Small-quantity lipid-based nutrient supplements for children age 6-24 months: a systematic review and individual participant data meta-analysis of effects on developmental outcomes and effect modifiers

Abstract

Background: Small-quantity (SQ) lipid-based nutrient supplements (LNSs) provide many nutrients needed for brain development.

Objectives: We aimed to generate pooled estimates of the effect of SQ-LNSs on developmental outcomes (language, social-emotional, motor, and executive function), and to identify study-level and individual-level modifiers of these effects.

Methods: We conducted a 2-stage meta-analysis of individual participant data from 14 intervention against control group comparisons in 13 randomized trials of SQ-LNSs provided to children age 6-24 mo (total n = 30,024).

Results: In 11-13 intervention against control group comparisons (n = 23,588-24,561), SQ-LNSs increased mean language (mean difference: 0.07 SD; 95% CI: 0.04, 0.10 SD), social-emotional (0.08; 0.05, 0.11 SD), and motor scores (0.08; 95% CI: 0.05, 0.11 SD) and reduced the prevalence of children in the lowest decile of these scores by 16% (prevalence ratio: 0.84; 95% CI: 0.76, 0.92), 19% (0.81; 95% CI: 0.74, 0.89), and 16% (0.84; 95% CI: 0.76, 0.92), respectively. SQ-LNSs also increased the prevalence of children walking without support at 12 mo by 9% (1.09; 95% CI: 1.05, 1.14). Effects of SQ-LNSs on language, social-emotional, and motor outcomes were larger among study populations with a higher stunting burden (≥35%) (mean difference: 0.11-0.13 SD; 8-9 comparisons). At the individual level, greater effects of SQ-LNSs were found on language among children who were acutely malnourished (mean difference: 0.31) at baseline; on language (0.12), motor (0.11), and executive function (0.06) among children in households with lower socioeconomic status; and on motor development among later-born children (0.11), children of older mothers (0.10), and children of mothers with lower education (0.11).

Conclusions: Child SQ-LNSs can be expected to result in modest developmental gains, which would be analogous to 1-1.5 IQ points on an IQ test, particularly in populations with a high child stunting burden. Certain groups of children who experience higher-risk environments have greater potential to benefit from SQ-LNSs in developmental outcomes.This trial was registered at www.crd.york.ac.uk/PROSPERO as CRD42020159971.

Keywords: child undernutrition; complementary feeding; executive function; language development; motor development; nutrient supplements; social-emotional development.

FIGURE 1

Study inclusion flow diagram. IPD, individual participant data; LNS, lipid-based nutrient supplement; RCT, randomized controlled trial; SQ, small-quantity.

FIGURE 2

Forest plot of the effect of small-quantity LNSs on the prevalence of children in the lowest decile of language scores. Individual study estimates were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled estimates were generated using inverse-variance weighting in both fixed- and random-effects models. Individual trial estimates for the SHINE trial are split by comparison in the figure to reflect the crossover design. For calculating the pooled estimates, the trial was analyzed with LNS intervention arms combined and non-LNS intervention arms combined. ASQ:I, Ages and Stages Questionnaire Inventory; BSID, Bayley Scales of Infant Development; CDI, MacArthur-Bates Communicative Development Inventory; DMC, Developmental Milestones Checklist; EASQ, Extended Ages and Stages Questionnaire; LNS, lipid-based nutrient supplement; PR, prevalence ratio; SOC, standard of care; WSH, water, sanitation, and hygiene intervention.

FIGURE 3

Forest plot of the effect of small-quantity LNSs on the prevalence of children in the lowest decile of social-emotional scores. Individual study estimates were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled estimates were generated using inverse-variance weighting in both fixed- and random-effects models. Individual trial estimates for the SHINE trial are split by comparison in the figure to reflect the crossover design. For calculating the pooled estimates, the trial was analyzed with LNS intervention arms combined and non-LNS intervention arms combined. ASQ:I, Ages and Stages Questionnaire Inventory; DMC, Developmental Milestones Checklist; EASQ, Extended Ages and Stages Questionnaire; LNS, lipid-based nutrient supplement; MDAT, Malawi Developmental Assessment Tool; PR, prevalence ratio; PSED, Profile of Social and Emotional Development; SOC, standard of care; WSH, water, sanitation, and hygiene intervention.

FIGURE 4

Forest plot of the effect of small-quantity LNSs on the prevalence of children in the lowest decile of motor scores. Individual study estimates were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled estimates were generated using inverse-variance weighting in both fixed- and random-effects models. Individual trial estimates for the SHINE trial are split by comparison in the figure to reflect the crossover design. For calculating the pooled estimates, the trial was analyzed with LNS intervention arms combined and non-LNS intervention arms combined. ASQ:I, Ages and Stages Questionnaire Inventory; BSID, Bayley Scales of Infant Development; DMC, Developmental Milestones Checklist; EASQ, Extended Ages and Stages Questionnaire; KDI, Kilifi Developmental Inventory; LNS, lipid-based nutrient supplement; MDAT, Malawi Developmental Assessment Tool; PR, prevalence ratio; SOC, standard of care; WSH, water, sanitation, and hygiene intervention.

FIGURE 5

Forest plot of the effect of SQ-LNSs on the prevalence of children walking without support at age 12 mo. Individual study estimates were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled estimates were generated using inverse-variance weighting in both fixed- and random-effects models. LNS, lipid-based nutrient supplement; PR, prevalence ratio.

FIGURE 6

Pooled effects of SQ-LNSs on all continuous developmental outcomes stratified by study-level stunting burden of children at age 18 mo in control groups. Individual study estimates for interaction effect were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled subgroup estimates and statistical testing of the pooled interaction term were generated using inverse-variance weighting random effects. LNS, lipid-based nutrient supplement; P-diff, P value for the difference in effects of small-quantity lipid-based nutrient supplements between the 2 levels of the effect modifier.

FIGURE 7

Pooled effects of SQ-LNSs on 6 primary developmental outcomes stratified by individual-level household SES. Individual study estimates for interaction effect were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled subgroup estimates and statistical testing of the pooled interaction term were generated using inverse-variance weighting fixed effects. **P-Int < 0.01, *P-Int < 0.1 P-int, P value for the interaction indicating the difference in effects of small-quantity lipid-based nutrient supplements between the 2 levels of the effect modifier in fixed-effects models. LNS, lipid-based nutrient supplement; SES, socioeconomic status.

FIGURE 8

Pooled effects of SQ-LNSs on 6 primary developmental outcomes stratified by individual-level child baseline acute malnutrition. Individual study estimates for interaction effect were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled subgroup estimates and statistical testing of the pooled interaction term were generated using inverse-variance weighting fixed effects. **P-Int < 0.05 P-int, P value for the interaction indicating the difference in effects of small-quantity lipid-based nutrient supplements between the 2 levels of the effect modifier in fixed-effects models. LNS, lipid-based nutrient supplement.

FIGURE 9

Pooled effects of SQ-LNSs on 6 primary developmental outcomes stratified by individual-level child baseline stunting. Individual study estimates for interaction effect were generated from log-binomial regression controlling for baseline measure when available and with clustered observations using robust SEs for cluster-randomized trials. Pooled subgroup estimates and statistical testing of the pooled interaction term were generated using inverse-variance weighting fixed effects. LNS, lipid-based nutrient supplement.