Effect of a package of health and nutrition services on sustained recovery in children after moderate acute malnutrition and factors related to sustaining recovery: a cluster-randomized trial

ABSTRACT

Background: Children who recover from moderate acute malnutrition (MAM) have high rates of relapse in the year after nutritional recovery. Interventions to decrease these adverse outcomes are needed to maximize the overall effectiveness of supplemental feeding programs (SFPs).

Objective: We evaluated the effectiveness of a package of health and nutrition interventions on improving the proportion of children who sustained recovery for 1 y after MAM treatment. We further explored factors related to sustained recovery.

Design: We conducted a cluster-randomized clinical effectiveness trial involving rural Malawian children aged 6–62 mo who were enrolled on discharge from an SFP for MAM. We enrolled 718 children at 10 control sites and 769 children at 11 intervention sites. In addition to routine health and nutrition counseling, the intervention group received a package of health and nutrition interventions that consisted of a lipid nutrient supplement, deworming medication, zinc supplementation, a bed net, and malaria chemoprophylaxis. A survival analysis was used to determine the effectiveness of the intervention as well as to identify factors associated with sustained recovery.

Results: Of 1383 children who returned for the full 12-mo follow-up period, 407 children (56%) and 347 children (53%) sustained recovery in the intervention and control groups, respectively. There was no significant difference in relapse-free survival curves between the treatment and control groups (P = 0.380; log-rank test). The risk factors for relapse or death after initial recovery were a smaller midupper arm circumference on SFP admission (P = 0.01) and discharge (P < 0.001), a lower weight-for-height z score on discharge (P < 0.01), and the receipt of ready-to-use supplementary food as opposed to ready-to-use therapeutic food during treatment (P < 0.05).

Conclusion: The provision of a package of health and nutrition services in addition to traditional SFP treatment has no significant effect on improving sustained recovery in children after treatment of MAM. This trial was registered at clinicaltrials.gov as NCT02351687.

INTRODUCTION

Children with moderate acute malnutrition (MAM) are generally treated for several weeks in a community-based supplementary feeding program (SFP) that provides one of a variety of supplementary foods (1). Children are generally discharged from SFPs as recovered after achieving an anthropometric threshold or after receiving food for a fixed duration of time (2).

Children who have been systematically followed after discharge have shown that relapse back to MAM, the development of severe acute malnutrition (SAM), and other poor outcomes are very common. A study in Niger followed children who were successfully discharged from an SFP for 6 mo and showed that 20% of children relapsed during that time (3). A study in Malawi showed that only 63% of children who were successfully treated for MAM remained well nourished for 12 mo (4). Other studies have also shown high relapse rates in children who were discharged after treatment of SAM (5, 6). A more recent study from Burkina Faso of a mixed population of children (90% with MAM and 10% with SAM) reported relapse rates that were closer to 15%, although greater than one-third of those children were lost to follow-up (LTFU), possibly suggesting an even higher rate of relapse or death (7).

In addition to high relapse rates, almost all studies reported that common childhood illnesses are prevalent in individuals who relapse or die after initial recovery (4, 5). Illnesses such as fever, cough, malaria, and diarrhea occur frequently during the initial 3 mo after discharge and decrease thereafter (5). These occurrences suggest that the same common infectious diseases that often afflict children in resource-limited settings (8) may be associated with relapses in children recovering from acute malnutrition.

In this study, we assessed whether a package of simple and affordable health and nutrition interventions that were added after achieving anthropometric criteria for nutritional recovery from MAM could improve the proportion of children who sustained recovery for 1 y after treatment. In addition, we used this cohort to explore factors that are related to sustained recovery, which remains an elusive but important goal in supplemental feeding programs.

METHODS

Subjects and setting

Children aged 6–62 mo who had recovered from MAM as defined by a midupper arm circumference (MUAC) ≥12.5 cm without bipedal edema (9) were recruited from 21 rural SFP clinics in southern Malawi from April 2014 to June 2015. Children were excluded if they had a chronic debilitating illness or had a history of a peanut, milk, or soy allergy. Children were also excluded if they had received therapy for acute malnutrition ≤1 mo before admission into the SFP to focus on the sustained recovery from an initial discrete episode of MAM. Children whose MUAC dropped below 11.5 cm or who developed edema during the initial treatment of MAM were also excluded from the study because they were considered to have progressed to SAM and were treated as such.

The study communities predominantly consisted of impoverished, subsistence-farming families who were living in mud and thatch homes. Maize is the staple crop in the region and is gathered 1 time/y after a single rainy season that generally stretches from December through March annually; this season is also notable for high rates of diarrhea and malaria along with substantial increases in food insecurity and rates of acute malnutrition.

Study design

The study was a cluster-randomized, controlled, clinical effectiveness trial (clinicaltrials.gov; NCT02351687). Randomization was performed across clinic sites rather than at the individual level to minimize the risk of sharing and cross contamination. Sites were randomized to individual or control groups on the basis of historical patterns of SFP enrollment at each site in pairs such that, e.g., the largest site was randomized to the intervention or control arm, and the second-largest site was made the opposite; likewise, e.g., the third-largest site was randomized, and the fourth-largest site was assigned the opposite. The primary outcome was the proportion of children who sustained recovery, which was defined as maintaining an MUAC ≥12.5 cm without bipedal edema for 12 mo after initial recovery from MAM.

Participants and field researchers were not blinded to the allocation of groups because it was inherently evident whether children had received the package of interventions. However, after each child’s anthropometric and clinical data were entered into the computer database, the group allocation was blinded during the outcome assessment and statistical analyses. Unblinding did not occur until after the end of the trial and all statistical analyses had been completed.

When calculating the sample size, a correction factor was used to adjust for any implementation or populations differences in sites. An intracluster correlation coefficient of 0.007 and a CV of cluster size of 0.65 were calculated from pilot data and similar studies that were conducted previously at the same study sites (2, 4). A power of 80% and α = 0.05 were used to identify a sample size that was sufficient to detect a 10–percentage point difference in the proportion of children who sustained recovery between the control and intervention groups. With the estimation that 63% of children would sustain recovery for 12 mo without any additional interventions (4), a minimum mean of 58 participants/cluster was calculated to be necessary across the 21 clusters. The sample size was calculated with the use of Stata Version 13.0 software (StataCorp LP). An additional 279 participants were ultimately enrolled as a buffer to account for the possibility of a high LTFU rate.

To identify whether an association existed between immune function and sustained recovery after the treatment of MAM, serum complement component 3 (C3) [as a proxy indicator for immune recovery (10)] was measured in a random sample of 145 children at the time of SFP discharge and 1 mo later. Whole blood was drawn into heparinized tubes and kept cold in an insulated plastic box with freezer packs in the field. On returning from field clinics, the blood was centrifuged at 2000 × g for 10 min at <25°C to isolate the plasma, which was frozen at −80°C. Samples were transferred frozen to the Core Laboratory at St. Louis Children’s Hospital for the analysis of serum complement C3 concentrations (Roche Cobas 6000, Roche Diagnostics, Indianapolis, Indiana).

Intervention

The control group received nutrition counseling at the time of discharge from an SFP, which consisted of messages regarding proper complementary feeding, caretaker recognition of common childhood illnesses, and appropriate health-seeking behaviors. The treatment group received the same counseling plus 5 additional interventions as follows: 1) 40 g/d of a lipid-based nutrient supplement (LNS) that provided 200 kcal and 1 Recommended Dietary Allowance of almost all micronutrients for 8 wk on discharge from an SFP (Table 1); the LNS consisted of 28% peanut paste, 18% nonfat dry skimmed milk powder, 24.5% palm oil, 21.2% sugar, 6.8% custom micronutrient mix, and 1.5% emulsifier; 2) a single dose of albendazole (200 mg for subjects <2 y old and 400 mg for subjects ≥2 y old) for deworming at the time of discharge from an SFP; 3) a 14-d course of 20 mg ZnSO₄ starting at the time of discharge from an SFP; this dose has been recommended after an episode of diarrhea and has also been shown to decrease the progression of environmental enteric dysfunction (12); 4) a single insecticide-treated bed net to reduce risk of malaria at the time of discharge from an SFP; and 5) sulfadoxine-pyrimethamine for malaria chemoprophylaxis (13) at a monthly dose of ∼25 mg/kg (sulfadoxine component) during the peak of the rainy season (December to February) because most adverse outcomes in children who recover from MAM occur during this time (4).

TABLE 1.

Nutritional content of the LNS that was provided to participants in the intervention group¹

	LNS	IOM RDA (1–3 y of age) (11)
Total weight, g	40.0	—
Energy, kcal	216.5	—
Protein, g	5.3	13.0
Fat, g	15.2	—
Minerals
Biotin, mg	11.1	8.0
Calcium, mg	310.1	500.02
Copper, mg	0.4	0.3
Iodine, μg	98.4	90.0
Iron, mg	8.2	1.3
Magnesium, mg	27.6	80.0
Manganese, mg	1.7	1.22
Phosphorus, mg	541.1	460.0
Potassium, mg	368.2	3000.02
Selenium, μg	24.4	20.0
Zinc, mg	3.5	0.9
Vitamins
Folic acid, μg	213.0	150.0
Niacin, mg	8.2	6.0
Pantothenic acid, mg	2.7	2.02
Riboflavin, mg	0.7	0.5
Thiamin, mg	0.6	0.5
Vitamin A, μg	452.1	300.0
Vitamin B-6, mg	0.6	0.5
Vitamin B-12, μg	1.2	0.9
Vitamin C, mg	36.0	15.0
Vitamin D, μg	11.9	5.02
Vitamin E, mg	9.2	6.0
Vitamin K, μg	34.1	30.02

¹IOM, Institute of Medicine; LNS, lipid nutrient supplement; RDA, Recommended Dietary Allowance.

²Adequate Intakes. These values are the approximations of the needed nutrient intakes when not enough data exists to create RDAs.

These interventions have all individually been proven safe, effective, and affordable in this context to improve the overall health of children but are often not universally implemented because of resource and logistical limitations (12–15). Thus, the package was provided specifically to this high-risk population to increase the likelihood of sustaining recovery especially during the first few months after discharge from an SFP.

Subject participation

At enrollment and at each subsequent visit, nutrition researchers and senior pediatric nurses conducted standard anthropometric measurements and assessed children for edema. Weight was measured with the use of an electronic scale to the nearest 5 g; length was measured with the use of a rigid length board to the nearest 0.1 cm; and MUAC was measured with a standard insertion tape to the nearest 0.1 cm. Edematous malnutrition (kwashiorkor) was diagnosed by examining for bilateral pitting edema.

After the enrollment criteria were confirmed, informed consent was obtained from all caregivers. Information on demographic characteristics, health history, and household food insecurity was collected. Health-history questions included caregiver-observed symptoms of illness during the previous 2 wk, immunization status, the use of any nutritional supplements, the use of malaria prophylaxis, the timing of the most recent deworming, and the use of a bed net. Household food security was assessed with the use of the validated 9-item Household Food Insecurity Access Scale (16).

All children were reassessed with follow-up visits at 1, 3, 6, and 12 mo after discharge from an SFP. Additional monthly visits were scheduled during the height of the rainy season (December through February) when malaria prophylaxis was also provided at the intervention sites. All children from both intervention and control groups were scheduled to return for 3 visits during the rainy season, thereby avoiding any imbalance in follow-up visits between control and intervention arms. The same procedures for collecting anthropometric measurements and health-history questions at the 1-, 3-, 6-, and 12-mo follow-up visits were also used during the rainy season follow-up visits. Caregivers were also educated that they could bring their children for additional evaluations at any time during the course of the study if they were concerned about their child’s nutritional status or if they had been referred by a community health worker.

Each child was classified as follows: having sustained recovery, which was defined as having an MUAC ≥12.5 cm at every follow-up visit for 12 mo; having relapsed to MAM, which was defined as having an MUAC of 11.5–12.4 cm at any point during the follow-up period; having developed SAM, which was defined as having an MUAC <11.5 cm or bipedal edema (kwashiorkor) at any point during the follow-up period; having died; or having been LTFU, which was defined as defaulting on a scheduled visit and never returning. Poor outcomes included relapsing to MAM, developing SAM, or death. If a child experienced ≥2 poor outcomes over the course of the follow-up period, the most severe category was assigned as the final outcome.

The definition of sustained recovery required consistently maintaining an MUAC ≥12.5 cm (without edema) throughout the entire time period of the study. This criterion included regularly scheduled visits, the scheduled visits during the rainy season, as well as any unscheduled visits to the nutrition clinic. If a child came to the clinic for an unscheduled extra follow-up visit or a scheduled rainy season visit that occurred before the end of a full 1-, 3-, 6-, or 12-mo follow-up period and did not have a MUAC ≥12.5 cm without edema, he or she was considered to have relapsed for that time period. For example, if a child relapsed to MAM at a rainy season visit that fell between the 3- and 6-mo follow-up visits, his or her outcomes would be classified as sustained recovery for 3 mo but relapsed to MAM by 6 mo.

Ethical oversight

The study was approved by the University of Malawi’s College of Medicine Research and Ethics Committee and Washington University’s Human Research Protection Office. Permission to conduct the study was obtained by each site’s district health officer, district nutritionist or both.

Statistical analyses

All data were double entered into an Access database version 2013 (Microsoft Corp.) and verified against original paper data forms when discrepancies were identified. Anthropometric indexes were based on the WHO’s 2006 Child Growth Standards (17) and were calculated with the use of WHO Anthro software version 3.2.2 (WHO). Rates of MUAC and length gain were calculated in millimeters per day, and weight gain was calculated in grams per kilogram per day. Child characteristics and final outcomes between control and intervention groups were compared with the use of a chi-square test for dichotomous outcomes, and Student’s t test was used for comparing continuous variables. Correction factors were included to account for clustering at the health clinic level. P < 0.05 was considered statistically significant. Relapse-free survival curves were developed with the use of the Kaplan-Meier method and were compared with the use of the log-rank test.

A Cox proportional model for a multivariate analysis was used to identify risk factors for a failure to sustain recovery for the full 12-mo follow-up period. The outcome variable in the Cox regression model was either a poor outcome (defined as relapsed to MAM, developed SAM, or died) compared with a sustained recovery or unknown outcome (LTFU). The assumption of the proportionality of hazards was checked with the use of Schoenfeld residuals. None of the covariates that were included in the model violated the proportional hazards assumption. Covariates that were included in the full regression model were based on identified risk factors for poor outcomes after recovery from acute malnutrition in previous studies. These covariates included MUAC and weight-for-height z score (WHZ) on SFP admission and discharge (2, 4, 6, 7), symptoms of illness during the 2 wk before SFP admission (6), mother’s HIV status (4), household food security (Household Food Insecurity Access Scale score) (4), and whether the mother was alive (4). Whether the child was in the intervention or control group was also included because this was the primary variable of interest. Additional control variables included sex, age, the type of food received during SFP treatment [whey ready-to-use supplementary food (RUSF), soy RUSF, or ready-to-use therapeutic food (RUTF)], admission during the harvest season (April through August), and any services the child had already received that were also provided in the intervention such as whether the child slept under a bed net, whether the child received malaria prophylaxis, whether the child was dewormed, and whether the child received additional supplements during the study. All statistical analyses were conducted with the use of Stata version 13.0 software (StataCorp LP).

RESULTS

Between April 2014 and June 2015, 1497 children recovered from MAM at 21 SFP study clinics and were enrolled in the study. During the analysis, 10 children were excluded because of incorrect enrollment, which left 1487 children for the final analysis with 718 children at 10 control sites and 769 children at 11 intervention sites (Figure 1). The majority of participants (57%) were enrolled toward the end of the lean season (March through May). Completed follow-up visits were fairly consistent across the calendar year and ranged from 93% to 98% with the lowest percentage occurring during the rainy season because of difficult traveling conditions (Supplemental Table 1). The mean number of follow-up visits was 3.8 and 3.7 for the intervention and control groups, respectively. A few characteristics differed between the control and intervention groups, which were later controlled for in the regression models (Table 2).

FIGURE 1.

Flow of participants through the cluster-randomized controlled clinical trial. LTFU, lost to follow-up; MAM, moderate acute malnutrition; SAM, severe acute malnutrition; SFP, supplementary feeding program.

TABLE 2.

Enrollment characteristics of intervention and control groups¹

	Intervention (n = 769)	Control (n = 718)
Total clusters (clinic sites), n	11	10
Sex, F, n (%)	472 (61)	435 (61)
Age, mo	17.01 ± 9.332	16.43 ± 9.08
On admission to initial treatment in an SFP
Type of treatment food received, n (%)
Whey RUSF	153 (20)**	105 (15)
Soy RUSF	155 (20)	128 (18)
RUTF	460 (60)**	485 (68)
MUAC, cm	12.10 ± 0.26	12.08 ± 0.27
WHZ	−1.77 ± 0.66	−1.76 ± 0.73
HAZ	−2.73 ± 1.24	−2.62 ± 1.37
Primary caretaker is mother, n (%)	736 (97)	682 (97)
Mother alive, n (%)	756 (99)*	696 (98)
Father alive, n (%)	735 (97)	679 (96)
Mother known to be HIV positive, n (%)	114 (18)*	138 (22)
Fever in 2 wk before admission, n (%)	486 (67)	479 (71)
Diarrhea in 2 wk before admission, n (%)	468 (63)	459 (66)
Admission during harvest season (Apr–Aug), n (%)	207 (27)**	241 (34)
HFIAS score	8 ± 6	10 ± 6
On discharge from initial treatment in an SFP
MUAC, cm	12.78 ± 0.27	12.79 ± 0.27
MUAC gain, mm/d	0.29 ± 0.21	0.30 ± 0.21
WHZ	−0.94 ± 0.73	−0.88 ± 0.74
WHZ change	0.83 ± 0.51	0.88 ± 0.60
Weight gain, g · kg–1 · d–1	2.77 ± 1.90	2.98 ± 2.37
Length gain, mm/d	0.30 ± 0.22	0.27 ± 0.22
Time to recovery, d	31.50 ± 20.60	31.92 ± 20.64
Child sleeps under bed net, n (%)	463 (60)***	584 (81)
Child takes malaria prophylaxis, n (%)	46 (6)	51 (7)
Child takes any supplements, n (%)	390 (51)***	459 (64)
Child received deworming medication in past month, n (%)	122 (17)**	159 (24)

¹*P < 0.05, **P < 0.01, ***P < 0.001. HAZ, height-for-age z score; HFIAS, Household Food Insecurity Access Scale; MUAC, midupper arm circumference; RUSF, ready-to-use supplementary food; RUTF, ready-to-use therapeutic food; SFP, supplementary feeding program; WHZ, weight-for-height z score.

²Mean ± SD (all such values).

Of the 1487 children who were included in the final analysis, 754 children (51%) sustained recovery throughout the 12-mo follow-up period, whereas 541 children (36%) relapsed to MAM, 73 children (5%) developed SAM, 15 children (1%) died, and 104 children (7%) were LTFU (Table 3). Many children experienced multiple relapses; of those who relapsed to MAM only, 26%, 10%, and 5% of children relapsed 2, 3, and ≥4 times, respectively. In addition, of those who developed SAM, 69% of children also relapsed to MAM ≥1 time. Children who relapsed to MAM multiple times required longer treatment of those relapses during the follow-up period than did children who relapsed only once (P < 0.001). Furthermore, the MUAC dropped significantly lower in children who relapsed to MAM multiple times than in children who relapsed only once (P < 0.001). Risk of relapse or death was highest during the period immediately after discharge from an SFP with ∼50% of all relapses (to either MAM or SAM) occurring within the first 3 mo of initial recovery from MAM (Figure 2).

TABLE 3.

Comparison of primary outcomes from SFP discharge to 1, 3, 6, and 12 mo of follow-up between intervention and control groups¹

	Follow-up, mo
	0–1			0–3			0–6			0–12
	Intervention (n = 769)	Control (n = 718)	P	Intervention (n = 769)	Control (n = 718)	P	Intervention (n = 769)	Control (n = 718)	P	Intervention (n = 769)	Control (n = 718)	P
Sustained recovery, n (%)	604 (78)	531 (74)	0.038	530 (69)	455 (63)	0.024	491 (64)	421 (59)	0.039	407 (53)	347 (48)	0.076
Relapsed to MAM, times, n (%)	153 (20)	161 (22)	0.233	209 (27)	215 (30)	0.238	230 (30)	234 (33)	0.265	281 (37)	260 (36)	0.895
1	147 (19)	156 (22)	0.212	176 (23)	183 (25)	0.242	163 (21)	167 (23)	0.339	175 (23)	149 (21)	0.349
2	6 (1)	5 (1)	0.851	30 (4)	32 (4)	0.592	51 (7)	55 (8)	0.441	63 (8)	76 (11)	0.113
3	0 (0)	0 (0)	NA	3 (0.4)	0 (0)	0.094	14 (2)	12 (2)	0.826	29 (4)	24 (3)	0.656
≥4	0 (0)	0 (0)	NA	0 (0)	0 (0)	NA	2 (0.3)	0 (0)	0.172	14 (2)	11 (2)	0.665
Developed SAM, n (%)	6 (1)	13 (2)	0.077	14 (2)	21 (3)	0.161	18 (2)	24 (3)	0.244	27 (4)	46 (6)	0.010
Died, n (%)	1 (0.1)	1 (0.1)	0.961	4 (0.5)	2 (0.3)	0.463	7 (1)	2 (0.3)	0.117	13 (2)	2 (0.3)	0.007
Death only	0 (0)	1 (0.1)	0.301	2 (0.3)	2 (0.3)	0.945	4 (1)	2 (0.3)	0.463	7 (1)	2 (0.3)	0.117
Relapse then death	1 (0.1)	0 (0)	0.334	2 (0.3)	0 (0)	0.172	3 (0.4)	0 (0)	0.094	6 (1)	0 (0)	0.018
LTFU, n (%)	5 (1)	12 (2)	0.064	12 (2)	25 (3)	0.018	23 (3)	37 (5)	0.034	41 (5)	63 (9)	0.009

¹Intracluster correlations for the primary outcome sustained recovery were 0.001, 0.000, 0.005, and 0.009 for 1, 3, 6, and 12 mo, respectively. LTFU, lost to follow-up; MAM, moderate acute malnutrition; NA, not applicable; SAM, severe acute malnutrition; SFP, supplementary feeding program.

FIGURE 2.

Percentage of total relapses to MAM or SAM for control and intervention groups by number of months from initial SFP discharge. MAM, moderate acute malnutrition; SAM, severe acute malnutrition; SFP, supplementary feeding program.

Of the 1383 children who remained in the study for the full 12 mo of follow-up, the proportion of children who sustained recovery was higher in the intervention group [n = 407 (56%) and 347 (53%) for the intervention and control groups, respectively] (Table 3). However, an analysis with the use of the Kaplan-Meier method showed no significant difference between the relapse-free survival curves for the intervention and control groups (P = 0.380; log-rank test) (Figure 3). The rate of experiencing a poor outcome (death, relapse to MAM, or development of SAM) in the children who completed the 12-mo follow-up was 41.7 and 42.9/100 children-years for the intervention and control groups, respectively.

FIGURE 3.

Relapse-free survival curves for 12 mo after supplementary feeding program discharge for control and intervention groups that were estimated with the use of the Kaplan-Meier method. Relapse-free survival was defined as free from relapse to moderate acute malnutrition, the development of severe acute malnutrition, or death. Difference between intervention and control group curves, P = 0.380 (log-rank test).

Secondary outcomes, including linear growth and illness during the 12-mo follow-up period, were similar across both intervention and control groups. The intervention package, which included a bed net and malaria chemoprophylaxis during the rainy season, did not result in a significant reduction in relapses during the rainy season; in the intervention and control groups, 547 children (71%) compared with 487 children (68%), respectively, sustained recovery during the rainy season (P = 0.167); 156 children (20%) compared with 148 children (21%), respectively, relapsed to MAM (P = 0.876); and 7 children (1%) compared with 13 children (2%) developed SAM (P = 0.132), respectively (data not shown in tables or figures).

When controlling for baseline characteristics and other possible confounders in a Cox regression model, children in the intervention group were more likely to sustain recovery throughout the follow-up period (HR: 0.82; 95% CI: 0.74, 0.91; P < 0.001) (Table 4). In addition to being in the control group, the factors that had a protective effect against relapse or death were having a larger MUAC on SFP admission (HR: 0.95; 95% CI: 0.92, 0.99; P = 0.01), having a larger MUAC on SFP discharge (HR: 0.94; 95% CI: 0.91, 0.97; P < 0.001), and having a higher WHZ on discharge (HR: 0.79; 95% CI: 0.67, 0.93; P < 0.01). Children who received RUSF as opposed to RUTF during treatment had a higher likelihood of experiencing relapse and death after initial recovery [HRs: 1.26 (95% CI: 1.04, 1.53; P < 0.05) and 1.22 (95% CI: 1.01, 1.47; P < 0.05) for whey RUSF and soy RUSF, respectively] (18).

TABLE 4.

Cox regression model exploring risk factors for experiencing a poor outcome (relapse to MAM, develop SAM, or death) compared with sustained recovery during 12 mo of follow-up after discharge from an SFP¹

Risk factor	Adjusted HR (95% CI)	P
Study arm (reference: control group)	0.82 (0.74, 0.91)	<0.001
Female (reference: male)	1.01 (0.85 1.19)	0.975
Age, mo	0.99 (0.99, 1.01)	0.655
Received whey RUSF during an SFP (reference: received RUTF)	1.26 (1.04, 1.53)	0.019
Received soy RUSF during an SFP (reference: received RUTF)	1.22 (1.01, 1.47)	0.041
MUAC on SFP admission, mm	0.95 (0.92, 0.99)	0.010
WHZ on SFP admission	1.17 (0.95, 1.42)	0.132
HAZ on SFP admission	0.97 (0.91, 1.04)	0.435
Fever in 2 wk before SFP admission (reference: no fever in 2 wk before SFP admission)	0.91 (0.77, 1.07)	0.243
Diarrhea in 2 wk before SFP admission (reference: no diarrhea in 2 wk before SFP admission)	0.90 (0.77, 1.04)	0.153
HFIAS score	0.99 (0.98, 1.01)	0.590
MUAC on SFP discharge, mm	0.94 (0.91, 0.97)	<0.001
WHZ on SFP discharge	0.79 (0.67, 0.93)	0.005
Admitted to an SFP during harvest season from April through August (reference: admitted outside of harvest season, from Jan through March and September through December)	1.06 (0.87, 1.29)	0.62
Mother known to be HIV positive (reference: mother known to be HIV negative or unknown status)	0.94 (0.77, 1.13)	0.538
Child sleeps under bed net (reference: child does not sleep under bed net)	0.94 (0.74, 1.22)	0.670
Child receives malaria prophylaxis (reference: child does not receive malaria prophylaxis)	1.36 (0.92, 2.00)	0.119
Child dewormed in past month (reference: child was not dewormed in past month)	0.97 (0.78, 1.22)	0.813
Child receives supplements (reference: child does not receive supplements)	0.98 (0.84, 1.14)	0.788

¹Cox regression model was used with cluster-adjusted robust SEs to account for clustering at the clinic level. The outcome variable was defined as experiencing a relapse (to either SAM or MAM) or death compared with sustaining recovery (alive and maintaining an MUAC >12.4 cm). The coefficients in the Cox regression were related to the hazard. An HR >1.00 indicated an increased hazard for relapsing or death, whereas an HR <1.00 indicated a protective effect against relapsing or death during the 12-mo follow-up period. All variables listed are original variables that were included in the full model; no variables were removed. HAZ, height-for-age z score; HFIAS, Household Food Insecurity Access Scale; MAM, moderate acute malnutrition; MUAC, midupper arm circumference; RUSF, ready-to-use supplementary food; RUTF, ready-to-use therapeutic food; SAM, severe acute malnutrition; SFP, supplementary feeding program; WHZ, weight-for-height z score.

Although illness before SFP admission and the use of supplements were associated with sustained recovery in the univariate analysis (Supplemental Tables 2 and 3), no significance was shown after controlling for other variables in Cox regression models. Cross-sectional logistic regression modeling was also performed; when controlling for baseline characteristics with the use of this type of analysis, children were more likely to sustain recovery throughout the follow-up period (OR: 1.40; 95% CI: 1.06, 1.85; P < 0.05) (Supplemental Table 4).

Of the 145 children who provided blood samples that were taken at the time of SFP discharge and 1 mo after discharge, nearly all serum complement C3 concentrations (96%) were within the normal range (80–160 mg/dL).

DISCUSSION

In this cluster-randomized, controlled, clinical trial, we showed that the provision of a package of basic health and nutrition interventions to children who were recovering from MAM did not have a significant increase in the proportion who sustained recovery during 1 y of follow-up.

Only 53% of all children who were enrolled in the study sustained recovery from MAM for 1 y after SFP treatment. This percentage is lower than in our previous work in which we observed 63% of children sustaining recovery (4). The difference could be attributed to the fact that the previous study defined relapse as having both an MUAC <12.5 cm and WHZ <−2 (4), whereas the current study used only the MUAC as the operational definition for relapse. The previous study also used the WHZ as the criterion for SFP admission and discharge, whereas the current study enrolled participants after an SFP that used the MUAC criterion. Because of the increasing use of the MUAC for admission and discharge in the management of acute malnutrition and the fact that MUAC identifies younger children and children at higher risk of adverse outcomes, the present study’s operational definition and subsequent results are likely to be more reflective of the long-term outcomes that are expected from children with MAM (19).

We observed a diversity of poor outcomes in children who did not sustain recovery. Some children experienced one short, mild episode of moderate malnutrition and quickly recovered after re-enrollment in an SFP to remain free from MAM or SAM thereafter. Other children repeatedly relapsed with more severe episodes of acute malnutrition that required a long treatment without any sustained recovery. These vastly different trajectories highlight the fact that, although children in SFPs are all classified with the same type and severity of malnutrition (i.e., MAM), not all children with MAM are at same risks of poor short- and long-term outcomes. This result suggests that a uniform approach for treating all children with MAM may not be best for ensuring that all of them reach sustained recovery.

Approximately 1% of all children died during the follow-up period, which was roughly in alignment with the expected annual mortality rate for children between 1 and 5 y of age in Malawi (20). The proportion of deaths at the end of 12 mo was higher in the intervention group (2%) than in the control group (0.3%) (Table 3). However, note that the randomization of individual sites as intervention or control sites did lead to a particularly large site (Muloza, Mulanje District) that was located adjacent to the border with Mozambique being assigned to the control arm. Because of the inability of village health workers and our field team to cross the border legally to follow up on subjects who missed clinic visits, we observed an especially high LTFU rate from this clinic, which was reflected in the control arm overall. Similarly, the identification of children who died was limited because most deaths were only identified by home visits that were conducted to follow defaulters.

Several risk factors for relapse or death after initial MAM recovery were identified that may provide insight for future interventions. The strongest predictors of relapse or death were inferior anthropometric measurements during SFP treatment. This result supports previous findings that the severity of malnutrition at admission to feeding programs is linked to increased risk of mortality and relapse in children after SAM treatment (6) or MAM treatment (2, 4). Children with more severe malnutrition often have more comorbidities (21). It is likely that children who present with more severe malnutrition (i.e., lower MUAC) have additional underlying biological deficiencies that take longer to recover than can be identified by simple anthropometric measures. Although most SFP protocols provide the same treatment to children with MAM regardless of the MUAC, our results suggest that treatment and follow-up procedures may benefit from individualization.

A higher discharge MUAC and WHZ were also identified as being protective against relapse and mortality after SFP discharge. We previously showed that higher discharge anthropometric measures were more important than the duration of treatment in sustaining recovery (2) because children may be more resilient when exposed to new infections that might otherwise precipitate relapse (22). In the current study, our findings corroborate the potential benefit of treating high-risk children with MAM to higher anthropometric targets to reduce relapse and mortality after discharge. However, this treatment would certainly increase the cost, which is an essential factor to consider before any programmatic change (2).

We also observed that children who received RUTF as opposed to RUSF during the initial treatment of MAM were less likely to relapse or die after initial recovery. RUTF is designed to meet the nutritional needs of children who are suffering from SAM. Although the exact nutritional needs of children with MAM are less clearly defined, they presumably lie somewhere between those required for recovery from SAM and those of healthy children (23, 24). It is possible that the higher nutrient content in the RUTF led to a more complete physiologic recovery, which, in turn, lowered risk of relapse.

Approximately 50% of all relapses occurred within the first 3 mo of discharge from an SFP; presumably, the underlying physiologic and social factors that led to the acute malnutrition in the first place may not have been fully reversed by the time of SFP discharge, which left the child susceptible to relapse. Previous studies have shown that the immune function is often compromised during acute malnutrition (10, 25, 26), and these deficiencies may linger beyond traditional anthropometric recovery (27). However, in our population, we did not find abnormal serum complement C3 concentrations at the time of recovery, thereby indicating normal function in at least this single immunologic cascade.

This study was conducted in a population in whom seasonal spikes in household food insecurity, chronic poverty, environmental enteric dysfunction, and a pervasively high burden of infectious diseases are widespread at the genesis of MAM. Thus, these results may not be generalizable beyond these conditions such as in emergency settings, various cultural contexts, and higher-income populations. A key limitation of the study is that the time between follow-up visits may have led to an underestimation of the relapse rate because it is possible that some children relapsed and recovered in the interval between visits. We were not able to measure compliance with the instructed use of the bed nets and LNS that was provided in the intervention package.

The inability to ascertain with certainty the clinical status of children who were LTFU may have led to an underestimation of relapse and mortality rates, especially at sites with a large proportion of subjects coming from Mozambique as previously described. Finally, our assessment of immune recovery by measuring only the serum C3 concentration, although practical, did not look broadly for immunologic compromise. The overall status of cell-mediated immunity in this population was not thoroughly investigated.

In conclusion, the failure to achieve a significant difference in relapse-free survival curves and the high proportion of children who failed to sustain recovery proves the package of interventions are not sufficient to bring about a drastic reduction in relapse rates after recovery from MAM. Further directed studies are warranted to identify the biological and sociologic differences between children who sustain recovery and those who experience varying degrees of poor outcomes after MAM treatment to better refine SFP protocols and the additional services that are provided that can further reduce relapse rates. Because of the global burden of MAM, generating such evidence is key to identifying the most effective treatment protocols to improve both the short- and long-term outcomes of children who recover from MAM.

ABBREVIATIONS

C3

complement component 3

LNS

lipid-based nutrient supplement

LTFU

lost to follow-up

MAM

moderate acute malnutrition

MUAC

midupper arm circumference

RUSF

ready-to-use supplementary food

RUTF

ready-to-use therapeutic food

SAM

severe acute malnutrition

SFP

supplementary feeding program

WHZ

weight-for-height z score