The care, stimulation and nutrition of children from 0-2 in Malawi—Perspectives from caregivers; “Who’s holding the baby?”

Introduction Universal access to quality early childhood development and care is a fundamental part of Sustainable Development Goal 4.2.1. Research from diverse settings, including that in low and middle income countries, now demonstrates the positive impact that interventions to promote play, stimulation, early communication and responsiveness can have, not just on child development, but on long term gains in education and economic growth. International agencies have recently produced the Nurturing Care Framework for Early Child Development in order to promote this and to encourage countries to move the focus from child survival to child thrival. Research on the best methods of integration of these programmes into present care practices, particularly in African settings is still very limited. Methods We used qualitative methods to gain an understanding of care practices (play, developmental stimulation, early communication and responsive feeding) for children 0–2 years. We conducted 18 focus groups (FGDs), six PFGDs (Participatory Research focus groups), 18 in depth interviews (IDIs) and 20 observations with caregivers in rural and urban settings of Malawi. We used a topic guide, audio-recorded the FGDs and IDIS and transcribed them in Chichewa or Yao and then into English. We coded data using an inductive approach to thematic analysis. We placed the data within a framework with the emerging major and minor themes. We conducted quality assurance for translations and coding frameworks through cross comparison of data and used respondent validation to check our results. Results Malawian caregivers see children’s play as a consequence of good health and wellbeing, less an interaction that a parent undertakes to promote wellbeing and learning. Non-verbal communication and responsiveness with infants is clearly present and caregivers have many one-on-one interactions with young infants. Furthermore, many caregivers have good knowledge of feeding recommendations but cannot always follow these due to constraints on money. When children become mobile (toddlers), play is an activity done more often between children or with other caregivers such as siblings or grandparents. Community members consider that caring for children from 0–2 years is a woman’s domain. Despite this, both men and women acknowledge the importance of maternal wellbeing in enabling good care of children. The present socioeconomic situation of most families in our study means that income generation and food security come first. Many mothers spend most of their time managing the responsibilities of daily living and have limited time to dedicate to children’s play, responsiveness and communication with their children. Discussion Programmes promoted as part of the Nurturing Care Framework which provide advice on developmental stimulation, play, early communication and responsive feeding should ensure that topics within these programmes are culturally appropriate for the setting. Furthermore, programmes must not be an added burden to parents but be supportive to parents managing many responsibilities of daily living. Multi-sectorial approaches where both men and women are provided with knowledge but are also supported through programmes which address family finances, safe spaces for children whilst families are working, and family mental health and relationships may enable programmes to work more effectively. As our aim was to understand perceptions, subjective experiences and meanings of care practices in Malawi, we undertook a multi layered mixed methods qualitative approach to gain a richer data set, improve trustworthiness and dependability of the data and to enable triangulation. Our epistemological position is one of interpretive subjectivism based on real world phenomena and an understanding that the world does not exist independently of our knowledge of it. Within this, we also come from a position of relativism–a view that reality is subjective and differs from person to person[27]. Using different data collection methods enabled us to understand whether data from different perspectives including that which is “said” and “perceived” (focus groups and interviews) fit a similar coding framework and reality to that which was “observed” (participant observations). We therefore aimed to explore perceptions, experiences, understandings and observed reality of participants within different contexts to provide us with a wider perspective on the situation [28] which we did not feel we could specify in advance. Table 1 outlines these different methodologies used as well as the numbers sampled. We conducted 18 focus group discussions (FGDs) with mothers, fathers, grandparents to generate open discussion on community views and norms and to understand and hear arguments as well as mutual views [29]. We also undertook 18 in depth interviews (IDIs) with a similar sampling framework. These were conducted in a narrative format and provided an opportunity to gain detailed individual perspectives and discussion of more personal issues [30]. We conducted six participatory research focus groups (PFGDs) for a wider discussion of the specific advice provided within the WHO/UNICEF Care for Child Development (CCD) programme. Within these sessions, we conducted the CCD training and recorded feedback and discussion of training materials. We conducted these groups firstly as they helped to shape what we needed in the interviews and focus groups. We then conducted interviews and focus groups concurrently. We conducted the observational work lastly as the interviews and focus groups did not provide enough real time evidence on what caregivers did with their children. Key: FGD = Focus group discussion, IDI = In-depth interview () = total number of participants in FGD or participatory focus groups We purposively sampled from different groups of individuals in diverse areas in Malawi to gain perspectives of those from different religions, income generating approaches, gender and age. During October 2012, health surveillance assistants conducted a census of all families with children under the age of two years in the areas studied. From this sample of 420 possible child-carer dyads, HSAs then randomly sampled (through use of computer-generated randomisation) participants for the FDGs, IDIs and participatory focus groups (PFGDs) from each of the areas as described below. We utilised a purposive sampling framework in that we were keen to gain views from different sexes, ages of participants, locations. We then sampled mothers, fathers and grandparents from three regions of Mangochi through HSAs. Areas included; Malombe (a fishing village), Katema (an agricultural village with a majority Christian population) and Nankhumba (a mixed fishing and agricultural) of Mangochi district which also vary in the prominent religious groups present within the community (Christian/Muslim). We also purposively sampled mothers, fathers and grandparents from Chilomoni (urban Blantyre) who varied in their distance from the road and how urbane/semi urbane they were (Table 1 sampling matrix). We undertook sensitisation and information sessions prior to sampling from any community with community leaders. We provided information leaflets in Chichewa to leaders and to Health Surveillance Assistants in the areas. These groups then provided leaflets to families of children of 0–2 prior to recruitment. We conducted focus groups (FGDs), in depth interviews (IDIs), observations and PAR groups between September 2012 and May 2013. KC and FC conducted all FGDs and IDIs in either Yao (Mangochi district) or Chichewa (Blantyre district) or in some cases in English with health workers who preferred to speak in English. KC and FC were both young and female and had both had over 2 years’ experience in qualitative research interviewing techniques in mental health. KC had a diploma level qualification and FC had a BSc. Both had been and were working as qualitative research associates in the area of maternal mental health and child protection. Neither had a relationship with either caregivers or professionals prior to study commencement and had no previous experience with the Care for Child Development training (CCD) or early child development research. We created a topic guide for FGDs and IDIs that we piloted on four occasions prior to the start of the study to check that we were enabling the most pertinent information to be gathered (S1 Text: Topic guides for FGDs and IDIs CCD Malawi). FGDs took place in a central location in each village (school, home, church or health centre) and took approximately one and a half hours and IDIs took place in the caregiver’s home and took approximately forty minutes. An additional note-keeper and supportive research assistant was present from the local area to enable ease of transcribing and remembering who was who when speaking during the focus groups. These were all audio recorded, transcribed and translated. We compared transcriber quality of translation by comparing a handful of the same transcripts done twice by both transcribers to enable consistent translation. The team discussed any inconsistencies. We recorded observations as field notes and typed these up after each session. We used materials from the CCD package during the PFGD group sessions in order to provoke discussion and gain feedback about the materials. We audio recorded this feedback and discussions. Observations: We used observational methods to generate a description of the daily life of parents and caregivers caring for their children under the age of 2 years in rural and urban settings in Malawi. Research assistants used a semi-structured interval-sampling approach to generate descriptions of interactions[31]. The research assistant recorded the time, context and participants present and then at one minute intervals (S2 Guide for structured observations), the key interactions between the child and caregivers at one minute intervals were written down followed by a written interpretation note every minute and then a review note every ten minutes[32]. We went through a series of steps in conducting a thematic content analysis[33, 34]. Our first step was to familiarise ourselves with the data by reading and reviewing the transcripts, taking notes and sharing ideas and views on the data. At this stage, we generated some ideas for codes. We continued with data collection until we felt we gained saturation of themes within the data and no new themes were emerging. We coded transcripts as we went along to enable us to see when we reached saturation. Data (written field notes and verbatim) from the observations was added to the database after the focus groups and interviews and the coded using the same coding framework. We conducted thematic content analysis by using the software package NVIVO 10. Data from interviews, focus groups and observations all fed into the thematic framework that we identified. This enabled us to triangulate the data with more certainty. The next step was to create a codebook. To do this, members of the team coded a number of transcripts from the focus groups, and interviews separately to create a list of themes and sub-themes that were emerging from the data. We created the coding framework iteratively from the data to prevent selectivity in the use of data and aimed to allow semantics within the data to drive codes. We coded sections of text several times if there were a number of different semantics within the text and coded as many potential patterns and themes as possible at this stage. The codes were shared and were discussed and modified within the research team (JP, FC, KC) prior to a final decision about our coding framework and coding tree. Three coders (MG, SM and MK) coded all data once a final codebook had been created. We then crosschecked between coders to make sure we (MG, SM and MK) were coding similarly. We then searched for potential themes within our listed codes and sub-codes and collated the data within these identified themes. We used mind maps and lists to create piles of themes. We compared data within themes and across sources (rural/urban, female/male, grandparents/parents) at analysis. We compared data from IDIs and FGDs with observational data to identify areas of similarity in themes and areas where there was less cross-over. We then devised a set of candidate major and minor themes which we defined and refined once we placed within Bronfenbrenner’s framework. We have utilised the Ecological Systems Theory conceptualised by Bronfenbrenner as a suitable framework through which to examine the themes which emerged from our data [35]. Bronfenbrenner’s theory or child development is commonly illustrated as a nested system of ‘environments’ presented as concentric circles (Fig 1) [36]. These interrelated systems include the ‘microsystem’–the interpersonal relationship and physical setting directly experienced by the child (family, peer group, school and neighbourhood), the ‘mesosystem’–connections between these, the ‘exosystem’–connections between settings that do not contain but may directly influence the child; and the ‘macrosystem’ which is described as the overarching pattern and fits all of the above [37]. For adaptation of the Care for Child Development package (see Table 2), we distilled the themes and subthemes from the framework in a way that could also be utilised for package development for our setting. An example of this might be the major themes of “child protection” and “maternal health and well-being”. After analysis of the data, we reviewed emerging themes and made consensus decisions about which of these could be best practically integrated into a 12 module-training package. As a developmental paediatrician, MG is a strong supporter of interventions to improve early child development of children and was conscious of this bias during the study period. She is a British Caucasian researcher who is aware that her cultural bias may have affected the analysis of the research. JP and KM also both had strong positive views of interventions to improve early child development due to his previous studies and work in public health in the field of early child development and nutrition. KC, FC, SM and MK had no experience of previous interventions and had important roles as neutral interviewers and members of the analysis team. We have utilised the COREQ guidelines to enable us a framework to check our reporting of the methodology and results of this study [38] (S1 Table: Supplementary Table 1 ISSM_COREQ_Checklist). We gained prior informed consent from the community leaders as well as written consent from each individual who took part prior to any data collection. We provided each individual with an information leaflet relating to the study, which we read out to him or her again prior to consent. They then signed (or fingerprint stamped with a witness) an individual informed consent form prior to being enrolled in the study. We did not give written transcripts back to the participants. We did gain some respondent feedback at the end of the study through a series of four feedback groups where we explained the emerging themes from our research. These sessions included professionals and caregivers in each location where the research took place. During these sessions we asked participants to provide their views, feelings and experiences in light of our results. We took notes on the comments received during the groups and reviewed our themes and sub-themes in light of these sessions. They also enabled us to feel more confident that our research results were credible with members of the community. This study was approved by the University of Liverpool Research Ethics Committee (RETH000536) as well as the College of Medicine Research Ethics Committee in Malawi (P.03/12/1193).