Types of studies.

Randomized controlled trials, randomized at the level of the individual or cluster, were included. In anticipation of non-availability of sufficient trials of this nature, we planned to include data from the following additional study designs: (i) Non- randomized trials at the level of the individual or cluster. To be eligible, such trials should have a concurrent comparison group (as defined later), preferably with adjustment for baseline characteristics and confounders; (ii) Controlled before-after studies where allocation to the different comparison groups was not made by the investigators. Outcomes of interest should have been measured in both intervention and control groups before the lipid-based nutrient supplements intervention was introduced, and again after the intervention had been introduced.

We planned to include cluster-randomized trials, non-randomized cluster trials, and controlled before-after studies with at least two intervention sites and two control sites.

Eligible studies were included irrespective of the date, language or publication status.

Types of participants.

Infants and children between the ages of 6 and 59 months with moderate acute malnutrition, defined as a weight-for-height between ≥ -3 SD (Z score) and < -2 SD (Z score) of WHO standard or a mid-upper arm circumference ≥ 115 mm and < 125 mm. The studies were included irrespective of setting or disease status of the study population.

Types of interventions.

Intervention: Lipid-based nutrient supplementation, as defined by the authors was considered the intervention. The supplements could have been produced centrally or locally and prepared with either a commercial or non-commercial intent. We included studies using any variations in the calories provided, milk content, type of macronutrients like type of oil (e.g. omega 3 fatty acids) or protein quality, maximum duration of supplementation and the definition of recovery.

Comparison: The comparison groups comprised either of the following: (i) Lipid-based nutrient supplements vs. specially formulated supplementary micronutrient fortified foods (e.g. fortified cereal-legume blends); (ii) Lipid-based nutrient supplements vs. no supplementary foods or no intervention; (iii) Lipid-based nutrient supplements vs. multiple micronutrient powders for point-of-use fortification of foods (improved adequacy of local diet); and(iv) Lipid-based nutrient supplements vs. counselling on feeding best possible local diet—home-based foods.

Studies employing simultaneous co-interventions like health education and/or drugs (e.g. deworming or antimalarials) were included if the only difference between the intervention and comparison arms pertained to the provision of lipid-based nutrient supplements.

Types of outcome measures.

Primary outcomes included (i) Proportion of children recovered (author defined); (ii) Duration of recovery (> 1 month; 1 to 2 months; > 2 months); (iii) Mortality; (iv) Weight; (v) Height; (vi) Weight-for-height z-score; (vii) Mid-upper arm circumference; (viii) Height-for-age; and (ix) Relapse (%, readmission into treatment programme).

Secondary outcomes included (i) Default rate (%, author defined); (ii) Morbidity (author defined; incidence of diarrhoea, acute respiratory illness, others); (iii) Haemoglobin (g/dL); (iv) Micronutrient status (e.g. vitamin A, iron); (v) Any adverse effects (author defined; e.g., allergic reactions as diagnosed by clinical assessment, including atopic dermatitis, urticaria, oedema (oral), ophthalmic pruritus, allergic rhinitis, asthma, anaphylaxis); (vi) Change in body composition; (vii) Cardio-metabolic risk factors, including blood pressure, glucose tolerance, and serum lipids; (viii) Change in infant and young child feeding practices; (ix) Environmental impact of lipid-based nutrient supplementation for example, waste from packaging; (x) Intra-household sharing of the provided supplement; and (xi) Cost-effectiveness.

Selection of studies.

The title and abstract of the studies identified in the search were scanned independently by two authors (TG and HSS) to exclude literature that clearly did not meet the inclusion criteria. With the remainder, we examined the full text articles against the eligibility criteria using a structured form. Any uncertainty about inclusion was resolved through discussion and by consensus between all authors.

Data extraction and management.

Two authors (TG and HSS) independently extracted relevant data including details of methods, participants, setting, context, interventions, outcomes and results, publications and investigators using a structured form. We contacted the study authors where reported information was unclear or contradictory, or where important data were missing.The abstracted data on the forms was independently evaluated by the other author. Any discrepancies were resolved through consensus and mutual discussion.

Assessment of risk of bias in included studies.

Two review authors (TG and HSS) independently assessed the risk of bias for each controlled trial using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions [34].

For randomized controlled trials, non-randomized controlled trials and controlled before-after studies these criteria included:

1. sequence generation (selection bias);
2. allocation sequence concealment (selection bias);
3. blinding of participants and personnel (performance bias);
4. blinding of outcome assessment (detection bias);
5. incomplete outcome data (attrition bias); and
6. selective outcome reporting (reporting bias).

For cluster- randomized trials, particular biases to consider included

1. recruitment bias;
2. baseline imbalance;
3. loss of clusters;
4. incorrect analysis; and
5. comparability with individually randomized trials [34].

The judgment for each entry involved assessing the risk of bias as ‘low risk’, as ‘high risk' or as ‘unclear risk’, with the last category indicating either lack of information or uncertainty over the potential for bias. Plots of ‘risk of bias’ assessments were created in RevMan software. Any disagreements were resolved through mutual discussion or group discussion, if required.

Measures of treatment effect.

Risk ratio (RR) estimations with 95% confidence intervals (CI) were used for binary outcomes. Data on continuous outcomes was analysed using either mean differences (MD) or standardized mean differences (SMD) if continuous outcomes were measured with similar, but not identical, instruments across studies. In order to maximize the data input for the pooled outcome measures, we utilized the post-intervention values (means and standard deviations (SDs) in preference to the changes from baseline. We used Review Manager 5.3 to manage the data and conduct the analysis. All results are presented with 95% confidence intervals.

Unit of analysis issues.

In factorial trials and in multi-arm designs yielding two or more intervention groups (different lipid-based nutrient supplements used) and a single control group, the data in the intervention groups, including the variation in the intervention characteristic, was pooled and compared against the single control group to prevent unit-of-analysis error. Ackatia-Armah (C) [35–38] had one intervention arm receiving lipid-based nutrient supplements and three control arms receiving different types of cereal legume blends. The data from these three control groups was pooled for purpose of analyses. LaGrone [39, 40] and Matilsky [22] compared two different types of ready-to-use therapeutic foods with cereal legume blend (corn soy blend plus plus). Here the data in the two intervention groups was pooled to avoid unit of analysis issues.Nikema et al. [41] had a multiarm design comparing the effect of lipid-based nutrient supplements (ready-to-use supplementary foods) with a cereal legume blend and nutritional counselling. These were treated as separate comparisons and are presented as such; i.e., lipid-based nutrient supplements versus specially formulated fortified food and lipid-based nutrient supplements versus nutritional counselling.

Handling of missing data.

We preferred to use intention-to-treat analyses (author reported), if a per-protocol analysis was also reported. If the author had not reported an intention-to-treat analysis, we used the reported analysis but judged the study to be at risk of bias due to this criterion. In case of missing data, we contacted trial authors for information, wherever possible. Where this could not be done, or the authors did not respond, the missing values were imputed where feasible with the help of a statistician.

Assessment of heterogeneity.

We intended to assess contextual heterogeneity on the basis of information collected on the context in which the intervention was implemented. We assessed for variability in the participants, interventions and outcomes studied to identify clinical heterogeneity, and for variability in study design to describe methodological diversity.

Statistical heterogeneity was identified by visual inspection of the Forest plots and the Chi square tests, and measured as recommended by the Cochrane Handbook for Systematic Reviews of Interventions (section 9.5.2) [34]. A rough guide used for interpretation was: (i) 0% to 40%: might not be important; (ii) 30% to 60%: may represent moderate heterogeneity; (iii) 50% to 90%: may represent substantial heterogeneity; and (iv) 75% to 100%: considerable heterogeneity.

In view of few included trials, a P value of 0.10 from the Chi2 test was used as cut off for statistical significance.

Assessment of reporting biases.

The presence of reporting bias in the extracted data was evaluated quasi-statistically using a funnel plot [42]. Formal statistical tests for funnel plot asymmetry, namely the Begg and Egger’s methods [43, 44] were not conducted because of insufficient number of trials.

Data synthesis.

We performed statistical analysis using the Review Manager software (RevMan 5.3). In concordance with the current recommendations [34], we conducted the meta-analysis of included randomized controlled trials and other trials separately. Pooled estimates (RR with 95% CIs) of the evaluated outcome measures were calculated by the generic inverse variance method. Results are not depicted as ‘not statistically significant’ or ‘nonsignificant’, but we have reported the CI together with the exact P value. We included studies with large variation with respect to populations, interventions, comparators, outcomes and settings. Thus, the true effect is likely to be related, but not the same for all studies. We therefore chose the random-effects model. If it was not possible to synthesise the data for some outcomes (e.g duration to recovery), we provided a narrative synthesis of the results.

The data was finally synthesised using a ‘summary of findings’ table that provided key information concerning the quality of evidence, the magnitude of effect of the interventions examined, and the sum of available data on all primary outcomes for a given comparison. For each primary outcome, quality assessment of the results was also carried out using the GRADE approach, which specifies four levels of quality (high, moderate, low and very low) where the highest quality rating is for a body of evidence based on randomized trials. Quality was assessed separately for each outcome.

Subgroup analysis and investigation of heterogeneity.

We planned to explore differences in recovery by the following subgroup analysis, if data permitted: age of the study group (6 to 24 mo, > 2years), breastfeeding practises in under 2 population, dietary intake of the study population, exposure to HIV infection (yes/no). antiretroviral therapy for HIV infected population. setting in refugee or internally displaced camps (yes/no). living in an emergency affected country (WHO definition) and not in a camp (yes/ no), belongs to food insecure group (yes/no). height-for-age of the study population. treatment provided in inpatient or outpatient setting. dose (calories provided per kg per day) of the formulation. type of formulation (milk based or not). type of macronutrients provided and duration of intervention.

The available data, however, allowed us to conduct subgroup analyses only for age of study group, dose of formulation, type of formulation (ready-to-use therapeutic foods vs. ready-to-use supplementary foods) and duration of intervention.

Sensitivity analysis.

We undertook sensitivity analyses with respect to the risk of bias parameters, namely: (i) sequence generation (selection bias), (ii) allocation sequence concealment (selection bias), (iii) blinding of participants and personnel (performance bias), (iv) blinding of outcome assessment (detection bias), (v) incomplete outcome data (attrition bias), (vii) selective outcome reporting (reporting bias) and (viii) other potential sources of bias.

The protocol of this review was registered at PROSPERO (CRD42016036458).

Results of the search.

The search results are summarised in Fig 1. We screened 8661 records, of which 38 were potentially eligible. Of these 13 publications (9 studies) were included in the final analyses [22, 35–41, 45–49] and 21 references (from 19 studies) [14, 17, 50–68] were excluded. One study is ongoing [69] and three references (from two studies) await classification [70–72] (S6 Table, S7 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study flow diagram.

https://doi.org/10.1371/journal.pone.0182096.g001

Included studies.

A total of 9 trials, incorporating data on 9270 children with moderate acute malnutrition were included in the final analysis (Table 1). Four trials were cluster randomized controlled trials, four were randomized controlled trials and one was a non-randomized controlled trial. All the trials were conducted in Africa. None of the trials were conducted in an emergency/refugee setting. One trial was conducted in a hospital setting [45] while the rest were community-based trials. Only one trial included children younger than 2 years exclusively [41], while most included children between the ages of 6 months to 5 years. Authors used different definitions for recovery (Table 1; S2 Table). The intervention was ready-to-use supplementary foods in seven of the nine included trials, while the other two provided ready-to-use therapeutic foods. Three of the lipid-based nutrient supplements used were milk based. The comparator was specially formulated micronutrient fortified food in all of the nine included trials (Cereal Legume Blend in 8 and United Nations Feeding Programme Supplementations in one). One trial [41] also used nutritional counselling as a control arm which was analyzed separately. The duration of supplementation varied (from 8 to 16 weeks). Calories provided varied from 250 to 1200 kcal/day in 7 trials and from 40 to 75 kcal/kg/day in 2 trials. The important characteristics of populations, studies and interventions are summarized in Tables 1 and 2 and details of individual studies can be seen in the section S2 Table. S3 Table compares the PROGRESS-Plus equity categories in the included studies. A descriptive analysis of the PROGRESS-Plus factors that were reported is presented in S4 Table. None of the studies had reported directly on social determinants of health.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of populations from the included trials.

https://doi.org/10.1371/journal.pone.0182096.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Details of intervention.

https://doi.org/10.1371/journal.pone.0182096.t002

Excluded studies.

Reasons for the exclusion of individual trials are listed in S5 Table. Of the 21 publications from 19 trials [14, 17, 50–68] the most common reason for exclusion was that study participants did not have moderate acute malnutrition (12/21). In two trials, the intervention did not comprise lipid-based nutrient supplements, three were observational studies, one study used participants with severe acute malnutrition after the stabilization phase, in two studies the outcomes were not of interest, and in one trial both intervention arms received lipid-based nutrient supplements with no control group.

Allocation (selection bias).

The risk of bias for the eight randomized controlled trials was low for random sequence generation. In one non-randomized trial [45], the risk of bias was high.

Allocation concealment was adequately performed in five of the nine included trials for which the risk of bias was judged to be low. The details of allocation concealment were not clear from the published text for two trials, in which the selection bias was judged to be unclear. For two trials, where allocation concealment was not done [41, 45], the risk of bias was considered to be high.

Blinding (performance bias and detection bias).

Lipid-based nutrient supplements are ready to eat supplements, supplied in packets to the participants. The control group, on the other hand, received supplements which required cooking for consumption or received nutritional counselling. Owing to the nature of these interventions, it was not possible for the participants to be blinded. However, some investigators ensured that field personnel and those performing anthropometry or other investigations were blinded to the allocation of study participants. Such trials were considered at low risk for performance bias [22, 39, 40, 48]. For the remaining six trials, the risk of performance bias was judged to be high.

The risk of bias for blinding for outcome assessment was low for two trials [39, 40, 48], unclear for two trials [35–38, 22] and was considered high for the remaining five trials.

Incomplete outcome data (attrition bias).

Attrition bias was low for eight out of the nine included trials. In one trial [45], the overall attrition was high and differed significantly in the two comparison groups.

Selective reporting (reporting bias).

Selective reporting was not detected in any of the included trials.

Other potential sources of bias.

Two studies were judged to be at high risk of bias for other reasons. In Karakochuk et al. [47], the quantity of the corn soy blend ration was much higher than ready-to-use supplementary foods, because it was assumed that there would be significant household food sharing of the corn soy blend ration. Due to communication problems, during the course of the study in Nackers et al. [49], children weighing > 8 kg at any time during their treatment were provided with three daily packs of ready-to-use therapeutic foods instead of two as planned in the study protocol. Consequently, these children were excluded from the analysis.

Recruitment bias.

Of the four cluster randomized controlled trials, recruitment bias was judged to be low in three. In Delchevalerie et al. [46], the risk of bias was considered unclear as the sequence of recruitment was not clear from the published text.

Loss of clusters.

None of the trials reported any loss of clusters.

Baseline imbalance.

The risk of bias due to baseline imbalance was considered high in Nikiema et al. [41], where the baseline comparability of the clusters was not reported. However, the lipid-based nutrient supplements group had a slightly better nutritional status and lower morbidity at baseline. This risk of bias for the other three trials was considered to be low.

Incorrect analyses.

The cluster effect was clearly taken into account in two of the four trials. In the remaining two, author clarifications are awaited.

Proportion of children recovered.

All nine trials (9270 children) reported this outcome. The meta-analyses of the eight randomized controlled trials showed that the overall recovery rate was 8% higher in the lipid-based nutrient supplement group (RR 1.08; 95% CI 1.02 to 1.14, 8 studies, 8934 participants, low quality evidence). One non-randomized controlled trial also showed higher recovery rate in the lipid based nutrient supplement group (Table 3, S8 Table Analysis 1.5; Fig 4). There was significant heterogeneity between trials (I² = 80%, p<0.00001). On stratified analyses based on study designs, effect size was a little lower for cluster randomized controlled trials (4 trials, 4328 participants, RR 1.06, 95% CI 0.97, 1.16) as compared to randomized controlled trials (RR 1.11, 95% CI 1.01 to 1,21, 4 studies, 4606 participants) and non-randomized controlled trials (RR 1.23, 95% CI 1.06 to 1.42, 1 study, 336 participants, very low quality of evidence). The test for subgroup difference showed low heterogeneity (I² = 29.7%, p = 0.24).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Forest plot Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods Outcome: 1.5 Recovery from moderate acute malnutrition (ALL).

https://doi.org/10.1371/journal.pone.0182096.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of findings: Lipid-based nutritient supplements compared to Specially formulated micronutrient fortified foods for treatment of moderate acute malnutrition in children (6 months to 59 months) with Moderate Acute Malnutrition.

https://doi.org/10.1371/journal.pone.0182096.t003

In subset analyses to explore heterogeneity, the type of macronutrient provided was associated with the recovery rate; ready-to-use therapeutic foods had a higher recovery rate than ready-to-use supplementary foods (RR 1.23, 95% CI 1.09, 1.38; one study for ready-to-use therapeutic foods; RR 1.06, 95% CI 1.01, 1.11 for ready-to-use supplementary foods; P = 0.02 for subgroup differences) (S8 Table, Analysis 1.3; Fig 5). Lipid-based nutrient supplements containing milk also had a higher recovery rate (RR 1.15, 95% CI 1.03, 1.29) than those without milk (RR 1.06, 95% CI 1.01, 1.11; overlapping CIs) (S8 Table, Analysis 1.7). Age group of the participants (Table 3, Analysis 1.2), duration of supplementation (S8 Table, Analysis 1.4) calories provided per day (S8 Table, Analysis 1.6; Fig 6) did not explain the heterogeneity of effect.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.3 Recovery from moderate acute malnutrition (SUBGROUP: by type of supplement).

https://doi.org/10.1371/journal.pone.0182096.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.6 Recovery from moderate acute malnutrition (SUBGROUP: Calories provided).

https://doi.org/10.1371/journal.pone.0182096.g006

LaGrone et al. [39, 40] performed binary logistic regression analysis to evaluate the effect of various factors on recovery in the two treatment groups. Children with HIV recovered less frequently [53 of 84 (63%) compared with 2310 of 2628 (88%); P <0.0001]. HIV positive children also had a higher risk of developing severe wasting [19 of 31 (61%) compared with 126 of 318 (40%), P < 0.03] and development of kwashiorkor was less frequent in HIV-positive children [4 of 31 (13%) compared with 116 of 318 (36%); P <0.01]. Recovery rate was higher in children receiving antiretroviral therapy (19 of 24 (79.2%)versus 31 of 54 (57.4%); RR: 1.38; 95% CI: 1.01, 1.88). These results were not affected by the type of supplementary food the child received.

Overall time taken to recovery was a little lower in children receiving lipid-based nutrient supplements. Three trials provided the mean duration (days) taken to recover from moderate acute malnutrition. Meta-analyses of these trials showed slightly shorter duration to recovery, but with wide confidence intervals (n = 2020; MD -4.77, 95% CI -12.54 to 3.0) (Table 3, S8 Table, Analysis 1.9). LaGrone [39, 40] does not provide detailed data; however, they mention that the mean duration of treatment required to achieve recovery was 23 days in the total study cohort. Of these, the children who received corn soy blend plus plus took an average of 2 days longer to recover (ANOVA, P < 0.003). Also, more than half of the children recovered in the first two weeks of therapy itself. Nackers et al. [49] mention the length of stay (to recovery) as median and range. This was significantly lower for the lipid-based nutrient supplements group (median 4 weeks; range 2 to 16 weeks), when compared with the corn soy blend cohort (median 6 weeks; range 2 to 16 weeks). Matilsky et al. [22] also mention the length of stay as median and 25th and 75th centile. This was lower in both the lipid-based nutrient supplements groups (median 14 days; IQR 14, 42), when compared to corn soy blend (median 28 days; Interquartile Range 14, 56). The mean period of food supplementation for children receiving corn soy blend was 4.0 weeks and the mean period for children receiving soy/peanut Fortified Spread or milk/peanut Fortified Spread was 3.3 weeks. Vanelli [45] report that the children in the intervention group (who received United Nations World Food Programme Supplementation portions and lipid-based nutrient supplements) recovered faster (mean duration 5.54 weeks) than those treated with Food Programme Supplementation regimen only (mean 8.16 weeks).

No recovery.

The number of children who did not recover was lower in the lipid-based nutrient supplements group (RR 0.70, 95% CI 0.58 to 0.85, 7 randomized controlled trials, 8364 participants) (Table 3, S8 Table, Analysis 1.8; Fig 7). The effect size was almost the same for the various study designs; for cluster randomized controlled trials (3 trials, 3758 study participants, RR 0.77, 95% CI 0.63, 0.94), for randomized controlled trials (4 trials, 4606 study participants, RR 0.54, 95% CI 0.41, 0.70) and for 1 non-randomized controlled trial (1 trial, 336 study participants, RR 0.51, 95% CI 0.27, 0.98) (45). There was little heterogeneity of effect among trials (I² = 48%, p = 0.07) or among subgroups (I² = 0%, p = 0.37).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Forest plot: Lipid-based nutrient supplement versus specially formulated micronutrient fortified foods, outcome: 1.2 No recovery.

https://doi.org/10.1371/journal.pone.0182096.g007

Mortality.

All nine trials (9270 children) reported this outcome. In the eight randomized controlled trials, the overall mortality was not affected by the type of supplement provided (RR = 0.91; 95% CI 0.54, 1.42) (Table 3, S8 Table, Analysis 1.12; Fig 8). There was no heterogeneity of effect between trials (I² = 0%, p = 0.90). The effect size was similar for cluster randomized controlled trials (4 trials, 4328 study participants, RR 0.83, 95% CI 0.24, 2.91), randomized controlled trials (4 trials, 4606 study participants, RR 0.92, 95% CI 0.52, 1.63) and non-randomized controlled trial (1 trial, 336 study participants RR 0.56, 95% CI 0.05, 6.08). The tests for subgroup difference showed no heterogeneity (I² = 0%, p = 0.70).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Forest plot: Lipid-based nutrient supplement versus specially formulated micronutrient fortified foods, outcome: 1.6 Mortality.

https://doi.org/10.1371/journal.pone.0182096.g008

Long term mortality after supplementation was stopped (6 months to 1 year later) was studied by 3 trials (2859 study participants) (RR 0.62, 95% CI 0.28, 1.37). The results were variable across different study designs. The 1 cluster randomized controlled trial (570 study participants) showed reduction in long term mortality in the lipid-based nutrient supplement group (RR 0.36, 95% CI 0.13, 0.98) while the 2 randomized controlled trials did not show any difference (RR 1.25, 95% CI 0.16, 9.49). There was significant heterogeneity of effect among the subgroups (Table 3, S8 Table, Analysis 1.13; Fig 9).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Forest plot: Lipid-based nutrient supplement versus specially formulated micronutrient fortified foods, outcome: 1.7 Post Discharge Mortality.

https://doi.org/10.1371/journal.pone.0182096.g009

Deterioration to severe acute malnutrition.

Data from 6 trials (7124 children) showed that children with moderate acute malnutrition receiving lipid-based nutrient supplements had a lower risk of developing severe acute malnutrition when compared to those receiving Specially Formulated Foods. The meta-analyses of the five randomized controlled trials showed lower risk of developing severe acute malnutrition with lipid based nutrient supplement (RR 0.87, 95% CI 0.73, 1.03). The effect size was almost similar for the two cluster randomized controlled trials (2633 study participants; RR 0.82, 95% CI 0.58, 1.14) and the three randomized controlled trials (4155 study participants; RR 0.89, 95% CI 0.73, 1.09). The effect size was larger and more imprecise for the non-randomized trial (45). There was no significant heterogeneity of effect among the included trials (Table 3, S8 Table, Analysis 1.10; Fig 10)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.10 Deterioration to severe acute malnutrition.

https://doi.org/10.1371/journal.pone.0182096.g010

Transferred to inpatient facility.

The patients were transferred to inpatient facilities either because of severe acute malnutrition or development of medical complications. Data for this outcome was reported in 5 trials (2 cluster randomized controlled trials and 3 randomized controlled trials) (Table 3, S8 Table, Analysis 1.11; Fig 11). Children receiving lipid-based nutrient supplements had a lower chance of transfer to in-patient treatment (4939 study participants; RR 0.57, 95% CI 0.24, 1.34). The effect estimate was imprecise with wider confidence intervals for the 2 cluster randomized controlled trials (1695 study participants; RR 0.54, 95% CI 0.07, 4.30) but more precise in the 3 randomized controlled trial (3244 study participants; RR 0.50, 95% CI 0.31, 0.81). There was no significant heterogeneity of effect among the included trials (I² = 58%, p = 0.07) or across the subgroup (I² = 0%, p = 0.94).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.11 transferred to Inpatient.

https://doi.org/10.1371/journal.pone.0182096.g011

Relapse after discharge.

In two randomized controlled trials the risk of relapse was almost 20% in both the groups (n = 2289, RR = 0.99, 95% CI 0.82, 1.19) (Table 4, S8 Table, Analysis 1.23) (Fig 12).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 12. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.23 Relapse after discharge.

https://doi.org/10.1371/journal.pone.0182096.g012

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Summary of findings: Lipid-based nutrient supplements compared to Specially formulated micronutrient fortified foods for treatment of moderate acute malnutrition in children (6 months to 59 months).

https://doi.org/10.1371/journal.pone.0182096.t004

Weight gain.

Seven trials reported data on weight gain in the study participants. Of these, 6 trials reported this as g/kg/day. In quantitative synthesis from 5 trials (Table 3, S8 Table, Analysis 1.14; Fig 13), weight gain was higher with lipid-based nutrient supplements (5054 study participants; MD = 0.62 gm/kg/day; 95% CI = 0.18, 1.06). The effect size was similar but with wider confidence intervals for the cluster randomized controlled trials (2 trials, 1939 study participants; MD = 0.49 gm/kg/day; 95% CI = -0.69, 1.66) as compared to randomized controlled trials (3 trials; 3115 study participants; MD = 0.71 gm/kg/day; 95% CI = 0.24, 1.18). Matilsky et al. [22] compared the weight gain (g/kg/day) at each two week interval in the lipid-based nutrient supplements and control group and depicted the data as a figure (not included in the meta-analysis). Rates of weight gain were greater (P< 0.05) with lipid-based nutrient supplements compared to corn soy blend, except in the 6 to 8 week interval.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 13. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.14 Weight Gain (g/kg/d) [g/kg/day].

https://doi.org/10.1371/journal.pone.0182096.g013

One cluster randomized controlled trial [35–38] provided information on the absolute weight gain after the supplementation period and showed a differential weight gain of 0.23 kg (95% CI 0.14, 0.32) with lipid-based nutrient (Tables 3, S8, Analysis 1.15; Fig 14).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 14. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.15 Weight Gain Total [kg].

https://doi.org/10.1371/journal.pone.0182096.g014

Weight-for-height z-score.

Three trials provided weight-for-height z-scores at the end of the study period (Table 3, S8 Table, Analysis 1.16; Fig 15). Of these two were randomized controlled trials and one was a cluster randomized controlled trial. Children receiving lipid-based nutrient supplement had a higher weight-for-height z-score (n = 5442; MD = 0.10, 95% CI 0.05, 0.14). The cluster randomized controlled trial showed no evidence of better effect (n = 1369; MD = 0.00, 95% CI -0.47, 0.47), while the two randomized controlled trials showed a benefit (n = 4074, MD = 0.11, 95% CI = 0.04, 0.17). One trial studied the impact of lipid-based nutrient supplements on the change in weight-for-height z-scores. This cluster randomized controlled trial [35–38] showed that while both groups had improvement in weight-for-height z-scores, the lipid-based nutrient supplements group effect was higher (MD = 0.28, 95% CI0.15, 0.41) (Table 3, S8 Table, Analysis 1.17; Fig 16).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 15. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.16 weight-for-height z-score End.

https://doi.org/10.1371/journal.pone.0182096.g015

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 16. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.17 Weight-for-length z-score Gain.

https://doi.org/10.1371/journal.pone.0182096.g016

Height/ length.

One trial studied the change in height (cm). This cluster randomized controlled trial (35–38) showed that while both groups showed increase in height, lipid-based nutrient supplement group had better improvement (MD = 0.16cm, 95% CI-0.02, 0.34) (Table 4, S8 Table, Analysis 1.18; Fig 17). Two trials evaluating rate of change in height (mm/day)found no difference between the two groups (MD = 0.00, 95% CI -0.03, 0.02) (Table 4, Analysis 1.19; Fig 18). Matilsky et al. [22] compared the height gain (mm/day) at each two-week interval and depicted the data as a figure (not included in the meta-analysis). Rates of height gain were similar in children receiving either lipid-based nutrient supplements or corn soy blend.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 17. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.18 Length Gain.

https://doi.org/10.1371/journal.pone.0182096.g017

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 18. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.19 Length Gain (mm/d).

https://doi.org/10.1371/journal.pone.0182096.g018

Height-for-age z-scores.

Two trials (n = 2731) were evaluated this outcome. Height-for-age z-scores at the end of the study period were a little higher with the lipid-based nutrient supplements (MD = 0.16, 95% CI = -0.03, 0.34) (Fig 19). There was one cluster randomized controlled trial (n = 1369, MD = 0.20, 95% CI -0.29, 0.69) and one randomized controlled trial (n = 1362; MD = 0.15, 95% CI -0.05, 0.35) for this analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 19. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.20 height-for-age z-score End.

https://doi.org/10.1371/journal.pone.0182096.g019

Mid upper arm circumference.

One trial studied the change in mid-upper arm circumference (cm). In this cluster randomized controlled trial [35–38] lipid-based nutrient supplement group had higher increase (MD = 0.29 cm, 95% CI 0.17, 0.41) (Table 4, S8 Table, Analysis 1.21; Fig 20). Four trials evaluating rate of change in mid-upper arm circumference (mm/day) showed slightly better improvement with lipid-based nutrient supplement (n = 4474; MD = 0.04, 95% CI 0.02, 0.06) (S8 Table, Analysis 1.22; Fig 21). The cluster randomized controlled trial (MD = 0.04, 95% CI 0.01, 0.07) and three randomized controlled trials showed similar effect sizes (MD = 0.04, 95% CI 0.02, 0.07). Matilsky et al. [22] compared the mid-upper arm circumference gain (mm/day) at each two week interval and depicted the data as a figure (not included in the meta-analysis). Rates of mid-upper arm circumference gain were greater (P< 0.05) with lipid-based nutrient supplement in comparison to corn soy blend, except with soy/peanut Fortified Spread at 6–8 week interval.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 20. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.21 mid-upper arm circumference Gain [cm].

https://doi.org/10.1371/journal.pone.0182096.g020

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 21. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.22 Mid-upper arm circumference Gain (mm/day) [mm/day].

https://doi.org/10.1371/journal.pone.0182096.g021

Default rate.

Default rate was reported in 7 trials (7570 study participants). Of these, three were cluster randomized controlled trials and four were randomized controlled trials. Lipid-based nutrient supplements had a higher risk of default (RR = 1.32, 95% CI 0.73, 2.40). There was significant heterogeneity of effect across the studies. Cluster randomized controlled trials showed high risk of default (n = 2964; RR = 2.59, 95% CI 0.79, 8.53), while randomized controlled trials showed lower risk of default with lipid-based nutrient supplements (n = 4606; RR 0.88, 95% CI 0.56, 1.40) (Table 4, S8 Table, Analysis 1.24; Fig 22).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 22. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.24 Default Rate.

https://doi.org/10.1371/journal.pone.0182096.g022

Hemoglobin (g/dL).

In one cluster randomized controlled trial, lipid-based nutrient supplement group showed slightly higher hemoglobin (n = 1154, MD = 0.25 g/dL; 95% CI 0.06 to 0.44) (Table 4, S8 Table, Analysis 1.25; Fig 23). Two trials, one cluster randomized controlled trial and one randomized controlled trial, provided information on the change in hemoglobin levels. The cluster randomized controlled trial showed that although both the lipid-based nutrient supplement and control groups showed a positive change in hemoglobin levels, the effect was higher with lipid-based nutrient supplements (1154 study participants; MD = 0.72 gm/dL, 95% CI 0.45 to 0.99). The one randomized controlled trial included, however, showed similar change in hemoglobin levels in the two groups (203 study participants; MD = 0.01 gm/dL, 95% CI -0.15 to 0.17). On pooling, there was a higher change with the lipid-based nutrient supplement group with imprecise confidence intervals (1357 study participants; MD = 0.36, 95% CI -0.34 to 1.05) (Table 4, S8 Table, Analysis 1.26; Fig 24).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 23. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.25 Hemoglobin (Final).

https://doi.org/10.1371/journal.pone.0182096.g023

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 24. Forest plot: Lipid-based nutrient supplement versus Specially formulated micronutrient fortified foods, outcome: 1.26 Change in Hemoglobin [gm/dL].

https://doi.org/10.1371/journal.pone.0182096.g024

Adverse effects.

None of the included trials reported systematically on the adverse effects of the dietary supplements provided. All of them reported 'no significant adverse effects'.

Morbidity.

In one trial, there was a higher risk of vomiting (RR 1.37, 95% CI 1.09, 1.72) and a slightly higher risk of diarrhea (RR 1.10, 95% CI 0.98, 1.24) (Table 5) in the lipid-based nutrient supplement group.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Summary of findings: Lipid-based nutrient supplements compared to specially formulated micronutrient fortified foods for treatment of moderate acute malnutrition in children (6 months to 59 months) with moderate acute malnutrition.

https://doi.org/10.1371/journal.pone.0182096.t005

Change in body composition.

Only one cluster randomized controlled trial [35–38] reported on the change in body composition (total body water, fat mass and % fat mass). At baseline, total body water, fat mass, and percent fat Mass across all study participants was 7.03 ± 1.12 kg, 3.94 ± 0.76 kg, and 29.0 ± 6.32%, respectively. At 12 weeks, fat mass increased by 0.35 kg, 0.29 kg, 0.25 kg, and 0.41 kg (p = 0.02) among children in the Supplementary Plumpy, corn soy blend plus plus, Misola, and locally milled flours groups, respectively; however, the change in percent fat mass did not differ between groups (p = 0.13). Gains in total body water, fat mass, and percent fat mass were greater in children who recovered from moderate acute malnutrition (p<0.0001).

Infant and young child feeding practices.

There was no detailed data studying the impact of lipid-based nutrient supplementation on infant and young child feeding practices. One trial [35–38] scheduled home visits (not for all included study participants). During the 8 hours of in-home observation (n = 164), > 70% of mothers in all groups breastfed the study child at least once.

Intra-household sharing of the provided supplement.

In Matilsky et al. [22] caretakers receiving any supplement believed that the foods were a medical treatment and only four of them reported sharing the food with others. During home visits Ackatia-Armah et al. (35–38) noted that the supplement was almost always consumed by the study child (> 94% of observation days) and was shared with other children during fewer than 4% of home visits.

There was no data on cardio-metabolic risk factors, environmental impact of lipid-based nutrient supplementation and cost effectiveness in the included trials.