Ectogestation ethics: The implications of artificially extending gestation for viability, newborn resuscitation and abortion

Abstract Recent animal research suggests that it may soon be possible to support the human fetus in an artificial uterine environment for part of a pregnancy. A technique of extending gestation in this way (“ectogestation”) could be offered to parents of extremely premature infants (EPIs) to improve outcomes for their child. The use of artificial uteruses for ectogestation could generate ethical questions because of the technology’s potential impact on the point of “viability”—loosely defined as the stage of pregnancy beyond which the fetus may survive external to the womb. Several medical decisions during the perinatal period are based on the gestation at which infants are considered viable, for example decisions about newborn resuscitation and abortion, and ectogestation has the potential to impact on these. Despite these possible implications, there is little existing evidence or analysis of how this technology would affect medical practice. In this paper, we combine empirical data with ethical analysis. We report a survey of 91 practicing Australian obstetricians and neonatologists; we aimed to assess their conceptual understanding of “viability,” and what ethical consequences they envisage arising from improved survival of EPIs. We also assess what the ethical implications of extending gestation should be for newborn and obstetric care. We analyze the concept of viability and argue that while ectogestation might have implications for the permissibility of neonatal life‐prolonging treatment at extremely early gestation, it should not necessarily have implications for abortion policy. We compare our ethical findings with the results of the survey.


| EC TOG E S TATI ON AND VIAB ILIT Y
The idea of an artificial womb has been extensively explored over the last century in science fiction, literature, and film. 1 Philosophers and ethicists have also written on this topic, with articles in both academic literature and the mainstream press exploring the ethics of gestating embryos and fetuses outside the female body. 2 However, discussion has taken a more practical turn following the publication in 2017 of a report on a technique for supporting extremely premature newborn lambs for up to 4 weeks in a "Biobag." 3 In the landmark study from Philadelphia, lambs were delivered by cesarean section at a level of lung maturity equivalent to humans at 23 weeks gestation (17 weeks early). Blood vessels in the umbilical cord were connected rapidly to a low resistance oxygenator circuit, which provided artificial intravenous nutrition. The lambs were supported within a sealed fluid-filled bag, the artificial amniotic fluid continuously exchanged to prevent infection. Eight of 13 lambs were sustained for 20-28 days using this system. 4 While the technology for "full ectogenesis" (undertaking the full pregnancy outside the human body) appears to be a long way off, the Philadelphia study suggests that it may soon be possible to support human fetuses for part of a pregnancy. This technique is sometimes referred to as partial ectogenesis, but we will use the term "ectogestation," since this refers more accurately to a period of external gestation rather than external creation.
It has been proposed that ectogestation could be offered to extremely premature infants (EPIs) to improve outcomes. 5 Some of the most premature infants have high rates of mortality and morbidity with existing forms of intensive care. 6 Their immature lungs are easily dam-aged by mechanical ventilation, and gas exchange may be difficult or impossible. 7 If a baby who is about to be born extremely prematurely (because of preterm labor or severe maternal illness) could be transferred into an artificial womb for a period of time, this may allow sufficient growth and maturity to substantially improve their health outcomes.
The development of ectogestation would have potential ethical implications for both obstetrics and neonatology. These include questions arising directly from the implementation of ectogestation-for example, regarding how the therapy should be evaluated, when or if randomized trials involving humans would be ethical, when the benefits would outweigh the risks of such a therapy, and the implications for resource allocation. 8 However, the development of ectogestation could also generate ethical questions through its potential impact on viability.
Viability can loosely be described as the ability of a fetus or infant to survive independently of its pregnant mother, although the precise definition is contested. 9 We will return to the definition in Section 3, but the basic idea is that prior to a certain point in pregnancy fetuses are not yet viable as they would not survive if they were delivered. However, beyond the point of viability, fetuses may be liveborn and may survive. 10 Many medical decisions during the perinatal period are based on the gestation at which infants are considered viable. For example, resuscitation and intensive care are considered if newborn infants are viable, whereas these are usually withheld if the newborn is not yet viable. In many jurisdictions, abortion is permitted in the first half of pregnancy, but termination of pregnancy is not legally permitted 1 Smajdor, A. (2007). The moral imperative for ectogenesis. Cambridge Quarterly of Healthcare Ethics,16(3), 336-345. 2 Ibid; Cannold, L. (1995). Women, ectogenesis and ethical theory. Journal of Applied Philosophy, 12(1), 55-64;Aristarkhova, I. (2005). Ectogenesis and mother as machine.
Retrieved from: https ://www.thegu ardian.com/lifea ndsty le/2017/sep/04/artif ical-womb-women-ectog enesis-baby-ferti lity.  In Section 2 of the paper, we report a detailed survey of obstetric and neonatal medical specialists, those most likely to be aware of changes in the outcome of EPIs, and to be regularly encountering questions relating to neonatal resuscitation and abortion. The aim is to investigate professionals' conceptual understanding of "viability," and what ethical consequences they identify arising from improved survival of EPIs using existing or future technology. In this paper, we will sometimes use the term "resuscitation" as short-hand for "resuscitation and provision of intensive care/life-prolonging treatment." However, we do not mean to restrict discussion to simply initial delivery room management of extreme preterm infants.  1471-0528.2009.02228.x. 21 There are multiple other papers that discuss health professionals' views on viability and fetal anomaly. These have not been included as the focus of this paper is on morphologically normal fetuses and the evolution of viability during pregnancy.

| Participants and procedure
An online survey was developed to determine the views of specialist doctors on questions relating to medical treatment at the borderline of viability. Participants were asked to consider several scenarios relating to neonatal and obstetric management of preterm labor and EPIs in the light of current and future technologies.
Before distribution, the survey was piloted on a group of medical students and doctors in obstetric/neonatal practice. Suggestions were incorporated into the final survey.
The study participants consisted of specialist doctors (consultants or registrars/fellows in obstetrics and neonatology) practicing in Victoria, Australia. Victoria has a population of 6.4 million, with approximately 82,000 births per year. 22 Neonatal intensive care is provided within a small number of specialized centers in metropolitan Melbourne. Abortion law differs between states in Australia. In Victoria, under the Abortion Law Reform Act 2008, abortion is permitted at the request of a woman up to 24 weeks. Past 24 weeks, two medical practitioners must agree that "abortion is appropriate in all the circumstances." 23 Participants in the four Victorian specialist children's/maternity hospitals with tertiary or quaternary neonatal intensive care units were contacted by email during June-July 2018.
We identified leading neonatologists/obstetricians in each institution who forwarded the invitation to colleagues regularly providing care to EPIs or their mothers. Non-responders were sent two reminder emails. The questionnaire was anonymous. In return for completing the survey, participants had the opportunity to enter a draw to win a book voucher.

| Design
The survey was conducted using the online platform, Qualtrics. It consisted of four main sections (Appendix A: Figure A1). Section A explained the purpose and structure of the survey (including details of ethics approval) and asked potential participants to confirm their status as medical specialists and consent to the use of their data.
Section B invited participants to reflect on current techniques for the treatment of women in extremely premature labor and any resulting infants. According to whether participants identified as specialists in obstetrics or neonatology, they were given questions with slightly different wording. For example, "Would you support Caesarean section" (for neonatologists) versus "Would you offer Caesarean" (for obstetricians). The responses were analyzed together. The initial questions related to a scenario with a woman in preterm labor at a non-specified gestation (the fetus was otherwise normal) and participants were asked about the lowest or highest gestation that they would be willing to provide or support Caesarean section, abortion, resuscitation or non-resuscitation (if the infant was born in a fair condition). They were then presented with scenarios (in random order) involving women in premature labor at gestations 22+3, 23+3 and 24+3 weeks. For each gestation, they were asked to estimate the chance of survival if resuscitation was attempted, and select their level of agreement (on a Likert scale from strongly agree to strongly disagree-see Figure 1) with statements that the infant was viable, that it was in its best interests to be resuscitated, and that they would support termination of pregnancy, resuscitation or non-resuscitation at parental request.
Participants were asked about the definition of viability. They were asked separately about how each of the following elements were relevant to their understanding of "viability": the proportion of surviving infants, dependence on technological support for survival (including whether or not technology needed to be available to the treating team for an infant to be regarded as viable), and the presence/absence of disability. Respondents were then asked to indicate their level of agreement with statements about recent developments in neonatal intensive care and how these advances impact on viability and the provision of various medical interventions.
Section C introduced the concept of ectogestation by briefly describing the science behind this technique and invited participants to consider a hypothetical scenario where this technology had been shown to yield 75% survival with no or mild disability in 75% of surviving infants when applied at 22+3 weeks (

F I G U R E 1 Example of survey question
Please indicate how much you agree or disagree with the following statements. All participants who consented to the survey and answered at least one question were included in the analysis for the questions that they responded to, regardless of whether they completed the whole survey. Statistical analysis was carried out using RStudio Versions 3.5.1 and 3.5.3 25 . The findings were descriptively presented as frequency (% of respondents for each question) for discrete variables and mean (standard deviation) for continuous variables. We generated graphs and tables to summarize these descriptive results. Graphs were generated using the package "ggplot2." 26 The association between medical specialty (obstetrics versus neonatology) for the Likert-scale questions were examined using chi-square tests for the difference between two proportions, by comparing those who agreed or disagreed, merging the "Strongly" and "Somewhat" categories, and discarding the "Neither agree nor disagree" responses. Chi-square tests were also used to compare the proportion of obstetricians and neonatologists who selected particular gestations in Section B, compared to those who selected any other gestation including free-text responses. We considered ordinal logistic regressions as an alternative to chi-square tests to eliminate the need to merge categories. However, we ultimately decided against this strategy due to the sparsity of the data making the proportional odds assumption difficult to check.
To compare the responses based on medical specialty for continuous variables, the t test was used. We elected not to examine for possible relationships between demographic characteristics and survey responses due to the small numbers in the study.

| Sample
We contacted 143 professionals for the survey. There were 91 respondents who answered at least one question of the survey: 50 neonatologists and 41 obstetricians (response rates: 63% and 65%, respectively). Most respondents were between 30 and 60 years old.
Two-thirds were female. Sixty-five percent were consultants, and 59% had at least 7 years of experience working with extremely premature infants (Table 1).
Thirty-seven percent of respondents identified as being religious; 70% of these identifying as Christian. Religious respondents varied in the importance of religion to their lives (Table 1).
Most participants were moderately (36%) or strongly (49%) prochoice, whilst some identified as strongly pro-life (2%), moderately pro-life (10%) or undecided (2%). There was no significant difference between obstetric and neonatal respondents in their demographic characteristics, religion or views on abortion.

| Viability
Most surveyed doctors related the concept of viability to it being possible for an infant to survive at a given gestation (67%), indicated that the presence of disability was not relevant (78%) and included survival with medical intervention (100%) (

F I G U R E 2 Ectogestation scenario
In 2017, a paper in Nature described a technique that involved supporting extremely premature newborn lambs in a liquid environment outside the uterus for a period of up to four weeks (https://www.ncbi.nlm.nih.gov/pubmed/28440792). The lambs were delivered by caesarean section at a level of lung maturity equivalent to ~23 week gestation human infants. Blood vessels in the umbilical cord were connected rapidly to a low-resistance oxygenator circuit, which also provided artificial intravenous nutrition. The lambs were supported within a sealed fluid-filled bag (a "Biobag"), the fluid continuously exchanged to prevent infection.
For the following questions, please assume that this technique has been evaluated in humans and proven to improve mortality and morbidity rates for extremely premature infants. Please also assume the costs of this technique are equivalent to other forms of neonatal intensive care.
Imagine that this technique if applied to infants at 22+3 weeks gestation has been shown to yield 75% survival, with no or mild disability in 75% of surviving infants.
A mother has gone into extremely premature labour at 22+3 weeks gestation. She has a dilated cervix and bulging membranes and delivery is thought to be imminent.
Participants were asked for their agreement with statements about viability and developments in medical practice. Seventy-six percent of doctors agreed that the gestation at which an infant is considered viable had changed in the last 10 years. Most agreed that improvements in neonatal intensive care in the last decade changed how they felt about resuscitation of 23-week infants (63%), although obstetricians were more likely to disagree than neonatologists (46% vs. 13%; χ 2 = 10.7, p < .01; Appendix A: Figure A2). A minority of respondents (25%) indicated that these advances changed how they felt about abortion being offered to infants of the same gestation. Note: Note that aside from professional specialty, answering demographic questions was optional. a Not all percentages add to 100% due to rounding.
The majority (53%) disagreed with the statement that laws on abortion should change as a consequence of changes in the viability of EPIs, while 24% neither agreed nor disagreed, and 24% agreed.
Most participants identified 23 or 24 weeks as the lowest gestation they would support Caesarean section for fetal reasons; however, neonatologists preferred lower gestations (71% selecting 22 or 23 weeks compared to 30% of obstetricians χ 2 = 15.2, p < .01).
Obstetricians varied in the highest gestation that they would offer abortion at parental request when preterm birth is imminent.
Although 47% of those who selected a gestation chose 24 weeks, eight chose "other." Estimated survival chances for infants born at 22+3, 23+3 and 24+3 weeks are presented in Appendix B: Table B1. For each gestation, the mean estimate was higher for neonatologists than obstetricians. For 22+3-week infants, responses ranged from 0% to 75%.
A large proportion (94%) of respondents agreed that a 24+3- week infant was viable (Figure 3). There was less agreement at 23+3 weeks. Sixty-nine percent of neonatologists and 89% of obstetricians did not believe that a 22+3-week infant was viable. Overall, obstetricians were less likely than neonatologists to indicate that the hypothetical newborns were viable.
There was little consensus when it came to supporting termination of pregnancy: for each gestation presented, responses ranged from strongly agree to strongly disagree ( Figure 4). Overall, participants showed decreasing support of termination with increasing gestation.

| Ectogestation
Eighty-eight percent of doctors agreed that in a hypothetical example of preterm labor and possible ectogestation at 22 weeks, they regarded the infant as "viable." See Figure 5 whether doctors supported not using ectogestation if parents did not wish to use this technology was a point of contention, with 54% agreeing that they would support non-provision of life-prolonging treatment, 32% disagreeing and the remaining 14% neutral. Forty-nine percent disagreed with abortion being an option in this scenario, and 13% neither agreed nor disagreed. Neonatologists were more likely to disagree than obstetricians (63% vs. 32%; χ 2 = 7.1, p < .01).
Most respondents (61%) indicated that Caesarean in this scenario should be optional at 22 weeks; there was clear disagreement (93%) with a statement suggesting that this practice should become mandatory (Appendix A: Figure A3).
Participants were shown several statements regarding their general attitude towards ectogestation (Appendix A: Figure A4). Over half neither agreed nor disagreed that this technology should become common practice (55%), but 33% thought it should. Overall 41% of respondents agreed that this technology would influence their views on abortion being performed at 22 weeks, while 48% disagreed. There was uncertainty as to whether abortion law should change if ectogestation were to lower the age of viability; however, TA B L E 2 Respondents' selections of conceptual elements of a definition of "viability"

Number of respondents a (%)
The proportion of infants who survive It is possible for infants to survive if born at this gestation 57 (67) The majority (>50%) of infants born at this gestation will survive 24 (28) The vast majority (>80%) of infants born at this gestation will survive 2 (2) obstetricians were more likely to disagree with this idea compared to neonatologists (54% vs. 31%; χ 2 = 4.4, p = .04).

| Discussion of empirical findings
This survey asked Victorian obstetrics and neonatology specialists to consider questions relating to medical decision-making at the borderline of viability. It is the first empirical study of neonatologists'/ obstetricians' conceptual understanding of viability. It is also the first survey of professionals' views of ectogestation, and its potential impact on medical practice.
Key findings of the survey were that: doctors appeared to define viability differently to how they applied the concept; respondents indicated a belief that ectogestation would shift the gestational age of viability; they were divided as to whether ectogestation should become common practice; and they were unsure whether this technology should result in restrictions in access to abortion. The survey was limited by its small sample size (n = 91) and geographic setting (all respondents being from Victorian hospitals). Our study asked doctors' views but did not provide an opportunity for participants to explain their reasoning (for example, we did not ask their views on consistency or the moral status of the fetus). Qualitative research would be helpful for providing further insight into doctors' views on ectogestation, and the decision-making process at the borderline of viability.

| Viability
When asked specifically about the criteria for the concept of viability, doctors appeared to support a view that the gestational age of viability should reflect the possibility of survival, regardless of disability. This contrasted with how the doctors applied the concept of viability. The clear majority did not believe that a 22+3-week infant is viable (although they acknowledged that survival is possible at this gestation).

| Ectogestation
Doctors who we surveyed agreed that ectogestation (in a hypothetical case example) would change the point of viability ( Figure 6).

| Termination of pregnancy
The question of whether to offer termination of pregnancy to a woman in premature labor at her request proved controversial. When presented with the 22+3-, 23+3-and 24+3-week scenarios, for each case some obstetricians and neonatologists strongly agreed with termination of pregnancy, whilst others strongly disagreed. There was also a range of responses relating to how the concept of viability maps onto abortion, and whether viability should have any influence on this practice. In comparison, the majority of physicians in the Canadian study would "rarely" or "never" offer post-viability termination of pregnancy in the absence of a "lethal" fetal abnormality. 31 In our study, obstetricians were more likely to support termination of pregnancy than neonatologists. Differences in attitudes between obstetricians/neonatologists may reflect different professional experience, as well as differences in perceived professional role: an obstetrician has a primary responsibility to the woman, a neonatologist to the infant.

| E THIC AL ANALYS IS
The above results provide some insights into the way that practicing obstetricians and neonatologists understand and apply the concept of viability in the light of medical advances. However, these results, while valuable, do not settle the ethical question of how viability should be applied to neonatal and obstetric decisions. We will first, briefly, analyze the concept of viability. We will then examine the ethical implications of changes in viability for medical practice, drawing on one key ethical argument used in debates about viability. nosis is sufficiently poor that resuscitation is not an option (the "Lower Threshold"). In between these two thresholds, resuscitation and intensive care provision is seen to be optional. The Lower Threshold in developed countries is usually regarded as being at 22-24 weeks (varying between countries). 34 It is thus close to, but not coinciding with Absolute Viability ( In obstetrics, viability is invoked in some countries' abortion law as both a hard limit after which abortion is not legal (notably in the United States 37 ), and implicitly as the underpinning of a limit, without being expressly referred to. In some jurisdictions, viability is not ex- It is worth noting first that many on both sides of the abortion debate reject viability as significant for the permissibility of abortion.

| Viability
Those with strongly pro-life views argue that an early embryo has full moral status and therefore would disallow termination of pregnancy even before the point of viability; those who are strongly pro-choice often believe that the ethical significance of a woman's autonomy means that termination should be an option beyond viability, or may argue that even the late-term fetus lacks moral status.  abortion policies use cut-off points prior to viability. It is beyond the scope of this paper to discuss the ethical basis for such policies; however, changes in viability would not be expected to influence policy in those countries.

Gestational age
Absolute Viability (AV) The youngest gestation with a known survivor with or without disability anywhere in the world 21 weeks and 4 days a Median Viability (MV) The gestation at which >50% of infants will survive with or without disability with the medical intervention available to them Somewhere between 23 and 25 weeks in a neonatal intensive care unit in a developed country b Median Intact Viability (MIV) The gestation at which >50% of infants will survive, and those infants will be free from major disability with the medical intervention available to them Around 26 weeks c , although depends on definition of major disability (and also varies with intensity of interventions provided and how often treatment is withdrawn because of predicted disability) Natural Viability (NV) The gestation at which >50% of infants will survive (with or without disability) in the absence of major medical intervention (e.g., mechanical ventilation and intensive care) This definition is hard to apply to high-income countries as after a certain gestation it would be regarded as unethical to not provide intervention. For the sake of this paper, we will set aside arguments that viability confers moral status, 43 to focus on the argument from con-

| Reflective equilibrium
It is potentially useful to re-examine our survey responses in the light of the above analysis. We described four different theoretical definitions of viability. Our survey respondents understood viability in the abstract as reflecting the possibility of survival (Absolute Viability); however, when asked specifically whether an infant at a specific gestation was "viable" they appeared to use a higher threshold, (possibly reflecting Median Viability or Median Intact Viability). It may be that they were answering the latter question with the Upper Threshold (and abortion) in mind.
Our respondents appeared to anticipate that ectogestation would shift the Lower Threshold for neonatal resuscitation. In the hypothetical case scenario of an infant at 22+3 weeks gestation, the majority endorsed the provision of intensive care with ectogestation, if parents desired this, although few supported treatment at this gestation with current technology and outcomes. However, very few professionals indicated that Caesarean section for ectogestation should be mandatory, implicitly accepting that ectogestation itself would be ethically optional. Survey respondents were divided in their views on whether ectogestation should influence laws on abortion; however, a majority of obstetricians did not think that the law should change, which might be a reflection of support for the argument that we have developed above.
Alternatively, it may reflect a view that access to abortion should not be linked to viability.

| CON CLUS IONS
This paper sought to explore the ethical implications of ectogestation. We have focused on the significance of ectogestation for viability, since this technique could make it possible for extremely premature infants to survive outside the uterus who would previously have been unable to do so.
We have combined empirical and analytical approaches to examining the implications of ectogestation for viability, neonatal resuscitation and abortion. We surveyed practicing medical specialists in Victoria on ectogestation and medical decision-making at the borderline of viability. Surveyed doctors appeared to apply the concept of viability in a way that was different from their theoretical understanding. Professionals believed that ectogestation would shift the gestational age of viability; however, they were divided as to whether this technology should become common practice and were unsure whether it should result in restrictions in access to abortion.
Our ethical analysis has clarified the concept of viability. We suggested that ectogestation would alter the Lower, but not the Upper Threshold for neonatal resuscitation and provision of life-prolonging treatment, increasing the potential permissibility of resuscitation at extremely low gestational age, but not making it mandatory.
For abortion, we concentrated on the argument from consistency.
Our aim was not to defend this argument; nevertheless, if consistency is the ethical basis for abortion policy, our analysis suggests that ectogestation would not necessarily warrant changes in cut-off gestations for abortion.
Moving forward, qualitative research would be useful to understand the reasons behind professionals' views on viability, abortion and newborn care. This would assist in informing further ethical deliberation and reflective equilibrium on ectogestation and related advances in neonatal care.

ACK N OWLED G M ENTS
We wish to thank Leon Di Stefano (Walter and Eliza Hall Institute) for his help with statistical analysis, in particular, for providing the code used to generate the Likert plots included in this paper. We are also grateful for the invaluable support for recruitment in our His particular interests are bioethics, the psychology of perinatal care, communication and language use in counseling.