where women and children can only access medical specialists via a primary care provider, sometimes referred to as ‘gatekeeper’ [86,88,89]. Britt and colleagues pointed out that many insurers used “. . .primary-care physicians to act as gatekeepers who must approve referrals to specialists and sub-specialists. . .” [90] “. . .to limit the use of so-called ‘unnecessary’ referrals. . .” [86] and “. . .keep costs down. . .” [90]. In their study about referrals for genetic counselling by GPs in the Netherlands, Aalfs and colleagues [91] explained that “. . .as stated in an agreement between the Dutch Society of Clinical Genetics and the Dutch health insurance companies, referral for genetic counselling to one of the eight academic centres for clinical genetics was the task of GPs exclusively. This means that every patient who wants to be referred for genetic counselling has to visit their GP first.” Risk selection as a practice: Detecting and assessing risk, and making decisions about the delivery of care Part of the included papers address risk selection as a practice of detecting and assessing risk and making decisions about the delivery of care. Overall, this process is perceived in two contrary ways; on the one hand risk selection is defined as an objective and straightforward process; on the other hand risk selection is viewed as a subjective and complex process. As an objective process, risk is considered predictable and detectable using many data sources, including screening and diagnostic methods, such as risk scoring [60,92], a partogram [28], fetal fibronectin bedside testing for diagnosing preterm labour [27], and abdominal palpation and ultrasound for determining fetal presentation [93] and fetal growth [94]. In their paper about obstetrician-gynaecologists’ management of mental health conditions, Leddy and colleagues [95] explained that “the purpose of screening is not to determine that complete realm of psychological needs of a patient, but instead is a means by which to identify patients who may require further assessment, monitoring or referral.” Many papers predominantly consider risk selection as a process with a dichotomous outcome, risk classified as either present or absent. Detected risks can relate to the mother and the child, and differ in their nature, severity and urgency [65,66,96,97]. As a subjective process, authors acknowledge risk selection’s complexity, determined by health care organisation, care providers and women [98]. Organisation characteristics include the number of care providers involved in care provision, location, communication, collaboration, and geography [30,58,69,93,99–113]. Health care providers’ perceived risk, knowledge, expertise, confidence, personal views, awareness and attitude, financial considerations and women’s characteristics and preference, amongst others, are described as decisive factors [17–19,26,46,86–88,95,99,100,105,107,109,114–118]. Providers’ behaviour is considered influenceable via, for example, education [97,119–121], and guidelines [19,97,115,122]. Women bring in factors such as timing of presentation, sense of control, views and beliefs, sense of safety, perceived norms, perceived availability of options, and demographic characteristics such as age, level of educational and income [114,115,123–130]. Because a consulting care provider either needs additional resources or does not, the decision to refer is inherently a “threshold phenomenon” [108]. The contributing factors in the decision making process are weighed differently by different care providers, resulting in varying referral-thresholds, thus practice variation. According to several papers [19,63,108,117,131], this is especially the case for intermediate levels of risk, also referred to as the “grey zone". In contrast to clear high or low risks, these risks “. . .may be near the referral threshold, and therefore disproportionately susceptible to the marginal influences of numerous personal, social, cultural, and financial considerations. . .” [108]. The practice of risk selection in terms of detecting and assessing risk is not reserved to a certain profession, but rather performed by all professionals involved in care provision. The ONE Risk selection in maternal and newborn care | June 8, 2020 7 / 22 75 3 TOWARDS A BETTER UNDERSTANDING OF RISK SELECTION