br Conclusions br Conflict of interest br Acknowledgments Th

Conclusions
Conflict of interest
Acknowledgments This work was funded by the Department of Veteran Affairs Clinical Science Research and Development Office, Grant CSRD 10CX000691-01 (PI Charlene Weir).
Introduction Multiple micronutrient powder (MNP) has been described as an effective and scalable way to combat childhood and infant anaemia. The WHO, for instance, recommends histone methyltransferase inhibitor wide daily MNP supplementation in settings where anaemia has a prevalence of 20% or greater amongst under-twos. Anaemia affects nearly half (46.6%) of under-twos in Colombia’s lowest socio-economic strata. This rises to nearly 70% of 6–12 month olds (own estimates using ENSIN 2010 data (Instituto Colombiano de Bienestar Familiar, 2015)). Colombia in general, and its youngest and most disadvantaged children in particular meet the WHO’s criteria for MNP adoption. In an earlier paper (Attanasio et al., 2014) we evaluated a daily MNP supplementation intervention targeted at children aged 12–24 months from the poorest households in Colombian small towns. We found no effect on childhood anaemia. In this paper, we investigate why MNP supplementation had no impact.
The MNP intervention The MNP we used was ‘Sprinkles’ developed by the Sprinkles Global Health Initiative. Each 1g sachet, to be sprinkled over a child’s meal daily, contains 12.5mg iron (ferrous fumerate), 5mg zinc, 300µg vitamin A, 30mg vitamin C and 160µg folic acid. This safely meets the average child’s daily requirement of iron, Vitamin A, Vitamin C and Folic Acid and the majority of their zinc requirement (World Health Organization, 2005). Our intervention supplemented the children daily for 18 months. Our intervention and evaluation used the structures of Familias en Acción (FeA), Colombia’s conditional cash transfer programme which targets the lowest socio-economic strata (SISBEN level one) which corresponds to roughly the poorest 16% of the population. The MNP intervention was delivered by home visitors who were local women and in 63% of cases were Madre Líderes, locally elected female representatives of FeA beneficiaries. Home visitors were paid $15,000 Colombian Pesos (US$40) a month. Before the intervention began, home visitors in MNP intervention areas received a five hour training session, led by a mentor.
Methods
Results There is evidence of differential attrition and missing haemoglobin measurements between treatment groups. A significantly higher proportion of the PS group (12.3%) and MNP group (10.2%) than of the control group (5.6%) were either completely lost from the study at follow-up or were missing haemoglobin measurements (p<0.05). Nonetheless, attrition and missing haemoglobin are uncorrelated with baseline haemoglobin and other key characteristics. Table 1 shows that our analysis sample were well balanced in terms of key baseline characteristics across treatment and control. Therefore, we do not expect differential attrition to bias our estimates.
Discussion The results presented in Sections “Validity of study design” to “No dietary substitution” suggest that for Colombian children aged 12–24 months from poor backgrounds and living in small towns, MNP supplementation is ineffective when delivered in real world conditions and these analyses suggest that Saltatory replication was due to a lack of efficacy. This finding holds even for children we might have expected to benefit the most – younger children and those who were anaemic at baseline. Key to understanding this, when similar interventions have been successful, is the progression of anaemia, and its causes, with age in our population. Nearly 70% of the poorest Colombian children aged 6–12 months are anaemic (Fig. 3), just 10 percentage points below rates in South and Southeast Asia and Sub-Saharan Africa. However, while anaemia appears very persistent in the Asian and African countries, in Colombia, the rate falls very quickly, almost halving in seven months and falling to 20%, a rate comparable to the USA, by 30 months. This trend appears similar to, but more pronounced than, other countries in South America; and is very similar in the evaluation sample we have analysed.