Last Friday, there was wide reporting in the media of a study which shows “that enabling women to have early abortions at home during the pandemic is safe and effective.” SPUC’s Alithea Williams pointed out that the study was authored by the medical directors of the three major abortion providers, and that there were immediate problems to using it as the basis for a permanent at home abortion policy. 

Now, Dr Greg Pike, research fellow at the Bios Research Centre, has raised some additional points using his academic expertise. He says: “This study lacks so much important information that it is difficult to be sure how reliable are any of the findings, and therefore conclusions.”  Of the authors’ claim that “Compelling evidence from 52,145 women shows telemedicine abortion is safe, effective & improves care”, he says: “Given the uncertainties in the paper this is an unsupportable claim, and the authors have some explaining to do.”

Dr Pike raises a number of issues with the study.

The study group

The study is designed to compare a telemedicine delivery service model with a traditional service model for medical abortion in England. They compare 22158 women seeking early medical abortion (EMA) under the traditional model (ie, before the government changes), and 29984 under the telemedicine-hybrid model (T-H, after those changes). The T-H model was actually split between 18435 (61%) who were provided EMA entirely via telemedicine (phone consultation with gestational age determined by reported last menstrual period, and with drugs posted or picked up), and 11549 (39%) who had an in person visit with ultrasound. Dr Pike says: “The breakdown of the T-H cohort is confusing and not explained.  Why compare the traditional model with a mixed group of some telemedicine and some in person?  The in person group within the T-H cohort is really similar to the traditional group…one is left surmising that the mixing would likely dilute any differences between telemedicine and traditional service delivery models.”

Success rates

The study says that “rates of successful medical abortion are high under both service delivery models – 98.2% in the traditional cohort and 98.8% in the telemedicine-hybrid cohort”.

“One of the biggest outstanding questions arising from this paper,” says Dr Pike, “is how the authors claim to know the success or otherwise of abortion for every woman.

“They show data for each and every woman, and therefore make claims about 98%+ success rates.  But the paper does not make clear how they know this. Did the abortion providers follow up every woman to seek self-reported outcomes?  By phone, in person, online?  They do not say.  Indeed, the authors don’t even say how completion was measured – via pregnancy test, ultrasound?  How do they know what the outcomes were for every single woman – something pretty much unheard of even in dedicated trials let alone via telemedicine for a whole of population group?

“This issue is important because it relates to one of the key claims of the paper, and yet it is nothing short of staggering that they have such data from everyone when even clinical trials can’t get close.”  

Dr Pike says that it is very difficult to believe that all the women were even contactable after the abortion, let alone that there was reliable evidence of abortion completion.  And the authors admit that – in the section on strengths and limitations, they say, “The main limitation of this study is that we were unable to actively follow-up patients post-abortion and therefore only significant adverse events can be reported with confidence.”

Dr Pike concludes: “This is a clear admission that they did not follow-up patients post-abortion.  So how can they report so definitively about completion rates?”

Adverse events

Dr Pike raises similar questions about complications arising from early medical abortion. “How did they assess adverse events? There is no explanation in the methods section, but there is a lengthy self-defense in the strengths and limitations section.  The authors start out by saying, ‘There is a potential gap in the consistency of reporting incidents, due to some complications not meeting the threshold of serious incidents, multiple routes of entry into the NHS and informal communication between the NHS and abortion providers.’  This sounds like a fair statement about the problems with picking up adverse events from a study in which patients were de-identified, so no cross checking could be undertaken with hospitals or other health records (a huge task in any case).  Despite identifying a gap, the authors then say “the risk management and reporting systems within the NHS are well defined, with serious incidents being routinely shared.” And that abortion providers are in touch with the Care Quality Commission, which in turn makes known adverse events to providers. 

“This is all very diffuse and unconvincing.  Without any means of properly checking health records for individual women, assessment of adverse events is certainly underestimated, potentially by a large amount.

“In addition, the authors have set the bar very high at blood transfusions, IV antibiotics in hospital, major surgery, and death.  There are other complications that can be serious and important indicators of harm. Moreover, patients could present as a complication of miscarriage, not EMA.”

Gestational age

The study claims that telemedicine means that abortions are happening earlier in the pregnancy. However, Dr Pike says, their results do not prove this. “The GA (gestational age) in the telemedicine group will be underestimated because of the way GA was determined.  It was based on the woman’s recall of last menstrual period up till the date of posting drugs.  The date of actual drug taking was unknown and hence GA at the time of drug use could be many days later (taking into account postal time during the pandemic plus any delay for personal reasons such as ambivalence).  Moreover, despite what the authors claim, GA via last menstrual period is known to be underestimated.  The claim that telemedicine led to medical abortion at an earlier GA is therefore unsustainable.”

Waiting times

The claim that waiting times are reduced is also called into question by the same consideration. The calculation of the waiting time was taken from time of first contact to date of posting. The efficiency of the postal service during lockdowns is a factor not considered by the authors.

Acceptability

The study says that survey data tells the authors that 96% of women were “satisfied” or “very satisfied” with their care, or rated their experience as “good” or “very good”. However, there are problems with this. Firstly, the authors report survey data for 2453 respondents, but they do not say whether that was for the whole cohort or only the mixed T-H cohort. Dr Pike explains: “If the respondents were from the whole cohort, then the response rate was a tiny 4.7%; if the respondents came only from the T-H cohort, then the rate was still a very small 8.2%. Any conclusions from such small percentages are entirely unreliable.  In fact even reporting an outcome from such a tiny self-selected number would be embarrassing for most researchers.  If this tiny feedback is indicative of post-abortion follow up regarding outcomes, then the paper has a serious problem.”

Conflict of interest

Agreeing with SPUC’s analysis, Dr Pike says: “Finally, the authors have an obvious conflict of interest as directors of abortion provider organisations.  That conflict is both in terms of image (providers must be seen to provide a safe and effective service), as well as financial (corporate for the organisation they run, as well as personal as their job depends on their role as directors maintaining the organisation’s operations - which also depends on the right image).”

Dr Pike concludes: “This study lacks so much important information that it is difficult to be sure how reliable are any of the findings, and therefore conclusions.  Nevertheless, the authors provide a tweetable abstract as follows: ‘Compelling evidence from 52,145 women shows telemedicine abortion is safe, effective & improves care.’  Given the uncertainties in the paper this is an unsupportable claim, and the authors have some explaining to do.”

The Government’s consultation on DIY abortion runs until 26 February. Guidance on responding to it can be found here.