Altogether, this study will further our knowledge on abortion safety by allowing the collection of relatively representative data on a large number of recent induced abortions in two contrasted sites in sub-Saharan Africa, one in a rural and small-town setting in West Africa and one in the slums of a capital city in East Africa. Sociodemographic and safety characteristics of abortions across the two sites may vary widely, thus enlarging the scarce empirical-base on this topic in the region.
Also, this study will advance our knowledge of the two network-based methods applied and refined to measure abortion safety. While ATPR has now been used in a diversity of settings with success (although using ad hoc corrections procedures), this method has failed puzzlingly in a few other contexts. There is a need to explore the different biases of the ATPR in depth and systematically, and in relation with various study contexts. This project offers the opportunity to do so. To achieve this aim, it will use a mixed-method formative study, the generalized NSUM model, varying points of comparison (self-report, RDS and national estimates), as well internal validation tests involving non-sensitive quantities. The current network-generator question based on secret sharing also still raises some questions worth exploring further. Finally, the level of stigma, which may vary across sites, may also affect ATPR implementation.
The study should also bring light on whether, where and how RDS can be used to collect quantitative data on abortion safety, and whether it is possible to reach equilibrium when constituting samples of abortion seekers. The pre-test is expected to bring much light on this issue; the multiplex RDS provides an interesting avenue to expand on current applications and could be well suited to conducting RDS in cases the practice of interest does not bring individuals directly in contact.
In the reminder of the discussion, we will cover the ethical challenges raised by our study and its limitations.
The ethical challenges involved in RDS on abortion are numerous, a topic which we discuss in-depth in this section. The main benefit of this study is that it will generate information on abortion safety in contexts where access to abortion services is restricted: first, using new network-based approaches, we will measure the degree of abortion safety in the two sites. Second, through this experiment, we hope to fine-tune two alternative network-based approaches, so that population-based information on the safety of abortion can be in the future more easily collected in restrictive contexts. To date, information on abortion safety is lacking in contexts where access to safe abortion services is restricted. Such information is needed at the national, regional and global level to help governments improve the entire spectrum of reproductive health services in their efforts to reduce maternal morbidity and mortality, from improving access to safe abortion services where pertinent, to providing harm reduction programs and quality post abortion care programs for illegal abortions. An additional benefit of the study for women lies in their exposure to a discourse that destigmatizes abortion. This study moreover provides the respondents with access to professional help in case they want to prevent unplanned pregnancies: we will indeed arrange for referral for contraceptive counselling for women who ask it in each site.
We considered the following risks for women who participate in the study. First, women will be asked to disclose a practice that is illegal in many cases. To protect women, we will ensure absolute privacy and confidentiality during the interviews, and collect absolutely no identifying information at any point during the study (see details below). Second, we will ask women to disclose a stigmatized practice in the course of this study; the participation to the study itself could thus be stigmatizing if women's entourage are aware of the object of the study. Therefore, while the objectives and content of the study will be clearly explained to participating women once in a private and confidential place, the study will be presented to all others as a women's health survey. Moreover, the results of the study itself will contribute to de-stigmatize this practice, by showing how common unsafe abortions are. Regarding these two risks (illegality and stigma), we should underline that RDS, one of the network-based approaches applied in this study to obtain information on the safety of abortion, has been applied more than 100 times before for other stigmatizing, illegal and health threatening practices, such as men having sex with men or prostitution in HIV research, or sexual violence and illicit drug use [31,32,33]. These studies have been successfully completed, which is an indicator of respondents feeling safe and able to respond openly. A final potential risk is that women could feel psychologically distressed during the interviews when evoking their abortions. However, the literature clearly shows that psychological distress is not a usual outcome of abortions; in the opposite, abortions usually bring relief to women who are pregnant at an inappropriate time . Moreover, in this study, we will only collect data on the conditions of the abortions as they relate to their safety; we will not explore the reasons for pregnancy termination or the unintended pregnancy. If the woman appears to the interviewer that she requires additional support or if she herself asks for more information or seeks support, we will arrange for referral to psychological counselling in each site. For each site, there is an established collaboration with the local hospital so that referrals can be made smoothly. As the study will be announced to all except the participants as a "women health" study, the abortion of the women will not be revealed in the referral process.
This study will protect the women participating to the study against the aforementioned risks through a number of measures, which will be identical across the two study sites, except in case of additional requirements by national ethics committees. The measures are the following:
For the ATPR/self-report survey, a random sample will be drawn from the HDSS sampling frame of residents: this mode of selection will be explained to the potential participant. For participants to in-depth interviews as well as RDS participants, interviewers will first work with a few key informants (for example: professionals in social work or community health workers). They will be asked whether they can locate women who had abortions. These key informants will be trained along RDS procedures, to ensure they approach these women in a confidential manner, explain the survey or in-depth interview, and give them a contact where the women can reach a fieldworker if interested in the study. For this project, we will reach out to key informants connected to varying population groups, from schoolgirls, young people out of school, to married women, young mothers, older reproductive age women and migrant women. Care will also be taken in working only with key-informants with a trusted record of collaboration; the longstanding implementation of the research teams in the study areas ensures that a variety of trustworthy key informants can be mobilized.
Once contacted by potential seeds or IDI respondent, interviewers will agree with them on a secure place to meet. The interviewer proposes a confidential place. Once the potential respondents come to the secure interview place, interviewers apply the informed consent and screening procedures; in case the woman accepts to participate and meets the eligibility criteria, she will be enrolled. The eligibility criteria for IDI will be age, residence. The eligibility for RDS will be age, residence and having had an abortion in the last 3 years (or being close to somebody who had an abortion for the multiplex RDS). To screen for eligibility for the RDS, a large spectrum of reproductive health behaviours and events will be investigated, so that women who end up not being selected will not know what the study is about. Women who accept to participate in the first RDS interviews ("seeds") will be requested at the end of the interview to propose the study to other potential participants; interested potential participants will contact the study team, who will then apply eligibility and consent procedures. Willing participants will be trained to give out referral coupons to potential participants, to keep track of women who were approached but are not interested to participate and asked to come back for a second interview (to document the recruitment process).
Inclusion of adolescents
Since the study population includes individual from 15 years onwards, the project engages adolescents. As adolescents are minor, the standard practice in both Burkina Faso and Kenya is to obtain an assent form for them and consent form from their parents. But according to the fact that this topic is sensitive allowances can be made by local ethics committee to permit the adolescent to give the consent. In the event that these allowances are not obtained, the study will be limited to the 18 to 49 year-olds. Privacy and confidentiality for this group will be reinforced.
Informed consent will be sought from all participants (in-depth interviews, ATPR/self-report survey, RDS survey). The consent form will be read to participants and explained. The informed consent form will explain that the study is on the help of network members in case of family or reproductive problems for IDI and ATPR participants and on the safety of abortion for RDS participant. It will be made clear that the woman can stop the study at any time. The benefits (no direct benefit, see compensations below) and potential harm of the study (i.e. talk about private and potentially stressing reproductive events) will be made clear to participants. The measures taken by the study to minimize risks to the participants will be explained: how the study ensures full confidentiality and avoids accidental disclosure will be described. For participants who agree to take part in the study, informed consent will be documented as a verbal consent on a tablet.
As RDS participants will be asked to go twice to a chosen location to be interviewed, fieldworkers will offer them compensation twice for their transportation costs. Compensations will be site specific. They may be monetary (i.e. reimbursement of transportation costs and/or credits for phone time) depending on local habits and national ethics committees' rules.
No identifying information recorded
All women recruited as study participants will be assigned unique ID numbers and no names or addresses will be recorded on the interview tools. There will be no identifying information on the consent form nor on the questionnaires, nor on the audio recording of the in-depth interviews. We will also ask for contact information (phone numbers) and permission to be contacted again for the second RDS interview. At this occasion, the contact information will be written down directly in an encrypted format on a separate form also containing the participant’s ID number, assigned at the same time. The encrypting key will be computer generated, and saved by the fieldworkers on their tablets and then at the University of Geneva in a computer file itself safeguarded on a computer protected by a password.
Privacy during interviews
All discussions and interviews will be held in a perfectly private space, with no one else assisting except the interviewer and the participant. For household surveys (ATPR/ self-report tool) and in-depth interviews, the following procedures will be enforced. The interview will only take place if a perfectly private place (like a secluded room, or a secluded spot in the courtyard or near the house) can be secured. If another person walks in the space, the interview is interrupted and starts again only once the person has left. The IDI and RDS interviews will be held in a private space indicated over the phone once the person calls; the place has to meet the same criteria as above. Fieldworkers will have a safe place to offer if the respondent does not know where to meet.
Selecting, training and contracting fieldworkers
Fieldworkers will go through a value clarification procedure, involving the collection of data on the level of stigma they attach to abortion at the beginning and at the end of the training. They will be trained to ensure that they will in no way use pressure or coerce women to participate. They will be trained on how to find first RDS participants ("seeds"). Fieldworkers will be trained in screening participants for eligibility criteria in a non-threatening way and in obtaining informed consent. If the allowance has been made by the country team for adolescents to consent for themselves, then this information will be provided to the trainees. Fieldworkers will moreover be trained in maintaining neutrality during the interview and in making women feel safe and comfortable talking about private reproductive issues with them. This will be even more important if adolescents are included in the study. They will be trained to comply with women who want to stop or refuse to answer one question during the interviews. To increase rapport and trust during the interview, all interviewers will be females; they will not live in the communities nor be known in the communities. Interviewers will be required by contract to maintain perfect confidentiality about the information shared with them by participants. Interviewers will be trained in registering absolutely no identifying information along with the data collected, either via tablets for the quantitative data or consent form or on audio records for the qualitative material. They will be trained to explain to participants the use of tablets for data recording. They will also be trained to explain how the anonymity of the data is fully preserved during the study.
Quantitative data collection software
Data collection for the ATPR and RDS component of the study will be conducted on handheld android tablets using the mobile data collection software SurveyCTO collect. SurveyCTO is a platform that facilitates offline electronic data collection. Electronic data collection will be used to allow for increased security of collected data; tablets will be password-protected and completed consent forms and questionnaires will be uploaded to a secure server at the end of each day. Interviewers will also carry paper copies of the instruments however, in the case of power or mechanical failure. Paper copies will be entered as soon as tablets are functional again and will be destroyed thereafter either by shredding or through incineration. Interviewers will pick up and return their tablets daily to the HDSS offices where they will be stored in locked cabinets. Only the lead scientist on each site will have access to these cabinets. Interviewers’ passwords will allow them to access the data collection tool but not to see or edit previously entered questionnaires on the tablet. Backups are automatically stored on each tablet in case they were are unable to transmit the data immediately and automatically to the remote server. The backup data will remain on the password-protected tablets until the end of field activities, and until all the data have been synced with the SurveyCTO Server, at which time each tablet will be securely and permanently wiped clean. When not in use, all tablets will be secured in lockable cabinets or containers/cases in the HDSS field offices accessible only to the lead scientist of each site.
Encryption and decryption keys will be generated by security software that comes with the SurveyCTO package and which is automatically available to the project through the study’s SurveyCTO account. The encryption key will be programmed into all questionnaires to ensure that the responses are encrypted prior to transmission to the remote server. The study team will have the private decryption key, which will be saved by the study investigators in a file on a secure computer accessible only to the PIs. Once a questionnaire is completed, all the sensitive data it contains (identification number and nicknames or initial for network numbers, mention of abortions) will be automatically encrypted during the transmission to our secure, private, remote server space within the SurveyCTO Server. Thereafter encrypted data will be downloaded onto computers by the study investigators using a desktop application called SurveyCTO Sync.
Data will be downloaded from the study server servers to cold room computers by the research team or persons appointed by them where they will be decrypted using the private key (during transmission it undergoes double encryption again). Thereafter all personal identifying information will be removed from the dataset before it is made available to other members of the research team for analysis. Although local backups of the data are saved on the computer and servers, just in case of a crash, it remains encrypted and therefore unreadable until the private decryption key is entered. At the end of the study period, all the data on the tablets will be wiped, final data collection tools, datasets and stata do files for importing data will be downloaded from the server and the server wiped and closed.
Decryption and storage of raw data will occur on one secure computer, accessible only to the PIs, for the duration of the study. All decrypted data will be stored on regularly backed up secure drives on password protected computers within the study institutions. Data will only be accessible to members of the study team for analysis. A master data set containing data from all countries will be held centrally at the University of Geneva, with the Geneva research members having access to this data set, and made available to HDSS sites as needed for the central products. Country specific data files will be available to researchers from the HDSS sites for future analyses, beyond the currently promised products. WHO, the University of Geneva and each center running the HDSS sites are the primary owners of each sites study data. All study data will be destroyed after 10 years.
Limitations of the study
Despite their promise to measure abortion safety in highly restricted contexts, network-based survey approaches will inevitably miss a portion of abortions, those of women who had a direct access to the health system and those who used common-knowledge methods. However, given that information on the safety of abortion comes so far almost exclusively from hospital-based data (women who seek care after a complication, estimated at 6.9 million in 2012 that is 27% of all unsafe abortion globally) , the network-based sample of abortions which can be expected through network approaches (about 70–80% of all abortions in highly restrictive settings) represents a great improvement.
Moreover, while a four sites study would have been ideal (one rural and one urban site in two countries), this design is a compromise between the requests for generalization (more than one country) and budgetary constraints. Sub-Saharan Africa was chosen as the region with the highest incidence of least safe abortions.
A last concern is the reliance of the study on key informants for the formative stage and implementation of RDS. Thoughtful and purposive recruitment of these informants is key to the success of this study, since key informants who are too similar are likely to recruit women with overlapping network interactions and some parts of the networks will be missed. We outlined above the various steps taken to limit risks in this regard: the choice of Health and Demographic Surveillance Systems, whose researchers know the study populations, and the high capacity of the staff involved in fieldwork.