Implementing and expanding safe abortion care: An international comparative case study of six countries

We conducted a comparative case study‐based investigation of health sector strategies that were useful in expanding or establishing new abortion services. We selected geographically diverse countries from across the human development index if they had implemented new abortion laws, or changed interpretations of existing laws or policies, within the past 15 years (Colombia, Ethiopia, Ghana, Portugal, South Africa, and Uruguay). Factors facilitating the expansion of services include use of a public health frame, situating abortion as one component of a comprehensive reproductive health package, and including country‐based health and women's rights organizations, medical and other professional societies, and international agencies and nongovernment organizations in the design and rollout of services. Task sharing and the use of techniques that do not require much infrastructure, such as manual vacuum aspiration and medical abortion, are important for rapid establishment of services, especially in low‐resource settings. Political will emerged as the key factor in establishing or expanding access to safe abortion services.


| INTRODUCTION AND BACKGROUND
Since the 1994 International Conference on Population and Development (ICPD) established the importance of universal access to sexual and reproductive health, 1 many countries have worked to increase access to obstetric care and contraception. Many have made similar efforts to add safe abortion to family planning services, either by reforming restrictive abortion laws to allow services or by strengthening programs to promote broader access to existing services. 2 In many places, this has expanded abortion services or introduced them into the formal healthcare system. The expansion or establishment of a national safe abortion program is complex, involving development of regulations, technical guidance, and clinical protocols; training providers; obtaining medications and equipment; and disseminating information to raise awareness among service providers and users. This process requires the sustained coordination of a broad array of governmental and nongovernmental actors. 3 While the ICPD emphasized rights, it is also well documented that increasing women's access to safe abortion is a key step in reducing maternal mortality from unsafe abortion. Yet the liberalization of laws and policies cannot achieve this goal unless sustained efforts are made to make services widely available. However, little is known about implementing abortion services at the health system level. Some articles have described the context and the actors involved in abortion reform in specific countries, including Uruguay, 4 Colombia, 5

and
Nepal. 6 Some studies employed in-depth interviews to identify successes and problems in specific country contexts. 4,7 However, these studies were not designed to inform the development of a practicebased guide for the implementation of health services. One multicountry report did offer a framework for the successful implementation of abortion services; it relied primarily on a review of published and "grey" literature although it included some interviews with key informants. 8 Several articles describe specific difficulties in guaranteeing equitable access to services in South Africa, Zambia, and India, and the interventions designed to address these problems. [9][10][11] Comprehensive reviews of a single country's program, including health system level implementation steps, successes, and difficulties exist for India, 12 Mexico City, 13 and South Africa, 14 but there are no comparative data to guide future program implementation. A recent five-country case study developed to inform the WHO guideline on task sharing found limited documentation of the implementation processes undertaken. 15 The objective of the present comparative case study-based investigation is to provide a systematic description of the health sector's roles in the implementation of new abortion laws. We focus on strategies and elements of service design useful for those expanding or establishing safe abortion services.

| MATERIALS AND METHODS
We employed a multiple case study methodology, which is ideal for answering open-ended questions that deal with processes, individuals, institutions, organizations, and events. Questions such as "how" and "why" deal with operational links that need to be traced over time; they are explanatory, and optimally addressed by triangulating multiple sources of information, as is typical of the case study method.
Multiple case studies start with an in-depth analysis of each case, but also establish parallels and describe differences between cases, thus leading to lessons that have wider applicability. 16 We considered countries for inclusion in this comparative study if they had implemented new abortion laws or policies, or changed interpretations of existing laws or policies, within the past 15 years. Within this group, we then selected a diverse geographic distribution of countries across the human development index (HDI), as we expect workforce constraints to disproportionately affect low HDI countries, and implementation steps to be more complex in countries with weaker health infrastructure.

| THEORETICAL FRAMEWORK
Various theoretical frameworks and models have been developed to guide implementation processes, to understand what influences outcomes, and to evaluate implementation efforts. 17,18 We drew primarily on the Integrated Promoting Action on Research Implementation in Health Services (i-PARIHS) framework, which posits successful implementation to be a function of the innovation to be implemented and its intended recipients in their specific context, with facilitation as the "active ingredient" aligning innovation and recipients. 19 In this study, we use the i-PARIHS framework to ground the broad steps of abortion program implementation and to analyze each country's case. For example, the innovation construct acknowledges that each country has its own law with its particularities and must then develop its own rules, regulations, and technical guidelines. Recipients of the innovation could include public health officials, medical societies, healthcare providers, and ancillary staff that will enact the innovation, as well as nongovernmental organizations (NGOs), advocacy groups, and the women who are the intended beneficiaries of the new abortion services and whose characteristics and access to information may vary widely from one country to another. The facilitation construct includes specific strategies and actions that "activate" implementation and can be understood more broadly to comprise the ways in which political will is translated into concrete interventions. Facilitation is the key to each country's success, and encompasses the main people, decisions, and actions from which we can draw lessons applicable to other countries. We use this i-PARIHS framework to organize and discuss the findings in this article and in the six case studies that follow. 19

| RECRUITMENT AND DATA COLLECTION
For each country case, we relied on two types of data sources. First, we conducted an in-depth survey of each country's health system and legal landscape related to abortion and systematically reviewed published and unpublished data, including the WHO's Global Abortion Policies Database, and used this desk review to prepare for in-country fieldwork. Then, in collaboration with in-country partners, identified through the WHO and Global Doctors for Choice networks and investigator contacts, we identified key stakeholders and experts in the field, and organized a series of in-country, semistructured, in-depth interviews. We developed an interview guide from the findings of the literature and desk reviews, and tailored it for each country context (Supporting Information Table S1). In each country, we interviewed 8-13 respondents, including healthcare providers, public health and government officials who had been involved in establishing or expanding the service, academics, and members of NGOs and legal and feminist advocacy groups. Respondents provided written informed consent and were guaranteed confidentiality. Several respondents from each country served as in-country coauthors, in doing so giving up their anonymity as participants of the study, although no quotations provided as respondents are directly attributed to them. Interviews lasted 1-2 hours, were recorded, translated into English where required, and transcribed, with identifying details removed. Interviewees who had functioned as in-country partners also served as coauthors of the case studies.

| DATA ANALYSIS
Data analysis comprised a multistep iterative thematic analysis.
All authors read a common selected subsample of transcripts, and together created a codebook, identifying main groupings based on occurrence both across and between countries. The codebook domains were structured to follow the i-PARIHS framework and refined until all authors agreed on the final categorization of codes. 19 Open coding was performed by DB-P and JMSG using Dedoose, a qualitative software program. 20 Analysis involved ongoing dialogue on code application, and some double-coding, both measures intended to check bias and differential code application. Intercoder reliability varied between countries but remained above 80%. Code application frequencies were used to create a weighted outline of important themes in each country, and relevant excerpts were drawn from these outlines.

| LIMITATIONS
We restricted our focus to health sector implementation following legal change over a 15-year timeframe and only briefly provide the legal and sociopolitical contexts, although we recognize that each country's story was specific and complex. Our sample is subject to selection bias as respondents were recruited through the networks of the WHO, Global Doctors for Choice, and the investigators. This connective tissue of shared professional contacts and affiliations may have inflated concordance among interviewee perceptions of abortion implementation, both within and between countries. Because our focus was on implementation, we did not specifically seek opinions from those opposed to abortion, and such individuals may have viewed the successes and difficulties of implementation differently than those interviewed. The consensus in the findings regarding both useful practices and difficulties from differently situated respondents in varied countries reassures us that these findings have utility, particularly for countries about to embark on establishing or expanding abortion services.

| CONTEXT
Study contexts varied culturally, geographically, politically, and economically; nonetheless, several commonalities are worth noting. Despite wide variation in maternal mortality rates, the proportion attributed to unsafe abortion was disturbingly high across the board 21-27 and led to active participation from the national health sectors. Religion has been a significant force in all six, and it continues to be associated with antiabortion stigma and refusal of care based on conscience (referred to hereinafter and in all case studies as conscientious objection). [28][29][30][31][32][33] However, while conservative and religious opposition did contribute to prolonging the time to legalization in Uruguay, 4 and did lead to compromises in other countries, such as maintenance of abortion in the criminal code and authorization of conscientious objection, 28,[34][35][36][37][38] such opposition did not prevent the legal and programmatic changes that ultimately led to expansion of access to safe abortion. Another notable similarity is that liberalization of abortion regulations arose in periods following emergence from dictatorship, conflict, or other significant political change, such as the collapse of apartheid.
In Ghana, Ethiopia, Portugal, and Uruguay, the impact of unsafe abortion on maternal mortality spurred the national health systems to establish or expand legal, induced abortion services to reduce maternal morbidity and mortality. Although varied in their design and human resources capacity, the national health services are key providers of abortion services in these four countries.
In contrast, the legalization of abortion services in South Africa and Colombia was embedded within human rights discourses.
While abortions also accounted for a high proportion of maternal mortality in these countries, the fight for legalization was framed in terms of women's rights in the post-conflict and post-apartheid contexts 39,40 and the health sectors were less centrally involved in advocacy leading up to legal change or the subsequent establishment of abortion services. [41][42][43][44][45][46]

| INNOVATION
We identified two main frameworks used to advocate for legal change and effectively position abortion services within the healthcare system. One was to focus on the public health necessity of reducing maternal death and morbidity-a tactic utilized in all our case countries. 27,31,33,47,48 In South Africa, proponents of the new law used both women's rights and equity arguments as well as those surrounding maternal health. 40 Colombia creatively straddled the two positions by framing abortion in terms of health as a human right. 39 The chief innovation, by definition, was legal change to permit abortion on a woman's request with no requirement for justification, or broadening interpretation of application of existing legal grounds.
While the South African law imposed the fewest additional steps for women seeking the procedure, 40 the most important guarantor of timely access was the Portuguese stipulation that the National Health Service had to ensure provision of the procedure within 5 days of the woman's request. 49 Liberal interpretations of existing grounds comprised broad application of maternal health, inclusion of mental health as a health condition in Colombia, Ethiopia, and Ghana, [50][51][52] and acceptance of the woman's word that she had been sexually assaulted in Ethiopia and Ghana. 51,52 The second strategy located abortion within a comprehensive reproductive healthcare package. Prior to the official expansion of abortion services discussed here, many countries with high levels of morbidity and mortality from unsafe abortion had agreed to the necessity of the harm reduction approach involving postabortion care (PAC).
Provision of PAC enabled them to skirt the issue of providing legal and safe abortions and to focus on salvage after harm had been done.
The new comprehensive package combined safe abortion services, the provision of contraception, as well as PAC. [52][53][54][55][56] Another innovation was to expand the categories of health workers eligible to provide the procedure, as the delivery of safe abortion services can be limited by a shortage of trained, willing providers. This is particularly problematic in regions with high prevalence of unsafe abortion and associated mortality, but also important in contexts with subnational disparities in socioeconomic status, geographic distribution of services, and access to private and public healthcare services. 32,57,58 In 2015, WHO released guidelines offering recommendations for task sharing and task shifting, so that a broader range of health workers could provide abortion-related care. 59 Task sharing was introduced in the three lower HDI countries (Ghana, Ethiopia, and South Africa), where scarcity of physicians is a crosscutting concern.
In the three higher HDI countries, in contrast, task sharing was limited, and physicians (in Colombia) or more specifically obstetricians/ gynecologists (in Uruguay and Portugal) maintained authority over the provision of abortion services. Ghana and Ethiopia have developed new "cadres" of clinicians to address physician scarcity, many trained to provide first-trimester abortions through medication or manual vacuum aspiration (MVA). 51,52 In Ethiopia, Integrated Emergency Surgical Officers are trained to provide second-trimester procedures. 52 Closely related to task sharing was, for first-trimester abortions, the widespread use of low technology procedures requiring little infrastructure: MVA and medical abortion using combination mifepristone-misoprostol. Midwives, nurses, and other cadres of providers newly included under the task-sharing rubric, were readily able to learn and employ these techniques. 52,56,57,[60][61][62][63][64] However, in two of the higher HDI countries (Portugal and Uruguay), the Ministry of Health also decided to provide medical abortion almost exclusively. This decision was made for similar logistical reasons, as medical abortion requires less infrastructure and training, and for political reasons, as it was anticipated that there would be less provider resistance to medical rather than surgical abortion. This focus on medical abortion allowed for quick and widespread establishment of services across the national territories. In other countries, MVA had been the method used initially to train providers; medical abortion was added subsequently.
In all six countries the public sector is supposed to assure that abortion care is free, has nominal charges, or is covered by insurance. However, for various reasons, private provision (including by NGOs) in Colombia, Ghana, Portugal, and South Africa is still significant and may be costly. 40,48

| RECIPIENTS
The key recipients were similar in all six countries, although their respective roles varied. The Ministry of Health/National Health Service was necessarily involved and convened such critically important allies as medical societies and NGOs to formulate technical guidelines, training, and certification. In the African countries, this collaboration also included international agencies and bilateral donors.
The Ministry of Health/National Health Service provided leadership and ownership of the new programs and sought technical guidance and support from other essential sectors. In Ethiopia, Ghana, Portugal, and Uruguay this proved key to the successful establishment of the program. In South Africa, the role of the Ministry of Health has been inconsistent, according to our informants. Responsibilities for implementation have devolved to provincial health departments there, with resulting uneven performance, varying according to local political will and resources. 58,65 Similar challenges were encountered in Colombia. 66 For the most part, physicians and their professional associations were supportive and played significant roles. In Portugal and Uruguay the obstetricians/gynecologists had been important advocates in the period leading up to legal change and were poised to provide the care, obviating the need for much additional training or certification. 47,54 The professional medical societies collaborated in creating clinical protocols and designing training and certification in five of these. Their technical content and endorsement proved essential. Participation by the nursing and midwifery societies was key in those countries engaged in task sharing, although some of these were apprehensive about possible loss of turf and they encountered some resistance from members concerned about workload and associated remuneration. 31,67,68 In two of the African cases, informants reported that the UN agencies, WHO and UNFPA, had been important partners and that their rigorously developed guidelines for abortion services, including medical abortion, as well as for task sharing had proved highly useful. The Latin American countries studied also drew upon the WHO guidelines, which they modified. South Africa is the outlier here as there are still no official guidelines. UNFPA also funded clinical training, which was often provided by Ipas in the African countries.
In Africa, international NGOs were critically needed partners who contributed technical content for clinical protocols and training programs, and/or provided abortion services, and MVA kits. Some interviewees expressed concern that reliance on international NGOs would not prove to be a sustainable strategy and considered National Health Service responsibility to be essential to foster ownership and avoid reliance on nongovernmental players, whose priorities might change. Indeed, in many places, National Health Service responsibility for service provision continues to be partially achieved by contracting out to NGOs, with Colombia and South Africa relying significantly on provision by national and international NGOs, respectively. 61,69 Uruguay is the only one of these countries that limits abortion to facilities that belong to the integrated national health system; there is significant private provision alongside public sector provision in all the other countries. 4 Other civil society groups were significant as well. Some were health oriented, such as the Portuguese Family Planning Association, Iniciativas Sanitarias in Uruguay, and Oriéntame and ProFamila in Colombia. In some countries, legal advocacy organizations and activist feminist groups who had been involved in pressuring for legal change stayed involved in oversight and public education. Since the goal in all these cases was to increase access to safe services, cost becomes a potential deterrent. The provision of free or low-cost care, or insurance covered care reduces this barrier, while the concomitant persistence of fee-charging private services complicates it.
Use of the public health framework also facilitated adoption of the service and lessened opposition. While monitoring and evaluation are standard cornerstones for assessing programmatic efficiency and effectiveness, they proved particularly useful in assuaging fears about these new programs. In countries as different as Ghana and Portugal, data demonstrated the rapid decline in abortion associated death that followed availability of safe procedures. 24,26,31,40,74,75 While many complained that data were overly aggregated or problematically coded, precluding fine-tuned analysis, interviewees also reported that they learned from these data how to constructively modify the program, and had been able to refute incorrect stereotypes about characteristics of those seeking abortion.
Closely related to this public health framing was to present abortion as one component of a comprehensive reproductive health package. In Ghana and Ethiopia, this built directly on the previously established PAC approach. In Colombia and Uruguay, multidisciplinary approaches to options counseling were readily integrated into the discourse on women's sexual and reproductive rights. In all countries, these strategies defused some of the opposition by aligning abortion with counseling and immediate provision of contraception.
As previously described, the use of medical abortion was an important facilitator in several ways; because it requires far less training or infrastructure, it was readily introduced into low-resource settings. It also proved more palatable to ambivalent clinicians in high-as well as low-resource settings, as the patient's role in taking medication and the avoidance of instrumentation obviated some of their discomfort. Furthermore, its use enabled the rapid establishment of services in high-resource settings such as Portugal and Uruguay.
Moreover, use of medical abortion, as well as of MVA, facilitated task-sharing. This proved highly significant in Ethiopia and Ghana where the chronic lack of clinical personnel due to limited resources and the "brain drain" were exacerbated by clinician reluctance to perform abortion, and invocation of conscientious objection as a means to avoid doing so. 43,76 The expanded group of willing clinicians resulting from task sharing partially remedied this dearth of providers.

| REMAINING CONCERNS
Stigma prevails everywhere and causes suffering and consequent avoidance of sanctioned legal care. This in turn leads women to delay in obtaining abortions, or to resort to clandestine extra-legal, often unsafe alternatives. It is associated with conscientious objection, both because a woman refused care by an objector experiences heightened stigma, and because some clinicians claim to be objectors to avoid experiencing stigma themselves, rather than because of profound moral conviction. 77 Conscientious objection, in turn, compounds the lack of providers and impedes access. 11,28,31,36,78 The South African case illustrates the devastating impact on access that can accompany unregulated conscientious objection. 11,69 In contrast, Ethiopia does not permit individual conscientious objection within the public health system, and there is discussion of stricter regulation in Ghana. 37,52 The Portuguese model shifts responsibility from the individual to the societal level; the National Health Service obligation to provide an abortion within 5 days of a woman's request is accomplished by funding a clinician to travel, or for the patient to travel if no willing clinicians are available. 49,79 Restrictive laws encumber access and perpetuate stigma. Five of the six countries in this study require multiple consultations and approvals, and/or permit abortion under limited specified circumstances. Although many respondents supported the integration of counseling regarding the abortion decision and subsequent contraceptive use, some expressed concern if the requirement for counseling was structured so as to delay the abortion. All of the mandated consultations and approvals can deter women from seeking legally available care or impose delays. 4,36,51,80  Second-trimester care remains inadequate, with too few physicians trained and willing to provide services. While task sharing can mitigate some provider shortages, appropriate referral services must be available when advanced providers are needed to perform later and/or complicated procedures. 81,82 While medical abortion has greatly facilitated the introduction of abortion services as described, some have expressed concern that patients often do not have a choice of method. Others consider method choice to be a secondary concern compared to the benefits of widespread availability and safety.
However, some gynecologists continue to use sharp curettage for uncomplicated first-trimester procedures, which has not been standard of care for many years. 3,83,84 It has proven particularly difficult to engage or monitor private physicians and those operating outside of the National Health System.
The cost of private provision remains a barrier. Moreover, the lack of understanding of the legal grounds for abortion that prevails among many women and clinicians leads some women to expensive private or uncertified providers, some of whom may offer dangerous alternatives. 4,69-71

| LESSONS LEARNED
Political will emerges as the key factor, common to all six countries, in establishing or expanding access to safe abortion services. While we will reprise here the lessons learned about useful ingredients to facilitate service provision, based on the key informant interviews, none of the details matter as much as a Ministry of Health or National Health Service determined to provide safe abortion care. This is consistent with the position of the WHO that: "Ideally, leadership would be situated in the ministry/department of health or another institution with the mandate to influence and mobilize national action". 3 South Africa demonstrates the paradoxical counterexample-it has the least restrictive and most rights-based law of the six countries examined, and yet the most limited and problematic implementation. 40 With political will as the precondition, the other components that proved useful in this sample of six countries include those discussed below.
Framing the need for safe services in public health terms proved to be the most strategic, least contested way to engage the sectors needed for implementation as well as the general public. In Uruguay, harm reduction functioned as a transitional public health strategy to prepare the health system for eventual legal reform.
Many of the clinicians interviewed had become supportive of liberalized access to safe abortion because of their own devastating experiences with death and severe morbidity in women driven to illegal and unsafe abortion. Will they convert this commitment to concern for women's dignity and autonomy as the fraction of mortality attributable to unsafe abortion drops as the new programs replace unsafe with safe procedures? Does Colombia's theoretical model of health as a human right offer a way to meld the public health and rights approaches?
Allied to this is the conceptualization of abortion as one component of a comprehensive reproductive health package. The bundling of services positions prevention (counseling, contraception, and other related sexual and reproductive health care) alongside abortion, with PAC as the service of last resort. 85 Effective implementers drew on the expertise of a broad range of stakeholders including international agencies and NGOs, country-based health and women's rights organizations, and medical and other professional societies. The Ministries of Health were able to garner the technical wisdom of these diverse sectors, which encompassed clinical care, training, values clarification, and public education. The NGOs were also able to mobilize their respective members in support of the new program. Oftentimes, after the initial rollout, these groups later pressured the Ministry of Health to maintain and improve services. While there have been tensions when NGO services are more highly resourced than government services, all sectors agreed that the success of these collaborative efforts depended on strong leadership by the Ministry of Health/ National Health Service and its clarity as to their respective roles.
The guidelines of the WHO recommend comprehensive monitoring and evaluation of safe abortion services. 3 Such programs are necessary and should fall within the purview of the Ministry of Health.
When surveillance is insufficient or data overly aggregated, at times NGOs and universities have stepped in to remedy these gaps. Data demonstrate the improvement in outcomes following the institution of safe services and can refute misconceptions, as well as guide program improvement.
Task sharing and the use of low-technology techniques of abortion, such as MVA and medical abortion, are particularly important in lowresource and rural settings, consistent with previously reported findings. 8,51,52 They also facilitate rapid implementation and are thought to be more widely accepted among providers. However, they do not obviate the need for a systemic guarantee of a smooth transition to care by gynecologists when necessary. The need for second-trimester procedures and the associated training and infrastructure must be anticipated and included in the design of services.
Cost can always be a deterrent; providing free or low-cost public sector services is essential to increase uptake. Both women and clinicians need to be well informed about the legal landscape and care options so that women avail themselves of safe, high-quality services.
We have learned of many specific ways of implementing expanded access to safe abortion care. We hope that countries about to undertake similar efforts can learn from these experiences and adapt those measures that seem relevant and potentially useful. The bottom line, however, is clear: political determination to provide safe care and stop women's deaths and suffering is the essential ingredient.

AUTHOR CONTRIBUTIONS
WC: Developed initial proposal and interview instrument, conducted Ghana interviews, drafted manuscript, provided Ghana findings, and collated edits and reviews. BMS: Contributed toward initial