Misjudging early embryo mortality in natural human reproduction

Introduction

In 2002, a judicial review was considered by the Honourable Mr Justice Munby of the Queen’s Bench Division (Administrative Court) to determine whether the supply of Levonelle (commonly referred to as a morning-after pill (MAP)) by pharmacists amounted, in principle, to a criminal act under section 58 and/or section 59 of the Offences against the Person Act 1861^1,a. Counsel for the claimant argued that the MAP could act by preventing the implantation of a fertilised egg in the uterus and that its use for this purpose therefore constituted an act intended to bring about a miscarriage. Counsel argued that the supply and use of the MAP for this purpose ought therefore to be regulated in the same way as surgically- or medically-induced abortion, as required by the Abortion Act 1967. A judgment in favour of the claimant would have had a significant social impact on the supply of the MAP.

The Claimant was John Smeaton, on behalf of the Society for the Protection of Unborn Children (SPUC). The Defendant was The Secretary of State for Health, and the two Interested Parties were Schering Health Care Ltd and the Family Planning Association.

Munby J. held that “the prescription, supply, administration or use of the morning-after pill does not – indeed cannot – involve the commission of any offence under either section 58 or section 59 of the 1861 Act”^b.

The case included evidence from scientific expert witnesses and in his judgment, which Munby J. conceded was “necessarily very long”^c, he described and scrutinised aspects of this evidence in detail.

The primary purpose of this article is to shed light upon one aspect of the scientific expert witness evidence, uncritically accepted and deliberately emphasised by Munby J., which was inaccurate and glaringly inconsistent: namely, the extent of natural embryo mortality in humans in the first week after fertilisation. The nature of the error, the inconsistencies in the expert evidence, and their sources are reviewed and explained.

Whether any legal implications arise from the inconsistent expert witness testimony and judicial error is for others to consider.

Expert witness statements

Copies of original expert witness statements used in this critique were made available freely on request from the archives of the Claimant. Civil Procedure Rules concerning the use of witness statements for other purposes indicate that where a witness statement has been put in evidence at a hearing held in public, its use is not restricted to the purpose of the proceedings in which it is served^d. Regarding subsequent use of disclosed documents: they may be used only for the purpose of the proceedings in which they are disclosed, except where they have been read to or by the court, or referred to, at a hearing that has been held in public^e. The substantive hearing was held in public and lasted three days, starting on 12^th February 2002^f. All witness statements referred to in this article were read and referred to by Munby J. No court order has been made restricting or prohibiting the use of these expert witness statements. Finally, it is in keeping with principles of open justice (and academic enquiry) that evidence placed before courts be available for public scrutiny, as confirmed in recent case law².

A full transcript (redacted of personal information) of all these statements is available on request from the author. In producing the transcript from original copies of court papers, every effort has been made to reproduce the content as accurately as possible. Errors and idiosyncrasies in spelling, grammar and style have been retained. Any errors of transcription are entirely the responsibility of this author, and will be corrected on notification. Where consent to publish has been obtained from the witnesses, full transcripts are directly available in the Underlying data³ as described in the Data availability section. Where consent to publish has not been obtained, only those passages quoted by Munby J. or directly referenced in this article are shown, the remainder being redacted. Transcripts are intended to enable readers to draw their own conclusions about the contents of the judgment and this article.

The scientific and/or medical expert witness statements were provided to the court by eight “very eminent medical experts”^g as summarised in Table 1. Munby J. commended submissions from all parties, both written and oral, as being “uniformly of the very highest quality”^h. However, although Munby J. stated that “they were all agreed as to the basic physiology”ⁱ, close reading indicates that, with regard to the extent of embryo mortality, this was not the case.

The incidence of natural human embryo mortality

Before proceeding, it will be helpful to summarise the key biological stages in normal early human reproduction: coitus, ovulation, fertilisation, embryo development and implantation. Munby J. does a commendable job of synthesising the biological evidence in paragraph 126 of his judgment. Coitus introduces sperm into the lower female reproductive tract and ovulation releases an egg into the upper female reproductive tract. Fertilisation occurs in the fallopian tube when sperm and egg meet and one sperm penetrates the egg: “This can be described as Time 0”^j. The fertilised egg develops and becomes a blastocyst after 5–6 days. Around day 7, the blastocyst begins to implant into the lining of the uterus. During implantation, the embryo produces human chorionic gonadotrophin (hCG), detection of which “represents the first reliable opportunity to identify the existence of an embryo”^k. Approximately 2 weeks after fertilisation, a woman will miss her menstrual period, the first clear external indication of the presence of a developing embryo.

Complex biologic processes do not work perfectly all of the time, including human reproduction. A recent re-analysis has concluded that pre-implantation embryo loss is approximately 10–40% and that total loss from fertilisation to birth is approximately 40–60%⁴. In addition, a review of scientific data that contribute to quantitative claims regarding natural pregnancy loss provides a detailed background against which claims made by the expert witnesses regarding the incidence of natural human embryo mortality may be evaluated⁵. Making sense of these numerical estimates is not easy. To aid understanding, Figure 1 summarises conclusions from these articles alongside the claims of the expert witnesses and numerical estimates from an influential and valuable analysis by Henri Leridon⁶. Details of how the figure was constructed are in the legend.

Figure 1.

Estimates of embryo survival from fertilisation until (A) birth or (B) four weeks after fertilisation. Numerical values derived directly from witness statements are shown as solid points. Open points have been inferred to facilitate graphical representation. Two sets of reference values have been included for comparison. The first set is derived from Table 3 of Jarvis (2016) by averaging probabilities from three independent studies. The second is from Table 4.20 of Leridon (1977). The figure clearly indicates that the extent of early embryo mortality obtained from Braude (PB/2) and emphasised by Munby J. in his judgment is substantially different from all the other witness statement estimates and those published by Jarvis and Leridon. Drife’s estimate for total pregnancy loss from fertilisation to birth is also excessive compared to the other values. The explanation for the large discrepancy in pre-implantation mortality is that Braude’s estimate is derived from in vitro laboratory and clinical IVF data, and not from natural reproduction.

Four of the eight expert witnesses provided numerical estimates and/or comment on the extent of human embryo mortality: Drife, Brown, Braude and McLean. It is in these quantitative claims regarding embryo mortality that an inconsistency in the evidence and a judicial error of interpretation is apparent. I shall consider each witness statement in turn and then explain the nature and source of the scientific error in Munby J.’s judgment.

Munby J.’s error

In paragraph 129 of Munby J.’s judgment there are two quantitative claims, both attributed to Professor Braude, that invite close scrutiny: (1) “not much more than 25% of successfully fertilised eggs reach the blastocyst stage of development”, and (2) “fewer than 15% of fertilised eggs will result in a birth”. Although Munby J. does not place these statements in quotation marks, they are taken directly from the book chapter submitted by Braude as evidence (Exhibit PB/2)¹⁷. In this chapter, the 25% claim is supported by a citation to a review on the early development of human embryos in vitro co-written by Dr Virginia Bolton and Professor Braude²³. A “<15%” claim is found in the same review.

Bolton & Braude’s review of the development of the preimplantation human embryo in vitro represents IVF as “a remarkably inefficient therapeutic procedure” and explores biological and technical reasons for the “unacceptably high rate of embryonic loss” associated with IVF treatment. Within this context they describe the work of Fehilly et al. (1985)²⁴ who…

197 expanded blastocysts developing from 784 embryos is 25.1%, i.e., “not much more than 25%”.

Fehilly et al. (1985) conducted their study primarily to determine whether embryos frozen at the cleaving stage (day 3–4 after fertilisation) or at the blastocyst stage (day 5–6 after fertilisation) would be more effective at producing subsequent pregnancies after embryo thawing and transfer into the womb. Based on the data described above, with regard to embryo development to the blastocyst stage, they commented that “only one quarter of them expanded in vitro”. This data is the source for Munby J.’s claim that not much more than 25% of fertilised eggs reach the blastocyst stage.

In the following paragraph in their review, Bolton & Braude offer one reason for the low survival rate of the in vitro embryos: “Suboptimal culture conditions are undoubtedly responsible for a proportion of this embryonic failure”²³. Elsewhere, Braude reports that “Experiments in our laboratories have suggested that the in vitro handling of oocytes can produce chromosomal aberrations at alarmingly high frequencies”²⁵ and strikingly, that “When Bob Edwards and indeed my own group were researching these early stages, blastocyst culture was awful (about 15% of embryos made it to that stage)”²⁶. More recently, Dr Bolton has repeated that “Embryo culture conditions in vitro are likely to be suboptimal compared with those in vivo”²⁷. Put simply, human embryos created by fertilisation in vitro did not, and do not fare well. Hence, the use of in vitro data to define the fate of natural embryos in vivo is both biologically and quantitatively risky⁵.

Munby J.’s second quantitative claim, that “fewer than 15% of fertilised eggs will result in a birth”, also comes from Bolton & Braude’s review via Braude & Johnson’s chapter. A further quotation from p. 93 makes the point:

Table I in Bolton & Braude provides a quantitative summary of the success of embryo replacement using data from seven clinical IVF units. The data are sub-divided according to whether 1, 2 or 3 embryos were replaced (i.e., transferred) into the womb of the woman undergoing treatment. The clinical pregnancy rate among the patients thus treated is reported as a percentage of embryos replaced as 12.7%, 12.1% and 9.6% respectively. These values are somewhat lower than 15% and clearly refer to clinical pregnancies. Nevertheless, this IVF data appears to be the source for the judicial claim that “fewer than 15% of fertilised eggs will result in a birth”.

There are only two other cited studies in the key paragraph of Braude & Johnson’s chapter (Exhibit PB/2)¹⁷. These are the study by Wilcox et al.¹⁵ and a paper by Professor Lesley Regan²⁸. Wilcox et al. conclude that “The total rate of pregnancy loss after implantation, including clinically recognized spontaneous abortions, was 31 percent” and a re-analysis of that data suggests that approximately 50% of the fertilised eggs in that study may have been lost up to birth⁴. Regan’s paper addresses the incidence of spontaneous abortion of clinical pregnancies with a focus on recurrent abortion. It contains preliminary findings from a prospective study of pregnancy loss and, interestingly, concludes that the “overall incidence of spontaneous abortion in this prospective study is considerably lower than those reported in previous studies (10.3% overall; 5.6% for primigravidae)”²⁸. Therefore, neither Wilcox et al. (1988) nor Regan (1987) is a credible source for Munby J.’s claim that “fewer than 15% of fertilised eggs will result in a birth”.

It is clear therefore that the two values emphasised by Munby J. in his judgment referred firstly, to embryo mortality in vitro, and secondly, to the survival of IVF embryos following their transfer into women as part of fertility treatment. It is surely reasonable to suppose that the context for the use of the MAP, subject to the judicial review, was not in vitro embryos in a laboratory or women undergoing fertility treatment, but rather naturally conceived embryos and women at risk of pregnancy. The values emphasised by Munby J. have no bearing on the case at all.

An understandable error?

It is charitable to assume that Munby J. did not realise that the figures he emphasised were biologically misrepresentative and therefore irrelevant to the case. Nevertheless, are there any reasons to believe that he could have spotted and avoided this error?

There are clues in Braude & Johnson’s chapter that might alert an attentive reader to the potentially misleading nature of some of its quantitative claims. In the introduction, they explicitly state that their knowledge of human development was drawn from various sources, including “studies of live preimplantation pre-embryos in vitro as part of therapeutic infertility programmes”, and in the critical paragraph from which Munby J. quotes, they preface their remarks with the phrase: “From our knowledge of human development in vitro and those limited studies of early human development in vivo”. Unfortunately, their subsequent commentary does not make it clear which of the evidence they report is from in vivo, and which from in vitro studies. Scrutiny of the references, as noted above, is revealing. Since Braude & Johnson aimed “to summarize as simply as possible some of our current knowledge about human early development”, it is perhaps understandable that they glossed over such detail. Consequently, a non-expert reader may be forgiven for concluding that their account related to normal early human development under natural conditions, particularly since no indication is given in the chapter that there are any substantive differences between in vitro and in vivo embryonic development.

It is unlikely that Braude & Johnson wrote the chapter, published in 1990, with a view to it being submitted as expert witness testimony in a judicial review in 2001, let alone that it should be directly quoted by a judge. Perhaps it would have been wise only to submit a contemporaneous witness statement addressing matters of direct relevance to the case. It would be harsh to even hint that, in submitting the chapter as expert witness evidence, Braude intended to mislead the judge on this particular matter, and it may be asking too much to expect even a learned judge to see through such a tangle of scientific evidence. Nevertheless, irrespective of whether witness, judge or both were culpable, the outcome is clear: Munby J. misjudged the extent of human embryo mortality.

Munby J. had more evidence available to him than just Braude & Johnson’s chapter (Table 1). The emphasis he placed on human embryonic mortality must therefore be read in the context of that further expert witness testimony. Arguably, the contradictions between the various statements (see Figure 1) should have alerted him to these issues. Other than to acknowledge its existence, Munby J. makes no reference to McLean’s detailed and lengthy statement, prepared on behalf of the Claimant. This is regrettable, since the statement offers credible warnings about the relevance and reliability of estimates of embryo mortality in the scientific literature. The similar numerical estimates for pre-implantation loss offered independently by both Drife and Brown are in stark contrast to the one he chose to emphasise from Braude & Johnson’s chapter. Furthermore, in selectively weaving quotations from the witness testimonies, Munby J. gives the misleading impression that Braude and Brown were in agreement about the extent of early embryo mortality, despite evidence to the contrary.

One is left wondering whether Munby J. realised what he was doing.

Conclusion

Braude & Johnson’s chapter was written with more than just biology in mind: the critical paragraph is in a section headed “Ethics and the biology of pre-embryos” and the conclusion touches on religious, ethical and regulatory issues. It thus appears that there was an ethical and, by extension, a legislative agenda underlying this chapter. This agenda is more explicit in a magazine article²⁹ cited by Braude & Johnson.

Natural human embryo mortality has often been linked to the ethical status of human embryos. For example, in their brief article, Roberts & Lowe state that “If Nature resorts to abortion … by discarding as many as 3 in every 4 conceptions, it will be difficult for anti-abortionists to oppose abortion on moral and ethical grounds.”¹² Ronald Green, Professor Emeritus of Religion at Dartmouth College, points out, incorrectly, that “between two-thirds and three-quarters of all fertilized eggs do not go on to implant in the womb” and asks: “In view of this high rate of embryonic loss, do we truly want to bestow much moral significance on an entity with which nature is so wasteful?”³⁰ A report of the Ethics Committee of the Royal College of Obstetricians and Gynaecologists in 1983 states: “Knowing as we do that in the natural process large numbers of fertilised ova are lost before implantation, it is morally unconvincing to claim absolute inviolability for an organism with which nature itself is so prodigal”³¹. This link has been considered by many others^32–34. Thus, McLean’s assertion, in evidence, that early embryo loss is not only of biological interest but also of political and legislative significance^pp, was clearly correct. How specific estimates of embryo mortality inform an ethical calculus is, perhaps, not so clear. Nevertheless, for those who consider it germane, McLean’s exhortation that “It is therefore important to obtain as accurate an estimate as is possible for the occurrence of early human embryo loss”, must surely be correct too.

Did the quantitative bias in Munby J.’s description of embryo mortality have a significant influence on his legal judgment? If not, why then, one may ask, in such a “very long” judgment, would he deliberately choose to emphasise this particular point? What purpose might such an observation serve? Perhaps these are questions for other, legal minds. However, the biology is a different matter. Judgments carry weight and can influence opinion. Even on biological matters, legal scholars may rely upon the sayings of learned judges rather than scientific evidence.

In the first four editions of her popular undergraduate textbook^35–38, Emily Jackson discusses Munby J.’s judgment and comments on embryo loss: “Approximately 75 per cent of all naturally fertilized eggs will be lost before the woman’s next period, and it would be counterintuitive to describe these losses as miscarriages”^qq. Although she does not offer the judgment as her source, her words closely reflect Drife’s evidence, repeated verbatim in the judgment^rr. Her addition of the word “naturally” makes explicit the implied, and incorrect, sense in the judgment. It is notable that in the most recent (2019) edition of her book³⁹ the explicit 75% claim is omitted: “The majority of naturally fertilized eggs will be lost before the woman’s next period, and it would be counterintuitive to describe these losses as miscarriages.”^ss

Another legal scholar, Jonathan Herring, in all eight editions of his textbook^40–47, quotes directly from paragraph 129 of Munby J.’s judgment in order to outline an ethical argument:

The first reference (^†) correctly attributes these words to Professor Brown, as found in the judgment. Strangely, the second reference (^§) is to an article by Professor John Harris on assisted reproductive technological blunders⁴⁸ that has no bearing on the issue. Clearly, it is a direct quotation from Braude & Johnson (1990)¹⁷ and paragraph 129 of Munby J.’s judgment.

It is unfortunate that these biological errors have found their way into standard legal textbooks. Irrespective of the philosophical merits of arguments such as those outlined above, most would agree that in order to arrive at defensible ethical and legal conclusions it is necessary to begin with reliable and relevant evidence. It would be helpful if Munby J. were to clarify whether he realised that the numerical estimates he emphasised in paragraph 129 of his judgment were derived from in vitro circumstances, and were therefore not representative of natural in vivo situations. It would also be instructive to know whether Munby J. still believes that “not much more than 25% of successfully fertilised eggs reach the blastocyst stage of development”. If the answers to these queries are negative, a clarification from him regarding the biology would be welcome, in the hope that the widely held, unsubstantiated and excessively pessimistic view of natural human embryo survival may, little by little, be replaced with one that is more closely based on available, relevant, scientific evidence⁵.

Data availability

Notes

^a Paragraphs from the judgment (and witness statements) are referenced in footnotes as follows: Munby [1]. The full judgment is available here: http://www.bailii.org/ew/cases/EWHC/Admin/2002/610.html

^b Munby [346].

^c Munby [2].

^d The Civil Procedure Rules, r.32.12(2)(c)

^e The Civil Procedure Rules, r.31.22(1)(a)

^f Munby [32].

^g Munby [125] & [191].

^h Munby [32].

ⁱ Munby [126].

^j Munby [126(iii)].

^k Munby [126(ix)].

^l Munby [131-5] & [137]; WSJD [3-4], [8], [11-12] & [16].

^m Munby [134]; WSJD [8]. Footnote 1 (shown at the end of the quotation) is found in WSJD but omitted from the judgment, and reads as follows: “1 Drife, JO. British Medical Journal 1983; 286:294.”

ⁿ The BMJ has no record of whether this response to a reader question (or similar short items) would have been peer-reviewed in 1983 (personal email correspondence from BMJ editorial office, 23^rd June 2020).

^o WSJD [8]. This claim only makes sense if his estimate of post-implantation loss of ‘between one third and one half’ is interpreted as 42% (average of 33% and 50%). However, in his BMJ piece, Drife states that out of 36 women with detectable hCG, only 24 will miss a period, indicating a loss at this stage of 12/36 = 33%. Thus, his estimate of post-implantation loss is inflated in WSJD compared to the BMJ.

^p The BMJ figure incorporates an estimate (‘between 10% and 30%’) for the rate of pregnancy loss after a woman knows she is pregnant. To be internally consistent in the BMJ piece, a clinical pregnancy failure rate of 15% must be used. Applying this estimate to WSJD results in a total loss between fertilisation and birth of 79%.

^q Munby [129]; WSNB [22].

^r WSNB [22]. Footnote 1 in this passage reads as follows: “Edmonds D.K., Lindsay K.S., Miller J.F., Williamson E., Wood P.J. Early Embryonic Mortality in Women, Fertility and Sterility 1982 Vol 38 447–453”.

^s The precise value reported by Edmonds et al. (1982) was 61.9%.

^t Munby [129].

^u It is both likely and appropriate that Munby J. would have regarded Professor Johnson also as a “very eminent medical expert”.

^v WSPB [8].

^w WSPB [15]. Braude’s description of this data is slightly inaccurate. The study followed 221 women who were attempting to get pregnant, and the value of 22% (43/198) actually refers to the number of hCG positive menstrual cycles (198 out of 707 monitored cycles) that did not manifest as a clinical pregnancy, i.e., 43. 22% therefore refers to embryo loss from the onset of implantation (as indicated by the positive hCG test) up to, but not including, clinical diagnosis (as indicated by an absent menstrual period or subsequent positive pregnancy test).

^x WSPB [15].

^y Braude & Johnson (1990), p. 218. The passage contains three endnotes as follows:

Endnote 16 reads: “V. N. Bolton, and P. R. Braude, in A. McLaren and G. Siracusa, eds, Current Topics in Developmental Biology, Vol 23, Recent advances in mammalian development, Academic Press, 1987, pp. 93-114.”

Endnote 17 reads: “A. J. Wilcox, C. R. Weinberg, J. F. O’Connor, D. D. Baird, J. P. Schlatterer, R. E. Canfield, E. G. Armstrong and B. C. Nisula, Incidence of early loss of pregnancy’, New Eng. J. Med., 319, 1988, 189-94.”

Endnote 18 reads: “L. Regan ‘A prospective study of spontaneous abortion’, in: R. W. Beard and F. Sharp, eds, Early Pregnancy Loss, 18th Study group of the Royal College of Obstetricians and Gynaecologists, Springer-Verlag, 1987, pp. 23-37.”

^z Given the matter under consideration in the judicial review, it is notable that, in this passage, Braude & Johnson clearly use the word ‘pregnancies’ to refer to women carrying a fertilised egg before implantation.

^aa Munby [126(vii)] & [126(viii)].

^bb WSJM1 [30–34].

^cc WSJM1 [31] & [33].

^dd WSJM1 [32].

^ee This value of 50% ultimately derives from the 8 pre-implantation embryos, of which 4 were abnormal and therefore assumed to be destined to perish before the end of the first 2 weeks after fertilisation. It is therefore based on a very small sample size.

^ff It is a numerical coincidence that McLean’s value of 29% (10/34) is so close to this value of 30%.

^gg A zygote is the newly fertilised ovum: a one-cell embryo. Confinement is the time of childbirth.

^hh For example, Roberts & Lowe use Hertig’s work as a source for their claim that coitus at the time of ovulation results in a fertilisation rate of 50%. This contrasts with Hertig’s own conclusion, followed by Leridon, that ‘when conditions are optimal about 15 per cent of oocytes fail to become fertilized’ (Hertig, 1967).

ⁱⁱ WSJM1 [33].

^jj WSJM1 [33].

^kk A PubMed (https://www.ncbi.nlm.nih.gov/pubmed) search on <"early pregnancy factor"[All Fields] NOT Review[ptyp]> performed on 3^rd June 2020 identified 142 articles published between 1977 and 2000, and 35 from 2001 to the present day. A search on <“embryo derived platelet activating factor"[All Fields]> identified only 26 articles, published between 1985 and 2004. Only two of these were published after 1992.

^ll Munby [126(ix)].

^mm WSJM1 [34].

ⁿⁿ “Biochemical pregnancy”: this term is sometimes used to refer to pregnancies that are detected solely by elevated hCG rather than by clinical observations such as a missed menstrual period.

^oo This phrase is found both in WSJM1 [34] and the abstract of Walker et al. (1988).

^pp WSJM1 [34].

^qq Jackson, E., Medical Law: Text, Cases, and Materials. 1^st ed. (2006), p. 624; 2^nd ed. (2010), p. 693; 3^rd ed. (2013), p. 703; 4^th ed. (2016), p. 737.

^rr “…around 75% of all conceptions are followed by an apparently normal period. These losses … are not covered by the term ‘miscarriage’.” from WSJD [8] and Munby [134].

^ss Jackson, E., Medical Law: Text, Cases, and Materials. 5^th ed. (2019), p. 770.

^tt Herring, J., Medical Law and Ethics. 1^st ed. (2006), p. 250; 2^nd ed. (2008), p. 285; 3^rd ed. (2010), p. 310; 4^th ed. (2012), p. 318; 5^th ed. (2014), p. 320; 6^th ed. (2016), p. 332; 7^th ed. (2018), p. 331; 8^th ed. (2020), p. 374.