Cash Transfers and Child Nutrition: Evidence from sub-Saharan Africa

This paper summarises the arguments and counterarguments within the scientific discussion on cash transfers and child nutrition. The main purpose of the research is to assess the effectiveness of cash transfers in improving nutritional outcomes in vulnerable children in sub-Saharan Africa. Systematisation of the literary sources indicates that studies have justified cash transfer as social-income support that addresses a vital social determinant of health (income) for children in low-and-middle-income countries. The methodological basis of this study is a systematic review that searched a wide range of academic and grey literature databases, including PubMed, Cochrane Library and Google Scholar. This study included cluster-randomised controlled trials (R.C.T.s), randomised controlled trials, quasi-experimental studies, mixed-methods studies, and non-randomised cluster trials. Studies included in this systematic review were screened for their eligibility. The systematic review uses the Cochrane data collection form to extract data from the included studies. It was not feasible to statistically combine the results of the studies due to the heterogeneity of most of the studies. Preferably, the review employs a narrative synthesis to present the estimated effects of cash transfers on children’s nutritional outcomes. The systematic review presents the results of data synthesis, of which eleven studies met the inclusion criteria. Overall, the evidence from the systematic review indicates that cash transfer programmes targeted at children effectively improve anthropometric and nutritional outcomes. Further research is needed to spell out the multiple pathways to how cash transfers improve children’s nutritional outcomes. Moreover, this systematic review shows the importance of cash transfers in improving child nutrition. Policymakers should continue to employ institutional mechanisms to strengthen the nutritional status of children, especially the vulnerable ones since cash transfer intervention is a temporary measure.


Introduction
Globally, over 200 million children between 0-5 years are not accomplishing their potential for socio-emotional development due to income poverty, poor health, and nutrition (Grantham-McGregor et al., 2007). Current evidence showed that at least more than 600 million children are affected by different dimensions of child poverty (Save the Children International and Africa Platform for Social Protection, 2017).
Poverty is an anomaly that affects all parts of the world (Ekezie et al., 2017). It comes in various forms such as hunger and malnutrition, engagement in precarious work, childhood marriages, death during infancy and limited of vulnerability than adults and the role that cash transfers can play in providing enough nutrition and access to and utilising social services (Sanfilippo et al., 2012).
Several concepts for developing a conceptual framework have been used to hypothesise and design the pathways between C.T. programmes and child nutrition (de Groot et al., 2015. The most prominent approach is to initiate from the determinants of child nutrition and hypothesise the effects of a cash transfer programme on those determinants (de Groot et al., 2015. This approach is appropriate as it describes how C.T. can affect the root causes of child nutrition and thus helps shed light on the channels of impact (de Groot et al., 2015). Gaarder and colleagues presented a vital analysis of health conditional from eleven conditional cash transfer (C.C.T.) programmes (Gaarder et al., 2010).
De Groot and colleagues stipulated three significant pathways through which C.T. may influence the primary determinant of child nutrition by providing supplementary financial sources available in households for food security, health, and care. This study now summarises how these pathways may be influenced by C.T. using De Groot and colleagues' conceptualisation (de Groot et al., 2015.

Enhanced child nutrition through improved resources for food availability
One of the aims of many C.T. programmes is to improve the food security situation among beneficiaries (Hjelm, 2016). Vulnerable people in developing countries usually face high degrees of food insecurity, affecting families living in poverty (Hjelm, 2016). Children are especially vulnerable to food insecurity, as nutritious food is vital for child development (Hjelm, 2016). C.T. programme increases family disposal income and, consequently, the resources available for family food security (de Groot et al., 2015. If families use the cash to buy nutritious food or invest in food production, family food security and diet diversity improve (de Groot et al., 2015. In Latin America, C.C.T. programmes have strong evidence of improving child health and nutritional status (Segura-Pérez et al., 2016). In sub-Saharan Africa, CT programmes have shown to be an efficient mechanism for increasing families' calorie intake (Burchi & Strupat, 2016). Nevertheless, de Groot and colleagues asserted that the presence of food, food prices and economic shocks could moderate these pathways (de Groot et  programmes give cash to vulnerable people, and they directly affect poverty reduction (Doetinchem et al., 2008). Beneficiaries can make their own decisions about how to spend the money. Likewise, it is supposed to positively impact beneficiaries' socioeconomic wellbeing, such as improving a household dwelling (Doetinchem et al., 2008). In C.C.T.s, the conditionality motivates vulnerable people to invest in their human capital to eradicate an inter-generational poverty cycle. Health is one of the most significant elements enabling future generations to overcome poverty (Doetinchem et al., 2008). Beneficiaries (Browne, 2013). According to Browne (2013), the hypotheses in the ToC include: ❖ C.C.T. programmes lead to a rise in the uptake of preventive health services among vulnerable people who are presently underutilising these.

❖
An increase in access to healthcare services will improve health status, and particularly an increase in utilising public health services will have this effect.

❖
Money from social cash transfer programmes affects health basically by ensuring services accessibility and increased food consumption.

❖
Women in poor households have limited health knowledge, and that a transfer of information to them will generate behaviour changes.

❖
Enforcing conditions and observing compliance are significant to increase uptake of services to the needed level.

❖
Some programmes have accepted that the conditionalities are insufficient to ensure adequate child nutritional investment and have included a food supplement.

✓
Improved child nutrition through increased resources for care . Parental stress can affect child outcomes. A total decrease in family stress level may also impact caregiver behaviour and precisely influence child health. Additionally, CT can offer mintage incentives for expectant mothers to engage in precarious work, which has implications for birth outcomes; following increases resources for care, care for mothers and children may improve.

Types of studies
This study uses the EPOC (EPOC, 2017a) inclusion criteria to determine the study designs are vital for evaluating the impact of C.T. programmes on young children's nutritional outcomes. The following study designs were eligible for this study: ➢ Randomised controlled trials (R.C.T.s) and cluster-randomised controlled trials (C-RCTs), ➢ Quasi-experimental; ➢ Mixed methods; ➢ Quantitative analysis.

Types of participants
This impact evaluation aims to systematically assess the effects of C.C.T. and U.C.T. programmes on children's nutritional outcomes. I restricted the study population to children between the ages of 0-18 years old living in poor households in sub-Sahara Africa as defined by the World Bank (World Bank, 2019).

Types of interventions
This study considered relevant articles that evaluated the impacts of cash transfers on different children's nutritional outcomes. For cash transfer interventions to be included in this study, they had to meet the following criteria: ➢ the transfers had to be conditional or unconditional; ➢ regularly provided (monthly, once in two months, quarterly); ➢ provided to reduce poverty, increasing access to health services, food security and education; ➢ transfer to beneficiaries through electronic means, face to face or any other convenient means; ➢ transfer to households with orphans or vulnerable children; ➢ must be a non-contributory cash transfer; ➢ help prevent acute malnutrition in young children.

Outcome measures
The outcomes included in this study in terms of nutritional outcomes ensure comparability with the systematic review of the impact of cash transfers on nutritional outcomes in low-and-middle-income countries (Pega et al., 2017). Nutritional outcomes, including but not limited to: ➢ mid-upper arm circumference; ➢ weight-for-height; ➢ height-for-age; ➢ food consumption; ➢ diet diversity; ➢ underweight; ➢ stunted.

Search methods for identification of studies
Electronic searches I carried out the initial searches to identify studies for this systematic review as part of a holistic review of the role of both C.C.T.s and U.C.T.s in improving child nutritional status. The searches for relevant studies were conducted in different databases to determine their eligibility. I used different search terms to search for the studies initially selected for the systematic review. The search terms included, but not limited to: ➢ vulnerability; ➢ children; ➢ poor household; ➢ food security; ➢ food nutrient.
To avoid selection bias, I carried out comprehensive and rigorous searches for relevant studies in the academic (Pubmed, MEDLINE., Scopus) and grey literature (African Health Journals, African Journals Online, Google Scholar) databases.

Selection of studies
The searches for relevant studies were conducted on academic and grey literature databases and the websites of major international organisations that are stakeholders and actors in international development. Additional searches were conducted on past reviews, books and reference lists of the included studies.
I screened the titles and abstracts of studies initially identified from the searches conducted from the resources mentioned above for relevance. Through this process, irrelevant studies were eliminated with the retaining of others for further screening. For studies to meet the inclusion criteria of this systematic review, I screened the full text of studies retained to identify their eligibility. It was through the screening of the full text of relevant studies that duplicates were removed.

Data extraction and management
I extracted data from each included study with EPOC data collection form (EPOC, 2017b). The data collection form is meant for intervention review, and it is for studies with randomised trials and non-randomised trials. I extracted the following information from the included studies using the EPOC data collection form: ➢ study citation (including author(s)' name and date of publication); ➢ year and duration of the study (impact evaluation of cash transfers on children's outcomes); ➢ ages of study participants; ➢ characteristics of interventions (amount of the transfer, conditionality, purpose of transfer); ➢ a sample size of treatment group; ➢ a sample size of the control group; ➢ type of study (randomised trials, non-randomised trials or mixed methods); ➢ study setting (country and geographical location); ➢ methods of impact estimation (multilevel logistic regression models, propensity score matching); ➢ outcomes measured.

Assessment of the risk of bias in the included studies
There are different assessment tools for assessing the quality of relevant studies in a systematic review. However, I opted for the Mixed Methods Appraisal Tool (MMAT) Version 2018 to evaluate the methodological quality for studies included in this systematic review (Hong et al., 2018). The reason for using MMAT is that "it permits to appraise the methodological quality of five categories of studies: qualitative research, randomised controlled trials, non-randomised studies, quantitative descriptive studies, and mixed methods studies" (Hong et al., 2018:1).
Most of the included studies in this systematic review fell within the five categories of the study types mentioned above. To assess the risk of bias of the included studies, I applied the MMAT tool in each category of study types under review. I reported the risk of bias of individual studies at the methodological level. I judged each potential study source of bias as high, moderate, and low.

Data synthesis
Owing to the heterogeneity in the designs, interventions, sample size and outcomes mentioned in the included studies, it was not feasible to statistically combine the results of the studies. Instead, I used narrative synthesis to present the estimated effects of C.T.s programmes on children's health and nutritional outcomes. Before using narrative synthesis to present the results from the included studies, I initially used manual coding to identify salient information. The second round of coding was used to identify categories and themes. I used the final coding stage to identify similarities and differences of themes, and I presented the coded information in text and tables. Figure 1 presents a flowchart of the study selection process. The systematic search for relevant studies initially identified 3,803 articles through electronic databases, websites, and reference lists of the included studies. Among the identified studies, a total of 11 studies met the inclusion criteria for the systematic review.

Figure 1. Flowchart of study selection
Source: Author's design from included studies.
The initial searches for relevant articles on academic databases produced 1,412 titles. To complete the search criteria for the systematic review, this study conducted searches on grey literature databases and the websites of prominent organisations that are stakeholders in international development. Through these searches, the study identified an additional 391 articles. During the screening of the titles and abstracts of these identified studies, I excluded 1,746 articles because they did not meet the study's inclusion criteria.
The remaining 57 articles were screened for full-text review to determine their eligibility. One relevant study (Tonguet-Papucci et al., 2017), was identified from the reference lists of one of the 57 articles bringing the total articles for full-text screening to 58 articles. Of the 58 articles, 47 were considered ineligible and excluded because of some reasons ranging from population, intervention, duplicate data and irrelevant outcomes. outcomes measured by these studies were grouped into (i) anthropometric and (ii) nutrition. While some studies measured one outcome, others measured two or more outcomes.

Included studies
The description of the characteristics of the included studies is on Table 3. However, more detailed descriptions of the main features of the included studies are discussed below.

Types of study designs
Of the eleven studies included in this systematic review, four were cluster-RCTs (

Participants
The included studies involved 136,022 participants in their analyses of the impacts of C.T. programmes on children's health and nutritional outcomes. The included studies participants include 36,234 households, 9,783 children, 239 villages, and 89,766 others consisted of political wards (120) mother and child living pairs (3,443) and individuals 86,203. (Figure 6).

Interventions
In all the studies, the interventions were targeted at poor households with children, except in South Africa, where the intervention was targeted at the KwaZulu-Natal which was not the poorest province in South Africa but was considered to have the highest incidence of deprivation in terms of access to social services and perceived wellbeing (Agüero et al., 2007). This systematic review included one C.C.T. and 10 U.C.T.s. The main characteristics of the C.C.T. and U.C.T. interventions are detailed in Table 2. National governments operated most of the C.T. interventions included in this systematic review except for the interventions in Uganda and Somalia. The beneficiaries of the interventions were poor households with children, while the main objectives of the interventions were to reduce household poverty and enhance children's nutritional status. Different mechanisms were used to transfer the cash to their beneficiaries. Among these were face-to-face direct cash payment, bank transfer, pay point, mobile money transfer and electronic transfer of funds to cards. The U.C.T.s were without any primary conditionalities attached to them, while the only C.C.T. had conditions attached to it. The main conditions attached to the C.C.T. were health clinic attendance and children must go to school. One unique aspect of the C.T.s was in Uganda, where C.T. was primarily meant to be C.T.s for a child's early childhood development (ECD) centres. However, the conditionality attached to the intervention was removed because of some irregularities observed in the implementation process (Gilligan et al., 2013).   Table 3 shows the details of the possibility of each type of bias in individual study. I used the MMAT risk of bias tool prepared by (Hong et al., 2018) to assess the methodological quality of each study included in the systematic review. From the assessment, I considered the overall risk of bias in this study to be moderate. The majority of the included studies were cluster-RCTs and R.C.T.s whose designs were used to assess the effects of C.T. programmes on children's nutritional outcomes. All the studies in the cluster-RCTs and R.C.T.s category have straightforward research questions, and the data collected by these studies were able to answer their research questions. Aside from one study, Houngbe et al. (2017), the rest of the studies demonstrated how they performed their randomisation to a reasonable level. The randomisation in most of the studies was done at the household, village, and individual levels. All randomised controlled trials provided a baseline. None of the randomised controlled trials gave required details on assessors blinded to the intervention provided. One study did not explain whether study participants adhere to the assigned intervention (Gilligan et al., 2013). One study in the mixed methods category presented minor limitations. It was considered to be presenting a moderate risk of bias (Abdoulayi et al., 2016). The study domains of shortcomings were in columns 8 and 9 in Table 3

Height-for-age (stunting)
Six studies (two cluster-RCTs, two mixed methods and one each of R.C.T. and quantitative descriptive study) assessed the effects of five U.C.T. programmes and one C.  (Table 4). In Burkina Faso, there was no change between the treatment and control groups over the 24 months follow-up. The odds of stunting in the two groups at the end of the U.C.T. programme were similar . In DR Congo, the cash transfer intervention did not positively affect child height gain because there was no catch-up in H.A.Z. for both treatment and control groups (Grellety et al., 2017). Children remained stunted during the impact evaluation (Grellety et al., 2017). Still on a negative note, in Malawi, despite 49% of the treated sample being stunted at baseline, there were no overall impacts of the U.C.T. programme on the prevalence of stunting. The evaluators also did not find any impact amongst the subgroups (Abdoulayi et al., 2016). In South Africa, the impact of C.S.G. on H.A.Z. was positive when the treatment began at the youngest age. However, the value of the effect decreased with the age of initial treatment (Agüero et al., 2007). When treatment was given to children at the age of two, the impact was still positive but no longer statistically significant. The study found no positive effect when the intervention was less than 20% of the nutritional window but found positive gains when the treatment covered approximately two-thirds of the nutritional windows (Agüero et al., 2007). In Tanzania

Weight-for-height (wasting)
Seven studies from Burkina Faso, DR Congo, Lesotho, Malawi, Tanzania, Uganda and Zambia used cluster-RTCs, quasi-experimental, R.C.T. and mixed methods approaches to assess the effects of C.C.T. and U.C.T. programmes on W.H.Z. or wasting ( Table 4). The evidence from these studies was mixed, with studies from DR Congo, Lesotho, Malawi and Zambia showing positive effects of U.C.T.s on child wasting. On the contrary, the effects of C.C.T. and U.C.T.s on child wasting in Tanzania, Burkina Faso and Uganda were of no effect. In the study in DR Congo, the W.H.Z. mid-upper arm circumference for age Z-score (MAUCZ-age and mid-upper arm circumference for height Z-score (MUACZ-ht) changes were significantly higher than zero for the treatment group compared to the control group's changes in Z-score that were not positive (Grellety et al., 2017).  (Table 4). Food and cash interventions did not reduce the prevalence of stunting. However, food intervention had 9.5 pp impact reduction in stunting over the cash intervention.
Cash and food intervention did not reduce the prevalence of wasting, but cash transfers led to a significant 5.2 pp reduction in severe wasting compared to food group.
No U.C.T. impact but severe underweight prevalence was 3.8% lower in the U.C.T. group than in the food group. Source: Author's compliation from included studies.

Impact of cash transfers on nutritional outcomes
Six studies reported the estimates of the effects of U.C.T. programmes on child nutritional or related outcomes. In Burkina Faso, an evaluation of the seasonal U.C.T. programme on high-nutritional-value foods in young Burkinabe children showed positive results (Tonguet-Papucci et al., 2017). The mixed-effects Poisson regression models used to analyse differences in the dietary diversity scores (D.D.S.s) and the quantity of food taken a day revealed that the seasonal U.C.T. programme improved the diet of children aged 14-27 months. The results of the impact evaluation showed that the cash allowed large numbers of the treatment children to consume milk and dairy products (25% against 7.41%; P = 0.007), flesh foods (26% against 14.8%; P = 0.01), and egg (31.3% against 11.1%; P = 0.003) compared with the control children . In terms of energy and nutrient intake, the impact evaluation did not significantly affect the intake of solid, semisolid, and soft foods given to the treated and controlled children. Nevertheless, the study found that treated children consumed more fat (P < 0.01) than the controlled group and could consume more protein (P = 0.06). Children who were the beneficiaries of the seasonal U.C.T.s consumed more energy from fats (P < 0.01) and fewer carbohydrates (P < 0.01) than the non-beneficiary children . Additionally, treated children had significantly higher consumption of vitamin B-12 (P < 0.001), riboflavin (P < 0.05), and vitamin E (P < 0.05) than controlled children (Tonguet-Papucci et al., 2017).
Results from the DR Congo study showed that there was a significant increase in Households Dietary Diversity Scores (HDDSs), Food Consumption Scores (F.C.S.s) and Dietary Diversity Scores (D.D.S.s) in both treatment and control children (Grellety et al., 2017). However, the increment in the treatment children was very much higher than the control children. The increase measured to between 2.6 times for the index child's dietary diversity to 5.2 times the control children value for the D.D.S. (Grellety et al., 2017). In Malawi, the estimated effect of the SCTP on children's nutritional outcomes showed a negative impact (Abdoulayi et al., 2016). The study's findings revealed that only 4% of the children in treatment households took part in a nutrition programme at baseline (Abdoulayi et al., 2016). This declined to 3% at endline compared to an increase of 5% among children in control households (Abdoulayi et al., 2016).
The cash-based intervention in Somalia implemented to prevent acute malnutrition in children displaced by armed conflict showed mixed results (Grijalva-Eternod et al., 2018). The study's findings revealed that the cash intervention had a significant increase in child D.D.S. of 0.53 (95% CI 0.01; 1.05). In terms of acute malnutrition, the incidence of acute malnutrition was lowered in beneficiary children, but the effect was not significant. Overall, the study found that food security and children's dietary diversity significantly improved in cash transfers. However, the improvements did not correlate positively with the increase in children's nutritional status or with a declined risk of developing acute malnutrition (Grijalva-Eternod et al., 2018).
Using R.C.T. and D.I.D. model applied to survey instruments in Uganda, Gilligan et al. (2013) reported that the U.C.T. programme led to significant gains for the children aged 1-7 in the frequency of consumption of starches, meat and eggs and dairy. The magnitude of these effects was large, representing 66% gains in the incidence of meat and egg consumption and a 100% gain in the frequency of dairy consumption (Gilligan et al., 2013). On whether the U.C.T. impacted child anemia, the study's findings showed that U.C.T. led to weakly significant reductions in the prevalence of anemia and the prevalence of moderate/severe anemia among aged 54-83 months at endline. The impacts on the incidence of any anemia were similar across children aged 54-71 months and children aged 72-83 months at endlinea decrease of around ten percentage points (Gilligan et al., 2013). The impacts of U.C.T. on the prevalence of moderate/severe anemia appeared focused on children aged 54-71 months at endline. A decrease of approximately ten percentage points, cash had minor impacts on the prevalence of moderate/severe among children aged 72-83 months at endline (Gilligan et al., 2013). In Zambia, the study that used R.C.T. with data from households to evaluate the C.G.P. showed that the intervention had a substantial and statistically significant impact on Infant and Young Child Feeding (IYCF), a gain of 22 percentage points (Handa et al., 2014).

Conclusion
Eleven studies were included in this systematic review. The majority of the studies used cluster-RCTs, R.C.T.s, quasi-experimental and mixed methods design. The main findings of this study were of two domains which are anthropometric and nutritional outcomes.
Six studies reported the effects of five U.C.T. programmes and one C.C.T. programme on child height. The results showed no programme effect across countries except in South Africa, where the utmost estimated H.A.Z. increase was higher, approximately 0.45 for children who began treatment earlier in life. The evidence of seven studies' assessments of C.T. programmes on child wasting was mixed. Studies from DR Congo, Lesotho, Malawi, and Zambia showed positive effects of U.C.T.s on child wasting, while studies from Burkina Faso, Tanzania and Uganda showed no programme effects. In terms of underweight, three studies showed a positive impact on children's underweight and three other studies did not find any positive impact.
Five studies reported an estimate of the effect of U.C.T. programmes on child nutritional or related outcomes. The results were positive, negative, and of no effect with the positive effects overshadowing the negative and no effect outcomes. This study found that four U.C.T. programmes and one C.C.T. programme did not have any programme effect on children stunting. Nevertheless, one U.C.T. programme showed a positive impact on children stunting. However, when these children were two-year-old and given the treatment, the effect was still positive but no longer statistically significant (Agüero et al., 2007). Few studies showed mixed results of cash transfer interventions on child's wasting and underweight.
Cash transfers support vulnerable households on food security (de Groot et Table 5, the quality of evidence of children's anthropometric outcomes is shallow, while that of the nutritional outcomes showed a moderate quality of evidence. The overall existing evidence presented in this study finds that C.T. programmes can improve children's anthropometric and nutritional outcomes. Table 5. Summary of outcomes and quality of evidence using GRADE Six of the ten studies that looked at children's health showed positive outcomes. However, two studies showed adverse outcomes on children's morbidity. The research presents a remarkably satisfactory indication of the possible effects. The possibility that the effect will be considerably different is low.

Moderate ⊕⊕⊕⊖
The research presents a remarkably satisfactory indication of the possible effects. The possibility that the effect will be considerably different is moderate.

Low ⊕⊕⊖⊖
The research presents a remarkably satisfactory indication of the possible effects. Nevertheless, the possibility that the effect will be considerably different is high.

Very low ⊕⊖⊖⊖
The research presents a remarkably satisfactory indication of the possible effects. However, the possibility that the effect will be considerably different is very high.
Note: considerably different = a large enough difference that it might affect a decision. To the best of my knowledge, this is the first systematic review that systematically assessed national governments, pilot interventions and humanitarian C.T. programmes on children's nutritional outcomes in sub-Saharan Africa. According to the findings of this review, C.T. programmes have the potential of improving children's nutritional status. Future experimental research is needed to support the current evidence of this study. The future studies should focus more on children in rural areas where poverty is more pronounced than the urban areas.