key: cord-0707355-aejjb24p
authors: Gerver, Sarah M.; Guy, Rebecca; Wilson, Kate; Thelwall, Simon; Nsonwu, Olisaeloka; Rooney, Graeme; Brown, Colin S.; Muller-Pebody, Berit; Hope, Russell; Hall FFPH, Victoria
title: National surveillance of bacterial and fungal co- and secondary infection in COVID-19 patients in England – Lessons from the first wave
date: 2021-06-08
journal: Clin Microbiol Infect
DOI: 10.1016/j.cmi.2021.05.040
sha: d1be0a359ee9cf6093abcf89c02c5c89505cad7c
doc_id: 707355
cord_uid: aejjb24p

OBJECTIVES: The impact of bacterial/fungal infections on the morbidity and mortality of persons with COVID-19 remains unclear. We have investigated the incidence and impact of key bacterial/fungal infections in COVID-19 persons in England. METHODS: We extracted laboratory-confirmed SARS-CoV-2 cases (01/01/2020-02/06/2020) and blood and lower-respiratory specimens positive for 24 genera/species of clinical relevance (01/01/2020-30/06/2020) from Public Health England’s national laboratory surveillance system. We defined coinfection and secondary infection as a culture-positive key organism, isolated within 1-day, or 2-27 days, respectively of the SARS-CoV-2 positive date. We described the incidence and timing of bacterial/fungal infections and compared characteristics of COVID-19 patients with and without bacterial/fungal infection. RESULTS: One-percent (n=2,279/223,413) of COVID-19 persons in England had co/secondary infection, >65% were bloodstream infections. The most common causative organisms were Escherichia coli, Staphylococcus aureus and Klebsiella pneumoniae. Cases with co/secondary infections were older than without (median 70-years [IQR:58-81] vs 55-years [IQR:38-77]), and a higher percentage of cases with secondary infection were of Black or Asianst ethnicity than cases without (6·7% vs 4·1%, and 9·9% vs 8·2%, respectively, p<0·001). Age-sex-adjusted case fatality rates were higher in COVID-19 cases with a coinfection (23·0% [95%CI:18·8%-27·6%] or secondary infection (26·5% [95%CI:14·5%-39·4%]), than those without (7·6% [95%CI:7·5%-7·7%])(p<0·005). CONCLUSIONS: Co/secondary bacterial/fungal infections were rare in non-hospitalised and hospitalised persons with COVID-19, varied by ethnicity and age, and were associated with higher mortality. However, the inclusion of non-hospitalized persons with asymptomatic/mild COVID-19 likely underestimated the rate of secondary bacterial/fungal infections. This should inform diagnostic testing and antibiotic prescribing strategy.

J o u r n a l P r e -p r o o f Introduction With the second peak of SARS-CoV-2 infections, combined with the winter season in the northern 55 hemisphere, there is an urgent need to better understand the effect of bacterial and fungal coinfections 56 on the outcomes of COVID-19 patients. Given the potential for severe disease, hospitalisation and 57 mechanical ventilation, (1) there is concern regarding the additional morbidity posed by coinfections 58 amongst COVID-19 patients. 59

Bacterial/fungal infections can be a serious complication following influenza infection, with 60 prevalence between 2% and 65% in influenza patients. ( 

The prevalence of bacterial/fungal infections in COVID-19 patients and associated morbidity and 62 mortality is currently unclear. Early studies indicate a lower prevalence than for pandemic influenza, 63 with a pooled prevalence of 7% from 18 studies (range:0-45%). (3) To date, these studies have been 64 small-scale, often single-hospital settings. One study in two UK National Health Service (NHS) acute 65 hospitals found 3·2% (n=27/836) of COVID-19 patients had a confirmed bacterial infection identified 66 0-5 days post hospital admission, increasing to 6·1% (n=51/836) over the entire admission.(4) 67

Prevalence of bacterial/fungal infections in COVID-19 varies by setting and is higher in severe 68 COVID-19 patients. A US study found 11·9% of COVID-19 patients receiving invasive mechanical 69 ventilation had bacteraemia versus 1·8% of those who did not,(5) whilst a meta-analysis found 14% 70 of COVID-19 patients in intensive care units (ICUs) had bacterial coinfection versus 4% in mixed 71 hospital settings. (3) There are also indications that coinfection increases mortality, one meta-analysis 72 particularly given England's second wave coinciding with seasonal increases in respiratory infections, 81 to better characterise the frequency, profile and impact of bacterial/fungal infections in COVID-19 82

We have undertaken a national retrospective cohort study, using laboratory surveillance data, to 84 describe bacterial/fungal infections in all persons with COVID-19 diagnosed in England until the end 85

of May 2020. This information should inform diagnostic testing, antibiotic prescribing and infection 86 prevention and control (IPC) precautions strategies. 87 88

We undertook a national retrospective cohort study using Public Health England's (PHE) routine 90 laboratory surveillance data. We defined our cohort as laboratory confirmed COVID-19 cases 91

reported to the PHE Second Generation Surveillance System (SGSS) Communicable Disease 92

Reporting (CDR) module with a specimen date falling within weeks 1-22 of 2020 (01 January 2020-93 02 June 2020). Of the testing pathways feeding into SGSS, we used data from Pillar 1 (testing in NHS 94 and PHE laboratories for hospitalised patients and healthcare workers) and Pillar 2 (commercially run 95 laboratories testing the non-hospitalised public). Only the first laboratory confirmed report of SARS-96

CoV-2 was used per person. 97

Our outcome variable was co-or secondary bacterial/fungal infection, which was defined as a 98 pleura, bronchoalveolar lavage, sputum, endotracheal aspirate and pleural fluid. For data on 108

Clostridioides difficile infections (CDI), we extracted all confirmed diagnoses in patients (≥2 years 109 old) for the same time period from the Mandatory Surveillance of Healthcare Associated Infections 110 (HCAI) Data Capture System (DCS) on 22 July 2020. 111

Data for all bacterial/fungal organisms were grouped into rolling 14-day organism and site-specific 112 (blood/respiratory) episodes, except for CDI which due to its different data source were collected in 113 static 28-day episodes. Further exceptions were Mycoplasma species identified from blood, where 114 only culture results in patients aged <16 years were included. 115

The bacterial/fungal and SARS-CoV-2 datasets from SGSS and the CDI dataset from the HCAI DCS 116 were appended to each other. Multiple observations from all data sources from individual patients 117

were grouped together under a single unique identifier using a multi-stage algorithm (Supplementary 118

Information Table SI2) . 119

The first specimen date of each bacterial/fungal episode was used to determine when a co/secondary 121 infection occurred in relation to the SARS-CoV-2 positive specimen date. Categorisation as co-or 122 secondary infection was as described in Figure 1 . Episodes which started more than one day before 123 the SARS-CoV-2 specimen date were excluded. Episodes were organism-and site-specific. For 124 persons with multiple episodes, only the first was retained when presenting data at an individual-level; 125 however, all episodes were included when presenting data at species-level. 126

To assess associations between patient characteristics and co/secondary infections, additional data 128

were sourced via data linkage on unique patient identifiers. Ethnicity was obtained from the from the NHS Spine.(10) 132

Case fatality rates (CFRs) were calculated among NHS Spine-matched cases, stratified by Results 145

During the study period 223,413 persons had laboratory confirmed SARS-CoV-2 infections. Among 147 these, 2,279 (1·0%, 95%CI:0·97%-1·06%) had a co/secondary infection with a key bacterial/fungal 148 organism (Table 1 ). There were 879 (38·6%; 95%CI:36·56-40·60%) coinfections and 1,400 (61·4%; 149 95%CI:59·40-63·44%) secondary infections. Most (66%) co/secondary infections were bloodstream 150 infections (BSI), occurring in 0·7% of all COVID-19 cases (95%CI:0·64-0·71%). BSI coinfection 151 were 6·5 times greater than respiratory coinfections while BSI secondary infections were 2·0 times 152 greater than respiratory secondary infections.

10-14 (March 2020), following a similar trajectory to COVID-19 diagnoses ( Figure 2a ). However, 156 these trends diverged after week 14 when co/secondary infections peaked and then rapidly declined, 157

whereas diagnoses of COVID-19 cases continued to increase. However, Figure 2b 

COVID-19 cases without co/secondary infection were younger (26% ≤40 years) and more likely to be 165 female (55·8%) than COVID-19 cases with a co/secondary infection (4-6% ≤40 years, ≤40% female 166 (P<0·001)) ( Table 2) . 167

Differences also existed in ethnicity; while the majority were White/White British a greater 168 percentage of COVID-19 cases with coinfection were White/White British (78.8%) versus those 169 without co/secondary infection or with secondary infection (<67%). Furthermore, more COVID-19 170 cases with secondary infection were of Black/Black British ethnicity (6·7%) or Asian/Asian British 171 ethnicity (9.9%) compared to both those without co/secondary infection (4·1% and 8.2%, 172 respectively) or with coinfection (4·4% and 5.7%, respectively) ( Table 2) . 173 respiratory and bloodstream specimens. The most common causative organisms of co/secondary BSI 181

were Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae, Enterococcus faecium and 182 non-pyogenic streptococci while for respiratory co/secondary infections Pseudomonas aeruginosa and 183

Haemophilus influenzae replaced E. faecium and non-pyogenic streptococci in the top ranking (Table  184 3, all species Supplementary Information Table SI2) . 185 E. coli, S. aureus and K. pneumonia were most common in both co/secondary infections; however, 186 streptococci were ranked top five amongst coinfections versus E. faecium and P. aeruginosa for 187 secondary infections (Table 3) . The majority of key bacterial/fungal species were predominantly secondary infections; however, the 192 majority of reported Streptococcus pneumoniae, non-pyogenic and pyogenic streptococci were 193 coinfections (66·7%, 62·6% and 84·1%, respectively). 194

The median age varied by organism and co/secondary infection status (Table 3) . 195

Of all secondary infection episodes, the median time to onset of bacterial/fungal infections is 13 days 196 (IQR:8-18). However, S. pneumoniae and H. influenzae have a shorter times to onset (4 (IQR:2-7) and 197 be required to limit our COVID-19 population to hospitalised patients only. People diagnosed with 206 COVID-19 in the community, who did not require hospital admission, would likely not have been 207 tested for other bacterial/fungal infections, resulting in an under-detection of co/secondary infections 208 in our cohort. That said, our cohort was determined by testing eligibility, which changed rapidly in the 209 early months of the epidemic. Between 5 th -27 th March 2020 testing was restricted to hospitalised 210 patients only,(12) which coincides with the higher incidence of co/secondary infection detected. 211

However, even this incidence, when largely restricted to hospitalised patients, is considerably lower 212 than reported from hospital-based studies. 213

Alternatively, the lower incidence of co/secondary infections in our cohort may due to under-214 detection. It is possible that early in the COVID-19 epidemic, acute service pressures reduced testing 215 for other organisms.(13) This may partially explain the reduction in laboratory reports on hospital 216 respiratory and blood samples of a wide range of organisms to SGSS detected in weeks 11 and 12 217 (Fig S1a, Fig1b) . For COVID-19 cases specifically, the continued microbiological investigation for 218 other organisms, particularly respiratory pathogens, may have often ceased following a positive 219 SARS-CoV-2 result. One study in two English acute hospitals reported that whilst 77% of their 220 COVID-19 patients had blood samples taken, only 15% had respiratory samples taken,(4) this may 221 explain our finding that >65% of coinfections were bacteraemias. 222

Whilst recognising that our study design may have underestimated the true incidence, given the 223 comprehensive scale of our study, we have shown that laboratory confirmed bacterial/fungal 224 infections in COVID-19 cases in England's first wave was considerably lower than feared, and much 225 lower than observed in seasonal and pandemic influenza. We have been restricted to using routine surveillance data for this study, and this has inherent 244 limitations. SGSS does not include clinical information nor prescribed medications, as such we cannot 245 determine what importance the timing/presence of antimicrobial or immunomodulatory therapy may 246 have had. Our outcome variable is dependent on which COVID-19 cases received additional 247 microbiological testing, and for what and when this testing was undertaken. This bias may have led to 248 an underestimation of the incidence. The data on bacterial/fungal disease is also from voluntary 249 surveillance, dependent on hospital laboratories reporting to SGSS. Whilst this data source has very 250 high completion,(18) it is possible that some testing and reporting was reduced during the COVID-19 251 response, partially explaining the reduction in laboratory reports detected, leading to missed reports of 252 bacterial/fungal infections in COVD-19 cases. A more robust study design would be to conduct 253 screening for these organisms in a cohort of COVID-19 patients.

including the full range of syndromes, from asymptomatic cases in the community to severely unwell 256 patients on ventilators with prolonged hospitalisation. The likelihood of bacterial/fungal infections is 257 likely to differ between these patient groups. We did not have access to additional information on 258 hospital admission, including to intensive care, or the use of ventilators or antimicrobials during this 259 analysis, but further work utilising data linkage will address some of these limitations and allow us to 260 determine which bacterial/fungal infections were likely healthcare associated. 261

Finally, whilst we included age-sex-adjusted 28-day all-cause CFR as an outcome and found higher 262

CFRs in patients with co/secondary infection these data should be treated with caution. It may still 263 reflect the differences in patient characteristics such as comorbidities, in the different subgroups and 264 as such acts as a signal warranting more sophisticated analysis. 265

During the first wave of COVID-19 in England, the overall incidence of key bacterial/fungal 267 co/secondary infections was low. Incidence of co/secondary infections peaked early in the pandemic, 268 potentially stemming from delays to healthcare seeking and temporary disruption to IPC measures. 

The percentage without any ethnicity information in the COVID-19 patients without a key bacterial/fungal co/secondary infection and patients with a secondary infection was more than double that observed in the coinfection group.

J o u r n a l P r e -p r o o f Table 3 is a count of all patient episodes of co/secondary bacterial/fungal pathogens and not per patient; therefore, numbers will differ between patient-episodes of bacterial/fungal pathogens and the number of COVID-19 patients.

NAnot applicable. IQR -Interquartile range (25 th and 75 th percentiles). Medians and IQR rounded to 0 decimal places. b can be found in place of for median and IQR calculated where there was only 1 value, i.e. median and IQR was the value for that individual patient-episode.

J o u r n a l P r e -p r o o f 

Clinical outcomes of COVID-19 in Wuhan, China: a large cohort study

The frequency of influenza and bacterial coinfection: a systematic review and meta-analysis. Influenza and other respiratory viruses

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

Bacterial and fungal coinfection among hospitalized patients with COVID-19: a retrospective cohort study in a UK secondarycare setting. Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases

Clinical Characteristics of Covid-19 in

Antibiotic guideline recommendations for empirical treatment of pneumonia in adults during the COVID-19 pandemic

Office for National Statistics. Populations by sex, age group and Index of Multiple Deprivation (IMD) quintile

The NHS Spine

Estimates of the population for the

Government launches new drive on coronavirus tests for frontline NHS staff

Recommendations from RCPath and Professional Bodies (IBMS, ACP and ACB)

Prioritiation/deferral of Pathology Laboratory Work (in light of SARS-CoV-2 (COVID19) epidemic)

bf69ed866a7ca3da/Prioritisation-deferral-of-pathology-laboratory-work.pdf

Monitoring respiratory infections in covid-19 epidemics

How coronavirus lockdowns stopped flu in its tracks. Nature. 2020:ePub ahead of print

Severe COVID-19 and healthcare-associated infections on the ICU: time to remember the basics? The Journal of hospital infection

J o u r n a l P r e -p r o o f