A home-based nutritional intervention for preschool children: Evidence from India

A community-based cluster randomised controlled trial design was used for this study in rural southwestern India. All 253 participants were aged 3–5 years, had moderate (weight-for-age <-2 to –3 standard deviations) or severe underweight (<-3 standard deviations), and mothers who were registered1 centres. Control and intervention groups were randomised at a 1:1 ratio (see Box 1). Measurements were taken every month throughout the 12-month intervention period.

Both groups gained weight at a steady rate across the 12-month period – as expected – but the intervention groups’ growth trajectory was steeper. Average weights within the intervention group remained higher and accelerated away from the control group at each measurement, concluding with an average weight difference of 0.82 kilograms between the groups (p<0.001) at the end of the study. In the intervention group, the percentage of children moderately underweight reduced from 41.5% to 24% in 12 months, while severe underweight reduced from 8.6% to 3.16%. Only minimal changes were observed in the control group.

Cluster randomisation was valid for this setting. This method usually requires a larger sample size to see an effect – compared to individual randomisation, which is the gold standard – but in this case an appropriate sample size calculation was applied that also accounted for attrition. The researchers then enrolled slightly more children than needed, even though there was no dropout and zero mortality by study conclusion, so this study was appropriately powered.

The data are compelling and show a clear benefit of this intervention within this population. This trial would need to be expanded to a broader group to infer results about a larger population, but the methodology was robust, and the effect size was large – allowing us to be confident in these findings.

Although both arms of the study had comparable weights at baseline and most other characteristics did not differ significantly, more children in the intervention group were below the poverty line (9.4%) than those in the control group (0.8%). Conversely, there were more middle-income households within the control group (23.8%) than the intervention group (15.7%). As weights were comparable, this difference may be insignificant. However, it is plausible that the lower socio-economic status group adhered more strictly to the intervention as they placed more value in it. By contrast, higher-income households may have seen this as less of a priority. This could affect our interpretation until further results from a wider study group are published.