This 3-armed randomized controlled trial study was performed on 100 primiparous mothers whose premature infants were born during the 11 months from January 2020 to December 2020 in the NICU of academic center (Rouhani Teaching Hospital, Babol, Iran). NICU of Rouhani Hospital has level III intensive care affiliated with Babol University of Medical Sciences and high-risk pregnancy referral center.

The mothers who referred to the hospital were mostly local residents of the area and all of them were Persian-speaking individuals. Primiparous mothers based on the time of admission of their premature infants in the NICU of Ayatollah Rouhani Hospital were selected. Interventions for mothers were performed on the third day of neonatal hospitalization.

From the beginning of the study to the required sample size, Eligible individuals with eligibility criteria were included in the study. The mothers of the neonates signed an informed consent for inclusion in the study prior to the start of and before randomization.

The mother’s willingness to breastfeed, possibility of mother’s presence for breastfeeding at least once a day in the hospital, cesarean section mothers, mothers of normal hearing were included. Mothers due to infants with G.A < 34 weeks, mother due to neonates with significant congenital anomalies, mother refuses to participate, mothers with history of drug abuse or oral galactagogues and mother-child dyads with known contraindications for breastfeeding were excluded.

The reason for exclude infants (< 34) sucking are not Coordinated with swallowing and breathing [31], infants are fed with gavage or syringe, while late premature infants (34-37) are feed via the breast. Therefore, if we consider infants < 34 weeks, their feeding methods will be different.

For all cesarean section mothers, spinal anesthesia was used with markin injection (bupivacaine). For each mother, 2.2 cc of spinal anesthesia was injected. The effect of this drug is about 2 hours. 30 units of oxytocin per liter of Ringer serum were injected during labor in the operating room (15-20 minutes). Also, after transferring the mother to the ward, 30 units of oxytocin in one liter of Ringer serum were injected into the mother for 8 hours. Opioids were not used for cesarean section mothers.

Primiparous mothers were randomly allocated to one of three study groups: control (A) with 33 participants, playing lullaby for mother (B) with 33 participants and playing lullaby for a mother while holding a photo of her own baby (C) with 34 participants, through a using a computer random number generator and with a strategy of permuted blocks with size of six. To maintain randomization during the study, the allocation concealment method was performed. For allocation concealment so that the next person to be assigned to which group would not be known. They were only consecutively revealed in the final expulsion phase by the study monitor assigned to the center. The first author was approached the participants.

Blinding for nurses who implemented the intervention was not possible given the nature of the intervention. All outcomes in this study were measurement by laboratory instruments. Due to the characteristics of the study intervention, blinding was applied only to a group of researchers who were laboratory evaluators and data analysts. Incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding.

This study followed the CONSORT guidelines for reporting randomized controlled trials (Fig.1).

Participant flow diagram according to consolidated standards of reporting trials

The sample size was determined based on a study by Keith et al. [16] on the effect of lullaby interventions on increasing breast milk volume and fat. The mean difference (standard deviation) in the two groups was considered based on the same study of 46.2 (6.5) and 51.8 (12.2). The sample size was calculated based on the effect size. This effect size was obtained based on the mean difference before and after the intervention according to the same study.

The minimum sample size in the three groups (A, B, C) was calculated using G Power software version 3 [32], with a test power of 80% and an error of 0.05 and effect size of 0.53. A total of 90 people were required for the three groups (30 individuals for each group), and with 10% of the loss, 100 people were determined as the final sample size.

$$n=\frac_+_\right)}^2^2}_1-_2\right)}^2}$$

$$_}=1.96\kern0.5em 0.84=_\kern0.5em 42.3=_1^2$$

$$148.8=_2^2& 46.2=_1\end}$$

$$51.8=_2& \mathrm=\sqrt_1^2+_2^2}}\end}$$

Correlation between groups = 0.5.

The mothers in the control group first washed their hands and sat on a comfortable chair in the breastfeeding room after a routine breastfeeding training class conducted by a nursing expert working in the NICU. The operation of the electric milking pump in the ward was explained to the mothers of all three groups. Because the quality of breast milk changes during the day and the amount of milk fat in the morning is less, Keith et al. chose to do this process around noon because the stability of milk fat is higher at this time [16]. We also chose this time in our study, and besides, it was also in line with the time of mothers’ training. For all mothers in the control group, after the end of the routine breastfeeding training class, the milk was collected by an electric milking pump and measured in a graduated container based on milliliters under the supervision of researcher.

Mothers of groups B and C, in addition to participating in the routine breastfeeding training class, listened to traditional Iranian lullabies with a female voice through a headphone for 30 minutes [33] with an intensity of sixty-five to seventy-five decibels [34]. Mothers listened to lullabies for 15 minutes. At the end of the first 15 minutes, they started with electric milking pump and in the second 15 minutes, milking continued while the mother was listening to lullabies [28] in the presence of the researcher. In group C, in addition to the routine breastfeeding training class and playing lullabies, mothers look at the photo or their baby. The photo of the baby was taken by the researcher with a 16-megapixel camera at a distance of 30 cm without flashing, while the baby was sleeping. It was taken with the permission of the baby’s parents and doctor. The photos of the babies were pre-printed and given to the mothers of the babies in group C only during the intervention. The timeframe was 0-6 days (one session every day). None of the mothers in the three groups were trained separately.

The primary outcome was the effect of lullaby on the volume of breast milk in mothers of premature infants. For this purpose, Breast milk in all three groups was collected on the first and sixth day of the study around noon (11:30-12:00 AM) using a Mamivac electric pumping machine made in Germany and the volume of breast milk was measured by Medela graduated glass container (capacity 100 cc) made in Switzerland based on milliliters in all three groups. Milk volume was measured by the sixth researcher and recorded in a questionnaire.

The secondary outcome was to determine the effect of the lullaby on cholesterol, triglyceride, total protein and albumin concentration of breast milk in mothers of preterm infants. For this purpose, the researcher collected 2 ml of breast milk from all three groups on the first and the sixth day of the study and transferred to the Biochemistry Laboratory of the Babol University of Medical Sciences to compare the biochemical indicators of breast milk in terms of fat (cholesterol and triglyceride), albumin and total protein using a Spectrophotometer with accurate measurement of 1 mg/DL performing by a fixed laboratory expert according to the instructions on the standard test kit of Pars Azmoon (made in Iran) and based on mg/DL (Table 1).

Infant demographic information (sex, age, birth weight) was collected from the hospital record, maternal characteristics (age, weight, and economic status) by self-reporting.

The statistics advisor performed the data analysis blindly using SPSS Version 20 and a p-value< 0.05 was considered significant. Descriptive information was shown as frequency, percentage, mean and standard deviation.

To evaluate the related statistical tests, first, the normality of the data was evaluated using the Kolmogorov-Smirnov test and parametric tests were used due to its normality. To evaluate the relationship of two grouped variable, the chi-square test and paired t-test were used to examine the mean difference between the groups before and after the intervention. ANOVA was also used to compare the mean differences of quantitative variables between the three groups. Finally, the ANCOVA model was used to evaluate the effects of intervention for breast milk compounds with adjusted for baseline values [35].