Assessing the Intergenerational Linkage between Short Maternal Stature and Under-Five Stunting and Wasting in Bangladesh

Short maternal stature is identified as a strong predictor of offspring undernutrition in low and middle-income countries. However, there is limited information to confirm an intergenerational link between maternal and under-five undernutrition in Bangladesh. Therefore, this study aimed to assess the association between short maternal stature and offspring stunting and wasting in Bangladesh. For analysis, this study pooled the data from four rounds of Bangladesh Demographic and Health Surveys (BDHS) 2004, 2007, 2011, and 2014 that included about 28,123 singleton children aged 0–59 months born to mothers aged 15–49 years. Data on sociodemographic factors, birth history, and anthropometry were analyzed using STATA 14.2 to perform a multivariable model using ‘Modified Poisson Regression’ with step-wise backward elimination procedures. In an adjusted model, every 1 cm increase in maternal height significantly reduced the risk of stunting (relative risks (RR) = 0.960; 95% confidence interval (CI): 0.957, 0.962) and wasting (RR = 0.986; 95% CI: 0.980, 0.992). The children of the short statured mothers (<145 cm) had about two times greater risk of stunting and three times the risk of severe stunting, 1.28 times the risk of wasting, and 1.43 times the risk of severe wasting (RR = 1.43; 95% CI: 1.11, 1.83) than the tall mothers (≥155 cm). These findings confirmed a robust intergenerational linkage between short maternal stature and offspring stunting and wasting in Bangladesh.


Introduction
Undernutrition remains highly prevalent in most low-and middle-income countries (LMICs), especially in countries from South Asia. In 2016, globally among children under five years of age 39%, or 155 million, were stunted and 50%, or 52 million, were wasted [1]. In Bangladesh in 2014, the prevalence of stunting and wasting among children under five years was estimated at 36% and 14%, respectively [2]. Maternal undernutrition is a significant contributor to child undernutrition in LMICs [3]. A recent study using a pooled analysis of data from 137 developing countries showed that, in 2011, 14.4% of stunting among 44.1 million children aged under two years (6.4 million cases) was attributable to maternal undernutrition [4].
Maternal height is an indicator of intergenerational linkages between maternal and child nutrition and health [5]. As human height is inherited from parents, genetic factors should mainly determine the relationship between maternal height and offspring growth [6]. However, other factors such as metabolic programming, epigenetics, and the intergenerational transmission of poverty also play for determinants of child undernutrition [21] based on available data from BDHS 2004BDHS , 2007BDHS , 2011BDHS , and 2014.

Primary Outcomes
The primary outcomes of this study were stunting and wasting. Stunting and wasting were measured using two standard indices of physical growth of the children based on recommended guidelines of the World Health Organizations child growth standards (WHO 2006) [22]. The indices to measure stunting and wasting were height-for-age Z-score and weight-for-height Z-score, respectively. A child's height-for-age was calculated by dividing his/her height by the median height for a child considering his/her age and sex. Similarly, a child's weight-for-height was calculated by dividing his/her weight by the median weight for a child of that height and sex. Then, every computed number was standardised as a Z-score with a mean of 0 and a standard deviation (SD) of 1. A child was considered stunted or severely stunted if the height-for-age Z-score was less than 2 SD or less than 3 SD below the World Health Organizations determined mean scores for height-for-age. Similarly, a child was considered as wasted or severely wasted if the weight-for-height Z-score was less than 2 SD or less than 3 SD below the World Health Organizations determined mean scores for weight-for-height.

Main Exposure
The main exposure of this study was the maternal stature measured as height in cm. In this analysis, maternal height was used as both a continuous and a categorical variable. Continuous maternal height was used to measure the effect of every 1-cm increase of maternal height on the primary outcome (offspring stunting and wasting), while maternal height was categorized to assess the effect of the shortest maternal height category on offspring stunting and wasting compared to the tallest maternal height category. Therefore, this study categorized maternal height into four categories as follows: <145.0 cm (short stature), 149.9-145.0 cm, 154.9-150.0 cm, and ≥155.0 cm (tall stature, reference group). These catagories of maternal height were adapted from similar earlier studies [12,13,23].

Covariates
This study used covariates that are well-known risk factors for child undernutrition [24]. In this analysis, eleven selected covariates were selected as maternal, child, household, and community level variables. The maternal covariates were woman's age at birth, her education, and her occupation. The child covariates were child age, sex, birth order, and birth interval. The household and community level covariates were household wealth, husband's education, the location (rural or urban), and region of residence. Year of survey and recall period were also included as covariates. The variable "recall period" was calculated as the difference from the date of the interview to the child's date of birth. A composite household wealth index was created from a list of household assets and facilities using the principal component analysis (PCA) to weight the individual items [25]. The household wealth index was calculated as the sum of the weighted scores and divided into quintiles for categorical analyses.

Statistical Analysis
Data analysis was performed using STATA 14.1 (Stata-Corp, College Station, TX, USA). For survey data analysis, command 'svy' were employed to adjust for the cluster sampling design in frequencies and cross-tabulations. This study applied multivariable models using Poisson Regression with robust error variance and sampling weights [26] to estimate the association between maternal height and child anthropometry adjusted for other covariates. In this method, the associations were expressed as relative risks (RR) with 95% confidence intervals (CIs). Two-tailed Wald tests were performed for measuring the level of significance, which was calculated as p < 0.05 or the exact p value when p < 0.001.
There were several steps in the model building process. First, an unadjusted univariable analysis was performed for the each of the covariates with the outcomes, and the covariate was selected as a factor. All the covariates that were statistically significant at the level of significance p < 0.25 were used to create the baseline model. Then the collinearity among the covariates was checked using the Stata command 'collin'. This analysis showed a collinearity between birth order and birth interval. To minimize the collinearity, it was decided to exclude birth interval and keep birth order in the model, as birth order is a significant predictor of stunting among Bangladeshi children [27]. With the base model, a multi-stage backward elimination modeling technique was used to get the final model for assessing the significant factors for the primary outcome. In this process, the covariates that were not confounders, or not statistically significant at the level of significance p < 0.05, were excluded from the base model. In the final model, the assumption of linearity with fractional polynomial was checked for the continuous variables to choose the appropriate parameterization [28]. Moreover, the final model was checked for any interaction between main exposure and all significant covariates. The level of significance was considered at p < 0.01 for the interactions. All the possible interactions were considered in the final model and the non-significant interactions were gradually excluded using a backward elimination method. The significant interactions were added as effect modifiers in the full model. Then, the final models with both the factors and effect modifiers were tested for goodness-of-fit using both Pearson and deviance chi-square statistics. If these tests were not significant (p > 0.05), the model was considered as the best fit for Poisson regression [29]. Table 1 presents the percentage of stunted, wasted, severely stunted, and severely wasted children under-five and their maternal, birth, and socio-demographic characteristics. In this study, we analyzed data from a total of 25,635 children under-five, among whom there were 10,701 (42.1%) stunted children, 3902 (15.4%) wasted children, 4091 (15.8%) severely stunted children, and 866 (3.4%) severely wasted children. One in five stunted children and one in four severely stunted children had short mothers. About two-thirds of the mothers had at least a primary level of education. One-third of the children were their mothers' first child. Most of the children resided in rural areas. The distribution patterns were similar across all the categories of maternal, child, and household covariates among the stunted, severely stunted, wasted, and severely wasted children (Table 1). The distributions of these covariates were also comparable across the maternal height categories among all children under-five (Table S1). Note: * Stunted: height-for-age Z-score < −2 SD ** Severely stunted: height-for-age Z-score < −3 SD + Wasted: weight-for-height Z-score < −2 SD ++ Severely wasted: weight-for-height Z-score < −3 SD. Table 2 presents the mean maternal height in cm among the children under-five who were, or were not, stunted, wasted, severely stunted, or severely wasted. Among the 25,635 children under five included in the analysis, the mean height of the mothers was 150.8 cm (95% CI: 150.7, 150.9 cm). However, the average height among the mothers of stunted and severely stunted children was 149.3 cm (95% CI: 149.2, 149.4 cm) and 148.4 cm (95% CI: 148.3, 148.6 cm), respectively. The mean difference in maternal height was 2.6 cm (95% CI, 2.5, 2.8 cm, p < 0.001) lower among stunted than non-stunted children. There was also a significant difference in maternal height between severely and non-severely stunted children (Mean difference in height: 2.8 cm; 95% CI: 2.6, 3.0 cm, p < 0.001). The average height among the mothers of wasted children was 150.3 cm, (95% CI: 150.2, 150.5 cm), and 150.3 cm for severely wasted, (95% CI: 149.9, 150.7 cm). Note: * Stunted: height-for-age Z-score < −2 SD ** Severely stunted: height-for-age Z-score < −3 SD + Wasted: weight-for-height Z-score < −2 SD ++ Severely wasted: weight-for-height Z-score < −3 SD.

Association of Maternal Stature with Offspring Stunting and Severe Stunting
Our results showed that both stunting and severe stunting were significantly associated with maternal, child, and household and community level covariates ( Table 3, Table S2). The adjusted models estimated that a one-centimeter increase of maternal height was a protective factor for child stunting (RR = 0.960; 95% CI: 0.957, 0.962) and severe stunting (RR = 0.941; 95% CI: 0.935, 0.946). Compared to the children of the tall mothers (≥155 cm), the children from the shortest mothers (>145 cm) had about twice the risk of stunting (RR = 2.10; 95% CI: 1.97, 2.23) ( Table 3). Mothers with short stature had about three times the risk of severe stunting (RR = 2.97; 95% CI: 2.65, 3.33) compared to tall mothers (Table S2).

Interaction between Maternal Stature and Household Wealth and Its Effect on Stunting
When we tested the interaction between maternal height and household wealth in the model that measured the association between maternal height and offspring stunting adjusted for other covariates, we a found a robust significant interaction between maternal height categories and household wealth (p < 0.001) (Table S3). From this adjusted model, we measured the combined effect of household wealth and maternal height on the prevalence of stunting (Figure 1). We found that the prevalence of stunting gradually increased from the wealthiest to the poorest wealth quintile among the children of the tall stature mothers (≥155 cm). However, for the children of the short stature mothers (<145 cm), the prevalence of stunting remained almost the same across the wealth quintiles from the wealthiest to the poorest. When we tested the interaction between maternal height and household wealth in the model that measured the association between maternal height and offspring stunting adjusted for other covariates, we a found a robust significant interaction between maternal height categories and household wealth (p < 0.001) (Table S3). From this adjusted model, we measured the combined effect of household wealth and maternal height on the prevalence of stunting (Figure 1). We found that the prevalence of stunting gradually increased from the wealthiest to the poorest wealth quintile among the children of the tall stature mothers (≥155 cm). However, for the children of the short stature mothers (<145 cm), the prevalence of stunting remained almost the same across the wealth quintiles from the wealthiest to the poorest.

Association of Maternal Stature with Offspring Wasting and Severe Wasting
We also found a robust significant association between maternal height and wasting after adjusting for covariates ( Table 4, Table S4). The adjusted models showed that a one-centimeter increase in maternal height was associated with a significant reduction in relative risk for child wasting (RR = 0.986; 95% CI: 0.980, 0.992) and severe wasting (RR = 0.984; 95% CI: 0.971, 0.997). Children of the short statured mothers (<145 cm) were significantly more likely to suffer from both wasting (RR = 1.28; 95% CI: 1.14, 1.43) and severe wasting (RR = 1.43; 95% CI: 1.11, 1.83) compared to tall mothers (≥155 cm).

Association of Maternal Stature with Offspring Wasting and Severe Wasting
We also found a robust significant association between maternal height and wasting after adjusting for covariates (Table 4, Table S4). The adjusted models showed that a one-centimeter increase in maternal height was associated with a significant reduction in relative risk for child wasting (RR = 0.986; 95% CI: 0.980, 0.992) and severe wasting (RR = 0.984; 95% CI: 0.971, 0.997). Children of the short statured mothers (<145 cm) were significantly more likely to suffer from both wasting (RR = 1.28; 95% CI: 1.14, 1.43) and severe wasting (RR = 1.43; 95% CI: 1.11, 1.83) compared to tall mothers (≥155 cm).

Main Findings
Based on our analysis of a nationally representative sample, maternal height was inversely associated with stunting and wasting among children under-five in Bangladesh. Our results provided robust evidence that children of mothers of short stature have a substantially higher risk of stunting and modest risk of wasting compared to the children of taller mothers. The finding that the intergenerational effect of undernutrition was not ameliorated by current household wealth after adjustment for other maternal and child level factors is an important finding. In Bangladesh and other LMICs where there is a high level of child undernutrition, the intergenerational effects of maternal undernutrition may lag the effects of interventions aimed at reducing child stunting. Monitoring maternal height may give a good indication of the reducing effects of intergenerational malnutrition.

Strength and Limitations
The strength of this study was that we used a nationally representative sample and pooled several rounds of surveys, which gave us a large enough sample size to estimate the effect [30]. Another strength was that we used the Poisson regression model with a robust variance that was suggested as one of the less biased approaches to obtain the correct estimates of the risk ratio for a dichotomous outcome like stunting or wasting [31]. It is a crucial study for Bangladesh where the high prevalence of child stunting and wasting remain as public health challenges despite the government and non-government policies and programs trying to improve maternal and child nutrition. Our study specifically assesses the intergenerational effect of maternal height on under-five child stunting and wasting after adjusting for well-established determinants of maternal and child undernutrition, such as age, birth history, social determinants like education, household wealth, residence, and location. However, our limitation was that we did not control previously established immediate causes of child malnutrition, such as inappropriate infant and young child feeding practices, household food insecurity, infectious disease, and poor health-seeking behavior in our analysis [32]. Although genetic or epigenetic factors are the intergenerational determinants jointly related to maternal and offspring nutrition, we cannot assess the effect of these factors because these data were not available in the BDHS. Moreover, it was beyond our scope to describe the mechanism of how maternal height and child growth faltering were associated, as we only tested the hypothesis that maternal height and child anthropometric outcome were associated in Bangladesh.

Association of Maternal Stature with Offspring Stunting and Wasting
Our findings of a robust inverse association between maternal height and stunting and wasting among children under-five were similar to other studies [12,13]. A large national survey (NFHS, 2005(NFHS, -2006 in India reported that each one-centimeter increase of maternal height was inversely associated with under-five stunting (RR = 0.971; 95% CI: 0.968, 0.0973) and wasting (RR = 0.989; 95% CI: 0.984, 0.994) [13]. Similarly, another large study that pooled data from 54 low-income countries showed that for every one-centimeter increase of maternal height significantly reduced the risk of stunting (RR = 0.968; 95% CI: 0.967, 0.968) and wasting RR = 0.994; 95% CI: 0.993, 0.995) among children under-five [12]. Even though this multi-country pooled analysis included BDHS 1997-2007, the present study analyzed the most recent nationally representative surveys for Bangladesh (BDHS 2004(BDHS -2014. The findings from this study confirmed the intergenerational association between maternal stature and child undernutrition in Bangladesh.

Maternal Short Stature and the Risk of Offspring Stunting and Wasting
This study revealed a robust association between short stature of the mothers and the risk of their offspring stunting after adjustment for socioeconomic status, while this risk was greatest for the mothers whose stature was <145 cm. These findings have been similar to the results of previous studies [12,13]. The findings from other studies that investigated the intergenerational and other pathways of growth faltering can help explain the association between short maternal height and the risk of stunting. The intergenerational linkage between maternal short stature and the offspring's growth faltering in utero can be explained by biomechanical (i.e., maternal organ size) and biological mechanisms (maternal nutrition stock, and fetal programming). Prior research has shown that short women are more likely to have narrower pelves which affect the uterine environment for optimum fetal growth and leads to the birth of low birth weight (LBW) babies [33].
Moreover, maternal short stature is an indicator of her cumulative net nutrition and biological deprivation over periods of rapid growth [34]. Poor nutritional status of women in pregnancy adversely affects placental growth that causes inadequate nutrient transfer and oxidative stress to the fetus. Nutrient deficiencies in utero cause epigenetic modification (i.e., DNA methylation) to alter fetal programming that results in fetal growth faltering, and delivery of LBW or SGA babies [35]. LBW or SGA infants born with nutrient deficiencies and immature immune systems are more susceptible to infection, while infection increases the risk of acute undernutrition by mucosal damage, impaired absorption of essential nutrients and loss of weight [36].
Along with exposure to intergenerational factors and infectious diseases, dietary factors like insufficient quality of complementary feeding have a paramount role in growth faltering in the first two years of life [37]. Poor diet quality with low dietary diversity is a predictor of micronutrient inadequacies [38], while the synergistic interaction between micronutrient deficiencies and infections causes growth faltering among children under-five in LMICs [39,40]. Evidence also suggests that household food insecurity is a predictor of child undernutrition and linear growth faltering in diverse settings of LMICs [41,42]. Acute growth faltering is reflected as wasting in infancy and childhood, which can lead to long-term linear growth faltering or stunting [43], if not treated with immediate inputs such as appropriate infant feeding, dietary diversity, disease prevention, or health facilities and underlying causes like household food security. Hence, the intergenerational effects of growth impairment in utero interact with other immediate and underlying determinants of undernutrition influencing growth faltering from birth to five years of age.

Interactions between Household Wealth and Maternal Short Stature on Child Stunting
This study showed an important interaction between maternal stature and household wealth and its association with child stunting. The interaction showed that amongst short mothers there was a similarly high prevalence of stunting across all wealth quintiles from the wealthiest to the poorest families. However, this study found an effect of social inequalities on stunting, which has been reported in earlier studies [44,45], among the children of the tall mothers but not among the children of the short statured mothers. These findings demonstrate that the intergenerational effect of undernutrition remains important beyond the current wealth status of the household. Moreover, these findings are consistent with the recent concept of "the stunting syndrome" that proposes an intergenerational transmission of poor nutrition from mother to child, in which, short women, who were stunted in childhood, are more likely to have stunted children and thus, generating an intergenerational cycle of poverty [46]. These findings suggest that without addressing intergenerational factors, immediate determinants of food, care, health, or economic input will not be sufficient to overcome the long-term negative impact of intergenerational transmission of poor nutrition on child linear growth.

Policy Implications
In Bangladesh, a major contributor to social disparities in child stunting is maternal short stature [45]. The country has the highest proportion of women in the world with short stature (height < 145 cm), with more than one in ten women being short in stature or stunted (height < 145 cm) [47]. The high burden of maternal short stature implies that there has been little improvement in attaining adequate nutrition for women in adulthood for optimum fetal and child nutrition. A report from the MINIMat trial, which targeted pregnant women in rural Bangladesh, showed that combined food and micronutrient supplements in pregnancy were not effective in reducing child growth failure, where nearly one-third of the newborns had low birth weight [48]. This study also assumed that the high proportion of LBW might be related to the average short stature of the mothers (149.8 cm) who participated in this trial.
Furthermore, in Bangladesh, where inappropriate infant and young child feeding, poor sanitation, and infectious diseases are highly prevalent, LBW children who are born with compromised immunity are at high risk of infections, wasting, and stunting [36,49]. Evidence suggests that one-fourth of child growth faltering occurs in utero and continues until two years of age, and then slowly continues until five years of age in LMICs, including Bangladesh [50]. Therefore, a life-cycle approach is needed for nutrition-health related interventions for young girls, adolescents, preconception, and the first thousand days of life, which are 'critical windows of opportunity' for child nutrition [51,52]. Recently, United Nations System Standing Committee on Nutrition recommended improving maternal nutrition, with a special emphasis on short stature women, through improving preconception or conception diet quality to ensure optimum nutrition in utero to break the intergenerational cycle of growth faltering of the fetus that leads to LBW and stunting at a later stage of life [53]. Evidence also suggests that nutrition interventions such as promotion of nutrition and improved water and sanitation practices can improve nutrition and growth to break the vicious cycle of growth faltering in childhood [46]. As nutrition interventions have long term consequences on adult height, they should also target girls, especially adolescent girls, to ensure they get adequate nutrition to achieve the optimum height in adulthood to minimize short stature. Hence, the current study findings draw attention to program managers and policymakers to focus on improving nutrition throughout the lifecycle to reduce the risk of intergenerational transmission of undernutrition from mothers to their child to prevent stunting as well as wasting in the countries like Bangladesh.

Conclusions
In summary, short maternal height is associated with an increased risk of stunting and wasting among children under-five in Bangladesh. This finding suggests an intergenerational linkage between maternal and child chronic undernutrition that will need addressing for sustained improvements in maternal and child nutrition to reduce under-five stunting in the current context of Bangladesh.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6643/11/8/1818/s1, Figure S1: Conceptual framework for assessing association between maternal height and offspring stunting and wasting. (adapted from UNICEF 2013), Table S1: Frequency distribution (weighted) of the covariates across the maternal height categories in child anthropometry datasets, Table S2: Association of maternal height (cm) with severe stunting of the under-five children adjusted for other covariates, unadjusted and adjusted Relative Risk with 95% Confidence Interval, Table S3: Association between stunting and maternal height (cm) adjusted for significant interaction between maternal height and other covariates, unadjusted and adjusted Relative Risk with 95% Confidence Interval, Table S4: Association of maternal height (cm) with severe wasting of the under-five children adjusted for other covariates, unadjusted and adjusted Relative Risk with 95% Confidence Interval.