A cluster RCT to measure the effectiveness of cash-based interventions on nutrition status, Sindh Province, Pakistan

Location: Pakistan

What we know:  Evidence of the impact of cash-based interventions on nutrition outcomes is limited.

What this article adds: A four arm (standard cash (SC), double cash (DC), fresh food voucher (FFV), control) parallel longitudinal cluster randomised controlled trial was implemented over 6 months by Against Against Hunger and ENN. Primary outcomes were were weight-for-height z scores (WHZ) <-2 and mean WHZ in children under 5 years.  Preliminary results found a significant decrease in risk of being wasted with DC only and significant improvements in mean WHZ for DC and FFV.  All three interventions saw a significant decrease in both stunting and mean height for age. There was a significant decrease in mean haemoglobin (Hb) concentration for children and mothers in the FFV, and for mothers only in the SC arm. A pathway causal analysis is underway that will greatly aid interpretation, with results available mid-2017.

Background

The Research into Food Assistance for Nutrition Impact (REFANI) Consortium comprises two operational partners; Action Against Hunger (lead agency) and Concern Worldwide, and 2 academic/research partners; ENN and University College London (UCL). REFANI is a 3-year research project funded by UK aid, and co-financed through funding from the European Commission (EU & ECHO). The overarching aim of REFANI is to increase the evidence base of cash-based interventions (CBIs) on nutrition outcomes in humanitarian settings by addressing a number of evidence gaps. The use of CBIs among humanitarian agencies to prevent wasting in children is increasing but questions remain as to how to incorporate CBIs into emergency programmes to maximise their success in terms of improved nutrition outcomes.

The REFANI Pakistan study is a collaboration between Action Against Hunger and ENN, set in Dadu district, Sindh province. Dadu District is largely agrarian, dependent on crop production, livestock keeping, and agriculture labour. The majority of the population is highly vulnerable to shocks, especially the poorest households, and there is a lack of alternative income sources which are further constrained by lack of opportunities. Dadu District experiences frequent flooding and droughts, and high temperatures (above 45°C).

Methods

We undertook a four arm parallel longitudinal cluster randomised controlled trial (cCRT) – registered trial number ISRCTN10761532. The protocol has been published.¹ Three CBIs were implemented: two unconditional cash transfers – a ‘standard cash’ (SC) amount of 1500 Pakistan Rupees (PKR) and a ‘double’ cash (DC) amount of 3000 PKR, and one fresh food voucher (FFV) with a value of 1500 PKR, which could be exchanged for specified fresh foods (fruits, vegetables and meat). A fourth arm acted as the control group and received no additional intervention beyond the basic activities implemented by Action Against Hunger that were provided to all groups.

The SC was set to equal the amount disbursed by Pakistan’s national safety net programme – the Benazir Income Support Programme (BISP). The cash components were disbursed on a monthly basis either by mobile banks that travelled to a central location for some of the participating villages or through central banks that served a number of villages. The FFVs were disbursed to participating households at village level. All three interventions were delivered with verbal messages that children should benefit from the transfers.

The interventions were implemented over six consecutive months (July to December 2015) and targeted to mothers from poor/very poor households with a child 6-48 months at baseline. The implementation and the use of the CBIs were monitored both quantitatively and qualitatively through monthly questionnaires or three monthly focus group discussions and key informant interviews.

The main research question assessed the effectiveness of different CBIs at reducing the risk of undernutrition during the lean season. The primary outcomes were weight-for-height z scores (WHZ) <-2 and mean WHZ in children under 5 years. The study also encompassed a mixed-methods process evaluation to help interpret the results and a costs and cost-effectiveness analysis (results not presented here).

Results

The results presented here are a summary of the short term impact of CBIs on nutrition outcomes. The full analysis of both short and medium-term term impacts is forthcoming. The group with the higher amount of cash (DC) saw a significant decrease in risk of being wasted (WHZ <-2 and <-3) compared to the control group. There were no significant differences in risk of being wasted for either SC or FFV arms. Both the DC and FFV arms saw significant improvements in mean WHZ compared to the control arm. All three interventions saw a significant decrease in both stunting (height-for-age z-score (HAZ) <-2 and <-3) and mean HAZ compared to the control group. In the FFV arm, there was a significant decrease in mean haemoglobin (Hb) concentration for children and mothers, and for mothers only in the SC arm.

Lessons learned

The results have identified a number of questions that need yet to be answered and for now need careful interpretation. In terms of risk of being wasted, we need to have a better understanding of why children in the DC arm were significantly less wasted. This will be attempted through a pathway analysis whereby different pathways in the causal framework will be quantified. It was not possible to establish what the minimum level of cash required to have had a significant effect. We can only say that this threshold falls somewhere between the amounts allocated in the SC and the DC interventions.

Regarding mean WHZ, it appears that children in the FFV arm were getting fatter but not taller, especially if interpreted with the lack of improvement in being wasted. As well as this, the Hb levels of children and mothers in the FFV arm were significantly lower compared to the control group. We had hypothesised that the FFV would impact growth and micronutrient status through increasing dietary diversity. However, whilst all three arms showed a significant improvement in mother and child dietary diversity, this improvement was lowest in the FFV arm (highest in the DC arm). These results suggest that something unplanned was occurring in the FFV arm. It is possible that the vouchers themselves were too restricted, being dependent on what the vendors stocked, such as chicken being the only available meat. There were also anecdotal reports regarding vendors over-charging for food items redeemed against the vouchers as a way to cover their own administration fees in recovering the voucher costs. In this respect, the actual transfer amount given may have been lower.

Another question that needs to be addressed was the lack of improvement in Hb status in light of the improvements in ponderal growth and prevalence of being wasted as seen in the DC arm. The explanation for this will also be attempted through a pathway analysis.

The study setting presented a number of difficulties affecting data collection. The baseline survey took longer than expected since recruitment of female enumerators was difficult and the data collection coincided with Ramadan and reduced working hours. Added to this, temperatures reached 52^oC, which not only affected research team working ability but also had an effect on the haemocues.

Conclusion

The results illustrate the impacts of different CBIs on nutrition status. However, the theory of change for ‘how’ CBIs may influence nutrition outcomes in children is complex. In the REFANI Pakistan study, we illustrate that it is not a straight-forward task to simply interpret impact results as working or not; understanding the pathways and processes through which CBIs are implemented is essential to understand how best to implement them.  Such analyses will be completed mid-2017.

For more information, contact: Bridget Fenn

Footnotes