Study design and participants
The present study used a quasi-experimental design with a control group. Data were collected from February 2016 to September 2016 in a single maternity hospital located in an urban area in Kanagawa, Japan. This is the same hospital as in our preliminary study [
15].
The eligibility criteria for participation were as follows: (1) between 20 and 39 years of age, (2) having singleton and cephalic position and planning spontaneous delivery as low-risk pregnancy, (3) between 38 and 39 weeks of gestation, and (4) Asian and can read and write Japanese. Women were excluded from the study if they have (1) ongoing medications related to their gestation, (2) medical or pregnancy complications, (3) mental illness, (4) a medical history of assisted reproductive technology treatment, (5) a BMI above 25 before pregnancy, (6) planned on having an induced labor, or (7) already conducted breast stimulation for more than 10 min per day.
When eligible women at 34 weeks of gestation visited the hospital for a health check-up, they were recruited and they provided written informed consent. The participants were nonrandomly assigned into 2 groups. The first half was assigned to the control group and the last half to the intervention group. After participating in the study, the participants received monetary compensation by bank transfer (¥2000 or about $20 per day) based on the number of days participating in the intervention.
The warranted sample size was 15 subjects for each group in reference to previous reports on plasma oxytocin level in relation to short breast stimulation in pregnant women requiring 10–20 people [
10‐
14]. Considering a dropout rate of 30% from the lack of saliva for assay using the improved saliva collection protocol from our previous study [
15], the sample size needed was 22 women for each group.
All the participants and healthcare providers at the study settings and a biochemist who assessed the laboratory outcomes were masked regarding the participant allocations.
Procedures
The intervention period was 3 consecutive days. Women visited a quiet room at the hospital after eating lunch on the first and third intervention days. The participants sat in a semi-Fowler’s position in a chair and were free to change their posture.
On the first day in the intervention group after the basal saliva collection, the women received an explanation of the breast stimulation method. The researcher or research assistant demonstrated the procedure for about 5 min to standardize the breast stimulation procedure as follows: (1) rhythm, confirmation of 69 beats per minute using a lighting sign of metronome in mute; (2) overall posture, explanation regarding the proper positioning of the arm or fingers for stimulation; (3) finger position, indication of the proper position on a breast model; (4) pressure, indication using a pressure-measuring instrument (Perineometer, OWOMED, Gyeonggido, South Korea) (the recommended stimulation pressure was < 10 mmHg, which is gentle and does not damage the nipple); (5) final demonstration, stimulation of the participant’s breast for about 5 counts according to the stimulation rhythm.
Each breast was stimulated for 15 min on each side and alternating for a total of 1 h per day, as simultaneous stimulation might cause uterine tachysystole [
6,
7]. The nipple was stimulated directly by the participant using either the left or right hand. Pure lanolin nipple cream was used for a more comfortable stimulation [
5,
9]. The cream was given to the participants who could apply it liberally. As an example of the breast stimulation procedure using the right hand, stimulation was performed by pinching only the nipple using the thumb and forefinger. The thumb position was in the “12 or 9 o’clock” direction of the nipple, with the forefinger placed opposite of the thumb.
On the second day, breast stimulation was performed at home before noon so that the participants can have access to treatment within hospital hours in case of emergency. All the necessary materials for breast stimulation were lent to the women. To confirm compliance of the participants and their safety, the women sent emails to researcher 10 min before the breast stimulation and within 10 min after the intervention indicating whether they had vaginal bleeding, premature rapture of membrane, or less fetal movement.
On the third day, the same procedure on the first day was performed at the hospital.
In the control group, on the first and third days after the first saliva collection, the women continued to watch a silent train movie. On the second day, they resumed their usual life activities. On the third day, they repeated the stimulation procedure performed on the first day.
Outcome measurements
The primary outcome was a higher oxytocin level at 30 min on the third day in the intervention group than in the control group. All saliva samples on the first and third days were collected from 12:30 to 16:00 to control for diurnal effects [
16,
17]. Saliva samples (
n = 4) were collected before the intervention and at 30, 60, and 75 min after the start of the breast stimulation on the first and third days. This means that the samples were collected at the 30, 60, and 75 min time frame from the onset of breast stimulation. The total number of samples for each participant was 8. The participants were given the following instructions to avoid interference with saliva oxytocin assay [
18]: (1) exercise or alcohol is prohibited on the day before the intervention; (2) caffeine intake is not allowed on the day of intervention; (3) smoking, brushing, eating, and drinking (except drinking water) are prohibited 1 h before saliva collection; (4) lipstick is not allowed.
Saliva collection was started after gargling [
18]. Basal saliva oxytocin level was determined after 10 min of rest time while watching the movie “The World by Train” without the narration (silent movie). Both the intervention group and the control group watched the same movie which is familiar to the Japanese population. The participants were instructed to wait for 3 min for the saliva to accumulate naturally in their mouth. Then, saliva samples were collected in pre-chilled 2.0 mL polypropylene tubes (Eppendorf, NY, USA) that were subsequently stored on ice using a cut straw (passive drool). This process was repeated 3 times. The target amount of saliva to be collected was 2.0 mL [
15,
19]. If the amount of saliva collected was small, the participants were instructed to self-massage their submandibular gland line.
After saliva collection, the samples were immediately stored in a freezer at − 80 °C. Oxytocin level was assayed using the method of Carter et al. (2007) [
20]. We added the protease inhibitor aprotinin (500 KIU/mL) to inhibit metabolic breakdown of the peptide after thawing the saliva [
21]. Oxytocin level was determined by enzyme-linked immunosorbent assay (ENZO Life Sciences, NY, USA) (
http://static.enzolifesciences.com/fileadmin/files/manual/ADI-900-153A_insert.pdf). The technical report by Salimetrics LCC stated that the acceptable intra-assay and inter-assay coefficients of variation are < 10 and < 15%, respectively. In the present study, the intra-assay and inter-assay coefficients of variation from the dilutions of test samples were < 2.68 and < 5.49%, respectively. Where possible, all samples were run in duplicate (assay rate: 81.7%), but some samples had insufficient volume of saliva and were run as a single assay. There were also samples that contained less saliva or more mucin and did not reach the amount of saliva required for a single assay. These cases were treated as missing data.
For the secondary outcome, a higher rate of spontaneous labor onset was found in the intervention group than in the control group. Data on labor onset were obtained from the medical records. Moreover, data on labor induction, labor augmentation, gestational weeks at delivery, mode of delivery, and labor duration were gathered.
For another secondary outcome, there was a difference in the basal oxytocin level between the oxytocin receptor gene SNPs. Recently, it has been reported that specific oxytocin receptor gene SNP has decreased oxytocin sensitivity. It has also been confirmed that in the GG type of rs53576, the duration of labor at the first stage is prolonged [
22]. In addition, it has been found that the GG type of rs2254298 is related to a significantly low plasma oxytocin level [
23]. Buccal mucosa samples were obtained using a sample collection swab (Epicentre, Wisconsin, USA). All the samples were analyzed for the presence of 2 oxytocin receptor gene polymorphisms (i.e., rs53576 and rs2254298) by genotyping using the TaqMan SNP assay.
Negative events included uterine hyperstimulation, abnormal fetal heart rate, premature rupture of membranes, weak uterine contractions, Apgar score (1 and 5 min), meconium stained liquor, neonatal intensive care unit admission immediately after birth, and stillbirth. Cardiotocography was used to record the FHR and uterine activity 30 min before stimulation and until 30 min after intervention.
Statistical analyses
The baseline characteristics and outcomes were compared between the intervention group and the control group using an independent
t-test, the chi-square test, or Fisher’s exact test. In Table
2, we used a linear mixed model with a first-order autoregressive (AR1) covariance structure to estimate the change in the oxytocin level from baseline. The dependent variable was change in the oxytocin level from baseline. The independent variables were gestational weeks on the first day of intervention, age, education, rs53576 and rs2254298, group, time point, interaction of group and time. All independent variables were treated as fixed effects. Bonferroni correction was performed to avoid multiple comparison problems. SNPs were divided into groups GG and AA/AG, and an independent
t-test was performed for baseline oxytocin level analyses. Statistical analyses were performed using SPSS Statistics version 24.0 J (IBM Corp.).