4 Discussion
Owing to the important implication of SARS-CoV-2 co-infection for COVID-19 management, we found a large proportion of co-infection with other respiratory pathogens among COVID-19 patients in Qingdao, China. Meanwhile, we determined independent factors associated with co-infection by univariate and multivariate analysis. Besides, negative conversion of SARS-CoV-2 RNA was considered as associated variable of prognosis to evaluate the impact of co-infection on COVID-19 patients. Our findings suggested the distribution of co-infection in COVID-19, and provided the evidence that co-infection of only bacteria, only viruses and mixed of them could variously affect the COVID-19. A reported rate of COVID-19 co-infection with 39 pathogen detection was 94.2% (virus 31.5%, bacteria 91.8%) from Zhu (Zhu et al., 2020), as well as other reported rates of co-infected pathogens from 13.5% to 20.7% (Kim et al., 2020; Wang et al.). In our study, the co-infection rate of COVID-19 patients was 76.4% (virus 23.6%, bacteria 63.6%), which stayed similar level in comparison with other studies. In order to further verify whether the high rate of COVID-19 co-infection related to SARS-CoV-2 infection, we collected pneumonia cases in fever clinics considered as suspected cases of COVID-19, including 178 febrile outpatients with pneumonia who admitted to the local hospitals in Qingdao at the same time. As shown in Figure 3 , the common pathogens in COVID-19 patients and pneumonia cases were almost the same, including SP, HI, MC and IFV-B, IFV-A. However, there was a significant difference for rates of co-infection between COVID-19 patients and pneumonia cases (P<0.05), and the rate of co-infection in COVID-19 patients was four times of the co-infection rate of pneumonia cases (19.1%).
To our best knowledge, this has been the first study focused on independent factors associated with SARS-CoV-2 co-infection. Based on co-infection was no associated with disease severity, we further analyzed in terms of separately bacteria and viruses to determine characteristics of co-infection in COVID-19 patients. Among all co-infection patients, 83.3% cases were detected for bacterial co-infection, which were more than twice for viral co-infection (31.0%). For co-infection of bacteria, the most common bacterial pathogens were SP and HI. Results from multivariate Logistic model revealed that over 70% of neutrophils proportion was an independently factor of co-infection of bacteria, which positively associated with bacterial co-infection (OR: 4.563; 95%CI: 1.116-18.648). Moreover, for co-infection of viruses, the most common viral pathogen is INF-B. After multivariate Logistic regression analysis, fever and chest tightness were independently factors of co-infection of viruses. Fever (OR: 4.506; 95%CI: 1.044-19.441) was positively associated with co-infection of viruses, whereas chest tightness (OR: 0.106; 95%CI: 0.015-0.743) was negatively associated. These findings suggest the need to conduct comprehensive microbiologic surveys and clinical evaluation for other respiratory pathogens in COVID-19 patients, and clinicians should pay more attention on the co-infection for confirmed COVID-19 cases, which have great implications for COVID-19 treatment. Additionally, these independent factors may help clinicians to identify keys for co-infection prevention in COVID-19 patients.
At present, there has been limit study reporting the impact of co-infection on COVID-19 patients, which mostly focused on the descriptive characteristics of co-infection. However, this study has firstly provided evidence that co-infection could impact on COVID-19, which presents the association with negative conversion of SARS-CoV-2 RNA. Results suggested that co-infection was associated with a promoted shedding of SARS-CoV-2 in COVID-19 patients. Compared with COVID-19 patients without co-infection, patients with co-infection could promote the duration of negative conversion of SARS-CoV-2 RNA, and the effect of promotion varies from different types of co-infection pathogens. Results from multivariate Cox regression revealed that among all types of co-infection, The strongest promotion for negative conversion was detected with co-infection of only viruses (HR: 4.039; 95%CI: 1.238-13.177), and the weakest was found for co-infection of only bacteria (HR: 2.909; 95%CI: 1.308-6.471). Interestingly, the promotion in co-infection of mixed bacteria and viruses was between co-infection of only bacteria and only viruses, and its HR was 3.242 with 95%CI ranging from 1.171 to 8.977. However, there is no clear explanation for these findings. One of the potential explanation for this phenomenon may attribute to combination therapy. Although there has been no treatment guideline for co-infection in COVID-19, and the recommendations from different organizations are also inconsistent, combination therapy with non-anti-SARS-CoV-2 agents in co-infected COVID-19 patients has been seriously considered. In China, antibiotic therapy was recommended under different situations for COVID-19 patients in whom co-bacterial infection cannot be ruled out. Empirical antibiotic, such as amoxicillin, azithromycin, or fluoroquinolones, was recommended for mild cases, but broad-spectrum antibiotic covering all possible pathogens was suggested for severe cases (Jin et al., 2020). Based on the limited data of the present work, it remains unclear which antimicrobial agents should be empirically prescribed in patients with COVID-19. In addition, antimicrobial stewardship program should be implemented to prevent the rising rates of antimicrobial resistance could be caused by an increase in inappropriate antibiotic use for viral pneumonia (Huttner et al., 2020). Besides combination therapy, another potential explanation may attribute to much antagonistic effect of bacteria for SARS-CoV-2 than it of other viruses. Our findings suggest that there may be interaction in viral or bacterial replication and amplification in COVID-19 co-infection. As of now, there has been no evidence explain this phenomenon. Wilks et al proposed that defence system of host, as a supraorganism, contained commensal bacteria and immune system to against bacterial and viral pathogens (Wilks et al., 2012). Several researchers supported the view that the microbiota could inhibit viral replication, and affect virally induced pathogenesis (Domínguez-Díaz C et al., 2019; Khan R et al., 2019, Shi Z et al., 2018). Moreover, viruses in multiple infections can interact with each other in different ways, with different results such as antagonism (Mascia et al., 2016). These views may be useful in explaining our findings.
It is notable that there are several limitations of this study. A relatively small number of COVDI-19 cases were evaluated in comparison with other studies. There were 55 discharged patients of COVID-19 in Qingdao during the study period. Although we firstly assessed the association between co-infection and negative conversion of SARS- CoV-2 RNA, further studies regarding the impact of co-infection on COVID-19 prognosis should be warranted. In our study, we try to identify the interaction between SARS-CoV-2 and other respiratory pathogens by the duration of negative conversion. The timeline of viral load of SARS-CoV-2 and other respiratory pathogens during the disease period may be also an effective variable for better understanding the interaction with co-infected pathogens. Due to lack of continuous Ct values of all pathogens, we are unable to analyze in this aspect, but future studies should pay more attention doing this work.