Early version, shared for feedback — the wording and the
figures may still change. Tell me what is wrong with it.
Census Data
2021 Census Data for England and Wales: Trans* Health
The published figures understate how bad trans* health outcomes are, because trans* people are overrepresented in younger age groups
trans* here means everyone the census
counted as having a gender identity different from their sex registered at birth
— transgender men, transgender women, nonbinary people, and everyone else who
said so. The asterisk is a reminder that the group is wider than transgender
men and transgender women. Where a figure is about one of those groups on its
own, it says which.
The Census 2021 asked everyone in England and Wales aged 16 or over about
their gender identity, and separately about their general health. The
published cross-tabulation of the two says that trans* people report worse
health than cis people. That is true, but it understates the problem.
This is because trans* people are on average much younger than cis people, and
health usually gets worse with age. 48% of the
trans* population is under 35, so on age alone, they should look
healthier than a cis population. But they don't. This is happening
despite a substantial age advantage.
Taking the age effect out needs age, health and gender identity broken down
together, which the basic published tables don't offer. The Office for
National Statistics' dataset
CT21_0298 does this, showing age
by general health by gender identity. Here we break down what it shows.
1.42×
How much more likely trans* people are to report bad health, all ages together
4.35×
Among 16 to 24 year olds alone
Three populations of very different shapes
These charts show who is in each group. There are three populations: cis
people, trans* people, and the 6% of the
population who left the voluntary gender identity question blank, which likely
contains both cis and trans* people.
The populations are broken down by age groups. The top line
of charts shows the absolute scale of each population in comparison to each
other, all three on one scale — which is why the trans* bars barely register:
cis people outnumber trans* people 173 to one. The
second line shows the age distribution as a percent of that population, which
is the comparison that matters.
This illustrates that the trans* population is comparatively young:
24% of trans* people are 16 to 24, against
13% of cis people, and the trans* population thins
out fast after age 45. The people who skipped the question run the other way —
they are the oldest of the three, with 15% aged
75 or over against 10% of cis people.
Cis people
Trans* people
Did not answer
The gap in health, by age band
Trans* people report bad or very bad health at a higher rate in every age
band. The gap is not constant, though, and its shape is the most striking
thing in this dataset: it is widest among the young. Trans*
16 to 24 year olds are 4.35 times as likely as cis people
of the same age to report bad health. By 75+ the ratio is 1.08
— barely a difference at all.
That gradient does not appear in the ONS's published tables, and it is the opposite
of what an ageing effect would normally look like.
The non-responders sit between the two, and consistently above cis
people, especially in the older age ranges. That means that the
6% of people who did not answer the gender
identity question are, as a group, in worse health than the people who said
they were cis — so the people who did not answer were not a random subset of
the cis population. Since that group certainly contains trans* people, and
since we cannot know how many or which, it is a genuine source of uncertainty
in both directions.
The chart below shows bad or very bad health across age. For
all three groups, the proportion of people in bad or very bad health increases
with age, but the trans* line starts far higher and does not converge until
the oldest band. The dashed line is everyone who skipped the question.
Trans* people are not one group, and the differences inside the trans*
population are larger than the difference between trans* and cis people
overall.
The next chart breaks out the trans* identities that were separately coded by
the census. Each trans* identity has its own colour, and is named at
the end of its line. Nonbinary people and people reporting other identities
have worse health than transgender men and transgender women at every age.
† the line stops where the census counted too few people in the remaining age bands to report a rate.
Transgender men and transgender women track each other closely with those who
didn't report a specific identity also following a similar pattern. Nonbinary people
and people who described their identity in some other way sit far above
both with worse health outcomes at every age where the census counted enough of them to say so.
Age and health together
The difference sits at the good end of the scale too. Among
16 to 24 year olds, 65% of cis people report very
good health against 39% of trans* people —
26 percentage points — while the gap in bad or
very bad health is 5 points. What is happening is
that very good health is also becoming good or fair, as well as people
reporting the worst category.
These charts show the breakdown of all five health responses by gender
identity by age group. The height of each column is that age band's
share of the population for that population, so the columns reproduce the age
structure. The stack inside each column splits it by health, with the worst
outcomes on the baseline.
Read the columns and you get the age difference. Read the orange at the bottom
of each column and you get the health difference. The trans* panel is both
younger and, band for band, worse off.
Very good
Good
Fair
Bad
Very bad
Within single age bands. The below panels details health
outcomes for people of the same age across the different populations, by % of
that population, for every age band. This is similar to the chart above but
allows easier comparison between populations at the same age.
Very good
Good
Fair
Bad
Very bad
What this data cannot tell you
The census is a good source and a limited one, and the limits matter more than
usual here.
The question was voluntary. 2,914,625
people aged 16 or over did not answer it. They are shown throughout as their
own group, but that group is a mix of cis and trans* people in unknown
proportion — it cannot be added to either. Their health is worse than cis
people's at every age, so the non-response is clearly not random, and which
way it biases the trans* figure is unknowable from this data.
ONS classes these as official statistics in development,
and has said some respondents may have misread the question, particularly
people with lower English proficiency. If that happened, the trans* column
contains people who are not trans*, whose health will look more like the
general population's — so the effect, if anything, is to dilute the gap
shown here rather than to manufacture it. The
ONS
quality information has the detail.
Counts are perturbed to protect anonymity, and the smallest
cells are affected most. The census counted 12
nonbinary people aged 75 or over. Nothing on this page draws a conclusion
from a cell that small: the identity breakdowns stop where the counts do,
and the by-identity age panels are restricted to
16 to 54.
Health here is self-reported — one subjective question, not
a clinical assessment.
This is an association. The data cannot say why the gap
exists. Minority stress, discrimination inside healthcare, poverty, the
health effects of long waits and refused treatment, and the possibility that
poor health affects who answers a voluntary question are all consistent with
these numbers. The census cannot separate them.
Further charts
The same constructions as above, broken down by gender identity rather
than a single trans* group. These carry the small-count problem, so read them
with the caveats in mind — but the differences between identities are large
enough to be worth showing.
Who is in each age band. The first chart on the page again,
trans* people only: every row is that age band's share of the trans*
population, so the rows still add up to the whole, and each bar is split into
the five identities the census recorded. The row underneath is all ages
together — 100% by construction, so it is drawn to fill the width rather than
sharing the bands' scale.
Transgender women
Transgender men
Nonbinary people
No specific identity
Other identities
Age and health, by identity. The same construction as above,
one panel per identity. All seven still share one scale so the shapes are
comparable.
Very good
Good
Fair
Bad
Very bad
Within age bands, by identity. Among 16 to 24 year olds,
65% of cis people report very good health against
24% of nonbinary people.
Very good
Good
Fair
Bad
Very bad
Source
Office for National Statistics, Census 2021 custom dataset
CT21_0298 — Age by general health by
gender identity, created 20 June 2024. All usual residents
aged 16 or over, England and Wales. Licensed under the
Open
Government Licence v3.0. Every figure and chart on this page is generated
directly from that spreadsheet.