An Event-Related Potential Study of Attention and

Recognition Memory in Infants With Iron-

Deficiency Anemia

Matthew J. Burden, PhDa, Alissa J. Westerlund, BAb, Rinat Armony-Sivan, PhDc, Charles A. Nelson, PhDb, Sandra W. Jacobson, PhDa, Betsy Lozoff, MDd, Mary Lu Angelilli, MDe, Joseph L. Jacobson, PhDa,f

PEDIATRICS Volume 120, Number 2, August 2007 e337

ABSTRACT

OBJECTIVES. The purpose of this work was to determine whether iron-deficiency

anemia in infancy represents a risk factor for deficits in attention and memory

development using event-related potentials.

METHODS. Artifact-free event-related potential data were obtained at 9 and/or 12

months from 15 infants with iron-deficiency anemia and 19 who were iron

sufficient during a test of the infant’s ability to discriminate a highly familiar

stimulus, the mother’s face, from a stranger’s face.

RESULTS.A midlatency negative component associated with attention and a lateoccurring

positive slow wave associated with memory updating were identified at

both ages in the iron-deficiency anemia and iron-sufficient groups. Consistent

with the age-appropriate pattern of development at 9 months, the iron-sufficient

group showed a greater attentional response (negative component) to the mother

and a greater updating of memory for the stranger (positive slow wave). This

pattern of responses was not evident in the iron-deficiency anemia group until 12

months, suggesting a delay in cognitive development.

CONCLUSIONS. These data suggest that iron-deficiency anemia adversely affects the

allocation of neurophysiologic resources to attention and recognition memory

during the processing of information about familiar and unfamiliar stimuli. This

delay in cognitive development may reflect alterations in efficiency of central

nervous system functions that seem related to early iron deficiency.

IRON-DEFICIENCY ANEMIA (IDA) affects an estimated 1 to

2 billion people worldwide, primarily women and

children.1 Although the prevalence of IDA and iron deficiency

(ID) has markedly declined in US infants in the

last 30 years, poor, minority, and/or immigrant infants

and toddlers remain at increased risk.2 Infants with IDA

or other indications of chronic, severe ID have shown

lower cognitive test scores than infants with iron sufficiency

(IS) in all but 1 of 14 studies from countries

around the world.3,4 All of the available follow-up studies,

with intervals following iron therapy in infancy

ranging up to young adulthood, report persisting lower

scores (see review in ref 5). A recent meta-analysis estimated

the long-term effects on intelligence quotient to

be 1.7 points lower for each 10 g/L decrease in hemoglobin

(Hb).6 These lower scores on intelligence quotient

tests, though consistently observed, give little indication

of the neural mechanisms that underlie the cognitive

deficits.

Assessing specific neurocognitive functions with

brain-based measures in human infants is challenging.

Event-related potentials (ERPs) provide a noninvasive

means for measuring transient changes in the brain’s

electrical activity in response to stimuli, allowing the

evaluation of attention and memory in very young infants.

ERP studies informed by animal models of dietinduced

ID have shown impaired recognition memory in

human infants with prenatal ID because of maternal

diabetes during pregnancy.7–9 The study reported here

examined infant recognition memory in relation to IDA

during its period of peak postnatal prevalence (6–24

months). This study was part of an integrated crossspecies

program project on the brain and behavior effects

of early ID in human infants, nonhuman primate infants,

and developing rodents. As part of a hypothesisdriven

neurodevelopmental approach, the hippocampus

and/or related behaviors were assessed in all 3 species.

Recognition memory was assessed using ERP in the human

infant project.

We hypothesized that postnatal IDA would alter attention

and recognition memory because of the effect of

iron on hippocampal development and function. To examine

this hypothesis, we recorded ERPs in a procedure

specifically focused on the infant’s ability to recognize a

highly familiar stimulus, the mother’s face, and to discriminate

the mother’s face from a stranger’s face. Two

ERP components of particular interest have been identified

in studies of infant memory: a midlatency negative

component (NC) associated with attention and a lateoccurring

positive slow wave (PSW) associated with

memory updating. Previous research using this paradigm

indicates that infants respond differentially to photographs

of their mother versus a stranger, showing a

larger NC to mother and a larger PSW to stranger at 6

months of age.10 Developmental changes in NC and PSW

responses continue over the course of infancy into early

childhood and are often characterized by shifts in response

patterns.11,12 These changes are complex and

nonlinear, presumably reflecting age-related changes in

the allocation of attentional resources and interest.13

In this study, we compared NC and PSW in infants

with IDA versus those who were IS at 9 and 12 months.

We predicted that infants with IDA would show ERP

responses that were less developmentally appropriate

than those of IS infants.

METHODS

Participants

Black infants and caregivers were recruited for the study

during routine 9-month visits to Children’s Hospital of

Michigan, between April 2002 and August 2005, during

which a complete blood cell count was taken.14 Because

_90% of the clinic population was black, and the remaining

population was of varying ethnic origins, recruitment

was restricted to blacks. All of the methods

and procedures for this study were approved by the

Wayne State University and University of Michigan human

investigation committees. Written informed consent

was obtained from each infant’s mother or primary

caregiver. Based on hematologic measures from infants

screened in the first year of the study, the prevalence of

ID in this population ranged from 2.5% to 14.4%, depending

on the criteria used.14 Of 881 infants screened,

408 were not included because they met _1 of the

following exclusionary criteria: ethnicity other than

black (73); birth weight _2.5 kg (130); maternal diabetes

or birth weight _4.0 kg (67); multiple birth (29);

perinatal complications (85); hospitalization more than

once or _5 days (108); chronic health problem (60);

medicinal iron (22); in foster care (17); or mother _18

years old (43). The mothers of the qualified infants (n _

473) were asked to consider enrolling their infants in a

study on the neurodevelopmental effects of ID to be

conducted at the Child Development Research Laboratory,

Wayne State University, at ages 9 and 12 months.

For those willing to consider participation (n _ 412), an

extra tube of blood was drawn for additional iron-status

analyses. Infants and their mothers were invited to participate

in the study based on provisional criteria related

to infant iron status (B.L., unpublished data, 2007).

These criteria were considered provisional because not

all of the ID indicators were available before the

9-month neurodevelopmental assessment. Among the

242 infants invited to participate at 9 months, 113

healthy, term black infants (47%) participated in the

final neurodevelopmental assessment sample. Of these,

87 (77%) returned for 12-month neurodevelopmental

testing. Iron supplements were provided to all of the

infants in the study. However, the study cannot determine

the effects of iron therapy on 12-month infant

function because of uncertainty about the degree to

which the iron was actually administered to the infants

and the relatively small number of infants for whom

blood work was available at 12 months (58%) (B.L.,

unpublished data, 2007).

The analyses presented here compared infants with

IDA to infants with IS. IDA was defined as Hb _105 g/L

with abnormal scores on _2 measures of the following

indicators of ID: mean corpuscular volume _74 fL, red

cell distribution width _14%, zinc protoporphyrin/

heme ratio _69 _mol/mol of heme, ferritin _12 _g/L,

and transferrin saturation _12%. The comparison group

of infants with IS had Hb _115 g/L and _1 abnormal

iron indicator. One infant with IS and 3 infants with IDA

were not assessed with ERP because of a refusal to wear

the electrode cap or equipment malfunction. Artifactfree

ERP data were obtained from more than half of the

infants who were assessed, a proportion that compares

favorably with other infant studies using this methodology.

15 Thirteen (48%) of 27 infants with IDA and 15

(60%) of 25 infants with IS provided artifact-free ERP

data at 9 months (_1

2 _ .73; P _ .39, not significant), and

9 (43%) of 21 with IDA and 11 (61%) of 18 with IS

provided artifact-free data at 12 months (_1

2 _ 1.28; P _

.26, not significant). Table 1 summarizes the Hb and iron

indicators for the 2 groups. To consider the impact of

IDA on development across this age period, we first

report results for the 14 infants (7 IDA and 7 IS) who

provided artifact-free ERP data at both the 9- and 12-

month assessments. We then provide results for the

larger number of infants at each age to corroborate the

longitudinal findings.

Electroencephalogram Collection and Procedure

ERPs are recorded at the scalp and are time locked to a

particular event (eg, a visual stimulus), reflecting summated

postsynaptic potentials from synchronously firing

neurons.16 The amplitude of the ERP signal reflects the

population of active neurons, and latency reflects the

timing of that neuronal activation. Two of the most

important components identified in ERP studies of infant

memory are an NC occurring _300 to 800 milliseconds

after stimulus onset, which is characterized by a welldefined

negative peak (millivolts) with maximum amplitude

over frontocentral scalp and a PSW occurring

_1000 milliseconds after stimulus onset, which rises

steadily over the course of several hundred milliseconds

and is measured in arbitrary units based on voltage _

time.17,18 The NC has been characterized as an obligatory

attentional response,19–21 which differentiates between a

familiar and unfamiliar stimulus.10,22 PSW is believed to

reflect the extent to which a stimulus is encoded and

subsequently updated in memory.10,22,23

In this study, ERPs were recorded from 16 scalp electrodes

mounted in a close-fitting cap (Electro-Cap International,

Eaton, OH) using the 10–20 system.24 The

electrodes comprised midline (Fz, Cz, and Pz), frontal

(F3 and F4), central (C3 and C4), temporal (T3, T4, T5,

and T6), parietal (P3 and P4), and right occipital (O2)

scalp sites plus the left and right mastoids (M1 and M2)

and a ground electrode. Cz was the reference lead during

acquisition. Electrooculogram (EOG) was recorded from

bipolar miniature electrodes placed vertically above and

below the right eye for the purpose of artifact detection.

Impedance for all of the scalp and EOG electrodes was

kept below 10 kOhm. Electroencephalogram and EOG

were acquired using a Model 15 Grass Neurodata Acquisition

System (Grass Instruments, Quincy, MA) and amplified

at a gain of 20 000 for scalp leads and 5000 for

EOG. The bandpass was 0.1 to 30 Hz, and a 60-Hz notch

filter was engaged. Data were sampled every 5 milliseconds

(200 Hz). ERP data were digitized online and then

edited by computer algorithm. Before averaging, trials

with excessive artifact (ie, electroencephalogram more

than _125 mV) were rejected. Data were then rereferenced

to an averaged mastoid reference, and eye movement-

related artifact was corrected.25 Using 100 milliseconds

before stimulus onset as the baseline, individual

trials were baseline corrected and then averaged for each

participant within each stimulus type.

In the mother-stranger paradigm, the infant was presented

with randomly alternating digital photographs

repeated with equal probability of his/her mother’s face

and an unfamiliar stranger’s face. The stranger was the

mother of another infant participating in the study, with

whom the infant was not familiar. The infant was seated

on the mother’s lap _60 cm from a 13-in computer

monitor; the faces on screen were 10-cm wide (visual

angle: 9.5°) and 15-cm high (visual angle: 14°). Stimuli

were presented for 500 milliseconds with randomly alternating

intertrial intervals, ranging between 500 and

1000 milliseconds; ERP data were collected for 1500

milliseconds after stimulus onset for each trial. To maximize

processing of as many stimuli as possible, an experimenter

observed the infant from behind a screen

and paused whenever the infant looked away from the

screen, repeating trials as necessary. A maximum of 100

total trials was presented to the infant (50 trials for

mother and 50 for stranger), with an equal number of

trials for each stimulus (see Table 2 for more information).

The NC in this study occurred between 300 and 650

milliseconds after stimulus onset; the PSW, at 800 to

1500 milliseconds. The NC and PSW were most prominent

over frontocentral electrodes (Fz, F3, F4, Cz, C3,

and C4); thus, the analyses focused on overall frontocentral

scalp activity. Data were included in the ERP

analyses if they met all of the following criteria: (1) _10

good trials for each condition (ie, mother and stranger);

(2) activity at mastoids not exceeding _10 _V; (3) data

did not appear contaminated by EOG or mastoids; and

(4) activity at no channel exceeding _125 _V (_250 _V

for EOG). These criteria were designed to eliminate outliers

and other questionable values and were applied in

a review of individual ERP waveforms by investigators

who were blind with respect to iron status. After removing

trials not meeting these criteria, brain activity at each

electrode was averaged over an equal number of trials

for mother and stranger. If there was a discrepancy in

the number of good trials between mother and stranger

conditions, the excess trials from 1 condition were randomly

removed by the computer program to equalize

the number of trials between conditions.

Control Variables

The IDA and IS groups were compared on the following

sociodemographic variables (Table 1): infant gender

and age; age of primary caregiver (mother); socioeconomic

status assessed on the Hollingshead Scale

(A.B. Hollingshead, PhD, Four-Factor Index of Social

Status, unpublished manual, 1975); the Home Observation

for Measurement of the Environment,26 a semistructured

interview that assesses quality of intellectual

stimulation and emotional support provide by the

caregivers; maternal education (highest grade); 2

measures of maternal intellectual competence, the

Peabody Picture Vocabulary Test-Revised27 and the

nonverbal Raven Standard Progressive Matrices28; maternal

depression assessed in the Beck Depression Inventory29;

breastfeeding (present/absent); and

whether the infant was taking formula at the time of

study entry. Only 2 (5.9%) of the mothers (1 from

each group) had begun to give their children cow’s

milk when asked at the 9-month assessment.

Data Analysis

Table 2 shows that the IS and IDA groups were very

similar at both ages in number of ERP trials viewed,

number of trials available for rereferencing, and number

of trials cross-averaged by condition. Data for F4

were missing for 1 infant in the IDA group at 9 months

because of failure of that electrode during the session.

Because the other electrodes were all acceptable for

that infant, data for F4 were imputed from Fz and F3

based on the high proportion of variance accounted

for by those electrodes in relation to F4 for both

mother and stranger conditions for NC and PSW in the

sample as a whole, (R2 range: 0.65– 0.85; all P values

_ .0001). Peak amplitude and corresponding latency

of NC and area of PSW were analyzed by 2 _ 2 _ 2 _

6 repeated-measures analysis of variance. The between-

subjects factor was group (IDA and IS), and the

within-subjects factors were condition (mother and

stranger), age (9 and 12 months), and electrode (Fz,

F3, F4, Cz, C3, and C4).

The IDA and IS groups were similar in sociodemographic

background, except that the infants with IDA

were slightly younger at study entry, and their mothers

had fewer years of education and were more depressed

(see Table 1). For the 14 infants with artifactfree

data at both ages in the longitudinal analyses,

there was no difference in maternal education (mean:

12.1 vs 12.3 years for IDA versus IS, respectively; t12 _

0.49; P _ .80). However, the age difference at study

entry (ie, the 9-month assessment) showed a suggestive

trend for these 14 infants (mean: 9.3 vs 9.8

months for IDA versus IS, respectively; t12 _ 2.07; P _

.06), and more maternal depression (Beck Depression

Inventory total score) was evident in the IDA group

(mean: 12.0 vs 2.4 for IDA versus IS, respectively; t12

_ 2.38; P _ .035). Further examination of these maternal

depression data revealed that 3 of the 7 mothers

in the IDA group had total scores ranging from 17 to

28, which is classified as “moderately depressed,”

whereas the next highest score from any mother from

either group was only 5. Given the limited degrees of

freedom available in this small longitudinal sample,

the potential confounding effects of age at study entry

and maternal depression were examined in separate

repeated-measures analyses of covariance.

RESULTS

Longitudinal Analyses

NC

For NC amplitude across the 6 frontocentral electrodes,

there was a significant 3-way interaction of age _ iron

group _ condition (F1,12 _ 7.73; P _ .017; see Fig 1).

Results were essentially the same in the analyses of

covariance controlling for age at study entry (F1,11 _

5.00; P _ .047) and for maternal depression (F1,11 _

6.82; P _ .024). To better understand this 3-way interaction,

condition _ iron group effects were tested separately

at 9 and 12 months. At 9 months, there was a

significant condition _ iron group interaction (F1,12 _

6.61, P _ .024). The pattern at 9 months, as shown in Fig

1, was for a larger NC to mother in the IS group but

larger NC to the stranger in the IDA group. A posthoc

pairwise comparison test revealed that this mean difference

in NC to the mother versus stranger was significant

for the IS group (mean difference: 4.14 _V; SE: 1.23; F1,6

_ 11.36; P _ .015), though not for the IDA group, in

which there was more within-group variability (mean

difference: 6.07 _V; SE: 3.78; F1,6 _ 2.58; P _ .159). At

12 months, the response pattern was the reverse of that

seen at 9 months; that is, the IDA group showed a larger

NC to the mother at 12 months, and the IS group shifted

to a larger NC to the stranger (see Fig 1). There were no

main effects of age, iron group, or condition or other

significant interactions among these variables. There

were no effects for latency to NC by condition, iron

group, or age.

PSW

There was also a significant 3-way interaction for age _

iron group _ condition for PSW area (F1,12 _ 45.01; P _

.001; see Fig 2). The same pattern of results was found in

the analyses of covariance controlling for age at study

entry (F1,11 _ 43.58; P _ .001) and for maternal depression

(F1,11 _ 24.51; P _ .001). At 9 months, there was a

significant condition _ iron group effect (F1,12 _ 20.27;

P _ .001), with the IS group showing a larger PSW to the

stranger than to the mother compared with the IDA

group, which showed the opposite pattern. A posthoc

pairwise comparison test revealed that this mean difference

in PSW fell short of statistical significance for the IS

group (mean difference: 2628; SE: 1415; P _ .113) but

was significant for the IDA group (mean difference:

6145; SE: 1340; P _ .004). At 12 months, there was a

significant condition _ iron group interaction (F1,12 _

7.32; P _ .019), which revealed a pattern opposite to

that seen at 9 months; that is, the IDA group showed a

significantly larger PSW to the stranger than to the

mother, and the IS group showed the reverse pattern

(see Fig 2). A posthoc pairwise comparison at 12 months

showed that this difference was significant for the IDA

group (mean difference: 3909; SE: 1019; F1,6 _ 14.71; P

_ .009) but not for the IS group (mean difference: 2226;

SE: 2025; F1,6 _ 1.21; P _ .314).

Cross-sectional Analyses at Each Age

We also performed cross-sectional analyses including

subjects with data available at only 1 age (ie, 13 IDA vs

15 IS at 9 months; 9 IDA vs 11 IS at 12 months). The

effects seen with these larger numbers were similar to

those presented in the longitudinal analyses above. The

ERP waveforms for these data are shown at 9 months

(Fig 3) and 12 months (Fig 4). For NC amplitude at 9

months, there was a suggestive trend for the condition _

iron group interaction seen in the longitudinal subgroup

at 9 months (F1,26 _ 3.2; P _ .087). As in the longitudinal

subgroup, there were no effects for NC latency by

condition or iron group. For PSW area at 9 months, the

condition _ iron group interaction (F1,26 _ 12.1; P _

.002) was similar to the one seen in the longitudinal

subgroup. Posthoc tests revealed that the IDA group

showed a larger PSW to the mother than to the stranger

(mean: 6057 vs 1662; F1,12 _ 7.5; P _ .018), whereas

there was a suggestive trend for the opposite pattern in

the IS group (mean: 6437 vs 4702; F1,14 _ 4.0; P _ .066).

There was also a main effect for differences by electrode

(F5,22 _ 3.5; P _ .017), suggesting some random variability

across the electrodes. Therefore, we conducted

additional analyses comparing only frontal electrodes

(Fz, F3, and F4) versus only central (Cz, C3, and C4)

electrodes. The results showed that the condition _ iron

group interaction was evident both for frontal (F1,26 _

11.6; P _ .002) and central electrodes (F1,26 _ 7.5; P _

.011), suggesting that collapsing across frontocentral

electrodes to create a region of interest was valid despite

some variability across these electrodes.

For NC amplitude at 12 months, there was a significant

condition _ iron group interaction (F1,18 _ 5.2; P _

.035), which resembled the pattern seen in the longitudinal

subgroup at 12 months. For PSW area at 12

months, the condition _ iron group interaction showed

a suggestive trend (F1,18 _ 4.0; P _ .060), with results

resembling the pattern seen in the longitudinal subgroup.

A posthoc analysis for the IDA group revealed

significantly less PSW to the mother than to the stranger

at 12 months across frontocentral electrodes (mean:

4781 vs 8091; F1,8 _ 12.6; P _ .008), consistent with the

longitudinal findings. No differences were found in this

regard for the IS group. Thus, the pattern of effects

emerging overall with the larger sample sizes at 9 and 12

months supports the longitudinal findings presented

above.

DISCUSSION

A number of consistent patterns associated with normal

development have emerged in the infant ERP literature.

Seminal studies from de Haan and Nelson10,22 have revealed

a larger NC amplitude in response to the mother’s

face as compared with a stranger’s face at 6 months of

age, indicating that around this age infants discriminate

between faces and allocate relatively more attentional

resources to the mother’s face. This pattern was seen in

the IS infants at 9 months, but the infants with IDA did

not show the normal developmental response for young

infants to attend more to the mother’s face. At 12

months, there was a significant reversal of the NC amplitude

pattern, with the NC of the IS group now larger

to the stranger than to the mother and the IDA group

showing the reverse pattern.

Although the explanation for this reversal is unclear,

similar phenomena have been observed in ERP studies

somewhat later in development, extending from infancy

to 3 years of age.12,30 Findings from Carver et al12 suggest

that the larger NC amplitude to mother at the younger

age, as seen in our study, may reflect the greater salience

or importance of the mother as an attachment figure at

that point in development. The reversal seen in the IS

group in our data at 12 months may reflect rapidly

evolving infant interests and changes in attention, which

could occur dramatically in response to emerging awareness,

such as stranger and separation anxiety. The increased

fear of strangers is likely to have an effect on the

allocation of attentional resources when the infant is

confronted with the face of a stranger, which could

account for the increased NC amplitude in response to

the stranger’s photograph in the IS infants at 12 months.

It is possible that IDA infants did not show the expected

greater attentional response to the mother at the earlier

age because of their tendency to be more wary, fearful,

and hesitant in unfamiliar circumstances and/or with

strangers.31 It is also possible that infants with IDA

showed a developmentally delayed response in focusing

differentially on the mother, because they responded at

12 months much in the same manner as the infants with

IS at 9 months.

A second important element of ERP associated with

infant memory processes is the late-occurring PSW,

which is believed to reflect processing of a partially encoded

stimulus in memory.23 Six-month-old infants typically

show a larger PSW to the stranger’s face in comparison

with the mother’s face, which likely reflects the

updating of a memory trace for an unfamiliar stimulus.22

Consistent with these findings, infants with IS showed

the expected developmental pattern at 9 months, with a

larger PSW to strangers. Infants with IDA did not show

this pattern until 12 months, suggesting a delay in their

cognitive development. Although we are aware of no

other ERP study using a visual mother-stranger paradigm

to assess effects of IDA, auditory recognition deficits have

 been found in studies of infants of diabetic

mothers at high risk for prenatal brain ID.32,33 For example,

using an auditory ERP paradigm to assess recognition

of the mother’s versus a stranger’s voice, Siddappa

et al32 found that infants of diabetic mothers who were

presumed to have brain IS displayed a negative slow

wave to the stranger’s voice, whereas those who were

presumed to have brain ID did not discriminate between

the mother’s and stranger’s voices. In our study, infants

in both groups seemed capable of discriminating between

the stimuli, but the patterns of ERP activity in

response to the mother and stranger suggested a delay in

cognitive development in the infants with IDA. In light

of findings from previous studies showing recognition

memory in infancy to be predictive of cognitive and

language deficits in childhood,34,35 the developmental

delay in recognition memory seen here may be an early

indicator of later cognitive problems. Such a finding

would be consistent with previous research relating IDA

in infancy to poorer neurobehavioral function in childhood.

The mechanisms underlying this delay in cognitive

development associated with IDA are not clear, but there

is an increasing body of relevant research in animal

models. Several important developing central nervous

system processes, such as myelination, dendritogenesis,

synaptogenesis, and neurotransmission, are highly dependent

on iron-containing enzymes and hemoproteins.

36 Recent studies on ID and brain development

have demonstrated short- and long-term effects on myelination,

neuroanatomy, neurotransmitter function,

and neurometabolism, especially in the striatum and the

hippocampus.5 The hippocampus is of particular interest

given its role in the discrimination of novel from familiar

stimuli in recognition memory,37 the domain found here

to be affected in infants with IDA. Moreover, studies in

rodent models of gestational and/or lactational ID provide

evidence of reduced neuronal metabolism (as indexed

by cytochrome c oxidase activity38) and altered

dendritic structure in the hippocampus,39 with associated

poorer performance on spatial memory tasks40,41 and on

highly specific dorsal hippocampal tasks, such as trace

conditioning.42

One strength of ERP is its capacity to reveal subtle

developmental differences in neurophysiologic processes

underlying cognition that would not be detected by behavioral

observation. Thus, the results of this study suggest

a delay in cognitive development among infants

with IDA in processing information about mother and

stranger that was not clearly evident in their overt behavior.

Given that the same pattern of findings was seen

after controlling for infant age at study entry and maternal

depression (ie, the sociodemographic variables on

which the IDA and IS groups differed), we infer that the

delay in cognitive development associated with IDA is

more likely attributable to biological insult than to environmental

factors, though this inference is necessarily

tentative given the small sample size. Another limitation

to this study is our inability to assess the effectiveness of

iron treatment for infants with IDA because of incomplete

hematologic data at the 12-month assessment resulting

from attrition and/or compliance problems. Additional

studies are needed to address these issues and to

assess the degree to which this delay in cognitive development

may presage adverse effects on subsequent cognitive

and behavioral development in children with IDA

in infancy.

ACKNOWLEDGMENTS

This work was supported by a program project grant

from the National Institutes of Health (P01 HD39386,

Brain and Behavior in Early Iron Deficiency, Dr Lozoff,

principal investigator), a National Institutes of Health

National Research Service Award to Dr Burden, and a

grant from the Joseph Young, Sr, Fund from the State of

Michigan to Dr Jacobson.

We are grateful to the study families for their participation;

to our colleague John Beard (Pennsylvania State

University) for performing additional iron-status measures

and advising on their interpretation; to our collaborator

Rosa Angulo-Barroso for contributions to the design

of the study; and to Tal Shafir, Renee Sun, Jigna

Zatakia, Margo Laskowski, Brenda Tuttle, Douglas

Fuller, Agustin Calatroni, and Yuezhou Jing for work in

the data collection and analyses. All investigators in the

Brain and Behavior in Early Iron Deficiency Program

Project contributed to our thinking about early hippocampal

insults.

REFERENCES

1. World Health Organization/United Nations Children’s Fund/

United Nations University. Iron Deficiency Anemia: Prevention,

Assessment and Control: Report of a Joint WHO/UNICEF/UNU Consultation.

Geneva, Switzerland: World Health Organization;

1998

2. Brotanek JM, Halterman J, Auinger P, Flores G, Weitzman M.

Iron deficiency, prolonged bottle-feeding, and racial/ethnic disparities

in young children. Arch Pediatr Adolesc Med. 2005;159:

1038–1042

3. Grantham-McGregor S, Ani C. A review of studies on the effect

of iron deficiency on cognitive development in children. J Nutr.

2001;131:649S–668S

4. Lozoff B, Georgieff MK. Iron deficiency and brain development.

Semin Pediatr Neurol. 2006;13:158–165

5. Lozoff B, Beard J, Connor J, Barbara F, Georgieff M, Schallert

T. Long-lasting neural and behavioral effects of iron deficiency

in infancy. Nutr Rev. 2006;64:S34–S43

6. Stoltzfus RJ, Mullany L, Black RE. Iron deficiency anaemia. In:

Ezzati M, Lopez AD, Rodgers A, eds. Comparative Quantification

of Health Risks: Global and Regional Burden of Disease Attributable

to Selected Major Risk Factors. Geneva, Switzerland: World Health

Organization; 2004

7. deRegnier RA, Nelson CA, Thomas K, Wewerka S, Georgieff,

MK. Neurophysiologic evaluation of auditory recognition

memory in healthy newborn infants and infants of diabetic

mothers. J Pediatr. 2000;137:777–784

8. Nelson CA, Wewerka S, Thomas KM, Tribby-Walbridge S,

deRegnier RA, Georgieff M. Neurocognitive sequelae of infants

of diabetic mothers. Behav Neurosci. 2000;114:950–956

9. Nelson CA, Wewerka S, Borscheid AJ, deRegnier RA, Georgieff

MK. Electrophysiologic evidence of impaired cross-modal recognition

memory in 8-month-old infants of diabetic mothers.

J Pediatr. 2003;142:575–582

10. de Haan M, Nelson C. Recognition of the mother’s face by

six-month-old infants: a neurobehavioral study. Child Dev.

1997;68:187–210

11. Bartrip J, Morton J, De Schonen S. Responses to mother’s face

in 3-week to 5-month-old infants. Br J Dev Psychol. 2001;19:

219–232

12. Carver LJ, Dawson G, Panagiotides H, et al. Age-related differences

in neural correlates of face recognition during the toddler

and preschool years. Dev Psychobiol. 2003;42:148–159

13. Webb SJ, Long JD, Nelson CA. A longitudinal investigation of

visual event-related potentials in the first year of life. Dev Sci.

2005;8605–616

14. Lozoff B, Angelilli M, Zatakia J, Jacobson SW, Calatroni A,

Beard J. Iron status of inner-city African-American infants.

Am J Hematol. 2007;82:112–121

15. Carver LJ, Meltzoff AN, Dawson G. Event-related potential

(ERP) indices of infants’ recognition of familiar and unfamiliar

objects in two and three dimensions. Dev Sci. 2006;9:51–62

16. Allison T, Wood C, McCarthy G. The central nervous system.

In: Coles M, Donchin E, Porges S, eds. Psychophysiology: Systems,

Processes, and Applications. New York, NY: Guilford; 1986:5–25

17. Nelson C, Monk C. The use of event-related potentials in the

study of cognitive development. In: Nelson CA, Luciana M,

eds. Handbook of Developmental Cognitive Neuroscience. Cambridge,

MA: MIT Press; 2001:125–136

18. DeBoer T, Scott L, Nelson CA. Event-related potentials in developmental

populations. In: Handy T, ed. Event-Related

Potentials: A Methods Handbook. Cambridge, MA: MIT Press;

2004:263–297

19. Courchesne E, Ganz L, Norcia A. Event-related potentials to

human faces in infants. Child Dev. 1981;52:104–109

20. Nelson C. Electrophysiological correlates of early memory development.

In: Reese HW, Franzen, MD, eds. Thirteenth West

Virginia University Conference on Life Span Developmental

Psychology: Biological and Neuropsychological Mechanisms. Hillsdale,

NJ: Lawrence Erlbaum; 1996:95–131

21. Richards JE. Attention affects the recognition of briefly presented

visual stimuli in infants: an ERP study. Dev Sci. 2003;6:

312–328

22. de Haan M, Nelson C. Brain activity differentiates faces and

object processing in 6-month-old infants. Dev Psychol. 1999;35:

1113–1121

23. Nelson CA. Neural correlates of recognition memory in the first

postnatal year of life. In: Dawson G, Fischer K, eds. Human

Behavior and the Developing Brain. New York, NY: Guilford Press;

1994:269–313

24. Jasper H. The ten-twenty electrode system of the International

Federation. Electroencephalogr Clin Neurophysiol. 1958;10:

371–375

e344 BURDEN et al

25. Gratton G, Coles MGH, Donchin E. A new method of off-line

removal of ocular artifact. Electroencephalogr Clin Neurophysiol.

1983;55:468–484

26. Caldwell BM, Bradley RH. Home Inventory Administration Manual.

3rd ed. Little Rock, AR: University of Arkansas for Medical

Sciences; 2001

27. Dunn LM, Dunn LM. Peabody Picture Vocabulary Test: Revised.

Circle Pines, MN: American Guidance Service; 1981

28. Raven JC. Standardisation of progressive matrices. Br J Med

Psychol. 1941;19:137–150

29. Beck A, Steer RA. BDI: Beck Depression Inventory Manual. San

Antonio, TX: Psychological Corp; 1987

30. Parker SW, Nelson CA; The Bucharest Early Intervention

Project Core Group. An event-related potential study of the

impact of institutional rearing on face recognition. Dev Psychopathol.

2005;17:621–639

31. Lozoff B, Black M. Impact of micronutrient deficiencies on

behavior and development. In: Pettifor J, Zlotkin SH, eds.

Nutrition-Micronutrient Deficiencies During the Weaning Period and

the First Years of Life. Basel, Switzerland: Karger; 2003

32. Siddappa AM, Georgieff MK, Wewerka S, Worwa C, Nelson

CA, Deregnier RA. Iron deficiency alters auditory recognition

memory in newborn infants of diabetic mothers. Pediatr Res.

2004;55:1034–1041

33. Wewerka S, Thomas K, Tribby-Walbridge S, Georgieff M, Nelson

C. Electrophysiological assessment of pre-explicit memory

in at-risk 6- and 8-month-old infants [abstract]. Pediatr Res

1998;43:223A

34. Rose SA, Feldman JF. Prediction of IQ and specific cognitive

abilities at 11 years from infancy measures. Dev Psychol. 1995;

31:685–696

35. Benasich AA, Tallal P. Auditory temporal processing thresholds,

habituation, and recognition memory over the 1st year.

Inf Behav Dev. 1996;19:339–357

36. Wigglesworth JM, Baum H. Iron dependent enzymes in the

brain. In: Youdim MBH, ed. Brain Iron: Neurochemical and

Behavioural Aspects. New York, NY: Taylor and Francis; 1988

37. Seress L. Morphological changes of the human hippocampal

formation from midgestation to early childhood. In: Nelson

CA, Luciana M, eds. Handbook of Developmental Cognitive Neuroscience.

Cambridge, MA: MIT Press; 2001:45–58

38. DeUngria M, Rao R, Wobken JD, Luciana M, Nelson C, Georgieff

MK. Perinatal iron deficiency decreases cytochrome c

oxidase (CytOx) activity in selected regions of neonatal rat

brain. Pediatr Res. 2000;48:169–176

39. Jorgenson LA, Wobken JD, Georgieff MK. Perinatal iron deficiency

alters apical dendritic growth in hippocampal CA1 pyramidal

neurons. Dev Neurosci. 2003;25:412–420

40. Felt BT, Lozoff B. Brain iron and behavior of rats are not

normalized by treatment of iron deficiency anemia during

early development. J Nutr. 1996;126:693–701

41. Felt BT, Beard JL, Schallert T et al. Persistent neurochemical

and behavioral abnormalities in adulthood despite early iron

supplementation for perinatal iron deficiency anemia in rats.

Behav Brain Res. 2006;171:261–270

42. McEchron MD, Cheng AY, Liu H, Connor JR, Gilmartin MR.

Perinatal nutritional iron deficiency permanently impairs hippocampus-

dependent trace fear conditioning in rats. Nutr Neurosci.

2005;8:195–206