key: cord-0749510-zwzqv44b
authors: Baskaran, Vadsala; Lawrence, Hannah; Lansbury, Louise E.; Webb, Karmel; Safavi, Shahideh; Zainuddin, Nurul I.; Huq, Tausif; Eggleston, Charlotte; Ellis, Jayne; Thakker, Clare; Charles, Bethan; Boyd, Sara; Williams, Tom; Phillips, Claire; Redmore, Ethan; Platt, Sarah; Hamilton, Eve; Barr, Andrew; Venyo, Lucy; Wilson, Peter; Bewick, Tom; Daniel, Priya; Dark, Paul; Jeans, Adam R.; McCanny, Jamie; Edgeworth, Jonathan D.; Llewelyn, Martin J.; Schmid, Matthias L.; McKeever, Tricia M.; Beed, Martin; Lim, Wei Shen
title: Co-infection in critically ill patients with COVID-19: an observational cohort study from England
date: 2021-04-16
journal: J Med Microbiol
DOI: 10.1099/jmm.0.001350
sha: fa2f87097b1948dae4cccfca27896694cd81affd
doc_id: 749510
cord_uid: zwzqv44b

INTRODUCTION: During previous viral pandemics, reported co-infection rates and implicated pathogens have varied. In the 1918 influenza pandemic, a large proportion of severe illness and death was complicated by bacterial co-infection, predominantly Streptococcus pneumoniae and Staphylococcus aureus . GAP STATEMENT: A better understanding of the incidence of co-infection in patients with COVID-19 infection and the pathogens involved is necessary for effective antimicrobial stewardship. AIM: To describe the incidence and nature of co-infection in critically ill adults with COVID-19 infection in England. METHODOLOGY: A retrospective cohort study of adults with COVID-19 admitted to seven intensive care units (ICUs) in England up to 18 May 2020, was performed. Patients with completed ICU stays were included. The proportion and type of organisms were determined at <48 and >48 h following hospital admission, corresponding to community and hospital-acquired co-infections. RESULTS: Of 254 patients studied (median age 59 years (IQR 49–69); 64.6 % male), 139 clinically significant organisms were identified from 83 (32.7 %) patients. Bacterial co-infections/ co-colonisation were identified within 48 h of admission in 14 (5.5 %) patients; the commonest pathogens were Staphylococcus aureus (four patients) and Streptococcus pneumoniae (two patients). The proportion of pathogens detected increased with duration of ICU stay, consisting largely of Gram-negative bacteria, particularly Klebsiella pneumoniae and Escherichia coli . The co-infection/ co-colonisation rate >48 h after admission was 27/1000 person-days (95 % CI 21.3–34.1). Patients with co-infections/ co-colonisation were more likely to die in ICU (crude OR 1.78,95 % CI 1.03–3.08, P=0.04) compared to those without co-infections/ co-colonisation. CONCLUSION: We found limited evidence for community-acquired bacterial co-infection in hospitalised adults with COVID-19, but a high rate of Gram-negative infection acquired during ICU stay.

During previous viral pandemics, reported co-infection rates and implicated pathogens have varied. In the 1918 influenza pandemic an estimated 95 % of severe illness and death was complicated by bacterial co-infection, predominantly Streptococcus pneumoniae and Staphylococcus aureus [1] .

As of 3 September 2020, over 25 million cases and 850, 000 deaths due to COVID-19 infection have been reported worldwide [2] . The symptoms associated with COVID-19 infection are relatively non-specific. Fever and lower respiratory tract symptoms, such as a cough or breathlessness, are common in patients who require hospital care and radiological changes consistent with pneumonia are evident in up to 97 % of these patients [3] . Confirmation of acute COVID-19 infection is reliant on a positive SARS-CoV-2 polymerase chain reaction (PCR) test result. The immune response to SARS-CoV2 infection includes a rise in IL-6 and C-reactive protein (CRP), with higher levels associated with more severe disease [4, 5] .

The contribution of secondary or co-pathogens to COVID-19 infection is not well understood. The lack of an effective antiviral agent against SARS-CoV2 combined with challenges in differentiating secondary bacterial co-infection from severe COVID-19 infection alone, has fostered the widespread use of empirical antibiotics in the immediate management of OPEN ACCESS patients hospitalised with COVID-19 infection. Over the spring wave of the pandemic, 83.1 % of hospitalised patients received empirical antibiotic treatment [6] .

The utility of specific biomarkers such as procalcitonin to guide antibiotic therapy in severe respiratory tract infection, and specifically COVID-19 infection, is as yet uncertain [7, 8] .

In the meantime, a better understanding of the incidence of co-infection in patients with COVID-19 infection and the pathogens involved is necessary for effective antimicrobial stewardship. The primary objective of this study was to determine the rate of co-infection in critically ill adults with COVID-19 infection in England. Secondary aims were to describe the organisms, the characteristics of patients with co-infection and the antibiotic susceptibilities of identified bacteria.

A retrospective observational multicentre study of co-infection in adults with confirmed COVID-19 requiring intensive care unit (ICU) admission was performed. Seven acute hospitals from across England participated in the study including large (>1,000 beds) tertiary hospitals and medium (500-1,000 beds) district hospitals: Nottingham University Hospitals NHS Trust, Newcastle Upon Tyne Hospitals NHS Foundation Trust, Brighton and Sussex University Hospitals NHS Trust, Guy's and St Thomas' NHS Foundation Trust, Salford Royal NHS Foundation Trust, University Hospitals of Derby and Burton NHS Foundation Trust and University College London Hospitals NHS Foundation Trust.

Case inclusion criteria were adults aged >16 years with completed ICU admissions (discharged from or died whilst in ICU) for COVID-19 pneumonia (i.e. requiring Level 2 or Level 3 care according to the classification by the Intensive Care Society, UK) from disease emergence to 18 May 2020. SARS-CoV-2 was confirmed using reverse transcriptase-polymerase chain reaction (RT-PCR) from a respiratory specimen. Participating sites were asked to enter data for either: (1) all identified patients, or (2) a random selection of at least ten patients from across their eligible cohort. Where more than one critical care area existed at a participating site, a random selection from across areas was requested to avoid selection bias. Exclusion criteria were defined as: COVID-19 infection diagnosed >48 h after hospital admission or a hospital admission in the last 14 days (hospital-acquired COVID-19) and patients transferred into ICU from a different hospital. Only the first admission to ICU was included.

Personal information was removed at the point of participating site data entry onto a secure online database platform (REDCap Cloud). Data were gathered from electronic medical records. Fields collected were: demographics (age, gender, ethnicity, presence or absence of co-morbidity as defined in the Intensive Care National Audit and Research Centre (ICNARC) report on COVID-19 in critical care (Online Resource 1, available in the online version of this article) and type 2 diabetes mellitus); hospital admission details (date, days of symptom onset prior to admission and radiology findings); ICU details (date of admission, mechanical ventilation during the first 24 h, advanced respiratory support (Online Resource 1), acute physiology and chronic health evaluation (APACHE II) score and outcomes); antibiotics received and all microbiology test results to the end of the ICU admission (including any identified antimicrobial resistance).

Diagnostic microbiology tests were performed as per standard testing protocols within NHS laboratories at individual participating sites. Microbiology results included in the analysis were: standard culture (blood, sputum, trachealaspirate, bronchoalveolar lavage (BAL), urine) and validated culture-independent tests such as respiratory viral PCR (see Online Resource 2) and urinary antigens. Co-infection was defined as present if a likely pathogen was identified in a clinical sample taken for diagnostic purposes. Culture results were excluded if they were considered to represent contamination or colonisation. Specifically, this applied to the following situations: blood cultures yielding common skin contaminants in a single sample (coagulase-negative staphylococci, Micrococcus spp., viridans group streptococci, Propionibacterium spp., Corynebacterium spp., Bacillus spp.) without a concurrent positive culture from an indwelling line tip [9] [10] [11] , Candida spp. cultured from respiratory and urinary catheter samples [12, 13] , respiratory samples yielding Gram-positive organisms typically present in the oropharyngeal flora [14] , growth of Enterococcus spp. in a single catheter urinary specimen [15] . Despite this effort, culture results from some non-sterile respiratory samples may represent colonisation, hence the term co-infection/ co-colonisation is used for respiratory samples. Radiology findings were defined based on the COVID-19 British Society of Thoracic Imaging reporting template [16] . Where both chest CT and CXR findings were available, chest CT findings were prioritised.

Demographics, clinical and disease characteristics were described using appropriate descriptive statistics for: (i) those with co-infection, and (ii) those without co-infection. Characteristics of patients in the study were also compared with the patients in the Intensive Care National Audit and Research Centre (ICNARC) report on COVID-19 in critical care, 22 May 2020. The proportion of co-infection (%) was determined at three time points: on admission, within 48 h, and during ICU admission (from day of ICU admission to ICU discharge or death in ICU). The co-infection rate was calculated per 1000 person-days based on the first co-infection detected in hospital per patient (person-time was determined from date of hospital admission to date of first co-infection, date of discharge from ICU or date of death in ICU, whichever came first for each patient). Univariate logistic regression analyses were conducted to determine the association between selected variables (age, gender, study site, ethnicity and co-morbidities) and the odds of (a) developing co-infection during admission, and (b) co-infection and 

Of 579 eligible patients during the study period, 254 patients with completed ICU episodes were studied (Fig. 1) . potential pathogens from day 8 onwards (Fig. 2) . The co-infection/co-colonisation rate >48 hours after admission was 27.0 per 1000 person-days (95 % CI 21.3-34.1). All were bacterial pathogens (n=122) except for two fungal organisms. The commonest potential co-pathogens identified were Gram-negative bacteria, including Klebsiella spp. (23 patients) and Escherichia coli (20 patients) .

No viral co-pathogens were detected. Of the two fungal co-pathogens, one was Aspergillus fumigatus from a tracheal aspirate culture obtained on day 5 in a 54 year old male. The other was Candida parapsilosis from a blood culture taken at day 7 in a 55 year old lady. Neither patient had any pre-existing co-morbidities.

On univariate analyses, patients aged 50-64 years were more likely to have a co-infection/ co-colonisation than those aged 18-49 years. No other significant association was found ( [19, 20] . In COVID-19, systematic reviews based on studies predominantly from China reported low estimates (<7 %) of bacterial co-infection [21] [22] [23] . In the UK, retrospective single-centre studies have observed low rates of bacterial co-infection [24] [25] [26] [27] .

In contrast to studies that relied on predominantly culturebased techniques, Kreitmann et al. identified early bacterial co-infection in 27.7 % (13/47) of their prospective cohort of ventilated patients using a multiplex PCR assay with only one case identified by conventional culture [28] . In France, a single centre study using three multiplex PCR assays performed on respiratory specimens or nasopharyngeal swabs in addition to standard culture techniques retrospectively identified bacterial co-infection in 28 % of 92 ICU admissions [29] . Variations in case definitions, diagnostic testing and geography may partly account for the differences observed between studies although overall, there is a suggestion that increased severity of disease, particularly when ICU care is required, is associated with increased rates of co-infection.

The prevalence of nosocomial infection is 20.6 % and increases with duration of ICU stay [30, 31] . Our observed co-infection/ co-colonisation rate is relatively high, consistent with a patient cohort with long ICU stays (median 10 days) and requiring high levels of respiratory support.

Consistent with reports from other studies, the commonest potential co-infecting bacteria identified within 48 h of admission was S. aureus [25, 28, 32] . In patients in whom early co-infection is suspected clinically, due consideration of S. aureus is warranted. However, the rate of S. aureus co-infection is markedly lower than that observed in pandemic influenza, suggesting it is a less significant issue with COVID-19 infection [18] . The predominant late pathogens observed were Gram-negative bacteria, particularly K. pneumoniae. These pathogens are commonly associated with hospital and ventilator-acquired pneumonia and have been reported as common co-pathogens in COVID-19 infections, particularly ICU cohorts [21, 22, [33] [34] [35] . The predominance of Gram-negative bacteria in these studies likely reflects nosocomial infection following prolonged ICU stay and empirical antibiotic use.

Viral co-pathogen was identified in one patient in our cohort; lower than the 3 % (95 % CI 1-6 %) viral co-infection rate reported in systematic reviews and in contrast to the 20.7 % viral co-detection rate reported by Kim et al. in Northern California [21, 36] . The 2019/20 influenza season in the UK ended in late March [37] . Other UK cohorts recruited during the spring wave of COVID-19 (March -May 2020) similarly reported very little or no viral co-infection [25, 35] .

This pragmatic multicentre study provides novel data on both community-acquired and nosocomial co-infection/ co-colonisation in patients with COVID-19 requiring ICU care in England. The ICU cohort represents those with severe disease who were subject to more rigorous microbiology sampling. A key limitation of the study is its retrospective observational design subject specifically to case selection, ascertainment and sampling biases. Inclusion of consecutive eligible patients was not feasible due to pandemic workload constraints. To minimise case selection bias, participating sites submitted a random sample of their eligible cohort, although random sampling methods were not standardised. The impact of ascertainment bias due to differences in the proportion of eligible cases submitted by each institution was reduced through the participation of multiple centres. The study cohort was comparable to the ICNARC cohort except for an under-representation of patients of Black, Asian and Minority Ethnicity (BAME). Our results may not be applicable to settings with larger BAME populations. Restriction of our cohort to those with completed ICU admissions excluded: (i) frailer patients in whom ICU care was deemed not appropriate, and (ii) patients with very long ICU stays. Co-infection, particularly nosocomial infection, may be higher in these patients.

A second key limitation is that although results likely to represent contamination were excluded, some pathogens found in respiratory tract samples may represent colonisation rather than active co-infection. However, as sputum samples sent from ICU should reflect clinical concern of lower respiratory tract infection (especially during the pandemic timeframe) and positive culture represents predominant presence of a pathogen rather than as part of mixed flora, we have taken these results to represent infection. If colonising pathogens were wrongly attributed as causing infection, the direction of bias would be towards falsely higher co-infection rates observed in our study.

Thirdly, reliance on culture dependent techniques may have falsely decreased co-infection rates. Antibiotic use prior to admission was low (13.8 %), increasing the reliability of culture-based methods on admission. However, detection of pathogens later into admission would have been influenced by sampling bias and the use of empirical antibiotics. Fourthly, although seven hospitals participated in this study, one study site contributed a third of cases; observed 48 h co-infection/ co-colonisation rate excluding this site was, however, similar to overall results.

Notwithstanding these limitations, our data indicate that early in hospitalisation, bacterial co-infection in COVID-19 is very uncommon and support the recommendations that empirical antibiotics should not be started routinely in primary care or at the point of hospital admission without clinical suspicion of bacterial infection [8] . The high rate of co-infection found late in illness among patients requiring ICU and involving nosocomial pathogens is concerning. It is plausible that reducing unnecessary early antibiotic exposure in patients with COVID-19 could reduce their risk of late, Gram-negative, potentially antibiotic resistant infections [38, 39] .

Since study completion, dexamethasone has been shown to decrease mortality in patients hospitalised with COVID-19 who require oxygen support or invasive mechanical ventilation [40] . Consequently, dexamethasone has become established as standard of care for these patients in many countries. This may increase the already high rate of bacterial co-infection we observed in ICU-treated patients. A high level of microbiological vigilance is recommended as part of the management of these patients. In the setting of seasonal changes in respiratory pathogens, ongoing surveillance for co-infections in patients hospitalised with COVID-19, ideally through prospective studies with standardised sampling protocols, is advised.

Data are available from the corresponding author on reasonable request. 

We wish to thank Mr Glenn Hearson for building the study database on the secure online database platform (REDCAP Cloud).

All included authors fulfil the criteria of authorship; V. B. and H. L., are joint first authors for this manuscript. W. S. L., V. B. and H. L., had substantial contributions to the study conception and design. All authors had substantial contributions to the data acquisition. V. B., performed the analyses. All authors had substantial contributions to the results interpretation. VB and HL wrote the original draft. All authors revised the manuscript critically for important intellectual content, provided the final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The co-pathogenesis of influenza viruses with bacteria in the lung

gclid= CjwK CAjw 8df2 BRA3 EiwA vfZW aJWn mCWZ BUjJ dJZG VdH4 hGEN u8or jqQT HDsIst5u_ gYXoQcl8sS_ ZxoCIxEQAvD_ BwE

Clinical characteristics of coronavirus disease 2019 in China

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Elevated levels of IL-6 and CRP predict the need for mechanical ventilation in COVID-19

C-Reactive protein correlates with CT findings and predicts severe COVID-19 early

National Institute for Health and Care Excellence N. COVID-19 rapid guideline: antibiotics for pneumonia in adults in hospital: NICE

Trends in blood culture contamination: a College of American pathologists Q-Tracks study of 356 institutions

Updated review of blood culture contamination

Blood culture contamination with enterococci and skin organisms: implications for surveillance definitions of primary bloodstream infections

Epidemiology of candiduria and Candida urinary tract infections in inpatients and outpatients: results from a 10-year retrospective survey

The significance of Candida in the human respiratory tract: our evolving understanding

): investigation of bronchoalveolar lavage, sputum and associated specimens

Overtreatment of enterococcal bacteriuria

Britiish Sociecty of Thoracic Imaging B. Covid-19 BSTI reporting templates and codes 2020

Intensive Care National Audit and Research Centre I. Online reports 2020

Critical illness from 2009 pandemic influenza A virus and bacterial coinfection in the United States

Critically ill patients with the middle East respiratory syndrome: a multicenter retrospective cohort study

Severe acute respiratory syndrome in Taiwan: analysis of epidemiological characteristics in 29 cases

Co-Infections in people with COVID-19: a systematic review and meta-analysis

Bacterial co-infection and secondary infection in patients with COVID-19: a living rapid review and meta-analysis

Bacterial and fungal coinfection in individuals with coronavirus: a rapid review to support COVID-19 antimicrobial prescribing

Low rate of bacterial co-infection in patients with COVID-19

Bacterial and fungal coinfection among hospitalized patients with COVID-19: a retrospective cohort study in a UK secondary-care setting

Improving antibiotic stewardship in COVID-19: bacterial co-infection is less common than with influenza

Investigation of subsequent and co-infections associated with SARS-CoV-2 (COVID-19) in hospitalized patients

Early bacterial co-infection in ARDS related to COVID-19

Bacterial and viral co-infections in patients with severe SARS-CoV-2 pneumonia admitted to a French ICU

Nosocomial infections in adult intensive-care units

International study of the prevalence and outcomes of infection in intensive care units

Bacterial and fungal coinfections in COVID-19 patients hospitalized during the new York City pandemic surge

Microbial etiologies of hospital-acquired bacterial pneumonia and ventilator-associated bacterial pneumonia

Incidence of co-infections and superinfections in hospitalised patients with COVID-19: a retrospective cohort study

Organisms causing secondary pneumonias in COVID-19 patients at 5 UK ICUs as detected with the FilmArray test

Rates of co-infection between SARS-CoV-2 and other respiratory pathogens

Comparison of 8 vs 15 days of antibiotic therapy for ventilator-associated pneumonia in adults: a randomized trial

Short-course empiric antibiotic therapy for patients with pulmonary infiltrates in the intensive care unit. A proposed solution for indiscriminate antibiotic prescription

Dexamethasone in Hospitalized Patients with Covid-19. Preliminary report

Professor Lim reports grants from National Institute for Health Research (NIHR), grants from Pfizer, outside the submitted work. Paul Dark is funded by NIHR Manchester BRC as sub-theme lead in Respiratory Infections. 

Five reasons to publish your next article with a Microbiology Society journal 1 . The Microbiology Society is a not-for-profit organization. 2. We offer fast and rigorous peer review -average time to first decision is 4-6 weeks. 3. Our journals have a global readership with subscriptions held in research institutions around the world. 4. 80% of our authors rate our submission process as 'excellent' or 'very good'. 5. Your article will be published on an interactive journal platform with advanced metrics.Find out more and submit your article at microbiologyresearch.org.