Children born during the hunger season are at a higher risk of severe acute malnutrition: Findings from a Guinea Sahelian ecological zone in Northern Ghana

Abstract Heightened food insecurity in the hunger season increases the risk of severe acute malnutrition (SAM) in childhood. This study examined the association of season of birth with SAM in a Guinean Sahelian ecological zone. We analyzed routine health and sociodemographic surveillance data from the Navrongo Health and Socio‐demographic Surveillance System collected between 2011 and 2018. January–June, the period of highest food insecurity, was defined as the hunger season. We defined moderate acute malnutrition as child mid‐upper arm circumference (MUAC) between 115 mm and 135 mm and SAM as MAUC ≤ 115 mm. We used adjusted logistic regression to quantify the association between the season of birth and SAM in children aged 6–35 months. From the 29,452 children studied, 24% had moderate acute malnutrition. Overall, 1.4% had SAM, with a higher prevalence (1.8%) in the hunger season of birth. Compared with those born October–December, adjusted odds ratios (aOR) and 95% confidence interval (95% CI) for SAM were increased for children born in the hunger season: January–March (1.77 [1.31–2.39]) and April–June (1.92 [1.44–2.56]). Low birth weight, age at an assessment of nutritional status, and ethno‐linguistic group were also significantly associated with SAM in adjusted analyses. Our study established that being born in the hunger season is associated with a higher risk of severe acute malnutrition. The result implies improvement in the food supply to pregnant and lactating mothers through sustainable agriculture or food system change targeting the hunger season may reduce the burden of severe acute malnutrition.


| INTRODUCTION
After decades of undernutrition, Africa is now seeing a surge in overnutrition (Bain et al., 2013;Davies et al., 2018;UNICEF World Health Organization & World Bank, 2020). Despite the surge in overnutrition and the double burden, undernutrition remains an important public health problem in many African countries and requires a multidisciplinary approach to address it (Doku & Neupane, 2015).
Like many countries in sub-Saharan Africa (SSA), Ghana has a high burden of malnutrition among children aged 0-59 months with stunting, underweight and wasting all being highly prevalent (19%, 11%, and 5%, respectively in 2014) (Black et al., 2008 In sub-Saharan African countries most communities rely on rainenabled agriculture due to minimal access to irrigation facilities (Abizari et al., 2017;Chikhungu & Madise, 2014;Fentahun et al., 2018;Hagos et al., 2014). This creates a seasonal pattern of access to food, where the "hunger season" exposes families to food insecurity and its profound effect on nutritional status, particularly in children under age 5 years (Belayneh et al., 2020;Fentahun et al., 2018). The hunger season, a period between planting and harvest when food runs out is usually the worst hit period. This comes with profound adverse health effects on mothers and children.
The hunger season is a challenging period for subsistence farmers and their families, who solely rely on what they grow.
The established underlying causes of malnutrition across all age groups likely follow a seasonal pattern, with consequences for morbidity and mortality across the life cycle (Bryson et al., 2021). Hence establishing the seasonal patterns in food availability and its effect on nutritional health or morbidity have policy implications in planning the distribution of scarce resources. While Demographic and Health Surveys do not measure the impact of seasonal variations in food supply and nutritional status, several epidemiological studies (though not at the country level) have provided insights (Sorgho et al., 2016).
Most of these studies were not, however, conducted during the first 1000daysPlus (preconception through pregnancy, delivery and postdelivery and infancy up to 5 years), which is considered by the World Health Organization as a critical period for optimizing growth. Further evidence accruing from the developmental origins of health and diseases concept further highlights the importance of the 1000day-sPlus period as a highly vulnerable period where adverse exposures such as malnutrition can have lasting effects on health in later life (Davies et al., 2018).
Previous studies from SSA have reported inconsistent findings regarding any link between seasonal fluctuations in access to food and the nutritional status of young children (Chikhungu & Madise, 2014). Some studies from Gambia (Tomkins et al., 1986) and Malawi (Chikhungu & Madise, 2014) have reported seasonal trends in child growth and demonstrated weight and height deficits during the rainy season (period for planting with minimal food availability). A study from northern Ghana had also observed a seasonal effect of low dietary diversity and hence low micronutrient intake during the hunger season for both mothers and children (Abizari et al., 2017). In contrast, a study in Kenya observed increased nutrient intake in the rainy season. The preharvest season is often seen as a vulnerable period for acute malnutrition among exposed children. The effect is especially felt in drought-prone areas which rely mostly on rainenabled agriculture or subsistence farming.
These current analyses of children aged 6-35 months from the Navrongo Health and Demographic Surveillance System (HDSS) (Oduro et al., 2012) was conducted as part of a broader INPreP -Improved nutrition preconception, during pregnancy and postdelivery study (https://www.southampton.ac.uk/global-health/ research/lifecourse-epidemiology/inprep/about-us.page). We set out to determine the seasonal variation in the burden of severe acute malnutrition. We further determined the association of season of birth with SAM focusing especially on the effect of being born during the hunger season and the risk of SAM. We also reported on the unique factors in the Guinean Sahelian ecological zone and established factors that influence the nutritional status of children (Bain et al., 2013;Madan et al., 2018).

| METHODS
We used data from the Navrongo Health and Demographic Surveillance system (HDSS) in Ghana (Oduro et al., 2012)

Key Points
• Children born in the hunger season are at a higher risk of severe acute malnutrition in this Guinea-Sahelian ecological zone.
• Children born with low birth weight were at a higher risk of severe acute malnutrition Being female and of the Nankam ethno-linguistic group increased ones odds of severe acute malnutrition.
• Our observations imply sustainable food supply to pregnant and lactating mothers through improved agriculture or food system change targeting the hunger season may reduce the burden of severe acute malnutrition.
weight were extracted from the child health records of the children or the maternal health record book. For this study, we extracted and utilized HDSS data from 2011 to 2018.

| Study setting
The Navrongo Health and Demographic Surveillance System (HDSS) (Oduro et al., 2012) operational area lie in an ecological belt that is food insecure due to climatic conditions. The area is highly dependent on rain-enabled seasonal subsistence agriculture for livelihoods.
We conducted this study in the Kasena-Nankana West district and  (Oduro et al., 2012).
Through key informants in the community, new houses are identified for which field workers visit and then determine and register the number of households as well as resident individuals. These are then subsequently visited every 6 months to update the records and collect new information when needed.

| Data management and statistical analyses
Routine data collected by the field workers are doubly entered and verified by a data manager for inconsistent entries using Visual Birth weight was categorized as low (birth weight <2.5 kg) or normal (birth weight ≥2.5 kg) and according to 11th edition of the World Health organizations' (WHO) international classification of diseases (WHO, 2019). Season of birth was grouped into four categories to reflect the weather patterns and food availability within a year, that is, January-March (postharvest season), April-June (immediate pre-planting), July-September (planting season), and October-December (immediate harvest season). The hunger season was thus defined as the period between January and June, corresponding to the period of severely reduced availability of food.
We computed household wealth index using principal component analysis (PCA) from 30 separate household items, from large assets (e.g., land and car ownership) to smaller household items (e.g., phone, fan ownership). The principal component analysis of the household assets or items was done to predict factor scores. These scores were then categorized into wealth quintiles: Q1 = poorest, Q2 = poorer, Q3 = poor, Q4 = less poor, and Q5 = least poor.
We used logistic regression to examine the association between seasonality and severe acute malnutrition. Unadjusted and adjusted odds ratios (aOR) with 95% confidence intervals (95% CIs) were computed. We included all potential confounding variables in our data in our multivariable logistic regression model. These variables include season of birth, sex of child, birth weight, maternal education, wealth quintile, age of child, maternal age, mother's marital status, place of residence, and ethnicity. We did not find any statistically significant interaction between variables. The test of statistical significance was set at 5% (p < 0.05).  In Note: p-value calculated using Pearson's χ 2 test to show differences between the groups of the variables in the row data.

| RESULTS
Abbreviations: JSS, junior high school; SAM, severe acute malnutrition. Previous studies have shown that the mother's diet in pregnancy and lactation may program her offspring where the fetus adapts to undernutrition by changing its metabolism, altering its production of hormones and sensitivity of tissues to them, redistributing its blood flow, and slowing its growth rate (Morrison & Regnault, 2016).
Children born in the harvest period of October-December had the lowest SAM rate, perhaps due to the good supply of food and hence higher nutritional status of the lactating mother.
Our results showed a strong association between birth weight and SAM among children in the population. The odds ratios of SAM in low birth weight children was nearly twice the odds of severe acute malnutrition in normal birth weight children.

This observation is consistent with similar studies conducted in
Tanzania (Rogawski McQuade et al., 2019), Malawi (Chikhungu & Madise, 2014), Ethiopia (Fentahun et al., 2018), India (Choudhary et al., 2019), and Bangladesh (Hillbruner & Egan, 2008). One of the major reasons for the association of low birth weight with SAM is suboptimal maternal nutrient intake. In the hunger season, there is greater likelihood that milk production among lactating mothers is reduced (Roba et al., 2016). Children born in the lean season also have a greater susceptibility to weight loss in the first two weeks of life and this may also contribute to the higher risk of SAM. This may be explained by low quantity and/or quality of breast milk during this time. Some studies observed that these disparities were persistent and associated with smaller but sus- In the study area, the hunger season is also associated with harsh climatic conditions including the seasonal north-east harmattan winds blowing from the Sahara and extremely high temperatures, which may affect dry season gardening due to lack of irrigated agriculture. These may act together to increase respiratory infections and further worsen nutrient intake in the said period. The Ghana Nutrition profile shows that younger children are at a higher risk of SAM but data on age-categorized sub-analysis was not provided (https://www.usaid.gov/sites/ default/files/documents/1864/Ghana-Nutrition-Profile). Largely consistent with our findings, a study from northern Ghana had previously established that malnutrition was higher among children aged 6-8 and 12-23 months (Saaka et al., 2015). The authors observed that the lower age groups (<12 months) were less likely to recover from severe acute malnutrition (Akparibo et al., 2017).
We observed that lower household socioeconomic status was associated with higher risk of severe acute malnutrition while relatively wealthier households having lower risk of SAM. Similar findings have been previously reported in studies from western Nigeria (Imam et al., 2020), Pakistan (Ahmad et al., 2020), and Jamaica (Thompson et al., 2017). A study from Nepal also observed that families with higher socioeconomic status were likely to present with lower SAM (Hossain et al., 2020). These findings were further corroborated by a study in Ghana that observed that lower family income was associated with high levels of SAM (Tette et al., 2016). A plausible reason is that higher economic status is associated with a higher purchasing F I G U R E 2 Yearly proportions of severe acute malnutrition cases in children aged 6-35 months in the Navrongo Health and Demographic Surveillance Site from 2011 to 2018 T A B L E 2 Crude and adjusted odds ratios of factors associated with severe acute malnutrition in children aged 6-35 moths in the Navrongo Health and Demographic Surveillance System from 2011 to 2018 power and hence access to food during the lean season.
This may thus maintain a good or healthy dietary diversity across the seasons.
We further observed that maternal education was not associated with SAM. This has previously being observed in a study from Bangladesh, where in addition to maternal education, variations in household income did not influence levels of SAM (Rahman et al., 2016). We observed female children were likely to be severely malnourished during the hunger season. Consistent with our findings was findings from a study from Nepal that previously reported that male children were less likely to have SAM compared with female children (Hossain et al., 2020). This was further observed by a study in Tanzania (Rogawski McQuade et al., 2019) but contrary to findings from 16 demographic health surveys reports from SSA (Wamani et al., 2007). A further systematic review also reported boys are more vulnerable to malnutrition and stunting compared with girls (Thurstans et al., 2020). These inconsistent findings may suggest the varying reasons currently provided for these differences are more speculative rather than informed by direct evidence. Further longitudinal studies may be required to explain the observed findings.
Overall our findings have important implications for policies and program implementers in the study area. These results suggest that children, particularly girls, born in the hunger season are vulnerable to SAM and hence poor health outcomes. The implications may extend into adulthood as severe malnutrition and poor growth have been associated with reduced cognitive function and risk of chronic diseases later in life (Liu et al., 2003).  (Mehta et al., 2013). This implies the scores perform differently in different populations. However, MUAC measurements have been validated and found to be viable, low-cost, low-burden alternative for community-level nutritional status assessment.

| CONCLUSION
We established that being born in the hunger season is associated with a higher risk of SAM in this Guinea-Sahelian ecological zone. The result implies improvement in the food supply to pregnant and lactating mothers through sustainable agriculture or food system change targeting the hunger may reduce the burden of SAM. Further work is also needed to understand the mechanisms underlying these observed associations, especially the role of the pregnancy/lactation period so as to guide the development of appropriate interventions.

ACKNOWLEDGMENTS
We are grateful to the chiefs, elders and people from communities within the catchment of the Navrongo sociodemographic

CONFLICT OF INTERESTS
The authors declare that there are no conflict of interests.

ETHICS STATEMENT
Ethical approval for the operations of the Navrongo HDSS was ob-