Mark Mascolini

HIV-positive people admitted to New York City hospitals with COVID-19 usually did not differ significantly from people without HIV in clinical status, course, and outcomes, according to results of a small case-control study [1]. But some trends favored the group without HIV, and the authors called for larger studies to get a clearer picture. This report is a medRxiv preprint that has not been peer-reviewed.

Researchers from New York University (NYU) noted earlier reports of worse COVID-19 outcomes among people with preexisting conditions such as diabetes, cardiovascular disease, and cancer. But to date little work has compared outcomes in people with versus without HIV infection. To address that gap, they conducted a case-control study of people admitted to any of four acute-care NYU hospitals in New York City between March 2 and April 23, 2020. All study participants had been discharged from the hospital, moved to hospice, or died at the time of the analysis.

Using greedy nearest neighbor matching [2], the researchers matched 21 people with HIV to 42 without HIV by admission date, age, body mass index, gender, tobacco history, and history of chronic kidney disease, hypertension, asthma, chronic obstructive pulmonary disease, and heart failure. As a result of matching the cases with HIV and controls without HIV did not differ significantly in age (average 60 vs 61.5), sex (90.5% men), race, tobacco use (current 9.5% vs 14.3%), or medical history.

Among 19 HIV-positive people with an available CD4 count, median count stood at 298, and 6 people had a last measured count below 200. Fifteen of 17 people with an available HIV load had a load below 50 copies. Everyone with HIV was taking combination antiretroviral therapy when admitted to the hospital.

Compared with controls, people with HIV had a significantly higher absolute lymphocyte count (average 1.09 vs 0.88, P = 0.043) and higher C-reactive protein (CRP), an inflammation marker (average 154.48 vs 96.1, P = 0.020). But the groups did not differ significantly in initial white blood cell count, hemoglobin, absolute neutrophil count, ferritin, D-dimer, troponin, creatine phosphokinase, procalcitonin, or creatinine. A higher proportion of people with HIV had an abnormal initial chest image showing consolidation, infiltrate, or opacity (90.5% vs 64.3%).

The HIV group had nonsignificant trends toward longer hospital stays and higher rates of intensive care unit (ICU) admission, mechanical ventilation (23.8% vs 11.9%), and discharge to hospice or death (28.6% vs 23.8%). People with HIV had nonsignificantly higher peak lactate dehydrogenase, ferritin, procalcitonin, and D-dimer and significantly higher peak CRP (average 185.13 vs 128.06, P = 0.024). Logistic regression identified no association between admission CRP and death, but highest peak CRP did weakly predict mortality in people with HIV (odds ratio 1.026, 95% confidence interval 1.002 to 1.051) or without HIV (odds ratio 1.018, 95% confidence interval 1.006 to 1.029). Last measured CD4 count did not predict mortality in people with HIV.

One person with HIV and 1 without HIV had both pulmonary embolism and ST-segment elevation myocardial infarction. No one had a stroke. Three people with HIV and 1 without HIV had sputum cultures indicating bacterial pneumonia. All 4 people died in the hospital despite antibiotic therapy.

The NYU team believes their findings “suggest that HIV status did not significantly impact clinical outcomes in patients with SARS-CoV-2 infection.” But they “did detect trends suggesting that outcomes may be worse in HIV-positive patients, as a greater percentage of HIV-positive patients required ICU-level care, mechanical ventilation, or expired or were discharged to hospice.” The researchers call for a larger study to see if these unfavorable trends apply to everyone with HIV and COVID-19.

References
1. Karmen-Tuohy S, Carlucci PM, Zacharioudakis IM, Zervou FN, Rebick G, Klein E, Reich J, Jones S, Rahimian J. Outcomes among HIV-positive patients hospitalized with COVID-19. medRxiv. https://www.medrxiv.org/content/10.1101/2020.05.07.20094797v1 (This preprint report has not been peer-reviewed.)
2. Ho D, Imai K, King G, Stuart EA. MatchIt: Nonparametric preprocessing for parametric causal inference. Journal of Statistical Software. 2011;42:28.