RESEARCH
Impact of a cash-for-work programme on food consumption and nutrition among women and children facing food insecurity in rural Bangladesh
Impact d'un programme de rémunération en espèces du travail sur la nutrition et la consommation de nourriture chez les femmes et les enfants soumis à l'insécurité alimentaire dans les zones rurales du Bangladesh
Influencia del programa Dinero por Trabajo en el consumo de alimentos y en la nutrición entre las mujeres y los niños que sufren una situación de inseguridad alimentaria en las zonas rurales de Bangladesh
CGN Mascie-TaylorI,*; MK MarksII; R GotoI; R IslamII
IDepartment of Biological Anthropology, University of Cambridge, Pembroke Street, Cambridge, CB2 3RA, England
IIChars Livelihoods Programme, UK Department for International Development, Bogra, Bangladesh
ABSTRACT
OBJECTIVE: To determine whether a cash-for-work programme during the annual food insecurity period in Bangladesh improved nutritional status in poor rural women and children.
METHODS: The panel study involved a random sample of 895 households from over 50 000 enrolled in a cash-for-work programme between September and December 2007 and 921 similar control households. The height, weight and mid-upper arm circumference of one woman and child aged less than 5 years from each household were measured at baseline and at the end of the study (mean time: 10 weeks). Women reported 7-day household food expenditure and consumption on both occasions. Changes in parameters were compared between the two groups.
FINDINGS: At baseline, no significant difference existed between the groups. By the study end, the difference in mean mid-upper arm circumference between women in the intervention and control groups had widened by 2.29 mm and the difference in mean weight, by 0.88 kg. Among children, the difference in means between the two groups had also widened in favour of the intervention group for: height (0.08 cm; P < 0.05), weight (0.22 kg; P < 0.001), mid-upper arm circumference (1.41 mm; P < 0.001) and z-scores for height-for-age (0.02; P < 0.001), weight-for-age (0.17; P < 0.001), weight-for-height (0.23; P < 0.001) and mid-upper arm circumference (0.12; P < 0.001). Intervention households spent more on food and consumed more protein-rich food at the end of the study.
CONCLUSION: The cash-for-work programme led to greater household food expenditure and consumption and women's and children's nutritional status improved.
RESUMÉ
OBJECTIF: Déterminer si un programme de rémunération en espèces du travail lors de la période d'insécurité alimentaire annuelle au Bangladesh a amélioré l'état nutritionnel des femmes et des enfants pauvres vivant dans les zones rurales.
MÉTHODES: Le panel d'étude a concerné un échantillon aléatoire de 895 ménages sur plus de 50 000 ménages engagés dans un programme de rémunération en espèces du travail entre septembre et décembre 2007, mais aussi 921 ménages de contrôle semblables. La taille, le poids et la circonférence du bras à mi-hauteur de la femme et d'un enfant de moins de 5 ans de chaque ménage ont été mesurés au début et à la fin de l'enquête (durée moyenne: 10 semaines). Dans les deux groupes, les femmes ont indiqué des dépenses et une consommation alimentaire pour 7 jours pour le ménage. Les modifications des paramètres ont été comparées entre les deux groupes.
RÉSULTATS: Au départ, les deux groupes ne présentaient aucune différence significative. À la fin de l'étude, la différence dans la circonférence du bras à mi-hauteur entre les femmes du groupe d'intervention et le groupe de contrôle avait augmenté de 2,29 mm et la différence dans le poids moyen de 0,88 kg. Chez les enfants, la différence dans les moyennes entre les deux groupes avaient également augmenté en faveur du groupe d'intervention comme suit: taille (0,08 cm; P < 0,05), poids (0,22 kg; P < 0,001), circonférence du bras à mi-hauteur (1,41 mm; P < 0,001) et scores z pour rapport taille-âge (0,02; P < 0,001), rapport poids-âge (0,17; P < 0,001), rapport poids-taille (0,23; P < 0,001) et circonférence du bras à mi-hauteur (0,12; P < 0,001). Les ménages du groupe d'intervention avaient dépensé plus d'argent pour acheter de la nourriture et avaient consommé davantage d'aliments riches en protéines à la fin de l'enquête.
CONCLUSION: Le programme de rémunération en espèces du travail a entraîné une augmentation des dépenses et de la consommation alimentaires des ménages. De plus, l'état nutritionnel des femmes et des enfants s'est amélioré.
RESUMEN
OBJETIVO: Determinar si el programa Dinero por Trabajo mejoró el estado nutricional de las mujeres y los niños más desfavorecidos de las zonas rurales de Bangladesh durante un periodo de un año de inseguridad alimentaria.
MÉTODOS: El estudio de cohortes comprendió una muestra aleatoria de 895 hogares, de entre los más de 50 000 inscritos en el programa Dinero por Trabajo, entre septiembre y diciembre de 2007, y 921 hogares de características similares como grupo de control. En cada hogar se midieron la altura, el peso y la circunferencia braquial de una mujer y de un niño menor de 5 años al inicio y al final del estudio (duración media: 10 semanas). En ambas ocasiones, las mujeres informaron sobre los gastos y el consumo alimentarios semanales de sus hogares. Se compararon los cambios paramétricos existentes en los dos grupos.
RESULTADOS: Al inicio del estudio no se observó ninguna diferencia significativa entre los grupos. Al final del estudio, la diferencia en la circunferencia media braquial entre las mujeres del grupo de intervención y el de control había aumentado 2,29 mm y la diferencia en el peso medio era de 0,88 kg. Entre los niños, la diferencia de las medias entre los dos grupos también había aumentado a favor del grupo de intervención: altura (0,08 cm, p < 0,05), peso (0,22 kg, p < 0,001), circunferencia braquial (1,41 mm, p < 0,001) y las puntuaciones z de la altura/edad (0,02; p < 0,001), peso/edad (0,17, p < 0,001), peso/altura (0,23, p < 0,001) y circunferencia braquial (0,12, p < 0,001). Al final del estudio, los hogares del grupo de intervención gastaban más en alimentos y consumían más alimentos ricos en proteínas.
CONCLUSIÓN: El programa Dinero por Trabajo tuvo como resultado un mayor gasto en y consumo de alimentos en los hogares y una mejora del estado nutricional de mujeres y niños.
Introduction
Cash-for-work programmes are usually associated with disasters and emergencies and have been implemented in Afghanistan, Bangladesh, Haiti, Indonesia, Pakistan and Uganda.1 Many organisations employed cash interventions after the Asian tsunami of 2004 and cash-for-work programmes were widespread in Aceh, Indonesia, and in Sri Lanka.2 However, when implemented incorrectly, these programmes can disrupt the local economy, artificially inflate wages and result in unsustainable shifts in the labour force.3
In north-western Bangladesh there is an annual period of food insecurity (the monga) that generally occurs between mid-September and mid-November and is primarily caused by unemployment and a lack of income before the large aman rice harvest.4 People who live on chars, which are large flat islands in the main river channels, are particularly affected by the monga as flooding occurs during the preceding months almost every year. When flooding is severe, there can be household damage, a loss of assets, disruption of agricultural activities and obstacles to rearing livestock. Chars dwellers may be confronted with river bank erosion, which can take away their homestead and any land they possess. The scarcity of work also has severe repercussions for the income of poor chars households and many are obliged to reduce their food intake during this period.5
The Chars Livelihood Programme (http://www.clp-bangladesh.org), funded by the United Kingdom Department for International Development, aims to lift over 55 000 extremely poor households out of poverty by providing income-generating assets. Households are provided with a raised earthen plinth on which their homes are reconstructed and homestead gardens can be established. In part, these plinths are created by a labour-intensive earthmoving process that involves members of poor chars households.6
In August and September 2007, widespread severe flooding across northern Bangladesh intensified and prolonged the impact of the monga for chars dwellers.6 In some households participating in the Chars Livelihood Programme, the cash-for-work intervention was implemented between September and December 2007 to coincide with the monga. The programme provided approximately 2.6 million person-days of paid work during which both men and women received 36 taka (about US$ 0.50) for each cubic metre of soil moved in the construction of the earthen plinths.
No previous research has been conducted into the impact of cash-for-work programmes on nutritional status. In Bangladesh there was concern among some nongovernmental organisations that the physical labour required by the local cash-for-work programme would cause women to lose weight and to neglect their children because the mothers needed to work quite long hours. If this occurred, the programme would be counterproductive and would probably be detrimental to the health of the people it was intended to help. The present panel study was designed to determine whether women and children aged less than 5 years from landless households who were living on chars and who participated in a monga season cash-for-work programme were able to maintain a better nutritional status than similar women and children living in the same geographical area who belonged to households not taking part in the programme.
Methods
Study design
Over 100 000 households were enrolled in the Chars Livelihood Programme and the subsidiary cash-for-work programme was open to all members of the community. The wage rate was specified in advance. The panel study involved 1009 households containing a child aged less than 5 years and an adult female (usually the mother) which were randomly selected from more than 50 000 taking part in the cash-for-work programme and 1051 similarly selected control households that were not taking part in the programme. The two groups were comparable in household size, the adults' occupations and the age of the adult female family member. Several households were not included in the final analysis because the observed z-scores for some anthropometric measurements were outside of the acceptable ranges, which were -6 to +6 for the height-for-age z-score, -6 to +5 for the weight-for-age z-score, and -5 to +5 for the weight-for-height and mid-upper arm circumference z-scores.8 The final analysis included 1816 households (895 intervention and 921 control households). Tests of statistical power, carried out using SPSS Sample Power version 2 (SPSS Inc., Chicago, United States of America), showed that these numbers were sufficient to demonstrate significant differences at P < 0.01 and to provide a power of 90%, which was chosen to reduce errors. Of the 1816 households, 1800 (99%) had access to safe drinking water but had no cultivable or homestead land. Each intervention household contained either an adult female family member who actively worked in the cash-for-work programme (i.e. an "active" woman) or an adult female family member who did not work for the programme (i.e. a "non-active" woman) though a male household member did. There were 426 active women and 469 non-active women. In addition, all 895 intervention households contained one or more children aged less than 5 years. If there was more than one child aged less than 5 years in the household, one was selected randomly for assessment in the study.
Data collection
The study was carried out for the Chars Livelihood Programme by trained staff working for Helen Keller International, Dhaka, Bangladesh. Both intervention and control households were recruited between mid-September and early October 2007 from the Gaibandha and Kurigram districts of north-western Bangladesh. A structured interview was completed by the same adult female at baseline and at the end of the study in all households. The study teams used standard interviewing techniques, including reverse chronological description of consumption. The questionnaire used contained items on the number of days on which specific foods were consumed in the 7 days before the interview, the household's dietary habits, morbidity, socioeconomic factors and the household's contingency plans and coping strategies for the monga. Information on food consumption and the cost of food was collected verbally. Food costs were verified independently by research staff who visited local markets to determine the costs of the different food types mentioned in the questionnaire.
The women's height, weight and mid-upper arm circumference were measured at both baseline and at the end of the study, which was typically in mid-December 2007 since the mean duration on the panel was 10 weeks (range: 8-12). In addition, the height, weight and mid-upper arm circumference of each woman's child less than 5 years or of another child aged less than 5 years from the same household were also measured. All measurements were made using standard anthropometric techniques.7 Although no particular provision was made for caring for children aged less than 5 years, all workers in the cash-for-work programme were assigned to a specific site in their respective villages and were given a 2-hour break during the day. Height-for-age, weight-for-age, weight-for-height and mid-upper arm circumference z-scores were determined for the children using international standards,8 and a child was classified as stunted, underweight, wasted or of low mid-upper arm circumference, respectively, if the z-score was < -2.0. Each woman's body mass index (BMI) was calculated and women were categorized according to their BMI as having either grade III chronic energy deficiency (BMI: < 16.0), grade II chronic energy deficiency (BMI: 16.0-16.9) or grade I chronic energy deficiency (BMI: 17.0-18.49) or a normal nutritional status (BMI: > 18.5).9 Data were verified by supervisors from Helen Keller International who revisited a random selection of 5% of households.
Statistical analysis
As this was a panel study, the analysis investigated within-woman and within-child changes between baseline and the end of the study. Sequential multiple regression analysis was used to test for changes in the mean differences in anthropometric and nutritional variables between intervention and control groups; corrections were made for the linear and quadratic effects of age in women and for the linear and quadratic effects of age and sex in children. For categorical data, χ2 tests were used to examine differences between the groups and McNemar's test was applied to paired samples.
Results
Baseline characteristics
Households
The baseline demographic characteristics of households participating in the study are presented in Table 1. There was no significant difference between intervention and control households in the mean age of the women or child in the household or in the proportion of children who were male, which was just over 50%. Illiteracy levels were very high in both genders, and higher in females than males. The significantly higher illiteracy rate found in intervention households was therefore partially attributable to the fact that significantly more intervention households had a female head. The main wage earner in most households was a labourer, who was usually employed on a daily basis.
Women
There was no significant difference in mean baseline height, weight, BMI or mid-upper arm circumference between women in intervention and control households. Overall, their mean height was 150.5 cm, their mean weight was 42.2 kg, their mean mid-upper arm circumference was 234.1 mm and their mean BMI was 18.6. Overall, 51.6% (937/1816) of all women had a BMI < 18.5. The percentages of women with different grades of chronic energy deficiency were very similar in the two groups: 30.4% and 34.6% of those in intervention and control households, respectively, had grade I chronic energy deficiency; 13.6% and 11.8%, respectively, had grade II chronic energy deficiency; and 7.2% and 5.7% respectively, had grade III chronic energy deficiency. In intervention households, there was no significant difference between non-active women and active women in mean baseline height, weight, BMI or mid-upper arm circumference or in the percentage with different grades of chronic energy deficiency.
Children
No significant difference in mean baseline height, weight or mid-upper arm circumference was observed between children from intervention and control households. Overall, their mean height was 82.8 cm, their mean weight was 10.1 kg and their mean mid-upper arm circumference was 141.7 mm. Their mean height-for-age z-score was -1.98, their mean weight-for-age z-score was -1.94, their mean weight-for-height z-score was -1.17 and their mean mid-upper arm circumference z-score was -0.98. Overall, 51.3% (932/1816) of the children were stunted, 47.6% (864/1816) were underweight, 18.6% (379/1816) were wasted and 9.4% (171/1816) had a mid-upper arm circumference z-score < -2.0.
Household food
There was no significant difference between intervention and control households in the mean quantity of pulses, eggs, fish, green leafy vegetables, meat, cereal, rice or fruit reported as being consumed in the 7 days before the baseline survey was carried out, nor was there any significant difference in mean expenditure on these food items.
Changes during the study
Women and children in the intervention group showed significantly greater improvements in all anthropometric and nutritional measures than the control group. Changes in anthropometric measurements between baseline and the end of the study are illustrated in Fig. 1.
Women
On average, women in the intervention group increased in weight and mid-upper arm circumference during the study, whereas those in the control group experienced decreases: the mean increase in weight in women in the intervention group relative to those in the control group was 880 g and the mean relative increase in mid-upper arm circumference was 2.29 mm. Significantly fewer women in the intervention group had a BMI < 18.5 at the end of the study: the percentages were 48.4% and 56.6% in intervention and control groups, respectively (χ2 = 11.7; P < 0.001). Correspondingly, the percentages of women with grade I, grade II or grade III chronic energy deficiency were all lower in the intervention than in the control group (data not shown). In the intervention group, BMI improved in 9.7% of women while it worsened in 0.9%; the corresponding figures in the control group were 2.4% and 9.3% for improved and worsened BMI, respectively. Overall, 3.4% of women in the intervention group improved from having chronic energy deficiency to a normal BMI, compared with no women in the control group. Conversely, 4.3% of women in the control group changed from having a normal BMI to having chronic energy deficiency, compared with 0.7% in the intervention group. There was no significant difference in weight gain between non-active women and active women in intervention households: the mean gains in these two subgroups were 0.95 kg and 0.80 kg, respectively. Nor was there a significant difference in the increase in mid-upper arm circumference (2.74 mm and 1.74 mm in the two subgroups, respectively) or in the increase in BMI (0.42 and 0.36, respectively).
Children
During the study, children from intervention households gained, on average, 0.7 mm in height, 210 g in weight and 1.39 mm in mid-upper arm circumference more than those from control households, after adjustments for age and sex (Fig. 1). Among children from intervention households, but not those from control households, there were also significant reductions in the percentages who were underweight or wasted or who had a low mid-upper arm circumference. For example, 7.3% of the children from intervention households improved from being underweight to having a normal weight, compared with only 3.3% of those from control households. Conversely, only 2.1% of children from intervention households became underweight, compared with 7.7% of those from control households (χ2 = 43.7; P < 0.001). In these two groups, respectively, 9.1% and 6.9% improved from being wasted to having a normal weight-for-height, while1.3% and 5.4% became wasted (overall χ2 = 25.6; P < 0.001). There was no significant difference between children from intervention households with active or non-active women.
Household food
Overall, 99.7% of intervention households reported spending some of the money earned in the cash-for-work programme on food and nearly three-quarters stated that a family member had paid for medical treatment. Intervention households spent significantly more on food than control households (Table 2), with marked percentage increases in the consumption of eggs, meat, fish, milk, pulses and green leafy vegetables (Fig. 2).
At the end of the study, the reported consumption of eggs, meat, fish, pulses, green leafy vegetables, milk and fruit was significantly greater and cereal consumption was lower among children from intervention households than among those from control households (Table 3). The differences in egg, meat and fish consumption were particularly marked. For example, 74.6% of children from intervention households had eaten fish on three days or more in the 7 days before the end of the study, compared with 2.7% of those from control households.
As shown in Fig. 2, food consumption reported by intervention households increased between baseline and the end of the study for seven of the eight food types examined (i.e. all except cereals), while consumption by children from control households decreased for all eight food types. Moreover, these changes were all significant (McNemar's paired test: P < 0.001 for all food types).
Discussion
The cash-for-work programme reported here may be viewed in the wider context of the longer-term conditional cash transfer programmes that have been increasingly used to transfer money to poor households on the condition that they comply with certain requirements: for example, attending health-care sessions, using food and nutritional supplements or enrolling children in school. The few conditional cash transfer programmes that have focused on nutritional status have produced conflicting results. One Brazilian study found a small negative impact,10 while a Mexican study showed that children in programmes involving both cash transfer and multi-micronutrient supplementation grew about 1 cm more than those who received neither intervention.11 However, it was not possible to differentiate the effect of cash transfer from that of multi-micronutrient supplementation. Recent analyses of Mexican data have shown that doubling the quantity of cash transferred was associated with, for example, a better height-for-age ratio and a lower prevalence of childhood stunting.12
The results of the Bangladeshi study reported here indicate that implementing a cash-for-work programme at a crucial time of the year led to significant gains in the nutritional status of both children aged less than 5 years and women. These improvements also occurred in women who took part in the physically demanding cash-for-work programme. Although this was a short-term intervention, it could have longer-term benefits: selling household assets may become unnecessary and individuals may not have to agree in advance to work for less money in the future. In addition, the earthen plinths constructed in the programme, which have a life expectancy of 25-30 years, improve household security.
Although the improvements observed in child nutritional status are encouraging, with the main gain being less wasting due to acute malnutrition in intervention households (Fig. 1), many individuals in both intervention and control households continued to suffer from malnutrition. In a recent nationwide survey,13 39.2% of children aged less than 5 years in Bangladesh were found to be stunted, while 45.7% were underweight and 11.9% were wasted. At the end of the study, the prevalence of underweight and wasting in intervention households resembled the national prevalences: 43.7% and 11.4%, respectively. However, the corresponding figures in control households were worse, at 50.8% and 16.5%, respectively. Notably, the prevalence of stunting at the end of the study was much higher than the national average in both groups: it was 60.6% in children from intervention households and 64.5% in those from control households. Moreover, it had increased during the study: by 10.7% in the intervention group and by 11.9% in the control group. The higher rates of chronic malnutrition may have been partly due to the seasonal changes in nutritional status that have been documented in Bangladesh13 or to increased levels of infection and morbidity occurring at the time of the year when the study took place.14
Although no specific nutritional education was provided, households taking part in the cash-for-work programme spent more, in particular, on protein-rich foods such as fish, meat, eggs and milk, as well as on fruit, which is rich in folate, potassium and vitamin C. Detailed information on the actual quantity of food consumed by women and their children was not available, as that would have necessitated collecting data about portion sizes or weighing the food consumed.15 However, the level of food consumption reported is plausible. The resulting more diverse diet should lead to increased intake of vitamins important for growth, development and general health.
In conclusion, our findings indicate that the cash-for-work programme led to a greater quantity and variety of food, particularly animal protein, being consumed during the annual period of food insecurity in Bangladesh and resulted in a significant improvement in the short term nutritional status of women and children living on chars. Even women who actively took part in the physically demanding work associated with the programme experienced an improvement in nutritional status. Over the longer term, the significant reduction of both acute and chronic malnutrition among chars dwellers will depend on lifting them out of poverty. 
Funding: This study was funded by the United Kingdom Department for International Development as part of the research and impact monitoring activities of the Chars Livelihood Programme.
Competing interests: None declared.
References
1. CARE starts "cash-for-work" programme for school cleaning in Haiti. Geneva: Reliefweb (CARE); 2008. Available from: http://www.reliefweb.int/rw/rwb.nsf/db900sid/EDIS-7K6MQB?OpenDocument) [accessed 3 September 2010] .
2. Doocy S, Johnson D, Robinson C. Cash grants in humanitarian assistance: a nongovernmental organization experience in Aceh, Indonesia, following the 2004 Indian Ocean Tsunami. Disaster Med Public Health Prep 2008;2:95-103. doi:10.1097/DMP.0b013e318170b5b4 PMID:18525372
3. Cuny F, Hill R. Famine, conflict and response: a basic guide. West Hartford: Kumarian Press; 1999.
4. Zug S. Monga - seasonal food insecurity in Bangladesh: bringing the information together. J Soc Stud 2006:111.
5. Conroy K, Marks M. The use of coping strategies by extreme poor households on the Jamuna Chars during monga. Bogra: Chars Livelihood Programme; 2008. Available from: http://www.clp-bangladesh.org/index2.php?option=com_docman&task=doc_view&gid=80&Itemid=99 [accessed 20 September 2010] .
6. Conroy K, Islam R, Marks M. The impact of the 2007 CLP infrastructure and employment programme. Bogra: Chars Livelihood Programme; 2008. Available from: http://www.clp-bangladesh.org/index2.php?option=com_docman&task=doc_view&gid=37&Itemid=99 [accessed 20 September 2010] .
7. Lohmann TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign: Human Kinetics Books; 1989.
8. World Health Organization child growth standards: length/height-for age, weight-for-age, weight-for-length, weight-for-height and body mass index-for-age: methods and development. Geneva: World Health Organization; 2006. Available from: http://www.who.int/childgrowth/standards/Technical_report.pdf [accessed 20 September 2010] .
9. Shetty PS, James WPT. Body mass index: a measure of chronic energy deficiency in adults. Rome: Food and Agriculture Organization; 1994.
10. Morris SS, Olinto P, Flores R, Nilson EA, Figueiró AC. Conditional cash transfers are associated with a small reduction in the rate of weight gain of preschool children in northeast Brazil. J Nutr 2004;134:2336-41. PMID:15333725
11. Berhman JR, Hoddinott J. Program evaluation with unobserved heterogeneity and selective implementation: the Mexican PROGRESA impact on child nutrition. (PIER Working Paper 02-006). Philadelphia: University of Pennsylvania; 2002.
12. Fernald LCH, Gertler PJ, Neufeld LM. Role of cash in conditional cash transfer programmes for child health, growth, and development: an analysis of Mexico's Oportunidades. Lancet 2008;371:828-37. doi:10.1016/S0140-6736(08)60382-7 PMID:18328930
13. HKI. Bangladesh in facts and figures: 2005 annual report of the Nutritional Surveillance Project, Dhaka. Dhaka: Helen Keller International Bangladesh; 2006.
14. HKI. Household and community level determinants of malnutrition in Bangladesh. Nutritional Surveillance project Bulletin No. 17. New York & Cambridge: Helen Keller International &Institute of Public Health; 2006.
15. Rousham EK, Mascie-Taylor CGN. Seasonality and child morbidity in rural Bangladesh. Am J Hum Biol 1995;7:369-79. doi:10.1002/ajhb.1310070313
(Submitted: 15 June 2009 - Revised version received: 1 July 2010 - Accepted: 10 August 2010 - Published online: 29 September 2010 )
* Correspondence to CGN Mascie-Taylor (e-mail: nmt1@cam.ac.uk).