Determinants of maternal low mid‐upper arm circumference and its association with child nutritional status among poor and very poor households in rural Bangladesh

Abstract Malnutrition among women is a long‐standing public health concern that has significant adverse consequences on the survival and healthy development of children. Maternal mid‐upper arm circumference (MUAC) could potentially represent a simpler alternative to traditional nutritional indicators. This study aimed to investigate the factors associated with low maternal MUAC (as an indicator of being underweight) and address the research question of whether maternal MUAC is significantly associated with children's nutritional status among poor and very poor households in rural Bangladesh. Data on 5,069 households were extracted from the Suchana programme baseline survey, which was carried out in 80 randomly selected unions (the lowest administrative unit of Bangladesh) in Sylhet and Moulvibazar districts between November 2016 and February 2017. The outcome variables were three child nutritional status indicators: wasting, stunting and underweight. Mothers were classified as underweight if their MUAC was less than 23 cm. Separate multiple logistic regression analyses were used to determine the factors potentially associated with maternal underweight status and explore whether maternal underweight status is significantly associated with children's nutritional status. The prevalence of maternal underweight status was 46.7%, and the prevalence of wasting, stunting and underweight among children under two were 10.5%, 44.4% and 31.9%, respectively. After controlling for various socio‐economic and demographic characteristics, maternal MUAC was significantly associated with children's nutritional status in rural Bangladesh.

were classified as underweight if their MUAC was less than 23 cm. Separate multiple logistic regression analyses were used to determine the factors potentially associated with maternal underweight status and explore whether maternal underweight status is significantly associated with children's nutritional status. The prevalence of maternal underweight status was 46.7%, and the prevalence of wasting, stunting and underweight among children under two were 10.5%, 44.4% and 31.9%, respectively.
After controlling for various socio-economic and demographic characteristics, maternal MUAC was significantly associated with children's nutritional status in rural Bangladesh.

K E Y W O R D S
anthropometry, Bangladesh, baseline survey, child nutrition, food insecurity, maternal nutrition, mid-upper arm circumference, public health, Suchana, underweight 1 | INTRODUCTION Women of child-bearing age are often nutritionally susceptible (Lartey, 2008). Although maternal and child nutrition have improved much globally, the prevalence of maternal and child malnutrition still remains very high in low-resource settings. Despite the successful implementation of many nutritional policies at the national and international level, maternal malnutrition still represents a major public health concern that has adverse consequences on the survival and healthy advancement of women, especially in low-and middle-income countries (Black et al., 2013;National Institute of Population Research and Training -NIPORT/Bangladesh, Mitra and Associates, & ICF International, 2016). Previous studies have shown maternal malnutrition is commonly associated with childhood malnutrition and contributes to low birth weight, suggesting new avenues of research and novel interventions are required to reduce maternal and child malnutrition (Ajieroh, 2009;Bhutta et al., 2008;Choudhury et al., 2017;Islam et al., 1994;Lartey, 2008).
The situation of maternal undernutrition in Bangladesh is no better than in other low-and middle-income countries. The Food Security Nutritional Surveillance Project reported that 18% of women are undernourished nationally based on mid-upper arm circumference (MUAC) measurements, with an alarmingly high prevalence of 40% in Sylhet region (HKI and JPGSPH, 2014). Sylhet is one of the most vulnerable regions in northeast Bangladesh and includes diverse terrain such as plains, hills, Haor (wetlands) and areas prone to flash flooding (HKI and JPGSPH, 2014). Several important indicators of healthcare, maternal nutrition and socio-economic status are low in this region of the country (HKI and JPGSPH, 2014;NIPORT/Bangladesh et al., 2016). In contrast to the national decrease observed in Bangladesh overall, child malnutrition has increased in Sylhet in recent years (NIPORT/Bangladesh et al., 2016).
Numerous factors are associated with women's undernutrition, including socio-economic and demographic characteristics, smoking habits, household food insecurity status, decision-making autonomy and the area of residence (Balarajan & Villamor, 2009;Corsi et al., 2011;Hasan et al., 2017;Kamal et al., 2015;Khan & Kraemer, 2009). Nationally representative studies in Bangladesh have reported associations between child or maternal malnutrition and the wealth index, which is a composite indicator of several household socio-economic characteristics (NIPORT/Bangladesh et al., 2016).
However, it remains unclear whether specific socio-economic indicators-such as the type of latrine, source of drinking water, household construction materials or the number of children-influence the nutritional status of women, especially those from poor and very poor households.
Household food security status is a critical factor associated with childhood malnutrition that contributes to maternal malnutrition (S. L. Young et al., 2014). However, only a small number of studies have tested this association nationally or among the general Bangladeshi population, and no studies have been conducted among vulnerable populations living in specific regions of Bangladesh. In addition to the household food security status, biological factors also affect child malnutrition, and the mother plays a significant role in the interaction between children and the environment (Victora et al., 2008). Intrauterine nutritional deficiency and a lack of handwashing reduce the energy balance of the child (Sawaya & Roberts, 2003), which may have harmful effects during the late prenatal stages and lead to short height due to long-term nutritional deficits (Victora et al., 2008). However, many other factors have also been associated with maternal and child nutrition, including poor maternal reproductive healthcare-related indicators, experience of domestic violence and a lack of handwashing with soap (Adhikari et al., 2020;Chattopadhyay et al., 2019;Nguyen et al., 2017;Nkoka et al., 2019).
However, there are some practical limitations to the use of BMI as a rapid assessment tool, as it can be difficult to measure the height and weight of debilitated, disabled or acutely ill individuals. Moreover, the instruments used to measure weight and height/length are not always precise. High-precision instruments are very costly in resource-poor settings or population-based surveys and require trained personnel, which is not always possible. MUAC represents a sensible alternative in such situations and is the most common anthropometric measure used to assess the nutritional status of children (Briend et al., 1987;Sultana et al., 2015;WHO Working Group, 1986), particularly in emergency and crisis settings. Therefore, MUAC is considered as a reliable method for determining the nutritional status of women (Sultana et al., 2015). Several studies have been conducted to assess the association between maternal MUAC and children's nutritional indicators in various settings and populations (Assefa et al., 2012;Gewa et al., 2012;Kpewou et al., 2020;Vasundhara et al., 2020). These studies measured the nutritional status of either lactating or pregnant women (representing the general population) in facilities or households but did not specifically focus on women from poor or very poor households. Thus, there is a lack of research on the relationships between maternal underweight status determined using MUAC and the multiple factors mentioned above in rural Bangladesh, particularly among women belonging to vulnerable households.

Key messages
• Low maternal MUAC may indicate worsening nutritional status among children.
• To reduce maternal underweight status, public health strategies should focus on water, sanitation and hygiene (WASH), household food insecurity and optimal maternal healthcare practices.
• MUAC measuring tapes are cheap and widely available and enable underweight women to be easily identified in low-resource settings.
To address this knowledge gap, this paper aimed to identify the factors associated with maternal MUAC and investigate whether these indicators are significantly associated with children's nutritional status among a socio-economically deprived population in rural Bangladesh. To our knowledge, this is the first study of maternal MUAC among poor and very poor communities. The results of this study are expected to inform the formulation of appropriate strategies to reduce the worsening nutritional status of babies yet to be born in the Sylhet region of Bangladesh based on the MUAC of pregnant women.

| Source of data
Data for this study were extracted from the baseline crosssectional survey of the Suchana programme, a large-scale development programme that is currently ongoing in Sylhet division Haque et al., 2020). A household wealth ranking was completed; all listed households were verified as per the Suchana programme inclusion criteria and, if selected, given an identification number. Finally, households with a mother-child pair were selected for the baseline survey using a systematic sampling method ; the age range for mothers was 15-49 years and children 0-23 months. If any selected households were unavailable due to non-response or other reasons ( Figure S1), we replaced the household with the next immediate household from the sampling frame in an anti-clockwise direction (Haque et al., 2020) to survey the required number of households by phase and age group, according to our randomly generated listing.
In this analysis, cases were excluded when the maternal age was <20 years.
Survey data collection was conducted using Android tablets complemented with custom-developed Java software. Separate standard operating procedures were followed for interviewing and anthropometry. The survey team underwent 40 days of training, including field-testing and adjustment of the data collection tools. The training included a brief programme overview; general rules, norms and guidance on survey implementation; the survey methodology, such as team composition, sampling and household selection process; a detailed discussion of the questionnaire form, interview techniques, administration of questionnaires and handling the Android PDAs; role play exercises aided by demonstrations to better understand techniques for asking sensitive questions and methods of ensuring the quality of the data. The dedicated anthropometry team was trained on anthropometric measurements and how to assemble, calibrate and maintain their tools, as well as the key points to check while taking measurements. A master trainer was involved in training the anthropometric team.

| Variables of interest
The conceptual framework of factors influencing the association between maternal undernutrition and child nutrition is depicted in Figure 1. The outcome variables for this study were indicators of child nutritional status, such as wasting, stunting and underweight. The length of children was measured using seca 416 infantometers (seca, Hamburg, Germany) with a precision of 0.1 cm. Seca 874 weight scales with a precision of 1 gm were used to measure maternal and child weight. Maternal MUAC was measured at the mid-point between the tip of the shoulder and the tip of the elbow (olecranon process and acromion process; Mehranfar et al., 2020). The left arm was bent to a right angle, the wide end of the tape was placed on the shoulder and the length to the elbow was measured. The midpoint of the left upper arm was determined (and ideally, marked with a pen).
Then, the arm was straightened, and the tape was placed at the mid-point to measure MUAC, ensuring the tape was not too tight or too loose. Three readings were taken, and the average was computed using commands in the database. Low maternal MUAC was defined as a MUAC < 23 cm, which correlates closely with low BMI (Ghosh et al., 2019;Sultana et al., 2015). Length-for-age (LAZ), weight-for-age (WAZ) and weight-for-length (WLZ) Z scores were calculated as indicators of child nutritional status according to the 2006 WHO Standards For Children, using Z-score scale = (observed value À average value of the reference population)/standard deviation value of reference population (Haque et al., 2019;Onis & Bloessner, 1997).
Stunting was defined as a LAZ < À2, wasting as a WLZ < À2 and underweight as a WAZ < À2.
Several variables were considered as explanatory variables based on an extensive literature review and previous descriptive studies.
The variables relevant to maternal nutrition included current age, age at marriage and first pregnancy, formal schooling (at least 1 year), at least four antenatal care check-up visits during last pregnancy, attendance by qualified health personnel during last delivery, use of contraceptives, type of delivery (vaginal or caesarean section), experience of domestic violence and handwashing with soap and water before eating food and after defecation or handling domestic animals.
The household demographic variables included household size, the sex and education of the household head, religion, type of latrine (improved or non-improved), washing utensils with tube well water in all seasons, type of floor (improved or non-improved) and household food insecurity status.
Maternal experience of domestic violence was defined as a husband threatening to divorce or remarry, or verbal or physical abuse by their husband or other family members (Haque et al., 2020). The Household Food Insecurity Access Scale (HFIAS) was employed to define food insecurity according to the Food and Nutrition Technical Assistance's guidelines and categorized as food secure, mildly food insecure, moderately food insecure or severely food insecure (Coates et al., 2007;Haque et al., 2017).

| Quality control
The team supervisors were responsible for responding to critical questions on use of the Android tablets and also verified that the respondents from non-responsive households were truly unavailable or had opted out of participation. The precision of all instruments was checked every morning, and weighing scales were calibrated daily.
The supervisors cleaned all instruments before going to the field sites, and the anthropometric team cleaned the instruments carefully after each use. In addition to the data collection team, an independent quality control team of four quality control officers made random visits to the data collection teams to observe the data collection, sampling and re-interviewing processes for a randomly selected sub-sample of respondents.

| Statistical analysis
All statistical analyses were performed using Stata 14 (StataCorp, College Station, TX, USA). Data were visualized using box plots, histogram, pie charts, bar diagrams or scatter plots. Descriptive statistics such as the mean/standard deviations for quantitative variables and frequencies/proportions for qualitative variables were used to summarize the data and segregated by maternal nutritional status. The chi-square test was used to assess the association between two categorical variables; independent t tests were used to determine the significance of the differences between the means of two groups for normally distributed data. Primarily, simple logistic regression analysis was used to establish the strength of associations between low maternal MUAC and other variables. Multiple logistic regression analysis was employed to identify the factors potentially associated with maternal underweight status and assess the association between maternal underweight status and indicators of child nutritional status after controlling for union as a cluster. Covariates with a p < 0.25 in the bivariate analysis were included in the multiple regression model F I G U R E 1 Conceptual framework of factors associated with maternal and child nutritional status via a stepwise forward selection method (Bursac et al., 2008;Haque et al., 2020). P < 0.05 was considered as the level of significance, and 95% confidence intervals were calculated to assess the directions and strength of the effects.

| Ethical considerations
This study was approved by the Research Review Committee and Ethical Review Committee, the two obligatory components of the institutional review board of icddr,b. Informed written consent was obtained from study participants.

| RESULTS
A total of 7,723 households were visited. Of these, 2,283 households were not surveyed for various reasons ( Figure S1). Moreover, 371 households were not included in the analysis, as maternal age was <20 years, because this adolescent age still has potential to grow further. In total, 5,069 households with mother-child pairs were analysed in this study. Overall, 2,367 (47%) mothers had a MUAC < 23 cm (Figure 2). Among the mothers with a low MUAC, the prevalence of stunting among their children was 48.8%, wasting was 13.4% and underweight was 38.9%. The descriptive findings for this dataset segregated by maternal nutritional status are shown in

| DISCUSSION
This study examined data collected during the baseline survey of the Suchana programme to identify factors significantly associated with maternal underweight status, as indicated by MUAC, and indicators of child nutritional status. After controlling for many relevant covariates, which were available in our data set, maternal MUAC < 23 cm was significantly correlated with wasting, stunting and underweight status among children. These findings are consistent with results from other settings (Assefa et al., 2012;Gewa et al., 2012;Kpewou et al., 2020;Vasundhara et al., 2020), suggesting that maternal MUAC is a significant predictor of childhood malnutrition.
F I G U R E 2 Age-stratified prevalence of maternal underweight status based on a maternal low midupper arm circumference <23 cm  (Balarajan & Villamor, 2009;Hasan et al., 2017;Kamal et al., 2015;Khan & Kraemer, 2009). Similarly, the results of several nationally representative studies indicated that mothers belonging to households in low socio-economic strata are significantly more vulnerable to being underweight (Balarajan & Villamor, 2009;Khan & Kraemer, 2009;Oddo et al., 2012). These studies used the household wealth index as an indicator of socio-economic status, which reflects a number of important indicators of economic status such as the type of latrine, source of water and construction material.  children's play and feeding children in polluted environments have been reported as constant, cumulative health risks during the vital window of child growth and development (Ngure et al., 2014). Our findings of moderate and severe food insecurity at the household level and the significant association between household food insecurity and maternal undernutrition are in agreement with earlier reports (Kamal et al., 2015) that these variables are predictors of malnutrition in children Corsi et al., 2011).
In terms of the maternal health outcomes associated with undernutrition in this study, receiving ANC during pregnancy, the type of birth attendant during the last delivery and contraceptive use have been widely scrutinized as factors that contribute to both maternal and child health Perry et al., 2017;Rahman et al., 2016;Seward et al., 2017;Siddique et al., 2018).
This study suggests that a better understanding of several who played critical roles in generating and refining the data collection tools. We would also like to acknowledge Kazi Istiaque Sanin, Assistant Scientist as an internal reviewer and Andrea Devlin of Science Editing Experts for editing and linguistic revision of the manuscript.
This study was made possible by the committed contribution of FCDO.

CONFLICTS OF INTEREST
The authors have declared that no conflict of interest exists.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.