The exact incidence of bacterial and fungal co-infections in COVID-19 is unknown, and while there are anecdotal reports of documented bacterial superinfections, the incidence seems to be much lower than in other viral diseases as severe influenza (15, 16). In our population, the prevalence of bacterial and fungal co-infections was 17.3%. These findings are in line with the results (1–12%) reported in other epidemiological reports from China and the US (1, 3, 17). Among 16,654 Italian patients who died of COVID-19 during the first wave of pandemic, bacterial and fungal superinfections were reported in 11.0% of cases (18). In another Italian study (19), a very high prevalence of BSI was found in patients admitted in ICU with COVID-19 (45 episodes/71 patients, 63.3%), and a 24-day mortality rate of 25.0% was observed. In our study, the episodes of BSI occurred in ICU were much less frequent (13/101, 12.8%).
In our population, most of the cases were caused by Gram positive bacteria (55.2%), and only 26.8% by Gram negative bacilli. In addition, the rate of methicillin resistance was high (51.4%9). These two data can be explained by the fact that 81.4% of patients admitted to our hospital with COVID-19 received empirical antibiotic therapy and that in the 84.1% of the cases it was represented by a third-generation cephalosporin (most active on Gram negative bacilli). The rationale for antibiotic treatment in patients with COVID-19 was based on the experience with bacterial superinfection in influenza, where most studies reported initial co-infection or secondary bacterial pneumonia (11–35% of cases) in hospitalized patients (20). Generally, most patients admitted with COVID-19 pneumonia had fever, hypoxic respiratory failure, and increased of blood value of C Reactive Protein (CRP) for increased inflammatory activity. These signs and alterations worried clinicians who preferred to prescribe empirical/pre-emptive antibiotic therapy, at least during the first 3–5 days of hospitalization. A recent review (21) showed that 88.3% of COVID-infected patients (476/539) were treated with broad-spectrum antibiotics including third-generation cephalosporins, quinolones, carbapenems.
In our population, we observed a higher prevalence of infection caused by MDR microorganisms in patients who had received empirical antibiotic therapy. Fortunately, only 1 case of GNB CR was isolated (1 P. aeruginosa strain). This data is very significant, considering the high prevalence rates of GNB CR infection in Italian hospitals. We think this may be the result of an antimicrobial stewardship program put in place in our hospital starting in 2014 (22, 23). This observation raises an interesting question as to what considerations should be made before prescribing empirical antibiotic therapy in COVID-19 infected patients, assessing its advantages and disadvantages. Bio-humoral markers such as procalcitonin should be used by clinicians to discriminate when to start empirical antibiotic therapy, or when to stop it, in patients hospitalized for severe COVID-19 disease.
Based on our findings, the idea of prescribing all patients with a higher-spectrum antibiotic therapy (e. g. anti-MRSA antibiotic therapy) does not seem justified, as we believe that it was the effect of antibiotic therapy that created the selection of strains resistant to methicillin. The use of anti-MRSA antibiotics could select even more resistant microorganisms (i. e. vancomycin-resistant Enterococcus faecium or Coagulase-negative Staphylococcus resistant to linezolid). On the other hand, in centres where an antimicrobial stewardship program was active, no increase in MDR microorganisms BSI was observed during the first wave of the COVID-19 pandemic (24). In the study of Guisado-Gil AB et al., the frequency of empirical treatment of patients with COVID-19 infections was quite lower (33.7 %), resulting in a lower incidence of invasive MDR microorganisms BSI and a lower incidence of candidemia. Thus, when the probability of a bacterial infection is low antibiotics should not be prescribed and any empiric antibiotic therapy should be discontinued immediately. Antibiotics should be reserved for patients with the most severe respiratory presentations (25, 26). In our study, multivariate analysis of factors related to an increased risk of bacterial infection after hospitalization showed that patients with cancer, with a hospital stay > 10 days, with a SOFA Score > 5, and who were admitted early to the ICU or who required NIV rapidly had an increased risk of bacterial superinfection. In these patients, the addition of antibiotic therapy could be considered if fever persisted or the clinical respiratory picture worsened. Finally, we observed that estimated cumulative risk of developing at least 1 episode of superinfection was of 15% and 27% after 15 and 30 days at risk, respectively.
Equally, attention should be paid when assessing cases of possible invasive fungal infection, particularly cases of invasive pulmonary aspergillosis (IPA) and P. jirovecii pneumonia. Several studies had reported the occurrence of COVID-19 associated with IPA. The largest series was shown by Zhu et al. (27) in Jiangsu Province, China (January 22 - February 2, 2020) in which 23.3% (60/243) COVID-19 patients had co-infection with Aspergillus. Several studies in Europe (28–30) reported different incidence rates of IPA among COVID-19 patients admitted in ICU: 19.6% in Netherland, 20.6% in Belgium and 33.3% in France. In our population, 5 cases (1.2%) of IPA (3 probable and 2 possible) and 3 (0.7%) cases of P. jirovecii pneumonia have been observed. All the fungal infections occurred in patients admitted in ICU and considering only these setting (n = 101), the incidence of IPA and P. jirovecii pneumonia was very low (4.9 and 2.9%, respectively). The data showed that the incidence of pulmonary fungal infections in our population is lower than other studies. The diagnosis of aspergillosis and pneumocystosis is not easy in patients with COVID-19 pneumonia because one of the diagnostic criteria (the radiological one) is of scarce use because it can be confused with the alterations due to COVID-19. Moreover, being ubiquitous and opportunistic fungi, it is not uncommon to find them in the respiratory material of hospitalized and intubated patients. In addition, broncho-alveolar lavage is an invasive procedure highly at risk of infection for clinicians when performed in a patient with COVID-19 and is therefore rarely performed. For these reasons and in light of low sensitivity of serum galactomannan, more sensitive blood tests should be performed, e. g. PCR for Aspergillus, β-D-glucan, and the Aspergillus-specific lateral flow device and lateral flow assay.
Invasive candidiasis was also rare in our population (3 cases of candidemia and 1 case of subcutaneous abscess, 1.0%). Based on our data, and in consideration of the possible side effects of antifungal drugs as voriconazole, we do not believe that the introduction of antifungal prophylaxis in patients with COVID-19 related pneumonia is justified.