Long-chain polyunsaturated fatty acids and, in particular, docosahexaenoic acid (DHA) are associated with a number of positive effects on maternal and infant health (1). Interest in prenatal exposure to DHA has been fueled by findings showing improved pregnancy outcomes (e.g., gestation duration and birthweight) in both observational studies and randomized clinical trials (2,3,4,5,6). However, it has been hypothesized that prenatal exposure to DHA may also affect later development through fetal programming of the central nervous system and various other physiologic pathways. This possibility is supported by a number of observational studies that associate DHA status during pregnancy with positive long-term effects on the offspring (7,8,9,10,11,12,13). Several randomized trials of maternal DHA supplementation during pregnancy and/or lactation have been conducted with mixed results; the larger trials have not reported advantages for DHA supplementation during the first 18 mo (14,15,16), although one small study found positive effects on problem solving (17). Of studies that followed infants into the preschool period, two reported significant benefits on IQ and neurodevelopmental measures (18,19) while another did not (20). Positive effects of prenatal or postnatal DHA supplementation or status on attention in infancy and early childhood has been documented in several (9,21,22,23), but not all (24) studies. Many of the studies showing null effects of prenatal supplementation have used global standardized tests for evaluating developmental outcome; these tests may not be sensitive to the effects of such supplementation in specific cognitive domains (25).

We report here on the results of a phase III, double-blind, placebo-controlled randomized clinical trial (RCT, registered at as NCT00266825) of infants born to a large sample of mothers prenatally supplemented with DHA. The results of the other primary aims of the study (i.e., compliance, safety evaluation, and pregnancy outcomes) are documented elsewhere; supplementation had many positive effects, including the reduction of high-risk prematurity and increased birth length and weight (2). This study addresses the hypothesis that maternal DHA supplementation can enhance development, in particular visual attention, as assessed in infancy.


Behavioral Measures

Peak look duration yielded a significant main effect for Age, F(2, 167.218 = 11.624, P < 0.001) as look duration during habituation declined from 4 to 9 mos, but no significant main effects or interactions involving DHA Group emerged. The analysis of looks to habituation yielded a marginally significant effect of DHA Group, F(2,165.373) = 3.69, P = 0.056, with infants from supplemented mothers (M = 6.8 looks, SD = 2.83) showing slightly fewer looks to habituation (i.e., slightly faster attainment of habituation) than infants from mothers on placebo (M = 7.31 looks, SD = 3.35).

Heart Rate

Heart rate (HR, expressed in beats/min) data were analyzed with mixed model methods on the prestimulus period (the 2-sec interval prior to stimulus onset), latency (the interval after stimulus onset but prior to the onset of looking), looking, and poststimulus (the 2-sec period after the stimulus has been withdrawn). Data for these phases for Placebo and Supplemented groups at all three ages, with Looking broken into HR-Defined phases of orienting (OR), sustained attention (SA), and attention termination (AT) are shown (see Figure 1 ; definitions and explanations of these three phases of attention are provided in the “Methods” section under “Measures Analyzed”).

Figure 1
figure 1

Infant HR during the various phases of habituation trials analyzed with mixed models. For ease of exposition, values are collapsed across multiple looks during the habituation session. Although the graph shows the well-documented and highly robust changes in HR with age over the first year and the deceleration seen during infant looking while in SA, there are no differences between Supplemented (solid line) and Placebo (dashed line) groups at any point during the trial. The top pair of solid/dashed lines is data from 4-month-olds, the middle pair are from 6-month-olds, and the bottom pair is from 9-month-olds. The downward-pointing arrow represents the onset of the stimulus; the box represents encapsulates the period during which infants were looking at the stimulus. HR, heart rate; SA, sustained attention; PreStim, prestimulus period (before onset of stimulus); OR, orienting; AT, attention termination; Postlook, postlook period (after look is terminated but before withdrawal of stimulus). Data points represent successfully completed habituation sessions where HR data could be successfully coded: n = 159 (n = 72 and n = 87 placebo and supplemented, respectively) at 4 mo, n = 172 (n = 71 and n = 101) at 6 mo, and n = 156 (n = 68 and n = 88) at 9 mo.

PowerPoint slide

In each of the analyses for infant HR during Prestimulus, Latency, Looking, and Poststimulus periods, data revealed expected significant main effects for Age (all P’s < 0.001), which was due to the widely- and previously-reported decline in HR with infant age. No main effects or interaction involving DHA Group emerged at any point for HR analyses across looking during habituation.

HR Defined Phases of Attention

The proportion of time spent in OR increased significantly with age, F(2,176.512) = 4.58, P = 0.011), and the proportion of time spent in AT decreased with age, F(2, 180.826 = 3.88, P = 0.022). The main variable of interest, however, was proportion of time spent in SA, which reflects the relative amount of infants’ looking spent engaged with and actively processing the habituation stimulus. This analysis yielded a significant effect of Age, F(2, 179.027 = 3.31, P = 0.039) as SA decreased overall from 4 to 9 mo, but this effect was moderated by a significant DHA Group X Age interaction, F(2, 179.027 = 3.51, P = 0.032). SA decreased significantly with age in the Placebo group, F(2, 75.172 = 3.91, P = 0.024) but not in the Supplemented group, F(2, 99.545 = 2.40, P = ns). Modeled data for SA from 4 to 9 mo are shown ( Figure 2 ).

Figure 2
figure 2

Modeled data for proportion of time infants spent looking in SA as a function of randomized group assignment. Data are averaged across 4, 6, and 9 mo and reflect the significant Age X DHA Group interaction. Infants in the Placebo group (dashed line) showed a decrease in SA with age, infants in the Supplemented group (solid line) maintained levels of SA across the first year. Data shown are from completed habituation sessions where HR could be successfully coded: n = 159 (n = 72 and n = 87 placebo and supplemented, respectively) at 4 mo, n = 172 (n = 71 and n = 101) at 6 mo, and n = 156 (n = 68 and n = 88) at 9 mo. DHA, docosahexaenoic acid; HR, heart rate; SA, sustained attention.

PowerPoint slide

After observing this improvement in the quality of attention in infants from supplemented mothers on the habituation task, we examined whether the effect persisted after adding various covariates into the analyses. We repeated this analysis, controlling for parental verbal ability (as measured on the Peabody Picture Vocabulary Test (PPVT)), household income, maternal education, and additional DHA taken during pregnancy, and gestational age at enrollment. The DHA Group X Age interaction remained significant in each case.

Task Completion and Fussiness

An additional finding emerged from the analysis of infant habituation. From this task, there is some data loss due to fussiness or crying at each age. The proportion of loss varies widely across laboratories and across ages, although in this laboratory it tends to be between 10% and 20%. When we examined the distribution of infants whose data were unused due to behavioral state issues, we observed that these infants were significantly more likely to be from the Placebo group overall, especially at 6 and 9 mo. It is important to keep in mind that testers were blind to assignment group when these determinations were made. The number and percentage of infants excluded due to fussiness/crying as a function of randomized assignment are shown (see Table 1 ); the P-values reported are from χ2 tests conducted on observed cell counts.

Table 1 Subject loss (attrition) due to fussiness/crying on visual habituation tasks at 4, 6, and 9 mo of age as a function of membership in the two randomized groups


This project represents one of a very few follow-up studies on the effects of prenatal maternal supplementation on infant attention during the first year. At 4, 6, and 9 mo, infants from mothers supplemented with prenatal DHA were not different from infants from mothers in the placebo group on purely behavioral or HR measures, although infants from supplemented mothers showed a marginal trend to habituate more quickly across all ages. More importantly, however, infants from supplemented mothers maintained a consistent level of SA (a higher-quality attentional state strongly associated with stimulus processing) from 4 to 9 mo, while SA dropped off across the first year in infants from nonsupplemented mothers. Although this outcome measure is not a standardized index, and the interpretation of this pattern of change is not definitive, we think it important to note that this specific profile (i.e., the maintenance of consistent levels of SA across the first year), has been previously reported to be associated with higher preschool vocabulary and intelligence scores at 4 y (26). It is of interest that a behavioral measure of SA was also the only neuropsychological domain assessed at 5 y to be enhanced by maternal DHA supplementation during lactation (22).

Prenatal DHA supplementation did not affect measures of look duration or HR. An unexpected finding to emerge from this trial was the observation that attrition from the visual habituation task attributable to fussiness (i.e., a presumed indicator of regulation of behavioral state) was significantly lower for infants of DHA supplemented mothers overall (and in particular at 6 and 9 mo), suggesting another possible effect of early DHA status on infant development.

Lower HR has been reported in infants who are supplemented with DHA and arachidonic acid (ARA) and with fish oil (23,27), however, we did not find an effect of prenatal supplementation with DHA on HR. All children in the study were receiving DHA and ARA at the time they were tested, either from infant formula or human milk feeding. Although no findings in this area are yet definitive, this pattern of results is consistent with effects attributable to the presence of long-chain polyunsaturated fatty acid or DHA in the individual’s diet, rather than to an early programming effect. Our group has shown previously that fetal HR variability is increased by prenatal DHA supplementation with 600 mg/d of DHA (28) and higher HR variability is linked to cognitive measures such as arousal and attention (29), but to our knowledge a link between fetal HR variability and cognitive function in infancy has not been investigated.

This trial has its limitations. Blood levels did show that the prenatal supplementation did affect DHA levels in both maternal and cord blood at delivery; however, we did not control for postnatal dietary intake, although we recorded it at regular intervals in the first 12 mo of life. As noted above, all infants in the study received DHA and ARA from either human milk or modern infant formulas that include DHA and ARA. Despite postnatal consumption of DHA and ARA, the pattern of effects seen here following prenatal DHA supplementation (i.e., differences observed on early attention outcomes, but not on standardized developmental tests) echo those for a postnatal feeding trial of DHA and ARA supplementation that yielded strong effects on cognition and language when children were followed into the preschool period (23,30). Our data suggest, therefore, that there are benefits to prenatal DHA supplementation in our United States population over and above those of receiving DHA and ARA after birth. In addition, our decision not to invite children born <34 wk to participate in follow-up could be criticized, however, we did not wish to conflate any longer-term direct effect of DHA on these children’s developmental outcome with the indirect effect of early preterm birth. The strength of the study is the relatively large size of the groups studied compared with most studies of infant development, which reduces the likelihood of a Type II error for some of the outcomes that were not affected by DHA supplementation.

In summary, prenatal maternal DHA supplementation conferred advantages for the infants on attentional tasks (SA and behavioral state) during the first year of life. The pattern of effects seen here parallels that found for a postnatal feeding trial with DHA and ARA that yielded strong effects on cognition and language when children were followed into the preschool period (23,30), and suggests that benefits of prenatal DHA supplementation might persist into the preschool period despite the fact that all in the cohort were fed a source of DHA and ARA during the first year of life.



The Consolidated Standards of Reporting Trials (CONSORT) Diagram for the RCT is shown here (see Figure 3 ). Subjects were consented at enrollment during pregnancy for all follow-up measures. Details regarding the enrollment, randomization, blinding, Data Safety and Monitoring Board function, data checking and integrity, inclusion/exclusion criteria, compliance, and demographics of the sample are reported in the primary paper from this RCT (2). Informed consent was obtained from all participants and the study was approved by the University of Kansas Medical Center Human Subjects Committee. We invited all infants born to women in the Kansas University DHA Outcomes Study pregnancy trial to participate in follow-up. In making those invitations, we made the strategic decision to exclude infants born <34 wk gestation (n = 8), because premature infants show impoverished performance on visual habituation tasks (31) and experience significant delays on standardized tests in toddlerhood (32) and because we predicted that this group would be differentially distributed between the placebo and supplemented groups. We reasoned that, by excluding early preterm, the follow-up would provide a more direct test of the effects of prenatal DHA supplementation on later infant development, rather than reveal effects that might be moderated by reductions in prematurity.

Figure 3
figure 3

Consolidated Standards of Reporting Trials (CONSORT) diagram for follow-up of randomized clinical trial (RCT) on docosahexaenoic acid (DHA) prenatal supplementation. The attrition from delivery to the first-year tasks was 31.9% for the placebo group and 22.1% for the supplemented group.

PowerPoint slide

Subjects received either 3 capsules/d of an orange-flavored marine algae-oil source of DHA (200 mg DHA/capsule, DHASCO, from DSM Nutritional Products, Parsippany, NJ; formerly Martek Biosciences) from enrollment at a mean 14.5 wk gestation until birth (treatment), or 3 capsules containing half soybean and half corn oil (placebo, also orange-flavored). DSM Nutritional Products donated the capsules for the study but had no role in the study design, analysis, interpretation, or dissemination.

This clinical trial had two general aims. The first aim was to determine the effect of prenatal DHS supplementation on pregnancy outcomes. These outcomes (for which the study was powered) are reported in a previous publication (2) and the hypotheses were supported: supplementation increased gestation, birth weight, and birth length. In addition, DHA was observed to reduce the number of early preterm deliveries (<34 wk gestation). The second primary aim of the trial was to determine the effects of prenatal DHA on development of infants born to these mothers. The current report focuses on visual habituation from 4 to 9 mo of age. Per standard clinical trial methodology, testers remained blind to assignment group for all determinations, data coding, and analysis.

The demographic characteristics of the sample not followed up vs the characteristics of the sample that was followed after birth are shown (see Table 2 ). Compared with the children in the study not in the follow-up sample, those in the follow-up sample had mothers who were more compliant with capsule intake. However, the cohort included all major United States racial/ethnic groups with a wide range of education and income. The demographic characteristics of the follow-up sample broken out by Placebo vs. Supplemented groups are also shown (see Table 3 )

Table 2 Participants in the RCT follow-up vs those who did not participate
Table 3 Comparison of infants followed whose mothers received prenatal DHA supplement vs those whose mothers received a placebo

Longitudinal Measures

We chose postnatal measures based on the extant literature showing DHA affecting behavioral measures of visual attention (9). Visual habituation was administered at multiple time points to provide data on developmental trajectories to ensure assessment at points of maximum developmental sensitivity (33).

Visual Habituation and HR

Infants were evaluated at 4, 6, and 9 mo of age (corrected for gestational age) on a visual habituation protocol that was augmented with simultaneous measurement of HR. This outcome is well-suited to the first year but less appropriate beyond 12 mo, when infants become increasingly mobile (34). Visual habituation is a well-known measure of nonassociative visual learning, in which the infant’s visual and cardiac responses are assessed to repeated stimulus presentations. In this procedure, the infant is seated in a darkened room facing a screen on which visual stimuli are shown. The stimulus is shown repeatedly, and observers code infants’ looking to the stimuli over the repetitive presentations and HR is simultaneously collected during the session. Look duration decreases over the course of these repetitions. The decline in looking (habituation) reflects the infant’s learning and memory for the presented stimulus, and HR reflects the quality of the infant’s attention during looking; HR deceleration during looking is associated with engagement and active processing of the stimulus shown. The presentations continue until the infant’s looking declines (habituates) to a predetermined criterion. Details of the testing situation and recording of infant looking are reported elsewhere and the protocol was identical to that used in an RCT on postnatal feeding (23,30).

The stimuli used were two-dimensional faces of adults showing neutral expressions; the same set was used in a previous RCT involving zinc and iron (35). Along with allowing for the calculation of infant HR during the session, this protocol also allows for the derivation of different types or phases of attention during looking (36); most notable among those phases is SA, which reflects active processing of the stimulus. As in previous reports (23,30), the primary measures of interest were look duration during habituation, which reflects how quickly the stimulus is learned (34); and the proportion of time looking spent in SA, which indicates the proportion of time spent engaged and processing the stimulus (37,38).

Statistical Analysis

Analyses. Given that longitudinal data were available at 4, 6, and 9 mo of age, we conducted mixed-model analyses (which use all available data) with Subjects as a random factor, Age as a within-subject factor, and DHA group as a between-subject factor (preliminary analyses involving infant gender did not yield significant effects or interactions). Covariance was left unstructured as a conservative default. After initial tests were performed, appropriate demographic covariates were entered into analyses in order to rule out alternative plausible explanations for significant outcomes.

Analyses of look duration variables from visual habituation were conducted only on data from sessions that were complete and judged (by blinded observers) to have yielded usable data; analyses of HR and HR-defined phases from visual habituation were conducted on complete and usable habituation sessions but further required HR data from sessions judged (again by blinded HR coders) to be usable. Infants’ data were also excluded for reasons unrelated to fussiness (experimenter error, equipment failure, and parental interference). The number of sessions analyzed, are presented in the CONSORT diagram.

Measures analyzed. The measures analyzed from the visual habituation paradigm were derived from three basic categories. The first category included behavioral measures of peak look duration and number of looks to habituation; look duration has been reported to be affected by DHA status in one study of prenatal maternal supplementation (9) but not in a subsequent clinical trial of postnatal feeding (23). The second category was infants’ HR during the various points of the habituation protocol, which has been shown to be affected by postnatal supplementation (23,28). The third category reflected a coupling of behavior and HR (39,40), and included the proportion of time spent in HR-defined phases of attention. During periods of looking in the habituation procedure, infants typically show robust and sustained HR decelerations. Considerable evidence suggests that active engagement and processing of the visual stimulus occurs when the infant’s HR is decelerated (38). The use of HR during infant looking allows attention to be parsed into separate phases of SA (the period of HR deceleration seen during infant looking), OR (the phase of looking prior to the occurrence of deceleration), and AT (the phase during which the infant remains looking after SA but after HR has returned to baseline levels). Details on the computation of these variables are available in numerous previously-published reports (23,30).

Statement of Financial Support

This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Development (R01 HD047315), the Office of Dietary Supplements, and the Kansas Intellectual and Developmental Disabilities Research Center (P30 HD002528). DSM Nutritional Products (Parsippany, NJ, USA) donated the DHA and placebo capsules for the study.


None of the authors declare a potential conflict of interest.