Dose–response relationship between docosahexaenoic acid (DHA) intake and lower rates of early preterm birth, low birth weight and very low birth weight

Abstract

As previously reported, intention-to-treat findings from our phase III randomized clinical trial found that a supplement of 600 mg docosahexaenoic acid (DHA)/day during the last half of pregnancy reduced the incidence of early preterm birth (ePTB, <34 weeks gestation) and very low birth weight (VLBW < 1500 g) offspring. Given the potentially immense clinical significance of these findings, the goal of this secondary analysis was to (1) identify maternal characteristics related with capsule intake (i.e. DHA dose exposure) and (2) determine if DHA dose was associated with low (<2500 g) and very low birth weight after controlling for any relevant maternal characteristics. Three hundred forty-five pregnant mothers were recruited from hospitals in the Kansas City metropolitan area between 2006 and 2011. Most participants (n = 299) were from the phase III trial mentioned above, but we also included 46 participants from a second smaller, randomized trial that utilized an identical intervention design and was conducted concurrent to the larger trial. Both trials assigned participants to either 3 daily capsules of vegetable oil without DHA (n = 169) or 3 daily capsules of 200 mg DHA each (n = 176). Total capsules consumed was recorded by pharmacy supervised capsule count or participant self-report when needed. Maternal age, education, race and gestational age at delivery as well as infant birth weight were available for both trials. A Bayesian linear model indicated capsule intake increased with maternal age (p = 0.0100) and years of education (p = 0.0002). A Bayesian bivariate mixture-model associated capsule intake with simultaneous lower probability of ePTB, low birth weight (LBW, <2500 g) and VLBW (p = 0.0327). This, in conjunction with the positive findings in the clinical trial, support the need for future research to examine intervention methods to improve capsule compliance strategies in younger and less educated mothers.

1. Introduction

A recent meta-analysis concluded that supplementation with omega-3 fatty acids during pregnancy can reduce early preterm birth (ePTB, gestation <34 weeks) and very low birth weight (VLBW, <1500 g) [1] but no dose response study has determined the optimal intake of DHA to achieve these outcomes. As example of the variable doses explored thus far, two recent randomized controlled trials (RCTs) [2,3] provided 800 and 600 mg DHA/d, respectively, and found a reduction in ePTB and VLBW; while women provided 400 mg DHA/d in a third RCT did not show a reduction in either outcome [4]. Because subjects were similar among the studies (i.e., women were healthy with a singleton pregnancy, less than 20 weeks gestation and between the ages of 18 and 35 years at enrollment), comparison suggests that to improve ePTB and VLBW women may benefit from substantially more DHA than the 200–300 mg currently added to some commercially available prenatal supplements in the US.

Perhaps equally important to the original protocol driven dosing scheme, a common finding in clinical trials is that participants are variably compliant with their assigned treatment. Some participants carefully consume every capsule as prescribed while others miss days. In this secondary data analysis we exploit that variability to explore the effect of DHA dose. Our data include two RCTs [3,5] that provided DHA (600 mg/d) or placebo. We seek to answer (1) what maternal factors predict the number of capsules consumed (i.e. DHA dose exposure) and (2) does dose exposure relate to the incidence of ePTB, low birth weight (LBW, <2500 g) or VLBW (i.e. DHA dose response). We capitalized on data from 345 pregnancies studied concurrently in the Kansas City metropolitan area and extrapolate findings to generate ideas on what may be optimal DHA dosing strategies during pregnancy.

2. Patients and methods

2.1. Design overview

This is a secondary data analysis from studies, KUDOS [3] and HOPE [5], that were two-armed randomized placebo-controlled clinical trials to assess the efficacy of prenatal DHA on outcomes other than ePTB and VLBW. In both studies pregnant women were assigned equally to placebo (corn and soybean mixture) or 600 mg DHA/d (DHASCO, DSM, Baltimore, MD) beginning after 12 and before 20 weeks gestation and continuing until the end of their pregnancy.

2.2. Participants

In both studies healthy women with a singleton pregnancy were recruited from outpatient clinics and hospitals in the Kansas City metropolitan area between 2006 and 2011. We included mothers and infants for whom birth outcomes were available and present the summarized demographics across both studies in Table 1.

Table 1.

Participant characteristics.

Variable, mean (SD) or %	HOPE (n = 46)	KUDOS (n = 299)	Combined (n = 345)
Race, African American	28%	38%	37%
Age, years	26.8 (4.2)	25.4 (4.8)	25.6 (4.7)
Education, years	14.3 (3)	13.7 (2.8)	13.8 (2.8)
Randomized to 600 mg DHA	48%	52%	51%

2.3. Capsule compliance (i.e. dosing)

This paper investigates the dose-response relationship in which “dose” is the number of provided DHA capsules consumed per week. Briefly, in both studies participants were provided with an investigational product bottle monthly that contained 100 capsules; they were instructed to consume 3 capsules per day (i.e. 21 capsules per week). Women were also provided with a postage paid envelope to return their capsule bottle monthly to the Investigational Pharmacy where unconsumed capsules were counted and recorded. A minority of the participants who did not return bottles or returned only a portion of all their bottles were encouraged to self-report intake from missing returns. If a bottle was not returned and the participant could not reliably report intake, an estimated average was determined from previous bottle returns if at least 2 other returns were available. Capsule intake was determined by pharmacy capsule count, participant self-report, and estimated average among 66%, 23% and 11% of all returns, respectively. Thirty participants had missing capsules intake that could not reliably be estimated. The handling of these missing data is discussed in Section 2.5.

2.4. Determination of gestational age

Expected date of delivery (40 weeks or 280 days gestation) was determined by ultrasound late in the first or early in the second trimester of pregnancy. The expected date of delivery recorded was used to determine gestation duration even if a later ultrasound indicated a different expected date of delivery [3].

2.5. Statistical analyses

We conducted a Bayesian linear model analysis with capsules per week predicted by maternal age and education in years (Fig. 1). Deviance information criteria (DIC) compared all possible models with and without the predicted variables. The final model selected was best (i.e. lowest DIC). The final model allowed us to infer which if any of the predictors were associated DHA dose. Importantly, the model also allowed us to impute missing capsules (n = 30) and provided for multiple imputation that was simultaneously used for prediction of birth outcomes. The model that used capsules to predict birth outcomes utilized a Bayesian bivariate mixture that resulted in a three-component normal distribution of all gestational ages and birth weights validated by Gajewski et al. [6]. The model was utilized in an analysis reported previously [7]. The components’ parameters were regressed on the logit scale on capsules. All participants receiving placebo were transformed automatically to 0 capsules for this latter model. We simultaneously investigated the effect of education on birth weight and gestational age but found it not significant. The parameters predict the mixtures associated with all components. Births in component 1 have on average 40 weeks gestational age and weigh 3.5 kg; those in component 2 have 38 weeks and 3.1 kg, and those in component 3 have 33 weeks and 1.9 kg. We used the results to calculate 95% credible intervals for the relationship of weekly capsule intake and ePTB, VLBW and LBW rates. All models were computed in WinBUGS 1.4 (Imperial College & MRC, UK) and code can be obtained from BJG (Table 2).

Fig. 1.

Bayesian model for birth outcomes, capsules, and mother demographics. Two Bayesian models were fitted. The first Bayesian model was a linear model that regressed capsules taken on maternal age (p = 0.0010) and education (p = 0.0002). This first model also imputed capsules taken if missing. The second model was a mixture model of 3 normally distributed components with each component having a different mean gestational age and birth weight. The model was used to simultaneously predict gestational age and birth weight (p = 0.0327) based on the number of DHA capsules consumed. The capsules of DHA taken assigned 0 to placebo subjects.

Table 2.

Summary of capsules per week and birth outcomes.

Variable, mean (SD) or %	HOPE (n = 44–46)	KUDOS (n = 279–299)	Combined (n = 315–345)
Average capsules consumed per week	18.2 (4.5)	16.2 (5)	16.5 (5)
Gestational age at delivery (weeks)	39.7 (1.1)	39.2 (2.1)	39.3 (3)
Birth weight (kg)	3.43 (0.45)	3.28 (0.57)	3.3 (0.56)

3. Results

3.1. Capsule compliance and maternal characteristics

We wanted to make sure that the final model for predicting capsules was indeed the best among several models. Three different models were considered and the best among these three models included both age and education as predictors of capsules taken (DIC = 1868.1). The second-best model with education (DIC = 1871.5) and the third best model with age (DIC = 1882.1) were less robust.

Once we defined a reasonable model we used the model to understand the relationship between maternal age and education with capsules. Capsule compliance was significantly associated with both maternal age and education (Table 3) and displayed in Fig. 2 with plots illustrating the credible intervals for the relationship partitioned by older (32 years) versus younger (20 years) mothers. To further illustrate effect size, the median improvement in compliance was 1.5 more capsules per week for each 10-year increase in maternal age and 1.8 more capsules per week for each 4-year increase in education.

Table 3.

Summary of parameter estimates for the Bayesian linear model of capsules/week on maternal age and education.

Variable	Pr(B > 0)	5%-tile	B Median	95%-tile
Age	0.0100	0.05	0.15	0.26
Education	0.0002	0.27	0.45	0.63

Fig. 2.

Relationship between total capsules consumed regressed on mother’s years of education and age (first Bayesian model, 90% credible intervals).

3.2. Capsule compliance and birth outcomes

The relationships between capsule compliance and ePTB, VLBW, and LBW are shown in Figs. 3–5, respectively. ePTB, VLBW, and LBW declined significantly (p = 0.037) as capsule intake increased. Approximately 50% compliance (defined as 10 capsules/week or ~285 mg/d) reduced median ePTB and VLBW rates by nearly half, from 3.7% to 1.9% for ePTB and from 0.85% to 0.44% for VLBW. However, both ePTB and VLBW appear to decline further with still greater compliance.

Fig. 3.

Relationship between mean early preterm birth and capsules of DHA per week shown by the second Bayesian model described in Fig. 1 (p = 0.037).

Fig. 5.

Relationship between mean low birth weight and capsules of DHA per week defined by the second Bayesian model described in Fig. 1 (p = 0.037).

4. Discussion

After capsule intake is converted into DHA dose per day, the apparent dose response of DHA to reduce rates of ePTB and VLBW appears to improve continuously up to an intake of nearly 600 mg/day. At that intake, our model suggests an ePTB rate of 1%, lower than the rate of 3.4% [8] found in the US. While a limitation of our analysis is that we cannot know with certainty if even higher doses might further reduce rates of ePTB, VLBW and LBW, it seems extremely unlikely that DHA can prevent all ePTB, VLBW and LBW births. For example, some of these events are related to anatomical deviations that DHA would not be expected to affect.

A strength of the analysis is that we controlled for maternal age and education in modeling the relationship between DHA dose and both ePTB and VLBW after learning in earlier models that both are associated with more optimal birth outcomes. We are not the first to link education and age to more favorable health behaviors and outcomes during pregnancy. For example de Jong-Van den Berg et al. found that folic acid supplements awareness and intake during pregnancy were significantly associated with maternal education, age, pregnancy planning, family income, ethnicity, and race [9]. A study performed in Quebec linked higher maternal education with more favorable birth outcomes [10]. Mothers who had not completed high school had higher rates of preterm birth (adjusted OR 1.48, 95% CI 1.44–1.52), SGA birth (OR 1.86, 95% CI 1.82–1.91) and stillbirth (OR 1.54, 95% CI 1.36–1.74) compared to more educated mothers with a college degree. A possible explanation is that more educated mothers are better tuned to and receptive of information about health benefits of supplements and conscious of the importance of abiding to health guidelines established by science leading to more favorable birth outcomes.

It is now well documented from our work and others that DHA can reduce ePTB and VLBW [1–3]. Public health efforts are necessary to encourage improved DHA intake during pregnancy, whether with supplements containing DHA, higher intake from food such as seafood low in mercury, or both. However, our results in conjunction with others that find poorer birth outcomes, suggest that younger, less educated women deserve additional attention. It might be hypothesized they are less likely to take on the message to improve DHA intake even though they are at greater risk of poor birth outcomes. Efforts to understand and remove their barriers could be worthwhile not only to improve their DHA intake but to improve other health behaviors related to pregnancy and infant outcomes.

To conclude, higher maternal age and education predicted higher capsule intake and therefore higher DHA dose in women assigned to a supplement with DHA. Even after controlling for maternal age and education, higher DHA dose during pregnancy predicted lower rates of both ePTB and VLBW. It is important to reach young and less educated pregnant women early in pregnancy to highlight the importance of DHA intake during pregnancy and intervene within strategies to improve DHA intake.

Fig. 4.

Relationship between mean very low birth weight and capsules of DHA per week shown by the second Bayesian model described in Fig. 1 (p = 0.037).

Abbreviations:

DHA

docosahexaenoic acid

ePTB

early preterm birth (<34 weeks gestation)

VLBW

very low birth weight