Introduction

A major challenge of the post-2015 Global Development Agenda is the disparity of mortality data availability and quality between high-income and low and middle-income countries (LMICs), due to poor reliability of available tools for cause of death (CoD) assessment in the latter settings. Currently, CoD determination in LMICs is often overlooked or reliant on verbal autopsies, a method subject to a high degree of inconsistencies and misdiagnosis [1,2]. To overcome this problem, minimally invasive autopsy (MIA) methods, lately referred to as minimally invasive tissue sampling (MITS), have been recently developed and validated [3–7] and currently being implemented as a tool for mortality surveillance purposes [8–10].

MITS is a post-mortem procedure consisting of obtaining samples from key organs and body fluids, using biopsy needles, which are subsequently analysed using histopathological and microbiological methods. [5,6,11]. Increasing interest in implementing MITS in LMICs is based mainly on its higher feasibility compared to the complete diagnostic autopsy (CDA) and the need to deploy better methods for CoD investigation to supersede the suboptimal currently utilized ones. A key aspect of the feasibility differential between the two methods is that, with one exception so far studied [12], 7 studies suggest that MITS have potentially higher consent rates compared to the CDA [13–19]. It is assumed that, as MITS hardly affects the physical integrity of the body, the impact on families’ state of mind is lower, and healthcare providers’ discomfort in interacting with relatives to request a post-mortem examination reduces [16,20,21].

Knowledge about consenting to MITS methods was initially drawn from assessments of acceptability of post-mortem procedures in general, conducted predominantly in high-income countries [13,22–24]. Studies conducted in LMICs, or among immigrant populations residing in high-income countries, suggested that MITS methods might be acceptable even in settings where post-mortem procedures were assumed to be unfeasible, such as rural, limited-resource areas with no previous community awareness of post-mortem procedures, or among communities with religious values unfavourable to the concept of invasive post-mortem manipulation including the removal of body organs [13,15,19,25]. However, an important limitation of most of these studies is that they rely on hypothetical case scenarios based on respondents´ assessment on whether and why they would agree to MITS, even though they had never experienced it. One study conducted with parents of both dead and living children, had an opportunity to explore acceptability among parents of deceased children who had undergone a MITS, although the vast majority of study participants had not experienced a MITS, therefore the conclusions were drawn on a vast majority of hypothetical understanding of MITS [26]. It is unclear whether the findings obtained from that approach, which captures intentions to accept an intervention, could be extrapolated to actual scenarios during the course of the real-life implementation of the procedure [27]. Several reports indicated that barriers to implementation of interventions on different public health fields were minimized when recommendations drawn from acceptability assessments were followed [28–32]. However, most of these studies did not assess acceptability during the actual intervention, and thus it is unclear how the predicted barriers translated into de facto impediments to acceptability.

Acceptability encloses two distinct concepts, namely, anticipated acceptability, resulting from assessments prior to the intervention implementation, and experienced acceptability, which is assessed during the implementation of the intervention [27]. To our knowledge there are no published studies confronting and questioning the components of each of these two acceptability concepts, the factors contributing to each of them and the transferability of one into the other.

The aim of this study was to compare anticipated versus experienced acceptability of MITS procedure among relatives of deceased children in southern Mozambique, taking into account the specific components of acceptability and factors explaining it, in order to provide evidence for context-specific best practices in the implementation of mortality surveillance systems requiring the post-mortem manipulation of corpses to establish the CoD.

Methods

Study procedures

Anticipated acceptability variables derived from a secondary analysis of data from the ethnographic component of the CaDMIA study on local attitudes and perceptions related to death and willingness to know the cause of death described in detail elsewhere [15,40]. From September 2013 to April 2015, deaths in the previous 1–30 days within or outside the health facilities (HF) were notified by HF staff, DHSS field workers, or community informants to social behavioural sciences (SBS) research team members, who then invited the closest relative present at the time of death to a one-on-one semi-structured interview. If consent was granted, the interviews took place at the health-facility (waiting area at the morgue, the paediatric ward or the maternity ward) or at the respondent’s home or workplace, depending on their preference. The interview followed a guide of open-ended questions encouraging respondents to talk about whether and why they would consider or not the performance of MITS (explained to participants as the use of a semi-automated needle to obtain very small pieces of each organ, without the need to open up the body, and analysing such samples in the lab for CoD investigation) on their deceased child, should MITS be available at the HF (S1 Appendix).

Experienced acceptability variables were drawn from a mixed-methods study on the feasibility to conduct mortality surveillance using MITS by examining factors influencing their acceptability, practicality and implementation [26]. From December 2016 to December 2017, all deaths of U5 children occurring at the District hospital in the previous 24 hours were notified by HF staff to a mortality surveillance staff, who then approached the relatives to request consent to perform MITS on the deceased child. All eligible cases were considered for inclusion in the experienced acceptability assessment. One in every three consecutive cases was selected for observation of the informed consent (IC) process. The non-participant observation technique was conducted by a SBS team member who kept field notes to register actions and reactions of healthcare providers, the project staff and relatives of the deceased child throughout approach to families, consent request, the MITS process and body release. Registration of these events was based on an observation guide (S2 Appendix). If the observer arrived at the enrolment site after the interaction with the relatives was over, the observer conducted a guided informal conversation with the project’s staff who had conducted the IC process. Additionally, for all cases, information was extracted from clinical records, which included socio-demographic data, consent outcome and reasons for not performing MITS. A sub-sample of ten families who had refused MITS were sequentially visited at home and invited for a follow-up semi-structured interview to be conducted at home or a location where the respondent would feel comfortable. During the same period, a matched number of relatives who had consented to MITS were interviewed following the same procedure (S2 and S3 Appendices). All interviews were conducted at the respondent’s home.

Fig 1 provides a schematic representation of the different study components.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Schematic representation of the different study components.

https://doi.org/10.1371/journal.pone.0259621.g001

Ethical considerations

Studies contributing to this analysis received ethical approval from the Manhiça Health Research Centre’s IRB (ref: CIBS_CISM10/13), the Clinical Research Ethics Committee of the Hospital Clinic of Barcelona (ref: 2013/8676) and was deemed exempt from further IRB review and approval by the Emory University IRB (ref: IRB00086895). For both anticipated and experienced acceptability interviews, written informed consent was obtained from the family members of the deceased children. During the experienced acceptability study, consent to use observation data for research purposes was part of the overall consent for the CHAMPS study. Relatives who refused MITS did provide consent to be included in all other CHAMPS surveillance activities (and were referred to as “non-MITS cases”). Additional consent was requested to conduct follow-up interviews with selected HF staff and relatives.

Results

Participants’ characteristics

Of the 35 cases enrolled in the anticipated acceptability study, 15 were relatives of deceased children (4 stillbirths, 4 neonates, 3 infants and 4 children aged 12 to 60 months of age) [15].

During the experienced acceptability assessment, 114 cases of U5 children deaths were relayed to the surveillance team; these cases consisted of 32 stillbirths, 45 early neonatal deaths (up to 7 days old), 3 late neonatal deaths (from 8 to 28 days old), 14 infants between 29 days and 11 months, and 20 children between 12 and 59 months. Overall, 35 cases were directly observed, 11 were recorded through a follow-up informal conversation with MITS consent team once the interaction with the relatives was over, and 68 were registered through case-record forms (CRF) supported by consultations with the MITS consent team.

Additionally, 10 relatives who had refused MITS and an equal number of relatives who had accepted MITS were specifically interviewed.

The characteristics of participants and cases are summarized in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Cases characteristics (family members and deceased children) contributing to the anticipated and experienced acceptability analysis.

https://doi.org/10.1371/journal.pone.0259621.t001

Anticipated acceptability of MITS

Of the 15 relatives of recently deceased children who were asked whether they would hypothetically accept a MITS on their dead child, 14 (93%) replied affirmatively. Only one respondent (the mother of a neonate who had died within 30 to 40 days prior to the interview) kept silent, suggesting a potential refusal, although she freely discussed perceived advantages and disadvantages of the procedure later on in the interview.

When relatives were asked about reasons for MITS acceptance, albeit theoretically (Table 2), one relative considered that a CoD investigation was mandatory. All but one respondent were motivated by their interest in clarifying the CoD. One of the underlying drivers, expressed by the majority, was the strong demand as a parent to know what caused the death, indicated by some resentfulness for not having received explanations from the health facility about their loss. In some cases the driver was the need to answer to funeral-goers who would eventually query on the CoD, which is a common topic of conversation while passing the condolences to the family. Parents felt uncomfortable when unable to provide an answer, especially if hospital help had been sought.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Parents and guardians accounts of anticipated drivers to accept the performance of MITS on their deceased children.

https://doi.org/10.1371/journal.pone.0259621.t002

For some relatives MITS was seen as an answer to prevent future disease or death, be it at community level to “save others”, as put by one mother, or at family level. Thus, immediate action could be taken to prevent additional fatal events linked to the disease that killed the child in the case of contagious diseases, or to equip mothers with knowledge and skills to protect their future babies from dying, particularly in the case of sudden deaths occurring in apparently healthy babies.

In the view of some respondents the procedure might help to address commonly raised speculations about perceived negative actions leading to children’s death, such as health professionals’ negligence or witchcraft.

Although the majority of relatives were in favour of MITS, participants identified some barriers that would endanger the execution of this approach (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Parents and guardians accounts on anticipated barriers to the performance of MITS on their deceased children.

https://doi.org/10.1371/journal.pone.0259621.t003

Most of such barriers regarded clashes between religious, as well as traditional norms regarding children’s deaths and the requisites to perform MITS. There was the concern that MITS could be challenging or even impossible to perform in small children, who must be buried immediately “while the body is still hot”. The perceptions of these respondents were intrinsically linked with the requirement to maintain secrecy about the occurrence of stillbirths and early neonatal deaths. Knowledge of these deaths is restricted to the immediate family and close acquaintances, and female elder relatives who are responsible for the burial ceremonies decide the post-mortem actions to be taken without the involvement of the child´s parents. A few participants mentioned complex decision-making processes that would interfere with the final outcome of consent to MITS. A mother mentioned that although she would favour the performance of MITS on her dead child, her husband, who was absent at the time of the interview, would be the ultimate decision-maker. Other mothers revealed that they had no knowledge regarding what would be decided regarding their child’s funeral (or any other post-mortem procedure), leaving that decision up to the church leaders and the child´s grandmothers. One participant noted that priority given to CoD determination over cause of illness investigation among the living was unnatural, therefore a potential barrier for parents to consent. In this regard, a woman went further to say that “not only the results would not bring the child back to life but also they would evoke further pain.” Fig 2 illustrates the above findings on factors explaining the willingness to agree to MITS.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Dimensions of anticipated acceptance of MITS.

Components, drivers and mediating factors explaining MITS acceptance among relatives of deceased children before the implementation of MITS.

https://doi.org/10.1371/journal.pone.0259621.g002

Experienced acceptability of MITS

As shown in Fig 3, out of the 114 eligible CHAMPS cases, the project’s staff requested MITS’s consent to the families of 101 deceased children. In 13 cases, the relatives were not approached by the project’s staff.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. MITS Consent flow diagram.

Expected sequence of events from informed consent attempt, request and granting or refusal to MITS procedure on deceased children (0–5 Y) in Manhiça District Hospital, Mozambique (2017).

https://doi.org/10.1371/journal.pone.0259621.g003

The most common reason for not approaching relatives was the staff’s lack of courage to request consent, given the relatives’ verbalized disappointment with the clinical management of the child while alive. The observation team captured some evidence of those cases where relatives had attributed the CoD to the health-facility’s mismanagement or neglect leading to staff’s discomfort in requesting consent for MITS due to fears of an angry response by relatives (Table 4). Timing between death notification to parents and body release or discarding was also reason for not approaching relatives, as either the families or the health-facility staff took the child’s body too quickly while the mortality surveillance team was still organizing the logistics to proceed with the IC and MITS. The third reason captured by observations was the mortality surveillance staff’s perception that the mother’s state of mind was not suitable to go through the consent process. One of them was unconscious and the other one was in extreme distress. Finally, in one case the study team realised that the child’s body was degraded and was therefore not eligible to undergo MITS.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Barriers for health professionals to approach family members to request consent to MITS.

https://doi.org/10.1371/journal.pone.0259621.t004

Of the 101 families that were approached, 83 agreed to the procedure on their deceased child (of which 27 were stillbirths, 28 early neonates, 2 late neonates, 10 infants, 16 older children), while 18 families refused it (of which, 12 were early neonates, 1 late neonate, 2 infants and 3 older children). Thus the overall acceptability rate was 82% (100% for stillbirths, 70% for early neonates, 66% for late neonates, 83% for infants and 84% for older children).

The subsample of interviewed family members who had agreed to MITS alluded mainly to their desire to know the CoD, adding that in their opinion MITS seemed to be the only way to meet their expectations. One parent considered that MITS could help resolve their suspicion of mismanagement of the child while under intensive care. Finally, one family expected that MITS would help end the recurrence of miscarriages or stillbirths in the family. Details of these views are presented in Table 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Reasons for accepting a MITS as reported by parents/caretakers of deceased children (n = 10).

https://doi.org/10.1371/journal.pone.0259621.t005

As illustrated by some relatives, an important factor mediating acceptance was the presence of the main decision maker during IC; another facilitator for consent was the fact that they had heard about the intervention before (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Dimensions of experienced acceptance of MITS.

Components, drivers and mediating factors explaining MITS acceptance among relatives of deceased children during the implementation of MITS.

https://doi.org/10.1371/journal.pone.0259621.g004

Among the 18 families who categorically refused MITS, observations and informal conversations with HF staff revealed that a recurrent reason was the urgency to release the body in time for a burial (Table 6). Undergoing a MITS would potentially lead to postponing the burial to the following day, incurring extra costs to host the mourners, who only depart from the grieving house after the burial. On one hand, the observations revealed the urgency to transport the body while still fresh to enable the caretaker (the grandmother) to piggyback the body as if still alive to be allowed on a public transport van at no extra cost. Otherwise, transporting the body after it had reached post-mortem rigidity would imply arrangements for specific, often unaffordable, transportation services for corpses. On the other hand, complex decision-making processes were involved. For instance, one mother considered the decision had to be taken by the father, although she did not attempt to contact him for reasons unknown to the team. In some cases, although the main care-taker (mostly the mother but in one case a minor, older sister) showed interest in MITS, other relatives (namely fathers, aunts and grandparents) refused the procedure.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Reasons for refusing MITS based on direct observation and health staff accounts of refusal cases (n = 18).

https://doi.org/10.1371/journal.pone.0259621.t006

Observations also captured cases of relatives changing their minds after weighing the initially bearable predicted waiting time against the experienced waiting time between the death notification and the outset of MITS performance. In such cases, the relatives who had initially waited ended up giving up because the health-facility was not ready for the procedure to be undertaken as immediately as they initially thought it would be.

In one interview with a father of a deceased child, he explained that he refused the procedure because excessive manipulation of dead bodies, including the removal of body parts, went against his traditional values.

Taking into account the entire cascade, from the identification of eligible cases, followed by missed opportunities to request consent, and also those who during the process changed their minds from acceptance to refusal and finally those who underwent MITS, the proportion of cases in which MITS was performed accounted for 73% of the eligible cases.

Discussion

To our knowledge, this is the first study that applies the direct observation of experienced acceptability of relatives to perform a MITS on their deceased child in order to triangulate with findings from other classic qualitative enquiry techniques. We found a high level of anticipated acceptability with 93% of relatives willing to accept the performance of the post-mortem procedure on their deceased child when a hypothetical situation was presented to them. Although slightly lower, the level of experienced acceptability was also high (83%). Anticipated acceptability findings agree with studies that assessed willingness to accept MITS in other settings [12,15,19,25].

The small sample size in the anticipated acceptability study (CaDMIA) could also have influenced the relatively lower probability of capturing more hypothetical refusals. This limitation was also considered by authors of an earlier study in Kenya [12].

Notwithstanding, in theoretical scenarios acceptability can be an abstract concept, not entirely equivalent to what the actual outcome of consent would be. First, the outcome of an IC process can only be comprehensively assessed in real-case scenarios. Second, consenting constitutes only one of several components of acceptability; some of the other components were identified when experienced acceptability was examined in this study (e.g., agreeing to the procedure but not being able to comply to it, opposing views among relatives, and health facility staff’s hesitance to approach them) [27,41]. These findings add more value to the existing knowledge on the components of acceptability of MITS, which had previously mostly been assessed through questioning relatives of deceased individuals, health professionals and community members in hypothetical case scenarios on the willingness to accept the procedure [15,19].

Recently, there have been attempts to assess the acceptability concept based on theoretical insights, mostly related to health care interventions that focus on treatment [27]. However, due to the life-saving potential of treatment, the knowledge generated by such analyses cannot be transferrable to the understanding of the acceptability of a post-mortem intervention.

The drivers of both anticipated and experienced acceptability were oriented to the expectation that MITS can explain the unknown, ease some suspicions, and reassure relatives with the hope of avoiding further misfortune [41,42]. Importantly, although lightly captured, power relationships between the health care provider and the relatives—which was expressed by the perception that MITS was mandatory—may play a role in acceptability (particularly in anticipated acceptability) [32,42–44]; this may suggest that in the context of consent on deaths occurring in the community, whereby a power shift from the health-facility staff to the relatives and community authorities may be observed, the acceptability to the procedure might be lower.

In alignment with previous studies, the theoretical scenarios highlighted barriers and facilitators to MITS acceptability, which were heavily attached to cultural norms and values (namely, the importance to bury the child as early as possible and under utmost secrecy–especially for neonates, the low value attached to learning about the CoD with no apparent advantage to the family, and incompatible family-level decision making models) [15,19]. The real-case scenarios further revealed practical concerns originating both at family and health-facility level that accounted for the outcome of consent, such as the negative impact of MITS on the timing, financial, and transportation arrangements in preparation for the funeral ceremonies; the high dependence of MITS on health-facility logistics and staff preparedness; and further complexities of the decision-making situation (particularly regarding mother’s low decision-making power). The latter factor resonates with results obtained in a CHAMPS site in South Africa [41]. Of note, the comparison between the hypothetical and real case scenarios of the decision-making process regarding the performance of MITS on small children allowed the problem to be viewed from different angles of the family dynamics during the decision making process following the death of a child. Specifically, the anticipated acceptability study highlighted that upon the death of small children (neonates and infants) grandmothers take the lead in the traditional post-mortem events, sometimes not even involving the mother. Drawing from the experienced acceptability study, it was suggested that the addition of an external element (such as a request to perform MITS) to the routine post-mortem events, adds some complexity to the process, which justifies the need not only to involve but more importantly for the final decision to be contingent on the fathers’ pronouncement, in their capacity of a legitimate legal representative of the child and the protagonist of power within the family in the context of patriarchal societies [45]. The requirement to sign papers (informed consent), a procedure most grandmothers will not be familiar or comfortable with, adds to this dependency on the child’s father.

While the high proportion of consenting relatives supports the suggestion that MITS is a feasible approach to provide cause of death information, it should be noted that it is still an invasive procedure performed on a dead body of a child, which interferes with the above-discussed sensitive, logistic, traditional and religious requirements and values attached to a death event. Health professionals are not detached from such values either. A previous study on MITS acceptability suggests that, since it is less invasive compared to CDA, MITS might be associated with reduced discomfort for healthcare providers when requesting it to grieving relatives [16]. However, the real-case scenario acceptability analysis revealed important assumptions and reservations from health professionals, which impeded their approaching some of the families. This points to the importance of paying attention to healthcare providers’ self-efficacy and disposition to interact with grieving families regardless of how invasive the procedure is. An earlier identified scenario when MITS are not necessarily preferred over CDAs should also be considered with regards to healthcare professionals preparedness to discuss post-mortem examination with families [12].

Of note, the expectations on MITS were overwhelmingly high in this setting, despite the limitation that in a small proportion of cases MITS does not provide conclusive results [4,5,9]. This limitation did not seem to have been fully understood by the relatives. It would be important to discern the extent to which a better awareness of this limitation would influence acceptability. Further studies of relatives’ perceptions when they receive inconclusive results would shed more light onto this component of acceptability. The value and meaning attached to the CoD results themselves needs further examination. In this study, while the fear of breaching the confidentiality was captured from the only participant who anticipated a MITS refusal, parents who experienced the procedure expected to discuss results with funeral-goers. This is an unexpected finding, which should be reflected upon with caution especially in case of endemic stigma-prone diseases [15]. This also raises the question on how to manage this expectation over the results, since the funeral happens within days, at the longest, while cause of death results are only delivered months after the death. It is also true that in this setting, besides the vigil and the funeral, the mourning process is marked by a number of remembrance ceremonies, both religious and traditional, spanning across months and even years after the death event. Therefore it is thought that grieving relatives can still digest and discuss the cause of death results that are eventually delivered several weeks to months after the MITS is conducted. However, parents who consented to MITS did not allude to this possibility, probably because they had already experienced the entire MITS process including the lengthy turn-around of results feedback, discarding the overly high expectation of the possibility of feeding back the MITS diagnosis to mourners. This can in turn constitute a barrier to MITS acceptability.

This study highlights that irrespective of the religious background, the timing factor is a significant concern to post-mortem examinations targeting small children in this rural African setting. The relevance of timing is reflected in the observation that even those relatives who had formally consented did change their minds when experiencing delays due to MITS. Issues of timing and secrecy attached to child death events, particularly regarding stillbirths and early neonatal deaths, are critical for implementation of MITS, affecting not only the IC act but also all other MITS related processes, such as case identification and notification, body keeping at and release from health-facilities, transportation and feeding back results to relatives, all of which should be time- and privacy- sensitive.

Lack of health-facility preparedness was an important factor for not accepting MITS. Although this limitation is expected, given the fragile health system, it is likely that relatives may consider this inability to conduct MITS immediately as a lack of respect towards grieving families, and that they perceive it as unaligned with implementers claim on the contribution of MITS to the greater good [46]. Thus, programs that rely on health system infrastructure should contribute to the strengthening and maintenance of such infrastructure so that services can be adequately delivered in parallel to meeting programs’ goals. This is especially important for mortality surveillance platforms, which should be presented as initiatives contributing to mortality reduction.

The study findings also call for the need to continue to build up the knowledge base on the practical experiences of MITS implementation across different contexts. To further advance this knowledge, the recognition that acceptability of a MITS transcends a binary concept captured through relatives’ disposition or not to the procedure is crucial. Conduction of qualitative analyses to further deconstruct the concept of MITS acceptability into meaningful and measurable components is needed [47].

These findings can guide the development of more robust instruments to assess MITS acceptability in advance of and during its implementation. For example, the incorporation of more tangible questions that include variables representing practical concerns to all people involved in the procedures, from relatives to health care providers. Furthermore, they may contribute to improve the preparedness of study staff, health facilities, and health care providers to undertake the procedure in the most respectful manner possible to the family.

Conclusions

There were high levels of relatives’ anticipated and experienced acceptability to MITS on their deceased child driven by interest in knowing the CoD. The study identified two components of anticipated acceptability (willingness or not willingness to accept) and four components of experienced acceptability (accepting, inopportunity to voice acceptance due to hesitation by health professionals, relatives changing their minds from accepting to not accepting, and refusing). Secrecy of the event, confidentiality of the results and complex decision-making processes for consent were barriers to anticipated acceptability, while family- and health facility-level logistics and practical aspects were barriers to experienced acceptability. Besides personal, relational, social and cultural barriers to MITS, health-system and logistical impediments should be considered before the procedure’s implementation. Studies on anticipated MITS acceptability provide insights that cannot fully inform implementation. Improvement in the tools used for both anticipated and experienced acceptability assessments are needed. Finally, health programs should ensure maximum alignment between their objectives and the individual-, family- and community-level values and priorities; similarly, health-facilities must also be prepared before the implementation of sensitive procedures like MITS, and therefore ethical considerations should include prerequisites for new programs embedded on existing, often weak, health system platforms to make balanced investments between their scientific objectives and the immediate impact on health services quality towards child mortality reduction.

Supporting information

Acknowledgments

The authors would like to thank the families who were engaged in the study, the community leaders and the Manhiça District Hospital’s collaborators as gatekeepers for the activities to take place, the observers (Mr. Atanásio Matusse and Ms. Dulce Machava), interviewer (Mrs. Yolanda Uamusse), transcriber (Mr. Elso Ofumane) and the colleagues responsible for data entry and management (Mr. Bento Nhacale and Mr. Arlindo Malheia).