85 Viral co-infections in SARS-CoV-2 4 absolute numbers are small making comparison difficult2,5–8,10,23. In our population 1.3% of SARS-CoV-2 positive patients also tested positive for influenza virus, a bit higher than some pooled estimates from meta-analysis (0.3-0.8)23–25. In our study, 1.2% had an RSV co-infection, lower than some previously reported numbers1,8,26. Finally, 0.8% of our population had an adenovirus co-infection. Previous studies reported both similar26 and lower1,8 numbers of adenovirus co-infections. Of course, the prevalence of co-infections is mostly determined by the prevalence of these infections in the community. Most earlier research (supplementary table 1 and 2) consisted of small studies and only a few reported outcome data or compared outcomes between people with/ without co-infection. Ding and colleagues27 identified 5 patients co-infected with influenza and SARS-CoV-2, out of 115 SARS-CoV-2 positive hospitalised patients. These 5 patients did not need ICU admission or IMV and all were discharged alive. Chekuri and colleagues28 did compare SARS-CoV-2 mono-infected patients to coinfected patients. In this co-infected group, other human coronaviruses were most common, followed by rhinovirus/enterovirus. No influenza co-infections were found. Compared to the SARS-CoV-2 only group, the co-infected group was less likely to require oxygen treatment and was not associated with worse outcomes in an adjusted analysis. Drake and colleagues29 found 138 influenza co-infected patients, including children, in which a prolonged duration of hospital admission was found, although this was not corrected for likelihood of being tested. Finally, Alosaimi and colleagues2 identified 30 co-infected patients out of 48 hospitalised (14 ICU) SARSCoV-2 positive patients and found that influenza co-infection was associated with mortality. Our study had several strengths. Firstly, this is the largest study of people with COVID-19 undergoing additional testing for endemic respiratory viruses, reporting 583 confirmed co-infections and 6382 confirmed SARS-CoV-2 mono-infections. Secondly, we recruited patients over an 18-month period. Finally, we report outcome data for the majority of patients (>98% available for IMV and >99% for mortality) There are some weaknesses in our study. Firstly, there is a risk of selection bias because the subjects that were tested differed from untested patients (Supplementary table 3), particularly in severity of illness: being more unwell increased the probability of testing for co-infections. Patients were tested or not tested for a variety of reasons, like illness severity, but also laboratory capacity. In order to correct for these differences and extrapolate our results from the tested to the complete hospital population, an inverse probability weighting analysis was performed, which confirmed the results from the unweighted multivariable logistic regression analysis. Influenza coinfection remained associated with receipt of IMV in the re-weighted population, with an OR that was larger than in the unweighted analysis (4.32 vs 1.80, respectively) but with wider confidence intervals. As in the unweighted analysis, neither RSV nor adenovirus co-infection had a significant association with IMV. Furthermore, adenovirus and RSV co-infection do not show the same effect on IMV as influenza
RkJQdWJsaXNoZXIy MTk4NDMw