Summary of main results
Our NutriClim study in the Nouna HDSS of 2018 found that children 8–59 months old had a mean HAZ of − 1.27 and a mean WHZ of − 0.47. On average, the children had 4 meals per day. 99% of them integrated cereals, starchy roots, tubers and their products into their meals, which they mainly consumed in the early morning (before 9 am), at noon (12–2 pm) and in the evenings (5–8 pm). Maize, leaves, and oils and fats were consumed the most. Vitamin A-rich leaves make up 94% (Baobab leaves) and oils and fats 92% (shea butter) of the children’s diet, with 10% consuming eggs. 79% of the younger children (8–23 months) consumed mainly mother’s milk. Children consumed a median of 7 out of 10 food groups, and 13 out of 36 food items over a week. DDS and FVS increased linearly with age.
Associations of HAZ and WHZ with dietary habits were in general not robust in the fully adjusted models. In the adjusted model for socio-economic factors, we observed a negative trend in HAZ with an increase in FVS and a positive trend in WHZ for an increase in DDS. Additionally, we showed that after adjustment for socio-economic factors, the associations between dietary factors and HAZ were mitigated. The same results were observed in the sensitivity analysis using stunting (HAZ < − 2) and wasting (WHZ < − 2) as binary outcomes. Four distinct dietary patterns featured by energy source were identified, labelled as “leaves-based diet”, “beans and poultry-based diet”, “maize and fish-based diet”, and “millet and meat-based diet”. Higher DPSs were associated with increasing age. We found that HAZ was negatively associated with the leaves-based diet in the unadjusted model, but not in the fully adjusted model or with any other dietary pattern. By contrast, WHZ was positively associated with the beans and poultry-based diet before adjustments for socio-economic factors and significantly associated with the maize and fish-based diets in the fully adjusted model. Socio-economic factors had a positive association with HAZ, but not with WHZ.
Dietary habits among young children in the Nouna HDSS
The diet of children in the Nouna HDSS relies heavily on foods planted by the household and provided by nature. The main agricultural products in Burkina Faso are cereals (millet, sorghum, maize, rice, fonio), oil seeds (cotton, peanuts, sesame, niébé (beans), soy, voandzou); roots and tubers (igname, patate, manioc, potato); fruits and vegetables (mango, agrumes, tomato, onion, green beans); and sugar canes . The highest portion of the diet is provided by locally produced maize and rice in the form of porridge and combined with a leaves-based sauce and shea butter that can be collected from the surrounding trees. Although only 39% of the mothers reported exclusive breastfeeding for the first 6 months after childbirth, children were on average breastfed until 23 months of age. This is below the global target to have 70% of children exclusively breastfed until 6 months of age by 2030 . In low resource populations, especially from the developing world, diets are often mainly based on starchy staples with little or no animal products and few fruits and vegetables. Yet, plant-based diets tend to have a low quantity of micronutrients, with few being easily absorbed . Subsequently, a closer observation of nutrient intake and uptake would be recommended in the study region.
Our results showed that the children had a median DDS of 7 out of 10 food groups, and a median FVS of 13 out of 36 selected and 117 selectable food items over a week. Other studies such as by Sié et al.  reported an average of 6 out of 11 food groups for children in the Nouna HDSS over a 7-day recall period, while Nikièma et al.  counted only 2 out of 9 food groups for children in the same age group in rural Houndé, Burkina Faso, and with only 25% meeting their minimum dietary diversity of at least 4 food groups. However, the latter recall was done only over the previous 24 h, confirming a low dietary diversity over a 24 h DR compared to over a week. Yet, although a high DDS is considered a good indicator for dietary adequacy , emphasis should also be placed on food and nutrient quantity and diversity . This is often neglected in the literature. Our results showed that higher DDS and FVS of the children were mainly driven by more frequent intakes of energy-dense foods such as cereals, oils and fats, and beans (Fig. 1 and supplementary Table 4), which are essential for weight gain. Therefore, this might explain a positive relationship of DDS and FVS with WHZ. At the same time, foods rich in essential micronutrients (e.g. fruits and animal-based products), which are important for healthy linear growth, only partially contributed to DDS and FVS (Fig. 1 and supplementary Table 4). This may explain some of the inverse association of DDS and FVS with HAZ. Since dietary diversity cannot provide direct information on quantities and nutrient uptake, the association of DDS and FVS with HAZ and WHZ should be interpreted with caution .
Associations between dietary habits and nutritional status
While dietary pattern analyses do not allow observing the effect of specific nutrients on disease risks, it does provide insight on the associations of the overall diet as a combination of food items . Melaku et al.  found a positive association with HAZ for diets with a high intake of dairy, vegetables and fruits (ß-coef. = 0.19). All three food groups were hardly consumed by the children in our sample and did not meet the requirements to be assigned to a dietary pattern. Yet, loading scores for vegetables and fruits were higher in the beans and poultry-based and the maize and fish-based diets, which were not associated with HAZ, but with stunting as a binary outcome compared to the other two diets. A positive association between fish consumption and stunting among 6–23 months old children (OR = 0.947) was also found in Zambia by Marinda et al. . Dairy consumption is especially important for children as it provides important nutrients such as protein, calcium and vitamin A for bone growth. Yet only every fifth child over 24 months of age in the Nouna HDSS reported to have consumed some form of milk (maternal or animal milk) (supplementary Table 4). Considering the lack of dairy, vegetables and fruits in the leaves- and millet- and meat-based diets, they might show important impacts on reducing child stunting. As we controlled for number of days a food item was consumed, the reliability of our results increased.
In our sample of children aged < 5 years, FVS was negatively associated with HAZ and DDS was positively associated with WHZ. This was also the case after controlling for various socio-demographic factors and when tested with stunting and wasting as binary outcomes. Accordingly, wasting and WHZ improved significantly with increasing DDS and FVS, while results for stunting and HAZ showed to be highest in the middle DDS tertile and worsened with increasing FVS (33% and − 1.60 SD). Similar results were reported by Arimond and Ruel  for Mali, a country bordering Burkina Faso. They reported that the middle DDS tertile compared to the lowest DDS had the highest mean HAZ (− 0.11 SD). On the contrary, a higher DDS was related with higher HAZ (0.23 SD), which we cannot confirm for our results.
However, global results on the association of DDS and FVS with HAZ and WHZ seem to be contradictory. Steyn et al.  reports that most studies showed positive correlations between HAZ and WHZ with DDS and FVS, with the exception of WHZ, which showed a negative correlation among children aged 1–3 years. Hatloy et al.  reported no correlation between WHZ and DDS and FVS in rural Mali. In another study in Sri Lanka among children aged 6–59 months of age, dietary diversity was positively associated with HAZ, but not with WHZ, or stunting and wasting as binary outcomes . In Ethiopia, however, DDS was not significantly associated with HAZ among children under 5 years of age . Yet, Sié et al.  report a positive association with stunting and DDS in contrast to wasting, which showed no association, as “dietary diversity scores are reflective of longer-term nutritional habits”. This assumption cannot be confirmed with our study since we show the opposite. Such results indicate the need for further, comprehensive research in the area of child nutritional status and dietary adequacy and to provide attention to single results.
Nutritional status of young children in the Nouna HDSS
Child growth can also be defined by stunting (HAZ < − 2) and wasting (WHZ < − 2) as an indicator for children’s nutritional status. Stunting and wasting continue to be high among children aged < 5 years in the Nouna HDSS . This is a critical time span to develop optimal health, growth and neuro-development [18, 19, 42]. During the lean season in the Nouna HDSS in 2018, 26% of the children were stunted (mean − 1.27 SD) and 7% wasted (mean − 0.47 SD), which are similar results for two study villages in the Nouna HDSS as conducted by Sié et al.  in July 2017. They reported 21% stunting and 10% wasting among children in the same age group. Yet, according to Beiersmann et al. , seasonal differences in child growth development need to be considered. They found that in June 30% and in December 45% of the children < 3 years were stunted in 2009. On the contrary, more children < 3 years were affected by wasting in June (26%) compared to December (16%) in the same year. The results mirror our own findings as the lean season represents low food stocks leading to acute undernutrition, while those affects can only be observed on chronic undernutrition once several months have passed [2, 43]. Hence, this might explain a lack of association between dietary diversity and stunting when measured at the same point in time. The dietary habits of the lean season might show even higher stunting prevalence in the Nouna HDSS a few months later.
Public health research and policy implications
First, we would like to encourage further research on seasonal variations in diets and their link to child growth in order to even better measure the impact of diets on chronic undernutrition. Secondly, we observed a need for nutrition interventions and counselling that do not only provide solutions to continuous child undernutrition, but also prevent food insecurity and provide adaptation and copying mechanisms to a changing environment . Thirdly, further research is needed on the macro- and micronutrient intake to receive a precise feedback on nutrient deficiencies and uptake in order to structure sustainable and successful interventions. In this regard and fourthly, nutrition-related research requires interdisciplinary approaches and cohort studies, which allow combining several aspects rather than looking at single impacts and to consider for temporal differences between seasons and years . Lastly, by clustering the research area varying environmental impacts such as weather variability and extremes can be taken into account as an additional possible explanation for stagnating or even increasing child undernutrition and food insecurity (FAO et al., 2018; Nelson & International Food Policy Research Institute, 2010), which will be elaborated on in a subsequent paper. This will contribute to a grounded understanding of causes of and threats to undernutrition. We would like to encourage policy makers to specifically target stunting among children, to support the understanding of households and mothers on the importance of a nutritious and diverse diet bevor, during and after pregnancy, and to promote dietary diversity to further reduce and subsequently eliminate wasting.
Strengths and limitations
Our study findings had several strengths and weaknesses. Firstly, in order to control for the association of DDS and FVS with HAZ and WHZ, we controlled for various socio-demographic factors. By doing so, we were able to emphasize the link and importance of dietary diversity, which is not very well represented in the literature and unique for our study area. However, our measures of socio-demographic factors might be imperfect and not complete. Secondly, the period of retrospective dietary assessments are greatly discussed in the literature as longer recall periods are assumed to be prone to missing information, which was limited in this study by applying a 24 h DR plus a 7-day FFQ.
Thirdly, we did not examine the relationship between food quantities, dietary diversity and child growth. We did not attempt to measure food quantities for the following reasons: portion sizes within this age group cannot be standardized; the time for such detailed assessment is burdensome for the study participants; foods are regularly consumed from shared bowls; and the low level of formal education of mothers limits the precision of recalls . To better understand the relationship to food amounts, though, we conducted a desk review in Pubmed and google scholar (search terms in various combinations used: dietary diversity/ diversity, quantity/ amount/ food quantity/ calories, and HAZ/ stunting). Surprisingly, we did not find studies that examined this combination of indicators. Most studies focus on the use of nutrition indicators to measure the quality of the diet and the need to consider for energy and nutrient quantities, but do not report the amount of the foods consumed. Both, dietary diversity and dietary pattern scores, correlate positively with energy intake and thus, food amounts. Therefore, the link between portion sizes and dietary diversity as well as dietary patterns would be worth studying.
Fourthly, we were not able to control the association of dietary habits with stunting and wasting for different time-lags . Somé and Jones  looked at household dietary diversity by 4 seasons in Burkina Faso in 2014. They found out that dietary diversity was significantly higher during the beginning of the lean season, during the lean season and the highest during the harvest seasons compared to the post-harvest season. The post-harvest season covers the time from around December to April, and which represents a possible 6 months prior to the time when the data collection for NutriClim took place. As Somé and Jones did not use anthropometric data in their analyses, further research looking at stunting and its lagged association with dietary diversity would be highly recommended.