resulted in underestimation of HFI in NDHS 2011. Reporting biases of tobacco use, HFI due to cultural barriers and stigma have been discussed in detail elsewhere [34]. Notwithstanding above limitations, our study further strengthens the reported association of adult tobacco use with HFI, highlighting the previously less recognised non-health consequences of tobacco use on the poor households.

Tobacco use among economically weaker populations may be a potential hindrance to the success of poverty alleviation and food assistance programs. Tobacco users need to be educated not only about health gains but also financial benefits of quitting tobacco. Tobacco control programs should provide free counselling and medications to help quit tobacco to improve their socio-economic status and food security. In a broader context, the governments should promote food availability and strictly implement tobacco control laws.