Survey on sexual assault in surgery may not be representative

18 October 2023
What was claimed

One in three female surgeons have been assaulted by a colleague in the past five years.

Our verdict

This comes from a survey of surgical workers (not just surgeons) that may not be representative, because the responses came from those who chose to participate.

1 in 3 female surgeons assaulted by a colleague

Two thirds of female surgeons recently admitted that they had suffered sexual harassment at the hands of senior male colleagues. A third said that they had actually been groped or assaulted whilst working in the operating theatre.

Many media reports last month claimed about a third of women who work in surgery have been sexually assaulted by a colleague within the last five years.

Versions of that statistic have been quoted since, and it was used last week at the Labour party conference by shadow attorney general Emily Thornberry.

But the survey this figure comes from may not have been representative, and statisticians have told Full Fact it is likely to have overstated the true rate of sexual misconduct.

To be clear, even if the media reports did overstate the incidence of sexual misconduct, this does not mean that it is not a serious and widespread problem among those who work in surgery. Indeed we already have clear evidence that it is common in society as a whole, and for women in particular. This survey itself shows that many surgery workers have experienced it, even if it doesn’t reliably tell us what the rate is.

The claim is based on a study published in the British Journal of Surgery, which was announced with press releases from the University of Exeter, the University of Glasgow and the University of Surrey.

It appeared on the front page of the Times, as well as in the Sun, Guardian, MailOnline, Mirror, Telegraph, ITV News, Metro, Evening Standard, PA Media and the BMJ, and on Sky News, BBC Woman’s Hour, TalkTV, LBC, GB News, and many other places.

Following contact from Full Fact, the Mirror and the Guardian have since amended their articles to make it clear that the survey’s findings may not be representative.

The group that organised the research also made the same claim on X (formerly Twitter).

Some reports, including those from BBC News and the Independent, added more context by saying the claim applied to women “who responded” to the researchers. (Some of the other reports listed above also added this context at some point, but did not give it in the headline or when first using the figure.) This is an important distinction, because the survey was voluntary and shared on social media, so it may not be representative of people who work in surgery as a whole.

The research protocol (a written plan approved before a study begins), which does not appear to be published but was shared with Full Fact by the authors, describes its method as “convenience sampling”. It also says: “Unfortunately, it is outside the scope of this particular study to strategically obtain a true representative sample.” 

We’ve written about how self-selecting surveys are reported many times before.

To avoid confusion, research institutions should be clear about what their research has found, which includes being clear about what it doesn’t show, if appropriate. The media should take care to check the facts behind what they’ve been told to ensure their reporting is accurate.

Honesty in public debate matters

You can help us take action – and get our regular free email

How the survey was done

The research was conducted by a group of researchers and clinicians called the Working Party on Sexual Misconduct in Surgery. (We contacted the group via the University of Exeter press office and have not received a comment for this fact check, although one of its members explained the survey’s methodology to us and answered a number of technical questions.)

From September last year, the working party arranged for groups including the Royal College of Surgeons of England (RCS) and the British Orthopaedic Trainees Association (BOTA) to email a link to a survey to a list of their members.

The subject of the email was “Brief Survey about Sexual Misconduct in Surgery, endorsed by the RCS, BOTA, and NHS England Safeguarding”, and part of its text said: “It is important to get responses representing the breadth of surgical experience, including from those who have never witnessed or experienced sexual misconduct, to accurately assess the current situation.” As the study explained, this line was included “to help mitigate selection bias and capture the widest possible sample”.

The survey was also shared on social media. According to a Facebook post from BOTA, it appears that it was originally due to close on 23 October 2022. However, in December, the link to the survey was also shared on X by the RCS, BOTA, the Association of Surgeons in Training and several other people and organisations.

What it found

The survey, which was anonymous, asked people for their age, gender and occupation, and whether they had witnessed or experienced types of behaviour defined by the study as a form of sexual misconduct.

In total, 1,704 people participated, with 1,434 providing enough data to be included, and 756 of them being part of the NHS England surgical workforce.

The researchers analysed the data they collected in two basic ways. The data collected from people in the NHS surgical workforce was weighted to reflect the balance of job type and gender in the NHS workforce. And all the eligible data was also analysed without weighting, to show the simple proportions of different groups that gave each answer.

In practice, the results from both analyses were similar. In the unweighted data, asking about misconduct involving colleagues in the last five years, 29.9% of women and 6.9% of men reporting being the target of sexual assault. Even higher proportions reported witnessing this behaviour.

The 29.9% figure (34% in the weighted data) became the basis for the ‘one in three’ claim widely reported in the media.

The study also shares other data, weighted and unweighted, including more details on the frequency of people’s experience, the types of misconduct involved, and the number of people who said they’d experienced or witnessed rape.

The limitations

By our estimate, based on the figures provided, more than 200 women and around 50 men who participated in this survey said they had been sexually assaulted by a colleague in the past five years.

This fact alone is enough to suggest that sexual misconduct in healthcare is a serious and widespread problem.

However, in order to use this data to measure how widespread the problem is, we would need to know that the people answering the survey were a fairly typical sample of the surgical workforce as a whole—and there are good reasons to expect that they were not.

The main risk is self-selection bias, where people with an interest in a subject are more likely to answer survey questions about it.

In this case, on seeing an email or a social media post about this survey, people with experience of sexual misconduct might understandably be more motivated to click on the link and spend time answering questions. People without this experience might be more likely to ignore the survey altogether.

When a survey is shared on social media, it is also hard to know whether the people who saw it were themselves a typical sample of the people in this workforce. It is possible that the people who followed the accounts sharing the survey were more likely to be motivated by this subject and have higher rates of misconduct to report.

Research like this is sometimes called ‘convenience sampling’, because it uses data from respondents who are convenient to reach (instead of a random sample, which would be more difficult).

The study’s research protocols acknowledged this weakness by saying that it was “outside the scope of this particular study to strategically obtain a true representative sample”.

We have spoken to statistical experts who told us the survey results are likely to overestimate the true rate of sexual misconduct in the surgical workforce.

Dr Catey Bunce, an applied medical statistician and a statistical ambassador for the Royal Statistical Society, said: “I think it's natural that people who have an interest in this topic are more likely to have responded to this. Given the design of the study, it is likely that the figure will be an overestimate but clearly any incident would be of concern.”

David Hussey, principal statistician at the National Centre for Social Research, said: “There are two fundamental issues here. First, the survey is based on a convenience sample, so it’s not appropriate to extrapolate the findings to the whole surgical workforce. Second, there's a strong risk of selection bias due to the sensitive topic and the recruitment methods used.

“It's impossible to estimate the extent of it, but there are some indications that selection bias is present. Overall, the survey findings are likely to overstate the prevalence of sexual misconduct in the UK surgical workforce.”

In short, while it is true that the survey found that about one in three of the women who responded said they had been sexually assaulted by a colleague in the past five years, this doesn’t mean that one in three of all the women who work in surgery would say the same. Yet this is what most of the reporting claimed.

The Royal College of Surgeons of England told us: “Our understanding is that to help mitigate selection bias and capture the widest possible sample, those invited to complete the survey were encouraged to participate regardless of whether they had or had not witnessed, or been targets of sexual misconduct.

“The research is published in a peer-reviewed journal which is respected in the field of surgery. There has been a significant response to the survey from the surgical community, and our members, in particular, who have shared experiences of sexual misconduct. It is clear that the behaviours are happening at an unacceptable level.”

Ms Thornberry’s claims

In her conference speech last week, Ms Thornberry repeated the figures from this survey without explaining that its findings may not be representative, and made some other errors in how she described it.

The survey was intended to show the experience of the surgical workforce, not all of whom are surgeons, as Ms Thornberry and (some media reports) suggested.

The survey did find that nearly two thirds of female respondents had experienced sexual harassment by a colleague, by its definition, in the past five years—but not that this had been at the hands of senior and male colleagues specifically.

Finally, the survey did find that a third of female respondents had been sexually assaulted, but not necessarily in the operating theatre. Ms Thornberry’s office told us she had mistakenly omitted the word “even” from the text of her speech before saying “whilst working in the operating theatre”, and that this would be corrected in the subtitled film of the speech. It did not respond on the other points.

What else do we know?

Sexual harassment and sexual assault are sometimes defined in different ways, which means that research on the subject can be hard to compare. This study supplied some definitions that it hoped might be useful for people responding.

We may soon have more representative data on general rates of “unwanted behaviour of a sexual nature” in the NHS in England, after a question on the subject was added to the NHS Staff Survey this year. However, the results of this survey are not expected until spring 2024.

For now, we have not been able to find good data that measures the rates of sexual misconduct in surgery in the UK.

However, one survey of almost all the residents in accredited general surgery programmes​​ in the US in 2018 found that about 20% of women and 4% of men had experienced sexual harassment, either from a colleague, a patient or a patient’s family, since their residency began.

Data from the most recent Crime Survey for England and Wales (CSEW) shows that about 3.3% of women experienced sexual assault in the year to March 2022, and 27% have experienced it at some point in their adult lives. For men, the respective figures are 1.2% and 5.7%.

A Government Equalities Office survey on sexual harassment at the beginning of 2020, found that about 29% of people had experienced it in the workplace the previous year. In this survey, the rate reported by women was only slightly higher than by men.

We approached the media outlets, universities and professional associations mentioned in this article for comment.

Image courtesy of Olga Kononenko

We took a stand for good information.

As detailed in our fact check, the Mirror and the Guardian have amended their articles to make it clear that the survey’s findings may not be representative.

Emily Thornberry’s office told us she had mistakenly omitted the word “even” from the text of her speech before saying “whilst working in the operating theatre”, and that this would be corrected in the subtitled film of the speech. It did not respond on the other points.

We subsequently wrote to Ms Thornberyy to ask her to take steps to add important missing context to versions of her speech which have been published online by Labour.

Ms Thornberry did not respond. 

Don’t put up with bad information.

Add your name and join the fight for higher standards.

Full Fact fights bad information

Bad information ruins lives. It promotes hate, damages people’s health, and hurts democracy. You deserve better.