Cigarette smoking causes five million deaths per year worldwide, and it is estimated that the annual death toll from smoking will climb to ten million deaths by 2030, with seven million deaths in developing countries1, Reference de Beyer and Brigden2. Cigarette smoke damages the lower respiratory tractReference Aubry, Wright and Myers3, increases oxidative stress and increases the risk of bronchitis, chronic obstructive lung disease, cancer and death1. Tobacco companies have gradually shifted their market from high-income to low-income countries, where many people are poorly informed about the health risks of tobacco use and anti-smoking policy is relatively weakReference de Beyer and Brigden2. Although much research has focused on the relationship between smoking and adverse outcomes such as cancer, respiratory and cardiovascular disease, the problem of smoking and its relationship to malnutrition and poverty have not been well characterisedReference de Beyer and Brigden2. Tobacco use may have adverse consequences for nutrition, health and household budgets, especially among families living in poverty in developing countries.
Smoking exacerbates the effects of poverty, as expenditures for tobacco may divert household income from food, clothing, housing, health and educationReference Jenkins, Dai, Ngoc, Kinh, Hoang and Bales4, Reference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5. The amount of money spent on tobacco is especially problematic in low-income countriesReference Jenkins, Dai, Ngoc, Kinh, Hoang and Bales4, Reference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5. For example, in Vietnam in 1996, smokers spent an average of $US 49.05 on cigarettes per year, which was 1.5 times that spent on education, five times that spent on health care and about one-third that spent on food per capita in the household each yearReference Jenkins, Dai, Ngoc, Kinh, Hoang and Bales4. In the poorest households in Indonesia, more money was spent on tobacco than on education and health care combinedReference Reid6. Indonesia is the fifth largest market for tobacco in the world, with 182 billion sticks consumed per yearReference Achadi, Soerojo and Barber7. The absolute domestic consumption of tobacco increased by 159% between 1970 and 1980, coincident with the mechanisation of the cigarette industry in Indonesia in the early 1970sReference Achadi, Soerojo and Barber7. In the 1990s, an estimated 50% of men and 2.6% of women smoked cigarettes in Indonesia, usually kretekReference Barraclough8, and at present over 62% of Indonesian adult males smoke regularlyReference Achadi, Soerojo and Barber7. The prevalence of smoking is also increasing among adolescents in IndonesiaReference Smet, Maes, de Clerq, Haryanti and Winarno9.
Although smoking is thought to exacerbate poverty in developing countries, it is not well known whether smoking contributes to malnutrition among children. We hypothesised that among poor urban families in Indonesia in households where the father is a smoker: (1) children are at higher risk of malnutrition, and (2) household income spent on cigarettes is associated with proportionally lower expenditures on food compared with households where the father is a non-smoker. In order to examine these hypotheses, we characterised smoking and malnutrition among poor urban families in Indonesia.
Methods
The study subjects consisted of households that participated in a major nutritional surveillance system (NSS) in Indonesia that was established by the Ministry of Health, Government of Indonesia and Helen Keller International (HKI) in 1995Reference Bloem, Moench-Pfanner and Panagides10. The NSS was based upon the conceptual framework on the causes of malnutrition of the United Nations Children's FundReference de Pee, Bloem, Semba and Bloem11, with the underlying principle to monitor public health problems and guide policy decisionsReference Mason, Habicht, Tabatabai and Valverde12. The NSS was based upon stratified multistage cluster sampling of households in sub-districts of administrative divisions of the country and in slum areas of large citiesReference Bloem, Moench-Pfanner and Panagides10. The NSS in Indonesia involved the collection of data from approximately 40 000 randomly selected slum households every quarter. The NSS involved five major urban slum areas in the cities of Jakarta, Surabaya, Makassar, Semarang and Padang. New households were selected every round. Data were collected by two-person field teams. A structured coded questionnaire was used to record data on children aged 0–59 months, including anthropometric measurements, date of birth and sex. The mother of the child or another adult member of the household was asked to provide information on the household's composition, parental education and weekly household expenditures, along with other socio-economic, environmental sanitation and health indicators. The field teams measured and recorded the weight of each child aged 0–59 months to a precision of 0.1 kg and the length/height to a precision of 0.1 cm. Birth dates of the children were estimated using a calendar of local and national events and converted to the Gregorian calendar. Z-scores of weight-for-height (wasting), weight-for-age (underweight) and height-for-age (stunting) were calculated using EpiInfo software (Centers for Disease Control and Prevention), which uses the reference population of the US National Center for Health Statistics. Children with Z-scores below − 2 standard deviations (SD) from the median for weight-for-height, weight-for-age or height-for-age were considered wasted, underweight or stuntedReference de Onis, Semba and Bloem13. Severe wasting, underweight and stunting was defined by respective Z-score less than − 3SD. HKI provided training to new field teams, field supervisors and assistant field officers, and refresher training prior to each new round of data collection. During each round, a monitoring team from HKI visited all field sites to check and calibrate the equipment and supervise data collection. A quality control team from HKI revisited 10% of households without prior warning within two days of data collection by the field teams and re-collected data on selected indicators, including anthropometric measurements. Data collected by these quality control teams were later compared with the data collected by the field teams to check the accuracy of the data collection.
From 1999 to 2003, the NSS included questions on paternal and maternal smoking and weekly expenditures on cigarettes. In each household, data were gathered regarding expenditures the previous week on rice, other staple foods (cassava, sago, etc.), eggs, vegetables and other plant sources of food (bean curd, tempeh), fruits, cooking oil, beef, chicken, fish, sugar, instant noodles, milk, snacks, clothes, housing, education, cigarettes, savings, social activities, medicine, production activities, recreation, transportation, pocket money, water and other (gasoline, electricity, telephone, soap, seasonings, etc.). Expenditure and price variables were collected in Indonesian rupiah. For this analysis, expenditures are presented in $US to control for the fluctuation of the rupiah. Monthly exchange rates from 1999–2003 were established using historic data publicly available through the Bank of Canada14. Average exchange rates by data collection round were calculated in Excel® (Microsoft Corporation) based upon the months in which data were collected for each round. Expenditure and price variables in $US per round were created and calculated within SPSS software (SPSS Inc.) using the appropriate exchange rates by round.
The study protocol complied with the principles enunciated in the Helsinki Declaration15. The field teams were instructed to explain the purpose of the NSS and data collection to each child's mother or caretaker, and, if present, the father and/or household head; data collection proceeded only after written informed consent. Participation was voluntary and all subjects were free to withdraw at any stage of the interview. The NSS was approved by the Ministry of Health, Government of Indonesia, and the plan for secondary data analysis was approved by the Institutional Review Board of the Johns Hopkins University School of Medicine.
Malnutrition in children was defined using criteria of the World Health Organization (WHO) for stunting, underweight and wastingReference de Onis, Semba and Bloem13. In analyses where child malnutrition was the outcome and there was more than one child in the household, the youngest child in the household was used as the index of child malnutrition for that particular household (i.e. households were not counted more than once). Maternal and paternal age was divided into quartiles. Maternal and paternal education was categorised as 0, 1–6 (primary), 7–9 (junior high) and ≥ 10 years (high school or greater). The proportion of mothers and fathers who had achieved >12 years (high school graduate) was small and thus included in the category ≥ 10 years. Weighting was used to adjust for urban population size, and all results are weighted. Univariate and multivariate logistic regression models were used to examine the relationship between paternal smoking and the risk of wasting, underweight and stunting in the youngest 0–59-month-old child in the household. P < 0.05 was considered significant.
Results
From 1999 to 2003 there were 179 370 households surveyed, of which there were 175 859 households (98.0%) with information collected on paternal smoking. The prevalence of paternal smoking was 73.8%. The characteristics of households in which the father was a smoker were compared with those of households in which the father was not a smoker (Table 1). In households where the father was a smoker, children were younger, the level of both paternal and maternal education was lower, the proportion of households with >4 individuals eating from the same kitchen was higher, and the mean weekly per capita expenditure was higher than in households where the father was a non-smoker.
Table 1 Characteristics of households where the father is a smoker or a non-smoker among poor urban families in Indonesia
* Mean (standard error of the mean).
The prevalence of child wasting was 10.0%. The prevalence of wasting within categories of specific household-related risk factors was compared (Table 2A). Risk factors associated with child wasting included child's age in the 12–23 month age category, male gender, lower maternal and paternal education, maternal and paternal non-smoking, and lower weekly mean household expenditure per capita.
Table 2A Risk factors, including paternal smoking, for child wasting (weight-for-height Z-score <−2) among poor urban families in Indonesia
* Mean (standard error of the mean).
Maternal age and the number of individuals in the household sharing the same kitchen were not significantly associated with wasting. In a univariate model (model 1) and a multivariate model adjusting for child gender and child age (model 2), and in a final model adjusting for child gender, child age, maternal age, maternal education and weekly per capita household expenditure (model 3), paternal smoking was associated with a lower risk of child wasting (Table 3).
The prevalence of underweight children was 34.1%. The prevalence of underweight within categories of specific household-related risk factors was compared (Table 2B). Risk factors that were associated with the child being underweight included older child age, female gender, older maternal age, lower maternal and paternal education, lower per capita weekly household expenditure and >4 individuals sharing the same kitchen. In a univariate model (model 1) and in multivariate models adjusting for child gender and child age (model 2), and a final model adjusting for child gender, child age, maternal age, maternal education and weekly per capita household expenditure (model 3), paternal smoking was not associated with the child being underweight (Table 3).
Table 2B Risk factors, including paternal smoking, for child underweight (weight-for-age Z-score <−2) among poor urban families in Indonesia
* Mean (standard error of the mean).
The prevalence of child stunting was 28.1%. The prevalence of stunting within categories of specific household-related risk factors was compared (Table 2C). Risk factors that were associated with child stunting included older child age, older maternal age, lower maternal and paternal education, paternal and maternal smoking, lower per capita weekly household expenditure and >4 individuals sharing the same kitchen. In a univariate model (model 1), a multivariate model adjusting for child gender and child age (model 2), and a final model adjusting for child gender, child age, maternal age, maternal education and weekly per capita household expenditure (model 3), paternal smoking was significantly related to increased risk of child stunting (Table 3).
Table 2C Risk factors, including paternal smoking, for child stunting (height-for-age Z-score <−2) among poor urban families in Indonesia
* Mean (standard error of the mean).
Table 3 Multivariate models for paternal smoking and other risk factors for child malnutrition among poor urban families in Indonesia
OR – odds ratio; CI – confidence interval; ref – reference category.
The relationship between paternal smoking and severe malnutrition was also characterised. The prevalence of severe wasting (weight-for-height Z-score < − 3), severe underweight (weight-for-age Z-score < − 3) and severe stunting (height-for-age Z-score < − 3) was 1.0, 6.3 and 7.0%, respectively. Using a similar approach for wasting, underweight and stunting as above, in multivariate models adjusting for child gender, child age, maternal age, maternal education and weekly per capita household expenditure, paternal smoking was associated with an increased risk of severe wasting (odds ratio (OR) = 1.17, 95% confidence interval (CI) 1.03–1.33, P = 0.018) and severe stunting (OR = 1.09, 95% CI 1.04–1.15, P < 0.001).
The proportions of weekly per capita household expenditures on food, cigarettes and other items in households in which the father was a smoker versus households in which the father was not a smoker are shown in Fig. 1. In households where the father was a smoker, 22% of weekly expenditures per capita were spent on cigarettes (Fig. 1A), and a smaller proportion was spent on foods such as animal foods, vegetables and fruits, rice and other staples, snacks and baby food, sugar and oil, and instant noodles, than in households in which the father was not a smoker (Fig. 1B).
Fig. 1 Weekly per capita expenditures for households with a smoking father (A) and a non-smoking father (B). Proportion of expenditures spent on major food types and other commodities. ‘Other foods’ was defined as ready-to-eat foods and foods not produced in the household. ‘Snacks’ were defined as commercially packaged products made in a factory. The category ‘Other’ consisted of electricity, gasoline, telephone, soap, seasonings, etc.
Discussion
The present study shows that, in poor urban households in Indonesia, paternal smoking was associated with an increased risk of stunting in children. Paternal smoking was also associated with an increased risk of severe malnutrition among young children, notably severe wasting and severe stunting. Paternal smoking was most strongly associated with stunting but not risk of underweight among children, and this may be due to the more chronic effect of a lower-quality diet in households where the father was a smoker. The proportion of weekly per capita household expenditures on quality foods such as eggs, fish, fruits and vegetables was reduced in households where the father was a smoker. The slightly protective effect of paternal smoking and wasting (weight-for-age Z-score < − 2) may be a chance finding, as paternal smoking was associated with a significantly increased risk of severe wasting (weight-for-age Z-score < − 3).
These findings suggest that the adverse effects of tobacco use include increasing the risk of malnutrition among young children of the household, as a large proportion of household income is diverted towards cigarettes with a lesser proportion spent on food. The present study is consistent with observations from Bangladesh that in poor families in which the father smoked, a large proportion of weekly income was spent on tobacco, diverting money that might be spent on foodReference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5. These findings also corroborate findings from the National Family Health Survey II in India of 92 486 households in which household tobacco use increased the risk of malnutrition among childrenReference Bonu, Rani, Jha, Peters and Nguyen16.
The per capita expenditure on tobacco in the lowest-income households may be increasing in Indonesia, from 9% of total expenditures in 1981 to 15% of total expenditures in 1996Reference de Beyer, Lovelace and Yürekli17. In the present study, cigarettes accounted for an average of 22% of weekly per capita household expenditures in poor urban households where the father was a smoker. The mean weekly per capita household expenditure in poor urban households was $US 3.56, thus an estimated $US 0.78 of weekly per capita household expenditure was spent on cigarettes that could have been spent on food. These data are consistent with a study of poor families in Bangladesh which showed if money were not spent on cigarettes and were used for food and other necessities, over 50% more money would be available to purchase food for the householdReference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5.
Among poor urban households in Indonesia, nearly three-quarters of fathers were smokers. The high prevalence of smoking among men in this study is comparable to the high prevalence of smoking among men in many countries in southeast Asia, including Vietnam (72.8%)Reference Jenkins, Dai, Ngoc, Kinh, Hoang and Bales4, Bangladesh (70.3%)Reference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5, Cambodia (65% in urban areas)Reference Smith, Umenai and Radford18, Malaysia (49.2%)Reference Morrow and Barraclough19 and the Philippines (54.0%)Reference Morrow and Barraclough19. The overall male smoking prevalence in this region is 62.3% – the highest in the worldReference Stanton20. In contrast, only 1% of women in the present study reported that they smoked cigarettes, which is also consistent with a relatively low prevalence of smoking among women in other countries in southeast Asia, such as Vietnam (4.3%)Reference Jenkins, Dai, Ngoc, Kinh, Hoang and Bales4, Bangladesh (3.3%)Reference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5, Malaysia (4.0%)Reference Morrow and Barraclough19 and the Philippines (12.6%)Reference Morrow and Barraclough19.
The strengths of this study are the detailed data collection on demographic factors, anthropometry and household expenditures on cigarettes, types of food and other items in a large number of households. The inferences from this study are limited to the urban poor in Indonesia, as rural households were not included, and the proportion of household expenditures may be different in wealthier households. Further work is needed to characterise the relationship between paternal smoking and child nutritional status in rural households and to corroborate these findings in other settings in southeast Asia.
The results from the present study support the growing belief that tobacco control, poverty alleviation and child health promotion should not be looked upon as mutually exclusive effortsReference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5, Reference Bonu, Rani, Jha, Peters and Nguyen16. The WHO has presented three main ways by which tobacco exacerbates poverty on the household level: first, expenditure of tobacco takes over money that could otherwise be spent on basic necessities; second, smoking leads to increased health care needs, lost productivity and premature death of wage earners; and third, those employed in tobacco-related work experience particularly low wages and high health risks21. In the hand-rolled kretek sector employment has remained relatively stable, but the work is labour-intensive and wages are only 63% of average manufacturing-sector wagesReference Achadi, Soerojo and Barber7. While previous studies have inferred that household health is linked to household smoking expenditure and would improve if the money spent on cigarettes were instead spent on foodReference Efroymson, Ahmed, Townsend, Alam, Dey and Saha5, Reference Shah, Vaite and Efroymson22, Reference Shah, Vaite and Efroymson23, the present study corroborates and extends these arguments by showing that paternal smoking is associated with increased child malnutrition.
In Indonesia, kretek cigarettes, which contain about two-thirds tobacco, one-third cloves and various additives and flavours, account for nearly 90% of the cigarettes consumedReference Lawrence and Collin24. Kretek cigarettes are available for purchase individually or in small, less expensive packs, and they are particularly accessible to the poorReference Reynolds25. Indonesian and multinational tobacco companies advertise heavily on billboards, television, cinemas and at sporting events, with tobacco ranked among the largest advertising spending categories in the countryReference Reynolds25. There are few restrictions on the tobacco industry's conduct, advertising and promotion in IndonesiaReference Achadi, Soerojo and Barber7, and Indonesia is the only country in southeast Asia that has not signed the WHO Framework Convention on Tobacco ControlReference Shibuya, Ciecierski, Guindon, Bettcher, Evans and Murray26, which would require implementation of advertising limitations and the banning of tobacco sales to youths27. In addition, relatively weak tobacco control legislation passed in 1999 was further weakened with an amendment in 2003 to drop sanctions against the tobacco industry for violation of tobacco control regulations, such as not including health warningsReference Achadi, Soerojo and Barber7. The heavy advertising and marketing of cigarettes in Indonesia may be a contributing factor to the high prevalence of smoking among Indonesian men.
Child growth is internationally recognised as the best global indicator of physical well-being in children, as children with wasting, underweight or stunting are at higher risk of deficient or delayed mental development and increased infectious disease morbidity and mortalityReference de Onis, Semba and Bloem13. Long-term consequences of child malnutrition include poor school performance, diminished intellectual achievement, reduced adult size and reduced work capacityReference de Onis, Semba and Bloem13. Among poor urban families in Indonesia, children may be needlessly going hungry because money that could be spent on necessities like food is being diverted to cigarettes. Smoking is potentiating malnutrition among children in the family, exacerbating povertyReference de Beyer, Lovelace and Yürekli17, and may have long-term implications for the health of future generations of children in Indonesia and other countries with poverty and widespread tobacco use.
Acknowledgements
Sources of funding: The study was supported by the United States Agency for International Development. This paper was conceived and written independently of the funding agency; the funding agency had no role in the study design, data collection, analysis, interpretation, writing, or decision to submit the manuscript for publication.
Conflict of interest declaration: There are no conflicts of interest.
Authorship responsibilities: R.D.S. conceived the study hypothesis, supervised data analyses and wrote the manuscript. K.L.K. and S.d.P. contributed to the data analysis. M.O.R. conducted the data analysis. M.S. was responsible for data management of the Indonesian NSS. M.W.B. established the Indonesian NSS. All co-authors made substantial contributions to data analysis and interpretation and the writing of the manuscript.
