For both women's groups and community health worker interventions, coverage is thought to be a critical factor for a positive impact on neonatal mortality rates [14]. Simple quantitative data indicate considerable success in the scale-up of the women's groups in rural Bangladesh. We showed an approximately five-fold increase in population coverage, a 2.5-fold increase in the women's group membership among women of reproductive age and a 10-fold increase in the proportion of women who gave birth and reported attending a women's group. Scale-up was achieved without financial incentives for women and without any increase in managerial staff. Nevertheless, enhancement of already strong operational capabilities and institutional knowledge within our Perinatal Care Project (PCP) was crucial for the success of scale-up. Furthermore, PCP works under the auspices of BADAS, whose positive reputation as the largest non-governmental healthcare provider in Bangladesh facilitated acceptance of the intervention.
The specific impact of the scale-up on mortality rates is yet to be determined through an ongoing cluster-randomised trial [17], which includes detailed analysis of process indicators of delivery and receipt of the intervention. Nevertheless, documenting the process and success of scale-up is itself important if development initiatives are to foster participation and remain accountable to communities, and to avoid cumbersome organisational structures that are detached from their grassroots bases [23]. This paper does not presume to provide the ultimate solution to such challenges, but rather shares experiences of a single intervention in a resource-poor setting. Though limited to just nine unions in rural Bangladesh, these experiences and the described principles and processes of scale-up and its measurement are likely to be relevant to the wider development community, notwithstanding the need for intervention- and context-specific alterations and for programme flexibility.
Experience of delivery of the intervention on a smaller scale enabled PCP to meet the majority of requirements for successful scale-up as described by Simmons et al. (2006)[24] and summarised by Gilson and Schneider (2010)[19]. For example, clear messages about the objectives and advantages of women's groups were communicated to the community, while mapping exercises, household visits and community orientation meetings ensured early involvement through personal contact. Similarly, recognition of the importance of locally recruited and trained facilitators to manage groups, the participatory nature of the intervention, and active diffusion of key messages through household visits and community, governmental and non-governmental networks were central to the scale-up process. Concurrent systematic monitoring of the process and outcomes of scaling up further met Simmons et al's (2006)[24] and Gilson and Schneider's (2010)[19] recommendations to use evidence to guide and evaluate scale-up.
The scaling-up process described is a logical sequence and represents more or less the order in which various stages were initiated. In practice, however, the linear presentation is artificial and it is important to recognise that scaling-up is a dynamic, non-linear, iterative process whereby the various phases may occur simultaneously, in different orders, have feedback loops and may have to be repeated or revisited [18]. Feedback systems that allow for periodic programme corrections and continued innovation are central to successful scale-up and require managerial flexibility and strategic flair [19, 25].
Training and capacity building of staff is crucial for a thorough understanding of the group process and a sense of ownership at the grassroots level. Recruitment and selection of appropriate staff is critical, as are planned strategies to cope with staff turnover. Local recruitment was vital in maintaining effective relationships with the local community through familiarity with local customs, knowledge and beliefs. Recruitment criteria, however, cannot be rigid in areas of low literacy, low education and where there are gender barriers to employment. The resource-poor, rural context of our study area and an initial lack of appreciation by project managers for the need for family understanding of facilitators' roles increased vulnerability to staff turnover at the beginning of scale-up. This necessitated greater flexibility in recruitment criteria and processes, including a probationary period of field exposure, which ultimately minimised disruption caused by staff turnover. Effective systems of supervision, review and refresher training of project staff are likely to enhance delivery and sustainability of any intervention and therefore the success of any scale-up initiatives.
Various definitions of scaling-up exist, relating to complexity, impact and interactions with other organisations, but the concept of scaling-up as expansion of coverage is the most common [15, 23]. Geographical expansion is associated with particular challenges such as increasing distance from project headquarters, and larger areas and organisational structures to manage. Notwithstanding the introduction of groups into new villages in the existing unions, we did not expand geographically. Rather, our experience was an increase in the intensity of intervention in the same areas to increase population coverage. Some of the issues around effective geographical expansion and increased intensification are the same, however. In each case, scale-up does not mean exact duplication of pre-existing strategies. In our scale-up, lessons were learned from the previous, smaller-scale implementation of women's groups but existing structures and procedures were not always copied. For example, the schedule of group meetings and content were revised before scale-up to emphasise participation of women of reproductive age, and especially pregnant women. As a flexible replication of similar women's groups interventions in Nepal and India we endorse the recommendation that replication as a path for scaling up is only likely to work if done flexibly [23].
Strategies to measure the success of scale-up depend on specific objectives and anticipated outcomes. In relation to population coverage or intervention expansion, specific indicators of success, and how they are measured, are essential. There is no one-size-fits-all strategy for monitoring and evaluation and different methods have differing degrees of complexity and resource demands [26, 27]. Data from three different sources were used to measure the success of our scale-up, a particular strength of this initiative.
The capacity needed for such monitoring and evaluation systems should not be underestimated, however. Complex systems add to the cost of interventions considerably and are vulnerable to the limitations of population survey methods, such as recall and reporting biases [18, 27]. Furthermore, direct measurements of certain phenomena are not always straightforward or possible. For example, intervention coverage among pregnant women in the women's group intervention could not be measured directly as it was not possible to accurately measure the pregnancy status of all reproductive-aged women. The proportion of births that occur to women who report attendance at a women's group was therefore used as a proxy measure of coverage among pregnant women. This estimation depends on accurate reporting of women's group attendance and on all pregnancies ending in delivery, which is unlikely to be true. Nevertheless, provided that limitations are acknowledged and measurement methods are consistent pre- and post-scale-up, utilising proxy measures and actual data are more informative than entirely modelled estimates. Alternative methods for estimating coverage of an intervention among pregnant women with varying degrees of complexity, cost and data demands are described and discussed elsewhere [28].
Although the sustainability of women's groups is yet to be formally evaluated, unpublished data suggest that a high proportion of groups in Nepal and Malawi are running long after withdrawal of project financial support. Successful delivery of the intervention as planned during the first 18 women's group meetings may be considered indicative of institutional sustainability in terms of ongoing capacity to lead training, maintain the infrastructure, equipment and supplies and provide the necessary inputs and environment for implementation. Similarly, fairly consistent attendance levels at the women's groups (Figure 2) and reasonable population exposure estimates indicate programmatic sustainability from a community participatory perspective. Continued capacity building of women's group members, perhaps with a focus on facilitation methods for example, may obviate the need for paid facilitators and thus enhance the long-term financial sustainability of the intervention, whilst political sustainability can only be achieved through enhanced communication and advocacy efforts at local, national and global levels.
The ability to influence national policy from local initiatives is therefore a final determinant and indicator of successful scale-up [23, 25]. Before the women's groups became active, considerable time was spent identifying available services and other governmental and non-governmental organisations in the selected unions to communicate and build understanding and trust within the local community. These efforts enhance the participatory nature of scale-up as opposed to being purely driven by the implementing institutions or experts. In the women's groups, establishing links with other key players was beneficial in a number of ways, not least by involving traditional birth attendants and community health workers in raising awareness of the intervention among pregnant women encountered during their routine activities. We cannot objectively assess the success of wider stakeholder engagement in terms of bridging gaps and influencing policies, but these efforts are undoubtedly important for the acceptability and long-term sustainability of scale-up.