Sex and ‘gender identity’ are two different concepts so it makes good sense to collect data on both when needed and to take care not to conflate them. Never more so than in the NHS where a mix-up could be dangerous for patients and where the collection of accurate datasets regarding sex differences and sex-linked health issues really matters.
Despite this the NHS has been gradually phasing out the collection of data on sex for years and replacing it with concept of self-declared gender identity. This change has been driven by LGBT representatives who have been influencing the decision-making about whether collecting data on sex is necessary or important. We have found no evidence that data experts or stakeholder representatives for sex-based rights have been consulted at all. This article describes how, when and where the changes happened and what needs to be done next.
Pre-2002: NHS sets up the capability to record sex and gender separately but fails to use it.
The idea that the NHS should recognise and collect both types of data (sex AND gender) was established over two decades ago. The NHS followed the categorisation first seen in the Government’s Data Standard Catalogue in 2002.
NHS data relating to sex and gender originally came in two forms: PERSON GENDER AT REGISTRATION (sex registered at birth) and PERSON GENDER CURRENT (the gender someone currently identifies with). There used to be a data element called SEX OF PATIENTS as was used specifically for ward allocation.
The rather confusing use of the word ‘gender’ in the names of first two categories was flagged by a patient safety team in 2009 who recommended “The term ‘Gender’ is now considered too ambiguous to be desirable or safe because different locations and systems use it to mean different things. It is therefore desirable to use the two distinct terms ‘Sex’ and ‘Current Gender’.”
2016: NHS settles on two clear categories, sex and stated gender…
All these early NHS data categories were eventually retired around 2016/7 and the NHS data dictionary now uses:
PERSON PHENOTYPIC SEX (Male or Female) is defined as “A classification of the observed sex of a person, relating to the biological, physiological and physical characteristics that differentiate men and women, determined by endocrine influences. [It] is observed by a person (such as a care professional), and is not self-stated”
PERSON STATED GENDER (Male or Female) is defined as “self declared, or inferred by observation for those unable to declare”.
However, while the use of the word ‘sex’ helps distinguish it from the ambiguous concept of ‘gender’, it unhelpfully introduces the idea that ‘sex’ is only something a professional can ‘observe’ whereas ‘gender’ is something self-declared by the patient. This first shift in the definitions means that ‘sex’ has now become unnecessarily medicalised. This is odd when we all know what sex we were born (even the people who don’t identify with it) and everyone can and does self report their own sex when asked. We don’t need a medical professional to tell us what sex we are, except in the very rarest of circumstances.
…but fails to use them
But having created a specific data category to record ‘sex’ it was almost never used in any of the mandatory NHS national datasets or NHS patient demographic database.
(a) National NHS datasets:
The NHS collates a wide range of clinical and administrative datasets. They are used to provide a comprehensive national picture of the use of NHS services and are used by policy makers and commissioners for service planning and to monitor government policies and legislation.
Despite sex being a fundamental demographic variable of relevance to policy making and planning PERSON PHENOTYPIC SEX is absent from almost all NHS datasets. The data category PERSON PHENOTYPIC SEX only appears in three NHS datasets; the maternity services dataset and two neonatal datasets.
Whereas for all the other national datasets PHENOTYPIC SEX is never included. PERSON STATED GENDER is found in 20 active datasets and PERSON GENDER CODE CURRENT is found in another 16 (Note: This code is flagged to get changed to PERSON STATED GENDER the next time these datasets get updated).
(b) The NHS patient demographic database:
PHENOTYPIC SEX data is also missing from the PDS database. PDS is the national electronic database of all NHS patient demographic data, such as name, address, date of birth and NHS number. It supports quick and accurate identification of a patient, contact and communication with a patient, and linkage of data and records across care settings and information systems. The area was first investigated by Anne Harper Wright, who in 2018 requested a copy of what was recorded in her own demographic records. PERSON PHENOTYPIC SEX was unpopulated whereas PERSON GENDER CODE said female. This has operational consequences. The PDS database is used to contact patients when they need to attend check-ups etc. This is why female-born transmen are warned that they will not receive automatic invitations for cervical screening. Screening invitations have to be sent out based on the entry in the ‘gender’ category because data on ‘sex’ is simply not available in the database. (PDS guide, page 26)
2014: The stealth introduction of the concept of gender identity, starting with HIV/AIDS data
The meaning of the word ‘gender’ in PERSON STATED GENDER is ambiguous and when most people are asked “what is your gender” they will answer with their sex, male or female. Some people might answer in terms of what they wish their sex was and/or how they might present to others, so the data captured will be a mixture of mostly biological sex and sometimes gender identity. However, 2014 marked another shift in category definitions such that GENDER IDENTITY started to be recorded specifically.
This change wasn’t rolled out everywhere. At first it only appeared in one of the NHS national datasets: the HIV and AIDS reporting data set.
The version of this dataset rolled out in 2014 contained a new data category called GENDER IDENTITY CODE FOR HIV with three available options to select from: male, female or transgender. However, alongside this new category it was still recording PERSON GENDER AT REGISTRATION (the precursor to PERSON PHENOTYPIC SEX). The importance of collecting accurate data about an individual’s sex for a disease that is more common in males than females, and commonly attributed to male-to-male sexual contact was, back then, still clear.
In 2016 the HIV/AIDS dataset got another refresh. This time much more extensive.
The GENDER IDENTITY CODE FOR HIV was dropped and replaced with GENDER IDENTITY CODE (SEXUAL HEALTH). This change expanded the options to select from to include ‘non-binary’ and ‘other’, and modified the standard binary choices of ‘male’ and ‘female’ to ‘male (including trans man)’ and ‘female (including trans woman)’. This was a very significant change away from a existing administrative standard based on the sex binary of male and female towards an ideological concept based on a gender identity spectrum. However, the terminology of sex (male and female) WAS retained, giving the misleading impression that the sex of a transwoman is female and that sex itself is a spectrum. This conflation of sex and gender identity is confusing and misleading.
At the same time, PERSON GENDER AT REGISTRATION (sex at birth) was dropped from the dataset at the request of the LGBT Foundation and ‘transgender community’. The Requirement Specification documentation suggested this was because they had been told it may not be legal to ask. This was clearly incorrect advice accepted without the necessary due diligence by the NHS. WPUK has obtained a QC opinion that confirms that collecting data on sex is lawful, so long as it is a proportionate means of achieving a legitimate aim. Notably, the high court ruling about the guidance accompanying the sex question in the Census was also in line with this opinion. A simple question (as used in the Census) such as ‘What is your sex? Male or female’ can be asked when accurate information on sex is required. Illegality should never be used as justification for not collecting accurate data on sex directly.
Extract from Requirement Specification: “gender birth – work with the LGBT Foundation and the transgender community has informed that it is not good practice or possibly even legal to ask for gender at birth therefore we propose to remove this field from the dataset”
Instead of simply and clearly asking about someone’s sex a new category was introduced called GENDER IDENTITY SAME AT BIRTH (yes/no). This serves two purposes, one worthwhile, the other less so. It usefully identifies the subset of patients who identify as transgender, and as such is a useful addition. However, it can also be used (in some cases) to deduce sex at birth indirectly from a patient’s GENDER IDENTITY data. However, replacing a direct question about sex with an indirect method to deduce sex is a poor substitute and means data quality is inevitably compromised. For example:
- There is a risk people won’t understand a question that uses unfamiliar terms such as ‘gender identity’ and ‘gender assigned at birth’ whereas the general public readily understands the question, “What is your sex?”.
- Sex can only be inferred from 2 out of the 5 possible answers to the gender identity question. Sex data will be lost for gender diverse patients who self-declare their identity as ‘non-binary’ or ‘other’.
- It also means that if a patient wants medically-relevant information regarding their birth sex to be recorded they will have no choice but to provide a ‘gender identity’ – whether they have one or believe in this concept or not. Choosing the option “asked but declined to provide a response” means the opportunity to indirectly capture sex data to be deduced is lost. This new format clearly puts people who do not believe in a concept of gender identity at a particular disadvantage compared to people who do hold the belief. Lack of belief in gender identity is a protected belief (see EAT: Forstater vs CGD) and the NHS has a legal obligation to minimise discrimination suffered by all protected groups, not just some.
2018: Extending the concept of gender identity to a second data set
In 2018 a second dataset was updated to include GENDER IDENTITY CODE (SEXUAL HEALTH) and GENDER IDENTITY SAME AT BIRTH. This was the GUMCAD Sexually Transmitted Infection Surveillance System Data Set.
The Change Specification document states the reason for the update was to be more inclusive of trans and non-binary people.
But in the same document another suggested rationale is given for incorporating the GENDER IDENTITY CODE. Worryingly, here the authors refer to ‘gender’ as if they understand it to mean ‘sex’. This is clearly at odds with what the GENDER IDENTITY CODE actually is.
“Patient gender is relevant to all aspects of sexual health service provision. Most aspects of service provision, patient management practices and health outcomes are gender specific (eg cervical cytology screening). The majority of clinical audits, assessments of service provision and of public health burden would, therefore, require data to be gender specific”
The rationale for introducing GENDER IDENTITY SAME AS BIRTH (which they refer to in the table below as patient gender at birth) again refers (correctly) to how important information about birth sex is for this dataset, and for transgender people in particular. However, since birth sex is important it is unclear why they haven’t asked it directly using the category of PERSON PHENOTYPIC SEX. The fact that this important sex data will be unavailable for non-binary members of the transgender community appears to have been overlooked.
“This is a required data field to identify and monitor the sexual health of transgender men and women, a group which has disproportionately higher risks of poor sexual health.”
This is all the more surprising when a number of other data categories for this same dataset rely on knowing the sex of sexual partners. Sex clearly matters in this dataset but even here a direct question about sex is not consider necessary.
2021: Removing the sex category from the Mental Health Services data set
In 2021 a third dataset has now been earmarked for updating. This time it’s the Mental Health Services Data Set. Changes will come into affect in October 2021. We have written about these changes in more detail here.
This time PERSON STATED GENDER CODE will be replaced with GENDER IDENTITY CODE. This is currently marked in the NHS dictionary as ‘being used for development purposes and has not yet been approved”. Unlike GENDER IDENTITY CODE (SEXUAL HEALTH) it appears that this new category will not be restricted for use in specific medical areas (note that there is no GENDER IDENTITY CODE (MENTAL HEALTH) in the NHS dictionary. The creation of this generic category for gender identity may reflect a future intention to use it more widely across all dataset as and when they each get updated.
The MHSDS Change requirement list for version 5 lists the justifications and benefits of the change.
- The change brings MHSDS into alignment with other datasets (the introduction of gender identity elsewhere is being used to justify its inclusion as the new best practice).
- The change is to designed to “support(s) a better understanding of individuals’ gender identity alongside their biological sex”. However, this change does not support a better understanding of sex, quite the opposite. While in some cases it will be possible to deduce birth sex from the answers given to the two gender identity questions this will not always be the case. Information regarding the birth sex of individuals who consider their gender identity to be ‘non-binary’ or ‘other’ will be lost.
- The justification of the change is to “link with the changes being made to the 2021 census”. This justification is now defunct. The 2021 census was required by the high court to collect data on sex (not gender identity) alongside information about whether gender identity matches sex at birth. The new MHSDS format will not link with the data categories in the 2021 census.
- Another of the stated benefits was to “allow monitoring and reporting against the NHS obligations under the Equalities Act”(sic). Sex is a protected characteristic under the Equality Act but gender identity is not. Collecting data on gender identity without a question on sex does not aid equality monitoring. How can the NHS comply with its legal obligations under Public Sector Equality Duty to show “due regard” to eliminate discrimination due to sex if it’s not even monitoring it properly?
What is becoming clear is that over the last 20 years the ability to accurately record data on the material reality of sex has been slowly eroded within the NHS. The concept of gender identity is now taking its place. We have found no evidence that data experts or stakeholder representatives for sex-based rights have been consulted as these changes were introduced. Instead, it is LGBT representatives who have been influencing the decision-making about whether collecting data on sex is necessary or important. Our stakeholder experience with ONS regarding sex in census means it’s likely that no-one was considering or speaking up for the importance of accurate data on sex. The idea that sex should be replaced by gender identity has probably never been challenged within any of the working groups responsible for these changes, and it shows.
What next?
What happens next really matters. The Department of Health and Social Care (DHSC) has commissioned NHS England to develop recommendations for a unified information standard for the nine characteristics protected under the Equality Act (2010). It is vital that the case for why accurate data on sex matters is heard in the process. We are doing work in a number of areas to make sure that happens, some of this includes:
(1) Engaging with the Department of Health
We have asked DHSC for an update and have recently been told NHS has delivered its recommendations for change and that stakeholder engagement will take place before implementation. We have asked to be notified when this stakeholder engagement begins and take we and other should be invited to take part. We have also requested to view the 50 recommendations under Freedom of Information law.
“The Unified Information Standard for Protected Characteristics (UISPC) literature review was received by the Department from NHSEI. The Review made over 50 recommendations, which the team are reviewing before taking to Ministers, and which will need due consideration for implementation, involving significant stakeholder engagement. Therefore, it is a little too early to say what will be implemented and what that timetable for implementation will look like”.
(2) Engaging with Office for Statistical Regulation
We have recently written to and met with the Director General of the OSR to raise our concerns with the changes that have already been made. The value of collecting accurate data on sex has clearly been overlooked and as such this may represent a breach of the Code of Practice for Statistics for Trustworthiness, Value and Quality. We have asked OSR to investigate and to help make sure the Code of Practice is followed during the future development of the Unified Information Standard. In particular, the following standards must be strictly upheld:
Q1.5 The nature of data sources used, and how and why they were selected, should be explained. Potential bias, uncertainty and possible distortive effects in the source data should be identified and the extent of any impact on the statistics should be clearly reported.
Q2.1 Methods and processes should be based on national or international good practice, scientific principles, or established professional consensus.
Q2.6 Statistics producers should collaborate with topic and methods experts and producers of related statistics and data wherever possible.
Q3.5 Systematic and periodic reviews on the strengths and limitations in the data and methods should be undertaken. Statistics producers should be open in addressing the issues identified and be transparent about their decisions on whether to act.
V1.4 Statistics producers should engage publicly through a variety of means that are appropriate to the needs of different audiences and proportionate to the potential of the statistics to serve the public good. An open dialogue should be maintained using proactive formal and informal engagement to listen to the views of new and established contacts. Statistics producers should undertake public engagement collaboratively wherever possible, working in partnership with policy makers and other statistics producers to obtain the views of stakeholders.
(3) Engaging with EHRC
As a public body the NHS is legally obliged to follow Public Sector Equality Duty. This means it must show due regard to each of the protected characteristics and take steps to minimise discrimination in any of its policies. Good progress is being made to improve equality monitoring capabilities for the protected characteristic of gender reassignment by introducing new questions on gender identity and recording whether that gender identity is different from the sex they were born. However, it is unclear how equality monitoring for the protected characteristic of sex can be done effectively without a clear and direct question about sex also.
We think the EHRC could, and should, play an important role in ensuring equality monitoring and data capture processes work for all. While people’s identities are important to their own lives, and collecting data on identity can often be useful, this should not be done at the expense of collecting accurate data on sex. We have written to and met with senior staff at EHRC asking them to investigate and bring clarity in this area. Does the current data collection format indirectly discriminate on the grounds of sex? How does it impact people who share the protected characteristic of gender critical belief? Is it, as some LGBT organisations had advised, ‘illegal’ to collect data on sex?
To stay informed about progress in this important area please subscribe to our newsletter.
If you would like to help us continue the work we do you can also set up a regular donation.