PUTTING RISK IN ITS PLACE The complexity of risk selection in maternal and newborn care BAHAREH GOODARZI

PUTTING RISK IN ITS PLACE The complexity of risk selection in maternal and newborn care Bahareh Goodarzi

ISBN: 978-94-6419-693-1 Cover: Lauren Rebbeck Illustration, Lay-out: Ilse Modder, Printed by: Gildeprint, © 2023 B. Goodarzi All rights are reserved. No part of this thesis may be reproduced, distributed, or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or otherwise without prior written permission of the author or of the publisher holding the copyright of the published articles.

VRIJE UNIVERSITEIT Putting risk in its place The complexity of risk selection in maternal and newborn care ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. J.J.G. Geurts, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Geneeskunde op vrijdag 10 februari 2023 om 11.45 uur in een bijeenkomst van de universiteit, De Boelelaan 1105 door Bahareh Goodarzi geboren te Teheran, Iran

promotoren: prof.dr. A. de Jonge prof.dr. R. de Vries copromotor: dr. L. Holten promotiecommissie: prof.dr. P.J.M. Elders prof.dr. E.A.P. Steegers dr. P. Offerhaus dr. P. Verdonk prof.dr. S. Downe

oppositie: prof.dr. C.J.M. de Groot prof.dr. M.A.C. Versluis dr. E. de Miranda paranimfen: dr. Maaike Muntinga Rodante van der Waal, ME

Fish don’t see water they swim in. (David Foster Wallace, 1995 & Derek Sivers, 2011) The master’s tools will never dismantle the master’s house. (Audre Lorde, 1979) So long as we are divided because of our particular identities we cannot join together in effective political action. (Audre Lorde, 1983)

Table of content Summary Samenvatting Chapter 1 General introduction Chapter 2 Risk and the politics of boundary work: Preserving autonomous midwifery in the Netherlands Chapter 3 Towards a better understanding of risk selection in maternal and newborn care: A systematic scoping review Chapter 4 Models of risk selection in maternal and newborn care: exploring the organization of tasks and responsibilities of primary care midwives and obstetricians in Risk selection across the Netherlands Chapter 5 Maternal characteristics as indications for routine induction of labor: A nationwide retrospective cohort study Chapter 6 (Un)warranted variation in local hospital protocols for neonatal referral to the pediatrician: An explorative study in the Netherlands Chapter 7 General discussion Appendices About the Author Acknowledgements PhD Portfolio 10 14 21 39 69 93 109 129 159 198 199 202

Summary In this dissertation, we studied risk selection in maternal and newborn care (MNC). Chapter 1, the general introduction to this dissertation, begins by pointing to an interesting phenomenon: while contemporary risk selection has contributed to the lowest maternal and perinatal death rates in history, a misalignment between the care that is necessary and the care that is delivered remains, leading to unwarranted variation in care, the underuse and overuse of medical interventions, care disparities, and disrespectful care. To better understand this misalignment, I explore the genealogy of contemporary risk selection in MNC, starting with the origins of the term ‘risk’ and moving on to describe how changes in societal understandings of risk in high-income countries have shaped contemporary risk selection in MNC. The chapter concludes with the central research question of this dissertation: what is the nature of, and variation in, risk selection? To answer this research question, we conducted five studies, analysing (1) the concept of risk selection in MNC, and (2) the history, (3) the organisation, and (4 and 5) the practice of risk selection in MNC in the Netherlands. In chapter 2, we report the results of a scoping review of the concept of risk selection in MNC. We systematically searched the scientific literature to study how is risk selection conceptualized internationally. We included 210 papers, published over the past four decades, originating from 24 countries. We used inductive thematic analysis to identify key dimensions of risk selection. This study lays the groundwork for a shared conceptualisation of risk selection. We found that while different definitions of risk were used in the papers included in our review those definitions had at least two common elements: first, risk is often used as a proxy for establishing the needs of women and children, and second, risk is generally associated with pathology. Our analysis identified three main dimensions of risk selection: (1) risk selection as an organisational measure to optimally align women’s and children’s needs and resources, (2) risk selection as a practice to detect and assess risk and to make decisions about the delivery of care, and (3) risk selection as a tool to ensure safe care. We found that these three dimensions have three themes in common: risk selection (1) is viewed as both requiring and providing regulation, (2) has a provider centred focus and (3) aims to avoid underuse of care. The literature lacks a comprehensive understanding of risk selection that includes consideration of disparities in access to care, the needs of women in terms other than pathology, and the problem of the overuse of care. In chapter 3, we examine the systemof risk selection used in the Netherlands in its historical context, studying why and how midwives in the Netherlands gained their position as gatekeepers to specialist care. The existing system of risk selection in the Netherlands was based on the work of the 1983-1987 Study Group for the Revision of the Kloosterman List 10 SUMMARY

(SGKL), a committee comprised of representatives from all professions and organisations involved in Dutch MNC. We analysed the minutes of the SGKL’s meetings and conducted interviews with eight key-informants who were involved in the SGKL’s decision process. We used theories of professional jurisdiction and cultural theories of risk to analyse the factors that played a role in redefining the division of tasks and responsibilities in risk selection. The results of this study offer insight into how the authority to undertake risk selection is negotiated at the political and regulatory level. Our study contributes a better understanding of the arguments underlying the design of the system of risk selection, showing the political nature of risk selection, with decisions being determined not only by professionals’ understanding of risk but also by concerns with protecting their interests. These understandings and concerns can differ not only between but also within professions that seek to police and extend their boundaries. Negotiations are shaped by a dynamic interaction between these beliefs and interests, creating the possibility for otherwise unexpected transprofessional coalitions. These insights can reframe policy discussions in MNC, offering the possibility to view disputes in MNC as occurring between beliefs and interests, instead of between professional groups. In chapter 4, we describe the results of a nationwide survey amongst all primary care midwifery practices and obstetric departments undertaken to better understand how risk selection is organized in the Netherlands. We identified three MRS, which were distributed differently across regions: 1) primary care midwives assess risk and initiate a consult or transfer of care without discussing this first with the obstetrician; 2) primary care midwives assess risk and make decisions about consult or transfer care collaboratively with obstetricians; 3) models with other characteristics. We found no significant difference between MRS and levels of satisfaction. An approach that is evidence and value based is recommended as a way to promote optimal organization of risk selection. In chapters 5 and 6, we report on our studies of the effectiveness of the practice of risk selection. We researched the risk factors that inform the selection process and the protocols used to support that process. The results of these studies offer insight into the ways the accuracy of risk selection can be improved. In chapter 5 we studied how maternal age, parity, ethnicity, socioeconomic status, and gestational age inform risk selection regarding induction of labour. Parity and age are increasingly used as single indicators for inducing labour in otherwise healthy women. We analysed whether the use of additional maternal characteristics associated with perinatal mortality would more accurately identify births that would benefit from induction. We conducted a nationwide retrospective cohort study among a healthy Dutch population consisting of all singleton pregnancies in midwife-led care after 37 weeks of gestation in the period 2000-2018. We examined the association of maternal age, parity, ethnicity 11 SUMMARY

and socioeconomic status with fetal and neonatal mortality, for each gestational age at term. The association of single determinants was examined using descriptive statistics and univariable and multivariable logistics regression analyses. The associations of multiple determinants were examined using inter-categorical analyses with the inclusion of interaction terms in the multivariable logistic regression analyses. The study showed that the probability of perinatal mortality depends on the number of determinants taken into account. Importantly, we learned that decision making about the use of induction of labour to prevent fetal and neonatal mortality based on a single determinant may lead to the overuse or underuse of induction of labour. A value based health care strategy, addressing social inequity, and investing in better screening and diagnostic methods by employing an individualised and multi-determinant approach may be more effective at preventing fetal and neonatal mortality. In chapter 6 we present the results of our nationwide cross-sectional study into variation in the content of Dutch hospital protocols used in risk selection for neonatal referral to the paediatrician. Studies indicate that there is unwarranted variation in the risk selection process used in the referral of neonates. This may be caused by variation in the protocols used for neonatal referral by local hospitals. We compared the protocols of obstetric and neonatal departments in all hospitals in the Netherlands ― between regions, between neonatal and obstetrics departments and within neonatal and obstetrics departments ― for the six most common indications for neonatal referral. We found considerable interhospital, inter-department, and intra-department variation in recommendations for type of referral, admission, screening/diagnostic tests, treatment, and discharge. Furthermore, our results generally showed lower referral thresholds in neonatal departments compared to obstetric departments, and higher referral thresholds in the eastern region of the Netherlands. Our recommendations to reduce unwarranted variation in local protocols include: developing evidence-based, multidisciplinary guidelines to support local protocols, basing agreements onmultidisciplinary consensus only when evidence is lacking or remains inconclusive, paying attention to implementation of guidelines, and describing deviation from evidence-based guidelines because of specific local circumstances in the protocol. Chapter 7 is the general discussion. It reviews the findings of our research, showing how the subjectivity of, and blind spots in, risk selection limit its effectiveness. The advancements in knowledge and technology in MNC have contributed to the lowest maternal and perinatal mortality rates in history. And yet, MNC is plagued by the continued existence of unwarranted variation in care, the underuse and overuse of medical interventions, care disparities, and disrespectful care point. The contemporary understanding of risk selection is based on the presumption that risk is objective and is centred around risks generated from within the body because (1) risk is understood predominantly in terms of pathology, (2) risk is influenced by professionals’ beliefs and interests, and (3) risk is 12 SUMMARY

relative. The subjectivity of risk can lead to ineffective risk selection due to blind spots that impede optimal alignment between women’s’ needs and MNC. This misalignment is visible in (1) the lack of women-centred care, (2) the disregard for overuse of care, and (3) the little attention to primary prevention. These undesirable outcomes of MNC point to the pressing need for further studies of the scope and effect social determinants of health and discrimination on birth outcomes. It is also imperative that we move beyond the assumption that risk emerges only from the pregnant and birthing body and begin to examine the risks created by MNC itself and its framing and managing of pregnancy and birth as risky. Studies into the subjective and complex character of risk selection must use intersectional analyses taking into account the impact of converging determinants and the role of power on MNC outcomes. Our research points the way to the optimization of risk selection via a broader conception of the sources of risk, an appreciation of the subjective and complex character of risk, a move to centring the focus on women and on prevention, and finally, by giving strong support to physiological approaches to pregnancy and birth. This broader conception of risk selection will offer the possibility to deploy the advancements in MNC to further reduce maternal and perinatal morbidity and mortality by avoiding preventable risks and improving care equitably, not offering some children a better start, but all children an equal start. 13 SUMMARY

Samenvatting In dit proefschrift bestudeerden wij risicoselectie in de geboortezorg. Hoofdstuk 1, de algemene inleiding van dit proefschrift, begint met de beschrijving van een interessant fenomeen: hoewel de hedendaagse risicoselectie heeft bijgedragen tot de laagste sterftecijfers onder zwangeren en hun kinderen in de geschiedenis, blijft er een wanverhouding bestaan tussen de zorg die nodig is en de zorg die wordt verleend, wat leidt tot onwenselijk variatie in de zorg, onder- en overgebruik van medische interventies, ongelijkheden in de zorg en respectloze zorg. Om deze wanverhouding beter te begrijpen, verken ik de genealogie van de hedendaagse risicoselectie in de geboortezorg. Eerst beschrijf ik de oorsprong van de term ‘risico’ en vervolgens hoe veranderingen in de maatschappelijke opvattingen over risico’s in hoge-inkomenslanden de hedendaagse risicoselectie in de geboortezorg hebben gevormd. Het hoofdstuk besluit met de centrale onderzoeksvraag van dit proefschrift: wat is de aard van en de variatie in risicoselectie in de geboortezorg? Om deze onderzoeksvraag te beantwoorden, hebben wij vijf studies uitgevoerd, waarin wij (1) het concept van risicoselectie in geboortezorg en (2) de geschiedenis, (3) de organisatie (4 en 5) en de praktijk van risicoselectie in de Nederlandse geboortezorg hebben geanalyseerd. In hoofdstuk 2 rapporteren wij de resultaten van een scoping review naar het begrip van risicoselectie in de geboortezorg. Wij hebben systematisch in de wetenschappelijke literatuur gezocht hoe risicoselectie internationaal wordt geconceptualiseerd. Wij hebben 210 artikelen geïncludeerd, die gepubliceerd waren in de afgelopen vier decennia en afkomstig waren uit 24 landen. Wij gebruikten een inductieve thematische analyse om de belangrijkste dimensies van risicoselectie te identificeren. Deze studie legt de basis voor een gemeenschappelijke begrip van risicoselectie. Wij ontdekten dat, hoewel er verschillende definities van risico werden gebruikt in de studies die deel uitmaakten van ons onderzoek, deze definities ten minste twee gemeenschappelijke elementen hadden: ten eerste wordt risico vaak gebruikt als een benadering voor het vaststellen van de behoeften van zwangeren en hun kinderen en ten tweede wordt risico over het algemeen geassocieerd met pathologie. In onze analyse identificeerden we drie belangrijke dimensies van risicoselectie: (1) risicoselectie als een organisatorische maatregel om de behoeften van zwangeren en hun kinderen en de beschikbare middelen optimaal op elkaar af te stemmen, (2) risicoselectie als een praktijkinstrument om risico’s op te sporen en te beoordelen en om beslissingen te nemen over de zorgverlening en (3) risicoselectie als een hulpmiddel om veilige zorg te garanderen. Wij vonden dat deze drie dimensies drie thema’s gemeen hebben: risicoselectie (1) wordt gezien als regulerend en regulatie behoevend, (2) is zorgprofessional-gecentreerd en (3) heeft tot doel onderbenutting van zorg te voorkomen. In de literatuur ontbreekt een alomvattend begrip van risicoselectie dat rekening houdt met 14 SAMENVATTING

ongelijkheden in toegang tot zorg, met de behoeften van zwangeren in andere termen dan pathologie en met het probleem van overmatig gebruik van zorg. In hoofdstuk 3 onderzoeken wij het systeem van risicoselectie in Nederland in de historische context. We analyseren waarom en hoe verloskundigen in Nederland hun positie als poortwachters naar specialistische zorg hebben verworven. Het bestaande systeem van risicoselectie in Nederland was gebaseerd op het werk van de Werkgroep Bijstelling Kloostermanlijst (WBK) tussen 1983-1987, een commissie bestaande uit vertegenwoordigers van alle beroepsgroepen en organisaties die betrokken zijn bij de geboortezorg. Wij analyseerden de notulen van de vergaderingen van de WBK en voerden interviewsmet acht sleutelinformanten die betrokkenwaren bij het besluitvormingsproces van de WBK. Wij gebruikten theorieën over professionele jurisdictie en culturele theorieën over risico om de factoren te analyseren die een rol speelden bij het herdefiniëren van de taak- en verantwoordelijkheidsverdeling tussen de betrokkenen bij risicoselectie. De resultaten van dit onderzoek bieden inzicht in hoe op politiek en regelgevend niveau wordt onderhandeld over de bevoegdheid om risicoselectie uit te voeren in de geboortezorg. Onze studie draagt ​bij aan een beter begrip van de argumenten die ten grondslag liggen aan de organisatie van risicoselectie in de geboortezorg. Ook geeft het inzicht in de politieke aard van risicoselectie, waarbij beslissingen niet alleen worden bepaald door het begrip van risico’s van professionals, maar ook wordt ingegeven door bescherming van hun belangen. Deze opvattingen en zorgen kunnen niet alleen verschillen tussen, maar ook binnen beroepsgroepen die de grenzen van hun vak willen bewaken en verleggen. Onderhandelingen over de taak- en verantwoordelijkheidsverdeling van risicoselectie worden gevormd door een dynamische interactie tussen deze overtuigingen en belangen van de betrokken partijen, waardoor de mogelijkheid ontstaat voor onverwachte transprofessionele coalities tussen deze partijen. Deze inzichten kunnen beleidsdiscussies in de geboortezorg herdefiniëren, wat de mogelijkheid biedt om geschillen in de geboortezorg te zien als het gevolg van een conflict in overtuigingen en belangen, in plaats van tussen beroepsgroepen. In hoofdstuk 4 beschrijven wij de resultaten van een landelijk onderzoek onder alle eerstelijns verloskundige praktijken en obstetrie afdelingen. Wij hebben deze studie uitgevoerd om beter te begrijpen hoe risicoselectie in Nederland is georganiseerd. Wij identificeerden drie risicoselectie-modellen, die verschillend verdeeld waren over de regio’s: 1) eerstelijns verloskundigen beoordelen het risico en verwijzen voor een consult of overdracht van zorg zonder dit eerst met de gynaecoloog te overleggen; 2) eerstelijns verloskundigenbeoordelenhet risicoennemenbeslissingenover consultatieenoverdracht van zorg in overleg met de gynaecoloog; 3) modellen met andere kenmerken. Wij vonden geen significant verschil tussen de risicoselectie-modellen en mate van tevredenheid. Om een optimale organisatie van risicoselectie te bevorderen, bevelen wij een aanpak aan die 15 SAMENVATTING

gebaseerd is op evidence en op de uitgangspunten van waardegedreven zorg. In de hoofdstukken 5 en 6 doen wij verslag van onze studies naar de effectiviteit van de praktijk van risicoselectie. Wij hebben onderzoek gedaan naar de risicofactoren die het selectieproces beïnvloeden en naar de protocollen die worden gebruikt om dat proces te ondersteunen. De resultaten van deze studies bieden inzicht in de manieren waarop de nauwkeurigheid van risicoselectie kan worden verbeterd. Inhoofdstuk5hebbenwijonderzoekgedaannaarhoedefactoren leeftijd,pariteit,etniciteit, sociaaleconomische status en zwangerschapsduur van de zwangere samenhangen met risicoselectie met betrekking tot het inleiden van de bevalling. Pariteit en leeftijd worden in toenemende mate gebruikt als enige indicatoren voor het inleiden van de bevalling bij gezonde zwangeren. Wij onderzochten of het gebruik van bijkomende kenmerken van de zwangere die geassocieerd worden met perinatale sterfte nauwkeuriger de zwangeren zou kunnen identificeren die baat zouden hebben bij een inleiding. Wij voerden een landelijke retrospectieve cohortstudie uit onder een gezonde Nederlandse populatie bestaande uit alle zwangeren van een eenling en bij de eerstelijns verloskundige onder controle waren na 37 weken zwangerschap in de periode 2000-2018. Wij onderzochten voor elke zwangerschapsduur in de á-terme periode de associatie tussen leeftijd, pariteit, etniciteit en sociaaleconomische status van de zwangere met foetale en neonatale sterfte. De associatie van enkelvoudige factoren werd onderzocht aan de hand van beschrijvende statistieken en univariabele en multivariabele logistische regressieanalyses. De associaties van meerdere factoren werden onderzocht door middel van inter-categorische analyses door het toevoegen van additionele factoren als interactietermen in de multivariabele logistische regressieanalyses. De studie toonde aan dat de kans op perinatale sterfte afhangt van het aantal factoren waarmee rekening wordt gehouden. Wij lieten zien dat besluitvorming op basis van één enkele factoren kan leiden tot over- of ondergebruik van inleiding van de bevalling. Een op waarden gebaseerde zorgstrategie, het aanpakken van sociale ongelijkheid en het investeren in betere screenings- en diagnostische methoden door gebruik te maken van een geïndividualiseerde benadering waarin rekening wordt gehouden met meerdere factoren zou effectiever kunnen zijn in het voorkomen van foetale en neonatale sterfte. In hoofdstuk 6 presenteren wij de resultaten van onze landelijke cross-sectionele studie naar variatie in de inhoud van Nederlandse ziekenhuisprotocollen die gebruikt worden bij risicoselectie voor neonatale doorverwijzing naar de kinderarts. Uit onderzoek blijkt dat er onwenselijke variatie bestaat in het risicoselectieproces bij de verwijzing van neonaten. Dit kan worden veroorzaakt door variatie in de protocollen die door lokale ziekenhuizen worden gebruikt voor neonatale verwijzing. Wij vergeleken de protocollen van obstetrie en neonatale afdelingen in alle ziekenhuizen in Nederland tussen regio’s, tussen neonatale 16 SAMENVATTING

en obstetrie afdelingen en binnen neonatale en obstetrie afdelingen, voor de zes meest voorkomende indicaties voor neonatale verwijzing. Wij vonden aanzienlijke variatie tussen ziekenhuizen, tussenafdelingenenbinnenafdelingeninaanbevelingenvoortypeverwijzing, opname, screening/diagnostisch onderzoek, behandeling en ontslag. Bovendien lieten onze resultaten over het algemeen lagere verwijsdrempels zien op neonatale afdelingen in vergelijking met obstetrie afdelingen en hogere verwijsdrempels in de oostelijke regio van Nederland. Wij bevelen aan om onwenselijke variatie in lokale protocollen te verminderen door het ontwikkelen van evidence-based, multidisciplinaire richtlijnen ter ondersteuning van lokale protocollen. Verder bevelen wij aan afspraken alleen te baseren op multidisciplinaire consensus wanneer wetenschappelijke onderbouwing ontbreekt of niet overtuigend is en om aandacht te besteden aan implementatie van richtlijnen en om argumenten te beschrijven voor het afwijken van evidence-based richtlijnen vanwege specifieke lokale omstandigheden in het protocol. Hoofdstuk 7 is de algemene discussie van dit proefschrift. Het geeft een overzicht van de bevindingen van onze studies, waaruit blijkt hoe de subjectiviteit van en blinde vlekken in risicoselectie de effectiviteit ervan beperken. De vooruitgang in kennis en technologie in de geboortezorg heeft bijgedragen tot de laagste sterftecijfers onder zwangeren en hun kinderen in de geschiedenis. En toch blijft er sprake van onrechtvaardigde variatie in zorg, het onder- en overgebruik vanmedische interventies, onwenselijke gezondheidsverschillen en respectloze zorgverlening. Het hedendaagse begrip van risicoselectie is gebaseerd op de veronderstelling dat risico objectief is en is gecentreerd rond risico’s die ontstaan in het zwangere en barende lichaam, omdat (1) risico voornamelijk wordt opgevat in termen van pathologie, (2) risico wordt beïnvloed door overtuigingen en belangen van professionals en (3) het idee dat risico relatief is. De subjectiviteit van risico kan leiden tot ineffectieve risicoselectie als gevolg van blinde vlekken die een optimale afstemming tussen de behoeften van zwangeren en de geboortezorg in de weg staan. Deze verkeerde afstemming is zichtbaar in (1) het gebrek aan persoonsgerichte zorg, (2) de veronachtzaming van overgebruik van zorg en (3) de geringe aandacht voor primaire preventie. Deze ongewenste uitkomsten van geboortezorg wijzen op de dringende noodzaak van verder onderzoek naar de reikwijdte en het effect van sociale factoren van gezondheid en discriminatie op geboortezorguitkomsten. Het is ook noodzakelijk dat wij verder gaan dan de veronderstelling dat risico’s alleen voortkomen uit het zwangere en barende lichaam en beginnen met het onderzoeken van risico’s die worden gecreëerd door de geboortezorg zelf, waarin zwangerschap en geboorte worden gedefinieerd en behandeld als risicovolle processen. Wij bevelen aan dat toekomstige studies naar het subjectieve en complexe karakter van risicoselectie intersectionele analyses gebruiken die rekening houden met de impact van convergerende factoren en de rol van macht op geboortezorguitkomsten. 17 SAMENVATTING

Ons onderzoek wijst de weg naar optimalisering van risicoselectie via een bredere opvatting over risico, de erkenning van het subjectieve en complexe karakter van risico’s, het centreren van de focus op zwangeren en op preventie en, tenslotte, op de inclusie van een fysiologische benadering van zwangerschap en geboorte. Deze bredere opvatting van risicoselectie zal de mogelijkheid bieden om de morbiditeit en mortaliteit van moeders en kinderen verder terug te dringen door vermijdbare risico’s te voorkomen en de zorg op rechtvaardige wijze te verbeteren, door niet sommige kinderen een betere start, maar alle kinderen een gelijke start te bieden. 18 SAMENVATTING


Chapter 1 Research replicates systems of power, since the types of questions that are asked, and the sort of information worthy of collection, often reflects the biases of the powerful. Public health has moved from an environmentalist approach, making the world safer for people, to a people education model, in which people need to make themselves safer in the world. (Barbara Katz Rothman, 2016) Maternal and newborn care is a political arena and risk is the currency in the physiology vs pathology debate. 20

General introduction Pregnancy and birth are primarily physiological processes.1 The vast majority of births result in a healthy mother and newborn. However, complications can occur. The likelihood of experiencing a complication and the severity of that complication varies between countries, care settings, and women2 and their children.[1,2] To ensure women and their children receive the right care, at the right place, and at the right time, an effective system for selecting the most appropriate care and care provider is necessary.[1,2] The current selection system is informed by the type and degree of risks associated with pregnancy and birth.[1] This risk-based selection system is responsible, in part, for the reduction in maternal and perinatal mortality. And yet, over the past decade, five separate Lancet series — on Midwifery (2014), Maternal health (2016), Stillbirth (2016), Caesarean section (2018), and Miscarriage (2021) ― called attention to outcomes of maternity and newborn care (MNC) that indicate a misalignment between necessary care and delivered care: unwarranted variation in care, the underuse and overuse of medical interventions, care disparities, and disrespectful care.[3–9] Unwarranted variation is variation that cannot be explained by variation in needs of childbearing women and their children.[10,11] Underuse of care refers to care that is delivered “too little too late”, while overuse of care refers to care that is delivered “too much too soon”.[8] Care disparity refers to unwarranted variation due to institutional disadvantages and injustice.[12] Care is respectful when it is individualized, culturally and contextually appropriate, delivered with respect for people’s fundamental rights, and responsive to their changing needs.[8] MNC in the Netherlands is no different in this regard. Women in the Netherlands have reported experiencing disrespectful care, such as verbal abuse and medical interventions without consent.[13] Furthermore, studies indicate disparities in the outcomes of care offered to Dutch women, including ethnic differences in maternal and perinatal mortality.[14–16] Research has also shown increased rates of referral from midwife-led to obstetrician-led care and increased use of medical interventions in birth ― in spite of the lack of clear evidence that rising referral rates lead to better outcomes ―, and variation in care by region of the country rather than the needs of women.[17,18] Indeed, policymakers in the Netherlands have concluded that the existing system of risk selection is not effective.[19,20] 1 It has been argued that it is not correct to use the term ‘physiological’ to describe the healthy and normal nature of pregnancy and birth, because it refers to a science that deals with the ways that living things function. Therefore, it has been suggested to use the term ‘low risk’ instead.[114 p 1] However, the term ‘physiological’ should not be confused with the term ‘physiology’. Indeed, the latter refers to the scientific domain, however, the former is defined as characteristic of, or appropriate to, the health and normal functioning of an organism.[115] 2 When we use the term ‘woman’, I also refer to individuals with a uterus who are not woman identified, including trans men and non-binary individuals. 21

To optimize the system of risk selection, midwives and obstetricians in the Netherlands have begun to look for a better way to determine the most appropriate care and care provider. This endeavor is hindered, however, by disagreement about what risk selection is and how it can be best organized and practiced.[21] International scientific literature does not offer a way out of these disagreements. Consider, for example, the varied ways risk selection is operationalized in research: as the planning and arranging of care,[22,23] as a cognitive skill,[24,25] and as a clinical tool.[26–28] Scholars also disagree about the level of risk that justifies medical interventions, as illustrated in the recent debate over offering induction of labour to all women at term in the International Journal of Obstetrics and Gynecology (2019).[29,30] The lack of clarity about risk selection hampers the development of an evidence base for its optimization. Current strategies for risk selection in MNC have been shaped by the ways risk is perceived in relation to pregnancy and birth, but also by the way risk is understood in society in general. Therefore, in this introduction, I begin with a description of the genealogy of contemporary risk selection in MNC, starting with the origin of the term ‘risk’. I then explore changes in societal understandings of risk in high-income countries, focusing on the way these changes have shaped contemporary risk selection in MNC. I conclude with a presentation of the aims of the research presented in this dissertation. The origin of the term risk Although people have long been preoccupied with risks in life, the word ‘risk’ is relatively new. One of the first recorded human practices of dealing with risk in a systematic way dates from circa 3200 B.C. in the Tigris-Euphrates valley, where consultants used signs from the gods to manage harmful, uncertain, or difficult decisions.[31] There is no agreement about the etymological origin of the word ‘risk’.[32–37] But there is agreement that the understanding of the word risk changed over time. It seems that the word first appeared as ‘risque’ in the 1674 Thomas Blounts Glossrophia,[38] originating from the Latin word ‘risigus’ or ‘riscus’.[32,33,36] According to some, it comes from the classical Greek ‘ριζα’, a navigational term meaning ‘root’, ‘stone’ or ‘cut of the firm land’ used by sailors entering uncharted waters as a metaphor for ‘difficulty to avoid in the sea’.[35,36] In Spanish, risk means ‘steep abrupt rock’, also suggesting danger for those at sea. According to others, the origin is linked to the Arabic word ‘rizq’, a term for the acquisition of wealth and good fortune.[32,34] With the advent of the printing press in the 15th century, the word spread to other countries.[32,33,39] In contemporary society, risk is understood to mean a combination of knowledge and uncertainty. When there is a risk, there must be uncertainty: something that is unknown 22 1 CHAPTER 1

or has an unknown outcome. But for an uncertainty to constitute a risk, something must be known about it.[40] Safety is often defined as the antonym of risk, where the level of safety is conceived in relation to the level of risk: the lower the risk, the higher the safety. [41] The construction of the contemporary understanding of risk In pre-modern times, risk was associated with a harmful natural event. There was little that people could do in the face of these events. They could try to estimate the likelihood of a harmful event and take measures to reduce its impact, but the outcome was a product of chance, often ascribed to the will of gods, often rationalized by the belief that bad things happen to bad people.[42] With the advent of the scientific revolution in 17th century Europe, knowledge that once was regarded as belonging exclusively to gods came to be seen as discoverable with use of the scientific method. The emphasis shifted from subjective to objective knowledge and the assumption that the social and natural world follow laws that may be measured, calculated, and therefore predicted. Also, risk came to be associated with human fault and responsibility, seen as something that could be managed and avoided.[42] Members of an increasingly secular society no longer placed their destiny in the hands of gods. The natural events of life were seen as controllable and indeterminacy could be reduced with the calculation of probabilities.[43] Probability calculations always contain a degree of ‘uncertainty’. That is when notion of risk arises: risk is present, not when an event will happen, but when it may happen.[37,42] Society continues to strive for the transformation of uncertainty into certainty.[44] In the early modernist notion, risk was a neutral phenomenon. In the fields of economy and insurance for example, risk was perceived as the probability of something happening combined with the sum of associated losses and gains.[42,45] This idea of risk as a neutral concept dominated until the beginning of the 19th century. By the end of the 19th century, the distinction between risk and uncertainty and between good risk and bad risk was lost, and the focus shifted to negative outcomes.[42,46] In contemporary society, the concept of risk is related to a desire to control and predict the future. Whereas pre-modern society was concerned with events that occurred on a regular basis, post-modern society is preoccupied with improbable events. The existence of technologies that offer to screen, predict and detect risk, gave rise to the idea that harm and loss can be prevented before something is actually at stake. The ability to foresee 23 1 GENERAL INTRODUCTION

risk and thus prevent harm, alters the nature of responsibility. The decision to act, or not, in the face of predicted risk will be regretted if harm or loss occurs that could have been averted.[33,42] Thus, the choice to accept risk and possible negative consequences became controversial. In this context, the decision to act to avoid predicted risk is always right and immune to failure and critique, because even if things go wrong one had acted correctly by choosing to avert risk and its consequences.[33,47] Risk and risk selection in maternal and newborn care Medical interventions, broadly defined, have always been present in MNC. However, the nature of these interventions changed over time. In pre-modern times, MNC was predominantly provided by women in the community who were skilled in midwifery practices. These ‘wise women’s’ knowledge was based on observation and experience combinedwithwhat they learnedasmothers, whichwas handeddown throughgenerations of attending women.[48] Midwives did not use instruments to assist at birth, but they had a variety of techniques, including herbal remedies, chants, invocations, charms positioning, movement, and hands-on manoeuvres to prevent mortality.[49] These techniques were ‘researched’ long before it was understood as such; they were tried, evaluated, adjusted as more subtle variables became clear, and moved into practice or abandoned.[50,51] Pregnancy and birth have become a much safer process in terms of mortality — especially in high-income countries — as a result of scientific discoveries about hygiene and nutrition, the organization of hospitals, and the specialties of obstetrics and neonatology.[49,50,52– 55] During the 19th and 20th century, medical interventions became more mechanical and more automated, including blood transfusions,[56] induction of labour,[57] and vacuumassisted delivery.[58] Advancements in medicine made in the 20th century brought new ways of preventing and treating maternal and perinatal mortality, such as antibiotics [56], and anti-D immunoprophylaxis.[59] Modern science, technology, and hospitals radically changed the perception of the nature of pregnancy and birth. In pre-modern societies pregnancy and birth were seen as natural, physiological, and healthy events until proven otherwise.[60–62] Modernity introduced a biomedical and technocratic discourse, first in the Western-industrial countries,[63] and later elsewhere.[64] Amajor turning point was the acceptance of the Descartes’ mind-body dualism in the 17th century. Philosopher René Descartes conceived the body as a machine governed entirely by the laws of physics, that could be taken apart and reassembled in an effort to fully understand its structure and function.[53,65–67] In this discourse, male bodies were considered the norm and conceptualized as ‘advanced’, ‘purified’, and ‘celestial’. Female bodies were regarded as ‘natural’, ‘primitive’, ‘terrestrial’, and 24 1 CHAPTER 1

‘polluting’.[53,68–71] From this patriarchal point of view, pregnancy and birth were viewed as inherently imperfect, ‘pathological’, ‘abnormal’ and ‘unnatural’, and ‘untrustworthy mechanical processes’ [72] that could and should be controlled.[52,73] Thebiomedical and technocratic discourse altered the role and status ofMNCprofessionals. Midwives’ knowledge and practice were increasingly disputed and devalued, and they were often accused of witchcraft.[53,67,74] Neonatology claimed authority over diseases of the newborn,[75] obstetricians assumed expertise over the pathologies of pregnancy and birth and the use of medical interventions, and midwives were granted the domain of physiological a pregnancy and birth.[50,55,76,77] Furthermore, MNC professionals were now deemed responsible and accountable for their decisions, confirming the belief that risks can be managed and adverse outcomes are the fault of those who made the decisions about care.[52,73] The pre-modern blaming system based in religion and sin was replaced by a secular system of blame.[52,78–80] Within the discourse that views pregnancy and birth as events that require medical management, risk ― a term used predominantly in relation to pathology ― is central in decisions about care provision.[52,73] Rothstein and colleagues (2007) refer to this phenomenon as “risk colonization”,[81] where risk has increasingly come to define the object, methods, and rationale of MNC.[82] This is accompanied by what Scamell and Alaszewski (2012) refer to as “an ever-narrowing window of normality”,[52] where normality is signified only through an absence of risk, and pregnancy and birth can only be labelled as physiological in retrospect.[52,60–62,69] Risk is located inside the body, which can be detected and treated from the outside-in, legitimizing continuous monitoring. [60,83,84] Reynolds (1991) labels this process as the “one-two punch”: punch one is destroying natural processes after rendering them dysfunctional with technology, and punch two is fixing it with technology and rebuilding them as a medical process.[72] The labelling of an increasing number of normal bodily processes as risky is also referred to as ‘medicalization’ or ‘pathologization’. This includes defining morbidity and mortality that result from structural inequalities, such as housing and nutrition, as individual medical conditions that can be treated with medical interventions.[85–88] Paradoxically, the same science, technology, hospitals, and care professionals that monitor, prevent, and treat risks from inside the body, can increase the risks of maternal and perinatal morbidity and mortality.[53,89–92] This is referred to as iatrogenesis.[93] Consider, for instance, the development of the forceps, an advancement in medical technology won at the expense of underprivileged, poor and black women, who were involuntarily experimented upon in horrific circumstances.[94,95] Historians have noted that the development of pain management was likewise built on discriminatory ideas. There was a fear that white upper-class women ― who were thought to be more fragile 25 1 GENERAL INTRODUCTION

and sensitive to pain than women of colour and lower-class women ― were avoiding childbirth, creating the possibility they would eventually become a minority.[96,97] With the advent of hospitals, the most common cause of maternal deaths became puerperal fever due to highly virulent strains of Streptococcus pyogenes transferred to birthing women from doctors and nurses in hospitals.[56,98] The use of the medicine Diethylstilbestrol (DES) to prevent miscarriage is another example of iatrogenesis. Between 1940 and 1970 DES was routinely prescribed until DES was shown to cause cancer and fertility problems in women who had been exposed to this medication in utero. Two current examples are the routine use of electronic fetal monitoring during birth and the routine use of ultrasonography in the third trimester, both practices that in many places have become a standard part of contemporary MNC despite a lack of evidence of their clinical effectiveness. Studies show that electronic fetal monitoring is a poor tool for identifying or predicting fetal and neonatal morbidity and mortality, and it is associated with unnecessary caesarean sections, leading to excess maternal and perinatal morbidity and mortality.[99–103] Likewise, routine ultrasonography used in the third trimester as a means to reduce adverse perinatal outcomes in low risk pregnancies does not, in fact, decrease the rate of perinatal mortality.[104] Research into ways to reduce ineffective care and care disparity point towards practices such as value based health care,[105] women-centred care,[1,106] shared decision making,[107] respectful care,[108,109] and continuity of care.[110] However, efforts to implement these practices have been stymied by the risk focused, biomedical, and technocratic discourse of the current system of selection. That system is based on the premise that risks mainly reside inside the pregnant and birthing body, and that harms can and should be predicted, detected and treated using medical interventions. Towards optimizing risk selection Contemporary MNC in high-income societies includes care during the preconceptual, prenatal, intrapartum and postnatal periods offered by many different professionals. Care is provided within a risk focused, biomedical, and technocratic discourse, where the understanding of risks related to pregnancy and birth informwhat care should be provided and who should provide it. The optimization of risk selection will require a wider lens that addresses risks that come from outside the pregnant and birthing body, and the risks induced by MNC itself. This more complete understanding of risk selection is necessary to develop an evidence base that supports optimal risk selection. The Netherlands provides a unique context to study risk selection. In most high-income countries MNC is routinized within a hospital setting and midwives have limited autonomy. 26 1 CHAPTER 1

[50,70,76,77,111] In the Netherlands, midwives, obstetricians, and neonatologists have their own areas of expertise and practice autonomously. Two-thirds of Dutch midwives work in primary care practices.[112] They are the main care providers for women during pregnancy and birth and serve as gatekeepers. They autonomously conduct risk selection, deciding when referral is necessary to obstetrician-led care or the neonatologist in the hospital. When specialist care is no longer required, women are referred back to their primary care midwife.[113] Objective of this dissertation The overarching objective of this dissertation is to enhance the understanding of risk selection in contemporary MNC, as a means to reduce maternal and newborn morbidity and mortality. The studies were guided by the following research question: What is the nature of, and variation in, risk selection? To answer this research question, we conducted five studies, analysing (1) the concept, (2) the history, (3) the organisation, and (4 and 5) the practice of risk selection in MNC in the Netherlands. We used the following research questions: 1. How is risk selection conceptualized internationally? 2. Why and how did midwives in the Netherlands gain their position as gatekeepers to specialist care? 3. How is risk selection in the Netherlands organized? 4. How do maternal ethnicity, age, parity, socioeconomic status, and gestational age inform risk selection regarding induction of labour? 5. What is the variation in the content of Dutch hospital protocols used in risk selection for neonatal referral to the paediatrician? Outline of this dissertation Chapter 1: Understanding risk selection The lack of clarity about risk selection impedes the evaluation and comparison of models of risk selection (MRS) across various settings. The results of a scoping review into the concept of risk selection are presented in Chapter 1. We systematically searched the scientific literature, examining papers spanning the last four decades. We used inductive thematic analysis to identify key dimensions of risk selection. The results of this study contribute to a shared conceptualisation of risk selection. Chapter 2: The history of risk selection Deciding on a more effective strategy for providing the most appropriate care and care 27 1 GENERAL INTRODUCTION

provider requires an understanding of the arguments underlying the design of a system of risk selection. The existing system of risk selection in the Netherlands was based on the work of the 1987 Study Group for the Revision of the Kloosterman List (SGKL), a committee comprised of representatives from all professions and organisations involved in Dutch MNC. This committee formally granted community-based primary care midwives the authority to autonomously conduct risk selection. Until now, there was no thorough explanation of how and why the SGKL granted midwives this authority. In Chapter 2, we examine the system of risk selection used in the Netherlands in its historical context. We analysed the minutes of the SGKL’s meetings and conducted interviews with eight key-informants who were involved in the SGKL’s decision process. We used theories of professional jurisdiction and cultural theories of risk to analyse the factors that played a role in redefining the division of tasks and responsibilities in risk selection. The results of this study offer insight into how the authority to undertake risk selection is negotiated at the political and regulatory level. Chapter 3: The organization of risk selection The lack of knowledge about the best way to organize risk selection has opened the door to experiments with different MRS in the Netherlands. As a result of experiments with MRS, obstetricians in some regions became routinely involved in the assessment and decision making process, although their tasks, responsibilities, and authority varied considerably. In other regions, primary care midwives’ tasks, responsibilities and authority were extended to women with a high-risk profile. Little is known about these MRS. In chapter 3, we studied the MRS using a nationwide survey amongst all primary midwifery care practices and obstetric departments. The results of this study offer an in-depth understanding of MRS that is necessary to compare and evaluate risk-based selection systems. Chapter 4 and 5: The effectiveness of the practice risk selection Lack of knowledge about the effectiveness of the practice of risk selection hinders efforts to optimize the practice of risk selection. We studied the effectiveness of the practice of risk selection by researching the risk factors that inform the selection process and the protocols that are used to support that process. The results of our studies offer insight into the ways the accuracy of risk selection can be improved. Parity and age are increasingly used as single indicators for induction of labour of otherwise healthy women. In chapter 4, we studied the effectiveness of this risk selection practice by analysing whether consideration of additional maternal characteristics associated with perinatal mortality in addition to gestational age in the selection process would more accurately identify births that would benefit from induction of labour. We examined perinatal mortality rates for each gestational week at term, by maternal ethnicity, age, parity, and socioeconomic status, investigating the interaction between these 28 1 CHAPTER 1