Effects of breastfeeding training programmes for midwives on breastfeeding outcomes: a systematic review and meta-analysis

Of the 10 articles included, nine were published in English and one in Chinese. They were published between 2000 and 2019. Five RCTs were included [25, 31‐34]. One study was reported in three separate papers each reporting on different outcomes [31‐33]. Five reports described quasi-experimental studies [26‐30].

The settings of these studies were varied, including large and small hospitals in urban and rural areas. Two studies, reported in four articles, were carried out in Sweden [25, 31‐33], two in the United Kingdom [26, 27], one in the United States [29], one in Croatia [30], one in Bangladesh [34], and one in Taiwan [28].

Each of the included studies assigned participants to intervention or control groups in different ways. Five studies assessed the KAP towards breastfeeding of 386 midwives and nurses (sample sizes ranged from 28 to 135) [25‐29]. Of these studies, two recruited midwives only [26, 27], while the other three recruited both nurses and midwives [25, 28, 29].

Four studies were quasi-experimental studies [26‐29]. Moran et al. collected data from four sites (A, B, C, and D) [26]. As site A did not implement the training course, midwives at site A were assigned to the control group. All midwives at site B attended the course and were thus assigned to the experimental group. Despite the course being delivered at sites C and D, only some of the midwives had attended the course at the time of data collection. Therefore, midwives from sites C and D were assigned to both control (those who had not yet taken the course) and intervention (those who had taken the course) groups. In the study by Law et al. 108 midwives received breastfeeding training and 27 final-year student midwives constituted a control group [27]. Wang and Ku and Al-Nuaimi et al. recruited midwives from two hospitals using convenience sampling [28, 29]. One hospital was selected as the intervention group and delivered a breastfeeding training course. The other hospital served as the control group. In contrast, Ekström et al. carried out an RCT [25]. Their sampling frame consisted of 10 municipalities that were paired based on size and breastfeeding duration. The municipalities were then pairwise randomised to either an intervention group or a control group.

An additional three studies, reported in five articles, measured breastfeeding initiation, rates, and duration and included 3,463 mothers cared for by midwives (sample size ranged from 480 to 2,330) [30‐34]. Zakarija-Grkovic et al. recruited mother/infant pairs before and after breastfeeding training of the maternity staff [30]. In their study, the breastfeeding training course was conducted twice: in May 2008 and February 2009. The control group was recruited from February to May 2008, and the intervention group was recruited from April to August 2009. Shamim et al. carried out their study in Bangladesh, which is divided into 64 districts, 493 sub-districts, and nearly 4,500 unions [34]. They randomly selected three sub-districts from the five sub-districts constituting the Panchagarh district. Of the 26 unions in the three selected sub-districts, nine unions were randomly selected and randomised into three groups. Outcomes were then compared between mothers in a control group (CG), who lived in unions where services were provided by midwives with no training; those in the intervention group (IG), who were living in unions where services were provided by midwives with breastfeeding training; and mothers in a supervision group (SG), who were living in unions where services were provided by midwives with breastfeeding training and supervision. Ekström et al., Blixt et al., and Ekström and Stina reported on the same study [31‐33]. In this study, ten municipalities in Sweden were randomised to either an intervention group or a control group. Midwives in the intervention group were given breastfeeding training, while midwives in the control group were not. Mothers in the intervention or control municipalities were assigned to the intervention or control group, respectively.

All included studies used multiple instruments to measure various outcomes at different data collection time points. Moran et al. and Law et al. assessed breastfeeding knowledge and skills using the Breastfeeding Support Skills Tool (BeSST), a questionnaire including 20 open and ten closed questions based on four video clips [26, 27]. In the validation study, the BeSST demonstrated excellent internal reliability (Cronbach's α = 0.89) and inter-rater reliability (intraclass correlation coefficient = 0.96) [26, 35]. Wang and Ku used the Breastfeeding Knowledge Scale to assess breastfeeding knowledge and practice, which comprised 62 questions, each with a correct answer [28]. Total scores ranged from 0 to 62. The Kuder-Richardson Formula 20 (KR-20) of the scale was 0.83. Al-Nuaimi et al. used two validated questionnaires developed by the American Academy of Pediatrics to collect data about midwives’ breastfeeding knowledge and practices, but they did not report the reliability and validity of the scale [29].

To evaluate attitudes towards breastfeeding among midwives, Wang and Ku used the nine-item Breastfeeding Attitude Scale, which is assessed using a six-point scale ranging from strongly disagree (1 point) to strongly agree (6 points) [28]. The higher the score, the more positive the midwife’s attitude towards breastfeeding. The Cronbach’s α of the scale was 0.78. Ekström et al. used the Breastfeeding Attitudes Instrument, consisting of four dimensions relating to breastfeeding (regulating, facilitating, disempowering, and antipathy) [25]. The Cronbach’s α of the scale was 0.51. The internal consistencies for each factor were 0.80 (regulating), 0.60 (facilitating), for 0.62 (disempowering), and 0.29 (antipathy) [25]. Al-Nuaimi et al. used a seven-item questionnaire, but they did not report the reliability and validity of the scale [29].

The three studies that were reported on in five articles used self-designed questionnaires to test the secondary outcomes of breastfeeding initiation, rates, and duration [30‐34]. Among them, Blixt et al. and Ekström et al. also collected data from maternal and/or neonatal medical records [31, 32].

Of the five studies included in this review that measured the primary outcome, four measured the effects of a breastfeeding training programme on the knowledge and skills of midwives, with all reporting a positive effect (p< 0.05) [26‐29]. The pooled results of these four studies (305 participants), combined as an SMD, was 1.33 (95%CI, 0.98 to 1.68) suggesting statistically significant beneficial effects (p < 0.01) of the breastfeeding training programme, with moderate heterogeneity (I² = 36%) (See in Fig. 2).

Three studies examined changes in midwives’ attitudes, and all of these concluded that breastfeeding training programmes could improve their attitudes towards breastfeeding (p< 0.05) [25, 28, 29]. However, these studies could not be combined for meta-analysis due to heterogeneity of scales and scoring methods.

The secondary outcomes were measured in three studies reported in five different articles [30‐34]. Three studies assessed breastfeeding initiation [30, 33, 34]. One study reported that mothers in the intervention group reported earlier initiation (within 24 h after delivery) and higher frequency (within 24 h) of breastfeeding compared with mothers in the control group (p< 0.05) [33]. Similarly, Shamim et al. reported that the intervention group had a statistically significantly greater proportion of mothers reporting breastfeeding initiation ≤ 1 h after birth (340/353, 96.3%) compared with the control group (383/437, 87.6%) (p< 0.05) [34]. In contrast, Zakarija-Grkovic et al. reported that the proportion of mothers who initiated breastfeeding in the control group (387/388, 99.7%) was higher than that in the intervention group (378/385, 98.2%) (p< 0.05) [30]. However, meta-analysis of these three studies was not possible due to high heterogeneity (I² = 98%).

The rates of exclusive breastfeeding, assessed by two included studies [30, 34]. Zakarija-Grkovic et al. reported that the proportion of newborns exclusively breastfed during the first 48 h after birth increased from 6.0% to 11.7% (p< 0.05) [30]. Unfortunately, this effect did not persist, with no differences seen in breastfeeding rates at discharge or at three, six, or 12 months postpartum between the groups. Similarly, Shamim et al. also reported no statistically significant difference in the rate of exclusive breastfeeding within six months between the groups [34].

Other benefits have also been reported: after breastfeeding training for midwives, mothers reported a statistically significantly longer duration of exclusive breastfeeding (p< 0.05) [31, 33], fewer breastfeeding challenges (p< 0.05) (e.g. insufficiency in breast-milk) [32, 33], higher satisfaction with the breastfeeding counselling received (p< 0.01) [32], and fewer infants receiving breast milk substitutes in the first week of life without medical reasons (p< 0.05) in the intervention group compared with the control group [31, 33, 34].

The breastfeeding training programmes in the included studies incorporated a range of different components. The three main components were breastfeeding theoretical knowledge, supportive skills, and counselling skills, which were incorporated by five included articles (describing three studies) in their breastfeeding training programmes [25, 29, 31‐33]. The remaining five studies only included breastfeeding theoretical knowledge and supportive skills [26‐28, 30, 34].

Breastfeeding training programmes containing all three components improved midwives’ knowledge [29], skills [29], and attitude towards breastfeeding [25, 29]. Additionally, such programmes were associated with the use of fewer breast milk substitutes in the first week of life without medical reasons, later introduction of milk substitutes after discharge from the hospital [31], earlier initiation of breastfeeding, higher frequency of breastfeeding, longer duration of breastfeeding, and fewer breastfeeding challenges [33], such as insufficient breast milk supply [32].

However, programmes that included only breastfeeding theoretical knowledge and supportive skills components were also associated with improved knowledge, skills, and attitude towards breastfeeding of midwives [26‐28]. However, Zakarija-Grkovic et al. reported that the proportion of newborns exclusively breastfed during the first 48 h after birth increased, but there were no differences in breastfeeding rates at discharge or at three, six, or 12 months after birth between the intervention and control groups [30]. Shamim et al. similarly reported no statistically significant difference in the rate of exclusive breastfeeding between the groups [34].

The durations of the breastfeeding training programmes reported by the included studies varied from two hours to seven days. Of the ten included articles, Ekström et al., Ekström et al., Blixt et al., and Ekström and Stina had the longest training durations [25, 31‐33]. They provided a seven-day process-oriented training programme for participants in the intervention group. Positive effects were reported for both primary and secondary outcomes. However, other long training durations (e.g. five days) led to higher proportions of mothers reporting early initiation of breastfeeding, but no statistically significant difference was observed in the rate of exclusive breastfeeding between the groups [34].

Two studies implemented the 20-h WHO/ UNICEF Breastfeeding Management Course and reported improved breastfeeding knowledge and skills among midwives and an increased proportion of newborns exclusively breastfed during the first 48 h after birth [26, 30]. However, no statistically significant differences were reported in breastfeeding rates at discharge or at three, six, or 12 months after birth between the intervention and control groups [30].

The remaining three studies provided midwives with short training sessions (two hours to eight hours) [27‐29]. These studies reported positive effects on the KAP of midwives, but secondary outcomes were not assessed.

Of the eight included studies, only four studies provided teaching materials to the midwives involved in the training programmes [26, 28‐30]. Moran et al. provided midwives with a comprehensive, fully referenced workbook that could be used to cascade the training within their local maternity care services [26]. Zakarija-Grkovic et al. used standard course materials including guidelines for course facilitators, outlines for course sessions, and PowerPoint slides for the course [30]. The teaching materials used by Wang and Ku were the National Health Bureau, Department of Health, Executive Yuan Guidelines for Breastfeeding Teaching Materials in Taiwan and the Breastfeeding Question and Answer Manual [28]. Al-Nuaimi et al. developed two educational materials based on up-to-date evidence, including recommendations from the WHO and the National Institute for Health and Care Excellence (NICE), addressing the importance of breastfeeding initiation and child growth and development from birth to five years of age [29].

The remaining four studies, reported on in six articles, did not provide information about the teaching materials, if any were included in their programmes [25, 27, 31‐34].

Lectures and group discussions were the most common training programme formats and were used by three studies, reported on in five articles [25, 27, 31‐33]. Role play was used by two studies [27, 34]. Additionally, group facilitation, group work, case studies, demonstrations, and field trips were also used in one training programme [34]. However, four studies did not provide any information about the format of the breastfeeding training programme implemented [26, 28‐30]. All formats were found to be effective in improving both primary and secondary outcomes.

Breastfeeding training programme providers varied widely among the included studies. The course provided by Moran et al. was organised in the UK by the UNICEF UK Baby Friendly Initiative Team and was taught by midwives and health visitors employed by UNICEF [26]. Wang and Ku invited breastfeeding teachers from the National Health Bureau, Department of Health, Executive Yuan to teach the course [28]. Shamim et al. organized a 5-day training course for the trainers by Training and Assistance for Health and Nutrition and Eminence [34]. The teachers of the course of Zakarija-Grkovic et al. were a neonatologist, a gynaecologist, a paediatrician, an economist/representative of a voluntary parenting group, a psychologist, and a general practitioner who was also a board-certified lactation consultant, midwife, and community nurse [30].

However, no details were provided in the six remaining articles regarding the characteristics of the person(s) providing the intervention in the four studies they reported [25, 27, 29, 31‐33].