- Top of page
Objectives To test the earlier suggested hypothesis that intake of long-chain n-3 fatty acids from fish oil may delay the timing of spontaneous delivery and to test if alpha-linolenic acid, provided as flax oil capsules, shows the same effect.
Design Randomised controlled trial including women reporting low dietary fish intake. The women were allocated in the proportions of 1:1:1:1:1:1:2 into six treatment groups and a control group, respectively, from week 17–27 of gestation. The treatment groups received fish oil, in various doses, or flax oil, and the control group did not receive any treatment.
Setting The Danish National Birth Cohort.
Sample A total of 3098 women allocated into six treatment groups and one control group.
Methods The six intervention groups were offered fish oil capsules in doses of 0.1, 0.3, 0.7, 1.4 and 2.8 g of eicosapentaenoic acid and docosahexaenoic acid per day or 2.2 g of alpha-linolenic acid (ALA) per day from week 17–27 of gestation until delivery.
Main outcome measures Timing of spontaneous delivery.
Results No differences in timing of spontaneous delivery was detected in the fish oil groups or the flax oil group, compared with the control group. The difference in timing of spontaneous delivery in the group receiving the highest fish oil dose compared with the control group was 0.8 days (95% CI: –2.3 to 1.0). Only a minority of the women in the intervention groups took capsules until delivery.
Conclusion Possible explanations for these findings include no true effect of n-3 fatty acids on spontaneous delivery or a quick-acting effect not detectable in this trial.
- Top of page
Today, preterm delivery (birth before day 259 of gestation) is a major problem in obstetrics, and it accounts for 70% of the perinatal mortality1 and 50% of the congenital neurologic disability, including cerebral palsy.2 In Denmark, preterm deliveries account for approximately 7% of all liveborn infants.3
When significantly higher birthweights and longer gestation durations have been observed among Faroese women compared with Danish women,4,5 it was suggested that the difference was due to the higher consumption of marine diet by the Faroese women,5 which contains large amounts of the long-chain n-3 fatty acids, namely eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA). These fatty acids were suggested to delay the onset of spontaneous delivery by interfering with arachidonic acid (AA) in the formation of prostaglandins E and F, which are involved in the parturition process.5,6 Three randomised trials where pregnant women were supplemented with either fish oil or a placebo7–9 and one trial where DHA-enriched eggs were provided,10 found an effect of long-chain n-3 fatty acids on gestation length. Observational studies, where dietary intake of fish in pregnancy was assessed,11,12 have supported the hypothesis. However, other intervention13–15 and observational16–20 studies did not find an association.
An observational study showed positive association between level of long-chain n-3 fatty acids in erythrocyte and length of gestation in pregnant Danish women, but no such association was found in Faroese women, who have a higher intake of marine diets compared with the Danish women.11 This led to the hypothesis that the effect of n-3 fatty acids on duration of pregnancy meets a saturation point, above which no further effect of long-chain n-3 fatty acids can be detected.12,21 The hypothesis was supported in two randomised trials in the UK and USA, where pregnant women were supplemented with low doses of long-chain n-3 fatty acids and significant effects on gestation length were found,7,10 and in one observational study, where gestation length increased across the intake range from 0 to 0.15 g of fatty acids per day above which no further increases in gestation length was seen.11,12 No information of baseline fish consumption exists in these trials, but it is assumed to be lower than in the Scandinavia.
Flax oil contains large amount of alpha-linolenic acid, an 18-carbon n-3 fatty acid. Through elongation and desaturation, alpha-linolenic acid can be converted into EPA and DHA and may therefore theoretically have similar effects of the long-chain n-3 fatty acids on spontaneous delivery.
Our aim was, in a randomised trial, to study the effect of fish oil supplementation on timing of spontaneous delivery among healthy, pregnant Danish women with a low dietary intake of fish and to examine the dose–response relationship of the effect of fish oil supplementation on the timing of spontaneous delivery. In addition, we wanted to investigate if alpha-linolenic acid has an effect on timing of delivery as expected from the effects attributed to the long-chain n-3 fatty acids.
- Top of page
The hypotheses were tested in a randomised controlled trial. The study included seven groups: five fish oil groups, a group receiving flax oil and a control group that did not receive any treatment. The study population was a subgroup of the Danish National Birth Cohort (DNBC), which includes 100 000 pregnant women and their offspring.22 Information on socio-economic and lifestyle factors was collected through four telephone interviews; two during pregnancy and two when the child was 6 and 18 months, respectively.22 Inclusion criteria for the randomised study were limited fish intake, no use of fish oil capsules in pregnancy and being in gestation week 17–27. From approximately 16 000 women (recruited from October 2001 and onwards), those within the lowest 20% of fish consumption, as reported in the first telephone interview, performed in week 12 of gestation, were selected. In this group, block-wise stratified randomisation was carried out in strata defined by fish intake (no/low), smoking status (never/ever), pre-pregnant body mass index (BMI: <30/≥30) and women's height (</>160 cm). The women were randomised to either one of the six treatment groups or the control group in the proportions 1:1:1:1:1:1:2, respectively. A flow chart of the trial is shown in 1Figure 1. A total of 3098 women were randomised, distributed as on average 387 women in each of the treatment groups and 774 women in the control group. This sample size gave a statistical power of 90% (P < 0.05) to detect a difference of 4 days in mean gestation length in each pair-wise comparison with the control group, which was the difference in gestation length found between Faroese women and Danish women.4 After randomisation, the women in the treatment groups were invited to take part in the study by a mailed information letter including an enrolment form, by which the women had to give their written consent. If the women accepted to participate, the capsules were distributed by mail. The package included a written information on when and how to take the capsules and a timetable to tick off when the capsules had been taken. The timetable was enclosed as a reminder and its use was optional. The participants were asked to start taking the capsules on the day they received them and to take them until the expected date of delivery or until delivery.
Figure 1. Flow chart of the randomised controlled trial. Women were randomised into six intervention groups and one control group in the ratio 1:1:1:1:1:1:2. ’Participants’ are defined as women who accepted the invitation to receive the capsules.
Download figure to PowerPoint
The doses in the fish oil groups were 0.5 g three times per week, 0.5 g per day, 1 g per day, 2 g per day or 4 g per day of Futura Fish Oil™ (32% EPA, 22% DHA and 13.4 mg D-alpha-tocopherol per gram in 0.5- or 1-g capsules; Dansk Droge A/S, Ishoej, Denmark), which corresponds to 0.1, 0.3, 0.7, 1.4 and 2.8 g of long-chain n-3 fatty acids per day, respectively. The sixth group received four 1-g capsules of flax oil (Prima Flax™, Bioriginal Food & Science Corp., Saskatoon, Canada), which supplied the participants with 2.2 g of alpha-linolenic acid per day. The women in the control group were allocated to any treatment and were not contacted at all. The groups were named according to the dosage of fatty acids in the capsules they received, i.e. 01, 03, 07, 14 and 28. The group that received flax oil were named C18 and the control group was named CG. The study was approved by the scientific ethical committee, Copenhagen, Denmark (KF 11-044/01). All the participants gave a written informed consent according to the Helsinki II Declaration.
On average, 37 days after delivery (range: 6–140 days), all the participants who had delivered a liveborn child were sent a questionnaire concerning dates of start and stop of taking the capsules and daily dosage. They were asked to state the date of expected delivery and if they had stopped taking the capsules before birth.
As a part of the routine data collection in DNBC, all the participants were sent a 360-item food frequency questionnaire (FFQ) in week 25 of gestation. The questionnaire was semiquantitative and covered dietary intake in the previous 4 weeks, including detailed information on the participants’ fish consumption. Daily intakes of long-chain n-3 fatty acids were estimated using information on n-3 fatty acid content in fish species from the Danish food tables23 and average standard recipes and portion sizes. The FFQ was validated against a 7-day-weighed food record.24
On a consecutive sample of participants living in particular areas (n= 62, 6–15 in each treatment group), blood samples were collected. The blood samples were drawn on average 87 days (32–129 days) after the women had received the capsules. After sampling, the erythrocytes were washed and the lipids were extracted according to the Folch procedure.25 Membrane phospholipids were methylated according to Hamilton and Hamilton.26 Fatty acid methyl esters (FAME) were separated and quantified by gas chromatography (Hewlett-Packard Inc., Waldbronn, Germany). FAME were identified by comparing their retention times with that of standards (NU-Check-Prep Inc., Elysian, MN, USA).
Gestational age length was primarily assessed from the last menstrual period (LMP) on the basis of information recorded in the recruitment form and in the first interview. The date of birth was extracted from the Danish Civil Registration System. When the LMP-based estimate was uncertain due to irregular or abnormally long or short menstrual cycles ordue to use of contraceptive pills within the last 3 months prior to LMP, gestational age was based on information on expected date of delivery provided by the women in the second interview, which took place around gestation week 30; most of the latter estimates are expected to be based on early ultrasound scanning. If this information was not valid either, due to missing interviews or due to unrealistic gestational age estimates (more than 308 days), information on estimated gestation length assessed at delivery by the midwife and reported to the National Patient Registry was used. Information on mode of onset of delivery was extracted from the National Patient Registry, which contains all codes for hospital admissions. Elective births comprised induced labour and caesarean sections before onset of labour.
Gestational age at delivery (in days) in each treatment group was compared with gestational age in the control group by t test. Effect of the capsules on timing of spontaneous delivery was examined by Cox regression analysis regarding elective deliveries as censoring events. Dichotomous variables were analysed by means of chi-square test. Analyses were primarily performed as intention to treat according to the randomisation. A second series of analysis was undertaken where the women in the treatment groups, who accepted to participate, were compared with the women in the control group. Intakes of dietary long-chain n-3 fatty acids as well as ratios of erythrocyte EPA + DHA to AA as suggested to show the best association with fish consumption27 in the six treatment groups were compared by means of analysis of variance. Association between erythrocyte EPA + DHA/AA and length of gestation was analysed by means of linear regression. All the analyses were carried out using SAS version 9.1 for Windows (SAS Institute Inc., Cary, NC, USA).
- Top of page
A total of 3098 women were eligible for the study. No differences was seen between the groups with respect to the four variables used in the randomisation (smoking, fish consumption, BMI and women's height) or to the women's age, parity and weeks of gestation at randomisation (1Table 1).
Table 1. Baseline characteristics in the seven randomisation groups (total 3098 women)
|Participation rate (%)||60.2||60.0||59.2||58.2||49.6||46.3||—|
|Overweight, n (%)||128 (32.9)||129 (33.5)||129 (33.5)||128 (33.4)||130 (33.1)||128 (32.9)||259 (33.5)|
|Smoked during pregnancy, n (%)||79 (20.3)||78 (20.3)||78 (20.3)||79 (20.6)||78 (19.9)||79 (20.3)||160 (20.7)|
|Height of <160 cm, n (%)||45 (11.6)||44 (11.4)||42 (10.9)||47 (12.3)||43 (10.9)||44 (11.3)||89 (11.5)|
|No fish consumption, n (%)||56 (14.4)||54 (14.0)||54 (14.0)||56 (14.6)||55 (14.0)||53 (13.6)||112 (14.5)|
|Age in years, mean (SD)||28.4||28.7||28.4||28.9||28.8||28.8||28.5|
|Weeks of gestation, mean (SD)||22.6||22.5||22.5||22.5||22.7||22.7||22.4|
|Primiparous women, n (%)||257 (66.2)||267 (69.5)||244 (63.5)||247 (64.7)||246 (62.9)||258 (66.3)||513 (66.4)|
|Twins, n (%)||7 (1.8)||7 (1.8)||4 (1.0)||10 (2.6)||9 (2.3)||6 (1.5)||9 (1.2)|
|Abortions, n (%)||5 (1.3)||6 (1.6)||10 (2.6)||6 (1.6)||4 (1.0)||8 (2.1)||11 (1.4)|
Mean gestation length in the five fish oil groups and the flax oil group was compared with mean gestation length in the control group (Student's t test). Cox regression analyses, with elective births as censoring event, were performed, and hazard ratios are shown in 23Tables 2 and 3. Only liveborn singleton pregnancies were included in the analyses. No significant differences in gestation length was detected between any of the treatment groups and the control group, neither when comparing mean gestational age or hazard rates of spontaneous delivery. This applied both to the intention-to-treat analyses (Table 2) and to the analyses based on the participants only (Table 3).
Table 2. Intention-to-treat comparisons of gestation length in each of the six intervention groups with the control group. Mean gestational age at delivery and hazards rate ratios with elective deliveries as censored data are shown. P values indicate comparisons between the respective intervention group and the control group
|Group||n||Gestational age||SD||P value*||Hazards rate ratio||95% CI||P value**|
|CG||748||280.6||11.7|| ||1.00|| |
Table 3. Comparisons of gestation length in each of the six intervention groups with the control group; nonparticipants were excluded. Mean gestational age at delivery and hazards rate ratios with elective deliveries as censored data are shown. P values indicate comparisons between the respective intervention group and the control group
|Group||n||Gestational age||SD||P value*||Hazards rate ratio||95% CI||P value**|
|CG||748||280.6||11.7|| ||1.00|| |
Approximately 75% of the women in the cohort completed the FFQ. The participants in this subgroup had a mean dietary intake of 0.3 g of long-chain n-3 fatty acids. No differences in dietary intake of any of the fatty acids was found between the groups (4Table 4). To analyse if our data support the earlier proposed hypothesis that a positive effect of n-3 fatty acids on length of gestation is below dietary intake of 0.15 g per day,12 we excluded all women having a dietary intake of EPA and DHA (estimates based on FFQ) above this value from the analyses (5Table 5). In the analyses including all randomised women (intention to treat), shorter gestations were found in groups 07 and 14. The hazard ratio was, however, only significant in group 07. In the analyses including only the women who agreed to participate and the control group, no differences was detected in the gestation length.
Table 4. Mean intake of selected fatty acids from the diet (exclusive supplements) in the seven groups. Intakes were estimated by the FFQ in week 25 of gestation
|Group||n||EPA (g/day)||DHA (g/day)||EPA + DHA (g/day)||ALA (g/day)||AA (g/day)|
|Mean (SD)||Mean (SD)||Mean (SD)||Mean (SD)||Mean (SD)|
|01||295 (76.1)*||0.080 (0.06)||0.20 (0.17)||0.29 (0.24)||1.60 (0.72)||0.085 (0.06)|
|03||300 (77.9)*||0.075 (0.07)||0.20 (0.16)||0.27 (0.23)||1.60 (0.77)||0.080 (0.05)|
|07||303 (78.7)*||0.086 (0.11)||0.22 (0.23)||0.31 (0.33)||1.58 (0.70)||0.086 (0.06)|
|14||284 (74.2)*||0.081 (0.07)||0.20 (0.17)||0.29 (0.24)||1.63 (0.80)||0.081 (0.05)|
|28||293 (74.6)*||0.095 (0.19)||0.24 (0.46)||0.34 (0.65)||1.63 (0.82)||0.084 (0.06)|
|C18||277 (71.2)*||0.084 (0.08)||0.21 (0.18)||0.30 (0.26)||1.57 (0.80)||0.085 (0.06)|
|CG||587 (75.8)*||0.085 (0.10)||0.22 (0.23)||0.31 (0.33)||1.62 (0.82)||0.083 (0.05)|
Table 5. Intention-to-treat comparisons of gestation length in the six treatment groups and the control groups. The analysis was restricted to only women with estimated intake of n-3 fatty acids < 0.15 g per day (FFQ, week 25)
|Group||n||Mean gestational age||SD||P value*||HR||95% CI||P value**|
|CG||203||281.1||10.9|| ||1.00|| |
|Intention to treat|
Eighty-one percent of the participants (equally distributed in the treatment groups) fulfilled and returned the questionnaire about compliance. The women who had given birth before or at expected date of delivery stopped taking the capsules on average 12 days before delivery.
Ratios of erythrocyte EPA + DHA to AA in groups 01, 03, 07 and 14 were compared with group 28. The women in group 28 had significantly higher EPA + DHA/AA ratio in erythrocytes than the women in groups 01, 03, 07 and C18, while no difference was detected between groups 28 and 14 (6Table 6). No linear association was detected between the EPA + DHA/AA ratio and the length of gestation (data not shown).
Table 6. Level of EPA, DHA and AA in erythrocyte phospholipids assessed in a subsample of women in the six treatment groups. All intervention groups were compared with group 28
|Group||n||EPA||DHA||EPA + DHA||AA||Ratio (EPA + DHA)/AA||P value*|
|Mean (SD)||Mean (SD)||Mean (SD)||Mean (SD)||Mean (SD)|
|01||9||0.46 (0.21)||4.70 (0.96)||5.16 (1.06)||10.09 (0.94)||0.51 (0.08)||>0.001|
|03||9||0.43 (0.17)||4.27 (0.80)||4.55 (0.72)||9.16 (1.20)||0.52 (0.10)||>0.001|
|07||16||0.68 (0.19)||4.95 (1.02)||5.63 (1.18)||9.40 (1.46)||0.60 (0.09)||0.003|
|14||12||1.16 (0.35)||5.82 (1.34)||6.99 (1.58)||9.31 (1.88)||0.75 (0.07)||0.409|
|28||4||1.11 (0.11)||5.66 (0.83)||6.76 (0.92)||8.51 (0.64)||0.79 (0.08)||—|
|C18||13||0.48 (0.23)||4.06 (0.84)||4.53 (0.99)||9.35 (0.93)||0.49 (0.09)||>0.001|
- Top of page
Contrary to our expectation, we did not find any effect of fish oil or flax oil supplementation on timing of spontaneous delivery.
The strengths of the study were its large size and its design. We had the opportunity to select a large group of women based on their fish intake assessed in the first trimester. Moreover, we had access to a number of potential confounders collected routinely as a part of DNBC, some of which were included in the stratified block-wise randomisation. The data on birth outcome are regarded as unbiased since data were obtained from the National Patient Registry on birth outcome. Use of fish oil supplements (as well as other supplements) throughout pregnancy was assessed, which enabled us to make sure that the women were not taking other fish oil supplements than the ones we provided. The women were recruited from the entire country of Denmark. Risk of contamination bias across the treatment groups (cointervention bias) and in the control group was regarded as negligible as women allocated to one treatment were unlikely to know about treatment in the other groups. This justified having a control group that did not receive a placebo. Since no control treatment was introduced, there was no possibility of a shortening effect on length of gestation of the placebo, a principal possibility discussed in relation to one earlier trial in the field, where significantly shorter gestation durations were detected in women receiving olive oil capsules compared with women receiving fish oil capsules.9
A weakness compared with some of the earlier trials that showed an effect,8,9 where women were recruited by face-to-face meetings and contact with the participants was kept through repeated meetings, was that we had no personal contact with the women, thereby only limited opportunities to encourage the women to take the capsules.
Taking into account the results from earlier intervention studies, two interpretations of the findings from the present study are considered. The first assumption is that there is no true effect of long-chain n-3 fatty acids on timing of spontaneous delivery, and the second is that there is an effect, but it could not be revealed with the design applied in this study. The first interpretation is in line with the findings of no association in two earlier intervention trials13,15 and four observational studies.16–19 This would require reinterpretation of the two largest earlier intervention trials in the field,8,9 which showed a delayed delivery in women receiving fish oil compared with women receiving olive oil. An alternative explanation of the findings from these studies is that the differences seen between the two groups were caused by a shortening effect of olive oil, rather than a prolonging effect of the fish oil. Furthermore, if the hypothesis is not true, associations between short gestations and low fish intake found in observational studies4,12 must be explained by alternative mechanisms such as other substances in the marine diet responsible for the association with birth outcome and other unmeasured and/or uncontrolled confounding.
It has been suggested that the potential effect of long-chain n-3 fatty acids on gestation length may be a quick-acting effect,28 which is supported by some evidence from animal experiments.29 The women in this study were asked to stop taking the capsules on the date of expected delivery if they had not yet given birth. Approximately 50% of the women gave birth before or at expected date of delivery. Of these, 50% accepted to participate, and of these 27% stopped taking the capsules before birth. Accordingly, only a minority (approximately 23%) of the women took capsules until delivery. If the effect of the long-chain n-3 fatty acids is quick acting and of short duration, this could explain why we did not see an association.
Another possible reason that we did not find an association could be due to lack of effect of the fish oil preparation provided. We measured fatty acids in erythrocytes in a small subsample of the study population, in which we found a good correlation between the ratio of EPA + DHA to AA in erythrocytes and the groups the women were allocated to. However, no association could be detected between the biomarker and the length of gestation, which is in contrast to an earlier biomarker study.11 The discrepancy could be because there is no effect of the n-3 fatty acid on the timing of delivery, because the association could not be detected within this range of intake or, provided that the effect is fast, because a large proportion of the women stopped taking capsules after they had given the blood sample and before they delivered.
In summary, this study exhibited no association between intake of long-chain n-3 fatty acids or alpha-linolenic acid and timing of spontaneous delivery. This may reflect lack of any such effect. On the other hand, if there is an effect, it is more likely to be fast and rely on continued supplementation. We recommend to optimising and measuring compliance in future trials in this field, by achieving personal contact with the participants. Further, a new intervention trial may include two placebo groups, one consisting of olive oil and the other of low linoleic acid containing safflower oil.
- Top of page
Financial support for the DNBC was obtained from the Danish National Research Foundation, Pharmacy Foundation, Egmont Foundation, Augustinus Foundation and the Health Foundation. The dietary component was supported by March of Dimes Birth Defects Foundation, European Union (QLK1-2000-00083), the Danish Medical Research Foundation, the Heart Foundation and the Health Foundation. Dansk Droge A/S and Bioriginal Food Science Corp. donated the fish oil and the flax oil capsules, respectively. Dansk Droge A/S also were in charge of distribution of the capsules to the participants. The managerial team of the DNBC consisted of Jørn Olsen (chair), Mads Melbye, Anne Marie Nybo Andersen, Sjurdur F Olsen, Thorkild IA Sørensen and Peter Aabye.