Thursday 13 August 2020
An analysis of the impact of HIV on COVID-19 death from primary care data has been published online concluding that ‘HIV infection was associated with a markedly raised risk of COVID-19 death’1. As the data have been posted as a pre-print, we write to provide an initial comment while we wait for the paper to undergo peer review. We note the caveat attached to all pre-prints “new medical research that has yet to be evaluated and so should not be used to guide clinical practice.”
An analysis of people hospitalised with COVID-19 from the national ISARIC database2, also recently posted as a pre-print, showed a 1.63-fold higher risk of death for people living with HIV, compared to a 2.3-fold excess risk reported by the primary care analysis. Of note, unlike the primary care analysis, the ISARIC analysis showed no impact of ethnicity on outcomes after hospitalisation, a discrepancy that requires more research.
Whilst we welcome any data that can help inform policies and advice for people living with HIV, especially as findings to date have been conflicting, we are concerned that the conclusions drawn by the primary care analysis may be too strong considering the limitations.
The analysis is based on primary care coding (data sourced from GP surgeries). The errors in communications about shielding from the Department of Health has led us to believe that primary care codes for HIV may be incomplete or inaccurate3.
It is possible that people with HIV included in the primary care data may have different risk factors for poor COVID-19 outcomes (such as co-morbidities) to people not included.
People with missing weight data were assumed to be not obese which, if incorrect, could impact the results. Adjusting fully for confounders (i.e other factors that could explain a difference between groups of people) is a problem for incomplete datasets.
People with poorer HIV outcomes could be over-represented in the study. No data on HIV treatment, CD4 (current or past) or viral load was available. The study assumes that since most people with diagnosed HIV are undetectable on treatment, the studied population was likely to be too. This may be incorrect.
COVID-19 deaths may be overestimated if someone who has tested positive for COVID-19 dies of another cause. If testing rates differed for people with and without HIV this could affect the results.
This paper, alongside the ISARIC analysis, adds to evidence suggesting a higher risk of COVID-19 death for people with HIV, as described in the updated BHIVA/THT statement4, an important signal warranting further investigation. However, the lack of complete data, in particular the absence of HIV specific data on treatment, previous and current CD4 and viral load, is why we encourage collaboration between groups to link results. In addition, BHIVA is planning to collect data from all HIV clinics, which we hope will better determine what drives risk of worse COVID-19 outcomes for people living with HIV.
At present we continue to advise people with HIV to follow the advice outlined by BHIVA and THT4 and call for the data that has emerged since the original Department of Health guidance on people at higher risk of COVID-19 to be used to guide any subsequent updates to that advice. Importantly, most people with HIV who died had other co-morbidities associated with COVID-19 death. Finally, we encourage groups producing data with potentially significant conclusions to work with HIV societies and community organisations to develop appropriate messaging for people providing HIV care and, most importantly, for people living with HIV.
For further information, please contact Jo Josh on +44 (0)7787 530922 or firstname.lastname@example.org.
1. https://www.medrxiv.org/content/10.1101/2020.08.07.20169490v1 accessed 12th August 2020
2. https://www.medrxiv.org/content/10.1101/2020.08.07.20170449v1 accessed 12th August 2020