A randomised controlled trial evaluating the effect of a brief motivational intervention to promote breastfeeding in postpartum depression

Postpartum depression (PPD) is the most frequent psychiatric complication during the postnatal period. According to existing evidence, an association exists between the development of PPD and the maintenance of breastfeeding. A brief motivational intervention (bMI), based on the motivational interview, seems effective in promoting breastfeeding. The objective of this study was to analyse the impact of a bMI aiming to promote breastfeeding on the development of PPD and explore the mediating/moderating roles of breastfeeding and breastfeeding self-efficacy in the effect of the intervention on developing PPD. Eighty-eight women who gave birth by vaginal delivery and started breastfeeding during the immediate postpartum period were randomly assigned to the intervention group (bMI) or control group (breastfeeding education). Randomisation by minimisation was carried out. The breastfeeding duration was longer in the intervention group (11.06 (± 2.94) weeks vs 9.02 (± 4.44), p = 0.013). The bMI was associated with a lower score on the Edinburgh Postnatal Depression Scale, with a regression β coefficient of − 2.12 (95% CI − 3.82; − 0.41). A part of this effect was mediated by the effect of the intervention on the duration of breastfeeding (mediation/moderation index β = − 0.57 (95% CI − 1.30; − 0.04)). These findings suggest that a bMI aiming to promote breastfeeding has a positive impact preventing PPD mainly due to its effectiveness in increasing the duration of breastfeeding.

Among interpersonal therapies, the brief motivational intervention (bMI) is based on the motivational interview, which could be defined as a collaborative communication style designed to increase personal motivation to achieve one's goals while exploring the individual's ambivalences to change within an atmosphere of acceptance 46 . This type of intervention has already been shown to be effective in increasing the duration of BF and the level of BSE [47][48][49][50][51][52][53] . Nevertheless, to the best of our knowledge, only two studies 54,55 analysed the use of this type of therapy to improve help-seeking when women experience negative feelings associated with PPD.
The evidence suggests that a complex relationship exists among the duration of BF, the presence of PPD 11,21,25,27 , and the level of BSE 29,31 . Therefore, analysing whether an intervention that acts on BF and BSE also has an impact on PPD and determining the causal relationships and the magnitude of the effect among these factors are relevant and pertinent.

The study
Aims. This study aimed to analyse the impact of a bMI aiming to promote BF on the development of PPD and explore the mediating/moderating role of BF and BSE in the effect of the intervention on the development of PPD.
For the aims described above, we hypothesise the following: • Women who receive a bMI would have fewer depressive symptoms in the third month postpartum.
• The effect of the bMI on the BF rates acts as a mediator of the relationship between the bMI and the presence of depressive symptoms postpartum. • The increase in maternal BSE targeted by the bMI acts as a moderator of the relationship between the bMI and the presence of depressive symptoms postpartum.

Design.
A randomised, controlled, multicentre, parallel-group clinical trial was carried out.

Participants.
The study was carried out in two public hospitals in southwestern Spain between February 2018 and January 2019. The participants in this study were women who gave birth via vaginal delivery to healthy newborns and who started BF in the hour after delivery regardless of the number of pregnancies or previous deliveries. The exclusion criteria were as follows: (a) mother-newborn separation after delivery due to admission to the neonatal care unit; (b) psychiatric disorder or cognitive damage previously diagnosed in the woman (including the diagnosis of depression before or during pregnancy); (c) impossibility of follow-up with the newborn-mother dyad; (d) women without adequate prenatal care that included prepartum depression screening; and (e) communication barriers.
Sample size calculation. The sample size was established based on the effects obtained in similar studies on BF and EBF [48][49][50] . With a common alpha risk of 0.05, a beta risk of 0.20, and an estimated loss to follow-up of 10%, we calculated that we needed to include 44 participants in each group.
Procedures. During the immediate postpartum period (first 2 h after delivery), participation in the study was requested from all women who met the inclusion criteria and did not present any of the exclusion criteria based on the clinical history recorded and completed at admission to the hospital until the sample size was reached.
As described in the published study protocol 56 , after recruitment and signing of informed consent, an initial interview was carried out to collect the baseline variables of the participating women. Subsequently, through a randomisation process by minimisation 57 , the women were blindly assigned to one of the two study groups. The midwife responsible for the recruitment and collection of the baseline data was unaware of the randomisation sequence until the investigator in charge of the randomisation reported the result. This researcher obtained the www.nature.com/scientificreports/ data required for the anonymised randomisation. The collection of the follow-up data was carried out by another researcher blinded to the intervention/control group status of the participants. After the randomisation process, the women assigned to the intervention group received a single bMI as follows: a semi-structured interview based on open questions, reflections, and summaries according to the principles of motivational interviewing 46 . The content of the bMI included the exploration of motivations, ambivalences, barriers, and facilitators of the development of BF and the woman's BSE ( Table 1).
The women in the control group received a single educational session that addressed the correct guidelines for achieving successful BF. The session consisted of providing information regarding BF and tips for success (Table 1) using the information leaflet distributed and approved by the Spanish Association of Paediatrics "Congratulations on Your New Motherhood" as a guide 58 .
Both sessions were carried out during the immediate postpartum period in individual sessions by the same midwife with extensive experience in BF who was specifically trained in motivational interviewing by expert psychologists. The duration of the two types of sessions ranged between 20 and 30 min. The women in both groups were given written information regarding BF through the brochure "Congratulations on Your New Motherhood!" 59 and information regarding BF support groups in their area.
The women in both study groups received booster calls at the 1st postpartum month, lasting no longer than 15 min. The calls in the intervention group followed the principles of the motivational interview, and in the calls to the women in the control group, the women were encouraged to continue BF, and resolutions to possible doubts were offered. At the 3rd month, the women were contacted by phone by another researcher for the collection of the results variables. In both cases, telephone contact was attempted up to 3 times for 1 week before considering the absence of response as a loss of the study.
Outcome measures. The variables collected at baseline included the following: (i) sociodemographic variables: age, stable partner (yes/no), educational level (no education, elementary, high school or higher), employment status, and socioeconomic level (< € 13,000/year; € 13,001-21,700/year; and > € 21,700/year); (ii) clinical variables: body mass index, pregnancy weeks, type of delivery, previous children, and newborn weight; and (iii) factors related to BF success: BF experience and BF training. Age, education level, employment situation, and previous experience with BF were the variables used in the randomisation process by minimisation.
As a main variable, the presence of PPD was assessed through the Edinburgh Postnatal Depression Scale (EPDS), a tool that has been translated and validated in Spanish 60 . The EPDS contains questions regarding how the woman has felt during the last 7 days 61 . Women with scores greater than 10 have a high probability of meeting the diagnostic criteria for depression. The EPDS was collected at the third month postpartum. In this study, a score of 11 or higher on the EPDS was considered a cut-off point for the diagnosis of PPD with a sensitivity and specificity of 81 and 88%, respectively 2 .
In this study, information regarding BF was collected at the time of the PPD assessment. Information on the week duration and maintenance of BF (exclusive or not exclusive) was collected in the third month postpartum.
The preintervention and third-month-postpartum BSE levels were also analysed using the Breastfeeding Self-Efficacy Scale Short-Form (BSES-SF) 62 . This 14-item scale has good internal consistency with a Cronbach's alpha of 0.92. This scale identifies women at the greatest risk of discontinuing BF, assesses BF behaviours and cognitions to individualise confidence-building strategies, and evaluates the effectiveness of various interventions. This scale has been validated in Spanish. Informed consent was obtained from all participants. The confidentiality of the participants was maintained at all times. Figure 1 shows the CONSORT diagram of the study.
The women participating in this study who were diagnosed with PPD by the EPDS were referred to psychiatric services for proper care.
Data collection. The baseline variables were collected from the clinical history and initial interview before the randomisation of the participants to each study group. The data from the third month were collected by telephone calls between May 2018 and January 2019. Figure 1 shows the CONSORT diagram of the study.  To analyse the effect of the intervention on the development of PPD, the scores on the EPDS scale were compared between both study arms. To quantify the strength of the association between the bMI and the development of PPD, a multinomial logistic regression was performed, which allowed us to obtain estimates of the odds ratio (OR) and adjusted OR (aOR) when introducing the maintenance of BF at the time of PPD assessment as an independent variable in the model. All estimates were conducted with their corresponding 95% confidence intervals (95% CIs). All statistical procedures were performed using IBM SPSS Statistics for Windows version 24.0 (IBM, Armonk, New York).
Finally, to determine the role of the increase in BSE and duration of BF in the association between bMI and PPD, mediation and moderation analyses were performed according to the Hayes method 63,64 through the PROCESS macro 64 2016 version in IBM SPSS Statistics for Windows version 24.0 (IBM, Armonk, New York).

Results
Participation was requested from 88 women who met the inclusion criteria, and none of them declined to participate. During the follow-up period, in total, six losses occurred, all of which were due to the impossibility of phone contact (Fig. 1). These women were somewhat younger, with a mean age (± standard deviation) of 27.83 (± 8.04) years vs. 33.18 (± 4.76) years among those who completed the study (p = 0.013); had a lower level of income (income < € 13,001/year: 66.7% vs. 9.2%, p = 0.002); and were mostly first-time mothers (83.3% vs. 37.8%, p = 0.016).
The mean age of the women was 32.82 years (± 5.17). More women in the intervention group had higher education (34.1% vs 18.2%) and previous experience with BF (63.6% vs 56.8%). No differences were found in the other baseline variables. There were no differences in the baseline score on the BSES-SF as follows: 59.14 (± 9.35) in the bMI group and 59.41 (± 8.45) in the control group (p = 0.886) ( Table 2). www.nature.com/scientificreports/ The duration of BF (including both exclusive and nonexclusive) was 11.06 (± 2.94) weeks in the bMI group compared to 9.02 (± 4.44) weeks in the control group (p = 0.013). In the mothers who completed the follow-up, the baseline BSES-SF scores were practically the same in both groups. However, the intervention group had 4.13 points more than the intervention group at the third month postpartum (Table 3). When comparing the intragroup scores, we found that in the intervention group, the self-efficacy score was increased by 6.24 points, while in the control group, there were no significant differences.
The EPDS score in the control group, with a median [interquartile range] of 8 [6][7][8][9][10][11], was higher than that in the group that received the bMI, with a median of 5.5 [1.75-9] (Fig. 2). When considering scores above 10, which indicate a high probability of a PPD diagnosis, the percentage of women with a probable diagnosis of PPD was higher in the control group (30% vs. 11.9%) (Fig. 2). When the magnitude of the association was measured, we found that the women who received the intervention were less likely to score above 10 on the EPDS, with an OR of 0.32 (95% CI 0.1; 0.99) (p = 0.050). This association magnitude hardly varied when adjusting this effect for BF at the time of the evaluation (aOR of 0.33 [95% CI 0.10; 1.08] p = 0.068).
The mediation/moderation analyses showed that the effect of the bMI on the variation in BSE did not mediate the effect of the intervention on the score on the EDPS scale, with a β mediation/moderation index of 0.35 (95% CI − 0.09; 0.82) (Fig. 3). However, this mediation/moderation analysis showed that the effect of the intervention on the BF time acted as a mediator of the effect of bMI on the EPDS score, with a β moderation/mediation index of − 0.57 (95% CI − 1.30; − 0.04) (Fig. 3).
The interaction analyses showed that educational level and previous experience with BF did not interfere with the effect of the bMI on the EPDS score (F = 0.672, p = 0.514; and F = 1.541, p = 0.218, respectively). In addition, analysis of the possible moderating effect of the increase in BSE on the effect of the bMI on PPD showed that it did not act as a moderator as follows: it had a β mediation/moderation index of − 0.25 (95% CI − 0.20; 0.16) (p = 0.786).

Discussion
Our results show that women who received a bMI to promote BF during the immediate postpartum period with telephone reinforcement at 1 month postpartum scored lower on the EPDS at the third postpartum month than those who received an educational session concerning BF at the same time. Our analyses show that the women who received bMI had a lower probability of scoring above 10, which is a score associated with a high probability of the existence of PPD. This probability is slightly modified when considering the maintenance of BF at the time of the PPD assessment. The existing evidence concerning effective interventions for the prevention of PPD has been analysed in several systematic reviews [65][66][67] . In a 2013 systematic review 65 , it was found that postpartum visits by nurses or midwives, telephone peer support, and interpersonal psychotherapy were effective in reducing symptoms associated with depression. Furthermore, a recent meta-analysis 66 found that cognitive behavioural therapy and interpersonal therapy were effective interventions to prevent PPD. The bMI can be classified as an interpersonal therapy 65 , which could explain the effect of the intervention on the EPDS score.
To better understand the results obtained by the bMI on the development of feelings associated with PPD, mediation/moderation analyses were carried out, which showed that a part of the effect of the bMI on the EPDS score was mediated by the effect of the bMI on the duration of BF. The women in the intervention group had a longer duration of BF, which has been associated with a lower probability of developing PPD in some studies 11,21,28 . Although the evidence concerning the association between BF and PPD is unclear, some authors note that women with PPD have a shorter duration of BF [25][26][27] , and other authors note that women who breastfeed are less likely to develop PPD 11,21,28 . The mediating effect of BF duration on the relationship between bMI and EPDS scores found in our study reinforces the hypothesis of these authors, who found a protective effect of BF on PPD 23,68 . Moreover, in our study, the previous presence of depressive symptoms was ruled out. However, a large part of the effect of bMI on PPD is due exclusively to the intervention. A bMI aiming to promote breastfeeding could indirectly act in the development of negative feelings associated with depression, producing a protective effect 34 . Similar interventions based on motivational interviewing and Bandura's theory of self-efficacy have found that there is an increase in the ability to seek help by women and a decrease in the risk of developing PPD during the postpartum period 34,54 .
Additionally, the applied bMI achieved a greater increase in BSE than the control intervention, and a higher level of BSE has been associated with a lower EPDS score [29][30][31][32] . However, the mediation/moderation analyses showed that although the bMI produced a greater increase in BSE than the control intervention, this increase did not mediate the effect that bMI had on developing PPD. These analyses indicate that the bMI also does not moderate the effect of developing PPD. The reason for this discrepancy with previous evidence could be that a higher BSE is also associated with a longer duration of BF [69][70][71][72] , and the latter is truly related to the development of PPD.
Therefore, the bMI applied in this study is effective in preventing the development of PPD partially due to the intervention and partially due to its effect on the duration of BF. To our knowledge, this is the first to analyse the impact of an intervention directly aiming to provide BF support on the development of PPD using mediation/moderation analyses. This approach allowed us to estimate the direct and indirect causal effect paths of the intervention on the EPDS score. The role as mediator of BF that we observed in our results allows us to suppose that other interventions aiming to increase the duration of BF could have the same collateral effect on the psychological well-being of women. However, more research is needed to clarify whether other interventions aiming to promote BF could have the same effect on PPD. Furthermore, although the bMI appears to be effective in promoting BF [47][48][49][50] , additional studies are still needed to analyse this efficacy in different situations. If these results are confirmed, the convenience of including this type of intervention in routine healthcare practice www.nature.com/scientificreports/ could be reinforced. The bMI applied in this study is easily integrated into clinical practice. The application of this intervention by nursing and midwifery through its integration into the usual care of mothers and babies could be feasible and easily achievable since it only requires an investment in the specific training of professionals 73 .
Increasing the duration of BF could have an important impact on public health due to the associated benefits, such as the reduction in infant and female morbidity and mortality 22,23 , to which could be added a reduction in the burden of disease associated with PPD. This approach could make the widespread implementation of programs to promote BF through bMI a cost-benefit practice, as has occurred in other clinical settings 73 . Limitations. The findings of this study are limited by its design since the intervention was applied in women without a history of certain psychiatric disorders (depression, depression and anxiety during pregnancy, or postpartum depression) 7,10 . Such history is one of the most relevant risk factors for the development of PPD; thus, the effectiveness of this intervention could be diminished in women with a psychiatric history. Furthermore, an assessment of baseline EPDS score was not carried out in this study. Although the program of medical checkups to follow up the pregnancy of the national health system contemplates depression symptoms evaluation, the determination of depression through the clinical history may not have been effective in all cases. Additionally, it was not applied in women who gave birth by caesarean section or had complications in pregnancy, which are two other factors associated with PPD 13,14 . Therefore, it could be necessary to investigate whether this bMI aiming to promote BF has an impact on the development of PPD in women with these characteristics. Finally, this intervention aimed to promote BF; thus, given its design, it would not be feasible to use this bMI to help prevent PPD in women who are unwilling or unable to engage in BF. A modification in the design of the bMI would be required to adapt it to preventing the development of PPD specifically. Despite our efforts to achieve a balanced distribution of the variables strongly associated with the outcome using randomisation by minimisation, between-group baseline differences in education level and previous experience with BF were found. Although these differences were not very important, these factors were introduced into the adjusted models to prevent their effect on the results. Furthermore, some characteristics of the women classified as lost to follow-up, such as primiparity or low socioeconomic status, are associated with a greater probability of developing PPD. More studies are needed to evaluate how these factors could affect the possible implementation of intervention programmes.

Conclusions
A bMI for the promotion of BF applied during the immediate postpartum period in women who initiate BF in the first hour of life of the newborn and reinforced by one telephone call per month is associated with a lower score on the EPDS at the third postpartum month.
Part of the effect of this bMI on the EPDS score is due to the intervention itself. However, a significant part of the effect is mediated by the effect of the bMI on the duration of BF, which could support evidence associating a longer duration of BF with a lower probability of PPD development 11,21,28 .
Given the importance of both improving the duration of BF and preventing PPD, more research is needed to support the findings found in this study. In addition, future studies should analyse the efficacy of the bMI in women with greater susceptibility to the development of PPD, such as women with a history of psychiatric disorders, those who deliver by caesarean section, and those with complications during pregnancy.