Circulating brain injury biomarkers for prediction of outcome after elective neurosurgery - a scoping review.

OBJECTIVE: There is a need for refined methods to detect and quantify brain injuries that may be undetectable by magnetic resonance imaging and neurologic examination. This review evaluates the potential efficacy of circulating brain injury biomarkers for predicting outcomes following elective neurosurgical procedures. METHODS: A comprehensive search was conducted using the Cochrane, PubMed, and Scopus databases. RESULTS: Analysis of 23 relevant studies revealed that specific biomarkers, including glial fibrillary acidic protein, neurofilament light chain, neuron-specific enolase, S100B, and tau, are significantly associated with the extent of brain injury and could potentially predict postsurgical outcomes. The evaluated studies described intracranial tumor surgeries and miscellaneous neurosurgical interventions, and demonstrated the complex relationship between biomarker levels and patient outcomes. CONCLUSIONS: Circulating brain injury biomarkers show promise for providing objective insights into the extent of perioperative brain injury and improving prognostication of postsurgical outcomes. However, the heterogeneity in study designs and outcomes along with the lack of standardized biomarker thresholds underscore the need for further research.

Factors for increasing positive predictive value of pneumothorax detection on chest radiographs using artificial intelligence.

This study evaluated the positive predictive value (PPV) of artificial intelligence (AI) in detecting pneumothorax on chest radiographs (CXRs) and its affecting factors. Patients determined to have pneumothorax on CXR by a commercial AI software from March to December 2021 were included retrospectively. The PPV was evaluated according to the true-positive (TP) and false-positive (FP) diagnosis determined by radiologists. To know the factors that might influence the results, logistic regression with generalized estimating equation was used. Among a total of 87,658 CXRs, 308 CXRs with 331 pneumothoraces from 283 patients were finally included. The overall PPV of AI about pneumothorax was 41.1% (TF:FP = 136:195). The PA view (odds ratio [OR], 29.837; 95% confidence interval [CI], 15.062-59.107), high abnormality score (OR, 1.081; 95% CI, 1.066-1.097), large amount of pneumothorax (OR, 1.005; 95% CI, 1.003-1.007), presence of ipsilateral atelectasis (OR, 3.508; 95% CI, 1.509-8.156) and a small amount of ipsilateral pleural effusion (OR, 5.277; 95% CI, 2.55-10.919) had significant effects on the increasing PPV. Therefore, PPV for pneumothorax diagnosis using AI can vary based on patients' factors, image-acquisition protocols, and the presence of concurrent lesions on CXR.

Predictive value of dynamic arterial elastance for vasopressor withdrawal: a systematic review and meta-analysis.

BACKGROUND: Dynamic arterial elastance (Eadyn) has been investigated for its ability to predict hypotension during the weaning of vasopressors. Our study focused on assessing Eadyn's performance in the context of critically ill adult patients admitted to the intensive care unit, regardless of diagnosis. MAIN BODY: Our study was conducted in accordance with the Preferred Reported Items for Systematic Reviews and Meta-Analysis checklist. The protocol was registered in PROSPERO (CRD42023421462) on May 26, 2023. We included prospective observational studies from the MEDLINE and Embase databases through May 2023. Five studies involving 183 patients were included in the quantitative analysis. We extracted data related to patient clinical characteristics, and information about Eadyn measurement methods, results, and norepinephrine dose. Most patients (76%) were diagnosed with septic shock, while the remaining patients required norepinephrine for other reasons. The average pressure responsiveness rate was 36.20%. The synthesized results yielded an area under the curve of 0.85, with a sensitivity of 0.87 (95% CI 0.74-0.93), specificity of 0.76 (95% CI 0.68-0.83), and diagnostic odds ratio of 19.07 (95% CI 8.47-42.92). Subgroup analyses indicated no variations in the Eadyn based on norepinephrine dosage, the Eadyn measurement device, or the Eadyn diagnostic cutoff to predict cessation of vasopressor support. CONCLUSIONS: Eadyn, evaluated through subgroup analyses, demonstrated good predictive ability for the discontinuation of vasopressor support in critically ill patients.

Innovation through Artificial Intelligence in Triage Systems for Resource Optimization in Future Pandemics.

Artificial intelligence (AI) systems are already being used in various healthcare areas. Similarly, they can offer many advantages in hospital emergency services. The objective of this work is to demonstrate that through the novel use of AI, a trained system can be developed to detect patients at potential risk of infection in a new pandemic more quickly than standardized triage systems. This identification would occur in the emergency department, thus allowing for the early implementation of organizational preventive measures to block the chain of transmission. MATERIALS AND METHODS: In this study, we propose the use of a machine learning system in emergency department triage during pandemics to detect patients at the highest risk of death and infection using the COVID-19 era as an example, where rapid decision making and comprehensive support have becoming increasingly crucial. All patients who consecutively presented to the emergency department were included, and more than 89 variables were automatically analyzed using the extreme gradient boosting (XGB) algorithm. RESULTS: The XGB system demonstrated the highest balanced accuracy at 91.61%. Additionally, it obtained results more quickly than traditional triage systems. The variables that most influenced mortality prediction were procalcitonin level, age, and oxygen saturation, followed by lactate dehydrogenase (LDH) level, C-reactive protein, the presence of interstitial infiltrates on chest X-ray, and D-dimer. Our system also identified the importance of oxygen therapy in these patients. CONCLUSIONS: These results highlight that XGB is a useful and novel tool in triage systems for guiding the care pathway in future pandemics, thus following the example set by the well-known COVID-19 pandemic.

Models for predicting the risk of illness in leprosy contacts in Brazil: Leprosy prediction models in Brazilian contacts.

OBJECTIVE: This study aims to develop and validate predictive models that assess the risk of leprosy development among contacts, contributing to an enhanced understanding of disease occurrence in this population. METHODS: A cohort of 600 contacts of people with leprosy treated at the National Reference Center for Leprosy and Health Dermatology at the Federal University of Uberlândia (CREDESH/HC-UFU) was followed up between 2002 and 2022. The database was divided into two parts: two-third to construct the disease risk score and one-third to validate this score. Multivariate logistic regression models were used to construct the disease score. RESULTS: Of the four models constructed, model 3, which included the variables anti-phenolic glycolipid I immunoglobulin M positive, absence of Bacillus Calmette-Guérin vaccine scar and age ≥60 years, was considered the best for identifying a higher risk of illness, with a specificity of 89.2%, a positive predictive value of 60% and an accuracy of 78%. CONCLUSIONS: Risk prediction models can contribute to the management of leprosy contacts and the systematisation of contact surveillance protocols.

Improving Cancer Probability Estimation in Nondiagnostic Bronchoscopies: A Meta-analysis.

BACKGROUND: In patients with peripheral pulmonary lesions (PPLs), nondiagnostic bronchoscopy results are not uncommon. The conventional approach to estimate the probability of cancer (pCA) after bronchoscopies relies on dichotomous test assumptions, using prevalence, sensitivity, and specificity to determine negative predictive value. However, bronchoscopy is a multidisease test, raising concerns about the accuracy of dichotomous methods. RESEARCH QUESTION: By how much does calculating pCA using a dichotomous approach (pCAdichotomous) underestimate the true pCA when applied to multidisease tests like bronchoscopy for the diagnosis of PPL? METHODS: In this meta-analysis of cohort studies involving radial endobronchial ultrasound for PPL, Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines were followed, constructing 2 × 2 contingency tables for calculating pCAdichotomous. For the multidisease test approach, 3 × 3 contingency tables for calculating probability of malignancy for a test that can have multiple test different categories of results and can diagnose multiple diseases (pCAmultidisease) using the likelihood ratio (LR) method for nondiagnostic results (LR(T0)) was used. Observed malignancy rates in patients with nondiagnostic results were compared with pCAdichotomous and pCAmultidisease. RESULTS: In 46 studies (7,506 patients), malignancy was the underlying diagnosis in 76%, another specific disease in 13%, and nonspecific fibrosis or scar in 10%. The percentage of patients with nondiagnostic results who had malignancy matched pCAmultidisease across all studies. In contrast, pCAdichotomous consistently underestimated cancer risk (median difference, 0.12; interquartile range, 0.06-0.23), particularly in studies with a higher prevalence of nonmalignant disease. The pooled LR(T0) was 0.46 (95% CI, 0.40-0.52; I2 = 76%; P < .001) and correlated with the prevalence of nonmalignant diseases (P = .001). INTERPRETATION: Conventional dichotomous methods for estimating pCA after nondiagnostic bronchoscopies underestimate the likelihood of malignancy. Physicians should opt for the multidisease test approach when interpreting bronchoscopy results.

Is Streptococcus pyogenes a pathogen or passenger in uncomplicated acute sore throat? A systematic review and meta-analysis.

OBJECTIVES: The objective was to estimate the probability that finding a Streptococcus pyogenes (Group A Streptococcus) in a throat swab in a patient with a sore throat reflects the aetiology. We also investigated to what extent this is influenced by age, carrier rates of S. pyogenes and climate zone. METHODS: We conducted a comprehensive search of Medline and Scopus up until October 2023 for case-control studies reporting the prevalence of S. pyogenes in patients with a sore throat and healthy controls. We only included studies with separate data for children and adults. We used the positive and negative etiologic predictive values (P-EPV and N-EPV) to estimate the probability of a link between a sore throat and a finding of S. pyogenes. RESULTS: We included 15 studies in our meta-analysis. The overall P-EPV for children and adults were 63% (49-74%) and 92% (87-95%), respectively. The P-EPV rose to 83% (64-93%) for children and 94% (90-97%) for adults when only patients with 3-4 Centor criteria were included. The overall N-EPV was 97% (96-98%) for children and 96% (95-97%) for adults. CONCLUSION: Detecting S. pyogenes in adult patients with an uncomplicated acute sore throat is useful to rule in S. pyogenes as the likely aetiologic agent. The P-EPV significantly increased for children when those with 3-4 Centor criteria were selected. A negative throat swab is always useful for both children and adults to rule out S. pyogenes as the cause of sore throat.

Automated Prediction of Proximal Middle Cerebral Artery Occlusions in Noncontrast Brain Computed Tomography.

BACKGROUND: Early identification of large vessel occlusion (LVO) in patients with ischemic stroke is crucial for timely interventions. We propose a machine learning-based algorithm (JLK-CTL) that uses handcrafted features from noncontrast computed tomography to predict LVO. METHODS: We included patients with ischemic stroke who underwent concurrent noncontrast computed tomography and computed tomography angiography in seven hospitals. Patients from 5 of these hospitals, admitted between May 2011 and March 2015, were randomly divided into training and internal validation (9:1 ratio). Those from the remaining 2 hospitals, admitted between March 2021 and September 2021, were designated for external validation. From each noncontrast computed tomography scan, we extracted differences in volume, tissue density, and Hounsfield unit distribution between bihemispheric regions (striatocapsular, insula, M1-M3, and M4-M6, modified from the Alberta Stroke Program Early Computed Tomography Score). A deep learning algorithm was used to incorporate clot signs as an additional feature. Machine learning models, including ExtraTrees, random forest, extreme gradient boosting, support vector machine, and multilayer perceptron, as well as a deep learning model, were trained and evaluated. Additionally, we assessed the models' performance after incorporating the National Institutes of Health Stroke Scale scores as an additional feature. RESULTS: Among 2919 patients, 83 were excluded. Across the training (n=2463), internal validation (n=275), and external validation (n=95) datasets, the mean ages were 68.5±12.4, 67.6±13.8, and 67.9±13.6 years, respectively. The proportions of men were 57%, 53%, and 59%, with LVO prevalences of 17.0%, 16.4%, and 26.3%, respectively. In the external validation, the ExtraTrees model achieved a robust area under the curve of 0.888 (95% CI, 0.850-0.925), with a sensitivity of 80.1% (95% CI, 72.0-88.1) and a specificity of 88.6% (95% CI, 84.7-92.5). Adding the National Institutes of Health Stroke Scale score to the ExtraTrees model increased sensitivity (from 80.1% to 92.1%) while maintaining specificity. CONCLUSIONS: Our algorithm provides reliable predictions of LVO using noncontrast computed tomography. By enabling early LVO identification, our algorithm has the potential to expedite the stroke workflow.

Radiological Cutoff Values for Diagnosis of Lymph Node Metastasis in Colorectal Cancer With Multilevel Analysis.

Background/Aim: A cutoff value for lymph node diameter in colorectal cancer lymph node metastases has not been established. This prospective study aimed to investigate the direct association between swollen lymph nodes identified on preoperative computed tomography (CT) and pathological findings and proposed a cutoff value. Patients and Methods: We enrolled patients scheduled to undergo curative surgery with lymph node dissection for colorectal adenocarcinoma who underwent preoperative contrast-enhanced CT and had swollen lymph nodes ≥7 mm in diameter. Two gastrointestinal surgeons intraoperatively identified the target lymph nodes to assess the association between lymph node diameter and pathological findings. The diagnostic performance for lymph node metastasis was determined using multi-level logistic modelling. Results: A total of 109 patients were enrolled, and 225 swollen lymph nodes were pathologically evaluated. Using a cutoff value of ≥9 mm for the short diameter, the positive and negative predictive values, sensitivity, and specificity were 100.0% (99.6%-100.0%), 99.9% (99.1%-100.0%), 62.0% (45.6%-76.0%), and 84.9% (67.0%-94.0%), respectively. Conclusion: The cutoff value for improving the positive predictive value for the preoperative lymph node metastasis diagnosis in colorectal cancer patients should be at least 9 mm in diameter.

Atrial Fibrillation Screening During Routine Automated Office, Home, and Ambulatory Blood Pressure Measurement: A Diagnostic Test Accuracy Systematic Review and Meta-Analysis.

BACKGROUND: Atrial fibrillation (AF) is often asymptomatic and undiagnosed. As AF and hypertension often coexist, opportunistic AF detection during routine automated blood pressure (BP) measurement appears to be an attractive screening method. METHODS: A systematic literature search was conducted to identify studies assessing the diagnostic test accuracy of office, home, or 24-hour ambulatory BP measuring devices with AF detection algorithms versus reference electrocardiography. Analyses were performed per participant (AF status based on several BP readings; most office/home devices) or per reading (AF status based on individual readings; all ambulatory devices). A meta-analysis stratified by device type (office/home/ambulatory) was conducted to calculate pooled measures of diagnostic accuracy. Sensitivity/meta-regression analyses were also performed. RESULTS: Among 3096 records initially retrieved, 23 diagnostic test accuracy studies were included. Data derived from 11 093 individuals (weighted age 69 years, males 56%, hypertensives 79%, diabetics 24%, and AF prevalence 17%) indicated a pooled sensitivity 0.97 (95% CI, 0.92-0.99), specificity 0.93 (95% CI, 0.90-0.95), and accuracy 0.93 (95% CI, 0.89-0.95), with generally consistent results using office, home, or ambulatory BP devices (slightly lower specificity with the latter). The positive and negative predictive values were 0.70 (95% CI, 0.60-0.80) and 0.99 (95% CI, 0.98-1.00), respectively. Sensitivity analyses indicated lower specificity in studies implementing reading versus participant analyses. Most studies presented a low risk of bias and minor applicability concerns. CONCLUSIONS: There is considerable and consistent evidence suggesting high diagnostic accuracy of AF detection algorithms implemented in automated BP monitors during routine BP measurements in and out of the office. AF diagnosis requires verification (electrocardiography) before treatment is administered.

Association between Initial Fibrinogen Levels and the Need for Massive Transfusion in Emergency Department Patients with Primary Postpartum Hemorrhage: A Retrospective Study from a Single Center in Korea.

BACKGROUND: This study aimed to evaluate the association between initial fibrinogen levels and massive transfusion (MT) in emergency department (ED) patients with primary postpartum hemorrhage (PPH). METHODS: This retrospective study was conducted in the ED of a university-affiliated, tertiary referral center from January 2004 to August 2023. Patients were divided into two groups: the MT group, which included those who received a transfusion of 10 or more units of packed red blood cells within the first 24 h, and the Non-MT group. RESULTS: Out of the 364 patients included in the study, 97 (26.6%) required MT. Fibrinogen, shock index, and lactate were independently associated with MT (odds ratio [OR] 0.987; 95% confidence interval [CI] 0.983-0.991; p < 0.001, OR 7.277; 95% CI 1.856-28.535; p = 0.004, and OR 1.261; 95% CI 1.021-1.557; p = 0.031, respectively). The area under the receiver operating characteristic curve for fibrinogen, shock index, and lactate in predicting MT was 0.871 (95% CI 0.832-0.904; p < 0.001), 0.821 (95% CI 0.778-0.859; p < 0.001), and 0.784 (95% CI 0.738-0.825; p < 0.001), respectively. When the cutoff value of fibrinogen was 400 mg/dL, both the sensitivity and negative predictive values for predicting MT were 100.0%. When the cutoff value of fibrinogen was 100 mg/dL, the specificity and positive predictive values were 91.8% and 70.7%, respectively. CONCLUSION: The initial fibrinogen levels were independently associated with the need for MT in ED patients with primary PPH.

Validity of Prescription-Defined and Hospital-Diagnosed Hypertension Compared with Self-Reported Hypertension in Denmark.

Purpose: Hypertension is an important risk factor in cardio-epidemiological research, but data quality remains a concern. We validated different registry-based definitions of hypertension. Patients and Methods: The cohort included all first-time responders of the Danish National Health Surveys (2010, 2013, or 2017). Prescription-defined hypertension was defined as ≥1 or ≥2 filled prescriptions of antihypertensive specific drugs in ≥1 or ≥2 different antihypertensive drug classes within 90, 180, or 365 days before survey response. Hospital-diagnosed hypertension was defined from hypertension diagnoses within five years before the survey response. Considering self-reported hypertension as the reference, we calculated the positive predictive value (PPV), the negative predictive value (NVP), the sensitivity, and the specificity of prescription-defined and hospital-diagnosed hypertension. Results: Among 442,490 survey responders, 127,247 (29%) had self-reported hypertension. For prescription-defined hypertension with 365-day lookback, the PPV was highest for ≥2 prescriptions in ≥2 drug classes (94%) and lowest for ≥1 prescription in ≥1 drug class (85%). The NPV was highest for ≥1 prescription in ≥2 drug classes (94%) and lowest for ≥1 prescription in ≥2 drug classes (80%). The sensitivity was highest for ≥1 prescription in ≥1 drug class (79%) and lowest for ≥2 prescriptions in ≥2 drug classes (30%). The specificity was ≥94% for all algorithms. The PPV and specificity did not change noteworthy with length of lookback period, whereas the NPV and the sensitivity generally were higher for longer lookback. The algorithm ≥1 prescription in ≥2 drug classes with 365-day lookback was among the best balanced across all measures of validity (PPV=88%, NPV=94%, sensitivity=75%, specificity=96%). For hospital-diagnosed hypertension, the PPV was 90%, the NPV was 76%, the sensitivity was 22%, and the specificity was 99%. Conclusion: Compared with self-reported hypertension, the algorithms for prescription-defined and hospital-diagnosed hypertension had high predictive values and specificity, but low sensitivity.

Test characteristics for combining non-invasive liver fibrosis staging modalities in individuals with Hepatitis C virus.

Non-invasive methods have largely replaced biopsy to identify advanced fibrosis in hepatitis C virus (HCV). Guidelines vary regarding testing strategy to balance accuracy, costs and loss to follow-up. Although individual test characteristics are well-described, data comparing the accuracy of using two tests together are limited. We calculated combined test characteristics to determine the utility of combined strategies. This study synthesizes empirical data from fibrosis staging trials and the literature to estimate test characteristics for Fibrosis-4 (FIB4), APRI or a commercial serum panel (FibroSure®), followed by transient elastography (TE) or FibroSure®. We simulated two testing strategies: (1) second test only for those with intermediate first test results (staged approach), and (2) second test for all. We summarized empiric data with multinomial distributions and used this to estimate test characteristics of each strategy on a simulated population of 10,000 individuals with 4.2% cirrhosis prevalence. Negative predictive value (NPV) for cirrhosis from a single test ranged from 98.2% (95% CB 97.6-98.8%) for FIB-4 to 99.4% (95% CB 99.0-99.8%) for TE. Using a staged approach with TE second, sensitivity for cirrhosis rose to 93.3-96.9%, NPV to 99.7-99.8%, while PPV dropped to <32%. Using TE as a second test for all minimally changed estimated test characteristics compared with the staged approach. Combining two non-invasive fibrosis tests barely improves NPV and decreases or does not change PPV compared with a single test, challenging the utility of serial testing modalities. These calculated combined test characteristics can inform best methods to identify advanced fibrosis in various populations.

Concordance and generalization of an AI algorithm with real-world clinical data in the pre-omicron and omicron era.

All viruses, including SARS-CoV-2, the virus responsible for COVID-19, continue to evolve, which can lead to new variants. The objective of this study is to assess the agreement between real-world clinical data and an algorithm that utilizes laboratory markers and age to predict the progression of disease severity in COVID-19 patients during the pre-Omicron and Omicron variant periods. The study evaluated the performance of a deep learning (DL) algorithm in predicting disease severity scores for COVID-19 patients using data from the USA, Spain, and Turkey (Ankara City Hospital (ACH) data set). The algorithm was developed and validated using pre-Omicron era data and was tested on both pre-Omicron and Omicron-era data. The predictions were compared to the actual clinical outcomes using a multidisciplinary approach. The concordance index values for all datasets ranged from 0.71 to 0.81. In the ACH cohort, a negative predictive value (NPV) of 0.78 or higher was observed for severe patients in both the pre-Omicron and Omicron eras, which is consistent with the algorithm's performance in the development cohort.

Validation of human immunodeficiency virus diagnosis codes among women enrollees of a U.S. health plan.

BACKGROUND: Efficiently identifying patients with human immunodeficiency virus (HIV) using administrative health care data (e.g., claims) can facilitate research on their quality of care and health outcomes. No prior study has validated the use of only ICD-10-CM HIV diagnosis codes to identify patients with HIV. METHODS: We validated HIV diagnosis codes among women enrolled in a large U.S. integrated health care system during 2010-2020. We examined HIV diagnosis code-based algorithms that varied by type, frequency, and timing of the codes in patients' claims data. We calculated the positive predictive values (PPVs) and 95% confidence intervals (CIs) of the algorithms using a medical record-confirmed diagnosis of HIV as the gold standard. RESULTS: A total of 272 women with ≥ 1 HIV diagnosis code in the administrative claims data were identified and medical records were reviewed for all 272 women. The PPV of an algorithm classifying women as having HIV as of the first HIV diagnosis code during the observation period was 80.5% (95% CI: 75.4-84.8%), and it was 93.9% (95% CI: 90.0-96.3%) as of the second. Little additional increase in PPV was observed when a third code was required. The PPV of an algorithm based on ICD-10-CM-era codes was similar to one based on ICD-9-CM-era codes. CONCLUSION: If the accuracy measure of greatest interest is PPV, our findings suggest that use of ≥ 2 HIV diagnosis codes to identify patients with HIV may perform well. However, health care coding practices may vary across settings, which may impact generalizability of our results.

Diagnostic accuracy of a simple qualitative ultrasound assessment for the diagnosis of high-risk gastric contents in the parturient. A prospective observational cohort study.

STUDY OBJECTIVE: Interpretation of gastric ultrasound relies on the use of a clinical algorithm that combines qualitative analysis of the gastric antrum contents with the calculation of the volume of fluid contents. This reference method may be difficult to apply in the parturient. We therefore aimed to assess the diagnostic accuracy of a simple qualitative assessment in the supine position for the diagnosis of high-risk gastric contents in the parturient. We also assessed the diagnostic accuracy of a composite scale and another clinical algorithm based on a mathematical model different to that used in the reference method. DESIGN: Prospective observational cohort study. SETTING: University hospital, Lyon, France. PATIENTS: Adult women admitted to the delivery room. INTERVENTIONS: Qualitative and quantitative gastric ultrasound examination within the first hour following admission. MEASUREMENTS: With respect to the reference method, the diagnostic accuracy of a simple qualitative assessment for the diagnosis of high-risk gastric contents was assessed. The diagnostic accuracy of a composite scale and another clinical algorithm, and the agreement between each approach were also assessed. MAIN RESULTS: A total of 235 parturients were included and analyzed. The simple qualitative assessment led to conclusive ultrasound assessment in 233 (99%) women, while the reference method led to conclusive assessment in 213 (91%) women (P < 0.05). The sensitivity and the specificity of the simple qualitative assessment were 97% (95%CI: 93 to 99%) and 96% (95%CI: 90 to 99%), respectively. These were not significantly different from those of the composite scale and the clinical algorithm. The four approaches showed almost perfect agreement with each other. CONCLUSIONS: These results suggest that simple qualitative assessment may be useful in clinical practice to help the anesthesiologist in the assessment of gastric contents status and risk of aspiration.

Variáveis Clínicas e Laboratoriais na Admissão Hospitalar são Preditores de Fibrilação Atrial Nova em Pacientes Internados com Pneumonia por COVID-19 / Clinical and Laboratory Data Up on Hospital Admission are Predictors of New-Onset Atrial Fibrillation in Patients Hospitalized Due to COVID-19 Pneumonia

Resumo Fundamento Fibrilação atrial nova (FAN) ocorre em pacientes internados por COVID-19. Há controvérsias quanto ao valor preditivo de dados clínicos e laboratoriais à admissão hospitalar para ocorrência de FAN. Objetivos Analisar, à admissão hospitalar, variáveis com potencial preditivo para ocorrência de FAN em pacientes com pneumonia por COVID-19. Método Estudo observacional, retrospectivo, caso-controle. Foram avaliados prontuários eletrônicos de pacientes consecutivos ≥ 60 anos, hospitalizados com pneumonia por COVID-19 entre 1º de março e 15 de julho de 2020. Comparações feitas pelos testes `t' de Student ou qui-quadrado. Foi empregado modelo de risco proporcional de Cox para identificação de preditores de FAN. Considerou-se o valor de p < 0,05 como estatisticamente significativo. Resultados Entre 667 pacientes internados por COVID-19, 201 (30,1%) foram incluídos. FAN foi documentada em 29 pacientes (14,4%) (grupo 1). Grupo 2 foi composto por 162 pacientes que não apresentaram FAN. Dez pacientes excluídos por estarem em FA na admissão hospitalar. Houve diferenças entre os grupos 1 e 2, respectivamente, no tempo de permanência em UTI (11,1±10,5 dias vs. 4,9±7,5 dias; p=0,004), necessidade de ventilação invasiva (82,9% e 32,7%; p<0,001) e mortalidade hospitalar (75,9% vs. 32,1%; p<0,001). No modelo de Cox, idade > 71 anos (hazard ratio [HR]=6,8; p<0,001), leucometria ≤ 7.720 cels.μL-1 (HR=6,6; p<0,001), natremia ≤ 137 mEq.L-1 (HR=5,0; p=0,001), escore SAPS3 > 55 (HR=5,6; p=0,002) e desorientação (HR=2,5; p=0,04) foram preditores independentes de FAN. Conclusões FAN é uma arritmia comum em idosos hospitalizados com pneumonia por COVID-19. Parâmetros clínicos e laboratoriais avaliados na admissão são preditores de FAN durante internação.

Abstract Background New-onset atrial fibrillation (NOAF) occurs in patients hospitalized due to COVID-19. It is still unknown whether clinical and laboratory data assessed upon hospital admission have predictive value for NOAF. Objectives To analyze, upon hospital admission, variables with predictive potential for the occurrence of NOAF in patients with COVID-19 pneumonia. Methods Observational, retrospective, case-control study. Electronic medical reports of consecutive patients, 60 years of age or older, hospitalized due to COVID-19 pneumonia between March 1st and July 15th, 2020, were reviewed. Non-paired Student or chi-squared tests compared variables. A Cox proportional hazard model was employed to identify independent predictors of NOAF. P value < 0.05 was considered statistically significant. Results Among 667 patients hospitalized due to COVID-19, 201 (30.1%) fulfilled the inclusion criteria. NOAF was documented in 29 patients (14.4%), composing group 1. Group 2 was composed of 162 patients without NOAF. Ten patients were excluded due to the AF rhythm upon hospital admission. In groups 1 and 2, there were differences in overall in-hospital survival rate (24.1 % vs. 67.9%; p<0.001), length of stay in ICU (11.1 ± 10.5 days vs. 4.9 ± 7.5 days; p=0.004) and need for mechanical ventilation rate (82.9% vs. 32.7%; p<0.001). In the Cox model, age > 71 y/o (HR=6.8; p<0.001), total leukocyte count ≤ 7,720 cels.μL-¹ (HR=6.6; p<0.001), serum [Na+] ≤ 137 mEq.L-¹ (HR=5.0; p=0.001), SAPS3 score > 55 (HR=5.6; p=0.002), and disorientation (HR=2.5; p=0.04) on admission were independent predictors of NOAF. Conclusion NOAF is a common arrhythmia in elderly hospitalized patients with COVID-19 pneumonia. Clinical and laboratory parameters evaluated on admission have a predictive value for the occurrence of NOAF during hospitalization.

External validation of a convolutional neural network algorithm for opportunistically detecting vertebral fractures in routine CT scans.

The Convolutional Neural Network algorithm achieved a sensitivity of 94% and specificity of 93% in identifying scans with vertebral fractures (VFs). The external validation results suggest that the algorithm provides an opportunity to aid radiologists with the early identification of VFs in routine CT scans of abdomen and chest. PURPOSE: To evaluate the performance of a previously trained Convolutional Neural Network (CNN) model to automatically detect vertebral fractures (VFs) in CT scans in an external validation cohort. METHODS: Two Chinese studies and clinical data were used to retrospectively select CT scans of the chest, abdomen and thoracolumbar spine in men and women aged ≥50 years. The CT scans were assessed using the semiquantitative (SQ) Genant classification for prevalent VFs in a process blinded to clinical information. The performance of the CNN model was evaluated against reference standard readings by the area under the receiver operating characteristics curve (AUROC), accuracy, Cohen's kappa, sensitivity, and specificity. RESULTS: A total of 4,810 subjects were included, with a median age of 62 years (IQR 56-67), of which 2,654 (55.2%) were females. The scans were acquired between January 2013 and January 2019 on 16 different CT scanners from three different manufacturers. 2,773 (57.7%) were abdominal CTs. A total of 628 scans (13.1%) had ≥1 VF (grade 2-3), representing 899 fractured vertebrae out of a total of 48,584 (1.9%) visualized vertebral bodies. The CNN's performance in identifying scans with ≥1 moderate or severe fractures achieved an AUROC of 0.94 (95% CI: 0.93-0.95), accuracy of 93% (95% CI: 93%-94%), kappa of 0.75 (95% CI: 0.72-0.77), a sensitivity of 94% (95% CI: 92-96%) and a specificity of 93% (95% CI: 93-94%). CONCLUSION: The algorithm demonstrated excellent performance in the identification of vertebral fractures in a cohort of chest and abdominal CT scans of Chinese patients ≥50 years.