Outcomes of COVID-19 related hospitalisation among people with HIV in the ISARIC Clinical Characterisation Protocol UK Protocol: prospective observational study

Background. There is conflicting evidence about how HIV infection influences COVID-19. We compared the presentation characteristics and outcomes of people with and without HIV 95 hospitalised with COVID-19 at 207 centres across the United Kingdom. Methods. We analysed data from people with laboratory confirmed or highly likely COVID- 97 19 enrolled into the ISARIC CCP-UK study. The primary endpoint was day-28 mortality 98 after presentation. We used Kaplan-Meier methods and Cox regression to describe the 99 association with HIV status after adjustment for sex, ethnicity, age, indeterminate/probable 100 hospital acquisition of COVID-19 (definite hospital acquisition excluded), presentation date, 101 and presence/absence of ten comorbidities. We additionally adjusted for disease severity at 102 presentation as defined by hypoxia/oxygen therapy.

comorbidities at admission (a series of binary variables to indicate the presence or absence of 2 4 0 each of chronic cardiac disease, chronic pulmonary disease, chronic renal disease, diabetes, 2 4 1 obesity, chronic neurological disorder, dementia, liver disease [mild, moderate or severe], 2 4 2 malignancy, and chronic haematological disease). We also included adjustment for the 2 4 3 baseline date to account for changes in mortality over the period of interest. Where entries on 2 4 4 comorbidity (presence or absence) were partially missing from the study CRF, we assumed 2 4 5 that missing data indicated the absence of the specific comorbidity; however, participants 2 4 6 with missing entries on all comorbidities were excluded from these adjusted analyses. Finally, we fitted a further model with additional adjustment for hypoxia at presentation, 2 4 8 defined as oxygen saturation (SpO2) <94% on air or a record of receiving oxygen, as a 2 4 9 marker of presenting disease severity, in order to assess whether any increased/decreased risk 2 5 0 of mortality in PWH could be explained by a different stage of disease advancement at 2 5 1 hospitalisation. A series of sensitivity analyses were performed for the main mortality 2 5 2 outcome: i) we repeated the analyses after censoring follow-up on the day of discharge for 2 5 3 those discharged before day 28; ii) we included those with definite hospital-acquired COVID- 19 (baseline = date of symptom onset); iii) we used symptom onset date as the baseline date 2 5 5 for all (rather than admission date where applicable); iv) we excluded PWH lacking a record 2 5 6 of ART; v) we calculated propensity scores for HIV-positive status using a logistic regression 2 5 7 model based on sex, ethnicity, age (in quadratic form), indeterminate/probable hospital 2 5 8 acquisition of COVID-19, smoking status, baseline date, and ten comorbidities, and included 2 5 9 the propensity score in a Cox regression model for death at 28 days; and vi) we considered a 2 6 0 binary endpoint of 14-day mortality and performed logistic regression (with the same 2 6 1 confounder adjustment as described above). In the HIV-positive group, we used a Cox 2 6 2 proportional hazard model to investigate the associations of presenting characteristics with 2 6 3 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020.08.07.20170449 doi: medRxiv preprint 1 0 day-28 mortality. Analyses were conducted in Stata v16·1 (Statcorp, College Station, TX, 2 6 4 USA). characteristics of patients excluded from the analysis did not differ by sex, ethnicity or age; in 2 7 3 particular, the characteristics of those excluded due to missing data on HIV status closely 2 7 4 resembled those reported to be HIV-negative (Supplementary Table 1). Among PWH, one 2 7 5 person was diagnosed with HIV during the admission and 103 (89·6%) had an ART record. The regional distribution of study participants with HIV compared to the total UK population  HIV-negative group. PWH had lower prevalence of chronic cardiac disease, chronic 2 8 6 pulmonary disease, chronic neurological disorders, dementia, malignancy and 1 1 evidence of chest infiltrates did not show significant differences between the two groups. PWH presented with lower total white blood cell and platelet count, but higher lymphocyte 2 9 9 count and C-reactive protein (CRP) ( Table 3). Other laboratory parameters showed no 3 0 0 significant differences. During admission, whereas there was no significant difference in the proportion of 3 0 4 participants who received oxygen between the two groups, significantly higher proportions of 3 0 5 PWH were admitted to critical care and received non-invasive and invasive ventilation   Table 4). In the HIV-positive group, 26 (22·6%) were known to have died 3 1 5 compared to 13,955 (29·1%) of the HIV-negative group; the cumulative incidence of day-28 3 1 6 mortality was 25·2% vs. 32·1%, respectively (p=0·12, log-rank test, Figure 4 A). Whilst for age revealed higher mortality among PWH in the two younger age groups (<50 years and  In unadjusted models (Table 4), the cumulative hazard of day-28 mortality was 26% lower in stratified Kaplan-Meier analyses, adjustment for age resulted in a change in the direction of the association (adjusted HR 1·39, 95% CI 0·94-2·09; p=0·10) ( of COVID-19 and ten co-morbidities, the risk of mortality was 49% higher in PWH (adjusted 3 2 8 HR 1·49; 95% CI 0·99-2·26; p=0·06). Following additional adjustment for disease severity at presentation (based on a record of hypoxia or oxygen therapy), the risk of mortality was 63% 3 3 0 higher in PWH (adjusted HR 1·63; 95% CI 1·07-2·48; p=0·02) ( Table 4). After day 28, there 3 3 1 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020 were no deaths recorded in the HIV-positive group whereas 586 deaths occurred in the HIV-3 3 2 negative group. Sensitivity analyses showed consistent results (Supplementary Table 5). In particular, 3 3 5 censoring follow-up on the day of discharge for those discharged before day 28, including 3 3 6 patients with definite hospital acquired COVID-19, using symptom onset as the start of 3 3 7 follow-up, or excluding PWH lacking an ART record did not significantly alter the model. A  In the HIV-positive group, relative to patients who survived by day 28, patients who died 3 4 3 were slightly older and had a higher prevalence of obesity and diabetes with complications 3 4 4 (Table 5 and Supplementary Tables 6 and 7). An ART record was more often missing among  In this study of 115 HIV-positive and 47,979 HIV-negative people, we found evidence 3 5 0 suggesting a 63% increased risk of day-28 mortality among PWH hospitalised with COVID- 19 compared to HIV-negative individuals in the same dataset, after adjustment for sex, 3 5 2 ethnicity, age, baseline date, ten key comorbidities, and disease severity at presentation (as  Numerically, HIV-positive people who died were more likely not to have a record of being 4 1 2 on ART than those who remained alive at day 28. However, it cannot be stated with certainty 4 1 3 that those lacking an ART record were untreated nor therefore that lack of ART played a role 4 1 4 in the adverse outcomes. Our experience of working with large HIV datasets is that we often 4 1 5 find that people with missing data have worse mortality outcomes, simply because mortality our study being off ART despite an absent record was overall low. improved or normalised CD4 cell counts. 6 We found no evidence of increased lymphopenia 4 2 7 among PWH in our study. Furthermore, compared to HIV-negative people, PWH were more 4 2 8 likely to experience systemic symptoms with fever and also showed higher CRP levels. These observations are likely to be reflective of the younger age of PWH in our study, 27 and at the likelihood of PWH in our study being severely immunosuppressed was overall low. After careful considerations and multiple adjustments for demographics, comorbidities and 4 3 5 disease severity on admission, our initial analyses of the outcomes of patients hospitalised 4 3 6 with COVID-19 in the UK show a signal towards an increased risk of day-28 mortality due to 4 3 7 HIV-positive status. The data for this study were collected during the peak of the UK  for PWH should be placed on early HIV diagnosis, prompt ART initiation, and optimised 4 6 2 screening for and control of comorbidities including obesity and diabetes. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020  used as an indicator of disease severity. All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

LEGENDS TO FIGURES
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020   All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020  All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020   All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 11, 2020. . https://doi.org/10.1101/2020  All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The CO-CIN data were collated by ISARIC4C Investigators. ISARIC4C welcomes 5 3 4 applications for data and material access through our Independent Data and Material Access Committee (https://isaric4c.net). Annette Lake, Claire Petersen, and Scott Mullaney. Lectureship at the University of Liverpool. LT is supported by the Wellcome Trust (grant 5 6 1 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.