The Mothers, Omega-3, and Mental Health Study: a double-blind, randomized controlled trial

hb(Tpc

Research www.AJOG.org

OBSTETRICS

The Mothers, Omega-3, and Mental Health Study:a double-blind, randomized controlled trialEllen L. Mozurkewich, MD; Chelsea M. Clinton, MD; Julie L. Chilimigras, MPH; Susan E. Hamilton, MS;Lucy J. Allbaugh, MS; Deborah R. Berman, MD; Sheila M. Marcus, MD; Vivian C. Romero, MD;Marjorie C. Treadwell, MD; Kristie L. Keeton, MD; Anjel M. Vahratian, PhD;Ronald M. Schrader, PhD; Jianwei Ren, MD; Zora Djuric, PhD

OBJECTIVES: Maternal deficiency of the omega-3 fatty acid, docosa-exaenoic acid (DHA), has been associated with perinatal depression,ut there is evidence that supplementation with eicosapentaenoic acidEPA) may be more effective than DHA in treating depressive symptoms.his trial tested the relative effects of EPA- and DHA-rich fish oils onrevention of depressive symptoms among pregnant women at an in-reased risk of depression.

STUDY DESIGN: We enrolled 126 pregnant women at risk for depres-sion (Edinburgh Postnatal Depression Scale score 9-19 or a history ofdepression) in early pregnancy and randomly assigned them to receiveEPA-rich fish oil (1060 mg EPA plus 274 mg DHA), DHA-rich fish oil(900 mg DHA plus 180 mg EPA), or soy oil placebo. Subjects completedthe Beck Depression Inventory (BDI) and Mini-International Neuropsy-
chiatric Interview at enrollment, 26-28 weeks, 34-36 weeks, and at
trial. Am J Obstet Gynecol 2013;208:313.e1-9.

http://dx.doi.org/10.1016/j.ajog.2013.01.038

6-8 weeks’ postpartum. Serum fatty acids were analyzed at entry and at34-36 weeks’ gestation.

RESULTS: One hundred eighteen women completed the trial. Therewere no differences between groups in BDI scores or other depressionendpoints at any of the 3 time points after supplementation. The EPA-and DHA-rich fish oil groups exhibited significantly increased post-supplementation concentrations of serum EPA and serum DHA respec-tively. Serum DHA- concentrations at 34-36 weeks were inverselyrelated to BDI scores in late pregnancy.

CONCLUSION: EPA-rich fish oil and DHA-rich fish oil supplementationdid not prevent depressive symptoms during pregnancy or postpartum.

Key words: depression, docosahexaenoic acid, eicosapentaenoic
acid, fish oil, supplementation
Cite this article as: Mozurkewich EL, Clinton CM, Chilimigras JL, et al. The Mothers, Omega-3, and Mental Health Study: a double-blind, randomized controlled

t

Depressive symptoms are associatedwith significant morbidity in preg-

nancy and postpartum. Perinatal depres-sion may be associated with impairedmother-infant bonding and may also be

From the Departments of Obstetrics and GynecPsychiatry (Ms Hamilton, Ms Allbaugh, and Dr Mand the Department of Obstetrics and GynecoloGynecology (Dr Mozurkewich), and the Clinical aAlbuquerque, NM.

Received Nov. 21, 2012; revised Jan. 6, 2013; a

This study was supported by National InstitutesMedicine) and a University of Michigan Clinical RMichigan Clinical Research Unit. This study was(P30 CA046592). This study was also supportedTranslational Sciences, NIH, through grant 8UL1Naturals Corporation donated both active suppl

E.L.M. was an invited speaker at the Nutracon 2Anaheim, CA, March 7-8, 2012, and received re

Presented at the 33rd annual meeting of the Soc

The racing flag logo above indicates that this art

Reprints not available from the authors.

0002-9378/$36.00© 2013 Mosby, Inc. All rights reserved.

associated with adverse outcomes ofpregnancy, such as preterm birth andlow birthweight.1 Although antidepres-sant medications of the selective sero-tonin reuptake inhibitor category are

y (Drs Mozurkewich, Clinton, Berman, Romerous), and Family Medicine (Drs Ren and Djuric), USt Joseph Mercy Health System, Ypsilanti (Dr KTranslational Science Center (Dr Schrader), Uni

pted Jan. 16, 2013.

ealth (NIH) grant R21 AT004166-03S1 (Nationaarch Initiatives grant and by the University of Mi

o supported (in part) by the NIH through the Unipart by the National Center for Research Resou000041, for the Clinical and Translational Scienents and placebos to the trial.

Conference, sponsored by the Global Organizursement for travel expenses. The remaining au

for Maternal-Fetal Medicine, San Francisco, C

was rushed to press for the benefit of the scient

APRIL 2013 Americ

readily prescribed, these medicationshave been associated with both majorcardiovascular malformations and poorneonatal adaptation syndromes.2 Forhis reason, there has been interest to test

eadwell, and Vahratian and Ms Chilimigras),ersity of Michigan Medical School, Ann Arbor,on), MI, and the Department of Obstetrics andity of New Mexico Health Sciences Center,

enter for Complementary and Alternativean General Clinical Research Center, now theity of Michigan’s Cancer Center Support Grants and the National Center for Advancingenter, University of New Mexico. The Nordic

n for EPA and DHA Omega-3s (GOED),rs report no conflict of interest.

eb. 11-16, 2013.

community.

olog , Trarc nivgy, eetnd vers

cce

of H l Cese chigals versin rceTR ce C

em

012 atioimb tho

iety A, F

icle ific

an Journal of Obstetrics & Gynecology 313.e1

http://dx.doi.org/10.1016/j.ajog.2013.01.038

eDpfppt

dvDtscvavtfiEcnbi

cpdgoPhtmrsdiop2

soNMaMt1

ammsp0nepepcaasttr

Research Obstetrics www.AJOG.org

alternative medicine modalities thatmight prevent or treat this debilitatingcondition without associated deleteriouseffects.

Over the past decade, there has beenconsiderable interest in the omega-3fatty acids, eicosapentaenoic acid (EPA)and docosahexaenoic acid (DHA), aspossible preventive or therapeutic mo-dalities for depression. In animal studies,dysregulation of the innate immune re-sponse has been associated with diets lowin omega-3 fatty acids; this immune dys-function has also been implicated inmood disorders.3 Human observationalvidence has suggested that deficiency ofHA may predispose to perinatal de-ression.4,5 DHA is preferentially trans-

erred from the maternal to fetal com-artment in the third trimester ofregnancy, leaving many mothers rela-ively DHA deficient.6

Several early trials have suggested thatfish oil supplementation may be benefi-cial in treating perinatal depression.7-9

However, several larger, blinded trials ofmaternal DHA supplementation havefailed to show a benefit for this interven-tion.10-12 One potential reason for this

isparity in the observational and inter-entional studies is that EPA rather thanHA may be the more active fatty acid in

he prevention or treatment of moodymptoms. Three systematic reviewsomparing DHA and EPA for the pre-ention or treatment of mood disordersmong nonpregnant and pregnant indi-iduals have suggested that EPA ratherhan DHA may have beneficial ef-ects.13-15 Likewise, a recently random-zed controlled trial comparing 1 g ofPA, 1 g of DHA, and coconut oil pla-ebo for mild to moderate depression inonpregnant individuals found EPA toe superior to DHA and placebo in treat-

ng depressive symptoms.16

We carried out this study to directlycompare EPA-rich fish oil, DHA-richfish oil, and soy oil placebo for the pre-vention of depressive symptoms amongpregnant women at an increased risk fordepression. We hypothesized that theEPA- or DHA-rich fish oil supplementa-tion would reduce the Beck DepressionInventory (BDI) score at 6 weeks post-
partum by 50% compared with placebo.
313.e2 American Journal of Obstetrics & Gynecolo

MATERIALS AND METHODSDetails of ethics approvalThis study was approved by the institu-tional review boards of the University ofMichigan Health System (Ann Arbor,MI) and St Joseph Mercy Health System(Ypsilanti, MI). Enrolled participantsprovided written informed consent.

Study designThe protocol for this study has been previ-ously described.17 We solicited permissionto determine eligibility from pregnantwomen presenting for prenatal care at 2health systems in southeastern Michigan:the University of Michigan Health Systemand St Joseph’s Mercy Hospital HealthSystem. We used the Edinburgh PostnatalDepression Scale (EPDS), a widely used 10item measure of perinatal mood, to screenpotential subjects for depression risk.18 In-lusion criteria included past history of de-ression, an EPDS score 9-19 (at risk forepression or mildly depressed), singletonestation, a maternal age of 18 years orlder, and a gestational age of 12-20 weeks.otential subjects were excluded if theyad a history of a bleeding disorder,hrombophilia requiring anticoagulation,

ultiple gestation, bipolar disorder, cur-ent major depressive disorder, currentubstance abuse, lifetime substance depen-ence, or schizophrenia. Women were also

neligible if they were currently takingmega-3 fatty acid supplements or antide-ressant medications or eating more thanfish meals per week.For a final determination of eligibility,

tudy staff with training in clinical psychol-gy administered the Mini-Internationaleuropsychiatric Interview (MINI). TheINI is a structured interview for di-

gnosis of Diagnostic and Statisticalanual of Mental Disorders-IV and In-

ernational Classification of Diseases,0th revision, psychiatric disorders.19

We used the MINI to exclude currentmajor depressive disorder (MDD), bi-polar disorder, current substance abuseor dependence, suicidal ideation, orschizophrenia.17

Sample size considerationsOur sample size calculation was based onthe premise that supplementation with
either EPA-rich or DHA-rich fish oil
gy APRIL 2013

would result in a 50% reduction in themean BDI score at 6 weeks postpartum.The test characteristics of the BDI as adepression screen among pregnant pop-ulations have been well characterized.20,

21 The BDI has been used in perinatalpsychiatric studies as a measure of sever-ity of depressive symptoms.22,23

Sample size calculations were gener-ated in nQuery Advisor version 6.01(STATCON, Witzenhausen, Germany).The sample size was chosen based on theexpected BDI scores using postpartumBDI scores from a previous study ofpregnant women.24 For this calculationwe used a mean BDI score of 8.4 and SDof 6.4 expected among postpartummothers at risk for depression. Assuminga significance level of P � .05, a 1-waynalysis of variance test, a variance ofeans (variance of the individual groupeans) of 3.920, a SD of 6.4, and an effect

ize (the index of the separation ex-ected among the observed means) of.0957, we planned to recruit 105 preg-ant women (35 pregnant women inach of the 3 groups) to have 80%ower to detect at least 1 group differ-nce in the mean BDI score. The sam-le size was increased by 20% to ac-ount for anticipated study dropoutsnd women who were expected to startntidepressant treatment. Our deci-ion to hypothesize a 50% reduction inhe mean BDI score between interven-ion and control groups was based onesults from published studies.7,8

Women who met all eligibility criteriaand who agreed to participate were ran-domly assigned to receive one of the fol-lowing: (1) EPA-rich fish oil supplementa-tion (1060 mg EPA plus 274 mg DHA); (2)DHA-rich fish oil supplement (900 mgDHA plus 180 mg EPA); or (3) a soy oilplacebo (control arm). Randomizationwas carried out using a random numbertable maintained in the University ofMichigan Investigational Drug Service.

The intervention supplements andplacebo were provided by Nordic Natu-rals Corporation in Watsonville, CA.The EPA-rich fish oil (ProEPAXtra,Nordic Naturals) contained an approxi-mate 4:1 ratio of EPA to DHA (1060 mgEPA plus 274 mg DHA), whereas the
DHA-rich oil (ProDHA, Nordic Natu-

www.AJOG.org Obstetrics Research

rals) contained DHA and EPA in an ap-proximate 4:1 ratio (900 mg DHA plus180 mg EPA). The placebos were formu-lated to be identical in appearance toboth the EPA- and DHA-rich supple-ments and contained 98% soybean oiland 1% each of lemon and fish oil. Thesupplements were molecularly distilledand free of industrial contaminants,mercury, and organochlorines.

Because the EPA and DHA capsuleswere not identical in appearance, weused a double-dummy design to main-tain blinding. The EPA group received 2large EPA-rich fish oil capsules and 4small placebo capsules formulated to ap-pear identical to the DHA-rich fish oil cap-sules. The DHA group received 2 large pla-cebo capsules formulated to appear

FIGUREConsolidated Standards of Reportin

This figure describes the recruitment and determMozurkewich. Mothers, Omega-3, and Mental Health Study

identical to the EPA-rich fish oil capsules

and 4 small DHA-rich fish oil capsules.The placebo group received 2 large and 4small placebo capsules daily. Adherence tothe protocol was assessed by self-report aswell as by capsule counts. Subjects wereasked to return unused capsules to studystaff at each study visit. We also assessedomega-3 highly unsaturated fatty acid lev-els before and after supplementation as ameasure of compliance.

Enrolled subjects attended 4 study visits.The BDI and MINI were administered ateach of these 4 time points: visit 1 at 12-20weeks’ gestation, visit 2 at 26-28 weeks’gestation, visit 3 at 34-36 weeks’ gestation,and visit 5 at 6-8 weeks’ postpartum. Atvisits 2, 3, and 5, the MINI was readminis-tered to diagnose MDD. Subjects who metcriteria for MDD at any time during study

Trials 2010 flow diagram

tion of eligibility process for this trial.J Obstet Gynecol 2013.

participation continued in the study but

APRIL 2013 Americ

were also referred to mental health provid-ers for standard care. Obstetrical and men-tal health providers were free to prescribeantidepressant medications, if indicated.

Maternal blood was drawn at enroll-ment as well and at 34-36 weeks’ gesta-tion after a fast of at least 4 hours. Um-bilical cord blood (mixed arterial andvenous) was collected from infants bornto mothers who participated in thestudy. All samples were centrifuged be-fore separation into the 6 aliquots andwere stored at �70 degrees C.

Fatty acid analyses were carried out onthawed serum aliquots. Total serum fattyacids were first extracted with Folch re-agent and then converted to fatty acidmethyl esters. Quantitation was done us-ing gas chromatography with mass spec-

g

ina. Am

tral detection as previously described.25


udy.


Results were expressed as percentage of to-tal fatty acids.

The Fisher exact test was used to com-pare categorical variables. For continu-ous variables, 1-way analysis of variance(ANOVA) was used to compare groups.The change in fatty acid levels was as-

TABLE 1Baseline characteristics

Parameter

Age, mean (SD)...................................................................................................................

Gestational age at enrollment (wks), mean (SD...................................................................................................................

Parity, mean (SD)...................................................................................................................

Racial characteristics, n (%)..........................................................................................................

White..........................................................................................................

African-American..........................................................................................................

Hispanic-Latina..........................................................................................................

Asian..........................................................................................................

American Indian or Alaska Native..........................................................................................................

Native Hawaiian or other Pacific ethnicity...................................................................................................................

Past history of depression (self-reported), n (%...................................................................................................................

Baseline BDI score, mean (SD)...................................................................................................................

EPDS screen, mean (SD)...................................................................................................................

ANOVA, analysis of variance; BDI, Beck Depression Inventory;a One-way ANOVA; b Fisher exact test.

Mozurkewich. Mothers, Omega-3, and Mental Health St

TABLE 2Intent-to-treat analysis

Parameter

EPA-richfish oil(n � 39)

Mean BDI visit 2, n (SD) 8.7 (4.2)...................................................................................................................



MDD visit 2, n (%) 4 (10)...................................................................................................................

MDD visit 3, n (%) 2 (5)...................................................................................................................

MDD visit 5, n (%) 3 (8)...................................................................................................................

Started antidepressant, n (%) 6 (15)...................................................................................................................

On lowest antidepressant dose,n (%)

3 (50)

...................................................................................................................

The BDI included scores in which 0-9 means minimal depresymptoms, 20 or greater means moderate to severe depressivat 34-36 weeks’ gestation; and visit 5 was at 6-8 weeks’ posANOVA, analysis of variance; BDI, Beck Depression Inventory; Dmajor depressive disorder; SD, standard deviation.a One-way ANOVA adjusted for BDI at enrollment; b Generaliz

Mozurkewich. Mothers, Omega-3, and Mental Health Study.


sessed using the paired Student t tests, theWilcoxon signed rank test, and theKruskal-Wallis nonparametric test. Analy-sis of covariance (ANCOVA) was used toevaluate the role of serum EPA and DHAas predictors of the BDI score.

This study’s identifier is NCT00711971.

EPA-rich fish oil(n � 39)

DHA-rich fish(n � 38)

29.9 (5.0) 30.6 (4.5).........................................................................................................................

15.9 (2.6) 17.0 (2.3).........................................................................................................................

0.87 (0.83) 1.08 (0.94).........................................................................................................................

.........................................................................................................................

33 (85) 29 (76).........................................................................................................................

4 (10) 4 (11).........................................................................................................................

0 (0) 4 (11).........................................................................................................................

1 (3) 1 (3).........................................................................................................................

0 (0) 0 (0).........................................................................................................................

1 (3) 0 (0).........................................................................................................................

32 (82) 30 (79).........................................................................................................................

8.41 (5.65) 7.79 (5.29).........................................................................................................................

7.93 (4.63) (n � 32) 7.56 (4.49) (n �.........................................................................................................................

, docosahexaenoic acid; EPA, eicosapentaenoic acid; EPDS, Edin

Am J Obstet Gynecol 2013.

A-richh oil� 38) Placebo (n � 41) Significance

(4.6) 6.3 (3.9) .051a

..................................................................................................................

(6.3) 7.4 (5.5) .81a

..................................................................................................................

(4.8) 5.9 (6.1) .78a

..................................................................................................................

11) 0 (0) � .16b

..................................................................................................................

11) 3 (7)..................................................................................................................

5) 2 (5)..................................................................................................................

18) 4 (10) .56c

..................................................................................................................

100) 3 (75) [1 unknowndose]

.07c

..................................................................................................................

symptoms, 10-19 means mild to moderate depressiveptoms. Visit 2 was at 26-28 weeks’ gestation; visit 3 was

um.docosahexaenoic acid; EPA, eicosapentaenoic acid; MDD,

timating equation model; c Fisher exact test.


gy APRIL 2013

RESULTS

Between October 2008 and May 2011,2657 women granted permission to de-termine eligibility and were screened fordepression in early pregnancy. Of these,161 women who met initial entry criteriaunderwent a final determination of eligi-bility with the MINI. Thirty-five womenwere excluded after undergoing theMINI, with reasons for exclusion beingcurrent major depressive disorder, bipo-lar disorder, and substance dependence.There were 126 women who enrolled inthe study and who were randomly as-signed to receive EPA-rich fish oil, DHA-rich fish oil, or placebo. There were 8women who discontinued trial partici-pation. One of the subjects in the DHA-rich fish oil group experienced a second-trimester pregnancy loss attributed tocervical insufficiency. The remaining 7subjects were lost to follow-up and infor-mation was not available on the studyoutcomes for these subjects.

There remained 39 women who re-ceived EPA-rich fish oil, 38 women whoreceived DHA-rich fish oil, and 41women who received placebo supple-mentation. During the course of the

Placebo (n � 41) Significance

30.4 (5.9) .85a

..................................................................................................................

16.2 (2.3) .15a

..................................................................................................................

0.85 (1.20) .55a

..................................................................................................................

..................................................................................................................

34 (83) .49b

..................................................................................................................

2 (5)..................................................................................................................

3 (7)..................................................................................................................

1 (2)..................................................................................................................

1 (2)..................................................................................................................

0 (0)..................................................................................................................

33 (80) .96b

..................................................................................................................

7.15 (5.21) .58a

..................................................................................................................

) 7.34 (4.40) (n � 29) .87a

..................................................................................................................

h Postnatal Depression Scale; SD, standard deviation.

oil

......... .........

)......... .........

......... .........

......... .........

......... .........

......... .........

......... .........

......... .........

......... .........

......... .........

)......... .........

......... .........

32......... .........

DHA burg

DHfis(n

7.0.........

6.9.........

5.7.........

4 (.........

4 (.........

2 (.........

7 (.........

7 (

.........

ssivee symtpartHA,

ed es

trial, there were 16 ongoing subjects who

apwTtrvawsatitap

tp(aT

.a

udy.


discontinued taking the study supple-ment before the final study visit at 6-8weeks’ postpartum. These subjects wereanalyzed in the intent-to-treat analysis.Reasons for discontinuation includedside effects (mostly nausea, belching, andfishy aftertaste as well as forgetting totake capsules or becoming too busy totake study capsules). The Consort flowdiagram of subject recruitment and par-ticipation is in the Figure. The baselinecharacteristics of the study participantsare detailed in Table 1.

BDI scores at 34-36 weeks’ gestationnd at 6-8 weeks’ postpartum were com-ared between groups in an ANCOVAith the baseline score as a covariate.here were no significant differences in

he change in the BDI scores between en-ollment and the 34-36 weeks’ gestationisit or the 6-8 week postpartum visitmong the randomized groups. Thereere no differences in the mean BDI

cores among the groups at entry or atny of the study visits after supplemen-ation. There were no statistically signif-cant differences among the groups inhe proportion of women who startedntidepressant medications or in antide-ressant dose requirements (Table 2).There were no significant differences in

he proportion of subjects who com-lained of gastrointestinal side effectsnausea, belching, and fishy aftertaste)mong the randomized treatment groups.

TABLE 3Measures of capsule compliance

Parameter

Stopped before visit 2 (self-report)...................................................................................................................

Compliance between visits 1 and 2 (pill countmean (SD) [number missing]c...................................................................................................................





Visit 1 was at enrollment at 12-20 weeks, visit 2 was at 26-2ANOVA, analysis of variance; DHA, docosahexaenoic acid; EPa Fisher exact test; b One-way ANOVA; c Missing means that


here were no significant differences be-

tween the randomized groups in measuresof adherence (Table 3). Based on self-reported adherence, a per-protocol analy-sis was carried out, excluding those sub-jects who reported stopping their capsulesearly (n � 16) and those whose capsulecontinuation was unable to be ascertained(n � 4). There were no significant differ-ences in BDI scores in the per-protocolanalysis (Table 4).

Supplementation significantly increasedserum EPA in the EPA group and signifi-cantly increased serum DHA in the DHA-rich fish oil group (Table 5). To evaluatethe relationship between maternal serumEPA and DHA and BDI scores at visits 3and 5, we fit the following variables in anANCOVA model: BDI at enrollment,group allocation, smoking, body massindex at enrollment, admission to havingstopped taking capsules, admission to

TABLE 4Per-protocol analysis

Parameter

EPA-rich fishoil (n � 31),mean (SD)

DHA-roil (nmean

BDI visit 2 7.9 (3.7) 6.5 (4....................................................................................................................

BDI visit 3 7.4 (5.0) 5.9 (4....................................................................................................................

BDI visit 5 6.6 (5.2) 5.1 (4....................................................................................................................

ANOVA, analysis of variance; BDI, Beck Depression Inventory;standard deviation.a One-way ANOVA controlling for BDI score at enrollment.


DHA-rich fish(n � 38)

3 0.........................................................................................................................

0.67 (0.25) [3] 0.78 (0.20) [4]

.........................................................................................................................

4 2.........................................................................................................................

0.58 (0.32) [5] 0.65 (0.27) [6]

.........................................................................................................................

5 4.........................................................................................................................

0.67 (0.35) [4] 0.62 (0.30) [2]

.........................................................................................................................

eks, visit 3 was at 34-36 weeks, and visit 5 was at 6-8 weeks’ posapentaenoic acid; SD, standard deviation.

ubject did not return capsules to be counted at this time point.



APRIL 2013 Americ

have started antidepressant medications,serum EPA at 34-36 weeks, serum DHAat 34-36 weeks, and total omega-3 highlyunsaturated fatty acids at 34-36 weeks.

The initial evaluation found the vari-ables BDI at enrollment, admission tohaving stopped taking capsules, serumEPA, and serum DHA to be significantpredictors of the BDI score at visit 3.Because of skewed distributions, welog transformed serum EPA and DHAvalues and performed regression anal-yses to evaluate the relationship ofthese variables with the BDI scores atvisits 3 and 5.

Because EPA and DHA were highlycorrelated, they were modeled sepa-rately. The BDI score at visit 3 was signif-icantly predicted by serum DHA (P �05), BDI at enrollment (P � .001) anddmission to having stopped taking cap-

fish3),)

Placebo(n � 32),mean (SD) Significance

6.3 (4.1) .29a

..................................................................................................................

7.2 (5.2) .51a

..................................................................................................................

6.2 (6.5) .56a

..................................................................................................................

, docosahexaenoic acid; EPA, eicosapentaenoic acid; SD,

Placebo(n � 41) Significance

3 .24a

..................................................................................................................

0.68 (0.27) [1] .15b

..................................................................................................................

4 .77a

..................................................................................................................

0.63 (0.31) [2] .63b

..................................................................................................................

7 .75a

..................................................................................................................

0.64 (0.36) [3] .85b

..................................................................................................................

artum.

ich� 3(SD

7).........

9).........

4).........

DHA

oil

......... .........

),

......... .........

......... .........

),

......... .........

......... .........

),

......... .........

8 we ostpA, eic

the s



udy.


sules (P � .01). Serum DHA was in-versely related to BDI scores. The modelincluding DHA, BDI score at visit 1 andadmission to having stopped capsulesaccounted for 31% of the variance in the

TABLE 6Maternal outcomes

Parameter

Gestational age at delivery (wks), mean (SD)...................................................................................................................

GDM, n (%)...................................................................................................................

Gestational hypertension or preeclampsia, n (...................................................................................................................

Induced labor, n (%)...................................................................................................................

Estimated blood loss (mL), mean (SD)...................................................................................................................

Cesarean section, n (%)...................................................................................................................

Spontaneous vaginal delivery, n (%)...................................................................................................................

Operative vaginal delivery, n (%)...................................................................................................................

ANOVA, analysis of variance; DHA, docosahexaenoic acid; EPa One-way ANOVA; b Tukey’s studentized range test; c The DH

TABLE 5Maternal fatty acids

Group Visit Serum fatty

EPA-rich fish oil 1 EPAc

..................................

DHAc

..................................

Total omega-...........................................................

3 EPAc

..................................

DHAc

..................................

Total omega-...................................................................................................................

DHA-rich fish oil 1 EPAc

..................................

DHAc

..................................

Total omega-...........................................................

3 EPAc

..................................

DHAc

..................................

Total omega-...................................................................................................................

Placebo 1 EPAc

..................................

DHAc

..................................

Omega-3 hig...........................................................

3 EPAc

..................................

DHAc

..................................

Omega-3 hig...................................................................................................................

ANOVA, analysis of variance; DHA, docosahexaenoic acid; EPa One-way ANOVA with Tukey’s multiple comparison test; b St




BDI score at visit 3. Serum EPA did notsignificantly predict the BDI score at visit3. Neither serum EPA nor DHA signifi-cantly predicted BDI scores at 6-8 weekspostpartum.


DHA-rich fish o(n � 38)

39.1 (1.5) 40.4 (0.9).........................................................................................................................

7 (18) 1 (3).........................................................................................................................

8 (21) 2 (5).........................................................................................................................

15 (38) 12 (32).........................................................................................................................

507 (481) 508 (325).........................................................................................................................

10 (26) 12 (32).........................................................................................................................

25 (64) 25 (66).........................................................................................................................

4 (10) 1 (3).........................................................................................................................

osapentaenoic acid; GDM, gestational diabetes mellitus; NS, not

oup was significantly greater than both the EPA group and the pl

s n Mean

42 0.29.........................................................................................................................

42 4.24.........................................................................................................................

ghly unsaturated fatty acidsd 42 22.10.........................................................................................................................

38 0.41.........................................................................................................................

38 4.38.........................................................................................................................


41 0.31.........................................................................................................................

41 4.66.........................................................................................................................


37 .38.........................................................................................................................

37 6.05.........................................................................................................................


42 .34.........................................................................................................................

42 3.85.........................................................................................................................

unsaturated fatty acidsd 42 22.86.........................................................................................................................

41 0.36.........................................................................................................................

41 3.70.........................................................................................................................

unsaturated fatty acidsd 41 27.46.........................................................................................................................

osapentaenoic acid; NS, not significant; SD, standard deviation.

t t test and signed rank test; c Expressed as percent of total fatty a



gy APRIL 2013

Supplementation with DHA-rich fishoil significantly lengthened gestation(40.4 weeks) compared with EPA (39.1weeks) and placebo (39.1 weeks) (P �.0001) but did not result in significant


39.1 (1.5) � .0001a-c

..................................................................................................................

2 (5) .06d

..................................................................................................................

5 (12) .12d

..................................................................................................................

8 (20) .20d

..................................................................................................................

454 (296) .81a

..................................................................................................................

11 (28) .084d

..................................................................................................................

28 (70) .88d

..................................................................................................................

1 (3) .32d

..................................................................................................................

ificant; SD, standard deviation.

o group; d Fisher exact test.

SDSignificancevs placeboa

Significancevs visit 1b

0.18 NS —..................................................................................................................

2.30 NS —..................................................................................................................

3.72 NS —..................................................................................................................

0.43 NS � .05..................................................................................................................

2.42 NS NS..................................................................................................................

9.86 NS NS..................................................................................................................

0.24 NS —..................................................................................................................

2.29 NS —..................................................................................................................

7.73 NS —..................................................................................................................

.23 NS NS..................................................................................................................

3.77 � .05 � .05..................................................................................................................

9.71 � .05..................................................................................................................

.22 — —..................................................................................................................

1.77 — —..................................................................................................................

5.02 — —..................................................................................................................

0.40 — NS..................................................................................................................

1.62 — NS..................................................................................................................

9.55 — NS..................................................................................................................

d Expressed as percent of total highly unsaturated fatty acids.

il

......... .........

......... .........

%)......... .........

......... .........

......... .........

......... .........

......... .........

......... .........

A, eic sign

A gr aceb

acid

......... .........

......... .........

3 hi.......... .........

......... .........

......... .........

3 hi......... .........

......... .........

......... .........

3 hi.......... .........

......... .........

......... .........

3 hi......... .........

......... .........

......... .........

hly.......... .........

......... .........

......... .........

hly......... .........

A, eic

uden cids;

udy.


differences in labor inductions. Therewere no significant differences in anyother maternal outcomes (Table 6).

There were 119 babies born to con-tinuing trial participants, included 1set of twins. The twin gestation was un-diagnosed at the time of enrollment;both neonates were analyzed based onintent-to-treat principles. Supplemen-tation resulted in higher birthweight inthe DHA group compared with theEPA and placebo groups (P � .001).Five minute Apgar scores were signifi-cantly higher in the DHA group thanthe EPA and placebo groups com-bined, an effect without clinical signif-icance. There were no differences inany other neonatal outcome measureamong the groups (Table 7).

Maternal DHA-rich fish oil supple-mentation significantly increased umbil-ical cord serum DHA proportion. Um-bilical cord serum EPA and DHAconcentrations in babies born to moth-ers who had received EPA-rich fish oilwere not significantly different from um-bilical cord serum concentrations frombabies born to mothers in the placebogroup (Table 8).

COMMENTOur study did not support the hypothe-sis that EPA-rich fish oil or DHA-richfish oil would prevent depressive symp-toms in pregnant women at risk for de-pression. However, DHA levels did pre-dict BDI score at 34-36 weeks. Strengthsof our study are that it was carried out in

TABLE 7Neonatal outcomes

ParameterEPA-ric(n � 40

Birthweight (g), mean (SD) 3402 (5...................................................................................................................

One minute Apgar, mean (SD) 7.3...................................................................................................................

5-minute Apgar, mean (SD) 8.6...................................................................................................................

Cord arterial pH, mean (SD) 7.26...................................................................................................................

NICU admission, n (%) 6 (1...................................................................................................................

ANOVA, analysis of variance; DHA, docosahexaenoic acid; EPData are missing on 1 neonate who was not delivered at eithea One-way ANOVA, Kruskal-Wallis test for outliers; b Tukey’s m

exact test.


a population of women at risk for de-

pression; we chose to carry out our studyamong women with predisposition todepression so that our primary outcomewould be sufficiently frequent. Theblinded design of the trial was a strength;our analyses of side effects showed thatthe proportion of women who reportedgastrointestinal side effects with the in-tervention did not differ between thegroups, suggesting that the placebo in-tervention was appropriately masked.The comparison of fish oils that werehigh in either DHA or EPA was astrength as well. To our knowledge this isthe first study to compare fish oils of

TABLE 8Fatty acids in umbilical cord serum

Group mean

EPA (n � 33) EPAb

.....................................................

DHAb

.....................................................

Total omega-3 highlyunsaturated fatty acid

...................................................................................................................

DHA (n � 33) EPAb

.....................................................

DHAb

.....................................................


...................................................................................................................

Placebo (n � 36) EPAb

.....................................................

DHAb

.....................................................


...................................................................................................................

ANOVA, analysis of variance; DHA, docosahexaenoic acid; Edeviation.a One-way ANOVA with Tukey’s multiple comparison test; b E

total highly unsaturated fatty acids.

sh oil DHA-rich fish oil(n � 38)

3774 (438).........................................................................................................................

) 7.9 (1.8).........................................................................................................................

) 9.1 (0.2).........................................................................................................................

9) (n � 34) 7.27 (0.05) (n � 31).........................................................................................................................

2 (5).........................................................................................................................

osapentaenoic acid; NICU, neonatal intensive care unit; SD, standy hospital.

le comparisons test; c The DHA group was greater than both the



APRIL 2013 Americ

these differing compositions amongpregnant women.

A weakness of our study was that it hadstatistical power to detect differences indepressive symptoms but not in diagno-ses of major depressive disorder. Adher-ence to the protocol in our study was notideal; 14% of continuing subjects dis-continued taking their capsules duringtheir trial participation. The nonsignifi-cant increases in DHA and EPA concen-trations in the EPA-rich fish oil groupcompared with placebo suggest subopti-mal compliance in the EPA group, al-though compliance measures did not

s percent of total fatty acids

SDSignificancevs placeboa

0.49 0.50 NS.................................................................................................................

7.59 3.82 NS.................................................................................................................

25.12 7.22 NS

..................................................................................................................

0.43 0.34 NS.................................................................................................................

9.94 4.82 � .05.................................................................................................................

29.87 7.04 � .05

..................................................................................................................

0.50 0.49 —.................................................................................................................

6.23 2.58 —.................................................................................................................

22.44 7.03 —

..................................................................................................................

icosapentaenoic acid; NS, not significant; SD, standard

sed as percent of total fatty acids; c Expressed as percent of


3309 (555) � .001a-c

..................................................................................................................

7.9 (1.9) .19a

..................................................................................................................

8.9 (0.5) � .01a,d

..................................................................................................................

7.27 (0.06) (n � 31) .54a

..................................................................................................................

4 (10) .39e

..................................................................................................................

deviation.

nd placebo groups; d Mann-Whitney (Wilcoxon) test; e Fisher

, a

.........

.........

sc

.........

.........

.........

sc

.........

.........

.........

sc

.........

PA, e

xpres

h fi)

50)......... .........

(2.2......... .........

(0.8......... .........

(0.0......... .........

5)......... .........

A, eic dardr stu

ultip EPA a



stoc

Esstswdpcr


differ according to group assignment.However, we found no evidence of ben-efit in the per-protocol analysis. Thisstudy may represent outcomes thatmight be encountered in actual clinicalpractice.

We found that serum DHA concentra-tions at 34-36 weeks significantly pre-dicted BDI scores in late pregnancy butdid not predict scores at 6-8 weeks’ post-partum. All groups experienced im-provement in the BDI scores at 6-8weeks’ postpartum, suggesting a possibleplacebo effect. The soy oil placebos usedin this study contained a very smallamount of fish oil for masking as well asa small amount of alpha linolenic acid,which may be endogenously convertedto DHA and EPA, thus possibly servingas an active placebo. However, the DHAand EPA concentrations did not signifi-cantly increase over time in the placebogroup, and available literature suggeststhat the endogenous conversion of lin-olenic acid to EPA and DHA is verylow.26

Our findings are in agreement withthose of Makrides et al,11 who found nobenefit for DHA-rich fish oil for preven-tion of depressive symptoms amongpregnant women who were not selectedbased on predisposition for depression.These findings are also compatible withthose of Freeman et al,27 who found noadditional benefit for EPA-predominantfish oil compared with interpersonalpsychotherapy alone. Our findings donot agree with those of Su et al9 and ear-lier open-label studies by Freeman etal,7,8 which were suggestive of benefit.One possible reason that our results dif-fered from those of Freeman and Su isthat we tested lower doses of EPA andDHA than were used in those studies.However, Freeman’s dose-ranging trialfound that higher doses of EPA plusDHA were not more effective than lowerdoses.8

Our trial was a prevention trial carriedout among women who without a diag-nosis of MDD at entry. It is possible thatomega-3 fatty acids are effective fortreatment rather than prevention ofperinatal depression.

The prolongation of pregnancy noted
in the DHA group is consistent with

prior reports.11,28 Our study did nothow an increased need for labor induc-ions in the supplemented group, butur sample size may have been insuffi-ient to demonstrate such a difference.

In summary, we found no benefit forPA-rich fish oil or DHA-rich fish oilupplementation to prevent depressiveymptoms in pregnancy and postpar-um. We demonstrated that maternalerum DHA concentrations at 34-36eeks were significantly predictive ofepression scores at the same timeoint. Further research is needed tolarify the mechanism underlying thiselationship.

ACKNOWLEDGMENTSThe Nordic Naturals Corporation donated bothactive supplements and placebos to the trial.The Nordic Naturals Corporation had no role inthe interpretation or reporting of results. TheData Monitoring Committee included the fol-lowing: Barbara Luke, ScD, Michigan State Uni-versity; Charles Neal, MD, University of Hawaii;Mel Barclay, MD (deceased), University of Mich-igan; and Dwight Rouse, MD, Brown University.No funding or compensation was provided forservice on the Data Monitoring Committee. As-sistance for this study’s database constructionwas provided by Ms Michelle Housey, who wasa student at the University of Michigan School ofPublic Health. Her assistance to this study wasfunded by Dr Marjorie Treadwell’s University ofMichigan Outstanding Clinician Award. DeliaVazquez, MD, University of Michigan Depart-ment of Pediatrics, assisted with the study de-sign and enrollment infrastructure and had1-2% effort on the R21 grant that supported thistrial during 2 years of the trial period. The prin-cipal investigator would also like to acknowl-edge Ms Babett Riblett, RN; Ms. Mary Beaupre,LPN; Ms Joanne Wiklund, RN; Ms Lori Cri-mando, NP Joanne Bailey, CNM; Joyce Baker,CNM; and Mary McGuiness, CNM. All of thoselisted previously are employed by the Universityof Michigan and assisted with subject recruit-ment. Dr Jennifer Williams assisted with subjectrecruitment and data collection from the Inte-grated Health Associates/St Joseph MercyHospital site. None of these persons receivedany compensation or funding for their roles inthis study. f

REFERENCES1. Marcus SM. Depression during pregnancy:rates, risks and consequences—Motherisk Up-date 2008. Can J Clin Pharmacol 2009;16:e15-22.2. Sie SD, Wennink JMB, van Driel JJ, et al.Maternal use of SSRIs, SNRIs and NaSSAs:
practical recommendations during pregnancy
gy APRIL 2013

and lactation. Arch Dis Child Fetal Neonatal Ed2012;97:F472-6.3. Freeman MP, Rapaport MH. Omega-3 fattyacids and depression: from cellular mecha-nisms to clinical care. J Clin Psychiatry 2011;72:258-9.4. Hibbeln J. Seafood consumption, the DHAcontent of mothers’ milk and prevalence ratesof postpartum depression: a cross-national,ecological analysis. J Affect Disord 2002;69:15-29.5. Golding J, Steer C, Emmett P, Davis JM, Hib-beln JR. High levels of depressive symptoms inpregnancy with low omega-3 fatty acid intakefrom fish. Epidemiology 2009;20:598-603.6. Larque E, Gil-Sanchez A, Prieto-SanchezMT, Koletzko B. Omega 3 fatty acids, gestationand pregnancy outcomes. Br J Nutr 2012;107(Suppl 2):S77-84.7. Freeman M, Hibbeln J, Wisner K, WatchmanM, Gelenberg A. An open trial of omega-3 fattyacids for depression in pregnancy. Acta Neuro-psychiatrica 2006;18:21-4.8. Freeman M, Hibbeln J, Wisner K, BrumbachB, Watchman M, Gelenberg A. Randomizeddose-ranging pilot trial of omega-3 fatty acidsfor postpartum depression. Acta PsychiatrScand 2006:31-5.9. Su KP, Huang SY, Chiu TH, et al. Omega-3fatty acids for major depressive disorder duringpregnancy: results from a randomized, double-blind, placebo-controlled trial. J Clin Psychiatry2008;69:644-51.10. Ramakrishnan U, Stein AD, Parra-CabreraS, et al. Effects of docosahexaenoic acid sup-plementation during pregnancy on gestationalage and size at birth: randomized, double-blind,placebo-controlled trial in Mexico. Food NutrBull 2010;31(Suppl 2):S108-16.11. Makrides M, Gibson RA, McPhee AJ, et al.Effect of DHA supplementation during preg-nancy on maternal depression and neurodevel-opment of young children: a randomized con-trolled trial. JAMA 2010;304:1675-83.12. Ramakrishnan U. Fatty acid status and ma-ternal mental health. Matern Child Nutr 2011;7(Suppl 2):99-111.13. Martins JG. EPA but not DHA appears to beresponsible for the efficacy of omega-3 longchain polyunsaturated fatty acid supplementa-tion in depression: evidence from a meta-anal-ysis of randomized controlled trials. J Am CollNutr 2009;28:525-42.14. Jans LA, Giltay EJ, Van der Does AJ. Theefficacy of n-3 fatty acids DHA and EPA (fish oil)for perinatal depression. Br J Nutr 2010;104:1577-85.15. Sublette ME, Ellis SP, Geant AL, Mann JJ.Meta-analysis of the effects of eicosapenta-enoic acid (EPA) in clinical trials in depression.J Clin Psychiatry 2011;72:1577-84.16. Mozaffari-Khosravi H, Yassini-Ardakani M,Karamati M, Shariati-Bafghi SE. Eicosapenta-enoic acid versus docosahexaenoic acid inmild-to-moderate depression: a randomized,
double-blind, placebo-controlled trial. [pub-


lished online ahead of print August 18, 2012].Eur Neuropsychopharmacol 2012.17. Mozurkewich E, Chilimigras J, Klemens C,et al. The mothers, omega-3 and mentalhealth study. BMC Pregnancy Childbirth 2011;11:46.18. Buist A, Condon J, Brooks J, et al. Accept-ability of routine screening for perinatal depres-sion. J Affect Disord 2006;93:233-7.19. Sheehan DV, Lecrubier Y, Sheehan KH, etal. The Mini-International Neuropsychiatric In-terview (M.I.N.I.): the development and valida-tion of a structured diagnostic psychiatric inter-view for DSM-IV and ICD-10. J Clin Psychiatry1998;59(Suppl 20):22-33.20. Holcomb Jr W, Stone L, Lustman P, GavardJ, Mostello D. Screening for depression in preg-nancy: characteristics of the Beck Depression
Inventory. Obstet Gynecol 1996;88:1021-5.
21. Beck CT, Gable RK. Comparative analysisof the performance of the Postpartum Depres-sion Screening Scale with two other depressioninstruments. Nurs Res 2001;50:242-50.22. Flynn H, Blow F, Marcus S. Rates andpredictors of depression treatment amongpregnant women in hospital-affiliated obstet-rics practices. Gen Hosp Pyschiatry 2006;28:289-95.23. Steer R, Scholl T, Beck A. Revised BeckDepression Inventory scores of inner-city ado-lescents: pre- and postpartum. Psychol Rep1990;66:315-20.24. Marcus S, Lopez J, McDonough S, et al.Depressive symptoms during pregnancy: im-pact on neuroendocrine and neonatal out-comes. Infant Behav Dev 2011;34:26-34.25. Djuric Z, Ren J, Blythe J, VanLoon G, Sen A.
A Mediterranean dietary intervention in healthy
APRIL 2013 Americ

American women changes plasma carotenoidsand fatty acids in distinct clusters. Nutr Res2009;29:156-63.26. Gibson RA, Muhlhausler B, Makrides M.Conversion of linoleic acid and alpha-linolenicacid to long-chain polyunsaturated fatty acids(LCPUFAs), with a focus on pregnancy, lacta-tion and the first 2 years of life. Matern Child Nutr2011;7(Suppl 2):17-26.27. Freeman MP, Davis M, Sinha P, Wisner KL,Hibbeln JR, Gelenberg AJ. Omega-3 fatty acidsand supportive psychotherapy for perinatal de-pression: a randomized placebo-controlledstudy. J Affect Disord 2008;110:142-8.28. Makrides M, Duley L, Olsen SF. Marine oil,and other prostaglandin precursor, supplemen-tation for pregnancy uncomplicated by pre-ec-lampsia or intrauterine growth restriction. Co-
chrane Database Syst Rev 2006:CD003402.

Documents

The Mothers, Omega-3, and Mental Health Study: a double-blind, randomized controlled trial