We conducted a nationwide voluntary open e-survey to recruit a convenience sample of independent midwives working in France during the first lockdown. This quantitative study followed the Checklist for Reporting Results of Internet E-Surveys (CHERRIES) to report our data .
Screening and recruitment
The link to the survey was disseminated by e-mail and Twitter to all the independent midwives who had signed up to receive newsletters from either the French national college of midwives (Collège National des Sages-Femmes de France) or the French union of midwives (Organisation Nationale Syndicale des Sages-Femmes). A nonprobability chain-referral sampling was obtained by e-mail. All midwives gave their consent before participating in the study, as described in the ethics approval section. Participation was voluntary, without any incentive or reward. The survey was open during the last week of the first lockdown period, from 29 April to 15 May, 2020. Only one entry was accepted from any IP address, to limit the risk of multiple participation.
The questionnaire was anonymous. Participants gave their informed consent by participating in the study, which they could stop at any time by leaving the site, thereby withdrawing their permission. The start of the questionnaire clearly stated the objectives of the study, the estimated length of time completing the survey would take, where and for how long data were stored, who the investigators were, the Ethics Committee registration number, and the procedure for submitting objections to such research to the national authorities. This study was approved by the Ethics Committee of the Hospices Civils de Lyon (decision n°20-48 dated 23rd March 2020) and registered at the Comission nationale de l’informatique et des libertés (CNIL, MR-004 n° 2217640 dated 17 April 2020).
The questionnaire was constructed by the authors and then revised during a meeting with the Accord group (a group whose objective is to Assemble, Coordinate, Understand, Research, and Debate in Primary Care) . It was tested on 10 midwives to verify the items’ usability, technical functionality, clarity, and reliability. It was then administered with open-source software (LimeSurvey).
The survey contained 19 questions on 4 pages. Six focused on the midwives’ characteristics (i.e., age, gender, practice setting: solo or group, and if the latter, whether all the co-workers were also midwives), 2 concerned practice adaptations (cancelled or postponed pregnancy consultations and/or Early Prenatal Interviews), 6 concerned making referrals for women, 4 concerned collaborations with hospitals, and 1 concerned a loss of opportunity for their patients. Loss of opportunity was defined as an absence or delay in care that could damaging their management (definition included in the question).
First, we assessed the difficulties for midwives in making referrals for the women in their care. We estimated these difficulties by using multiple-choice questions on a Likert-like scale (i.e., “no difficulty”, “difficulties, same as before”, and “new difficulties during the pandemic”; see Appendix S1) for each of the six dimensions measured to assess difficulty of access to ambulatory care (i.e., access to psychologists, social workers, family physicians, medical test laboratories, and sonographers).
Then, we assessed the difficulties for midwives in collaborating with hospitals. We estimated these difficulties using multiple-choice questions on a Likert-like scale (i.e., “nonexistent”, “worse than before”, “same as before”, or “better than before”; see Appendix S1) for each of the four dimensions assessed to measure their difficulty of collaborating with the hospital (i.e., transmission of health results, requests for medical expertise, organisation of unscheduled hospital care, and adoption of common protocols).
Finally, we assessed the midwives’ opinions of whether the pandemic had resulted in a “loss of opportunity” for the women in their care, by asking them directly if they considered that “all women (at different levels)” had experienced such a loss, specifically defined in the question as an absence or delay of care that of care that risked damaging their management)”.
Only completed questionnaires were analysed. All statistical analyses were performed with R software, version 4.0.3 . Depending on their distribution (normal or not), quantitative variables were expressed as means ± standard deviations (SD), and then compared with a Welch two-sample t-test, or as medians [25th-75th percentiles] and then compared with a Wilcoxon rank sum test. Qualitative variables were expressed as counts (percentages) and then compared with Fisher’s exact test.
We specified the 95% confidence intervals (95% CI) by the exact binomial method  and used multivariate analysis to examine factors that predicted loss of opportunity from the midwives’ perspective. All independent variables (i.e., experience, gender, office practice, and crisis area) were included in the multivariate model. All statistical tests were two-sided.