Patterns and determinants of breastfeeding and complementary feeding practices over the first year of life in a rural Gambian population

Introduction

The World Health Organization (WHO) recommends exclusive breastfeeding (EBF) until six months of age, followed by introduction of safe and nutritionally adequate foods in addition to continued breastfeeding until 24 months of age¹. This duration has been identified as a key intervention for reducing mortality during childhood. However, a large proportion of infants begin consuming non-breast milk foods (NBMFs) much earlier^2–6. A comprehensive characterization of early life diet is of particular importance in low-income populations that experience marked seasonality associated with annual food insecurity, heavy maternal workload, and fluctuations in infectious disease burden. These factors influence early life growth, the outcome of which is linked to infant/childhood – and later life – morbidity and mortality^7–9.

Globally, undernutrition is the underlying cause of more than 45% of deaths in children under 5 years of age¹⁰. This can occur directly through acute malnutrition, or through childhood diseases (such as malaria, diarrhea, pneumonia), which are more likely to become fatal in undernourished children¹¹. In The Gambia in West Africa (Figure 1), more than 25% of children under five years of age are affected by chronic undernutrition^12–14. The prevalence of undernutrition has increased in recent years, largely impacting rural areas of the country⁷. Chronic malnutrition is most prevalent among children 6–17 months of age in The Gambia; aligning with the period of transition from exclusive breastfeeding (EBF) to incorporation of liquids and foods other than maternal milk into the diet¹⁵.

Figure 1. Map of The Gambia (West Kiang region shaded in gray).

Figure 1 Footer: (MRC: Medical Research Council Keneba field station. Note: There are a total of 36 villages currently registered within the West Kiang Demographic Surveillance System. Only the villages included in the present analysis are depicted on the map.).

Suboptimal dietary intake can result in a myriad of adverse health consequences, including increasing susceptibility to infection. Mortality rates for children under five from diarrhea and pneumonia in The Gambia are high, and peak during ages when complementary foods are first introduced (~ five months)¹⁶. Infection rates and food security fluctuate across the year and are associated with the country’s marked seasonality, with a long dry “harvest” season (November to June), and a short wet “hungry” season (July to October)¹⁷. Moore (2016) reports that The Gambia’s strong seasonality creates a setting in which month of birth is a strong proxy of nutrition, infectious diseases, and mortality in early life. Additionally, long-term effects on immune programming in rural Gambian populations likely have a nutritional origin¹⁸.

Breastfeeding is nearly universal in The Gambia, and the only sustainable option for many. Prior research shows that more than 95% of infants in The Gambia receive maternal milk throughout the first twelve months of life and nearly half of mothers continue to breastfeed until their infant is 24 months of age^19–22. Existing research reports that cessation of EBF in The Gambia occurs between 3 to 6 months of age^2,22–26, generally earlier than the WHO recommended duration.

Earlier introduction of NBMF in populations in the West Kiang Region of The Gambia (Figure 1) may relate to fluctuations in maternal agricultural workload in this particular subsistence economy. Crop growth predominantly occurs during the wet season whereas crop harvesting takes place at the end of the wet season and into the start of the dry season. As such, the wet season is characterized by reduced food availability and increased maternal agricultural workload^27,28. It is also associated with increased morbidity and mortality^29–31. Food stores improve, physical labor related to subsistence practice is lessened, and morbidity prevalence is reduced during the dry season.

Women in the West Kiang region of The Gambia work for long periods (around 15 hours per day) in agricultural fields, conducting physically intensive tasks such as land preparation, planting, weeding, watering, harvesting, and transporting³². Rural Gambian women report that engaging in laborious work under harsh environmental conditions sometimes comes at the expense of the child’s well-being, including reduction of time allocated towards child health and nutritional needs³³. Physical activity level of mothers in this area has been reported to return to pre-pregnancy levels after the first month post-partum³⁴; thus, an increase in agricultural workload may relate to a decision to begin introducing complementary foods earlier. Infants are often separated from their mothers from morning to evening during periods of heavy workload (the amount of physical effort required to carry a child to the field has been cited as a reason to introduce NBMFs at earlier ages^35,36 and are commonly left in the care of other children or an elderly family member^33,37–44. When infants receive care from others, earlier cessation of EBF may be a necessary alteration to infant feeding^45,46.

Annual cropping patterns in The Gambia also determine food supply throughout the year, and thus the availability of certain ingredients for complementary foods. Traditional local complementary foods in rural Gambia include cereal dishes and other staples, such as rice (local name, ‘mani’), millet (‘sanyo’, ‘suno’), and maize (‘tubanyo’), which are among the staple crops grown in The Gambia. Cereal crops are generally harvested between September and December in rural Gambia (most observations cited were made in Keneba, a remote subsistence farming village in the West Kiang Region), resulting in particularly plentiful food supplies in November and December⁴⁷. Food shortages often occur in July and August. For example, the availability of groundnuts, the chief cash crop in the country and a primary ingredient of ‘tiakere churo’ (rice and groundnut porridge), decreases in August and September when the new crop has been planted but is not yet harvestable⁴⁷. When food is scarce during the wet season, ‘jidiyo,’ a watery/thin gruel made with powder or any pounded grain, may be preferentially used over a thicker, ‘mono,’ porridge for infants⁴⁷. Meat, fish, and animal milk are used less frequently as complementary foods^23,24,48. Nutritional composition of raw ingredients and dietary information for recipes of locally prepared dishes (e.g., porridges) have been reported in previous studies^49–53.

Infant feeding practices and access to nutritious foods have been associated with household socioeconomic position (SEP) across various populations, which can add complexity to infant feeding decisions. In The Gambia, children from rural communities had significantly higher odds of not meeting the requirement for minimum acceptable diet (a core indicator for assessing infant and young child feeding practices developed by the WHO; includes minimum dietary diversity and minimum meal frequency) compared with their urban counterparts^12,54,55. Prevalence of early life malnutrition and mortality are also higher in rural areas of the country compared to urban areas¹⁵. This may relate to a number of factors, such as general household poverty⁵⁶. Over-dependence on subsistence rain-fed agriculture and consumption (insufficient dietary diversity) combined with high poverty, low literacy levels, and high prevalence of morbidity were also identified as major predisposing factors to undernutrition in rural areas of the country¹⁵. Because malnutrition is a significant contributor to infant and child morbidity and mortality, it is critical to regularly assess early life diet.

In this study, we aim to comprehensively characterize breastfeeding patterns and the timing of introduction of specific local complementary foods in a rural Gambian population across infants’ first twelve months of life. We also explore potential environmental and sociodemographic predictors of EBF duration in order to identify factors that may influence infant feeding decisions in this population.

Methods

Subsample inclusion/exclusion criteria

A subsample (N=194) of mother-infant pairs from the larger HERO-G cohort (N=238) were included in the present analysis based on completeness of collected data over the first 12 months of life. Specifically, mother-infant pairs with no available infant feeding data (N=10), and those missing infant feeding data from three consecutive visits (the equivalent of one month) during the first 6 months of life (N=34) were excluded from this analysis (Figure 2). Instances of no available infant feeding data were attributable to unavailability of mother at the time of questionnaire distribution, either related to maternal travel or undocumented reasons. In many cases, missing infant feeding data from > three consecutive reports of infant feeding practices occurred if mothers were traveling or working and thus unavailable to provide responses to the questionnaire.

Figure 2. Flow diagram of included and excluded HERO-G participants in the HERO-G subsample.

To determine the statistical power of a sample size of 194 in examining associations between EBF duration and environmental, sociodemographic, and maternal and infant factors using multiple linear regression (F-test), a post-hoc power analysis was conducted using G*Power 3.1⁵⁹. With a medium effect size (f²=0.15) and a significance level of α = 0.05, a sample size of 194 has a power (1-ß err prob) of 0.99. Baseline characteristics of the HERO-G subsample are described in Table 1. There were no significant differences between baseline characteristics of the full HERO-G cohort and the HERO-G subsample.

Results

Infant feeding

The mean (±SD) age for introducing any food or liquid other than breast milk was 5 months (±1.5) with 59 (30.4%) infants EBF ≥6 months and 135 (69.6%) of infants EBF < 6 months. The mean duration of EBF for infants categorized as EBF <6 months was 4.4mo (±1.4) and those categorized as EBF ≥6 months was 6.5 months (±0.4). At 1 month of age, all infants were EBF. By 3 months of age, 5.6% of infants had been given water and 5.6% given semi-solids, which increased to 62.9% and 47.1% by 6 months of age, respectively. At 12 months of age, 98.7% of infants still received breast milk. Non-breast milk liquids given before 6 months of life included tea with milk (8.2%), powdered milk (3.1%), cow’s milk (1.5%), and tinned milk (1.0%). At 6 months of age, 0.01% of infants received solid foods compared to 66.9% at 12 months of age. Broad feeding practice categorizations (provision of maternal milk only, maternal milk plus non-breast milk liquids, and maternal milk plus non-breast milk semi-solids) are depicted according to age in months in Figure 3. Specific common NBMFs (including liquids and semi-solids) given over the first year of life are depicted by monthly reports in Figure 4.

Figure 3. Rural Gambian infant feeding practices by age.

Figure 4. Common NBMFs given over the first year of life.

Discussion

Here, one-third of rural Gambian infants were EBF until the WHO recommended 6 months of age, which corroborates results from both investigations of breastfeeding practices in another rural Gambian population from the West Kiang region over the past decade^21,44 and reports from decades prior^25,40. Similar findings have been documented in other populations across sub-Saharan Africa^66,67. Breastfeeding duration is relatively long in this rural Gambian population, with nearly all infants still receiving maternal milk at 12 months of age, the latest age at which the questionnaire was administered. This percentage aligns with other reports in the literature^20–22.

The most common NBMFs incorporated into the infant diet during the first 12 months of life in this analysis were grain-based porridges, which include some of the country’s staple crops as ingredients. In particular, gruels made with ‘suno’ (millet), ‘sanyo’ (millet), and ‘tiakere churo’ (groundnut porridge) were reported most commonly. It is well-established that local agriculture is central to the diets of communities in this area, including infant dietary intake^{21,40,68–70}. Around 70% of small-scale agricultural production in The Gambia is done by women^28,63,71,72, including subsistence work conducted throughout pregnancy and shortly after giving birth³⁴. Thus, the NBMF types commonly reported in the present study highlights the key connection between maternal agricultural workload and complementary feeding.

Assessment of potential predictors of EBF duration showed that seasonality is an important driver of the timing of introduction of NBMFs into the infant diet. This may relate to maternal agricultural workloads, which fluctuate with the annual rains and may influence infant caregiving and subsequently infant feeding practices. Birth month was the only significant predictor of EBF duration in our multiple linear regression model, with May, the month before the start of the wet (“hungry”) season, predicting earlier introduction of NBMFs. During the wet season in The Gambia, mothers commonly spend much of the day separated from their infants, resulting in less frequent breastfeeding and alternative caregivers responsible for infant feeding^43,73. The results from the present analysis support that earlier introduction of NBMF coincides with the annual period where maternal agricultural workload intensifies. Infants born in May were weaned, on average, by 3.54 months of age, suggesting that those born right before the start of the wet season are likely to cease EBF before the end of the wet season.

This finding may, however, relate to the uneven number of births across different months in the HERO-G cohort. The greatest number of births occurred in January, February, and March in the HERO-G subsample, with the greatest number of infants receiving their first NBMF in the month of June (which aligns with the median age of weaning in this study). This may be an artifact of the clustering of recruitment to women for this study at a particular time in the year. Thus, it is possible this finding is a byproduct of the proportion of births during the early months of the year as opposed to directly stemming from maternal subsistence activities and seasonality. Future works should consider investigating ‘month’ as a continuum rather than category, as using a coefficient of cyclic variation (as in Fourier analysis) could provide an even deeper understanding of the magnitude and temporal patterns at play⁷⁴.

Household SEP was not a significant predictor of breastfeeding practices in the multiple linear regression model. This differs from factors established in the literature from other populations, where SEP is associated with earlier EBF cessation due to influences such as limited access to resources such as transportation, access to roads, lower income to purchase nutritionally adequate and hygienic complementary foods, and lower education levels^75–78. The present finding challenges that of Issaka et al., (2017), where Gambian children from poor households had significantly higher odds of not meeting the recommendations for timing introduction and type of solids, semi-solids, or other soft complementary foods compared to children from wealthy households.

There were some limitations to this study. First, it important to note that calculation of household SEP was less robust here than in other studies due to the reduction of items from the PCA. Acknowledgement of possible bias must be given to the exclusion of livestock ownership in the PCA. Because cattle are valuable as sources of food, income, and transportation and traction (e.g., field plowing), ownership of a cow contributes a high score to the SEP calculation. The calculation used to assess household SEP, however, takes multiple variables into consideration at once, which is more appropriate here than using variables such as maternal education or livestock ownership as single predictor variables.

Employment and household income are variable and difficult to measure in settings such as rural Gambia. For example, broad occupation information may not accurately capture the individual income for self-employed farmers in rural Gambia if it cannot take into account factors such as crop type, seasonality, and transitory market conditions⁷⁹. Thus, asset scores are often calculated in order to measure relative wealth. Interviews or focus groups, to supplement questionnaires, might enhance the ability of future studies to provide additional context for the interpretation of the causes and consequences of infant feeding decisions in a highly seasonal environment. Revisiting details of household SEP, especially as it relates to maternal agricultural employment in the contemporary setting throughout pregnancy and over the course of lactation in the West Kiang region – a topic that has been rigorously investigated in earlier studies – would provide further insight into the influence of livelihood on dietary patterns.

Other characteristics of this population must be considered in the evaluation of the impact of SEP on EBF practices. In this region, there is low education attainment and prevalent household crowding. Formal education may be less relevant as a driver of infant feeding decisions in this region, as locally designed and run programs such as the Baby Friendly Community Initiative (BFCI) (developed by the National Nutrition Agency) in The Gambia provide local support and education to improve infant and young child feeding practices, including promoting EBF to 6 months of age⁸⁰. The villages included in the HERO-G study are among those implementing the BFCI in the West Kiang region. The WHO (2009) reports significant success from implementation of BFCI initiatives in The Gambia with marked improvement in early initiation of breastfeeding and EBF to 4 months of age. Specifically, the adoption of certain BFCI strategies contributed to the increase in the national average of EBF from 0% in 1989, to 17.4% in 1998, to 36% in 2000⁸¹; and 41% in 2006²⁰.

Next, collection of retrospective dietary intake data may introduce bias through recall error^82–84. However, studies comparing maternal recall methods found that 24- and 48-hour recall periods resulted in overestimated EBF durations, and report that 7 to 10 day recall intervals are adequate for general assessment of food intake, and may more accurately capture the complexity of infant feeding patterns compared to shorter intervals^84–86. Future analyses may benefit from analyses of dietary diversity scores, details of ingredients and preparation processes for homemade complementary foods, assessment of food volumes, focus groups on local feeding decisions/motivations, in particular the transfer of knowledge between generations/sources of information that influence decisions to introduce NBMFs. Future study designs may consider adopting methods to assess any specific relationships between infant complementary feeding practices and shifts in availability of certain food ingredients and/or quantified maternal workload across the seasons.

Studies incorporating detailed analyses of food preparation and microbial composition of local complementary foods in this region would also strengthen the present analysis. In older research (>40 years ago), reports indicate that it was common practice in The Gambia for infant foods to be prepared in large quantities in the morning, sufficient enough for several meals throughout the day and so it is available if another caregiver is responsible for feeding the infant while the mother is away^34,87. It is important to note that this may be out of date. This form of food preparation is convenient and efficient but has an associated downside of high levels of pathogenic bacterial contamination due to extended and poor storage conditions^87–93. After preparation, common local complementary foods are stored at ambient temperatures, which allows the child to be fed on demand⁸⁷. Additionally, these traditional complementary foods and some of their commonly used ingredients have been shown to contain high abundance of pathogens such as Staphylococcus aureus and Escherichia coli, which are associated with gastrointestinal infection^92–94. In addition, foods prepared in the wet season contained higher levels of potential pathogens compared to foods prepared during the dry season, presumably because the environmental conditions during the wet season are very hospitable for bacterial growth⁸⁷. There is also evidence that some gruels are prepared using contaminated water that contains potentially pathogenic coliform bacteria from both animal and human sources⁷⁰, though this too is out of date research.

Though not incorporated into the present analysis, infant health and growth attainment may also impact a mother’s decision to wean her infant earlier. Research on the directionality of these associations, however, provides mixed results. For example, mothers may cease EBF or modify feeding practices because their infants are ill, perhaps related to hospitalization, type of illness, or perception that maternal milk is not meeting the infant’s immunological needs⁹⁵. Others may wean infants perceived as healthy (e.g., fewer morbidities) earlier, which may result in increased incidence of morbidity as it relates to reduced immunological protection from maternal milk. Infants perceived as growing well may receive NBMFs earlier if they appear to demand more feeding; mothers may also EBF for longer durations if they regard breastfeeding as causal to healthy growth²¹. This should be explored in future studies, and the present study sets a strong foundation for such research.

Because malnutrition is a significant contributor to infant and child morbidity and mortality, it is critical to regularly assess early life diet. Although breastfeeding is a common practice in The Gambia, high rates of undernutrition in children under five warrant continued investigations of particular feeding patterns and their possible determinants. Environmental conditions in The Gambia temporally affect many aspects of diet, health, and behavior. Birth month has been identified as a strong proxy of nutritional status, infectious diseases, and mortality in early life in this environment, but to our knowledge has not previously been investigated as a predictor of exclusive breastfeeding duration. We found evidence to suggest that the age at which NBMFs are introduced to the infant diet has a temporal pattern, potentially related to annual shifts in maternal agricultural workload in this population. Further research is warranted to further explore the significance of the observed seasonally driven effects of dietary patterns on infant health and growth outcomes in this region. The present exploration of longitudinal infant feeding data provides context for future research on such topics.