Trend of risk and correlates of under-five child undernutrition in Bangladesh: an analysis based on Bangladesh Demographic and Health Survey data, 2007–2017/2018

Abstract

Objectives

The objectives of this study are to identify the trend of undernutrition risk among under-five children (U5C) in Bangladesh and the trend of its correlates.

Design

Multiple cross-sectional data sets from different time points were used.

Setting

Nationally representative Bangladesh Demographic and Health Surveys (BDHSs) were conducted in 2007, 2011, 2014 and 2017/2018.

Participants

In the BDHSs, the sample sizes for ever-married women (age: 15–49 years) were 5300 in 2007, 7647 in 2011, 6965 in 2014 and 7902 in 2017/2018.

Outcomes

Extant indicators of undernutrition (stunted, wasted and underweight) have been considered as the outcome variables.

Materials and methods

Descriptive statistics, bivariate analysis and factor loadings from factor analysis have been used to determine the prevalence of undernutrition over the years and find the trend of risk and its correlates.

Results

Risks of stunting among the U5C were 41.70%, 40.67%, 36.57% and 31.14%; that of wasting were 16.94%, 15.48%, 14.43% and 8.44%; and that of underweight were 39.79%, 35.80%, 32.45% and 22.46% in 2007, 2011, 2014 and 2017/2018, respectively. From the factor analysis, it has been found that the top five potential correlates of undernutrition are the wealth index, the education of the father and mother, the frequency of antenatal visits during pregnancy, the father’s occupation and/or the type of place of residence in the last four consecutive surveys.

Conclusion

This study helps us gain a better understanding of the impact of the top correlates on child undernutrition. To accelerate the reduction of child undernutrition more by 2030, Government and non-government organisations should focus on improving education and household income-generating activities among poor households and raising awareness among women about the importance of receiving antenatal care during pregnancy.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The main strength of the study is that nationally representative data have been used with large sample and enough information on child malnutrition for four consecutive years.

• Despite the above strengths, the study might have a recollection bias.

• The data do not have information on some important factors related to a kid’s development like the amount of diet given to children, the mobility pattern of children, etc.

• There is a lot of missing data for some existing variables in the data set, such as antenatal visits during pregnancy and duration of breast feeding; this study did not investigate the impact of missing data.

Introduction

Malnutrition has always been one of the most widespread causes of concerns pertaining to a person’s health which has negatively affected the health and welfare of countless people across the globe.^1–3 Millions of women and children worldwide, particularly in low and middle-income countries, continue to suffer from malnutrition, which includes undernutrition, overweight, obesity and deficiencies in some micronutrients.⁴ Malnutrition has been defined by the WHO as any insufficiencies, over or unevenness in a person’s intake of energy and/or nutrients.⁵ The undernutrition status of children is an imperious indicator of poverty in a country; and it is obvious that poverty, undernutrition and disease are intertwined.^{6 7} However, WHO has referred to malnutrition (undernutrition and overnutrition) as the sole supreme risk to the world’s public health, especially for the developing countries.⁵ Malnutrition is a primary cause of about half of the children’s deaths globally.⁸ Also malnutrition has been identified as a major risk factor for childhood mortality in some studies conducted in Bangladesh.^9–11 Besides, Bangladesh belongs to countries with highly prevalent child malnutrition (stunting more than 54% of preschool-age children, underweight in 56% and wasting in 17%)¹² and malnutrition-related mortality rate is about 53 per 1000 live birth according to Bangladesh Demographic and Health Survey (BDHS) 2011.¹³ Previous studies have shown that the prevalence of stunting in children has been steadily decreasing.^{5 14} Though the stunting rate in Bangladesh has significantly been reduced over the past two decades (from 51% in 2004 to 31% in 2017/2018),¹⁵ the global recommended rate of 3.9% for stunting is still not being met by the average annual rate of reduction.¹⁶ Consequently, the country is still behind the progress in achieving the Sustainable Development Goal (SDG) target 2.2 of reducing the stunting and wasting level below 20% and 7% by 2025, respectively. In addition, by 2030, the target is to end all forms of malnutrition, including achieving, by 2025, the internationally agreed targets on stunting and wasting among under-five children (U5C) which are 15.5% and 5%, respectively.¹⁷ However, if we look back to the global target in reduction of child malnutrition and policies to implement,^{16 18–20} Bangladesh failed to fulfil the target implementing all proposed policies. To enable a sharp decline of malnutrition rate in children it is essential to monitor evidence-based effectiveness of implemented policies.

A wide range of studies have been conducted on child malnutrition in Bangladesh and/or its associated risk factors.^21–24 Some literature has shown that factors such as low birth weight, family size, lack of parental education, breastfeeding status, previous birth interval, mother’s body mass index (BMI), low dietary intake, incidence of diarrhoea and household economic status are significantly related to infant mortality and child malnutrition.^21–28 Gender and birth order were significant correlates of malnutrition in previous studies.²³ Socioeconomic factors and diseases are also found to be significant correlates of child malnutrition.²⁵ Another study has revealed that malnutrition is higher in rural areas in comparison to urban areas.²⁶ Gender inequality owing to socioeconomic structure was found to be another important determinant of malnutrition in children in another study.²⁷ Still, another study suggested that antenatal care seeking behaviour of mother and related knowledge of healthcare are significant differential of child malnutrition, especially for infants.²⁸ In most of the previous studies^20–28 they have used binary, multiple and multilevel generalised linear regression analysis to identify the correlates associated with child malnutrition in Bangladesh. However, regression analysis encounters multiple limitations in determining trends when dealing with correlated covariates.^{29 30} If one independent covariate in a multiple regression equation exhibits a high correlation with one or more other independent covariates, it indicates the presence of multicollinearity.²⁹ Multicollinearity is a problem because it undermines the statistical significance of independent covariates.^{29 31} The aforementioned regression analysis is unable to address the issue of multicollinearity.^32–34 In such cases, factor analysis scores can be used along with correlation analysis to solve the issue of multicollinearity.^{35 36} Moreover, scores/loading derived from the covariance matrix of factor analysis can be employed to detect patterns by identifying the relationships between the observed variables and the underlying factors.^{37 38} Remarkably few research studies have used factor loading to examine the pattern of correlates in other disciplines, such as psychology,³⁹ social science⁴⁰ and dietary patterns in nutrition.^{41 42} To the best of our knowledge, there is no research study that has explored the trend of undernutrition (stunting, wasting and underweight) among U5C in Bangladesh, as well as the trend of its correlates using factor analysis.

Therefore, this study aimed to investigate the trend of risk in stunting, wasting and underweight among U5C in Bangladesh from 2007 to 2017/2018 and to identify the trend of some selected socio-demographic and economic correlates using factor loading from factor analysis. Outcome of this study will provide evidence to the health department and policymakers of Bangladesh about effectiveness of implemented policies in reducing child malnutrition towards strengthening their mode of action against it and modifying existing nutrition policy through more meaningful nutrition intervention programmes if needed.

Materials and methods

Data source and study design

The data have been used from the BDHS 2007, 2011, 2014 and 2017/2018 conducted under the authority of the National Institute for Population Research and Training of the Ministry of Health and Family Welfare.⁴³ Two-stage stratified sampling design was used to collect data in BDHS 2007, 2011, 2014 and 2017/2018. Data screening for Bangladeshi children under the age of five has been shown in figure 1.

Figure 1.

Data screening for under-five children in Bangladesh (BDHS-2007, 2011, 2014 and 2017). BDHS, Bangladesh Demographic and Health Survey.

Outcome variables

According to the 2006 WHO standard for children, the Z scores for height-for-age (HAZ), weight-for-height (WHZ) and weight-for-age (WAZ) were calculated as indicators of a child’s nutritional status. This indicator was calculated as follows: Z-score = (observed value − median value of the reference population)/SD value of the reference population.^{44 45} Stunting was defined as an HAZ < −2, wasting as a WHZ < −2 and underweight as a WAZ < −2. These three indicators have been used as outcome variables in this study for analysis.

Explanatory variables

First, we collected possible explanatory variables of malnutrition from several literature reviews.21–28 30–63 Among them, a few explanatory variables were absent from the last four consecutive survey data sets. In addition, some of them were accessible but have numerous missing values (see online supplemental appendix table S1). Finally, we selected 14 common explanatory variables which were available in all four BDHS data sets. Description and coding plan of the selected study variables are given in table 1.

Table 1.

Description and coding plan for the selected study variables

Variables	Description	Analysis coding
Outcome variable
Stunting	Stunting is defined as low height-for-age (HAZ ˂ −2).	0=not-stunted 1=stunted
Wasting	Wasting is defined as low weight-for-height (WHZ < −2).	0=not-wasted 1=wasted
Underweight	Underweight is defined as low weight-for-age (WAZ < −2).	0=not-underweight 1=underweight
Explanatory variables
Age of child (months)	Age in months of the child is calculated by subtracting the century day code (CDC) for the date of birth from the CDC for the date of interview and dividing the result by 30.4375.	1 = <12 months 2=12–23 months 3=24–35 months 4=36–47 months 5=48–59 months
Sex of child	Sex of child.	1=male 2=female
Preceding birth interval (months)	The preceding birth interval is the difference between birth date of child and birth date of preceding child in months.	1=first birth 2=1–23 3=24–48 4=48+
Age of mother at first birth (years)	Age of the respondent at first birth is calculated from the century month code (CMC) of the date of first birth and the CMC of the date of birth of the respondent.	1 = <20 2=20–30 3=30+
Mother’s body mass index (BMI) (kg/m2)	Body mass index is defined as respondent weight in kilograms divided by the square of her height in metres.	1=underweight (<18.5 kg/m2) 2=normal (18.5–24.9 kg/m2) 3=overweight (25.0–29.9 kg/m2) 4=obese (≥30 kg/m2)
Frequency of antenatal visits during pregnancy	Number of antenatal visits during pregnancy.	0=no visit 1=antenatal visits during pregnancy
Breastfeeding status	Duration of breast feeding.	1=ever breast fed (0–93) 2=never breast fed (94) 3=still breast feeding (95)
Mother’s educational level	Mother’s level of education.	0=no education 1=incomplete primary 2=complete primary 3=incomplete secondary 4=complete secondary 5=higher education
Mother’s employment status	Whether the respondent is currently working.	0=not working 1=working
Father’s educational level	Father’s level of education.	0=no education 1=incomplete primary 2=complete primary 3=incomplete secondary 4=complete secondary 5=higher education
Father’s occupation	Father’s present occupation.	1=did not work 2=worker 3=skilled worker 4=agricultural 5=small business 6=services big business
Division	Administrative division.	1=Barisal 2=Chittagong 3=Dhaka 4=Khulna 5=Mymensing 6=Rajshahi 7=Rangpur 8=Sylhet
Type of place of residence	Type of place of residence.	1=urban 2=rural
Wealth index	The wealth index is a total assessment of a household’s standard of life*.	1=poorest 2=poorer 3=middle 4=richer 5=richest

*The list of variables which were commonly used to calculate wealth index: type of flooring, refrigerator, water supply type of vehicle, sanitation facilities, persons per sleeping room, electricity, ownership of agricultural land, radio, domestic servant, television, country-specific items, telephone.

Statistical analysis

We summarised the outcome variables of undernutrition (stunting, wasting and underweight) and explanatory variables by frequency distribution (n and %). The bivariate analysis χ² test was used to examine the association of each of the outcome variables with the explanatory variables. Associations were considered significant at the 5% level. To extract the trend of correlates of the undernutritional status of U5C in Bangladesh, factor analysis has been used. A factor can be considered a theoretical measure representing a group of actual measures that correlate highly with one another and lowly with the others in a set. The loadings are the polychoric correlations between the actual measures and the theoretical measures or factors. For ordinal variables, polychoric correlation assesses similarity between various raters.⁴⁶ A factor is identified through an examination of the variables that load highly (ie, correlate highly) on it. Principal Component Analysis has been used for extraction of factors and orthogonal rotation (varimax option) to derive non-correlated factors.^{37 47 48} One factor was created with the data from undernourished children to rank the absolute loadings properly. In the factor analysis, any child with missing covariate data was completely dropped from the analysis. A total of 886 (17%), 1083 (14%), 2947 (42%) and 3296 (42%) observations were entirely dropped due to missing covariates in BDHS 2007, 2011, 2014 and 2017/2018, respectively. The high number of missing observations in 2014 and 2017/2018 was due to 41% missing data specifically related to the covariates of antenatal visits during pregnancy and the duration of breast feeding. The factor analysis included the remaining complete cases: 4414 (83%), 6564 (86%), 4018 (58%) and 4606 (58%). Statistical analysis was performed using the Stata/SE V.15.0 (Stata, USA) and R V.4.2.2.

Factor analysis

Factor analysis, a multivariate analysis technique, is frequently used to reduce the dimensions of data by locating the underlying latent factors that account for the covariance among the observed variables. It can be used for pattern detection by identifying the relationships between the observed variables and the underlying factors. By reducing the number of variables to a smaller set of underlying factors, factor analysis can simplify complex data structures, and help interpreting the relationships among the variables. Factor analysis is widely used in fields such as psychology, marketing research, social sciences and many others to uncover the underlying structure of data and to gain insights into complex relationships among variables.³⁷

Factor loading

The observable random vector X, with p components, has mean µ and covariance matrix ∑. The factor model postulates X is linearly dependent on a few unobservable random variables F_1, F₂,…, F_m, called common factors, and p additional sources of variation ℇ₁,ℇ_2,… ℇ_p called errors. In particular, the factor analysis model is

X₁-µ₁=l₁₁F₁+ l_12 F2+…+ l_1mF_m+ ℇ₁

X₂-µ₂=l₂₁F₁+ l_22 F2+…+ l_2mF_m+ ℇ_2.

… … …

X_p-µ_p=l_p1F₁+ lp_p2 F2+…+ l_pmF_m+ ℇ_p

Or, in matrix notation, X-µ=L F+ ℇ

(p×1) (p×m) (m×1) (p×1)

The coefficient l_ij is called the loading of the ith variable on the jth factor, so the matrix L is the matrix of factor loading.⁴⁸

Patient and public involvement

None.

Results

As shown in table 2, in this study, we included 5300, 7647, 6965 and 7902 U5C records from BDHS 2007, 2011, 2014 and 2017/2018, respectively, after excluding all unusable records (figure 1). From figure 2, it has been shown that the prevalence of child nutrition has improved steadily over the past decades. The prevalence of stunting among U5C has been declined from 41.7% (95% CI: 40.4% to 43.0%) in 2007 to 40.7% (95% CI: 39.5% to 41.7%) in 2011 to 36.6% (95% CI: 35.4% to 37.7%) in 2014 and then to 31.1% (95% CI: 30.10% to 32.2%) in 2017/2018. The prevalence of wasting has been declined from 16.9% (95% CI: 15.9% to 17.9%) in 2007 to 15.5% (95% CI: 14.6% to 16.2%) in 2011 to 14.4% (95% CI: 13.6% to 15.2%) in 2014 and then to 8.4% (95% CI: 7.8% to 9.1%) in 2017/2018 and underweight from 39.8% (95% CI: 38.5% to 41.1%) in 2007 to 35.8% (95% CI: 34.7% to 36.8%) in 2011 to 32.4% (95% CI: 31.3% to 35.5%) in 2014 and then to 22.5% (95% CI: 21.5% to 23.4%) in 2017/2018.

Table 2.

Characteristics of the study subjects, BDHS: 2007, 2011, 2014, 2017/2018

Characteristic	BDHS:2007	BDHS:2011	BDHS:2014	BDHS:2017/2018
	Total (N=5300) n (%)	Total (N=7647) n (%)	Total (N=6965) n (%)	Total (N=7902) n (%)
Age in months
<12	1049 (19.8)	1484 (19.4)	1344 (19.3)	1729 (21.9)
12–23	1095 (20.7)	1442 (18.9)	1456 (20.9)	1624 (20.5)
24–35	1073 (20.3)	1441 (18.8)	1406 (20.2)	1526 (19.3)
36–47	1048 (19.7)	1679 (22.0)	13.76 (19.8)	1463 (18.5)
48–59	1035 (19.5)	1600 (20.9)	1383 (19.8)	1560 (19.8)
Sex of child
Male	2672 (50.4)	3907 (51.1)	3571 (51.3)	4125 (52.2)
Female	2628 (49.6)	3740 (48.9)	3394 (48.7)	3777 (47.8)
Preceding birth interval (months)
First birth	1720 (32.5)	2703 (35.3)	2714 (39.0)	3041 (38.6)
<24	525 (9.9)	587 (7.7)	476 (6.8)	527 (6.8)
24–47	1525 (28.7)	1879 (24.6)	1388 (19.9)	1488 (18.6)
48+	1530 (28.9)	2478 (32.4)	2387 (34.3)	2846 (36.0)
Age of mother at first birth (years)
<20	3947 (74.5)	5691 (74.4)	5085 (73.0)	5537 (70.1)
20–30	1325 (25.0)	1911 (25.0)	1838 (26.4)	2316 (29.3)
30+	28 (0.5)	45 (0.6)	42 (0.6)	49 (0.6)
Mother’s BMI
Underweight	1669 (31.6)	2116 (27.7)	1596 (22.5)	1167 (14.8)
Normal	3132 (59.3)	4529 (59.2)	4067 (58.5)	4688 (59.5)
Overweight	416 (7.8)	823 (10.8)	1097 (15.8)	1636 (20.8)
Obese	64 (1.3)	131 (1.3)	220 (3.2)	388 (4.9)
Frequency of antenatal visits during pregnancy
No antenatal visits	1648 (37.4)	2169 (33.1)	848 (21.0)	389 (8.1)
Antenatal visits	2766 (62.6)	4390 (66.9)	3196 (79.0)	4303 (91.9)
Duration of breast feeding
Ever breast fed	2234 (42.3)	2347 (35.5)	535 (13.2)	665 (14.1)
Never breast fed	33 (0.6)	127 (2.0)	46 (1.2)	31 (0.7)
Still breast feeding	3023 (57.1)	4154 (62.5)	3481 (85.6)	4004 (85.2)
Mother’s educational level
No education	1420 (26.9)	1449 (18.9)	1076 (15.5)	566 (7.2)
Primary incomplete	1136 (21.5)	1356 (17.7)	1105 (15.8)	1423 (18.0)
Primary complete	506 (9.6)	974 (12.8)	829 (11.8)	870 (11.0)
Secondary incomplete	1441 (27.3)	2834 (37.0)	2762 (39.7)	3333 (42.2)
Secondary complete	359 (6.8)	426 (5.6)	457 (6.6)	398 (5.0)
Higher	418 (7.9)	608 (8.0)	736 (10.6)	1312 (16.6)
Mother’s employment status
Yes	1263 (23.8)	6918 (90.5)	1747 (25.1)	4701 (59.5)
No	4076 (76.2)	729 (9.5)	5217 (74.9)	3201 (40.5)
Father’s educational level
No education	1787 (33.7)	2141 (28.0)	1736 (24.9)	1284 (16.3)
Primary incomplete	967 (18.3)	1243 (16.3)	1168 (16.8)	1553 (19.7)
Primary complete	507 (9.6)	981 (12.8)	932 (13.4)	1117 (14.2)
Secondary incomplete	1033 (19.5)	1755 (22.9)	1653 (23.7)	2093 (26.6)
Secondary complete	332 (6.3)	474 (6.2)	465 (6.7)	409 (5.2)
Higher	668 (12.6)	1047 (13.7)	1009 (14.5)	1425 (18.0)
Father’s occupation
Did not work	80 (1.5)	131 (1.8)	44 (0.6)	199 (2.5)
Worker	1772 (33.5)	2501 (32.7)	2023 (29.1)	2395 (30.3)
Skilled worker	963 (18.2)	1465 (19.2)	1535 (23.6)	2153 (27.3)
Agricultural	950 (18.0)	1255 (16.4)	1046 (15.1)	952 (12.1)
Small business	889 (16.8)	1258 (16.5)	1398 (20.1)	1496 (18.9)
Services-big bus’s.	635 (12.0)	1030 (13.5)	796 (11.5)	707 (8.9)
Division
Barisal	696 (13.1)	837 (10.9)	812 (11.5)	1170 (14.8)
Chittagong	1093 (20.6)	1516 (19.8)	1320 (19.0)	830 (10.5)
Dhaka	1222 (21.2)	1272 (16.7)	1213 (17.4)	1284 (16.3)
Khulna	630 (11.9)	894 (11.7)	774 (11.1)	1125 (14.2)
Maymensing	-	-	-	833 (10.5)
Rajshahi	856 (16.2)	916 (12.0)	875 (12.7)	941 (11.9)
Rangpur	-	993 (13.0)	865 (12.4)	817 (10.3)
Sylhet	903 (17.0)	1219 (15.9)	1106 (15.9)	902 (11.5)
Type of place of residence
Urban	1850 (34.9)	2342 (30.6)	2188 (31.4)	2694 (34.1)
Rural	3450 (65.1)	5305 (69.4)	4777 (68.6)	5208 (65.9)
Wealth index
Poorest	1050 (19.8)	1682 (22.0)	1515 (21.7)	1767 (22.4)
Poorer	1089 (20.6)	1489 (19.5)	1307 (18.8)	1597 (20.2)
Middle	996 (18.8)	1456 (19.0)	1379 (19.8)	1430 (18.1)
Richer	991 (18.7)	1493 (19.5)	1420 (20.4)	1565 (19.8)
Richest	1174 (22.1)	1527 (20.0)	1344 (19.3)	1543 (19.5)

Mymensingh division was a part of the Dhaka division in 2007, 2011 and 2014 and Rangpur was part of the Rajshahi division in 2007.

BDHS, Bangladesh Demographic and Health Survey; BMI, body mass index.

Figure 2.

Trend of risk of stunting, wasting and underweight among under-five children in Bangladesh (2007–2017).

Among the U5C, 50.4%, 51.1%, 51.3% and 52.2% were men and 49.6%, 48.9%, 48.7% and 47.8% were women in 2007, 2011, 2014 and 2017/2018, respectively. About 37.4%, 33.1%, 21.0% and 8.1% mothers had no antenatal visits during pregnancy and 26.9%, 18.9%, 15.5% and 7.2% mothers had no formal education in 2007, 2011, 2014 and 2017/2018, respectively. Of the respondents, 65.1% in 2007, 69.4% in 2011, 68.6% in 2014 and 65.9% in 2017/2018 came from rural areas, and 19.8% in 2007, 22.0% in 2011, 21.7% in 2014 and 22.4% in 2017/2018 had the lowest wealth index.

From online supplemental appendix table S2, we found that age in months of the children, preceding birth interval (months), age of mother at first birth (years), mother’s BMI, frequency of antenatal visits during pregnancy, duration of breast feeding, mother’s educational level, mother’s employment status, father’s educational level, father’s occupation, administrative division, type of place of residence and wealth index are found to be significantly associated with undernutrition among U5C in Bangladesh. It has also been observed that among the stunted children in 2007, 9.6% were of age <12 months, 19.2% were of age 12–23 months, 25.2% were of 24–35 months, 24.9% were of age 36–47 months and 21.1% were of age 48–59 months. Among the stunted children in 2011 those figures were 9.9%, 22.6%, 21.6%, 25.0% and 20.9%, respectively. Furthermore, among the stunted children in 2014 those percentages were 9.0%, 22.2%, 23.4%, 24.7% and 20.8%, respectively, and in 2017/2018 those figures were 14.1%, 23.5%, 23.9%, 20.5% and 18.0%, respectively. Similar pattern has been found among wasted and underweight children. It has also been found that preceding birth interval, age of mothers at first birth, mothers BMI, frequency of antenatal visits during pregnancy, mothers educational status, father’s educational status, administrative division, type of place of residence and wealth indices were found to be significantly associated with all kinds of undernutrition in the consecutive surveys.

Factor loadings derived from single-factor factor analysis have been reported in tables 3–5. It has been found that top-most-five correlates of stunting in 2007 were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and place of residence; in 2011 the correlates were household wealth index, mother’s educational level, father’s educational level, frequency of antenatal visits during pregnancy and place of residence; in 2014, they were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and place of residence; and in 2017/2018, they were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and father’s occupation (table 3).

Table 3.

Factor loadings (pattern matrix) and rank of absolute loading of each variable among under-five stunted children in Bangladesh (BDHS: 2017/2018, 2014, 2011 and 2007)

Socio-demographic and economic variables	BDHS:2017/2018		BDHS:2014		BDHS:2011		BDHS:2007
	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading
Wealth index	0.77	1	0.78	1	0.80	1	0.77	1
Father’s education level	0.73	2	0.78	2	0.76	3	0.75	2
Mother’s education level	0.73	3	0.73	3	0.77	2	0.75	3
Frequency of antenatal visits	0.56	4	0.64	4	0.62	4	0.63	4
Father’s occupation	0.45	5	0.43	6	0.47	6	0.37	7
Place of residence	−0.42	6	−0.45	5	−0.52	5	−0.45	5
Mother’s BMI	0.41	7	0.37	7	0.43	7	0.42	6
Age of mother at first birth (years)	−0.40	8	0.37	8	0.39	8	0.33	8
Mother’s employment status	−0.34	9	−0.17	11	0.04	12	−0.17	10
Preceding birth interval (months)	−0.24	10	−0.29	9	−0.25	9	−0.25	9
Age in months	−0.16	11	−0.25	10	−0.15	10	−0.08	11
Sex of child	−0.08	12	0.01	14	−0.02	13	−0.01	13
Division	−0.06	13	−0.15	12	−0.15	11	−0.04	12
Duration of breast feeding	−0.02	14	0.09	13	−0.01	14	0.00	14

BDHS, Bangladesh Demographic and Health Survey; BMI, body mass index.

Table 4.

Factor loadings (pattern matrix) and rank of absolute loading of each variable among under-five wasted children in Bangladesh (BDHS: 2017/2018, 2014, 2011 and 2007)

Socio-demographic and economic variables	BDHS:2017/2018		BDHS:2014		BDHS:2011		BDHS:2007
	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading
Wealth index	0.79	1	0.78	1	0.82	1	0.76	2
Father’s education level	0.78	2	0.78	2	0.77	3	0.77	1
Mother’s education level	0.77	3	0.73	3	0.79	2	0.76	3
Frequency of antenatal visits	0.58	4	0.64	4	0.61	4	0.65	4
Father’s occupation	0.49	5	0.43	6	0.56	5	0.55	5
Mother’s employment status	−0.40	6	−0.17	11	0.04	14	−0.12	11
Age of mother at first birth (years)	0.38	7	0.37	7	−0.53	7	0.41	8
Place of residence	−0.38	8	−0.45	5	−0.53	6	−0.48	6
Mother’s BMI	0.29	9	0.37	8	0.50	8	0.46	7
Preceding birth interval (months)	−0.29	10	−0.29	9	−0.27	9	−0.24	9
Age in months	−0.24	11	−0.25	10	−0.16	10	−0.06	12
Duration of breast feeding	−0.07	12	0.09	13	−0.04	13	−0.16	10
Sex of child	−0.04	13	0.01	14	−0.05	12	−0.01	13
Division	0.01	14	−0.15	12	−0.03	14	−0.01	14

BDHS, Bangladesh Demographic and Health Survey; BMI, body mass index.

Table 5.

Factor loadings (pattern matrix) and rank of absolute loading of each variable among under-five underweight children in Bangladesh (BDHS:2017/2018, 2014, 2011 and 2007)

Socio-demographic and economic variables	BDHS:2017/2018		BDHS:2014		BDHS:2011		BDHS:2007
	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading	Factor loading	Rank of absolute loading
Wealth index	0.79	1	0.78	2	0.81	1	0.75	1
Father’s education level	0.74	2	0.79	1	0.78	2	0.74	2
Mother’s education level	0.74	3	0.75	3	0.76	3	0.73	3
Frequency of antenatal visits	0.62	4	0.64	4	0.62	4	0.62	4
Father’s occupation	0.44	5	0.46	6	0.50	6	0.43	5
Mother’s BMI	0.44	6	0.38	8	0.47	7	0.42	6
Age of mother at first birth (years)	0.43	7	0.43	7	0.43	8	0.32	8
Place of residence	−0.38	8	−0.47	5	−0.55	5	−0.42	7
Mother’s employment status	−0.30	9	−0.14	10	0.02	14	−0.19	10
Preceding birth interval (months)	−0.23	10	−0.31	9	−0.24	9	−0.32	9
Duration of breast feeding	−0.20	11	−0.051	13	−0.09	10	−0.05	11
Division	−0.06	12	−0.14	11	−0.08	11	−0.04	12
Sex of child	−0.06	13	0.04	14	−0.03	13	−0.01	14
Age in months	0.01	14	0.11	12	−0.06	12	−0.03	13

BDHS, Bangladesh Demographic and Health Survey; BMI, body mass index.

The top-most-five correlates of wasting in 2007 were father’s educational level, household wealth index, mother’s educational level, frequency of antenatal visits during pregnancy and father’s occupation; in 2011 the correlates were household wealth index, mother’s educational level, father’s educational level, frequency of antenatal visits during pregnancy and father’s occupation; in 2014 they were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and place of residence; and in 2017/2018, they were wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and father’s occupation (table 4).

The top-most-five correlates of underweight in 2007 were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and father’s occupation; in 2011 the correlates were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and type of place of residence; in 2014 they were father’s educational level, household wealth index, mother’s educational level, frequency of antenatal visits during pregnancy and type of place of residence; in 2017/2018, they were household wealth index, father’s educational level, mother’s educational level, frequency of antenatal visits during pregnancy and father’s occupation (table 5).

Discussion

In the last few decades, a lot of policies and interventional programmes have been implemented by the government of Bangladesh and non-governmental organisations to improve the nutritional status of U5C. But, it is essential to work more with child malnutrition to fulfil the target of SDG which directs reduction of malnutrition among young children by integrating child malnutrition and related strategies. However, from this study it has been found that the risk of stunting among U5C has declined from 41.7% in 2007 to 31.1% in 2017/2018. It declined by 10.6 percentage points over the past 10 years. The level of underweight has declined from 39.8% in 2007 to 22.5% in 2017/2018. Particularly in the last 10 years, underweight has declined by 17.3 percentage points. It is still at a critically high level of about 15%, according to WHO.⁴⁵ The prevalence of wasting came down from 16.9% in 2007 to 8.4% in 2017/2018, that is, it declined by 8.5 percentage points (decreased 50%) in the last 10 years.

The findings of the bivariate and factor analysis revealed that the wealth index, the educational levels of the father and mother, the frequency of antenatal visits during pregnancy and the father’s occupation and/or the type of place of residence have remained among the top five leading correlates that were highly associated with undernutrition over the years. The wealth index was found to be the most significant among the top five correlates of undernutrition. The poorest households have actively contributed to U5C malnutrition, which is a natural result of their failure to ensure adequate food security, maintain proper hygiene, provide treatment facilities, etc. A similar discussion was also reported in some previous studies conducted in Bangladesh^{49 50} and other countries.^51–55 In Bangladesh, 18.7% of the population remains below the national poverty line, with 10% experiencing extreme poverty.⁵⁶ The government of Bangladesh has implemented various poverty reduction programmes over the years to reduce wealth index inequalities, such as the Social Safety Net Programmes (SSNPs),⁵⁷ World Food Programme⁵⁸ Food and Agricultural Organisation,⁵⁹ etc. Some of the prominent SSNPs implemented by the Government of Bangladesh include Vulnerable Group Feeding, Open Market Sales, Cash for Work, Food for Work, Vulnerable Group Development, Gratuitous Relief and the 100 days employment guarantee scheme. Though, SSNPs in Bangladesh have played a role in poverty reduction, their coverage is so limited compared with the extent of extreme poverty in the country, particularly in certain regions like the haor areas in the northeast, Noakhali’s chars, river erosion-prone areas and parts of the north-central region.^{60 61} Although the programmes mainly target rural poverty, some indicators suggest that urban poverty is more severe.⁶¹ As a result, the government may fail to address wealth index inequalities related to child malnutrition, which remains a critical issue over the years. Poverty leads to increased illiteracy and malnutrition among children, as it hinders access to education and proper healthcare.⁶² Contrarily, the children from rich families have greater access to health facilities, better environmental conditions including drinkable water, sanitation and access to enough food may account for the decreased risks of child stunting associated with higher wealth index. Also, it is possible that parents from the wealthiest families have a higher level of education and are more accountable for their children’s health compared with parents who come from the poorest households.⁶³ The key findings of the current study imply that raising household wealth status may significantly reduce the chance of undernutrition in children.

Parental education has another significant impact on the level of undernutrition among U5C and our results are consistent with those of earlier studies.^64–66 Education is a valuable type of human capital that enhances the productivity, health and nutrition of individuals within a nation while also mitigating population growth.⁶⁷ In Bangladesh, promoting inclusive education is challenging due to geographical factors like indigenous children, tea garden children, coastal area children, border area children and children in monga and flood-affected areas. Over the course of time, these children will become parents. These groups are facing numerous barriers to receiving quality education over the years.⁶⁸ Sultana et al²⁴ have discussed that educated parents are more aware about health, nutrition, proper child care, health services, hygiene, proper food for children, etc. Educated parents generally refer to those who have received formal education or have a certain level of academic or intellectual knowledge. They also contribute positively to promoting household income, which can support providing the above-mentioned resources to their children. Other authors have discussed that educated mothers are better able to cope with inadequate family income and provide healthcare facilities efficiently. They are also more likely to limit family size, promote health-stimulating activities and afford quality healthcare for their children.49 50 69 70 However, current study shows that a significant percentage of children of higher educated parents were found to be malnourished over the years. On the other hand, prevalence of malnourished U5C among employed mothers was remarkably higher in 2017/2018 than previous years, which was consistent with the findings of previous studies conducted in Bangladesh⁷¹ and in other countries.⁷² Theoretically, a working mother has two major effects on the home that we can classify into the income effect and the substitution effect. The income effect can make it possible for kids to buy healthy food and essential medications. On the other hand, maternal care time for children is reduced due to the substitution effect. If the income effect is greater than the substitution effect, the children’s nutritional status would be better; it would be worse for the child’s nutritional status if the substitution effect is higher than the income effect. In addition, nowadays, most of the families are nuclear and the employed parents are obligatory bound to leave their children at home under illiterate maid. This is important considering the fact that as much as 77% of companies in Bangladesh do not offer childcare facilities to their female employees at workplaces.⁷³ Therefore, the government and policymakers may rethink about the social infrastructure of Bangladesh to intervene and establish a day-care system for employed mothers.

Bangladesh has made great strides in improving the health of its population, but maternal health and healthcare facilities still face vulnerabilities.⁷⁴ Children whose mothers had received antenatal care more times during their pregnancies had lower rates of undernutrition. According to earlier studies, antenatal care was associated with a reduction in child stunting.^{75 76} Both mother and child can avoid numerous health issues like infections, anaemia, iron deficiency and more by receiving antenatal care during pregnancy.⁶⁵ Type of place of residence showed a similar pattern for all the three kinds of child undernutrition over the years. The place of residence for stunting was in the fifth position in 2007 which went back to sixth position in 2017/2018. In the past decade a lot of infrastructure developmental activities, like the National Strategy for Accelerated Poverty Reduction II,⁷⁷ the Asian Development Bank (ADB) country strategy and programme, 2006–2010 for Bangladesh,⁷⁸ the ADB country strategy and programme, 2011–2015 for Bangladesh,⁷⁹ etc, were implemented in rural areas regional divisions in Bangladesh which opened scope for better access to services and institutions, including education, healthcare, livestock, microfinance and cooperatives. Besides, microfinance agencies, including non-governmental organisations (NGOs) expanded their services to the poor in the root level areas of regional divisions significantly in the past decades. All these helped to minimise the regional inequalities in stunting among U5C in Bangladesh.

Maternal BMI is an important correlate of both maternal and child nutrition but it did not have a place in the top five correlates. According to our findings, the trend of prevalence of low BMI (<18.5) among women of childbearing age has declined over the years but it is still found to be highly associated with the three forms of child’s undernutrition. This outcome is consistent with earlier research done in Bangladesh^{24 80–82} and also in low-income and middle-income countries.^{83 84} For a mother to be able to breast feed and recover from the physical and possibly mental stress of being pregnant and giving birth as well as deal with raising and caring for children, she needs to be in good nutritional condition.

Conclusion

This study explored how socioeconomic status and demographic correlates were significantly related to child undernutrition as well as examined the trend of the top five leading correlates that changed from 2007 to 2017/2018. The potential top five correlates are associated with undernutrition among U5C in Bangladesh, namely the wealth index, the educational level of the father and mother, the frequency of antenatal visits during pregnancy, the father’s occupation and/or the type of place of residence. For investment and policy guidance, it is essential to understand the levels and trends of the major correlates of undernutrition among U5C. It is also important to monitor the trends of leading correlates of undernutrition over time to see where interventions are working and where more attention is needed. Therefore, the findings of this study strongly highlight the necessity of improving other factors which are directly or indirectly related to these top leading five-factors including social infrastructure to achieve better nutritional status among U5C in Bangladesh.

Recommendations

It is very important to integrate child malnutrition and related strategies in order to meet the SDG aim of reducing malnutrition among young children. To accelerate the reduction of undernutrition, the authors think that any comprehensive interventional programme taken by the government and public–private owner organisations at the community level should be focused on improving education and household income-generating activities among poor households and raising women’s awareness about receiving antenatal care facilities during pregnancy. In addition to the existing policy for improving the nutritional status of U5C, the government, the NGOs or policymakers should take necessary steps for controlling these related factors by new intervention strategies.

Data availability statement

No data are available. Data are available in a public, open access repository. Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The National Institute of Population Research and Training (NIPORT) of the Ministry of Health and Family Welfare (MOHFW) were in charge of conducting the Bangladesh Demographic and Health Survey (BDHS). Responsible authority followed a standard ethics to conduct the survey and make it open for researchers.