key: cord-0820544-6f93ppei
authors: Meyer, Hans-Jonas; Wienke, Andreas; Surov, Alexey
title: Extrapulmonary CT findings predict in-hospital mortality in COVID-19. A systematic review and meta analysis
date: 2021-10-15
journal: Acad Radiol
DOI: 10.1016/j.acra.2021.10.001
sha: b6cf2f2d7a0ccd43a93e92d8794f86127559e8a5
doc_id: 820544
cord_uid: 6f93ppei

BACKGROUND: : Several prognostic factors have been identified for COVID-19 disease. Our aim was to elucidate the influence of non-pulmonary findings of thoracic computed tomography (CT) on unfavourable outcomes and in-hospital mortality in COVID-19 patients based on a large patient sample. METHODS: : MEDLINE library, Cochrane and SCOPUS databases were screened for the associations between CT-defined features and mortality in COVID-19 patients up to June 2021. In total, 22 studies were suitable for the analysis and included into the present analysis. Overall, data regarding 4 extrapulmonary findings could be pooled: pleural effusion, pericardial effusion, mediastinal lymphadenopathy, and coronary calcification. RESULTS: : The included studies comprised 7859 patients. The pooled odds ratios for the effect of the identified extrapulmonary findings on in-hospital mortality are as follows: pleural effusion, 4.76 [95% CI 2.97-7.62]; pericardial effusion, 1.21 [95% CI 0.74-1.98]; coronary calcification, 2.68 [95% CI 1.78-4.04]; mediastinal lymphadenopathy, 2.02 [95% CI 1.18-3.45]. CONCLUSIONS: : Pleural effusion, mediastinal lymphadenopathy and coronary calcification have a relevant association with in-hospital mortality in COVID-19 patients and should be included as prognostic biomarker into clinical routine. FUNDING: : None

ICU, and expanding more aggressive treatment, e.g. with extracorporal membrane oxygenation.

The purpose of the present systematic review and meta-analysis was to calculate the impact of CT-derived extrapulmonary features with in-hospital mortality in COVID-19 patients.

MEDLINE library, Cochrane and SCOPUS databases were screened for CT findings and in-hospital mortality in COVID-19 patients up to June 2021. The paper acquisition is summarized in figure 1 .

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement was used for the analysis [13] .

The studies were screened for potential prognostic CT findings. Most studies reported prognostic relevance of pleural effusion, pericardial effusion, mediastinal lymphadenopathy and coronary calcification. These findings were used for further analyses.

The following search words were used: -COVID-19‖ AND -Computed Tomography‖ OR -CT‖ AND -mortality‖ OR -severe course‖ OR -death‖.

The primary endpoint of the systematic review was the odds ratio of CT findings on mortality.

Studies (or subsets of studies) were included if they satisfied the following criteria: (1) COVID-19 diagnosis by PCR-RT, (2) reported CT findings (3) reported odds ratio or hazard ratio with confidence interval (CI). Exclusion criteria were (1) systematic reviews, (2) case reports, (3) non-English language, (4) other imaging modalities than CT.

In total, 22 studies were suitable for the analysis and included into the present study .

Data extraction was performed by HJM followed by an independent evaluation of extractions for correctness (AS). For each study, details regarding study design, year of publication, country of origin, patient number, patient characteristics (age and sex), diagnosis, treatment, CT findings, timepoint of the CT acquisition, survival outcome results and adjustment factors were extracted.

The quality of the included studies was assessed by the Newcastle-Ottawa Scale (NOS) [36] . Study quality assessment was conducted by two authors (HJM, AS), and mainly included the selection of cases, comparability of the cohort, and outcome assessment of exposure to risks. A score of 0-9 was assigned to each study, and a study with score ≥6 was considered to be of high quality.

The meta analysis was performed using RevMan 5.3 (2014; Cochrane Collaboration, Copenhagen, Denmark). Heterogeneity was calculated by means of the inconsistency index I 2 [37, 38] . DerSimonian and Laird random-effect models with inverse variance weights were performed without any further correction [39] . Possible publication bias was assessed with funnel plot and Egger test.

Additionally, sub-analyses were performed to test possible heterogeneities caused by different CT techniques (16 slices vs 128 and 256 slices) and by the origin of the studies.

Of the included 22 studies, 21 were retrospective (95.4%) and one was prospective (4.6%) [20] . Table 1 gives an overview of the included articles.

The overall risk of bias can be considered as low, indicated by the high NOS values among the studies ranging from 5 to 8 points (table 2) . Two studies [15, 25] did not report the exact patient recruitment duration, which can result in a potential bias. The exact CT timing was not reported sufficiently in several studies, which can be another bias. Funnel plot displays a publication bias (figure 2).

The included studies comprised 7859 patients. There were 4713 men (60%) and 3146 women (40%), with a mean age of 59.6 years ranging from 44.2 to 70 years. In all studies, COVID-19 was diagnosed with RT-PCR.

Overall, 19 studies (86.4%) investigated patients during the first wave of the pandemic, two studies (9.1%) did not report the exact time-period [15, 25] . Most studies investigated patients between March to April 2020. Only one study analyzed cases after the first wave with inclusion of patients between September and October 2020 [14] .

In 15 studies (68.2%) performed CT at the presentation of the admission or within 24 hours after admission. In 6 studies (27.3%) the timepoint of the CT was not clearly stated within the manuscript. In one study [30] investigating coronary calcifications, CTs of the patients were included up to 5 years prior to admission.

Thirteen studies (59.1%) were performed in Asia, six studies (27.3%) in Europe and three studies (13.6%) in North America. Table 3 provides an overview of the investigated sub-analyses according to study origin.

Overall, 15 studies with 3623 patients analyzed the effect of pleural effusion on inhospital mortality in COVID-19 patients. The pooled odds ratio for the association between pleural effusion and in-hospital mortality was 4.6 [95% CI 2.97-7.12, Tau² = 0.36, Chi²= 29.8, df= 14, I²=53%] (figure 3a).

On the next step, the reported data were analysed in accordance on study origin.

Three studies with 772 patients were performed in Europe. The pooled odds ratio No sub-analyses could be performed for perdicardial effusion. 

In 6 investigations with 1906 patients the effect of pericardial effusion on in-hospital mortality in COVID-19 patients was studied. In 5 studies the threshold value of the short axis diameter was 10 mm, in one study [15] no threshold value was provided. 

This is the first meta analysis regarding associations between non-pulmonary CT findings and mortality in COVID-19 patients. As shown, there was a statistically significant association of pleural effusion, coronary calcifications and mediastinal lymphadenopathy with in-hospital mortality, whereas no significant association was identified with pericardial effusion. These findings highlight the importance of extrapulmonary findings in COVID-19 infection.

COVID-19 has a high mortality for patients with an unfavourable course. Thus, a short-term mortality of up to 20% was reported in COVID-19 patients admitted to the intensive care unit (ICU) [2] [3] [4] [5] [6] 40] . As mentioned above, established prognosis parameters are age above 60 years and male sex, shorter period between symptom onset and emergency room presentation [2] [3] [4] [5] [6] 40] . Moreover, the extension of pulmonary consolidation on CT images is also considered as prognostic relevant [2, 41] . The consolidations are indicative of a disease progression and are most prominent in day 10 of the disease [8] .

The present analysis highlights the importance of CT for prognostic purposes beyond the quantification of pulmonary consolidations. Notably, the reported odds ratios are good comparable to the established risk factors, such as higher age over 60 years and male sex [5] , which highlights the importance of the extrapulmonary CT findings.

In a recent meta analysis regarding CT findings in COVID-19, the time dependence of different CT findings was investigated [42] . Pleural effusion and lymphadenopathy were more frequently identified in later disease stages [42] . So far, pleural effusion was found in 5% of patients in early stages and in 16% of patients in advanced stages [42] . Similarly, mediastinal lymphadenopathy was observed in 4% of patients in early stages and in 15% of patients in advanced stages [42] . The time dependence may be a potential confounder of the present analysis. However, in most studies.

CTs were performed at hospital admission.

Risk stratification of COVID-19 patients is very crucial for treatment planning.

Important clinical parameters were identified, and several scores were proposed to predict mortality in COVID-19 [40] . A recent study could show that a score based on serologically parameters comprising white blood cells, C-reactive protein, lymphocyte ≥0.8 × 10 9 /L, and lactate dehydrogenase ≥400 U/L was highly accurate to predict survival with an area under the curve of 0.95 [40] .

Very early on during the pandemic, it was shown that cardiovascular co-morbidities, especially coronary heart disease are an important risk factor for a severe COVID-19

course [43] . The present analysis can agree with this with a significant association between coronary calcification on CT images as an imaging finding of coronary heart disease. Clearly, there are cases of patients with amnestic known coronary heart disease without calcified plaques, which are not covered by this approach. Secondly, thoracic CT without contrast media application and electrocardiogram triggering cannot be considered as a diagnostic gold standard for cardiac imaging. However, the sole presence of calcified plaques in CT performed for COVID-19 evaluation can be suspicious for an unfavorable outcome. Moreover, promising data indicated that quantification of coronary plaques using scores is even better to predict unfavorable outcomes [19, 29] . As a shortcoming of the present study, we could not pool the results of these plaque scores further due to differences in the included studies. To harmonize the data, we only include the dichotomized analyses of presence of coronary plaque or not.

Pleural effusion was the strongest predictor for mortality in the presented results. It was early on discussed as a rare finding in patients with COVID-19 [44] and was stated to be more characteristic for other disorders, as pleural effusion is very common in critical ill patients. Clearly, there are many causes of pleural effusion, including viral pleuritis, congestive heart failure or cancer [33] .

As early on stated, extrapulmonary imaging findings may indicate the occurrence of severe inflammation as identified by Li and colleagues [9] . Another explanation was given that pleural effusion might indicate a bacterial superinfection as a severe complication of the COVID-19 infection.

In a comprehensive meta analysis of prognostic factors in COVID-19, pleural effusion showed even a higher odds ratio for severe course and mortality than pulmonary consolidation (OR of 3.31 versus 2.46) [45] . However, this analysis only included studies up to April 2020, which could explain the different results to the present study.

Unfortunately, most studies did not report the size of the pleural effusion. It is yet unknown, whether the size has also a significant effect on mortality of COVID-19 patients.

Mediastinal lymphadenopathy is also an imaging finding, which was considered as rare in COVID-19 [7] . It was also discussed as a possible sign for bacterial superinfection. The frequency of mediastinal lymphadenopathy was reported to be up to 29% [26] . Of note, the frequency can differ according to the threshold value of enlargement. All included studies used 10 mm in short axis. In a recent study investigating 650 patients employed the threshold value of 10 mm, the frequency of mediastinal lymphadenopathy was 8.6% [26] . The identified odds ratio for 30-days mortality in this mentioned study was 2.38 [95% 1.13-4.98] [26] .

The only investigated CT finding not associated with mortality was pericardial effusion. In a small study based on 54 patients, a significant difference was identified between a severe and a critical patient group in regard of pericardial effusions (n=1, 2.6%, n=5, 33.3%, p<0.01) [46] . The presumed reason for the pericardial effusion was inflammatory effusion [46] . Contrary to the present results, the authors used echocardiography, which might be more sensitive for detection of pericardial effusion in comparison to CT. According to Wang et al., cardiac injury caused by COVID-19 infection may also provoke pericardial effusion [3] .

However, the present data can lead to the assumption that pericardial effusion is not a significant predictor for mortality.

Notably, the investigated studies included only patients of the first wave of the pandemic, which has a relevant impact on the results [47] , as the mortality rates are declining since then. Due to less experience in care of COVID-19 patients and less knowledge of the disease in general, the course of COVID-19 patients might be worse than in recent days of the pandemic. Moreover, the possible effect of the current vaccination campaigns on COVID-19 mortality cannot be addressed by the present analysis.

The present analysis could show substantial differences between different origins of the investigated studies. This could explain the heterogeneity in this meta analysis.

Interestingly, for pleural effusion, the pooled odds ratio was higher for studies from China compared to those in the studies performed in USA, Europe and Iran.

Furthermore, for coronary calcifications, no relevant difference between the studies from USA and Europe were identified. Finally, for mediastinal lymphadenopathy, however, the pooled odds ratio was low in the sub-analysis for studies from Iran, whereas in the sub-analysis for studies from Europe, it was high. These findings are difficult to explain. Presumably, different virus subtypes may play a role. Another explanation may be the different beginning of the pandemic throughout the countries.

China was the origin of COVID-19 and had less experience with this disease compared to the other world regions. There were also differences according to gender ratios and mean ages of the investigated patients. 

Michigan: an observational study in 32 hospitals

Reggio Emilia COVID-19 Working Group. The value of computed tomography in assessing the risk of death in COVID-19 patients presenting to the emergency room

Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China

Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China

High risk groups for severe COVID-19 in a whole of population cohort in Australia

Risk factors of critical & mortal COVID-19 cases: A systematic literature review and meta-analysis

Chest CT in COVID-19: What the Radiologist Needs to Know

Coronavirus Disease 2019 (COVID-19): A Systematic Review of Imaging Findings

The Clinical and Chest CT Features Associated With Severe and Critical COVID-19 Pneumonia

Vaccine confidence in the time of COVID-19

Recent updates on COVID-19: A holistic review

Impact of vaccination by priority group on UK deaths, hospital admissions and intensive care admissions from COVID-19

The PRISMA 2020 statement: an updated guideline for reporting systematic reviews

Factors Predicting Outcome in Intensive Care Unit

Patients: Using Clinical, Laboratory, and Radiologic Characteristics

Clinical and Radiologic Characteristics of COVID-19 in Patients With CKD

Computed tomographic features of coronavirus disease-2019 (COVID-19) pneumonia in three groups of Iranian patients: A single center study

Adverse Initial CT Findings Associated with Poor Prognosis of Coronavirus Disease

The Association of CT-measured Cardiac Indices with Lung Involvement and Clinical Outcome in Patients with COVID-19

Coronary and total thoracic calcium scores predict mortality and provides pathophysiologic insights in COVID-19 patients

Epicardial adipose tissue is associated with extent of pneumonia and adverse outcomes in patients with COVID-19

Coronary artery calcification in COVID-19 patients: an imaging biomarker for adverse clinical outcomes

Predictive value of initial CT scan for various adverse outcomes in patients with COVID-19 pneumonia

Can CT performed in the early disease phase predict outcome of patients with COVID 19 pneumonia?

Chest Computed Tomography (CT) as a Predictor of Clinical Course in Coronavirus Disease

Mediastinal lymphadenopathy may predict 30-day mortality in patients with COVID-19

Imaging data in COVID-19 patients: focused on echocardiographic findings

CT-derived Chest Muscle Metrics for Outcome Prediction in Patients with COVID-19

Impact of clinical and subclinical coronary artery disease as assessed by coronary artery calcium in COVID-19

Coronary artery calcification and epicardial adipose tissue as independent predictors of mortality in COVID-19

CT Features and Shortterm Prognosis of COVID-19 Pneumonia: A Single-Center Study from Kashan

Predictive value of CT in the short-term mortality of Coronavirus Disease 2019 (COVID-19) pneumonia in nonelderly patients: A case-control study

Pleural effusion as an indicator for the poor prognosis of COVID-19 patients

Clinical characteristics of COVID-19 complicated with pleural effusion

Coronary calcium scoring assessed on native screening chest CT imaging as predictor for outcome in COVID-19: An analysis of a hospitalized German cohort

The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses -2008

Systematic reviews of diagnostic test accuracy

Meta-DiSc: a software for meta-analysis of test accuracy data

Meta-analysis in clinical trials

Development and validation of a simple-to-use nomogram to predict the deterioration and survival of patients with COVID-19

Predictors of COVID-19 severity: A literature review

Meta-analysis: COVID-19 diagnosis in chest CT-master key for radiologists

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in

China: a retrospective cohort study

Frequency and Distribution of Chest Radiographic Findings in Patients Positive for COVID-19

Prognostic factors for severity and mortality in patients infected with COVID-19: A systematic review

Cardiovascular manifestations in severe and critical patients with COVID-19

Changes in COVID-19 in-hospital mortality in hospitalised adults in England over the first seven months of the pandemic: An observational study using administrative data

Abbreviations: CT = Computed tomography, Abbreviation: NOS = Newcastle-Ottawa Scale