Effect of Short-Term Supplementation with Ready-to-Use Therapeutic Food or Micronutrients for Children after Illness for Prevention of Malnutrition: A Randomised Controlled Trial in Uganda

Background Globally, Médecins Sans Frontières (MSF) treats more than 300,000 severely malnourished children annually. Malnutrition is not only caused by lack of food but also by illnesses and by poor infant and child feeding practices. Breaking the vicious cycle of illness and malnutrition by providing ill children with nutritional supplementation is a potentially powerful strategy for preventing malnutrition that has not been adequately investigated. Therefore, MSF investigated whether incidence of malnutrition among ill children <5 y old could be reduced by providing a fortified food product or micronutrients during their 2-wk convalescence period. Two trials, one in Nigeria and one in Uganda, were conducted; here, we report on the trial that took place in Kaabong, a poor agropastoral region of Karamoja, in east Uganda. While the region of Karamoja shows an acute malnutrition rate between 8.4% and 11.5% of which 2% to 3% severe malnutrition, more than half (58%) of the population in the district of Kaabong is considered food insecure. Methods and Findings We investigated the effect of two types of nutritional supplementation on the incidence of malnutrition in ill children presenting at outpatient clinics during March 2011 to April 2012 in Kaabong, Karamoja region, Uganda, a resource-poor region where malnutrition is a chronic problem for its seminomadic population. A three-armed, partially-blinded, randomised controlled trial was conducted in children diagnosed with malaria, diarrhoea, or lower respiratory tract infection. Non-malnourished children aged 6 to 59 mo were randomised to one of three arms: one sachet/d of ready-to-use therapeutic food (RUTF), two sachets/d of micronutrient powder (MNP), or no supplement (control) for 14 d for each illness over 6 mo. The primary outcome was the incidence of first negative nutritional outcome (NNO) during the 6 mo follow-up. NNO was a study-specific measure used to indicate progression to moderate or severe acute malnutrition; it was defined as weight-for-height z-score <−2, mid-upper arm circumference (MUAC) <115 mm, or oedema, whichever came first. Of the 2,202 randomised participants, 51.2% were girls, and the mean age was 25.2 (±13.8) mo; 148 (6.7%) participants were lost to follow-up, 9 (0.4%) died, and 14 (0.6%) were admitted to hospital. The incidence rates of NNO (first event/year) for the RUTF, MNP, and control groups were 0.143 (95% confidence interval [CI], 0.107–0.191), 0.185 (0.141–0.239), and 0.213 (0.167–0.272), respectively. The incidence rate ratio was 0.67 (95% CI, 0.46–0.98; p = 0.037) for RUTF versus control; a reduction of 33.3%. The incidence rate ratio was 0.86 (0.61–1.23; p = 0.413) for MNP versus control and 0.77 for RUTF versus MNP (95% CI 0.52–1.15; p = 0.200). The average numbers of study illnesses for the RUTF, MNP, and control groups were 2.3 (95% CI, 2.2–2.4), 2.1 (2.0–2.3), and 2.3 (2.2–2.5). The proportions of children who died in the RUTF, MNP, and control groups were 0%, 0.8%, and 0.4%. The findings apply to ill but not malnourished children and cannot be generalised to a general population including children who are not necessarily ill or who are already malnourished. Conclusions A 2-wk nutrition supplementation programme with RUTF as part of routine primary medical care to non-malnourished children with malaria, LRTI, or diarrhoea proved effective in preventing malnutrition in eastern Uganda. The low incidence of malnutrition in this population may warrant a more targeted intervention to improve cost effectiveness. Trial Registration clinicaltrials.gov NCT01497236


Stephanie Roll
Institute for Social Medicine, Epidemiology and Health Economics, Charité University Medical Center, Berlin, Germany   Aim of the study To determine the effectiveness of 14 days nutritional supplementation (RUTF or MNP) given concurrently with the appropriate medical treatment, amongst children diagnosed with malaria and/or diarrhoea and/or LRTI in reducing the incidence of acute malnutrition and the incidence of the 3 study diseases compared to a control group.

Primary objective
The primary objective is to determine the effectiveness of 14 days of RUTF supplementation versus no supplementation (control group) on the incidence of acute malnutrition

Primary Hypothesis
Supplementation for 14 days with RUTF, concurrently with appropriate medical treatment for malaria, and/or LRTI, and/or diarrhoea reduces the incidence of acute malnutrition compared to a control group during 6 months follow-up, amongst children 6-59 months of age with malaria and/or, LRTI and/or diarrhoea at time of recruitment.

Secondary objectives
Secondary objectives include the following.
-Incidence of malnutrition Determine the efficacy of 14 days MNP supplementation versus a control group and RUTF versus MNP on the incidence of acute malnutrition during 6 months follow-up, -Frequency of morbidity Determine the effect of 14 days supplementation on the frequency of morbidity (3 study diseases) during 6 months follow-up, comparing both intervention groups to the control group and comparing RUTF to MNP) -Weight changes immediately after intervention period, measured at day-14 and day-28 Measure mean weight change (total change and rate (grams/kg/day) of change), comparing both intervention groups separately (RUTF and MNP) to the control group Determine frequency of 'no weight gain and/or weight loss', comparing both intervention groups separately (RUTF and MNP) to the control group Assess if provision of 14 days of RUTF or MNP supplement promotes a sustained weight gain, measured after 28-days of follow-up, comparing both intervention groups separately (RUTF and MNP) to the control group -Evaluation of several subgroup to assess potential effect modifiers.

Study design
This is a partially blinded randomized controlled trial with three study groups. Children 6 to 59 months of age presenting at the OPD in Kaabong and diagnosed with one or more of the three study diseases (malaria, diarrhoea and LRTI) are eligible for the study.
Children participating in this study will be randomised to one of 3 study groups to: A) Receive 14 days of RUTF supplementation with standard care and treatment or B) Receive 14 days MNP supplementation with standard care and treatment or C) Be included in a control group receiving standard care and treatment but not receiving nutritional supplementation (control group) Individual follow-up will be 6 months. During this time, children in the RUTF or MNP group will receive 14 days nutritional supplement every time diagnosed with at least one of the three study diseases, not exceeding more than 14 days supplementation in any 28 day period.

Primary endpoint
The primary endpoint is "negative nutritional outcome (NNO)" of a child within the 6 months follow-up period. The incidence of a negative nutritional outcome will be defined in two different ways according to the baseline nutritional status: • for children with no malnourishment at time of entry into study, "negative nutritional outcome" is defined as progression to moderate or severe malnourishment, Thus, the event of a child reaching a "negative nutritional outcome" at least once within 6 months of follow-up period will be compared between the three study groups.

Intention to treat population
The intention to treat population (ITT) includes all patients randomised that received at least one dose of study medication and who have baseline data recorded.

Per protocol population
The per protocol population (PP) includes all patients of the ITT population which were treated following the specifications of the study protocol without major protocol deviations. Excluded from the per protocol population will be patients that meet at least one of the following criteria: − Violation of one of the following exclusion criteria: exclusive breastfeeding, hospitalisation for serious complicated illness (e.g. chronic illness). − Not treated according to the allocated intervention group o For the RUTF group (for each 14-day supplementation period) less than 12 days of supplement taken by the child less than 12 sachets consumed by the child less than 12 empty sachet counted o For the MNP group (for each 14-day supplementation period) less than 24 days of supplement taken by the child less than 24 sachets consumed by the child less than 24 empty sachet counted o For all three groups (RUTF, MNP, control): having taken RUTF or MNP outside the study allocation (e.g. as part of measles treatment) − Patient misses 3 or more scheduled visits − Patient misses 2 or more consecutive scheduled visits − Patient misses the last scheduled visit (6 months visit) − Participation in other feeding programmes

Completed the study
Patients will be considered as having 'completed the study' if they have attended at least the first and last scheduled visit at 6 months.

Baseline drop-out
All patients enrolled who dropped out before the start of randomized treatment.

Analysis of baseline data
All available data at baseline will be analysed descriptively for each intervention group and in total. Results will be summarised as frequencies and percentages for nominal data; as means, standard deviations, and range for continuous data; and as medians, quartiles, and ranges for ordinal data. Differences regarding baseline variables will be checked for clinical relevance between the three treatment groups.
The flow of participants will be described (e.g. number screened, recruited, randomized, emergency unblinding, etc.). Number and reasons for drop-out, lost-to-follow-up, or withdrawal will be presented. If possible respondents and non-respondent will be compared.

Compliance
Compliance to RUTF and MNP will be assessed through questionnaire and collection of returned supplement sachets. Compliance is measured as the percentage of dispensed sachets returned empty and will be described by mean percentage (with range and standard deviation) for each treatment group.

Primary analysis of primary endpoint
The primary endpoint is "negative nutritional outcome" of a child within the 6 months follow-up period. The incidence rate (in formulae below referred to simply as 'rate') of "negative nutritional outcome" will be computed as the numbe r of events divided by total observed person-time in each treatment group. As an "event" the first time a child reaches a negative nutritional outcome will be used. The primary analysis of the primary endpoint will be performed on the intention to treat population with no imputation of missing data.
In the primary analysis, RUTF will be compared to the control group The hypotheses in this step are where rate RUTF = rate of "negative nutritional outcome" in the children in the RUTF group, and rate Control = rate of "negative nutritional outcome" in the children in the control group.
The comparison of rates among the two treatment groups (RUTF vs. control) will be performed by a Poisson regression model, adjusted for nutritional status at enrolment (fixed effect) and will be performed within one model including the three intervention groups with contrasts for each two-group comparison. The result will be presented as rates for each intervention group, and a rate ratio for the pair-wise comparison (each with 95% confidence intervals).Tests will be two-sided with a significance level of 0.05.

Treatment group comparisons
Although the study is primarily powered for the comparison between the RUTF and control group the other groups will be compared as well, keeping in mind that the sample size might not be large enough to obtain significant results. Since there are three interventional groups to be compared, a hierarchical test procedure will be used to account for multiplicity.
i) In a first (primary) step, RUTF will be compared to the control group which is described above as analysis of primary outcome If the RUTF group is not significantly different at a 0.05 level (two-sided test), the hierarchical procedure is stopped (all following analyses will be considered explorative). If the RUTF group is significantly different at a 0.05 level, the hierarchical procedure is continued with step 2.
ii) In the second step, MNP will be compared to the control group The hypotheses in this step are where rate MNP =rate of "negative nutritional outcome" in the children in the MNP group, and rate Control =rate of "negative nutritional outcome" in the children in the control group.
If the treatment factor is not significant at a 0.05 level (two-sided test), the hierarchical procedure is stopped (all following analyses will be considered explorative). If the treatment factor is significant at a 0.05 level, the hierarchical procedure is continued with step iii).
iii) In the third step, RUTF will be compared to MNP (non-inferiority test). The hypotheses in this step are H 0 : rate RUTF -rate MNP > -δ vs. H 1 : rate RUTF -rate MNP ≤ -δ where rate RUTF =rate of "negative nutritional outcome" in the children in the RUTF group, rate MNP =rate of ""negative nutritional outcome" in the children in the MNP group, and δ=inferiority margin.
As non-inferiority margin, a δ of 2% (absolute value) will be used (i.e. allowing a slightly worse result in the MNP compared to the RUTF group). The third step will be tested at a 0.025 level (one-sided test), by a onesided 97.5% confidence interval of the difference in rates between the RUTF and the MNP group. The general problem of assay sensitivity in non-inferiority comparisons can be addressed in this study by the 3arm design including a control group receiving no nutritional supplementation. Thus, superiority of the two nutritional interventions (RUTF and MNP) versus no intervention can be evaluated prior to the non-inferiority comparison.
These step-wise comparisons of rates among treatment groups will be performed by the same Poisson regression model, adjusted for nutritional status at enrolment (fixed effect) as described above, i.e. performed within one model including the three intervention groups with contrasts for each two-group comparison. The results will be presented as rates for each intervention group, and rate ratios for the pair-wise comparison (each with 95% confidence intervals).

Further adjustments
In case of relevant differences in baseline characteristics additional adjustment(s) for these factors will be performed within the model described in 5.4.1 to address potential confounding.

Subgroups
Stratified analysis will be done for three pair-wise comparisons (RUFT vs. control, MNP vs. control, RUTF vs. MNP (as described in 5.4.1) according to the following subgroups (measured at baseline): − age (binary: <36 months, ≥36 months) − gender − breastfeeding practices (no, partial breast feeding) − underlying illnesses (only malaria, only diarrhoea, only LRTI, 2 or more diseases) − socio-economic status (low, moderate, high) − food availability (binary or in tertile) − length of illness before visiting MSF clinic (binary or in tertile) − recruitment period (binary or in tertile) To assess the presence of effect modification for these factors, models with interaction terms will be analysed.

Missing endpoint data
The amount, pattern, timing and distribution of missing data of the primary endpoint will be checked by a person blinded to treatment allocation. A decision based upon the result of this check will be made as to whether and how imputation of missing primary endpoint data will be performed as a sensitivity analysis on the ITT population for the analysis in 5.4.1.

Per protocol analysis
The analysis as described in 5.4.1 will be repeated for the per protocol population.