Introduction
The micronutrient iron is notable for its role in the development and function of all body tissues and is especially important for brain development. Iron deficiency is one of the most common forms of malnutrition, affecting an estimated 4 in 10 children under 5 years of age globally (Stevens et al., Reference Stevens, Finucane, De-Regil, Paciorek, Flaxman, Branca, Pena-Rosas, Bhutta, Ezzati and Nutrition Impact Model Study2013) and 15.1% of toddlers in the United States (Gupta et al., Reference Gupta, Perrine, Mei and Scanlon2017). Developmentally, infants are especially at risk of becoming iron-deficient (ID) at approximately 6–12 months of age, when prenatal iron stores become depleted (Georgieff, Reference Georgieff2017). During the critical 6–24-month postnatal window of rapid brain development, iron deficiency is particularly damaging, as several areas of the brain require iron for normal development (Cusick & Georgieff, Reference Cusick and Georgieff2016). The high demand for iron in infancy coincides with the period of rapid growth and development of brain structure and functions that require iron, including the hippocampus, cortical regions, neuronal and glial energy metabolism, myelin synthesis, and neurotransmission (Lozoff & Georgieff, Reference Lozoff and Georgieff2006). Iron is also essential for serotonin, norepinephrine, and dopamine neurotransmitter synthesis (Lozoff & Georgieff, Reference Lozoff and Georgieff2006). Animal studies show associations between altered brain metabolism, myelination (Beard et al., Reference Beard, Wiesinger and Connor2003; Kwik-Uribe et al., Reference Kwik-Uribe, Gietzen, German, Golub and Keen2000; Oloyede et al., Reference Oloyede, Folayan and Odutuga1992; Yu et al., Reference Yu, Steinkirchner, Rao and Larkin1986), and neurotransmitter function (Lozoff, Beard, et al., Reference Lozoff, Beard, Connor, Barbara, Georgieff and Schallert2006) and early-life iron deficiency. Early ID also is associated with alterations to the developing hippocampus (de Deungria et al., Reference de Deungria, Rao, Wobken, Luciana, Nelson and Georgieff2000), with pervasive and long-lasting iron deficiency-induced metabolic (Rao et al., Reference Rao, Tkac, Townsend, Gruetter and Georgieff2003) and dendritic structure changes (Jorgenson et al., Reference Jorgenson, Wobken and Georgieff2003). Neurophysiologic studies of the effects of iron deficiency have found differences in the speed of neural transmission in the auditory system (Li et al., Reference Li, Wang and Wang1994; Roncagliolo et al., Reference Roncagliolo, Garrido, Walter, Peirano and Lozoff1998), recognition memory (Burden et al., Reference Burden, Westerlund, Armony-Sivan, Nelson, Jacobson, Lozoff, Angelilli and Jacobson2007; Siddappa et al., Reference Siddappa, Georgieff, Wewerka, Worwa, Nelson and Deregnier2004), longer auditory brainstem response, and longer visual evoked potentials latencies (Algarin et al., Reference Algarin, Peirano, Garrido, Pizarro and Lozoff2003). Infants at high risk for ID show poorer recognition memory, possibly due to iron’s effects on the hippocampus and central nervous system (Nelson et al., Reference Nelson, Wewerka, Thomas, Tribby-Walbridge, deRegnier and Georgieff2000). Iron deficiency during infancy is associated with children’s socioemotional and behavioral problems and lower cognitive abilities (Georgieff, Reference Georgieff2011; Lozoff, Beard, et al., Reference Lozoff, Beard, Connor, Barbara, Georgieff and Schallert2006; Pivina et al., Reference Pivina, Semenova, Dosa, Dauletyarova and Bjorklund2019). Thus, inadequate iron can negatively impact neurodevelopment across several domains and in different brain regions (Beard & Connor, Reference Beard and Connor2003; Beard et al., Reference Beard, Felt, Schallert, Burhans, Connor and Georgieff2006; Felt et al., Reference Felt, Beard, Schallert, Shao, Aldridge, Connor, Georgieff and Lozoff2006).
Iron deficiency may also be more likely to occur in the context of early-life adversity (ELA) (Walker et al., Reference Walker, Wachs, Grantham-McGregor, Black, Nelson, Huffman, Baker-Henningham, Chang, Hamadani, Lozoff, Gardner, Powell, Rahman and Richter2011). ELA can include exposure to poverty, a harsh family environment, parental separation, caregiver mental illness, major illness, death of a family member, or other forms of psychosocial adversity (Lupien et al., Reference Lupien, McEwen, Gunnar and Heim2009). ELA is also detrimental to optimal neurodevelopment, placing children at risk for psychopathology later in life (Lupien et al., Reference Lupien, McEwen, Gunnar and Heim2009).
There are biological and sociological reasons for their co-occurrence (Grantham-McGregor & Ani, Reference Grantham-McGregor and Ani2001). Sociologically, children experiencing ELA in the form of poverty may be at higher risk of iron deficiency due to food insecurity or diets low in iron (e.g., Skalicky et al., Reference Skalicky, Meyers, Adams, Yang, Cook and Frank2006). Biologically, conditions of ELA may co-occur with conditions that increase an infant’s exposure to infection and inflammation that impact iron status. Hepcidin, a hormone that responds to body iron status and inflammation to regulate intestinal absorption and tissue distribution, mediates iron absorption and sequestration. Hepcidin expression is modulated during infection and inflammation to decrease iron availability to invading pathogens. Iron supply for red blood cell precursors is also restricted during inflammation and infection (Ganz & Nemeth, Reference Ganz and Nemeth2009).
Adversity might also put a child at risk for developing iron deficiency due to physiological processes arising from ELA. The stress of ELA can contribute to dysregulated neuroendocrine pathways that can disrupt nutrient absorption and utilization even in the context of adequate nutrient intake (Monk et al., Reference Monk, Georgieff and Osterholm2013; Osterholm & Georgieff, Reference Osterholm and Georgieff2015; Suchdev et al., 2017). Chronic disruptions to the stress response shape neurobiology and physiology during sensitive periods of development, leading to alterations in stress and immune system activity (Nusslock & Miller, Reference Nusslock and Miller2016). The hypothalamic-pituitary-adrenocortical (HPA) plays a significant role in how an organism responds to stress and how stressful experiences are biologically embedded (Gunnar et al., Reference Gunnar, Doom, Esposito and Lerner2015; Gunnar & Vazquez, Reference Gunnar, Vazquez, Cicchetti and Cohen2006). Early-life exposure to stress and increased exposure to cortisol or its releasing hormone, corticotropic-releasing hormone, is thought to program the developing HPA axis and brain (Korosi & Baram, Reference Korosi and Baram2008). Disruptions to stress-mediating systems that arise from ELA can produce multi-system reactions (Danese & Baldwin, Reference Danese and Baldwin2017). These include elevated cytokine levels, glucocorticoid resistance, and dysregulation of the HPA axis, among other stress-mediating systems (Miller et al., Reference Miller, Chen and Parker2011). Psychological stress can negatively impact nutritional processes such as iron absorption, synthesis, and availability (Monk et al., Reference Monk, Georgieff and Osterholm2013). Animal studies show evidence for changes in iron metabolism following stress exposure. Adult rodents exposed to psychological stress exhibit decreases in serum iron, hemoglobin, ferritin, and erythropoietin (Wei et al., Reference Wei, Zhou, Huang and Li2008). In another study, exposure to acute and chronic stressors in rodents reduced whole blood iron concentration (Teng et al., Reference Teng, Sun, Shi, Hou and Liu2008). Additionally, research from non-human primates shows that experimental stress to the pregnant monkey produced compromised iron status in the infant (Coe et al., Reference Coe, Lubach and Shirtcliff2007).
Few studies in humans examine the effects of stress on iron status. One experiment in Navy SEAL trainees found that a week of psychological stress resulted in an acute disruption to iron status (Singh et al., Reference Singh, Smoak, Patterson, LeMay, Veillon and Deuster1991). From a population perspective, more impoverished countries have higher rates of iron deficiency (Black et al., Reference Black, Walker, Fernald, Andersen, DiGirolamo, Lu, McCoy, Fink, Shawar, Shiffman, Devercelli, Wodon, Vargas-Barón and Grantham-McGregor2017). However, studies assessing the association between ELA and iron status are primarily limited to adversity exposure in pregnancy. In studies of pregnant mothers, mothers exposed to higher levels of objective stressors and mothers who reported higher levels of stress in pregnancy were more likely to have offspring with worse iron status, including lower cord blood ferritin (Armony-Sivan et al., Reference Armony-Sivan, Aviner, Cojocaru, Fytlovitch, Ben-Alon, Eliassy, Babkoff, Lozoff and Anteby2013; Campbell et al., Reference Campbell, Tamayo-Ortiz, Cantoral, Schnaas, Osorio-Valencia, Wright, Tellez-Rojo and Wright2020) and a higher cord blood zinc protoporphyrin/heme ratio (McLimore et al., Reference McLimore, Phillips, Blohowiak, Pham, Coe, Fischer and Kling2013). Rendina and colleagues (Rendina et al., Reference Rendina, Blohowiak, Coe and Kling2018) also found that pregnant women with higher stress levels had 1-year-old infants with an increased risk of low plasma ferritin.
Currently, it is unknown how postnatal ELA impacts iron status in infants. The current study examines the extent to which cumulative adversity experienced during the first year of life relates to infants' iron status. Our outcomes of interest included infants' iron status at 12 months of age (i.e., iron deficiency with or without anemia) and hematological markers of poor iron status (ferritin, hemoglobin, mean corpuscular volume (MCV), free erythrocyte protoporphyrin (FEP)). All infants studied here were randomized to the control (no-added iron) arm of an iron deficiency anemia preventive trial from 6 to 12 months (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003). We hypothesize that higher levels of adversity in the first year of life will be associated with poor iron status at 12 months and individual hematological markers of poor iron status at 12 months of age.
Method
Participants
The current sample was part of a Chilean cohort that participated in a randomized controlled trial in infancy to prevent iron deficiency anemia (IDA; see descriptive statistics in Table 1). The study is fully described (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003). Infants were recruited from 1991–1996 at community clinics in four adjacent working-class communities in Santiago, Chile. In these communities, infants were primarily fed breastmilk and powdered milk (“Leche Purita”) that was not iron-fortified. Screening infants for anemia was not a regular part of pediatric care, and routine iron supplementation was not the policy in Chile at the time of the study (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003). Infants randomized (double-blind) into one of three iron supplementation groups: (1) iron-fortified formula (12.7 Fe mg/L, provided by Abbott-Ross Laboratories) or vitamins with iron (15 mg/day) if primarily breastfed, (2) low-iron formula (2.3 Fe mg/L, provided by Abbott-Ross Laboratories), and (3) the control group of no-added iron formula or vitamins without iron if primarily breastfed. Infants received the intervention from 6 to 12 months of age.
Participant characteristics
Note. *Higher scores reflect greater adversity.
Inclusion criteria in the trial were birth weight >3.0 kg, singleton term birth, vaginal delivery, stable caregiver, and residence in the target communities. Exclusion criteria were a major congenital anomaly, birth complications, phototherapy, hospitalization longer than 5 days, illness, or iron therapy, another infant less than 12 months of age in the household, daycare for the infant, and a caregiver who was illiterate or psychotic, which was self-reported or reported by family members on a recruitment questionnaire. Until mid-1994, exclusive breastfeeding (<250 mL cow milk or formula/day) was also an exclusion criterion. However, given secular increases in breastfeeding, the study was modified to enroll qualifying infants even if they had not started any bottle feeding. Between 1994 and 1996, to increase the size of the no-added iron group, infants who were consuming ≥250 mL/day cow milk or formula were randomly assigned in a 1-to-3 ratio to either (a) high-iron formula or (b) unmodified cow milk plus multivitamins without iron. Infants who were taking <250 mL/day of formula (defined as “exclusively breastfed”) were randomly assigned in a 1-to-2-ratio to liquid multivitamin preparation that (a) did contain iron or (b) did not contain iron (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003). The no-added iron group consisted entirely of infants recruited after 1994.
Infants received capillary hemoglobin (Hb) screening at 6 months. Infants with low Hb (≤103 g/L) were further assessed by venipuncture. IDA at 6 months was defined as venous Hb ≤ 100 g/L and 2 of 3 iron measures in the deficient range: MCV <70 fL, FEP >100 μg/dL red blood cells, and ferritin <12 μg/L. At 6 months, infants with iron deficiency anemia (IDA) were excluded from the preventive trial and given medicinal iron (n = 73 of 2,027 screened). All other (nonanemic) infants were invited to participate in the preventive trial.
The current study analyzed only those infants who were not anemic at 6 months to meet trial enrollment criteria, completed the trial, had iron status measured at 12 months, and were randomized to the control group, which received no iron supplementation from the study (n = 534). We selected only the no-added iron control group to understand how ELA is associated with iron status without supplementation. There were no differences between those who did or did not complete the study by group assignment, infant characteristics (birth weight, gestational age, sex, growth, and temperament [described below]), or family characteristics (household size, father absence, parental education, maternal depressed mood, and child development support provided in the home) (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003). Table 1 displays the descriptive characteristics of the sample. The Institutional Review Boards at the relevant institutions in the US and Chile approved the study. The study’s procedures were consistent with the Declaration of Helsinki. Additional details of the supplementation study have been previously published (Lozoff et al., Reference Lozoff, Wolf and Jimenez1996).
Measures
Iron status
At 12 months, all participants provided a venous blood sample to determine iron status based on Hb (Hb, g/L), MCV (fL), FEP, and serum ferritin (μg/L). The focal iron status variable of interest was iron status, defined as iron sufficient (IS), ID without anemia, or iron deficiency with anemia. For the categorical iron status assessment, infants were ID if they had 2 of 3 iron measures in the abnormal range (described in the Participants section above) (Oski, Reference Oski1993). Infants were categorized as having IDA with 2 of 3 iron measures in the abnormal range and Hb < 110 g/L. Infants without ID or IDA were categorized as IS. Iron status was coded as an ordinal variable, with IS coded as 1, ID without anemia coded as 2, and iron-deficient anemic coded as 3. We also analyzed continuous measures of Hb, MCV, FEP (FEP was reverse-scored and log-transformed), and ferritin (ferritin was log-transformed due to right-skew) in a latent variable with structural equation modeling (SEM) (detailed below). All iron measures are reported in Table 1.
ELA
ELA was a composite variable created by the sum of three indices (further explained below) of ELA in infancy: (1) stressful life events that occurred during the infants' first year, (2) frequent maternal depressive symptoms, and (3) low support for child development in the home. When their infants were 6–12 months old, mothers completed self-report questionnaires to assess these various adversities. All measures were extensively pilot tested in Chile before conducting the study. Spanish versions were used for all measures and found to have good reliability and high equivalence to the English versions (Wu et al., Reference Wu, East, Delker, Blanco, Caballero, Delva, Lozoff and Gahagan2019). A native Spanish speaker translated measures into Spanish before a Chilean psychologist back-translated the measures to verify comparability with the English version (Wu et al., Reference Wu, East, Delker, Blanco, Caballero, Delva, Lozoff and Gahagan2019). The original translations were also adjusted to accommodate subtle regional differences in Chilean Spanish (Ceballo et al., Reference Ceballo, Ramirez, Castillo, Caballero and Lozoff2016). The sum of each separate ELA index was standardized (z-score) and then summed to form an overall ELA index, with higher scores indicating higher levels of early-life adversity (Table 1).
Stressful life events
Mothers reported stressful life events with a modified Social Readjustment Rating Scale (Holmes & Rahe, Reference Holmes and Rahe1967) when the target infant was 11 months old, capturing stressful life events in the last year. The measure was a maternal self-report of 30 possible stressors (e.g., chronic illness of a family member, marital separation, financial instability, etc.) occurring during the last year. Items were coded “0” if the event did not happen and “1” if the event did happen, with scores summed across the 30 items (range: 0–30).
Maternal depressive symptoms
Maternal depressive symptoms were assessed via the Center for Epidemiological Studies Depression scale (CES-D) (Radloff, Reference Radloff2016) when the infant was 7 months old. The CES-D has been widely used in cross-cultural research and has demonstrated reliability and validity across ethnic groups within the US and internationally (Naughton & Wiklund, Reference Naughton and Wiklund1993; Roberts et al., Reference Roberts, Rhoades and Vernon1990). Research-trained psychologists administered the scale to mothers via private interview. The 20-item scale asks about the frequency of depressive symptoms within the past 3 months, with response options ranging from “rarely or none of the time” (coded as 0) to “most or all the time” (coded as 3). Cronbach’s alpha of the CES-D items at infancy was .85 (Wu et al., Reference Wu, East, Delker, Blanco, Caballero, Delva, Lozoff and Gahagan2019). Scores on the CES-D reflect the frequency of maternal depressive symptoms in the last 3 months. We used the raw sum score of maternal depressive symptoms in this analysis, with higher scores indicating more frequent depressive symptoms (range: 0–60).
Support for child development in the home
We assessed support for child development with the Home Observation for Measurement of the Environment Inventory (HOME), which evaluates the quality of stimulation and support available to a child in their home environment (Caldwell & Bradley, Reference Caldwell and Bradley1984). HOME is well-established and has been used in several Latin American countries (Bradley & Corwyn, Reference Bradley and Corwyn2016). Scores from the HOME in studies conducted in Chile are comparable to studies conducted in the United States (Bradley & Corwyn, Reference Bradley and Corwyn2016; Bulnes et al., Reference Bulnes, Cajdler, Edwards and Lira1979; Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003). The HOME assessment was conducted by a trained researcher through home observation when the infant was 9 months old. The HOME measures support for child development, including variety in daily stimulation, provision of play materials, organization of the environment, and the parent’s responsivity and involvement with the child. The HOME score was reverse-scored, such that higher scores indicate less support for child development.
Covariates
Covariates determined a priori related to study outcomes include gestational age, birth weight (obtained from hospital records), and weight from 6 to 12 months (measured monthly on an electronic scale to the nearest 10 g by the study team). Researchers recorded the amount of formula or cow milk that the infant ingested at weekly home visits from 6 to 12 months. To control for differences in feeding behavior (formula vs. breastfeeding), we used the average daily intake (mL/day) of cow milk or formula as a covariate. This average daily intake of milk/formula (milliliters per day) is inversely related to breastfeeding status. Of the 359 participants with data on breastfeeding at 1 year, 44.6% were still breastfeeding. The mean age at weaning from breastfeeding was 5.9 months (SD: 3.3 months). Sex (0 = female, 1 = male) was included as a covariate given previous research has found that male infants are more likely to have poorer iron status in the no-added-iron group even after control for birth weight and growth (Lozoff, Beard, et al., Reference Lozoff, Beard, Connor, Barbara, Georgieff and Schallert2006), which also supports evidence that sex differences in iron status are independent of more rapid postnatal growth in males (Domellof et al., Reference Domellof, Lonnerdal, Dewey, Cohen, Rivera and Hernell2002).
Analytic approach
First, ordinal logistic regression models were used to predict iron status (the severity of iron deficiency at 12 months; coded as 0–2 for IS, iron-deficient, and iron-deficient anemic) from ELA exposure (the sum standardized z-score of stressful events, maternal depressive symptoms, and low support in the home). We controlled for the impact of growth velocity on infant iron status by including change in body weight from 6 to 12 months in the ordinal logistic regression model. All analyses included all covariates.
Then, we conducted a separate analysis with SEM to test the extent to which ELA was associated with the continuous hematological measures of iron status at 12 months. As previously described (Lozoff, Kaciroti, et al., Reference Lozoff, Kaciroti and Walter2006), we created a latent variable (iron status) with MCV, hemoglobin, FEP (reverse-coded, log-transformed), and ferritin (log-transformed). A higher value on this continuous latent variable indicates better iron status. We conducted linear regressions within the SEM framework to predict the latent iron status variable from ELA, controlling for birth weight, gestational age, sex, and daily cow milk/formula intake. We fit the structural equation model with the SEM package Lavaan in R studio Version 1.456 (Rosseel, Reference Rosseel2012) with full information maximum likelihood estimates. We assessed the goodness of fit of the model using the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Missing data in exposure variables were imputed using multiple imputation techniques (Rubin, 1987) with IVEWARE software within SAS using available demographic, anthropometric, and environmental data in infancy and early childhood (as described in (Doom et al., Reference Doom, Gahagan, East, Encina, Delva and Lozoff2020; Newman, Reference Newman2016)).
Results
Of the 534 infants analyzed here, 23% were IS at 12 months, approximately half were ID (51.1%), and 25.8% met the criteria for iron deficiency anemia (Table 1). Drawing from historical data and estimates from Latin American countries from 1995 to 2011 for children <5 years, the mean hemoglobin concentrations from the present sample fall within the expected population hemoglobin concentration ranges (Stevens et al., Reference Stevens, Finucane, De-Regil, Paciorek, Flaxman, Branca, Pena-Rosas, Bhutta, Ezzati and Nutrition Impact Model Study2013). The mean maternal depressive symptom score was 14.6 (SD = 2.6), mothers reported an average of 4.8 stressful life events in the last year (SD = 2.6), and the mean HOME support for child development was 39.6 (SD = 7.6). The overall ELA score mean was −.47 (SD = 1.9) and ranged from −4.7 to 6.6, with higher scores indicating higher levels of adversity.
The ordinal logistic regression results showed a significant association between ELA in infancy and 12-month iron status (p = .003), such that higher levels of psychosocial adversity were associated with increased odds of iron deficiency and iron deficiency anemia (Table 2, Figure 1). For every one-unit increase in the ELA score, the odds of being ID or iron-deficient anemic at 12 months was 1.16 (i.e., an increase of 16%), holding constant all other variables. Table 3 displays the correlation matrix of key variables.
Early-life adversity and iron status at 12 months. Note. Early-life adversity was significantly associated with iron status at 12 months. IDA = iron deficiency anemia; ID = iron deficiency; IS = iron sufficient.
Predictors of iron deficiency at 12 months
Note. Iron deficiency is coded as an ordinal categorical variable, where 0 = iron sufficient; 1 = iron-deficient; 2 = iron-deficient anemia. aHigher ELA scores reflect greater early-life adversity. bCoded as 0 = female; 1 = male.
Correlation matrix of key variables
Note. CES-D = Center for Epidemiological Studies Depression scale; HOME = Support for child development in the home; ELA = early-life adversity; Hb = hemoglobin; MCV = mean corpuscular volume; FEP = free erythrocyte protoporphyrin. aHigher values represent worse socioeconomic status. bCoded as 0 = iron sufficient; 1 = iron-deficient; 2 = iron-deficient anemia. *p < .05; **p < .01; ***p < .001.
For the structural equation model, model fit indices demonstrated that the model was an acceptable fit for the data (comparative fit index: .96, root mean square error of approximation: .059, standardized root mean square residual: .032). Full model results are displayed in Table 4 and Figure 2. SEM with the latent variable of hematological iron measures demonstrated a significant association between ELA and 12-month iron status composite, such that higher levels of ELA were associated with poorer iron status (Figure 2 and Table 3; standardized β = −.09; β = −.375; 95% CI = −708 to −.005; p = .047). Table 5 provides additional information on the ELA characteristics by iron status at 12 months.
Higher early-life adversity scores were associated with worse continuous iron status at 12 months. Note. Model fit indices suggest model fit was acceptable (CFI: .96, RMSEA: .059 (90% CI .041–.079), SRMR = .032). All coefficients are standardized. Dotted lines indicate non-significant paths. ELA = early-life adversity; Hb = hemoglobin; MCV = mean corpuscular volume; FEP = free erythrocyte protoporphyrin (reversed); Gest Age = gestational age in weeks; BW = birthweight (g). Sex coded as 0 = female; 1 = male. Formula intake was average intake in ml/day 6–12 months of no-added iron cow milk. *** = Correlation is significant at the 0.001 level (2-tailed). ** = Correlation is significant at the 0.01 level (2-tailed). * = Correlation is significant at the 0.05 level (2-tailed).
Predicting hematological iron status at 12 months in a structural equation model
a Coded as 0 = female, 1 = male. Bolded p values indicate statistical significance at the following levels: *p < .05; **p < .01; ***p < .001.
Early-life adversity characteristics by iron status at 12 months (Mean (SD))
Note. *High scores reflect greater adversity.
Discussion
This study’s results demonstrate that greater ELA exposure in the first year of life was associated with poorer iron status at 12 months of age, analyzed both as a categorical iron status variable and as a continuous hematologic composite. This finding is consistent with animal and prenatal human models of stress and iron status, such that postnatal ELA was associated with iron status in infancy. To our knowledge, this is the first study to examine postnatal stress in the first year of life and iron status in infants.
This study found evidence for poor iron status with clinical cutoffs (ID, iron deficiency anemia) and a continuous composite of iron status across multiple iron markers. The continuous hematologic assessment is critical. There is a risk for harmful effects of compromised iron status that may arise from exposure to ELA even before an infant reaches a clinical iron-deficient state. During the rapid growth period of infancy, the body prioritizes iron to red blood cells and other organs over the brain (Georgieff, Reference Georgieff2017; Zamora et al., Reference Zamora, Guiang, Widness and Georgieff2016). Thus, brain iron is reduced even before the infant becomes anemic (Georgieff et al., Reference Georgieff, Landon, Mills, Hedlund, Faassen, Schmidt, Ophoven and Widness1990; Petry et al., Reference Petry, Eaton, Wobken, Mills, Johnson and Georgieff1992). After iron treatment, red blood cells become iron-replete before the brain (Geguchadze et al., Reference Geguchadze, Coe, Lubach, Clardy, Beard and Connor2008; Rao et al., Reference Rao, Tkac, Townsend, Gruetter and Georgieff2003). Therefore, even before clinical diagnoses of iron deficiency or anemia, disruptions to iron status are likely to influence early brain development (Rao & Georgieff, Reference Rao and Georgieff2002). Inadequate iron can negatively impact neurodevelopment across several domains and in different brain regions, as evidenced by animal studies (Beard & Connor, Reference Beard and Connor2003; Beard et al., Reference Beard, Felt, Schallert, Burhans, Connor and Georgieff2006; Felt et al., Reference Felt, Beard, Schallert, Shao, Aldridge, Connor, Georgieff and Lozoff2006). Maintaining sufficient iron early in life is vital to optimal neurodevelopment and reducing psychopathology risk (Georgieff, Reference Georgieff2017). ELA, an exposure that poses a threat to both neurodevelopment and psychopathology, is found here to be additionally associated with iron insufficiency. Studying ELA and iron status simultaneously may help identify which children could benefit from iron supplementation and additional psychosocial services.
Several explanations exist for the associations found. The association between ELA and iron deficiency found in this study may arise from a bidirectional associations between HPA axis functioning and iron status early in life. For example, there is evidence that iron deficiency (ID) and iron deficiency anemia (IDA) impacts the developing HPA system. (Felt et al., Reference Felt, Peirano, Algarin, Chamorro, Sir, Kaciroti and Lozoff2012; Golub et al., Reference Golub, Hogrefe, Tarantal, Germann, Beard, Georgieff, Calatroni and Lozoff2006; Saad et al., Reference Saad, Morais and Saad1991; Yehuda & Yehuda, Reference Yehuda and Yehuda2006). Non-infectious psychosocial stress may operate along similar pathways of infection and inflammation (Cohen et al., Reference Cohen, Janicki-Deverts, Doyle, Miller, Frank, Rabin and Turner2012), leading to the risk of IDA even in the context of adequate iron intake. Non-infectious psychological stress responses utilize many of the same pathways as infectious stress to alter basic processes of nutrient metabolism, including absorption and prioritization. Like infection, psychosocial stress activates and, if chronic, dysregulates the HPA axis and increases proinflammatory cytokines (Hantsoo et al., Reference Hantsoo, Kornfield, Anguera and Epperson2019). Through glucocorticoid regulation and the HPA axis, ELA and chronic psychological stress may lead to iron sequestration and decreased iron absorption. Hepcidin may mediate these effects. Hepcidin, its receptor, and iron channel ferroportin work in concert to control the dietary absorption, storage, and tissue distribution of iron. Hepcidin is upregulated in response to inflammatory states to decrease iron availability and control infection. During infection and inflammation – and potentially psychosocial stress – hepcidin and ferroportin expression are modulated to reduce iron availability. Iron supply for red blood cell precursors is also restricted, contributing to the anemia associated with infections and inflammatory conditions (Ganz & Nemeth, Reference Ganz and Nemeth2009). ELA and HPA dysregulation are associated with increases in proinflammatory cytokines IL-6 and CRP that regulate hepcidin. Dysregulated neuroendocrine pathways that arise from contexts of psychosocial stress can disrupt iron absorption and utilization even in the context of adequate intake (Monk et al., Reference Monk, Georgieff and Osterholm2013; Suchdev et al., 2017). The hepcidin-mediated reduction in gut absorption may also worsen total body iron deficiency in chronic stress. Thus, when increased inflammation is due to psychological stress, the impact of increased hepcidin activity is increased risk of IDA, without the benefit of controlling infection. Currently, the treatment response to ID or IDA is to increase iron with supplementation. However, suppose stress and neuroendocrine dysregulation is altering iron prioritization and loading through hepcidin. In that case, the answer is a non-nutritional solution, which fundamentally shifts how ID and IDA are assessed and treated.
Alternatively, there may be bidirectional associations between HPA axis functioning and iron status early in life. A prior study on adults found that patients with iron deficiency experienced reduced cortisol secretion in response to an adrenocorticotropic hormone challenge. Other studies have found an impact of ID on HPA functioning in prenatal and postnatal models (Saad et al., Reference Saad, Morais and Saad1991). In a study of rhesus monkeys randomly assigned to prenatal iron deprivation, infants born to mothers that were iron deprived showed elevated cortisol levels in response to novel contexts at 4 months (equivalent in age to older infancy/toddlerhood in humans) even though the animals were never anemic (Golub et al., Reference Golub, Hogrefe, Tarantal, Germann, Beard, Georgieff, Calatroni and Lozoff2006). This suggests that the effects of iron deprivation on the HPA axis can occur in the absence of anemia. Pregnant guinea pigs on an iron-deficient diet during gestation and lactation had offspring with significantly elevated basal cortisol levels at postnatal day 24 than the iron-sufficient control group (Shero et al., Reference Shero, Fiset, Plamondon, Thabet and Rioux2018). Postnatal studies of iron deficiency in humans are sparse but also suggest that ID could have long-term impacts on the developing neuroendocrine system. In one study, infants treated for IDA showed lower morning salivary cortisol levels as children (Yehuda & Yehuda, Reference Yehuda and Yehuda2006), suggestive of a later hypo-responsive HPA system. Another study found that 10-year-old children who had IDA in infancy exhibited a blunted cortisol response to venipuncture and catheter placement (Felt et al., Reference Felt, Peirano, Algarin, Chamorro, Sir, Kaciroti and Lozoff2012). Given the potential for a bidirectional connection between iron status early in life and the developing stress response system, infants exposed to higher levels of ELA be experiencing changes to HPA axis functioning to impact later iron status. Or their stress response systems may be altered from worse iron status to make them more sensitive to conditions of ELA. Without more research, it is impossible to disentangle the two. Unfortunately, this study does not have measures of the stress response system (such as cortisol), inflammation markers, or hepcidin measures. Future studies should consider focusing on biomarkers of the stress response to understand if the stress-mediating systems are involved in the associations between ELA exposure and iron status in infancy.
The association found between ELA exposure and infant iron status may be due to family background factors. In this sample, mothers who were younger, less educated, and of lower socioeconomic status were also more likely to have higher ELA scores (Table 3). These factors could influence nutrition during pregnancy and nursing and the money available for infant food. Families experiencing higher levels of stress and depression may be less able to provide iron-rich complementary food for infants. Conversely, the lack of ability to provide food for their family may also drive rates of stress and depression higher. This alternate explanation for the associations found herein may also inform interventions focusing on nutrition and stress mechanisms.
Several additional considerations of the current study should be noted. For example, ELA was measured as a sum of three forms of family-level adversities, some of which may have been present during mothers' pregnancies. Prenatal adversity was not assessed in this study and may have had residual effects on children’s iron status at 1 year (Armony-Sivan et al., Reference Armony-Sivan, Aviner, Cojocaru, Fytlovitch, Ben-Alon, Eliassy, Babkoff, Lozoff and Anteby2013; Campbell et al., Reference Campbell, Tamayo-Ortiz, Cantoral, Schnaas, Osorio-Valencia, Wright, Tellez-Rojo and Wright2020; McLimore et al., Reference McLimore, Phillips, Blohowiak, Pham, Coe, Fischer and Kling2013; Rendina et al., Reference Rendina, Blohowiak, Coe and Kling2018). A significant limitation to the present study is the lack of data on maternal iron status during pregnancy or postpartum, maternal dietary intake of iron throughout pregnancy and lactation, and infant iron stores at birth. Iron deficiency is common among women of childbearing age, and poor maternal iron status has been linked to depression and adversity. It is thus possible that maternal iron status may have contributed to both infant iron status during pregnancy and to infant iron status at 12 months, indirectly through higher levels of maternal depression (Beard et al., Reference Beard, Hendricks, Perez, Murray-Kolb, Berg, Vernon-Feagans, Irlam, Isaacs, Sive and Tomlinson2005; Black et al., Reference Black, Quigg, Hurley and Pepper2011; Corwin et al., Reference Corwin, Murray-Kolb and Beard2003). As a significant amount of iron is stored before birth via placental transfer in the third trimester (Winzerling & Kling, Reference Winzerling and Kling2001), disruptions to maternal iron status or placental transfer arising from maternal psychosocial adversity may be partially responsible for the postnatal adversity-iron associations found in this study (Monk et al., Reference Monk, Georgieff and Osterholm2013). However, suppose iron transfer and iron stores were inadequate during pregnancy. In that case, research suggests that infants would have developed iron deficiency anemia before recruitment into the iron supplementation trial at 6 months of age (Chaparro, Reference Chaparro2008). As infants with iron deficiency anemia at 6 months were excluded from the iron supplementation trial (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003), the strength of the association might have even been greater if infants with IDA were included.
The current study also did not have information on mothers' alcohol consumption, maternal obesity, or gestational diabetes during pregnancy (Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003), which could have impacted infant iron status (Rao & Georgieff, Reference Rao and Georgieff2002). However, the current study excluded unhealthy pregnancies, growth-restricted or small for gestational age infants, and infants <3 kg, who often have rapid catch-up growth. These exclusions limit the confounding of iron status from those conditions. We also do not have information on the introduction of complementary foods or types of foods other than formula and breastmilk fed to infants, limiting our ability to ascertain if differences in iron status arose from differences in other food consumption.
The current study cohort was recruited from low- and middle-class neighborhoods (Doom et al., Reference Doom, Richards, Caballero, Delva, Gahagan and Lozoff2018). Thus, the ability to assess how adversity contributes to iron status in a full range of socioeconomic contexts was not possible. It is also important to remember that eligibility into the preventive trial may have limited the generalizability of study findings. Children with anemia at 6 months were treated with iron and excluded, as were children with major health problems. Thus, all infants studied here were healthy and nonanemic at 6 months.
It is also important to note the particular context in Chile at the beginning of the study in 1991. From a nutritional and pediatric practice perspective, feeding practices and government-provided formula have since changed. In the current study, infants were generally well-nourished at enrollment because general undernutrition was nearly eradicated in Chile at the time of the study. As part of a legally-mandated initiative to combat universal undernutrition, health clinics provided unmodified powdered milk as part of the National Complementary Feeding Program. The feeding program was available to all children from birth to age five regardless of their territorial location, nationality, or socioeconomic status. Accordingly, the prevalence of IDA in Chile’s population of otherwise healthy, well-nourished infants not receiving routine iron supplementation was conservatively estimated at between 20% and 30% (Brito et al., Reference Brito, Olivares, Pizarro, Rodriguez and Hertrampf2013; Lozoff et al., Reference Lozoff, De Andraca, Castillo, Smith, Walter and Pino2003; Stevens et al., Reference Stevens, Finucane, De-Regil, Paciorek, Flaxman, Branca, Pena-Rosas, Bhutta, Ezzati and Nutrition Impact Model Study2013). In addition to the successful public health campaigns for exclusive breastfeeding noted above during the study, the decades after the study saw changes in the government-distributed powdered milk product, “Purita.” In 1999, Purita was fortified with iron, zinc, copper, and vitamin C, though caregivers still needed to add sugar and other additives to the milk. A series of cross-sectional studies in Chile using data collected before and after the national introduction of iron-fortified milk found that the introduction of the iron-fortified product was associated with lower anemia and improved iron status in 11–18-month-olds (Brito et al., Reference Brito, Olivares, Pizarro, Rodriguez and Hertrampf2013). In 2021, a new infant starter formula was introduced that reduced the need for additives in bottle preparation and more closely matched national and international nutrient recommendations for infant formula (Chile Ministerio de Salud, 2021).
Socio-politically, this study occurred just after the return to democracy, after 13 years of military dictatorship in which government-enforced violence reached high levels (National Research Council, 1985). This may have been a particularly salient time for maternal depressive symptoms. During this time, sociopolitical violence experienced by pregnant women was associated with a five-fold increase in pregnancy complications (Zapata et al., Reference Zapata, Rebolledo, Atalah, Newman and King1992), and domestic violence was associated with increased rates of depression in Chilean women (Quelopana, Reference Quelopana2012). Analysis of a subset of study mothers (N = 215) when children were age 5 years showed that greater economic hardship and more stressful life events were positively associated with domestic abuse, which was itself associated with higher reports of depressive and posttraumatic stress disorder symptoms in study mothers (Ceballo et al., Reference Ceballo, Ramirez, Castillo, Caballero and Lozoff2016). Given the widespread sociopolitical violence in Santiago in the years preceding the start of this study, the levels of maternal depressive symptoms reported here may potentially reflect heightened levels of depressive symptoms compared to other historical periods. In addition, studies of depressive symptoms in Chilean women have found that women of lower socioeconomic status are at greater risk of developing depression (Jadresic & Araya, Reference Jadresic and Araya1995). Thus, the historical context during data collection may limit the generalizability of the study’s findings to other contexts and serve as an example for future studies on ELA and infant iron status to take the sociopolitical context into account when examining associations between ELA and iron status. Nonetheless, the rates of anemia in this sample and mean hemoglobin concentrations are comparable with those in populations in the Latin American region in the early to mid-1990s and more recently in 2011 (Stevens et al., Reference Stevens, Finucane, De-Regil, Paciorek, Flaxman, Branca, Pena-Rosas, Bhutta, Ezzati and Nutrition Impact Model Study2013), suggesting that the study’s findings may continue to be relevant for more contemporary cohorts.
Future directions
Though the current burgeoning literature on the associations between adversity and iron status is intriguing, more work is needed to clarify how these exposures relate to each other and their downstream effects on children’s socioemotional and cognitive development. ELA is associated with a host of later psychopathologies and neurobehavioral development that mirror exposure to early iron deficiency. Adversity is not randomly assigned or changed easily in human studies. Still, animal models that study iron and stress exposures with proper control groups could help understand how the HPA axis influences iron status early in life. Human studies that take advantage of natural experiments (e.g., natural disaster studies) and experiments to either reduce adversity burden or increase iron status could also be used. For instance, if maternal depressive symptoms are a modifiable area of ELA through interventions (e.g., Rojas et al., Reference Rojas, Fritsch, Solis, Jadresic, Castillo, Gonzalez, Guajardo, Lewis, Peters and Araya2007) this may be one way to reduce the ELA burden on children and reduce the risk of child iron deficiency in the first year of life. Future work should also investigate whether ELA influences the response to iron treatment as well as the risk of becoming ID early in life. If physiological changes arising from ELA impact iron prioritization and sequestration, then treating iron deficiency through oral supplementation may be less effective in contexts of ELA. As this study is the first that examines how ELA is associated with iron status in infancy, it lays the groundwork for future work to elucidate why this association exists, and how to intervene to prevent subsequent risk to socioemotional and cognitive development. A better understanding of the synergies of ELA and early nutritional adversities will encourage dual-focused interventions, rather than interventions singularly focused, and could help numerous children around the world reach their developmental potential.
Acknowledgements
We are grateful to the families who participated in the research presented here. Funding from K01HL143159 (PI: Doom), R01HD14122 (PI: Lozoff), R01HD33487 (PI: Lozoff and Gahagan), R01HL088530 (PI: Gahagan), R03-HD-097295 (East), and T32HD101392 (Reid). The sponsors had no role in the study design, the collection, analysis, or interpretation of data, the writing of the report, or the decision to submit the manuscript for publication. We thank the dedicated research study team and all the individuals and families who participated in the research.
Conflicts of interest
None.
Open access
