Ask users for gender or sex

[sarawilcox]

What

Use this issue to discuss when and how to ask about people's gender and sex.

Related info

GOV.UK Design System's gender or sex backlog issue

[sarawilcox]

This is one of the areas that we found that people needed more guidance on in our guidance on How to write good questions for forms #145.

There is already some good work being done elsewhere on this, for example, at Disclosure Scotland and at the Office for National Statistics.

https://blogs.gov.scot/digital/2018/07/05/sex-and-or-gender-working-together-to-get-the-question-right/

[sarawilcox]

See also GDS backlog: alphagov/govuk-design-system-backlog#69 (comment)

[henocookie]

Current iteration of the question asking for user's sex on the NHS coronavius status checker

[Zeno001]

That is perfect. Any idea when it'll be rolled out to the live checker?

[sarawilcox]

Thanks @Zeno001. For now, we've decided not to add it to the checker as the data is being captured in other ways. We'll continue to test this pattern in other services for the time being.

[Zeno001]

Sorry, are you referring to the coronavirus status checker? How is sex now being captured?

[amyj2110]

Summary of research carried out on the Coronavirus Status Checker service around how to ask for someone's sex

Disclosure Scotland
Spoke to a former user researcher and content designers about how they asked people for their sex.

They said that asking questions around sex and gender can be very loaded and traumatising questions for some people. They recommended explaining up front what sort of questions would be asked and why; rather than springing it on people, it may be better to forewarn them.

You should explain why you are asking for the information; giving a good reason may convince some people to answer the question. Words like "we need to understand" are better than "we're interested in", and if asking for sex, you should explain as fully as possible what genetic or medical differences you might be talking about. Trans people may also answer with their current sex, rather than their sex at birth.

In their opinion, if you are making a strong case that someone needs to answer this question, adding a "prefer not to say" option may undermine this.

If you ask a gender question, note that a free text box may be used for anti-trans abuse. You should have the comms team lined up to deal with any such abuse that might come down on the team putting this together.

They recommended a simple question: "How do you identify? Male / Female / Another way". Do not use "unspecified" as an answer. There are potentially a lot of ways for someone to define their gender, and it is better to let someone tell you how they define themselves rather than provide all the set options.

Office for National Statistics
Spoke to the topic lead for gender identity and sexual orientation.

The 2021 census team has done a lot of research into this area, and recommended to use their own approach. This involves asking a simple sex question (Male / Female) with an additional voluntary question for over-16s that gives the option to say if the person's gender is different from their sex registered at birth, and, if different, to record their gender, and include a "prefer not to say" option. They have conducted large-scale quant testing around "acceptability" of the question, and cognitive testing for "understanding".

The key issue is what data is actually required and why. This should be made clear to the users answering your question. You should give people a way to avoid the question.

Asking about "sex at birth" could bring up some problems. Trans people may feel that this is a private matter and that could potentially increase non-response rates.

Documents by ONS
Update on Sex and GI (December 2018) - see section 4 on Gender Identity which covers the development of the overall approach for the 2021 Census and testing informing this.
Census 2021 - list of all testing.
2019 Rehearsal questionnaire
Government Statistical Service page on sex and gender
Government Statistical Service Harmonised principles for demography

[sarawilcox]

Yes, @Zeno001. We're not rolling out a question about sex in the coronavirus status checker for now because the data is being captured elsewhere. I don't know the details but I believe that the data analysts are collecting and analysing data from multiple sources. They've said that they're able to get the data they need without us adding a question about sex to the status checker. We're only asking for data we need to collect.

[Zeno001]

We know the effects of the coronavirus are highly sex-dependent so surely asking for the sex of the particular person you're gathering the other information on is vital?

[sarawilcox]

From the ONS update December 2018

[Zeno001]

Thanks, Sara.

That's from the census, but why ask a question about gender identity in a survey on the effects of the coronavirus? You say you only ask for data you need, so what's the need here?

[sarawilcox]

Thanks @Zeno001. We did some research around this, which we're documenting in case it's helpful to other teams. At one point it was suggested that we might need to include questions about sex and/or gender in the status checker. In the event, we're not including either.

[Zeno001]

Is the research published anywhere? But that seems an odd decision, given, as I said, we know of the sex-dependent effects (and no reason to suppose there would be any gender-specific effects).

[Zeno001]

As an illustration, the ONS list deaths involving coronavirus (COVID-19) by sex (ie male or female) only.

For example, see: Coronavirus (COVID-19) roundup - Office for National Statistics

These data show a clear sex bias. They do not mention 'gender' anywhere as it's irrelevant.

Similarly, Public Health England: Coronavirus (COVID-19) cases in the UK

Sex is important here: gender isn't.

[sarawilcox]

We are currently doing more research into this issue across various services.

Worth considering whether you can get your system to do the hard work so you don't have to ask a question that might be a barrier for some users. See this blog post by @GrilloPress: https://grillopress.github.io/2019/03/18/do-the-hard-work-to-make-it-inclusive.html

[davidhunter08]

How the ‘Sign up to be contacted for coronavirus vaccine studies’ service asked users for sex and gender:

[Zeno001]

Sex 'registered at birth' is just someone's sex so there's no need for the superfluous words. Just ask, "What is your sex? Female | male.

Gender is not equatable to sex (because the former is biological and the latter is a social construction relating to expectations, behaviours and attributes acquired through life) so it's unclear what the point is of stratifying participants by their 'gender', particularly since the question is essentially binary and I suspect not many will fill in the optional field. Will the write-in answers even be coded?

[Tosin-Balogun]

Sex 'registered at birth' is just someone's sex so there's no need for the superfluous words. Just ask, "What is your sex? Female | male.

What if the user was born intersex?

[Zeno001]

Intersex conditions are anomalous developments of dimorphic sexual classes[1] and are not a third sex class. Those with a Difference of Sex Development are still either male or female.

---

1 Gender incongruence in children, adolescents, and adults
Susan Bewley, Damian Clifford, Margaret McCartney and Richard Byng
British Journal of General Practice 2019; 69 (681): 170-171. DOI: https://doi.org/10.3399/bjgp19X701909

[amyj2110]

@Zeno001 , thank you so much for engaging with this thread. It looks like you've got a lot of experience that will help us make this work a lot better.

Would you be open to meeting myself and some other members of the service manual team to talk through your thoughts? Thanks!

[Zeno001]

I really have no specific relevant experience here (although I have worked for a multinational company as R&D Manager running software and hardware projects in electronic security but never really touched much on UX) but I do have an interest in ensuring that the term sex is not conflated with other characteristics where any confusion between them has real-world implications that can cause unintentional harm or unintended discrimination.

If you think a chat would be useful to you, please let me know.

[amyj2110]

@Zeno001 that's absolutely fine. And yes, would be good to chat.
Please feel free to email me at amy.james2@nhs.net and we can set up a meeting.

[Zeno001]

This new paper by Bewley et al. in the bmj is important and needs to be carefully considered:

Sex, gender, and medical data | The BMJ

Medical care requires an understanding of the difference between sex and gender categories; untangling them is crucial for safe, dignified, and effective healthcare of all groups. Avoidable harm may result when they are conflated—for example, if sex specific laboratory reference ranges are used for people whose gender is recorded but not their biological sex.7 -9 Furthermore, assuming that a patient’s recorded sex equals their gender creates problems for people who do not identify with socially constructed gender roles. Disaggregated data on sex and gender are necessary to ensure that public services are organised for the benefit of the whole population, not just the majority. Finally, confidentiality and respect for all groups are essential to ensure safety in reporting, and ensure that data are as accurate as possible and can be used to benefit all groups.[1]

Anyone using data primarily collected for another purpose, including clinical researchers using NHS or census datasets, needs to understand the original purpose and mode of data collection. Ambiguous data collection methods that conflate sex and gender risk erroneous research findings, poor service planning, and lower quality medical practice. Gender and sex should not be used interchangeably. We risk harming patients if we do not understand the difference.[1]

And this response to it provides a useful insight and background:

Sex, gender, and medical data: a way forward | The BMJ

---

1 Bewley S, McCartney M, Meads C, et al. Sex, gender, and medical data. BMJ 2021;372:n735. doi:10.1136/bmj.n735

[sarawilcox]

Good Twitter thread here. Thanks @deanvipondNHS.
https://twitter.com/AlexPetrovnia/status/1417550024360304649?s=20

[Zeno001]

There are certainly a few sensible things in that thread, but most should be self-apparent from either a clinician (for whom sex is important) or a UK GDPR (only asking for personal information you need to know and have already established a lawful basis for processing) point of view.

The author is certainly correct that sex is not the same as 'gender'. However, I would take issue with a couple of points:

If you’re asking questions about uterine health, ask if your participant has a uterus, not for their gender. If you’re researching gender bias, ask about gender. If you simply need sex statistics, ask about sex.
Think critically about what information you want.

'Gender bias' is usually sex bias and it is sex that is the protected characteristic under discrimination law, the Equality Act 2010.

According to Jo's Cervical Cancer Trust, Half of women don’t know what the cervix is. I suggest it will be much the same with uterus and asking about the term could well exclude a large proportion of women and girls. If it's important you know whether the individual is likely to have a uterus, ask about their sex: only those who are female have a uterus, cervix, etc.

One of the most common mistakes I see in this is listing “male” and “female” (and sometimes “intersex”) as genders. None of these are genders, all three of them are sex descriptors, and should only be used if you are asking a question about sex specifically.

Female and male are certainly the two sexes but intersex is not a sex or sex descriptor: Intersex describes conditions that are anomalous developments of dimorphic sexual classes.[1] They are a set of hormonal, chromosomal, etc medical conditions, not a sex.

So in conclusion, here’s the simplest possible options that I would recommend, when asking questions about sex and gender.

For sex, list options: male, female and intersex.

For gender, list options: man, woman, nonbinary (write in).

The problem here is that woman and man have legal definitions as the protected characteristic of sex under the Equality Act 2010. Section 212 defines:

“man” means a male of any age;
“woman” means a female of any age.[2]

Using them as categories for 'gender' (which has no legal definition) can only cause confusion and wrong answers.

Oh! One more quick note: in general, sex when applied to humans is NOT a particularly helpful categorization. Overall, it’s best practice to avoid asking about sex (instead ask specific anatomy or physiology questions) altogether if possible.

This is a bizarre thing to say. Sex is a fundamentally useful categorisation: apart from having a legal meaning in terms of equality law, it is important in the assessment, diagnosis and treatment of patients. Getting it wrong by asking irrelevant or confusing questions could cost lives.

The author says:

Gender describes an individuals’ personal gender identity.

There may be a limited set of circumstances where a clinician might want to know how the patient feels about themselves, but it is far more likely that for the vast majority of patient encounters, it is sex that is important to know and having the wrong information causes harms.

---

1. Bewley S, Clifford D, McCartney M, et al. Gender incongruence in children, adolescents, and adults. Br J Gen Pract 2019;69:170–1. doi:10.3399/bjgp19X701909 https://bjgp.org/content/69/681/170

2. Equality Act 2010

[TJFDM]

As part of the registration process for registering with a GP, users are asked on the current paper GMS1 form 'What is your gender' and given 2 options - Male or Female.

We are now looking to introduce a digital way of registering with a GP and of course, we want to ensure that the questions we ask are as inclusive as possible, but also ensure that we capture the right information to enable GP practices to locate the person's existing NHS record (where one already exists).

During discovery we cycled through a few different options, using examples here in this thread and using the research by the National Booking Service. Our service had to not only enable GPs to match to existing records where they can, but it also needs to ensure that patients are invited for screening.

The current iteration has our question set looking like this:

We have tested these pages with those of cis-gender, transgender and those who are non-binary. Feedback as follows:
Overall:

• Users are unaware why we are asking the question and expect this to be explain in real-terms to them, with the potential for examples to show exactly how the information will be used

Screenshot 1:
Participant 5 (cis-gender)

• Overall believed that the question is inclusive of all
• elt that giving people the option to self-identify was strange
  Participant 2 (trans):
• trans) didn't expect this question to be asked as part of the process and would like more information about why the information is being captured
• Would like to be able to provide their pronouns as part of this question too.
• Felt represented in the options provided and would select the relevant option for them
• Didn't feel that 'in another way' was grammatically correct and 'another way of self-description' would be better and wider understood
  Participant 3 (trans)
• "I'm glad its there [inclusing transwoman] - it should be on every form"
• Felt that the options provided were representative and that most forms only provide male/female options
• They would have been happy to provide this information in this form with the question being asked in this way
  Participant 12 (non-binary)
• Having an option of 'non-binary' made them happy
• Felt it was good that transman and transwoman were not separate checkboxes
• The question and options did not confuse gender with sex
• "I would say its one of the best I've seen"
• It made them feel 'really comfortable' to answer the question
• Pleased that 'another way' provided a free text option

Screenshot 2:
Participant 5 (cis-gender)

• Felt represented in the options provided
• Thought this was a standard question on forms

Participant 2 (trans)

• Immediately wanted to know why this was being asked
• Didn't feel that the information given on the why was sufficient (it isn't)
• Felt that it is 'a good question to ask - it is important in providing services'

Participant 3 (trans)

• Didn't really pay much attention to the question
• Felt represented in the options
• Not sure that they selected the right option for their situation, but they moved on very quickly
• Felt the question was necessary in this scenario

Participant 12 (non-binary

• Wanted to know why it was being asked
• Felt it was confusing for non-binary people, "you can't be assigned non-binary at birth"
• Intersex is not being catered for in this question, and people can be assigned the wrong sex at birth
• No option for 'rather not say'
• There is already trans community mis-trust of the NHS and putting this information into a database might be a step too far, may be better asking this information once they are registered and speaking to a GP

Hope this is helpful, we still have work to do but I think that question 1 did made those trans and non-binary people feel comfortable in answering the question and that it was inclusive. There is still some work to do around effectively communicating why we need to know such information when registering with a GP and also explaining why a combination of question 1 and 2 could be required.

[Zeno001]

I've made many of the same points above but it's worth reiterating:

Intersex is not a sex or an identity. See Differences in sex development - NHS

Sex is not 'given' at birth as if it was in some way an arbitrary or capricious choice that was made at that time and/or is mutable. Female and male are the two sexes and the only two sexes. See Sex, gender and gender identity: a re-evaluation of the evidence | BJPsych Bulletin | Cambridge Core

Sex

Humans are sexually dimorphic: there are only two viable gametes and two sexes, whose primary and secondary sexual characteristics determine what role they play in human reproduction. Sex is determined at fertilisation and revealed at birth or, increasingly, in utero. The existence of rare and well-described ‘disorders (differences) of sexual differentiation’ does not negate the fact that sex is binary. The term ‘assigned at birth’ suggests a possibly arbitrary allocation by a health professional, rather than the observed product of sexual reproduction.

Sex is what is important for a doctor to know (for any number of safety, diagnostic and treatment reasons) and I believe it is the person's sex that is recorded in their medical records. Only those who are female need to be invited for cervical screening and only those who are male need to be invited for prostate screening. This is entirely independent of how they choose to 'identify'.

For further information on this important topic, please see Sex, gender, and medical data | The BMJ

'Gender', 'gender identity' and 'non-binary' are unhelpful terms as their meanings are not widely or consistently understood.

'Cisgender' is considered a derogatory and unhelpful term by many.

'Gender' and 'Gender identity' (whatever is meant by these terms) are not 'assigned at birth'.

Not everyone has a 'gender' or 'gender identity', but everyone has a sex.

Woman and man are the terms used in the Equality Act 2010 to refer to those who are of the female sex or the male sex so using these terms for 'gender' can only cause confusion. See s.11 Equality Act 2010 and s.212 Equality Act 2010

Your screens state "We do not discriminate on the grounds of gender", but discrimination under the Equality Act is on protected characteristics: 'gender' is not a protected characteristic but sex is. And it is sometimes entirely lawful - and indeed desirable and necessary - to discriminate on the grounds of sex. This includes the provision of single-sex wards and medical personnel of the same sex as the patient for some intimate procedures. If you need to state this, it should say something like "We do not unlawfully discriminate on the grounds of sex", but this would be a matter for lawyers.

Some questions that I think need to be answered:

Why isn't the sex (female or male) of the person being asked for?

What are the harms that will be caused by wrongly recording an individual's sex?

What impact analysis has been done if a question on 'gender' or 'gender identity' is asked instead?

What proportion of those who are trans don't know and understand what sex they are?

What value is there from asking for and recording answers to a question on 'gender' or 'gender identity'?

[TJFDM]

Thanks @Zeno001 lots of good points raised - as stated, we are in the early stages of this work and it is a prototype that we are testing continually, however, I think these were important points to share with the wider community in the essence of working in the open. This is why we do user research.

Please be assured that we're not working in isolation on this, we are working with a lot of the same people that Emma and their team worked with. The recording of gender on NHS records is not a simple thing and as Emma has previously posted, this has not changed, and therefore we need to ask questions in a way that allows us to make a match on PDS, as well as consider any other services someone may need.

I'm not sure if you ever had the change to speak to @amyj2110 and the wider team but I will check in and see if there's any learnings we can take from those conversations. I do think that the participants we had some great feedback and some clear points to work on too, which we will of course be doing.

[Zeno001]

Amy and I had a brief email conversation but she never got back to me about suitable dates and times for a meeting.

[sarawilcox]

This issue is an epic on the service manual team's roadmap.

[sarawilcox]

Recent comments on GOV.UK gender or sex issue: alphagov/govuk-design-system-backlog#69 (comment)

[sarawilcox]

A personal blog post about the question about sex and the national coronavirus vaccination booking service. Let's talk about sex by Emma Parnell on Medium.

[ambchang]

On 111 Online, users have to answer some mandatory demographic questions - age, location, sex at birth - before they can then start to answer triage questions about their symptom. The binary answers of Male/Female on the ‘sex at birth’ question is determined by how Pathways, the triage system underlying our service, is classified.

We’ve received a lot of live feedback from users have said that the ‘sex at birth’ question (screenshot below) is difficult to answer and/or often distressing. We have since conducted 2 rounds of discovery research with 15 trans and gender non-conforming users to better understand their experiences and needs around sex and gender. The first round was solely interview-based while the second was a mix of an interview and showing some initial design ideas.

We tested 3 triage journey ideas with users:

1. Asking a tweaked 'sex at birth' question followed by a gender identity question before users start their triage
2. No 'sex at birth' question + a gender identity question only asked if a user is referred to an onwards service
3. No 'sex at birth' question

Note: The gender question was directly used from the version tested by the 'book a coronavirus vaccination' national service (before it was later removed as the PDS API matching algorithm was updated)

Interview feedback on the current ‘sex at birth’ question (above):

• There were differing views around how inclusive the wording of the sex at birth question is, and the extent to which it meets users’ needs
  • Some users appreciated the content explaining why the question is asked, and why sex at birth is needed rather than gender identity
  • Users felt like 111 online had no interest in understanding people’s current circumstances but only the sex they were assigned with at birth
  • Many users pointed out that the current question excluded intersex people
  • “I would prefer if it said sex assigned at birth because sex is assigned at birth it is not just binary.”
• The upfront positioning of the question before knowing a user’s symptom can often be distressing, triggering users to feel dysphoria
  • Most users suggested the question would be more relevant if it was moved to the point in a clinical triage where it was necessary to know
  • However, some users preferred this upfront positioning as they expected the answer to affect the triage questions
  • Being transparent about why specific information is needed and when can increase people’s trust in using the service
  • “I don't want to lie on a form but for me not lying on a form is work because I need to imagine what the person writing the form wanted to know.”
• Users already have mistrust in the NHS which they describe as having a gate-keeping role (e.g. when it comes to accessing gender-affirming care)
• Users are worried that only asking for sex at birth leads to incomplete information about their biological, leading to greater clinical risk and/or incorrect clinical decisions
  • Users would feel comfortable answering specific questions about body parts, hormones, or pregnancy as necessary if they felt relevant to their symptoms
• Despite being a difficult question to answer, some users found it important for 111 online to have information about their biological sex as they recognised that some conditions differ between different biological bodies
• Some users said they wouldn’t answer the question as intended just to proceed along the journey
• If answering this question for someone else, most users agreed that there’s a risk someone may not know that other person’s sex at birth. Users suggested an option to state that you don’t know someone’s sex at birth
  • Some users were concerned that proxy use may force a trans person to be unnecessarily outed in order to use the service

Collation of findings from Idea 1 ('sex at birth' question followed by gender identity question - above screenshot) and Idea 2 (No 'sex at birth' question but gender identity asked at point of referral):

• Users felt that biological sex is relevant in the context of a medical triage, but they appreciate being able to explain their gender, particularly for users who had recently transitioned - it felt affirming and empowering
  • Users hoped that this information could be used by healthcare professionals to address them with the correct pronouns and title
  • However, a few users said that gender identity information may not always be relevant in this triage context so there should be an option for those who don’t want to share this information
• However, if users provided their gender identity (and pronouns) they expected that information to be passed to onward referral services
  • If information is collected and not used, it sets false expectations and leads to more distressing encounters with the healthcare system
• Some users were unsure as to why the gender question was being asked and how the information would be used, this made it difficult to give the most accurate answer possible
  • Some users expected this information to be used to help inform the triage, particularly if someone had different answer for the sex vs gender question
  • One user thought that gender data would be used to improve services for trans and gender non-conforming people at a local and national level
• Most users expected to be able to select multiple options for the gender question
  • Some users expected more options (e.g. ‘queer’, entering your own identity, or the ability to give information about whether they were taking hormone replacement therapy)
  • Some users were concerned that they didn’t know what gender is currently on their medical records. In fact, they could have their gender recorded differently across various medical records across the system
• Some users were concerned about the discrepancy between self-identification (which is good to have) vs finding the right medical record
  • Those who read the explanatory text in the gender question generally understood that the answer would be used to match GP records, but one user questioned whether a different method of matching records could be used

[sarawilcox]

Emma Parnell's blog post from 2021: Let's talk about sex*

[Zeno001]

That's a confusing blog post, @sarawilcox. Can you say why you re-posted it and what you think it brings to the discussion?

[Zeno001]

This is interesting: Trans patients who change sex in medical records face ‘unintended negative consequences’

‘Unintended negative consequences’
The SNP government ordered a review of their records management policy for health and social care, with a draft of the new guidelines sent out to NHS staff for consultation in December.

The proposed update states that the long-standing trans policy, in place for at least the last decade, could have “unintended negative consequences to [a patient’s] overall health where they have chosen a name and gender identity that differs from their current legally designated name and sex assigned to them at birth.”

It adds: “Decisions based on, for example test results, can differ between those with chromosome XX and those with chromosome XY due to the physiological and biological differences.

“Therefore there could be a clinical risk if the biological sex is not known by the clinician, as treatment may be required to be based on the patient’s sex at birth instead of their gender identity.”

The draft guidance calls for patients to be made aware of the dangers when requesting a change to their sex in health records, which was absent from the previous version of the policy.

Those who change their sex in medical records are also not always automatically called for relevant cancer screening programmes, such as breast and cervical screening for biological women who identify as male.