In this study, the items used to observe knowledge, self-efficacy, and action competence proved to measure three different latent constructs. Furthermore, we showed that the breastfeeding training programme provided in the Breastfeeding Trial improved health visitors’ knowledge, self-efficacy and action competence related to providing breastfeeding support, even though scores in the latter two constructs already were high at baseline. Thus, our initial hypothesis was supported: the training programme could improve knowledge, self-efficacy and action competence. This confirms the findings of two previous studies investigating self-efficacy in Danish health visitors following training. Although study populations were a third in size compared to the present, and self-efficacy was measured by slightly different instruments, the relative increase in self-efficacy in the intervention group was similar to the one we identified with equally high baseline values (19, 26). However, in these studies the self-efficacy in the control groups also increased, watering down the effect of the training programme, and thus leaving the intervention effects about half of the effect found in present study. One study (19) reported baseline knowledge scores similar to the present, and none of the studies included measures of action competence.
In a review by Mulcahy et al., in which studies that take a coaching style approach to teaching health professionals effective communication were included, findings suggest that interactive teaching components along with theoretical knowledge are more effective than formal classroom-based interventions. (33). The training programme evaluated in the present study included many of the effective aspects pointed to by the review and focused on teaching health visitors to enhance mothers’ breastfeeding self-efficacy, to support the families’ unique wishes for breastfeeding, and to help them maintain breastfeeding even when facing challenges, consistent with findings from a needs assessment performed prior to the present trial (34). It did not, however, include individual coaching. Yet, all municipalities had a local representative who joined monthly dialogue meetings with the instructors for sharing of experiences and asking questions, thus resembling group-based coaching. The training especially emphasised supporting women of young age and women with a low educational attainment (34), in line with the recommendation by Entwistle et al., who urged the training of health professionals “…to focus on the knowledge and skills required for them to assess and promote the self-efficacy expectations of new mothers in relation to their breastfeeding experience” (35) (p.239).
For the knowledge and action competence constructs, we found small improvements between baseline and follow-up in both trial arms. An explanation could be that knowledge can be easily accessible, and so health visitors in the control group, if in doubt about specific questions when answering the baseline questionnaire, could have felt spurred on to actively acquire the knowledge needed to respond correctly at follow-up. Improved action competence could follow from improved knowledge. Conversely, for self-efficacy to improve there may be a need for more than theoretical and practical input, as argued by Dykes (21). The training programme integrated elements argued to stimulate confidence in breastfeeding counselling, including embodied knowledge, vicarious knowledge, and practise-based knowledge (21). Thus, improving practise-based and vicarious knowledge could feed into self-efficacy and even more so to action competence, because it provides a more solid foundation for health visitors to act from.
Most of the knowledge provided in the training programme in the Breastfeeding Trial is identical to the training provided in the health visitor education. In the adjusted models for knowledge (covariate estimates provided in Additional File 3), each extra year since education as health visitors demonstrated no impact on knowledge gain at follow-up. Thus, experience was not necessarily a facilitator for improvement of knowledge. Dykes argue that knowledge stem for four different sources: embodied-, vicarious-, practice-based- and formal theoretical knowledge (21). And so, if for instance experience from a health visitor’s practice counteract the formal theoretical knowledge presented in the training programme, the latter might not be stored in the memory as effectively. Mulcahy et al. argue in a recent systematic review that “new information can only be connected to existing knowledge if the knowledge to support the new information exists within the learner” (33) (pp. 11–12). Thus, the intervention may have been more useful for newly educated nurses, who already held similar knowledge and needed to boost their memories.
At baseline, a larger proportion of health visitors in the intervention group than in the control group felt in doubt about their abilities to provide breastfeeding support (Table 3). Because the study was randomised and thus characteristics theoretically equally distributed between the groups, this may reflect that the sheer anticipation of the intervention could have left health visitors in the intervention arm more insecure about their abilities to provide breastfeeding support. Supporting this hypothesis, we found no difference in the levels of knowledge at baseline (Table 4), which can be argued to be less affected by the anticipation of the training. The same pattern was found in another study (26). Contrary, a similar study found baseline levels to be higher in the intervention group, however, in that particular study, baseline measurements were collected after the training had been provisioned (19).
Our findings show that the training programme was effective, even though it did not include an individual coaching element as identified important by Mulcahy et al. (33) and thus may be a more cost-effective approach to post-graduate training. A next step will be to investigate whether out findings translates into enhanced breastfeeding self-efficacy and -practices in the group of mothers receiving breastfeeding support from the trained health visitors, as it has been found by others (36, 37). Similarly, it might be relevant to investigate which of the three constructs influence the practical breastfeeding support more, in conjunction and individually.
4.1 | Strengths and Limitations
The present study has many strengths, including the randomised nature of the study, as well as the pre and post-test design. Moreover, the large number of study participants and a baseline response rate above 85% benefits the generalizability of the findings. The confirmation of the factor constructs of each of our reported constructs adds to the strengths. And finally, our study contributes to overcoming the paucity of evidence (4) in the field of training conducted in large groups to improve the self-reported breastfeeding support competencies of health professionals.
However, this study also has some limitations. Firstly, we did not collect data on health visitors’ own breastfeeding experience, even if it has been found to be the foundation of breastfeeding support provided by health professionals (38). However, in a previous study this source of information about breastfeeding was found to be of lesser importance compared to theoretical knowledge (19).
Secondly, we assessed the health visitors’ self-reported self-efficacy, action competence and knowledge. Thus, we cannot know if this will lead to changes in supportive behaviour as perceived by mothers. However, previous studies have found that health visitors’ knowledge about breastfeeding is the most important predictor of their actual supportive behaviour (19, 39), providing us with evidence that underlines the usefulness of our findings.
Thirdly, regarding participation and contamination, there is a possibility that the Hawthorne effect (40) by which study participants change their behaviours due to being observed, may have impacted the control arm to focus more on breastfeeding. The positive changes in knowledge between baseline and follow-up for health visitors in the control arm support this.
Fourthly, because we collected baseline data after randomisation had been done, health visitors in each trial arm knew if they would get the training programme or not. This seemed to affect the levels of self-efficacy and action competence at baseline. Moreover, conducting the follow-up collection six months after the training programme had been completed could have washed out effects of for instance knowledge, as shown in Tanaka and Horiuchi, who detected lower proportions of correct replies after just three months compared to immediately after training (41).
Lastly, we did not account for multiple testing and therefore we cannot exclude the possibility of random findings of effectiveness. Nevertheless, although not reported, the changes in self-efficacy and action competence sum scores were highly statistically significant (< .0001 and < .001, respectively) and so accounting for the multiple tests conducted in this study by applying even conservative methods like the Bonferroni correction would not increase the probability a null-finding above 5% for the self-efficacy result, at the least.