Deep Learning With Chest Radiographs for Making Prognoses in Patients With COVID-19: Retrospective Cohort Study

BACKGROUND

An artificial intelligence (AI) model using chest radiography (CXR) may provide good performance in making prognoses for COVID-19.

OBJECTIVE

We aimed to develop and validate a prediction model using CXR based on an AI model and clinical variables to predict clinical outcomes in patients with COVID-19.

METHODS

This retrospective longitudinal study included patients hospitalized for COVID-19 at multiple COVID-19 medical centers between February 2020 and October 2020. Patients at Boramae Medical Center were randomly classified into training, validation, and internal testing sets (at a ratio of 8:1:1, respectively). An AI model using initial CXR images as input, a logistic regression model using clinical information, and a combined model using the output of the AI model (as CXR score) and clinical information were developed and trained to predict hospital length of stay (LOS) ≤2 weeks, need for oxygen supplementation, and acute respiratory distress syndrome (ARDS). The models were externally validated in the Korean Imaging Cohort of COVID-19 data set for discrimination and calibration.

RESULTS

The AI model using CXR and the logistic regression model using clinical variables were suboptimal to predict hospital LOS ≤2 weeks or the need for oxygen supplementation but performed acceptably in the prediction of ARDS (AI model area under the curve [AUC] 0.782, 95% CI 0.720-0.845; logistic regression model AUC 0.878, 95% CI 0.838-0.919). The combined model performed better in predicting the need for oxygen supplementation (AUC 0.704, 95% CI 0.646-0.762) and ARDS (AUC 0.890, 95% CI 0.853-0.928) compared to the CXR score alone. Both the AI and combined models showed good calibration for predicting ARDS (P=.079 and P=.859).

CONCLUSIONS

The combined prediction model, comprising the CXR score and clinical information, was externally validated as having acceptable performance in predicting severe illness and excellent performance in predicting ARDS in patients with COVID-19.

Impact of Bilateral Infiltrates on Inflammatory Biomarker Levels and Clinical Outcomes of Children With Oxygenation Defect

OBJECTIVES

The 2015 definition for pediatric acute respiratory distress syndrome did not require the presence of bilateral infiltrates. We tested the hypothesis that pediatric patients meeting oxygenation criteria for pediatric acute respiratory distress syndrome but without bilateral infiltrates would have different inflammatory biomarker levels and clinical outcomes than those with bilateral infiltrates.

DESIGN

Secondary analysis of a prospective cohort study.

SETTING

Twenty-two PICUs.

PATIENTS

Four-hundred forty-six patients age 2 weeks to 17 years intubated for respiratory failure with oxygenation index greater than or equal to 4 or oxygenation saturation index greater than or equal to 5 on the day of intubation or the day after.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Patients with bilateral infiltrates, either on the day of intubation or within the following 2 days, were compared with children who never developed bilateral infiltrates. Two analyses were performed to test 1) whether bilateral infiltrates are associated with elevated interleukin-1 receptor antagonist or interleukin-8 and 2) whether bilateral infiltrates are associated with worse clinical outcomes. Patients with bilateral infiltrates more often had a primary diagnosis of pneumonia (41% vs 28%; p = 0.02) and less often asthma (8% vs 23%; p < 0.01). After controlling for age, gender, and primary diagnosis, interleukin-1 receptor antagonist was higher on study days 1 and 2 in patients with bilateral infiltrates. There was no difference in interleukin-8 levels. After adjusting for age, gender, Pediatric Risk of Mortality score, and severity of oxygenation defect, presence of bilateral infiltrates was associated with longer duration of mechanical ventilation in survivors (hazard ratio, 0.64; 95% CI, 0.49-0.82; p < 0.01); this association was independent of primary diagnosis. Overall mortality was 9%; mortality was higher in those without bilateral infiltrates (14% vs 8%; p = 0.04).

CONCLUSIONS

Children meeting pediatric acute respiratory distress syndrome oxygenation criteria with bilateral infiltrates on chest radiograph experience a more intense early inflammatory response. Bilateral infiltrates are associated with longer time on the ventilator independent of oxygenation defect severity.

Giant coronary artery aneurysms: review and update

Giant coronary artery aneurysms are rare, with a reported prevalence of 0.02% to 0.2%. Causative factors include atherosclerosis, Takayasu arteritis, congenital disorders, Kawasaki disease, and percutaneous coronary intervention. Most giant coronary artery aneurysms are asymptomatic, but some patients present with angina pectoris, sudden death, fistula formation, pericardial tamponade, compression of surrounding structures, or congestive heart failure. Clinical sequelae include thrombus formation, embolization, fistula formation, and rupture. Surgical correction is generally accepted as the preferred treatment for giant coronary artery aneurysms. We present an illustrative case of a giant 70 × 40-mm coronary artery aneurysm in a 56-year-old man who declined surgery and died one month later. In addition, we provide a review of the medical literature on giant coronary artery aneurysms.

Rapid viral diagnosis for acute febrile respiratory illness in children in the Emergency Department

BACKGROUND

Pediatric acute respiratory infections (ARIs) represent a significant burden on pediatric Emergency Departments (EDs) and families. Most of these illnesses are due to viruses. However, investigations (radiography, blood, and urine testing) to rule out bacterial infections and antibiotics are often ordered because of diagnostic uncertainties. This results in prolonged ED visits and unnecessary antibiotic use. The risk of concurrent bacterial infection has been reported to be negligible in children over three months of age with a confirmed viral infection. Rapid viral testing in the ED may alleviate the need for precautionary testing and antibiotic use.

OBJECTIVES

To determine if the use of a rapid viral detection test for children with an acute respiratory infection (ARI) in Emergency Departments (EDs) changes patient management and resource use in the ED, compared to not using a rapid viral detection test. We hypothesized that rapid viral testing reduces antibiotic use in the ED as well as reduces the rate of ancillary testing and length of ED visits.

SEARCH METHODS

We searched CENTRAL (2014, Issue 6), MEDLINE (1950 to July week 1, 2014), MEDLINE In-Process & Other Non-Indexed Citations (15 July 2014), EMBASE.com (1988 to July 2014), HealthStar (1966 to 2009), BIOSIS Previews (1969 to July 2014), CAB Abstracts (1973 to July 2014), CBCA Reference (1970 to 2007) and ProQuest Dissertations and Theses (1861 to 2009).

SELECTION CRITERIA

Randomized controlled trials (RCTs) of rapid viral testing for children with ARIs in the ED.

DATA COLLECTION AND ANALYSIS

Two review authors used the inclusion criteria to select trials, evaluate their quality, and extract data. We obtained missing data from trial authors. We expressed differences in rate of investigations and antibiotic use as risk ratios (RRs), and expressed difference in ED length of visits as mean differences (MDs), with 95% confidence intervals (CIs).

MAIN RESULTS

No new trials were identified in this 2014 update. We included four trials (three RCTs and one quazi-RCT), with 759 children in the rapid viral testing group and 829 in the control group. Three out of the four studies were comparable in terms of young age of participants, with one study increasing the age of inclusion up to five years of age. All studies included either fever or respiratory symptoms as inclusion criteria (two required both, one required fever or respiratory symptoms, and one required only fever). All studies were comparable in terms of exclusion criteria, intervention, and outcome data. In terms of risk of bias, one study failed to utilize a random sequence generator, one study did not comment on completeness of outcome data, and only one of four studies included allocation concealment as part of the study design. None of the studies definitively blinded participants.Rapid viral testing resulted in a trend toward decreased antibiotic use in the ED, but this was not statistically significant. We found lower rates of chest radiography (RR 0.77, 95% CI 0.65 to 0.91) in the rapid viral testing group, but no effect on length of ED visits, or blood or urine testing in the ED. No study made mention of any adverse effects related to viral testing.

AUTHORS' CONCLUSIONS

There is insufficient evidence to support routine rapid viral testing to reduce antibiotic use in pediatric EDs. Rapid viral testing may or may not reduce rates of antibiotic use, and other investigations (urine and blood testing); these studies do not provide enough power to resolve this question. However, rapid viral testing does reduce the rate of chest X-rays in the ED. An adequately powered trial with antibiotic use as an outcome is needed.

Screening for lung cancer

BACKGROUND

This is an updated version of the original review published in The Cochrane Library in 1999 and updated in 2004 and 2010. Population-based screening for lung cancer has not been adopted in the majority of countries. However it is not clear whether sputum examinations, chest radiography or newer methods such as computed tomography (CT) are effective in reducing mortality from lung cancer.

OBJECTIVES

To determine whether screening for lung cancer, using regular sputum examinations, chest radiography or CT scanning of the chest, reduces lung cancer mortality.

SEARCH METHODS

We searched electronic databases: the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2012, Issue 5), MEDLINE (1966 to 2012), PREMEDLINE and EMBASE (to 2012) and bibliographies. We handsearched the journal Lung Cancer (to 2000) and contacted experts in the field to identify published and unpublished trials.

SELECTION CRITERIA

Controlled trials of screening for lung cancer using sputum examinations, chest radiography or chest CT.

DATA COLLECTION AND ANALYSIS

We performed an intention-to-screen analysis. Where there was significant statistical heterogeneity, we reported risk ratios (RRs) using the random-effects model. For other outcomes we used the fixed-effect model.

MAIN RESULTS

We included nine trials in the review (eight randomised controlled studies and one controlled trial) with a total of 453,965 subjects. In one large study that included both smokers and non-smokers comparing annual chest x-ray screening with usual care there was no reduction in lung cancer mortality (RR 0.99, 95% CI 0.91 to 1.07). In a meta-analysis of studies comparing different frequencies of chest x-ray screening, frequent screening with chest x-rays was associated with an 11% relative increase in mortality from lung cancer compared with less frequent screening (RR 1.11, 95% CI 1.00 to 1.23); however several of the trials included in this meta-analysis had potential methodological weaknesses. We observed a non-statistically significant trend to reduced mortality from lung cancer when screening with chest x-ray and sputum cytology was compared with chest x-ray alone (RR 0.88, 95% CI 0.74 to 1.03). There was one large methodologically rigorous trial in high-risk smokers and ex-smokers (those aged 55 to 74 years with ≥ 30 pack-years of smoking and who quit ≤ 15 years prior to entry if ex-smokers) comparing annual low-dose CT screening with annual chest x-ray screening; in this study the relative risk of death from lung cancer was significantly reduced in the low-dose CT group (RR 0.80, 95% CI 0.70 to 0.92).

AUTHORS' CONCLUSIONS

The current evidence does not support screening for lung cancer with chest radiography or sputum cytology. Annual low-dose CT screening is associated with a reduction in lung cancer mortality in high-risk smokers but further data are required on the cost effectiveness of screening and the relative harms and benefits of screening across a range of different risk groups and settings.

Two scimitar veins in an adult: repair through a right thoracotomy without cardiopulmonary bypass

Scimitar syndrome is a rare but serious congenital condition that consists of anomalous pulmonary venous drainage of the right lung to the inferior vena cava. The appearance on chest radiography resembles a curved Turkish sword, or scimitar. Scimitar syndrome is associated with other anomalies, including hypoplasia of the right lung, dextroposition of the heart, anomalous systemic arterial supply to the right lung, and atrial septal defect. Clinical manifestations in infants include severe tachypnea and cyanosis. Adult patients may present with dyspnea and increased fatigue. Few adult cases have been reported in the medical literature. Classical surgical repair involves diversion of the scimitar venous flow into the left atrium with a baffle, requiring the use of cardiopulmonary bypass and deep hypothermic circulatory arrest. Herein, we report the case of a 42-year-old woman with 2 scimitar veins who underwent corrective surgery at our center without the use of cardiopulmonary bypass. We also comment on the importance of a patient's lung hypoplasia in the decision to repair the defect through a right thoracotomy.

Large mediastinal teratoma originating from the aortic adventitia

Many mediastinal tumors do not cause local symptoms, so they are discovered incidentally upon thoracic imaging. We present the case of a patient who had a large, mature, cystic, mediastinal teratoma. The primary tumor arose from the ascending aortic adventitia and was in a highly unusual location-the medial mediastinum. The teratoma, which was diagnosed in our patient on thoracic computed tomography for follow-up evaluation of empyema, was resected completely. To our knowledge, such a teratoma, arising from the adventitia of the ascending aorta in the medial mediastinum, has not previously been reported in the English medical literature. We review diagnostic methods and therapeutic approaches to such mediastinal tumors. We conclude that surgical resection is the method of choice for treating these tumors, because it enables radical therapy and tissue diagnosis after extirpation.