An important finding of this study is that VA deficiency at school age is a serious public health problem in the intervention areas, since 47.2% in Kaya and 37.1% in Bogandé had low serum retinol at baseline, whereas the cut-off for a severe public health problem is 20% low serum retinol according to WHO [33
]. This confirms previous findings in school-children in Niger [32
], with a 45% baseline rate of low serum retinol. In the 15 RPO schools in Kaya and in the 8 RPO schools in Bogandé, the rate of low serum retinol was down to 13% and 15%, respectively, so that the VA deficiency went from a severe to a moderate public health problem, after an average of 28 and 51 RPO fortified meals in Kaya and Bogandé respectively, over a year. These findings are in accordance with previous studies showing the efficacy or effectiveness of RPO among preschool children [27
], pupils [34
], and reproductive age women [4
]. As suggested by Wasanwisut [35
], the intervention was considered effective since the deficiency rate was down to 15% or less in all intervention groups.
The VA supplied by the RPO supplement over the test year amounted to approximately 42 mg RAE in Kaya and 76.5 mg in Bogandé, which is close to the 60 mg provided by a single VA capsule if we use 6:1 as conversion factor for β-carotene to retinol. Had we used the newly recommended conversion factor of 12:1 [36
], the total amount of VA provided as RPO would have represented around half of the dosage of a VA capsule. It is interesting to note that the RPO and the single VA capsule had a nearly equivalent impact on serum retinol in Bogandé school children. RPO was not found more effective than retinol supplements in our study, however, which is at variance with others [34
]. It may be simply a matter of dosage or duration of the RPO fortification in our study, or else, it may be due to the fact that the interval between VA capsule administration and the endline serum retinol measurement was slightly shorter (5 months) than the interval between the last RPO meal and the endline serum retinol measurement (5.5 – 6 months).
Mean final serum retinol in Kaya pupils was nearly twice as high as that of Bogandé RPO pupils, in spite of the fact that the former had received roughly half as much RPO as the latter. This may reflect the fact that in Kaya, pupils were still receiving RPO supplements when endline blood samples were collected for retinol determination, whereas in Bogandé, RPO supplementation was interrupted for school recess 5.5 to 6 months before blood sampling, so that VA stores could be more depleted.
In Bogandé, the rate of VA deficiency at baseline was lower than in Kaya. This is unquestionably due to the different timing of the survey, which took place in the rainy season with plenty of green leaves and mangoes in Bogandé, and during the dry and lean season in Kaya. This may also be why in Kaya, the few children who had received a VA capsule in the course of the National Micronutrient Days 6 months or more before the baseline study did not have a better VA status compared to other pupils.
In both sites, our findings support previous studies showing that initially deficient subjects derived the most benefit from the VA supplement, whether in the form of RPO or a single VA capsule [4
]. In initially deficient pupils of Bogandé, serum retinol increased by 87.5% with the VA capsule and by 63.1% with RPO meals. In VA replete subjects, there was no further increase in serum retinol; there was even a tendency for the reverse, with 12% of the normal pupils at baseline showing a low serum retinol value at endline (VA capsule or RPO treatment). Such paradoxical findings were reported previously, but with synthetic VA, not with food supplements [6
]. It may simply reflect regression to the mean, but further research on the potential adverse effect of VA supplementation among non-deficient children is warranted.
In Kaya and in Bogandé as well, the rate of low serum retinol remained quite high (between 13% and 17%) after the intervention, however. This high residual rate, whether with the VA capsule or RPO "treatment", again shows that a dosage of approximately 60 mg RAE sustains normal VA status for less than 6 months. Among pre-schoolers of the same Kaya area, Zagré et al [4
] had reported that 6 months following a VA capsule distribution among preschoolers with a coverage rate around 90% in Burkina Faso, the rate of low serum retinol was 84.5%. In under-five children of Niger, it was shown that three months following VA capsule administration, the rate of low serum retinol was practically back to baseline level of 38% [11
]. Although this was not the purpose of the present study, we could observe that the benefit of VA capsules was indeed short-lived. In one of the intervention sites (Kaya), 18% of the pupils had received a VA capsule 6 months prior to our baseline survey, but their VA status was not different from that of other pupils. So providing some 60 mg of VA either through RPO fortification of school meals or through a single VA capsule over the test year only alleviates the VAD problem in school children. At least two VA capsules per year if not more, or a higher level of RPO fortification of school meals, or a combination of VA supplementation and fortification, would be required.
Factors other than VA intake may also contribute to low serum retinol values, and these should not be overlooked. One of those factors is underlying infection, which is known to reduce serum retinol [41
] and makes serum retinol non-specific of VA deficiency. We could have used a more sensitive and specific indicator of VA deficiency such as the modified relative dose-response [43
], but the required retinol analog was only available at high cost. We collected information on the occurrence of illness in the previous fortnight, but this variable showed no significant association with serum retinol at either time. Another factor that may explain the persistence of more than 10% low serum retinol values is the presence of concurrent nutritional deficiencies which may act as limiting factors. Zinc deficiency is widely prevalent among children worldwide, and it is known to affect growth [46
] and to interact with VA [47
]. The fact that taller children had a higher serum retinol response in both areas indeed suggests that zinc or protein-energy malnutrition may interfere with VA status improvement. These observations lead to advocate for more global nutritional approaches to micronutrient malnutrition rather than single nutrients, and therefore dietary diversification strategies, along with public health measures to control infection.
Boys are reportedly at higher risk of VA deficiency [48
] although the reasons for their higher vulnerability are largely unexplained. Baseline data did not disclose a better VA status of girls, but their response to RPO in Kaya was significantly higher than that of boys. In Bogandé, sex was not a significant determinant of serum retinol in the co-variance analyses including all three treatment groups. In separate linear regression of endline or change of serum retinol for the capsule and RPO groups, it was found that female sex was associated with a higher response, with the VA capsule but not with RPO supplementation. No explanation for this difference can be proposed.
The interpretations of the findings in Bogandé were obscured by the much better VA status of the negative control group of pupils at baseline. Indeed, the rate of low serum retinol was of the same magnitude as that found in the positive control and RPO groups, but after the intervention. These wide differences underline the disparities that may be found within the population of a relatively small area. Other indices of a better socio-economic status of the negative control group pupils are their significantly higher height-for-age, and the fact that among the 8 schools selected at random to serve as controls, more than half were in villages actively involved in trade. The link between better socio-economic status and better health and nutrition status is well documented [33
]. In all three school groups, BMI was lower at endline than at baseline as the last harvest had been poor, which underlines the vulnerability of the area to food insecurity.