Our approach to understanding the situation around unsafe abortion in Zambia is multi-disciplinary (health economics, policy analysis, sociology, anthropological demography, medicine, statistics). This necessitated a rigorously integrated multi-method strategy incorporating primary data collection of both qualitative and quantitative data.
Our research strategy is one of a comparative case study (women seeking safe abortion compared to women seeking post-abortion care) using mixed methods. It therefore balances the need to assemble detailed understandings of individual contexts and complexities with the necessity of producing results that can be generalised to other contexts.
The research process was designed to be iterative and interpretative, with a continuous interplay between collection and analysis. The study was based in a large teaching hospital. Three strands of primary data collection were carried out: a quantitative survey of women who had received hospital-based abortion-related services; qualitative in-depth interviews with women who had received hospital-based abortion-related services; and qualitative interviews with policymakers.
Our research instruments are included in the resources of the Consortium for Research on Unsafe Abortion in Africa. You can access them there, or just get in touch with us.
Our research was carried out in five stages, each building on the preceding one:
Stage One: Pilot Study
Prior to the main data collection phase, a pilot study was carried out. This enabled us to conduct a detailed audit of hospital case notes to inform sampling and interview guides; develop, translate and rigorously test the qualitative question guide and quantitative questionnaire; and refine interviewer skills.
Stage Two: Quantitative survey
This aimed to establish the distribution of out-of-pocket expenses for women and their households incurred using hospital-based safe abortion and PAC services.
Over 4 months all women that were identified as having undergone either a safe abortion or having received PAC were approached for inclusion in the quantitative survey. The survey questionnaire, that aimed to determine key characteristics of women that undergo either safe or unsafe abortion as well as the costs incurred, was based on questionnaires used elsewhere and produced in English (the language of the formal education system), Nyanja and Bemba (the most commonly used languages in Lusaka).
Treatment records were accessed, with permission, in order to validate individual reports of direct hospital costs. Interviews were conducted privately with women prior to discharge and after any prescriptions etc. had been given.
Analysis: Comparative (safe abortion versus PAC) quantitative analysis of costs (indirect and direct) for women and their households. Key socio-demographic determinants of the two groups will be studied using logistic regression. Analyses will make allowance for uncertainty in the estimates of costs and consequences, and non-market items (e.g.: opportunity costs) will be imputed using nationally-available secondary datasets in order to produce a range of estimates. Loss of productivity due unsafe abortion-related complications will be estimated also using disability adjusted life-years. Analysis of treatment records will allow for classification of the degree of seriousness of post-abortion complication, and will measure the economic impact of different levels of PAC. Moreover, access to treatment record will allow to perform a simple cost-effectiveness analysis of SA versus PAC. Data on unsafe abortion mortality will be imputed using national estimates.
Stage 3: In-depth qualitative interviews (QL1)
This aimed to establish the range of reasons why women sought abortion, and why they used or did not use safe abortion services, and to explore the social costs and benefits of their trajectories, and the policy implications.
Individual fine-grained narratives are not easily captured in a questionnaire-type survey, especially on such a sensitive area. Subsequently, in-depth qualitative interviews were conducted with a sub-sample of 40 women, drawn from the quantitative sample. This sub-sample was drawn to maximize heterogeneity in sociodemographic characteristics, including age, marital status, ethnic group, education, employment status, residence, in financial circumstances, and in clinical intervention and outcomes. Permission was sought to audio-record the interviews. Interviews were conducted in a private room within the hospital in the language of the interviewee’s choice.
Analysis: Data were organised using NVivo software. Preliminary content analysis of transcripts was carried out independently by the lead researchers and interviewers and then discussed in a team analysis workshop. This informed the coding basis for second cycle thematic analysis for a narrative summary on socio-economic issues surrounding abortion-related services.
Stage 4: Follow up in-depth qualitative interviews (QL2)
Stage 4 aimed to examine the ongoing socio-economic trajectories in a sub-group of interviewees.
Where permission was given, and follow-up was possible, we re-interviewed as many of women as possible at around 6 months after their discharge from hospital. The location of these re-interviews was determined by the women themselves, but where possible, linked to a return visit to the hospital or family planning clinic.
In order to build on existing rapport, these interviews were carried out by the same interviewer as the first interview (Stage 3). Again, interviews were in-depth and audio-recorded with permission. These follow-up interviews produced important data in a situation where very little indeed is known about what happens to women post-abortion in this context and about their coping strategies to deal with the socioeconomic costs over time.
Analysis: The analysis will use a longitudinal qualitative analysis summary matrix to capture change and continuity over time. Qualitative longitudinal research is emerging as an important tool for sociological investigation into change and transition, but is still rarely used in health research in low income country settings. It has yet to be employed specifically for examining post-abortion experiences.
Stage 5: Policymaker interviews and feedback
Policy maker interviews aimed to improve the quality of the research interpretation and findings, and the likelihood of policymaker impact.
Recorded in-depth interviews were carried out using a pre-tested interview guide, and were audio-recorded with permission and transcribed. Policymakers were recruited purposively to represent a range of (inter)national perspectives and approaches, including government, NGO and civil society.
Analysis: Thematic analysis of interviews using the policy triangle (context, content, actors, processes) framework will be conducted to provide retrospective and prospective profiles of and for policy. Findings will be integrated iteratively into the research findings and policy briefs.