Methods

Prior to the current study, a focus group of seven women were interviewed in order to test the interview guide. One of the design consequences of this initial focus group interview was to conduct single interviews as to minimize influence from the other women. Therefore it was decided to perform semi-structured individual interviews in the current study (12,13).
Informants were purposefully sampled with four groups of pregnant women based on the stage of pregnancy. The women in the first group were interviewed at a pregnancy consultation at 28 to 37 weeks of gestation concerning the mode of delivery. The second group comprised women at term, scheduled for elective caesarean delivery. The third group consisted of women in post-term pregnancies who were attending a planned examination prior to induction of labour.
The women were asked about their preferences regarding timing of prophylactic antibiotic treatment for caesarean delivery and asked to elaborate on their choices (Appendix S1). In addition, we noted their age, parity, and body mass index (BMI), but further demographics were not collected to ensure anonymity of participants.

Sampling strategy and structure of the interviews

Initially the women were contacted in person at the hospital when they arrived for a scheduled appointment. The aim of the current interview study was briefly outlined. If they volunteered to participate, they were given the informational pamphlet intended for participants in a possible future randomised trial on prophylactic antibiotics prior to or after cord clamping at caesarean delivery (Appendix S1). Thereafter they were given enough time to read the information undisturbed. The women were interviewed using an interview guide, to ensure a focused interview and that the same topics were discussed with all women (see Table S1). However, the interview structure remained flexible, allowing for follow-up questions to be asked if the women’s answers were unclear or needed elaboration. It allowed the researcher to explore unanticipated issues and to consider additional questions for the subsequent interviews. The women were both thoroughly asked about their preferences regarding antibiotic treatment and also about their attitudes to participation in a future RCT. Only the first issue about the preferences is discussed in this study.
The interviews were conducted from February to June 2020. All interviews were audio-recorded, and subsequently transcribed verbatim and anonymized by the first author. The women had no prior information regarding the local antibiotic prophylaxis procedures, until after the interview.
Additionally, post hoc interviews were performed where pregnant women in various stages of pregnancy were asked about their views on the neutrality and objectivity of the informational pamphlet (see Table S2).

Analyses

The analyses were performed by two authors, including the interviewer, using a systematic text condensation approach described by Malterud (12) to make a cross-sectional analysis of the interviews and to ensure a systematic interpretation aided by NVivo version 12 by QSR International (15).
First, themes were noted during a read-through of the interview-transcripts. Second, coding was then conducted by identifying and sorting meaning units - a text paragraph from the interview-transcript contributing to illustrate the issue - on which the main categories were based (Table 2). The main categories were divided into two or three sub-categories, but these sub-categories were adjusted if necessary, to ensure nuances of the answers were covered. When the two authors had agreed on the final sub-categories, they were used for the third step - condensation. Each sub-category was condensed in a process described as making false quotations, which is an essential step for the systematic text condensation approach. Finally, the condensates were synthesized and recontextualised. Our recapitulations were validated with the meaning units from the interview to ensure accordance between the data and our recontextualization and representative quotations were used as substantiation. The process remained flexible as the authors discussed the interviews as they progressed and if new themes emerged the above process was fully or partially repeated.