The shows background characteristics of the participants, including iron status in infancy and at 10 years. The mean age at test was 10.2 years for both the FIDA and control groups (range 9.7 to 11.0 years). All participants had normal or corrected-to-normal vision; two participants (both in the control group) were left-handed. Only one child had IDA at 10 years, defined by age-appropriate NHANES II and III cutoffs24,25
as Hb < 112 g/l and two or more abnormal iron measures (mean corpuscular volume <76 fl, erythrocyte protoporphyrin >70 μg/dl red blood cells [1.24 nmol/l], and serum ferritin <10 μg/l), and four children had iron deficiency without anemia, resulting in < 5% overall prevalence of iron deficiency.
Due to the block design used in task administration, accuracy as indexed by d’ was analyzed using a 2 Condition (d’ lag 1, d’ lag 4, OR d’ lag 2, d’ lag 5) × 2 Group (controls, FIDA) RM ANOVA. Because not every subject had d’ scores and relevant covariate information for all four blocks, n = 40 for controls and n = 47 for FIDA for the d’ analysis of the first two blocks and controls = 39 and FIDA = 48 for d’ analysis of the last two blocks. After controlling for the background factors that met criteria for the inclusion in the model— maternal depression and SES— analysis of the first two blocks of trials (d’ lag 1, d’ lag 4) revealed no significant effects. After controlling for stress, analysis of the second two blocks of trials (d’ lag 2, d’ lag 5) revealed a main effect of Accuracy, F(1,84) = 11.959, p = .001, whereby d’ for lag 2 (M = 2.66, SD = .87) was significantly larger than d’ for lag 5 (M = 2.44, SD = .86). Of note, there was no Group × Accuracy interaction for either set of trials, where F(1,83) = 0.926, p = .339 for the first two blocks of trials and F(1,84) = 2.151, p = .146 for the second two blocks of trials.
For reaction time analysis, n = 41 for controls and n = 50 for FIDA for the first two runs and controls n = 42 and FIDA n = 49 for the analysis of the last two runs. Analysis of reaction time data for the first two blocks of trials (new words, lag 1, lag 4) revealed a main effect of Group, F(1,89) = 4.171, p = .044, whereby the controls (M = 1003ms, SD = 205) showed significantly faster reaction times than the FIDA group (M = 1085ms , SD = 176). Additionally, there was a main effect of Condition, F(2,88) = 18.355, p < .001, whereby reaction times for new words (M = 998ms, SD = 189) were significantly faster than those for words presented after a lag of 1 (M = 1076, SD = 206) or a lag of 4 (M = 1070, SD = 213). Analysis of the second two blocks of trials (new, lag 2, lag 5) revealed no significant effects. Importantly, there was no Group X Condition interaction in either block of trials, where F(2,89) = 0.463, p = 0.576 for the first blocks and F(2,86) = 0.636, p = .481 for the second blocks of trials.
Preliminary analyses were conducted in order to determine if there were any electrophysiological differences between lags 1, 2, 4, and 5. Participants were included in this initial set of analyses only if they contributed usable data for both lag 1, 4 and lag 2, 5 trials (25 controls, 29 FIDA). Analyses of each ERP component revealed no significant differences in amplitude or latency between any of the lag conditions. However, old/new effects were evident. All subsequent analyses were collapsed across condition.
Participants who did not meet the minimum requirement of 10 trials for each of the three frontal electrodes (Fz, F3, and F4) were excluded from analysis of the FN400 component. Participants were also excluded if they had missing values for any covariate that met inclusion criteria for a given analysis. Therefore, controls n = 40 and FIDA n = 48 for amplitude analyses and controls n = 40 and FIDA n = 49 for latency analyses.
After controlling for the three background factors that met criteria for inclusion in the analysis—sex, SES and maternal depression—there was a main effect of Electrode, F(2,82) = 4.027, p = .024, whereby Fz (M = −24.4 μV, SD = 12.8) showed more negative amplitudes than F3 (M = −22.4 μV, SD = 11.9) and F4 ( M = −19.8 μV, SD = 12.2). There was also a Condition × Group interaction for FN400 peak amplitude, F(1,82) = 3.918, p = .051. Post-hoc tests indicated that for controls only, new words (M = −11.7 μV, SD = 6.0) elicited more negative overall amplitude than old words (M = −10.2 μV, SD = 6.6) ().
Figure 1 Topographic distribution of scalp potentials to new and old words for A, controls and B, FIDA at 344 milliseconds poststimulus. Grand averaged ERP waveforms for C, controls and D, FIDA over the left frontal electrode (F3). The x-axis represents latency (more ...)
Both HOME score and maternal depression met criteria for inclusion in the analyses of FN400 latency. After controlling for these covariates, there was a main effect of Group, F(1,85) = 9.652, p = .003, whereby controls (M = 360.0ms, SD = 31.2) showed significantly faster overall latency than the FIDA group (M = 383.7ms, SD = 40.2) (). In addition, there was a main effect of Condition (), F(1,85) = 4.841, p = .031, whereby old words (M = 368.8ms, SD = 40.4) elicited significantly faster FN400 latency than new words (M = 378.1ms, SD = 43.8).
Mean latency to peak amplitude of the FN400 component, collapsed across condition. The x-axis represents ‘group’ and the y-axis represents latency in milliseconds (ms). Error bars represent +1 standard error of the mean.
Participants who did not meet the minimum requirement of 10 trials for each of the three midline electrodes (Fz, Cz, and Pz) were excluded from analysis of the P300 component. Therefore, controls n = 42 and FIDA n = 50. After controlling for maternal education and gestational age, there was a main effect of Group for P300 mean amplitude, which was qualified by a Group × Electrode interaction, F(2,87) = 4.878, p = .030. Post hoc comparisons revealed that controls showed larger P300 mean amplitudes than FIDA at electrodes Fz and Cz ().
Figure 3 Topographic distribution of scalp potentials collapsed across memory condition for A, controls and B, FIDA at 699 milliseconds poststimulus. C, Grand averaged ERP waveforms for the mean amplitude of the P300 component collapsed across condition. The (more ...)