Beyond measurement: the drivers of disrespect and abuse in obstetric care

Abstract Concerns about disrespect and abuse (D&A) experienced by women during institutional birth have become critical to the discourse on maternal health. The rapid growth of the field from diverse points of origin has given rise to multiple and, at times, confusing interpretations of D&A, pointing to the need for greater clarity in the concepts themselves. Furthermore, attention to measurement of the problem has been excessive when viewed in relation to the small amount of work on critical drivers of disrespect and abuse. This paper raises some key issues of conceptualisation and measurement for the field, puts forward a working definition, and explores two critical drivers of D&A – intersecting social and economic inequality, and the institutional structures and processes that frame the practice of obstetric care. By identifying gaps and raising questions about the deeper causes of D&A, we point to potentially fruitful directions for research and action.


Introduction: an evolving field of inquiry
Women's experience of institutional childbirth has garnered unprecedented global attention in recent years. Neither research nor policy advocacy are, of course, new to this field. In particular, feminist researchers and advocates have been conducting studies and raising issues for some time. [1][2][3] The new burst of current interest is at least partially influenced by systematic engagement by the World Health Organisation (WHO), especially its Department of Reproductive Health Research. [4][5][6] Triggered by a 2010 landscape analysis 7 and supported by a 2014 statement by the WHO, 4 the issue of disrespect and abuse (D&A) during intrapartum care across low and middle-income countries (LMICs) has now sparked new empirical research across different continents, an advocacy agenda and a growing number of interventions. [8][9][10][11][12][13] Nonetheless, this is still an evolving field of inquiry and action with uncertain definitions, varying terminology, the absence of agreed validated tools and measures, and many unanswered questions. Definitional and measurement uncertainties can be partially attributed to ongoing conceptual fuzziness. Lack of consensus on terminology ("obstetric violence" versus "disrespect and abuse" versus "mistreatment", "humanised childbirth" versus "respectful maternity care") is probably the result of distinctly different points of origin and influence of the field's protagonists.* While understandable, attention to measurement and terminology has been excessive when viewed in relation to the small amount of work on critical drivers of disrespect and abuse. 14,15 Uncovering and exploring these drivers is, we believe, essential to deeper understanding, more focused policy initiatives, as well as to more meaningful measurement.
*Disrespect and abuse are the terms we choose to use for reasons that are outlined later in the paper. This paper begins by raising some key issues of conceptualisation and measurement for the field at present. It develops a working definition, and then goes on to explore two critical drivers of D&Aintersecting social and economic inequality, and the institutional structures and processes that frame the practice of obstetric care. By identifying gaps and raising questions about the deeper causes of D&A, we point to potentially fruitful directions for research and action.

Concepts and language
Definitions and terminology For a field that has been growing at such a rapid pace, there have been surprisingly few attempts to actually define the phenomenon being studied. An early definition of D&A: " … any form of inhumane treatment or uncaring behaviour toward a woman during labor and delivery" 11 appeared open to including multiple aspects of the problem but too broad for sustained application. Another attempt 16 is more nuanced in including both women's experiences and agreed norms for what may be "humiliating or undignified" treatment but may perhaps be insufficient in how it handles actually harmful obstetric practices.
The surprising lack of a clear, agreed conceptual definition may be due to two major reasons. The first is the presence of a diversity of stakeholders with varying perspectives on what is important. Stakeholders range from feminist activists and advocates concerned over violations of women's human rights and bodily integrity, to clinicians and health workers focusing on their conditions of work, responsibilities for pregnancy outcomes in often difficult institutional contexts, and their professional expertise. Stakeholders also include those responsible for setting norms and standards (for effective, ethical and quality obstetric care) at national and international levels, who may be concerned to meet broad development goals such as the Sustainable Development Goals and their target on maternal mortality, and to ensure that the concerns of all stakeholders are treated in an evenhanded manner. The second reason derives from the first: the field, as a result of stakeholder diversity and even antagonism, includes a mix of women's subjective perceptions, inter-personal behaviour, objective medical practices and normative standards that have been difficult to bring together. Depending on who is addressing the issue, different aspects are included or excluded, and what the phenomenon is called varies considerably.
Various terms have been used to describe the poor treatment and care women receive from providers in institutional settings. These have included humanised childbirth, obstetric violence, disrespect and abuse, respectful maternity care, and mistreatment, to name the most common. But what do different terms really mean and how are they understood? How should terminology be evaluated to understand what each term may be useful for? While each term in itself may appear to be neutral and open, judging what is the most useful terminology needs to situate the term in the historical context from which it has arisen, how it is actually being used by its protagonists, and what gets excluded, included, shaded, downgraded or ignored in the process.
A look at the field's origins highlights the evolution of terms. As early as the 1980s, concerns about the overmedicalisation of maternal care in Latin America coalesced into a movement to "humanise childbirth" 1,17,18 in institutions and to address violence in routine obstetric practice. Led largely by feminist academics, health practitioners and activists in Brazil, it drew support from other social movements on women's and indigenous peoples' rights. By the 1990s a strategic regional network had been established with a focus on reducing routine use of ineffective, non-evidence based and potentially harmful obstetric interventions. Action and research have articulated how such obstetric practice undermines women's autonomy and intrinsic childbirth capabilities, unnecessarily interferes with the physiological process of labour, and can result not only in unpleasant birth experiences but also poorer birth outcomes. 17,18 As the issue gained political traction, the term "obstetric violence" 1 was introduced and began to be recognised in laws on violence against women and by statutory human rights bodies. 19,20 The obstetric violence discourse understands violations of women's human rights as being enmeshed with institutional obstetric practice, recognising the power inequality inherent in patient-provider interactions, and the inequities that underpin differences in experiences of marginalised versus more empowered groups.
Grounded strongly in reproductive rights and powerful for advocacy, 'obstetric violence' is the preferred terminology in Latin American feminist work. Within a broader Venezuelan law on violence against women, it is defined as … the appropriation of the body and reproductive processes of women by health personnel, which is expressed as dehumanized treatment, an abuse of medication, and to convert the natural processes into pathological ones, bringing with it loss of autonomy and the ability to decide freely about their bodies and sexuality, negatively impacting the quality of life of women". 19 But researchers and advocates also use the earlier terminology of 'humanised childbirth' strategically, cautioning against the antagonising effect that the language of obstetric violence can have on the community of clinical practitioners with whom they must engage if change is to take hold. 1,18 The word violence typically carries with it intentionality, which is not a consistent marker for abuse in the context of obstetric care. For instance, providers may claim that preventing a baby from going into distress may be the reason for using forceful fundal pressure during a prolonged second stage of labour. They may not intend to cause harm. Nonetheless, the action may be abusive because it may not be in the best interests of the woman or the baby.
In the obstetric violence discourse, the routine or excessive use of certain obstetric practices is itself seen as the problem because it prioritises the needs of institutions and providers above those of women. In our ongoing qualitative work on D&A in institutions, we have found several examples, such as labour tables being arranged so that women's pelvises face entryways, thus offering no privacy to women but greater convenience to providers to monitor labour progress. This has made us probe the ways disrespect is inherent to institutionalised obstetric practice.
The recent discourse on D&A moves beyond the language of obstetric violence to speak more directly to health system contexts where inadequate access to comprehensive obstetric care is still a challenge to maternal survival, and material and human resource shortages are endemic. This D&A discourse has emerged from the institutional imperatives of norm-setting organisations such as the WHO that are committed to safe deliveries as the way to reduce unconscionably high rates of maternal mortality and morbidity. Following the work of d'Oliviera et al. 21 and Castro and Erviti 2 describing disrespect and abuse, Bowser and Hill 7 proposed the terminology and early typologies for D&A, problematising it as a barrier to institutional skilled birth attendance.
While feminist researchers continue to use the language of D&A, WHO has shifted to the term "respectful maternity care" (RMC), framing the issue as a long-neglected component of healthcare quality, while also recognising it as a violation of human rights. More recently, it has been argued that "abuse" should be replaced by "mistreatment", 6 a term that is not only more acceptable and less provocative, but further separates the issue from individual intentionality and links it to the realm of healthcare quality.
Respectful Maternity Care presents a more positive approach to the agenda. This may be legitimate in order to investigate the subject while obtaining buy-in from providers. Noting that the mere absence of D&A is not enough, RMC advocacy has been important in spelling out women's entitlements to care using human rights covenants to establish what standards ought to be. 10 But the terminology of mistreatment and RMC do not capture abuse when it is intentional and bordering on violence, such as the deliberate denial of pain relief. 21,22 Human rights instruments by themselves have limited capacity to identify underlying causes or effective levers for change.
In sum, the language of D&A and obstetric violence galvanise women but may be less helpful when it comes to conducting research or investigating the problem with providers. Women resonate to it, providers do not. Respectful Maternity Care, on the other hand, is more inclusive and less hostile to providers, but may also exclude important elements of intentional abuse, and may obfuscate underlying causes.
For our purposes of being both inclusive and incisive, we prefer the terminology of D&A despite the above limitation for provider buy-in. In the context of obstetric care, we define disrespect as the violation of a woman's dignity as a person and as a human being on the basis of her economic status, gender, caste, race, ethnicity, marital status, disability, sexual orientation, or gender identity. Disrespect is often revealed in the biased normative judgements that health workers make about women and the resulting acts of omission or commission. Abuse refers to actions that increase the risk of harm to the woman and are not in the best interests of her health or well-being. Such actions may be learned and reproduced through the practices of institutional medicine. They may or may not be intended to cause harm and are often justified by resource constraints that can become a cover for prioritising the convenience of health providers over the well-being of the woman.
We identify three important advantages to this definition. It captures both intentional behaviours and unintended consequences. It is open to addressing institutionalised medical practices as well as socioeconomic inequalities. And it allows us to identify both manifestations and underlying drivers of the problem.
This definition appears to meet the criteria spelled out by Vogel et al.: 6 "Any definition needs to adequately capture the health, human rights, legal and sociocultural dimensions of this problem. It should consider a range of possible acts (whether intentional or not), the risks (or potential risks) of harm or suffering to women, and that these events can occur in different levels of care."

Measurement
Disrespect and abuse is particularly challenging to define and capture empirically. Most empirical work has captured forms and manifestations of D&A using predefined typologies drawn from the initial landscape analysis 7 or those expanded and refined by a subsequent mixed-methods systematic review by Bohren et al. 23 A conceptual challenge of the existing evidence-based typologies is that they include a combination of interpersonal interactions, standards, and health system constraints, which are difficult to aggregate or even compare. They may serve as a useful checklist for further research, but we believe it may be premature and difficult to interpret prevalence measures based on aggregating them. Variations are high and difficult to explain. For instance, across prevalence studies, self-reported D&A (of any type) ranged from 20% to 98%. 24,25 Further, 94.8% reported non-consensual care in one Ethiopian study 26 versus 0.06% in a Tanzanian study. 27 Efforts at original tool and scale development are currently under way, which are discovering context-specific types of D&A from the ground up. [28][29][30] A second challenge for empirical measures is the difference between results based on selfreporting of D&A versus observations. Most current research has used women's perception as a starting point, capturing their and other community members' attitudes through qualitative research [31][32][33] and using their recall of experiences along predefined domains in quantitative surveys. [24][25][26][27] But what is considered disrespectful or abusive by women will shift based on personal, social and cultural expectations, as well as levels of awareness of human rights. Such an approach relies on what women see and don't see, and are satisfied with or not. It misses what women may experience but may not perceive to be D&A. This is particularly important in contexts of stark inequalities and power imbalances, or if D&A takes the form of routine or non-evidence based obstetric practices that women may not know about or be able to question. Freedman and colleagues 16 consider the role of normative standards and the implications of normalisation by women and providers in unequal societies and resource-strained health systems to present a framework for D&A that works toward greater recognition of the problem by all stakeholders.
Given the limitations of self-reporting, the importance of observation is increasingly being recognised, yielding higher reports of D&A compared to women's self-reports, and revealing important insights on what gets normalised by women. 34 But observation too has limitations. While measures have been taken to reduce the Hawthorne effect, it may still play a role in modifying provider behaviours. There are also difficult ethical challenges related to bystander intervention when documenting D&A for the purposes of understanding prevalence.
In the next section of the paper we focus on two key drivers of D&A: socioeconomic inequalities and institutional structures and processes.

Socioeconomic inequalities
The role of socioeconomic inequalities in underpinning women's experiences of obstetric care has been recognised. Feminist literature has documented abusive interactions between women and providers, situating them within broader gendered inequalities. Through this lens, how women perceive, internalise or justify the poor care they receive reflects entrenched gender discrimination and oppression within homes, communities and wider societies. 35 Women's requests for attention, comfort or pain relief during labour are not merely ignored or met with anger by overstretched healthcare providers; they are often ridiculed because notions of women's entitlements among providers can be virtually non-existent. 22 Govender and Penn-Kekana 36 found gender bias and discrimination to be particularly acute in sexual and reproductive health service provision. It is not surprising then that in contexts of considerable gender inequity, labour and birth become sites for moral judgements about women's sexuality: versions of the taunt "you enjoyed it then and you complain now" have been documented across time and diverse contexts. 2,21,22,33,37 Yet, gender does not operate alone: it cross-cuts other socio-economic inequalities. Intersecting inequalities may not only influence how providers perceive and interact with particular categories of women in labour, but also impinge upon their clinical decision-making and obstetric practices. Diniz et al. illustrate patterns of excessive obstetric intervention along socioeconomic lines. They found that poorer, racial and ethnic minority women who accessed public services were more likely to be given unnecessary episiotomies, while their predominantly richer, Caucasian counterparts who accessed the private sector were more likely to receive caesarean sections. 38 Poor women who overwhelmingly make up the patient profile of public teaching hospitals in many countries, may also be disproportionately subjected to unnecessary procedures such as multiple pelvic examinations for the purposes of teaching. 39 The structures and processes that give rise to such manifestations of D&A are not adequately studied. In this section, we hypothesise four overlapping routes through which socioeconomic inequality exerts its influence on D&A.
i. Historical evolution of bias First, interactions between the members of different socioeconomic groups, and between particular groups and the State, tend to be coloured by their collective memory of prior relationships and transactions. These histories and the power inequalities produced through them are not fixed for all time. Experiences of marginalisation and discrimination are continually reproduced or redefined, as groups negotiate the webs of power that define their lives. Based on ethnographic work in the conflict-riddled Indian state of Assam, Chattopadhyay argues that marginality is graded and modifiable as relationships between different groups and with the State evolve over time. 40 Histories of social relationships may seem unlinked to the practice of obstetrics; but they are not. They shape women's interactions with healthcare providers. Histories of conflict between particular groups and the State can engender a sense of mistrust and/or resistance to medical advice and instruction in public (state-run) facilities. 40 They can also foster, among healthcare providers, a set of beliefs, biases (both implicit and explicit) and prejudices about how particular groups behave and "must be handled". 37 Behavioural responses to such perceptions can create additional vulnerabilities for the women who seek obstetric care.
Socially and economically marginalised groups are known to concurrently suffer poorer access to maternal care and be more likely to face discrimination when they seek it. 41 While conducting verbal autopsies of maternal deaths in a deprived district in southern India, we found considerable complexity to provider bias. 42 For instance, a nurse's perception of how unpleasant a labouring woman smelled was attributed to unbridled sexual behaviour particularly associated with the woman's caste. In Smith-Oka's ethnographic research in Mexico, she suggests that urban middle-class physicians' use of the term ranchitos to describe their rural-poor obstetric patients "is often impregnated with meaning related to skin color and degree of indigenous heritage", combined with hostile beliefs about their sexuality and fertility. 37 Across Latin America and in India, systematic documentation of religious, ethnic and racial minority women's interactions with providers speak of the "triple burden" they face when seeking institutional childbirth. 41,43,44 Poverty, fertility and gender can form a powerful axis for discrimination in maternal health service provision, which may be further layered with racial, ethnic, religious, caste or other biases. Prejudices against certain categories of women (multi-gravid or obese women, women with histories of repeated abortions or HIV) can seep into the fabric of healthcare organisations, making disrespectful interactions less of a random individualcentric act and more of an operational norm. Our analysis of the parturition registers in public teaching hospitals and municipal corporation-run maternity hospitals in Karnataka (south India) revealed careful recording of unnecessary personal information and pejorative labels about the women who had sought care. Solnes-Miltenburg et al. found nurses dismissing the gestational age provided by women in Tanzanian hospitals, recording their own often inaccurate estimates instead. 45 Reports of women delivering without any assistance in facilities has often followed women voicing their readiness to give birth, but providers not believing them. 30 ii. Normalising and tolerating D&A Second, disempowered or marginalised groups can get so accustomed in some contexts to being treated with disrespect, being forced to negotiate the health system on less favourable terms, or receiving worse care that they may fail to recognise or may normalise and tolerate disrespectful care. 14,41 Providers, in turn, may selectively engage in behaviours and practices that are not in the best interests of the patient, where the selection is biased or discriminatory. 14,41 These two processes can work together to create an environment in which disrespectful obstetric care thrives.
iii. Role of 'fallback positions' Third, notwithstanding the fact that healthcare institutions tend to privilege the needs and convenience of providers over women (this is discussed in the next section), not all women experience D&A or to the same degree. Women who have stronger "fallback positions" 46 deriving from their caste-class or other positions, social capital and/or familiarity with institutional services, are better able to navigate healthcare institutions and negotiate respectful care. 47 iv. Linguistic and cultural barriers to communication Breakdowns in communication can become a fourth mechanism through which social inequalities impinge upon D&A. Communication is severely tested when healthcare providers do not speak the same language as the women who seek their care. In our ongoing qualitative research, we found that health workers, who had to ask people who knew the language of migrants to serve as impromptu translators, often sensed that their instructions were not delivered in the spirit in which they were intended. But communication can work at cross-purposes even when all parties share the same language. Communication theorists point to the culturally coded ways in which the same language can be used to imply different things. 48 These barriers can create distance between healthcare providers and women. Even well-intentioned providers get frustrated when women and their attendants cannot fully understand their medical advice or instruction. In Latin America, approaches to culturally adapt institutional birth practices have their roots in early humanisation efforts 18 and continue with more recent programming to respond to consistently higher mortality and lower rates of institutional birth among indigenous groups. 49 The care provided to indigenous people who are often at the lower ends of social and economic hierarchies tends to be non-evidence-based, risky and even harmful, including physical immobilisation, lack of privacy, multiple vaginal and cervical manipulations, routine episiotomy, and fundal pressure. 41,50 It is also often rife with cultural bias and insensitivity, making the experience of childbirth physically and emotionally traumatic by forcing indigenous women who are used to delivering in squatting or other positions to deliver lying down, not allowing companionship, or insisting on unauthorised discharge of the placenta, which can carry deep cultural and spiritual significance. 41 Little wonder that they resist going to health services. Training providers in indigenous languages and incorporating safe traditional birth practices comprise some of these measures, which may ease potential trigger points for D&A. But to ensure that care is truly respectful they must break down and tackle deep-rooted biases that providers may hold against the very communities they are mandated to reach.
Policies to institutionalise childbirth targeting groups that have hitherto endured inequitable access need to be more cognizant of these complexities; and D&A research should be similarly attuned to capture and explain differences in experience.

Institutional structures and processes
In this section we explore how D&A is reproduced through the institutional structures and processes of the health-care system. We first examine the feminist critique of the system, and follow this by focusing on: teaching, training and ongoing medical education along with the organisational dynamics of health care provision; the challenges of resource scarcity in relation to adherence to standards and protocols; and questions of accountability.
Organised medicine has been critiqued by feminists as being gender-biased, and this is reflected both in the Latin American work discussed earlier, as well as research and advocacy by feminists in the North. 18,51,52 The field of obstetrics has particularly come under scrutiny as exemplifying how integral gender power and control are to the practice of medicine. A range of obstetric practices such as the persistence of the lithotomy position for birth or routine interventions that restrict women's mobility are seen as weakening women's agency and control over their own bodies. Obstetrics itself has been critiqued for not placing women at the centre. Instead, its procedures and methods appear to be organised to serve the convenience of providers, who are often male.
Moral judgements about who is a "good" or "bad" patient, and values about pregnancy and childbearing can be imbibed during medical training and expressed through disrespect, especially towards socially disadvantaged groups. These may be more easily identified as D&A. But there are other obstetric practices that simply prioritise the needs or convenience of providers or are justified on the basis of safety, and therefore may not be acknowledged as D&A. Diniz et al. 38 have argued that when practices cannot be proven to be intentionally abusive at an individual level, population rates can be used at institutional levels to determine whether the practice is in women's best interests.
In settings of stark social distance between patients and providers, both older 22 and emerging research documents women's near complete submission and compliance during the birth process. 14,30 In such contexts, not pushing as instructed or changing birth positions are viewed as transgressions that warrant castigation according to providers and sometimes women themselves. 22,30,33 As Bradley et al. 14 note, denying women social support during labour and birth can be another way providers try to maintain power and control and exert their authority over women. Such practices violate women's dignity even if their submissiveness pre-empts more extreme forms of intentional abuse. In the highly technical field of medicine, knowledge is power. Not providing information on the progress of labour is another way that providers may prevent women from being active participants in birth, and can assert power by controlling biomedical knowledge. 14,52 Useful as this critique is, feminists may have tended to overemphasise intentional gender bias as a driver of D&A to the detriment of understanding unintentional abuse and how such practices are learned and reproduced through teaching and training. Our experience in the field tells us that we need to understand the extent of the latter, and how abuse gets built into systems and structures even if unintentional at the level of the individual. Two aspects of the problem call for further research -first, the psychological concept of implicit bias wherein providers may be unconscious of their own deep-seated biases, 53 and second, how bias is built into systems and structures. In this paper we do not go further into the presence of implicit bias but explore how bias is built into the structures of medical education and training, and into the ways in which the provision of obstetric care is organised.
A number of questions come up from our work and the work of other authors in this Special Issue and elsewhere. 22,39,54,55 Do medical teaching (including internships and residencies) and training (including continuing education) at multiple levels include knowledge and sensitisation on gender, caste, and other forms of socioeconomic inequality, and how these work and must be countered? Or does the training itself reinforce preexisting biases or create new ones? How seriously and effectively are providers taught about ethics, responsibility and accountability? Are communications and language skills taught, especially when there is likely to be considerable socioeconomic distance between providers and pregnant women and their families? Can such central issues be left to be handled by continuing medical education, or is that likely to be too late as we ourselves tend to believe?
In our own ongoing work in India we have found that internships and residencies are where actual practical learning occurs. It is where attitudes are imbibed, procedures are practised, and where new doctors start learning whether and how corners can be cut in terms of adherence to standards and protocols. In the teaching hospitals that we have been studying, senior doctors and heads of unit are unquestioned authorities. But there is also considerable learning between older and newer interns, and between interns and older nurses. In both these ways, informal norms on "how things are done" or should be done are passed on to newer generations of providers, and the cycle of D&A continues.
The role of medical education and training in shaping the attitudes of providers may be well recognised, especially in high-income countries, 56,57 but it is inadequately explored in the context of D&A. In our ongoing work referred to above, we found that the teaching of empathetic and caring behaviour is relegated to informal moral discourse by clinical superiors, or that it may be selectively learned by observation if such role models exist. However, these areas of interpersonal patient interaction are never formally taught to undergraduate or postgraduate students nor evaluated during medical education and training. Similar observations have been documented elsewhere. 58 The process of dehumanisation in medical training has been the subject of critique as well, with calls for more research on emotion in medical training. 57 Obstetric violence advocates believe this translates very directly to a lack of compassion and empathy in the way doctors treat women during labour and birth.
Studies on nursing care, especially from South Africa, draw attention to the "othering" process built into nursing education, which in many countries has strong colonial roots, imbued with implicit instructions to civilise and morally instruct their patients. 22,59 In this context, the domain of caring can be seen as diluting their professional identities of being more educated, knowledgeable and skilled than the patients they serve. 22 Provider distancing as part of their medical education and training was flagged early as a potential contributor to D&A, 7 but little empirical work has followed in this area.
Cultural aspects of medicine and obstetrics can interact with organisational dynamics to exacerbate D&A. Cross-disciplinary work has highlighted nurses' and midwives' inferiority in medical hierarchy and lack of power within their own professional and organisational structures as contributing to their need to dominate and control even more disempowered patients. 22,36,59 At the institutional level, Madhiwalla et al.'s work in Indian hospitals 54 illustrates how providers' subservience to medical hierarchy can lead to the reproduction of obstetric practices endorsed by senior staff, even if they run counter to evidence and formal protocol. Women's subordination to institutional bureaucracy has also been highlighted by this and other studies, showing how lack of adherence to administrative rules such as booking times or carrying specific documentation can lead to harmful delays or even a complete denial of care. 14,22 Some behaviours and practices are triggered or worsened by resource scarcity within the health system, but others are not necessarily constrained by such pressures. Indeed, high patient loads may not be viewed negatively in public teaching hospitals, where poor women are viewed as the necessary "teaching material" for students to learn, to practice and even to experiment. 39 Obstetric violence advocacy has been very critical of this, highlighting cases of unnecessary and egregious bodily harm caused to women in order to serve teaching needs. 2,39 Tracing decades of multi-pronged approaches to humanise childbirth in Brazil, Diniz and colleagues 39 found that teaching hospitals are the most resistant to change in practice, and that a stronger emphasis on gender and ethics in medical training is a prerequisite to enabling an environment of respect in institutions. Cuban medical education involves specific training in public health ethics, community development and social responsibility which is noted to contribute to Cuban doctors' dedication to serving vulnerable populations and instilling a culture of compassion. 60,61 Much of D&A research has pointed to the stressors of under-resourced health systems in influencing behaviours, but there has been less published work on how policy and programme interventions generate specific behavioural incentives and disincentives, 62,63 aggravating or mitigating D&A. For instance, under wider financial reform to improve public sector accountability in South Africa, it was found that making nurses responsible for ward finances had negative impacts on such areas of care as the provision of pain relief to obstetric patients. 64 Research from Mexico and India 50,65 has shown that, in a context of active demand-generation for institutional maternal services through cash incentives to women, health providers respond to high patient-loads by engaging in painful and potentially harmful practices to hasten labour and expedite delivery. These include abuses such as manual cervical dilation, perineal stretching, cervical sweeping, and fundal pressure. While the policies themselves may be aiming for greater efficiency, they have pressured providers to respond to the unprecedented increase in facility deliveries in highly questionable ways that become routinised as informal norms. The result from women's perspectives is poor maternal satisfaction with the birthing process, challenges to bodily integrity and functionality, and even the baby's maturity at birth. If such results were used to measure effectiveness rather than efficiency, they would very likely give us a very different and more women-friendly understanding.
Cutting corners on meeting global or national norms and standards are often rationalised by providers as responses to resource shortages. They may even be justified by them as essential to the woman's survival, e.g. fundal pressure may be justified as necessary to save the woman's life in the context of maternal anaemia / stunting by preempting maternal exhaustion. While medical education allows a degree of flexibility based on the provider's judgement and prognosis, non-adherence to standards is intended to be the exception, not to become the rule. This rule of thumb is easier to follow in the case of the individual woman. But what if a large majority of women have the same characteristics? Or if resource constraints are long-standing and severe?
The answer has to lie in a set of built-in procedures for less-than-optimal but still acceptable solutions that are clearly spelled out by an authoritative body. Medical superintendents and providers in general need to be trained to be knowledgeable about this. This can be supplemented by requirements for reporting and supervision that ensure that the modification of standards doesn't remain where it is currently in all too many LMIC contexts: ad hoc, rampant, and unsupervised. How to guide and support institutions and providers to move up towards full compliance with standards must also be a central responsibility for health system authorities.
This leads directly to the challenge of accountability to ensure that systems and structures are set up in a manner that reduces D&A. At present the main accountability mechanism in many contexts is the maternal death review conducted in a variety of different ways. Evidence exists that punitive approaches can be self-defeating and less effective than supportive approaches such as those used by the UK's National Health Service 66 that foster collective learning from mistakes. But punitive approaches are still in use and may encourage practices not in women's interests such as late referrals without adequate safeguards or stabilisation, 66,67 if not outright prevarication in order to avoid blame and punishment. There are also few if any accountability mechanisms for maternal morbidity which is itself poorly measured, or for D&A. It is clear that a great deal more research, advocacy and policy development are needed in this regard. It would be a fitting area for the work of the International Accountability Panel of the UN's Every Woman, Every Child, Every Adolescent initiative. 68

Implications for research and policy
The recent WHO recommendations on intrapartum care for a positive birth experience 69 integrate interpersonal and clinical aspects of care, representing a much-needed and cohesive set of benchmarks against which to evaluate behaviour and practice. This will hopefully prompt more investigation into how these different dimensions of care interact to produce either positive experiences for women or D&A.
The argument of this paper is that standards such as these provide an essential basis of accepted norms in the progression towards genuinely women-centred care. Yet, much remains to be done to improve our collective understanding of the problem. Current research is heavily invested in measuring the burden of the problem and comprehensively capturing its manifestations. While these have useful policy benefits, attempts to delve deeper into an exploration of the causes and drivers of the phenomena are urgently needed. Such examination is vital to a fundamental understanding of why D&A is so pervasive and to effectively inform approaches to address it.
The absence of agreed definitions and understanding of the phenomena of inappropriate and harmful obstetric care is striking, even though measurement proceeds apace. We have in the paper put forward a definition that is, we believe, robust in addressing conceptual concerns and causal challenges, while being supple enough to be usable in multiple national and local contexts. But equally germane to the effective application of new standards is the fact that actual behaviours and practices in obstetric care are conditioned by social and economic inequalities, and by the structures and processes of institutionalised medicine. The paper draws on a selection of evidence to document and analyse how these two major drivers work to reproduce D&A. A synthetic focus on drivers is relatively new to the field, although some of the different elements have been highlighted before, particularly by feminist researchers and advocates concerned about gender biases in health care.
As D&A research is situated in a context of increasing institutional childbirth, much remains to be understood about how women's own histories and relationships with the health system, their capacity to navigate institutions, and provider biases against different socioeconomic groups play a role in how D&A is patterned. More empirical work is needed that investigates different forms of socioeconomic inequality and how they impact the extent and forms of D&A at different levels of care, as well as women's own perceptions and responses to them.
Tempting as they may be, easy policy fixes through training or dissemination of norms are likely to be limited in their efficacy. Training and capacity building are one side of efforts, but organisational culture also plays a big role. What kind of changes can instil a caring and respectful attitude? How can incentive structures be changed to ease the uptake of more women-friendly norms? Efforts to humanise childbirth emphasise the need to decentralise power through the way cadres of providers are organised. The idea is to enable physicians, obstetricians, midwives and nurses to work on a more equal basis, which in turn can foster a shift from enforced compliance to collaboration in interactions with patients. Is this applicable in different contexts? More research is also needed on how larger policies that impact the health sector influence the nature of obstetric practice and patient-provider interactions, and how D&A can be systematically countered through policy or programme changes.
Beyond the specific research and policy questions raised in the paper, a shift in orientation is needed. That shift stems from the recognition that social and economic inequalities interact in fundamental ways with the institutional scaffolding of medical education, learning and practice, to shape how providers behave and how women experience the consequences of that behaviour.

Résumé
Les préoccupations causées par le manque de respect et la maltraitance que les femmes connaissent lorsqu'elles accouchent dans un centre de santé sont devenues essentielles dans le discours sur la santé maternelle. La croissance rapide du domaine depuis divers points d'origine a donné lieu à de multiples interprétations, prêtant parfois à confusion, de ce manque de respect et de cette maltraitance, ce qui montre qu'il faut clarifier les concepts eux-mêmes. De plus, l'attention accordée à la mesure du problème a été excessive lorsqu'on la compare avec la petite somme de travail sur les facteurs essentiels du manque de respect et de la maltraitance. Le présent article soulève des questions clés de conceptualisation et de mesure pour ce domaine, propose une définition de travail et étudie deux facteurs essentiels du manque de respect et de la maltraitance : l'intersection des inégalités sociales et économiques, et les structures et processus institutionnels qui encadrent la pratique des soins obstétricaux. En identifiant les lacunes et en posant des questions sur les causes plus profondes de ce phénomène, nous indiquons des directions utiles pour la recherche et l'action.