Literature Search and Data Extraction
The databases PubMed and Scopus were reviewed from December 1, 2019 to
July 3, 2020, to identify all relevant COVID-19 primary publications.
The keywords and Medical Subject Headings (MeSH) terms selected included”Novel coronavirus 2019”, “COVID-19” and “SARS-CoV-2”, and
the target population was specified with the terms “pediatric”,
“children”, “infant” , “neonate”, “adolescent”. The last search
was performed on July 3rd, 2020 and the search was not limited by
language (translation performed with Google Translate) or geographic
region.
Eligibility criteria and Study Selection
Inclusion criteria for screening
Study selection methodology entailed initially screening articles using
title and abstract and subsequent full-text screening. All available
peer-reviewed original articles (case reports, case series, cohort
studies, cross-sectional studies etc.) pertaining to pediatric COVID-19
published in the literature in the aforementioned time frame were
included in this study. Selected articles must have subjects with
SARS-CoV-2 infection confirmed via real-time reverse transcriptase
polymerase chain reaction (RT-PCR) using upper respiratory swabs. The
pediatric population comprised of ages 0-18 years (including neonates).
The selected articles included variables on demography, risk factors,
clinical manifestations, laboratory and radiological findings,
treatment, and outcomes.
Articles were excluded due to the inability to extract pediatric data
from adult data separately. The publication types excluded were review
articles and studies such as letters, correspondences or comments that
had no extractable primary pediatric data. Articles with irrelevant
clinical study focus (e.g. epidemiology, modeling, animal data,
post-mortem data) and non-clinical study focus (e.g. genetics,
diagnostic techniques or virology) were excluded as well.
Inclusion criteria for Meta-analysis data
Studies that qualified for initial screening for data extraction, as
mentioned above, were further filtered with stringent criteria. The
inclusion criteria were optimized for selection of articles with
sufficient data and sample size for data synthesis.
Data Extraction and Quality Assessment
Four investigators worked on title/abstract, full-text screening and
data extraction in pairs (AA and ME, SG and NT) on a shared data
extraction form. Any disagreements were resolved by consulting one of
the investigators from the other pair. The investigators extracted data
on demographics, co-morbidities, signs and symptoms, laboratory
investigations, radiological investigations, treatment and outcomes of
COVID-19 in pediatric patients. The Joanna-Briggs Institute (JBI)
checklists were utilized for the critical appraisal of case series and
cross-sectional studies10. The investigators assigned
2 points for ‘Yes’, 1 point for ‘Unclear’ and 0 points for
‘No/Inapplicable’. The average score of the 2 investigators generated
the final JBI score. Each checklist had a different cumulative score
which was scaled out of 10 (Table 1). A score >7 reflected
a high-quality study, 5-7 moderate-quality and <5 a
low-quality study.
Statistical Analysis
Percentages were calculated to describe the distribution of the
categorical dichotomous variables. For continuous data, the pooled
prevalence with mean and 95% confidence intervals (CI) were calculated.
For studies reporting the mean with 95% CI or the range of the data,
the formula, (upper limit-lower limit)/4, was used to extract the
standard deviation (SD).
The meta-analysis was conducted on Comprehensive Meta-Analysis version
3.3.070 software. The random-effect model was implemented to estimate
the pooled prevalence and 95% CI. Pooled percentage, proportion and
corresponding 95% CI were calculated to summarize the weighted effect
size for all binary variables. The measure of heterogeneity reported
included the Cochran’s Q statistics, I2 index with the
level of heterogeneity defined as low < 25, moderate
> 50, and high > 75, and the tau square
(T2) test. Publication bias was assessed with a funnel
plot and Egger’s test.
Results