from the fields of general medical practice, obstetrics, midwifery, anthropology, and psychology, (4) two reviewers conducted the selection of publications, (5) we used a charting form and qualitative content analysis approach for the data extraction, and (6) we report the results and consider the meaning of the findings as they relate to the purpose of the study and research question. The study protocol was not registered. Search strategy The literature search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement extension for scoping reviews [36] (S1 Table). A comprehensive search was performed in the bibliographic databases PubMed,, Cinahl (via Ebsco) and the Cochrane Library, in collaboration with a medical librarian (LS). Search terms included controlled terms (MesH in PubMed, Emtree in Embase and Cinahl Headings). We used free text term only in the Cochrane database. The search was conducted from inception to April 16th 2019. Considering the breadth of the subject, we limited the scope of our review to risk selection based on medical risk factors, excluding risk selection based on social risk factors. Hence, the following terms, including synonyms and closely related words, were used as index terms or free-text words: “risk”, “selection”, “maternal and newborn care”, and “quality of care”. The search was performed without date or language restrictions. Duplicate articles were excluded. The search results were imported and merged in the reference database Mendeley [37]. The full search strategies for all databases can be found in S2 Table. Selection criteria, data extraction and analysis We used a systematic two-stage screening process to assess the relevance of the papers identified in the search [38]. In the first stage, two researchers (BG and AW) independently screened the papers’ title and abstract for inclusion. To ensure inter-reviewer agreement, BG and AW met weekly to discuss uncertainties, and they specified and expanded the inclusion and exclusion criteria. In the second stage, the papers’ full texts were assessed for eligibility. To ensure reviewer agreement, BG and AW assessed the first 20 full-texts independently, which resulted in complete agreement on inclusion. BG assessed the remaining papers by herself. A search update was conducted. AW randomly screened 500 of the additionally identified papers’ title and abstract independently and BG and AW assessed the first 15 papers’ full-texts independently, which reconfirmed inter-reviewer agreement. BG assessed the remaining title and abstracts and full-texts of the papers identified in the search update by herself. We excluded papers published prior to the year 1981, and non-research papers such as statements, opinions, book chapters, guidelines, protocols, conference posters and presentations to enhance feasibility. We restricted the language to English and Dutch. Studies conducted in low income countries and war zones were excluded to enhance comparability. Only studies were included focusing on the relation between the selection of medical risks, and referral between medical specialists delivering MNC. An inductive thematic synthesis approach [34,39,40] was used to identify how the concept of risk selection was approached in the included papers. Because we were interested in the operationalization of risk selection, we focused on the background, methods and discussion sections of the papers, and we did not assess the study results and the methodological quality of the papers. Following in-depth reading, we mapped the main focus of each paper using an excel sheet, which we then organised into main categories. We then searched for overarching themes (S3 Table). Data saturation was reached after data extraction of 125 papers. Screening of the remaining papers did not result in new main categories, confirming data saturation. ONE Risk selection in maternal and newborn care ONE | June 8, 2020 3 / 22 71 3 TOWARDS A BETTER UNDERSTANDING OF RISK SELECTION