PICK-CHAMP workshops were the first activities carried out in the sites, with each site completing all workshops in a timeframe of 2–3 months. Participants for community leader workshops were drawn from among religious leaders, local political leaders (including representatives of minority parties), traditional leaders such as village chiefs (where applicable), traditional medical practitioners, maternal and child health providers, trusted elders, and leaders of local nongovernmental organizations. Participants for community member workshops were selected to ensure representation across ethnic groups, genders, religion, age, marital status, educational level, and socioeconomic status. Some sites are implementing CHAMPS in communities where they have carried out other programs; in such instances, the sites relied on existing social networks to advertise the workshops and recruit participants, supplementing those reached through those networks with invitations to others in order to achieve the representation among participants described above. For sites implementing CHAMPS in new communities where no earlier work had been carried out, staff first met with local leaders to introduce the program and solicit their support. Through such meetings, those leaders then introduced staff to individuals and organizations in the local community; participants for both workshops were identified through these processes using snowball sampling methods. The workshop curriculum was written in English and local CHAMPS staff in each country trained in SBS translated the content into the predominant language(s) of the local communities.

Subsequent to the workshops, CHAMPS staff entered participants’ responses into databases and analyzed the findings to inform formative research and the start of community engagement activities. In 6 of 7 CHAMPS sites, these activities preceded the initiation of mortality surveillance. The South Africa site was an exception, as it had previously piloted MITS and continued carrying out the procedure for deaths that occurred in a hospital facility as the site transitioned into the CHAMPS network.

Demographic characteristics for each participant were gathered at registration. These characteristics included age, sex, marital status, religion, education, employment, and history of parenting (did the participant have any children; if so, what were their ages and had the participant experienced the death of any children). Participants were each assigned a unique identifying number and their responses were coded with that number so that responses could be analyzed by demographic variables without being tied to a participant’s name. In many instances, participants did not have the literacy level required to provide written responses to questions. This was the case in all community member workshops in every country; in these workshops participants offered verbal responses to questions raised in the exercises and CHAMPS staff members took notes of those responses, identifying the participant who offered a particular response through her or his unique identifier, which was printed on a nametag that participants wore. In some community leader workshops, participants wrote responses to questions on index cards that were labeled with their unique identifier. In other community leader workshops, verbal responses were recorded in the same way as in the community member workshops. Responses for each workshop were entered into a separate Access database file (Microsoft, Redmond, Washington) and each site provided the program office with those files. The responses were then exported into Excel files (Microsoft) for analysis.

Participants’ responses in each workshop were analyzed in light of the topics described above (see discussion on “Design and Rationale” in this section). In addition, responses for selected exercises were analyzed for alignment or tension with mortality or pregnancy surveillance (the “Alignment Tension Assessment Instrument” in the Supplementary Data notes how each exercise was analyzed). For community member workshops, the general alignment and tension score ranged from 0 to 4. For community leader workshops, the scores ranged from 0 to 8. Each score was combined, yielding a composite alignment score range from 0 to 12 and composite tension score of 0 to 12. Alignment and tension scores were also generated specifically for mortality surveillance and pregnancy surveillance. The range for the alignment and tension score for pregnancy surveillance was 0–12 and the range for MITS was 0–24.

The rationale behind the alignment/tension scoring is grounded in a number of assumptions informed by research in the field. Each of the exercises seeks to understand community perceptions related to pregnancy and childhood illness/death so that the acceptability of CHAMPS pregnancy and mortality surveillance activities as the first step of improving pregnancy and pediatric health outcomes could be assessed. Exercises 1–3 in both workshops assume that concepts of health and illness are generated in cultural contexts [12, 13]. While CHAMPS objectives are logical for practitioners and researchers coming from clinical and public health cultural frameworks informed by disciplines such as epidemiology and medicine, the cultural frameworks of local communities where CHAMPS is being implemented may not understand the purpose or logic of those same objectives or prioritize them in the same way. If participants could develop a number of different messages from their own framework that demonstrated the value or purpose of CHAMPS objectives, this would demonstrate an alignment with CHAMPS objectives. Similarly, if participants identified factors that impacted the course of pregnancy or child health that are readily addressed by clinical or public health programs, this would indicate an alignment between community perceptions and CHAMPS objectives.

Finally, alignment/tension scores for mortality surveillance carried more weight than those for pregnancy surveillance, and scores from community leaders carried more weight than those from community members. Greater weight was given to assessments of alignment/tension of mortality surveillance (including MITS) over pregnancy surveillance based on the literature that showed lower acceptance of autopsy procedures. Participants’ responses in the PICK-CHAMP workshops reinforced this finding; the levels of acceptability for mortality surveillance were far lower than those for pregnancy surveillance based on a number of different factors that are discussed below. This held true in every country. Finally, alignment/tension scores derived from community leaders’ responses were given greater weight than those derived from community members because of the greater extent of social relations and the greater number of internal and external ties of these leaders, each of which translated into greater social capital to influence community norms and perceptions [14, 15].

The Supplementary Data include the instrument for scoring alignment/tension. This instrument provides detailed information on the calculation of alignment/tension in the various exercises, the calculation of the alignment and tension scores for mortality and pregnancy surveillance, and the composite alignment/tension score for each site.

While the CHAMPS network currently consists of sites in 7 countries, findings from PICK-CHAMP workshops in 2 countries—Mali and Sierra Leone—are not included in this article because the activities were carried out there as stand-alone community engagement activities prior to ethical review and clearance of the broader SBS protocol. For the other 5 sites, PICK-CHAMP was included as part of SBS activities approved by institutional review boards in each country. In addition, the IRB at Emory University approved the SBS protocol that included PICK-CHAMP, and the US Centers for Disease Control and Prevention (CDC) Human Research Protection Office reviewed it and established reliance agreements with the IRBs at Emory and in each country where CDC staff is engaged [11].

The tension between mortality surveillance/MITS and religious beliefs and religious practices was revealed in exercise 6. In that exercise, participants were asked to indicate whether they agreed, disagreed, or were uncertain in regard to 20 statements dealing with maternal and child health/surveillance; 10 of those statements related specifically to mortality surveillance. The percentage of responses to all 10 statements helped determine each site’s overall alignment/tension score for mortality surveillance. Two of the 20 statements in that exercise specifically refer to religious beliefs and practices in the context of MITS. Participants’ responses to each of those 2 statements are listed in Table 3.

In both statements, an “agree” response was interpreted to indicate some level of tension with mortality surveillance and MITS. There was no common pattern to which religious tradition was correlated with participant responses that indicated tension. In Bangladesh, for example, a high percentage of Muslim participants provided responses to each statement that were in tensions with mortality surveillance (100% of responses for statement 9 and 71% of responses for statement 10); conversely, in Ethiopia 19% of Muslim participants offered a response to statement 9 in tension with mortality surveillance and only 1% of Muslim participants offered such a response to statement 10. Responses by Christian participants in South Africa to statement 9 indicated tension at a rate (58%) almost twice that of responses by Christian participants in Ethiopia (30%); the discrepancy was even greater for statement 10, with tension at a rate of 64% in South Africa and 23% in Ethiopia. These kinds of findings reinforce findings in anthropological and religious studies which demonstrate that religion is not monolithic but highly variable across cultural contexts and challenge any assumption that religious objections to health initiatives could be addressed with a generalized set of messages that referenced either the doctrines or sacred texts of a specific religion [16].

By themselves, these statements do not demonstrate that communities are irrevocably opposed to MITS (in fact, overall alignment with mortality surveillance was high in Bangladesh even though tension related to both of these statements was also high); rather, they serve as a reminder that many community members will not prioritize CHAMPS activities over religious beliefs/practices at the time of death. Therefore, to successfully implement CHAMPS in these communities, CHAMPS activities must not be seen as contrary to religious beliefs or interfere with such practices. As sites began to implement MITS, this was addressed in concrete ways by ensuring that the time for conducting MITS was minimized so as not to delay burial, and the distance the body was transported was minimized because of religious beliefs among some that transport represented an unnecessary action that delayed burial. A total of 8 statements specifically asked about mortality surveillance in the context of community priorities. Statements 9 and 10 (Table 3) examined the context of religion; the 6 others examined other topics. With the exception of Ethiopia, responses that indicated tension were far more frequent in the 2 statements that referenced religion in comparison to the 6 statements that do not refer to religion (the percentage of responses indicating tension were basically the same in Ethiopia regardless of the issue referenced). See Figure 2 for the percentages of statements indicating tension.

In exercise 4 of the community leaders workshops, participants were asked to name the ways that CHAMPS could support them in their efforts to improve maternal and child health in their communities. Participants’ responses were coded as “feasible” (CHAMPS could directly support such efforts at the present time), “not feasible” (the suggestion was beyond the scope and capacity of CHAMPS and would remain so over the duration of the program), or “other” (CHAMPS could not support such efforts at the present time but could possibly offer support in the future through partnerships that would be built over the course of the project). In Ethiopia, only 32% of the suggestions made by participants were actually feasible and in Mozambique, 48% were feasible. These scores indicate a lower alignment with CHAMPS objective and activities, as well as high expectations or needs in the community that CHAMPS cannot address. The percentages of feasible activities in the other 3 sites were higher: 57% in Bangladesh, 64% in Kenya, and 66% in South Africa.

For all sites, the majority of activities surrounding pregnancy were not in tension with CHAMPS activities. In exercise 5 of the community leaders’ workshops, participants were asked to name the things that occur in the community when a woman finds out she is pregnant; they were then asked to identify whether pregnancy surveillance would conflict with carrying out those activities. In 3 sites, the percentage of activities that conflicted was <5%; South Africa had the highest percentage of activities that conflicted with pregnancy surveillance, at 26%. Conflict in relation to mortality surveillance was higher in some sites. Participants were asked to identify the things that occur in the community when a child dies and then asked to identify whether mortality surveillance (including MITS) would conflict with carrying out these activities. While conflict in Bangladesh and Mozambique was quite low (at 3% in each country), the percentage of activities that conflicted with mortality surveillance in South Africa was 29% and in Ethiopia, the percentage was 60%.

There was also tension related to beliefs about the acceptability of the MITS procedure in some sites. In exercise 6 of the community members’ workshop, participants were asked to provide their opinion on the claim that removing tissue samples from a deceased child was wrong even if it helps to determine cause of death (exercise 6, statement 1). Twenty-three percent of community members in South Africa agreed that such a procedure was wrong. In the Kenyan and Mozambican sites, approximately 10% of community members expressed this belief. Another statement in the same exercise also demonstrated tensions related to mortality surveillance from other factors. Twenty-one percent of all participants expressed a general belief that no parent should ever consent to tissue removal after a child’s death. This level of disagreement among some members in the community might indicate that parents who consent would be stigmatized in some local communities by at least some of their neighbors. This number was highest in Ethiopia, with 38% of participants expressing this belief.

While the percentages cited above reflect the aggregate findings across all the PICK-CHAMP workshops conducted in that particular country site, there were instances of great discrepancies across different communities within the same site where workshops were conducted. This was most pronounced in South Africa. An analysis of participants’ responses to the following 5 statements addressing mortality surveillance in exercise 6 of the community members’ workshop demonstrates a broad range of responses:

In one site (Bramfischer), 86% of participant responses indicated tension with mortality surveillance, whereas only 18% of responses in Thulani indicated tension. Further, in some of the clusters within the South Africa catchment, participants’ responses were widely divergent with responses to some statements indicating tension while responses within the same site to other statements indicated alignment. Figure 3 shows the range of responses in each of the clusters across the catchment areas in South Africa.

The disparities across the communities in South Africa where PICK-CHAMP was carried out is even more striking when one considers that these communities comprise the CHAMPS surveillance site in the country and comprise a small geographic area of neighborhoods within the same urban subdistrict of Johannesburg.

PICK-CHAMP workshop findings on alignment and tension provided insight into the factors related to both acceptability (eg, in assessing participants’ responses to the following: “It is wrong to remove small samples of tissue from a child after she has died even if tests done on that tissue could tell you how she died”) and practicality (eg, in assessing participants’ responses to the following: “I would not agree to any kind of procedure if it meant that I could not bury my child according to my faith or tradition”). Understanding the drivers of acceptability and practicality informs the resources and logistics issues that must be addressed in relation to implementation.

PICK-CHAMP has also informed subsequent community engagement activities. PICK-CHAMPS findings have been used to inform approaches to 3 community engagement objectives: (1) identify and create partnerships with trusted individuals and organizations working on maternal and child health; (2) working with these partners, develop and sponsor feasible community activities in support of maternal and child health; and (3) demonstrate respect and appreciation for families who participate in CHAMPS activities, and for all children in the community through culturally appropriate activities and symbols. In exercise 6 in the community leaders’ workshop and exercise 7 in the community members’ workshop, participants named trusted organizations and individuals with whom CHAMPS could partner. CHAMPS SBS teams subsequently followed up with those who were named and established partnerships as feasible. In exercise 4 in the community leaders’ workshop, participants listed the ways that CHAMPS could support them as they carried out their roles and responsibilities in support of maternal and child health. Each of these responses was coded as feasible (CHAMPS could readily support such efforts), unfeasible (such efforts were beyond the capacity of CHAMPS to support), or other (supporting such efforts was not currently feasible but might be through partnership with local community networks). Those efforts labeled as “other” were prioritized in developing partnerships and the subsequent maternal and child health activities that CHAMPS is sponsoring or supporting.

Finally, the PICK-CHAMP findings that address religious and cultural norms helped inform the kinds of activities that could appropriately convey messages of respect and appreciation to families and communities. For example, some CHAMPS sites have an imam, pastor, or priest from the community available as part of the MITS team if parents request the religious leader’s involvement. In addition, one site also provides burial clothes for the child’s body after the procedure is completed. These acts convey 2 messages: (1) they demonstrate that the program respects the families’ religious beliefs and practices, and (2) they address critical issues of timing of the MITS procedure in a way that minimizes disruption to religious and community rituals.

Assessing disparities in alignment/tension across different communities within the same country allows CHAMPS SBS teams to prioritize community engagement activities in certain high-tension areas or to tailor the types of activities in certain communities to address suspected sources of tension based on PICK-CHAMP findings. For example, multiple initiatives are under way in Ethiopia, the country with the highest reported tension. Efforts have been made to establish a community advisory board consisting of key stakeholders. Community leaders, including religious leaders, teachers, and health professionals, attended a training of trainers initiative to provide health education in local communities; the initiative included modules on CHAMPS objectives and activities as well as feasible approaches for improved child health by addressing nutrition and water, sanitation, and hygiene. These trainers are now carrying out workshops on these topics alongside local health professionals to build trust with the community. Finally, the site is deploying “theatre for development,” a well-established method for community-based participatory action, to engage communities in new platforms to discuss CHAMPS activities, especially MITS.