Increased mortality in COVID-19 patients with fungal co- and secondary infections admitted to intensive care or high dependency units in NHS hospitals in England.

Dear Editor,

We read with interest the recent systematic review and meta-analysis by Lansbury et al. which aimed to evaluate the burden of bacterial and fungal co-infections in patients with COVID-19, highlighting the challenges in diagnosing certain fungal infections in this setting and the need for a high level of suspicion of fungal infections in critically-ill COVID-19 patients. Severe COVID-19 disease does predispose patients to fungal co- and particularly secondary infections, with reports describing COVID-associated pulmonary aspergillosis (CAPA) and invasive yeast infections (particularly candidaemia) predominating.2, 3, 4, 5, 6, 7, 8, 9, 10 Additionally, a devastating wave of COVID-associated mucoromycosis (CAM) impacted the Indian subcontinent in late spring, with tens of thousands of cases reported in India alone. Anecdotal reports and individual case series suggested that each of these fungal complications of COVID-19 increases the risk of poor outcomes, including increased hospital lengths of stay (LoS) and excess mortality. , , , A recent French multi-center study confirmed significant increased mortality in patients with possible or probable CAPA. Here we report data to quantify the impact of fungal co- and secondary infections on outcomes of patients admitted to intensive care units (ICU) in NHS hospitals in England with laboratory confirmed COVID-19 and likely or proven fungal disease, concentrating on CAPA and COVID-associated candidaemia (CAC), since few UK cases of CAM have been reported as of November 2021.

De-duplicated data on patients with candidaemia or aspergillosis (March 2020-March 2021) were collated from the UK Health Security Agency (UKHSA) UK National Mycology Reference Laboratory (MRL); candidaemia patients were also collated from the Communicable Disease Reporting module of UKHSA's Second Generation Surveillance System. Laboratory confirmed cases of COVID-19 were retrieved from the PHE National Incident Coordination center Epidemiology Cell (NICCEC). Data on Intensive Care Unit (ICU) admissions in NHS hospitals in England was collected from Secondary Uses Service (SUS) +, NHS Digital, and sorted into total ICU stays, before being linked to the laboratory confirmed COVID-19 cases to create a dataset of patients admitted to ICU with COVID-19.

Candida data was first linked to the NICCEC data to determine which of the candidemia patients also had a COVID-19 diagnosis. Patients with diagnoses of candidemia that were more than 10 days before or 60 days after a COVID-19 diagnosis were excluded, except where the patients had remained hospitalised for the entire period between diagnoses. Retained records were then linked to the ICU dataset of COVID-19 admissions. Data linkage across the datasets was performed in 3 stages to capture patients with missing key identifiers using: (i) NHS number and date of birth, (ii) date of birth and soundex of both surname and forename, (iii) NHS number and soundex of both surname and forename. After data linkage and exclusion, 34,550 patient records were available for analysis, 271 of which highlighted patients with CAC. Remaining patients with COVID-19 with no evidence of either Aspergillus or candidaemia co/secondary infection formed the reference group. CAC cases were categorised as (i) preceding COVID-19 diagnosis (N = 6; recovery of Candida sp. from blood between 10 and 1 days pre-COVID-19 test positivity); (ii) co-infection (N = 7; blood culture positivity within 1 day either side of COVID-19 test positivity) and (iii) secondary infection in 258 patients (blood culture positivity 2 to 60 days post COVID-19 diagnosis in 255 patients, plus 3 cases of candidaemia >60 days post COVID-19 diagnosis in patients who had remained hospitalised for the entire duration). Among the ICU cohort, 60-day crude case fatality rate (CFR) for patients with CAC was 57.2% (95% CI 48.5–66.9; 155/271 patients), compared to 39.9% (95% CI 39.3–40.6; 13,684/34,279 patients) for patients with COVID-19 alone (p < 0.001). When adjusted for age and sex, both CFRs decreased, to 37.5% (95% CI 25.8–50.7) in patients with CAC and 25.5% (95% CI 24.8–26.2) in patients with COVID-19 alone (p = 0.058). Finally, median LoS in ICU patients with CAC was substantially longer (25 days; IQR 14–43; N = 270) than median LoS among ICU patients with COVID-19 alone (7 days; IQR 3–16; N = 34,057; p < 0•001). Median time from ICU admission to the Candida sample was 14 days (IQR 8–21), with patients then staying in ICU for a median of 10 days (IQR 2–27) from the fungal diagnosis to discharge.

For CAPA, accurate evaluation of incidence is confounded by multiple factors, including lower likelihood of bronchoalveolar testing of severely ill COVID-19 patients in ICU settings (due to risk of transmission from aerosols), geographic differences in biomarker testing and reporting (which influence incidence estimates), lack of a universally employed diagnostic algorithm and the fact that recovery of an Aspergillus spp. isolate from a COVID-19 respiratory sample is not proof of invasive disease. , , Given these constraints, a different approach to estimating excess mortality was employed. Initial data linkage and exclusion was performed against the NICCEC and SUS+ ICU data as above, starting with all UK patients from whom an isolate of Aspergillus fumigatus had been referred to the MRL from any respiratory secretion (sputum, tracheal aspirate, directed and non-directed bronchial lavages) from March 2020-March 2021. After linkage 34,398 patients with COVID-19 were available, of which 119 had Aspergillus fumigatus isolated from a respiratory specimen. For these 119 patients, the MRL Information Management System was interrogated for fungal biomarker testing (serum galactomannan [GM] or (1–3)-β-d-glucan, respiratory sample GM, Aspergillus-specific PCR) that could contribute towards a diagnosis of CAPA. A resulting 28 patients were stratified as possible/probable CAPA (multiple positive biomarker tests), 22 patients as unlikely to have CAPA (biomarker tests all negative), and 69 patients for whom biomarker tests were unavailable (unknown CAPA). The crude CFR for COVID-19 cases admitted to ICU with an Aspergillus fumigatus sample was 54.6% (95% CI 42.2–69.9; 65/119 patients). Adjusted for age and sex, this reduced to 44.4% (95% CI 25.6–66.9), higher than the COVID-19 ICU patients with no fungal infection (25.5%; p = 0.073). Looking to mortality by CAPA classification, crude CFRs were 67.9% (95% CI 40.8–100.0; 19/28) for possible/probable CAPA and 50.5% (95% CI 37.0–67.4; 46/91) for unlikely or unknown CAPA combined. Numbers were too small to look at age and sex adjusted CFRs by these categories. Finally, median LoS was significantly longer in possible/probable CAPA patients (29.5 days, IQR 23.5–43.5, n = 28) than in matched COVID-19 patients in ICU (median LoS 7 days; IQR 3–16; n = 34,057; p < 0.001). Median time from ICU admission to the Aspergillus sample was similar between possible/probable CAPA patients (16.5 days, IQR 11.5–26.5) and unknown/unlikely CAPA patients (17 days, IQR 12–24). This was also the case for time to discharge after the fungal sample, with the median time being 6.5 days (IQR 1.5–20) and 6 days (IQR −1–21) respectively.

This analysis has limitations. Cases of candidemia and CAPA are likely to be significantly underestimated as they relied upon laboratories reporting and/or referring isolates to the MRL. Laboratories may have undertaken in-house biomarker testing to aid in the diagnosis of CAPA, which would have helped to better stratify the patients in the “unknown CAPA” cohort. Linkage to SUS+ ICU data could only be done by NHS number and date of birth, so any COVID-19 patients missing these data could not be included. Nonetheless, these estimations for excess mortality caused by the two most common fungal co-/secondary infections encountered with severe COVID-19 align with prior smaller reports from individual centers (nationwide and worldwide) and the recent French study on CAPA in that mortality and LoS are increased in patients with COVID-19 and fungal co-/secondary infection. , , , This data analysis further emphasises the importance of raised awareness, testing, rapid diagnosis, surveillance and prompt and appropriate treatment of fungal infections in severely ill patients with COVID-19.

Declaration of Competing Interest

The authors have no competing interests to declare.

Funding

No specific funding support was received for this study.

CRediT authorship contribution statement

Andrew M. Borman: Conceptualization, Methodology, Data curation, Visualization, Formal analysis, Writing – original draft, Writing – review & editing, Validation. Holly Fountain: Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing, Validation. Rebecca Guy: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Ella Casale: Methodology, Project administration, Data curation, Writing – review & editing, Writing – original draft. Sarah M Gerver: Formal analysis, Writing – original draft, Writing – review & editing. Suzanne Elgohari: Formal analysis, Writing – original draft, Writing – review & editing. Colin S Brown: Conceptualization, Writing – original draft, Writing – review & editing. Susan Hopkins: Conceptualization, Writing – original draft, Writing – review & editing. Victoria J Chalker: Conceptualization, Methodology, Project administration, Writing – original draft, Writing – review & editing. Elizabeth M. Johnson: Conceptualization, Methodology, Writing – original draft, Formal analysis, Validation, Writing – review & editing.