What mechanisms drive uptake of family planning when integrated with childhood immunisation in Ethiopia? A realist evaluation

Background: Maternal and child health are key priorities among the Sustainable Development Goals, which include a particular focus on reducing morbidity and mortality among women of reproductive age, newborns, and children under the age of five. Two components of maternal and child health are family planning (FP) and immunisation. Providing these services through an integrated delivery system could increase the uptake of vaccines and modern contraceptive methods (MCMs) particularly during the post-partum period. Methods: A realist evaluation was conducted in two woredas in Ethiopia to determine the key mechanisms and their triggers that drive successful implementation and service uptake of an intervention of integrated delivery of immunisations and FP. The methodological approach included the development of an initial programme theory and the selection of relevant, published implementation related theoretical frameworks to aid organisation and cumulation of findings. Data from 23 semi-structured interviews were then analysed to determine key empirical mechanisms and drivers and to test the initial programme theory. These mechanisms were mapped against published theoretical frameworks and a revised programme theory comprised of context-mechanism-outcome configurations was developed. A critique of theoretical frameworks for abstracting empirical mechanisms was also conducted. Results: Key contextual factors identified were: the use of trained Health Extension Workers (HEWs) to deliver FP services; a strong belief in values that challenged FP among religious leaders and community members; and a lack of support for FP from male partners based on religious values. Within these contexts, empirical mechanisms of acceptability, access, and adoption of innovations that drove decision making and intervention outcomes among health workers, religious leaders, and community members were identified to describe intervention implementation. Conclusions: Linking context and intervention components to the mechanisms they triggered helped explain the intervention outcomes, and more broadly how and for whom the intervention worked. Linking empirical mechanisms to constructs of implementation related theoretical frameworks provided a level of abstraction through which findings could be cumulated across time, space, and conditions by theorising middle-range mechanisms. Process-focused, theory based realist evaluation presents a useful framework when seeking to answer questions of what works, for whom, and under what circumstances. A central tenet of realist evaluation is that interventions work, or do not work, based upon the decisions that actors make in response to available resources. These decisions constitute mechanisms which are triggered in some contexts and not in others. We explored this relationship between context, mechanism, resources/intervention, and resulting oucome(s) using the context-mechanism-outcome (CMO) configuration of realist evaluation [14]. Where a CMO relates to a specific category of actor, a context-actor-mechanism-outcome (CAMO) is useful, and where the CMO relates to an intervention or component of an intervention, then a context-intervention-actor-mechanism-outcomes (CIAMO) configuration is better able to specify what works, for whom, why and where. We used two further theory based approaches alongside the CMO/CAMO/CIAMO heuristic in our interrogation of the uptake of FP within the integrated service delivery model. The first of these was the development of an initial programme theory (Fig. 1). Discussions were held with intervention designers and implementers in a workshop 15 months into the implementation of the intervention. This exercise focussed on implementers’ understanding of how the intervention and its components were expected to work, how they were currently perceived to be working, and how CMO/CAMO/CIAMO configurations could be used to determine and explain factors enabling or hindering the intervention. In constructing the initial programme theory we were able to elucidate factors that intervention designers and implementers perceived as the major drivers of the intervention [15]. These were adherence to clinical and counselling guidelines among HEWs, and community and religious leader support for FP. The initial programme theory also described potential barriers and mitigating factors to intervention implementation including the lack of tracking for referrals to higher level facilities for FP made by HEWs at health posts, long wait times at health posts in densely populated communities, and the lack of HEW training on implant removals. This initial programme theory was used in developing themes for interview guides and in identifying stakeholders for empirical interviews used to develop CMO/CAMO/CIAMOs. Initial programme theory Our second approach was to map our findings against the constructs of implementation related theoretical frameworks. Our reasons for using this approach hinge upon two methodological axes which are that the major challenge for evaluation is the cumulation of findings across time, space and conditions [14] and that for realist evaluation, such generalisation or transferability of findings occurs through abstraction. Abstraction is achieved through linking to theories including those from cognitive psychology, and behavioural science [16] and more recently also including a range of theories relating to behaviour change in health systems [17]. We assumed that the use of the constructs of implementation related frameworks rather than broader behavioural theories, for example, would provide more insight on mechanisms driving implementation outcomes and opportunites for cumulation of findings across evaluations. We considered several theories and concluded that our initial programme theory had best fit with constructs of three theoretical frameworks. These were acceptability of FP and MCMs by both health workers and community members [18], adoption and diffusion of innovations particularly with respect to health workers [19, 20], and access by women [21]. The constructs described by these frameworks align with the concept of mechanisms within realist evaluation as they describe factors that drive or lead to decision making among different actors. By mapping our findings against these constructs, we aimed to identify transferrable theories which could be used by implementers in similar contexts. Sekhon et al describe acceptability using seven constructs in the Theoretical Framework of Acceptability (TFA): affective attitude, burden, perceived effectiveness, ethicality, intervention coherence, opportunity costs, and self-efficacy [18]. Rogers’ diffusion (and adoption) of innovations framework consists of five contructs: relative advantage, compatibility, trialability, observability and complexity [20]. Finally, Penchansky and Thomas describe access using five constructs: availability, accessibility, accommodation, affordability and acceptability [21]. We linked the empirical mechanisms identified in the data to a construct(s) of the frameworks and critiqued the potential of this approach as an aid to cumulation of findings across studies. BGRS is one of nine regional states in Ethiopia. It is predominantly rural and consists of twenty woredas (districts) and 398 kebeles (smallest administrative unit) [22]. Assosa and Bambasi are part of BGRS and encompass 74 and 40 kebeles respectively. Within BGRS, there are five native Ethnic groups (Bertha, Gumuz, Mao, Komo and Shenash) and other dwellers (predominantly Oromo and Amhara). The region has relatively low levels of literacy (60.9% of women and 30.3% of men are illiterate), and high religiosity, with Islam and Orthodox Christianity being the predominant religions (51.3 and 28.2% respectively) [22, 23]. At the time of the 2016 Demographic Health Survey (DHS), full immunisation coverage in BGRS was 57.4% compared to nearly 89.2% in the capital of Addis Ababa and a national average of 38.3% [23]. The proportion of women aged 15 to 49 using any FP method in BGRS was 28.5% compared to 55.9% in Addis Ababa. Knowledge about FP was only slighty lower in BGRS compared to Addis Ababa (97.6% vs 100.0% respectively). There was also more male involvement in decision-making about FP in BGRS compared to Addis Ababa: 9.8% compared to 2.4% of women reported that their male partner was the main decision maker about FP, while 14.4% compared to 25% of women said they made the decisions about FP. However, the majority of respondents said that decision making was done jointly (75.9% vs 72.2%) [23]. In BGRS, the Health Extension Programme (HEP) plays a key role in health service delivery by providing primary health services at health posts in rural communities. It was adopted by the government of Ethiopia in 2003 to achieve universal health coverage among rural populations by 2009 [24]. The HEP is driven by model families, the Health Development Army (HDA) and HEWs [24]. Model families are male and female headed households that have received specific training on the HEP and that follow best practices for health and hygiene. They serve as role models within the community [24]. The HDA is an organised community based movement aimed at improving health sector capacity by engaging with communities and community leaders [24]. HEWs, commonly women, typically staff health posts in pairs and provide services such as community integrated management of childhood illness, immunisations, injectable contraceptives, implant insertions (but not removals), as well as basic curative services such as first aid and malaria treatment. HEWs are the lowest level health cadre in Ethiopia, usually with an education up to Grade 10, supplemented with a 1 year didactic and practical training in different health care packages. Among other responsibilities, HEWs conduct household visits and outreach activities and refer cases to health centers as needed. Semi-structured interviews (SSIs) with key stakeholders involved in the delivery and uptake of the intervention were conducted to identify contextual factors that triggered the mechanisms driving intervention outcomes. Purposive sampling was used for SSIs to select key stakeholders involved in, or with an interest in, the intervention including implementing partners, government officials, HEWs, and community leaders. Participants were selected to offer a range of perspectives and opinions of the intervention. HEWs selected were involved in the delivery of childhood immunisation and/or FP services and were from health posts where the intervention was perceived to be more, or less, well received based on project monitoring data. An interview and discussion guide for SSIs was developed specifically for this study and was informed by the initial programme theory. Broad themes encompassed workload, socio-cultural norms, and healthcare access, and questions specific to particular participant groups and specific aspects of the intervention within the study context were included. This ensured that key issues captured within the initial programme theory were included in the interviews. Please see supplementary file S1 for the interview guide that was used. CMOs developed with the implementers were also included in interview and discussion guides [25]. Interviews were conducted in October 2017 and March 2018 in Amharic and Afan Oromo by local research assistants with guidance and oversight from a London School of Hygiene & Tropical Medicine researcher and an implementation supervisor. All interviews were recorded, transcribed verbatim and then translated into English. Translated transcripts were imported into NVivo 11.2 for coding and analysis. Quotes were anonymized, but the type of respondent attributable to each quote was retained to aid analyses. Key themes were identified based on the interview guides and supported by quotes from interview transcripts. Coding and analysis was based on an intial framework of: interventions; actors; context; mechanisms; outcomes and initial CAMO and CIAMO configurations. These categories were populated inductively with themes and sub-themes as they were identified from the data. We developed CMO/CAMO/CIAMO configurations from the analysis of stakeholder interviews. Overarching contexts were identified as well as contextual and intervention triggers for specific mechanisms driving outcomes. The outcomes included in the CAMOs and CIAMOs were both outputs and outcomes. We then linked the identified mechanisms with constructs of the acceptability, adoption and diffusion of innovations, and access frameworks. Finally, we used the CMO/CAMO/CIAMO configurations to construct a revised programme theory.