There may be fewer people living with undiagnosed HIV in the UK than previously thought

Public Health England’s estimate of the number of people who have HIV without being aware of it has been lowered, for the second year running. However, this does not necessarily reflect an improvement in the real world situation, but is due to changes in the way this information is estimated.

Public Health England (PHE) now estimates that 13% of people living with HIV are undiagnosed – considerably lower than the 24% figure that was provided two years ago. According to their current methods and data, the undiagnosed proportion is no higher than 13% in any of the populations that are most affected by HIV in England (gay men, black African men, black African women, injecting drug users). However, amongst heterosexual men and women who are not of black African ethnicity, 20% are thought to be undiagnosed.

One consequence of there apparently being fewer undiagnosed people than previously thought is that the figure for the total number of people living with HIV is lower than before. In the report issued today, PHE estimates that 101,200 are living with HIV.

Another consequence is that the UK appears to be closer to meeting the UNAIDS 90-90-90 targets than before – the first of the targets is for more than 90% of people living with HIV to be diagnosed.

The challenge of counting people who don’t know they have HIV

By its nature, information on undiagnosed infection is challenging to produce.

The number of people living with diagnosed HIV is relatively simple to calculate – there are reliable data on new diagnoses and on people attending HIV clinics. These include some information about the age, gender and exposure group each person belongs to.

However, the number of people with undiagnosed HIV has to be estimated. One important source of data used to be anonymised blood samples which had been taken in sexual health clinics and from pregnant women. However, these samples are no longer collected, as almost all people attending these settings get tested for HIV anyway. This has left the epidemiologists to make more complex calculations and extrapolations.

For example, when there are data on new HIV diagnoses made in a particular setting, this also provides information about infections that had – until then – been undiagnosed. Further, it is assumed that in addition to people who have their infection diagnosed in a clinic, other people will have refused or not been offered a test and will have left a clinic with an undiagnosed infection.

So it is not the case that the epidemiologists now have much better sources of data to work with and that we can be much more confident in the current estimates than in those produced a few years ago. In fact, some key sources of data are no longer available.

Although it is simpler to look only at the headline figures, it is more informative to look at the ‘credible range’ around those figures. For example, PHE say that there are 101,200 people (97,469-105,700) living with HIV. The figures in the brackets indicate the range of figures within which the true number is thought to lie. There could be as few as 97,469 people living with HIV, while there could be as many as 105,700.

When the difference between the lower and upper numbers is quite small, this indicates that we can be relatively confident in the reliability of the main figure. When the difference is large, this shows that there is more uncertainty. In general, this is because there are fewer sources of reliable data which can be used to inform the estimate.

Gay men: undiagnosed HIV

PHE report on three different models which have attempted to estimate undiagnosed HIV in gay, bisexual and other men who have sex with men. This range of methods is not available for any other population group.

According to the PHE’s primary method, the multi-parameter evidence synthesis, there are a total of 47,040 gay men (44,219-50,860) living with HIV. Of these, 5830 (3199-9639) are undiagnosed. In other words, 12% (7-19%) are undiagnosed.

Another method is based on the CD4 counts of gay men diagnosed with HIV. (For example, a man diagnosed with HIV at a CD4 count of 400 is likely to have been living with undiagnosed HIV for the previous four years.) According to this method, 4700 gay men (2300-8400) may have undiagnosed HIV.

PHE note that these two methods are based on different types of data. Therefore the fact that the two estimates are broadly similar can give us confidence in their reliability. Of note, the headline figure produced by each method is within the credible range of the other method. “It is likely that the true number of gay/bisexual men living with undiagnosed HIV lies between the two [figures],” comment PHE.

A third method provides another estimate, but it relates to 2013 rather than 2015. According to the synthesis progression model, 19% of gay men living with HIV (9-28%) were undiagnosed. Again, 19% is within the credible range of the first method, which was 7-19%.

Heterosexual men and women: undiagnosed HIV

PHE gives the following estimates for the number of people living with HIV, including those undiagnosed. These are based on PHE’s primary method, the multi-parameter evidence synthesis.

If these figures are accurate, they suggest that concerns in previous years about a large proportion of African people, especially African men, being undiagnosed could have been unfounded.

However, it does suggest that undiagnosed infection is a particular problem for people who do not belong to readily identifiable risk groups – heterosexuals who are not African. With HIV testing rarely being offered in a universal manner, this problem is likely to persist.

Late diagnosis

The data on late diagnosis could also tell us about undiagnosed HIV, as people who are diagnosed late (with a CD4 cell count below 350 cells/mm³) must have been living with undiagnosed HIV for several years. Moreover, these figures reflect observed data, rather than being modelled and estimated.

Overall rates have fallen over the past decade, from 56% diagnosed late in 2006 to 39% diagnosed late in 2015.

But these figures do suggest inequalities between demographic groups – whereas 30% of gay men were diagnosed late, 60% of black African men and 52% of black African women were diagnosed late. As has been consistently reported over the years, an HIV diagnosis is more likely to be late for heterosexuals, older people, black people and individuals living outside of London.

This inequality is in contrast with the estimates of the proportion undiagnosed, which did not find large differences between gay men and other groups.

Commenting on this part of the report, Dr Chloe Orkin, chair of the British HIV Association (BHIVA) said: “The fact that 39% of newly diagnosed patients are being diagnosed late remains deeply troubling. This phenomenon is most significant among heterosexual men and women who are not recognised as being at risk and therefore remain undiagnosed for too long. We must continue to push for all doctors in emergency departments, GP surgeries and general hospital settings to be adequately funded and fully empowered to offer and undertake HIV testing.”

New HIV infections and diagnoses

The report includes an estimate of annual new infections (incidence) in gay men – the figure is unchanged from year to year at 2800 (credible range 1700-4400). This suggests that current levels of testing, treatment and condom use are failing to curb the spread of new infections.

As previously reported, the number of new diagnoses is also essentially unchanged from year to year. Last year, 6095 people received a new HIV diagnosis, including 3320 gay men, 1350 women, 1010 heterosexual men and 160 people who inject drugs.

People seen for HIV care

Just under 89,000 people attended HIV medical services in the UK in 2015. Standards of care are very high – 97% of newly diagnosed individuals were linked to care within three months of diagnosis; 93% of people were retained in care from year to year; 96% of people attending care received antiretroviral treatment; 94% of people on treatment had an undetectable viral load.

Or, in terms of the 90-90-90 targets:

• 87% of people living with HIV are thought to be diagnosed
• 96% of diagnosed people are on treatment
• 94% of people on treatment have an undetectable viral load.

People are starting treatment earlier: 66% of those beginning treatment in 2015 had a CD4 cell count above 350 cells/mm³, including 41% with a CD4 cell count above 500 cells/mm³. This compares with 22% and 10% respectively, a decade ago.

Diversity of the HIV epidemic

“People living with diagnosed HIV in the UK represent a diverse group and assumptions about the characteristics of those living with HIV need to be challenged,” say PHE.

Due to changing migration patterns, black African people no longer make up the overwhelming majority of heterosexual men and women living with HIV. Whereas 61% of heterosexuals attending HIV medical services in 2015 were of black African ethnicity, 25% were white, 7% were of other black ethnicities and 4% were Asian.

At the same time, 14% of gay men attending HIV medical services were of non-white ethnicity. (And two in five newly diagnosed gay men were born abroad).

Over a third of those attending care were over the age of 50.