Data on patients newly diagnosed with HIV and those accessing HIV care is collected and processed using the HIV and AIDS Reporting System (HARS) dataset. The dataset includes laboratory reports of patients’ CD4 blood counts.
HARS data is submitted every 3 months by all outpatient HIV service providers.
The data on diagnoses of HIV at a late stage is collected, validated, de-duplicated and matched using HARS by the HIV & STI department of PHE’s National Infections Service.
Data is as reported by service providers but relies on ‘record linkage’ to integrate data and ‘de-duplication’ to prevent double counting of the same individual. Every effort is made to ensure accuracy and completeness of the data, including web-based reporting with integrated checks on data quality. However, responsibility for the accuracy and completeness of data lies with the service providers that submit the data. Service providers have to follow a data sharing policy to ensure data is consistent and patient confidentiality is protected.
Data supplied for previous years is updated every year to account for clinic/laboratory resubmissions and improvements to data cleaning.
The confidence intervals for the estimate of the percentage of all newly diagnosed adults with late HIV diagnosis are available if you download the data.
In the period from 2015-17, the percentage of late HIV diagnoses for people of White ethnicity aged 15 years and over was 36.9%. This is a reliable estimate based on the number of people diagnosed with late stage HIV infection as a proportion of those who were newly diagnosed, but it is not possible to be 100% certain of the true rate of late diagnosis for this population due to the effect of natural variation.
It’s 95% certain, however, that the rate of diagnosis of HIV at a late stage of infection lies somewhere between 35.7% and 38.2% of White people who were newly diagnosed with HIV. In statistical terms, this is a 95% confidence interval. This means that 95 times out of 100 the estimate would fall in this range (i.e. between the upper and lower confidence interval). But 5 times out of 100 it would fall outside this range.
The smaller the number of new diagnoses, the more uncertain the estimate and the wider the confidence interval. To illustrate this, in the period from 2015-17 there were fewer diagnoses of HIV infection among people from the Black Other ethnic group than for White people, so we can be less certain about the estimate of 51.3% diagnosed with HIV infection at a late stage for the Other Black ethnic group. This greater uncertainty is expressed by the wider confidence interval of between 44.0% to 58.5% of people from the Other Black ethnic group, who were newly diagnosed at a late stage of HIV infection.
Statistically significant differences are where the 95% confidence interval around the estimate for an ethnic group does not overlap with the confidence interval around the estimate for the group it’s being compared with.
Percentages are rounded to 1 decimal point. Download the data for unrounded figures.
HIV annual data tables and report present national data. Local level data is available for related indicators in the Sexual and Reproductive Health Profiles tool but the local data cannot be broken down by ethnicity.
Further technical information
To reflect the 2008 British HIV Association (BHIVA) treatment guidelines, the definition of a late HIV diagnosis was updated from a CD4 count of less than 200 cells/mm3 within 91 days of diagnosis to less than 350 cells/mm3 within 91 days of diagnosis. The data for all of the years presented here reflects the updated definition of a late-stage infection (of fewer than 350 cells/mm3). The method for calculating confidence intervals at 95% significance was based on the exact binomial (Clopper-Pearson) interval. This method produces a slightly wider confidence interval, and is thus slightly more conservative in determining significant difference, compared to the Wilson score interval. Further information on HIV surveillance can be found on GOV.UK in the HIV: surveillance, data and management collection.