Unhealthy feeding practices in early childhood: prevalence and associated factors in Bangladesh

Abstract

Background

Feeding practices during early childhood are critical to health and development. In low- and middle-income countries, many young children are exposed to sweetened beverages and processed foods, which may contribute to poor dietary quality. This study assessed the prevalence and correlates of unhealthy feeding practices (UFPs) among Bangladeshi children aged 6–23 months.

Methods

Data were derived from the 2022 Bangladesh Demographic and Health Survey. The analysis included 2608 children aged 6–23 months living with their mothers. Three infant and young child feeding indicators were considered: consumption of sweetened beverages, unhealthy foods and zero fruit or vegetable intake.

Children having none of these three UFPs were considered as ‘no UFP’; those who have at least one UFP are classified as ‘1 UFP’ and subsequently having two is ‘2 UFP’ and three is ‘3 UFP’.

Results

Among the 2608 children, 1239 (47.51%) reported having at least one UFP. In adjusted ordered logistic regression, older child age was associated with higher odds of having a higher UPF score. Maternal mass media exposure was also associated with increased UFP score (adjusted OR (aOR) 1.31; 95% CI 1.11 to 1.55). High women’s empowerment was associated with lower odds of being in a higher UFP category (aOR 0.78; 95% CI 0.64 to 0.95) compared with low empowerment.

Conclusion

UFPs were common among Bangladeshi children aged 6–23 months. Higher child age and maternal media exposure were associated with higher UFP scores, while women’s empowerment was associated with lower odds of higher UFP scores. These findings support integrating age-appropriate complementary feeding counselling, supporting women’s decision-making and promoting marketing of healthy foods.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• Provides a national estimate of unhealthy feeding practices (UFPs) among Bangladeshi children aged 6–23 months using Bangladesh Demographic and Health Survey 2022 data.

• Distinguishes children with unhealthy behaviours of varying extent (0 UFP, 1 UFP, 2 UFP, 3 UFP), a subgroup not previously examined nationally.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Highlights the need for longitudinal and intervention studies to understand behavioural and structural drivers of poor feeding.

• Supports integrating age-specific nutrition counselling into maternal and child health services and informs policy actions to regulate food marketing and promote healthier complementary feeding environments.

Introduction

Early childhood nutrition is a cornerstone of lifelong health and development.¹ The first 1000 days of life represent a critical period during which inadequate nutrition can lead to developmental delays, micronutrient deficiencies and increased vulnerability to infections.² The period between 6 and 23 months is particularly important, as it marks the transition from exclusive breastfeeding to complementary feeding, which is a window of both opportunity and vulnerability.³ Feeding practices during this stage are central to ensuring optimal growth and cognitive development. The infant and young child feeding (IYCF) indicators, developed by the WHO and UNICEF, provide standard measures for assessing feeding practices during this period.⁴ In low and middle-income countries (LMICs), particularly in urban areas, there is an increasing trend of unhealthy feeding practices (UFPs).⁵ In response to the rapid nutrition transition, WHO and UNICEF recently updated the IYCF indicators to better capture unhealthy dietary behaviours in young children. Three indicators are particularly relevant to dietary quality in this age group: sweetened beverages consumption, unhealthy or sentinel foods consumption (UFC) and zero fruit or vegetable (ZvF) intake.⁶ These reflect growing concern over the dietary quality of complementary foods, which contributes to both undernutrition and overnutrition.⁷

UFPs have been linked to poor dietary adequacy, childhood malnutrition, obesity and the early onset of non-communicable diseases (NCDs).⁸ Evidence from the Child Health and Mortality Surveillance surveillance (2016–2023) across seven high-burden countries, including Bangladesh, found that malnutrition accounted for 39.5% of under-five deaths, 89% of which were associated with infections such as pneumonia, sepsis or diarrhoea.⁹ The long-term effects of poor nutrition extend beyond physical growth, impairing cognitive ability, educational achievement and productivity in adulthood, with measurable economic implications.10,12

The WHO complementary feeding guidelines discourage the provision of sugary drinks, which add calories but provide little nutritional benefit. The European Society for Paediatric Gastroenterology, Hepatology and Nutrition recommends limiting free sugar intake to <5% of total caloric intake in children under 2 years.¹³ Adequate consumption of fruits and vegetables is equally important for micronutrient adequacy and healthy growth. However, recent data show that UFP, that is, frequent intake of energy-dense and nutrient-poor foods is increasing in many LMICs.¹⁴ A multicounty analysis across sub-Saharan Africa reported that 62.4% of children aged 6–23 months exhibited at least one unhealthy feeding behaviour.¹⁵

Despite notable socioeconomic gains in Bangladesh, progress in complementary feeding indicators has remained stagnant.¹⁶ A study in rural Bangladesh found that 12.2% of children aged 6–23 months consumed unhealthy foods, while 63.2% had no fruit or vegetable intake on the day preceding the survey.¹⁷ A study using data from the 2019 Multiple Indicator Cluster Survey reported that 44.1% of Bangladeshi children belonging to the same age group had ZvF consumption.¹⁸ Using one UFP indicator alone may not accurately represent the burden of such practices. Hence, there is a need to explore all three indicators of unhealthy feeding behaviour from a public health and policy perspective. The current study used the updated WHO-UNICEF IYCF measurement indicators (derived from 24-hour recall) to explore the extent of UFP in Bangladeshi children aged 6–23 months and its associated factors.

Methods

Data overview

The study used secondary data extracted from the Bangladesh Demographic and Health Survey (BDHS) 2022. The BDHS is a nationally representative survey that provides data on a wide range of health and demographic indicators. BDHS data are naturally hierarchical with data collected using two-stage stratified random sampling.⁶ The dataset and results of this survey have been published on the DHS programme website (https://www.dhsprogram.com/). The survey employed a stratified, two-stage cluster sampling method. First, clusters or enumeration areas (EAs), small, well-defined geographic units from the most recent national census, were randomly selected to cover the entire country. Each EA contains a manageable number of households. In the second stage, systematic sampling was used to select households within each EA. Details of the methodology of the BDHS 2022 survey can be found elsewhere.¹⁹ The publicly available dataset contains information about 8784 under-5 children; however, after excluding children aged <6 months, children aged >23 months and children who do not live with their mothers, 2608 participants were included in the study (figure 1).

Figure 1. Selection of the study participants.

Operational definitions of outcome variables

This study applied the 2022 WHO-UNICEF IYCF framework, which introduced three indicators to assess UFPs among children aged 6–23 months.⁶

Sweet beverage consumption: percentage of children 6–23 months of age who consumed a sweet beverage during the previous day.

UFC: percentage of children 6–23 months of age who consumed selected sentinel unhealthy foods during the previous day.

ZvF: percentage of children 6–23 months of age who did not consume any vegetables or fruits during the previous day.

Using these three indicators, we calculated a composite index, that is, “UFP score” which is used as the outcome variable in this study and is defined as:

• 0 (No) unhealthy feeding practice: children with none of the three unhealthy predefined feeding practices.

• 1 Unhealthy feeding practice: children with the presence of one of the three predefined UFPs.

• 2 Unhealthy feeding practice: children with the presence of any two of the three predefined UFPs.

• 3 Unhealthy feeding practice: children with the presence of all three predefined UFPs.

Independent variables

The independent variables were selected based on prior literature review, theoretical knowledge and availability in the secondary dataset.⁹ These include child’s age (in complete months), mother’s age (in complete years), father’s age (in complete years), number of family members, number of under five children, child’s sex (male/female), residence (urban/rural), maternal education (no education/primary/secondary/post-secondary), father’s education (no education/primary/secondary/post-secondary), current marital status (married/widowed/divorced/separated), father earns (no/yes), mother earns (no/yes), maternal media exposure (no/yes), wealth index, women empowerment, religion, birth order (first or second/third or above), antenatal visit (optimal (≥8), not optimal), caesarean section (no/yes), delivery by skilled birth assistant (SBA) (no/yes), early initiation of breast feeding (no/yes), skin-to-skin contact (no/yes).

Maternal media exposure was derived from three variables, following previously established methodology.^{15 20} These variables include “frequency of reading newspapers or magazines”, “frequency of listening to the radio” and “frequency of watching television”. Each variable has three response categories, participants who responded, “not at all” to all three variables were categorised as having no media exposure or “no”, otherwise “yes”.

The wealth index is an asset-based measure that estimates relative household economic status based on ownership of assets, services and amenities, without requiring direct assessment of income or expenditure.²¹ Based on this index, participants were categorised into five wealth quintiles: poorest, poor, middle, rich and richest.

Skilled birth attendants (SBAs) were defined as professionally trained health workers capable of managing normal labour and delivery and performing essential obstetric interventions, including doctors, nurses, midwives, community SBAs and paramedical or sub-assistant medical officers.²²

The Survey-based Women’s Empowerment Index, developed in 2017 using DHS data from 34 African countries, is the first individual-level measure enabling cross-country and temporal comparisons of women’s empowerment. It captures three domains: social independence (education, access to information, age at key life events and spousal disparities), decision-making (participation in household decisions as a measure of instrumental agency) and attitudes toward violence (reflecting intrinsic agency and acceptance of gender norms related to violence). Details on index computation are available elsewhere.²³ In the current study, the variable “women empowerment” had three categories: low, medium and high.

Antenatal care (ANC) was defined as care received from a skilled provider at least eight times during the most recent pregnancy. Women with ≥8 visits were classified as having “optimal ANC”, while those with fewer were categorised as “not optimal”.²⁴

Data analysis

All statistical analyses were conducted using STATA V.19 (Stata, College Station, TX, USA). Descriptive statistics were used to summarise the characteristics of the study population. Continuous variables were expressed as medians with IQRs, and categorical variables were presented as frequencies and percentages. An UFP score was created, where 0 UFP indicates having none of the three UFPs; those who have at least one UFP are classified as ‘1 UFP’ and subsequently having two is ‘2 UFP’ and three is ‘3 UFP’.

An ordered logistic regression model was applied. Unadjusted and multivariable-adjusted ORs (aORs) with 95% CIs were reported. Statistical inference was based on effect estimates and their precision.

We assessed multicollinearity among covariates using variance inflation factors (VIFs). Marital status showed evidence of multicollinearity (high VIF) and was therefore removed from the final multiple models. A VIF of >10 was considered as indicative of multicollinearity. All tests were two-sided, and a p value of <0.05 was considered significant (α=0.05).

Public and patient involvement

Patients or members of the public were not involved in the design, conduct, reporting or dissemination of this research.

Results

Among the 2608 children aged 6–23 months who were considered for this study, 403 (15.45%) reported no UFP, 1239 (47.51%) reported 1 UFP, 837 (32.09%) 2 UFP and 129 (4.95%) reported having 3 UFP (figure 2).

Figure 2. Percentage of different unhealthy feeding practices scores among the children aged 6–23 months. UFP: Unhygienic feeding practices.

Online supplemental table 1 summarises the participant’s characteristics across different UFP categories (no UFP, 1 UFP, 2 UFP, 3 UFP). The median (IQR) age of children was 14 (9–19) months, while mothers and fathers were aged 25 (21–30) and 32 (28–38) years, respectively. Boys and girls were almost equally represented (50.6% vs 49.4%), and 32.3% of participants resided in urban areas. Boys and girls were almost equally represented. More than half the mothers completed secondary-level education (52.5%). Almost all mothers were currently married (99.04%). More than half the mothers reported mass media exposure (56.3%) and high empowerment (54.8%). Optimal ANC was uncommon (5%; n=2550), and 69.7% reported delivery by a SBA (n=2600).

Online supplemental table 1- Sociodemographic and maternal characteristics of participants by the presence of different UFP scores.

Table 1 presents the ordered logistic regression model assessing factors associated with different UFP scores. Child age was associated with higher odds (aOR 1.05; 95% CI 1.04 to 1.07) of having a higher UFP category. Maternal mass media exposure was also associated with increased UFP score (aOR 1.31; 95% CI 1.11 to 1.55); while high-level women’s empowerment was associated with lower odds of having a higher UFP score (aOR 0.78; 95% CI 0.64 to 0.95) compared with low empowerment. Current marital status was not considered for the final ordered logistic regression model due to multicollinearity (table 1).

Table 1. Ordered logistic regression of UFP score (0–3): unadjusted and adjusted ORs.

Variable	Unadjusted OR (uOR, 95% CI)	P value	Adjusted OR (aOR, 95% CI)	P value
Child’s age (in complete months)	1.05 (1.03 to 1.06)	<0.001	1.05 (1.04 to 1.07)	<0.001
Mother’s age (in complete years)	0.98 (0.97 to 1.00)	0.014	1.00 (0.98 to 1.03)	0.752
Father’s age (in complete years) (n=2582)	0.99 (0.98 to 1.00)	0.005	0.99 (0.97 to 1.00)	0.083
Number of family members	0.98 (0.95 to 1.01)	0.206	1.00 (0.97 to 1.03)	0.970
Number of under five children	0.90 (0.81 to 0.99)	0.028	0.93 (0.84 to 1.04)	0.231
Child’s sex (Ref: Male)
Female	1.09 (0.95 to 1.26)	0.220	1.13 (0.97 to 1.31)	0.107
Residence (Ref: Urban)
Rural	0.92 (0.79 to 1.08)	0.300	0.88 (0.73 to 1.05)	0.144
Maternal education (Ref: No education)
Primary	1.03 (0.73 to 1.45)	0.879	0.89 (0.61 to 1.29)	0.532
Secondary	1.10 (0.80 to 1.53)	0.556	0.90 (0.62 to 1.31)	0.571
Post-secondary	0.88 (0.62 to 1.25)	0.483	0.81 (0.52 to 1.24)	0.333
Father’s education (Ref: No education)
Primary	1.30 (1.04 to 1.62)	0.020	1.21 (0.95 to 1.55)	0.119
Secondary	1.26 (1.01 to 1.57)	0.036	1.21 (0.93 to 1.56)	0.150
Post-secondary	0.92 (0.73 to 1.17)	0.517	0.96 (0.70 to 1.31)	0.784
Marital status (Ref: Married)
Widow	–		–
Divorced	–		–
Separated	–		–
Father earns (Ref: No)
Yes	1.20 (0.72 to 1.99)	0.481	1.14 (0.68 to 1.91)	0.611
Mother earns (Ref: No)
Yes	1.12 (0.95 to 1.33)	0.171	1.15 (0.96 to 1.38)	0.118
Maternal media exposure (Ref: No)
Yes	1.22 (1.05 to 1.41)	0.008	1.31 (1.11 to 1.55)	0.001
Wealth index (Ref: Poorest)
Poor	1.05 (0.84 to 1.31)	0.665	0.97 (0.76 to 1.23)	0.784
Middle	1.18 (0.94 to 1.48)	0.145	1.12 (0.87 to 1.44)	0.375
Rich	1.18 (0.94 to 1.47)	0.157	1.17 (0.89 to 1.52)	0.257
Richest	0.94 (0.75 to 1.18)	0.584	0.95 (0.69 to 1.29)	0.735
Women empowerment (Ref: Low)
Medium	1.16 (0.94 to 1.44)	0.177	1.16 (0.92 to 1.45)	0.202
High	0.76 (0.63 to 0.95)	0.004	0.78 (0.64 to 0.95)	0.012
Religion (Ref: Islam -no)
Yes	1.18 (0.90 to 1.54)	0.233	1.12 (0.85 to 1.49)	0.423
Birth order (Ref: first or second)
third or above	0.86 (0.73 to 1.01)	0.068	0.92 (0.73 to 1.14)	0.431
Antenatal visit (Ref: Not optimal)
Optimal	0.83 (0.59 to 1.16)	0.272	0.99 (0.69 to 1.41)	0.994
Caesarean section (Ref: No)
Yes	0.93 (0.80 to 1.07)	0.325	0.92 (0.76 to 1.11)	0.389
Delivery by SBA (Ref: No)
Yes	0.88 (0.75 to 1.03)	0.102	0.86 (0.70 to 1.07)	0.171
Early initiation of breast feeding (Ref: No)
Yes	0.98 (0.84 to 1.15)	0.842	0.95 (0.79 to 1.13)	0.537
Skin-to-skin contact (Ref: No)
Yes	0.94 (0.78 to 1.13)	0.492	0.97 (0.80 to 1.18)	0.756

ANC, Antenatal care; aOR, Adjusted OR; SBA, skilled birth assistant; UFP, Unhealthy feeding practice; uOR, Unadjusted OR.

Discussion

This study explored the prevalence and determinants of UFP among Bangladeshi children aged 6–23 months using data from the 2022 round of the BDHS. The analysis applied the recently updated WHO-UNICEF IYCF framework, which introduced three new indicators to assess dietary quality during complementary feeding. These indicators capture key aspects of poor dietary behaviour that contribute to both undernutrition and the emerging risk of diet-related NCDs in early life. We operationalised UFP as an ordinal score (0–3) based on the presence of the number of UFP indicators, using a 24-hour recall method.

The findings revealed that UFPs were highly prevalent among children in Bangladesh. Most of the children have at least one (1 UFP) or two (2 UFP) UFP indicators. The study identified several significant predictors that may help to explain these patterns. Older child age was associated with increased odds of having a higher UFP score. Similar age-related trends have been documented in studies from Nepal, Indonesia and other LMICs.25,27 As children grow older, they transition from parental control to greater autonomy in their food choices and become increasingly exposed to the external food environment, including processed snacks, sugary beverages and fast foods marketed toward families and young children. The availability and affordability of such foods in both urban and rural markets may also facilitate this shift.²⁸

Maternal media exposure emerged as an unexpected predictor, showing a positive association with higher UFP scores. This finding aligns with evidence from Indonesia, where increased access to media has been linked to higher consumption of processed foods.²⁹ While media exposure can enhance maternal knowledge of health and nutrition, the widespread marketing of sugary drinks, snack foods and fast-food products through television and digital platforms may outweigh the benefits of informational content. In Bangladesh, exposure to aggressive marketing strategies targeting young children and mothers is increasingly common, and evidence suggests that food advertisements often promote nutritionally poor products.³⁰ Therefore, maternal media exposure may serve as a proxy for exposure to persuasive advertising rather than for health education. This underscores the need for stricter regulation of food marketing directed toward children and families.

Women’s higher level of empowerment was associated with a lower UFP score compared with low empowerment. This finding aligns with a study conducted in sub-Saharan Africa.³¹ The women empowerment-related factors, such as increased decision-making autonomy and greater influence over household choices, may support healthier feeding behaviours during complementary feeding.

This study’s major strength is the use of a large, nationally representative dataset, allowing generalisation of findings to Bangladeshi children aged 6–23 months. The standardised BDHS methodology ensures data reliability and comparability across countries. By applying updated WHO-UNICEF IYCF indicators and introducing the composite scoring of UFP, the analysis captures both traditional and emerging dietary risks in early childhood. However, the cross-sectional design limits causal inference, and maternal recall of 24-hour dietary intake may have introduced bias. The BDHS dataset also lacks information on food availability, caregiver attitudes and exposure to food marketing. The use of binary indicators may oversimplify dietary complexity. Additionally, the survey employs a 24-hour recall method that fails to capture quantity or volume and is often subject to day-to-day variability. Additionally, the BDHS IYCF module operationalises these exposures as binary indicators and does not provide reliable portion-size measures for the relevant items in our analytic dataset. Future longitudinal studies should examine how parental, social and environmental factors shape feeding behaviours over time.

Conclusion

UFP remains common among Bangladeshi children aged 6–23 months. This study assessed its burden by using an ordinal UFP score. Older child age was associated with higher UFP score, as was maternal mass media exposure. However, a high level of women’s empowerment was associated with lower score. Our findings highlight the need for age-appropriate complementary feeding counselling in routine maternal and child health service. Given the positive association between maternal media exposure and higher UFP score, policies to curb marketing of unhealthy foods and beverages to young children and families may prove to be beneficial.

Data availability statement

Data are available in a public, open access repository.