“Our voices matter”: a before-after assessment of the effect of a community-participatory intervention to promote uptake of maternal and child health services in Kwale, Kenya

Slow progress on the Millennium Development Goals redirected focus towards addressing the Primary Health Care ideals laid out in the Alma Ata Declaration of 1978 [1‐6]. Key amongst these was the involvement of communities in the planning and implementation of health interventions targeting them. Such a participatory approach is essential for negotiated decision-making, shared commitment, social accountability, enhanced ownership and ultimately, assured sustainability of these interventions [7, 8]. For this reason, the World Health Organization currently recommends active community participation during the development and implementation of interventions aimed at specifically improving maternal and child health (MCH) outcomes [9].

For MCH services in particular, community participation is an important component of a human rights-based approach to promoting health and well-being [10]. It emphasizes the need to take into consideration patterns of individual behavior that could ultimately affect uptake of health services. It also seeks to address inherent structural determinants of health outside of the formal health system, including socio-cultural factors, which could ultimately influence the health-seeking behavior of individuals within a particular community [11‐15].

There is an increasing appreciation of the importance of addressing demand-side factors to improve uptake and utilization of MCH services [16, 17]. Addressing factors that influence demand for these services requires the development of interventions that aim to be not just effective, but also locally responsive and culturally appropriate [18, 19]. Such interventions are anchored on the understanding that consumers of health services, be they individuals or communities, ought to be partners in improving the delivery of these services and ultimately, enhancing health outcomes [20]. As a result, it is important that they participate in the design, planning and implementation of health interventions targeting them to ensure their buy-in and assure future sustainability.

The Dialogue Model (DM) is a structured, community-participatory intervention that is anchored on the critical pedagogy theory advocated by the Brazilian educator/philosopher Paulo Freire [21‐23]. It utilizes problem-posing techniques to stimulate societal reflection and raise critical awareness (conscientization) of the situation to provoke reflective action. Ultimately, it seeks to promote a deeper understanding of local factors that lead communities, households and individuals to seek to remain healthy so that external interventions can be effective [24].

This approach recognizes the central role that communities play in individual decision-making towards positive behaviour change and leverages on social capital i.e. the networks developed within communities that are intended to achieve common good based on trust, cooperation and reciprocity. Social capital assumes that individual behaviour change is driven by the desire for seeking validation and belonging within a wider community. As a result, an individual is likely to adopt behaviors that endear them to their wider social network [25, 26]. On the other hand, wider community actions are assumed to work in the best interests of individuals in that society. As such, these community actions are more likely to be adopted by individuals towards the wider common good.

A DM approach is especially relevant in parts of the developing world where uptake and utilization of critical health services has been sub-optimal [27]. In these settings, there is usually a power dynamic at play between health care workers (HCWs) on one hand, and consumers of health services on the other. The HCWs are typically considered dominant purveyors of health information while patients are expected to remain as passive recipients. As a result, HCWs often assume that what they have advised has been heard, understood, accepted and will be done. Unfortunately, this is never the case in many instances [28]. This approach therefore, provides consumers of health services with a stake in determining how their health should be managed. It has previously been used to promote various social change initiatives in health and development [29, 30].

Kwale County in coastal Kenya consists of residents who are predominantly rural (20% urbanization), Muslim (80%), from the Digo community (80%) with a very high poverty rate (75%). The 2014 Kenya Demographic and Health Survey (KDHS) had previously revealed a high total fertility rate of 4.7, low family planning (FP) utilization with a contraceptive prevalence rate (CPR) of 42% and high unmet need at 21% in this setting [31]. Additionally, only 49% of women reported having delivered in a health facility. Just like the national average, 96% of women reported having received antenatal care (ANC) during their last delivery although only ~ 60% of these reported having made the recommended ≥4 ANC visits. These findings reflect a slight improvement over time [32‐36].

With this background, the DM intervention was implemented in Matuga sub-county, Kwale with the aim of promoting uptake of select MCH services specifically, FP, ANC and facility-based delivery. It consisted of regular DM sessions in community units (CUs) linked to health facilities where an ongoing, multi-country, operational research project (the Missed Opportunities in Maternal and Infant (MOMI) health) was being implemented. The broad objective of the MOMI project was to reduce maternal and infant mortality through implementing a set of context-specific interventions combining facility and community-based strategies [37].

Between October 2013 and December 2015, a total of 570 DM sessions were held in 12 CUs associated with 10 intervention facilities in Matuga sub-county, Kwale. In the 2013/14 annual work plan, these facilities were estimated to have a total catchment population of 120,574 out of which 27,732 (23%) constituted WRA (Table 1). Of these, 1 was a district/county referral hospital (Level 4), 2 were health centers (Level 3) while 7 were dispensaries (Level 2). The 10 remaining facilities that did not receive the intervention comprised of 1 health center and 9 dispensaries and were estimated to serve a total catchment population of 60,966 out of which 14,021 (23%) were WRA.

Overall, the median (IQR) number of DM sessions held per facility per month was 2 (1–3) with the range varying per facility (Table 2). In 27 months, Kizibe and Mwaluphamba dispensaries had 19 (70%) and 14 (52%) months respectively, during which no DM sessions were held in their associated CUs. Vyongwani dispensary held at least 1 DM session during all the months of intervention implementation. The highest number of DM sessions held in 1 month was 20 in the 2 CUs associated with Tiwi Rural Health Training Centre while the least number of DM sessions held in a facility was 4 in the CU associated with Mwaluphamba (Fig. 1).

In the year before the intervention (October 2012–September 2013), the total number of WRA receiving FP commodities as reported by the 10 intervention facilities was 10,138 (Mean (M) = 1014, Standard Deviation (SD) = 381). The highest number was reported in Mkongani Model Health Centre at 1655 and the lowest in Magodzoni Dispensary at 541 (Fig. 2). Between January – December 2016, the total number of WRA receiving FP commodities was 11,628 (M = 1163, SD = 400). The highest number of WRA receiving FP commodities at this time point was reported in Mkongani at 1951 (18% increase) while the lowest was in Mazumalume dispensary at 669 which was a 10% decline. At 88%, Magodzoni dispensary reported the highest proportionate increase in number of WRA taking up FP services. Vyongwani dispensary reported the largest decline in FP uptake (19%) between these two time-points. Overall, FP uptake increased by 15% across the 10 intervention facilities.

Additionally, prior to the intervention (October 2012 – September 2013), the 10 intervention facilities reported a total of 4441 (M = 444, SD = 220) new ANC attendees, once again highest in Mkongani at 807 and lowest in Vyongwani at 106 (Fig. 2). Post-intervention implementation (January – December 2016), the total number of new ANC attendees reported by these facilities was slightly higher at 4511 (M = 451, SD = 264, a 2% increase. Just like at baseline, the highest and lowest numbers of new ANC attendees in 2016 were reported in Mkongani and Vyongwani at 999 and 153, respectively. The largest proportionate increase in uptake of ANC was reported in Matuga Dispensary (54%) while Magodzoni dispensary reported the largest decline (48%).

e total number of facility-based deliveries reported by the 10 intervention facilities at baseline was 1853 (M = 185, SD = 216). This number was highest for Kwale District Hospital at 571 and lowest for Mazumalume at 21. In 2016, the total number of facility-based deliveries had nearly doubled to 3229 (M = 323, SD = 384), again highest and lowest in Kwale and Mazumalume at 1165 and 35, respectively. Overall, uptake of facility-based deliveries increased by 74%. Ng’ombeni Dispensary reported the largest proportionate increase (530%) in facility-based deliveries while Mwaluphamba reported a 3% decline (Fig. 2).

Using a paired sample t-test, there were statistically significant increase in the number of WRA taking up FP services pre- (M = 1014; SD = 381) versus post- (M = 1163; SD = 400); t (18) = − 0.603, P = 0.04) as well as facility-based deliveries pre- (M = 185; SD = 216) versus post- (M = 323; SD = 384); t (14) = − 0.698, P = 0.03). The difference seen in number of new ANC attendance pre- (M = 444; SD = 220) versus post- (M = 451; SD = 264) was not statistically significant (t (18) = − 0.046, P = 0.43) [Table 3].

To understand the effect of concurrent interventions in the area on the outcomes of interest, similar comparisons pre- versus post-intervention implementation were conducted for the 10 remaining facilities that did not receive the intervention. Amongst these, there was an 8, 11 and 8% change in uptake of FP, ANC and facility-based deliveries, respectively (Table 4). However, these differences were not statistically significant for either of the outcomes, including uptake of FP services pre- (M = 720; SD = 259) versus post- (M = 776; SD = 396); t (18) = − 2.657, P = 0.33), new ANC attendance pre- (M = 120; SD = 86) versus post- (M = 134; SD = 87); t (18) = − 0.246, P = 0.27) as well as facility-based deliveries pre (M = 65; SD = 90) versus post- (M = 70; SD = 79); t (18) = − 0.097, P = 0.33).

In this before-after pragmatic study, we found a significant increase in the uptake of FP services and facility-based deliveries in facilities associated with CUs where we implemented a structured, community-participatory intervention targeted at improving uptake of these services in Kwale County, Kenya. We also found a slight increase in the uptake of ANC services, but this was not statistically significant. Our findings suggest that implementing a structured, community-participatory intervention could contribute to enhancing uptake of select MCH services in a rural Kenyan setting.

It is worth noting that the enhanced uptake of FP services and facility-based delivery reported in this study corresponds with high utilization rates for these MCH services that we have previously reported in this setting from findings of a household survey [40, 41]. In our previous work, we reported a high CPR of 54%, low unmet need for FP at 16% and a facility-based delivery rate of 78%. These findings also follow a general trend seen in recent KDH surveys that show an overall increase in uptake of FP and facility-based delivery in this setting [42‐44].

Our findings also reflect potential gains derived from decentralization of health services in Kenya. This decentralization has made available resources and devolved decision-making to a local system of governance that is better placed at identifying locally-responsive solutions to public health issues [45]. In this regard, the County Government of Kwale has made significant investments aimed at strengthening the local health system so as to create an enabling environment to ensure uptake and utilization of MCH services [46‐48]. While the bulk of these resources have gone into improving supply-side factors like putting up the necessary infrastructure, employing and enhancing the capacity of HCWs and improving the supply chain for medical commodities; a significant proportion has also been invested in building demand for health services through community-led initiatives.

The success of community-participatory approaches for MCH interventions hinges on the participatory model chosen [24, 49, 50]. Community-organized actions employ a model that relies on the intrinsic motivation of community members to develop and implement the interventions, in contrast with extrinsically-induced community participation that is driven primarily by external stakeholders. A key aspect of the DM intervention was that it was fully community-led without any active external influence from the project team, save for occasional supportive supervision visits. Prior to intervention implementation, we trained CHVs on how to effectively conduct DM sessions. After the training, we relied on them to obtain necessary buy-in from relevant community gatekeepers, organize and mobilize session participants, spearhead sessions and take responsibility for the final action plans developed.

Our findings need to be interpreted within the context of several limitations. Using a before-after pragmatic study design did not allow us to fully account for the potential confounding effect of concurrent interventions. Principally, just before we began implementing the DM sessions, the Government of Kenya enacted a policy of free maternity care in all public health facilities. Within several months of implementing the policy, the proportion of facility-based deliveries reported nationally had increased significantly [51]. We accounted for this limitation by conducting a similar comparison of outcomes in the remaining 10 non-intervention facilities which showed a slight, non-significant change in uptake of the same MCH services pre-versus post-intervention implementation.

In our case, non-intervention facilities served a small catchment population and were generally considered to be low-volume. For this reason, they were not prioritized in the initial setting up of CUs and selection/training of CHVs. Regardless, implementation of the free maternity care policy was rolled out countrywide without consideration of the facility workload. In fact, each facility was reimbursed a standard fee by the Kenyan government for every delivery reported and these resources were used to improve local service delivery. It was therefore, in the best interest of each facility to mobilize for additional deliveries. To further illustrate this point, while intervention facilities reported ~ 75% increase in uptake of facility-based deliveries pre-versus post-implementation, there was only a marginal increase in non-intervention facilities (8%), suggesting an effect over and above that of the free maternity policy. Finally, this policy did not target uptake of FP services and cannot account for the changes in FP uptake reported in the intervention facilities.

To our knowledge, apart from the national free maternity policy, there were no other interventions targeting health systems strengthening for MCH services in Matuga sub-county, Kwale during the period of DM implementation. Additionally, the County Government of Kwale did not establish any additional CUs associated with intervention and/or non-intervention facilities during this time. The scope of work for CHVs did not also change in the course of intervention implementation. As part of their community activities, CHVs typically provide referrals, and in some instances, physically escort WRA, including pregnant mothers, to the nearest facility within their jurisdiction.

The MOMI project also aimed to strengthen the capacity of county and sub-county health management teams to conduct supportive supervision and provided specific on-job training for HCWs on emergency obstetric care and FP services. There was also an element of service delivery expansion given that some of the intervention facilities were capacitated to better offer services that they were not capable of previously offering. Since they were implemented in the same intervention facilities, the effect of these additional interventions was most likely complementary to that of the DM intervention.

Additionally, we did not collect individual-level data on the MCH outcomes of interest to demonstrate individual-level behavior change. Data collection was at an aggregate facility level and as such, we were unable to verify whether the women reported as taking up these services had actually attended a DM session that influenced their choice. However, due to the proximity of CUs where we implemented the intervention to the health facility and the intensity of intervention implementation, it is less likely that the WRA who eventually accessed these services could have come from other communities apart from those associated with the intervention facilities. The rates of reporting into the DHIS-2 by both intervention and non-intervention facilities ranged from 90 to 100%. Additionally, the rates of in- and out-migration from this rural community are generally low and we can assume, with some level of certainty, that most WRA remained in their communities during the period of the study and likely interacted with CHVs implementing the DM intervention.

We also report absolute numbers of women receiving FP, ANC services and those who delivered in a health facility as the outcomes of interest. Since these numbers could have changed over time, the ideal outcome indicators would have been the proportion of women initiating FP, newly-attending ANC and delivering in a health facility out of the total number of WRA or number of pregnancies. Without accurate denominator information, a population-based survey would have been a better way of measuring this change which would also have allowed measurement of the level of exposure to the DM intervention. In this case, unlike the numerator data which was abstracted from the DHIS-2, the denominator in our case is an estimate derived from the annual workplan and would not accurately capture the information intended.

Finally, we sampled the villages and participants to DM sessions purposively, mainly at the convenience of the CHVs organizing the session. This approach may have introduced a selection bias in the manner that our intervention was implemented. Our approach was pragmatic given that we aimed to assess the effect of an intervention that relied on the presence of an active CU to be implemented. We also wanted to have an intervention that could be implemented within a real-world setting and represent the reality on the ground. It was therefore impractical to set up separate CUs that could not be maintained at the end of the project. We also aimed for the intervention being fully community-led with minimal external influence apart from occasional supportive supervision. It was informed by previous work that has demonstrated greater efficacy for participatory approach models that are fully community-led. Our study design accounted for the potential selection bias by comparing outcomes pre- versus post-intervention implementation in the same sample of facilities thereby reducing any inter-facility variability.

We found a significant increase in the uptake of FP services and facility-based deliveries in facilities associated with CUs where we implemented a structured, community-participatory intervention targeting to improve uptake of these services. These findings reflect a need for programs to include community participatory approaches as a key component as it enables them to implement interventions that are culturally-sensitive and locally-responsive. Ultimately, it is a useful approach in addressing demand-side factors for enhanced uptake of MCH services by providing communities with a stake in influencing their health outcomes.