Incidence, Risk Factors, and Outcomes of Patients with COVID ‐ 19 ‐ Associated Pulmonary Aspergillosis (CAPA) in Intensive Care Units: A Systematic Review and Meta-Analysis Of 29 Cohort Studies

Background Patients with Coronavirus disease 2019 (COVID-19) admitted to an intensive care unit (ICU) might develop COVID-19-related pulmonary Aspergillosis (CAPA). We aimed to identify studies systematically that describe the incidence and risks factors of CAPA, and to assess its outcome. Two independently We included observational cohort studies that investigated patients with to an We assessed the of all included studies using the Newcastle– Ottawa Scale). The meta-analysis registered with (CRD42021242179). Twenty-nine cohort studies with 2095 patients with COVID-19 admitted to an ICU and 264 patients who developed to CAPA were included (Pooled incidence: 0.14, 95% condence interval [CI] = 0.11–0.17). The overall mortality and case fatality rate of CAPA were 0.07 (0.05–0.09) and 0.51 (0.44–0.58), respectively. at after (range, 5.48–9.08). those signicantly median body mass index (27.32 vs. 28.97 kg/m 2 , P = 0.034), higher median level (127.94 vs. 88.23 µmol/L, P = 014), (41.0% risk ratio [RR] = 95% CI=1.08–3.63) (42.0% vs. 28.2%, RR = 1.61, 95% (RR = 1.66, 95% CI=1.31–2.12) and publication bias. Patients with COVID-19 admitted to an ICU might develop CAPA and have higher all ‐ cause mortality. We recommend conducting prospective screening for CAPA among patients with severe COVID-19, especially for those who receive mechanical ventilation over 7 days. of found the selected of medium (60.78%). representativeness of the cohort, determination of comparability of the cohorts on the design or and the adequacy of the follow up of the cohorts (these factors were scored as 34.48%, 27.59%, and 51.72%, respectively), which should be improved in future cohort studies. the investigate CFR, research study exclusion


Quality Assessment
The quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS) by the two researchers who were involved in the literature search (Chen WS and Zhang K), who assessed the quality of all included studies and discussed discrepancies until a consensus was reached with the corresponding author (Zhuang GH).

Data Extraction and Outcome Measures
Data were extracted for the following variables: study period, sample size, de nition of CAPA, age, sex, underlying disease, laboratory examination, exposure information of CAPA, and outcome. Mortality was de ned as the number of deaths diagnosed with CAPA divided by the number of COVID-19 cases in the cohort. The case fatality rate (CFR) was de ned as the number of CAPA deaths divided by the total number of CAPA cases in the cohort. The primary outcome was all-cause mortality and all-cause CFR during admission. The secondary outcome was the length of ICU stay for the patients with COVID-19.

Statistical Analysis
Random-effects meta-analysis [18] was used to reported the incidence, all-cause mortality, all-cause CFR, and risk ratios (RRs) with 95% con dence intervals (CIs). For pooling of the means of numerical variables, we computed missing means and standard deviations (SDs) from medians; ranges (minimum to maximum); and inter-quartile ranges (IQRs); using the methods proposed by Hozo et al. [19]. Pooled proportions were computed using the inverse variance method and the variance-stabilizing Freeman-Tukey double arcsine transformation [20]. Subgroup analysis was conducted based on the literature design (prospective or retrospective study), the type of research centre (Single or Multi-centre), and the clinical characteristics of the patients (e.g., acute respiratory distress syndrome (ARDS), acute respiratory failure (ARF)). The Mantel-Haenszel estimator was used to calculate the between-study heterogeneity statistic Q, which was used in the DerSimonian-Laird estimator. The I 2 statistic was used to present between-study heterogeneity, where I 2 ³ 50% was considered substantial. We assessed publication bias using Egger's test and by inspecting funnel plots. Statistical analyses were performed using the R software version 3.6.3 (R Foundation for Statistical Computing, 2016), with the package "meta" and Stata Special Edition 15.1 (Stata Corp, College Station, TX, USA). P-values less than 0.05 were considered statistically signi cant.

Results
Our search algorithm in the MEDLINE, OVID, PubMed and MedRXIV databases retrieved 1441 records. We screened based on the title and abstract and excluded 1377 studies. Sixty-four articles were assessed for eligibility. Finally, we identi ed 29 observational cohort studies that quali ed according to our prede ned inclusion and exclusion criteria (Figure 1), which included a total of 2095 patients with COVID-19 admitted to an ICU.

Quality control of included studies
The risk of bias included in the cohort studies was assessed using the Newcastle-Ottawa-Scale (NOS) as presented in Table 1. The overall score was 141 out of 232 (60.78%), which is considered to be indicative of moderate quality. All the eligible studies included patients with severe COVID-19 admitted to ICU.
Among the studies, 34.48% (10/29) did not describe the determination of exposure. In addition, 72.41% (21/29) studies did not control the comparability of the cohorts on the basis of the design or analysis. Although all studies (29/29) followed up the patients for long enough for de ned outcomes to occur, only 51.72% (15/29)

Incidence of CAPA
In the studies, 2095 patients with severe COVID-19 admitted to an ICU were investigated and 264 cases of CAPA were reported. The incidence of CAPA was calculated as 0.14 (95% con dence interval [CI] = 0.11-0.17, I 2 = 81.2%). In the sub-group analysis by study design, the incidence of CAPA was calculated at 0.14 (95% CI = 0.09-0.19, I 2 = 87.4%) in the prospective studies and 0.14 (95% CI = 0.10-0.18, I 2 = 74.1%) in the retrospective studies (Table 1). When strati ed analysis was undertaken by research centre (multi-centre or single centre), the CAPA incidence was 0.12 (95% CI = 0.10-0.15, I 2 = 77.3%) for the single centre studies and 0.20 (95% CI = 0.12-0.28, I 2 = 81.2%) for the multi-centre studies. We found a high heterogeneity for calculation of the incidence of CAPA and signi cant publication bias according to Egger's regression test (See additional le 1, supplementary 3 Figure A). Remarkably, we found that patients with COVID-19 admitted to an ICU would develop CAPA after approximately 7.28 days of mechanical ventilation (nine studies, 95% CI = 5.48--9.08, I 2 = 46%) ( Figure 2).

Colonisation by Aspergillus
In ve cohort studies including 365 patients with COVID-19 admitted to an ICU, 21 patients were assessed for Aspergillus colonisation. The calculated colonization rate of was 5.75% (21/365) (See additional le 1, supplementary 1), suggesting that clinicians should comprehensively analyse Aspergillus colonisation combined with other clinical evidence (such as imaging and in ammatory factors, etc.) to judge whether the patient might develop CAPA and require further anti-fungal treatment.

CFR of CAPA
Twenty-three studies were eligible for further analysis of all-cause CFR with 119 deceased patients and 145 surviving patients. We found that the pooled CFR  Outcome with CAPA or without CAPA Eleven studies were included. The analysis showed that patients with CAPA were associated signi cantly with a 1.66-fold higher risk for mortality (risk ratio  (Figure 3 A), a 1.52-fold elevated risk of mortality (nine studies, RR = 1.52, 95% CI = 1.12-2.07, I 2 = 24.7%) the for single centre studies, and a 2.03-fold elevated risk of mortality (two studies, RR = 2.03, 95% CI = 1.46-2.81, I 2 = 0%) for the multi-centre studies (Figure 3 B). No statistically signi cant difference was found in the length of ICU stay for CAPA and non-CAPA patients (MD = 5.58, 95% CI = -1.93 to 13.08, P = 0.145) ( Table 2).

Discussion
We conducted this systematic review of 29 cohort studies and observed that 14% of patients with COVID-19 admitted to the ICU developed CAPA. Meanwhile, the median time to develop CAPA among patients with COVID-19 who received mechanical ventilation was 7.28 days. Studies reported that the incidence of CAPA varied from 1.9-27.7% [14,43,45]. This wide divergence might result an underestimate of the risk of CAPA. Invasive pulmonary Aspergillosis is a secondary infection of severe respiratory viral infection [46,47] and these infections have been related with signi cant morbidity and mortality, even when appropriately diagnosed and treated [48]. CAPA not only occurred in immunosuppressed patients (e.g., those with acute myeloid leukaemia) [49], but also in immunocompetent patients with COVID-19 [48,50]. COVID-19 was independently associated with CAPA [36]. However, we observed signi cant heterogeneity in the pooled analysis of CAPA incidence. Many hospitals do not commonly perform bronchoscopy for CAPA diagnosis or use PCR tests for Aspergillus species [51]. Non-standardized de nitions of CAPA (See additional le 1, supplementary 4), differences in detection methods [52], and variations among research designs might have contributed to heterogeneity across the studies, which requires further investigation. Nevertheless, we still recommend that clinicians should be alert to the occurrence of CAPA in patients with COVID-19, especially in those receiving mechanical ventilation more than 7 days.
The patients with COVID-19 who had higher APACHE II scores [36] and complicated underlying diseases (such as hypertension, diabetes, and tumours) [53], were independently associated with CAPA. We found that patients with CAPA had a lower BMI and higher concentration of creatinine compared with those without. One study indicated that elevated serum creatinine level was an independent risk factor for neutropenia-related invasive Aspergillosis [54]. The roles of BMI and creatinine in CAPA deserve further investigation. In this study, no differences in use of Lopinavir/ritonavir, HCQ, Tocilizumab, ventilated prone, and ECMO were observed in the patients with CAPA (As shown in Table 2). However, patients with CAPA were more likely to have received corticosteroid therapy and renal replacement therapy during admission compared with the non-CAPA controls. Previous studies reported that the use of corticosteroids increased the risk of invasive pulmonary Aspergillosis signi cantly after severe viral infection (In uenza) [55]. Patients requiring dialysis and kidney transplantation were also at risk of invasive Aspergillosis [56]. Existing evidence supported the view that the use of corticosteroids and renal replacement therapy might play an important role in the progression of patients with COVID-19 with CAPA.
We pooled 11 cohort studies and observed that the patients with CAPA were associated signi cantly with a 1.66-fold higher risk of mortality (1.66, 95% CI = 1.31-2.12, I 2 = 22.5%) (Fig. 3) without signi cant heterogeneity and publication bias (See additional le 1, Supplementary 3 Figure D). The ndings supported the view that patients with CAPA had uniformly worse outcomes compared to those without, regardless of the de nitions used. Meanwhile, we investigated the outcome of CAPA and found that the overall mortality of CAPA in COVID-19 patients in the ICU was 0.07 (Pooled mortality = 0.07, 95% CI = 0.05-0.10). The occurrence of CAPA would exacerbate the deterioration of respiratory function, thus increasing the risk of death. We pooled the studies and found that the allcause CFR of patients with CAPA was 51%. Among all included studies, a very high proportion of patients with COVID-19 received mechanical ventilation. The proportion of patients receiving baseline mechanical ventilation in the 13 studies was more than 89% (11 studies, 100% mechanical ventilation). A systematic review has revealed that the CFR of patients with COVID-19 receiving mechanical ventilation was 45% (95% CI, 39-52%) [57]. Meijer and others conducted a small-sample cohort study (66 ICU patients with COVID-19) and found that the incidence of CAPA was 19.7% (n = 13/66), and the fatality rate of a single study was 46.2% (6/13); the author's team listed some of the articles that have been published and analysed for the CAPA fatality rate. The pooled CFR for CAPA was 47% (62/132) [32].
The NOS was used to evaluate the quality of the included articles. We found that the selected studies were of medium quality (60.78%). The most often overlooked factors were the representativeness of the exposed cohort, determination of exposure, comparability of the cohorts on the design or analysis, and the adequacy of the follow up of the cohorts (these four factors were scored as 0%, 34.48%, 27.59%, and 51.72%, respectively), which should be improved in future cohort studies. Meanwhile, we pooled the studies to investigate the incidence of CAPA incidence and CFR, and observed signi cant research heterogeneity. On the one hand, because of the lack of comparison of individual rates, heterogeneity occurred easily [57]. On the other hand, there was lack of a uni ed de nition of CAPA (See supplementary 4) and a standard diagnostic algorithm in the published studies. Although we strictly controlled quality based on the study entry and exclusion criteria, heterogeneity between studies was still unavoidable. This meta-analysis did not explore the risk factors for mortality in patients with CAPA because of insu cient literature support. Further well designed research and demonstration would be warranted in the future.

Conclusion
In this study, the pooled incidence of CAPA was estimated as 14% among patients with COVID-19 admitted to the ICU, with high rate of all cause-mortality and CFR. We recommend that clinicians establish standardized screening methods for CAPA and to identify high risk patients early, especially for patients who have been mechanically ventilated for more than 7 days. Host factors (BMI and creatinine level) and the treatment of COVID-19 (the use of corticosteroids and renal replacement therapy) would signi cantly affect the development of CAPA. The datasets synthesized during the current study are available from the corresponding author on reasonable request.  Figure 1 Flow Diagram of this study

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.