Effect of Motivational Interviewing on Exclusive Breastfeeding in Working Mothers: A Randomized Controlled Trial

Abstract

Objectives:

Globally the rate of exclusive breastfeeding at 6 months postpartum remains low. We aimed at evaluating the effect of an intervention involving motivational interviewing (MI) and implementation intentions (IMI) on promoting exclusive breastfeeding (EBF) in Chinese working mothers.

Methods:

A randomized controlled trial was conducted, in which 460 new mothers from 4 hospitals were recruited. Mothers in the intervention group (MI group) received MI via telephone at 8 weeks postpartum and completed online IMI questionnaire 1 week later. Mothers in the control group received usual care. The primary outcome was moderation effect of breastfeeding self-efficacy (BSE) on the between-group difference in the likelihood of continuing EBF at 6 months postpartum. Intention-to-treat (ITT) and sensitivity analyses (SA) were conducted.

Results:

Although the results based on ITT was insignificant, SA showed significant moderation effect of BSE (log odds = 1.23, 95% CI: 0.20-2.26).

Conclusions:

The present study revealed that single-session MI telephone-based intervention helped boost maternal motivation for EBF, especially for those with inadequate BSE. Public health interventions that help nurture breastfeeding friendly culture in the workplace and community should be coupled with individualized interventions to dampen early cessation of breastfeeding.

Introduction

As recommended by the World Health Organization (WHO), ¹ mothers should breastfeed their infants exclusively for a minimum of 6 months, and continue doing so for 24 months or beyond. Yet short duration of breastfeeding is common in developed countries. In the US, the initiation rate was 83.2% but a significant drop was observed in any breastfeeding (ABF) and exclusive breastfeeding (EBF) at 6 months postpartum, which were 55.8% and 24.9%, respectively. ² Similar observations were found in Korea, Canada, Japan, and Australia, of which the rates of EBF ranged from 29.1% to 37.5%.^{3 -6} The situation in Hong Kong is far from ideal. Although over 86.8% of the new mothers breastfed their infants at hospital discharge, the rate of ABF at 6 months postpartum was only 43.1%, and that of EBF further dropped to 22.2% during the same period. ⁷

The barriers of breastfeeding continuation are numerous, one of which is short maternal leave. ⁸ New mothers in Hong Kong only enjoy 14 weeks of paid maternity leave. ⁹ Ample evidence has demonstrated the impact of short paid maternity leave on the duration of breastfeeding. ¹⁰ Furthermore, accumulating evidence has shown that full-time employment is detrimental to prolonged breastfeeding.^11,12 About 75% of working mothers in Hong Kong return to work when their infants are between 3 and 6 months old, and it is difficult to combine breastfeeding with employment in workplaces. ⁸ Such barriers may decrease motivation to breastfeed, and the capacity to implement breastfeeding, even if they have the intention to do so.

Premature cessation of breastfeeding not only results from unfavorable maternal factors, ⁸ but also the unavailability of breastfeeding support by family members, ¹³ peers and/or lactation consultants, ¹⁴ and insufficient breastfeeding-friendly facilities/policies in the community.^13,15 Limited studies are found in evaluating the impact of interventions on breastfeeding exclusivity among working mothers. A non-randomized controlled trial has shown that anticipatory counseling and teaching breast milk expression before returning to work enhances working mothers to breastfeed exclusively at 6 months postpartum in the intervention group than the control group. ¹⁶ Providing training regarding practical skills of breast milk expression to working mothers in Turkey significantly improves the rate of EBF at 6 months postpartum. ¹⁷ The drawbacks of these studies are the small sample size and the use of convenient samples. More interventions to promote breastfeeding for working mothers are needed.

It is important to increase motivation for breastfeeding. New mothers who had an authentic belief that breastfeeding was indispensable to the health of their infants were motivated to breastfeed. ¹⁸ A recent review has identified that promoting maternal self-regulated motivation is required to attenuate premature breastfeeding discontinuation. ¹⁹ A positive relationship between maternal breastfeeding motivation and actual practice of breastfeeding was found in the early postpartum period. ²⁰ Motivational interviewing (MI) is one type of counseling strategy that focuses on clients’ self-regulated motivation in promoting health behavior change. Miller and Rollnick defined MI as “a collaborative conversation style for strengthening a person’s motivation and commitment to change.” ²¹ It is a client-centered, non-directive, goal-oriented counseling technique that facilitates clients to explore and resolve ambivalence; the process may lead to behavioral changes. ²¹ MI involves 4 processes: (1) Engaging the client to establish a helpful connection and a strong working relationship; (2) Focusing a specific direction in the conversation about change; (3) Evoking the client’s intrinsic motivation for change by creating discrepancies between the client’s present behavior and his/her important current/future goals; and (4) Planning to develop commitment and formulate a specific plan for changes. ²¹ MI is a potential means to increase motivation for breastfeeding among working mothers. MI nurtures the perception of autonomy and competence; it could enable individuals with ambivalence and weak intentions to endorse a more self-endorsed motivation toward adopting healthy behavior. ²¹ MI has been effective in creating behavioral changes in many health-related behaviors, such as diet, physical activities, binge drinking, and smoking cessation. ²¹ In the context of breastfeeding, some quantitative studies have shown that MI significantly increased exclusive breastfeeding several months postpartum among mothers who were homemakers or worked outside the home.^18,22 However, other studies have also reported non-significant findings on breastfeeding outcomes.^23,24 These studies did not specifically focus on the needs of working mothers. The finding is hence mixed and further research is needed.

The postpartum period is an opportune time in which new mothers are concerned about their health and that of their babies and are receptive to brief interventions for behavioral change,^{25 -27} including brief breastfeeding support. ²² Brief adaptation of MI has been documented. The effect of either a single session of brief telephone MI, stand-alone MI, or MI as a prelude in conjunction with other intervention components in promoting the uptake of health screening was found to be significant.^28,29 Previous studies have demonstrated that laypersons can serve as MI workers after receiving adequate training and supervision.^{30 -33} Ample evidence has shown that brief interventions were cost-effective in facilitating behavioral change.^32,34,35 The present study employed layperson-led brief MI to evaluate the efficacy of MI on breastfeeding continuation.

While motivation is important, it does not always lead to implementation. Implementation intentions (IMI) are distinguished from planning intention, which is the general intention to perform a behavior. ³⁶ IMI is specific about the intention for actually performing a behavior among those who have been motivated to do so. ³⁶ IMI are “If-then” plans, in which the “if” component indicates a critical scenario that may affect the performance of the behavior, while the “then” component illustrates appropriate responses to deal with the specified critical situation described by the “if” component. IMI can be generated by using a volitional help sheet, where a respondent is asked to indicate a situation in which a working mother is tempted not to have milk expression at the workplace (if-component) and to indicate a way to avoid such temptation (then-component), followed by asking her to draw a line linking the if-component and the then-component. ³⁷ IMI enables us to specify particular critical situations in that we want to act on behavior at when and in where, and to form strong associations between the critical situations and responses. IMI facilitates goal achievement in people with strong motivation to attain goals and with high self-efficacy. An earlier study has revealed that intention to EBF is associated with planning regarding action and coping (analog of IMI), which in turn is associated with the actual enactment of EBF. ³⁸ IMI is hence an important linkage between motivation to action; mothers who have started breastfeeding need both motivation and enhancement of IMI to prolong breastfeeding. On the other hand, MI skills would enhance clients’ perceived self-efficacy and motivation. Perceived self-efficacy regarding breastfeeding maintenance is linked to the action and coping planning of EBF. ³⁸

Another important factor for maternal breastfeeding continuation is breastfeeding self-efficacy (BSE). BSE suggests that past breastfeeding experiences, observational learning from competent role models, verbal appraisals from trusted individuals, and maternal emotional relaxation toward breastfeeding are pivotal in enhancing maternal confidence and self-efficacy in mothers who want to continue breastfeeding. ³⁹ A systematic review has demonstrated that mothers with stronger BSE are more likely to continue breastfeeding than those with weaker BSE. ¹⁹ The extant literature shows that BSE moderates the effect of MI on adopting health-related behaviors, such as reducing alcohol drinking, ⁴⁰ and improvement in severity of social anxiety disorder. ⁴¹ A recent midwife-led brief MI conducted in Spain revealed that breastfeeding participants with a higher magnitude in the change of BSE were more prone to continue breastfeeding at the end of the study. ²² MI plus IMI intervention might exert a stronger effect on breastfeeding participants who lack self-efficacy at baseline.

Objective

To our knowledge, this was the first study to assess the effect of combining MI and IMI on breastfeeding maintenance among new mothers with full-time employment who planned to return to work. The objective was to test if the between-group difference in the likelihood of continuing EBF at 6 months postpartum was moderated by baseline BSE scores.

Methods

Study Settings and Participants

This study was conducted in postnatal obstetric wards of 4 public hospitals in Hong Kong between 2021 May and 2022 November. Participants were recruited in these wards by trained research assistants (RAs) who briefed the participants about the study, ensuring comprehension, voluntary participation, anonymity, and confidentiality. Written informed consent was sought before the start of the study. Participants were asked to leave mobile phone numbers, which were used to contact them by our RA at 6 weeks postpartum. Multiple attempts were made to reach the prospective participants who had initially agreed to join the study via phone or WhatsApp. At 6 weeks postpartum, our RA administered baseline questionnaires to participants via telephone and then carried out randomization.

The target group was mothers who had given birth and stayed in postnatal obstetric units. Mothers were eligible to participate if they were: (a) Hong Kong Chinese mothers; (b) at least 18 years old; (c) Cantonese speaking; (d) singleton pregnancy; (e) having full-time employment and currently on maternity leave; (f) continuing any breastfeeding at the time of baseline (6 weeks postpartum). The exclusion criteria were: (a) mothers and their infants were separated after birth because their infants were admitted to the special care nursery or the neonatal intensive care unit; (b) mothers with major obstetric complications or serious medical problems. The CONSORT guideline was used to report this RCT (Supplemental Appendix 3).

Randomization and Masking

Participants were randomly assigned to either the intervention group (MI plus IMI) or the control group in a 1:1 ratio. Block randomization was used, of which the block size was 6. Computer-generated randomization allocation codes that were sealed in opaque envelopes were prepared by a staff with no involvement in the study. One envelope was drawn and opened by a trained telephone interviewer. Due to the nature of the present study, it was not possible to mask the interventional activities.

Interventions for the Intervention Group

Participants in the intervention group received a 20-30 min MI through telephone at 8 weeks postpartum by trained MI interviewers. At 9 weeks postpartum, participants who confirmed their intention to continue breastfeeding after returning to work received a link to an online tutorial promoting IMI. The tutorial was a volitional help sheet that consisted of “if-then” plans on how to combine breastfeeding and work. Participants were asked to select 3 scenarios or barriers of breastfeeding while working and to pick 2 responses at maximum in dealing with each of the chosen scenarios or barriers.

Intervention for the Control Group

Participants in the control group had access to standard-of-care services (e.g., outpatient breastfeeding support offered by midwives/lactation consultants).

Development of Intervention Materials

The procedure for developing the “if-then” plans is described in Supplemental Appendix 1.

Regarding training of MI (Supplemental Appendix 1), we had 2 certified MI trainers whose qualifications were recognized by the Motivational Interviewing Network of Trainers (MINT).

Measures

Background Information

At 6 weeks postpartum, we conducted baseline interviews via telephone to collect participants’ information using a structured questionnaire. The information included background information about the participants and their infants, information about breastfeeding (previous child and this pregnancy), and breastfeeding self-efficacy. If participants failed to answer our first call, we tried 2 more attempts at time periods different from the first ones. If all the attempts were in vain, we tried to contact the participants by WhatsApp, in which a shortened questionnaire (short version) was used. The short version comprised of questions about when milk came in, how likely was that the mother would combine breastfeeding and work till 6 months postpartum, number of feedings in the previous 24 hours (24 h) by direct breastfeeding/expressed milk either by hand or breast pump/formula feeding.

Breastfeeding Self-Efficacy at Baseline

This was assessed by the Breastfeeding Self-Efficacy Scale (BSES), ⁴² which comprised of 14 items. Each item was anchored by a 5-point Likert scale, ranging from “1 = not at all confident” to “5 = always confident.” The total score ranged between 17 and 70, with a higher score representing a higher level of breastfeeding self-efficacy. The Cronbach’s alpha of the present sample was 0.94.

Outcomes

At 6 months follow-up, participants were contacted again and asked if they had returned to work and if they were still breastfeeding their babies. If they responded that they were still breastfeeding their babies, we asked the participants the following information in the previous 24 h: (1) the number of milk feedings by direct breastfeeding, hand expression/pumping and infant formula; (2) if they had fed babies fluids and water; and (3) if they had fed babies solid food.

Exclusive breastfeeding at 6 months postpartum was defined as, when counting the milk-feeding portion of babies’ diet, participants feeding babies with breastmilk without infant formula. ⁸ Partial breastfeeding was when participants fed their babies breastmilk, infant formula, or other food or drinks. ⁸

Sample Size Calculation

The differences in breastfeeding rates between control and intervention groups were reported to be 6% to 20%. ⁴³ Therefore, we anticipated that the effect size to be 13%. According to the method suggested by Fleiss, ⁴⁴ the number of participants required was 219 per group and 438 participants in total, with a study power of 0.80 and a significance level of 0.05.

Statistical Analyses

Multiple imputation was used to deal with missing data. ⁴⁵ Predictive mean matching was used, in which each imputed value was defined by random selection of a complete case from the closest 5 prediction values. ⁴⁵ We multiply imputed variables that were missing at baseline. A total of ten imputed datasets with 15 iterations using the Markov Chain Monte Carlo method were computed. After that, parameter estimates and standard errors were obtained using multiple imputation combination rules. ⁴⁶ Comparisons of baseline variables between the study groups using the multiply imputed data and the complete cases yielded highly consistent results, demonstrating that the pattern of missingness was likely to be missing at random and justifying the use of multiple imputations to handle the missingness in our data.

Intention-to-treat (ITT) and sensitivity analysis (SA) were conducted to test if the score of BSES moderated the effect of MI plus IMI on breastfeeding exclusivity at 6 months postpartum. Model 1 of PROCESS macro was used to test the moderation effect. ⁴⁷ Since the macro was incompatible with the imputed dataset, the original dataset was used. The Johnson-Neyman procedure was used to detect which scores of BSES exerted a significant moderation effect. Furthermore, the conditional effect of BSE was assessed by conditioning the scores of BSES at the 16th, 50th, and 84th percentiles, which represented low, medium, and high levels of BSE in the sample.

Sensitivity analyses were carried out by repeating the analyses on the subgroup which involved participants who had completed baseline interviews (long version), had completed both the MI and IMI interventions, and reported to have paid employment at 6 months postpartum. A P-value less than 0.05 was considered statistically significant. Data analyses were carried out by SPSS version 27.0 (IBM SPSS Statistics for Windows. IBM Corp. Armonk, NY).

Results

We enrolled 623 participants in the postnatal wards. Subsequently, 460 participants were contacted at 6 weeks postpartum to complete baseline interviews and 163 participants did not respond (Figure 1). Compared to those who lost to baseline interviews (N = 163), those who completed the interviews (N = 460) had a longer duration of any breastfeeding of the last child (10.76 ± 27.82 vs 1.42 ± 7.18, P < .001), had babies with higher birthweight (3116.19 ± 358.47 vs 3033.93 ± 379.44, P = .01), were primiparous (29.10% vs 17.80%, P = .01), and had vaginal delivery (74.60% vs 65.60%, P = .03).

Figure 1.

Study flow chart.

Among the 460 participants (Table 1), 392 completed the long version and 68 completed the short version of baseline questionnaires, in which the mothers in the former group were older in age (33.17 ± 3.75 vs 32.09 ± 3.24, P = .01), and had more feedings by direct breastfeeding in the previous 24 h than the latter group (4.19 ± 3.61 vs 2.75 ± 3.05, P = .002).

Table 1.

Comparison of Maternal Characteristics Between Mothers Who Had Completed the Long Version Baseline Questionnaire and Those Who Had Completed the Short Version.

	Completion of baseline questionnaire at 6 weeks postpartum		P value
Maternal characteristics	Short version (n = 68)	Long version (n = 392)
Age (years)	32.09 ± 3.24	33.17 ± 3.75	.01
Duration of any breastfeeding of last child (weeks)	8.80 ± 25.80	11.10 ± 28.17	.53
Baby’s weight at birth (gram)	3125.29 ± 368.99	3114.61 ± 357.07	.82
Parity (%)
0	72.10	67.60	.39
1	27.90	29.30
2	0.00	3.10
Baby’s gender (%)
Female	51.50	51.00	1.00
Male	48.50	49.00
Preterm birth (%)
No	98.50	99.20	.47
Yes	1.50	0.80
Delivery method (%)
Vaginal	72.10	75.00	.65
Caesarean	27.90	25.00
When milk coming in (days postpartum)	4.11 ± 2.18	3.76 ± 2.38	.30
Number of feedings in the previous 24 h by
Direct breastfeeding	2.75 ± 3.05	4.19 ± 3.61	.002
Pumping	3.15 ± 2.34	2.75 ± 2.50	.22
Formula feeding	2.88 ± 2.72	2.31 ± 2.59	.10
How likely was that after return to work you continued breastfeeding till your baby was 6 months old? (%)
Definitely impossible/highly impossible	12.00	13.30	.15
Impossible	16.20	14.80
Definitely possible/highly possible	61.80	71.90
Feeding method at baseline (%)
Exclusive breastfeeding	27.90	39.30	.08
Mixed breastfeeding	72.10	60.70

The 460 participants were allocated to the 2 groups by randomization, in which 232 were in the control group and 228 were in the MI group. The completion rates of MI only, IMI only and MI+IMI in the intervention group were 70.2% (n = 160), 64.5% (n = 147) and 49.1% (n = 112), respectively. We excluded 15 and 19 participants in the control and MI groups, respectively, because at 6 months postpartum they reported they had quit their jobs 1 month after returning to work (i.e., homemakers) or lost to follow-up (Figure 1). As a result, the number of respondents at the 6 months follow-up was 217 and 209 in the control and MI groups, respectively. There were no significant differences in maternal sociodemographics, infant birth data, feeding methods that were measured at baseline, and duration of exclusive breastfeeding between the 2 groups based on ITT analyses.

Moderation Effect of BSES Score (ITT)

There was no significant moderation effect of BSES scores on the association between the 2 groups and breastfeeding exclusivity at 6 months postpartum (log odds = −0.02, 95% CI: −0.05, 0.02). This model did not include variables that had a significant association with BSES scores, which were parity (P = .02), feeding method at baseline (P = .001), number of infant formula feeding in the past 24 h at baseline interviews (P = .002), and the likelihood of continuing breastfeeding after returning to work (P < .001).

Sensitivity Analyses (SA)

The results of sensitivity analyses were in line with that of the ITT analyses, with a few exceptions described below. As shown in Table 2, compared to the control group, mothers in the MI group had less infant formula feedings in the past 24 h that was measured at baseline (1.68 ± 2.42 vs 2.36 ± 2.59, P = .02), a lower proportion of mothers opined that they were definitely or highly impossible to combine breastfeeding and working till 6 months postpartum (4.63% vs 15.95%, P = .01), higher proportion of the mothers practicing EBF at baseline (50.93% vs 38.30%, P = .04), and longer median (in weeks; interquartile range IQR) duration of EBF (13.00 [IQR: 0.00-24.72] vs 4.30 [IQR: 0.00-17.30], P = .02). Regarding the moderation effect of the BSES score being measured at baseline (Figure 2), a significant moderation effect was observed when the BSES score was lower than 42.00. Participants who scored below 42.00 accounted for 39.50% (n = 117) of the study population. The moderation effect of BSE was assessed by conditioning the scores at the 16th, 50th, and 84th percentiles of the continuous BSES scores distribution, which represented scores of 33.00 (low self-efficacy), 47.00 (medium self-efficacy), and 57.00 (high self-efficacy), respectively. A significant moderation effect was found in the low self-efficacy group (log odds = 1.23, 95% CI: 0.20-2.26), but not in the medium (log odds = 0.40, 95% CI: −0.21 to 1.02) and high (log odds = −0.19, 95% CI: −0.97 to 0.58) self-efficacy groups. That is, among the participants belonging to the low self-efficacy group, the odds of participants in the MI group continuing EBF at 6 months postpartum was 3.42 times (exp(1.23) = 3.42) that of those in the control group. This model did not include variables that had a significant association with BSES scores, which were those mentioned in the moderation effect of BSES scores in the intention-to-treat analysis.

Table 2.

Comparison of Maternal Characteristics Between Mothers in MI and Control Groups Using Sensitivity Analysis Approach.

Maternal characteristics	Treatment group		P value
	Control (n = 188)	MI (n = 108)
Age (years)	32.99 ± 3.81	33.42 ± 3.71	.34
Duration of any breastfeeding of last child (weeks)	9.93 ± 22.74	16.65 ± 39.86	.11
Baby’s weight at birth (gram)	3117.09 ± 342.28	3129.49 ± 371.71	.77
Parity (%)
0	68.62	60.19	.09
1	29.79	34.26
2	1.60	5.56
Baby’s gender (%)
Female	51.06	57.41	.33
Male	48.94	42.59
Preterm birth (%)
No	99.47	98.15	.30
Yes	0.53	1.85
Delivery method (%)
Vaginal	71.28	79.63	.13
Caesarean	28.72	20.37
Planned duration of any breastfeeding (month)	6.82 ± 4.32	7.39 ± 4.82	.30
Intention of exclusive breastfeeding (%)
Yes	59.04	66.02	.45
No/unsure	40.96	33.98
Planned duration of exclusive breastfeeding (month)	3.93 ± 4.45	4.15 ± 3.96	.67
Return to work (weeks postpartum)	13.20 ± 2.89	13.01 ± 2.44	.57
When milk coming in (days postpartum)	3.81 ± 2v53	3.54 ± 2.00	.33
At 6 weeks postpartum, number of feedings in the previous 24 h by
Direct breastfeeding	3.98 ± 3.40	4.88 ± 3.96	.50
Pumping	2.74 ± 2.39	2.81 ± 2.82	.85
Formula feeding	2.36 ± 2.59	1.68 ± 2.42	.02
Total score of BSES	44.88 ± 10.82	47.49 ± 11.21	.05
Marital status (%)
Single	0.00	1.85	.06
Married/cohabitation	100.00	98.15
Education level (%)
Primary school or below	0.00	0.00	.38
Lower secondary school	1.06	0.00
Upper secondary school or post-secondary education	43.24	32.78
Bachelor degree	41.28	50.56
Master degree or above	14.41	16.67
Monthly household income (HK$) (%)
<30 000	11.86	8.43	.27
30 000-39 999	14.89	12.22
≥40 000	73.24	79.35
Participating mothers had been breastfed by their own mother when they were babies (%)
No/unknown	65.85	64.72	.17
Yes	34.15	35.28
How likely was that after return to work you continued breastfeeding till your baby was 6 months old? (%)
Definitely impossible/highly impossible	15.95	4.63	.01
Impossible	17.55	10.19
Definitely possible/highly possible	66.49	85.19
Feeding method at baseline (%)
Exclusive breastfeeding	38.30	50.93	.04
Partial breastfeeding	61.70	49.07
Duration of exclusive breastfeeding (weeks) (median; IQR)	4.30; 0.00-17.30	13.00; 0.00-24.75	.02

Abbreviation: BSES, Breastfeeding Self-Efficacy Scale.

Figure 2.

Moderation effect of BSES score on the association between MI plus IMI and breastfeeding exclusivity at 6 months postpartum based on sensitivity analysis.

Discussion

The present study demonstrated that BSE moderated the effect of MI on EBF at 6 months postpartum in working mothers. Our observation paralleled Miller and Rollnick’s viewpoint that participants who initially had low confidence or readiness for change could benefit from MI intervention. ²¹ Our participants with low BSE at baseline represented those who were not fully committed to or confident about sustaining EBF at 6 months postpartum. MI could provide opportune moments for these participants to express what they gained from feeding their babies infant formula and disliked about breastfeeding, thus evoking “sustain talk” (ST; statements in favor of formula feeding). ST were cues for MI workers to use MI-consistent strategies to help the participants solve their ambivalence, build their confidence and evoke in-depth “change talk” (statements in favor of EBF), ⁴⁸ which were authentic self-reflections endorsing their desire, ability, confidence, and commitment toward EBF continuation. In contrast, participants who had medium or high levels of BSE would think that exploring the ambivalence was unnecessary. Under this situation, it is recommended to discuss with the participants their plans and actions about continuing EBF when they return to work. ²¹ The present study’s observation that the moderation effect of BSE on breastfeeding exclusivity is echoed by a recent study, ²² which showed that brief MI boosted BSE at 6 months postpartum in mothers who initially had low BSE, thus significantly lowering the chance of cessation of EBF at the end of the study. Our study finding about the moderation effect of self-efficacy on the use of MI in promoting breastfeeding provides the impetus for further research, which evaluates if delivering the 4 core tasks of MI (engaging, evoking, focusing, and planning) in a standardized manner for participants with a high level of confidence/self-efficacy dampens MI’s efficacy or not.

Though we observed that MI was effective in promoting EBF among mothers with low BSE, MI was not effective in doing so among all the participating mothers (Supplemental Appendix 2). We chose to deliver the intervention at 8 and 9 weeks postpartum because of 2 reasons. First, participants would likely have refused to take part in the intervention activities during the early postpartum period, for instance, at 2-3 weeks postpartum, when they were busy taking care of their own health and newborns. Second, since most of the mothers would return to work at around 10-12 weeks postpartum, administering the intervention activities before they return to work would be good to prepare them for continuing breastfeeding after returning to work. It was very likely that our intervention was implemented as intended, because we followed the protocol strictly and MI was carried out by trained workers. We anticipated that the timing and intensity of conducting MI, as well as the low uptake rate of the intervention were the reasons. Ample evidence showed that new mothers are vulnerable to giving up breastfeeding before 1 month postpartum, a period in which new mothers need emotional and practical support to sustain breastfeeding.^{49 -51} We delivered MI at 8 weeks postpartum, in which if new mothers had established EBF at a much earlier postpartum period they no longer required MI to boost their self-efficacy/confidence. Furthermore, literature about the effect of breastfeeding support could help shed light on why a single session of MI was unable to promote EBF at 6 months postpartum. A meta-analytic review demonstrated that the intensity and type of provider (lactation consultants or peers) of breastfeeding support had no significant effect on EBF at 6 months postpartum in new mothers, ⁵² which reflects that community-wide breastfeeding support is one of the indispensable factors to breastfeeding exclusivity. In Hong Kong, implementing a breastfeeding supportive policy for the female employees who want to continue breastfeeding is not compulsory for employers. Without breastfeeding-friendly workplaces, working mothers would lack confidence in negotiating with employers for lactation breaks or flexible working arrangements that are compatible with their lactation needs; and would worry about criticism from supervisors/colleagues. Under such circumstances, though using MI in the present study may help enhance participating mothers’ competence/confidence in breastfeeding their babies exclusively when they are off duty, using MI was unable to attenuate mothers’ ambivalence regarding breast milk expression while they are on duty because of their worries about harassment, discrimination and victimization that were associated with breast milk expression in workplaces.

We used validated questionnaires to measure BSE. Regarding training of MI, we followed a standard protocol to conduct rigorous training and ongoing supervision was available to ensure quality MI interviews. Low attrition was observed in the present study. Yet, our study had limitations. First, since our participants had relatively high household income and educational attainment, the study findings were not generalizable to working mothers who were socially disadvantaged. Second, the characteristics of working mothers who were eligible to join the study but refused would have been different from our study participants; and we were unable to collect the information of the former group. Third, we did not include working mothers whose babies were admitted to the neonatal intensive care unit (NICU)/special nursery. Their breastfeeding experience, attitudes, and self-efficacy of continuing breastfeeding upon return to work would be different from our participants. Fourth, our study was unable to untangle the individual effects of MI and IMI. The effects of MI or IMI on terminating early cessation of breastfeeding have important implications in terms of cost-effectiveness.

Implications

The present study trained laypersons to deliver MI. The possibility of training peers who have breastfeeding experience to conduct MI exists when there is a standard protocol for the training. MI emphasizes empathy and supports autonomy, which are essential for successful breastfeeding support because the literature reveals that breastfeeding continuation requires empathetic contact within the peer counselor-breastfeeding mother pair, and supporting maternal autonomy toward breastfeeding.^19,53

The null results of the present study should not be regarded as an indication that MI is not useful in promoting breastfeeding continuation. The potential of applying MI to promote sustained breastfeeding among socially disadvantaged mothers and mothers whose babies have been admitted to a special nursery/NICU remains unknown. Further studies are needed to evaluate the efficacy of MI in these populations.

Insufficient breastfeeding-friendly workplaces remain a huge obstacle for working mothers to continue breastfeeding. Public health interventions that help nurture a breastfeeding-friendly culture in the workplace and community are essential, in addition to individualized interventions for enhancing maternal breastfeeding self-efficacy.

Conclusions

Intervention comprising of MI and IMI was effective in promoting EBF at 6 months postpartum among working mothers with insufficient BSE. Public health interventions that help nurture a breastfeeding-friendly culture in the workplace and community should be coupled with individualized interventions to dampen early cessation of breastfeeding.