Micronutrient powders to combat anaemia in young children: do they work?

Point-of-use fortification of foods is recommended by the World Health Organization (WHO) [1] as an alternative to mitigate or overcome the constraints associated with supplementation and industrial food fortification. In this new approach, powders containing a mixture of vitamins and minerals are supplied as small, single-serving packets, the contents of which can be mixed into semi-solid food prior to consumption.

In a recent placebo-controlled trial among Kenyan children aged 12–36 months, we found no evidence that daily point-of-use fortification with either 3 mg iron as NaFeEDTA or 12.5 mg iron as ferrous fumarate was efficacious in improving haemoglobin concentration or plasma iron markers [2]. As discussed in a commentary [3], the decline in anaemia prevalence over time that we observed in all intervention groups may have been due to premedication with anti-malarial and anti-helminth medication at baseline, or due to the vitamin A or zinc content in the fortification powders. However, attribution of such time trends is impossible since the observed effects may also have been due to various other factors, including seasonal variation in nutrient status or infections, age changes or regression to the mean. Thus, these temporal changes in anaemia prevalence cannot be per se interpreted as evidence to support the use of micronutrient powders in an effective intervention strategy. Reviews of the impact of multi-micronutrient powders on indicators other than iron deficiency or anaemia have been inconclusive [4, 5], and unfortunately there is little evidence from randomised trials to show that point-of-use fortification leads to improved vitamin A or zinc status [6].

The effect of an intervention for any person is best defined as the difference between the outcomes as a result of receiving the intervention and those that would have resulted had the intervention been denied [7]. Under such a counterfactual view of causality, an effect can be operationally estimated as the difference in endpoints between groups concurrently randomised to verum (i.e. to the active substance under investigation) or control. In our trial, compared to placebo, point-of-care fortification with NaFeEDTA at the end of intervention led to a reduction in the prevalence of iron deficiency by 20.1% (from 44.6% to 24.5%). However, effects on iron deficiency should be interpreted with caution since a small group difference in the distribution of a continuous outcome can misleadingly result in a relatively large difference in prevalence of the dichotomised outcome, whether expressed as a relative or absolute difference. Thus, point-of-use fortification with NaFeEDTA improved geometric mean plasma ferritin concentrations by only 4 μg/L (from 29.7 to 33.7 μg/L), which may seem much less impressive than suggested by the observed 20.1% decrease in prevalence of iron deficiency, an outcome variable that was derived by dichotomising plasma ferritin concentration.

Similar issues arise in the interpretation of effect estimates of the meta-analyses that formed the basis of the WHO guidelines, which showed absolute effects on haemoglobin and ferritin concentrations that may seem much less impressive than the corresponding relative reductions in anaemia and iron deficiency [1]. For example, in the WHO meta-analysis, in children aged 2–12 years, point-of-care fortification with multiple micronutrient powders increased haemoglobin concentration by 3.4 g/L (95% CI 0.9–5.8 g/L) as compared to a reduction in anaemia risk by 34% (95% CI 12–51%) [1]. In trials and meta-analyses, priority should be given to effects on continuous outcomes. Relative effects on dichotomised outcomes can be useful to allow extrapolation of trial results over time and in different settings, but they should be given secondary importance, and extrapolated results should be re-converted to the continuous outcome to appreciate the true public health gains.