The diversity and composition of the neonatal gut microbiota is garnering interest as a target for the prevention of noncommunicable diseases. The disappearing microbiome hypothesis contends that reduced bacterial diversity over generations results in increased allergic and metabolic disease risk in children [1]. In 1-month old infants, low microbial diversity is associated with an increased risk of later atopic eczema [2], allergic sensitization, allergic rhinitis, peripheral blood eosinophilia [3], and asthma [4]. In addition, differential microbial composition is associated with an increased risk of noncommunicable diseases, including allergic sensitization [3], eczema [2], asthma risk [5], neurodevelopmental outcomes [6], and later adiposity in infants [7]. Hence, novel methods of altering the neonatal gut microbiome are of interest. The influence of poor maternal diet (high fat or low fiber) on offspring gut microbiota has been studied in animals, demonstrating that poor maternal diets disturb offspring gut microbiota [8,9]. To our knowledge, there are no human randomized controlled trials (RCTs) with the primary aim of testing whether the maternal diet can modify the diversity and composition of the infant gut microbiota. Dietary supplementation trials of perinatal prebiotic or probiotic supplements provide a premise for testing this aim, with some studies indicating that these supplements modify the composition of gut microbiota in mothers [10] and infants [11,12]. Importantly, though, a supplementation approach fails to address the quality of the underlying diet. In humans, the prenatal diet has been associated with the composition of the infant gut microbiota [13,14], but it is still unclear whether this relationship is modifiable. Human studies are needed to determine whether infant gut microbiota can be modified through perinatal dietary change.

In murine [8] and primate [9] models, poor-quality prenatal diets disturb vertical transmission of microbiota (from mother to offspring). For example, a prenatal diet devoid of dietary fiber reduced microbial diversity and the abundance of fiber-degrading taxa in mothers and offspring [8]. Low diversity compounded over 4 generations and could not be corrected via a high-fiber diet. Similarly, compared with a low-fat prenatal diet (13% of energy from soya bean oil), a high-fat prenatal diet altered the microbiota of vaginally born primates [9]. This alteration was persistent at 1 year and could not be corrected by weaning offspring onto a low-fat diet [9]. In humans, compared with a low-fat prenatal diet (24% of daily energy from fat), a high-fat (43%) diet during pregnancy was associated with an altered infant gut microbiome, including a depletion of Bacteroides persisting to 4 to 6 weeks of age [13]. Taken together, these results suggest that poor-quality diets (such as low fiber, high saturated fatty acid, and high sugar content) during pregnancy and lactation disturb vertical transmission. However, a causal relationship between the maternal diet and neonatal microbial acquisition is yet to be established in humans.

Healthy dietary patterns that are high in fiber and low in fat are associated with higher microbial alpha diversity in adults [15]. Population-based metagenomic analysis indicates that the dietary features that are associated with higher alpha diversity (as measured by the Shannon Index) are frequent fruit and vegetable consumption along with polyphenol-containing tea, coffee, and red wine [16]. Conversely, dietary features associated with low alpha diversity are sugar-sweetened soda, whole fat milk, savory snacking, and a high total energy intake. Across the developed countries, the mean daily intake of fiber for pregnant women is 18 (SD 4.4) g, this is below the recommended ranges (21-28 g depending on country) [17]. Similarly, mean saturated fat intakes of 32.2 (SD 9.1) g/day were 8.5% to 16.5% above the recommended ranges (depending on country) [17]. In Australia, pregnant women have poor diet quality; they neither know nor meet the Dietary Guidelines for all 5 food groups [18-20], but they are motivated and would like further nutritional education [18].

The Healthy Parents, Healthy Kids (HPHK) study (Trial registration: Australian New Zealand Clinical Trials Registry, ACTRN12616000936426) is a prospectively registered open-label, parallel group, RCT of an educational perinatal dietary intervention targeting gut microbiota from the third trimester of pregnancy until 4 weeks after birth. The primary aim is to evaluate whether the dietary intervention alters alpha diversity of the infant gut microbiota 4 weeks after birth. Secondary aims are to evaluate the efficacy of the intervention in altering microbiota, inflammatory and stress profiles, epigenetic regulation, and maternal diet and well-being. The feasibility and acceptability of the study intervention will also be evaluated. The HPHK study intervention design couples pedagogical theory and educational design (focused on self-assessment and self-efficacy) with Behavior Change Techniques (BCTs) [21] to support efficacy and dietary adherence. A sound educational design is an important, yet seemingly overlooked consideration; first, it ensures that participants are able to do the target behavior, second, it helps to mitigate against confusing a true null effect with insufficient learning, and third, it safeguards the literature against spurious findings from poorly designed interventions.

The dietary intervention will result in increased microbial diversity (Shannon index) in infants measured 4 weeks after birth, compared with the control group.

A between-group difference in microbial alpha diversity, measured using the Shannon Diversity Index (which accounts for species richness and evenness) at follow-up (4 weeks after birth) in the infant stool samples.

Melbourne-based women were recruited online or within the community (obstetric clinics, doctor’s surgeries, maternal and child health centers, childcare, playgroups, toy libraries, shopping centers, physiotherapy centers, sports centers, and radio). Online recruitment strategies included pregnancy forums, twitter, and paid Facebook advertisements that targeted Melbourne-based women aged between 18 and 40 years who met Facebook pregnancy-related demographic characteristics.

The randomization process used a concealed 1:1 group allocation ratio with randomly permuted block sizes to ensure allocation was unpredictable. External personnel prepared the randomization schedule and applied it to the Research Electronic Data Capture (REDCap) randomization module [28]. The study administrator used REDCap to randomly allocate each participant. After allocation, blinding was no longer possible because the team had to book study visits or a workshop.

As a gesture of appreciation, participants received an Aus $20 grocery store gift voucher at the initial study visit. In recognition of effort, completed participants entered a raffle to win an Apple iPad, this was drawn at the end of the study. When the study results are known, participants will receive a summary of results and an invitation to attend a presentation.

The objectives of the dietary intervention were that participants become educated, motivated, empowered, and equipped with the skills and self-efficacy to make long-term dietary change targeting the gut microbiota. The gut microbiota were targeted as the intervention’s change mechanism, and the target behavior was eating for the gut microbiota. We expected that the intervention would be feasible and accepted because when asked about support preferences, pregnant women wanted nutrition education, preferably in person and individually tailored [18,29].

Treatment as usual was used as an active control condition. Participants continued receiving dietary advice from a health care provider who was managing their pregnancy. The rationale for using treatment as usual was to be able to compare the intervention against standard treatment [41]. In addition, provision of a different treatment as an active control may have introduced factors that could have altered the gut microbiota or bias the results.

Participants from both groups reported on the dietary advice that they received from their health care provider. Both groups also reported on their dietary intake, including prebiotic and probiotic foods, and dietary supplements. All nonwithdrawn participants received the intervention materials in written form upon study closure after sample and data collection closed.

Data were collected at 4 time points: gestation week 26, 31, 36, and 4 weeks after birth (follow-up), as detailed in Table 1. Data collection included demographic, physical health and medications, mental health and social support, diet, lifestyle, and evaluative feedback. A probiotic food and drink questionnaire was administered at all time points for the intervention group, but only at baseline and follow-up for the control group to prevent the control group from becoming aware or prompted to increase intake of probiotic foods.

The study team were trained to collect baseline and follow-up anthropometrics and biological samples. Biological samples were collected as outlined in Table 2. Participants collected a baseline stool sample during gestation week 26, and a follow-up sample from themselves and their infant 4 weeks after birth. Stool samples were stored in the domestic freezer and transported on ice to the study visit scheduled during week 26 of gestation (or before week 29). The study team collected the follow-up samples during a home visit. Samples were transported to long-term storage (−80°C) on dry ice. At the conclusion of the study, stool samples were couriered on dry ice to the Australian Genomic Research Facility (AGRF) for DNA extraction and 16S rRNA sequencing. The V3-V4 hypervariable region of the 16S rRNA gene was amplified using a forward primer, 341F, 5’-CCTAYGGGRBGCASCAG-3’ and reverse primer, 806R, 5'-GGACTACNNGGGTATCTAAT-3'. Polymerase chain reaction amplicons were generated from approximately 100 ng of extracted DNA, and purified amplicons were sequenced using Illumina MiSeq, in accordance with the manufacturer specification and AGRF protocols.

Questionnaires were administered electronically to participants through REDCap [28], which is hosted on secure servers at MCRI. Information is kept confidential through a secure password-protected system and restrictive user-access permissions. Study team access to participant information was strictly limited to the purposes of running the study, such as organizing study visits, support calls, and recording biological sample collection. For analysis, the investigators have access to the final deidentified trial dataset.

Being a community-based health intervention, minimal harms are foreseen. No independent bodies were developed for data monitoring or auditing trial conduct. Any adverse events would be reported to the Human Research Ethics Committees (HRECs) in accordance with the safety reporting policy of the HREC. SD, JMC, and FNJ oversaw the implementation of the study, and the HRECs of The Royal Children’s Hospital and Deakin University review its progress.

Data sharing is not applicable, and no data have been reported.

The study results will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines [51]. Analyses will be performed according to a modified intention-to-treat principle, in which participants with at least 1 valid postbaseline follow-up are included.

This study was approved by the Royal Children’s Hospital Human Ethics Committee on the December 17, 2015 (HREC 35200), and Deakin University Human Ethics Committee on the February 16, 2016 (DUHREC 2016-036). Current protocol: Version 15, November 17, 2017. Protocol modifications will be detailed in subsequent papers.

Participants provided written informed consent and may have optionally consented to (1) be contacted about future related research and (2) to have data and samples used for future ethically approved research.

To our knowledge, there are currently no human trials testing the hypothesis that the diversity and composition of the infant gut microbiota is modifiable through the perinatal diet. Animal studies implicate poor maternal diets (high intakes of fat or low fiber) in the disturbance of gut microbiota in offspring [8,9]. Experimental studies are needed to determine whether this holds in humans. This is particularly important because diet quality during pregnancy appears to be poor, with many women failing to meet recommendations for fiber and energy and exceeding the recommendation for fat intake [17]. Data arising from this study may inform future interventions aiming to target the composition of the gut microbiota in early life. The results of this study may also be used to inform clinical and public health recommendations supporting the gut microbiota in early life.