Discussion
In this systematic review and meta-analysis, we included 955
laboratory-confirmed COVID-19 patients coinfected with influenza or RSV
and 6907 patients infected with SARS-CoV-2 alone from 12 retrospective
observational studies. We found
that co-infection with influenza was associated with a 2-fold increase
in the risk for ICU admission and for mechanical ventilation among
COVID-19 patients whereas evidence was limited on the role of RSV
co-infection. Co-infection with
influenza did not seem to increase the risk of death in COVID-19
patients.
The
2-fold increase in the risk of ICU admission and receiving mechanical
ventilation in COVID-19 patients coinfected with influenza could have
important implications for the clinical management.
COVID-19 patients who received
influenza positive tests before or upon admission might benefit from
early interventions that could prevent or slow disease deterioration.
This also highlights the importance of influenza vaccination program
that might have been suspended because of the COVID-19 pandemic28.
Earlier systematic reviews that compared co-infections of any viruses
(rather than co-infection of influenza as used in this study) with
mono-infection among COVID-19 patients did not observe any statistical
differences in ICU admission between the co-infection and mono-infection
groups 29–31.
This suggests that the increased risk of ICU admission in the
co-infection group observed in our study was likely to be
influenza-specific. One of the possible explanations is that influenza
co-infection could predispose patients to secondary bacterial
infections, which could lead to more severe clinical outcomes32,33. This explanation is supported by a recently
published study in Israel that observed substantial reductions in
pneumococcal diseases in young children during the COVID-19 pandemic;
the authors showed that the reductions in pneumococcal and
pneumococcus-associated diseases were not predominantly related to
reduced pneumococcal carriage and density but were strongly associated
with the disappearance of specific respiratory viruses such as influenza34. The role of
influenza co-infection in the increased severity of COVID-19 patients
was also supported by influenza probe
studies. In a large Brazilian cohort
study of over 53,000 COVID-19 patients, those who received a recent
influenza vaccine had 7% lower risks of requiring intensive care
treatment (95% CI: 2% to 13%) and 17% lower risks of requiring
invasive respiratory support (95% CI: 12% to 23%)35. Moreover, a
recently published animal model study found that simultaneous or
sequential co-infection by SARS-CoV-2 and the influenza A(H1N1)pdm09
strain caused more severe disease than mono-infection by either virus in
hamsters 36.
However
and interestingly, a systematic
review and meta-analysis by Guan
and colleagues reported that influenza co-infection lowered the
risk for critical outcomes
(composite outcomes including any of shock, being admitted to ICU and
requiring ventilatory support), with the pooled OR of 0.64 (0.43 to
0.97) based on five studies 37. As this was contrary
to the findings of our study, we closely compared the methodology and
the extractions between our study and that by Guan and
colleagues; in addition to the use
of the composite outcomes as mentioned above that differed from our
study, the meta-estimate by Guan and colleagues was largely driven by a
preprint that only had univariate
OR available for extraction.
Moreover,
there seemed to be a discrepancy in the extracted OR in the study by
Guan and colleagues — one of the included studies, by Stowe and
colleagues, 19 reported that COVID-19 patients with
influenza co-infection were around twice as likely to be ventilated (OR
2.15, 1.20 to 3.84) or to be admitted to ICU (OR 2.08, 1.17 to 3.70)
through multivariate analysis whereas the OR was extracted as 0.91 (0.41
to 2.02) in the report by Guan and colleagues.
After
removing the preprint and correcting the extraction above from the
included studies by Guan and colleagues, the updated OR in their study
reversed — 1.76 (1.06 to 2.92), which re-confirmed the robustness of
the meta-estimates in our study
(details are provided in Figure
S4).
Our meta-analysis revealed that co-infection with SARS-CoV-2 and
influenza had no observable effects on the overall mortality in both the
main and the two sensitivity analyses (i.e., excluding studies with
small sample size and excluding low-quality studies). Nonetheless, these
findings did not yet indicate that co-infection of influenza did not
increase the risks of mortality. We could not rule out type II error
(i.e., false negative) due to the limited statistical power; most
studies had less than 50 COVID-19 patients with co-infections and the
median number of deaths is 1 (IQR: 0-4). Moreover, the point estimates
from the main and sensitivity analyses were consistently above 1,
favouring the association between influenza co-infection and mortality.
Similar to influenza co-infection,
no effects on the overall mortality were observed for RSV co-infection
but this was only based on three small studies.
Our review
highlighted the gaps in the
knowledge on the role of RSV co-infection in COVID-19 disease severity.
A retrospective study among six children’s hospitals in the United
States revealed that one in six COVID-19 inpatients under 18 years old
had viral co-infections, with two thirds of viral co-infections being
RSV co-infections; the proportion of viral co-infections was even higher
in infants under one year — one in three infants hospitalized for
COVID-19 had viral co-infections, with almost three quarters of them
having RSV co-infections 38.
The substantial proportion of RSV
co-infection among paediatric
COVID-19 patients calls for further research on the possible synergistic
effects of the RSV and SARS-CoV-2 in this vulnerable age group.
At the time of writing of
this
manuscript, a large-scale
study from the UK (including about
7000 COVID-19 patients) was published, which reported that
that
influenza co-infection was
associated with increased odds of receiving invasive
mechanical ventilation among
COVID-19 patients (OR: 4.1, 2.0 to 8.5) in a multivariable regression
analysis, further confirming the findings of our study39. The authors of that study did not find any
significant associations for co-infection with other viruses including
adenovirus and RSV, confirming our speculation that the association on
viral co-infection and COVID-19 clinical severity is specific to
influenza. Ad-hoc inclusion of the
estimates from that study did not substantially change our
meta-estimates (Table S2).
We acknowledge some limitations of this study. Several factors could
contribute to heterogeneity in the findings of the studies, such as
study population (including age and sex), case definition, laboratory
method and statistical method; we were unable to explicitly account for
these variations. In the present
review, there were only four studies applying multivariate analyses to
adjust for common confounders between co-infections and severity
outcomes. Although restricting to
these high-quality studies did not substantially change any of our
findings, this highlighted the lack of high-quality studies on the role
of co-infections and COVID-19 disease severity. We were also unable to
account for bacterial co-infection in the analysis due to the absence of
data that had reliable laboratory confirmation on viral and bacterial
infections. Compared with other outcomes, mortality as the proxy for
disease severity was more likely to be affected by variations in
clinical treatment, which could confound the estimates if treatment
differed by status of co-infection. All of the included studies were
conducted in 2020 when the activity of influenza and RSV was on the
course to gradual reduction; it
remains unknown whether the resurgence of influenza and RSV could modify
the association reported in this review.
Despite these limitations, this
systematic review and meta-analysis includes a comprehensive summary
with extensive literature search, and critical appraisal and synthesis
of existing evidence on the role of influenza and RSV co-infection in
the clinical severity of COVID-19 patients.
While our study highlights the gaps
in research on this topic particularly for RSV, existing evidence
suggests that influenza co-infection could increase the disease severity
of COVID-19 patients, which could
have important implications for clinical management as well as influenza
vaccination campaign in the context of resurgence of influenza among
other respiratory viruses in the post-COVID-19 era.
Competing interests: YL
reported grants from the World Health Organization and Wellcome Trust,
outside the submitted work. All other authors declared that they have no
competing interests.
Funding: This work receive1234-d support from the Nanjing
Medical University Talents Start-up Grants (NMUR20210008).
Access to data: The study
data are available from the corresponding author upon reasonable
request.
Authorship contribution: BC – led data extraction, analysis
and interpretation and wrote the manuscript;SD
– contributed to data extraction
and quality assessment; XW – contributed to interpretation and
critically reviewed the manuscript; YL
– conceptualised the study,
provided critical oversight to data extraction, analysis and
interpretation and critically reviewed the manuscript. The corresponding
author (YL) had full access to all the data in the study and was
responsible for the decision to submit the manuscript for publication.