Early-Stage Primary School Children Attending a School in the Malawian School Feeding Program (SFP) Have Better Reversal Learning and Lean Muscle Mass Growth Than Those Attending a Non-SFP School^,

Abstract

In developing countries, schoolchildren encounter a number of challenges, including failure to complete school, poor health and nutrition, and poor academic performance. Implementation of school feeding programs (SFPs) in less developed countries is increasing and yet there is mixed evidence regarding their positive effects on nutrition, education, and cognition at the population level. This study evaluated cognitive and anthropometric outcomes in entry-level primary school children in Malawi with the aim of generating evidence for the ongoing debate about SFPs in Malawi and other developing countries. A total of 226 schoolchildren aged 6–8 y in 2 rural Malawian public primary schools were followed for one school year. Children attending one school (SFP school) received a daily ration of corn-soy blend porridge, while those attending the other (non-SFP school) did not. Baseline and post-baseline outcomes included the Cambridge Neurological Test Automated Battery cognitive tests of paired associate learning, rapid visual information processing and intra-extra dimensional shift, and anthropometric measurements of weight, height, and mid-upper arm circumference (MUAC). At follow-up, the SFP subcohort had a greater reduction than the non-SFP subcohort in the number of intra-extra predimensional shift errors made (mean 18.5 and 24.9, respectively; P-interaction = 0.02) and also showed an increase in MUAC (from 16.3 to 17.0; P-interaction <0.0001). The results indicate that the SFP in Malawi is associated with an improvement in reversal learning and catch-up growth in lean muscle mass in children in the SFP school compared with children in the non-SFP school. These findings suggest that the Malawian SFP, if well managed and ration sizes are sustained, may have the potential to improve nutritional and cognitive indicators of the most disadvantaged children.

Introduction

Global primary education initiatives, such as the United Nations’ declaration on universal primary education (1), have led to an increase in the number of children enrolling for primary education in developing countries, albeit their impact on completion rates, standard of achievement, and quality of education has been questioned (2). Although a variety of different causal factors may adversely influence educational outcomes, those specifically linked to the cognitive development and educational attainment of school learners include hunger, poor nutrition, and the health of preschool and school children (3–5). A few randomized controlled trials have reported positive findings regarding the effectiveness of school feeding on nutrition, education, and cognition (6–9), whereas others have failed to do so (10–14).

Despite the conflicting evidence surrounding the effectiveness of school feeding programs (SFPs)⁶, individual governments and the international community are increasingly promoting and adopting the introduction and implementation of SFPs as a strategy to improve children’s general health, school attendance, and academic performance and hunger alleviation (15–17). In most developing countries, SFPs have been initiated and supported by the World Food Program (WFP) of the United Nations, especially in those countries severely affected by childhood hunger and malnourishment (18).The WFP’s programs are aimed at improving the food security situation of children and thereby achieving desirable educational and cognitive indicators such as enrolment, attendance, attention, and nutritional benefits (19). To date, evaluation of the effectiveness of SFPs has mainly been confined to controlled studies of relatively small groups, and only a limited number of studies have attempted to address the overall effectiveness of SFPs at a national level (11, 20, 21). In terms of the effectiveness of SFPs as a means of improving health and education-related outcomes in children, there is a lack of overall consensus in the scientific literature (22–24).

Initially implemented as a pilot program in 1999, the Malawi government and WFP continue to support a SFP in Malawi. Under the funding of WFP, the program offers a free daily ration of corn soy blend (CSB) porridge to all school children in the selected schools and is administered by the district education managers and local schools. The program broadly aims at improving learning of pupils and educational indicators such as school dropout, enrolment, and attendance. In 2011, the program operated in 13 districts covering 681 public primary schools and targeting about 703,630 learners (WFP Malawi Standard Project Report 2011, unpublished data). The effectiveness of the Malawian SFP has been evaluated with respect to its impact on school enrolment, attendance, and relief from short-term hunger (25, 26). To date, however, no study has focused on the effect of the SFP on nutritional or cognitive status in participating Malawian children. This study evaluated the cognitive and anthropometric status of entry-level primary school children at 2 Malawian rural public schools (one an SFP school and the other a non-SFP school as a comparison) with the aim of generating evidence for the ongoing debate about SFPs in Malawi and other developing countries.

Methods

The study was conducted among rural primary school children at the beginning and end of the 2010–2011 academic year. The study participants were recruited at 2 public primary schools, namely Matawale and Songani in the traditional authority Kumtumanje-Zomba District, which is located in the eastern region of Malawi-southern Africa. The selected schools were located at a distance of 8 km apart from each other and were in the same educational management zone of Songani. They were located in communities with broadly similar socio-economic, environmental, agricultural, and food pattern characteristics. One school (Matawale, SFP school) was an existing participant in the national SFP (more than 2 y standing) and the other school (Songani, non SFP school) was not a participant. Standard 1 school children, aged between 6–8 y, were eligible for the study. We excluded children who were aged <6 and >8 y or whose age could not be established, and who had previously been in primary school or who came from the same household as another participant child. A list of potential participants was obtained from each school and parents/guardians of eligible children were invited for information meetings and recruitment. After the screening process, parents of eligible children were then asked to sign 2 copies of consent forms, and the children enrolled into the study gave their assent by turning up and participating in the study. Children were allowed to withdraw from the study, with no questions asked, at any stage if they indicated their reluctance to continue. Data on the age of the study participants were obtained from official documentation, such as hospital birth record, immunization record, and/or birth certificate. In the absence of these documents, the parent/guardian’s memory and calendar of events were used. Out of a total of 418 children who were screened in both schools, 114 and 112 children were recruited at the Matawale SFP and Songani non-SFP schools, respectively (Fig. 1). These children were recruited into the study for a period of one academic year (September 2010 to July 2011) and their measurements were taken and testing conducted at baseline (at the start of the academic year) and follow-up (at the end of the academic year).

FIGURE 1.

Flow chart of participants included in the study. ¹SFP, school feeding program. ²Children who have previously been in primary school, i.e., for whom this was not the first school year. SFP, supplementary feeding program.

Sample size was calculated based on an estimated biological weight gain that would differentiate SFP participants and non-SFP participants with respect to the use of the supplement (CSB porridge) over a given time period. Assuming that Malawian children would grow at an estimated −3 SDs of the WHO weight-for-age references, power calculations indicated that a sample size of 72 children per school was required to detect a mean weight gain difference of 0.9 kg with the same variability between the 2 schools after 8 mo at 95% CI (α = 0.05) and 80% power (β = 0.2). Allowing for an estimated attrition rate of 40% owing to anticipated dropouts during the Standard I school year, the targeted sample size for each of the 2 cohorts was set at 120 children. Ethical approval for the study was granted in the United Kingdom by the University of Ulster Research Ethics Committee (27 May 2010, ref. no. REC/10/0082) and in Malawi by Chancellor College Research Ethics Committee of the University of Malawi (6 September 2010, ref. no. CHAREC/R1/10/7). The Malawi Ministry of Education provided official permission and authorization to conduct the study in Malawian schools (letter dated 13th July 2010 ref. no. 7/1/10).

In each school, a list of all children deemed eligible for the study was presented to the research team for an initial eligibility check. Based on the provided list, pupils who appeared to have qualified for the study under the initial check were given invitation letters inviting their parents/guardians to come to the school grounds on a specified date for an information meeting, for further eligibility screening of the pupils, and to seek parental consent. All communications were conducted in a nationally known language, Chichewa.

Daily administration of government/WFP CSB porridge (SFP school only).

Throughout the period of study (September 2010 to July 2011), daily administration of the fortified porridge at the SFP school was carried out by community volunteers in adherence to government/WFP instructions. The project researchers were not involved in the administration of the fortified porridge. CSB flour fortified with various micronutrients was used as the supplement in the SFP and provided an estimated 350 kcal/100 g. Standard SFP procedures in Malawi normally outline that children in the SFP school are to receive 100 g of CSB porridge on each school day throughout the academic year. However, owing to national scarcity of CSB at the time of this study, instructions were handed down by the government/WFP for the daily ration of CSB flour to be reduced by a factor of 25%. Therefore, for the greater duration of the study reported in this paper, pupils at the SFP school received 75% of the normally recommended ration, with each child thus receiving an estimated 263 kcal/d and with all other nutritional components similarly reduced. The serving of the porridge took place at ∼0900 h each school day (half-way through the daily learning time). The nutrient composition of the CSB is shown in Table 1. The porridge was prepared at the SFP school by community volunteers (mostly women) who operated on a village-by-village rotational basis and who were members of a school feeding committee.

TABLE 1.

Nutrients provided by the CSB supplement compared with the percentage of the RNI for children aged 7–9 y¹

Nutrient	Amount per 100-g ration of standard CSB porridge	Amount in a 25% reduced ration of CSB porridge	Percentage RNI2 contributed by 25% reduced ration of CSB porridge
Energy, kcal	350	263	16.03
Vitamin A, μg	166	125	25.0
Thiamin, mg	0.1	0.1	11.0
Riboflavin, mg	0.4	0.3	37.0
Niacin, mg	4.8	3.6	30.0
Folic acid, μg	60.0	45.0	15.0
Vitamin B-12, μg	1.2	0.9	50.0
Vitamin C, mg	48.0	36.0	103.0
Iron, mg	8.0	6.0	33.0
Calcium, mg	100	75.0	11.0
Zinc, mg	5.0	3.8	33.0

¹CSB, corn soy blend; RNI, recommended nutrient intakes.

²Percentages calculated based on WHO/FAO vitamin mineral requirements (27).

³Based on FAO/WHO/UNO expert consultation technical report (28).

Anthropometric measurements.

Anthropometric measurements were carried out following recommended procedures (29). Children’s body weight was measured to the nearest 0.1 kg while they were dressed in light clothing without footwear using an electronic digital SECA 877 flat weighing scale (SECA). Height, without shoes, was measured to the nearest 0.1 cm by using a SECA Leicester height measure (with a fixed foot plate and movable headboard). Reusable Lasso-o tapes were used to measure the mid-upper arm circumference (MUAC) of the children. All anthropometric measurements were carried out by trained research assistants who had previous experience in anthropometry. These measurements were carried out on a veranda at each of the 2 schools. In the case of the SFP school, anthropometric measurements were carried out before the children had received their daily porridge portion to ensure that total body weight measurements were not affected.

Cognitive testing.

Cognitive testing was carried out using a series of tests from the Cambridge Neurological Test Automated Battery (CANTAB), which permits separate assessment of behavioral domains involved in cognition, such as learning, set-shifting, memory, and attention (30). CANTAB computerized testing has been used in children as young as 4 y of age (30) and has been shown to be sensitive to various conditions (31–35). The testing was carried out in a quiet test room, which assured a calm environment, by the lead researcher who was trained in the use of CANTAB tailored to small children. The 3 brain cognitive domains tested were memory, reversal learning, and attention, all of which have been related to aspects of under-nutrition (36–39). These were tested using, respectively, the following CANTAB subtests: paired associate learning (PAL), intra-extra dimensional shift (IED), and rapid visual information processing (RVP). The test-retest reliability coefficients of the most common outcomes of the selected tests are: PAL stages completed, r = 0.87; PAL total errors, r = 0.68; IED stages completed, r = 0.75; IED total errors, r = 0.40; RVP total hits, r = 0.64; and RVP latency, r = 0.64.

For PAL, a measure of visual memory and new learning, a series of boxes on the screen open one at a time to reveal a hidden pattern. After all the patterns have been shown, the patterns are presented one at a time and the individual must remember in which box was the pattern. Several measures of PAL were examined. Stages completed reflect the total number of patterns to be remembered, which increases from 2 to 8, and PAL stages completed on the first trial reflects number of patterns correct on the first trial (maximum = 10 trials/stage). The first trial memory score indicates the number of patterns correctly located after the first trial summed across all the stages completed. Because the total number of errors reflects the number of stages completed by an individual, PAL total errors are adjusted by summing the number of patterns not attempted and subtracting the number of patterns divided by the number of boxes from it and multiplying the result by the number of trials allowed for the stage. PAL mean trials to success calculates the total number of trials required to correctly locate all the patterns in all stages attempted and divides the result by the number of successfully completed stages.

For the IED, a test of rule acquisition and reversal (set-shift), a series of discrimination problems and their reversals are presented across 7 stages and require the participant to continue to touch the stimulus thought to be correct. Once discrimination is learned, the correct stimulus is reversed in the next stage. Stage 8 reflects a shift of the discrimination problem to a stimulus dimension that the participant has previously learned to be irrelevant, and a reversal of this discrimination problem is presented in Stage 9. Measures included the number of stages completed and errors made. IED pre-extra dimensional (Pre-ED) errors are summed across Stages 1–7 (prior to the shift) and IED extra dimensional shift (EDS) errors represent the total errors during Stages 8–9. Because participants may not complete all stages, a total errors adjusted score is calculated by adding 25 for each stage not attempted owing to failure.

The RVP, a measure of visual sustained attention, requires the participant to detect target sequences of 3 digits by pressing a key each time a target sequence is detected. Mean latency reflects the mean time taken to respond in milliseconds. Total false alarms reflect the number of times the participant responds outside the target sequence window and total misses reflect the number of occasions on which the participant fails to respond to a target sequence within the targeted time window.

At both baseline and follow-up in the SFP school only, cognitive testing of the study children was undertaken before they had received their daily portion of porridge with the aim of avoiding any acute effect on the child’s alertness or mental state as a result of food intake. To rule out any possible visual and color blindness problems before carrying out the CANTAB tests, children were screened for basic visual acuity and color identification/differentiation using Kay standardized pictures and the ISHIHARA’s Color blindness screening test (14-Plate edition), respectively. All participants were given testing instructions and procedures in Chichewa before they proceeded with initial CANTAB motor screening tests. All enrolled children were able to go through the designated tests and had no specific problems in terms of what they were told to do. Care was taken throughout testing to ensure that all children received the same instructions and that none received extra “coaching” or repeat testing (to avoid any potential bias in results arising from “test practice effects”).

Demographic/socio-economic questionnaire-based survey.

A team of 4 trained and experienced Malawian research assistants administered a piloted, structured, and translated questionnaire to the parents/guardians of each participant child. Data relating to the occupation (collected in categories of subsistence farming, small-scale business, regular job, and casual employment), religion (no religion, a Christian, or a Moslem), and educational attainments (no education, lower primary, upper primary, or secondary and above) of household heads were obtained as were data regarding household demographic composition (sex, marital status, i.e., single or married, and total number of people in a household), food security (measured in terms of availability of main staple food-maize at the time of the study), and attendance history of the child at a community based child-care center, which is a community-organized and -managed preschool day care facility. The questionnaire-based survey was completed at baseline.

Statistical analysis.

Statistical analyses of the data were conducted on an intent to treat, weighted means basis of the last observation according to their original randomized assignment using SPSS version 19. Continuous variables were assessed for normality using the Kolmogorov-Smirnov test in combination with histograms and box plots. Variables that indicated a skewed distribution were appropriately log-transformed prior to conducting parametric statistical analyses. Baseline differences between schools in relation to dichotomous variables were analyzed using chi-square tests and differences in continuous variables were analyzed using independent t tests. A repeated-measures ANOVA with effects for feeding, time, and their interaction was employed to assess changes in anthropometric measures of weight and height and all cognitive measures between the schools (i.e., SFP vs. non-SFP) over time. Owing to baseline differences between schools in relation to occupation and religion of the household head and the MUACs of children from the non-SFP school cohort, these variables were treated as potential confounders and were therefore adjusted for in the repeated-measures ANOVA. ANCOVA was carried out on follow-up MUACs with baseline MUAC, occupation, and religion of household head as covariates. A priori decision was made to adjust for age and sex in all the main effects analysis. Values in the text are presented as percentages and means ± SDs and α-level set at P < 0.05.

Results

A total of 418 children were deemed eligible for the study and were screened. Of that number, 226 children were enrolled, of whom 50.4% (n = 114) were from the SFP school and 49.6% (n = 112) from the non-SFP school. At follow-up, 190 children remained in the study, with a total of 36 (16%) dropouts owing to discontinuation at school (n = 31) or transfer to other schools (n = 5) (Fig. 1). A lower dropout rate (12.3%) was observed in the SFP school cohort compared with the non-SFP school cohort (19.6%) (Fig. 1). The sample comprised 50.4% males and 49.6% females and their overall age at commencement of the school year was 6.6 ± 0.5 y, with a mean age of 6.6 ± 0.5 y in the SFP school and 6.5 ± 0.5 y in the non-SFP school. With the exception of MUAC, which was higher in the non-SFP school, and occupation and religion of household head, there were no significant differences in cognitive and anthropometric measures at baseline. Baseline demographic characteristics are shown in Table 2.

TABLE 2.

Baseline socio-demographic characteristics of households of children enrolled in the study¹

Variable	Feeding school (n = 114)	Non-feeding school (n = 112)	P value
Sex of household head, n (%)
Male	89 (78.1)	95 (84.8)
Female	25 (21.9)	17 (15.2)	0.19
Marital status of household head,2 n (%)
Lone parent	27 (23.7)	22 (19.6)
Married	87 (76.3)	90 (80.4)	0.46
Education of household head, n (%)
Standard 1–4	23 (20.2)	17 (15.2)
Standard 5–8	53 (46.5)	46 (41.0)	0.20
Secondary +	29 (25.4)	43 (38.4)
None	9 (7.9)	6 (5.4)
Occupation of household head, n (%)
Subsistence farmer	41 (36.0)	24 (21.4)
Small-scale business	27 (23.7)	43 (38.4)
Regular job	16 (14.0)	18 (16.1)
Casual employment	30 (26.3)	27 (24.1)	0.04
Religion of household head, n (%)
Christianity	86 (75.4)	62 (55.4)
Moslem	28 (24.6)	50 (44.6)	<0.001
Households with staple food,3 n (%)
Yes	110 (96.5)	109 (97.3)
No	4 (3.5)	3 (2.7)	0.72

¹Reproduced with permission from (40).

²Includes widowed, divorced/separated, never married.

³The staple food is maize.

Effects of the CSB porridge on anthropometric outcomes.

Anthropometric measurements at baseline and follow-up are presented in Table 3. Children attending both schools significantly increased in weight and height between the 2 time points. MUAC significantly increased in the SFP school but did not undergo significant change in the non-SFP school. Considering the effects of feeding, time, and their interaction, an effect was noted in relation to MUAC (P-interaction < 0.0001). Post-hoc analysis indicated that there was a significantly greater increase in the MUAC measurement in the SFP school compared with the non-SFP school (P < 0.0001) after controlling for baseline MUAC. No further difference between the 2 schools in any of the other anthropometric outcomes was observed.

TABLE 3.

The effects of CSB feeding on the anthropometric outcomes of children in feeding and non-feeding schools after 1 school year¹

	Feeding school (n = 114)		Non-feeding school (n = 112)
Anthropometric measure	Baseline	Follow-up	Baseline	Follow-up	P-interaction
Weight, kg	18.1 ± 2.5	19.3 ± 2.5	17.7 ± 2.3	19.1 ± 2.1	0.09
Height, cm	109.7 ± 6.4	113.5 ± 6.0	109.1 ± 5.6	113.0 ± 5.1	0.75
MUAC, cm	16.3 ± 1.3	17.0 ± 1.2	17.1 ± 1.2	17.0 ± 1.1	<0.0001

¹Values are means ± SDs. CSB, corn soy blend; MUAC, mid-upper arm circumference.

Effects of the CSB porridge on cognition.

At baseline, the 2 schools did not significantly differ in any of the CANTAB learning, reversal learning, and attention outcomes. At follow-up, a feeding × time interaction effect was evident for IED Pre-ED errors (P-interaction = 0.02). Post-hoc analysis indicated that there was a significantly larger decrease in IED Pre-ED errors in the SFP school compared with the non-SFP school (Table 4).

TABLE 4.

The effects of CSB feeding on the CANTAB cognitive outcomes of children in feeding and non-feeding schools after 1 school year¹

	Feeding school (n = 114)		Non-feeding school (n = 111)
CANTAB outcome	Baseline	Follow-up	Baseline	Follow-up	P-interaction
CANTAB learning/memory outcomes
PAL first trial memory score	4.3 ± 3.9	7.1 ± 4.0	4.2 ± 3.6	6.7 ± 4.0	0.52
PAL mean trials to success	4.5 ± 1.5	3.4 ± 0.9	4.5 ± 1.4	3.5 ± 1.1	0.19
PAL stages completed	2.8 ± 1.3	3.8 ± 1.0	2.7 ± 1.4	1.8 ± 1.0	0.17
PAL stages completed on first trial	1.1 ± 0.9	1.8 ± 1.0	1.2 ± 1.0	3.5 ± 1.1	0.10
PAL total errors (adjusted)	75.6 ± 18.9	60.7 ± 19.2	74.8 ± 17.9	63.8 ± 19.7	0.13
CANTAB reversal learning outcomes
IED EDS errors	9.7 ± 13.4	15.1 ± 11.6	8.3 ± 12.3	10.8 ± 10.6	0.20
IED Pre-ED errors	30.4 ± 14.8	18.5 ± 13.0	31.3 ± 15.2	24.9 ± 13.5	0.02
IED stages completed	3.6 ± 3.2	6.3 ± 2.5	3.4 ± 3.2	5.4 ± 3.1	0.28
IED total errors (adjusted)	152 ± 68.4	85.1 ± 59.1	157 ± 69.9	111 ± 69.8	0.12
CANTAB attention outcomes
RVP A’ score	0.8 ± 0.1	0.9 ± 0.1	0.8 ± 0.1	0.9 ± 0.1	0.25
RVP mean latency	871 ± 198	841 ± 177	876 ± 189	820 ± 157	0.54
RVP total false alarms	38.2 ± 25.3	19.1 ± 11.8	41.4 ± 27.3	22.1 ± 15.3	0.89
RVP total misses	10.1 ± 5.0	9.7 ± 4.2	10.3 ± 5.2	8.9 ± 4.3	0.33

¹Values are means ± SDs. CANTAB, Cambridge Neurological Test Automated Battery; CSB, corn soy blend; ED, extra dimensional; EDS, extra-dimensional shift; IED, intra-extra dimensional shift; PAL, paired associate learning; RVP, rapid visual information processing.

Discussion

Findings from this study suggest that children participating in the SFP had anthropometric and cognitive benefits (reversal learning) compared with those children who attended a nonparticipating school. During the time period tested, a significant increase in MUAC was observed in the SFP school cohort but not in the non-SFP cohort (after adjustment for baseline differences in MUAC), and a significantly greater decrease in IED Pre-ED errors was observed in the SFP school cohort compared with the non-SFP school cohort. Despite the many challenges it faces, the Malawian SFP may have the potential to improve nutritional and cognitive outcomes among the targeted beneficiaries and the need for SFP is cross cutting even in schools that are not participating in SFPs.

In this study, our finding in relation to MUAC is in agreement with findings of previous studies carried out in other African countries, which have reported both significant and nonsignificant results (6, 10, 41, 42).

The comparison of follow-up and baseline data indicated a significant decrease in IED Pre-ED errors in both schools. Despite such being the case, the decrease in the SFP-school cohort was greater than the decrease in the non-SFP school cohort. This finding may indicate that the CSB had a positive effect on cognition in the SFP schoolchildren. Although IED Pre-ED errors were the only significant outcome, the pattern of outcomes suggests a preferential targeting of executive functions given that PAL and RVP were not significantly affected (although performance in the non-SFP school cohort was always in a worse direction than for the SFP school cohort). PAL is considered a measure of visual memory, whereas IED is a measure of learning/executive function. The increase in errors suggests a functional deficit in which these children did not learn the rules governing which pattern was correct and/or that they were more “perseverative,” failing to switch behavior as the rules switched, functions that have been largely tied to frontal cortex. It has been argued that CANTAB tests are subject to “practice effects,” enabling participants to “perform better” at repeat or follow-up (43, 44). Nonetheless, an increasing body of literature not only attests to their validity (45) and suitability for children (46) but also provides guidance that minimizes/avoids such effects (47, 48). Even allowing for some minor contribution of practice effects to the reductions in number of IED Pre-ED errors observed in both cohorts, the significantly greater reduction observed in the SFP-school cohort appears to indicate a positive effect on the part of the CSB porridge regarding this cognitive outcome. The high number of and lack of significant difference in IED EDS errors between the 2 schools may suggest that fewer children make it to levels 8 and 9, i.e., to the set shift. These few children, if we consider them “smarter,” “more attentive to the environment,” or especially “more flexible,” may be less susceptible to the effects of nutritional deficiency and thus the beneficial effects of the intervention do not show up under such conditions. As a result of the smaller number of children that reach that phase (set shift), it may also be harder to detect significant differences between the groups.

Although some studies have found that micro-nutrients, such as iron, vitamin A, vitamin B-12, and zinc, are associated with improved cognitive performance in school-age children (49–52), there appears to be no clear consensus in the literature regarding the effects of dietary supplementation in relation to cognition (22, 24, 46, 53–58). According to the report of the Malawi national micronutrient survey of 2009, micronutrient deficiencies of vitamin A and iron were present in 33 and 27% of Malawian school children, respectively (59). Bearing in mind that the CSB was fortified with several micronutrients (including iron, vitamin A, vitamin B-12, and zinc), our finding on cognitive function would suggest that children in the SFP school may have derived cognitive benefit from one or more of the micronutrients in the porridge.

School attendance and pupil retention are among the key scholastic objectives of the SFP in Malawi. Difficulties in collecting reliable and up-to-date school attendance data in African countries have been reported (60). In this study, it was not possible to obtain reliable day-to-day attendance data from the schools owing to the practical difficulties faced by teachers in gauging the attendance of large cohorts of pupils; however, it is perhaps worthy of note that our study cohort experienced a 19.6% drop-out rate among non-SFP school children compared with the drop-out rate of 12.3% among the SFP school children. This finding is consistent with the reported positive effects of SFPs on attendance and pupil retention (61).

There is ample evidence of the acceptance of CSB in various settings among Malawian children (62–64) and school children in other African countries (65). Additionally, there was a high degree of enthusiasm for eating the porridge that was received at the SFP school each day. Other factors are more likely to provide some plausible explanations for the general lack of significant differences between the 2 schools. For example, economic pressures at the national level during the study were such that, rather than receiving the standard daily ration of 100 g of CSB porridge, SFP schools were instructed to reduce the daily ration of CSB flour by a factor of 25% (Ministry of Education, letter dated 30/08/2010, ref. No. ED/VOL 182/08, unpublished data). This temporary directive was supposed to remain in force for 3 mo (September, October, and November 2010) but in fact remained in force for almost the whole academic year (from August 2010 to May 2011), and clearly may have reduced any anthropometric and cognitive benefits derived by the SFP-school children. Another factor that may have had a negative effect, albeit a minor one, on CSB-derived benefits is the issue of local influences, such as ad hoc unofficial holidays and community funerals, which resulted in closing the school with little warning and eliminating the provision of porridge on those days. Additionally, there is the possibility that some families whose children were participating in the SFP may have effectively regarded the daily CSB ration as a replacement for a part of the daily food provision that was previously being given to the children and thus cutting back on their own daily food provision to these children. In an earlier study conducted in Kenya, where the effects of a SFP were being assessed, it was reported that some families of SFP-participant children reduced the quantity of food being provided at home to those children (66).

Our study had limitations. The results are based on entry-level primary schoolchildren and we do not know whether the benefits would be the same with older children and more advanced classes (Standard 2–8). Nevertheless, we consider this study a pointer to what could be the potential impact of SFPs in Malawi bearing in mind that this, to the best of our knowledge, is the first study to use a computer-based, validated, and standardized methodology to assess cognition among schoolchildren participating in a SFP in Malawi. We included 2 schools out of the many possible schools and a larger sample size would have granted us more confidence on the generalizability of our findings. We conclude that the Malawian SFP may have the potential to improve nutritional and cognitive indicators, albeit there are plausible national cost and local implementation challenges that may influence the achievement of the intended benefits and that need to be addressed to maximize the benefits of these SFPs. Lack of major differences between the 2 cohorts at baseline suggest that children from the non-SFP targeted schools are in equal need of intervention.