1. Objectives

To evaluate the safety and effectiveness of different types of specially formulated foods for children with moderate acute malnutrition in low- and middle-income countries, and to assess whether foods complying or not complying with specific nutritional compositions, such as the WHO technical specifications, are safe and effective

2. How studies were identified

The following databases were searched in October 2012:

• CENTRAL (The Cochrane Library 2012, Issue 10)
• MEDLINE
• EMBASE
• LILACS
• CINAHL
• metaRegister of Controlled Trials (mRCT)
• BIBLIOMAP
• POPLINE
• ZETOC
• ClinicalTrials.gov
• United Nations System Standing Committee on Nutrition (UNSSCN): Moderate Malnutrition e-platform
• iLiNS project website (www.ilins.org)

Reference lists, conference proceedings and journals were also searched and the authors directly contacted relevant researchers, organisations and international groups

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials, cluster-randomized controlled trials, quasi-randomized trials, non-randomized controlled clinical trials, controlled before-and-after studies, interrupted time series studies

3.2 Study participants

Children aged six to 60 months in low- and middle-income countries with moderate acute malnutrition, treated either in hospital, a community clinic, or at home

(Moderate acute malnutrition is defined as weight-for-height between -3 and -2 standard deviations from the mean or between 70% and 80% of the mean, or mid-upper arm circumference (MUAC) between 115 and 125 mm, and no oedema)

(Studies of children with special requirements, such as those with cancer, were excluded)

3.3 Interventions

In comparison to treatment as usual or the provision of an alternative food, interventions included any type of food used for children with moderate acute malnutrition. Potential interventions included: i) improved adequacy of local diet, such as germinating, soaking, fermenting or malting foods, or preparing food at home with a given recipe (interventions using nutritional counselling alone were excluded); ii) lipid-based nutrient supplements, which are usually ready-to-use therapeutic foods with a high energy density; iii) blended food supplements, such as corn-soy blends, which are generally cooked at home as porridges; iv) complementary food supplements of any type, which can be mixed with or consumed in addition to the usual diet, providing only part of the total daily caloric needs (e.g., 270 kcal/day)

(Studies with co-interventions were only eligible if administered to both experimental and control arms)

3.4 Primary outcomes

• Recovered
• Not recovered
• Progression to severe acute malnutrition
• Died
• Defaulted (i.e., dropped out of the programme)
• Weight gain
• Weight-for-height Z-score (WHZ)
• MUAC
• Any adverse effect, including predisposition to obesity (assessed by measuring rapid weight gain and increase in fat mass compared to lean body mass) and diarrhoea

Secondary outcomes included nutritional adequacy of the diet, lean body mass increase, height gain, height-for-age Z-score (HAZ), and coverage of the population

4. Main results

4.1 Included studies

Eight randomized controlled trials, enrolling 10,037 children, were included in this review

• Children ranged from six months to five years of age, with two studies enrolling children under two years, and one study enrolling children under three years. Trial duration ranged from eight to 16 weeks, and two trials had an extended post-intervention follow-up of six and 12 months
• Growth reference standards varied between trials, but the definition of moderate acute malnutrition in all trials included weight-for-height, with five trials also using MUAC, either concomitantly (4 trials) or as an alternative criterion (1 trial)
• Five trials were factorial, providing data on more than two comparisons. Two trials assessed specially formulated food versus standard care, seven trials evaluated lipid-based nutrient supplements versus blended foods, and two trials assessed foods at complementary dosage. Three trials assessed lipid-based nutrient supplements versus a specific blended food, “Corn-Soy Blend plus plus” (CSB++), and one trial evaluated CSB++ versus an alternative blended food
• The definition of recovery varied both for the cut-off (WHZ>-1.5, WHZ>-2 or WHZ>85% of median Z-score) and timeline

4.2 Study settings

• Bangladesh, Burkina Faso, Ethiopia, Malawi (2 trials), Mali, Niger, and Sierra Leone
• Children were treated as outpatients in all trials
• Seven trials were conducted in rural settings and one trial was conducted in an urban setting (Dhaka, Bangladesh)

4.3 Study settings

How the data were analysed
Four main comparisons were performed: i) specially formulated complementary foods versus standard care; ii) lipid-based nutrient supplements versus blended foods; iii) lipid-based nutrient supplements versus specific types of blended foods; and iv) the nutrient composition of specially formulated foods in comparison to WHO recommendations. Trials were pooled in random-effects meta-analyses by type of intervention. The cluster-randomized trials adjusted for the effect of design and therefore generic inverse variance methods were not required. For dichotomous data, risk ratios (RR) and 95% confidence intervals (CI) were calculated, and for continuous data, mean differences (MD) and 95% CI were calculated. Subgroup analyses were planned to explore for sources of heterogeneity due to differences between studies and the populations studied, such as the prevalence of stunting and wasting, level of breastfeeding, level of food security, and the prevalence of HIV and TB. However, due to limited data, subgroup analysis was performed for only one comparison, namely lipid-based nutrient supplements full dose versus blended food full dose, on the primary outcome “recovered”. Sensitivity analyses were also conducted by excluding studies at high risk of bias, using a fixed-effects model, and altering imputation methods for missing data.

The energy density of each intervention food was compared to the WHO recommendation of a minimum of 0.8 kcal/g for food when ready to be consumed. Individual macronutrient and micronutrient contents per 1000 kcal of intervention foods were compared to the WHO Technical specifications for the management of moderate acute malnutrition in children aged 6 to 59 months. Most studies provided nutritional composition data, but where data were missing for selected ingredients, the USDA National Nutrient Database for Standard was used.

Results
Specially formulated complementary foods versus standard care
Recovered
Compared to nutritional counselling and standard medical care, the additional provision of specially formulated complementary foods increased the likelihood of recovery from moderate acute malnutrition by 29% (RR 1.29, 95% CI [1.20 to 1.38], p<0.00001; 2 trials/2152 children).

Not recovered, progression to severe acute malnutrition, died, defaulted
One trial reported on the outcomes of failing to recover, progression to severe acute malnutrition, and death, in which no significant differences between those given complementary foods and those given counselling alone were found (not recovered: RR 0.97, 95% CI [0.74 to 1.27], p=0.82; severe acute malnutrition: RR 0.78, 95% CI [0.59 to 1.03], p=0.083; death: RR 0.44, 95% CI [0.14 to 1.36], p=0.16; 1 trial/1974 children). Seventy percent (70%) fewer children dropped out of the complementary food arms in comparison to the counselling alone arm in one trial of 1974 children (RR 0.30, 95% CI [0.22 to 0.39], p<0.00001).

Anthropometry
Weight gain was statistically significantly higher in the group receiving complementary food, with an MD of 0.18 kg (95% CI [0.04 to 0.33], p=0.013; 1 trial/188 children). Final WHZ and WHZ gain were both statistically significantly greater with complementary food intervention (MD 0.20 Z-score, 95% CI [0.03 to 0.37], p=0.02; 2 trials/1546 children and MD 0.28 Z-score, 95% CI [0.06 to 0.49], p=0.01; 1 trial/178 children, respectively). No differences were observed between treatment groups for the outcome MUAC (MD 0.62 mm, 95% CI [-1.38 to 2.61], p=0.55; 1 trial/178 children). Height gain and final HAZ were not statistically different between groups.

Adverse effects
Trials did not report on adverse effects.

Lipid-based nutrient supplements versus blended foods
Recovered
Overall, lipid-based nutrient supplements were superior to blended foods for the recovery from moderate acute malnutrition by 8% (RR 1.08, 95% CI [1.04 to 1.13], p=0.0005; 7 trials/8861 children). At full dose, lipid based supplements remained statistically significantly superior (RR 1.10, 95% CI [1.04 to 1.16], p=0.00065; 5 trials/6367 children). Lipid-based supplements at complementary dose versus blended food at full dose also resulted in a statistically significantly greater number of recovered children (RR 1.10, 95% [1.01 to 1.20], p=0.03; 1 trial/1125 children), but no difference was found between lipid-based supplements at full dose and blended food at complementary dose (RR 1.00, 95% CI [0.94 to 1.06], p=0.90; 1 trial/1369 children). In subgroup analysis by country level of food security, a lower effect of lipid-based nutrient supplements at full dose versus blended foods at full dose was observed in countries with moderate food insecurity (RR 0.99, 95% CI [0.94 to 1.03], p=0.50; 1 trial/570 children) compared to countries with high food insecurity (RR 1.24, 95% CI [1.15 to 1.32], p<0.00001; 2 trials/1723 children). In other subgroup analyses by the prevalence of wasting in the country (medium versus low), and the prevalence of stunting in the country (high versus moderate/low), no differences in effect were noted. Sensitivity analyses also did not affect findings materially.

Not recovered, progression to severe acute malnutrition, died, defaulted
Lipid-based supplements reduced the risk of not recovering compared with blended foods overall (RR 0.69, 95% CI [0.54 to 0.87], p=0.0023; 5 trials/7031 children), and when at full dose (75 kcal/kg/day) compared to blended foods at full dose (RR 0.53, 95% CI [0.40 to 0.69], p<0.00001; 5 trials/4537 children). No difference was observed between groups in the risk of progression to severe acute malnutrition (RR 0.88, 95% CI [0.74 to 1.04], p=0.12; 5 trials/7031 children) or death (RR 0.94, 95% CI [0.55 to 1.58], p=0.81; 7 trials/8861 children). Defaulting did not differ between groups overall (RR 1.23, 95% CI [0.80 to 1.88], p=0.23; 6 trials/7601 children), but when both lipid-based supplements and blended foods were compared at complementary doses, lipid-based supplements increased the risk of defaulting (RR 1.69, 95% CI [1.07 to 2.69], p=0.025; 1 trial/1369 children).

Anthropometry
Overall, lipid-based supplements resulted in a greater daily weight gain than blended foods (MD 0.53 g/kg/day, 95% CI [0.14 to 0.93], p=0.0083; 4 trials/4241 children). This significant difference persisted when compared at full dose to full dose blended food (MD 0.69 g/kg/day, 95% CI [0.31 to 1.06], p=0.00036; 3 trials/3223 children), but when both foods at complementary doses were compared, there was no statistically significant difference between treatments (MD -0.10 g/kg/day, 95% CI [-0.47 to 0.27], p=0.59; 1 trial/1018 children). Final WHZ and WHZ gain were significantly greater among children receiving lipid-based nutrient supplements compared to those receiving blended foods (MD 0.12 Z-score, 95% CI [0.05 to 0.18], p=0.00054; 4 trials/6009 children, and MD 0.13 Z-score, 95% CI [0.03 to 0.22], p=0.009; 2 trials/3631 children, respectively). Gain in MUAC was statistically significantly higher with lipid-based supplements by 0.04 mm/day (95% CI [0.02 to 0.06], p=0.0004; 4 trials/4568 children), however height gain and HAZ were not different between groups. Long-term 12 month follow up of children in a Malawian trial demonstrated that those treated with soy/whey lipid-based nutrient supplements were significantly more likely to remain well-nourished than those treated with soy lipid-based nutrient supplements or CSB++ (p=0.02).

Adverse effects
The risk of vomiting was greater among those treated with lipid-based supplements compared to those treated with blended foods in one trial of 2712 children (RR 1.43, 95% CI [1.11 to 1.85], p=0.006); however, there was no difference observed in the risk of diarrhoea.

Specific comparisons between lipid-based nutrient supplements and types of blended foods
Lipid-based nutrient supplements compared to CSB++
No statistically significant differences were found between lipid-based nutrient supplements and CSB++, except for a greater risk of vomiting with lipid-based supplements (RR 1.43, 95% CI [1.11 to 1.85], p=0.006; 1 trial/1712 children).

Lipid-based nutrient supplements compared to CSB pre-mix
Lipid-based nutrient supplements did not significantly increase the likelihood of recovery compared to CSB pre-mix (RR 1.09, 95% CI [0.96 to 1.25], p=0.17; 3 trials/2158 children). However, failure to recover was significantly decreased with lipid-based nutrient supplements (RR 0.79, 95% CI [0.64 to 0.97], p=0.022; 2 trials/1588 children). No significant differences were observed between treatment groups for the outcomes progression to severe acute malnutrition, died or defaulted. Lipid-based nutrient supplements significantly improved daily weight gain and MUAC gain relative to CSB pre-mix (MD 1.09 g/kg/day, 95% CI [0.72 to 1.47], p<0.00001; 2 trials/852 children, and MD 0.04 mm/day, 95% CI [0.01 to 0.06], p=0.0016; 2 trials/838 children, respectively). No differences in WHZ or WHZ gain were noted.

Comparison among different types of blended foods
CSB++ was compared to Misola and to Home food (a mixed home ration of blended foods) in one trial of 925 children, in which no differences in the number of recovered children, deaths or in weight gain were found.

Comparison with the WHO technical specifications
Corn-soy blended foods did not comply with the 2012 WHO Technical specifications recommending a minimum energy density of 0.8 kcal/g. No food completely complied with the 2012 WHO Technical specifications for individual macro- and micronutrient contents, thus comparisons between foods that met the nutrient composition standards versus those that did not could not be performed.

5. Additional author observations*

The overall risk of bias in included trials was low, with only two studies considered to be at high risk of bias due to high rates of attrition. However, this review was limited by the low number of studies assessing foods for treating children with moderate acute malnutrition, and it is possible that different types of food interventions, such as improved adequacy of local diet, will prove to be beneficial in the future.

The provision of specially formulated complementary foods improved recovery and gains in weight in children with moderate acute malnutrition. While lipid-based nutrient supplements did not reduce deaths or progression to severe acute malnutrition in comparison to blended foods, improvements were seen in the number recovered, not recovered and in some anthropometric measures, although CSB++ may be equally as effective. Regarding adverse effects, vomiting was increased among those treated with lipid-based nutrient supplements.

Future research efforts should evaluate measures to improve the adequacy of the local diet, compare different types of blended foods, include outcomes such as lean body mass gain and long-term health, include more diverse populations, such as children in Latin America and Asia and children with known TB/HIV status, include integrative treatments, such as de-worming, and conduct systematic evaluations of adverse effects and cost-benefit analyses.