Contribution of point-of-care ultrasound in the prehospital management of patients with non-trauma acute dyspnea: a systematic review and meta-analysis

Abstract

Acute dyspnea is a common symptom whose management is challenging in prehospital settings. Point-of-care ultrasound (POCUS) is increasingly accessible because of device miniaturization. To assess the contribution of POCUS in the prehospital management of patients with acute nontraumatic dyspnea, we performed a systematic review on nontrauma patients of any age managed in the prehospital setting for acute dyspnea and receiving a POCUS examination. We searched seven databases and gray literature for English-language studies published from January 1995 to November 2023. Two independent reviewers completed the study selection, data extraction, and risk of bias assessment. The primary outcome was the assessment of the contribution of POCUS to feasibility, diagnostic, therapeutic, prognosis, patient referral, and transport vector modification. Twenty-three studies were included. The risk of bias assessment identified 3 intermediate-risk, 18 serious-risk, and 2 critical-risk studies. Three studies reported moderate to excellent feasibility for lung POCUS, and three studies reported poor to mediocre feasibility for cardiac POCUS. The median duration of the POCUS examination was less than 5 minutes (six studies). POCUS improved diagnostic identification (seven studies). The diagnostic accuracy of POCUS was excellent for pneumothorax (sensitivity = 100%, specificity = 100%, two studies), very good for acute heart failure (sensitivity = 71–100%, specificity = 72–95%, eight studies), good for pneumonia (sensitivity = 88%, specificity = 59%, one study), and moderate for pleural effusion (sensitivity = 26–53%, specificity = 83–92%, two studies). Treatment was modified in 11 to 54% of the patients (seven studies). POCUS had no significant effect on patient prognosis (two studies). POCUS contributed to patient referrals and transport vectors in 51% (four studies) and 25% (three studies) of patients, respectively. The evidence supports the use of POCUS for managing acute nontraumatic dyspnea in the prehospital setting in terms of feasibility, overall diagnostic contribution, and, particularly, lung ultrasound for acute heart failure diagnosis. Moreover, POCUS seems to have a therapeutic contribution. There is not enough evidence supporting the use of POCUS for pneumonia, pleural effusion, pneumothorax, chronic obstructive pulmonary disease, or asthma exacerbation diagnosis, nor does it support prognostic, patient referral, and transport vector contribution. A high level of evidence is lacking and needed.

Introduction

Acute dyspnea is a common symptom whose management is challenging in the prehospital setting [1]. In trauma patients, a specific context (thoracic or severe trauma) and precise causes enable management to be guided by explicit guidelines [1]. In nontrauma patients, the etiologies are diverse and include acute heart failure (AHF), pneumonia (with or without pleural effusion), chronic bronchitis or asthma exacerbation, and, to a lesser extent, pneumothorax [1]. These presentations often combine with a clinical presentation of cardiorespiratory decompensation, particularly in elderly patients, making the precise etiological diagnosis difficult [1,2]. However, accurate and early diagnosis has been shown to be a prognostic factor [1]. Point-of-care ultrasound (POCUS) is becoming increasingly accessible, especially in prehospital care, due to the miniaturization of devices and the emergence of portable and ultraportable ultrasound scanners [3]. POCUS is performed at the patient’s bedside using simple decisive criteria based on relevant guidelines and has been validated in the diagnosis, treatment, and referral of acute dyspnea, particularly for AHF [4]. Indeed, the POCUS-derived overall lung ultrasound (LUS) and cardiac ultrasound (CUS) learning curves are excellent, their feasibility seems high [5], and they have excellent performance in the in-hospital management of dyspnea [6]. Recent updates to AHF, pneumonia, and pleural effusion management guidelines place greater emphasis on the use of POCUS for accurate and early diagnosis and to target early management [1]. However, given the difficulty in conducting controlled and randomized studies with large sample sizes, there is a lack of evidence for the prehospital setting, where studies are mainly observational and have small sample sizes. No qualitative or quantitative synthesis of data in the prehospital setting has been published to date. The main aim of this study was to address this gap by conducting a systematic review to assess the contribution of POCUS in the prehospital management of patients presenting with acute nontraumatic dyspnea.

Methods

The study protocol was registered with the PROSPERO International Prospective Register of Systematic Reviews (CRD42022361127) and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [7].

Eligibility criteria

The study selection was defined on the basis of the PICOS statement [8], and addressed the following question:

1. Population: in patients (or models simulating patients) with acute nontraumatic dyspnea (determined by the absence of reported trauma) managed in the prehospital setting (i.e. any medical care provided before hospital arrival);

2. Intervention: undergoing POCUS assessment by trained physicians, paramedics, or nurses;

3. Comparator: compared with a reference standard;

4. Outcomes: what is POCUS feasibility (i.e. ability to perform and interpret POCUS) and its contributions to diagnosis (i.e. its impact on diagnosis accuracy), therapeutic outcomes (i.e. its impact on treatment decisions), prognosis (i.e. its impact on patient outcomes), patient referral (i.e. its impact on hospital admission decisions), and transport vector modification (i.e. its changes in mode or destination of patient transport following POCUS use);

5. Studies: in observational (prospective or retrospective) and randomized studies.

We included all English-language randomized controlled trials and observational studies involving patients with acute nontraumatic dyspnea managed in the prehospital setting and undergoing POCUS that were published in full-text from January 1, 1995 to November 28, 2023. We excluded case reports, case series, review articles, editorials, and expert opinions.

Search strategy

A comprehensive search was conducted across the following databases from January 1995 to November 28, 2023: Medline, Embase, PubMed, the Cochrane Library, BASE, ABES, and Google Scholar (Table S1, Supplemental Digital Content 1, http://links.lww.com/EJEM/A475). The search was conducted by a registered librarian (F.C.) and peer-reviewed by a second specialist, following the Peer Review of Electronic Search Strategies guidelines [8].

Study selection

Studies were screened via Rayyan software (http://rayyan.qcri.org). Titles were imported directly from the search database, and duplicates were removed. Two reviewers (O.T. and J.S.) were blinded and independently screened titles and abstracts to identify relevant studies and then assessed the full texts of relevant studies for inclusion. Disagreements were resolved by consensus with a third reviewer (T.M.) (Fig. 1). Systematic reviews were excluded, but the review of references led to the inclusion of additional studies. The primary outcome was the contribution of POCUS to feasibility, diagnosis, therapeutic, prognosis, patient referral, and transport vector modification. Studies were included in the meta-analysis if they contained sufficient and comparable quantitative data. Specifically, for diagnostic test meta-analysis, a 2 × 2 table of true positives, false positives, true negatives, and false negatives, a homogeneous design, index test, and comparator, which was the reference standard, was used.

Fig. 1.

PRISMA 2020 flow diagram of the POCUS-PH-DYSPNEA Study.

Data collection and processing

Two reviewers (O.T. and J.S.) were blinded and independently extracted data from the included studies for qualitative synthesis (Supplementary Table S2, Supplemental digital content 1, http://links.lww.com/EJEM/A475). These data were either extracted directly from the studies or calculated from other reported data. If these values could not be obtained, the authors were contacted directly. Studies were excluded if the corresponding author did not respond after 3 attempts.

Risk of bias assessment

Two reviewers (O.T. and J.S.) independently assessed the risk of bias (ROB). Nonrandomized studies were evaluated via the ‘Cochrane Risk Of Bias In Nonrandomized Studies—of Interventions’ (ROBINS-I) tool, and randomized controlled trials were assessed via the ‘Cochrane Risk of Bias Tool for Randomized Controlled Trials’ (RoB-2). The results of the assessment were synthesized via the ‘robvis’ tool (Fig. 2) [9,10]. All studies were also evaluated via the ‘Grading of Recommendations Assessment, Development and Evaluation’ (GRADE) scale [11]. For diagnostic accuracy studies, the ROB was also assessed via the ’Quality Assessment of Diagnostic Accuracy Studies-2’ (QUADAS-2) tool (Supplementary Table S3, Supplemental digital content 1, http://links.lww.com/EJEM/A475) [12].

Fig. 2.

Cochrane risk of bias In nonrandomized studies—of interventions (a) and Cochrane Risk of Bias Tool for randomized controlled trials (b) assessment results.

Evidence synthesis

For the qualitative synthesis, feasibility, diagnostic, therapeutic, prognosis, referral, and vector contribution were reported in qualitative thematic synthesis via simple descriptive statistics. Quantitative synthesis (meta-analysis) was performed if studies met the following criteria: (1) quantitative description of intervention; (2) comparable intervention assessment; and (3) complete data availability. If quantitative synthesis was feasible, individual study results were evaluated graphically with plotted sensitivity and specificity estimates on one-dimensional forest plots and the receiver operating characteristic space to visually assess heterogeneity. For a diagnostic test accuracy meta-analysis, for each study, 2 × 2 tables containing the number of true positives, true negatives, false positives, and false negatives were made according to the thresholds specified for the scales found. Forest plots (with 95% confidence intervals) of these thresholds, obtained via Review Manager (RevMan) Version 5.4.1, The Cochrane Collaboration, 2020, provided visual representations of the sensitivities and specificities reported in each study [8]. In a meta-analysis of diagnostic test accuracy studies, a pair of two outcome measures, such as sensitivity and specificity, must be analyzed simultaneously. However, they are generally inversely correlated and could be affected by a threshold effect. Therefore, separate pooling and summary points alone should be avoided and a summary line, such as a summary receiver operating characteristic curve, to show how the different sensitivities and specificities of primary studies are related to each other should be constructed instead. The summary receiver operating characteristic approach removes such effects, combined with the bivariate approach, incorporating any correlation between sensitivity and specificity via a random effects approach [13,14].

Results

Study selection

After the databases were queried, 3173 studies were retrieved. There were 70 duplicates. Selection on the basis of study titles and abstracts identified 62 studies for full-text review. Ten systematic reviews were excluded, and the review of references led to the inclusion of two additional studies. Twenty-three studies were included and extracted via qualitative synthesis (Fig. 1). Seven studies were included in the meta-analysis (Fig. 3) [15–37].

Fig. 3.

Point-of-care ultrasound contribution for management of patients with acute dyspnea in the prehospital setting: meta-analysis for diagnostic accuracy of lung ultrasound in acute heart failure. Forest plots of specificity (a) and sensitivity (b) for studies, and summary receiver operating characteristic curve for studies included in meta-analysis (c). (a) and (b): Study point estimates presented proportional to size. CI, confidence interval; FN, false negatives; FP, false positives; ROC, receiver operating characteristic; SROC, summary receiving operator characteristics; TN, true negatives; TP, true positives.

Characteristics of the studies

There was one randomized [28] and 22 observational studies, among which 5/23 (22%) had a control group (where POCUS was not performed) and all were prospective, often with several outcomes involving feasibility (n = 20), diagnostic (n = 15), therapeutic (n = 8), prognosis (n = 2), referral (n = 4), and change in vector contributions (n = 3) (Table 1). One study included a human lung model [30]. Scans were performed onsite (n = 19 studies), during ground transportation (n = 16), during air transport (n = 2), or at the office (n = 2) (Supplementary Table S4 and S5, Supplemental digital content 1, http://links.lww.com/EJEM/A475).

Table 1.

Contribution of point-of-care ultrasound in the management of patients with nontrauma acute dyspnea in the prehospital setting: study characteristics and outcomes

Author, year, country	Design	N a	Setting	Sonographer, Nb	Intervention	Reference standard	Outcome(s) (contributive or notc)
Russel, 2023, USA [15]	Observational, prospective, comparatived	353	On-site, ground transport	Paramedics N = 26	LUS	Review: discharge diagnosis	Diagnostic A (yes): AHF Therapeutic (maybe): AHF treatment, time to treatment
Kowalczyk, 2023, Poland [16]	Observational, prospective	44	On-site, ground transport	Paramedics N = 1	LUS	Review: discharge diagnosis	Feasibility (yes): interpretation, IRA
Gundersen, 2023, Denmark [17]	Observational, prospective, comparative	214	On-site	Physicians N = 40	LUS, CUS	Review: adjudication	Diagnostic A (yes): AHF, LVEF, PE, COPD-AE
Donovan, 2022, Australia [18]	Observational, prospective	95	Ground transport	Paramedics N = 44	LUS	Assessment: LUS (1 expert)	Diagnostic I (maybe): overall, AHF, pneumonia, PE, PNO Feasibility (maybe): interpretation, duration, quality, IRA Therapeutic (maybe): modification
Hermann, 2022, Austria [19]	Observational, prospective	24	On-site	Physicians N = 4	LUS, CUS	Review: real-time (1 expert)	Feasibility (maybe): time, quality, transmission, communication
Rodríguez-Contreras, 2022, Spain [20]	Observational, prospective, multicenter	82	At the office	Physicians N = 28	LUS	Assessment: CXR (radiologist)	Feasibility (maybe): interpretation, time Diagnostic A (yes): pneumonia
Ienghong, 2022, Thailand [21]	Observational, prospective, cross-sectional	169	On-site, ground transport	Physicians NR	LUS, CUS	Review: adjudication	Feasibility (maybe): IRA (overall, pneumonia, AHF) Diagnostic I (yes): overall, AHF, pneumonia; diagnostic A (yes): AHF
Pietersen, 2021, Denmark [22]	Observational, prospective,e quality control	NR	On-site, ground transport	Paramedics N = 100	LUS	Review: expert	Feasibility (maybe): interpretation, duration, quality, IRA (normal, PNO, interstitial, PE) Diagnostic I (maybe): PE Therapeutic (maybe): change, number needed to scan
Schoeneck, 2021, USA [23]	Observational, prospective	69	On-site, ground transport	Paramedics N = 22	LUS	Review: discharge diagnosis	Feasibility (maybe): interpretation, duration, quality, IRA Diagnostic A (yes): AHF
Nadim, 2021, Denmark [24]	Observational, prospective	41	On-site, ground transport	Paramedics N = 100	LUS	Review: expert	Feasibility (maybe): interpretation, time on site Referral (maybe): hospital admission Vector (maybe): renewed ambulance
Becker, 2018, USA [25]	Observational, prospective	34	On-site, ground transport	Paramedics N = 17	LUS	Adjudication	Feasibility (maybe): interpretation, quality, transmission, intraRA, IRA (overall, ED, expert)
Scharonow, 2018, Germany [26]	Observational, prospective, comparative	38	On-site, ground transport	Physicians NR	LUS, FAST	Review: discharge diagnosis	Feasibility (maybe): IRA (overall) Diagnostic A (maybe): AHF, LVEF, right ventricle stress, lung interstitial syndrome, PNO, abdominal fluid Therapeutic (maybe): modification Referral (maybe): modification Vector (maybe): transport priority, monitoring requirement
Zanatta, 2018, Italy [27]	Observational, prospective, comparative, case‒controlled	60	On-site, ground transport	Physicians NR	LUS	Review: discharge diagnosis	Diagnostic A (maybe): AHF, PE Therapeutic (yes): modification, more appropriate, furosemide dose reduction; intubation Referral (maybe): hospitalization, destination, transport delay Prognosis (yes): prediction of clinical, laboratory, hospitalization data
Strnad, 2016, Slovenia [28]	RCT	20	On-site, ground transport	Physicians N = 2	LUS	Unclear	Feasibility (yes): POCUS topography Therapeutic (yes): aim/threshold (oxygen), monitoring (B-lines)
Laursen, 2016, Denmark [29]	Observational, prospective, cross-sectional	40	On-site, ground transport	Physician N = 12	LUS	Review: adjudication	Feasibility (maybe): interpretation, time, IRA (AHF) Diagnostic I (yes): AHF Diagnostic A (yes): AHF
Charron, 2015, France [30]	Observational, prospective	100	On-site, ground transport	Physicians N = 9	CUS	Review: adjudication	Feasibility (maybe): quality, IRA (LVEF, RV, pericardia effusion, IVC) Diagnostic I (maybe): overall
Neesse, 2012, Germany [31]	Observational, prospective	62	On-site, ground transport	Physicians N = 4	LUS, CUS	Review: discharge diagnosis	Feasibility (maybe): interpretation, time, utility Diagnosis I (overall) (maybe) Therapeutic (maybe): modification, modification proportion, delay Referral (maybe) Vector (maybe): delay on transport
Lyon, 2012, USA [32]	Observational, prospective	NA	Helicopter transport	Physicians N = 4	LUS	Other	Feasibility (unconclusive) Diagnostic A (yes): PNO
Prosen, 2011, Slovenia [33]	Observational, prospective	218	On-site, ground transport	Physicians N = 10	LUS	Review: expert	Feasibility (maybe): overall, time, Diagnostic A (yes): AHF
Duchateau, 2011, France [34]	Observational, prospective	50	On-site	Physicians NR	Other	Unclear	Feasibility (maybe): interpretation, time, ease of use Therapeutic (maybe): modification, modification proportion
McBeth, 2011, Canada-USA-Italy [35]	Observational, prospective	8	On-site, in-flight	Other	E-FAST	Review: expert	Feasibility (maybe): interpretation, quality, transmission, communication
Fagenholz, 2007, Nepal [36]	Observational, prospective, case‒controlled	18	At the office (rescue clinic)	NR	LUS	Review: expert	Feasibility (maybe): interpretation, intraRA, IRA Diagnostic I (yes): overall Prognosis (yes): saturation prediction, saturation correlation
Lapostolle, 2006, France [37]	Observational, prospective	169	On-site, ground transport	Physicians N = 8	LUS, CUS, other	Review: discharge diagnosis	Feasibility (maybe): interpretation, time, diagnostic modification Diagnostic I (maybe): diagnostic modification

AE, acute exacerbation; AHF, acute heart failure; COPD, chronic obstructive pulmonary disease; CUS, cardiac ultrasound; CXR, chest X-ray; Diagnostic A, diagnostic accuracy (indices of diagnostic accuracy such as sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios provided); Diagnostic I, diagnostic identification (other diagnostic contribution, without indices of diagnostic accuracy such as sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios provided); FAST, focused assessment with sonography in trauma; intraRA, intrarate agreement (between point-of-care ultrasound and reference standard); IRA, interrate agreement (between point-of-care ultrasound and reference standard); LUS, lung ultrasound; LVEF, left ventricle ejection fraction; NR, not reported; PE, pleural effusion; PNO, pneumothorax; RCT, Randomized controlled study.

^aNumber of patients included (not all patients underwent POCUS, see Table S4).

^bNumber of different sonographers performing point-of-care ultrasound.

^cYes: when quantitative data answering the given outcome is available, statistically significant and there is a positive contribution (in bold), No: when quantitative data answering the given outcome is available statistically significant and there is a negative contribution, Maybe: when quantitative data answering the given outcome is available but not statistically significant, Unconclusive: when quantitative data answering the given outcome is available but not statistically significant.

^dPresence of a non-point-of-care ultrasound group.

^eProspective cohort with a retrospective quality control study.

Risk of bias, quality, and applicability

The quality of evidence assessment according to the GRADE scale rated studies as either very low (17.4%, n = 4), low (69%, n = 16), or moderate (13%, n = 3) [11]. ROB assessment identified three intermediate-risk studies, 18 serious-risk studies, and two critical-risk studies (Table 2, Fig. 2) [9,10].

Table 2.

Point-of-care ultrasound contribution for management of patients with acute dyspnea in the prehospital setting: study outcomes feasibility, diagnostic, therapeutic, prognosis, referral, and transport vector outcomes (aggregated data)

Data/outcomes	Resultsa	Quality of evidenceb	Risk of biasc
Feasibility (n = 20 studies, 1261 patients included, 1818 scans analyzed)
Overall feasibility/type of ultrasound (n = 3)	▪ LUS: 90–100% (Laursen 2016, Prosen 2011, n = 2)	Low/moderate	Serious
	▪ CUS: easy in 62% of cases (Duchateau 2011)	Very low	Serious
POCUS failure in patients with dyspnea (n = 1)	▪ In case of failure 80% of dyspneic patients vs. 23% when the technique was successful (P < 0.001) (Duchateau 2011)	Very low	Serious
Scans adequate for interpretation (n = 14)	▪ Proportion of interpretable scans: 41–100% (Kowalczyk 2023, Donovan 2022, Hermann 2022, Rodríguez-Contreras 2022, Pietersen 2021, Schoeneck 2021, Becker 2018, Laursen 2016, Neesse 2012, Prosen 2011, Duchateau 2011, McBeth 2011, Fagenholz, 2007, Lapostolle 2006)	Very low/low/moderate	Moderate/serious/critical
Scanning time (n = 9)	▪ Mean: 63–105 s (Kowalczyk 2023, Duchateau 2011, n = 2)	Very low/low	Serious
	▪ Median: 60–600 s (Hermann 2022, Rodríguez-Contreras 2022, Scharonow 2018, Laursen 2016, Neesse 2012, Prosen 2011, Lapostolle 2006, n = 7)	Very low/low/moderate	Moderate/serious/critical
Image quality (n = 7)	▪ Average to excellent (53–100%) (Donovan 2022, Hermann 2022, Pietersen 2021, Charron 2015, Duchateau 2011, McBeth 2011, Lapostolle 2006)	Very low/low	Moderate/serious/critical
Transmission quality (n = 4)	▪ 71–100% (Hermann 2022, Pietersen 2021, Nadim 2021, Becker 2018)	Very low/low	Moderate/serious
Communication quality (n = 1)	▪ 7/10 (McBeth 2011)	Very low	Critical
Positive lung ultrasound scans according to localization (n = 1)	▪ Compared to middle axillary line: medium right (67 vs. 25%, P = 0.017), basal right fourth, intercostal space (91 vs. 55%, P = 0.038); medium left (83 vs. 33%, P = 0.007); and basal left (fourth intercostal space) (92 vs. 58%, P = 0.039) (Strnad 2016)	Low	Intermediated
Impact of transport on feasibility (n = 1)	▪ Impact of helicopter rotor during transport: higher motion artifact during rotation, no effect on diagnostic identification (in LUS with M-mode) (Lyon 2012)	Low	Serious
POCUS agreement with diagnosis: LUS (n = 1)	▪ Prehospital LUS agreement with discharge diagnosis: 90.91% (Kowalczyk 2022) ▪ Prehospital LUS agreement with ED diagnosis: 88.64% (Kowalczyk 2022) ▪ Prehospital LUS diagnosis comparison with ED diagnosis (McNemar’s chi-square = 0, P = 1.0) (Kowalczyk 2022) ▪ Inter-rater agreement prehospital LUS and ED diagnosis, k = 0.822 (SE: 0.07; 95% CI, 0.68–0.96), strong agreement (Kowalczyk 2022) ▪ Inter-rater agreement prehospital LUS and final hospital diagnosis, k = 0.934 (SE: 0.03; 95% CI, 0.88–0.99), almost perfect agreement (Kowalczyk 2022)	Low	Serious
POCUS agreement with diagnosis: CUS (n = 2)	▪ Intermediate sonographer-expert agreement: normal profile assessment (Donovan 2022) ▪ Very low sonographer-expert agreement: inferior vena cava assessment (Charron 2015) ▪ Low sonographer-expert agreement: pericardial effusion assessment (Charron 2015) ▪ Low sonographer-expert agreement: left ventricle ejection fraction assessment (Charron 2015) ▪ Low sonographer-expert agreement: right ventricle size assessment (Charron 2015)	Low	Moderate/serious/critical
Diagnostic contribution (n = 15 studies, 1381 patients included, 2113 scans analyzed)
Overall diagnostic contribution (n = 6)	▪ Moderate inter-rater agreement between prehospital POCUS diagnosis and reference standard for normal ultrasound identification (Donovan 2022)	Low	Critical
	▪ 75.8% inter-rater agreement between prehospital POCUS diagnosis and reference standard for overall diagnosis (accuracy of prehospital diagnosis) (Ienghong 2022)	Low	Moderate
	▪ High (87.7%) inter-rater agreement between prehospital POCUS diagnosis and reference standard, for normal ultrasound identification, overall agreement (k = 0.44) (Pietersen 2022)	Low	Moderate
	▪ 89.7% POCUS diagnosis agreement with reference standard (100% consensus after discussion) (Charron 2015)	Low	Serious
	▪ 91–100% Intra-rater agreement between prehospital POCUS diagnosis and reference standard, 91% with expert 1 (P < 0.0001), 100% with expert 2 (P < 0.0001), Inter-rater agreement between prehospital POCUS diagnosis and reference standard 100% (SE for kappa statistic = 0.10, P < 0.0001) (Fagenholz 2007)	Low	Serious
	▪ 67% diagnostic accuracy improvement; 8% diagnostic accuracy decrease; 25% not contributive for diagnostic. When initial diagnosis is uncertain (n = 115) 89.5% diagnostic accuracy improvement (Lapostolle 2007)	Low	Serious
▪Acute left heart failure/pulmonary edema: diagnostic identification (n = 5)	▪ Without LUS: Se = 23% (95% CI, 0.14–0.34), Sp = 97% (95% CI, 0.92–0.99); with LUS: Se = 71% (95% CI, 0.44–0.88), Sp = 96% (95% CI, 0.76–0.99) (Russel 2023)	Low	Serious
	▪ Without POCUS: Se 58% (95% CI, 46%–69%), ROC AUC = 0.72 (95% CI, 0.66–0.78); With POCUS: sensitivity 65% (95% CI, 54%–75%) (P = 0.12), ROC AUC = 0.79 (0.73–0.84) (P < 0.001) (Gundersen 2022)	Low	Serious
	▪ Moderate inter-rater agreement prehospital POCUS diagnosis and reference standard for AHF diagnosis identification (Donovan 2022)	Low	Critical
	▪ 66.7% inter-rater agreement prehospital POCUS diagnosis and reference standard for AHF diagnosis identification (Ienghong 2022)	Low	Moderate
	▪ Excellent (87.5%) inter-rater agreement between the two initial auditors for the diagnosis of cardiogenic pulmonary oedema (k = 0.746) (Laursen 2016)	Low	Serious
Acute left heart failure/pulmonary edema: diagnostic accuracye (n = 8)	▪ Se = 71–100%, Sp = 72–95%, PPV = 77–96%, NPV = 94–100%, AUC = 0.72–0.79 (Russel 2023, Gundersen 2023, Ienghong 2022, Schoeneck 2021, Scharonow 2018, Zanatta 2018, Laursen 2016, Prosen 2011)	Moderate/low	Serious
Pneumonia: diagnostic identification (n = 3)	▪ Moderate inter-rater agreement prehospital POCUS diagnosis and reference standard for pneumonia with B-lines diagnosis identification (Donovan 2022) ▪ Good inter-rater agreement prehospital POCUS diagnosis and reference standard for pneumonia with consolidation diagnosis identification (Donovan 2022)	Low	Critical
	▪ 67% inter-rater agreement pre-hospital POCUS diagnosis and reference standard for pneumonia diagnosis identification (Ienghong 2022)	Low	Moderate
	▪ Correct in 66.7% of cases (Neesse 2012)	Low	Serious
Pneumonia: diagnostic accuracy (n = 1)	▪ LUS: Se = 88% (95% CI, 75–95%), Sp = 59% (95% CI, 43–72%), PPV = 68% (95% CI, 55–79%), NPV = 83% (95% CI, 66–92%), PLR = 2.12 (CI, 1.45–3.10), NLR = 0.21 (CI, 0.09–0.49) (Rodríguez-Contreras 2022)	Moderate	Serious
Pleural effusion: diagnostic identification (n = 3)	▪ Good inter-rater agreement prehospital POCUS diagnosis and reference standard for pleural effusion diagnosis identification (Donovan 2022)	Low	Critical
	▪ Almost perfect (96.3%) inter-rater agreement prehospital POCUS diagnosis and reference standard for pleural effusion, overall agreement (k = 0.69) (Pietersen 2021)	Low	Moderate
	▪ Detected in 100% of AHF cases and 20% of chronic bronchitis exacerbation cases (Neesse 2012)	Low	Serious
Pleural effusion: diagnostic accuracy (n = 2)	▪ LUS: Se = 26–53.3%, Sp = 83.3–92% (Gundersen 2022, Zanatta 2018)	Low/moderate	Serious
Pneumothorax: diagnostic identification (n = 1)	▪ Very high sonographer-expert agreement for pneumothorax assessment (Donovan 2022)	Low	Critical
Pneumothorax: diagnostic accuracy (n = 2)	▪ Se = 1 (NR), (Lyon 2012) Sp = 1 (NR) (Scharonow 2018, Lyon 2012)	Low	Serious
COPD and asthma exacerbation: diagnostic identificationf (n = 1)	▪ AUC = 0.87 (0.82–0.91) before and 0.93 (0.88–0.97) after POCUS (P < 0.001) (Gundersen 2022)	Low	Serious
Therapeutic contribution (n = 8 studies, 678 patients included, 916 scans analyzed)
Contribution of pocus on therapeutic management (n = 7)	▪ 11.7–54.4% (Russel 2023, Donovan 2022, Pietersen 2021, Scharonow 2018, Zanatta 2018, Neesse 2012, Duchateau 2011)	Very low/low/moderate	Moderate/serious
Pharmacological therapy: appropriateness (n = 1)	▪ More appropriate pharmacological therapy in POCUS group (P < 0.01) especially with A-profile (P < 0.001) (Zanatta 2018)	Moderate	Serious
Pharmacological therapy: dose of furosemide (n = 1)	▪ Lower mean dose of furosemide in the POCUS group with A-profile than the non-POCUS group (3.33 mg vs 15.29 mg, P = 0.036) (Zanatta 2018)	Moderate	Serious
Contribution of POCUS on CPAP (n = 2)	▪ In patients with A-profile: CPAP employed more in the POCUS group than in the non-POCUS group (P = 0.011) (Zanatta 2018) ▪ In patients with B profile: CPAP use comparable (Zanatta 2018) ▪ The FIO2 administered was not significantly different between groups (Zanatta 2018)	Moderate	Serious
	▪ Treatment (required oxygen) to obtain adequate oxygen saturation: CPAP group: 40% vs. control group: 100% (P < 0.001) (Strnad 2016) ▪ Treatment according to the total number of B-lines: CPAP group: 46.9 ± 14.8 vs. control group: 29.0 ± 16.2 (P < 0.001) (Strnad 2016)	Low	Intermediated
Contribution of POCUS on intubation (n = 1)	▪ No statistical evidence: 2 patients were intubated in the NUS group and 1 in the US group, without any statistical significance (ns) (Zanatta 2018)	Moderate	Serious
Number-needed-to-scan (n = 1)	▪ 8.5 (Pietersen 2021)	Low	Moderate
Time to treatment (n = 2)	▪ Median time to treatment = 169 min without, 21 min with LUS (Russel 2023)	Low	Serious
	▪ No delay on treatment (Neesse 2012)	Low	Serious
Prognosis contribution (n = 2 studies, 78 patients included, 52 scans anayzed)
Prediction of clinical data (n = 1)	▪ Comet-tail score was predictive of oxygen saturation (P = 0.002) (Fagenholz 2007) For every 1-point increase in comet-tail score oxygen saturation fell by 0.67% (95% CI, 0.41–0.93%, P < 0.001) (Fagenholz 2007)	Low	Serious
Prediction of laboratory tests (n = 1)	▪ Laboratory tests and blood gases analysis NS different (Zanatta 2018) In patients with A-profile: significant lower concentration of PCO2 (PCO2 : 42.62 vs 52.23 P = 0.049) (Zanatta 2018)	Moderate	Serious
Prediction of hospitalization rate (n = 1)	▪ Hospitalization rate comparable between the two study groups (Zanatta 2018)	Moderate	Serious
Contribution to referral/destination (n = 4 studies, 201 patients included, 162 scans)
Hospitalization (n = 2)	▪ No difference (rate comparable between the two study groups) (Zanatta 2018)	Moderate	Serious
	51% (21/41) admitted to hospital (Nadim 2021)	Low	Serious
Destination/overall referral (n = 1)	▪ Patient transport destination, patient transport priority or monitoring requirements (e.g. patient does not need to be accompanied by a physician), changed in 25% (17/68) of cases (Scharonow 2018)	Low	Serious
Delay on transport/referral (n = 1)	▪ No delay on transport (Neesse 2012)	Low	Serious
Contribution to vector/transport (n = 3 studies, 141 patients included, 132 scans)
Transport priority, monitoring requirement (n = 1)	▪ Patient transport destination, patient transport priority or monitoring requirements (e.g. patient does not need to be accompanied by a physician), changed in 25% (17/68) of cases (Scharonow 2018) ▪ Ultrasound-related changes in transport occurred in 16.4% (Scharonow 2018)	Low	Serious
Ambulance renewed (n = 1)	▪ None of the patients released at the scene requested a renewed ambulance within the first 48h following the intervention (Nadim 2021)	Low	Serious
Delay on transport (n = 1)	▪ No delay on transport (Neesse 2012)	Low	Serious

AHF, acute heart failure; AUC, area under the curve; CI, confidence interval; COPD, chronic obstructive pulmonary disease; CPAP, continuous positive airway pressure; CUS, cardiac ultrasound; ED, emergency department; FIO₂, fraction of inspired oxygen; LUS, lung ultrasound; NLR, negative likelihood ratio; NPV, negative predictive value; POCUS, point-of-care ultrasound; PLR, positive likelihood ratio; PPV, positive predictive value; SE, standard estimate; Se, sensitivity; Sp, specificity;

^aSignificance (reference: author, year).

^bGRADE scale.

^cROBINS-I scale, unless stated otherwise.

^dRoB-2 scale.

^eConcerning studies reporting acute heart failure data on lung ultrasound assessment for meta-analysis.

^fDiagnostic accuracy was not reported.

Review findings (qualitative synthesis)

Twenty studies (N = 1261 patients included, N = 1818 scans analyzed) assessed feasibility. The scanning time ranged from 33 s to 10 min (n = 9 studies, moderate-critical ROB) [16,19,20,26,29,31,33,34,37]. Overall feasibility was assessed as 90–100% for LUS (n = 2, serious ROB) [29,33] and 62% for CUS (n = 1, serious ROB) [34]. The proportion of scans assessed as adequate for interpretation was 41–100% (n = 14, moderate-critical ROB) [16,18–20,22,23,25,29,31,33–37]. Image quality was assessed as average to excellent (53–100%, n = 7, moderate-critical ROB) [18,19,22,30,34,35,37]. Prehospital POCUS agreement with reference standard diagnosis was very high for LUS (88–90%, strong to almost perfect agreement, n = 1) [16] and very low to intermediate for CUS (n = 2, moderate-critical ROB) (Supplementary Table S4, Supplemental digital content 1, http://links.lww.com/EJEM/A475) [18,30].

Fifteen studies (N = 1381 patients included, N = 2113 scans analyzed) assessed diagnostic contribution, 11 studies assessed diagnostic identification (944 patients included, 1469 scans analyzed), and nine assessed diagnostic accuracy specifically (999 patients included, 932 scans analyzed) (Supplementary Table S5, Supplemental digital content 1, http://links.lww.com/EJEM/A475). The overall diagnostic contribution of POCUS was rated as moderate (for normal LUS identification) to perfect (for diagnosis agreement with the reference standard) (n = 6, moderate-critical ROB) [18,21,22,30,36,37]. There was a 67% improvement in overall diagnostic accuracy, and when the initial diagnosis was uncertain, there was an 89.5% improvement in diagnostic accuracy (n = 1, serious ROB) [37] AHF identification was assessed as moderate to excellent (n = 5, moderate-critical ROB) [15,17,18,21,29]. Pneumonia identification was assessed as moderate to good (n = 3, moderate-critical ROB) [18,21,31]. The diagnostic accuracies were sensitivity (Se) = 88% [95% confidence interval (CI), 75–95%], specificity (Sp) = 59% (95% CI, 43–72%), positive predictive value (PPV) = 68% (95% CI, 55–79%), negative predictive value (NPV) = 83% (95% CI, 66–92%), positive likelihood ratio (PLR) = 2.12 (95% CI, 1.45–3.10), and negative likelihood ratio (NLR) = 0.21 (95% CI, 0.09–0.49) (n = 1, serious ROB) [18]. Pleural effusion identification was assessed as good to almost perfect (n = 3, moderate-critical) [18,22,31]. The diagnostic accuracy was Se = 26–53.3%, Sp = 83.3–92% (n = 2, serious ROB) [17,27]. Pneumothorax identification was assessed as very high (n = 1, critical ROB) [18]. The diagnostic accuracy was Se = 100% (NR) (n = 1) and Sp = 100% (n = 2, serious ROB) [26,32]. There was a modification in the area under the curve [0.87 (0.82–0.91) before and 0.93 (0.88–0.97) after POCUS] for chronic obstructive pulmonary disease and asthma exacerbation identification (n = 1, serious ROB) (Table 2 and Supplementary Table S5, Supplemental digital content 1, http://links.lww.com/EJEM/A475) [17].

Eight studies (N = 678 patients included, N = 916 scans analyzed) assessed therapeutic contributions. POCUS triggered a change in management or therapy in 11.7 to 54.4% of cases (n = 7, intermediate to serious ROB) [15,22,26,27,31,34] 11–42% of cases when ultrasound was performed by a paramedic, and in 17–54% of cases when it was performed by a physician (Table 2).

Two studies (N = 78 patients included, N = 52 scans analyzed) assessed prognosis contributions. LUS (comet-tail score) was predictive of oxygen saturation (P = 0.002) [36] and the concentration of PCO₂ (42.62 vs. 52.23 mmHg, P = 0.049) [27], but the hospitalization rate was comparable between groups (n = 1, intermediate ROB) [27] (Table 2).

Four studies (N = 201 patients included, N = 162 scans analyzed) assessed the contribution to patient referral. Hospital admission referrals were shown in 51% of patients (n = 1, serious ROB) [27], and there was no difference in the hospitalization rate (n = 1, serious ROB) [24] (Table 2).

Three studies (N = 141 patients included, N = 132 scans analyzed) assessed transport vector change contribution. A change in the vector was shown in 25% of the cases (n = 1, serious ROB) [26], there was no delay in transport (n = 1, serious ROB) [31], and no patient experienced ambulance renewal within 48 h (n = 1, serious ROB) [24]. Excluding the results of studies with a critical ROB (n = 2/23) based on the ROBINS-I/RoB-2 tool assessment [18,35] did not alter these conclusions (Table 2, Table 3 and Supplementary Table S5–S7, Central Illustration, Supplemental digital content 1, http://links.lww.com/EJEM/A475).

Table 3.

Point-of-care ultrasound contribution for management of patients with acute dyspnea in the prehospital setting: evidence of contribution and main results

Outcome	ECa	Main resultsb	Quantity of evidencec	Quality of evidenced
Feasibility	Yes	Median scanning time of 1–10 min (n = 7, NA), 41–100% scans adequate for interpretation (n = 14, NA), image quality average to excellent (53–100%, n = 8, NA); overall, 90–100% for LUS (n = 2, NR) and 62% for CUS (n = 1, NA), LUS concordance with diagnosis (n = 1, SS), POCUS results according to localization (n = 1, SS)	20 studies 1261 patients 1818 scans	Low (GRADE moderate-very low, moderate-critical ROB)
Diagnostic: overall	Yes	Moderate to perfect (n = 5, P < 0.0001), 67% overall diagnostic accuracy improvement, 89.5% when uncertainty (n = 1, NR)	11 studies 551 patients 1285 scans	Low (GRADE low, moderate-critical ROB)
Diagnostic: acute heart failure	Yes	Moderate to excellent (n = 5, SS); Se = 71–100%, Sp = 72–95%, Accuracy 72–97%, PPV = 77–96%, NPV = 94–100%, AUC = 0.72–0.79 (SS)	8 studies 814 patients 1447 scans	Moderate (GRADE moderate-low, moderate-critical ROB
Diagnostic: pneumonia	No	Moderate to good (n = 3, NA); Se = 88% (95% CI, 75–95%), Sp = 59% (95% CI, 43–72%), PPV = 68% (95% CI, 55–79%), NPV = 83% (95% CI, 66–92%) (n = 1, SS)	4 studies 316 patients 441 scans	Low (GRADE moderate, moderate-critical ROB)
Diagnostic: pleural effusion	No	Good to almost perfect (n = 3, NA); Se = 26–53.3%, Sp = 83.3–92% (n = 2, NA)	5 studies 401 patients 985 scans	Low (GRADE moderate-low, moderate-critical ROB)
Diagnostic: pneumothorax	No	Very high (n = 1, NA); Se = 100% (n = 1, NA), Sp = 100% (n = 2, NA)	3 studies 130 patients 234 scans	Low (GRADE low, serious-critical ROB)
Diagnostic: COPD or asthma exacerbationd	No	AUC 0.87 (0.82–0.91) before 0.93 (0.88–0.97) after POCUS (n = 1, P < 0.001)	1 study 214 patients 214 scans	Low (GRADE low, serious ROB)
Therapeutic	Yes	11–54% treatment modification (n = 7, NA), More appropriate pharmacological therapy (n = 1, P < 0.01), Lower dose of furosemide (n = 1, P = 0.036), Number needed to scan = 8.5 (n = 1, NA)	8 studies 678 patients 916 scans	Low (GRADE moderate-very low, moderate-serious ROB)
Prognosis	No	LUS predictive of oxygen saturation (n = 1, P = 0.002), laboratory tests (PCO2) (n = 1, P = 0.049), No difference in hospitalization rate (n = 1, NA)	2 studies 78 patients 52 scans	Low (GRADE moderate-low, serious ROB)
Referral	No	Hospital admission = 51% (n = 1, NA), No difference in hospitalization rate (n = 1, NA)	4 studies 201 patients 162 scans	Low (GRADE low-moderate, serious ROB)
Transport vector change	No	25% of cases (n = 1, NR), No delay in transport (n = 1, NA)	3 studies 141 patients 132 scans	Low (GRADE serious ROB)

AUC, area under the curve; CI, confidence interval; COPD, chronic obstructive pulmonary disease; CUS, cardiac ultrasound; LUS, lung ultrasound; NA, quantitative data available, but no comparison test available or comparison test results not reported; NS, quantitative data available, statistically nonsignificant; POCUS, point-of-care ultrasound; ROB, risk of bias, assessed with ROBINS-I interventional nonrandomized studies; RoB-2 for randomized studies; SS, quantitative data available in and statistically significant in at least one study (confidence intervals not reported, best P value reported when available),

^aEC, evidence of contribution (based on main results statistical significance, quantity of evidence, and quality of evidence relative to the outcome).

^b(Number of studies, statistical significance), for diagnostic contribution, if reported: diagnostic identification (concordance) and diagnostic accuracy.

^cNumber of studies, number of patients included (which does not correspond to the number of patients scanned, see Table S4), and number of analyzed scans when reported.

^dGrading of Recommendations Assessment, Development and Evaluation (GRADE) scale, and Cochrane Risk Of Bias In Nonrandomized Studies—of Interventions (ROBINS-I) or Cochrane Risk of Bias Tool for Randomized Controlled Trials (RoB-2) scale.

Quantitative synthesis (meta-analysis)

The only outcome to meet quantitative synthesis criteria was the diagnostic accuracy of AHF. Seven diagnostic test accuracies were included in the meta-analysis (one study did not contain complete data, and it was not provided by the authors) [27]. A specific ROB assessment was conducted via the QUADAS-2 protocol [10], revealing a low ROB in one study [33], but at least one ROB concern for the six other studies (Supplementary Table S3, Supplemental digital content 1, http://links.lww.com/EJEM/A475) [15,17,21,23,26,29]. Bilateral B-lines for AHF diagnosis had a sensitivity of 71–100%, a specificity of 72–95%, a positive predictive value of 77–96%, a negative predictive value of 94–100%, and an area under the curve of 0.72–0.79 for the diagnosis of AHF (n = 7, serious ROB) [15,17,21,23,26,29,33] (Fig. 3). The reference standards had heterogeneous thresholds. Excluding the results of the only study with a high ROB in most QUADAS-2 domains [21] did not alter these conclusions (Supplementary Table S5, Supplemental digital content 1, http://links.lww.com/EJEM/A475).

Discussion

This comprehensive review addressed a knowledge gap regarding the contribution of POCUS to the management of acute dyspnea in nontrauma patients in the prehospital setting. Our findings support POCUS as feasible, especially LUS, supporting an overall diagnostic contribution and the use of LUS for diagnosing AHF. The analysis suggested that POCUS has a therapeutic contribution, modifying patient management. The evidence supporting the use of POCUS for pneumonia, pleural effusion, pneumothorax, chronic obstructive pulmonary disease, or asthma exacerbation diagnosis, as well as its prognostic contribution, contribution to patient referral decisions, and transport vectors, is limited. A high level of evidence on the contribution of POCUS for managing acute nontraumatic dyspnea in the prehospital setting is lacking and needed.

POCUS using LUS demonstrated excellent feasibility, whereas CUS showed moderate feasibility. We attribute this difference to learning curve and technique (LUS is quicker to learn, easier to perform, and less technically demanding) [38,39], device resolution (the limitations of ultraportable devices impact CUS more significantly and CUS performed with low-resolution devices is less feasible) [39,40], and environmental factors mainly affecting CUS image quality [37,41].

Diagnostic contribution results align closely with a recent meta-analysis by Russell et al. [42], however, their meta-analysis included pooling, which we did not perform for methodological reasons [13,14,42]. These results also align with those reported for in-hospital settings. A 2022 systematic review by Kok et al. on in-hospital POCUS use in acute dyspnea management reported similar outcomes in terms of feasibility, scanning times, and diagnostic agreement, especially for LUS [6]. This consistency across settings underscores the robustness and reliability of LUS as a diagnostic tool [5].

POCUS has a moderate therapeutic effect and does not significantly contribute to patient prognosis, referral, or transport vector changes for acute dyspnea management in nontrauma patients in the prehospital setting. It modified treatment decisions in a significant number of studies, affecting up to half of the cases or one in every nine patients, demonstrating utility in guiding pharmacological therapy and adjusting furosemide dosages [27]. However, our analysis did not find significant prognostic value for POCUS in this setting, nor did we find a significant contribution to referral decisions or changes in the transport vector. The latter is important with respect to early identification of pneumothorax, which may rule out air transport for the use of ground transport, thereby improving patient outcomes [24,26,31].

When separating higher- and lower-quality studies, high feasibility for LUS is consistently reported only in studies with moderate ROB studies. Similarly, for acute heart failure diagnosis, overall findings show a wide range of sensitivity and specificity but studies with a low ROB studies consistently report high accuracy. Therapeutic contribution appears more variable regardless of study quality, ranging from 11 to 54% across all studies, with the highest impact (54%) reported in a single study with a moderate ROB. These comparisons highlight the importance of study quality when interpreting POCUS effectiveness in prehospital settings.

These findings are broadly consistent with those of LUS performed in trauma patient management in the prehospital setting, where while it has an overall high diagnostic contribution when used by trained sonographers (particularly in identifying pneumothorax and confirming correct endotracheal intubation), there is currently no definitive evidence of improved clinical outcomes or survival rates associated with its use [43].

The effectiveness of POCUS remains highly dependent on the operator’s skill [43]. To maximize the benefits of POCUS in prehospital clinical practice, it is essential to implement standardized guidelines and integrate ultrasound findings with comprehensive clinical assessment [39].

Future research directions should include large-scale randomized controlled trials in the prehospital setting, as well as standardized protocols to ensure that POCUS is applied reliably. Finally, the cost-effectiveness of implementing POCUS could also be studied.

Limitations

First, the term ‘prehospital’ lacks a dedicated Medical Subject Headings term in search engines and encompasses various types of prehospital organizations, depending on the country where the study was conducted. Second, the heterogeneity of included studies, particularly with respect to the methods and technologies used, weakens the overall strength of findings. The majority of studies (20/23, 86%) were observational and classified as having low or very low quality of evidence (GRADE score) or serious to critical ROB (ROBINS-I/RoB-2 score). However, these scoring systems are considered highly conservative in their performance assessment [8,9]. Operator expertise ranged from newly trained paramedics to experienced physicians, while ultrasound devices varied from older models to pocket-sized units. The inclusion of older studies presents a limitation as prehospital ultrasound practices have evolved significantly over the past two decades [37]. A comprehensive meta-analysis was not performed to prevent methodological bias [13,14]. This heterogeneity impacts the validity of meta-analysis results and limits generalizability. Third, the potential influence for publication bias should not be ignored.

Conclusions

Evidence supports the contribution of POCUS in managing acute nontraumatic dyspnea in the prehospital setting, demonstrating its feasibility and overall diagnostic utility, particularly in the use of LUS for diagnosing AHF. POCUS also appears to influence the therapeutic contribution. However, there is insufficient evidence supporting the use of POCUS for diagnosing pneumonia, pleural effusion, pneumothorax, chronic obstructive pulmonary disease, or asthma exacerbation. Furthermore, evidence is lacking regarding the prognostic value of POCUS, its impact on patient referral decisions, and its influence on transport vector choices. A high level of evidence in these areas is currently lacking and needed.

Acknowledgements

The authors thank François Calais (registered librarian, Franche-Comté University) for his help in performing the search request.

Conflicts of interest

There are no conflicts of interest.

Supplementary Material

ejem-32-087-s001.pdf ^{(889.1KB, pdf)}