Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2025 Sep 5;8(9):e2530677. doi: 10.1001/jamanetworkopen.2025.30677

Cardiopulmonary Point-of-Care Ultrasonography for Hospitalist Management of Undifferentiated Dyspnea

Kameswari Maganti 1, Catherine Chen 2, Ankush D Jamthikar 1, Payal Parikh 2, Naveena Yanamala 1, Partho P Sengupta 1,✉
PMCID: PMC12413653  PMID: 40911308

This quality improvement study evaluates associations of hospitalist-implemented cardiopulmonary point-of-care ultrasonography for patients admitted with undifferentiated dyspnea with hospital length of stay and hospitalization cost.

Key Points

Question

Is cardiopulmonary point-of-care ultrasonography (POCUS) associated with reduced hospital length of stay (LOS) and hospitalization costs in patients with undifferentiated dyspnea?

Findings

In this quality improvement study involving 208 patients with undifferentiated dyspnea, POCUS, compared with standard care, was associated with a reduction of 246 hospital bed–days and direct cost savings of $751 537, with an incremental cost-effectiveness ratio of $3055 per bed-day saved. However, despite training, only 20% of hospitalists used POCUS independently.

Meaning

These findings suggest that cardiopulmonary POCUS has substantial potential to enhance hospital efficiency by reducing LOS and costs, and future studies should address adoption barriers with tailored education and support to ensure consistent integration.

Abstract

Importance

The association of cardiopulmonary point-of-care ultrasonography (POCUS) with length of stay (LOS) and hospitalization costs for patients admitted to internal medicine wards remains uncertain.

Objective

To evaluate a collaborative implementation model involving hospitalists, sonographers, and a remote cardiologist for integrating cardiopulmonary POCUS into the assessment of adult patients (≥18 years) hospitalized with undifferentiated dyspnea, and to assess its association with LOS and hospitalization costs.

Design, Setting, and Participants

This quality improvement study employed a type 1 effectiveness-implementation hybrid design using a 6-month stepped-wedge cluster randomized approach, conducted at a tertiary care hospital in the US between December 7, 2023, and July 2, 2024, to compare the standard-of-care (control) with the intervention group. Patients were eligible for inclusion if they were older than 18 years, admitted to 1 of the 5 internal medicine teaching hospitalist teams, and presented with undifferentiated dyspnea.

Exposure

Structured cardiopulmonary POCUS examinations performed by hospitalists and/or sonographers, integrated into routine assessment of dyspnea.

Main Outcomes and Measures

Study outcomes (LOS and hospitalization costs) were presented using the reach, effectiveness, adoption, and implementation (RE-AIM) framework.

Results

The study reached 208 patients (median [IQR] age, 71 [59-80] years; 121 female [58%]), including 107 in the control group and 101 in the POCUS group. The implementation of cardiopulmonary POCUS was associated with a 30.3% (95% CI, 5.5%-48.9%) reduction in expected LOS (mean [SD] LOS, 8.3 [5.2] days for the POCUS group vs 11.9 [7.5] days in the control group). Based on cumulative assessments, POCUS use was associated with a total reduction of 246 hospital bed–days and direct cost savings of $751 537, with an incremental cost-effectiveness ratio of $3055 per hospital bed–day saved. POCUS altered medical decisions in 30 patients (35%). Adoption and implementation of POCUS by hospitalists remained limited despite comprehensive training, with only 20% of POCUS evaluations (17 patients) being performed independently, while the majority relied on sonographers.

Conclusions and Relevance

In this quality improvement study, cardiopulmonary POCUS implementation was associated with a significant reduction in LOS and hospitalization costs, highlighting its clinical utility and potential for improved hospital efficiency; however, limited adoption by hospitalists underscores the need for ongoing training, support, and professional incentives to strengthen competency and motivation. Multicenter studies are needed to evaluate tailored educational models and sustainable support systems to optimize long-term integration of POCUS into routine practice.

Introduction

Dyspnea is a frequently encountered presenting symptom in clinical medicine, with almost 1 in 10 hospitalized patients experiencing it at rest.1,2 Cardiopulmonary causes of dyspnea are particularly concerning because they are associated with poor clinical outcomes, frequent hospital readmissions, and increased mortality risk.3,4,5 Traditional dyspnea assessment relies on a history and physical examination to differentiate between cardiac and noncardiac causes.6 However, point-of-care ultrasonography (POCUS) as an extension of physical examination has demonstrated superior accuracy in bedside cardiopulmonary evaluations, outperforming traditional stethoscope and chest x-ray–based evaluations.7,8,9

POCUS is widely endorsed and increasingly integrated across medical disciplines, enhancing diagnostic accuracy, efficiency, patient engagement, and satisfaction.10,11,12,13,14,15,16,17,18,19,20,21,22,23 Despite its potential and endorsement by professional societies,24,25 the adoption of POCUS among clinicians remains inconsistent.26,27,28,29,30 Moreover, there is limited and conflicting evidence from clinical trials about the impact of POCUS on the length of hospital stay and hospitalists’ management decisions, including the transfer of participants to other services, and the cost of care in inpatient care settings.31,32,33

To address this gap, our study sought to evaluate a collaboration between hospitalists, a sonographer, and a cardiologist at a distance for implementing a hospitalist-driven bedside cardiopulmonary POCUS evaluation in patients admitted with undifferentiated dyspnea. We hypothesized that a collaborative model would overcome barriers for implementing cardiopulmonary POCUS during the assessment of patients hospitalized with undifferentiated dyspnea and would be associated with a reduction in length of stay (LOS) and overall cost of hospitalization.

Methods

This study was designed as a quality improvement initiative that followed the Standards for Quality Improvement Reporting Excellence (SQUIRE) reporting guideline for quality improvement studies.34 This study was approved by the institutional review board at the Rutgers Office of Research in New Jersey. We employed a type 1 hybrid effectiveness-implementation framework to support the implementation of cardiopulmonary POCUS at a tertiary care hospital in the US, while simultaneously collecting data on the outcomes associated with its delivery.35 Furthermore, a stepped-wedge cluster randomized implementation design was employed for 5 inpatient teams between December 7, 2023, and July 2, 2024 (Figure 1).

Figure 1. Step-Wedge Cluster-Randomized Design.

Figure 1.

Open in a new tab

The diagram illustrates the structure of a cluster-randomized design where multiple clusters (labeled 1 to 5) are assigned either to the control or point-of-care ultrasonography (POCUS) group. A total of 216 hospitalizations were identified for the study. Following the exclusion of rehospitalizations—7 from the control group and 1 from the POCUS group—resulting in 208 unique patient hospitalizations (107 in the control group and 101 in the POCUS group). Propensity score matching (for age, sex, cardiac arrythmia, pulmonary comorbidities, and venous thromboembolism) was then applied to obtain well-matched cohorts for analysis. CHF indicates congestive heart failure.

For implementing cardiopulmonary POCUS, the first training sessions were offered to the hospitalists by research sonographers using a live model and educational videos provided via the Butterfly Network learning management system. The project implementation began with a control group of physicians who did not use POCUS, followed by randomized clusters that adopted cardiovascular and pulmonary POCUS. All education was provided on a 1:1 basis starting in January 2024, just before the randomization.

At the time this study was implemented, cardiopulmonary POCUS was not routinely performed by all practitioners. Additionally, images or preliminary findings obtained in the emergency department were not consistently shared with the hospitalist team. The standard practice involved ordering a formal echocardiogram, which was performed by a sonographer and interpreted by a board-certified cardiologist. In this study, patients assigned to the POCUS group received the examination on the first day of admission, performed by a board-certified hospitalist during rounds and supported by a POCUS-trained sonographer to ensure consistent implementation. Patients were included consecutively when the echocardiography team was available (weekdays, from 8:00 am to 5:00 pm).

A study coordinator reviewed the admission dashboard each morning to identify patients admitted with suspected cardiopulmonary dyspnea for inclusion in either the control group or POCUS group. Patients were eligible for inclusion if they were older than 18 years, admitted to 1 of the 5 medicine teaching hospitalist teams, and presented with undifferentiated dyspnea. Exclusion criteria included (1) chest wall trauma, (2) limited code status with patient choosing to receive only specific resuscitation treatments or none, (3) patient refusal of POCUS evaluation, and (4) clinical instability (systolic blood pressure <85 mm Hg and heart rate >120 beats per minute). Written informed consent was waived because this was an implementation-efficacy study evaluating a standard-of-care initiative, which is consistent with the ethical guidelines for minimal-risk research. POCUS for procedural guidance was excluded from the study, and POCUS examinations were not billed.

Cardiopulmonary POCUS

POCUS examinations were conducted by a physician or sonographer using handheld ultrasonography probes (Butterfly iQ3) and a tablet or smartphone-based application to capture images, including essential details such as the patient’s unique identifier, the time and location of the examination, and the body areas and items evaluated. Images generated during POCUS examinations were uploaded to a cloud-based system, granting unrestricted access to the hospitalists.

Preliminary interpretations of the examinations were conducted by either the attending hospitalists or the sonographers. The protocol allowed hands-on support in cases where hospitalists consulted sonographers for assistance with image acquisition. The acquired ultrasonography images were uploaded to the research cloud, and the final interpretations were completed by 2 level III–trained cardiologists (US National Board of Echocardiography–certified experts in echocardiography with more than 9 cumulative months of training) who were blinded to the patient’s clinical data.

The POCUS evaluation (Figure 2) included (1) a cardiovascular evaluation and (2) pulmonary evaluation. The cardiovascular evaluation included parasternal long axis, parasternal short axis, apical 4-chamber, apical 2-chamber, and apical 3-chamber views to assess the cardiac chamber and valve function. Briefly, analysis and interpretation by expert readers provided qualitative grading of left or right ventricular dilation, reduction of left and right ventricular systolic function, regional wall motion abnormalities, and other relevant findings, including the presence of valvular abnormalities (regurgitant or stenotic). Left ventricular ejection fraction less than 50% was considered low by visual estimation. Subcostal images assessed the collapsibility of the inferior vena cava to provide an indirect assessment of right atrial pressure. The pulmonary evaluation was performed using a 6-zone lung ultrasonography examination (Figure 2). A US Food and Drug Administration–approved artificial intelligence tool (Auto B-line Counter) was used for automated B-line count from a 6-second lung ultrasonography clip in each zone. These automated B-line counts were further visually verified to ensure that the lines were defined as vertical, hyperechoic artifacts originating at the pleural line, extending to the bottom of the screen without fading, and moving with lung sliding. The presence of 3 or more B-lines in at least 2 lung zones was considered abnormal.

Figure 2. Components of Cardiovascular and Pulmonary Point-of-Care Ultrasonography (POCUS).

Figure 2.

Open in a new tab

A-C, An example of cardiac assessment (parasternal long-axis view [PLAX]; parasternal short-axis view [PSAX]; and apical 4-chamber view [A4C]). D, The lung fields used for pulmonary POCUS (R indicates right; L, left). E-F, An example of dilated inferior vena cava and more than 3 B-lines seen in the right (RT) upper pulmonary POCUS.

Data Sources and Measures

Data were collected from 3 sources: the POCUS smartphone application (capturing all POCUS-related variables), the hospital’s electronic health record (including patient demographics, admission diagnoses, formal imaging studies, care utilization, outcomes, and care team identifiers), and the hospital billing database (covering costs). The study outcomes were presented using the reach, efficacy, adoption, and implementation (RE-AIM) framework. This systematic framework has been successfully used to report the implementation of health care interventions (eTable 1 in Supplement 1).36 Following the pragmatic (ie, practice-oriented) use of RE-AIM, and because this was a short-term pilot study, maintenance was not evaluated. The primary effectiveness outcome of the study was hospital LOS. Secondary effectiveness outcomes included total direct hospitalization costs, serving as a measure of hospital resource utilization. The direct hospitalization cost was estimated by multiplying the total charges billed (including drug supply, laboratory services, radiology, room charges, operating room costs, respiratory care, therapy, and other miscellaneous expenses) by the hospital’s overall cost-to-charge ratio for 2024.37 The incremental cost-effectiveness ratio was calculated by dividing the difference in mean direct hospitalization costs by the difference in mean LOS per patient between the standard-of-care (control) and POCUS groups and presented as a reduction in direct cost per bed-day saved.38 To understand if a change in LOS was associated with readmission rates, we also collected data on 30-day and 90-day readmissions for both groups.

Adjudication of Heart Failure Diagnosis at Discharge

The discharge diagnosis of heart failure was determined by a cardiologist expert after reviewing the patient’s medical record. Medical record review provided full access to all adjunctive tests performed during the hospital stay, but the was blinded to the POCUS or control group of the study.

Adjudication of Association of POCUS With Clinical Decision-Making

After the POCUS was interpreted by 2 board-certified cardiologists as promptly as possible, Health Insurance Portability and Accountability Act–compliant text alerts were sent to the attending hospitalist. In addition, the study coordinator or attending cardiologists directly informed the hospitalist of the findings to ensure timely communication. Any subsequent changes in medical decision-making and management were determined at the discretion of the treating hospitalist. An expert hospitalist independently reviewed the medical records to assess the diagnostic approach before and after the use of POCUS. The association of POCUS with medical decision-making was measured as a percentage of patients with either (1) no change or (2) a change in medical decision (a normal study eliminating a cardiopulmonary source of dyspnea, a new diagnosis, or a change in therapeutic management, as determined by blinded medical record review.

Statistical Analysis

Main Analysis

Descriptive statistics are presented as the median (with IQR) for continuous variables and the count (with percentage) for categorical variables. Continuous variables were analyzed using a Mann-Whitney test for the POCUS and control groups, while categorical variables were assessed with a χ2 test. LOS is reported as median (IQR) with 1 decimal precision to avoid overlap with the IQR values. Because the empirical distribution of LOS for hospitalizations is typically right-skewed, plurimodal, and contains outliers, we first explored and assessed its fit to an underlying statistical model. Among the candidate models, the 2-component gamma mixture model (GMM) demonstrated the best fit, exhibiting the lowest Akaike information criterion (1022) and highest log-likelihood (−506) (eTable 2 and eTable 3 in Supplement 1).39,40 This model, first applied separately to the control and intervention groups, identified 2 latent subgroups based solely on LOS gamma distribution components, each parameterized by shape and scale (eTable 4 in Supplement 1), without relying on predefined clinical characteristics.39,40 The first subgroup (component 1) represented patients with shorter stays, and the second subgroup (component 2) captured patients with longer stays, with an overlap region reflecting those exhibiting characteristics of both patterns.

An intention-to-treat analysis was performed to assess the primary effectiveness outcome of LOS.41 To account for potential clustering in the outcome variable (LOS), we first fitted a generalized linear mixed-effects model with a gamma distribution and log link to the entire dataset, incorporating a random intercept for each of the five clinical teams. From this null model, we estimated the intraclass correlation coefficient (ICC) to quantify between-team variability.42,43 Based on a cluster size of 5 and an ICC of 0.006, the design effect (design effect = 1 + [cluster size – 1] × ICC) was calculated as 1.24, indicating a minimal clustering effect.44,45,46 Given the negligible variance inflation we retained a parsimonious approach using traditional (nonclustered) regression methods for analyzing the primary endpoint.

In the regression methods, we first fit a gamma-distributed generalized linear model with a log link to estimate the overall association of the intervention with LOS across the full population. Next, we employed a bayesian 2-component GMM to account for potential subpopulations. The model, implemented in Stan via CmdStanPy in Python version 1.2.5 (Python Software Foundation), estimated mixing proportions, ordered intercepts, covariate effects, and shape parameters. The rate parameter in each gamma component was modeled as a function of the intercept and intervention (rate = exponential function [ − (intercept + β × intervention)]), where β is the regression coefficient representing the effect size of the intervention on the (log-transformed) expected LOS. Four chains with 6000 postwarmup samples each showed good convergence (R̂, approximately 1.0). Full model details are in eTable 5 in Supplement 1.

For the secondary effectiveness analysis with cost outcomes, to address the unequal group sizes in the control and intervention groups, we performed a propensity score matching analysis using the Python package PsmPy version 0.3.13.47 The differences in the 2 matched groups for LOS and direct costs were also assessed by calculating and plotting the cumulative bed-days and costs for all patients across each research group.48 To compare cumulative cost distributions over time between study groups, a nonparametric bootstrap method was used to estimate the difference in area under the cumulative cost curves and corresponding P values (α = .05). To minimize the influence of extreme outliers, isolated LOS values exceeding 120 days were truncated in cumulative calculations. The statistical analysis was performed using Python version 3.8.18 and R version 4.4.0 (R Project for Statistical Computing) packages.

Sample Size Estimation

A GMM with 2 components (K = 2) to a single variable (LOS; d = 1) requires the estimation of 5 parameters: the shape (α1 and α2) and scale (β1 and β2) parameters for each cluster, along with the mixing proportion (π) that determines the relative size of each cluster. A commonly used rule is that the minimum required sample size (N:q) values should be in the range of 10:1 (10 observations per 1 estimated parameter) to ensure stable parameter estimation.49,50 Given that our model has 5 parameters, the recommended minimum sample size was 50. Based on published literature,16 a moderate effect size (d = 0.5) was assumed for combined cardiac and pulmonary POCUS. Given α = .05, power = 0.80, and a 1:1 allocation ratio, the minimum required sample size was determined to be 64 patients per group (128 patients total). Due to the expected outliers and a long-tailed LOS distribution, the final sample size was an adequate representation of multimodal LOS patterns and extreme values.

Results

Reach Outcomes

Between December 7, 2023, and July 2, 2024, we recorded a total of 216 patient hospitalizations, comprising 114 in the control group and 102 in the POCUS group (Figure 1). After excluding 7 rehospitalizations in the control group and 1 rehospitalization in the POCUS group, 208 unique hospitalizations remained (median [IQR] age, 71 [59-80 years]; 121 female [58%]), including 107 in the control group and 101 in the POCUS group (Table 1). Of the 101 patients, POCUS was not recorded in 17 patients due to a complete transthoracic echocardiogram having already been performed (8 patients), transfer from another team (5 patients), physician or patient refusal (3 patients), or limitation in positioning the patient for POCUS assessment (1 patient). Of all patients, 125 (60%) were admitted due to congestive cardiac failure.

Table 1. Baseline Characteristics for POCUS and Control Groups for the Original Study Design.

Variable Participants, No. (%) P value
Overall (N = 208) POCUS (n = 101) Control (n = 107)
Age, median (IQR), y 71 (59-80) 70 (58-80) 73 (61-81) .56
Sex
Male 87 (42) 33 (33) 54 (51) .01
Female 121 (58) 68 (67) 53 (49)
Body mass index, median (IQR)a 29 (25-35) 28 (25-35) 30 (25-34) .49
Risk factors
Smoking 65 (31) 26 (26) 39 (36) .13
Diabetes 46 (22) 14 (14) 32 (30) .009
Hypertension 154 (74) 73 (72) 81 (76) .69
Past cardiac conditions
Congestive cardiac failure 125 (60) 58 (57) 67 (63) .53
Coronary artery disease 33 (16) 13 (13) 20 (19) .34
Valvular heart disease 26 (13) 10 (10) 16 (15) .40
Valve replacement 10 (5) 5 (5) 5 (5) >.99
Cardiac arrhythmia 57 (27) 13 (13) 44 (41) <.001
Pulmonary hypertension 18 (9) 4 (4) 14 (13) .03
Noncardiac conditions
Pulmonary or pulmonary vascular disease 54 (26) 12 (12) 42 (39) <.001
Stroke 20 (10) 4 (4) 16 (15) .009
Cancer (type, active or remission, and metastasis) 21 (10) 7 (7) 14 (13) .22
Chronic kidney disease stage 3 at diagnosis 65 (31) 32 (32) 33 (31) >.99
Diagnosis at discharge
Heart failure and cardiomyopathy 135 (65) 63 (62) 72 (67) .55
Coronary artery disease 16 (8) 8 (8) 8 (8) >.99
Pulmonary or pulmonary vascular disease 69 (33) 34 (34) 35 (33) >.99
Kidney disease 20 (10) 13 (13) 7 (7) .19
TTE findings
TTE performed 135 (65) 66 (65) 69 (65) >.99
Time to TTE 3 (2-4) 3 (2-4) 3 (2-5) .22
Valvular heart disease (moderate or severe) 35 (17) 19 (19) 16 (15) .39
Left ventricular ejection fraction <50% 51 (24) 26 (26) 25 (23) .06
Open in a new tab

Abbreviations: POCUS, point-of-care ultrasonography; TTE, transthoracic echocardiography.

a

Calculated as weight in kilograms divided by height in meters squared.

Effectiveness Outcomes

Primary Outcome—LOS

The initial exploratory analysis, using GMMs in both the control and intervention groups, identified 2 latent patient groups based on the distributional patterns of LOS, independent of predefined clinical characteristics (Figure 3 A and B).51 These 2 subgroups included a primary subgroup with a short LOS and a secondary subgroup with a longer LOS. These 2 subgroups were present in the control group (short LOS: median [IQR], 7.1 [4.6-10.2] days; long LOS: median [IQR], 36.1 [22.8-53.7] days), but the LOS was substantially shorter for these 2 subgroups in the POCUS group (short LOS: median [IQR], 5.1 [3.8-7.3] days; long LOS: median [IQR], 16.9 [12.2-22.6] days). The total fitted GMM, closely aligned with the observed data in both groups, demonstrating the model’s robustness. The 2-component model had the lowest Akaike information criterion for both groups (POCUS, 488.08; control, 539.12), and goodness-of-fit statistics indicated a reasonable fit for both models (eTable 3 and eTable 4 in Supplement 1).

Figure 3. Comparison of Length of Stay (LOS), Total Direct Cost, and Cumulative Bed-Days With Cost Analysis.

Figure 3.

Open in a new tab

A and B, The LOS distributions for the control and point-of-care ultrasonography (POCUS) groups, analyzed using a gamma mixture model. The brown and light blue components represent distinct subpopulations, while the light blue line shows the combined density, with the blue bars indicating actual data points. C and D, The association of LOS with direct hospitalization for the controls and POCUS groups. E and F, The cumulative LOS and the cumulative direct hospitalization costs for patients in both POCUS (dark blue) and standard of care (orange) groups. The lighter blue curve represents cost savings in panel E and bed-days saved in panel F.

A primary outcome analysis using gamma generalized linear model demonstrated that the intervention was associated with a 30.3% (95% CI, 5.5%-48.9%; P = .01) reduction in expected LOS compared with controls (Table 2), with mean (SD) LOS decreasing from 11.9 (7.5) days in controls to 8.3 (5.2) days in the intervention group. Several sensitivity analyses were subsequently performed. In the first step we excluded 17 patients for whom POCUS was not performed in the intervention group, and analysis using a gamma generalized linear model showed a 28.4% (95% CI, 2.9%-47.5%; P = .03) reduction in LOS for the POCUS group in comparison with the control group (eTable 6 in Supplement 1). The baseline differences in characteristics (sex, diabetes, arrhythmias, pulmonary hypertension, and pulmonary or pulmonary vascular disease) showed no significant univariable association with LOS in the gamma generalized linear model .

Table 2. Primary Generalized Linear Mixed-Effects Model Analysis of the Intervention and Secondary Comparison of LOS and Cost Distributions.
Outcome POCUS Control P value
LOS, mean (SD), d 8.3 (5.2) 11.9 (7.5) .01a
Cumulative LOS in each group, d 713 959 .008b
Cumulative total direct hospitalization cost for all patients, $ 2 442 945 3 194 482 .02b
Open in a new tab

Abbreviations: LOS, length of stay; POCUS, point-of-care ultrasonography.

a

Gamma generalized linear model with a log link was used to estimate the intervention effect size (POCUS, 101 individuals; control, 107 individuals). The intercept was 2.48 (SE = 0.10; P < .001), and the intervention was associated with a significant reduction in the outcome (estimate = –0.36; SE = 0.144; P = .01).

b

Secondary analysis (POCUS, 84 individuals; control, 84 individuals) using nonparametric bootstrapping to compare cumulative LOS and cost curves between groups, estimating area differences and 2-tailed P values.

Following our gamma generalized linear model, we further confirmed our analysis using a 2-component GMM for the LOS distribution, pooled from both the POCUS and control groups, with the performance of POCUS included as a covariate in the model (eTable 5 in Supplement 1). In the short-stay subgroup, POCUS was associated with a reduction in LOS of 1.16 days (90% credible interval, 0.03-2.31 days), and in the long-stay subgroup, a reduction of 14.30 days (90% credible interval, 3.86-30.05 days). These intervals represent central 90% credible intervals, corresponding to the fifth to 95th percentiles of the posterior distributions.

Secondary Outcome—Hospitalization Costs

Analysis of the matched patient groups (eTable 7 in Supplement 1) revealed a strong correlation between direct hospitalization cost and LOS in both the control group (r = 0.90; P < .001) and the POCUS group (r = 0.95; P < .001) (Figure 3C and D). The cumulative LOS and cost curves demonstrated that POCUS resulted in a total reduction of 246 hospital bed–days (959 vs 713 days in control vs POCUS groups; P = .008), corresponding to direct cost savings of $751 537 ($3 194 482 vs $2 442 944 in control vs POCUS groups; P = .02) (Figure 3E and F). Incremental cost-effectiveness ratio analysis revealed that POCUS reduced direct costs by $3055 per hospital bed–day saved.

Readmissions

A total of 35 patients (18%) were readmitted within 30 days, with no significant differences observed between the POCUS and control groups (19 patients [23%] vs 16 patients [19%]; P = .57). Readmissions within 90 days were lower in the POCUS group compared with the control group; however, this difference was not statistically significant (14 patients [17%] vs 19 patients [23%]; P = .33).

Adoption Outcomes

In the intervention group, only 17 POCUS evaluations (20%) were performed independently by hospitalists, while the remaining 67 (80%) were conducted by sonographers. Qualitative data from interviews with sonographers and hospitalists identified the following themes about barriers to the adoption of POCUS by hospitalists: (1) insufficient training in image acquisition and interpretation, (2) time constraints during rounds, and (3) a lack of incentivization to incorporate POCUS as a standard of care.

Implementation and Fidelity Outcomes

Fidelity was measured through standardized assessments of image interpretation, clinical integration, and decision-making using the POCUS image archive and medical record review. The findings on POCUS included left ventricular systolic dysfunction in 23 patients (27%), right ventricular dysfunction in 15 patients (18%), significant valvular heart disease in 31 patients (37%), dilated inferior vena cava in 7 patients (8%), and B-lines in 50 patients (60%). Upon reviewing medical records, the performance of POCUS and its findings were documented in the hospitalization notes of 74 patients (88%). A change in the medical decision was recorded in 30 cases (35%), including 12 cases (14%) where a new diagnosis was made, 6 cases (7%) where a change in medical therapy was made, and the remaining 12 cases (14%) where a normal POCUS examination excluded cardiopulmonary sources of dyspnea. The reviewer’s certainty regarding the change in the medical decision, as assessed on a 1 to 5 Likert scale (5 being most certain and 1 being least certain), was a median (IQR) of 4 (3-5) out of 5. The change in the medical decision following POCUS in 35% of patients, when compared with remaining patients in the POCUS group, was associated with a reduction in LOS (median [IQR], 4.5 [3.0-9.5] vs 7.0 [5.0-11.0] days; P = .04).

Discussion

This quality improvement study utilized a stepped-wedge cluster randomized design to assess the association of POCUS with hospital LOS and direct costs among patients hospitalized with undifferentiated dyspnea. The analysis produced 2 key findings. First, the introduction of POCUS was associated with shorter hospital stays and reduced direct costs. Second, despite the provision of dedicated training, only 20% of internal medicine physicians performed POCUS, highlighting the need for further strategies to enhance adoption and utilization.

Randomized clinical trials (RCTs) on the use of POCUS in general hospitalized patients are few and have revealed conflicting results. For example, a recent systematic review33 identified only 2 randomized cardiac trials16,31 that included LOS as their primary outcome. The first RCT 31 conducted at a teaching hospital in Chicago found that POCUS did not significantly reduce the length of hospital stay for unselected general medicine inpatients compared with standard care; however, subgroup analyses revealed that POCUS-guided care effectively reduced LOS for patients explicitly referred for heart failure. The second RCT16 evaluated the association of lung POCUS with the hospital LOS for patients admitted with a heart failure diagnosis, excluding those with acute coronary syndrome, respiratory conditions, or obesity. Our study differs from the 2 studies16,31 in that it used a cardiac and lung POCUS examination, employing a stepped-wedge randomized design, which can be tailored for efficiency and pragmatic (ie, practice-oriented) needs. We attempted to adjust for the right-skewed, plurimodal distribution of LOS and the distributional properties of its outliers that pose challenges in statistical analysis52 and may be responsible for the conflicting results in previous RCTs. Given that the LOS distribution was not amenable to conventional parametric models because it would violate normality and independence assumptions, a 2-component gamma mixture regression model was found to provide the best model fit to the data. Indeed, recent studies have highlighted similar benefits of using GMMs as a statistical approach; however, they remain underutilized.39,40 Incorporating GMMs in future multicenter RCTs could potentially improve the detection of treatment effects within specific patient subgroups.

Notably, 60% of admissions in our study were due to heart failure—a leading cause of hospitalizations and frequent readmissions that significantly contribute to the economic burden on health care systems—with total care costs projected to reach $160 billion by 2030.53,54 Our findings are therefore relevant and align with results from another prospective cohort study,30 which demonstrated that the availability and selective use of POCUS by hospitalist teaching teams was meaningfully associated with cost reductions, even though the LOS was not significantly different. In our study, the use of POCUS was associated with an overall cost savings of $751 537 (P = .02), highlighting its potential to reduce the economic burden and improve health care efficiency.

The implementation of POCUS was challenging in our study with only 20% of internal medicine physicians consistently performed POCUS. Interestingly, the barriers identified in our study closely mirror those reported in a previous multicenter survey of hospital-based internists,55 which highlighted lack of training and time constraints during rounds as key obstacles to POCUS use. Personal attitudes about the utility of POCUS, the current lack of requirements by external organizations encouraging POCUS being integral to the practice of internal medicine, and the inability to bill were not considered important barriers by our participants. Our data suggest the need for further education and training, development of incentives, and continued collaboration between sonographers and expert physicians as potential strategies for implementing the use of POCUS in hospital medicine.

Limitations

This study is subject to certain limitations. This implementation study was conducted at a single tertiary academic center using trained sonographers and remote cardiologist interpretation—a model that may be difficult to scale in typical inpatient settings. The relatively low independent adoption rate among internists (20%) highlights the need to explore more scalable and generalizable implementation models.

From an analytical standpoint, we adopted a parsimonious approach and acknowledge that any residual clustering may not have been fully accounted for. Moreover, propensity score matching, although not ideal for clustered designs, was used secondarily to adjust for residual confounding and balance groups for cost estimation. Furthermore, although we employed a 2-component GMM and observed differences in the LOS for the patients with longer stay, the exact mechanism of LOS reduction and potential confounders in this cohort could not be accounted for. We have made the codes used for our pooled GMM analysis available.56 Future studies should further investigate this observation and consider more advanced methods that incorporate mixed-effects 2-component GMMs, as well as adopt larger, multicenter designs.

Additionally, the study did not assess follow-up outcomes, use of guideline-directed therapies, or advanced heart failure care such as rehabilitation or palliative services. The adjudication of POCUS impact was based on an unblinded retrospective medical record review, introducing potential for bias and confounding by indication. Future research should address additional clinical end points such as diagnostic accuracy, interrater reliability, time to intervention, morbidity, and patient-centered outcomes.

Conclusions

In our study, POCUS was associated with reduced hospital stays and lower costs, suggesting that it could enhance the value of care and improve triage and treatment of patients with undiagnosed dyspnea. These results support wider adoption of POCUS in hospitals to optimize patient outcomes and lower hospitalization costs. Larger multicenter trials are required to validate these findings. Given budget constraints, policymakers should evaluate the cost-effectiveness of integrating POCUS into standard care for patients with heart failure because such policies could substantially reduce morbidity and mortality.

Supplement 1.

eTable 1. RE-AIM framework to assess the effectiveness of POCUS implementation

eTable 2. Models for fitting length of stay in the overall population

eTable 3. Comparing gamma mixture models with 1, 2, and 3 components for the length of stay using the Akaike information criterion (AIC)

eTable 4. Gamma mixture model parameters for fitting length of stay in POCUS and control groups, and model evaluation

eTable 5. Two component gamma mixture model to assess the impact of intervention impact on the latent subgroups

eTable 6. Gamma generalized linear model for sensitivity analysis of length of stay, excluding 17 ITT patients in the intervention arm who did not receive POCUS

eTable 7. Matched baseline characteristics for POCUS and control groups

jamanetwopen-e2530677-s001.pdf (293.9KB, pdf)
Supplement 2.

Data Sharing Statement

jamanetwopen-e2530677-s002.pdf (13.6KB, pdf)

References

  • 1.Stevens JP, Dechen T, Schwartzstein R, et al. Prevalence of dyspnea among hospitalized patients at the time of admission. J Pain Symptom Manage. 2018;56(1):15-22.e2. doi: 10.1016/j.jpainsymman.2018.02.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Banzett RBB, Sheridan AR, Baker KM, Lansing RW, Stevens JP. ‘Scared to death’ dyspnoea from the hospitalised patient’s perspective. BMJ Open Respir Res. 2020;7(1):e000493. doi: 10.1136/bmjresp-2019-000493 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Abidov A, Rozanski A, Hachamovitch R, et al. Prognostic significance of dyspnea in patients referred for cardiac stress testing. N Engl J Med. 2005;353(18):1889-1898. doi: 10.1056/NEJMoa042741 [DOI] [PubMed] [Google Scholar]
  • 4.Nishimura K, Izumi T, Tsukino M, Oga T. Dyspnea is a better predictor of 5-year survival than airway obstruction in patients with COPD. Chest. 2002;121(5):1434-1440. doi: 10.1378/chest.121.5.1434 [DOI] [PubMed] [Google Scholar]
  • 5.Stevens JP, Baker K, Howell MD, Banzett RB. Prevalence and predictive value of dyspnea ratings in hospitalized patients: pilot studies. PLoS One. 2016;11(4):e0152601. doi: 10.1371/journal.pone.0152601 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Sunjaya AP, Homaira N, Corcoran K, Martin A, Berend N, Jenkins C. Assessment and diagnosis of chronic dyspnoea: a literature review. NPJ Prim Care Respir Med. 2022;32(1):10. doi: 10.1038/s41533-022-00271-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Arts L, Lim EHT, van de Ven PM, Heunks L, Tuinman PR. The diagnostic accuracy of lung auscultation in adult patients with acute pulmonary pathologies: a meta-analysis. Sci Rep. 2020;10(1):7347. doi: 10.1038/s41598-020-64405-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Staub LJ, Mazzali Biscaro RR, Kaszubowski E, Maurici R. Lung ultrasound for the emergency diagnosis of pneumonia, acute heart failure, and exacerbations of chronic obstructive pulmonary disease/asthma in adults: a systematic review and meta-analysis. J Emerg Med. 2019;56(1):53-69. doi: 10.1016/j.jemermed.2018.09.009 [DOI] [PubMed] [Google Scholar]
  • 9.Chiu L, Jairam MP, Chow R, et al. Meta-analysis of point-of-care lung ultrasonography versus chest radiography in adults with symptoms of acute decompensated heart failure. Am J Cardiol. 2022;174:89-95. doi: 10.1016/j.amjcard.2022.03.022 [DOI] [PubMed] [Google Scholar]
  • 10.Shen-Wagner J, Deutchman M. Point-of-care ultrasound: a practical guide for primary care. Fam Pract Manag. 2020;27(6):33-40. [PubMed] [Google Scholar]
  • 11.Soni NJ, Schnobrich D, Mathews BK, et al. Point-of-care ultrasound for hospitalists: a position statement of the Society of Hospital Medicine. J Hosp Med. 2019;14(1):E1-E6. doi: 10.12788/jhm.3079 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Mehta M, Jacobson T, Peters D, et al. Handheld ultrasound versus physical examination in patients referred for transthoracic echocardiography for a suspected cardiac condition. JACC Cardiovasc Imaging. 2014;7(10):983-990. doi: 10.1016/j.jcmg.2014.05.011 [DOI] [PubMed] [Google Scholar]
  • 13.Al Deeb M, Barbic S, Featherstone R, Dankoff J, Barbic D. Point-of-care ultrasonography for the diagnosis of acute cardiogenic pulmonary edema in patients presenting with acute dyspnea: a systematic review and meta-analysis. Acad Emerg Med. 2014;21(8):843-852. doi: 10.1111/acem.12435 [DOI] [PubMed] [Google Scholar]
  • 14.Ramsingh D, Runyon A, Gatling J, et al. Improved diagnostic accuracy of pathology with the implementation of a perioperative point-of-care ultrasound service: quality improvement initiative. Reg Anesth Pain Med. 2020;45(2):95-101. doi: 10.1136/rapm-2019-100632 [DOI] [PubMed] [Google Scholar]
  • 15.Ramsingh D, Singh S, Canales C, et al. The evaluation point-of-care ultrasound in the post-anesthesia unit-a multicenter prospective observational study. J Clin Med. 2021;10(11):2389. doi: 10.3390/jcm10112389 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Mozzini C, Di Dio Perna M, Pesce G, et al. Lung ultrasound in internal medicine efficiently drives the management of patients with heart failure and speeds up the discharge time. Intern Emerg Med. 2018;13(1):27-33. doi: 10.1007/s11739-017-1738-1 [DOI] [PubMed] [Google Scholar]
  • 17.Zanobetti M, Scorpiniti M, Gigli C, et al. Point-of-care ultrasonography for evaluation of acute dyspnea in the ED. Chest. 2017;151(6):1295-1301. doi: 10.1016/j.chest.2017.02.003 [DOI] [PubMed] [Google Scholar]
  • 18.Guazzi M. Alveolar-capillary membrane dysfunction in heart failure: evidence of a pathophysiologic role. Chest. 2003;124(3):1090-1102. doi: 10.1378/chest.124.3.1090 [DOI] [PubMed] [Google Scholar]
  • 19.Martindale JL, Wakai A, Collins SP, et al. Diagnosing acute heart failure in the emergency department: a systematic review and meta-analysis. Acad Emerg Med. 2016;23(3):223-242. doi: 10.1111/acem.12878 [DOI] [PubMed] [Google Scholar]
  • 20.Picano E, Frassi F, Agricola E, Gligorova S, Gargani L, Mottola G. Ultrasound lung comets: a clinically useful sign of extravascular lung water. J Am Soc Echocardiogr. 2006;19(3):356-363. doi: 10.1016/j.echo.2005.05.019 [DOI] [PubMed] [Google Scholar]
  • 21.Pivetta E, Goffi A, Lupia E, et al. ; SIMEU Group for Lung Ultrasound in the Emergency Department in Piedmont . Lung ultrasound-implemented diagnosis of acute decompensated heart failure in the ED: a SIMEU multicenter study. Chest. 2015;148(1):202-210. doi: 10.1378/chest.14-2608 [DOI] [PubMed] [Google Scholar]
  • 22.Volpicelli G, Elbarbary M, Blaivas M, et al. ; International Liaison Committee on Lung Ultrasound (ILC-LUS) for International Consensus Conference on Lung Ultrasound (ICC-LUS) . International evidence-based recommendations for point-of-care lung ultrasound. Intensive Care Med. 2012;38(4):577-591. doi: 10.1007/s00134-012-2513-4 [DOI] [PubMed] [Google Scholar]
  • 23.Howard ZD, Noble VE, Marill KA, et al. Bedside ultrasound maximizes patient satisfaction. J Emerg Med. 2014;46(1):46-53. doi: 10.1016/j.jemermed.2013.05.044 [DOI] [PubMed] [Google Scholar]
  • 24.Melgarejo S, Schaub A, Noble VE. Point of care ultrasound: an overview. American College of Cardiology. Published October 31, 2017. Accessed July 30, 2025. https://www.acc.org/Latest-in-Cardiology/Articles/2017/10/31/09/57/Point-of-Care-Ultrasound
  • 25.Kirkpatrick JN, Grimm R, Johri AM, et al. Recommendations for echocardiography laboratories participating in cardiac point of care cardiac ultrasound (POCUS) and critical care echocardiography training: report from the American Society of Echocardiography. J Am Soc Echocardiogr. 2020;33(4):409-422.e4. doi: 10.1016/j.echo.2020.01.008 [DOI] [PubMed] [Google Scholar]
  • 26.Huang GS, Alviar CL, Wiley BM, Kwon Y. The era of point-of-care ultrasound has arrived: are cardiologists ready? Am J Cardiol. 2020;132:173-175. doi: 10.1016/j.amjcard.2020.06.062 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Liu L, Pattock A, Kersey C, et al. Cardiologist perspectives on cardiac point-of-care ultrasound. Abstract presented at: 2021 American Heart Association Scientific Sessions; November 13 to 15, 2021; Boston, Massachusetts. Accessed July 30, 2025. https://www.ahajournals.org/doi/10.1161/circ.144.suppl_1.12653 [Google Scholar]
  • 28.Conner S, Chia D, Lalani F, et al. Minding the gap (s): hospitalists experience aspirational, safety, and knowledge deficits that prevent them from practicing POCUS. POCUS J. 2019;4(2):27-32. doi: 10.24908/pocus.v4i2.13693 [DOI] [Google Scholar]
  • 29.Thomas MK, Conner SM, Maw A, Soni NJ. Point-counterpoint: should point-of-care ultrasound be a required skill of hospitalists? J Hosp Med. 2023;18(12):1150-1155. doi: 10.1002/jhm.13208 [DOI] [PubMed] [Google Scholar]
  • 30.Tierney DM, Rosborough TK, Sipsey LM, et al. Association of internal medicine point of care ultrasound (POCUS) with length of stay, hospitalization costs, and formal imaging: a prospective cohort study. POCUS J. 2023;8(2):184-192. doi: 10.24908/pocus.v8i2.16791 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Lucas BP, Candotti C, Margeta B, et al. Hand-carried echocardiography by hospitalists: a randomized trial. Am J Med. 2011;124(8):766-774. doi: 10.1016/j.amjmed.2011.03.029 [DOI] [PubMed] [Google Scholar]
  • 32.Szabó GV, Szigetváry C, Szabó L, et al. Point-of-care ultrasound improves clinical outcomes in patients with acute onset dyspnea: a systematic review and meta-analysis. Intern Emerg Med. 2023;18(2):639-653. doi: 10.1007/s11739-022-03126-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Cid-Serra X, Hoang W, El-Ansary D, Canty D, Royse A, Royse C. Clinical impact of point-of-care ultrasound in internal medicine inpatients: a systematic review. Ultrasound Med Biol. 2022;48(2):170-179. doi: 10.1016/j.ultrasmedbio.2021.09.013 [DOI] [PubMed] [Google Scholar]
  • 34.Ogrinc G, Mooney SE, Estrada C, et al. The SQUIRE (standards for quality improvement reporting excellence) guidelines for quality improvement reporting: explanation and elaboration. Qual Saf Health Care. 2008;17(Suppl_1)(suppl 1):i13-i32. doi: 10.1136/qshc.2008.029058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Curran GM, Bauer M, Mittman B, Pyne JM, Stetler C. Effectiveness-implementation hybrid designs: combining elements of clinical effectiveness and implementation research to enhance public health impact. Med Care. 2012;50(3):217-226. doi: 10.1097/MLR.0b013e3182408812 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Harden SM, Smith ML, Ory MG, Smith-Ray RL, Estabrooks PA, Glasgow RE. RE-AIM in clinical, community, and corporate settings: perspectives, strategies, and recommendations to enhance public health impact. Front Public Health. 2018;6:71. doi: 10.3389/fpubh.2018.00071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.New Jersey Medicaid Management Information System (NJMMIS). Hospital information: inpatient cost to charge ratio—history, Robert Wood Johnson University Hospital. Accessed 18 February, 2025. https://www.njmmis.com/costratioinpatient.aspx
  • 38.Cornes MW, Aldridge R, Biswell E, et al. Improving care transfers for homeless patients after hospital discharge: a realist evaluation. Health Serv Deliv Res. 2021;9(17):1-186. doi: 10.3310/hsdr09170 [DOI] [Google Scholar]
  • 39.Williford E, Haley V, McNutt LA, Lazariu V. Dealing with highly skewed hospital length of stay distributions: the use of Gamma mixture models to study delivery hospitalizations. PLoS One. 2020;15(4):e0231825. doi: 10.1371/journal.pone.0231825 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Venturini S, Dominici F, Parmigiani G. Gamma shape mixtures for heavy-tailed distributions. Ann Appl Stat. 2008;2(2):756-776. doi: 10.1214/07-AOAS156 [DOI] [Google Scholar]
  • 41.Soares I, Carneiro AV. Intention-to-treat analysis in clinical trials: principles and practical importance. Rev Port Cardiol. 2002;21(10):1191-1198. [PubMed] [Google Scholar]
  • 42.Li F, Hughes JP, Hemming K, Taljaard M, Melnick ER, Heagerty PJ. Mixed-effects models for the design and analysis of stepped wedge cluster randomized trials: an overview. Stat Methods Med Res. 2021;30(2):612-639. doi: 10.1177/0962280220932962 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Wedel M, DeSarbo WS. A mixture likelihood approach for generalized linear models. J Classif. 1995;12(1):21-55. doi: 10.1007/BF01202266 [DOI] [Google Scholar]
  • 44.Sommet N, Morselli D. Keep calm and learn multilevel linear modeling: a three-step procedure using SPSS, Stata, R, and Mplus. Rev Int Psychol Soc. 2021;34(1):1-19. doi: 10.5334/irsp.555 [DOI] [Google Scholar]
  • 45.Muthén BO, Satorra A. Complex sample data in structural equation modeling. Sociol Methodol. 1995;25(1995):267-316. doi: 10.2307/271070 [DOI] [Google Scholar]
  • 46.Peugh JL. A practical guide to multilevel modeling. J Sch Psychol. 2010;48(1):85-112. doi: 10.1016/j.jsp.2009.09.002 [DOI] [PubMed] [Google Scholar]
  • 47.Chang TH, Stuart EA. Propensity score methods for observational studies with clustered data: a review. Stat Med. 2022;41(18):3612-3626. doi: 10.1002/sim.9437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Mustonen E, Hörhammer I, Absetz P, et al. Eight-year post-trial follow-up of health care and long-term care costs of tele-based health coaching. Health Serv Res. 2020;55(2):211-217. doi: 10.1111/1475-6773.13251 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Kline RB. Principles and Practice of Structural Equation Modeling. Guilford Publications; 2023. [Google Scholar]
  • 50.Memon MA, Ting H, Cheah J-H, Thurasamy R, Chuah F, Cham TH. Sample size for survey research: Review and recommendations. J Appl Struct Equation Modeling. 2020;4(2):1-9. doi: 10.47263/JASEM.4(2)01 [DOI] [Google Scholar]
  • 51.Sinha P, Calfee CS, Delucchi KL. Practitioner’s guide to latent class analysis: methodological considerations and common pitfalls. Crit Care Med. 2021;49(1):e63-e79. doi: 10.1097/CCM.0000000000004710 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Atienza N, García-Heras J, Muñoz-Pichardo JM, Villa R. An application of mixture distributions in modelization of length of hospital stay. Stat Med. 2008;27(9):1403-1420. doi: 10.1002/sim.3029 [DOI] [PubMed] [Google Scholar]
  • 53.Virani SS, Alonso A, Benjamin EJ, et al. ; American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee . Heart disease and stroke statistics-2020 update: a report from the American Heart Association. Circulation. 2020;141(9):e139-e596. doi: 10.1161/CIR.0000000000000757 [DOI] [PubMed] [Google Scholar]
  • 54.Wei C, Heidenreich PA, Sandhu AT. The economics of heart failure care. Prog Cardiovasc Dis. 2024;82:90-101. doi: 10.1016/j.pcad.2024.01.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Wong J, Montague S, Wallace P, et al. Barriers to learning and using point-of-care ultrasound: a survey of practicing internists in six North American institutions. Ultrasound J. 2020;12(1):19. doi: 10.1186/s13089-020-00167-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.POCUS-CARE. Github. Accessed August 7, 2025. https://github.com/rutgers-rwjms-innov-center/POCUS-CARE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eTable 1. RE-AIM framework to assess the effectiveness of POCUS implementation

eTable 2. Models for fitting length of stay in the overall population

eTable 3. Comparing gamma mixture models with 1, 2, and 3 components for the length of stay using the Akaike information criterion (AIC)

eTable 4. Gamma mixture model parameters for fitting length of stay in POCUS and control groups, and model evaluation

eTable 5. Two component gamma mixture model to assess the impact of intervention impact on the latent subgroups

eTable 6. Gamma generalized linear model for sensitivity analysis of length of stay, excluding 17 ITT patients in the intervention arm who did not receive POCUS

eTable 7. Matched baseline characteristics for POCUS and control groups

jamanetwopen-e2530677-s001.pdf (293.9KB, pdf)
Supplement 2.

Data Sharing Statement

jamanetwopen-e2530677-s002.pdf (13.6KB, pdf)

Articles from JAMA Network Open are provided here courtesy of American Medical Association

RESOURCES