CHEST

Original Research
CHEST ULTRASONOGRAPHY

Assessment of Left Ventricular Function by Intensivists Using Hand-Held Echocardiography*

Roman Melamed, MD; Mark D. Sprenkle, MD; Valerie K. Ulstad, MD; Charles A. Herzog, MD; and James W. Leatherman, MD, FCCP

Background: Bedside transthoracic echocardiography (TTE) provides rapid and noninvasive hemodynamic assessment of critically ill patients but is limited by the immediate availability of experienced sonographers and cardiologists. Methods: Forty-four patients in the medical ICU underwent near-simultaneous limited TTE performed by intensivists with minimal training in echocardiography, and a formal TTE that was performed by certified sonographers and was interpreted by experienced echocardiographers. Intensivists, blinded to the patients diagnosis and the results of the formal TTE, were asked to determine whether left ventricular (LV) function was grossly normal or abnormal and to place LV function into one of the following three categories: 1, normal; 2, mildly to moderately decreased; and 3, severely decreased. Results: Using the formal TTE as the gold standard, intensivists correctly identified normal LV function in 22 of 24 cases (92%) and abnormal LV function in 16 of 20 cases (80%). The statistic for the agreement between intensivist and echocardiographer for any abnormality in LV function was 0.72 (95% confidence interval [CI], 0.52 to 0.93; p < 0.001). Intensivists correctly placed LV function into one of three categories in 36 of 44 cases (82%); in 6 of the 8 cases that were misclassified, the error involved an overestimation of LV function. The statistic for agreement between the intensivist and echocardiographer with regard to placement into one of three categories of LV function was 0.68 (95% CI, 0.48 to 0.88; p < 0.001). Conclusions: Intensivists were able to estimate LV function with reasonable accuracy using a hand-held unit in the ICU, despite having undergone minimal training in image acquisition and interpretation. (CHEST 2009; 135:1416 1420)
Abbreviations: ED emergency department; LV left ventricle, ventricular; TTE transthoracic echocardiography

ritically ill, hypotensive patients should undergo C hemodynamic assessment as quickly as possible. Unfortunately, most invasive methods for assessing
*From the Divisions of Pulmonary-Critical Care (Drs. Melamed, Sprenkle, and Leatherman) and Cardiology (Drs. Ulstad and Herzog), Hennepin County Medical Center, Minneapolis, MN. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript received October 10, 2008; revision accepted January 13, 2009. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/site/misc/reprints.xhtml). Correspondence to: James W. Leatherman, MD, FCCP, Division of Pulmonary and Critical Care, Hennepin County Medical Center, Minneapolis, MN 55415; e-mail: leath001@umn.edu DOI: 10.1378/chest.08-2440
1416

hemodynamics involve delays in implementation. One diagnostic tool that can be used quickly and noninvasively is limited, point-of-care, bedside transthoracic echocardiography (TTE). Unlike formal For editorial comment see page 1407 TTE, the limited examination focuses primarily on exclusion of hemodynamically significant pericardial effusion and assessment of global left ventricular (LV) function.15 Assessment of LV function is of particular importance because decreased cardiac contractility is a common cause of hemodynamic instability in critically ill patients,6, including those with sepsis.7,8 The comprehensive TTE has a high sensitivity and specificity for defining a cardiac cause of shock when
Original Research

Downloaded From: http://journal.publications.chestnet.org/ on 12/07/2013

performed by experienced sonographers and interpreted by cardiologists trained in echocardiography.9 Unfortunately, sonographers and cardiologists are not always immediately available. Because of this limitation, limited TTE has become increasingly used in the emergency department (ED).1 4 Studies have shown that emergency medicine physicians can assess LV function with reasonable accuracy1 4and that the focused TTE results in earlier and more accurate diagnosis.3 Surprisingly, there have been very few studies5,10 of the use of limited TTE in the ICU. The current study examined the ability of intensivists with minimal training in echocardiography to accurately assess LV systolic function using a hand-held unit, with interpretation of a simultaneously performed comprehensive TTE by an experienced echocardiographer the reference standard.

Table 1Normal vs Abnormal LV Function

Formal TTE by Echocardiographer Limited TTE by Intensivist Normal findings Abnormal findings Normal Findings 22 2 Abnormal Findings 4 16

Materials and Methods

Study Population This study was performed in the medical ICU of Hennepin County Medical Center, an urban teaching hospital affiliated with the University of Minnesota. The study population included 44 patients who had a formal TTE ordered by their primary physician because of one or more of the following indications: hypotension (n 19); dyspnea-hypoxemia (n 17); unexplained tachycardia (n 4); increased troponin levels (n 3); and abnormal ECG (n 3). Training and Interpretation of Focused TTE by Intensivists Prior to beginning the study, intensivists involved in the study underwent 2 h of didactic instruction on echocardiography and 4 h of hands-on training by certified ultrasonographers in image acquisition and visual estimate of the LV function. In addition, independent study by participating intensivists was encouraged by providing digitally stored examples of TTEs in which LV function ranged from normal to severely decreased. Within 2 h of the formal TTE, patients underwent a focused bedside TTE by one of four intensivists who used a hand-held unit (SonoSite 180 [with C15/4-2 mHz MCX transducer]; SonoSite; Bothell, WA). Intensivists were blinded to the results of the formal study and to the patients underlying diagnosis. Intensivists attempted to acquire images in the parasternal (long and short axis), apical, and subcostal positions. On the basis of the limited examination, the intensivists were asked to place LV function into one of the following three categories: 1, normal; 2, mild-to-moderate decrease in contractility; and 3, severe decrease in contractility. Comparative Assessment Between Intensivist and Cardiologist The primary objective of the study was to compare the intensivists assessment of LV function using the limited TTE with that of an echocardiographers assessment of LV function from the comprehensive TTE. Formal echocardiograms (Sequoia C 512; Siemens; Malvern, PA) were obtained by certified sonographers and interpreted by experienced echocardiographers. All studies performed by intensivists were captured with a videocaswww.chestjournal.org

sette recorder attached to the portable ultrasound device for later review. Whenever there was discordance between the interpretation of the limited TTE by the intensivist and the formal TTE by the echocardiographer, a different echocardiographer determined whether the discordance was primarily due to poor image quality of the limited TTE or to misinterpretation by the intensivist. Statistical analysis was performed with a statistical software package (SPSS, version 11.5; SPSS; Chicago, IL). The statistic was used to allow for comparison in agreement between intensivist and cardiology interpretation of TTE. The institutional review board of the hospital approved the study without need for informed consent.

Results The 44 patients included 28 men (64%) and 16 women (36%) with a mean ( SD) age of 60 16 years. The most common primary diagnoses were sepsis (n 10), pulmonary edema (n 6), pneumonia (n 6), and seizure (n 5); the remaining 13 patients had other miscellaneous conditions. Eighteen of the 44 patients (41%) were intubated, and 8 patients (18%) required therapy with vasopressors. The time required to complete the limited TTE was 10 min in most cases. We first assessed the ability of intensivists to accurately discriminate normal from abnormal LV systolic function. Based on the results of the comprehensive TTE, there were 24 studies with normal LV function (category 1) and 20 with abnormal LV function (categories 2 and 3). Using the comprehensive TTE as the gold standard, intensivists correctly differentiated normal and abnormal LV function on the basis of the limited TTE in 38 of 44 cases (86%). Normal LV function was correctly identified in 22 of 24 cases (92%), and abnormal LV function was identified in 16 of 20 cases (80%) [Table 1]. The positive predictive value for the intensivists identifying any abnormality in LV function was 89%; the negative predictive value was 85%. The statistic for the agreement between intensivist and echocardiographer for any abnormality in LV function was 0.72 (95% confidence interval, 0.52 to 0.93; p 0.001). Next, we studied the ability of intensivists to assess the degree to which LV function was decreased. Of the 20 studies in which the comprehensive TTE showed decreased LV function, the degree of abnormality was deemed to be mild to moderate in 13
CHEST / 135 / 6 / JUNE, 2009

1417

Downloaded From: http://journal.publications.chestnet.org/ on 12/07/2013

Table 2LV Function by Category*

Limited TTE by Intensivist Category 1 Category 2 Category 3 Formal TTE by Echocardiographer Category 1 22 2 0 Category 2 4 9 0 Category 3 0 2 5

*Category 1, normal LV function; category 2, mild-to-moderate decrease in LV function; category 3, severe decrease in LV function.

studies and severe in 7 studies (Table 2). The intensivist correctly categorized 9 of the 13 patients (69%) with mild-to-moderate LV dysfunction; LV function was incorrectly assessed as normal in the remaining four cases (Table 2). Severe LV dysfunction was correctly recognized by the intensivists in five of seven cases (71%), with the two remaining cases misclassified as mild-to-moderate dysfunction. Overall, intensivists correctly placed LV function into the appropriate category in 36 of 44 cases (82%); in 6 of the 8 cases that were misclassified the error involved an overestimation of LV function (Table 2). The statistic for agreement between the intensivist and echocardiographer with regard to placement into one of three categories of LV function was 0.68 (95% confidence interval, 0.48 to 0.88; p 0.001). Intensivists often did not obtain good quality images from all four transducer positions. Nonetheless, the cardiologist who reviewed the video recordings of the limited TTEs felt that there were only four instances in which LV function could be assessed with only a low degree of certainty. With a single exception, when there was discordance between the intensivist interpretation of the limited TTE and the comprehensive TTE, the cardiologist who reviewed the limited TTE tapes sided with the formal reading rather than the intensivist. This indicated that the most common source of error was misinterpretation on the part of the intensivist rather than inadequate image quality. Discussion The goal of this prospective observational study was to determine whether intensivists who had undergone brief and focused training in echocardiography were able to assess LV function accurately using a miniaturized ultrasound unit. Using a nearsimultaneous comprehensive TTE as the gold standard, intensivists were able to differentiate normal and abnormal LV function correctly in 86% of cases; in 82% of cases the intensivists correctly classified LV function as normal, mildly to moderately im1418

paired, or severely impaired. When mistakes were made, intensivists most often overestimated LV function. Cardiac abnormalities are common in critically ill patients and may be unsuspected on clinical grounds despite their hemodynamic significance.5 Joseph et al9 found that a comprehensive TTE resulted in a change of management in 51% of patients who were in shock. Similarly, Kaul et al11 found that a formal TTE had a high level of agreement with the pulmonary artery catheter findings in patients with hemodynamic compromise and provided complementary information in cases with discordance. Unfortunately, a formal TTE with cardiologist interpretation is not always readily available in the ICU. Our study suggested that minimally trained intensivists using a hand-held unit can make a reasonably accurate assessment of LV function in the ICU setting, despite the potential challenges of performing optimal TTE in the ICU because of factors such as lung hyperinflation, anasarca, dressings or tubes, and inability to position the patient optimally. Surprisingly, few studies have specifically examined the ability of minimally trained intensivists to assess LV function in the ICU. Using the interpretation of expert cardiologists as the gold standard, Manasia et al5 found that surgical intensivists correctly interpreted 84% of limited TTEs, and this new information changed management in 37% of cases. Similarly, Vignon and associates10 found that residents with minimal training accurately differentiated normal and abnormal LV function in 88% of patients in a mixed medical-surgical ICU, with interpretation by intensivists experienced in TTE serving as the reference. Our study design was slightly different from these two studies in that we examined medical ICU patients exclusively and compared intensivist interpretation of the limited TTE with the results of a formal TTE interpreted by an experienced echocardiographer, but we found a similar degree of accuracy in assessment of LV function. These three studies strongly suggest that minimally trained intensivists using hand-held ultrasound can reliably assess LV function in the great majority of cases. Limited TTE performed by noncardiologists has also been shown to be reasonably accurate in settings other than the ICU.15,1214 In an ED study3 of 184 patients presenting with nontraumatic hypotension, limited bedside TTE resulted in an improvement in diagnostic accuracy. In a study similar to ours, Moore et al1 demonstrated that ED physicians with limited echocardiographic training were able to estimate LV function accurately in 50 hypotensive patients. The statistic for agreement between ED physicians and cardiologists for normal, moderately depressed, or severely depressed LV
Original Research

Downloaded From: http://journal.publications.chestnet.org/ on 12/07/2013

function was 0.61; the statistic for agreement between two different cardiologists who examined a random sample of 20 cases was 0.70.1 The latter finding indicates that disagreement regarding assessment of LV function between experienced echocardiographers is not uncommon. Even though there are obvious benefits to intensivistperformed limited TTE in the ICU, the misinterpretation of studies could lead to potentially serious errors in clinical decision making. For example, inotropic therapy might be withheld from patients with hemodynamic impairment resulting from decreased LV function if the TTE was misinterpreted as normal. Conversely, a patient with normal LV function whose hypotension was due to hypovolemia or vasodilation might be inappropriately treated with an inotrope if LV function was incorrectly deemed to be abnormal. In the present study, the most common error was failure to recognize a decrease in LV function rather than misinterpreting normal LV function as abnormal. It should also be recognized that limited TTE, as performed by minimally trained individuals, could fail to identify other important cardiac causes of hemodynamic compromise such as cor pulmonale, acute valvular abnormalities, and intracardiac shunts.11 Indeed, noncardiologists with minimal training often fail to identify important cardiac abnormalities such as valvular disease and regional wall abnormalities.14 Our study was limited to the evaluation of LV systolic function, but sometimes other information provided by TTE may be even more important in managing hemodynamically unstable patients. As previously mentioned, TTE is an excellent tool to diagnose acute cor pulmonale and severe valvular insufficiency. In addition, respiratory variation in inferior vena cava diameter may help predict the likelihood of fluid responsiveness, and measurement of flow across the aortic valve can be used to estimate cardiac output and calculate systemic vascular resistance.15,16 Future studies should examine the ability of intensivists to gain proficiency in these additional uses of TTE. The present study has several important limitations. First, we studied a relatively small number of patients and examined the performance of only a few intensivists. Second, although none of the intensivists had had any formal training in TTE before the study, all had had some prior exposure to formal TTEs during their clinical practice. Third, only a limited number of imaging planes were used in some cases, and we did not evaluate the comparative utility of different transducer positions. Finally, we did not assess the benefit of the administration of a contrast agent, which has been shown17 to improve image acquisition by novice sonographers.
www.chestjournal.org

In conclusion, medical intensivists with minimal training in TTE obtained adequate images with handheld ultrasound units and accurately differentiated normal and abnormal LV systolic function in the great majority of cases. Intensivists were somewhat less accurate in differentiating mild-to-moderate LV dysfunction from severe LV dysfunction, but severe LV dysfunction was never misclassified as normal. When errors occurred, they tended to involve the overestimation of LV function. Although these preliminary data support the use of intensivist-performed TTE in the ICU, they also suggest that the optimal use of this noninvasive tool in the ICU may require a more rigorous training program, the components of which include initial training in image acquisition by skilled ultrasonographers, extensive hands-on experience with focused TTE, and ongoing feedback from experienced echocardiographers to improve the intensivists accuracy of interpretation.

References
1 Moore CL, Rose GA, Tayal VS, et al. Determination of left ventricular function by emergency physician echocardiography of hypotensive patients. Acad Emerg Med 2002; 9:186 193 2 Randazzo MR, Snoey ER, Levitt MA, et al. Accuracy of emergency department physician assessment of left ventricular ejection fraction and central venous pressure using echocardiography. Acad Emerg Med 2003; 10:973977 3 Jones AE, Tayal VS, Kline JA. Focused training of emergency medicine residents in goal-directed echocardiography: a prospective study. Acad Emerg Med 2003; 10:1054 1058 4 Jones AE, Tayal VS, Sullivan DM, et al. Randomized, controlled trial of immediate versus delayed goal-directed ultrasound to identify the cause of nontraumatic hypotension in emergency department patients. Crit Care Med 2004; 32: 17031708 5 Manasia AR, Nagaraj HM, Kodali RB, et al. Feasibility and potential clinical utility of goal-directed transthoracic echocardiography performed by noncardiologist intensivists using a small hand-carried device (SonoHeart) in critically ill patients. J Cardiothorac Vasc Anesth 2005; 19:155159 6 Bossone E, DiGiovine B, Watts S, et al. Range and prevalence of cardiac abnormalities in patients hospitalized in a medical ICU. Chest 2002; 122:1370 1376 7 Maeder M, Fehr T, Rickli H, et al. Sepsis-associated myocardial dysfunction: diagnostic and prognostic impact of cardiac troponins and natriuretic peptides. Chest 2006; 129:1349 1366 8 Kan H, Fallinger CF, Fang Q, et al. Reversible myocardial dysfunction in sepsis and ischemia. Crit Care Med 2005; 33:28452847 9 Joseph MX, Disney PJ, Da Costa R, et al. Transthoracic echocardiography to identify or exclude cardiac cause of shock. Chest 2004; 126:15921597 10 Vignon P, Dugard A, Abraham A, et al. Focused training for goal-oriented hand-held echocardiography performed by noncardiologist residents in the intensive care unit. Intensive Care Med 2007; 33:17951799 11 Kaul S, Stratienko AA, Pollock SG, et al. Value of twodimensional echocardiography for determining the basis of hemodynamic compromise in critically ill patients: a prospective study. J Am Soc Echocardiogr 1994; 7:598 606
CHEST / 135 / 6 / JUNE, 2009

1419

Downloaded From: http://journal.publications.chestnet.org/ on 12/07/2013

12 Royse CF, Seah JL, Donelan L, et al. Point of care ultrasound for haemodynamic assessment: novice compared with an expert operator. Anaesthesia 2006; 61:849 855 13 Hellmann DB, Whiting-OKeefe Q, Shapiro EP, et al. The rate at which residents learn to use hand-held echocardiography at the bedside. Am J Med 2005; 118:1010 1018 14 Alexander JH, Peterson ED, Chen AY, et al. Feasibility of point-of-care echocardiography by internal medicine house staff. Am Heart J 2004; 147:476 481

15 Jardin F, Viellard-Baron A. Ultrasonagraphic examination of the vena cavae. Intensive Care Med 2006; 32:203206 16 Brown JM. Use of echocardiography for hemodynamic monitoring. Crit Care Med 2002; 13611364 17 Makaryus AN, Zubrow ME, Gillam LD, et al. Contrast echocardiography improves the diagnostic yield of transthoracic studies performed in the intensive care setting by novice sonographers. J Am Soc Echocardiogr 2005; 18: 475 480

1420

Original Research

Downloaded From: http://journal.publications.chestnet.org/ on 12/07/2013