Diagnostic Accuracy of Ultrasound in Intensive Care Patients with Undifferentiated Shock: A Systematic Review and Meta-analysis

Abstract

Purpose

This systematic review aimed to assess the accuracy of ultrasound in diagnosing shock types among intensive care patients.

Materials and methods

A comprehensive search of PubMed, Embase, Scopus, Cochrane Central Register, and Google Scholar was conducted for controlled trials published up to June 2023. Two intensivists independently screened articles for full-text reviews and abstracts, evaluating study quality using the QUADAS-2 tool. Prospective studies assessing ultrasound for diagnosing shock types in critically ill patients with undifferentiated shock were included.

Results

Among 7287 articles identified, four met the inclusion criteria for meta-analysis. Pooled positive likelihood ratios were 8.8 (95% CI: 2.4–32.37) for distributive shock and 137.56 (95% CI: 27.76–681.64) for obstructive shock. Summary receiver operating characteristic (SROC) curves showed an area under the curve (AUC) of 0.99 for cardiogenic and obstructive shock, 0.5 for hypovolemic and mixed shock, and 0.76 for distributive shock. Pooled negative likelihood ratios ranged from 0.05 (95% CI: 0.010 to 0.24) for cardiogenic shock to 0.22 (95% CI: 0.127–0.38) for mixed-etiology shock.

Conclusion

Ultrasound demonstrates high accuracy in diagnosing obstructive and cardiogenic shock among intensive care patients with undifferentiated shock. However, its utility for other shock types appears limited.

How to cite this article

Karigowda L, Gupta B, Elkady H, Deshpande K. Diagnostic Accuracy of Ultrasound in Intensive Care Patients with Undifferentiated Shock: A Systematic Review and Meta-analysis. Indian J Crit Care Med 2024;28(12):1159–1169.

Introduction

Shock is a leading cause of intensive care admission and is associated with high morbidity and mortality.¹ There are five major types of shock: distributive, cardiogenic, obstructive, hypovolemic, and mixed shock (a combination of one or more types). Early identification of the type of shock and prevention of prolonged hypotension has been shown to decrease morbidity and mortality.² Accurate identification of the underlying type of shock is essential, as it dictates entirely different clinical management pathways.³

Several systematic reviews and meta-analyses have examined the diagnostic accuracy of ultrasound in patients with circulatory failure admitted to the emergency department.^4–6 However, no systematic review has yet assessed the diagnostic accuracy of ultrasound specifically in intensive care patients with circulatory failure. This systematic review aims to address this gap by focusing on the diagnostic accuracy of ultrasound in identifying shock types in intensive care patients. The critical care patients included in the study may have received initial resuscitation in the emergency department, ward, or operation theater and remained in shock.

Materials and Methods

The systematic review protocol was registered with PROSPERO (CRD42023402223). The preferred reporting items for systematic review and meta-analyses (PRISMA) guidelines were followed in conducting this review.⁷

Selection Criteria

The inclusion criteria were as follows:

• Adult patients who presented to the intensive care unit with an undifferentiated shock state.

• Ultrasound assessment to identify the type of shock.

The following studies were excluded:

• Studies that used ultrasound as a diagnostic tool to determine shock outside the ICU.

• Patients in whom the type of shock had already been determined (shock due to trauma, sepsis, or any known type of shock).

• Case reports, case series, animal studies, retrospective studies, and review articles.

Search Strategy

We searched PubMed, Embase, Scopus, Cochrane Library, and Google Scholar. Additional searches included citation tracking of included studies and previous systematic reviews. Databases were searched from their inception to June 2023. The following MeSH terms were used: “Shock” and “Ultrasound” OR “Echocardiography” OR “POCUS” and “Critical care.”

The references of the included studies were assessed, and relevant review articles were evaluated for further pertinent articles. Three reviewers (LK, HE, and BG) independently reviewed the literature and included studies that met the inclusion criteria. Full-text reviews, quality assessments, and data extraction were independently conducted by LK and HE. The reviewers were not blinded to authorship, journal, or year of publication. Disagreements were resolved through consensus-based discussion, with BG acting as an adjudicator when necessary. Data extracted included study design, location, sample size, participant characteristics, intervention, reference standards, and outcome measures. Two authors (LK and HE) independently assessed the quality of the studies using the QUADAS-2 tool⁸ to identify bias.

We performed a systematic review and meta-analysis to determine the diagnostic accuracy of ultrasound in intensive care patients with undifferentiated shock.

Statistical Analyses

The studies were divided into five groups: cardiogenic, hypovolemic, distributive, obstructive, and mixed shock. Meta-DiSc® (version 1.4, XI Cochrane Colloquium, Spain) was used for all statistical analyses. The Spearman correlation coefficient of sensitivity and 1-specificity log were used to estimate heterogeneity due to the threshold effect. Heterogeneity due to non-threshold effects was estimated using the I² test. I² values ≤ 25% indicated low heterogeneity, 25–50% indicated moderate heterogeneity, and >50% indicated high heterogeneity. In the presence of heterogeneity, a random-effects model was used for further analysis.

For each study, true-positive (TP), false-positive (FP), false-negative (FN), and true-negative (TN) index parameters were calculated. The pooled sensitivity, specificity, negative likelihood ratio, and positive likelihood ratio were determined with 95% confidence intervals (CIs). A summary receiver operating characteristic curve (SROC) was generated for each parameter, and the Q value was calculated. The area under the curve (AUC) and standard error were computed to assess the diagnostic accuracy of ultrasound in detecting shock types. An AUC value close to 1 indicates good discriminative ability.

Results

Study Selection and Characteristics

After screening the titles and abstracts of the 5,099 studies, we selected four articles for review. The number of studies identified, screened, and included in the meta-analysis is shown in Figure 1. The baseline characteristics of the four studies are presented in Table 1. These studies were conducted in India, Canada, China, and Egypt. All of them had prospective study designs utilizing ultrasound to determine the type of shock in critically ill patients with undifferentiated shock. Four studies met the inclusion criteria and were eligible for meta-analysis. There were 375 patients with shock in these four studies.

Fig. 1.

Preferred reporting items for systematic reviews and meta-analysis (PRISMA) flow diagram

Table 1.

Characteristics of the studies included in the meta-analysis

Author	Year	Design, site and location (country)	No. of participants	Median or mean age(years)	Inclusion criteria	Exclusion criteria	Operator; equipment	Study outcome	Protocol	Comparator
Vaidya et al.9	2014	Prospective, ICU, India	100	51.5 (mean)	Age 18 SBP <90 and Unresponsiveness, altered mental status, syncope, respiratory distress profound asthenia fatigue and malaise severe chest or abdominal pain	Etiology for shock (trauma, external bleeding, pregnancy related complications)	Operator: Intensivist; Equipment: Sonosite M-Turbo and Voluson ultrasound machines. Straight linear array probe – 5–12 MHz, Curvilinear probe – 2–5 MHz and Sector array ultrasound probe 1–5 MHz	To classify shock, using ultrasonography as the modality of choice for imaging and to assess the diagnostic accuracy of ultrasound as a tool to classify shock	RUSH protocola	Clinical and biochemical studies, evaluation by physician and also depending on response to treatment
Majo et al.10	2004	Prospective, ICU, Canada	100	63 ± 14	SBP < 100 mm Hg Fall in BP > 25% and inotrope use Evidence of low output Pulmonary/venous congestion	Within 7 days after cardiac surgery	Operator: National Board of Echocardiography diplomates Equipment: GE Vingmed System V or Vivid 5 with 2.5 MHz transducer	ECHO detection of cardiac/non-cardiac cause for shock	No protocol. Pre-specified ECHO criteria of cardiac cause of shock	Clinical diagnosis with PAC parameters, ECG, biochemical markers, angiography, surgery, autopsy, patient chart, discharge diagnoses, death certificates. Cardiac cause – clinical positive/negative
P Geng et al.11	2022	Prospective, EICU, China	112	66.5 ± 13.5	Age >18 <95 years SBP < 90 mm Hg Signs of Hypoperfusion (altered mental status, resp distress, oliguria, fatigue, discomfort, mottling, elevated lactate and severe chest pain or abd pain Vasopressor dependent shock despite IV fluid challenge to achieve CVP of 8	Pre-existing hypotensive state from past medical history or reported by the patient. Transfer from another hospital with known diagnosis of shock No definite diagnosis of shock type established during hospitalization	Operator: Physician who had completed 20 hours of emergency ultrasound workshop including the THIRD protocol and 3 years of experience with > 300 US exams per year Equipment: Philips Sparq. High frequency 4–12 MHz linear probe, 2–6 MHz curvilinear probe and a 2–4 MHz cardiac probe	ECHO diagnosis of type of shock using THIRD protocol Agreement between ECHO and final diagnosis of shock subtype	THIRD protocolb	Three board certified physicians confirm final diagnosis of shock based on all relevant clinical data – history of presenting illness, signs, auxiliary examination results. Disagreement in diagnosis resolved by voting
Agmy et al.12	2017	Prospective, ICU, Egypt	63	Not available*	Tachycardia >100, non-palpable pulse, cool peripheries PP <20 mm Hg, pre-existing HTN, drop in SBP by 30%	NA	NA	Efficiency of transthoracic ultrasound (TUS) in hemodynamic assessment of shock	FALLS protocolc	History, clinical examination, laboratory investigations, chest X-ray, echocardiography, CT pulmonary angiography or other diagnostic tools

BP, blood pressure; CVP, central venous pressure; EICU, emergency intensive care unit; ^cFALLS, fluid administration limited by lung ultrasonography; HTN, hypertension; ICU, intensive care unit; NA, Not available; PAC, pulmonary artery catheter; PPV, positive pressure ventilation; ^aRUSH, rapid ultrasound for shock and hypotension; SBP, systolic blood pressure; ^bTHIRD, tamponade/tension pneumothorax, heart, inferior vena cava, respiratory system, deep venous thrombosis/aortic dissection; US, Ultrasound. ± indicates standard deviation. * Data are not available in the poster abstract

Quality Assessment

Figure 2 presents the results of the QUADAS-2 assessment. Vaidya et al.⁹ had a high risk of bias in the patient selection domain because it was not clear whether the patient selection was performed consecutively or randomly. The risk of bias with respect to applicability was high for Majo et al.¹⁰ because the index test was performed by experts and not intensivists, as in the other studies. Geng et al.¹¹ had a very low risk of bias in all domains. The index test and comparator were clearly defined in this study. The bias is low regarding applicability, as physicians trained in ultrasound diagnosed the type of shock in this study. In the study by Agmy et al.¹² there was only a conference abstract, the detailed definitions of the exclusion criteria and patient characteristics were not available, and the risk of bias with respect to patient selection and flow and timing is unclear.

Figs 2A and B.

Assessment of risk of bias of studies by QUADAS-2 tool

Summary Estimates

The SROC curves for each shock are shown in Figures 3A to 7A. The summary estimates of the sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio are shown in Table 2. The dots represent point estimates of sensitivity and 1-specificity for each included study, and the ellipses represent 95% CIs for sensitivity and 1-specificity. The pooled sensitivity for all types of shocks is shown in Figures 3B to 7B and pooled specificity is shown in Figures 3C to 7C.

Figs 3A to C.

(A) Summary receiver operating characteristic curves of cardiogenic shock; (B) Pooled sensitivity for diagnosis of cardiogenic shock; (C) Pooled specificity for diagnosis of cardiogenic shock

Figs 7A to C.

(A) Summary receiver operating characteristic curves of mixed shock; (B) Pooled sensitivity for diagnosis of mixed shock; (C) Pooled specificity for diagnosis of mixed shock

Table 2.

Sensitivities, specificities, likelihood ratios and diagnostic odds ratio by shock subtype

Shock type	No. of patients	Pooled sensitivity (95% CI)	Pooled Specificity (95% CI)	PLR (95% CI)	NLR (95% CI)	DOR (95% CI)	AUC
Cardiogenic	121	0.96 (0.91–0.99)	0.97 (0.94–0.99)	26.56 (13.39–52.64)	0.05 (0.01–0.25)	519.50 (104.02–2594.5)	0.97
Hypovolemic	38	0.90 (0.76–0.97)	0.99 (0.96–1.00)	39.49 (12.88–121.04)	0.09 (0.001–10.95)	707.98 (82.13–6102.9)	0.50
Distributive	85	0.81 (0.71–0.88)	0.91 (0.86–0.95)	8.8 (2.40–32.37)	0.20 (0.08–0.48)	53.892 (6.56–442.50)	0.64
Obstructive	25	0.89 (0.71–0.97)	1.00 (0.98–1.00)	137.56 (27.76–681.63)	0.15 (0.06–0.36)	999.29 (126.41–7899.5)	0.99
Mixed	45	0.79 (0.66–0.89)	0.98 (0.95–0.99)	48.17 (14.00–165.71)	0.22 (0.12–0.38)	318.70 (66.85–1519.3)	0.50

AUC, area under the curve; CI, confidence interval; DOR, diagnostic odds ratio; NLR, negative likelihood ratio; PLR, positive likelihood ratio

Figs 4A to C.

(A) Summary receiver operating characteristic curves of distributive shock; (B) Pooled sensitivity for diagnosis of distributive shock; (C) Pooled specificity for diagnosis of distributive shock

Figs 5A to C.

(A) Summary receiver operating characteristic curves of obstructive shock; (B) Pooled sensitivity for diagnosis of obstructive shock; (C) Pooled specificity for diagnosis of obstructive shock

Figs 6A to C.

(A) Summary receiver operating characteristic curves of hypovolemic shock; (B) Pooled sensitivity for diagnosis of hypovolemic shock; (C) Pooled specificity for diagnosis of hypovolemic shock

For all types of shocks, the positive likelihood ratio was above 25, except for distributive shock. The negative likelihood ratios ranged from 0.05 to 0.22. Among these shocks, the specificity and positive likelihood ratio for obstructive shocks were high. The diagnostic odds ratio was the highest for obstructive shock and the lowest for distributive shock. The SROC curve showed an AUC close to 1 for obstructive and cardiogenic shock. Table 2 depicts the summary estimates of the sensitivity, specificity, likelihood ratio, and diagnostic odds ratio for all types of shocks.

Discussion

This systematic review included studies with a total of 314 patients with shock and evaluated the diagnostic accuracy of ultrasound in diagnosing the etiology. A previous systematic review that investigated the diagnostic accuracy of ultrasound was conducted on patients presenting to the emergency department with shock. The limited number of studies on the diagnostic accuracy of ultrasound in critically ill patients with shock highlights the lack of sufficient evidence for this diagnostic tool to identify shock in an ICU setting. This review examined patients with shock in the ICU and all forms of ultrasound protocols to identify the type of shock.

Our meta-analysis showed pooled sensitivity for each type of shock ranged from 0.80 (mixed) to 0.96 (cardiogenic), and specificity ranged from 0.92 (distributive) to 1.00 (obstructive), and the AUC for each type of shock was quite variable anywhere from 0.50 to 0.99. The positive likelihood ratios exceeded 25 for all types of shock except distributive shock, and the negative likelihood ratios were as low as 0.05 (cardiogenic) and 0.22 (mixed).

The finding that ultrasound exhibits high diagnostic accuracy for both obstructive and cardiogenic shock is consistent with previous research in the fields of emergency and critical care medicine. Ultrasound has become an increasingly valuable tool for diagnosing various medical conditions, including different types of shocks. Numerous studies have reported the utility of ultrasonography in identifying cardiac tamponade (a form of obstructive shock) and cardiac function in patients with cardiogenic shock. These studies have emphasized the ability of ultrasound to provide rapid and non-invasive diagnostic information that can guide clinical management decisions.¹³ The high AUC values in the SROC curves for obstructive and cardiogenic shock are consistent with the concept that ultrasound can aid in the differentiation of these types of shock, which can be challenging based solely on clinical evaluation. The reported positive likelihood ratios above 25 for most types of shock indicate that a positive ultrasound result substantially increases the probability of shock. Similarly, the range of negative likelihood ratios from 0.05 to 0.22 suggests that a negative ultrasound finding is effective in “ruling out” shock. This is consistent with the idea that ultrasound can help exclude shock when applied appropriately. The high specificity and diagnostic odds ratio for obstructive shock supports the role of ultrasound in accurately identifying individuals without this type of shock. This is valuable for avoiding unnecessary interventions and guiding treatment decisions. The findings of this meta-analysis reinforce the existing body of literature that supports the use of ultrasound as a valuable diagnostic tool for identifying obstructive and cardiogenic shock. However, it is crucial to consider the quality of individual studies and the potential for bias, as highlighted in the QUADAS-2 assessment. Further research and standardization of ultrasound protocols in the diagnosis of shock can enhance their clinical utility and reliability in diverse healthcare settings. The significance of ultrasound lies in improving the diagnostic accuracy of the shock type, ultimately leading to more informed clinical decisions and improved patient outcomes.

In one systematic review of point-of-care ultrasound (POCUS) for identifying the type of shock in the emergency department, the sensitivity ranged from 85 to 100% and specificity ranged from 79 to 100%. The primary difference between our study and the previously mentioned systematic review is that we included only ICU patients with undifferentiated shock and not those in the emergency department. The SROC curve for obstructive shock in our study showed an AUC value of 0.99, similar to a previous systematic review. The AUC value for obstructive shock was slightly higher than that for other types of shock. This may be because obstructive shock is associated with ultrasound-detectable conditions such as tension pneumothorax and cardiac tamponade. Cardiogenic shock, with an AUC value of 0.94, was also accurately diagnosed using ultrasound in ICU patients. This might be due to the ability of ultrasound to detect conditions such as left or right ventricular failure or significant valvular disease. In contrast, the AUC values for distributive shock (0.59), hypovolemic shock (0.81), and mixed shock (0.82) were slightly lower.

Conclusion

This systematic review and meta-analysis showed that ultrasound has high diagnostic accuracy for detecting obstructive and cardiogenic shock in critically ill patients. However, it has moderate accuracy in detecting other types of shock. Therefore, ultrasound should be used in conjunction with clinical evaluation and other diagnostic tools to accurately determine the type of shock in critically ill patients. Further research and more high-quality studies are needed to determine the accuracy of ultrasound in identifying other types of shock in the ICU.

Ethical Declarations

Ethical Standards

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008 (5).

Data Availability

The original dataset is available from the corresponding author.

Disclaimer

This manuscript has been published as a preprint on Research Square (link: https://www.researchsquare.com/article/rs-4799965/v1) prior to its formal publication in Indian Journal of Critical Care Medicine journal.

Orcid

Lohith Karigowda https://orcid.org/0000-0001-5391-412X

Bhavna K Gupta https://orcid.org/0000-0002-3108-0408

Kush Deshpande https://orcid.org/0000-0003-0482-8158

Hatem Elkady https://orcid.org/0000-0002-8938-2820