Monitoring patients with acute dyspnea with serial point-of-care ultrasound of the inferior vena cava (IVC) and the lungs (LUS): a systematic review

Abstract

Purpose

The primary aim was to investigate if treatment guided by serial ultrasound of the inferior vena cava-collapsibility index (IVC-CI) and B-lines on lung ultrasound (LUS) could reduce mortality, readmissions, and length of stay (LOS) in acutely dyspneic patients admitted to a hospital, compared to standard monitoring. The secondary aim was to determine how the changes of B-lines and IVC-CI are correlated to vitals and symptoms.

Methods

A systematic search was conducted on PubMed, Embase, Cochrane, Google Scholar, Web of Science, Scopus, OpenGrey, ProQuest, and databases for ongoing trials. The risk of bias was assessed according to study design.

Results

Of the 8258 studies identified, 50 were selected for full-text screening, and 24 studies were chosen for data extraction (19 pre–post-, two non-randomized controlled-, two randomized controlled-, and one retrospective cohort study), covering 2040 patients. Most studies were single-center and had small study populations with only heart failure patients. The risk of bias was high. No studies evaluated how the difference between two ultrasound measurements correlated with the primary outcomes. Seven studies reported that a decline in either B-lines or IVC size, or an increased IVC-CI reduced mortality, readmissions, and LOS when correlated to a single ultrasound measurement. All studies showed changes in the IVC-CI and B-lines, but these were not related to vitals or symptoms.

Conclusion

B-lines and IVC-CI are dynamic variables that change over time and with treatment. A single ultrasound measurement can influence prognostic outcomes, but it remains uncertain if repeated scans can have the same impact.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s40477-021-00622-7.

Introduction

Acute dyspnea is one of the most frequent reasons for admittance to an emergency department (ED) [1]. Dyspnea is caused by a broad spectrum of conditions, e.g., pneumonia, heart failure, and pulmonary embolism [2]. Patients with dyspnea have high mortality and prolonged length of stay (LOS) compared to patients admitted with other symptoms [3, 4]. Dyspnea is a patient-reported outcome and causes anxiety among patients [5–7].

Ultrasound is used for diagnosis and monitoring and as a prognostic tool in dyspneic patients. Ultrasound of the inferior vena cava (IVC) is used either alone or as a part of focused cardiac ultrasound (FoCUS) or more comprehensive echocardiography to gain information about the patient’s fluid status [8] and to diagnose acute heart failure [9]. Focused lung ultrasound (LUS) is additionally applied to detect pneumothorax, consolidations (e.g., pneumonia), and pleural effusions as well as interstitial syndrome (e.g., lung edema) that is defined sonographically as the presence of multiple B-lines in several scanning zones bilaterally [10].

Studies have investigated the role of monitoring patients admitted with dyspnea with repeated ultrasound of the IVC [11, 12] or LUS [13, 14], either used separately or in combination [15], but the treatment in these studies was not based on a protocol adjusted to the ultrasound findings. Only patients with dyspnea due to heart failure were included, and no studies included all dyspneic patients regardless of the presumptive diagnosis. Furthermore, the studies correlated outcomes to only a single ultrasound of the IVC or B-lines and not to repeated scans. Therefore, it is essential to examine the potential of serial ultrasound assessment in patients with dyspnea due to various causes.

The primary aim of this systematic review was to investigate if treatment guided by ultrasound of B-lines on LUS alone or combined with ultrasound of IVC-collapsibility index (IVC-CI), could reduce mortality, readmissions, and LOS in acutely dyspneic patients admitted to a hospital, when compared to usual care. The secondary aim was to investigate how the dynamic changes of B-lines and IVC-CI are correlated to vital signs and symptoms.

Materials and methods

This systematic review follows the Joanna Briggs Institute methodology for systematic reviews [16] and the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines [17, 18] (PRISMA checklist, Supplementary Material 1).

The protocol was registered at the International Prospective Register of Systematic Reviews (PROSPERO), number CRD42018116608, and published to enhance transparency of the research process [19].

Search strategy

A systematic search was performed on PubMed, Cochrane, Embase, Scopus, Web of Science, and Google Scholar. The gray literature was sought in OpenGrey and ProQuest. Ongoing trials were sought in Clinical Trials and the International Clinical Trials Registry Platform (ICTRP). Reference lists of the included studies were searched manually to find additional eligible studies. The inbuilt function “related/similar articles” in PubMed was used to detect further studies. Finally, forward-searching was used in Scopus and Web of Science with the included studies. The general search string, and the specific search strings used in the different databases, are listed in Supplementary Material 2. The last search was run on 17 December 2019, and the snowballing process was completed on 14 January 2020.

Studies were eligible if they included patients 18 years or older with dyspnea who were admitted to a emergency or similar setting and examined with ultrasound of the IVC and/or LUS at least twice during the admission. If patients were admitted with dyspnea mainly due to trauma or were mechanically ventilated, the studies were excluded. This was because patients with trauma or those undergoing mechanical ventilation represent a cohort that can receive a different diagnostic work-up and is typically transferred initially to an intensive care unit or surgical ward. Furthermore, the IVC behaves differently when the patient is ventilated [20]. Outcomes were mortality, readmission, LOS, and changes in ultrasound parameters (IVC-CI and B-lines). Controlled trials (randomized and non-randomized), observational studies (cohort, case–control), case reports, and conference abstracts based on these types of study designs written in English, Danish, Swedish, Norwegian, and German were included for screening. Studies were not excluded solely on the grounds of study type because it narrowed the evidence. Instead, all included studies were critically appraised according to study design.

References were double-checked for duplicates in Endnote V8.2 (Clarivate Analytics, PA, USA) and Covidence (Covidence, Melbourne, Australia). Two reviewers (MDA and NJ) screened all identified studies independently, and any divergences were resolved through discussion and consensus agreement.

Data collection and processing

The data were extracted independently by MDA and NJ according to a predefined data collection form, and disagreements were resolved through discussion, eliminating the need for a third reviewer.

MDA and NJ critically appraised the included studies according to study design with standardized tools for randomized and non-randomized controlled trials provided by the Joanna Briggs Institute [16]. The tools consist of a series of questions and do not provide an overall score. No studies were excluded on the grounds of the appraisal. Instead, a general judgment was based on the strength and limitations of the individual studies. Interrater agreement was assessed with Cohen’s kappa and calculated with STATA V16.1 (StataCorp, TX, USA).

A meta-analysis was planned but was not feasible due to heterogeneity in the study designs, interventions, ultrasound protocols, and scanning intervals.

An overall assessment of the robustness of evidence was made using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach [21]. The GRADE approach rates quality of evidence as high, moderate, low, or very low for each examined outcome.

Results

Study selection

The search identified 8258 studies (Fig. 1). After removing duplicates, 5661 studies underwent screening of titles and abstracts against the eligibility criteria, yielding 50 studies. After the full-text screening, 24 studies met the inclusion criteria. Supplementary Material 3 provides a list of the excluded studies from the full-text screening and the reasons for excluding them. The studies were excluded for various reasons but mainly because ultrasound scanning was not repeated during the admission.

Fig. 1.

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow diagram of the study selection process. IVC inferior vena cava, FLUS focused lung ultrasound

Characteristics of the studies

The characteristics of the studies are presented in Table 1. The 24 included studies comprised 19 pre–post studies [11, 15, 22–38], two non-randomized controlled trials (NRCT) [12, 39], two randomized controlled trials (RCT) [13, 40], and one retrospective cohort study [41]. Seventeen studies were single-center [11–13, 22, 23, 25–28, 30, 32, 34, 36–40], four were multicenter [29, 33, 35, 41], and in three cases the authors did not report the number of sites [15, 24, 31]. Only a few studies reported about funding [15, 23, 28, 32–35, 38]. About half the studies were from either Italy [26, 29, 31, 32, 34, 36, 40, 41], or USA [25, 27, 28, 33, 35].

Table 1.

Characteristics of the included studies

Study; country; no. of sites	Funding	Design	Cohort	n	Age, years, mean	US protocola	Scanning interval	Mortality	Readmission	LOS	Changes in IVC-CI and B-lines
Studies with US of the IVC
Asahi et al. 2016 [11] Japan, 1 site	NR	Pre–post	AHF, chronic HF	74	75	Probe NR Mode NR Meas. point NR	Admission, 24 h	NR	NR	NR	72–52% if 1st IVC-CI ≥ 0.5, 28–43% if 1st IVC-CI < 0.5
Chuen et al. 2009 [22] UK, 1 site	NR	Pre–post	AHF	31	72	Probe NR M-mode Measurement point NR	0, 0.5, 1, 2, 3 h	NR	NR	NR	ICV-CI 12–14%
Cubo-Romano et al. 2016 [23] Spain, 1 site	Grants for a medical society in Spain and National Institutes of Health	Pre–post	AHF	97	78	Convex Mode NR 2 cm from RA	< 24 h, discharge	90 days mortality rate 25.4% if IVCmax ≥ 1.9 cm vs. 3.4% 180 days: 29.3% vs. 3.4%	90 days readmission rate 30.8% if IVCmax ≥ 1.9 cm vs. 10.7% 180 days, 38.0% vs. 14.3%	NR	IVC-CI 25.7–33.1%
DeVecchis et al. 2012 [41] Italy, 2 sites	NR	Retrospective	AHF	49 34 controls	77	Sector M-mode 3 cm from RA	Admission, < 48 h	NR	NR	NR	IVC-CI 28.2–30.0%
Fawzi et al. 2012 [24] Egypt, sites NR	NR	Pre–post	AHF	30	50	NR	Admission, day 5, day 10	NR	NR	NR	NRb
Goonewardena et al. 2008 [25] USA, 1 site	NR	Pre–post	AHF	75	61	Probe NR Mode NR 3 cm from RA	< 12 h, discharge	NR	30 days: 41% readmitted IVC-CI, at admission was 23% in those readmitted vs. 31% in those not readmitted At discharge 36% vs. 57%	NR	IVC-CI 27–45%
Guiotto et al. 2010 [26] Italy, 1 site	NR	Pre–post	AHF	24	72	Sector M-mode 3 cm from RA	Before, after 12 h and at the end of ultrafiltration	NR	NR	NR	IVC-CI 12.5–23.6%
Patnaik et al. 2016, USA [27] USA, 1 site	NR	Pre–post	AHF	50	68	NR	< 24 h, daily	NR	NR	No difference in LOS according to IVC-CI	IVC-CI 41.3–40.9%
Tchernodrinski et al. 2014 [28] USA, 1 site	None	Pre–post	AHF	70	55	Sector Mode NR 2 cm from RA	0.5 h before, 1–2 h, 2–3 h after i.v. furosemid	NR	NR	NR	IVCmax 2.3–0.21 cm
Yavasi et al. 2014 [12] Turkey, 1 site	NR	NRCT	AHF	47 50 controls	67	Conex Mode NR Distal to hepatic vein	Before + after 12 h of Tx	NR	NR	NR	IVC-CI: 22.80–39.75% Controls: no difference in IVC-CI
Studies with LUS
Cortellaro et al. 2017 [29] Italy, 4 sites	NR	Pre–post	AHF	41	77	Convex Transverse/oblique orient 11 zonesc	Admission, + 3, + 24 h	NR	NR	NR	B-score: 1.59–0.38 points
Daskalov et al. 2016 [30] Bulgaria, 1 site	NR	Pre–post	AHF	100	NR	NR	Admission, 4 ± 2 days	NR	NR	NR	Reduction in B-lines, exact numbers NR
Facchini et al. 2016 [31] Italy, sites NR	NR	Pre–post	AHF	50	75	Probe NR Orient. NR 28 zones Sum B-lines	Admission, 24 h	NR	NR	NR	B-lines: 53.4–31.7
Gargani et al. 2015 [32] Italy, 1 site	One coauthor was consultant for several medical indutries	Pre–post	AHF	118	70	Probe NR Orient. NR 28 zones Sum B-lines	Admission, discharge	4 died, B-lines relation NR	14 readmitted with more B-lines at admission + discharge	NR	B-lines: 48–20
Martindale et al. 2018 [33] USA, 2 sites	Supported in part by grants from a university and National Institutes of Health	Pre–post	Dyspnea, sBT ≥ 180 mmHg, AHF	20	68	Convex Sagittal orient 8 zones Sum B-lines	< 45 min, after improvement of dyspnea, discharge	NR	NR	NR	B-lines: 47–8 Correlation between dyspnea (VAS from 0 to 10) and no. of B-lines
Palazzuoli et al. 2018 [34] Italy, 1 site	One coauthor received grants from medical industry	Pre–post	AHF	162	80	Probe NR Orient. NR 8 zones Sum B-lines	< 12 h, discharge	10% died, had a higher B-line count at discharge	28% readmitted, had a higher B-line count at discharge	NR	B-lines: 31–20 B-lines had a correlation with RR
Platz et al. 2019 [35] USA, 2 sites	Supported by grants from National Institutes of Health, the British Heart Foundation and several medical indutries	Pre–post	AHF	349	75	Sector Sagittal orient 4 zones Sum B-lines	< 4 days, discharge	6 months, 36% (0–4 B-lines at discharge) vs. 55% (> 7 B-lines) died/readmitted	6 month, 36% (0–4 B-lines at discharge) vs. 55% (> 7 B-lines) died/readmitted	0–3 B-lines: 5 days 4–6 B-lines: 8 days ≥ 7 B-lines: 7 days	B-lines: 6–4
Strnad et al. 2016 [13] Slovenia, 1 site	NR	RCT	AHF	20	81	Micro convex Orient. NR 15 zones Sum B-lines	NR	CPAP: 58% survied to discharge Control: 75% survied to discharge	NR	CPAP: 13 days Control: 7 days	B-lines: CPAP group, 46.9–29.0 Control, 42.4–43.3
Vitturi et al. 2011 [36] Italy, 1 site	NR	Pre–post	Dyspnea	152	NR	Convex Sagittal orient Zones NR Pos. scan: > 8 B-lines	Admission, 48 h	NR	NR	NR	B-lines decreased, exact no. NR
Volpicelli et al. 2008 [37] Italy, 1 site	NR	Pre–post	AHF	81	75	Convex Sagittal orient 15 zones Sum B-lines	Admission, discharge	NR	NR	NR	B-lines: 8 positive zones to 0
Studies with both US of the IVC and LUS
Mozzini et al. 2018 [40] Italy, 1 site	NR	RCT	AHF	LUS: 60 CXR: 60	84	LUS: sector Orient. NR Zones NR Sum of B-lines IVC: convex M-mode 2 cm from RA	LUS: admission, 24 h, 48 h, 72 h, discharge IVC: admission, discharge	NR	NR	LUS group: 7 days CXR group: 8 days	B-lines from 23 to 50% IVC-CI: Inverse association between IVC-CI and no. of B-lines
Spevack et al. 2017 [15] Canada, sites NR	None	Pre–post	AHF	50	77	LUS: convex Oblique orient 8 zones Sum of B-lines IVC: sector mode NR 1–2 cm from RA	Admission, 1 day, discharge	NR	NR	NR	IVC-CI 30–25% B-lines: 11–8.3
Öhman et al. 2018 [39] Finland, 1 site	NR	NRCT	AHF	US-guided Tx: 20 Control: 100	76	LUS: sector Sagittal orient 6 zones Sum B-linesd IVC: sector M- or B-mode 1–2 cm from hepatic vein	Admission, daily, discharge	84% patients with resolution of congestion survied at 6 months vs. 61%	36% patients with resolution of congestion died/readmitted vs. 58% at 6 months	NR	Congestion decreased but no. of B-lines NRe
Öhman et al. 2018 [38] Finland, 1 site	None	Pre–post	AHF	60	76	LUS: sector Sagittal orient 6 zones Sum B-linesd IVC: sector M- or B-mode 1–2 cm from hepatic vein	0 h, 12 h, 24 h, 48 h, discharge	84% patients with resolution of congestion survied at 6 months vs. 60%	37% patients with resolution of congestion died/readmitted vs. 64% at 6 months	Resolution: 6.16 days Congestion: 7.22 days	Congestion decreased but no. of B-lines NRe

AHF acute heart failure, CPAP continuous positive airway pressure, CXR chest X-ray, HF heart failure, IVC-CI inferior vena cava-collapsibility index, IVCmax inferior vena cava max diameter, LOS length of stay, LUS lung ultrasound, NR not reported, NRCT non-randomized controlled trial, orient. orientation, RA right atrium, RCT randomized controlled trial, RR respiratory rate, sBT systolic blood pressure, Tx treatment, US ultrasound, VAS visual analog scale

^aProbe type, M- or B-mode (IVC), orientation of the probe (LUS), no. of scanning zones (LUS)

^bDiscrepancy between reported IVC-CI in text and table and therefore NR

^cScore: 0 points: < 3 B-lines in a zone, 1 point: ≥ 3 B-lines in ≥ 1 zone, 2 points: multiple B-lines

^dCongestion definition: 3 ≥ B-lines in ≥ 1 zone bilat. or > 5 mm of pleural fluid bilaterally

^eIVC is reported as a categorical scaled IVC index

The included studies involved 2040 patients (1796 subjects vs. 244 controls) with a mean age of 72 years (66% women). The studies had a sample size ranging from 20 to 349 patients with diagnosed or suspected heart failure. In most studies, the treatment intervention strategy was only described in broad terms, e.g., “usual heart failure treatment,” without mentioning specific medications, dosages, or other treatments. Furthermore, the studies did not link the changes in ultrasound findings to a specific treatment algorithm.

The scanning protocols also varied: ten studies with ultrasound of IVC only [11, 12, 22–28, 41], ten with LUS only [13, 29–37], and four with both [15, 38–40]. The ultrasound protocols for IVC scanning were nearly identical with IVC scanning done during a normal respiration cycle in the subcostal view 2 cm from the junction of the IVC to the right atrium. Measurements with both M- and B-mode were used, and the patient was in the supine or semi-supine position. The choice of probe was the phased array. Preset was not described. The minimum scanning interval ranged from 0.5 h to 5 days.

The LUS protocols varied regarding the number of zones scanned (from 4 to 28 zones) and how the B-lines were quantified: (1) sum of B-lines; (2) the number of positive zones (positive zone: 3 ≥ B-lines per zone); (3) a positive scan defined as 8 or more B-lines in total; (4) two different congestion scores—one depending on the number of B-lines in each zone and one depending on the number of B-lines and/or presence of pleural effusions bilaterally. Different types of probes were used (micro convex, convex, sector). The transducer’s orientation was either in the sagittal or oblique plane, and the patient was positioned supine or semi-supine. Preset setting was not reported except for two studies that used the abdominal preset [29, 33]. The time between the two LUS examinations varied from 3 h to 6 days.

Critical appraisal of the studies

An overview of the critical appraisal is provided in Figs. 2 and 3. Consensus was reached between the two reviewers with a substantial agreement; Cohen’s kappa was 0.69. Responses were either yes, no, or unclear. The option of ‘not applicable’ was not used. The answers “no” and “unclear” were both considered to introduce potential bias. The studies generally failed to test for intra- and intervariability and to report confidence intervals and sample size calculations. In addition, the NRCTs failed to measure outcomes in a reliable way and were missing a control group to compare the intervention against to strengthen the examination of causal validity. In the RCTs, it was unclear if true randomization was used, and it was uncertain if outcome assessors were blinded. According to the predefined critical appraisal tool for the RCT, unblinding of the treating physician or patient should be considered bias. However, this was not considered a bias in our review because we investigated the effect of ultrasound-guided treatment; thus, the physician at some point in time had to be informed of the ultrasound findings in order to react to them.

Fig. 2.

Critical appraisal of the quasi-experimental studies

Fig. 3.

Critical appraisal of the randomized controlled trials

Review findings

No studies reported on how differences between two ultrasound measurements correlated with the primary outcomes—death, LOS, or readmission. However, some of the studies correlated one or several timepoint measurements (e.g., the ultrasound measurement at admission or at discharge) with the outcomes. Seven studies reported on deaths in correlation with point estimates of either IVC [23] or B-lines [13, 32, 34, 35, 38, 39]. Patients with more B-lines had about 1.5-fold increased risk of dying [35, 38, 39] and readmission [32, 34, 35, 38, 39]. A diameter of ≥ 1.9 cm IVC_max was correlated with about 7–9 times higher mortality rate at 3 and 6 months [23]. A larger IVC diameter [23] and/or a smaller IVC-CI [25] correlated with a 2.5-fold increased rate of rehospitalization. No differences were observed in LOS related to IVC-CI [27]. Three studies [35, 38, 40] reported reduced LOS by 1–2 days if the patient had fewer B-lines, but another study [13] showed increased LOS even though there were fewer B-lines compared to a control group. The studies using both IVC and LUS ultrasound did not correlate the combined scanning results with mortality, readmission, or LOS [15, 38–40].

All studies reported variations in B-lines or IVC-CI between two or several timepoints during the admission. A decline in B-lines and an increase in IVC-CI was seen from admission to discharge. Only two studies correlated the findings to vital signs or symptoms. One study reported a positive correlation between dyspnea on a verbal scale from 0 to 10 and the number of B-lines [33]. Another study reported a positive correlation between B-lines and respiratory rate [34].

In the GRADE summary of findings (Tables 2, 3), the outcomes were associated with B-lines and IVC. In general, the evidence was rated as low to very low primarily due to the type of study design, small sample sizes, few events, restriction to patients with heart failure, and no control groups.

Table 2.

Summary of findings using the Grading of Recommendations Assessment, Development and Evaluation approach (GRADE) for outcomes associated with number of B-lines

Outcomes	No of participants (studies)	Certainty assessment						Comments
		Risk of bias	Inconsisten-cy	Indirectness	Imprecision	Publication bias	Overall certainty of evidence
Mortality associated with no. of B-lines	837 (4 pre–post studies, 1 NRCT, 1 RCT)	Seriousa	Seriousb	Seriousc	Seriousd	None	⊗◯◯◯ Very low	The studies suggested that fewer B-lines was associated with better survival
Readmission associated with no. of B-lines	809 (4 pre–post studies, 1 NRCT)	Seriouse	Seriousb	Seriousc	Seriousf	None	⊗◯◯◯ Very low	A lower B-line count at discharge was associated with fewer readmissions
LOS associated with no. of B-lines	549 (2 pre–post studies, 2 RCTs)	Seriousg	Seriousb	Serioush	Seriousd	None	⊗◯◯◯ Very low	A fewer no. of B-lines was associated with a decreased LOS in three studies
B-lines associated with vitals	162 (1 pre–post study)	Seriousi	Not serious	Seriousj	Seriousk	None	⊗◯◯◯ Very low	B-line had a strong correlation with the respiratory rate (Spearmans coefficient 0.75)
B-lines associated with degree of dyspnea	20 (1 pre–post study)	Seriousl	Not serious	Seriousj	Seriousk	None	⊗◯◯◯ Very low	Correlation between dyspnea (on a VAS from 0 to 10) and no. of B-lines, but no correlation between the magnitude of change in sonographic pulmonary edema and VAS scores

CI confidence interval, LOS length of stay, NRCT non-randomized controlled trial, RCT randomized controlled trial, VAS visual analog scale

^aDowngraded due to risk of bias (pre–post studies: no control group, not the same treatment, no sample size calculations. RCT randomization and allocation unclear, no blinding, no sample size calculations)

^bDowngraded due to inconsistency (different scanning protocols and outcome measurements)

^cDowngraded due to indirectness (only patients with heart failure, different scanning protocols, no control groups, different outcome measurements)

^dDowngraded due to imprecision (few events and more events in control group in RCT)

^eDowngraded due to risk of bias (no control group, not provided the same treatment, outcomes not measured in a reliable way, no sample size calculations)

^fDowngraded due to imprecision (few events)

^gDowngraded due to risk of bias (pre–post studies: no control group, no sample size calculations. RCT randomization and allocation unclear, no blinding, no sample size calculations)

^hDowngraded due to indirectness (only patients with heart failure, different scanning protocols, small control group, different outcome measurements)

ⁱDowngraded due to risk of bias (no control group, no sample size calculation)

^jDowngraded due to indirectness (only patients with heart failure, no control group)

^kDowngraded due to imprecision (small sample)

^lDowngraded due to risk of bias (no control group)

Table 3.

Summary of findings using the Grading of Recommendations Assessment, Development and Evaluation approach (GRADE) for outcomes associated with IVC-CI

Outcomes	No. of participants (studies)	Certainty assessment						Comments
		Risk of bias	Inconsisten-cy	Indirectness	Imprecision	Publication bias	Certainty of the evidence
Mortality associated with IVC-CI	97 (1 pre–post study)	Not serious	Not serious	Seriousa	Seriousb	None	⊗⊗◯◯ Low	Only one pre–post study with a small sample size conducted in a selected population of heart failure patients
Readmission associated with IVC-CI	172 (2 pre–post studies)	Not serious	Seriousc	Seriousa	Seriousb	None	⊗◯◯◯ Very low	Only two pre–post studies with a small sample size conducted in a selected population of heart failure patients with different outcomes (IVCmax and IVC-CI)
LOS associated with IVC-CI	50 (1 pre–post study)	Seriousd	Not serious	Seriousa	Seriousb	None	⊗◯◯◯ Very low	Only one pre–post study with a small sample size conducted in a selected population of heart failure patients and no differences in LOS associated with IVC-CI
IVC-CI associated with vitals	(0 studies)	–	–	–	–	–	–	IVC-CI associated with vitals not reported
IVC-CI associated with degree of dyspnea	(0 studies)	–	–	–	–	–	–	IVC-CI associated with degree of dyspnea not reported

CI confidence interval, IVC-CI inferior vena cava collapsibility index, LOS length of stay

^aDowngraded due to indirectness (only patients with heart failure, no control group)

^bDowngraded due to imprecision (small sample)

^cDowngraded due to inconsistency [uses different outcomes (IVCmax and IVC-CI)]

^dDowngraded due to risk of bias (outcomes not measured in a reliable way, no sample size calculations)

Discussion

We found that single LUS or IVC ultrasound alone or in combination have a possible role in monitoring acutely dyspneic patients and may offer prognostic outcomes (death, readmission, LOS) when related to a single ultrasound measurement. However, the association between outcomes and differences between serial ultrasound measurements remains uncertain. Ultrasound parameters (IVC, B-lines) changed during the course of admission, but their correlation to standard measurements was unclear. The studies showed risk of bias, and the body of evidence was low.

The review showed that a larger IVC size or a lower IVC-CI was correlated with higher mortality and more readmissions [23, 25]. However, these prognostic outcomes were only linked to point estimates of IVC-CI at a particular timepoint and not to the differences between two measurements, which was our objective in examining the necessity of repeated ultrasound examinations. Furthermore, the treating clinicians were not instructed on how to react to changes in the IVC.

IVC measurement is usually used in the ED to forecast fluid responsiveness [20]. Two meta-analyses that investigated the role of only one IVC measurement during admission came to opposite conclusions regarding its usefulness to guide fluid administration [42, 43]. It was found to be more reliable in mechanically ventilated patients, which was not the subject of interest in this systematic review. Another meta-analysis suggested cut-offs in the IVC-CI value and IVC_max size to predict if a patient with acute dyspnea is more likely to have acute heart failure than another condition; the heterogeneity and bias across the studies were both high, however [44]. A dynamic structure such as the IVC should have the potential to be followed over time, but the data should be interpreted with caution and always in the clinical context and together with other parameters such as vital signs, physical examination, and blood samples.

A low number of B-lines on LUS was related to better survival, fewer readmissions, and reduced LOS. This is supported by a systematic review investigating the role of B-lines as a prognostic tool in heart failure patients in emergency and outpatient settings [45]. It concluded that the number of B-lines decreased when heart failure was treated, and a high sum of B-lines might identify those with a higher risk of readmission or death. In a study where LUS was compared to low-dose computed tomography as the reference method, B-lines on an 8-zoned LUS could be used to diagnose patients with congestion; ≤ 1 B-line could rule out congestion while ≥ 3 B-lines bilaterally and/or the presence of bilateral pleural effusions could rule in congestion in patients 50 years or older and with suspected heart failure [46]. The evidence thus suggests that B-lines are reasonable to measure upfront in the diagnostic evaluation and later in discharge planning. However, it remains uncertain how far this relates to patients with dyspnea from any cause. B-lines are not always present on the first examination, and they could develop in the course of the admission, e.g., in patients receiving too much fluid. B-lines can also be seen in patients without heart failure, e.g., acute respiratory distress syndrome (ARDS), pulmonary infections, and interstitial lung diseases, underlining the need for studies with broader inclusion criteria [10].

All the included studies showed that ultrasound could provide dynamic measurements that alter over time. This means that treatment effects could theoretically be monitored by the dynamics of the ultrasound parameters. This is time-consuming, however, compared to just asking the patients about symptoms or interpreting the vital signs. Therefore, these ultrasound parameters must be shown to be linked to both symptoms and vital signs to determine if they provide additional clinical information. Furthermore, an ultrasound-based treatment algorithm must be related to the ultrasound variables to investigate the intervention’s effect.

As dyspneic patients have a poor prognosis [3, 4], it is essential to find better ways of monitoring them and guiding their treatment. It is too early to proclaim serial ultrasound as a feasible monitoring tool in the ED. This needs to be examined in a study designed to overcome some of the issues present in most of the included studies in this systematic review. The study has to be powered to answer the research question, with sample size calculated beforehand. Many of the studies had a hard endpoint (e.g., death) that could require a large sample size. Instead, a patient-reported outcome (such as dyspnea on a verbal scale with blinding of the outcome accessor) could be chosen. An evaluation of the effect of serial ultrasound-guided treatment should be based on a treatment protocol, and differences between two (or more) ultrasound measurements (period effect) at different timepoints need to be compared to a control group (treatment effect). Furthermore, the presumptive diagnosis is not always correct [47], and the inclusion of a broader group of patients with dyspnea is desirable, not just heart failure; such a study has already been initiated [48]. Ultrasound should always be interpreted in the clinical context. Instead of using exact estimates like the precise measurement of the IVC and the total number of B-lines, another approach is to use dichotomized variables, e.g., large/small IVC, collapsible/not collapsible IVC, and B-lines present uni- or bilaterally. This method is also more straightforward and faster and can be used by non-specialists with limited training [49, 50].

Limitations

First, we included studies published in the gray literature as well as abstracts with more limited information in order to limit selection bias. This could have led to unjustified heterogeneity, missing data, and reporting bias. Second, only a handful of the included studies reported prognostic outcomes (mortality, readmission, LOS); thus, the evidence is based on a few studies with a high risk of bias, and the results should be interpreted with caution. Third, the studies’ lack of comparability made a meta-analysis inappropriate, and we could not report precise effect estimates or their certainty. However, our qualitative synthesis of the results provided an overview of the literature and current evidence. Fourth, many different tools exist to judge the risk of bias in studies, and it may not be possible for one tool to cover all types of included study designs [51]. We prioritized tools from the same source to ensure consistency, but some of the predefined questions had to be interpreted differently from the original intention, e.g., the importance of blinding of the treating physician.

Conclusion

In conclusion, this systematic review showed that B-lines and IVC are dynamic variables that change over time and treatment. A single ultrasound examination can influence prognostic outcomes in patients with dyspnea, but it remains uncertain if several ultrasound examinations guided by a treatment protocol have the same effectiveness. The nature of the variables could hypothetically be used by the physician to plan patient care, predict clinical deterioration, and adjust treatment.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

MDA has conceived the study and received inputs and feedback from ATL, NJ, PHG, and CBL. MDA developed the search strategy in cooperation with AFH. MDA and NJ screened the included studies and independently assessed the quality and extracted the data. MDA drafted the manuscript. All co-authors read and approved the final manuscript.

Funding

The study is supported by the Department of Emergency Medicine at Slagelse Hospital and grants from the Naestved, Slagelse, and Ringsted Hospitals’ Research Fund (Grant Number 111.2219) and by a 1-year scholarship from the University of Southern Denmark. The funders have no role in the study’s design, in the collection, analysis, or interpretation of data, in the writing of manuscripts, or in decisions to publish results.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Code availability

Coding is available upon request.

Declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.