Autonomous Chest Radiograph Reporting Using AI: Estimation of Clinical Impact.

Background Automated interpretation of normal chest radiographs could alleviate the workload of radiologists. However, the performance of such an artificial intelligence (AI) tool compared with clinical radiology reports has not been established. Purpose To perform an external evaluation of a commercially available AI tool for (a) the number of chest radiographs autonomously reported, (b) the sensitivity for AI detection of abnormal chest radiographs, and (c) the performance of AI compared with that of the clinical radiology reports. Materials and Methods In this retrospective study, consecutive posteroanterior chest radiographs from adult patients in four hospitals in the capital region of Denmark were obtained in January 2020, including images from emergency department patients, in-hospital patients, and outpatients. Three thoracic radiologists labeled chest radiographs in a reference standard based on chest radiograph findings into the following categories: critical, other remarkable, unremarkable, or normal (no abnormalities). AI classified chest radiographs as high confidence normal (normal) or not high confidence normal (abnormal). Results A total of 1529 patients were included for analysis (median age, 69 years [IQR, 55-69 years]; 776 women), with 1100 (72%) classified by the reference standard as having abnormal radiographs, 617 (40%) as having critical abnormal radiographs, and 429 (28%) as having normal radiographs. For comparison, clinical radiology reports were classified based on the text and insufficient reports excluded (n = 22). The sensitivity of AI was 99.1% (95% CI: 98.3, 99.6; 1090 of 1100 patients) for abnormal radiographs and 99.8% (95% CI: 99.1, 99.9; 616 of 617 patients) for critical radiographs. Corresponding sensitivities for radiologist reports were 72.3% (95% CI: 69.5, 74.9; 779 of 1078 patients) and 93.5% (95% CI: 91.2, 95.3; 558 of 597 patients), respectively. Specificity of AI, and hence the potential autonomous reporting rate, was 28.0% of all normal posteroanterior chest radiographs (95% CI: 23.8, 32.5; 120 of 429 patients), or 7.8% (120 of 1529 patients) of all posteroanterior chest radiographs. Conclusion Of all normal posteroanterior chest radiographs, 28% were autonomously reported by AI with a sensitivity for any abnormalities higher than 99%. This corresponded to 7.8% of the entire posteroanterior chest radiograph production. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Park in this issue.

Left-sided heart disease and risk of death in patients with end-stage kidney disease receiving haemodialysis: an observational study.

BACKGROUND: Cardiovascular disease is the most common cause of death in patients with end-stage kidney disease on haemodialysis. The potential clinical consequence of systematic echocardiographic assessment is however not clear. In an unselected, contemporary population of patients on maintenance haemodialysis we aimed to assess: the prevalence of structural and functional heart disease, the potential therapeutic consequences of echocardiographic screening and whether left-sided heart disease is associated with prognosis. METHODS: Adult chronic haemodialysis patients in two large dialysis centres had transthoracic echocardiography performed prior to dialysis and were followed prospectively. Significant left-sided heart disease was defined as moderate or severe left-sided valve disease or left ventricular ejection fraction (LVEF) ≤40%. RESULTS: Among the 247 included patients (mean 66 years of age [95%CI 64-67], 68% male), 54 (22%) had significant left-sided heart disease. An LVEF ≤40% was observed in 31 patients (13%) and severe or moderate valve disease in 27 (11%) patients. The findings were not previously recognized in more than half of the patients (56%) prior to the study. Diagnosis had a potential impact on management in 31 (13%) patients including for 18 (7%) who would benefit from initiation of evidence-based heart failure therapy. After 2.8 years of follow-up, all-cause mortality among patients with and without left-sided heart disease was 52 and 32% respectively (hazard ratio [HR] 1.95 (95%CI 1.25-3.06). A multivariable adjusted Cox proportional hazard analysis showed that left-sided heart disease was an independent predictor of mortality with a HR of 1.60 (95%CI 1.01-2.55) along with age (HR per year 1.05 [95%CI 1.03-1.07]). CONCLUSION: Left ventricular systolic dysfunction and moderate to severe valve disease are common and often unrecognized in patients with end-stage kidney failure on haemodialysis and are associated with a higher risk of death. For more than 10% of the included patients, systematic echocardiographic assessment had a potential clinical consequence.

Abnormal routine blood tests as predictors of mortality in acutely admitted patients.

BACKGROUND: This study aimed to improve early risk stratification in the emergency department by creating a simple blood test score based on routine biomarkers and assess its predictive ability for 30-day mortality of acutely admitted patients. METHODS: This was a secondary analysis of data from the TRIAGE II study. It included unselected acutely admitted medical and surgical patients, who had albumin, C-reactive protein, creatinine, haemoglobin, leukocytes, potassium, sodium and thrombocytes levels analysed upon admission. Patients were classified according to the number of biomarker results outside the reference range into four risk groups termed "very low", "low", "intermediate", and "high" with 0-1, 2-3, 4-5 and 6-8 abnormal biomarker results, respectively. Logistic regression was used to calculate odds ratios for 30-day mortality and receiver operating characteristic was used to test the discriminative value. The primary analysis was done in patients triaged with ADAPT (Adaptive Process Triage). Subsequently, we analysed two other cohorts of acutely admitted patients. RESULTS: The TRIAGE II cohort included 17,058 eligible patients, 30-day mortality was 5.2%. The primary analysis included 7782 patients. Logistic regression adjusted for age and sex showed an OR of 24.1 (95% CI 14.9-41.0) between the very low- and the high-risk group. The area under the curve (AUC) was 0.79 (95% CI 0.76-0.81) for the blood test score in predicting 30-day mortality. The subsequent analyses confirmed the results. CONCLUSIONS: A blood test score based on number of routine biomarkers with an abnormal result was a predictor of 30-day mortality in acutely admitted patients.

Cardiac systolic dysfunction in past illicit users of anabolic androgenic steroids.

Background: Illicit use of anabolic androgenic steroids (AAS) is associated with left ventricle (LV) systolic dysfunction and increased LV mass (LVM), but whether these findings persist in former AAS users has yet to be elucidated. The objective was to assess LV systolic function, LVM and myocardial fibrosis in current and former illicit AAS users compared with non-users. Methods: Community-based cross-sectional study among men, aged 18­50 years, involved in recreational resistance training. We included 37 current and 33 former illicit AAS users, geometric mean (95%CI), 30 (21; 44) months since AAS cessation, and 30 non-users as controls. We assessed myocardial function and structure using advanced echocardiography and cardiac MRI with late-gadolinium enhancement. Results: Mean (SE) LV global longitudinal strain (GLS) was impaired in former AAS users compared with non-users, −16.7 (0.5) versus −18.2 (0.4) %, P < .05. Mean (SE) LV ejection fraction (EF) was decreased, 51 (1) versus 58 (1) %, P < .001 and LV GLS impaired, −14.5 (0.4)%, P < .001, in current AAS users compared with non-users. Measures of LVM were increased in current AAS users compared with the other two groups, P < .001. Plasma total testosterone was independently associated with reduced LVEF (P = .049) and increased LVM/body surface area (P = .005) in multivariate linear regressions. Focal myocardial fibrosis was not detected in any participants and diffuse myocardial fibrosis, assessed using post-contrast T1-mapping time, did not differ among the three groups. Conclusions: Past illicit AAS use is associated with impaired LV GLS, suggesting subclinical cardiac systolic dysfunction years after AAS cessation.

Iron Deficiency in COPD Associates with Increased Pulmonary Artery Pressure Estimated by Echocardiography.

OBJECTIVES: Iron deficiency (ID) might augment chronic pulmonary hypertension in chronic obstructive pulmonary disease (COPD). This observational study investigates the association between ID and systolic pulmonary artery pressure estimated by echocardiography in non-anaemic COPD outpatients. METHODS: Non-anaemic COPD patients (GOLD II-IV) with no history of cardiovascular disease were recruited from outpatient clinics. Iron deficiency was defined as ferritin<100µg/L. Pulmonary artery pressure was estimated from the tricuspid regurgitation maximum velocity (TR Vmax). Tricuspid regurgitation Vmax indicative of pulmonary hypertension was considered present for values ≥ 2.9 m/s. RESULTS: In a total of 75 included patients, 31 (41%) had ID. These patients had a significantly higher TR Vmax (3.02 vs. 2.77 m/s, p=0.01) and lower diffusion capacity of carbon monoxide (40% vs. 50% of predicted, p<0.01), though similar in age, sex, pack years, FEV1 and high-sensitive CRP (p>0.05). Ferritin inversely correlated with TR Vmax in ID patients (-0.37 (p=0.04)). The prevalence of TR Vmax ≥ 2.9 m/s was twice as high in patients with ID (58% vs. 29%) and odds ratio of pulmonary hypertension in ID (compared to no ID) was 3.3 (95% CI 1.3-8.6, p=0.015). CONCLUSION: Iron deficiency in non-anaemic COPD patients was associated with a modest increase in systolic pulmonary artery pressure and limitation of diffusion capacity.

Chronic obstructive pulmonary disease in patients with end-stage kidney disease on hemodialysis.

The objectives of this study were to assess the prevalence of chronic obstructive pulmonary disease (COPD) in hemodialysis patients with spirometry and to examine the effects of fluid removal by hemodialysis on lung volumes. Patients ≥18 years at two Danish hemodialysis centers were included. Forced expiratory volume in one second (FEV1 ), forced vital capacity (FVC), and FEV1 /FVC ratio were measured with spirometry before and after hemodialysis. The diagnosis of COPD was based on both the GOLD criteria and the lower limit of normal criteria. There were 372 patients in treatment at the two centers, 255 patients (69%) completed spirometry before dialysis and 242 of these (65%) repeated the test after. In the initial test, 117 subjects (46%) had airflow limitation indicative of COPD with GOLD criteria and 103 subjects (40.4%) with lower limit of normal criteria; COPD was previously diagnosed in 24 patients (9%). Mean FVC and FEV1 decreased mildly after dialysis (FVC: 2.84 to 2.79 L, P < 0.01. FEV1 : 1.97 to 1.93 L, P < 0.01) Hemodialysis did not affect the FEV1 /FVC ratio or number of subjects with airflow limitation indicative of COPD (113 vs. 120, P = 0.324; n = 242). COPD is a frequent and underdiagnosed comorbidity in patients on chronic hemodialysis. Spirometry should be considered in all patients on dialysis in order to address dyspnea adequately. Hemodialysis induced a small fall in mean FEV1 and FVC, which was more pronounced in patients with little or no fluid removal, but the FEV1 /FVC ratio and the number of subjects with airflow limitation indicative of COPD were not affected by dialysis.