External validation of the CHEDDAR score for suspected pulmonary embolism in patients with SARS-CoV-2 infection in an independent cohort.

The accuracy of the classic scores that help stratify the pretest clinical probability of pulmonary embolism (PE) in SARS-CoV-2 infection (COVID-19) is low. Therefore, to estimate the risk of PE in these patients, a new set of guidelines must be established. The recently published CHEDDAR score proposes a new diagnostic strategy to reduce the use of computed tomography pulmonary angiography (CTPA) in non-critically ill SARS-COV-2 patients with suspected PE. According to the nomogram, patients are segregated into low-risk (< 182 points) or high-risk (≥ 182 points) based on the best cut-off value to discard PE in the original cohort. We aimed to externally validate this diagnostic strategy in an independent cohort. We analyzed data from two retrospective cohorts of hospitalized non-critically ill COVID-19 patients who underwent a CTPA due to suspicion for PE. CHEDDAR score was applied. As per the CHEDDAR nomogram, patients were classified as having a low or high clinical pre-test probability. Of the 270 patients included, 69 (25.5%) had PE. Applying the CHEDDAR score, 182 (67.4%) patients could have had PE excluded without imaging. Among 58 patients classified as having high clinical pre-test probability, 39 (67.2%) had PE. Sensitivity, specificity, positive and negative predictive values, and AUC were 56%, 90%, 67%, 85%, and 0.783 (95% CI 0.71-0.85), respectively. We provide external validation of the CHEDDAR score in an independent cohort. Even though the CHEDDAR score showed good discrimination capacity, caution is required in patients classified as having low clinical pre-test probability with a D-dimer value > 3000 ng/mL, and a RALE score ≥ 4.

Baricitinib or imatinib in hospitalized COVID-19 patients: Results from COVINIB, an exploratory randomized clinical trial.

Baricitinib and imatinib are considered therapies for coronavirus disease 2019 (COVID-19), but their ultimate clinical impact remains to be elucidated, so our objective is to determine whether these kinase inhibitors provide benefit when added to standard care in hospitalized COVID-19 patients. Phase-2, open-label, randomized trial with a pick-the-winner design conducted from September 2020 to June 2021 in a single Spanish center. Hospitalized adults with COVID-19 pneumonia and a symptom duration ≤10 days were assigned to 3 arms: imatinib (400 mg qd, 7 days) plus standard-care, baricitinib (4 mg qd, 7 days) plus standard-care, or standard-care alone. Primary outcome was time to clinical improvement (discharge alive or a reduction of 2 points in an ordinal scale of clinical status) compared on a day-by-day basis to identify differences ≥15% between the most and least favorable groups. Secondary outcomes included oxygenation and ventilatory support requirements, additional therapies administered, all-cause mortality, and safety. One hundred and sixty-five patients analyzed. Predefined criteria for selection of the most advantageous arm were met for baricitinib, but not for imatinib. However, no statistically significant differences were observed in formal analysis, but a trend toward better results in patients receiving baricitinib was found compared to standard care alone (hazard ratio [HR] for clinical improvement: 1.41, 95% confidence intervals [CI]: 0.96-2.06; HR for discontinuing oxygen: 1.46, 95% CI: 0.94-2.28). No differences were found regarding additional therapies administered or safety. Baricitinib plus standard care showed better results for hospitalized COVID-19 patients, being the most advantageous therapeutic strategy among those proposed in this exploratory clinical trial.

External validation of the RIETE and SOME scores for occult cancer in patients with venous thromboembolism: a multicentre cohort study.

INTRODUCTION: Venous thromboembolism (VTE) may be the first sign of an undiagnosed cancer. The RIETE and SOME scores aim to identify patients with acute VTE at high risk of occult cancer. In the present study, we evaluated the performance of both scores. METHODS: The scores were evaluated in a retrospective cohort from two centers. The area under the receiver-operating characteristics curve (AUC) evaluated the discriminatory performance. RESULTS: The RIETE score was applied to 815 patients with provoked and unprovoked VTE, of whom 56 (6.9%) were diagnosed with cancer. Of the 203 patients classified as high-risk, 18 were diagnosed with cancer, representing 32.1% (18/56) of the total cancer diagnoses. In the group of 612 low-risk patients, 67.9% of the cancer cases were diagnosed (38/56). Sensitivity, specificity, negative and positive predictive values, and AUC were 32%, 76%, 94%, 9%, and 0.430 (95% confidence interval [CI], 0.38‒0.47), respectively. The SOME score could be calculated in 418 patients with unprovoked VTE, of whom 33 (7.9%) were diagnosed with cancer. Of the 45 patients classified as high-risk, three were diagnosed with cancer, representing 9.1% (3/33) of the total cancer diagnoses. In the group of 373 low-risk patients, 90.9% of the cancer cases were diagnosed (30/33). Sensitivity, specificity, negative and positive predictive values, and AUC were 33%, 88%, 94%, 20%, and 0.351 (95% CI, 0.27‒0.43), respectively. CONCLUSIONS: The performance of both scores was poor. Our results highlight the need to develop new models to identify high-risk patients who may benefit from an extensive cancer screening strategy.

Prediction of Venous Thromboembolism in Patients With Cancer Using Machine Learning Approaches: A Systematic Review and Meta-Analysis.

PURPOSE: Recent studies have suggested that machine learning (ML) could be used to predict venous thromboembolism (VTE) in cancer patients with high accuracy. METHODS: We aimed to evaluate the performance of ML in predicting VTE events in patients with cancer. PubMed, Web of Science, and EMBASE to identify studies were searched. RESULTS: Seven studies involving 12,249 patients with cancer were included. The combined results of the different ML models demonstrated good accuracy in the prediction of VTE. In the training set, the global pooled sensitivity was 0.87, the global pooled specificity was 0.87, and the AUC was 0.91, and in the test set 0.65, 0.84, and 0.80, respectively. CONCLUSION: The prediction ML models showed good performance to predict VTE. External validation to determine the result's reproducibility is necessary.