Assessment of Clinical Outcomes Among Children and Adolescents Hospitalized With COVID-19 in 6 Sub-Saharan African Countries

Key Points Question What are the clinical outcomes and associated factors among children and adolescents hospitalized with COVID-19 in sub-Saharan Africa? Findings In this cohort study of 469 children and adolescents hospitalized with COVID-19 in 6 sub-Saharan African countries, morbidity and mortality were substantially higher than reported among those in non-African settings and were independently associated with age younger than 1 year and select noncommunicable disease comorbidities. Meaning This study’s findings may have implications for clinical practice and health policy regarding pediatric COVID-19 in African countries; given their high risk of adverse outcomes, COVID-19 vaccination and therapeutic interventions are needed for African children and adolescents.

A fter 2 years of the COVID-19 pandemic, several studies [1][2][3][4][5][6][7][8][9] have reported that disease severity is substantially lower among children compared with adults.Of the more than 315 million cases and 5.5 million deaths reported to be associated with SARS-CoV-2 as of January 13, 2022, more than 29 million cases and 22 000 deaths are estimated among children and adolescents aged 0 to 19 years. 10][3][4][5][6][7][8][9] The African continent has a young population; children younger than 18 years constitute almost 50% of people. 117][18] Sub-Saharan Africa has a high prevalence of both communicable (eg, HIV infection and tuberculosis) and noncommunicable (eg, asthma, cancer, diabetes, hypertension, and sickle cell anemia) diseases that also occur among children. 19,20Combined with the high prevalence of comorbidities, limited availability of intensive care may have substantial consequences for COVID-19 outcomes in sub-Saharan Africa. 20,21In the multicenter African COVID-19 Critical Care Outcomes Study, 22 almost 50% of adults with COVID-19 died within 30 days of intensive care unit (ICU) admission, with up to 23 excess deaths per 100 patients compared with the global average.Limited critical care resources, organ dysfunction at admission, and select comorbidities accounted for this excess mortality.Both children and adults experience inadequate availability of and access to SARS-CoV-2 testing and high-quality intensive care in constrained sub-Saharan African settings. 15,22,23Access to hospital care is limited and varies within and across countries and regions. 24,25The burden of SARS-CoV-2 infection, including severe disease requiring hospitalization, is underestimated in sub-Saharan Africa [26][27][28] and is potentially more underestimated among children, who are less likely to be evaluated for infection. 15To address this issue, the present study assessed clinical manifestations, outcomes, and factors associated with outcomes among children and adolescents hospitalized with COVID-19 in 6 countries in sub-Saharan Africa.

Methods
This cohort study was a multicountry retrospective record review that pooled data from hospitalized children and adolescents aged 0 to 19 years with SARS-CoV-2 infection confirmed through reverse transcriptase polymerase chain reaction testing.Study review and approval, including waivers of informed consent and permission to use deidentified information from existing data sets or medical records, were obtained from institutional and/or national research ethics committees and/or regulatory bodies in participating countries (eTable 2 in Supplement 1).This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. 29

Settings and Participants
The study included all children and adolescents with confirmed SARS-CoV-2 infection who were admitted to 25 health care facilities in the Democratic Republic of the Congo (7 facilities), Ghana (2 facilities), Kenya (1 facility), Nigeria (2 facilities), South Africa (10 facilities), and Uganda (3 facilities) between March 1 and December 31, 2020.These countries were selected based on regional representation (eastern, western, central, and southern Africa) to participate in the study.For between-country comparisons of outcomes, western and central African regions were combined to maximize available sample size and statistical power.Data on race and ethnicity were not collected because the racial profile across the 6 countries was more than 90% Black or African descent, and the ethnic diversity across the 6 countries was too broad (almost 750 ethnic groups) for meaningful categorization or analysis.Detailed information about participating health care facilities (including names, locations, urban vs rural settings, and public vs private status) is available in eTable 1 and eFigure 1 in Supplement 1.

Variables
Using World Health Organization (WHO) pediatric COVID-19 case report forms, 30 demographic and clinical data were extracted from national or institutional COVID-19 data sets and/or hospital records.Data collected included age, sex, preexisting comorbidities, WHO-defined COVID-19 severity stage at admission, 31 and diagnosis of multisystem inflammatory syndrome in children (MIS-C) temporally associated with COVID-19. 30,32To accommodate partial or complete lack of laboratory and imaging data (eg, inflammatory markers and echocardiographic results) required for MIS-C diagnosis in our study settings, cases were characterized as suspected MIS-C when at least 2 required multisystem abnormalities that were clinically observable or measurable were documented in the medical records and/or databases from which study data were extracted.This requirement was in addition to fulfilling WHO criteria for the diagnosis of MIS-C that pertained to ruling out "other obvious microbial cause[s] of inflammation" 3 2 ( p 1 ) plus confirmation of COVID-19 through a positive result on reverse transcriptase polymerase chain reaction testing.

Statistical Analysis
Baseline demographic and clinical characteristics were summarized using frequencies and proportions; medians and IQRs were applied to categorical and continuous variables.For missing data on preexisting comorbidities, we performed multiple imputation using chained equations to generate 20 data sets.Most comorbidities had missingness less than 10%, with the exception of diabetes (23%), chronic lung disease (26%), cerebral palsy (28%), and cardiac disease (35%).
Multivariable proportional odds logistic regression analysis was used to identify factors associated with outcome severity among those with SARS-CoV-2 infection by including only factors that were considered clinically relevant and had a significance level of P < .15 in bivariable analyses.In our analyses, the proportional odds logistic regression model compared lower severity levels with higher severity levels (eg, category 1 vs categories 2-5, categories 1 and 2 vs categories 3-5, categories 1-3 vs categories 4 and 5, or categories 1-4 vs category 5).The proportional odds assumption was evaluated using χ 2 and parallel line tests.Using robust SEs, the bivariable and multivariable proportional odds logistic models were fitted to account for potential within-cluster correlation of outcomes owing to shared processes and quality of care.Adjusted odds ratios (aORs) and associated 95% CIs were used to characterize the association between factors and disease severity.
We examined factors associated with the probability of hospital discharge over time using a competing-risk analysis of the Fine and Gray proportional subdistribution hazards model 33 accounting for death.Factors with significance levels of P < .15 in bivariate models were included in a multivariable proportional subdistribution hazards model to estimate adjusted subdistribution hazard ratios (asHRs) and associated 95% CIs.Overall survival was estimated using the Kaplan-Meier method, and the log-rank test was applied to compare survival differences by sex, region, WHO COVID-19 severity stage, and number of comorbidities.
Two-sided P < .05 was considered statistically significant.All regression models were applied to the 20 imputed data sets, and estimates were combined according to Rubin rules. 34,35All analyses were performed using Stata software, version 16.1 (StataCorp LLC).

Secondary Outcome: Length of Hospital Stay
The median length of hospital stay was 9 days (IQR, 5-16 days) among patients who recovered and 8 days (IQR, 3-19 days) among those who died.Among 10 patients (1 with missing data) who remained hospitalized at the time of data collection, the median length of hospital stay was 18 days (IQR, 7-24 days).

Discussion
This multicountry cohort study of pediatric COVID-19 in sub-Saharan Africa revealed relatively high morbidity and mortality, with greater likelihood of more severe outcomes among children younger than 1 year and those with hypertension, chronic lung disease, or a hematologic disorder.Furthermore, in a competing-risk analysis of time to discharge, age younger than 1 year, the presence of 1 comorbidity, and the presence of 2 or more comorbidities were independently associated with reduced rates of hospital discharge.Overall, 34.6% of hospitalized children and adolescents were admitted to the ICU or required oxygen supplementation, and 21.2% of those admitted to the ICU required invasive ventilation.The region with the highest proportion of children and adolescents requiring ICU admission and/or oxygen supplementation (52.4%) was southern Africa, where there was better availability of high-quality critical care than in other sub-Saharan African countries. 25The proportion of children and adolescents requiring ICU admission or oxygen supplementation in this study was similar to or higher than the proportions reported in studies of non-African countries but was likely underestimated because of the limited availability of pediatric ICUs in much of sub-Saharan Africa.
previous Nachega et al 39 40 This finding was consistent with our finding of higher frequency of severe outcomes, including death, among infants and with the results of a study conducted by Oliveira et al 41 in Brazil.In a global systematic review of severe pediatric COVID-19 illness, Kitano et al 15 also reported that infants had the highest mortality, and the overall case fatality rate was significantly higher in lowand middle-income countries (0.24%) than in high-income countries (0.01%).In our study, mortality was high after ICU admission (31.9%) and substantially greater than the 0% to 0.5% mortality observed in pediatric studies conducted in highresource settings 1,[4][5][6]15 but closer to the mortality (approximately 50%) reported in the African COVID-19 Critical Care Outcomes Study involving adults. 22 e also found that hypertension, chronic lung diseases, and hematologic disorders were independently associated with severe clinical outcomes, including death.Preexisting comorbidities have been associated with worse COVID-19 prognosis in children and adults in other studies.[41][42][43][44][45] In 1 study, 45  In this study, HIV infection was not associated with worse outcome severity, possibly because of low numbers of children living with HIV. Th finding necessitates further research.Of note, published data on the association of HIV infection with COVID-19 outcomes among adults has been inconsistent.[46][47][48] However, a recent data review by the WHO found that HIV infection in adults was a risk factor associated with severe and critical illness at hospital admission and inhospital mortality after adjusting for age, sex, and underlying conditions.49 Our findings have several implications for clinical practice or health policy.The high morbidity and mortality among hospitalized African children and adolescents with comorbidities suggest that targeting these populations for prompt COVID-19 vaccination may be warranted when vaccines become available.Therapeutic interventions should be specifically evaluated among children and adolescents with severe COVID-19 illness and made available as appropriate.In addition, limitations in the quality and scope of pediatric general and critical care services in Africa need to be addressed to improve outcomes among children and adolescents with severe COVID-19 illness and other serious health conditions.

Limitations
This study has several limitations.Our findings of higher COVID-19-associated in-hospital mortality among children and adolescents in sub-Saharan Africa compared with those in non-African settings needs to be interpreted with consideration of important factors.First, we studied only hospitalized children and adolescents, whereas most published studies from China, Europe, and the US [1][2][3][4][5][6][7][8][9]15 included hospitalized, nonhospitalized, and asymptomatic patients. Ou data are not generalizable to outpatient populations.In addition, because of limited hospital resources in sub-Saharan Africa, there may be higher thresholds for hospitalization compared with those in more resource-rich settings, potentially producing a cohort of inpatients who had more severe illness.
Second, a high prevalence of concurrent endemic infections and noncommunicable diseases, malnutrition, and associated dietary deficiencies may have had implications for COVID-19 outcomes in sub-Saharan Africa.Third, MIS-C did not appear to be a major factor associated with mortality in our study; of 26 evaluable deaths, only 4 (15.4%) had confirmed or suspected MIS-C.However, MIS-C cases were likely underestimated owing to the limited availability of tests (eg, tests for inflammatory biomarkers).
Fourth, the limited availability of essential equipment and the narrower scope of pediatric intensive care (compared with adult care) in sub-Saharan Africa likely had implications for the high mortality observed in the present study cohort. 21,25The finding of greater use of intensive care, oxygen supplementation, and mechanical ventilation in the southern African region vs other regions as well as the lower risk of in-hospital death in this region likely reflected varied availability of resources across African regions.Furthermore, these results highlight the opportunity for improved outcomes afforded by greater availability of high-quality pediatric intensive care.
Fifth, the retrospective study design relied on record extraction of routinely collected and available data; however, fewer than 6% of extracted outcome data were missing, and we used multiple imputation techniques for missing data on comorbidities to minimize biased OR estimates.Sixth, limited availability of laboratory tests and diagnostic procedures may have produced underdiagnosis of COVID-19 and some associated features (eg, MIS-C) and precluded reporting and further analysis of immunological status among children living with HIV infection.
Seventh, our lack of a SARS-CoV-2-negative comparator group and the general limited access to diagnostic testing prevents us from drawing conclusions about the relative prevalence and severity of COVID-19 vs other pediatric diseases in sub-Saharan Africa.In addition, statistical modeling of outcomes for each region was limited by small numbers within regions.However, the power of our study comes from pooling data across health facilities and regions as well as adjustment for any potential facility-level differences that may have had consequences for outcomes.

Conclusions
In this cohort study of 6 countries in sub-Saharan Africa, morbidity and mortality rates among hospitalized children and adolescents with COVID-19 were substantially higher than those reported in non-African settings and were associated with age younger than 1 year and select noncommunicable disease comorbidities.These findings provide new data that may be used to inform pediatric COVID-19 health policy in Africa.With hundreds of millions of African children and adolescents at risk of adverse outcomes, COVID-19 vaccination and therapeutic interventions are much needed for this population.

Figure 2 .
Figure 2. Clinical Outcomes of Children and Adolescents With COVID-19 by Region
[36][37][38]Of note, our study included 62 hospitalized patients from what was previously the larg- est (N = 159) cohort of African children with COVID-19. 17In that study, 11 of 51 hospitalized children (21.6%) required ICU admission; of those, 4 children required mechanical ventilation, but none died.Overall, 8.3% of inpatients in the present study died.In comparison, among 766 patients with COVID-19 from a

Table 2 .
among 43 465 US children diagnosed with COVID-19 from March 2020 to January 2021, 28.7% had underlying medical conditions; the Univariable and Multivariable Ordinal Logistic Regression Model Using 5 Levels of COVID-19 Disease Severity as Primary Outcome a important risk factors associated with hospitalization or severe COVID-19 were type 1 diabetes, obesity, cardiac or circulatory congenital anomalies, hypertension, neuropsychiatric disorders, and complex chronic disease.In our study as well as the Oliveira et al 41 study from Brazil, an increase in COVID-19-associated mortality occurred as the number of preexisting comorbidities increased.