Probiotics in pregnancy: protocol of a double-blind randomized controlled pilot trial for pregnant women with depression and anxiety (PIP pilot trial)

Recruitment methods in the pilot will mimic those of a full RCT. Recruitment will take place in collaboration with two ultrasound centers (VCN/CVN and Fara), a lactation consultant practice (Fijn Voeden), two mental health centers (Praktijk Moeder and Mental health center Jolande Zewuster) and an academic hospital (Radboud university medical center) in the Netherlands. These different centers and clinics were chosen because they serve ethnically and socioeconomically diverse, low-class and high-class populations.

Potentially eligible women are identified through two mechanisms: (1) local research advertising and (2) identification in the research database. Women visiting echo center Fara (around 200 per month) will receive an e-mail invitation for an online questionnaire after their 20-week ultrasound scan. The e-mail will be sent by a Fara organizational committee member. At the second echo center (VCN/CVN), women who visit the centers for their 20-week echo will receive a flyer and can sign up to receive the online questionnaire by providing their e-mail addresses. Research assistants at the Radboud University will collect these e-mail addresses and send the women an e-mail invitation for the online questionnaire. Flyers with a link to the online questionnaire are placed in the waiting room at the academic hospital, the lactation consultant practice and the mental health centers. In this way, women can directly access the online questionnaire. The questionnaire forms part of another study investigating the psychosocial needs of women during pregnancy, which is approved by the Ethics Committee of the Faculty of Social Sciences ECSW (registration number: ECSW2016–1710-42, registration date 19 December 2017). Women voluntarily give informed consent before filling in the online questionnaire. The online questionnaire consists of the Edinburgh Postnatal Depression Scale (EPDS), State-trait Anxiety Inventory, State (STAI-S), questions on demographic variables (e.g. general health, maternal lifestyle, sociodemographic and economic factors) and open questions on psychosocial needs, barriers and facilitators to mental healthcare use. At the end of the online questionnaire the respondent will be asked whether she would want to be included in a research database. This would entail that the woman’s contact details are kept securely on the protected server of the Radboud University. Women agreeing to this give permission to be contacted as potential participants for research in the future. The data on women not agreeing to be contacted as potential participants for further research are not used for this study.

From the research database, the Coordinating Investigator (CI) will contact women who have, according to the STAI-S and EPDS, at least mild symptoms of depression or anxiety. The CI informs these women about the PIP pilot trial and inquires whether they would like to receive information about the PIP pilot trial. When a woman agrees to receive information, a participant information booklet is sent to her. If a woman is not interested in receiving information, her reasons for this are recorded on a Microsoft Excel spreadsheet chart.

One week after the participant information booklet is sent, the CI will contact the woman by telephone to inquire whether she is interested in participating in the trial. This call is also an opportunity for the woman to ask any remaining questions about the pilot trial. If the woman is interested in joining the study, she will be requested to send the signed informed consent form to the Radboud University. If a woman does not want to participate in the study, the CI will inquire about her reasons for non-participation (see “Qualitative outcomes”) and will record these on a Microsoft Excel spreadsheet chart. However, according to the principles of the Declaration of Helsinki, the CI will also inform the woman that she is not obligated to reveal her reason(s) for non-participation.

After signing the informed consent form, women will be requested to fill in a screening questionnaire online for assessment of their eligibility (see “Participants” for inclusion and exclusion criteria). If women are eligible, the CI will contact them by telephone to administer the Mini International Neuropsychiatric Interview (MINI). A woman who scores as “high risk” on the MINI questionnaire or is otherwise not eligible will be informed about exclusion from the pilot trial. As stated in the informed consent form, in the case of a high risk score the woman’s general practitioner will also be informed within a maximum of 12 h about the outcome of the MINI. The remaining eligible women will be informed about inclusion and a home visit will be scheduled. They will also be randomized into one of the treatment arms.

Recruitment through centers and clinics and by using a research database was deemed appropriate for two reasons. First, it would not put undue pressure on women to consent to participate in the pilot trial. Second, it would prevent unnecessary visits by pregnant women to the Radboud University during the screening phase, as many of the women visiting the University would not fall under the eligibility criteria (i.e. the prevalence rate of elevated symptoms of anxiety and depression in pregnant women is between 10 and 20%).

A total of 2135 women need to be approached for this trial to reach a sample population of 40 participants (see Fig. 2). This number was calculated based on the following. First, from our own experience, we assume a response rate of 25% to the e-mail invitation for the online screening. Second, based on previous studies, we can expect 15% of all women who filled in the online questionnaire to screen positive for at least mild symptoms of depression or anxiety and who are thus preliminarily eligible [22, 23]. Finally, we expect that 50% of all the preliminarily eligible women interested in participating will meet all participation criteria and give informed consent. The 50% rate is based on previous studies examining the effect of dietary and psychological interventions during pregnancy [24, 25]. The expected total duration of the study from the start (enrolment of the first participants) to the last participant finishing is 13 months. No target population per site is set; recruitment will continue at all sites until the target population of 40 women is achieved. Training/information meetings will be held with organizational committee members at all involved sites prior to the start of recruitment. Each site will be monitored by Radboud University staff through regular telephone calls and e-mail correspondence, and in the case of any recruitment problems, strategies for solving them will be discussed.

Prior to the first home visit, the participant fills in the baseline questionnaire (see Fig. 1). During the first home visit, the participants receive the probiotic/placebo product and kits from the CI who is blinded to the intervention. The participant collects the first stool and vaginal samples at home prior to the start of taking the probiotic/placebo product. These samples are kept in the participant’s freezer and handed over to the CI or research assistant at time point 1 (t1) or t4. After stool and vaginal fluid collection, the participant starts with the probiotic/placebo product intake once per day until time of delivery.

Participating women will be randomly allocated to either control group (placebo) or intervention probiotic group (multispecies probiotic Ecologic® Barrier, Winclove Probiotics B.V.) once daily for 8–14 constituent weeks. The probiotic food supplement is a freeze-dried powder in sachets and contains 2,5 × 10⁹ CFU/g. One dose (one sachet) consists of a total of 5,0 × 10⁹ CFU (2 g). The recommended dosage of the probiotic product of one sachet per day is in concordance with probiotic dosages used in previous studies examining Ecologic® Barrier and other multispecies probiotic products. The number of bacteria in a dose ranges from 10⁹ to 10¹⁰ CFU/dose, which is the normal amount of bacteria for studies in healthy volunteers and pregnant women [27‐29]. The probiotic product contains the following bacterial strains: Bifidobacterium bifidum W23, Bifidobacterium lactis W51, Bifidobacterium lactis W52, Lactobacillus acidophilus W37, Lactobacillus brevis W63, Lactobacillus casei W56, Lactobacillus salivarius W24, Lactococcus lactis W19 and Lactococcus lactis W58 (total cell count 2.5 × 10⁹/g). In addition to the bacteria, the probiotic product contains the carrier of the product: maize starch, maltodextrins, inulin and fructo-oligosaccharides (FOS).

Preclinical studies suggest that prebiotics, such as FOS and inulin, can exert antidepressant and anxiolytic effects [30, 31]. Therefore, in order to properly assess potential effects of bacteria on maternal mood, the placebo intervention includes the same carrier of maize starch, maltodextrins, inulin and FOS, but no bacteria. The placebo is indistinguishable from the probiotic sachets in color, taste and smell (Winclove Probiotics B.V., The Netherlands).

Participants in the probiotic and placebo intervention groups are provided with 108 sachets (one for each day of the intervention, total of 3,5 months). The contents of the sachets need to be dissolved in 100 mL water, milk or yoghurt. The solution containing the probiotic/placebo is to be ingested orally, preferably on an empty stomach. The participant will be advised to take the probiotic/placebo 2–3 h after antibiotic ingestion if taking antibiotics. The sachets are stored at room temperature in their original package.

The full RCT will not investigate probiotics as a single treatment, but will explore their properties as a food supplement in addition to patients’ current treatments, if applicable. Hence, the use of any kind of co-medication, therapy (e.g. cognitive behavior therapy) or alternative therapy (e.g. yoga, mindfulness, tai chi, etc.) is allowed and will be reported accordingly. Participants are requested to discontinue any other probiotic (such as probiotic yoghurts) two weeks prior to the start of the intake of the probiotic/placebo product and to refrain from taking any other probiotics during the whole trial period.

The primary outcome will be assessed as follows: feasibility will be assessed in terms of (A) recruitment (quantitative and qualitative), (B) retention (quantitative) and acceptability will be assessed in terms of (C) compliance (quantitative) and (D) participants’ impressions and experiences during the pilot trial (quantitative).

The primary outcome of a full RCT would include effectiveness of the probiotic, compared to placebo, in reducing anxiety and depressive symptoms during pregnancy. For that reason, key outcome measures of this pilot study will inform whether or not to proceed with a full trial after this study, and if so, the revisions that should be made to the study protocol before conducting a full RCT. These outcomes include (B) retention rate, (C) compliance (i.e. proportion of participants taking ≥ 80% probiotic/placebo and proportion of participants filling in electronic questionnaires at all time points) and (D) participants’ experiences of participation and perspectives on joining this and future probiotic trials. Table 1 shows the outcomes, progression criteria (where applicable), key outcome measures and time points for each outcome of this pilot study.

Data collection for qualitative outcomes will consist of face-to-face and telephone interviews conducted by the CI. At the end of the home visit at baseline (t0), the CI (a physician experienced in research and patient consultations) will ask participants face-to-face the timing and reasons for deciding to join the study (“When did you make the decision to join the PIP pilot trial?” “Why did you choose to join the PIP pilot trial?”). Responses will be noted and directly checked by participants for accuracy. To determine the reasons for non-participation , the CI will ask women who refuse to participate by telephone their reason(s) for non-participation. Responses will be directly transcribed by the CI and not audiotape-recorded. Microsoft Excel spreadsheet charts will be created with (refusing and participating) women (in the rows) and recruitment outcomes (in the columns).

In addition to the primary objective, this pilot trial includes six secondary objectives to ensure that certain data can be collected and to exploratorily assess whether maternal probiotic intake, compared to placebo, potentially has an effect on (1) depressive and/or anxiety symptoms, (2) maternal stress (i.e. reported and hair cortisol), (3) maternal vaginal and intestinal microbiota, (4) maternal bonding to offspring, (5) infant microbiota and (6) infant crying. In the absence of a core outcome set to measure potential effectiveness of probiotics on maternal mood, we choose outcome measures that have shown good psychometric properties and have been previously used to assess depressive symptoms, anxiety and stress in RCTs including non-pregnant [6, 28] and pregnant women [36]. Similarly, outcome measures used to assess maternal bonding to offspring and infant crying have been shown reliable in previous studies [36‐38]. Regarding the rationale for the choice of biological outcomes, please see “Discussion”. Given that the main aim of this pilot trial is to assess feasibility and acceptability, and not probiotic effectiveness, no criteria will be set for comparison of treatment success between the two treatment arms (probiotic versus placebo) for any of the secondary outcomes. Figure 1 shows the secondary outcomes measured throughout the study. Table 2 displays the secondary outcome measurement variables, participant level analysis metric and methods of aggregation at specific measurement time points of interest. Participants will fill in all questionnaires electronically, except for the Baby’s Day Diary, in order to improve data accuracy and the timelines for receiving data. For data collection forms (instructions and CRFs) see Additional file 2A-E.

The Mini International Neuropsychiatric Interview (MINI) is used to determine a positive diagnosis according to the main Diagnostic and Statistical Manual (DSM)-III-R/IV Axis I disorders. The Dutch translation [39] of the clinician-rated version of the MINI will be used [40]. The Dutch translation has had good validity in previous studies [41, 42]. Suicide risk will be measured using six screening questions from this tool. High suicidal risk is defined as a positive response on questions C4 or C5 or (C3 + C6) [40].

A strand of around 50 hairs per person will be collected from the posterior vertex [64] at 4 weeks for postpartum cortisol analysis, in concordance with previously validated methods of collection and analysis of hair. The CI will cut the hair with scissors, as close as possible to the participant’s scalp. The hair closest to the scalp represents the most recent part of pregnancy. The sampling site is not visible afterwards, as the hair growing above covers it. After collection, 1 cm of the strand of hair that was closest to the scalp will be removed. The strand will then be divided into two segments for analysis, representing (1) the period during the last 8 weeks of probiotic or placebo intervention (2 cm) and (2) 2 months before the intervention (2 cm). Hair grows approximately 1 cm a month and therefore 2 cm will cover 8 weeks of probiotic or placebo intake. Hair cortisol will be analyzed using the procedure described by Manenschijn, Koper, Lamberts & van Rossum in 2011 [64]. A correction factor will be applied to account for the potential influence of differences in hair weight. Participants will record on a data form any intake of oral corticosteroids during pregnancy and other potential confounding factors. The hair and accompanying form will be securely stored together with the hair sample at the Sponsor’s site.

Two vaginal samples will be collected to study changes in vaginal microbiota between baseline and 8 weeks after the start of the intervention. Participants will collect the samples by rotating swabs 3–4 cm deep into the vagina. Samples will be collected by the validated self-collection method using dry swabs [65]. The samples will be immediately stored at − 20 °C, collected by investigators at t1 or t4 and then stored at − 80 °C at the Radboud University. Microbial composition will be determined using polymerase chain reaction (PCR) sequencing methods at affiliated universities experienced in this type of analysis; the location for these analyses has yet to be determined.

Participants will collect fecal samples twice using collection methods applied in our previous studies [36, 38], in order to study changes in maternal intestinal microbiota between baseline and 8 weeks after the start of the intervention. For storage procedures, see “Maternal vaginal microbiota”.

Parents will collect two infant fecal samples at home on weeks 1 and 4 using similar collection methods to those applied in our previous studies [36, 38], to measure the relative abundance of infant intestinal bacteria. Samples will be immediately stored by the participants at − 20 °C, collected by investigators at t3 and taken back to the University for storage at − 80 °C. The microbial composition of the stool samples will be analyzed using PCR sequencing methods.

Sleep quality is measured using the Pittsburgh Sleep Quality Index (PSQI), a self-rated questionnaire that assesses sleep quality and disturbances during the previous months [68]. The PSQI has demonstrated good psychometric properties, including good internal consistency and validity in pregnant women [69]. We will use the Dutch translation of the PSQI, which has been commonly used in Dutch populations [36, 70]. The sum of scores ranges between 0 and 21 and scores ≤ 5 and > 5 indicates “poor sleepers” and “good sleepers”, respectively [68].

Other systematic reviews have noted the stressful events that can predict prenatal anxiety and depression [71, 72]. Based on these studies, we constructed a new stressful events questionnaire that assesses whether or not participants experienced one or more stressful events during the past 12 months. The questionnaire includes 8 items (i.e. death of a close relative/family member, financial problems, divorce etc.) on a dichotomous scale (yes/no). More reported stressful events indicates higher incidence of stressful events.

The primary outcomes, (A) recruitment, (B) retention rate, (C) compliance and (D) (1–5) participants’ impressions and experiences, will be descriptively analyzed and visualized in tables or graphs: number and percentages will be calculated for categorical variables and mean (or median) and SD (or IQR) will be calculated for continuous variables. For outcome (D) (1–3), participants’ impressions and experiences, the main themes and categories of participants’ responses will be identified (see “Primary qualitative outcomes” for description of conventional content analysis and the framework approach) and quantified [73, 74]. The prevalence of coded themes will be calculated as a percentage of the total number of identified coded themes. Demographic data on subjects (general health, socio-economic factors, etc.) will be tabulated.

Qualitative content analysis following principles of the framework approach will be undertaken for outcome (A) recruitment [73, 74]. To develop an analytical framework, two researchers will read through all answers from Microsoft Excel spreadsheet charts and will apply inductive “open coding” (i.e. label what they consider to be relevant from as many perspectives as possible). After coding the first transcripts, the labels applied will be compared between the two researchers to agree on a set of codes and these codes will be grouped to form themes and categories. The analytical framework, including the codes and themes, will consequently be used to index all transcripts. Data will be summarized by category and put into a Microsoft Excel spreadsheet chart to generate a matrix. The chart consists of participants (rows) and emergent qualitative themes (columns) and will be used to identify similarities and differences between participants’ expressed thoughts. The qualitative findings will be described in detail, supported by quotes from participants.

Although the aim of this pilot trial is to assess the feasibility and acceptability of running a definitive trial, the secondary psychological, behavioral and biological outcomes will be analyzed to mimic the definitive trial. All analyses of the secondary outcomes will hence be viewed as preliminary and thus interpreted with caution.

Outcomes will be analyzed per-protocol and by intention to treat (ITT). For the per-protocol analysis, participants who are found ineligible after randomization or deviated from the probiotics/placebo product intake (i.e. less than 80% intake) will be excluded from the main analysis. For the ITT analysis, all randomized participants, regardless of protocol adherence or dropouts (i.e. participants who withdraw consent for continued follow up), will be included in the main analysis. Participants will be analyzed according to the group to which they were originally allocated. In case of missing data, missing values will be imputed. Student’s t test will be used to analyze normally distributed data and the nonparametric Mann–Whitney U test will be used to analyze non-normally distributed data. For comparisons of more than two groups (e.g. differences in the relative abundance of bacteria at the level of phylum up to the level of order in microbial samples), analysis of variance with Tukey’s post-hoc test of the intervention and placebo group at the primary endpoint will be compared using a one-way analysis of covariance (ANCOVA) model, where the covariate will be the baseline score. In addition, the effect size will be computed to estimate the required sample size for the future large-scale trial. The significance level for statistical analysis is set at p < 0.05.