How can engagement of adolescents in antenatal care be enhanced? Learning from the perspectives of young mothers in Ghana and Tanzania

This cross sectional qualitative study involved focus group discussions (FGD) with primiparous adolescents who had accessed ANC while pregnant with their first child. We were particularly interested in what these young women valued and understood about their ANC experience, as this would provide insights into what factors enabled them to seek care. Data collection occurred within the context of existing community-based programs implemented by two large international non-governmental organizations (NGOs): Plan International (in Ghana) and World Vision (in Tanzania).

Data was collected from within specific program implementation areas in Volta and Eastern regions of Ghana and the Singida region of Tanzania. The research team had no role in the selection of country sites; rather, the implementing NGOs made this decision based on a combination of maternal, newborn, child, and adolescent health indicators targeted by their programs.

The second sampling stage involved a random selection of adolescents from ANC rosters in each health centre. Eligible participants were screened for the following inclusion criteria: 1) primiparous female aged 15–19 years at the time of their ANC visit; 2) gave birth within the past 12 months (live birth or stillbirth); 3) attended ANC at a target health centre at least once during this pregnancy; and 4) resided in the catchment area of a target health centre. Screening was based on data obtained from health facility ANC records. Six to eight participants were randomly selected and invited to participate in a FGD by a community health worker or community mobilizer. In some cases, we found on the day of the FGD that selected participants had actually given birth more than 12 months prior (maximum 21 months), which may have been due to inaccurate facility records. This was true for five participants in Tanzania, and one participant in Ghana. Since these individuals had travelled to the interview site and made arrangements to participate, they were included in the study.

The consent documents and FGD guide were developed in English and translated into the local languages of each country/region (Swahili in Tanzania, and Twi, Ewe or Krobo in Ghana). All members of the local research teams were proficient in these languages. Both instruments were pre-tested prior to use, and language was adapted to each context to ensure phrasing was culturally appropriate and easily interpreted by study participants.

We developed a semi-structured FGD guide that covered a range of pre-determined topics related to the study aims (Additional file 1). The guide consisted of open-ended questions and facilitator probes to encourage elaboration by participants where appropriate. The guide also included several hypothetical case examples, which the facilitator read aloud to participants. The facilitator then asked participants to offer their opinions of the situation. This technique was particularly useful for more sensitive issues, as participants could share their ideas freely without relating opinions directly to personal experiences.

In both Ghana and Tanzania, a local FGD facilitator and research assistants (all females) were recruited and trained by the Canadian research team. To ensure a comfortable and familiar environment conducive to open discussion, community health workers/mobilizers identified “neutral” spaces (i.e. far from health facilities and ANC providers) that were easily accessible to participants. Village-based indoor spaces were chosen to protect the privacy of the participants and to facilitate digital recording of the conversations. Facilitators were trained to guide discussion in a flexible manner, allowing participants to steer the conversation towards relevant topics of interest. They were also trained to ask follow-up questions and use appropriate probing techniques to encourage rich explanations, and to avoid prompting or asking leading questions [14]. Note-takers took detailed written notes to document observations of group interactions, facial expressions, and other non-verbal behaviours to provide additional context to the analysis.

All digital FGD recordings were transcribed and translated from Swahili, Twi, Ewe and Krobo to English by the in-country FGD facilitators. Transcription and translation of the audio recordings occurred simultaneously. Participant conversations were transcribed verbatim without condensing or summarizing responses, and notes of any important context were embedded within the transcripts. One bilingual member of each in-country research team conducted quality and consistency checks of each transcript using a standard quality assurance protocol and assessment form (see Additional file 2). This stage was critical to ensure the accuracy of translated transcripts from the local language into English and reduce the chances of accidental errors, misinterpretations, or omissions.

The analysis team held debriefing meetings with facilitators shortly after each FGD to discuss each session’s content. This provided an opportunity to discuss any challenges that emerged during FGDs, to identify emergent themes in the data, and to discuss important contextual information that would help frame the analysis and interpretation. This iterative approach served as a critical first step in the data analysis process.

Data analysis was guided by the principles of conventional content analysis [15]. The aims of this analysis were twofold: first, to describe the phenomenon under study (adolescent girls’ experiences and decision-making processes with respect to ANC), and secondly to draw linkages to existing theories and/or conceptual models by comparing and contrasting results with similar studies. Researchers adopted an inductive approach, whereby themes and patterns emerge from the data, rather than a deductive approach where data analysis aims to test the validity of specific pre-existing ideas and theories [16]. An initial coding framework was developed using QSR International’s NVivo 10 software [17] based on broad themes identified through an initial review of the transcripts.

Two experienced qualitative researchers coded each transcript independently. The ‘primary coder’ analyzed the transcripts and made minor changes to the coding framework as new themes emerged from the data. A ‘secondary coder’ then reviewed the primary coder’s analysis and made additions or changes based on their impression of the data. Each coder kept a ‘coding diary’ within NVivo to document ideas about emerging themes, surprising findings, and other details useful for analyzing and interpreting the data. Regular meetings were held to discuss and resolve any discrepancies. These discussions guided interpretation of emergent findings, and enabled the team to draw connections within and across FGDs.

Once coding was finalized, the research team independently ran queries, read and re-read the coded transcripts to synthesize themes and to identify interesting outliers and surprising features of the data. The analysis team then summarized findings and questions in briefing notes that were consolidated and circulated to all research team members and FGD facilitators. Group calls were held to discuss all parties’ impressions of the findings, to validate interpretations of the data, and to fill in any gaps in the analysis. This process was designed to enhance the internal validity of the analysis [15].

The Health Belief Model (HBM) was selected as the principal framework to guide analysis. Developed in the 1950s, the HBM is one of the most commonly used frameworks for understanding individual health behaviours [18, 19]. The HBM (Fig. 1) includes a number of core constructs that help to explain why individuals take action to prevent or control illness/health conditions, including: one’s beliefs about susceptibility to a condition and severity of that condition, considerations of the benefits and barriers to a health-promoting behaviour, cues to action, and self-efficacy [19]. While the HBM model served as a useful starting point for framing emergent study findings, we learned during early stages of analysis that it was not a perfect fit for the data. Given the multitude of factors affecting ANC utilization that are beyond control of individual adolescents, we adapted the HBM to incorporate additional constructs from The Theory of Planned Behaviour [20] and ecological models of behaviour change such as Social Cognitive Theory [21]. This adapted model was used as a guide to organize emergent findings from FGD transcripts.

Across all FGDs, participants expressed a shared perspective that they were not ‘meant’ to be pregnant yet, and that adolescent women faced a higher risk of obstetric complications compared to older women, which put the fetus at risk of adverse birth outcomes. Perceptions of increased obstetrical risks associated with their ‘abnormal’ situation are captured in the model as “perceived susceptibility and severity”. These notions of risk inform the perceived threat associated with being pregnant at a young age. The HBM postulates that the higher perceived susceptibility and severity, the more likely the associated health-promoting behaviour will be practiced. Participants clearly identified ANC as the best practice for mitigating risks associated with pregnancy.

According to the model, when perceived benefits of attending ANC are thought to outweigh perceived barriers, the likelihood that the target behaviour will be performed increases. In both countries, participants generally understood the benefits of ANC, however, there was variability in participants’ expression of enthusiasm for the benefits. In general, Ghanaian adolescents were more vocal about the benefits of ANC, while Tanzanian participants were more subdued in their responses. Perceived barriers to ANC uptake could be categorized into two categories in both countries: tangible costs, such as cost of transport to the health facility, and psychosocial costs, such as fear of discrimination by peers (Table 3).

The motivation to confirm one’s pregnancy was described as a pivotal determinant of whether, and at what point, adolescents sought ANC. In Ghana, having a professional confirm a pregnancy was seen as a means of ensuring that pregnancy could not be denied by the ‘responsible’ male. Participants who sought early pregnancy confirmation at a health facility could then learn about and enroll in ANC at an earlier gestational age. On the other hand, participants also suggested that associated pregnancy confirmation might prevent or delay ANC uptake because of perceived potential consequences including being forced to leave school (in Tanzania), not being allowed to continue living in the family home (both countries), or fear of possible serious ramifications if they opted for pregnancy termination (as expressed in Ghana):

“If I want to abort the baby and I go for the session and later abort it maybe the midwife might notice me later and question me and if I tell her the truth she might call the police to arrest me.” [GHA-6, R3]

Each individual-level construct shapes the adolescent’s belief in her ability to successfully utilize ANC (i.e her self-efficacy) (Fig. 2). Participants spoke about overcoming their shyness and fears in order to attend ANC and of weighing perceived benefits and barriers of attending. They expressed a belief that some people are just less “shy” by nature (or have higher self-efficacy to access health services).

Participant narratives suggest that adolescent girls’ health-seeking decisions are highly influenced by sociocultural factors (summarized in Table 4). Many participants alluded to a strict cultural definition of what constitutes “proper” timing for motherhood and acknowledged that by this definition, they were too young to become mothers. The stigma associated with teenage pregnancy and early childbearing represented a powerful deterrent to attending ANC, and led to passivity among adolescents when they did attend:

“When we were schooling, our mothers advised us not to take boyfriends, but because we were disobedient and took boyfriends we ended up getting pregnant. So when we go to the hospital, we are shy.” [GHA-5, R8]

In both countries, service provider attitudes played an important role in reinforcing adolescents’ ‘normative beliefs’ regarding the most socio-culturally acceptable time to get pregnant.

Positive social referents are individuals with the ability to influence the individual’s likelihood of acting to help themselves. Participants described two main types of social referents: a) formal authority figures and b) role models with relevant experience. Formal authority figures included fathers, male partners and health care providers (particularly nurses) – if these individuals believed ANC was important, it had an impact on the participants’ beliefs. However, participants placed the most emphasis on the opinions of those who had gone through childbirth themselves. If a mother, sister or female in-law had a positive prior ANC experience, then adolescents were reportedly more likely to attend.

In addition to individual-level and sociocultural factors, adolescent participants discussed several health systems-level factors that influence ANC uptake. These issues are not under the control of the individual yet have a marked impact on her ANC experience, including ease of accessing care as well as the quality of the services received. As summarized in Table 5, four themes emerged under this level: health system policies; patient-provider relationships; human and material resources; and health centre practices.

In Ghana, some participants did not consider ANC to be adolescent “friendly” and believed that more could be done to tailor services to their needs. Attitudes of health care providers, or adolescents' anticipation of provider attitudes towards them, were particularly salient in participants' reflections. Participants often felt stigmatized by service providers for being school-aged and pregnant, and several participants noted feeling ostracized by older women attending ANC sessions. These experiences shaped their belief that ANC was not intended for them. One participant expressed:

Actually, at the age when you are supposed to give birth, we have not attained it yet…. So when I see that I am the only young girl there and not being attended to immediately, that is what will make me think and ask myself, “Ah! Why is it that the adults are being attended to and I am not?” I can ask that and someone will tell me that [ANC] is for adults and not for children" [GHA-5, R4]

In the adapted HBM (Fig. 2) we posit that an individual adolescent mother is defined by a set of underlying personal characteristics (white box on the left), and this individual engages in a decision-making process about a health-related behaviour – in this case, whether to access ANC. The decision-making process is dependent on individual-level decision-making, (represented by the blue box), but is also embedded within a set of household- and community-level variables (surrounding red box). The latter set of variables was added based on our review of existing literature and emergent findings in the present study. We also added the health service delivery component to the model (red box at the bottom) to represent constructs from ecological models of health service utilization. These variables correspond to the most distal health systems and policy-level factors, which are largely out of individuals’ control yet indirectly influence adolescents’ uptake of ANC.