Malnutrition and Its Determinants Are Associated with Suboptimal Cognitive, Communication, and Motor Development in Tanzanian Children 1 3

Transcription

1 The Journal of Nutrition Nutritional Epidemiology Malnutrition and Its Determinants Are Associated with Suboptimal Cognitive, Communication, and Motor Development in Tanzanian Children 1 3 Christopher R Sudfeld, 4 * Dana Charles McCoy, 8 Günther Fink, 4 Alfa Muhihi, 9 David C Bellinger, 7,10 Honorati Masanja, 9 Emily R Smith, 4 Goodarz Danaei, 4,5 Majid Ezzati, 11 and Wafaie W Fawzi 4 6 Departments of 4 Global Health and Population, 5 Epidemiology, 6 Nutrition, and 7 Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA; 8 Center on the Developing Child, Schools of Education and Public Health, Harvard University, Cambridge, MA; 9 Ifakara Health Institute, DaresSalaam,Tanzania; 10 Department of Neurology, Boston ChildrenÕs Hospital, Boston, MA; and 11 Medical Research Council-Public Health England (MRC-PHE) Centre for Environment and Health, Departments of Epidemiology and Biostatistics, Imperial College London, London, United Kingdom Abstract Background: A large volume of literature has shown negative associations between stunting and child development; however, there is limited evidence for associations with milder forms of linear growth faltering and determinants of malnutrition in developing countries. Objective: The objective of this study was to assess the association between anthropometric growth indicators across their distribution and determinants of malnutrition with development of Tanzanian children. Methods: We used the Bayley Scales of Infant Development III to assess a cohort of 1036 Tanzanian children between 18 and 36 mo of age who were previously enrolled in a neonatal vitamin A trial. Linear regression models were used to assess standardized mean differences in child development for anthropometry z scores, along with pregnancy, delivery, and early childhood factors. Results: Height-for-age z score (HAZ) was linearly associated with cognitive, communication, and motor development z scores across the observed range in this population (all P values for linear relation < 0.05). Each unit increase in HAZ was associated with (95% CI: 0.05, 0.13), (95% CI: 0.07, 0.14), and (95% CI: 0.09, 0.16) higher cognitive, communication, and motor development z scores, respectively. The relation of weight-for-height z score (WHZ) was nonlinear with only wasted children (WHZ <22) experiencing deficits (Pvalues for nonlinear relation < 0.05). Wasted children had (95% CI: 20.97, 20.29), (95% CI: 20.64, 0.01), and (95% CI: 20.86, 20.23) z score deficits in cognitive, communication, and motor development z scores, respectively, relative to nonwasted children. Maternal stature and flush toilet use were associated with higher cognitive and motor z scores, whereas being born small for gestational age (SGA) was associated with a (95% CI: 20.30, 20.01) z score deficit in cognition. Conclusions: Mild to severe chronic malnutrition was associated with increasing developmental deficits in Tanzanian children, whereas only wasted children exhibited developmental delays during acute malnutrition. Interventions to reduce SGA, improve sanitation, and increase maternal stature may have positive effects on child development. This trial was registered with the Australian New Zealand Clinical Trials Registry as ACTRN JNutr2015;145: Keywords: infant, nutrition, malnutrition, body height, body weights and measures, child development, intelligence, cognition, psychomotor performance, sanitation Introduction In 2004 an estimated 219 million children under 5 y of age residing in developing countries did not meet their development 1 Funding for the parent trial was provided by the Bill & Melinda Gates Foundation supplied through the WHO. Funding for the follow-up study was provided by the Saving Brains Program, Grand Challenges Canada ( ). 2 Author disclosures: CR Sudfeld, DC McCoy, G Fink, A Muhihi, DC Bellinger, H Masanja, ER Smith, G Danaei, M Ezzati, and WW Fawzi, no conflicts of interest. potential (1). Suboptimal cognitive, language, and motor development may occur as a result of a combination of biological and psychosocial risk factors, which are often caused or exacerbated by poverty in developing country settings (2, 3). These developmental 3 Supplemental Tables 1 4 are available from the Online Supporting Material link in the online posting of the article and from the same link in the online table of contents at * To whom correspondence should be addressed. csudfeld@hsph. harvard.edu. ã 2015 American Society for Nutrition. Manuscript received April 23, Initial review completed June 1, Revision accepted September 8, First published online October 7, 2015; doi: /jn

2 deficits may persist throughout childhood and lead to poor schooling achievement and subsequent reductions in lifetime earnings (4, 5). As a result, alleviating early life adversity and corresponding developmental setbacks is currently being considered as one of the new sustainable development goals and may also support reaching long-term health and poverty reduction goals (6). A growing body of evidence from cross-sectional and prospective studies has shown that stunting [height-for-age z score (HAZ) 12 <22], an indicator of chronic malnutrition, is associated with reduced cognitive and motor development (7). In contrast, there is limited evidence regarding the independent association between wasting [weight-for-height z score (WHZ) <22], an indicator of acute malnutrition, and child development. Studies that thoroughly examine the shape of the relation between HAZ, WHZ, and other nutritional indicators across their distribution and child development are lacking. It is essential to determine whether mild forms of growth faltering are associated with suboptimal child development for the design of intervention trials and for studies that use linear growth faltering as a proxy for delay in other domains of child development (1). Despite the large volume of literature linking malnutrition to developmental deficits, there is comparatively limited evidence on the association between known risk factors for malnutrition and child development (8). Multiple studies have linked preterm birth and intrauterine growth restriction with developmental deficits in high income settings; however, data from low- and middle-income countries are much more limited (9, 10). To our knowledge, no studies have assessed the association between small-for-gestational-age (SGA) births and child development in the African context. Furthermore, no studies, to our knowledge, have examined the observational association between maternal height, water quality, and sanitation and child development in any low- and middle-income setting, despite their wellestablished links with malnutrition (11, 12). Determining the relation between underlying causes of malnutrition and child development is urgently needed to inform future intervention trials and studies looking to improve child growth and development concurrently. In this study, we examined the cross-sectional relation between child anthropometry and cognitive, communication, and motor development in a cohort of Tanzanian infants at mo of age. We hypothesized that mild forms of growth faltering are associated with significant but smaller delays in child development compared with stunting. In addition, to address the significant research gap on the relation between underlying causes of malnutrition and child development, we prospectively examined the relation between maternal, delivery, and child factors commonly associated with linear growth faltering and child development. Methods 12 Abbreviations used: BSID-III, Bayley Scales of Infant Development III; HAZ, height-for-age z score; LBW, low birth weight; LMP, last menstrual period; SGA, small for gestational age; SMD, standardized mean difference; WAZ, weight-for-age z score; WHZ, weight-for-height z score. Study design and data collection. This study used data from a cohort of children who were enrolled in a randomized, double-blind controlled trial of neonatal vitamin A supplementation conducted in Tanzania from August 2010 to March 2014 (13) (ACTRN ). Trial recruitment and data collection procedures have been presented elsewhere, but are summarized here briefly (13, 14). The trial participants included newborns in the Morogoro region of Tanzania, within the Ifakara Health InstituteÕs Health and Demographic Surveillance System, which covers ;2400 km 2, to allow for enrollment of infants born at a health care facility or at home. Newborns were eligible for random assignment if they were able to feed orally, were born within the previous 72 h, were not already enrolled in other clinical trials, if their family intended to reside in the study area for $6 mo postdelivery, and if their caregivers provided written informed consent. Trained study staff administered a baseline questionnaire to mothers in order to collect demographic, socioeconomic, and environmental information, as well as the date of each motherõs last menstrual period (LMP). Study staff also measured birthweight with the use of calibrated scales that had digital screens. Additional follow-up for trial participants with a focus on child development was conducted during the period from February to October Children were selected for participation in the follow-up study if they lived in Ifakara town or surrounding villages, if they were mo of age at the time of assessment, and if the caregiver consented to participate. Selected children and their caregivers were contacted at home, and those who provided informed consent were invited to central health clinics for child development and anthropometric assessment. At the clinic, trained female research nurses administered an adapted and translated (Swahili) version of the Bayley Scales of Infant Development III (BSID-III), including the cognitive, expressive language, receptive language, fine motor, and gross motor subscales (15). To ensure cultural appropriateness, 30 BSID-III items (13%) were adapted based on consensus of a local panel of child development experts. Adaptations included the replacement of unfamiliar images or terminology with more culturally relevant stimuli (e.g., changing a picture of an apple to a banana or redrawing cartoon images of white children in Western clothing to show black children in local attire). To maintain functional equivalence, replacement stimuli were selected to be of similar size, style, and complexity to the original stimuli. An additional 2 items on masculine/feminine pronoun usage (i.e., his compared with her) were removed from the BSID-III communication subscale because of their nonapplicability to the Swahili language. All BSID-III materials were translated and checked by local bilingual staff before field administration. A total of 4 research nurses were selected as data collectors based on quantitative performance evaluations from a pool of 6 who had completed a 3 wk training period. Biweekly field-based supervision and weekly staff meetings were used to prevent assessor drift and ensure continued quality of implementation. Overall, the adapted BSID-III showed adequate inter-rater reliability at the end of the training period (k = across subscales) and adequate internal consistency (a = across subscales). Length of children <24 mo of age was measured to the nearest 0.1 cm with the use of a length board (Seca) and height of children $24 mo was measured to the nearest 0.1 cm with the use of a portable stadiometer (Seca). Weight was measured to the nearest 0.1 kg with the use of a digital scale (Seca). HAZ, WHZ, and weight-for-age z score (WAZ) were calculated with the use of WHO child growth standards (16). At the clinic visit, we did not test the mother or child for HIV infection; as a result, the HIV-infection and exposure status of all study children is unknown. In the home, interviewers administered 6 yes or no response items from UNICEFÕs Multiple Indicator Cluster Survey Early Child Development module regarding childrenõs opportunities for cognitive stimulation within the home (17). A stimulation score was then created by summing the yes responses to the 6 items, with a score of 0 indicating low levels of cognitive stimulation and 6 indicating high levels of cognitive stimulation. Statistical methods. This study was restricted to 1036 children with BSID-III data at mo of age. BSID-III scores were analyzed for the 3 developmental subscales of the BSID-III: 1) cognition, 2) communication (comprising expressive and receptive communication skills), and 3) motor (comprising fine and gross motor skills). BSID-III raw scores by subscale were normalized into z scores, which facilitates comparability with other studies that use other child development assessment tools (7). We examined the cross-sectional association between child HAZ, WHZ, and WAZ and BSID-III cognitive, communication, and motor z 2706 Sudfeld et al.

3 scores at mo of age with the use of linear regression models to assess standardized mean differences (SMDs) in child development per 1 SD change in anthropometry z score. In order to first inform the shape and correct categorization of HAZ, WHZ, and WAZ variables, restricted cubic splines were used to assess the possible nonlinear relations nonparametrically (18). Knots were placed every 0.5 z scores across the full range of observed HAZ, WHZ, and WAZ distributions. Tests for nonlinearity used the likelihood ratio test, comparing the model with only the linear term to the model with the linear and the cubic spline terms. The spline analysis indicated that HAZ and WAZ were best modeled linearly for all child development outcomes. WHZ exhibited a significant nonlinear relation with BSID-III scores and was best captured dichotomously, with wasting defined as a WHZ $2 SDs below the WHO reference population median. Multivariate analyses for HAZ and wasting also included infant sex, infant age (in 2 mo categories), BSID-III assessor (n = 4), baseline maternal education (none/some primary, completed primary, or secondary), wealth index quintile, cognitive stimulation [High (top 25%) compared with low/medium (lowest 75%)], and randomized regimen (vitamin A or placebo). The wealth index quintile was defined by the first principal component of household assets and characteristics (bicycle, radio, mobile phone, television, motorcycle, car, or animal ownership; electricity; and roof type). Because of collinearity, models examining WAZ included these covariates but did not include HAZ and wasting. Next, we performed a prospective cohort analysis examining the association between pregnancy factors and child development by linking data collected in the original trial with the follow-up data collected at mo. We used linear regression models to calculate SMDs in BSID- III z scores for maternal factors including maternal age (<20, 20 30, or >30 y), maternal education (none/some primary, completed primary, or secondary), wealth quintile, maternal height (<150 cm, cm, or >155 cm), and parity (first born, 2nd 4th birth, or $5th birth). Multivariate models for maternal factors also included infant sex, infant age, BSID-III assessor, and randomized regimen. Delivery (preterm, low birth weight, etc.) and child anthropometry (HAZ, wasting, and WAZ) variables were not included in pregnancy models, because these factors are potential downstream mediators and adjustment for these factors will produce biased results. Similarly, we performed a prospective analysis of the association between delivery and child characteristics and child development. Linear regression models were used to calculate SMDs in z-scored BSID-III scores for child sex, low birth weight (LBW), preterm birth, SGA, sanitation, and water quality. Gestational age was calculated from maternal report of date of LMP and preterm birth was defined as <37 wk gestation. SGA was defined as birth weight <10th percentile for gestational age and sex with the use of International Fetal and Newborn Growth Consortium (INTERGROWTH) standards, AND LBW was defined as a birth weight of <2500 g (19). Sanitation quality was dichotomized as flush toilet compared with no flush toilet, whereas water quality was classified as improved or unimproved based on Joint Monitoring Program definitions (20). Multivariate models included adjustment for potential confounding factors including infant sex, infant age, BSID-III assessor, randomized regimen, maternal age at delivery, maternal education, wealth quintile, maternal height, and parity. Preterm and SGA were included in multivariate models together, whereas LBW was included alone because of overlap of these characteristics. As a supplemental analysis, we also present the association between all maternal, delivery, and child characteristics and HAZ. Because of the relatively high rate of nonattendance at the BSID-III clinic visit a total of 1943 of 2979 children selected for participation did not attend a clinic visit (65.2%) we also present a sensitivity analyses for maternal, delivery, and child characteristic analyses with the use of marginal structural models with stabilized inverse probability weights to account for censoring (21). P-trend in categorical analyses were calculated by treating the median value of each category as a continuous variable. Missing data were retained with the use of the missing indicator method. All P values were 2-sided with a P < 0.05 considered statistically significant. Statistical analyses were performed with the use of SAS v9.4. Ethics. The institutional review boards of the Harvard School of Public Health, Ifakara Health Institute, and National Medical Research Coordinating Council of Tanzania approved the parent trial and the extended follow-up study. The WHO Ethical Review Committee also approved the parent trial protocol. Results A total of 1036 caregiver child pairs attended a clinic visit for assessment of child development at mo of age. Characteristics of children who attended the clinic visit for development assessment compared with children who did not attend are presented in Supplemental Table 1. Mother infant pairs who participated in the study tended to have slightly lower maternal education and lower socioeconomic status, and were more likely preterm compared with nonparticipants. Despite statistical significance, these differences were small, which indicates a low risk of substantial nonresponse bias. Nevertheless, we examined the potential impact of censoring on estimates in a sensitivity analysis with marginal structural models. Pregnancy, delivery, and child characteristics of the study population are presented in Table 1. The mean 6 SD maternal age was y, most mothers completed primary school (82.0%), and the mean 6 SD TABLE 1 Characteristics of participants assessed with BSID-III at mo of age 1 Characteristics Values Maternal characteristics Maternal age at birth, y Maternal education None/some primary 110 (10.8) Completed primary 831 (82.0) Secondary or more 72 (7.1) Parity First birth 276 (30.0) 2nd 4th birth 497 (54.0) $5th birth 147 (16.0) Maternal height, cm Delivery/child characteristics Female 491 (47.4) Age at BSID-III assessment, mo Stimulation score LBW (,2500 g) 175 (16.9) Preterm (,37 wk) 95 (17.8) 2 SGA (INTERGROWTH,10%) 94 (17.6) 2 Improved water (93.9) Flush Toilet 262 (25.3) Randomly assigned to vitamin A 519 (50.1) Child anthropometry mo of age HAZ Stunted (HAZ,22) 375 (36.2) WHZ Wasted (WHZ,22) 15 (1.5) WAZ Underweight (WAZ,22) 92 (9.0) BSID-III Cognitive score Communication score Motor score Values are means 6 SDs or n (%); n = BSID-III, Bayley Scales of Infant Development III; HAZ, height-for-age z score; INTERGROWTH, International Fetal and Newborn Growth Consortium; LBW, low birth weight; SGA, small for gestational age; WAZ, weight-for-age z score; WHZ, weight-for-height z score. 2 In 533 infants with gestational age data. 3 Joint Monitoring Program definition. Malnutrition and development of Tanzanian children 2707

4 maternal height was cm. As for delivery characteristics, 16.9% of the child participants were LBW, 17.8% were preterm, and 17.6% were SGA. The mean HAZ of children at mo was , with 36.2% of children classified as stunted, the mean WHZ was , with 4.5% of infants exhibiting wasting, and the mean WAZ was , with 9.0% classified as underweight. Mean BSID-III raw scores by subscale at mo of age also are presented in Table 1. HAZ had a linear relation with all 3 developmental subscales (all P values for linear relation <0.05) (Figure 1). In contrast, the relation between WHZ and cognitive, communication, and motor development was nonlinear, with significant deficits seen at WHZ <21.5 (all P values for nonlinear relation <0.05) (Figure 2). Based on the shape of the WHZ relation and for comparability, we dichotomized WHZ with the use of the WHO definition of wasting (WHZ <22) in subsequent models. The WAZ had a significant linear relation with the 3 developmental domain scales (all P values for linear relation <0.05) (Figure 3). We also produced SMD estimates in BSID-III domain z scores for cross-sectional HAZ, wasting, and WAZ at mo of age (Table 2). Each unit increase in HAZ was associated with (95% CI: 0.05, 0.13), (95% CI: 0.07, 0.14), and (95% CI: 0.09, 0.16) higher cognitive, communication, and motor development z scores, respectively. Wasted children had (95% CI: 20.97, 20.29), (95% CI:20.64, 0.01), and (95% CI: 20.86, 20.23) z score deficits in cognitive, communication, and motor development, respectively. As for WAZ, each unit increase was associated with (95% CI: 0.03, 0.12), (95% CI: 0.06, 0.15), and (95% CI: 0.07, 0.15) greater cognitive, communication, and motor development z scores, respectively. Next, we examined the prospective association between maternal, delivery, and child factors during the first 1000 d of life and BSID-III z scores at mo of age. Associations between pregnancy characteristics and child development indicators are presented in Table 3. In a multivariate analysis, significant predictors of cognitive z scores were greater wealth (P-trend: 0.04) and decreased maternal stature (P-trend: 0.04). Greater wealth was the only maternal predictor significantly associated with communication z score (P-trend: 0.03). Decreased maternal stature was the only pregnancy factor significantly associated with child motor z score (P-trend < 0.01). The association between delivery and child characteristics and BSID- III z scores is presented in Table 4. Being born SGA was significantly associated with a (95% CI: 20.30, 20.01) z score deficit in cognitive development. Independent of wealth quintile, use of flush toilets during childhood was associated with significantly higher cognitive (P = 0.047) and motor z scores (P = 0.03). The associations between maternal, delivery, and child factors and HAZ at mo are presented in Supplemental Table 2. Socioeconomic status, maternal height, and SGA were significantly associated with HAZ. In sensitivity analyses that used marginal structural models with stabilized inverse probability weights to account for censoring, we found nearly identical point estimates for all maternal, delivery, and child factors (Supplemental Tables 3 and 4). Discussion In this study, we found a robust negative association between malnutrition and cognitive, communication, and motor development at mo, but found no evidence that these associations were limited to children with stunting. In spline analyses, HAZ and WAZ were linearly associated with all 3 FIGURE 1 Linear relations between HAZ and cognition (A), communication (B), and motor (C) BSID-III scores after multivariate adjustment for infant sex, infant age, maternal education, wealth quintile, stimulation tertile, BSID-III assessor, randomized regimen, and wasting; n = All P, 0.01 for linear relation. Graphs show z score prediction for girls aged mo, mother completed primary school, third wealth quintile, middle stimulation tertile, BSID-III assessor no. 2, received placebo, and nonwasted. BSID-III, Bayley Scales of Infant Development III; HAZ, height-for-age z score. developmental subscales, whereas WHZ exhibited a nonlinear relation with only children experiencing wasting (WHZ <22) having across-domain deficits. We also determined from a prospective analysis that known determinants of malnutrition, including wealth, maternal height, SGA, and flush toilet sanitation, were significant predictors of child development. Our findings that both HAZ and WAZ were significantly associated with cognitive, communication, and motor development 2708 Sudfeld et al.

5 and WAZ and developmental outcomes for cognitive, communication, and motor domains that extended to z scores >22 and even >0. Because of the low prevalence of acute malnutrition in the study population, it was expected that WAZ, an indicator of chronic and/or acute malnutrition, would parallel the HAZ findings. Our HAZ results suggest that children experiencing mild forms of linear growth faltering, not just those classified as stunted, may experience suboptimal developmental outcomes. The leading biological mechanism potentially explaining this FIGURE 2 Nonlinear relations between WHZ and cognition (A), communication (B), and motor (B) BSID-III scores after multivariate adjustment for infant sex, infant age, maternal education, wealth quintile, stimulation tertile, BSID-III assessor, randomized regimen, and HAZ; n = All P, 0.01 for nonlinear relation. Graphs show z score prediction for girls aged mo, mother completed primary school, third wealth quintile, middle stimulation tertile, BSID-III assessor no. 2, received placebo, and median value of HAZ. BSID-III, Bayley Scales of Infant Development III; HAZ, height-for-age z score; WHZ, weight-for-height z score. scores corroborates a large base of observational evidence that links chronic malnutrition with cognitive and motor development deficits (7). Nevertheless, few studies have examined the shape of the HAZ and WAZ relations with development. Our study determined a significantly linear relation between HAZ FIGURE 3 Linear relations between WAZ and cognition (A), communication (B), and motor (C) BSID-III scores after multivariate adjustment for infant sex, infant age, maternal education, wealth quintile, stimulation tertile, BSID-III assessor, and randomized regimen; n = All P, 0.01 for linear relation. Graphs show z score prediction for girls aged mo, mother completed primary school, third wealth quintile, middle stimulation tertile, BSID-III assessor no. 2, and received placebo. BSID-III, Bayley Scales of Infant Development III; WAZ, weight-for-age z score. Malnutrition and development of Tanzanian children 2709

6 TABLE 2 Cross-sectional associations between HAZ, wasting, and WAZ and BSID-III subscale scores at mo of age 1 Cognitive Communication Motor Age-adjusted P Multivariate-adjusted P Age-adjusted SMD P Multivariate-adjusted P Age-adjusted SMD P Multivariate-adjusted P Anthropometry HAZ (per unit increase) (0.06, 0.13), (0.05, 0.13), (0.08, 0.15), (0.07, 0.14), (0.08, 0.15), (0.09, 0.16),0.01 Wasting (WHZ,22) 2 Yes (20.84, 20.19), (20.97, 20.29), (20.67, 20.03) (20.64, 20.01) (20.97, 20.36), (20.86, 20.23),0.01 No Ref. Ref. Ref. Ref. Ref. WAZ (per unit increase) (0.04, 0.12), (0.03, 0.12), (0.07, 0.15), (0.06, 0.15), (0.07, 0.14), (0.07, 0.15), Values are standardized mean differences (95% CIs). BSID-III, Bayley Scales of Infant Development III; HAZ, height-for-age z score; Ref., reference; SMD, standardized mean difference; WAZ, weight-for-age z score; WHZ, weight-for-height z score. 2 Model includes HAZ, wasting (WHZ,22), stimulation category, infant sex, infant age, BSID-III assessor, baseline maternal education, baseline wealth quintile, and randomized regimen. 3 Model includes WAZ, stimulation category, infant sex, infant age, BSID-III assessor, baseline maternal education, baseline wealth quintile, and randomized regimen. relation is that early exposure to malnutrition and infection produces simultaneous growth faltering and deficits in neuronal growth, pruning, and connectivity within regions of the brain associated with motor functioning, memory, learning, and higher order cognition (22, 23). Children with malnutrition may also experience a reduced ability to engage their environments or a lack of responsiveness by caregivers, referred to as the functional isolation hypothesis (24). In this study, we were primarily interested in the biological effect of malnutrition on child development; as a result, we controlled for concurrent child stimulation to limit the impact of functional isolation on estimates. We determined in this cross-sectional analysis that mild growth faltering is associated with development deficits in Tanzanian children at mo, but studies examining the long-term relation between mild growth faltering in early childhood and development in late childhood and adolescence are needed. Long-term follow-up studies examining the relation between stunting or severe growth faltering and development have found mixed results (25, 26). In addition, very few children in our study population had HAZ >1, and additional studies are needed to determine whether a linear relation extends to HAZ >1. Some studies in well-nourished populations have found small or no association between HAZ and development, which may be because a significant proportion of participants have HAZ >1, a section of the HAZ distribution in which the magnitude of association may be reduced (27, 28). Two studies have examined the independent association between wasting or WHZ and child development after adjustment for HAZ or stunting (29, 30). Similar to our findings, a study of HIV-infected infants in Tanzania found that wasted infants had a psychomotor development deficit of z scores and a z score deficit in mental development after multivariate adjustment that included HAZ (29). Nevertheless, our wasting results need to be interpreted with caution, because only 15 infants experienced wasting. It is also unclear whether the cross-sectional relation between wasting and developmental deficit is a result of a true development deficit or if these infants did not have the energy or concentration to complete developmental assessment tests as a result of their acute malnutrition. A prospective study by Berkman et al. (30) found no association between weight-forlength z score or wasting (weight-for-length z score <22) during infancy and Wechsler Intelligence Scale for Children Revised scores at 9 y of age in Peruvian children. Additional evidence is needed to determine whether there are long-term developmental deficits associated with wasting during childhood, as well as whether treatment of severe and moderate acute malnutrition improves child development. We also determined that maternal, delivery, and child factors that are classically associated with linear growth faltering and wasting are also associated with child development (8, 11, 12, 30). We noted that SGA infants had significantly reduced cognitive scores. Multiple observational studies in settings in developed countries have reported that SGA infants have lower cognitive and schooling achievement compared with non-sga infants, but evidence from developing countries is more limited (31, 32). In a study of Guatemalan infants, SGA infants had decreased developmental scores compared with full-term normal-weight infants at 24 mo and 3 y of age (33). Furthermore, one study of term LBW infants and another examining LBW infants with adjustment for gestational age also support our finding that 2710 Sudfeld et al.

7 TABLE 3 Prospective associations between maternal characteristics and BSID-III subscale scores at mo of age 1 Cognitive Communication Motor n Age-adjusted P Multivariate-adjusted 2 P Age-adjusted P Multivariate-adjusted 2 P Age-adjusted P Multivariate-adjusted 2 P Maternal age,20 y (20.12, 0.09) (20.12, 0.11) (20.10, 0.10) (20.10, 0.11) (20.09, 0.11) (20.10, 0.11) y 632 Ref. Ref. Ref. Ref. Ref. Ref..30 y (20.07, 0.12) (20.05, 0.16) (20.06, 0.12) (20.03, 0.16) (20.05, 0.13) (20.04, 0.15) 0.29 Maternal education None/some primary (20.10, 0.15) (20.03, 0.23) (20.26, 0.00) (20.16, 0.09) (20.13, 0.11) (20.05, 0.20) 0.25 Completed primary 831 Ref. Ref. Ref. Ref. Ref. Ref. Secondary (20.11, 0.18) (20.13, 0.17) (20.13, 0.16) (20.17, 0.12) (20.22, 0.07) (20.22, 0.06) 0.27 Missing (20.31, 0.21) (20.34, 0.18) (20.49, 0.01) (20.52, 0.03) (20.25, 0.24) (20.26, 0.22) 0.87 Wealth quintile (poorest) 142 Ref. Ref. Ref. Ref. Ref. Ref (20.17, 0.11) 0.00 (20.13, 0.14) 0.02 (20.12, 0.15) 0.04 (20.09, 0.17) 0.01 (20.12, 0.14) 0.03 (20.10, 0.16) (0.02, 0.30) 0.18 (20.04, 0.32) 0.16 (0.02, 0.29) 0.15 (0.02, 0.29) 0.11 (20.03, 0.24) 0.12 (20.02, 0.25) (20.07, 0.19) 0.09 (20.04, 0.22) 0.10 (20.03, 0.23) 0.10 (20.03, 0.22) 0.07 (20.05, 0.20) 0.09 (20.03, 0.21) 5 (richest) (20.02, 0.23) 0.12 (20.01, 0.25) 0.16 (0.04, 0.28) 0.10 (20.03, 0.22) 0.08 (20.04, 0.20) 0.09 (20.04, 0.21) Missing (20.11, 0.37) 0.07 (20.18, 0.31) 0.17 (20.07, 0.40) 0.05 (20.19, 0.28) 0.10 (20.13, 0.33) 0.01 (20.22, 0.25) Maternal height , , ,150 cm (20.20, 20.01) (20.19, 20.01) (20.19, 20.01) (20.18, 0.00) (20.23, 20.05) (20.22, 20.04) cm (20.12, 0.06) (20.12, 0.05) (20.09, 0.08) (20.09, 0.08) (20.19, 20.02) (20.19, 20.02).155 cm 416 Ref. Ref. Ref. Ref. Ref. Ref. Parity First birth (20.11, 0.08) (20.11, 0.10) (20.09, 0.09) (20.12, 0.08) (20.10, 0.08) (20.10, 0.10) nd 4th birth 497 Ref. Ref. Ref. Ref. Ref. Ref. $5th birth (20.15, 0.08) (20.16, 0.08) (20.14, 0.09) (20.14, 0.09) (20.10, 0.12) (20.13, 0.10) 0.83 Missing (20.13, 0.12) (20.11, 0.10) (20.18, 0.07) (20.13, 0.13) (20.17, 0.07) (20.13, 0.12) Values are standardized mean differences (95% CIs); n = BSID-III, Bayley Scales of Infant Development III; Ref. reference. 2 Adjusted for infant age, infant sex, BSID-III assessor, randomized regimen, and all characteristics included in table. 3 P-trend. Malnutrition and development of Tanzanian children 2711

8 TABLE 4 Prospective associations between delivery and child characteristics and BSID-III subscale scores at mo of age 1 Cognitive Communication Motor n Age-adjusted P Multivariate-adjusted 2 P Age-adjusted P Multivariate-adjusted 2 P Age-adjusted P Multivariate-adjusted 2 P Sex M (20.08, 0.17) (20.09, 0.06) (20.14, 0.11) (20.14, 0.01) (20.06, 0.19) (20.07, 0.07) 0.91 F 491 Ref. Ref. Ref. Ref. Ref. Ref. LBW,2500 g (20.17, 0.04) (20.18, 0.03) (20.16, 0.04) (20.17, 0.04) (20.17, 0.03) (20.19, 0.01) 0.09 $2500 g 861 Ref. Ref. Ref. Ref. Ref. Ref. Gestational age Preterm (,37 wk) (20.14, 0.14) (20.13, 0.14) (20.20, 0.07) (20.17, 0.10) (20.12, 0.14) (20.10, 0.16) 0.63 Term ($37 wk) 438 Ref. Ref. Ref. Ref. Ref. Ref. Missing (20.08, 0.10) (20.17, 0.04) (20.20, 0.07) (20.12, 0.08) (20.08, 0.10) (20.16, 0.04) 0.21 Size for gestational age SGA (,10th percentile) (20.24, 0.05) (20.30, 20.01) (20.14, 0.14) (20.17, 0.11) (20.13, 0.14) (20.19, 0.08) 0.44 AGA ($10th percentile) 439 Ref. Ref. Ref. Ref. Ref. Ref. Missing (20.10, 0.09) (20.17, 0.04) (20.03, 0.15) (20.12, 0.08) (20.13, 0.14) (20.16, 0.04) 0.21 Sanitation Flush toilet , 0.20) (0.00, 0.19) (0.00, 0.17) (20.03, 0.15) (0.01, 0.17) (0.01, 0.19) 0.03 Nonflush toilet 774 Ref. Ref. Ref. Ref. Ref. Ref. Water quality Improved (0.01, 0.34) (20.03, 0.30) (0.05, 0.38) (20.02, 0.30) (20.08, 0.23) (20.13, 0.18) 0.77 Unimproved 65 Ref. Ref. Ref. Ref. Ref. Ref. 1 Values are standardized mean differences (95% CIs); n = AGA, appropriate for gestational age; BSID-III, Bayley Scales of Infant Development III; LBW, low birth weight; Ref., reference; SGA, small for gestational age. 2 Adjusted for infant sex, infant age, BSID-III assessor, randomized regimen, maternal age at delivery, maternal education, wealth quintile, maternal height, parity, and all characteristics included in table. Preterm and SGA were included in models together or with LBW alone. Estimates for sex, sanitation, and water quality include preterm and SGA adjustment Sudfeld et al.

9 intrauterine growth restriction is associated with child development deficits (34, 35). A limitation of our analysis is that preterm and SGA were defined by maternal report of LMP at entry into the trial, which likely led to some misclassification. Nevertheless, errors in maternal report of LMP at trial enrollment likely are not systematically related to child development scores and, as a result, would lead us to underestimate the association between SGA and preterm and developmental outcomes. We also determined that shorter maternal height and nontoilet sanitation were associated with cognitive and motor development deficits. To our knowledge, this is the first study to independently link maternal stature to cognitive and motor development during early childhood. Similarly, studies have found that diarrhea during infancy is associated with developmental setbacks (36 38), but the effect of water, sanitation, and hygiene differences on child development has not been noted previously. Nevertheless, because of the potential of residual confounding due to the strong link between improved sanitation and greater socioeconomic status, we may be overestimating the association between flush toilet access and child development. As a result, water, sanitation, and hygiene intervention trials including long-term child development follow-up are needed. Overall, chronic and acute malnutrition independently are associated with cognitive deficits, communication, and motor development in Tanzanian infants. HAZ exhibited a linear relation with child development, with incremental improvements in developmental scores for HAZ >22 and even >0. As a result, the most recent global estimate that 219 million children did not reach their developmental potential in 2004, which used stunting as a proxy for developmental delay, may be a significant underestimate (2). Millions of children with milder forms of linear growth faltering also may not be fulfilling their development potential. A multitude of risk factors and effective interventions that improve linear growth and prevent wasting are known (20), but for the majority of these, the impact on child development remains understudied. A few trials of energy supplementation have found improved cognitive development for undernourished children or those at high risk of malnutrition, but other trials have found no effect (39, 40). Existing evidence suggests that enhancements in the areas of parent training, cognitive stimulation, and violence prevention are needed to strengthen the impact of nutritional interventions on child development (41). Furthermore, our finding that pregnancy, delivery, and child factors were associated with both malnutrition and child development indicate the need to examine the effect of approaches that span preconception through early childhood. Randomized controlled trials of high-impact and comprehensive approaches and packages to reduce malnutrition and promote child development are urgently needed. Acknowledgments AM, HM, ERS, and WWF conducted the parent trial; CRS, DCM, GF, AM, DCB, GD, ME, and WWF designed and implemented the follow-up study; and CRS conducted the statistical analysis and drafted the manuscript. All authors read and approved the final manuscript. References 1. Grantham-McGregor S, Cheung YB, Cueto S, Glewwe P, Richter L, Strupp B, International Child Development Steering Group. Developmental potential in the first 5 years for children in developing countries. Lancet 2007;369: Committee on Integrating the Science of Child Development. From neurons to neighborhoods: the science of child development. Washington (DC): National Academy Press; Wachs TD. Necessary but not sufficient: the respective roles of single and multiple influences on individual development. Washington (DC): American Psychological Association; Brooks-Gunn J, Duncan GJ. The effects of poverty on children. Future Child 1997;7: Adair LS, Fall CH, Osmond C, Stein AD, Martorell R, Ramirez-Zea M, Sachdev HS, Dahly DL, Bas I, Norris SA, et al, COHORTS group. Associations of linear growth and relative weight gain during early life with adult health and human capital in countries of low and middle income: findings from five birth cohort studies. Lancet 2013;382: UN General Assembly [Internet]. Open Working Group on Sustainable Development Goals [cited 2002 Jul 9]. Available from: sustainabledevelopment.un.org/topics/sustainabledevelopmentgoals. 7. Sudfeld CR, Charles McCoy D, Danaei G, Fink G, Ezzati M, Andrews KG, Fawzi WW. Linear growth and child development in low- and middle-income countries: A meta-analysis. Pediatrics 2015;135:e Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, de Onis M, Ezzati M, Grantham-McGregor S, Katz J, Martorell R, et al, Maternal and Child Nutrition Study Group. Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet 2013;382: Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cognitive and behavioral outcomes of school-aged children who were born preterm: A meta-analysis. JAMA 2002;288: Moster D, Lie RT, Markestad T. Long-term medical and social consequences of preterm birth. N Engl J Med 2008;359: Ozaltin E, Hill K, Subramanian SV. Association of maternal stature with offspring mortality, underweight, and stunting in low- to middle-income countries. JAMA 2010;303: Fink G, Günther I, Hill K. The effect of water and sanitation on child health: evidence from the demographic and health surveys Int J Epidemiol 2011;40: Masanja H, Smith ER, Muhihi A, Briegleb C, Mshamu S, Ruben J, Noor RA, Khudyakov P, Yoshida S, Martines J, Bahl R, Fawzi WW. Effect of neonatal vitamin A supplementation on mortality in infants in Tanzania (Neovita): a randomised, double-blind, placebo-controlled trial. Lancet 2015;385: Bahl R, Bhandari N, Dube B, Edmond K, Fawzi W, Fontaine O, Kaur J, Kirkwood BR, Martines J, Masanja H, Mazumder S, et al. Efficacy of early neonatal vitamin A supplementation in reducing mortality during infancy in Ghana, India and Tanzania: study protocol for a randomized controlled trial. Trials 2012;13: Bayley N. Bayley Scales of Infant and Toddler Development. San Antonio (TX): The Psychological Corporation; WHO Multicentre Growth Reference Study Group. WHO child growth standards: length/height-for-age, weight-for-age, weight-forlength, weight-for-height and body mass index-for-age: methods and development. Geneva (Switzerland): WHO; UNICEF. The formative years: UNICEFÕs work on measuring early childhood development. New York: UNICEF; Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989;8: Villar J, Papageorghiou AT, Pang R, Ohuma EO, Cheikh Ismail L, Barros FC, Lambert A, Carvalho M, Jaffer YA, Bertino E, et al. International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st). The likeness of fetal growth and newborn size across non-isolated populations in the INTERGROWTH- 21st Project: the Fetal Growth Longitudinal Study and Newborn Cross- Sectional Study. Lancet Diabetes Endocrinol 2014;2: World Health Organization. Progress on sanitation and drinking water: 2010 update. Geneva (Switzerland): WHO; Robins JM, Hernán MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology 2000;11: Prado EL, Dewey KG. Nutrition and brain development in early life. Nutr Rev 2014;72: Fish I, Winick M. Effect of malnutrition on regional growth of the developing rat brain. Exp Neurol 1969;25: Levitsky DA, Barnes RH. Nutritional and environmental interactions in the behavioral development of the rat: long-term effects. Science 1972;176: Malnutrition and development of Tanzanian children 2713

10 25. Crookston BT, Forste R, McClellan C, Georgiadis A, Heaton TB. Factors associated with cognitive achievement in late childhood and adolescence: the Young Lives cohort study of children in Ethiopia, India, Peru, and Vietnam. BMC Pediatr 2014;14: Casale D, Desmond C, Richter L. The association between stunting and psychosocial development among preschool children: a study using the South African Birth to Twenty cohort data. Child Care Health Dev 2014;40: Huang C, Martorell R, Ren A, Li Z. Cognition and behavioural development in early childhood: the role of birth weight and postnatal growth. Int J Epidemiol 2013;42: Kordas K, Lopez P, Rosado JL, García Vargas G, Alatorre Rico J, Ronquillo D, Cebrián ME, Stoltzfus RJ. Blood lead, anemia, and short stature are independently associated with cognitive performance in Mexican school children. J Nutr 2004;134: McDonald CM, Manji KP, Kupka R, Bellinger DC, Spiegelman D, Kisenge R, Msamanga G, Fawzi WW, Duggan CP. Stunting and wasting are associated with poorer psychomotor and mental development in HIV-exposed Tanzanian infants. J Nutr 2013;143: Berkman DS, Lescano AG, Gilman RH, Lopez SL, Black MM. Effects of stunting, diarrhoeal disease, and parasitic infection during infancy on cognition in late childhood: a follow-up study. Lancet 2002;359: Levine TA, Grunau RE, McAuliffe FM, Pinnamaneni R, Foran A, Alderdice FA. Early childhood neurodevelopment after intrauterine growth restriction: a systematic review. Pediatrics 2015;135: de Bie HM, Oostrom KJ, Delemarre-van de Waal HA. Brain development, intelligence and cognitive outcome in children born small for gestational age. Horm Res Paediatr 2010;73: Villar J, Smeriglio V, Martorell R, Brown CH, Klein RE. Heterogeneous growth and mental development of intrauterine growth-retarded infants during the first 3 years of life. Pediatrics 1984;74: Grantham-McGregor SM. Small for gestational age, term babies, in the first six years of life. Eur J Clin Nutr 1998;52 Suppl 1:S Liu X, Sun Z, Neiderhiser JM, Uchiyama M, Okawa M. Low birth weight, developmental milestones, and behavioral problems in Chinese children and adolescents. Psychiatry Res 2001;101: Niehaus MD, Moore SR, Patrick PD, Derr LL, Lorntz B, Lima AA, Guerrant RL. Early childhood diarrhea is associated with diminished cognitive function 4 to 7 years later in children in a northeast Brazilian shantytown. Am J Trop Med Hyg 2002;66: Petri WA, Jr., Miller M, Binder HJ, Levine MM, Dillingham R, Guerrant RL. Enteric infections, diarrhea, and their impact on function and development. J Clin Invest 2008;118: Ngure FM, Reid BM, Humphrey JH, Mbuya MN, Pelto G, Stoltzfus RJ. Water, sanitation, and hygiene (WASH), environmental enteropathy, nutrition, and early child development: making the links. Ann N Y Acad Sci 2014;1308: Waber DP, Vuori-Christiansen L, Ortiz N, Clement JR, Christiansen NE, Mora JO, Reed RB, Herrera MG. Nutritional supplementation, maternal education, and cognitive development of infants at risk of malnutrition. Am J Clin Nutr 1981;34 Suppl 4: Grantham-McGregor S, Baker-Henningham H. Review of the evidence linking protein and energy to mental development. Public Health Nutr 2005;8: Yousafzai AK, Aboud F. Review of implementation processes for integrated nutrition and psychosocial stimulation interventions. Ann N Y Acad Sci 2014;1308: Sudfeld et al.