Systematic review of the efficacy and effectiveness of complementary feeding interventions in developing countries

[Table: see text]

Introduction: Complementary feeding interventions are usually targeted at the age range of 6–24 months, which is the time of peak incidence of growth faltering, micronutrient deficiencies and infectious illnesses in developing countries. After 2 years of age, it is much more difficult to reverse the effects of malnutrition on stunting, and some of the functional deficits may be permanent. Therefore, interventions that are effective at reducing malnutrition during this vulnerable period should be a high priority. Although several types of interventions can be targeted to this age range (e.g. micronutrient supplementation), a food‐based, comprehensive approach may be more effective and sustainable than programmes targeting individual nutrient deficiencies. For this review, a broad definition of ‘complementary feeding interventions’ is used so as to capture the full range of strategies that can be used.

Scope and methods of the review: The interventions described in this review generally include one or more components related to the Guiding Principles for Complementary Feeding of the Breastfed Child (PAHO/WHO 2003). The 10 guiding principles cover: (1) duration of exclusive breastfeeding and age of introduction of complementary foods; (2) maintenance of breastfeeding; (3) responsive feeding; (4) safe preparation and storage of complementary foods; (5) amount of complementary food needed; (6) food consistency; (7) meal frequency and energy density; (8) nutrient content of complementary foods; (9) use of vitamin‐mineral supplements or fortified products for infant and mother; and (10) feeding during and after illness. This review includes any relevant intervention that targeted children within the age range of 6–24 months. In some cases, the intervention may have included children older than 24 months, but in all studies at least some of the children were between 6 and 24 months. The assumption is that many of the children in these studies were breastfed, although a certain proportion will have terminated breastfeeding before 24 months. Although strategies for optimizing the duration of exclusive breastfeeding or increasing the total duration of breastfeeding may have a direct influence on several of the outcomes of interest, this review will not cover those strategies because another report will review those results.

The primary outcomes of interest for this review include growth, morbidity and child development. Micronutrient intake and micronutrient status were also included as outcomes because of their link to these key functional outcomes. Studies that assessed the impact of complementary feeding interventions on feeding practices only were not included because of time constraints and because it has been demonstrated previously that appropriately designed interventions can have a positive impact on feeding practices (Caulfield et al. 1999). For most intervention strategies and outcomes, the literature search was focused on the period from 1996 to 2006, as the previous review by Caulfield et al. (1999) covered the period from 1970 to 1997. For certain interventions not covered in the previous review (i.e. using amylase to increase energy density and interventions focused on iron status outcomes), studies dating back to 1990 were included. Only studies conducted in developing countries were included. The search was conducted using electronic methods, inspection of websites of key private voluntary organizations and the bibliographies of published papers, and personal contacts. The two authors of this review independently assessed the quality of each of the reviewed studies, and those scored as 2– (non‐randomized studies with a high risk of bias) were not included in the tabulation of results.

In total, 42 papers were included in the review. These papers report results from 29 efficacy trials and 13 effectiveness studies or programme reports from 25 developing countries. Interventions were considered efficacy trials if there was a high degree of assurance of delivery of the ‘treatment’, generally under carefully controlled research conditions (e.g. provision of a fortified complementary food with frequent follow‐up to assess adherence). Evaluations of interventions carried out in a programme setting, generally with less ability to control delivery of and adherence to ‘treatment’, were considered effectiveness studies.

To compare growth (weight and length) results across studies (when these results were reported as means ± SD), we calculated the treatment effect size for each outcome of interest using the formula:

When possible, the effect sizes for each outcome were averaged across interventions to obtain a rough estimate of overall impact. Effect size can be categorized as small (∼0.2), medium (∼0.5) or large (∼0.8).

Interventions were grouped into five categories depending on the main strategy used:

1. 1

   education about complementary feeding as the main treatment,

2. 2

   complementary food or a food product offering extra energy (with or without added micronutrients) provided as the only treatment,

3. 3

   provision of food combined with some other strategy, usually education for mothers,

4. 4

   fortification of complementary foods (centrally processed fortified foods or home‐fortification products) with micronutrients (with no difference in energy provided to intervention vs. control groups), and

5. 5

   increased energy density and/or nutrient bioavailability of complementary foods through the use of simple technologies.

Some studies had more than one intervention group and may thus be included in more than one of the categories. In these situations, only the results for the intervention groups that are relevant to the comparison in question are included in that section. Some of the interventions targeted only malnourished children, but most were aimed at all children in the target age range.

Growth: Nearly all of the studies assessed growth as an outcome. There were six efficacy trials and five effectiveness studies in which the main intervention strategy was education about complementary feeding. Taking these 11 studies together, educational interventions had a modest effect on weight (mean effect size = 0.28; range −0.06, 0.96) and linear growth (mean effect size 0.20, range 0.04, 0.64). The two educational interventions with the greatest impact on both weight and length gain (effect sizes of 0.34–0.96) were the projects in Peru (Penny et al. 2005) and China (Guldan et al. 2000). In both of these, a key message was to regularly provide an animal‐source food to the infant (chicken liver, egg or fish in Peru; egg in China). The other educational intervention with a relatively large impact on weight (though not on length) was a study in Bangladesh that targeted children with low weight‐for‐age at baseline (Roy et al. 2005). That intervention also promoted the home preparation of a complementary food mixture that included egg, meat or fish.

There were seven efficacy trials and one effectiveness study in which the only intervention strategy was provision of complementary food (often fortified). The results were somewhat inconsistent: there was a positive impact in Ghana and Malawi but no impact in South Africa, Indonesia or Brazil. The overall mean effect size was 0.60 (range −0.02, 2.99) for weight and 0.47 (range −0.04, 1.81) for linear growth, but these effects are inflated by the results from Nigeria (Obatolu 2003) (effect sizes: weight = 2.99, length = 1.81). Excluding that study, the mean effect size was 0.26 (range −0.02, 0.57) for weight and 0.28 (range −0.04, 0.69) for length. For the combination of provision of complementary food with some other strategy (usually education), there were two efficacy trials and six effectiveness studies. With these eight studies combined, the average effect size for weight was 0.35 (range 0.18, 0.66) and that for linear growth was 0.17 (range 0, 0.32). Two studies specifically evaluated whether provision of food plus education was more effective than education alone (Bhandari et al. 2001; Roy et al. 2005). In India (Bhandari et al. 2001), the food plus education group gained 250 g more weight and 0.4 cm more than the control group during the 8‐month intervention, whereas the education‐only group gained only 90 g more than the control group and did not have any advantage in length gain. In Bangladesh (Roy et al. 2005), results for the education‐only group were intermediate between those of the food plus education and control groups. Thus, in these two settings the inclusion of a food supplement was more effective than education alone.

The effect of fortification of complementary foods (with no difference in the amount of energy provided to intervention and control groups) on growth was evaluated in six efficacy trials, three of which involved home fortification using micronutrient supplements (powders or crushable tablets). The other three studies used cereal/legumes mixes or a milk formulation to which the micronutrients were added during processing. Only in the fortified‐milk study (conducted in India) was there a significant impact on growth. The average effect size for all six studies was 0.11 (range −0.22, 0.37) for weight and 0.12 (range −0.02, 0.45) for length. There were no effectiveness studies identified within this category.

There were five efficacy trials in which the main strategy was aimed at increasing the energy density of the usual complementary food. Only two of these trials had a significant impact on growth (John & Gopaldas 1993; Moursi et al. 2003). In the other three (Mamiro et al. 2004; Hossain et al. 2005a; Owino et al. 2007), there was no increase in energy intake, so the lack of impact on growth is not surprising. The average effect size across all trials was 0.35 (range −0.13, 1.37) for weight and 0.23 (range −0.25, 0.71) for linear growth.

1, 2 compare the effect sizes for growth across each category of intervention. The average effect sizes are in the small to medium range, which is in agreement with estimates from the previous review of interventions completed between 1970 and 1997 [effect size generally 0.10–0.50 (Caulfield et al. 1999)].

Morbidity: Only 10 of the intervention studies included data on morbidity outcomes. In most of these, there were no significant effects on morbidity. Most studies included morbidity as a secondary outcome and were not designed or powered to detect differences in morbidity. Two of the educational interventions showed a beneficial effect: a reduction in diarrhoea in Brazil (Vitolo et al. 2005) and a reduction in upper respiratory infection in Vietnam (Schroeder et al. 2002). The fortified‐milk study in India demonstrated a significant reduction in both diarrhoea and acute lower respiratory illness (Sazawal et al. 2007), and a study evaluating home fortification with a micronutrient powder (‘Sprinkles™’) in Pakistan showed beneficial effects on diarrhoea and fever (Sharieff et al. 2006). However, in three studies the interventions were associated with increased symptoms of morbidity. This was evident in food supplementation interventions in Bangladesh [during the first 2 months of the intervention (Roy et al. 2005)] and in India (Bhandari et al. 2001) and in an energy‐density intervention in Congo (Moursi et al. 2003). In India, the adverse effects on fever and dysentery could have been due to the reduction in breastfeeding that occurred in the intervention group. Unhygienic preparation and storage of complementary foods is another possible explanation for adverse effects of these interventions on morbidity.

Behavioural development: Only four studies, all efficacy trials, included data on behavioural development. The provision of a fat‐based fortified food product or micronutrients alone improved gross motor development in Ghana (Adu‐Afarwuah et al. 2007) but these types of interventions did not have any significant effect on developmental outcomes in South Africa (Oelofse et al. 2003) or India (Dhingra et al. 2004). Positive results of supplementation with extra energy in Indonesia were seen only in a subgroup (Pollitt et al. 2002).

Micronutrient intake: Only a few studies reported data on iron, zinc and vitamin A intakes. Education for mothers significantly increased child iron intake in Malawi, India and Peru, but did not have any significant effect on intakes in Brazil. Taking those four studies together, the intervention increased iron intake from complementary foods by 24% (range −7%, 60%) and zinc intake by 26% (range 9%, 53%). Despite those increases, mean iron and zinc intake from complementary foods was still well below recommended intakes in some sites. In Brazil (Santos et al. 2005) a large‐scale food supplementation programme failed to have an impact on intakes of these three micronutrients. There was also no impact of traditional processing of complementary foods in Tanzania (Mamiro et al. 2004). The largest impact on micronutrient intakes resulted from fortification strategies, which increased iron intake by 145–207% in Mexico and Ghana, zinc intake by 201–271% in Ecuador and Ghana, and vitamin A intake by 107% to more than 2300% in Ecuador and Ghana.

Anaemia and iron status: Four studies of educational interventions included data on anaemia and/or iron status. In India and China there was an increase in mean haemoglobin but in Nicaragua and Brazil there was no significant effect. The difference in impact across studies could be due to the specificity of the messages regarding enhancement of iron intake in the two former studies, compared with the latter two projects. Overall, for these four studies the average impact was an increase of 4 g L⁻¹ in mean haemoglobin and a reduction in the prevalence of anaemia of 5 percentage points.

In 12 studies, the target group was provided with a complementary food that was fortified with iron (and sometimes other micronutrients as well). The comparison group received either no additional food (five studies: two efficacy trials and three programme evaluations), or an unfortified complementary food (seven efficacy trials). For the former group of five studies, the average impact was an increase of 4 g L⁻¹ in mean haemoglobin and a reduction in the prevalence of anaemia of 13 percentage points. For the latter group of seven studies, the average effect was an increase of 6 g L⁻¹ in mean haemoglobin and a reduction in the prevalence of anaemia of 17 percentage points.

Another seven studies (five efficacy trials, two programme evaluations) evaluated the effect of home fortification of complementary foods using powders, crushable tablets or fat‐based products. Taking these seven studies together, the average impact was an increase of 8 g L⁻¹ in mean haemoglobin and a reduction in the prevalence of anaemia of 21 percentage points.

Some of the above studies included direct assessments of iron status, such as ferritin values. In most cases, the impact on the prevalence of iron deficiency was greater than the impact on anaemia, indicating that other factors such as malaria contribute to the persistently high rates of anaemia in certain populations.

Zinc status: Only five studies reported plasma zinc concentrations, all of which involved evaluation of a fortified complementary food (three efficacy trials, one programme evaluation), or a home‐fortification product (efficacy trial). The fortified foods provided 3–6.5 mg day⁻¹ zinc, and the daily home‐fortification ‘foodlet’ (crushable tablet) provided 10 mg day⁻¹. In the four studies using fortified foods, none demonstrated a significant difference between intervention and control groups in mean plasma zinc concentration or the percentage of children with low plasma Zn. In the foodlet intervention trial in South Africa, the group receiving daily micronutrients had significantly higher plasma zinc than the placebo group (Smuts et al. 2005). Overall, these results indicate that complementary foods fortified with multiple micronutrients, including zinc, have little impact on plasma zinc concentration, perhaps because of the relatively low bioavailability of zinc when consumed with cereal‐based or cereal/legume blend foods.

Vitamin A status: Seven intervention studies to evaluate the impact of a fortified complementary food (three efficacy trials, two programme evaluations) or home‐fortification products (two efficacy trials) included data on vitamin A status. There was a significant impact on mean serum vitamin A concentration in four of the five interventions using fortified complementary foods, and a reduction in the incidence of vitamin A deficiency in the two studies (of these five) that evaluated this outcome. There was no significant impact on serum vitamin A concentration in the two studies using home‐fortification products, which the investigators attributed to widespread participation in vitamin A supplementation programmes that occurred during the study time period. Taken together, these seven studies indicate that complementary foods fortified with vitamin A can reduce the incidence of vitamin A deficiency (by an average of ∼−13 percentage points in the two studies that reported this), although this impact may be obscured by concurrent vitamin A supplementation programmes.

Conclusions: The results of this review indicate that there is no single universal ‘best’ package of components in complementary feeding interventions because the needs of the target population vary greatly. The impact of such interventions is thus context specific, and depends on factors such as the initial prevalence of malnutrition, the degree of household food insecurity, the energy density of traditional complementary foods and the availability of micronutrient‐rich local foods.

Child growth was the most common outcome measured, but it may not be the most sensitive indicator of benefit because of other constraints that limit the extent to which a child's growth (particularly height) can respond to post‐natal interventions. The impact of these interventions on child growth was mixed. When the primary approach was education about child feeding, interventions that included a strong emphasis on feeding nutrient‐rich animal‐source foods were more likely to show an effect. When a complementary food was provided, with or without concurrent strategies such as nutrition education, the studies in Africa and South Asia generally showed positive effects, while those in other regions were more variable. This may be related to the relatively high prevalence of food insecurity in Africa and South Asia. In such contexts, providing additional food – not just education – may facilitate the ability of families to follow complementary feeding guidelines.

In several studies, the impact of providing a complementary food, in combination with nutrition education, was evident only in the younger children. This underscores the importance of beginning complementary feeding programmes during infancy, when nutrient needs relative to energy intake are the highest and the ability of the child to respond to a nutritional intervention is the greatest.

Because most interventions in which a complementary food was provided used fortified foods, it is not possible to determine whether the positive effects on growth are due to greater energy/protein/fat intake, greater micronutrient intake, or the combination. It is noteworthy that the interventions in which micronutrient fortification was the sole component (i.e. comparisons of fortified vs. unfortified complementary foods, or evaluations of home fortification) generally had little or no effect on growth. Further research on the biological mechanisms underlying growth effects, including the potential roles of milk protein and essential fatty acids, is needed.

Increasing the energy density of complementary foods may have a positive effect on growth when the traditional complementary food has a low energy density and infants are unable to adequately compensate by consuming a higher volume or being fed more frequently. However, before including this strategy in a complementary feeding programme, it is advisable to first demonstrate that increasing energy density of the traditional food will actually result in increased total daily energy intake (including energy intake from breastmilk). It should be noted that increasing energy density will not necessarily result in adequate micronutrient intake, so this strategy should be accompanied by other efforts to improve dietary adequacy.

The potential for an impact on growth appears to be greater with interventions using key educational messages, provision of complementary food with or without fortification, or increased energy density of complementary foods than with interventions based on fortification alone. Although the effect sizes for growth were generally modest (0.1–0.5), the potential impact is larger (0.5–0.6) if programmes are optimally designed and implemented. Furthermore, the impact on the lower tail of the distribution – that is, on stunting rates – could be considerably larger than the effect on the mean height z‐score. In general, effect sizes for growth of interventions providing complementary foods were greater for efficacy trials than for programmes. This is not surprising, given the logistical challenges of ensuring consistent delivery of food (and education) in large‐scale programmes.

Some of the complementary feeding interventions reviewed had a beneficial impact on morbidity rates, but there is the potential for adverse effects of strategies such as food supplementation and increased energy density. This may be due to excessive displacement of breastmilk and/or unhygienic preparation and storage of complementary foods. This highlights the need to couple complementary feeding interventions with counselling regarding continued breastfeeding, responsive feeding and hygienic practices.

There is very little information on the impact of complementary feeding interventions on behavioural development, but recent studies in infants have yielded promising results. It is important to include assessments of behavioural development in such evaluations, as these outcomes may be more sensitive to improvements in child nutrition than outcomes such as growth and morbidity.

With regard to micronutrient intake, the results of educational interventions indicate that it is difficult to achieve adequate iron intake from unfortified local foods at 6–12 months of age. Fortification (either processed complementary foods or home fortification) is the most feasible option in most circumstances given the cost of iron‐rich foods (such as liver or meat). Adequate zinc and vitamin A intakes can be achieved from local foods, but this requires very careful attention to dietary choices. Fortification can help ensure zinc and vitamin A intakes when nutrient‐rich local foods are costly or unavailable (e.g. seasonally).

The results also indicate that fortification can be highly effective at improving iron and vitamin A status. Although this could be accomplished by other strategies, such as iron or vitamin A supplementation, using complementary foods as the vehicle may be less risky [given recent concerns about adverse effects of iron supplements in certain situations (WHO & UNICEF 2007)] and more acceptable to caregivers. Further research is needed to understand why zinc‐fortified foods have generally little effect on plasma zinc concentrations.

Complementary feeding interventions, by themselves, cannot change the underlying conditions of poverty and poor sanitation that contribute to child malnutrition. They need to be implemented in conjunction with a larger strategy that includes improved water and sanitation, better health care and adequate housing. Nonetheless, the results of this review indicate that carefully designed programmes that include pre‐tested educational messages provided through multiple channels, with fortified foods or home‐fortification products made available depending on the needs of the target population, can substantially improve growth and micronutrient status and may also reduce morbidity and enhance behavioural development. The key challenge is how to implement high‐quality programmes that are sustainable when delivered on a large scale.