Background: Lung point of care ultrasound (POCUS) offers advantages over traditional imaging for diagnosing pulmonary conditions, with superior accuracy compared to chest X-ray and lower cost compared to computed tomography. Despite these benefits, widespread adoption is limited by operator dependency, moderate interrater reliability, and training requirements. Deep learning (DL) could potentially address these challenges, but the development of effective algorithms is hindered by the scarcity of comprehensive image repositories with proper metadata. Methods: We created an open-source dataset of lung POCUS images derived from a multi-center study involving 226 adult patients presenting to emergency departments with respiratory symptoms between March 2020 and April 2022. Images were acquired using a standardized scanning protocol (12-zone or modified 8-zone) with various POCUS devices. Three blinded researchers independently analyzed each image following consensus guidelines, with disagreements adjudicated to provide definitive interpretations. Videos were preprocessed to remove identifiers, and frames were extracted and standardized to 512×512 pixels using letterboxing to maintain aspect ratios. Results: The dataset contained 1,871 video clips comprising 324,027 frames extracted and standardized to 512×512 pixels. Half of the participants (50%) had COVID-19 pneumonia. Among all clips, 66% contained no abnormalities, 18% contained B-lines, 4.5% contained consolidations, 6.4% contained both B-lines and consolidations, and 5.2% had indeterminate findings. Pathological findings varied significantly by lung zone, with anterior zones more frequently normal and less likely to show consolidations compared to lateral and posterior zones. Discussion: This dataset represents a large, annotated lung POCUS repository and includes patients with and without COVID-19. The repository metadata and expert interpretations enhance its utility for DL applications. Despite limitations including potential device-specific characteristics and COVID-19 predominance, this repository provides a valuable resource for developing artificial intelligence tools to improve lung POCUS acquisition and interpretation.

Lung ultrasound (LUS) is a bedside tool for assessing pulmonary edema in patients at risk due to heart failure or impaired kidney function. However, automated LUS analysis remains challenging because of speckle noise, imaging artifacts, and operator-dependent acquisition variability. In this work, we present a deep learning framework for multi-class LUS video classification that explores two components: hierarchy-aware training, and anatomy-guided learning. Starting from a strong baseline, we introduce hierarchical training strategies and then introduce pleural line mask supervision to guide model attention toward anatomically relevant regions. We study four clinically relevant classes–healthy, B-lines, consolidations, and mixed B-lines with consolidations–using an open-access dataset of 1,886 videos from 219 patients, evaluated with patient-level five-fold cross-validation. Results show that hierarchy-aware training improves pathological separation relative to flat classification, while mask-guided attention supervision achieves the highest mean macro-F1 of 65.7% and produces more localized attention patterns. Transfer experiments on the external COVID-BLUeS dataset further show competitive and parameter-efficient adaptation while preserving pleural-focused attention behavior. These findings suggest that combining clinically structured objectives with anatomy-guided supervision is a practical approach to robust, interpretable LUS video analysis. Code and model implementations are available at https://github.com/Alya-Almsouti/LUS-video-classification.

BACKGROUND:Post-acute sequelae of SARS-CoV-2 infection, more commonly known as long COVID, has emerged as a major health problem. The pathogenesis of long COVID is unknown, but among the leading hypotheses is viral persistence. We aimed to investigate whether the use of the SARS-CoV-2 antiviral nirmatrelvir-ritonavir improved long COVID symptoms. METHODS:We conducted a double-blind, placebo-controlled, randomised trial involving adults who had developed persistent symptoms (≥12 weeks) associated with three major symptom phenotypes (cognitive, autonomic, or exercise) after acute SARS-CoV-2 infection at 69 US sites. Participants were eligible if they were 18 years or older and had a previous suspected, probable, or confirmed SARS-CoV-2 infection, as defined by the Pan American Health Organization. Eligible participants were also required to have either at least two moderate symptoms from the same phenotype or one severe phenotype-associated symptom, as identified with the Cluster Targeted COVID-19 Symptom Questions. Participants were randomly allocated in a double-blind manner in a 1:1:1 ratio using permuted blocks of size 30 to receive either 15 days of active intervention followed by 10 days of placebo (300 mg nirmatrelvir-100 mg ritonavir twice daily, then 100 mg ritonavir-placebo); 25 days of active intervention (300 mg nirmatrelvir-100 mg ritonavir twice daily); or 25 days of placebo-ritonavir (100 mg ritonavir-placebo). A clinically significant change in patient-reported outcomes at day 90 comprised the primary endpoint: Patient-Reported Outcomes Measurement Information System Cognitive Function Short Form 8a, Orthostatic Hypotension Questionnaire question 1, and a modified version of the DePaul Symptom Questionnaire Post-Exertional Malaise short form. Secondary outcomes were phenotype-specific performance measures. The study was registered at ClinicalTrials.gov (NCT05595369) and is complete. FINDINGS:Between July 27, 2023, and Sept 6, 2024, 1207 individuals were screened. Of these, 964 were randomly allocated and 959 participants, excluding four participants who were later found ineligible and one who did not initiate treatment, were enrolled in the three phenotypes: 332 to cognitive, 334 to autonomic, and 332 to exercise. In the 959 participants in the mITT population, 643 (67%) self-reported as female, 314 (33%) were male, and two participants had a sex of unknown or undifferentiated; 750 (78%) were White; and 108 (11%) were Hispanic, Latino, or Spanish. The median age was 49 years (IQR 38-59). No statistically significant benefits were observed for any phenotype for primary endpoints. For the cognitive phenotype, adjusted differences compared to placebo were 3·2% (95% CI -10·4 to 16·8, p=0·65) for the 25-day regimen and -2·2% (-15·5 to 11·1, p=0·74) for the 15-day regimen. For the autonomic phenotype, adjusted differences were -6·4% (-18·5 to 5·7, p=0·30) for the 25-day regimen compared to placebo and -0·1% (-12·5 to 12·3, p=0·99) for the 15-day regimen compared to placebo. For exercise, adjusted differences were -7·8% (-19·5 to 3·8, p=0·19) for the 25-day regimen compared to placebo and 0·9% (-11·4 to 13·2, p=0·88) for the 15-day regimen compared to placebo. There were no differences in secondary endpoints, and no safety signals were observed; there were no deaths, and 52 serious adverse events occurred in 42 (4%) of 963 participants over the course of the study. INTERPRETATION:Nirmatrelvir-ritonavir for 15 days or 25 days showed no evidence of benefit in long COVID in any of the three phenotypes studied. These findings suggest additional approaches to measuring the symptom burden and treating Long COVID are needed. FUNDING:National Institutes of Health.

This JAMA Insights explores the use of point-of-care ultrasound as a less expensive, more accessible alternative to chest radiography and computed tomography for diagnosing community-acquired pneumonia.

Importance Olfactory dysfunction is common after SARS-CoV-2 infection and has been associated with cognitive loss in other conditions. Formal testing is needed to characterize the presence, severity, and patterns of olfactory dysfunction. ObjectiveTo characterize long-term olfactory dysfunction after SARS-CoV-2 infection. Design, Setting, and Participants This prospective cohort study included adults enrolled in the Researching COVID to Enhance Recovery (RECOVER)-Adult study. All those with and a random sample of those without self-reported change or loss in smell or taste were offered olfactory testing, performed at 83 sites in 35 US states and territories. Participants included 2956 enrollees with prior infection (1393 with and 1563 without self-reported change or loss) and 569 without prior infection (9 with and 560 without self-reported change or loss in taste) who underwent olfactory testing a mean (SD) of 671.6 (417.8) days after the index date. Data were collected from October 29, 2021, to June 6, 2025.ExposureSARS-CoV-2 infection.Main Outcomes and Measures Olfactory function, as defined by age- and sex-standardized performance on the University of Pennsylvania Smell Identification Test (UPSIT), a well-validated test comprising 40 unique odors. ResultsThe study included 3525 participants with a mean (SD) age of 47.6 (15.2) years; of 3520 with data available, 2548 (72.4%) were female or intersex. Among 1393 infected participants with self-reported change or loss, 1111 (79.8%) had hyposmia on the UPSIT, including 321 (23.0%) with severe microsmia or anosmia. Among 1563 infected participants without self-reported change or loss, 1031 (66.0%) had hyposmia, including 128 (8.2%) with severe microsmia or anosmia. Participants with prior infection and self-reported change or loss scored at the 16th age- and sex-standardized UPSIT percentile, compared with the 23rd and 28th percentiles for those without self-reported change or loss with and without prior known infection, respectively. Younger women had scores corresponding to lower mean age- and sex-standardized percentiles. Among participants who self-reported change or loss in smell, those with abnormal UPSIT scores more often reported cognitive problems (742 of 1111 [66.8%]) than those with normal UPSIT scores (179 of 282 [63.5%]).Conclusions and RelevanceIn this cohort study of RECOVER-Adult participants, self-reported change or loss in smell or taste was an accurate signal of verified hyposmia, but a high rate of hyposmia among those with no reported change or loss was also observed. Formal smell testing may be considered in those with prior SARS-CoV-2 infection to diagnose occult hyposmia and counsel patients about risks.

Background Deep learning (DL) programs can aid in the acquisition of echocardiograms by medical professionals not previously trained in sonography, potentially addressing access issues in underserved communities. This study evaluates whether DL-enabled devices improve limited echocardiogram acquisition by novice clinicians not trained on sonography. Methods In this single-center randomized controlled trial (2023-2024), internal medicine residents (N=38) without sonography training received a personal ultrasound device with (N=19) or without (N=19) DL capability for two weeks while caring for patients on a hospital ward. Participants were allowed to use the devices at their discretion for patient-related care. The DL software provided real-time guidance for probe placement and image quality assessment. The primary outcome was time to acquire a five-view limited echocardiogram. Measurements occurred at randomization and after two weeks, with all scans performed on the same standardized patient. Secondary outcomes included image quality using the modified Rapid Assessment for Competency in Echocardiography (RACE) scale and participant attitudes. Results At baseline, both groups had comparable scan times and image quality scores. At follow-up, the DL group demonstrated significantly faster total scan times (152 seconds [IQR 115-195] vs. 266 seconds [IQR 206-324]; p<0.001; Cohen's D 1.7) and better image quality with higher RACE scores (15 [IQR 10-18] vs. 11 [IQR 7-13.5]; p=0.034; Cohen's D 0.84). Trust in the AI features did not differ between the groups post-intervention. Conclusions Ultrasound machines with DL features may improve image acquisition times and image quality by novices not trained in sonography. These findings suggest DL algorithms could help address critical gaps in image acquisition by healthcare professionals. ### Competing Interest Statement Dr. Kumar reports receiving consultant fees from Caption Health, which is unrelated to this body of work. The other authors do not have any financial interests to disclose. This study was an investigator-initiated study. None of the device manufacturers used in this study oversaw, influenced, or reviewed the data presented prior to submission. The authors vouch for the independent nature by which this investigation was conducted and disseminated, without influence from outside organizations. ### Clinical Trial NCT05900440 ### Funding Statement This study was not funded ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The Stanford University Institutional Review Board gave ethical approval for this work. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The datasets generated and/or analyzed during the current study are not publicly available due to sharing limitations from our IRB protocol, but are available from the corresponding author on reasonable request.

Generative Artificial Intelligence (Gen AI) shows significant promise as a technology that could improve healthcare delivery, but its implementation will be influenced by the spheres in which it is studied and the limited resources of hospitals. The Point authors argue that we should focus on is the cognitive abilities of GenAI or we risk being left out of a technological leap that will change the way doctors practice. The Counterpoint argues that we should focus on using GenAI to ease system burdens and address workflow issues, focusing our efforts on fixing the problems that would improve doctors' quality of life and increase time spent with patients.

ImportanceA substantial number of individuals worldwide experience long COVID, or post-COVID condition. Other postviral and autoimmune conditions have a female predominance, but whether the same is true for long COVID, especially within different subgroups, is uncertain.ObjectiveTo evaluate sex differences in the risk of developing long COVID among adults with SARS-CoV-2 infection.Design, Setting, and ParticipantsThis cohort study used data from the National Institutes of Health (NIH) Researching COVID to Enhance Recovery (RECOVER)–Adult cohort, which consists of individuals enrolled in and prospectively followed up at 83 sites in 33 US states plus Washington, DC, and Puerto Rico. Data were examined from all participants enrolled between October 29, 2021, and July 5, 2024, who had a qualifying study visit 6 months or more after their initial SARS-CoV-2 infection.ExposureSelf-reported sex (male, female) assigned at birth.Main Outcomes and MeasuresDevelopment of long COVID, measured using a self-reported symptom-based questionnaire and scoring guideline at the first study visit that occurred at least 6 months after infection. Propensity score matching was used to estimate risk ratios (RRs) and risk differences (95% CIs). The full model included demographic and clinical characteristics and social determinants of health, and the reduced model included only age, race, and ethnicity.ResultsAmong 12 276 participants who had experienced SARS-CoV-2 infection (8969 [73%] female; mean [SD] age at infection, 46 [15] years), female sex was associated with higher risk of long COVID in the primary full (RR, 1.31; 95% CI, 1.06-1.62) and reduced (RR, 1.44; 95% CI, 1.17-1.77) models. This finding was observed across all age groups except 18 to 39 years (RR, 1.04; 95% CI, 0.72-1.49). Female sex was associated with significantly higher overall long COVID risk when the analysis was restricted to nonpregnant participants (RR, 1.50; 95%: CI, 1.27-1.77). Among participants aged 40 to 54 years, the risk ratio was 1.42 (95% CI, 0.99-2.03) in menopausal female participants and 1.45 (95% CI, 1.15-1.83) in nonmenopausal female participants compared with male participants.Conclusions and RelevanceIn this prospective cohort study of the NIH RECOVER-Adult cohort, female sex was associated with an increased risk of long COVID compared with male sex, and this association was age, pregnancy, and menopausal status dependent. These findings highlight the need to identify biological mechanisms contributing to sex specificity to facilitate risk stratification, targeted drug development, and improved management of long COVID.

The global shortage of sonographers has created significant barriers to timely ultrasound diagnostics across medical specialties. Deep learning (DL) algorithms have potential to enhance image acquisition by clinicians without formal sonography training, potentially expanding access to crucial diagnostic imaging in resource-limited settings. This study evaluates whether DL-enabled devices improve acquisition of multi-view limited echocardiograms by healthcare providers without previous cardiac ultrasound training. In this single-center randomized controlled trial (2023-2024), internal medicine residents (N = 38) without prior sonography training received a portable ultrasound device with (N = 19) or without (N = 19) DL capability for a two-week clinical integration period during regular patient care on hospital wards. The DL software provided real-time guidance for probe positioning and image quality assessment across five standard echocardiographic views. The primary outcome was total acquisition time for a comprehensive five-view limited echocardiogram (parasternal long axis, parasternal short axis, apical 4-chamber, subcostal, and inferior vena cava views). Assessments occurred at randomization and after two weeks using a standardized patient. Secondary outcomes included image quality using a validated assessment tool and participant attitudes toward the technology. Baseline scan times and image quality scores were comparable between groups. At two-week follow-up, participants using DL-equipped devices demonstrated significantly faster total scan times (152 s [IQR 115-195] versus 266 s [IQR 206-324]; P < 0.001; Cohen's D = 1.7) and superior image quality with higher modified RACE scores (15 [IQR 10-18] versus 11 [IQR 7-13.5]; P = 0.034; Cohen's D = 0.84). Performance improvements were most pronounced in technically challenging views. Both groups reported similar levels of trust in DL-functionality. Ultrasound devices incorporating deep learning algorithms significantly improve both acquisition speed and image quality of comprehensive echocardiographic examinations by novice users. These findings suggest DL-enhanced ultrasound may help address critical gaps in diagnostic imaging capacity by enabling non-specialists to acquire clinically useful cardiac images.

Abstract Background Point-of-care ultrasound (POCUS) has emerged as an essential bedside tool for clinicians, but lack of access to ultrasound equipment has been a top barrier to POCUS use. Recently, several handheld ultrasound devices (“handhelds”) have become available, and clinicians are seeking data to guide purchasing decisions. Few comparative studies of different handhelds have been done. We conducted a cross-sectional study comparing 6 handhelds readily available in the United States (Butterfly iQ + ™ by Butterfly Network Inc.; Clarius™ by Clarius Mobile Health; Kosmos™ by EchoNous; TE Air™ by Mindray; Vscan Air™ SL and CL by General Electric; and Lumify™ by Philips Healthcare). A multi-specialty group of physician POCUS experts (n = 35) acquired three standard ultrasound views (abdominal right upper quadrant, cardiac apical 4-chamber, and superficial neck and lung views) in random order on the same standardized patients and rated the image quality. Afterward, a final survey of the overall ease of use, image quality, and satisfaction of each handheld was completed. Results Thirty-five POCUS experts specializing in internal medicine/hospital medicine, critical care, emergency medicine, and nephrology acquired and rated right upper quadrant, apical 4-chamber, and superficial neck and lung views with 6 different handhelds. For image quality, the highest-rated handhelds were Vscan Air™ for the right upper quadrant view, Mindray TE Air™ for the cardiac apical 4-chamber view, and Lumify™ for superficial views of the neck and lung. Overall satisfaction with image quality was highest with Vscan Air™, Lumify™, and Mindray, while overall satisfaction with ease of use was highest with Vscan Air™. The 5 most desirable characteristics of handhelds were image quality, ease of use, portability, probe size, and battery life. Ultimately, all 6 handhelds had notable advantages and disadvantages, with no single device having all desired qualities or features. Conclusions The overall satisfaction with image quality was rated highest with Vscan Air™, Lumify™, and Mindray TE Air™when acquiring right upper quadrant, apical 4-chamber, and superficial neck and lung views. No single handheld was perceived to be superior in image quality for all views. Vscan Air™ was rated highest for overall ease of use and was the most preferred handheld for purchase by POCUS experts.

rapid assessment for competency in echocardiography).

Background:Point-of-care ultrasonography (POCUS) machines may use deep learning, a subfield of artificial intelligence (AI), to improve image interpretation and acquisition in real time. The impact of AI on POCUS learning is unknown. Research Question:Do AI-enhanced devices equipped with deep learning aid in cardiac image acquisition and interpretation among POCUS novices? Study Design and Methods:We conducted a single-center investigation from 2021 through 2022. Internal medicine trainees (N = 43) with limited POCUS experience were randomized to receive a POCUS device with (Echonous; n = 22) or without (Butterfly; n = 21) AI functionality for 2 weeks while on inpatient rotations. AI device functionality included guidance for optimal probe placement to acquire an apical four-chamber (A4C) view and ejection fraction estimations based on deep learning. Participants used the devices at their discretion for patient-related care after randomization. The primary outcome was the time to acquire A4C images on a standardized patient. Secondary outcomes included A4C image quality using a validated scale, image quiz performance, and attitudes. Measurements were performed at randomization and at 2-week follow-up using the same standardized patient. Results:Both AI and non-AI groups showed similar scan times and image quality scores at baseline. At follow-up, the AI group showed faster scan times (57 s [interquartile range (IQR), 32-75 s] vs 85 s [IQR, 50-172 s]; P = .01), higher image quality scores (4.5 [IQR, 2-5.5] vs 2 [IQR, 1-3]; P < .01), and more accurately identified reduced systolic function on the image quiz (85% vs 50%; P = .02) vs the non-AI group. The AI group used the devices more than the non-AI group (median, 5.5 times [IQR, 4-10 times] vs 2 times [IQR, 0-4 times]; P < .01). Trust in the AI features did not change during the intervention. Interpretation:POCUS devices with deep learning functionality may improve A4C image acquisition and interpretation by novices. Future studies are needed to determine the extent that AI impacts POCUS learning.

This study compares performance on free-response clinical reasoning examinations of first- and second-year medical students vs 2 models of a popular chatbot.

Background Point-of-care ultrasound (POCUS) machines may utilize artificial intelligence (AI) to enhance image interpretation and acquisition. This study investigates whether AI-enabled devices improve competency among POCUS novices. Methods We conducted a randomized controlled trial at a single academic institution from 2021-2022. Internal medicine trainees (N=43) with limited POCUS experience were randomized to receive a POCUS device with (Echonous, N=22) or without (Butterfly, N=21) AI-functionality for two weeks while on an inpatient rotation. The AI-device provided automatic labeling of cardiac structures, guidance for optimal probe placement to acquire cardiac views, and ejection fraction estimations. Participants were allowed to use the devices at their discretion for patient-related care. The primary outcome was the time to acquire an apical 4-chamber (A4C) image. Secondary outcomes included A4C image quality using the modified Rapid Assessment for Competency in Echocardiography (RACE) scale, correct identification of pathology, and participant attitudes. Measurements were performed at the time of randomization and at two-week follow-up. All scanning assessments were performed on the same standardized patient. Results Both AI and non-AI groups had similar scan times and image quality scores at baseline. At follow-up, the AI group had faster scan times (72 seconds [IQR 38-85] vs. 85 seconds [IQR 54-166]; p=0.01), higher image quality scores (4.5 [IQR 2-5.5] vs. 2 [IQR 1-3]; p<0.01) and correctly identified reduced systolic function more often (85% vs 50%; p=0.02) compared to the non-AI group. Trust in the AI features did not differ between the groups pre- or post-intervention. The AI group did not report increased confidence in their abilities to obtain or interpret cardiac images. Conclusions POCUS devices with AI features may improve image acquisition and interpretation by novices. Future studies are needed to determine the extent that AI impacts POCUS learning.

Lung ultrasound (LUS) has received considerable interest in the clinical evaluation of patients with COVID‐19. Previously described LUS manifestations for COVID‐19 include B‐lines, consolidations, and pleural thickening. The interrater reliability (IRR) of these findings for COVID‐19 is unknown.

Introduction: Point-of-care ultrasound (POCUS) can detect the cardiopulmonary manifestations of COVID-19 and may predict patient outcomes in an expedient and cost-effective manner.Methods: We conducted a prospective cohort study at four medical centers from 3/2020 to 9/2020 to evaluate the relationship between POCUS findings and clinical outcomes with COVID-19. Our inclusion criteria included adult patients hospitalized for COVID-19 who received lung ultrasound (LUS) examinations at the bedside with a 12-zone protocol. All images were interpreted by at least two reviewers who were blinded to clinical outcomes.Results: N=99 patients met inclusion criteria. The median time from ED triage to LUS was 0.9 days (IQR: 0.1-2.9). LUS was rarely normal (11% of patients), with B-lines (90%) and subpleural consolidations (62%) representing the most prevalent findings. Findings associated with ICU admission included anterior B-lines (OR: 3.1 [95% CI: 1.2-9.7]), anterior consolidations (OR: 3.1 [95% CI: 1.1-9.9]) and lateral consolidations (OR: 4.1 [95% CI: 1.3-15.5]), while a normal scan was strongly protective against ICU admission (OR 0.08 [95% CI: 0.00-0.68]). LUS findings remained stable over a period of 28 days from symptom onset.Discussion: Anterior and/or lateral lung involvement on POCUS may portend a three to four-fold risk of critical illness among COVID-19 patients. The location, rather than the absolute appearance of POCUS findings, may be important harbingers of severe disease. POCUS findings did not change over a 28-day scanning period, suggesting that their detection at any time point may be clinically important.Funding Statement: None to declareDeclaration of Interests: None to declare. Ethics Approval Statement: This study was approved by the Institutional Review Boards of Stanford University and the University of California, San Francisco. A waiver of consent was obtained by both institutions.

IntroductionPoint-of-care ultrasound (POCUS) has the potential to transform healthcare delivery in the era of COVID-19 with its diagnostic and therapeutic expediency. It can be performed by clinicians already at the bedside, which permits an immediate and augmented assessment of a patient. Although lung ultrasound can be used to accurately diagnose a variety of disease states such as pneumothorax, pleural effusions, pneumonia and interstitial lung disease2, there are limited reports on the sonographic manifestations of COVID-19. There is an urgent need to identify alternative diagnostic modalities that can be immediately employed at the bedside of COVID-19 patients.MethodsThis study was conducted at two medical centers in the United States from 3/21/2020-6/01/2020. Any adult who was hospitalized with COVID-19 (based on symptomatology and a confirmatory RT-PCR for SARS-CoV-2) and received a pulmonary POCUS examination was included. Providers were instructed to use a 12-zone scanning protocol for pulmonary views and save 6 second clips of each lung zone. This study utilized several POCUS devices, including Butterfly IQ, Vave, Lumify, and Sonosite. The collected images were interpreted by the study researchers based on a consensus document developed by the study authors and previously accepted definitions of lung POCUS findings.ResultsA total of 22 eligible patients who received 36 lung scans were included in our study. Eleven (50%) patients experienced clinical deterioration (defined as either ICU admission, invasive mechanical ventilation, or death within 28 days from the initial symptom onset). Among the 36 lung scans collected, only 3 (8%) were classified as normal. The remaining scans had the following abnormalities: presence of B-lines (n=32, 89%), consolidations (n=20, 56%), pleural thickening (n=17, 47%), and pleural effusion (n=4, 11%). Out of 20 scans with consolidations, 14 (70%) were subpleural and 5 (25%) were translobar. A-lines were present in 26 (72%) of patients, although they were only observed in the majority of the collected lung zones in 5 (14%) of patients. Ultrasound findings were stratified by time from symptom onset to the scan based on the following time periods: early (0-6 days), middle (7-13 days), and late (14-28 days). B-lines appeared early after symptom onset and persisted well into the late disease course. In contrast, pleural thickening increased in frequency over time (early: 25%, middle: 47%, late: 67%). Subpleural consolidations also appeared in higher frequency later in the disease course (early: 13%, middle 42%, late: 56%).Discussioncertain lung ultrasound findings may be common in Covid-19, while others may appear later in the disease course or only occur in patients who experience clinical deterioration. Future efforts should investigate the predictive utility of consolidations, pleural thickening and B-lines for clinical deterioration and compare them to traditional radiological studies such as X-rays or CTs.

Introduction: Point-of-care ultrasound (POCUS) may detect the cardiopulmonary manifestations of COVID-19 and expediently predict patient outcomes. Methods: We conducted a prospective cohort study at four medical centers from 3/2020-1/2021 to evaluate POCUS findings and clinical outcomes with COVID-19. Our inclusion criteria included adult patients hospitalized for COVID-19 who received cardiac or lung POCUS with a 12-zone protocol. Images were interpreted by two reviewers blinded to clinical outcomes. Our primary outcome was ICU admission incidence. Secondary outcomes included intubation and supplemental oxygen usage. Results: N=160 patients (N=201 scans) were included. Scans were collected a median 23 hours (IQR:7-80) from emergency department triage. Triage POCUS findings associated with ICU admission included B-lines (OR 4.41 [95% CI:1.71-14.30]; p<0.01) or consolidation (OR 2.49 [95% CI:1.35-4.86]; p<0.01). B-lines were associated with intubation (OR 3.10 [95% CI:1.15-10.27]; p=0.02) and supplemental oxygen usage (OR 3.74 [95% CI:1.63-8.63; p<0.01). Consolidations present on triage were associated with the need for oxygen at discharge (OR 2.16 [95% CI: 1.01-4.70]; p=0.047). A normal lung triage scan was protective for ICU admission (OR 0.28 [95% CI:0.09-0.75; p<0.01) or need for supplemental oxygen during the hospitalization (OR 0.26 [95% CI:0.11-0.61]; p<0.01). Triage cardiac POCUS scans were not associated with any outcomes. Discussion: Lung POCUS findings detected early in the hospitalization may provide expedient risk stratification for important COVID-19 clinical outcomes, including ICU admission, intubation, or need for oxygen on discharge. A normal admission scan appears protective against adverse outcomes, which may aid in triage decisions of patients.