Epidemiology, Clinical Characteristics, and Mortality of Hospitalized Patients with Severe COVID-19 in Uganda, 2020–2021

To the Editor:

Little is known about the clinical epidemiology of severe COVID-19 in sub-Saharan Africa (SSA), a high human immunodeficiency virus (HIV)-burden region, where vaccine coverage remains poor and surges of coronavirus disease (COVID-19) have overwhelmed underresourced health systems (1–4). We sought to determine the epidemiology, clinical characteristics, and outcomes of patients with severe COVID-19 admitted to a national COVID-19 referral hospital in Uganda during 2020–2021.

Methods

This study was conducted at Entebbe Regional Referral Hospital (ERRH), a 200-bed public hospital in central Uganda, from March 22, 2020 to November 30, 2021. During this period, ERRH functioned as a national referral hospital for COVID-19; patients with COVID-19 nationwide were referred to the facility for management. There was no functional intensive care unit, invasive mechanical ventilation (IMV), or piped oxygen available at ERRH during the study period. Oxygen therapy was typically provided via nasal cannula and simple or non-rebreathing facemask using concentrators and/or cylinders, the latter filled from a pressure swing adsorption plant built on site. These devices were used in tandem for patients requiring high-level respiratory support. High-flow nasal oxygen or continuous positive airway pressure (CPAP) devices and chest radiography were infrequently available. All management decisions, informed by national COVID-19 treatment guidelines, were made by ERRH clinicians (5, 6).

Patients were included in this cohort study if they were: 1) ⩾5 years of age; 2) admitted to ERRH with COVID-19–related symptoms; and 3) had severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection confirmed by polymerase chain reaction (PCR) testing of a naso-/oropharyngeal swab sample at a Ministry of Health–accredited laboratory. Study participants were identified via review of hospital admission logs and medical charts, from which data were abstracted using a modified case record form developed by the International Severe Acute Respiratory and Emerging Infection Consortium and the World Health Organization (7). Most patients were identified retrospectively; a subset of patients, from whom blood samples were collected, were identified and enrolled prospectively.

Our primary outcome was the frequency of severe COVID-19 during the study period, defined as the proportion of all symptomatic COVID-19 admissions who fulfilled criteria for severe COVID-19. Based on World Health Organization guidelines, we considered patients to have severe COVID-19 if they met one or more of the following criteria: 1) oxygen saturation ⩽ 90%; 2) respiratory rate ⩾ 30 breaths/min; 3) showed signs of respiratory distress (chest indrawing, nasal flaring, grunting respirations); 4) received oxygen therapy (8). Although oxygen saturation was recorded for 99.0% of patients, a substantial proportion (67.6%) did not have their respiratory rate documented. As such, we assumed a missing-at-random mechanism and performed multiple imputation via chained equations (using all relevant clinical variables as predictors), generating five imputed datasets (mice R package) (9). After imputation, the pooled proportion of patients with severe COVID-19 and the corresponding 95% confidence interval (CI) were determined using Wilson scores (MIWilson R package) (10). Given the large proportion of missing respiratory rate data, we conducted three sensitivity analyses: one in which all missing respiratory rate and oxygen saturation values were assumed to be <30 breaths/min and >90%, respectively; a second in which we defined severe COVID-19 without a respiratory rate criterion; and a third in which severe COVID-19 was defined by receipt of oxygen therapy alone. For patients with severe COVID-19, we estimated cumulative incidence of hospital mortality considering discharge or transfer alive as a competing risk (cmprsk R package).

Each prospectively enrolled participant, their surrogate, or a parent/guardian provided written informed consent, and assent was obtained from patients aged 8–17 years. For retrospective chart review, informed consent was waived. Study protocols were approved by ethics committees at Uganda Virus Research Institute, Uganda National Council for Science and Technology, and Columbia University. Administrative clearance was provided by ERRH.

Results

We identified 1,418 patients admitted with PCR-confirmed SARS-CoV-2 infection. After excluding 14 patients with unknown demographics or vital status and 328 asymptomatic individuals admitted for isolation, we analyzed data from 1,076 patients with symptomatic COVID-19 (772 [71.7%] identified retrospectively and 304 [28.3%] prospectively) (Table 1). Most patients were young adult males, and 6% were healthcare workers.

Table 1.

Patient characteristics in first imputed dataset stratified by COVID-19 severity

Patient Characteristic	All Patients (N = 1,076)	Mild–Moderate COVID-19 (n = 765)	Severe COVID-19 (n = 311)	P Value*
Male sex	726/1,076 (67.5)	521/765 (68.1)	205/311 (65.9)	0.533
Age, yr	37 (29–48)	34 (26–41)	50 (40–62)	<0.001
Employed as healthcare worker	66/1,076 (6.1)	42/765 (5.5)	24/311 (7.7)	0.215
Hypertension	194/1,076 (18.0)	100/765 (13.1)	94/311 (30.2)	<0.001
Diabetes mellitus	67/1,076 (6.2)	20/765 (2.6)	47/311 (15.1)	<0.001
HIV infected	59/1,076 (5.5)	44/765 (5.8)	15/311 (4.8)	0.646
On ART before admission	43/59 (72.9)	30/44 (68.2)	13/15 (86.7)	0.361
Malaria infection†	14/1,076 (1.3)	8/765 (1.0)	6/311 (1.9)	0.247
Active or prior tuberculosis‡	6/1,076 (0.6)	4/765 (0.5)	2/311 (0.6)	1.000
Duration of illness before hospitalization§	5 (3–7)	4 (2–7)	6 (3–7)	<0.001
History of fever	371/1,076 (34.5)	213/765 (27.8)	158/311 (50.8)	<0.001
Night sweats	23/1,076 (2.1)	13/765 (1.7)	10/311 (3.2)	0.185
Headache	410/1,076 (38.1)	333/765 (43.5)	77/311 (24.8)	<0.001
Cough	768/1,076 (71.4)	495/765 (64.7)	273/311 (87.8)	<0.001
Shortness of breath	336/1,076 (31.2)	96/765 (12.5)	240/311 (77.2)	<0.001
Rhinorrhea	381/1,076 (35.4)	333/765 (43.5)	48/311 (15.4)	<0.001
Chest indrawing	42/1,076 (0.9)	—	42/311 (13.5)	—
Nasal flaring	20/1,076 (1.9)	—	20/311 (6.4)	—
Grunting respirations	4/1,076 (0.4)	—	4/311 (1.3)	—
Temperature, °C	36.5 (36.4–36.8)	36.5 (36.4–36.7)	36.5 (36.4–36.9)	0.212
Heart rate, beats/min	87 (78–99)	85 (77–96)	94 (84–110)	<0.001
Respiratory rate, breaths/min	19 (18–22)	18 (18–20)	24 (20–28)	<0.001
Systolic blood pressure, mm Hg	127 (117–139)	126 (117–137)	130 (117–142)	0.015
Oxygen saturation, %	97 (95–98)	98 (97–98)	91 (86–95)	<0.001
Altered mental status‖	9/1,076 (0.8)	2/765 (0.3)	7/311 (2.3)	0.003
Received oxygen therapy	280/1,076 (26.0)	—	280/311 (90.0)	—
Oxygen flow rate, L/min¶	8 (5–15)	—	8 (5-15)	—
Received corticosteroids	347/1,076 (32.2)	96/765 (12.5)	251/311 (80.7)	<0.001
Received corticosteroids with oxygen therapy	246/280 (87.9)	—	246/280 (87.9)	—
Received antibacterial agent	819/1,076 (76.1)	520/765 (68.0)	299/311 (96.1)	<0.001
Received antibacterial agent with oxygen therapy	277/280 (98.9)	—	277/280 (98.9)	—
Duration of hospitalization, d**	9 (7–12)	9 (7–12)	9 (6–13)	0.381
Hospital outcome
Death	81/1,076 (7.5)	3/765 (0.4)	78/311 (25.1)	<0.001
Discharge alive	969/1,076 (90.1)	755/765 (98.7)	214/311 (68.8)
Transferred to other health facility alive	26/1,076 (2.4)	7/765 (0.9)	19/311 (6.1)

Definition of abbreviations: ART = antiretroviral therapy; COVID-19 = coronavirus disease; HIV = human immunodeficiency virus.

Data are presented as n/N (%) or median (interquartile range).

^*Wilcoxon rank-sum, Fisher exact, or chi-square test.

^† Rapid diagnostic test or smear microscopy positive.

^‡ Diagnosis as documented by treating clinicians.

^§ Known for 973 patients.

^‖ Anything other than “alert” on AVPU (alert, responsive to voice, responsive to pain, unresponsive) mental status assessment.

^¶ Known for 209 patients.

^**Known for 1,068 patients.

After imputation, the proportion of symptomatic patients who met criteria for severe COVID-19 was 29.1% (95% CI, 26.9–31.5%), with peaks from March to August 2021 (Tables 1 and 2 and Figure 1A). Ninety percent of patients with severe COVID-19 received oxygen therapy, of whom approximately 88% and 99% received corticosteroids or antibacterial agents, respectively (Table 1). Patients with severe COVID-19 were older, more likely to have hypertension or diabetes, and presented later in the course of illness. We observed no associations between HIV, malaria, or tuberculosis and COVID-19 severity, though the latter two coinfections were rare.

Table 2.

Frequency of severe COVID-19 and related mortality during the study period stratified by case definition and approach to missing data

Case Definition Used to Define Severe COVID-19 and Approach to Missing Data	Proportion of Symptomatic Patients with Severe COVID-19 (95% CI)*	Hospital Mortality in Patients with Severe COVID-19 (95% CI)*
Modified WHO case definition† with multiple imputation of missing respiratory rate and oxygen saturation values	29.1% (26.9–31.5%)	25.1% (20.6–30.2%)‡
Modified WHO case definition† assuming all missing respiratory rate values <30 breaths/min and oxygen saturation values >90%	28.4% (25.8–31.2%)	25.1% (20.6–30.3%)
Modified WHO case definition† excluding respiratory rate criterion and assuming all missing oxygen saturation values >90%	28.3% (25.6–31.0%)	25.3% (20.8–30.5%)
Receipt of oxygen therapy alone	26.0% (23.5–28.7%)	27.1% (22.3–32.6%)

Definition of abbreviations: CI = confidence interval; COVID-19 = coronavirus disease; WHO = World Health Organization.

Frequency defined as the proportion of all symptomatic COVID-19 admissions who fulfilled criteria for severe COVID-19.

^*Confidence intervals generated using Wilson scores.

^† One or more of the following criteria: oxygen saturation ⩽ 90%, respiratory rate ⩾ 30 breaths/min, signs of respiratory distress (chest indrawing, nasal flaring, grunting respirations), receipt of oxygen therapy.

^‡ Reflects first imputed dataset; results from other four imputed datasets as follows: 24.4 (20.0–29.4), 25.1 (20.6–30.2), 24.4 (20.0–29.5), 25.3 (20.8–30.4).

Figure 1.

Frequency of severe coronavirus disease (COVID-19), hospital mortality, and oxygen delivery during the study period. (A) Frequency of severe COVID-19 over study period, reflecting data pooled across five imputed datasets and defined as proportion of all symptomatic COVID-19 admissions who fulfilled criteria for severe COVID-19 (N = 1,076); shading indicates Wilson score 95% confidence interval (CI). (B) Cumulative incidence of hospital outcomes for patients with severe COVID-19 in first imputed dataset; shading indicates 95% CI (n = 309; 2 patients with unknown duration of hospitalization excluded). (C) Hospital vital status over study period for patients with severe COVID-19 in first imputed dataset (n = 311). (D) Density plot of peak oxygen flow rates received by patients with severe COVID-19 during study period, stratified by hospital vital status; dashed lines indicate median oxygen flow rate for each group (n = 209).

Hospital mortality for all patients with severe COVID-19 was 25.1% (95% CI, 20.6–30.2%) (Figure 1B). Temporally, mortality was highest (40% [50/125]; 95% CI, 31.8–48.8%) during the nationwide Delta variant–driven surge in May and June 2021 (11, 12) (Figure 1C). Patients who died frequently received high-level oxygen therapy (median flow rate, 20 L/min) (Figure 1D), with mortality of 52.6% (40/76; 95% CI, 41.5–63.5%) among patients who received ⩾15 L/min during admission. Estimates of the proportion of symptomatic patients with severe COVID-19 and related mortality were consistent across multiple assumptions about missing data and when severe COVID-19 was defined by receipt of oxygen therapy alone (Table 2).

Discussion

In a large cohort of patients hospitalized with symptomatic COVID-19 in Uganda, most of whom were young adults, we observed a high frequency of severe illness. Despite provision of locally maximal oxygen therapy, corticosteroids, and empiric antibacterial agents, mortality among the severely ill was substantial, comparable to if not higher than that reported among patients hospitalized with severe COVID-19 in high-income countries (13–16).

Although early reports suggested that COVID-19–related morbidity and mortality in SSA were low, the frequency of severe COVID-19 and related mortality in our cohort were high (17–19). For patients who received ⩾15 L/min of oxygen therapy, outcomes were similar to those reported for adults with severe COVID-19 requiring IMV in high-income countries, with hospital mortality >50% (20). These findings emphasize the urgent need for improved vaccine coverage in Uganda and the African region, where only 20.8% and 17.3% of the population were fully vaccinated, respectively, as of June 10, 2022 (2).

More than 25% of all hospitalized patients received oxygen therapy, many at flow rates that are difficult to sustain without piped oxygen. As has been stressed throughout the pandemic, this observation highlights the profound need to improve oxygen delivery, a core component of acute care, throughout SSA (21, 22). Furthermore, because higher-level respiratory support (e.g., high-flow nasal oxygen, CPAP, IMV) necessitates consistent supplies of oxygen and electricity, strengthening basic health system infrastructure remains essential to raise the standard of intra- and interpandemic acute care in the region.

Our study has limitations. First, our findings are derived from a single center, albeit one that received patients with COVID-19 referred nationwide, with monthly admission rates that reflected national trends (12). Second, there were substantial missing data for respiratory rate, necessitating imputation. However, our findings were consistent across case definitions and assumptions about missing data. Third, although we determined hospital vital status for all analyzed patients, future studies are needed to characterize longer-term outcomes of COVID-19 in SSA.

The frequency of severe COVID-19 and related mortality in Uganda during 2020–2021 were substantial. Improvements in vaccine deployment and acute-care capacity are imperative to avoid preventable COVID-19–related morbidity and mortality in SSA.