BHIVA registry reports HIV is independently linked to worse presentation and outcomes from COVID-19

Simon Collins, HIV i-Base

An analysis of outcomes from COVID-19 in HIV positive people were collected as part of a BHIVA audit and the results was presented at the BHIVA conference. Although retrospective data was collected from October 2020 until the end of March 2021, participating clinics included all cases from January 2020.

Audit leads were invited to submit details of people attending their services with suspected/confirmed COVID-19, with data collected through the BHIVA audit system.

The analysis used two regression analyses looking at severity of symptoms at presentation (based on use of oxygen ventilation) and factors linked to worse outcomes (based on extended hospitalisation, death or continued symptoms >3 months).

Multivariate analyses covered three main categories: regular demographics (including employment), clinical/lifestyle risks (including a new comorbidity score), and HIV-related factors. Notably, a double weighting was given for uncontrolled compared to controlled comorbidities, and also for recent HIV viraemia (>200 copies/mL) compared to having an undetectable viral load on ART.

Overall, the registry included 1310 cases: approximately 50% were older than 50 years, 40% were women and 50% lived in London or the South. Ethnicity included 47% white, 37% black African and 15% other/unknown. An occupational risk was reported by 35% (mainly health and care work) and 16% had a recent household contact with a confirmed case.

Symptoms at presentation

Just under 80% or cases reported symptoms: mainly fever (47%), cough (51%), shortness of breath (36%) or anosmia (23%). Although within this group, only 60% were confirmed by PCR, this percentage was much higher (95%) for those without symptoms.

HIV demographics included median CD4 count of 611 cells/mm³ (IQR: 437 to 812) and but a significantly lower CD4 nadir of 257 cells/mm³ (IQR: 123 to 410). Median CD4:CD8 ratio was 0.8 (IQR: 0.57 to 1.15) with 9% being <0.4. Roughly 4% had a current AIDS diagnosis, 18% a previous AIDS event, and 15% had recent HIV viraemia.

The median comorbidity score was 1 (but ranged from 0 to 13), mostly controlled, with the most common being hypertension (25%), obesity (20%), dyslipidaemia (17%) and diabetes (12%). Median BMI was 28 kg/m² (IQR: 24 to 32) and 12% were current smokers.

Overall, 24% of cases required hospital admission, 8% to ICU, 16% needing oxygen support and 5% mechanical ventilation. This meant 230/1310 (17%) were categorised with severe presentation.

Although most factors were associated with poor presentation in the univariate analysis, only age (p=0.0001), being female (p=0.0002), black African ethnicity (p=0.0001), BMI (p=0.0002), comorbidity score (p=0.0001) and previous AIDS (p=0.005) highly significant in multivariate analysis.

Importantly, higher latest CD4 count was protective (p=0.04) suggesting an independent association of HIV.

Outcome results

Outcomes were available for 985/1154 participants (130 were still within the three month window) of which 85% (n=985/1154) were positive. However, 169/1154 (15%) had a poor outcome linked to persisting physical health problems.

In multivariate analysis, poorer outcomes were significantly associated with: older age (p=0.06), higher comorbidity score (p=0.0001), shortness of breath (p=0.0001) and anosmia (p=0.006). Again, CD4 count <200 cells/mm³ (p=0.02) was also significant, supporting an independent effect of HIV.

However, when severity of presentation (p=0.0001) was added to the model, only total comorbidity score (p=0.006) and shortness of breath (p=0.0006) remained significant, suggesting other factors were significant because they drove a poorer presentation.

comment

Although as the study can’t comment on prevalence and incidence of COVID-19 in HIV positive people, low CD4 count and previous AIDS independently linked to worse outcomes, supporting a causal role of HIV.

The study did not report on mortality (approximately 6%) because initially the registry only expected several hundred cases overall and this would have been too few for a separate outcome. However, the data on deaths will be included as a sensitivity analysis in the full paper.

Reference

Sabin C et al. Coronavirus (COVID)-19 in people with HIV in the UK: Initial findings from the BHIVA COVID-19 Registry. BHIVA BASHH 2021. Oral abstract O-008.
https://bhiva-bashh.org/sessions-posters/session-13-day-2

(This link should be open access from approximately 20 May 2021).