COVID-19 Severity and COVID-19–Associated Deaths Among Hospitalized Patients with HIV Infection — Zambia, March–December 2020

The effect of HIV infection on COVID-19 outcomes is unclear. Studies in South Africa (1) and the United Kingdom (2) found an independent association between HIV infection and COVID-19 mortality; however, other studies have not found an association between poor COVID-19 outcomes and either HIV status among hospitalized patients (3–5) or HIV-associated factors such as CD4 count, viral load, or type of antiretroviral therapy (ART) (6). The effect of HIV infection on COVID-19 outcomes remains an urgent question in sub-Saharan Africa, where many countries are experiencing dual HIV and COVID-19 epidemics, and capacity to treat severe COVID-19 is limited. Using data from patients with probable or confirmed COVID-19 admitted to specialized treatment centers during March–December 2020 in Zambia, the Zambian Ministry of Health and CDC assessed the relationship between HIV infection and severe COVID-19 and COVID-19–associated death. Among 443 patients included in the study, 122 (28%) were HIV-positive, and of these, 91 (89%) were receiving ART at the time of hospitalization. HIV status alone was not significantly associated with severe COVID-19 at admission or during hospitalization or with COVID-19–associated death. However, among HIV-positive persons, those with severe HIV disease were more likely to develop severe COVID-19 and were at increased risk for COVID-19–associated death. Ensuring that persons maintain HIV disease control, including maintaining ART continuity and adherence, achieving viral suppression, and addressing and managing underlying medical conditions, could help reduce COVID-19–associated morbidity and mortality in sub-Saharan Africa.

Zambia is a landlocked country in southeastern Africa, with an estimated population of 17.4 million* and a generalized HIV/AIDS epidemic with HIV prevalence among persons aged ≥15 years of 12.1% (7). Beginning in March 2020, patients with a diagnosis of probable or laboratory-confirmed COVID-19 in Zambia were admitted to one of five Zambia Ministry of Health specialized COVID-19 treatment centers located in the capital city of Lusaka (two treatment centers) and in Ndola, Kabwe, and Livingstone. Confirmed cases were those with positive reverse transcription–polymerase chain reaction or rapid antigen test results for SARS-CoV-2, the virus that causes COVID-19. Patients with diagnosed probable cases had radiologic evidence suggestive of COVID-19 with acute respiratory symptoms. Treatment centers have specifically designated isolation and treatment units staffed by clinicians and nurses trained in COVID-19 clinical management. All patients who received medical care for COVID-19 in these centers during March–December 2020 and who provided verbal consent to receive treatment were enrolled in the COVID-19 clinical outcomes study. Patient demographic, clinical, and survival time data were collected during hospitalization until patients died or were discharged. Data were primarily collected electronically in real time by trained staff members using a standardized case record form^†; among patients who had received COVID-19 care before the start of data collection, medical records were reviewed and abstracted into the case record form. Study staff members contacted discharged patients by telephone 28 days after admission to determine their health status.

Severe COVID-19 was defined as having an oxygen saturation <90%, respiratory rate >30 breaths/minute, or a need for oxygen therapy.^§ COVID-19 severity was assessed at admission and during hospitalization. HIV status was self-reported, and patients with unknown status or deemed eligible and consented for testing were tested at admission. Underlying medical conditions were self-reported and included cardiac disease, hypertension, diabetes, other pulmonary disease, active tuberculosis (TB), previous TB, asthma, kidney disease, liver disease, neurologic disorder, asplenia, malignant neoplasm, and current smoking. The number of underlying conditions was summed for each patient. Patients with severe HIV disease were defined as those meeting one or more of the following criteria: 1) severely anemic (hemoglobin <8.0 g/dL); 2) CD4 <200 cells/μL; 3) active TB, including patients taking anti-TB medication; or 4) underweight (body mass index [BMI] <18.5 kg/m²). HIV-positive patients who did not meet any of the conditions were considered to have controlled HIV infection.

The three primary outcomes assessed in the study were 1) severe COVID-19 at admission; 2) severe COVID-19 during hospitalization; and 3) death. Mixed-effects logistic regression models were used to assess associations between exposure variables (age, sex, number of underlying conditions, and HIV status) and the two severe COVID-19 outcomes (at admission and during hospitalization). Mixed-effects Cox proportional hazards regression models were used to examine time to COVID-19–associated death in relation to exposure variables. All models included a random-effects term for treatment center, and other covariates were sex, age, and number of underlying health conditions. Cox models were also adjusted for COVID-19 severity at admission. Similar mixed-effects logistic regression and mixed-effects Cox proportional hazards models were used to assess COVID-19 outcomes among HIV-positive persons stratified by HIV infection control status; covariates in these models included sex, age, and treatment center (random effects term). Data were analyzed using R (version 4.0.2; R Foundation). An alpha level of 0.05 was used to assess statistical significance. The study protocol was approved by the University of Zambia Biomedical Research Ethics Committee, reviewed by CDC, and conducted consistent with applicable federal law and CDC policy.^¶

Among 612 hospitalized patients who were eligible for the study, 443 (72%) had HIV status recorded. Among those patients, 122 (28%) were HIV-positive, and among the 102 HIV-positive persons who provided information on ART status, 91 (89%) were receiving ART (Table 1). Although sex and mean age did not differ by HIV status, among HIV-negative patients, the proportion of those aged ≥60 years was higher than the proportion of those aged <60 years (p = 0.002). HIV-positive patients also were more likely to be anemic (defined as hemoglobin [Hb] <12 g/dL for women and Hb <13 g/dL for men) (p<0.001) or severely anemic (Hb <8 g/dL) (p = 0.004) and to report having two or more underlying medical conditions than were HIV-negative patients (p = 0.017).

Age ≥60 years and having two or more underlying medical conditions were associated with severe COVID-19; however, HIV status alone was not associated with severe COVID-19 at admission or during hospitalization (Table 2). Similarly, male sex, age ≥60 years, and reporting two or more underlying medical conditions were significantly associated with COVID-19–associated death, but HIV status was not. However, among HIV-infected patients, compared with patients with controlled HIV, severe HIV disease was associated with severe COVID-19 at admission (adjusted odds ratio [aOR] = 3.91; 95% confidence interval [CI] = 1.69–9.69) or during hospitalization (aOR = 4.42; 95% CI = 1.83–11.66) and with increased COVID-19–associated death (aOR = 3.27; 1.21–8.79).

Acknowledgments

Study participants, hospital management, and staff members at Levy Mwanawasa Teaching Hospital, University Teaching Hospital, Kabwe General Hospital, Ndola Teaching Hospital, and Livingstone General Hospital.

Corresponding author: Peter A. Minchella, [email protected].

¹Zambian Ministry of Health, Lusaka, Zambia; ²CDC, Lusaka, Zambia.

References

1. Boulle A, Davies MA, Hussey H, et al. Risk factors for COVID-19 death in a population cohort study from the Western Cape Province, South Africa. Clin Infect Dis 2020. Epub August 29, 2020. https://doi.org/10.1093/cid/ciaa1198 PMID:32860699
2. Bhaskaran K, Rentsch CT, MacKenna B, et al. HIV infection and COVID-19 death: a population-based cohort analysis of UK primary care data and linked national death registrations within the OpenSAFELY platform. Lancet HIV 2021;8:e24–32. https://doi.org/10.1016/S2352-3018(20)30305-2 PMID:33316211
3. Sigel K, Swartz T, Golden E, et al. Coronavirus 2019 and people living with human immunodeficiency virus: outcomes for hospitalized patients in New York City. Clin Infect Dis 2020;71:2933–8. https://doi.org/10.1093/cid/ciaa880 PMID:32594164
4. Gervasoni C, Meraviglia P, Riva A, et al. Clinical features and outcomes of patients with human immunodeficiency virus with COVID-19. Clin Infect Dis 2020;71:2276–8. https://doi.org/10.1093/cid/ciaa579 PMID:32407467
5. Parker A, Koegelenberg CFN, Moolla MS, et al. High HIV prevalence in an early cohort of hospital admissions with COVID-19 in Cape Town, South Africa. S Afr Med J 2020;110:982–7. https://doi.org/10.7196/SAMJ.2020.v110i10.15067 PMID:33205724
6. Inciarte A, Gonzalez-Cordon A, Rojas J, et al. Clinical characteristics, risk factors, and incidence of symptomatic coronavirus disease 2019 in a large cohort of adults living with HIV: a single-center, prospective observational study. AIDS 2020;34:1775–80. https://doi.org/10.1097/QAD.0000000000002643 PMID:32773471
7. Joint United Nations Programme on HIV and AIDS. AIDSinfo. Geneva, Switzerland: United Nations, Joint United Nations Programme on HIV and AIDS; 2021. Accessed March 2, 2021. http://aidsinfo.unaids.org/
8. Griffith GJ, Morris TT, Tudball MJ, et al. Collider bias undermines our understanding of COVID-19 disease risk and severity. Nat Commun 2020;11:5749. https://doi.org/10.1038/s41467-020-19478-2 PMID:33184277

Abbreviations: ART = antiretroviral therapy; BMI = body mass index; CRP = C-reactive protein; Hb = hemoglobin; IQR = interquartile range; IU = international units; N/A = not applicable; SD = standard deviation; TB = tuberculosis; VL = viral load; WBC = white blood cell count.
* Reported as number (%) unless indicated otherwise. Denominator is total number by HIV status unless indicated otherwise. HIV status was self-reported and confirmed by HIV test at hospital admission (if the patient specifically consented to an HIV test).
^† P-values <0.05 were considered statistically significant. Comparison of proportions was via chi-square test, comparison of normally distributed variables was via Welch’s two sample t-test, and comparison of nonnormally distributed variables was via Wilcoxon test. P-values for age ≥60 years and two or more comorbidities are for binary variables (i.e., <60 versus ≥60 years and less than two versus two or more comorbidities).
^§ COVID-19 cases were confirmed via SARS-CoV-2 reverse transcription–polymerase chain reaction or SARS-CoV-2 antigen detecting rapid diagnostic test. Probable cases had acute respiratory symptoms with radiological evidence that was suggestive of COVID-19.
^¶ ART status was self-reported.
** Denominator was less than the total number by HIV status.
^†† Anemia was defined as Hb <12 g/dL for women and Hb <13 g/dL for men.
^§§ Severe anemia was defined as Hb <8 g/dL.
^¶¶ Comorbidities composite term is based on the number of self-reported comorbidities. Eligible comorbidities included chronic cardiac disease, hypertension, pulmonary disease, active TB, previous TB, asthma, kidney disease, liver disease, neurologic disorder, diabetes, current smoking, asplenia, and malignant neoplasm. Conditions commonly linked to COVID-19 severity (diabetes and hypertension) and related to HIV (active TB) are listed in the table.
*** Severe COVID-19 was defined as one or more of the following conditions: oxygen saturation <90%, respiratory rate >30 breaths/minute, and need for oxygen therapy.

Abbreviations: aHR = adjusted hazard ratio; aOR = adjusted odds ratio; BMI = body mass index; CI = confidence interval; Ref = referent; TB = tuberculosis.
* Underlying medical conditions included cardiac disease, hypertension, diabetes, other pulmonary disease, active TB, previous TB, asthma, kidney disease, liver disease, neurologic disorder, asplenia, malignant neoplasm, and current smoking.
^† Oxygen saturation <90%, respiratory rate >30 breaths/minute, or need for oxygen therapy.
^§ Mixed-effects logistic regression models were used for severe COVID-19 outcomes. Models were adjusted for sex, age (<60 and ≥60 years), comorbidities (fewer than two and two or more), and treatment center (random effects term). Cox proportional hazards models were used for COVID-19–associated death outcome. Models were adjusted for sex, age (<60 and ≥60 years), underlying medical conditions (fewer than two and two or more), COVID-19 severity at admission (mild and severe) and treatment center (random effects term). Reference categories for exposure variables in all models were female sex, age <60 years, fewer than two underlying medical conditions, and HIV-negative status.
^¶ HIV-positive patients were classified as having severe HIV disease if they met one or more of the following conditions: severely anemic (<8.0 g/dL), CD4 <200 cells/μL, active TB, and underweight (BMI <18.5 kg/m²).