Intakes of Garlic and Dried Fruits Are Associated with Lower Risk of Spontaneous Preterm Delivery^,

Abstract

Several studies have found associations between microbial infections during pregnancy and preterm delivery (PTD). We investigated the influence of food with antimicrobial and prebiotic components on the risk of spontaneous PTD. A literature search identified microbes associated with spontaneous PTD. Subsequently, 2 main food types (alliums and dried fruits) were identified to contain antimicrobial components that affect the microbes associated with spontaneous PTD; they also contained dietary fibers recognized as prebiotics. We investigated intake in 18,888 women in the Norwegian Mother and Child Cohort (MoBa), of whom 950 (5%) underwent spontaneous PTD (<37 gestational weeks). Alliums (garlic, onion, leek, and spring onion) [OR: 0.82 (95% CI: 0.72, 0.94), P = 0.005] and dried fruits (raisins, apricots, prunes, figs, and dates) [OR: 0.82 (95% CI: 0.72, 0.94); P = 0.005] were associated with a decreased risk of spontaneous PTD. Intake of alliums was related to a more pronounced risk reduction in early spontaneous PTD (gestational weeks 28–31) [OR: 0.39 (95% CI: 0.19, 0.80)]. The strongest association in this group was with garlic [OR: 0.47 (95% CI: 0.25–0.89)], followed by cooked onions. Intake of dried fruits showed an association with preterm prelabor rupture of membranes (PPROM) [OR: 0.74 (95% CI: 0.65, 0.95)]; the strongest association in this group was with raisins [OR: 0.71 (95% CI: 0.56, 0.92)]. The strongest association with PPROM in the allium group was with garlic [OR: 0.74 (95% CI: 0.56, 0.97)]. In conclusion, intake of food with antimicrobial and prebiotic compounds may be of importance to reduce the risk of spontaneous PTD. In particular, garlic was associated with overall lower risk of spontaneous PTD. Dried fruits, especially raisins, were associated with reduced risk of PPROM.

Introduction

Preterm delivery (PTD)⁶ accounts for most adverse pregnancy outcomes, including most neonatal deaths and a substantial proportion of neonatal complications (1, 2). According to the Child Health Epidemiology Reference Group of the WHO and UNICEF, PTD is one of the most important causes of neonatal deaths globally (3). In the US, the rate is reported to be 12–13% (4, 5), whereas rates range from 5.6 to 6.4% (6) in Scandinavian countries.

PTD is divided into spontaneous PTD and medically induced PTD. Spontaneous PTD is subdivided into preterm labor (PTL), i.e., labor starting with contractions, and preterm prelabor rupture of membranes (PPROM) (5, 7).

Spontaneous PTD is related to different infections and inflammatory conditions in the genital tract, estimated to account for ∼25– 40% of PTDs (8–12). In addition, bacterial infections are more strongly implicated in early PTD, in which the highest proportion of adverse outcomes is found (13). Concentrations of PGE₂ in amniotic fluid increases throughout pregnancy, and when reaching a critical threshold labor is induced concomitant with cervical dilation and delivery (14). The inflammatory process causes TNF-α and IL-1β to increase and stimulate PGE₂ production. Elevated concentrations of PGE₂ in amniotic fluid are a consistent indicator of PTD (15, 16). With this rationale, infections in any part of the body might influence the reproductive system. Infections have different routes of access to the intrauterine environment, the most frequent of which is ascending from the lower genital tract; a hematogenous route via the placenta and a retrograde route from the abdominal cavity are others. The function and size of the components of interest in this study are comparable to antibiotics when released from the gut into the blood flow and are presumably able to penetrate similar routes as infection, as described above.

Already in the gut these components modulate the microbiota through their activity. An intriguing aspect of this is that the lower genital tract is exposed to rectal bacteria, which function as a source of vaginal bacteria. Modulation of gut microbiota and rectal-vaginal transfer of bacteria is perhaps the most interesting in relation to potential causes of lower genital tract infections. An example is a recent study of how orally given lactobacilli is found to colonize both rectally and vaginally, affecting vaginal pH and Nugent score (microflora) (17). Similarly, modifications in gut microbiota, here suggested by antimicrobial or prebiotic components, adjust which bacteria are available to potentially colonize the vaginal tract.

Bacterial vaginosis (BV) is one of the main bacterial conditions in the lower genital tract associated with PTD (14, 15), but abnormal vaginal flora and anaerobic vaginitis also have different degrees of implication for PTD (18). Studies of vaginal infections have found microbial differences between the clinical subgroups PTL and PPROM. Gardnerella vaginalis, in particular, but also Ureaplasma urealyticum, Mycoplasma hominis, Bacteroides spp., Escherichia coli, Klebsiella spp., Haemophilus spp., and Staphylococcus aureus are increased in PTL, whereas U. urealyticum, as well as Bacteroides spp., Klebsiella spp., and Haemophilus spp. are associated with PPROM (19–21).

Trichomonas vaginalis infections are associated with PTD and low birth weight in a number of studies (22–24). Chlamydia trachomatis is strongly implicated in early PTD when concurrent with inflammation of the placenta (25–27).

Furthermore, PTD is associated with urinary tract infections during pregnancy, most commonly involving E. coli, but also other gram-negative and gram-positive bacteria such as S. aureus (8, 28–31).

Amniotic fluid infection in preterm labor is associated with BV and intermediate vaginal flora (32). Among the bacteria identified in the amniotic cavity are those in the genus Mycoplasma, the most common of which are U. urealyticum and, to a lesser extent, M. hominis. Other bacteria can also be detected in the amniotic cavity, i.e., Streptococcus agalactiae, E. coli, G. vaginalis, and S. aureus as well as most bacteria mentioned above related to lower genital tract infections, with the addition of fungus, such as Candida albicans (13, 33, 34).

Microbial invasion of the amniotic cavity and activation of the innate immune system are a partial explanation of the preterm parturition syndrome. Combined, the endotoxins from the microbes and proinflammatory cytokines from the innate immune system increase the production of cytokines and other inflammatory mediators such as prostaglandins, resulting in augmentation of uterine contractility in a process leading toward PTL, while simultaneously, additional inflammatory mediators influence the degradation of extracellular matrix in the fetal membranes, contributing to parallel processes leading toward PPROM (35, 36).

An increase in the PTD rate has been observed in some intervention studies after treatment for pathogens associated with PTD (37–40), which has been interpreted as a disruption in the lower genital tract microbial flora balance caused by antibiotics.

In recent literature, probiotics have been shown to be associated with a decreased risk of PTD (41). It has been suggested that some probiotic lactic acid bacteria may have antiinflammatory properties in addition to their more well known antimicrobial effects (42). Alliums, in particular garlic, have well-documented antimicrobial effects (43–45) and are high in prebiotic dietary fibers (46), which stimulate the presence of probiotics in the gut. These prebiotics are defined as “a selectively fermented ingredient that allows specific changes, both in the composition and/or activity in the gastrointestinal microflora that confers benefits upon host well-being and health” (47). This definition can be strictly applied to trans-galactooligosaccharide and inulin, as well as more loosely to fructooligosaccharide and lactulose (48). Dried fruits such as raisins, apricots, prunes, figs, and dates are high in these types of dietary fibers and exhibit antimicrobial activity against some microbes of particular interest for this study, such as E. coli, S. aureus, S. mutans, Enterococcus faecalis, and C. albicans (49–51).

We hypothesized that dietary components might have a protective effect against pathogenic microbes that cause pregnancy complications associated with PTD. The aim of this study was to investigate the association between maternal intake of antimicrobial and prebiotic-containing foods and spontaneous PTD.

Participants and Methods

Participants.

The Norwegian Mother and Child Cohort Study (MoBa) is a population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health (52). Participants were recruited from all over Norway during 1999–2008, and 38.7% of invited women consented to participate. The cohort includes 107,383 pregnancies (90,996 mothers and 109,027 children).

This study is based on version 4 of the quality-assured MoBa data files released for research in 2008. Medical Birth Registry of Norway data were included in the data set (53), together with data from 2 self-reported questionnaires answered during pregnancy. The first was a general questionnaire (questionnaire 1) answered in gestational week 15; the second was an FFQ (questionnaire 2) answered in gestational weeks 17–22. Questionnaire 1 covered pregnancy, maternal health, and lifestyle factors that were necessary as background information for strict inclusion, exclusion, and model covariates. Women aged 20–35 y with singleton pregnancies were included; cases underwent spontaneous PTD ≥22 to < 37 wk of gestation, and controls delivered at ≥39 to < 40 wk of gestation. We excluded pregnancies with preexisting medical conditions such as diabetes, hypertension, autoimmune disease, inflammatory bowel disease, systemic lupus erythematous, rheumatoid arthritis, and scleroderma, as well as pregnancies complicated by preeclampsia, hypertension, diabetes, small-for-gestational-age development (according to intrauterine growth curves), placental abruption, placenta previa, or cervical cerclage. We also excluded children born with serious malformations.

Informed consent was obtained from each MoBa participant upon recruitment. The study was approved by the Regional Committee for Medical Research Ethics and the Data Inspectorate in Norway.

Dietary information.

The MoBa FFQ (questionnaire 2) is a semiquantitative questionnaire designed to capture dietary habits and intake of dietary supplements during the first 4–5 mo of pregnancy (54) (http://www.fhi.no/dokumenter/253304bd64.pdf). A validation study has demonstrated that the MoBa FFQ is a valid tool for estimating habitual intake and ranking pregnant women according to high and low intakes of energy, nutrients, and foods compared with a dietary reference and selected biological markers, including iodine as a marker of milk and dairy intake (55, 56).

On the basis of this evaluation, consumption frequencies were converted into food amounts (g/d) based on standard Norwegian portion sizes. FoodCalc (57) and the Norwegian Food Composition table were used to calculate food and nutrient intakes.

We applied an extensive bipartite literature search first to identify microbes involved in preterm delivery (or pregnancy complications associated with preterm delivery) and second to identify which of the 225 food types available/included in the questionnaires contained known antimicrobial components active toward any of the microbes potentially involved in PTD and simultaneously contained prebiotics.

We identified several food types containing known antimicrobial components with potential effects against microbes associated with or implicated in PTD for analysis as follows: olive oil (for cooking), honey, dried fruits (raisins, apricots, prunes, figs, and dates), wine (white or red), tea (black and green tea analyzed separately), coffee, alliums (garlic, onion, leek, and spring onion), and homemade salad dressing with olive oil. Other food types containing antimicrobial compounds such as herbs, with potential effects on pregnancy complications, were identified in the literature search. However, questions about fresh or dried herbs were not included in the MoBa food questionnaire. We selected alliums and dried fruits for this investigation on the basis of their combined antimicrobial and prebiotic qualities.

The following description of antimicrobial activities in alliums and dried fruits, identified through the literature search, is by no means exhaustive. Alliums consist of garlic, onion, leek and spring onion. Garlic contains antimicrobial components that act against a broad spectrum of microorganisms, ranging from E. coli, S. aureus, E. faecalis, and Klebsiella pneumoniae, and has antifungal activity against Candida spp., as well as being active against Trichomonas vaginalis as shown from in vitro studies (43, 44, 58–60). Extracts of onions have known in vitro activity against a range of gram-positive bacteria as well as the gram-negative K. pneumoniae and exhibit antifungal activity as well (45). The dried fruits category consists of raisins, apricots, prunes, figs, and dates. Components of raisins (Vitis vinifera) contain active in vitro activity against S. mutans, a bacterium found in previous studies on BV and PTD (49). Extracts of apricots (Prunus armeniaca) exhibit in vitro antibacterial activity against both gram-negative (e.g., E. coli and K. pneumoniae) and gram-positive (e.g., S. aureus) bacteria (50). Figs (Ficus spp.) have in vitro antimicrobial activity against E. faecalis, E. coli, and Proteus mirabilis and are also active against C. albicans (51).

Statistical methods.

We used the statistical software IBM SPSS statistics 20 for all analyses. For maternal characteristics according to intake we used Pearson’s chi-square asymptotic 2-sided test of intake frequencies in groups for descriptive analyses (Table 1). Intake data for Tables 2–4 were analyzed by using binary logistic regression, and the following covariates were evaluated as potential confounding factors: maternal age, fetal sex, parity, prepregnancy BMI based on self-reported weight and height, previous spontaneous PTD, marital status, smoking before pregnancy, smoking during pregnancy, alcohol intake, family income, and educational level. In addition to potential confounders, differences in total energy intake (kJ/d) were assessed between cases and controls. The food types were examined by first dividing participants into nonconsumers and consumers, then into categories of intake in grams per day. For these analyses, intakes of food items with few nonconsumers were grouped into categories of “low,” “medium,” and “high” consumption, whereas intakes of food items with many nonconsumers were grouped into categories of “none,” “low,” and “high” consumers. P values <0.05 were considered significant.

TABLE 1.

Intake of foods with known prebiotic and antimicrobial components during pregnancy, according to maternal characteristics of 18,888 women in the Norwegian Mother and Child Cohort Study (MoBa), 2002–2007¹

		Alliums			Dried fruits
	n (%)	Garlic	Onion, leek, spring onion (raw)	Onion, leek, spring onion (cooked)	Raisins	Apricots	Prunes, figs, dates
		% of consumers2			% of consumers2
Maternal age
20–24 y	2414 (12.8)	64.4	64.1	80.7	38.9	8.7	13.0
25–29 y	7640 (40.4)	74.6	69.0	88.1	51.5	13.4	19.7
30–34 y	8834 (46.8)	79.2	71.3	89.9	58.2	19.0	24.4
P value	—	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Parity3
Nulliparous	8408 (44.5)	78.0	70.4	87.5	50.8	16.0	21.9
Multiparous
1 child	7571 (40.1)	74.7	69.2	88.9	55.2	14.9	20.1
2 children	2488 (13.2)	71.3	68.3	87.3	54.5	15.2	21.3
≥3 children	410 (2.2)	62.0	61.7	83.7	48.5	13.2	20.0
P value3	—	<0.001	0.015	0.054	<0.001	0.164	0.049
Missing	11 (0.1)	90.9	72.7	100.0	63.6	27.3	27.3
Fetal sex
Male	9163 (48.5)	75.0	69.8	87.6	53.0	15.1	21.2
Female	9725 (51.5)	75.9	69.2	88.4	53.0	15.7	20.9
P value3	—	0.142	0.353	0.065	0.943	0.291	0.588
Marital status
Married	9073 (48.0)	77.0	68.3	89.7	57.9	17.1	22.7
Cohabiting	9202 (48.7)	74.2	70.9	86.9	49.0	13.9	19.6
Single	364 (1.9)	69.8	65.7	77.2	37.1	12.9	16.5
P value	—	<0.001	<0.004	<0.001	<0.001	<0.001	<0.001
Missing	249 (1.3)	71.5	66.7	79.9	46.2	12.4	18.5
Prepregnancy BMI
<18.5 kg/m2	630 (3.3)	73.3	66.7	85.9	53.2	16.2	22.5
≥18.5 to <25.0 kg/m2	12,593 (66.7)	77.4	69.8	89.0	55.9	16.8	22.3
≥25.0 to <30.0 kg/m2	3735 (19.8)	73.1	69.6	87.3	47.8	12.9	19.0
≥30.0 to <35.0 kg/m2	1064 (5.6)	66.9	69.0	82.7	43.2	9.7	15.4
≥35.0 kg/m2	370 (2.0)	65.1	67.8	82.7	39.5	9.5	14.9
P value3	—	<0.001	0.430	<0.001	<0.001	<0.001	<0.001
Missing	496 (2.6)	72.8	66.5	84.1	50.8	14.7	20.8
Previous spontaneous preterm delivery
No	18,491 (4.7)	75.6	69.6	88.1	53.1	15.5	21.1
Yes	387 (22.5)	70.5	67.2	84.5	48.6	10.3	17.8
P value3	—	0.022	0.313	0.031	0.077	0.005	0.116
Missing	10 (0.0)	80.0	60.0	80.0	50.0	0.0	10.0
Smoking before pregnancy
None	13,410 (71.0)	75.9	68.9	89.0	57.0	16.9	22.3
Occasional	1880 (10.0)	81.1	74.7	89.3	53.1	15.8	23.5
Daily	3433 (183)	70.8	69.4	83.6	37.9	9.6	14.8
P value3	—	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Missing	165 (0.9)	72.1	64.8	81.8	45.5	14.5	19.4
Smoking during pregnancy
None	17,347 (91.8)	76.1	69.7	88.6	54.7	16.2	21.8
Occasional	485 (2.6)	73.8	68.5	83.3	41.0	10.5	14.6
Daily	1056 (5.6)	65.5	66.5	80.0	30.2	5.4	12.0
P value3	—	<0.001	0.360	<0.001	<0.001	<0.001	<0.001
Missing	—	—	—	—	—	—	—
Educational level
<12 y	1351 (7.2)	65.3	69.4	84.4	35.8	8.8	14.9
12 y	4786 (25.3)	70.5	70.0	87.5	44.9	10.4	17.7
13–16 y	8196 (43.4)	80.2	73.6	93.1	58.5	16.4	22.8
≥17 y	4175 (22.1)	87.7	75.0	95.0	66.9	25.4	28.7
P value3	—	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Missing	380 (2.0)	80.7	78.6	91.1	52.8	11.7	19.8
Alcohol intake during pregnancy
No	16,787 (88.9)	74.1	68.6	87.3	52.0	14.5	20.2
Yes	2101 (11.1)	86.4	76.6	93.0	61.0	22.8	27.7
P value3	—	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Household income4
Both partners: <300	5777 (30.6)	70.2	68.4	86.4	50.1	13.5	21.0
One partner: ≥300	8246 (43.7)	76.3	69.1	88.3	52.9	14.6	20.6
Both partners: ≥300	3808 (20.2)	84.6	73.6	91.9	59.6	20.6	23.1
P value3	—	<0.001	<0.001	<0.001	<0.001	<0.001	<0.014
Missing	1057 (5.6)	64.9	63.3	79.8	46.0	13.6	17.9

¹Total n = 18,888.

²Percentage of consumers in category, compared with nonconsumers, accounting for missing data.

³Pearson’s chi-square asymptotic 2-sided test of intake frequencies in groups.

⁴In Norwegian kroner (NOK) × 1000.

TABLE 2.

Risk of spontaneous PTD according to no versus any food intake¹

			Crude model		Adjusted model2
	Spontaneous PTD (n = 950)	Term controls (n = 17,938)	P value	Crude OR (95% CI)	P value	Adjusted OR (95% CI)
	n
Dried fruit
No	460	7363	<0.001	0.74 (0.65, 0.85)	0.005	0.82 (0.72, 0.94)
Yes	490	10,575
Raisins
No	503	8372	<0.001	0.78 (0.68, 0.89)	0.017	0.85 (0.72, 0.97)
Yes	447	9566
Apricots
No	823	15,154	0.074	0.84 (0.69, 1.02)	0.220	0.88 (0.73, 1.08)
Yes	127	2784
Prunes, figs, dates
No	765	14,147	0.221	0.90 (0.77, 1.06)	0.356	0.92 (0.78, 1.09)
Yes	185	3791
Alliums
No	105	1823	0.377	0.91 (0.74, 1.12)	0.780	0.97 (0.78, 1.20)
Yes	845	16,115
Alliums3
Low	498	8520	0.003	0.82 (0.72, 0.94)	0.005	0.82 (0.72, 0.94)
High	452	9418
Garlic
No	267	4360	0.008	0.82 (0.71, 0.95)	0.011	0.82 (0.71, 0.96)
Yes	683	13,578
Onions, raw
No	288	5470	0.907	1.01 (0.88, 1.16)	0.869	1.01 (0.88, 1.17)
Yes	662	12,468
Onions, cooked
No	128	2135	0.146	0.87 (0.72, 1.05)	0.634	0.95 (0.78, 1.16)
Yes	822	15,803

¹PTD, preterm delivery.

²Adjusted for parity, fetal sex, prepregnancy BMI, previous spontaneous PTD, marital status, smoking before pregnancy, smoking during pregnancy, educational level, alcohol, family income, and maternal age.

³Adjusted lower threshold: low ≤ median, high > median, median = 5.84; performed due to the low number of participants in the no-intake group.

TABLE 3.

Risk of spontaneous PTD according to low or high food intake¹

			Crude model		Adjusted model2
	Spontaneous PTD (n = 950)	Term controls (n = 17,938)	P value	OR (95% CI)	P value	Adjusted OR (95% CI)
	n
Dried fruits			<0.001		0.005
No intake	460	7363		1.00		1.00
Low intake: ≤1.64 g	250	5424	<0.001	0.74 (0.63, 0.86)	0.004	0.79 (0.67, 0.92)
High intake: >1.64 g	240	5151	<0.001	0.75 (0.64, 0.88)	0.014	0.81 (0.69, 0.96)
Apricots			0.150		0.306
No intake	823	15,153		1.00		1.00
Low intake: ≤0.82 g	70	1431	0.412	0.90 (0.70, 1.16)	0.764	0.96 (0.74, 1.24)
High intake: >0.82 g	57	1354	0.069	0.78 (0.59, 1.02)	0.126	0.80 (0.61, 1.06)
Raisins			0.001		0.055
No intake	503	8372		1.00		1.00
Low intake: ≤0.66 g	163	3537	0.004	0.77 (0.64, 0.92)	0.034	0.82 (0.68, 0.99)
High intake: >0.66 g	283	6029	0.001	0.78 (0.68, 0.91)	0.077	0.87 (0.74, 1.02)
Prunes, figs, dates			0.349		0.359
No intake	765	14,145		1.00		1.00
Low intake: ≤0.72 g	86	1650	0.752	0.96 (0.77, 1.21)	0.899	1.02 (0.80, 1.28)
High intake: >0.72 g	99	2143	0.149	0.85 (0.69, 1.06)	0.162	0.86 (0.69, 1.06)
Alliums			0.019		0.014
No intake	49	859		1.00		1.00
Low intake: ≤6.04 g	492	8515	0.934	1.01 (0.75, 1.37)	0.497	1.12 (0.82, 1.52)
High intake: >6.04 g	409	8564	0.253	0.84 (0.62, 1,14)	0.540	0.91 (0.66, 1.24)
Alliums3			0.012		0.018
No intake: ≤5.84 g	498	8520		1.00		1.00
Low intake: >5.84 to ≤10.55 g	227	4768	0.012	0.82 (0.70, 0.96)	0.022	0.83 (0.70, 0.97)
High intake: >10.55 g	225	4650	0.022	0.83 (0.70, 0.97)	0.020	0.82 (0.70, 0.97)
Garlic			0.026		0.029
No intake	267	4360		1.00		1.00
Low intake: ≤0.40 g	386	7525	0.030	0.84 (0.71, 0.98)	0.047	0.85 (0.72, 1.00)
High intake: >0.40 g	297	6053	0.011	0.80 (0.68, 0.95)	0.009	0.79 (0.66, 0.94)
Onions, raw			0.971		0.714
No intake	288	5470		1.00		1.00
Low intake: ≤1.315 g	344	6426	0.839	1.03 (0.87, 1.19)	0.594	1.05 (0.89, 1.23)
High intake: >1.315 g	318	6042	0.997	1.00 (0.85, 1.18)	0.811	0.98 (0.83, 1.16)
Onions, cooked			0.003		0.008
No intake	128	2135		1.00		1.00
Low intake: ≤4.986 g	355	5971	0.937	0.99 (0.81, 1.22)	0.454	1.09 (0.88, 1.34)
High intake: >4.986 g	467	9832	0.023	0.79 (0.65, 0.97)	0.173	0.87 (0.70, 1.07)

¹PTD, preterm delivery.

²Adjusted for parity, fetal sex, prepregnancy BMI, previous spontaneous PTD, marital status, smoking before pregnancy, smoking during pregnancy, educational level, alcohol, family income, and maternal age.

³Adjusted lower threshold: low ≤ median, high > median, median = 5.84; performed due to the low number of participants in the no-intake group.

TABLE 4.

Risk of early or late spontaneous PTD according to no or any food intake¹

	Early spontaneous PTD, weeks 28–31						Late spontaneous PTD, weeks 32–37
			Crude model		Adjusted model2				Crude model		Adjusted model2
Food group	Spontaneous PTD (n = 43)	Term controls (n = 17,691)	P	OR (95% CI)	P	Adjusted OR (95% CI)	Spontaneous PTD (n = 867)	Term controls (n = 17,691)	P	OR (95% CI)	P	Adjusted OR (95% CI)
	% of consumers						% of consumers
Alliums
Alliums, combined	76.7	89.8	0.007	0.37 (0.18, 0.76)	0.011	0.39 (0.19, 0.80)	89.5	89.8	0.764	0.97 (0.77, 1.21)	0.769	1.04 (0.82, 1.30)
Garlic	60.5	75.7	0.023	0.49 (0.27, 0.91)	0.020	0.47 (0.25, 0.89)	72.1	75.7	0.017	0.83 (0.71, 0.97)	0.025	0.83 (0.71, 0.98)
Onions, raw	69.8	69.5	0.971	1.01 (0.53, 1.94)	0.957	1.02 (0.53, 1.97)	69.3	69.5	0.905	0.99 (0.86, 1.15)	0.960	1.00 (0.86, 1.16)
Onions, cooked	76.7	88.1	0.026	0.45 (0.22, 0.91)	0.047	0.48 (0.23, 0.99)	87.0	88.1	0.320	0.90 (0.74, 1.11)	0.968	1.00 (0.81, 1.23)
Dried fruits
Dried fruits, combined	46.5	56.5	0.190	0.67 (0.37, 1.22)	0.274	0.71 (0.38, 1.32)	49.8	56.5	<0.001	0.76 (0.67, 0.88)	0.011	0.83 (0.72, 0.96)
Apricots	16.3	15.5	0.886	1.06 (0.47, 2.39)	0.767	1.13 (0.49, 2.61)	13.3	15.5	0.077	0.83 (0.68, 1.02)	0.225	0.88 (0.72, 1.08)
Raisins	46.5	53.3	0.372	0.76 (0.42, 1.39)	0.573	0.84 (0.45, 1.56)	47.4	53.3	0.001	0.79 (0.69, 0.90)	0.049	0.87 (0.75, 1.00)
Prunes, figs, dates	30.2	21.1	0.149	1.62 (0.84, 3.10)	0.146	1.64 (0.84, 3.21)	18.6	21.1	0.070	0.85 (0.71, 1.01)	0.138	0.87 (0.73, 1.05)

¹The standard subgroup of very early preterm delivery of 22–27 gestational weeks was insignificantly small and thus omitted from the table. PTD, preterm delivery.

²Adjusted for parity, fetal sex, prepregnancy BMI, previous spontaneous PTD, marital status, smoking before pregnancy, smoking during pregnancy, educational level, alcohol, family income, and maternal age.

Results

A total of 23,822 women were eligible for the study, and 18,888 (79.3%) had answered both the general health questionnaire and the MoBa FFQ and had presented valid food reports. Of the 18,888 deliveries, 950 (5.0%) were classified as spontaneous PTDs. The sample was described in more detail in a previous study (41). There were no observed differences in total energy intake (kJ/d) between cases and controls (P > 0.05).

Among the 18,888 women in the study, the consumption of alliums and dried fruits differed according to maternal characteristics (Table 1). Generally, the number of consumers increased with age, education, and household income; was more often higher in nulliparous than in multiparous women; was higher in nonsmokers than in smokers (during pregnancy); and was higher in normal-weight women than in the other BMI groups.

To evaluate the association for all alliums combined, we constructed a nominal food group variable above and below the median due to a low number of nonconsumers of this combined food type. In the crude models, allium (garlic, onion, leek, and spring onion) intake above the median (=5.84 g/d) and dried fruit (raisins, apricots, prunes, figs, and dates) intakes were associated with reduced risk of spontaneous PTD (Table 2). This finding was consistent after adjusting for maternal age, parity, fetal sex, prepregnancy BMI, previous spontaneous PTD, marital status, smoking before pregnancy, smoking in pregnancy, alcohol intake, household income, and educational level, with similar ORs: OR: 0.82 (95% CI: 0.72, 0.94), P = 0.005, and OR: 0.82 (95% CI: 0.72, 0.94), P = 0.005, respectively (Table 2). In the case of alliums, the strongest association was with garlic [OR: 0.82 (95% CI: 0.71, 0.96), P = 0.011], whereas the strongest association among the dried fruits was with raisins [OR: 0.85 (95% CI: 0.72, 0.97), P = 0.017] (Table 2).

When all dried fruit intakes were grouped into categories of “none,” “low,” and “high,” differences in spontaneous PTD risk suggested any intake versus no intake was important in the case of raisins, whereas a high intake might be of interest for association with apricots, prunes, figs, and dates. In the corresponding case for alliums, analysis indicated that there might be a dose-response and stronger association for a higher intake of garlic [i.e., OR: 0.85 (95% CI: 0.72, 1.00), P = 0.047, for low intake compared with OR: 0.79 (95% CI: 0.66, 0.94), P = 0.009, for high intake] (Table 3).

When spontaneous PTD was examined according to early or late onset, the association for intake of alliums was significant for early spontaneous PTD [OR: 0.39 (95% CI: 0.19, 0.80)], and garlic intake was significant for both early [OR: 0.47 (95% CI: 0.25, 0.89)] and late [OR: 0.83 (95% CI: 0.71, 0.98)] PTD (Table 4). An association was observed between dried fruits and PPROM [OR: 0.74 (95% CI: 0.65, 0.95)], with similar strength across all food groups, and with significance for raisins [OR: 0.71 (95% CI: 0.56, 0.92)] (Table 5).

TABLE 5.

Risk of PPROM according to no or any food intake¹

			Crude model		Adjusted model2
Food group	PPROM (n = 270)	Term controls (n = 17,936)	P value	Crude OR (95% CI)	P value	Adjusted OR (95% CI)
	% of consumers
Alliums
Alliums, combined	88.9	89.8	0.610	0.91 (0.62, 1.33)	0.820	0.96 (0.65, 1.41)
Garlic	70.4	75.7	0.044	0.75 (0.58, 0.96)	0.029	0.74 (0.56, 0.97)
Onions, raw	69.3	69.5	0.931	0.99 (0.77, 1.27)	0.993	1.00 (0.77, 1.30)
Onions, cooked	86.3	89.3	0.366	0.86 (0.61, 1.20)	0.633	0.92 (0.64, 1.31)
Dried fruits
Dried fruits, combined	47.8	56.5	0.004	0.71 (0.55, 0.90)	0.020	0.74 (0.65, 0.95)
Apricots	12.2	15.5	0.139	0.76 (0.53, 1.08)	0.170	0.77 (0.53, 1.12)
Raisins	43.3	53.3	0.001	0.66 (0.52, 0.84)	0.008	0.71 (0.56, 0.92)
Prunes, figs, dates	17.0	21.1	0.103	0.75 (0.55, 1.03)	0.104	0.76 (0.55, 1.06)

¹PPROM, preterm prelabor rupture of the membranes.

²Adjusted for parity, fetal sex, prepregnancy BMI, previous spontaneous PTD, marital status, smoking before pregnancy, smoking during pregnancy, educational level, alcohol, family income, and maternal age.

Discussion

In this study, we investigated intake of foods with naturally occurring antimicrobial components and high prebiotic content. Our working hypothesis was that intake of foods with antimicrobial or antiinflammatory properties might modulate and reduce conditions associated with and contributing to spontaneous PTD. The rationale for this was putative antimicrobial activity against the specific microbes and/or a reduction in overall inflammatory state through the promotion of beneficial microbes, which might keep a general inflammatory condition at the subthreshold level and prevent initiation of labor.

We conceptualized this study from a previously identified probiotic association with spontaneous PTD to further explore the relevance of the antimicrobial and/or antiinflammatory aspects (41). We observed a reduced risk of spontaneous PTD related to groups of alliums and dried fruits. In particular, garlic seems to be associated with a reduction in the risk of spontaneous PTD, especially early spontaneous PTD. We observed that intake of dried fruits, especially raisins, was associated with a reduced risk of spontaneous PTD, most notably PPROM. Cooked onions showed an association with early spontaneous PTD, suggesting that elimination of food-borne microbes from soil may be involved.

The most interesting food item in this study was garlic. Garlic’s innumerable biological activities make speculation about the specific type of effect behind the association challenging. However, it does contain the antimicrobial component allicin, which is reported to have a broad range of antimicrobial activity, including against a number of microbes of interest in this study, e.g., E. coli, S. aureus, E. faecalis, and K. pneumoniae, as well as activity against Candida spp. and T. vaginalis (43, 44, 58–60). This broad range of activity against these microorganisms involved in vaginal conditions, either subclinical or atypical vaginal microbiota, concurs with an effect of garlic on incidences of early spontaneous PTD and PPROM, because both of these types of spontaneous PTD are assumed to be often caused by infections.

Among the food groups associated with spontaneous PTD in our model, there are overlaps with regard to which microorganisms they have activity against. Whereas these microorganisms are not necessarily more often implicated in the outcome of spontaneous PTD, the overlap observed does increase the likelihood that they are involved in a subset of spontaneous PTD. It is interesting that at least garlic and some dried fruits have antifungal activity against C. albicans, because this organism has been detected in connection with intrauterine infection. Furthermore, allicin’s antifungal properties against Candida ssp. are powerful (43). It is exciting to note here that both garlic and raisins have activity against the periodontal pathogen Streptoccocus mutans, which has also been identified in BV in connection with PTD (43, 49).

An in-depth investigation of the mechanism underlying prebiotic enhancement of the effect of probiotics and antimicrobial range was beyond the scope of this study. However, some interesting aspects should be mentioned. E. faecalis has previously been found to contain bacteriocins with activity against Lactobacillus spp., such as L. acidophilus, whereas strains of L. rhamnosus were resistant (61), making this bacterium an interesting potential culprit in the etiology of BV. These results, together with the noted antibacterial effect of allicin against E. faecalis (58), suggest that this microbe may be implicated in spontaneous PTD.

It is interesting to note that the required amount of the foods containing the antimicrobial component is within normal dietary intake. The daily intake of garlic needed to create a detectable association is low, and results can be assumed to be transferable to various populations. Whether diets composed of food types with different antimicrobial components are a factor in the observed differences in spontaneous PTD prevalence between white Americans and African Americans in numerous studies (62) remains to be seen. One example is the screening and treatment of common genitourinary tract infections in African Americans, in whom treatment was reported to reduce PTD by as much as 40% (63), suggesting a relatively high degree of microbial involvement in PTD among these women. The less clear association in whites may have to do with the classification and detection of infection (18).

Donders et al. (18) observed that asymptomatic women with a vaginal flora with absence of lactobacilli, presence of M. hominis, or intermediate or partial BV had an increased risk of PTD, whereas those with full BV did not. This finding suggests that the less studied microbial subgroups and the specific composition of the vaginal flora are influential in pregnancy outcome and might also be part of the reason for the suboptimal success of some antibiotics in improving the prognosis of PTD. Is this why antibiotic treatment and intervention studies sometimes are unsuccessful or less successful, and does it indicate that antimicrobial dietary components might be a more successful option? Furthermore, infections in the decidua or chorioamniotic space may often escape detection, because tests frequently turn out to be false negative (34, 64). Although antibiotic administration might have potential benefits for pregnancy outcome, some antibiotics have severe side effects; other options thus merit investigation (65). Other interesting aspects of allicin are its activity against multidrug-resistant E. coli and that both methicillin-sensitive and -resistant strains of S. aureus are equally sensitive to allicin (43). An effect of a diet containing antimicrobial compounds also concurs with a general hypothesis that PTD is partly caused by an infectious or inflammatory state presenting as an increased level of systemic inflammation. Dietary antimicrobial compounds match this hypothesis if they are assumed to function as a “rescue mechanism” by lowering the overall inflammation level.

The major strengths of this study comprise the large sample size, the detailed data gathered, and the prospective design, which limits the problems associated with recall bias and reverse causality. However, although the MoBa FFQ has been thoroughly validated, all dietary assessment methods are prone to imprecisions and uncertainties, which attenuate the results. The reported intakes of the specific food items examined in our study were limited by the degree of detail in the FFQ, and we did not have information about intakes of, e.g., fresh or dried herbs. Furthermore, there may be common factors that simultaneously increase the risk of PTD and influence the intake of alliums and dried fruits during pregnancy. The participation rate in MoBa is 38.7%, and selection bias is a potential concern. MoBa participants include a higher proportion of nonsmokers and are older and better educated than nonparticipants. However, the relationships between exposure and outcomes are not necessarily distorted by exclusion of nonparticipants (66).

Further investigations, applying different study designs, are warranted to ascertain whether the lack of foods containing antimicrobial components is a risk factor for spontaneous PTD. Also, more studies are needed to improve our understanding of health complications during pregnancy and to facilitate effective health promotion strategies entailing nutritional interventions preconception or in very early pregnancy. Hopefully, observations made in this study will be an incentive to investigate a wide range of pregnancy complications and nonpregnancy conditions in relation to antimicrobial dietary components. In the era of multidrug-resistant bacteria that dodge the “magic bullets” of antibiotics, it is especially important to consider alternative, more innovative methods for the prevention and treatment of resource-demanding conditions such as PTD.

In conclusion, we found that some alliums and dried fruits have properties of potential influence on the risk of spontaneous PTD. Of particular interest are our findings that garlic is associated with overall reduced risk of spontaneous PTD, whereas raisins mainly are associated with lowered risk of PPROM.