Evidence – qualitative and quantitative – was generated from interviews with adolescents aged 10-19 years in Ethiopia, Malawi, and Zambia to understand how adolescent abortion-related care-seeking differs across a range of socio-legal national contexts. Our comparative study design includes countries with varying levels of restriction on access to abortion: Ethiopia (abortion is legal and services implemented); Zambia (legal, complex services with numerous barriers to implementations and provision of information); and Malawi (legally highly restricted). Most adolescents (98%) in Ethiopia obtained a medically safe abortion, with most adolescents (64%) in Zambia and almost all adolescents (94%) in Malawi obtaining a less medically safe abortion. A total of 313 facility-based interviews were carried out with adolescents aged 10-19 in 2018/19 in Ethiopia (n=99), Malawi (n=104), and Zambia (n=110). Adolescents were seeking public sector care for either safe abortion or post-abortion care for complications from an abortion initiated elsewhere. Adolescent recruitment was initiated by a study-trained senior nurse, who identified and invited eligible participants to participate in the study upon their readiness for discharge. Our research assistants (RAs) were all females in their twenties or early thirties and were recruited after the completion of intensive (two weeks) training from the project team that included role-playing and pilot interviews. We completed paid training for more RAs than the project required; performance during training and piloting were explicitly part of our RA recruitment process. Interviews with adolescents were conducted in a private setting in each facility by RAs fluent in all major local languages. Informed consent was obtained from adolescents aged 18 and above, while for those under 18, consent was sought from an accompanying parent or guardian with the respondent's assent. Unaccompanied respondents under 18 were considered emancipated minors, and their independent consent was obtained. Each adolescent in our study had one interview. A set of approaches were used to maximize the likelihood of disclosure of abortion-related behaviors, and to elicit the details, many of which were multiple attempts to end the pregnancy interwoven into a complex trajectory over weeks or months. For interviews where consent or assent was granted, they were recorded and subsequently transcribed (qualitative) and data entered (quantitative) by the RAs. In each interview, there were normally two RAs: one RA (Interviewer 1) conducted the interview in a conversational style to put the participant at ease and facilitate the narrative flow, whilst a second RA (Interviewer 2) completed the datasheet seated to the side of or behind Interviewer 1, so as not to influence or distract her from the conversation. Interviewer 2 was always positioned to be visible to the respondent [i.e.: not behind her]. As Interviewer 1 conducted the interview using a conversational style to follow and probe the responses, Interviewer 2 completed the datasheet (see below). Interviewer 1 did not write and was able to maintain eye contact and react to the adolescent’s body language, facial expressions, and emotions. Before closing the interview, Interviewer 1 re-introduced Interviewer 2 who would ask supplementary questions building on the conversation she had listened to. During training RAs had learnt and internalized the overall logic of the research project, so that when they were Interviewer 1 they knew the topics and questions that they needed to probe for. Our research project wanted to understand – in detail –adolescent abortion-related care-seeking. We wanted to generate quantitative and qualitative evidence quickly to minimize burden on respondents who may be feeling uneasy or uncomfortable after receiving abortion-related care. We needed a tool that would collect data quickly and facilitate disclosure. We refined a datasheet approach that had previously been used in a study of abortion care-seeking in Zambia (Coast and Murray 2016). Due to the number of closed questions, the datasheet shows superficial similarities to a survey questionnaire. Critically, however, the order of the questions was not pre-determined – RAs were able to decide which questions to ask depending on the conversation. Interviewer 2 added the information to the data sheet by following the flow of the conversation rather than a pre-determined order. The size of the datasheet (A3) allowed for notes on the tone or content of the discussion to be written on the margins or for detailed notes if consent to record was not given. RAs understood that “messiness” on the datasheet was not a problem; what mattered was capturing the complexity of adolescents’ experiences. Datasheets for each country are available. To minimize the burden on respondents and maximize likelihood of completion, we did not want to conduct separate qualitative and quantitative interviews. A survey instrument demanding specific question order response patterns would be inappropriate for understanding the complexity of abortion-related care-seeking. We needed an instrument that would allow for the simultaneous collection of quantitative and qualitative evidence. Narratives about abortion care-seeking are not linear, and we needed an approach that would allow us to respond to the adolescent’s narrative whilst ensuring that quantitative evidence was generated to allow for aggregation and comparison. In many survey instruments on abortion-related care-seeking there is little detailed questioning about concurrent and/or unsuccessful abortion attempts, and nearly all instruments have a linear chronological design. Our approach was purposely nonchronological and nonlinear to respond to adolescents’ narratives. Our datasheet allowed us to capture non-linear and concurrent trajectories to be recorded in detail in a way that would be impossible with a structured chronological questionnaire. Our prior experience meant that we knew that attempting to gain disclosure of abortion attempt(s) and care-seeking is difficult – for very good reason. Adolescents are afraid of admitting to behaviors that are stigmatized, might be potentially criminalized, and have punitive outcomes (e.g.: police involvement, abuse from health professionals or from parents or partners). We developed a flipchart in each setting as a way of helping to identify local (perceived) abortifacients (e.g.: medications, toxic substances). We produced a flipchart booklet of locally produced photographs of all the ways in which adolescents might attempt or have an abortion in each context. To generate the photographs, during training we had an intensive and iterative group discussion amongst the RAs and country team members – to generate a long list of all the possible abortion methods, irrespective of efficacy, that anyone had ever heard about. These discussions served a useful secondary purpose of eliciting, debating, and discussing RAs’ beliefs and understandings about abortion methods. The RAs then took photographs of each of these methods in community settings and created a laminated flipchart booklet. The photographs were context-specific, and all three countries had different pictures. The flipchart was a low cost and low technology option that yielded positive results in terms of facilitating the disclosure of abortion – and abortion methods – by adolescents. The flipchart helped with recall because many adolescents make multiple, sequential, and concurrent attempts to try and terminate a pregnancy. It helped to identify pharmaceuticals that adolescents did not know the names of. Finally, the flipchart helped respondents to define the steps, components, and timelines of their abortion trajectory. Our use of a flipchart – locally produced in each context – served multiple purposes. First, it served to normalize adolescent’s behaviours in relation to the stigmatized issue of abortion; when an adolescent sees and points at – but does not necessarily speak about – something that she used or did, it communicates to her that others have also done this. Second, it helped to identify more accurately – especially in relation to pharmaceuticals – what medication adolescents had used. Finally, it was a time-effective way of eliciting information; RAs did not have to verbally describe a range of methods and wait for a response. We know that adolescents often lack accurate information on abortion and the flipchart was an important way of enabling and adding detail to abortion disclosure.A substantial proportion (35%) of adolescent pregnancies in sub-Saharan Africa are unintended; 10-19 year olds account for 25% of all abortions in Africa, higher than in any other world region. Unsafe abortion is a major public health problem, not only in countries where access to safe abortion services is highly restricted legally. Even where it is available legally, access, provision and knowledge of services can be inadequate. Adolescents are more likely to have an unsafe abortion and to experience complications (including death) of unsafe abortion compared to older women, even in settings where safe abortion is available. Addressing the age distribution of unsafe abortion, focusing on adolescents, is important in identifying and tackling barriers to health-care access and delivery. The needs of adolescents for contraception and abortion services are substantial; the implications for adolescents of not having access to these services are life-long. Research on unsafe abortion has shown the magnitude and the substantial health systems costs of unsafe abortion and unmet need for contraception, but significant evidence gaps remain for adolescents. Adolescent sexual and reproductive health services (ASRHS) aim to increase access to, and quality of, appropriate healthcare for adolescents, and respond to evidence that shows the significant barriers encountered by adolescents in accessing and receiving sexual and reproductive health services. Each of the three study countries has high-level commitment to the development and provision of YFS, although the rate of roll-out is variable. Systematic research on the implementation of contraception and abortion services for adolescents is absent. The evidence that does exist presents few opportunities for drawing out generalisable understandings across settings. In all three study countries, the issue of unsafe abortion among adolescents is a policy priority. There is acknowledgement that adolescents represent a group that is at high(er) risk of an unwanted pregnancy, linked to lower levels of access to effective contraceptive services. The team composition, with established and active relationships with policymakers, including Ministries of Health, means that this proposal has been informed and developed with in-country knowledge of what evidence and research is needed in order to improve health systems' capacity to deliver ASRHS effectively. There is an urgent need to understand how contraception and abortion care services can be provided for, and accessed by, adolescents. We focus on abortion care as one key component of the service provision needed to prevent unwanted adolescent pregnancies, which includes access to contraception, including emergency contraception. The evidence we generate will have relevance for other sub-Saharan African countries with a wide range of legal contexts, and will be able to inform current debates about abortion legality, policy and service change. This research aims to understand adolescent trajectories to facility-based abortion care-seeking, including post-abortion care, to establish how the implementation of contraception and abortion services for adolescents in sub-Saharan Africa can be improved. It does this by comparing adolescent experiences in public facilities in three countries: Ethiopia, Malawi and Zambia. These three countries represent a range of abortion legal frameworks, from least restrictive (Ethiopia) to most restrictive (Malawi).

Interviews were collected with adolescents aged 10-19 years in Ethiopia, Malawi, and Zambia to understand how adolescent abortion-related care-seeking differs across a range of socio-legal national contexts