Abstract
Objective:
There is concern that high iron uptake during the 6- to 24-month critical period of rapid brain development carries potential risks due to iron’s pro-oxidative effects, especially for infants who are nonanemic. This study examined the neurocognitive functioning of 16-year-olds who were nonanemic as infants and received iron supplementation. Dosage effects and interactions between supplementation and 6-month hemoglobin level were also analyzed.
Methods:
We studied 562 Chilean adolescents (M age 16.2 years; 52.7% female) who participated in a randomized controlled iron supplementation trial. Between ages 6 and 12 months, 346 were randomized to receive iron-fortified formula (12.7 Fe mg/L) or, if primarily breastfed, liquid vitamins with 15 mg elemental iron as ferrous sulfate, and 216 were randomized to a no-added iron condition (unmodified cow milk without iron or liquid vitamins without iron if primarily breastfed). Neurocognitive measures were administered at 16y.
Results:
Compared to adolescents in the no-added iron condition in infancy, those in the iron-supplemented condition had poorer visual-motor integration, quantitative reasoning skills, and incurred more errors on neurocognitive tasks. Effect sizes of the differences were small, indicating slight decrement in abilities at adolescence. Consuming larger amounts of iron-fortified formula in infancy was associated with lower arithmetic achievement. Of adolescents who had high hemoglobin at 6 months (Hb ≥ 125 g/L), those in the iron supplemented condition had poorer performance on arithmetic, quantitative reasoning, and response inhibition tests than those in the no-added iron condition. Of adolescents who had marginally low 6-month hemoglobin (Hb > 100 and < 110 g/L), those in the no-added iron condition incurred more errors on a visual searching task than those in the iron-supplemented condition.
Conclusion:
The physiologic need for iron during the period of rapid and critical brain development in young infants should be considered vis-à-vis the risks associated with supplementing nonanemic infants with high levels of iron.
Keywords: iron supplementation, neurocognition, executive function, adolescents, Chile
Introduction
Iron deficiency is the world’s leading nutrient deficiency, affecting more than 2 billion people worldwide, including approximately 115 million children [1]. Iron deficiency anemia (IDA) during infancy is of particular concern due to the vulnerability of the developing brain, with iron an essential nutrient for normal brain development. While there are known adverse effects of IDA during infancy, overexposure to iron within the 6- to 24-month critical period of rapid brain development is also suspected to cause harm [2,3]. Iron homeostasis is tightly-regulated biologically. High iron intake, especially if nonanemic, may carry potential risks due to iron’s pro-oxidative effects, competition with the absorption of other trace elements, and role in immune and microflora disturbance [4].
In addition to evidence of gastrointestinal infections and decreased growth [4], there are concerns about possible adverse cognitive effects of high iron intake during infancy [5,6]. In a follow-up of a randomized controlled iron-deficiency anemia preventive trial involving Chilean infants, those randomized to an iron-fortified formula (12.7 mg/L, the recommended amount of iron in the U.S. at the time of the study [1991-1996] and one regularly consumed by U.S. infants) scored lower on several neurodevelopmental measures (i.e., spatial memory, visual-motor integration) at age 10 years compared to those randomized to a low-iron formula (2.3 mg/L) [7]. In a 16-year follow-up of that sample, the iron-fortified group continued to score lower than the low-iron group on 8 of the 9 tests given, statistically significant for arithmetic ability, reading comprehension, and spatial memory [8]. At both ages, effects varied by infant hemoglobin concentration, such that children with high 6-month hemoglobin (>128 g/L) randomized to the iron-fortified formula had lower scores compared with those randomized to the low-iron formula [7,8]. Given the interactions with baseline iron status, the authors [7,8], as well as others [2,3], have suggested that iron supplementation at previously recommended levels may be unnecessarily high and potentially harmful in nonanemic infants. In 2014, the European Society of Pediatric Gastroenterology, Hepatology and Nutrition (ESPGHN) recommended lower concentrations of iron (4-7 mg/L) in infant formula [9].
Currently there is debate about whether excess dietary iron is fully absorbed in infancy or whether protective mechanisms can effectively guard against excess iron reaching the brain [10,11]. Human infants may be particularly vulnerable due to undeveloped feedback mechanisms in the infant gut and brain that limit excess iron uptake [10]. Some evidence indicates that regulation of iron metabolism is immature prior to 6 months of age [11]. One study found that both hemoglobin and plasma ferritin levels increased in infants in the presence of iron supplementation even when the initial values were well above standard levels, suggesting that excess iron can be absorbed regardless of need [12]. Iron accumulates naturally in the brain with aging, and excess iron cannot be easily eliminated from the human body. It is conceivable that the intake of higher than needed levels of dietary iron in infancy could initiate dysregulation of iron transfer to the brain and thus contribute to long-term increases in brain iron concentrations. Elevated iron concentrations in various brain regions in adults have been associated with motor and cognitive abnormalities and neurodegenerative diseases, such as Parkinson’s and Alzheimer’s Disease [3,13,14].
The aim of the present study was to assess neurocognitive functioning of 16-year-olds who were nonanemic as infants and received iron supplementation at previously recommended levels. We analyzed outcomes of the Chilean sample described above and compared visual-motor integration, cognitive abilities, and neurocognitive functioning of adolescents randomized in infancy to iron supplementation or a no-added iron condition, focusing on participants who were nonanemic throughout the preventive trial. We do not include infants randomized to the low-iron formula in the Chile study because we are interested in studying the effects of supplementing nonanemic infants with higher than needed levels of dietary iron. Inclusion of nonanemic infants who received no-added iron in infancy is informative as it gauges the impact of the absence of supplemental iron for those for whom it may not be necessary. We also consider dosage effects associated with the amount of iron-fortified formula consumed, as well as the role of 6-month hemoglobin concentration on 16-year outcomes. We hypothesized that those randomized to the iron supplemented condition who consumed large amounts of iron-fortified formula and, separately, had high 6-month hemoglobin concentrations would fare most poorly. There is no other study of which we are aware that has examined long-term cognitive outcomes of infants with good iron status who were supplemented with iron.
Methods
Study Design and Participants
Participants were 562 Chilean adolescents (M age 16.2 years; 52.7% female) from predominantly working-class backgrounds who have been studied since infancy as part of a double-blind iron-deficiency anemia preventive trial [15]. Six-month-old infants were recruited from public health clinics in urban neighborhoods surrounding Santiago, Chile. The trial’s eligibility criteria included singleton birth, term vaginal delivery, birth weight ≥ 3.0 kg, and no hospitalization longer than 5 days, phototherapy, birth or perinatal complications, or major congenital anomalies or chronic illnesses. Infants were screened for iron deficiency anemia (IDA) prior to entry into the trial by a finger stick to determine hemoglobin (Hb) concentration (HemoCue, Leo Diagnostics, Helsingborg, Sweden). Infants with low Hb (≤ 103 g/L) and the next screened infant who was clearly nonanemic (Hb ≥ 115 g/L) were further assessed by venipuncture. Those with venous Hb ≤ 100 g/L and 2 of 3 iron measures in the deficient range (described below), along with a nonanemic control, were invited to join a neurophysiology study and treated with oral iron (n =135); they are not studied here. All other infants were randomized to receive study-provided formula or, if primarily breastfed, vitamins between ages 6 and 12 months. A total of 1657 infants completed the preventive trial, randomized as: 718 to an iron-fortified formula (12.7 mg/L) or a liquid multivitamin preparation (15 mg elemental iron per day as ferrous sulfate), 405 to a low-iron formula (2.3 mg/L); and 534 to a no-added iron condition (unmodified cow milk formula or a liquid multivitamin preparation without iron).
Iron status was assessed in venipuncture blood samples for all infants who completed the preventive trial at 12 months. At 18 months, 48% of children in the low-iron and no-added iron conditions received a repeat venipuncture [15]. Anemia was defined as Hb < 110 (g/L) at 12 and 18 months. Iron deficiency (ID) was defined as 2 of 3 iron measures in the deficient range (free erythrocyte protoporphyrin concentration [FEP] > 100 μg/dL red blood cells, mean corpuscular volume [MCV] < 70 fl, serum ferritin < 12 μg/L). IDA was defined as anemia + ID.
Throughout the preventive trial, breastfeeding and the amount of formula or cow milk ingested was recorded at weekly home visits. Age at first bottle and last breastfeeding (if weaned) was recorded in months. Average daily intake of milk/formula (milliliters per day) was calculated and was inversely related to the breastfeeding variables. Birth weight was obtained from hospital records. Weight was measured monthly on an electronic scale (to the nearest 0.01 kg) from 6 to 12 months by the study team.
For the current analysis, we endeavored to address concerns about excess iron in nonanemic infants. To do so, we compared the iron-supplemented and no-added iron groups. Children who received oral iron because they developed IDA were excluded: 201 were treated for IDA (32 of 718 in the iron-supplemented group and 169 of 534 in the no-added iron group). An additional 41 infants received oral iron outside of the preventive trial (30 iron-supplemented, 11 no-added iron) and were excluded. The resulting sample was 1,010 infants: 656 iron-supplemented, 354 no-added iron. (See sample flow chart, Figure 1).
Figure 1.
Original sample and sample selection for current study.
A total of 562 of the 1,010 infants were successfully followed up and studied at age 16 years: 346 in the iron-supplemented group and 216 in the no-added iron group. A higher proportion of the no-added iron group was represented at 16 years compared to the iron-supplemented group (61.0% vs. 52.7%; χ2 [1] = 6.38, P < .05). This likely occurred because we instituted better tracking measures for those enrolled later in the trial, which included more of the no-added iron group [15]. Adolescents who did or did not participate at 16 years did not differ on any of the background characteristics assessed at infancy.
Procedures at the Adolescent Follow-up
All measures were administered at a university nutritional research center, which was familiar to participants from their earlier study involvement. Spanish versions of the study measures were supplied by each measure’s publisher, and tests were administered by psychologists trained in the administration of such tests and according to standard instructions. All study components were approved by the relevant university institutional review boards (IRB Project Number #121649). Signed informed consent was obtained from parents at all time points; assent was obtained from children at age 16.
Measures
The Beery-Buktenica Test of Visual-Motor Integration (VMI) is a widely used standardized copy and tracing forms test that assesses the ability to integrate and coordinate visual and motor skills. Scores were standardized to a mean of 100 (range 45 – 120), with higher scores indicating better visuoperceptual integration skills. Several studies support the validity of the Beery-Buktenica test for detecting kinesthetic and motor deficits in children with developmental or acquired brain dysfunction [16].
The Wide Range Achievement Test-Revised (WRAT-R)–Arithmetic yields a score reflecting arithmetic computation and calculating abilities, including algebraic and geometric operations. Standardized scores were used and range from 45 to 75.
The Wechsler Intelligence Scale for Children-Revised (WISC-R) provides a matrix reasoning score, reflecting inductive and quantitative reasoning (score range: 7 – 32), and a verbal ability score, which measures vocabulary knowledge (score range: 9 - 43).
The Trail Making Test (TMT) assesses processing speed and mental flexibility in a visual searching and sequencing task [17]. Part A assesses visuoperceptual abilities and mental processing speed, and Part B assesses response inhibition and task switching. Time to complete Part B minus time to complete Part A assesses task switching independent of processing speed and provides a purer indicator of impulse inhibition. Analysis of TMT errors reveals frontal-mediated cognitive impairments associated with brain regions involved in executive functions and are also analyzed here [18].
The Wisconsin Card Sorting Test (WCST) is a complex decision-making task that assesses cognitive flexibility and information processing in a card game scenario [19]. The WCST has proven to be a valuable clinical tool for assessing the frontal lobe functions of strategic planning, organized searching, set shifting, and modulating impulsive responding [20]. Perseverative errors, or classifying cards using rules that were previously correct, have been noted in individuals with basal ganglia abnormalities or dysfunction of the prefrontal cortex [20]. Non-perseverative errors involve all errors other than perseverative errors. The following scores assess problem-solving abilities and mental flexibility: conceptual level responses are the number of correct responses that occur in runs of three or more and are thought to provide insight into sorting abilities; categories completed are the number of runs of 10 correct responses; and trials to first category are the number of trials needed to achieve the first 10 consecutive correct responses. A computerized version of the WCST was administered.
Covariates
Covariates determined a priori to be related to study outcomes include child sex, family socioeconomic status (SES; measured using the Graffar social class instrument) [21], maternal IQ (abbreviated Wechsler Adult Intelligence Scale), maternal level of education, nurturance and support in the home (measured on the Home Observation for Measurement of the Environment) [22], weight gain from birth to 12 months, amount of formula/milk consumed (mean ml/day 6-12 months), and mode of supplementation (formula vs. vitamin drops) (Table 1). Age at assessment and birthweight were originally considered as covariates but were eliminated due to their lack of association with any of the adolescent outcomes. Family background covariates were based on data collected during children’s infancy (ages 9 – 12 months) as there were little missing data at this time point. Blood lead levels during infancy were available for only 128 of the adolescents studied here; thus, it was not possible to consider blood lead concentration as a covariate. There were no significant differences in lead concentration between the supplementation groups or between children receiving formula vs. vitamin drops.
Table 1.
Descriptive statistics of participant characteristics and study variables.
| Variable | N | M or % | SD | Range |
|---|---|---|---|---|
| Received iron supplementation | 346 | 61.6% | ||
| †Via iron-fortified formula (12 mg/L) | 291 | |||
| †Via multivitamin with iron | 55 | |||
| Received no iron supplementation | 216 | 38.4% | ||
| †Via no-added iron cow milk | 169 | |||
| †Via multivitamin without iron | 47 | |||
| Hemoglobin, 6m (g/L) | 562 | 117.1 | 7.3 | 101 - 140 |
| †Formula consumed (ml/day 6-12m) | 562 | 390.3 | 210.2 | 0 – 905.3 |
| †Sex (% female) | 562 | 52.7% | ||
| †Socioeconomic status, infancya | 562 | 27.0 | 6.2 | 9 - 47 |
| †Maternal education, infancy (years) | 562 | 9.6 | 2.6 | 1 - 17 |
| †Home nurturing score, infancy | 562 | 30.1 | 4.6 | 12 - 42 |
| †Maternal IQ, infancyb | 562 | 84.2 | 9.3 | 52 – 108 |
| Birthweight (kg) | 562 | 3.6 | 0.4 | 3.0 – 5.0 |
| †Weight gain birth to 12m (kg) | 562 | 6.4 | 1.0 | 3.5 – 9.9 |
| Age at 16y assessment | 562 | 16.2 | 0.2 | 15.9 – 17.3 |
| Age 16y assessments | ||||
| Visual-motor integrationc | 562 | 89.6 | 9.2 | 47 - 104 |
| WRAT-R arithmetic achievement | 562 | 55.1 | 5.1 | 45 – 70 |
| WISC-R matrix reasoning | 562 | 21.4 | 4.1 | 7 – 32 |
| WISC-R verbal | 562 | 29.3 | 5.4 | 9 - 43 |
| TMT Part A, sec | 548 | 50.0 | 16.0 | 18 - 132 |
| TMT Part B, sec | 548 | 89.0 | 33.0 | 38 - 260 |
| TMT Part B – Part A, sec | 548 | 38.9 | 28.8 | −9 - 206 |
| TMT Part A errors | 548 | 0.15 | 0.41 | 0 - 2 |
| TMT Part B errors | 548 | 0.38 | 0.80 | 0 - 8 |
| WCST total number correct | 530 | 44.8 | 8.4 | 14 - 58 |
| WCST perseverative errors | 530 | 9.8 | 4.9 | 3 - 39 |
| WCST non-perseverative errors | 530 | 9.4 | 5.7 | 1 - 36 |
| WCST conceptual level responses | 530 | 39.7 | 11.5 | 3 - 58 |
| WCST categories completed | 530 | 3.1 | 1.1 | 0 - 5 |
| WCST trials to 1st category | 530 | 14.4 | 9.2 | 10 - 65 |
Indicates a covariate.
Higher scores indicate more socioeconomic disadvantage.
Assessed by the abbreviated Wechsler Adult Intelligence Scale.
Assessed by the Berry-Buktenica test of visual-motor integration.
WRAT-R = Wide Range Achievement Test - Revised. WISC-R = Wechler Intelligence Scale for Children - Revised. TMT = Trail Making Test. WCST = Wisconsin Card Sorting Test. sec = seconds.
Data analytic plan
Analysis of covariance (ANCOVA) was used to test differences in VMI, arithmetic and verbal skill, and neurocognitive functioning at age 16 between the iron supplemented and no-added iron groups. We report adjusted mean differences (and 95% confidence intervals) for each test score, as well as an effect size of the difference (d). To test dosage effects associated with consumption of iron-fortified formula, multivariable linear regressions were conducted with the amount of iron-fortified formula consumed (6-12 months) as the independent variable. To ensure that quantity of formula per se was not associated with study outcomes, we also regressed the amount of cow milk consumed (6-12 months) on study outcomes. To determine whether the effects of supplementation group on outcome varied by hemoglobin concentration at study enrollment (6 months), multivariable linear regressions were conducted on test scores using iron supplementation group (iron-supplemented vs. no-added iron), 6-month capillary hemoglobin values, and their interaction as independent variables. All analyses included covariates.
Results
Description of Sample and Equivalence of Supplementation Groups
Slightly more than half of the sample was female, and participants averaged 16.2 years of age (SD = 0.2) at testing. The two supplementation groups were equivalent in most family and background characteristics (e.g., sex, birthweight, first-year growth, 6-month hemoglobin level, age at the 16-year assessment) (Table 2). However, formula intake was higher in the iron-supplemented group than was intake of cow milk in the no-added iron group (434.5 ml/day vs. 318.9 ml/day, respectively). This resulted from a change in enrollment criteria during the infancy study -- allowing more intensely breastfed infants to qualify for the study at the same time the no-added-iron group was added [15]. The iron-supplemented group was also from more disadvantaged families than the no-added iron group (Table 2). All background factors, including these differences, were controlled statistically.
Table 2.
Participant characteristics by supplementation group.
| Iron supplemented (N = 346) |
No-added iron (N = 216) |
||
|---|---|---|---|
| Variable | M (SD) or % | M (SD) or % | P of difference |
| Hemoglobin, 6m (g/L) | 116.9 (7.6) | 117.4 (6.8) | .49 |
| Amount formula consumed (ml/day 6-12m) | 434.5 (210.2) | 318.9 (190.0) | .001 |
| Sex (% female) | 47% | 50% | .41 |
| Socioeconomic status, infancya | 27.6 (6.2) | 26.1 (6.1) | .01 |
| Maternal education, infancy (years) | 9.5 (2.7) | 9.8 (2.6) | .09 |
| Home nurturing score, infancy | 30.3 (4.7) | 29.9 (4.5) | .31 |
| Maternal IQ, infancyb | 83.9 (9.6) | 84.7 (8.7) | .31 |
| Birthweight (kg) | 3.6 (0.4) | 3.6 (0.4) | .19 |
| Weight gain birth to 12m (kg) | 6.4 (1.0) | 6.4 (1.0) | .47 |
| Age at 16y assessment | 16.2 (0.2) | 16.2 (0.1) | .31 |
Note.
Higher scores indicate more socioeconomic disadvantage.
Assessed by the abbreviated Wechsler Adult Intelligence Scale.
Differences by Supplementation Group
Compared to adolescents randomized to the no-added iron condition in infancy, those randomized to iron supplementation had poorer visual-motor integration, lower matrix reasoning ability, incurred more errors on the TMT Part B, and had poorer scores on all WCST measures, controlling for covariates (Table 3). Effect sizes associated with the statistically significant differences were relatively small (d ≥ 0.16 and ≤ 0.26).
Table 3.
Mean scores (95% confidence intervals) at age 16 years by supplementation group.
| Iron fortified (N = 339-346) | No added iron (N = 205-216) | ||||
|---|---|---|---|---|---|
|
|
|||||
| M (95% CI) | M (95% CI) | F (1, 535-553) | P | d | |
| Visual-motor integration a | 88.8 (87.8, 89.8) | 90.8 (89.6, 92.1) | 5.70 | .017 | .20 |
| WRAT-R arithmetic | 54.9 (54.1, 55.7) | 55.3 (53.2, 57.4) | < 1 | .733 | .03 |
| WISC-R matrix reasoning | 21.1 (20.7, 21.6) | 21.9 (21.4, 22.5) | 4.65 | .031 | .21 |
| WISC-R verbal | 29.1 (28.6, 29.7) | 29.5 (28.8, 30.3) | < 1 | .424 | .09 |
| TMT Part A, seconds | 49.5 (47.8, 51.3) | 50.9 (48.6, 53.1) | < 1 | .373 | .02 |
| TMT Part B, seconds | 90.0 (86.3, 93.6) | 87.2 (82.5, 91.9) | < 1 | .372 | .06 |
| TMT Part B – Part A, seconds | 40.1 (37.0, 43.2) | 36.8 (32.7, 40.8) | 1.56 | .212 | .06 |
| TMT Part A errors | 0.15 (0.10, 0.19) | 0.16 (0.10, 0.22) | < 1 | .655 | .01 |
| TMT Part B errors | 0.44 (0.35, 0.53) | 0.27 (0.16, 0.39) | 5.04 | .025 | .16 |
| WCST total number correct | 44.0 (43.1, 44.9) | 46.3 (45.1, 47.6) | 8.60 | .004 | .24 |
| WCST perseverative errors | 10.1 (9.6, 10.6) | 9.3(8.6, 10.0) | 3.01 | .083 | .11 |
| WCST non-perseverative errors | 10.0 (9.3, 10.6) | 8.3 (7.5, 9.2) | 8.96 | .003 | .26 |
| WCST conceptual level responses | 38.6 (37.4, 39.9) | 41.9 (40.2, 43.6) | 8.90 | .003 | .25 |
| WCST categories completed | 3.1 (2.9, 3.2) | 3.3 (3.1, 3.5) | 5.78 | .017 | .21 |
| WCST trials to 1st category | 15.2 (14.2, 16.2) | 12.8 (11.4, 14.1) | 7.70 | .006 | .21 |
Note. d = effect size of difference.
Assessed by the Berry-Buktenica test of visual-motor integration.
WRAT-R = Wide Range Achievement Test - Revised. WISC-R = Wechler Intelligence Scale for Children - Revised. TMT = Trail Making Test, Part A, Part B, and Part B minus Part A. WCST = Wisconsin Card Sorting Test. Means are adjusted for: sex, maternal IQ, maternal education, family SES, home environment, weight gain birth to 12m, formula intake, and mode of iron supplementation (formula vs. vitamin drops).
Dosage Effects of Iron-fortified Formula in Infancy
The regression analysis of mean daily intake of iron-fortified formula on 16-year outcomes indicated that higher intake of iron-fortified formula was associated with lower mathematical achievement as measured on the WRAT-R (β = −.11, B = −.062, SE = .030, P = .039) (Table S1). Amount of cow milk formula intake was not significantly related to any study outcome.
Interaction of Supplementation Group and 6-month Hemoglobin
Results of regressions testing the interaction between iron supplementation group and 6-month hemoglobin concentration showed statistically significant interactions for the WRAT-R arithmetic scores and TMT Part A errors, and a trend for WISC-R matrix reasoning and time to complete TMT Part B and TMT Part B – Part A (Table 4). Figure 2 shows test scores plotted by supplementation group (adjusted for covariates) at various levels of 6-month Hb (6-month Hb M = 117 g/L; SD = 7.3). Reading left to right along the x-axis, scores are shown for those below 1 standard deviation (SD) from the 6-month Hb mean (Hb ≥ 101 and ≤ 109); those within 1 SD below the 6-month Hb mean (Hb ≥ 110 and ≤ 116); those within 1 SD above the mean (Hb ≥ 117 and ≤ 124); and those above 1 SD from the mean (Hb ≥ 125 and ≤ 140). Follow-up analyses indicated that, when 6-month hemoglobin was high, (i.e., above 1 SD from the mean, or Hb ≥ 125 g/L), adolescents in the iron-supplemented group had lower scores than those in the no-added iron condition on the WRAT-R arithmetic and WISC-R matrix reasoning tests and had longer completion times for the TMT Part B (P < .05), and the TMT Part B – Part A (P = .051). A significant group difference also emerged at Hb levels of 117-124, with the no-added iron group having higher WRAT-R arithmetic scores than the iron-supplemented group.
Table 4.
Multivariable linear regression results of supplementation group X 6-month hemoglobin level predicting scores at age 16 years.
| β | B | SE | P | R 2 | |
|---|---|---|---|---|---|
| Visual-motor integration | −.07 | −0.78 | 0.84 | .358 | .022 |
| WRAT-R arithmetic | −.44 | −2.37 | 1.12 | .036 | .174 |
| WISC-R matrix reasoning | −.12 | −0.36 | 0.22 | .096 | .026 |
| WISC-R verbal | −.02 | −0.04 | 0.18 | .819 | .078 |
| TMT Part A, seconds | .04 | 0.72 | 1.48 | .627 | .015 |
| TMT Part B, seconds | .13 | 5.16 | 3.02 | .089 | .031 |
| TMT B – A, seconds | .13 | 4.38 | 2.62 | .095 | .045 |
| TMT Part A errors | .17 | 0.09 | 0.04 | .021 | .033 |
| TMT Part B errors | .09 | 0.08 | 0.07 | .262 | .039 |
| WCST total number correct | −.02 | −0.15 | 0.78 | .848 | .031 |
| WCTS perseverative errors | .05 | 0.30 | 0.46 | .511 | .029 |
| WCST non-perseverative errors | .07 | 0.48 | 0.53 | .370 | .039 |
| WCST conceptual level responses | −.02 | −0.23 | 1.07 | .830 | .034 |
| WCST categories completed | −.02 | −0.02 | 0.11 | .820 | .025 |
| WCST trials to 1st category | −.12 | −1.27 | 0.87 | .145 | .036 |
Note. β = standardized regression coefficient of the interaction between supplementation group and 6-month hemoglobin level. WRAT-R = Wide Range Achievement Test – Revised. WISC- R = Wechsler Intelligence Scale for Children – Revised. TMT = Trail Making Test. WCST = Wisconsin Card Sorting Test. Models controlled for sex, maternal IQ, maternal education, home environment, family SES, weight gain birth to 12m, formula intake, and mode of supplementation (formula vs. vitamin drops).
Figure 2.
Scores on study measures for the iron supplemented and no-added iron groups by 6-month hemoglobin level. Reading left to right along the x-axis, scores are shown for those below 1 standard deviation (SD) from the 6-month Hb mean (Hb ≥ 101 and ≤ 109); those within 1 SD below the 6-month Hb mean (Hb ≥ 110 and ≤ 116); those within 1 SD above the 6-month Hb mean (Hb ≥ 117 and ≤ 124); and those above 1 SD from the 6-month Hb mean (Hb ≥ 125 and ≤ 140). Scores are adjusted for sex, maternal IQ, maternal education, home environment, family SES, weight gain birth to 12m, formula intake, and mode of supplementation (formula vs. vitamin drops). Confidence intervals are shown as vertical bars.
When 6-month hemoglobin was relatively low (within 1 SD below the mean, 110-116), those in the no-added iron condition had lower WRAT-R arithmetic scores than those in the high-iron group. When 6-month hemoglobin was at the very low end (Hb ≤ 109), adolescents in the no-added iron condition incurred more errors on the TMT Part A than those in the iron-supplemented group. To determine the potentially protective role of high hemoglobin concentrations given no-added iron, one effect was found: the no-added iron group had fewer errors on the TMT Part A at the highest versus the lowest hemoglobin levels (0.07 compared to 0.39, respectively; F = 7.14, P = .01).
Discussion
This study found that adolescents who were nonanemic and randomized to receive iron supplementation in infancy had poorer visual-motor and quantitative reasoning skills and incurred more errors on neurocognitive tasks than those who were nonanemic and received no-added iron in infancy. Effect sizes of the differences were small, indicating that supplementation among nonanemic infants conferred relatively slight decrements in abilities at adolescence. A dosage effect of consuming higher quantities of iron-fortified formula was also found for poorer arithmetic skills.
The mechanisms underlying the differences observed in test performance between the two supplementation groups are unclear. This study was not designed to investigate specific neurobiological mechanisms, but we can speculate on the commonalities across the various findings. The iron-supplemented group’s lower scores on the Berry-Buktenica test of visual-motor integration suggests impairment in visuomotor coordination, which in and of itself could affect performance on a variety of neurobehavioral tests. Errors on the TMT Part B indicate failure to identify visual targets and difficulty inhibiting a familiar response, or frontostriatal-mediated executive functions [23]. Low scores on the Wisconsin Card Sort Test indicate poor problem-solving and set-shifting abilities and may indicate frontal lobe impairment [24]. Thus, findings across this study’s various measures converge to suggest that early-life iron supplementation given to nonanemic infants might portend to alterations in the parietal lobe, responsible for visuomotor integration, and the frontal lobe, evident here as difficulties in response inhibition, set shifting, and problem solving.
We also found that iron supplementation in tandem with high levels of infant hemoglobin was associated with poorer arithmetic and quantitative reasoning skills and longer times associated with the TMT Part B and the TMT Part B – Part A. This suggests that infants with high hemoglobin concentrations (≥ 125 g/L) are at specific risk when receiving high levels of iron supplementation. However, iron supplementation was unlikely to pose meaningful risks for infants who had low hemoglobin levels.
Of adolescents with low 6-month hemoglobin, those in the no-added iron condition had poorer visuoperceptual abilities (more TMT Part A errors) than those in the iron-fortified condition. A similar effect was found for arithmetic scores at hemoglobin levels between 110 and 116. These results indicate that the absence of iron supplementation may impair those with relatively low but still normal levels of hemoglobin as they enter the period of rapid growth from 6 to 12 months of age. Results related to errors on the TMT Part A also demonstrated that high hemoglobin in infancy was protective when consuming no supplemental iron. Altogether these findings highlight the importance of targeting supplementation to need.
To our knowledge, the present study is the first to investigate neurocognitive effects at adolescence derived from iron supplementation given to nonanemic infants. Unlike previous analyses involving this Chilean sample [7,8], the current analyses focused on nonanemic infants randomized to a high- or no-added-iron formula to directly assess the impact of consuming high levels of iron supplementation in infants with good iron status. There have been repeated concerns in the literature about iron supplementation in infants with good iron status who otherwise might not need supplemental iron [2–6]. Study of nonanemic infants in the no-added iron condition provides a useful comparison as it gauges the impact of the absence of supplemental iron for those for whom it may be unnecessary.
Two points are important to emphasize regarding our study. The first is that iron status was not determined for all infants at enrollment into the preventive trial. Rather, enrollment was based on hemoglobin at 6 months. Because we did not have additional iron indices at 6 months, some participants could have been iron-insufficient at entry into the trial. The current results pertain only to nonanemic infants and cannot directly address the issue of iron-repletion. Secondly, this was a population at high risk for iron deficiency in infancy, and iron supplementation was highly effective in reducing that risk. Infants randomized to the iron-supplemented condition in the preventive trial were diagnosed with iron-deficiency anemia at 7-times lower rates than those randomized to the no-added iron condition (4.5% vs. 31.7%, respectively). Thus, iron supplementation for the purpose of preventing iron-deficiency anemia remains important. The question is the optimal level of supplementation. Our study used formula fortified at the level of iron recommended by the American Academy of Pediatrics at the time of the study (12.7 mg/L). Earlier results using the Chilean preventive trial sample showed that 2.3 mg/L of iron fortification in formula was sufficient to prevent IDA while resulting in better outcomes than fortification at the high level of iron supplementation [7,8]. Iron fortification in the lower range (2 mg/L) has been widely endorsed recently, with recommendations to stage formula iron content by age and breastfeeding status [25].
Several additional limitations of the current study should be considered. We did not have information on maternal iron status during pregnancy or postpartum, infant iron stores at birth, or iron measures at 6 months for all infants. Infant iron status is influenced by proper fetal iron load during pregnancy, with adequate maternal and neonatal iron status reducing the need for additional iron supplementation [2]. Additionally, iron status may be influenced by non-nutritional factors as well as nutritional intake other than iron [10]. Such factors were considered and tested where possible. Blood lead levels were tested at 12 months in only a subset of those analyzed here. This prevented us from using lead as a control in analysis and warrants that results be interpreted cautiously. However, we reiterate that blood lead concentrations were equivalent in the two supplementation groups.
Other factors, such as malaria, hemoglobinopathies, and parasites causing blood loss were virtually nonexistent in the current sample [15]. However, other confounding factors could have affected individuals’ test performance. For example, visuoperceptual and quantitative abilities correlate with distractibility and memory [16,17], abilities not accounted for here. Higher exclusion due to IDA in the no-added iron group might also have affected results. Finally, effects of iron supplementation were examined only among those with adequate hemoglobin concentrations in infancy and at relatively high levels of iron supplementation. As stated earlier, lower levels of iron-fortification in infant formula are now recommended and are being routinely implemented [9,25]. Findings are important, however, for those choosing to exceed recommended iron fortification levels.
Study strengths include the randomized design coupled with long-term follow-up at adolescence. The focus on nonanemic infants is unique and directly addresses the impact of iron supplementation in infants with good iron status. The strict study eligibility criteria requiring healthy, term births decrease the likelihood of confounding by prenatal factors, and the preventive trial was administered by a university nutritional research center, yielding very high compliance [15]. Outcomes were assessed by well-validated measures and encompassed a broad range of developmental domains, including visual-motor integration, mathematical and verbal skills, and executive functions. We also statistically controlled for a comprehensive panel of covariates in all analyses, including infant growth to account for accelerated postnatal growth depleting iron stores.
Conclusion
The physiologic need for iron during the period of rapid and critical brain development in young infants should be considered vis-à-vis the risks associated with supplementing nonanemic infants with high levels of iron. Further study is needed to identify the safest level of iron supplementation to ensure adequate iron status while preventing adverse effects associated with too much iron. Further research is also needed to understand the mechanisms by which higher levels of iron supplementation might contribute to the suboptimal cognitive and neurocognitive outcomes found here.
Supplementary Material
Funding:
This work was supported by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R03-HD097295, R01-HD033487, T32-HD101392), and the National Heart, Lung, and Blood Institute (R01-HL-088530).
Footnotes
Disclosure Statement: The authors report no conflict of interest.
Clinical Trials number: NCT01166451
Data Availability Statement (DAS)
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1.Sundararajan S, Rabe H. Prevention of iron deficiency anemia in infants and toddlers. Pediatr Research. 2021. Jan;89(1):63–73. doi: 10.1038/s41390-020-0907-5 [DOI] [PubMed] [Google Scholar]
- 2.Georgieff MK, Krebs NF, Cusick SE. The benefits and risks of iron supplementation in pregnancy and childhood. Annual Rev Nutr. 2019; 39:121–46. doi: 10.1146/annurev-nutr-082018-124213 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hare D, Cardoso BR, Szymlek-Gay EA, Biggs B. Neurological effects of iron supplementation in infancy: Finding the balance between health and harm in iron-replete infants. Lancet Child Adolesc Health. 2018;2(2):144–156. doi: 10.1016/S2352-4642(17)30159-1 [DOI] [PubMed] [Google Scholar]
- 4.Lönnerdal B Excess iron intake as a factor in growth, infections, and development of infants and young children. Am J Clin Nutr. 2017;106(suppl 6):1681S–7S. doi: 10.3945/ajcn.117.156042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Iannotti LL, Tielsch JM, Black MM, Black RE. Iron supplementation in early childhood: Health benefits and risks. Am J Clin Nutr. 2006;84(6):1261–76. doi: 10.1093/ajcn/84.6.1261 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wessling-Resnick M Excess iron: Considerations related to development and early growth. Am J Clin Nutr. 2017;106(suppl 6):1600S–5S. doi: 10.3945/ajcn.117.155879 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lozoff B, Castillo M, Clark KM, Smith JB. Iron-fortified vs low-iron infant formula: Developmental outcome at 10 years. Arch Ped & Adol Med. 2012;166(3):208–15. doi: 10.1001/archpediatrics.2011.197 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gahagan S, Delker E, Blanco E, Burrows R, Lozoff B. Randomized controlled trial of iron-fortified versus low-iron infant formula: Developmental outcomes at 16 years. J Pediatr. 2019; 212:124–130. doi: 10.1016/j.jpeds.2019.05.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Domellöf M, Braegger C, Campoy C, Colomb V, Decsi T, Fewtrell M, Hojsak I, Mihatsch W, Molgaard C, Shamir R, Turck D. Iron requirements of infants and toddlers. J Pediatr Gastroenterology and Nutr. 2014. Jan 1;58(1):119–29. doi: 10.1097/MPG.0000000000000206 [DOI] [PubMed] [Google Scholar]
- 10.Lönnerdal B, Georgieff MK, Hernell O. Developmental physiology of iron absorption, homeostasis, and metabolism in the healthy term infant. J Pediatr. 2015;167(4):S8–14. doi: 10.1016/j.jpeds.2015.07.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Hare DJ, Ayton S, Bush AI, Lei P. A delicate balance: Iron metabolism and diseases of the brain. Frontiers in Aging Neuroscience. 2013;5:34–53. doi: 10.3389/fnagi.2013.00034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Dewey KG, Domellöf M, Cohen RJ, Landa Rivera L, Hernell O, Lönnerdal B. Iron supplementation affects growth and morbidity of breast-fed infants: Results of a randomized trial in Sweden and Honduras. J Nutr. 2002;132(11):3249–55. doi: 10.1093/jn/132.11.3249 [DOI] [PubMed] [Google Scholar]
- 13.Agrawal S, Berggren KL, Marks E, Fox JH. Impact of high iron intake on cognition and neurodegeneration in humans and in animal models: A systematic review. Nutr Reviews. 2017. Jun 1;75(6):456–70. doi: 10.1093/nutrit/nux015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hare DJ, Arora M, Jenkins NL, Finkelstein DI, Doble PA, Bush AI. Is early-life iron exposure critical in neurodegeneration? Nature Reviews Neurology. 2015;11(9):536–45. doi: 10.1038/nrneurol.2015.100 [DOI] [PubMed] [Google Scholar]
- 15.Lozoff B, De Andraca I, Castillo M, Smith J, Walter T, & Pino P Behavioral and developmental effects of preventing iron-deficiency anemia in healthy full-term infants. Pediatrics. 2003;112:846–54. http://pediatrics.aappublications.org/content/112/4/846.full.html [PubMed] [Google Scholar]
- 16.Sutton GP, Barchard KA, Bello DT, Thaler NS, Ringdahl E, Mayfield J, Allen DN. Beery-Buktenica Developmental Test of Visual-Motor Integration performance in children with traumatic brain injury and attention-deficit/hyperactivity disorder. Psych Assessment. 2011;23(3):805–9. doi: 10.1037/a0023370 [DOI] [PubMed] [Google Scholar]
- 17.Sánchez-Cubillo I, Periáñez JA, Adrover-Roig D, Rodríguez-Sánchez JM, Rios-Lago M, Tirapu JE, Barcelo F. Construct validity of the Trail Making Test: Role of task-switching, working memory, inhibition/interference control, and visuomotor abilities. J Int Neuropsych Soc. 2009;15(3): 438–50. doi: 10.1017/S1355617709090626 [DOI] [PubMed] [Google Scholar]
- 18.Mahurin RK, Velligan DI, Hazleton B, Mark Davis J, Eckert S, Miller AL. Trail making test errors and executive function in schizophrenia and depression. Clin Neuropsychologist. 2006;20(2):271–88. doi: 10.1080/13854040590947498 [DOI] [PubMed] [Google Scholar]
- 19.Axelrod BN, Goldman RS, Heaton RK, Curtiss G, Thompson LL, Chelune GJ, Kay GG. Discriminability of the Wisconsin Card Sorting Test using the standardization sample. J Clin Exper Neuropsychology. 1996;18(3):338–42. doi: 10.1080/01688639608408991 [DOI] [PubMed] [Google Scholar]
- 20.Levine DS. Neural network models of human executive function and decision making. In Goldberg E (Ed.), Executive functions in health and disease (pp. 105–128). Academic Press. 2017. doi: 10.1016/B978-0-12-803676-1.00005-2 [DOI] [Google Scholar]
- 21.Mendez-Castellano H, de Mendez MC. (1986). Social stratification and human biology: Graffar’s modified method. Venezuelan Arch Childcare Pediatr. 1986;49(3/4):93–104. [Google Scholar]
- 22.Elardo R, Bradley RH. The Home Observation for Measurement of the Environment (HOME) Scale: A review of research. Dev Review. 1981. Jun 1;1(2):113–45. doi: 10.1016/0273-2297(81)90012-5 [DOI] [Google Scholar]
- 23.Kopp B, Rösser N, Tabeling S, Stürenburg HJ, de Haan B, Karnath HO, Wessel K. Errors on the Trail Making Test are associated with right hemispheric frontal lobe damage in stroke patients. Behavioural Neurology. 2015. Oct;1–10. doi: 10.1155/2015/309235 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Stuss DT, Levine B, Alexander MP, Hong J, Palumbo C, Hamer L, Murphy KJ, Izukawa D. Wisconsin Card Sorting Test performance in patients with focal frontal and posterior brain damage: Effects of lesion location and test structure on separable cognitive processes. Neuropsychologia. 2000;38(4): 388– 402. doi: 10.1016/S0028-3932(99)00093-7 [DOI] [PubMed] [Google Scholar]
- 25.American Academy of Pediatrics Committee on Nutrition. Iron. In Pediatric Nutrition: Policy of the American Academy of Pediatrics, 7th ed. Kleinman RE, Greer FR, Eds. American Academy of Pediatrics: Elk Grove Village, IL USA. 2019. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.