The optimal growth and development of infants and young children are fundamental for their future and for that of the societies in which they live. An evaluation of the timing of growth faltering including 39 nationally representative surveys in developing countries indicates that the two periods of highest vulnerability for growth retardation are the intrauterine development and that from birth to 36 months of age (Shrimpton 2001). In infancy and early childhood common causes of undernutrition in developing countries are: 1) inappropriate feeding practices and behaviours such as lack of promotion of exclusive breastfeeding jointly with early introduction of weaning foods; and 2) receiving inadequate diets in quantity and/or quality (WHO 1999). In addition, particularly in low-income populations, contextual factors such as unsafe water, sanitation, and the quality of parental health and care, may undermine the outcomes of public health programmes designed to address undernutrition.
A comprehensive analysis of 26 selected health risk factors, estimates that 15% of the global burden of disease can be attributed to the joint effects of childhood and maternal underweight or micronutrient deficiencies (Ezzati 2002). Undernutrition and infection are closely related to the high morbidity and mortality in circumstances of high exposure to infectious diseases and inadequate diet. Furthermore, this association claims the lives of millions of children throughout the world. For millions more who survive, growth and development may be damaged, quality of life diminished, and future well being compromised. The World Health Report 2002 reports that 170 million children in poor countries are underweight, and over three million of them die each year as its result (WHO 2002).
Supplementary feeding, defined as the provision of extra food to children or families beyond the normal ration of their home diets, is an intervention aimed at improving the nutritional status, or preventing the nutritional deterioration, of the target population (Beaton 1982). Young children could consume the supplementary food at home, at a supervised feeding centre, or other places adapted for this purpose. These approaches carry many implications that should be considered when assessing the effectiveness of supplementary feeding. In the latter 'out of home' approach, the family needs to be motivated to participate daily, and someone has to be able to take the child to and from the centre each day, and therefore the centre should be in reasonably close proximity to their home. Further, personnel are needed to prepare and serve the food as well as record and monitor participation. The 'at-home' food approach permits a greater geographic separation between those distributing the food and the homes, and hence requires fewer staff. However, the impact on the intended beneficiaries is probably less efficient due to the fact that food delivered to the home may well be shared with other family members (Beaton 1982).
The impact of supplementary feeding on child physical growth merits careful evaluation in view of the reliance of many states and NGOs on these interventions to improve child health in developing countries. Randomised controlled trials (RCTs) are widely accepted to be the best method for minimizing systematic errors (bias) when assessing the effects of health care interventions (Chalmers 1983; Villar 1996). Effective randomisation guarantees that there is no bias in the allocation of participants and enables trial conclusions to be more reliable than other methods of treatment allocation (Pocock 1990).
We recognize that supplementary feeding for young children in developing countries may ameliorate the current situation and may contribute to a long-term improvement, but not, in and of itself, represents a solution to the fundamental health and nutritional problems faced by families living in poverty. In this context, diarrhoea and infectious diseases may undermine the beneficial effects of extra feeding. Improvements relating to food safety, housing, water supply, and sanitation are important steps towards tackling undernutrition and these important contextual factors need to be considered when examining the effects of nutritional interventions.
The development of appropriate combinations of interventions aimed at preventing or treating impaired growth in young children is a priority given the devastating effects of child malnutrition on human performance, health and survival. Supplementary feeding have been widely implemented but not systematically evaluated. This systematic review will assess the available published evidence and, therefore, may contribute to identify research priorities to address the right to food, health, and basic care of children living in chronic poverty.
To assess the effectiveness of community-based supplementary feeding in promoting the growth of young (pre-school) children in developing countries.
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (randomisation by cluster or by individuals). Individually-randomised trials were excluded if 20% or more participants were lost to follow-up in any of the comparison groups. Quasi-random designs were also excluded.
Types of participants
Children from developing countries born at term (≥ 37 completed weeks of gestation) aged 0-5 years old. Studies including children with malnutrition not resulting from insufficient dietary intake, e.g. cystic fibrosis, metabolic and endocrine disorders were excluded.
Types of interventions
Supplementary feeding or food supplementation defined as the provision of extra food to children or families beyond the normal rations of their home diets. The intervention has to be "community-based", i.e. young children could consume the supplementary food at home, at a supervised feeding centre, or other places adapted for this purpose such as health care centres and crèches. Trials in hospital and refugee settings were excluded.
Supplementary feeding could comprise:
- meals (local or imported foods)
- drinks (juices or milk)
- snacks (including both food and milk snacks)
Controls include either no treatment (home diet or no extra feeding) or placebo (e.g. low or no-protein and low-energy drinks)
Types of outcome measures
The following primary anthropometric (growth measurement) outcomes were considered at the end of the intervention:
1) Weight expressed in kg or weight-for-age (W-F-A) z-score.
2) Length/height expressed in cm or length/height-for-age (L-F-A) z-score.
3) Weight-for-height (W-F-H) z-score.
4) Prevalence of underweight (weight-for-age below -2 standard deviation (SD) from the reference median value of the NCHS/WHO).
5) Prevalence of stunting (height-for-age below -2 SD from the reference median value of NCHS/WHO).
6) Prevalence of wasting (weight-for-height below -2 SD from the reference median value of the NCHS/WHO).
Secondary outcome measures were:
7) Head circumference (front-occipital circumference) in cm.
8) Mid-upper-arm circumference (MUAC) in cm.
9) Skinfold thickness (subscapular, tricipital) in mm.
Search methods for identification of studies
The terms listed below were used to search the Cochrane Controlled Trials Register (CENTRAL), published in The Cochrane Library (Issue 2, 2005), MEDLINE 1966 to June week 3 2005, EMBASE 1980 to week 26 2005, CINAHL 1982 to June week 3 2005, LILACS 1982 to 2005, Social Science Citation Index 1956 to 2005, and Dissertation Abstracts International (late 1960s to 2005). When necessary, the search terms were modified to suit the requirements of particular databases. No language restrictions were applied.
The following electronic search strategy was developed:
1 exp Child/
2 exp Infant/
3 (child$ or infan$ or baby or babies or pre-school$ or preschool$).tw.
4 exp Growth/
5 (grow$ or anthropometr$ or weigh$ or height or length).tw.
6 1 or 2 or 3
7 4 or 5
8 ((supplement$ or extra) adj5 (food or feed$ or diet$ or
9 6 and 7 and 8
10 randomized controlled trial.pt.
11 controlled clinical trial.pt.
12 randomized controlled trials.sh.
13 random allocation.sh.
14 double blind method.sh.
15 single-blind method.sh.
17 (animal not human).sh.
18 16 not 17
19 clinical trial.pt.
20 exp clinical trials/
21 (clin$ adj25 trial$).ti,ab.
22 ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or
26 research design.sh.
28 27 not 17
29 28 not 18
30 comparative study.sh.
31 exp evaluation studies/
32 follow up studies.sh.
33 prospective studies.sh.
34 (control$ or prospectiv$ or volunteer$).ti,ab.
36 35 not 17
37 36 not (18 or 29)
38 18 or 29 or 37
39 9 and 38
Searching other resources
References of retrieved articles and relevant reviews were scanned for potentially eligible studies. Letters to the authors of the included trials were sent asking for help in clarifying relevant and missing data.
Data collection and analysis
Selection of studies
Titles and abstracts of articles retrieved by the electronic searches or by other methods were assessed independently by two review authors (YS and GC) to determine whether they met the inclusion criteria. They were not blinded to the names of the authors, institutions or journal of publication. Discrepancies were resolved by discussion. When necessary, a third reviewer (MDO) was consulted. See also Figure 1
|Figure 1. Eligibility of studies for inclusion in the systematic review|
Data extraction and management
Data were independently extracted by YS and GC. Data were recorded on data extraction forms. These included method of random allocation, participant characteristics, description and length of the intervention and co-interventions, data on outcomes related to child physical growth, and rates of withdrawals. Disagreements were discussed and, when necessary, a third reviewer (MDO) was consulted.
Two review authors (YS and GC) independently coded all studies. Citations and data were entered and organized in RevMan 4.2.7.
Assessment of risk of bias in included studies
Two review authors (YS and GC) independently assessed each included study on a number of criteria, in particular:
* Allocation concealment to intervention groups (protection against selection bias)
Allocation concealment was defined as below, as per the Cochrane Reviewers' Handbook (Alderson 2004)
(A) indicates adequate concealment of allocation (e.g. by consecutively numbered sealed opaque envelopes).
(B) indicates uncertainty about whether the allocation was adequately concealed (e.g. possibly where the method of allocation concealment is not reported).
(C) indicates that the allocation was definitely not adequately concealed (e.g. open random number lists).
* Blinding in outcome assessment (protection against detection bias)
Blinding was rated as below:
MET: assessor unaware of the assigned treatment when collecting outcome measures.
UNCLEAR: blinding of assessor not reported, and cannot be verified by contacting investigators.
NOT MET: assessor aware of the assigned treatment when collecting outcome measures.
* Losses to follow-up were classified as follows:
ADEQUATE: losses to follow-up less than 20% in each of the comparison groups.
UNCLEAR: losses to follow-up not reported.
Individually-randomised trials were excluded if 20% or more participants were lost to follow-up in any of the comparison groups.
* Intention-to-treat was rated as below:
MET: intention-to-treat analysis performed or possible with data provided.
UNCLEAR: intention-to-treat analysis not reported, and cannot be verified by contacting the investigators.
NOT MET: intention-to-treat analysis not done and not possible with data provided.
* Reliable primary outcome measure(s):
ADEQUATE: if anthropometric measures such as weight and length/height were well-described and methods were in line with established protocols. Important details were the training of the anthropometric team, the number of replicates of measurements, and type and calibration of equipment.
UNCLEAR: if anthropometric measures and methods of measurements were not clearly described.
INADEQUATE: if the anthropometric team was not trained, if the methods were not described in detail, and if equipment was not well-calibrated.
Measures of treatment effect
Should sufficient continuous data be combined in future updates of this review, they will be analysed if means and standard deviations are available and there is no clear evidence of skew in the distribution. Where scales measured the same clinical outcomes in different ways, standardised mean differences (SMD) will be compared across studies. Inverse variance methods will be used to pool SMDs, so that each effect size is weighted by the inverse of its variance in an overall estimate of effect size. Confidence intervals of 95% will be used for individual study data and pooled estimates.
Should sufficient binary data be combined in future updates of this review, they will be analysed by calculating odds ratios with 95% confidence intervals. If some primary studies report an outcome as a dichotomous measure and others use a continuous measure of the same construct, two separate meta-analyses will be used (one for odds ratios and another for SMDs). When a primary outcome study provides multiple measures of the same construct at the same point in time, an average effect size will be used to avoid dependence problems. When a primary outcome study reports multiple measures of the same construct at different points in time, we will use a single measure that is closest to a one-year follow-up.
Dealing with missing data
Authors of studies with missing data were contacted.
Assessment of heterogeneity
Clinical heterogeneity amongst the included studies is discussed below. Should a meta-analysis be possible in future updates, consistency of results will be assessed visually and by examining I
Assessment of reporting biases
Funnel plots (effect size against standard error) will be drawn if sufficient studies are found in future updates of this review. Asymmetry could be due to publication bias, but they can also be due to a relationship between trial size and effect size. In the event that a relationship is found, clinical diversity of the studies will also be examined as a possible explanation (Egger 1997).
Statistical analysis was carried out using the RevMan 4.2.7 software. Results were presented as weighted mean differences (WMDs) with their 95 per cent confidence intervals, as the outcomes assessed were continuous. No meta-analysis is appropriate in the current version of this review.
Subgroup analysis and investigation of heterogeneity
Pre-specified subgroups based on the age of the children (younger than 12 months), nutritional status at baseline (underweight, stunting or wasting), and duration of the intervention (more than 12 months), were planned in the protocol of this review. This analysis was not done due to the small number and clinical diversity of the included studies. However, it will be done in future updates if sufficient data are available.
If required in future updates, sensitivity analyses will be performed to evaluate whether the pooled effect sizes are robust across components of methodological quality. In line with the methodological criteria assessed, we will conduct sensitivity analysis for each major component of the quality form such as concealment allocation and blinded assessment.
Use of data on program costs
None the included trials reported data on the direct or indirect costs related to the intervention.
Description of studies
Results of the search
Of the fourteen RCTs considered as potentially eligible, only four met the inclusion criteria for this review. See also Figure 1
All studies included are published journal articles. Two studies were conducted in Indonesia (Indonesia 1991; Indonesia 2000), and the other two in Jamaica (Jamaica 1991) and Guatemala (Guatemala 1995). These studies were conducted from the early 70s to late 90s, and were clinically heterogeneous. Differences existed in age of participants, with two studies recruiting only very young (<24 month) children (Jamaica 1991; Indonesia 2000), and the others older children (Indonesia 1991; Guatemala 1995). Nutritional status at baseline differed also. Two studies were focused on children with impaired growth (Jamaica 1991; Indonesia 2000) whereas in the other two (Indonesia 1991; Guatemala 1995) no exclusion criteria were based on the nutritional status of the study population. Supplementation varied both in physical form (liquid [Guatemala 1995; Indonesia 2000]) or solid form [Indonesia 1991; Jamaica 1991]) and in method of delivery (home delivery [Jamaica 1991] versus central distribution [Guatemala 1995; Indonesia 1991; Indonesia 2000]). Duration of food supplementation varied from 3 months (Indonesia 1991), 12 months (Jamaica 1991; Indonesia 2000) to 7 years (Guatemala 1995). Timing of outcome measurement differed among the included studies. One of the studies conducted in Indonesia (Indonesia 1991) evaluated weight and length only twice, i.e. at the beginning and at the end of the intervention. In the Jamaican study (Jamaica 1991), children were measured at enrolment, and 6 and 12 months later. Weight and length of children were assessed at birth, 15 days of life and at 3, 6, 9, 12, 15, 18, 21, 24, 30, 36, 42, 48, 60, 72 and 84 months of age in the Guatemalan study (Guatemala 1995). In the more recent Indonesian study (Indonesia 2000), anthropometric variables were assessed at baseline and at 2, 4,6, 8, 10 and 12 month after the beginning of the intervention.
We group the included studies into two categories: a) studies without formal assessment of malnourishment at baseline, b) studies involving children formally assessed as malnourished.
a) Studies without formal assessment of malnourishment at baseline
Developmental effects of short-term supplementary feeding in nutritionally-at-risk Indonesian infants
Indonesia 1991 This cluster randomised study (Indonesia 1991) was designed to assess the effects of supplementary feeding on mental, motor and cognitive development on pre-school children at six tea plantations in West Java, Indonesia. According to the authors, children at the plantations were considered to be nutritionally-at-risk children but no clear clinical definition of their nutritionally status was provided. Measurements of body growth and dietary intake of participants were also recorded. Day-care centres (DCCs) within the plantations provided caretaking services and the food free of charge to children. Twenty DCCs were selected for the study based on their having more than 15 children whose ages ranged from 6 to 59 months. Afterwards, the children were divided into two age groups according to the types of psychological tests administered to them. DCC assignment to the two types of interventions was randomised and stepwise by pairs. The daily supplement consisted of twice-a-day snacks given 6d/week for 3 months. Weight and height measurements were taken at the beginning and at the end of the intervention and reported only for the young children group, i.e. infants from 6 to 20 months (n = 113). Weight and height z-scores were based on means and SDs provided by the NCHS.
Nutritional impact of supplementation in the INCAP longitudinal study: analytic strategies and inferences
Guatemala 1995 The Guatemalan study (Guatemala 1995) was the first nutritional intervention trial that used villages as units of analysis. This cluster randomised controlled trial was conducted between 1969-1977 in four rural east Guatemalan villages to test the impact of early nutrition supplementation on child growth and development. Originally, three pairs of supplementation villages were specified before randomisation, but budgetary constraints reduced the number to two, with dire consequences for statistical power. Selection of villages was done on the basis of similarities in sociocultural, anthropometric, dietary, and morbidity characteristics. One pair of villages was relatively large (about 900 people each) and one pair was small (about 500 people each). Two pairs of study villages (one large and one small villages) were randomly allocated to receive a skimmed milk-based, high-energy and high-protein supplement (Atole) or a no-protein, low-energy supplement (Fresco). Both drinks were enriched with vitamins and minerals in equal concentrations but differed in appearance and taste, making cluster randomisation sensible as neighbours would not detect differences locally. The supplements were consumed on a voluntary basis by all residents of the villages at feeding stations. Ingestion was measured and recorded only for target participants, namely, pregnant and lactating woman aged 15 years and older and children up to age of 7. Children in the studied population were born before the intervention began, others throughout the period of the seven-year long study. Therefore, depending on the date of birth, some of their mothers could also have been ingesting supplementary feeding for several years prior to pregnancy. Primary health care and vaccination services were available in all villages.
b) Studies involving children formally assessed as malnourished
Nutritional supplementation, psychosocial stimulation, and growth of stunted children: the Jamaican study
Jamaica 1991 The study from Jamaica (Jamaica 1991) aimed to assess the effects of home delivery of nutritional supplementation and psychosocial stimulation on the growth, development, and morbidity of 129 stunted children (height-for-age below -2 SD of the median of the NCHS/WHO reference values) aged 9-24 months. The inclusion criteria were: singleton pregnancy, birth-weight over 1.8 kg, standard of housing and maternal education below defined levels, and no obvious mental or physical handicap. Children were randomly assigned to one of four groups: food supplementation, stimulation, food supplementation plus stimulation, and control. According to our objectives, we compared the following groups: supplemented (n = 32) versus control (n = 33) group. In addition to the food supplement, cornmeal and skimmed milk powder were provided to the family in an attempt to reduce sharing of the child's supplement. Free medical care was available for all children.
Effects of an energy and micronutrients supplement on anthropometry in undernourished children in Indonesia
Indonesia 2000 The Indonesian study (Indonesia 2000) was conducted in six tea plantations in Panganlengan, West Java, to assess the consequences of food supplements on the growth, physical activity and various aspects of development in nutritionally-at-risk young children (length-for-age below -1 SD; weight-for-length between -1 and -2 SD of the median of NCHS/WHO reference values). At the time of the study, the majority of mothers employed on the plantation left their children aged 1 month to 6 years in Day-Care Centres (DCCs). Two age cohorts of children at 12 and 18 months of age were recruited from 24 community-run DCCs. These children were randomly assigned to three nutritional interventions (condensed milk plus micronutrients, skimmed milk plus micronutrients, and skimmed milk). According to our inclusion criteria, we considered two of the three intervention schemes in both study cohort: condensed milk (high energy) plus micronutrient tablet as the supplemented group (n = 38) and skimmed milk (low energy) plus micronutrients as the control group (n = 37).
For further details see Table of Included Studies.
Details and reasons for exclusion can be found in the table of excluded studies (n = 10). Overall, otherwise relevant studies were excluded either because supplementary feeding was part of complex 'packages' (e.g. food supplementation plus nutritional education plus cash payments for the families), the intervention groups did not fit in with our inclusion criteria or exceeded the threshold of permitted drop-out rate for inclusion within this review. We point out that the four-country randomised study conducted by Simondon et al. (Bolivia 1996; Congo 1996; New Caledonia 1996; Senegal 1996) assessed data individually for each study setting. Therefore, we decided to consider this multicentre trial as four different studies due to the fact that comparability of groups was not complete.
For further details see Table of Excluded Studies and Figure 1.
Risk of bias in included studies
All of the trials included in this review are of relatively low-quality. The method of randomisation was not described and the allocation concealment was not stated in any of these studies. The unit of randomisation differed among the included studies. Two studies considered individuals as the unit of analysis (Jamaica 1991; Indonesia 2000) and two used a cluster randomised technique (Indonesia 1991; Guatemala 1995). Sample sizes were not calculated for a specific magnitude of effect in three of the four included trials (Indonesia 1991; Jamaica 1991; Guatemala 1995). Blindness of growth outcome assessors was unclear in two studies (Indonesia 1991; Jamaica 1991) and the personnel in charge of the data collection were aware of treatments given to the children in the other two included trials (Guatemala 1995; Indonesia 2000). The rate of drop out was not clearly stated in three of the four included studies (Indonesia 1991; Guatemala 1995; Indonesia 2000). In general, methods for anthropometric measurements were well described in all the included studies. Details on the training of personnel and the number of replicates of measurements were not always provided.
Effects of interventions
All four RCTs included within this review considered different ways of reporting anthropometrical outcomes. We present results narratively, in the order planned at protocol stage, grouping the included trials into two categories: a) studies without formal assessment of malnourishment at baseline (Indonesia 1991; Guatemala 1995), b) studies involving children formally assessed as malnourished (Jamaica 1991; Indonesia 2000). No meta-analysis is currently appropriate due to the clinical heterogeneity among the included studies.
a) Studies without formal assessment of malnourishment at baseline
Indonesia 1991; Guatemala 1995
Both studies were cluster RCTs. The Indonesian (Indonesia 1991), included 11 DCCs in the intervention group (n = 75 children) and 9 DCCs in the control group (n = 38 children).
The Guatemalan study (Guatemala 1995) included four villages as unit of analysis (exact sample sizes were not provided). As implemented, the effects of interventions were ascertained through comparisons of results before and after the intervention in the two "Atole" and two "Fresco" villages. The before-after comparison was, however, possible only for selected variables collected with adequate sample size in 1968, before the intervention began, or for variables collected in the first months of the study which could not be immediately affected by supplementation.
Weight-for-age and height-for-age z-score
In the absence of a relevant intra cluster coefficient and in order to avoid unit of analysis errors, we have used the numbers of DCCs to calculate the effect size after three months of intervention. Improvement in the intervention group compared to the control group was not statistically significant [WMD 0.19 (95% CI -0,64 to 1.02)] ( Analysis 3.1) and [WMD 0.12 (95% CI -0.87 to 1.11)] ( Analysis 3.2) for weight and height z-scores, respectively.
Weight (kg) and length (cm)
Among other methodological difficulties, sample sizes were unavailable for children under five after the end of the intervention. Length of 3-yr-old children born between 1969 and 1973 was reported, using village as unit of analysis. This analytical approach was based on before-after comparison following three years of supplementation, by village size and type of supplement. This information comes from a table published in Guatemala 1995 which is reproduced in full in Table 1. According to this analysis, the difference in net change in the large villages was 2.55 cm and in the small villages was 2.35 cm. The mean of these differences is 2.45 +- 0.10 cm (mean +- SD).
No data for these outcomes were reported in either study.
b) Studies involving children formally assessed as malnourished
Jamaica 1991; Indonesia 2000
In the Jamaican study (Jamaica 1991), data on weight, length, head circumference, arm circumference and skinfolds were reported as means of the anthropometric measurements at 6 and 12 months after the intervention began. For the present analysis, we considered the results after 12 months of supplementation.
The most recent Indonesian study (Indonesia 2000) provided means of the effects of supplementation on weight, length, head and arm circumferences, stratified by cohort (12 or 18 month old ages), by time, by sex, and by supplement. Standard deviations (SDs) missing from the original papers have been supplied by one of the trial's authors (Aitchison 2005). The present authors have calculated single group means and SDs for intervention and control conditions, combining data presented by sex and age.
In the Jamaican study (Jamaica 1991), the mean difference in weight between the intervention and control groups was 290 g (95% CI -0.29 to 0.87) ( Analysis 1.1). From the clinical point of view, this means that children in the supplemented group grew better than children in the control group. For the 2000 Indonesian study (Indonesia 2000), the high-energy and protein group showed a smaller improvement of 160 g (95% CI -0.27 to 0.59) ( Analysis 2.1) compared to the low-energy and low-protein group. Neither of these results were statistically significant.
In the Jamaican study (Jamaica 1991), stunted children in the supplemented group gained more than one cm [WMD 1.3 (95% CI 0.03 to 2.57)] ( Analysis 1.2) compared to children in the control group. This positive result should be interpreted with caution due to the small numbers and baseline differences in birth weight between the study groups (21% of the children in the control group weighted 1.8-2.3 kg and 100% of the children in the intervention group weighted more than 2.3 kg at enrolment). For the 2000 Indonesian study (Indonesia 2000), the high-energy and protein group was not better off in terms of length compared to the low-energy and low-protein group [WMD -0.1 cm (95% CI -1.61 to 1.41) ( Analysis 2.2).
Head and arm circumferences (cm)
In the Jamaican study (Jamaica 1991), differences in means in head and arm circumferences were 0.40 (95% CI -0.21 to 1.01) ( Analysis 1.4) and 0.20 (95% CI -0.29 to 0.69) ( Analysis 1.5), respectively.
For the recent 2000 Indonesia study (Indonesia 2000), the mean difference for head circumference was 0.19 (95% CI -0.41 to 0.79) ( Analysis 2.3) and for arm circumference was 0.10 (95% CI -0.22 to 0.42) ( Analysis 2.4).
Triceps and subscapular thickness (mm)
In the Jamaican study (Jamaica 1991), differences in means for tricipital and subscapular skinfold thickness were reported as 0.20 (95% CI -0.51 to 0.91) ( Analysis 1.6) and 0.20 (95% CI -0.34 to 0.74) ( Analysis 1.7), respectively.
Neither of these results were statistically significant.
Undernutrition among pre-school children is a major public health problem in developing countries and, in an attempt to address this problem, supplementary feeding is frequently implemented. This intervention is thought to be efficacious in preventing or treating growth faltering by optimising the dietary intake of children in at-risk populations and thereby enhancing their general health.
In all, only four RCTs conducted over a 20-year period met the inclusion criteria for this review. Unfortunately, the paucity of high-quality data makes it difficult to draw any conclusions about whether or not this intervention is effective in promoting the physical growth of pre-school children living in chronic poverty. In terms of study design and implementation, a number of explanations could be offered for the limited evidence available. Supplementary feeding may be aimed at preventing, treating, or both preventing and treating impaired child growth. A thorough knowledge of the nutritional status of the targeted population is crucial for a valid use and interpretation of the outcomes of interest. There are a further range of problems posed in relation to the conduct of nutritional intervention studies. Ideally, experimental designs assume that the only relevant variable that distinguishes the experimental and the control groups is exposure to intervention. One of the most notorious risks associated with quasi-experimental studies is the non-equivalence of groups at baseline. Therefore, if the intervention and controls groups differ along the outcome variables before exposure, then differences observed after the exposure may not be attributed to the intervention. Random allocation of people to intervention is aimed at avoiding such potential differences between experimental and control groups, although, randomisation of potentially beneficial interventions in poor populations raises ethical concerns. A second issue relates to the implementation of the intervention. Difficulties with blinding in the delivery of supplemental food, might have effects on the outcome measurements. There is a possibility that the field-workers behave differently toward the participants across the study groups. For instance, outcome assessors might give greater encouragement to the mothers to attend the feeding centre in the intervention group and this, in turn, may affect participation. These findings not only highlight the importance of adequate methodological quality of trials but also the importance of complete reporting. Authors should, as a minimum, explicitly describe their approaches to random sequence generation, allocation concealment, blinding, and handling of exclusions after allocation.
We also note that among the studies excluded from this review, one multicentre RCT was excluded due to attrition, or drop outs of more than 20% across study participant groups (Bolivia 1996; Congo 1996; New Caledonia 1996; Senegal 1996). The total number of participants thus excluded from the review was 447. Reasons for dropping out of such a controlled trial can vary, but among the reasons are the practical difficulties of engaging impoverished families in additional routines. Issues related to the cultural meaning of supplemental food in terms of its effects on attendance and parental perception of the child's health and well being are also important factors to take into account when designing a multicentre study.
Finally, the paucity of RCT data evaluating this nutritional intervention may be partially due to the complexity of conducting controlled trials at a community level, especially in developing countries. RCTs are well adapted to testing single medications, vaccines, and nutrient supplements that can be double masked and compared with placebo. It is much more difficult to conduct community-based RCTs of behavioural or complex interventions such as supplementary feeding.
We argue that the problems to which supplemental feeding is addressed are entwined with the problems of poverty. Drinking of unsafe water, inadequate quantities of water for hygiene, and lack of access to effective sanitation contribute to about 1.5 million child deaths and around 88% of deaths from diarrhoea in poor populations (WHO 2002). In this instance, sanitary control of the environment is seen as a prerequisite to the effective utilization of supplementary feeding. Other relevant factors include: geographical variables (e.g. rural areas, transport and communication, basic health services and medical care), nutritional education, and parental knowledge and care. Whereas valid research is essential for the effective use of resources and expansion of public health care interventions, poor families and children cannot wait for future investigations. They must have access both to adequate amounts of food and appropriate health care and sanitation. Moreover, research efforts will be misled if they are framed in a political and social background that neglects other basic human rights and does not attend to overall quality of life.
Implications for practice
Based on the published evidence reviewed, it is difficult to assess the impact of supplementary feeding on growth in young children living under poor socio-economic and environmental conditions. The lack of adequate evidence may provide the basis for advocating more specific implementation and testing of early childhood nutritional interventions in under-resourced settings . We argue that unless further consideration is given to some of the issues raised by this review, it might be difficult to assess the effectiveness of combining interventions into an integrated model of child care.
Implications for research
In view of the paucity of high-quality data, well-conducted trials are needed. Issues of research design such as sample size calculation to detect a meaningful magnitude of effects and blinding need to be addressed in future studies. Ideally, researches should also agree on a minimum set of standardised anthropometric measurements in trials evaluating nutritional interventions to facilitate synthesis and interpretation of results.
Main flaws in the methodological quality of the included studies are summarised as follows:
Overall, further studies in this area would need to:
We thank the Department of Nutrition of the World Health Organization, the National Ministry of Health and Environment of Argentina and the Aubrey Sheiham Scholarship in Public Health Promotion and Primary Care for financial support to assist the completion of the systematic review. We would also like to thank the Cochrane Developmental, Psychosocial and Learning Problems Group for valuable advice and assistance and Shailen Nandy from the University of Bristol, UK, for his comments on the background section. We acknowledge Edgardo Abalos, sub-director of Centro Rosarino de Estudios Perinatales, Argentina, and Esther Coren from the Social Care Institute for Excellence, UK, for their contributions to draft versions of this review. We are also grateful to Dr Reynaldo Martorell for his help in clarifying details of the Guatemalan study. We appreciate very much the data provided by Dr Tom Aitchison regarding the Indonesian study published in 2000.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Last assessed as up-to-date: 14 June 2005.
Protocol first published: Issue 4, 2004
Review first published: Issue 4, 2005
Contributions of authors
All the authors of the systematic review (YS, MDO, and GC) contributed to the drafting of the protocol. YS developed the search strategy and performed the searches, YS and GC selected studies for relevance, YS and GC extracted the data, YS entered all the data into RevMan and drafted and completed the review. MDO provided expertise and guidance and contributed to the writing and editing of the review. GC extracted and double checked the data, provided methodological advice and helped to write the review.
Declarations of interest
Sources of support
- Centro Rosarino de Estudios Perinatales (CREP), Rosario, Argentina.
- National Ministry of Health and Environment (Beca "Ramón Carrillo-Arturo Oñativia", CONAPRIS) Buenos Aires, Argentina.
- Department of Nutrition of the World Health Organization, Geneva, Switzerland.
- The Aubrey Sheiham Public Health and Primary Care Scholarship, The UK Cochrane Centre, UK.
Medical Subject Headings (MeSH)
MeSH check words
Child; Child, Preschool; Humans; Infant
* Indicates the major publication for the study