Lipid-based nutrient supplements and all-cause mortality in children 6–24 months of age : a meta-analysis of randomized controlled trials

Lipid-based nutrient supplements and all-cause mortality in children 6–24 months of age: a meta-analysis of randomized controlled trials Christine P Stewart,1K Ryan Wessells,1Charles D Arnold,1Lieven Huybregts,2Per Ashorn,3Elodie Becquey,2Jean H Humphrey,4and Kathryn G Dewey1 1Institute for Global Nutrition, University of California, Davis, Davis, CA, USA; 2International Food Policy Research Institute, Washington, DC, USA; 3Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland; and 4Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA ABSTRACT Background: Undernutrition is associated with an elevated risk of mortality among children in lowand middle-income countries. Small-quantity lipid-based nutrient supplements (LNS) have been evaluated as a method to prevent undernutrition and improve infant development, but the effects on mortality are unknown. Objective: Our objective was to evaluate the effect of LNS on allcause mortality among children 6–24 mo old. Methods: We conducted a systematic review and meta-analysis of randomized controlled trials of LNS designed to prevent undernutrition, with or without other interventions. Literature was searched in May 2019 and trials were included if they enrolled children between 6 and 24 mo old and the period of supplementation lasted ≥6 mo. We extracted data from participant flow diagrams and contacted study investigators to request data. We conducted a meta-analysis to produce summary RR estimates. Results: We identified 18 trials conducted in 11 countries that enrolled 41,280 children and reported 586 deaths. The risk of mortality was lower in the LNS arms than in the non-LNS comparison arms (RR: 0.73; 95% CI: 0.59, 0.89; 13 trials). Estimates were similar when trials with maternal LNS intervention arms were added or when alternative formulations of LNS were excluded. The results appeared stronger in trials in which LNS were compared with passive control arms. Excluding these contrasts and only comparing multicomponent arms with LNS groups and comparison groups that contained all the same components without LNS attenuated the effect estimate (RR: 0.82; 95% CI: 0.61, 1.10). Conclusions: LNS provided for the prevention of undernutrition may reduce the risk of mortality, but more trials with appropriate comparison groups allowing isolation of the effect of LNS alone are needed. This study was registered at www.crd.york.ac.uk/PROSPERO as CRD42019128718. Am J Clin Nutr 2020;111:207–218. Keywords: lipid-based nutrient supplement, complementary feeding, child mortality, infants and young children, home fortification Introduction Globally, an estimated 5.5 million children <5yofage die each year, the majority from preventable causes (1). Undernutrition among infants and children <2yofage remains common in lowand middle-income countries and is associated with increased risk of mortality, including among children with mild to moderate degrees of undernutrition (2,3). Among severely or moderately malnourished children, readyto-use therapeutic and supplementary foods have been found to significantly improve recovery rates and reduce the risk of mortality (4,5). Lipid-based nutrient supplements (LNS) are available in various quantities and formulations for the prevention or treatment of malnutrition (6) and are designed to provide multiple micronutrients embedded in a food base that also provides energy, protein, and essential fatty acids. Large-quantity LNS are typically used for treatment of severe acute malnutrition and provided in dosages of ≥500 kcal/d (5). Medium-quantity LNS are generally offered as ready-to-use supplementary food for the treatment of moderate acute malnutrition in dosages of 250–500 kcal/d, and small-quantity LNS are generally given as 100–120 kcal/d for the prevention of undernutrition. Supported by Bill & Melinda Gates Foundation grant OPP49817 (to KGD). LH received funding from the Consultative Group for International Agricultural Research program on Agriculture for Nutrition and Health, led by the International Food Policy Research Institute. Supplemental Tables 1–3 and Supplemental Figures 1 and 2 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup. com/ajcn/. Address correspondence to CPS (e-mail: cpstew[email protected]). Abbreviations used: CSB, corn–soy blend; IYCF, infant and young child feeding; LNS, lipid-based nutrient supplement(s); MNP, micronutrient powder; WASH, water, sanitation, and hygiene; WSB, wheat–soy blend. Received July 12, 2019. Accepted for publication September 23, 2019. First published online November 7, 2019; doi: https://doi.org/10.1093/ ajcn/nqz262. Am J Clin Nutr 2020;111:207–218. Printed in USA. Copyright ©The Author(s) 2019. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 207 Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 208 Stewart et al. The primary objective of most trials of small-quantity and some trials of medium-quantity LNS has been to evaluate their efficacy for the prevention of malnutrition, and such trials have therefore focused on linear growth and risk of stunting as their primary outcome measures. Many trials also have evaluated indicators of child development as secondary outcomes. The combination of macroand micronutrients addresses multiple potential nutritional deficiencies and thus may also reduce child mortality, although none of the trials had been designed with this as an explicit objective. A recently published review considered the impact of small-quantity LNS on a variety of outcomes, including all-cause mortality (7). However, the analysis of the effect on mortality was limited because the authors only included 3 trials, 2 of which reported deaths when children were <6mo old and therefore too young to have been directly exposed to LNS. The recent publication of a number of additional large trials of LNS provides a larger sample size to examine this outcome. Thus, the objective for this meta-analysis is to evaluate the effect of smalland medium-quantity LNS on child mortality, using an expanded set of trials and restricting the analysis to children who were ≥6 mo old, i.e., eligible to receive LNS. We hypothesize that children 6–24 mo old who receive LNS for a minimum of 6 mo will have a lower mortality rate than those who do not receive LNS. Methods The study was registered through Prospero (CRD42019 128718) and a detailed statistical analysis plan was developed and posted publicly (https://osf.io/q76r8/) before beginning the analysis. The primary and only outcome considered in this analysis was all-cause mortality. Potential trials for inclusion in the meta-analysis were identified using a systematic review process that mirrored that of the recent meta-analysis of LNS (7). To capture additional, recently published trials, we repeated the search protocol described by Das et al. (7) Specifically, we searched 16 international and 9 regional databases in May 2019 using the keyword and the same controlled vocabulary search terms laid out by Das et al.; reference lists of newly included trials were reviewed to identify any additional trials. The titles and abstracts of all identified records were screened; full-text reports for potentially eligible trials were reviewed using the inclusion and exclusion criteria described below. When only meeting abstracts were identified, we contacted the authors to invite them to share their results. Inclusion and exclusion criteria We restricted this analysis to prospective randomized controlled trials conducted in lowor middle-income countries that were designed such that all enrolled children were eligible to receive ≥6 mo of supplementation between 6 and 24 mo of age. Trials were excluded if they focused primarily on the treatment, not prevention, of malnutrition. Generally, this was defined as those trials in which severe or moderate malnutrition was an inclusion criterion for children in the trial. We did not include trials that provided ≥500 kcal LNS/d because such quantities exceed the average energy needed from non-breast-milk sources at 6–12 mo (6) and thus may reduce breast-milk intake and/or lead to incomplete consumption of the daily ration, thereby reducing the dose of micronutrients and fatty acids received. Both of these consequences could compromise the potential benefits of LNS. Trials that did not formally exclude children with severe or moderate acute malnutrition, as part of the larger sample, were included in this analysis. Because many trials enrolled children before the supplementation age of ≥6 mo, we restricted our analysis to children who were still in the trial at the time of initiation of child supplementation. When possible, this restriction was applied through careful extraction of data from publications; when not possible, we reached out directly to investigators. Our primary comparison was of children in smalland medium-quantity LNS intervention groups with those in nonLNS control groups. If a single trial contained multiple relevant LNS interventions (e.g., varying dosages, formulations, or combinations in different arms), we combined these groups so that there was a single comparison per trial. In some cases, LNS were provided in combination with other interventions within a single intervention arm. All forms of small-quantity and medium-quantity LNS and interventions that included LNS were included as “LNS” in the primary analyses and were differentiated in prespecified sensitivity analyses described below. Non-LNS comparison groups varied by trial. For the primary comparison, we considered all non-LNS arms as a “Control” group for that trial, excluding intervention arms that received other types of (non-LNS) child supplementation, such as micronutrient powder or other fortified blended foods. Intervention arms that included maternal LNS supplementation, in addition to child LNS, were excluded from the primary analysis. Maternal LNS supplementation may influence infant mortality independently of child LNS supplementation, as suggested by the lower RR for neonatal mortality in a recent meta-analysis of maternal LNS trials (0.72; 95% CI: 0.47, 1.10) (8). Although this was not statistically significant, that analysis was limited in power owing to a relatively small sample size, because only 3 trials were included (8). Our meta-analysis focuses on child mortality after 6 mo of age, when a residual effect of maternal supplementation is less likely. Nonetheless, it is possible that maternal LNS supplementation may affect the risk of mortality after 6 mo, e.g., because of longer-term effects on immune function related to birth size (9). For this reason, we excluded the trials that included maternal LNS from our primary analysis, but included them in a separate sensitivity analysis described below. Risk of bias assessment Risk of bias in the included trials was assessed using the criteria in the Cochrane Handbook for Systematic Reviews of Interventions (10). Two reviewers evaluated each trial against the following criteria: random sequence generation, allocation concealment, blinding of participants and personnel, outcome assessment bias, incomplete outcome assessments, selective outcome reporting, and other sources of bias. Data extraction Data were extracted primarily from participant flow diagrams. When insufficient detail was available in the diagram, we Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 Meta-analysis of effect of LNS on child mortality 209 contacted investigators for additional information. If the child supplementation began at enrollment, our denominator for analysis was the number of children enrolled. All deaths after enrollment were considered events. If supplementation began after enrollment (e.g., trials that enrolled during pregnancy), the denominator was the number of children alive and still enrolled when supplementation began. Only deaths that occurred after supplementation began were considered as events. Data extraction was independently replicated. Data analysis Our parameter of interest was the pooled RR of all-cause child mortality comparing LNS with non-LNS intervention groups. We used a patient-weighted random-effects method proposed by Shuster and Walker (11,12) to include trials in which 1 or both arms had 0 events. We did not adjust our methods for the inclusion of cluster-randomized trials. A recent meta-epidemiological study comparing effect estimates between cluster-randomized trials and individually randomized trials found no systematic differences between the estimation methods and concluded that the 2 types of trials can be safely pooled together for binary outcome meta-analyses (13). Statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc.). We explored trial protocol heterogeneity by summarizing the methods, participants, interventions, monitoring approaches, and potential for bias of each trial. Trials with protocols that differed substantially from the rest triggered discussion of sensitivity analyses, which are described below. For statistical heterogeneity we present I2and τ2statistics. Based on a preliminary literature search, we expected to find ≥18 trials with ≥32,500 participants and 400 deaths. Assuming similarly sized trials with equal-sized treatment groups, we would expect to have 80% statistical power to detect an RR of 0.77 between intervention and control groups at the 95% confidence level. Sensitivity and secondary analyses The majority of trials used similar LNS distribution mechanisms [e.g., weekly or monthly rations provided by study staff, community health workers, or other health extension agents together with some infant and young child feeding (IYCF) counseling] and used similar formulations of LNS, specifically peanutand milk-based products providing ∼1RDA of most micronutrients. However, some trials integrated LNS supplementation together with other types of interventions, such as water, sanitation, and hygiene (WASH) or enhanced morbidity monitoring and treatment. The comparison groups also differed between trials. Most had an active control group for which there was a similar visit frequency or set of activities as for the LNS intervention group. However, some trials had a passive control group or nonintervention group, for which there were no contacts between study staff and research participants apart from enrollment and follow-up data collection visits. In addition, some trials provided alternative LNS formulations or supplementation to both the mother and child. These variations in trial design might affect the effect size of the mortality estimates. Therefore, we conducted a series of sensitivity analyses using the following inclusion and exclusion criteria applied to the data included in the primary analysis: 1) Inclusion of intervention arms with maternal supplement components. 2) Exclusion of intervention arms with LNS formulations that were not milkand peanut-based. 3) Separate comparison of multicomponent arms with LNS groups and comparison groups that contained all the same components without LNS (e.g., LNS +WASH compared with WASH) and exclusion of trials with only passive control arms. Finally, in a secondary analysis, we conducted a comparison of LNS intervention arms with non-LNS supplementation arms. Some trials included products such as micronutrient-fortified wheat–soy blend (WSB), corn–soy blend (CSB), Nutritabs, or micronutrient powders (MNPs). All of these analyses were considered exploratory. Results We identified 18 trials that met the inclusion criteria for the primary, secondary, or sensitivity analyses (Figure 1). The 18 included trials were conducted in 11 countries, enrolling 41,280 children and recording 586 deaths (14–32). A brief description of each trial is included in Table 1 and information about how each trial was included in each set of analyses is shown in Supplemental Table 1. Trials were considered to generally have a low risk of bias, with the exception of blinding of participants owing to the nature of the intervention (Supplemental Table 2 and Supplemental Figure 1). In the primary analysis, which included 13 trials with 34,051 participants, there was a 27% lower risk of all-cause mortality (RR: 0.73; 95% CI: 0.59, 0.89; I2=23.2%, τ2=0.044) (Figure 2). All but 2 of the trials had RR point estimates <1; however, only 1 trial was independently statistically significant (iLiNS-Zinc in Burkina Faso). The point estimate of the RR did not differ in sensitivity analyses in which maternal LNS intervention arms were added (Figure 3; 15 trials; 39,903 participants) or when only peanut +milk-based formulations of LNS intervention were included (Figure 4; 12 trials; 30,776 participants). In a sensitivity analysis in which passive control trials were excluded and multicomponent arms (e.g., WASH +Nutrition) were compared with reference groups with the same components without LNS, the point estimate of the RR shifted toward the null and was no longer statistically significant (0.82; 95% CI: 0.61, 1.10; I2=33.3%, τ2=0.10) (Figure 5;11trials; 23,373 participants). To understand this attenuation of the point estimate, we further examined whether the Water, Sanitation, or Handwashing interventions might have had an effect on mortality (Supplemental Table 3). In the Kenya trial, the WASH +Nutrition group had the lowest risk of death (10/1000), followed by the Active Control group (16/1000) and the Water group (19/1000). The Nutrition-only arm had a mortality rate of 24/1000 and all other groups were between 20/1000 and 30/1000. None of the Water, Sanitation, Handwashing, or combined WASH arms had a mortality rate lower than the Active Control arm, even though all arms except for the passive control arm had active visitation. For comparison, in WASH Benefits Bangladesh, Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 210 Stewart et al. 2107 records identified through database searching 1407 records after duplicates removed 1466 records screened 1349 excluded on the basis of title and abstract 17 duplicate records already included in Das et al. (7) 100 full text reports assessed for eligibility 17 trials (54 reports) included in the Cochrane review 5 ongoing trials 16 trials (76 reports) included in the review 2 ongoing trials with data included in the review (1 report) 9 trials (11 reports) excluded: non RCT, non LNS, malnutrition 7 ongoing trials (9 reports) excluded: no data 3 systematic reviews excluded Das et al. (7) 13 trials included in primary analysis (LNS vs. control) 10 –15 trials included in sensitivity analyses 7 trials included in secondary analysis (LNS vs. non LNS) FIGURE 1 Study flow diagram. LNS, lipid-based nutrient supplement; RCT, randomized controlled trial. none of the Water, Sanitation, or Handwashing arms, or the combined WASH or WASH +Nutrition arms, had a mortality rate lower than the control group (in this case, passive control); the only arm that had lower mortality was Nutrition (only). In the other WASH +Nutrition trial (the SHINE trial in Zimbabwe), the mortality rate was also lowest in the IYCF arm and was not improved by the addition of WASH. In a secondary analysis, we compared LNS interventions with other supplement comparison groups (non-LNS, e.g., WSB++, CSB, MNP, or Nutritabs). The results were not statistically significant (RR: 0.76; 95% CI: 0.37, 1.58), but the point estimate was similar to that of the primary analysis (Supplemental Figure 2; 7 trials; 8681 participants). There was a much smaller sample size and larger degree of heterogeneity in these analyses, however (I2=52.8%, τ2=0.384). Discussion These results suggest that LNS supplementation for a minimum of 6 mo among children aged 6–24 mo may reduce the risk of mortality. The estimated reduction of 27% in all-cause mortality between 6 and 24 mo of age in the primary analysis, based on data from 13 trials with 34,051 children, was robust to the inclusion of interventions combined with maternal LNS supplementation or to the exclusion of alternative formulations of LNS. Our primary analysis included all studies and intervention Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 Meta-analysis of effect of LNS on child mortality 211 TABLE 1 Description of each trial included in the analysis1 Infant supplement Anthropometrics at start of supplementation2 Authors Trial name Country Intervention groups Maternal supplement Age at start Duration Outcome ascertainment description Stunting, % Wasting, % Adu-Afarwuah et al. (14) Ghana Nutributter (20 g/d; 108 kcal/d) Sprinkles powder Nutritabs (MMN) N 6 mo 6 mo Weekly morbidity surveillance in intervention groups 5.4 5.4 Passive control (no intervention) Adu-Afarwuah et al. (15) iLiNS-DYADG Ghana LNS: women received SQ-LNS during pregnancy and for 6 mo postpartum, child received SQ-LNS (20 g/d; 118 kcal/d) Y 6 mo 12 mo Weekly morbidity surveillance, trial hotline for participants —— MMN: women received MMN during pregnancy and 6 mo postpartum, no child supplementation IFA: women received IFA during pregnancy and placebo for 6 mo postpartum, no child supplementation Ashorn et al. (16) iLiNS-DYADM Malawi LNS: women received SQ-LNS during pregnancy and for 6 mo postpartum, child received SQ-LNSs (20 g/d; 118 kcal/d) Y 6 mo 12 mo Weekly morbidity surveillance, trial contact with local hospitals —— MMN: women received MMN during pregnancy and 6 mo postpartum, no child supplementation IFA: women received IFA during pregnancy and placebo for 6 mo postpartum, no child supplementation Becquey et al. (17) PROMIS Burkina Faso SQ-LNS (20 g/d; 118 kcal/d) Active control N 6 mo 12 mo Monthly home visits by research staff — — Bisimwa et al. (18) Democratic Republic of the Congo RUCF: LNS (50 g/d; 275 kcal/d) UNIMIX: MMN-fortified corn–soy blend (70 g/d; 280 kcal/d) N 6 mo 6 mo Medical records or verbal autopsy 18.6 5.7 Historical control Christian et al. (19) JiVitA-4 Bangladesh Plumpy’Doz (23–46 g/d; 125–250 kcal/d) N 6 mo 12 mo Twice-weekly morbidity surveillance 25.4 5.8 Chickpea based LNS (23–46 kg/d; 125–250 kcal/d) Rice-lentil LNS (28–56 g/d; 125–250 kcal/d) WSB++: Wheat–soy blend++ (32–64 g/d; 125–250 kcal/d) Active control Dewey et al. (20) RDNS Bangladesh LNS-LNS: women received SQ-LNS during pregnancy and for 6 mo postpartum, child received SQ-LNS (20 g/d; 118 kcal/d) Y 6 mo 18 mo Home visits every 6 mo by research staff — — (Continued) Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 212 Stewart et al. TABLE 1 (Continued) Infant supplement Anthropometrics at start of supplementation2 Authors Trial name Country Intervention groups Maternal supplement Age at start Duration Outcome ascertainment description Stunting, % Wasting, % IFA-LNS: women received IFA during pregnancy and 3 mo postpartum, child received SQ-LNS (20 g/d; 118 kcal/d) IFA-MNP: women received IFA during pregnancy and 3 mo postpartum, child received MNP IFA-Control: women received IFA during pregnancy and 3 mo postpartum, no child supplementation Hess et al. (21) iLiNS-Zinc Burkina Faso LNS-Zn0: LNS (20 g/d; 118 kcal) containing 0 mg Zn/d and placebo tablet LNS-Zn5: LNS (20 g/d; 118 kcal) containing 5 mg Zn/d and placebo tablet N 9 mo 9 mo Weekly morbidity surveillance; caregiver interview at endline for passive control group 22.4 16.4 LNS-Zn10: LNSs (20 g/d; 118 kcal) containing 10 mg Zn/d and placebo tablet LNS-TabZn5 (20 g/d; 118 kcal) containing 0 mg Zn/d and Zn tablet containing 5 mg Zn/d Passive control (no intervention) Humphrey et al. (22) SHINE (HIV−)3Zimbabwe IYCF: child received SQ-LNS (20 g/d; 118 kcal/d) N 6 mo 12 mo Monthly reports from village health workers —— WASH: family received ventilated improved pit latrine, handwashing stations, soap, chlorine, child play space WASH and IYCF: child received SQ-LNS (20 g/d; 118 kcal/d), family received ventilated improved pit latrine, handwashing stations, soap, chlorine, child play space Active control (standard of care) Huybregts et al. (23) PROMIS Mali SQ-LNS (20 g/d; 118 kcal/d) Active control N 6 mo 18 mo Monthly home visits by research staff — — Iannotti et al. (24) Haiti Nutributter (20 g/d; 108 kcal/d) for 6 mo N 6–11 mo 3–6 mo Monthly trial visit 9.4 2.2 Nutributter (20 g/d; 108 kcal/d) for 3 mo, followed by standard of care Active control (standard of care) (Continued) Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 Meta-analysis of effect of LNS on child mortality 213 TABLE 1 (Continued) Infant supplement Anthropometrics at start of supplementation2 Authors Trial name Country Intervention groups Maternal supplement Age at start Duration Outcome ascertainment description Stunting, % Wasting, % Luby et al. (25) WASH-B Bangladesh Nutrition: child received LNS (20 g/d; 118 kcal/d) N 6 mo 18 mo Annual caregiver interview — — Water: family received chlorine for drinking water Sanitation: family received upgraded latrine, sani-scoop, and child potty Handwashing: family received handwashing stations with soap Water, sanitation, and handwashing: family received all water, sanitation, and hygiene interventions Water, sanitation, handwashing, and nutrition: family received all water, sanitation, and hygiene interventions, child received LNS (20 g/d; 118 kcal/d) Passive control (no intervention) Maleta et al. (26) iLiNS-DOSE Malawi 10 g/d milk LNS (56 kcal/d) 20 g/d milk LNS (118 kcal/d) 20 g/d no-milk LNS (maltodextrin substituted) (118 kcal/d) N 6 mo 12 mo Weekly morbidity surveillance, trial contact with local health providers 29.3 — 40 g/d milk LNS (236 kcal/d) 40 g/d no-milk LNS (maltodextrin substituted) (236 kcal/d) Active control Mangani et al. (27) LCNI-5 Malawi Milk-LNS (54 g/d, 285 kcal/d) Soy-LNS (54 g/d, 276 kcal/d) Corn–soy blend (71 g/d, 284 kcal/d) N 6 mo 12 mo Weekly morbidity surveillance, trial contact with local health clinics 36.7 1.7 Active control Matias et al. (28) Peru LNS: Nutributter (20 g/d, 110 kcal/d) N 6 mo 6 mo Not reported 10.0 0.8 MNP: Sprinkles powder Null et al. (29) WASH-B Kenya Nutrition: child received LNS (20 g/d; 118 kcal/d) Water: family received chlorine for drinking water Sanitation: family received upgraded latrine, sani-scoop, and child potty N 6 mo 18 mo Monthly reports from child health promotors; annual caregiver interview in passive control group —— (Continued) Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 214 Stewart et al. TABLE 1 (Continued) Infant supplement Anthropometrics at start of supplementation2 Authors Trial name Country Intervention groups Maternal supplement Age at start Duration Outcome ascertainment description Stunting, % Wasting, % Handwashing: family received handwashing stations with soap Water, sanitation, and handwashing: family received all water, sanitation, and hygiene interventions Water, sanitation, handwashing, and nutrition: family received all water, sanitation, and hygiene interventions, child received LNS (20 g/d; 118 kcal/d) Passive control (no intervention) Active control Phuka et al. (30) Malawi FS-25: micronutrient-fortified spread (25 g/d; 127 kcal/d) N 6 mo 12 mo Weekly morbidity surveillance — — FS-50: micronutrient-fortified spread (50 g/d; 256 kcal/d) LP: Likuni Phala, micronutrient-fortified maize–soy flour (71 g/d; 282 kcal/d) Prendergast et al. (31) SHINE (HIV+)3Zimbabwe IYCF: child received SQ-LNSs (20 g/d; 118 kcal/d) N 6 mo 12 mo Monthly reports from village health workers —— WASH: family received ventilated improved pit latrine, handwashing stations, soap, chlorine, child play space WASH and IYCF: child received SQ-LNS (20 g/d; 118 kcal/d), family received ventilated improved pit latrine, handwashing stations, soap, chlorine, child play space Active control (standard of care) Smuts et al. (32) South Africa SQ-LNS (20 g/d; 114 kcal/d) SQ-LNS-plus: SQ-LNS with additional MMN, DHA, AA, L-lysine, and phytase (20 g/d; 113 kcal/d) N 6 mo 6 mo Weekly morbidity surveillance 29.5 1.5 Active control 1AA, arachidonic acid; FS, fortified spread; IFA, iron–folic acid; IYCF, infant and young child feeding; LNS, lipid-based nutrient supplement; LP, Likuni Phala; MMN, multiple micronutrients; MNP, multiple micronutrient powder; RUCF, ready-to-use complementary food; SQ-LNS, small-quantity lipid-based nutrient supplement; WASH, water, sanitation, and hygiene; WSB, wheat–soy blend. 2Stunting is defined as length-for-age zscore ≤−2 and wasting is defined as weight-for-length zscore ≤−2. Trials that did not report baseline stunting or wasting prevalences are indicated by a dashed line. 3The SHINE trial was a single trial published in 2 reports [Humphrey et al. (22) and Prendergast et al. (31)] separately by HIV exposure status of the infant. Data were extracted separately from the 2 reports. Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020 Meta-analysis of effect of LNS on child mortality 215 Trial GHANA PROMIS (Burkina Faso) JiVitA-4 RDNS iLiNS-Zinc SHINE (HIV−) PROMIS (Mali) HAITI WASH-B (Bangladesh) iLiNS-DOSE LCNI-5 WASH-B (Kenya) SHINE (HIV+) SOUTH AFRICA Summary Authors Adu-Afarwuah et al. (14) Becquey et al. (17) Christian et al. (19) Dewey et al. (20) Hess et al. (21) Humphrey et al. (22) Huybregts et al. (23) Iannotti et al. (24) Luby et al. (25) Maleta et al. (26) Mangani et al. (27) Null et al. (29) Prendergast et al. (31) Smuts et al. (32) I² = 23, s² = 0.04 Country Ghana Burkina Faso Bangladesh Bangladesh Burkina Faso Zimbabwe Mali Haiti Bangladesh Malawi Malawi Kenya Zimbabwe South Africa Quantity (g/d) 20 20 23–56 20 20 20 20 20 20 10–40 54 20 20 20 LNS n 103 991 3160 815 2435 1955 565 202 1247 1612 422 1705 350 500 16,062 218 Deaths 0 11 17 2 33 20 16 1 8 63 9 29 6 3 Control n 97 1046 1438 864 785 1881 567 191 3758 320 209 6235 348 250 17,989 Deaths 0 5 12 9 25 25 21 0 34 15 8 133 5 2 294 RR (95% CI) — 2.32 (0.81, 6.66) 0.64 (0.31, 1.35) 0.24 (0.05, 1.09) 0.43 (0.25, 0.71) 0.77 (0.43, 1.38) 0.76 (0.40, 1.45) — 0.71 (0.33, 1.53) 0.83 (0.48, 1.45) 0.56 (0.22, 1.42) 0.80 (0.54, 1.19) 1.19 (0.37, 3.87) 0.75 (0.13, 4.46) 0.73 (0.59, 0.89) Weight 0.01 0.06 0.14 0.05 0.09 0.11 0.03 0.01 0.15 0.06 0.02 0.23 0.02 0.02 0.08 0.12 0.18 0.25 0.35 0.50 0.71 1.00 1.40 2.00 3.00 4.00 RR of mortality Favors LNS Favors control FIGURE 2 Effect of LNS with or without other interventions on all-cause mortality in children 6–24 mo of age. The SHINE trial presented results in separate reports for children born to HIV+and HIV−mothers and these have therefore been listed as 2 rows in this figure. LNS, lipid-based nutrient supplement. arms containing LNS and compared those with arms without LNS with a goal of maximizing the sample size and power to detect a difference. In a more restrictive sensitivity analysis in which we excluded passive control arms from the analysis and only contrasted LNS groups with comparison groups that contained all the same components without LNS, the estimated risk reduction was reduced to 18% and was no longer statistically significant. There are a few notable differences in comparing the primary analysis with the latter sensitivity analysis. First, the sensitivity analysis excluded the iLiNS-Zinc trial in Burkina Faso (21) and the Ghana trial (14) completely, which may explain some of the attenuation in effect in this sensitivity analysis. The iLiNS-Zinc trial was the only trial that independently had a significant mortality risk reduction (RR: 0.43; 95% CI: 0.25, 0.71) and it also provided surveillance and treatment of diarrhea and malaria in the intervention groups but not in the passive control group. The second difference was in the contrasts for the 3 WASH +Nutrition trials (WASH Benefits Bangladesh and Kenya and SHINE in Zimbabwe). To isolate the effects of LNS, this sensitivity analysis compared the LNS arm with the Active Control arm in WASH Benefits Kenya and SHINE and the LNS +WASH arm with the WASH arm in all 3 trials. It also excluded the independent water, sanitation, and handwashing arms from the 2 WASH Benefits trials. The difference in the point estimate in this sensitivity analysis compared with the primary analysis did not appear to be due to a protective effect of the WASH interventions. There was no risk reduction in those arms in either of the WASH Benefits trials. Although the combined LNS +WASH arm had a nonsignificantly lower mortality rate than the LNS arm in Kenya, this was not the case in the other 2 trials. The LNS +WASH arm in WASH Benefits Bangladesh had a higher mortality rate than the LNS group. Similarly, in the SHINE trial, the LNS +WASH group had a higher mortality rate than the LNS group. Lastly, this sensitivity analysis included a smaller sample size (23,373 compared with 34,051 in the primary analysis) and fewer deaths (349 compared with 512), which reduced the power to detect differences. Trial GHANA iLiNS-DYADG iLiNS-DYADM PROMIS (Burkina Faso) JiVitA-4 RDNS iLiNS-Zinc SHINE (HIV−) PROMIS (Mali) HAITI WASH-B (Bangladesh) iLiNS-DOSE LCNI-5 WASH-B (Kenya) SHINE (HIV+) SOUTH AFRICA Summary Authors Adu-Afarwuah et al. (14) Adu Afarwuah et al. (15) Ashorn et al. (16) Becquey et al. (17) Christian et al. (19) Dewey et al. (20) Hess et al. (21) Humphrey et al. (22) Huybregts et al. (23) Iannotti et al. (24) Luby et al. (25) Maleta et al. (26) Mangani et al. (27) Null et al. (29) Prendergast et al. (31) Smuts et al. (32) I² = 25, s² = 0.05 Country Ghana Ghana Malawi Burkina Faso Bangladesh Bangladesh Burkina Faso Zimbabwe Mali Haiti Bangladesh Malawi Malawi Kenya Zimbabwe South Africa Quantity (g/d) 20 20 20 20 23–56 20 20 20 20 20 20 10–40 54 20 20 20 LNS n 103 397 243 991 3160 1735 2435 1955 565 202 1247 1612 422 1705 350 500 17,622 Deaths 0 3 5 11 17 5 33 20 16 1 8 63 9 29 6 3 229 Control n 97 800 492 1046 1438 864 785 1881 567 191 3758 320 209 6235 348 250 19,281 Deaths 0 3 15 5 12 9 25 25 21 0 34 15 8 133 5 2 312 RR (95% CI) — 2.02 (0.41, 9.94) 0.67 (0.25, 1.84) 2.32 (0.81, 6.66) 0.64 (0.31, 1.35) 0.28 (0.09, 0.82) 0.43 (0.25, 0.71) 0.77 (0.43, 1.38) 0.76 (0.40, 1.45) — 0.71 (0.33, 1.53) 0.83 (0.48, 1.45) 0.56 (0.22, 1.42) 0.80 (0.54, 1.19) 1.19 (0.37, 3.87) 0.75 (0.13, 4.46) 0.73 (0.60, 0.88) Weight 0.01 0.03 0.02 0.06 0.12 0.07 0.09 0.10 0.03 0.01 0.14 0.05 0.02 0.22 0.02 0.02 0.08 0.12 0.18 0.25 0.35 0.50 0.71 1.00 1.40 2.00 3.00 4.00 RR of mortality Favors LNS Favors control FIGURE 3 Sensitivity analysis including arms with combined maternal +child supplementation. The SHINE trial presented results in separate reports for children born to HIV+and HIV−mothers and these have therefore been listed as 2 rows in this figure. LNS, lipid-based nutrient supplement. Downloaded from https://academic.oup.com/ajcn/article-abstract/111/1/207/5613093 by Tampere university library user on 07 February 2020