Operationalization of the abortion trajectory

The current study builds upon the abortion trajectory inventory developed by Finer et al. [5]. Data on this abortion trajectory is collected through a self-administered questionnaire in the form of a calendar that surveys the timing of steps in the process of obtaining an abortion. Event history calendars (EHCs) have been demonstrated to be effective tools for collecting retrospective data in other studies [19, 20]. In addition, the EHC method has been used in SRHR research and other surveys that address sensitive topics [21,22,23]. As in previous research, we rely on the ideal type of abortion trajectory. However, we do not assume a certain chronological order in these steps, since the dates surveyed can be indicated freely. We assumed that an abortion trajectory comprises different, not necessarily chronological, steps, resulting in multiple, sometimes overlapping stages. For pragmatic reasons, we only measure when a respondent enters a particular stage for the first time; the respondent can indicate this date on the calendar integrated into the online questionnaire. The following key dates are surveyed explicitly: the first day of LMP, the day on which one first suspected pregnancy, and the day on which a (first) positive pregnancy test was taken.

Since we want to minimize possible recall bias even more and avoid double-questioning our respondents, we do not survey the dates of other steps in the trajectory directly. The five participating centers already record these data in their patient records. We link these data to our survey data, allowing us to reconstruct the entire trajectory without double querying. The day on which the abortion decision is made can be hard to recall since it is often not a clear-cut moment in time. For this key date, we, therefore, rely on the day the abortion center was contacted (for the first time), and the appointment for the first consultation was made. In general, during the first consultation, the appointment for the abortion itself is made. Accordingly, the date of the completion of the questionnaire, which is equivalent to the first consultation, serves as a proxy for the day on which an abortion is first requested. The last key date, the abortion procedure, is recorded by the centers themselves and is added to our dataset through the link between the center’s databases and our survey data. This link is established based on the respondents’ date of birth, which is recorded in the patient record at the center and which respondents are asked to provide at the beginning of the survey. In rare cases where two or more respondents share the same date of birth, we can still accurately link survey responses to patient records based on the day the survey is completed. This link also provides access to the respondents’ background data, such as parity, relationship status, and education level.

Operationalization of ‘barriers’Literature review

To operationalize the concept of ‘barrier’, we applied a systematic approach based on an extensive literature review as a first step. Through a thorough reading of the articles cited above, we could compile a comprehensive list of possible barriers and reasons for delay. To keep an overview, the barriers were grouped by the stage of the abortion trajectory in which they are most likely to occur [6] (Supplementary material - Table 1). Because the questionnaires are completed at the time of the first consultation, barriers can only be surveyed up to this point in the abortion trajectory. Therefore, we were not able to survey possible barriers faced after presenting for abortion. Since our study focuses on the gestational age at which women present for abortion, this is not a limitation of our study.

Fieldwork in abortion clinics

In each of the participating abortion centers, one day of fieldwork was conducted. In this way, insight could be gained into the actual operation and daily practice of Flemish abortion centers. As a result, barriers that were identified through this insight could be added to the list – drawn up during the literature review of possible barriers (see infra) – and some barriers could be deleted that turned out not to be applicable in the Belgian context (Supplementary material - Table 1). Being present in the abortion centers for a day was also important for building contact with the center managers and HCWs. In this way, we also ensured willingness to cooperate in the next phases of adaptation and validation and the implementation of the questionnaire.

Focus group discussion with HCWs

Based on the theoretical knowledge gathered through the literature review, the practical knowledge gained through fieldwork, the content, and the script for the Focus Group Discussion (FGD) with HCWs could be shaped. The list of barriers compiled and edited through the literature review and fieldwork served as the starting point for this FGD. We conducted an online FGD in which eight HCWs and two researchers were present. To encourage participation in the online setting, we used PollEverywhere to obtain input from our respondents in a simple and accessible way.

The FGD started with a short explanation of the project and the goal of the FGD. We started by asking, in an open-ended way, what they think are possible barriers or factors causing delays in presenting for abortion. The list of possible barriers was subsequently displayed per stage in the trajectory. The HCWs were asked which of the displayed barriers they feel do not apply in the Flemish context, which are the most important barriers, and what barriers they have already experienced in practice that are not yet listed.

The FGD revealed some interesting results; the HCWs of the different centers agreed clearly on which barriers do not occur in the Flemish context and which are the most relevant ones. The role of context as a critical shaping factor was noted, but it was stressed that the impact should certainly not be underestimated. The HCWs indicated that some barriers that were extracted directly from the literature are not that clearly or ambiguously worded. They advised on how they would clarify or simplify them. Some listed barriers were found to be similar, so they could better be grouped. Other barriers were found to be examples of higher-level barriers and thus fit better as a subcategory. It was also noted that some barriers may occur at multiple steps in the trajectory, whether in a slightly different form or not. Furthermore, HCWs added some specific barriers to the already compiled list, such as “It dawned on me that continuing the pregnancy would not be without risk (both for health; socially and emotionally)” or “I felt ashamed that I was thinking about opting for an abortion”. With the results of this focus group in mind, the list of barriers for each step in the abortion process was thoroughly reworked by the researchers, who created items that could be included in the questionnaire (Supplementary material - Table 1).

Piloting questionnaire

We conducted a three-way pilot study. First, we had other researchers experienced in the field or in developing questionnaires who critically reviewed the questionnaire. They were asked to mainly pay attention to the flow of the questionnaire and the completeness and wording of the questions. Therefore, it was also decided that since the ‘asking for an abortion’ step (stage 4) had quite a long list of barriers, which are also clearly of different nature, to split it into two parts. The barriers associated with making an appointment in an abortion center are displayed first, followed by the barriers involved in physically getting to the center for the first consultation.

Second, we conducted a pilot study with women who had an abortion in the preceding year. The participants were first asked to complete the draft version of the survey on their smartphones. Some questions were displayed in different ways to ascertain which was the most accessible. Later, the researcher asked to comment on various aspects of the questionnaire, including length, content, readability, and clarity. The goal of this pilot was to ensure that every woman felt that the questionnaire captured her particular situation. Accordingly, a group of respondents was selected for this pilot consisting of four women between 23 and 37 years old. They commented extensively on which questions they felt were unclear or triggering and whether they felt they could tell their story by completing the questionnaire. Adaptations were made accordingly (Supplementary material - Table 1).

The respondents of the pilot also commented on the suggested approaches for recruiting women into the study. The participants were unanimous that it would be effective if they were personally asked to complete the questionnaire rather than through other recruitment methods (e.g., through the display of a QR code in the waiting room). Keeping the comments from the pilot in mind, we chose to work with tablets. This ensures that the request to participate in the study is in person and, at the same time, makes the completion itself very accessible.

Third, the draft of the questionnaire was sent to the coordinators of each of the abortion centers. They mainly checked whether the questions would be straightforward to interpret and answer. These three steps of piloting resulted in the final questionnaire.

Final questionnaire

The final questionnaire was administered online on the Qualtrics Survey Platform, a commonly used online survey platform that conforms to the EU General Data Protection Regulation (GDPR) guidelines. When accessing the questionnaire, a short informed consent is displayed first (Supplementary material – final questionnaire). Participation in the questionnaire is entirely voluntary and can be stopped at any time, and the data will be processed anonymously. A link to a more comprehensive version of the informed consent is provided for those who want more information. Once the respondent agrees to the informed consent, the actual survey starts.

In the first module of the questionnaire, three key dates in the abortion trajectory are queried (Table 1). The respondents are first asked if they remember the date of a particular step. They can indicate one of the following responses: ‘Yes, I know the exact date’, ‘No, I don’t know the date exactly, but I have an idea of when it was approximately.’ or ‘No, I don’t know’. If they indicate the first or second option, a calendar will be displayed in which they can easily indicate the exact date, on which the step queried, took place. Some steps have additional questions if a certain response is indicated (Supplementary material – final questionnaire).

Table 1 The modules of the questionnaire and the number of items per module

The second module of the questionnaire focuses on the barriers experienced in these moments (Table 1). The respondents are asked to indicate for each barrier if they felt it applied by indicating ‘Yes’ or ‘No’. For some barriers, additional questions are asked when the respondent indicates that this barrier applies to her particular situation. First, the barriers associated with suspecting pregnancy are displayed, followed by those that might occur when taking a pregnancy test. The barriers associated with the third stage, the decision-making stage, are preceded by two slider questions. On an 11-point scale, running from 0 (‘I am still very much in doubt’) to 10 (‘my decision is firm’), the extent to which the decision for abortion is already certain should be indicated. Additionally, a second 11-point scale ranging from 0 (‘not difficult at all’) to 10 (‘very difficult’) is used to indicate how difficult the decision was. The answer to this question creates a context for the coming questions and is also very useful later in interpreting the responses.

The third and final module of the questionnaire consists of items copied and adapted from existing and broadly used scales and surveys (Supplementary material - Table 1). As indicated before, we used several subscales of the ILAS scale. The ILAS scale is a validated and reliable research instrument developed by Cockrill et al. [24] that can be used in research examining abortion stigma and related outcomes (e.g., women’s health, relationships, and behavior). Given that this scale was designed to be completed after the abortion, while we queried the respondents before the procedure, some changes had to be made. In addition to the ILAS scale items, we also included two questions from the European Value Survey (EVS) on religiosity and political orientation. By including these questions, we gain a basic insight into the respondents’ contexts, which is needed to frame perceived barriers and stigma [25].

The final questionnaire consists of 27 to 33 questions, depending on which answer options are indicated (Table 1).

Translation

The newly developed questionnaire is crafted in Dutch, the dominant language of the research team and the research setting. These questions are translated into French and English. Data from abortion centers show that the majority of the patient population (99.8%) is sufficiently proficient in one of these languages. For the remaining 0.02%, a telephone translator was consulted. The translation of the questionnaire was performed through a committee approach, ensuring that the process remained steep in the target language and avoiding the loss of quality associated with back-translation [26]. The same translation method was used to translate the ILAS scale from English into Dutch and French, while the questions obtained from the EVS were already translated into the three research languages by the EVS itself.