Deep Impact of Ultrasound in the Intensive Care Unit: The “ICU-sound” Protocol

• Ultrasound examination is now being utilized in the intensive care unit to detect lung abnormalities and recruitment

• Transthoracic ultrasound examination can be used to diagnose a multitude of abnormalities, helped modify admitting diagnoses in 26% of patients, led to changes in medical therapy in 18% of patients, and prompted invasive procedures in 22% of patients

RAPID and accurate diagnosis and treatment are crucial and problematic for patients admitted to an intensive care unit (ICU). The inaccuracy of physical examination at admission to the ICU has been extensively reported. Different diagnostic imaging modalities have been developed, but most lack sensitivity, availability, and portability. Diagnostic accuracy can be increased when a brief echocardiographic study is added to extend the physical examination.1 

Ultrasonography has grown rapidly and gained widespread acceptance. In a recent study, up to 36% of patients admitted to an ICU with a noncardiac diagnosis had one or more occult cardiac abnormalities.2ICU patients often present with thoracic and abdominal pathologies, making ultrasound examination essential for prompt diagnosis and treatment and to prevent the patient's conditions from deteriorating or resulting in death.3,–,5Safe, accurate, rapid, and repeatable at the bedside, the use of ultrasound in the ICU has been extensively validated as it provides data that may not be obtained with other routine methods.6Furthermore, because of the patient's critical condition, diagnostic accuracy is essential at ICU admittance.

The “ICU-sound” protocol was developed for a rapid global assessment, which linked real-time ultrasonographic evaluation with clinical data. This study was designed to test the hypothesis that a head-to-toe ultrasound evaluation at ICU admittance could increase diagnostic accuracy. A prospective study was undertaken to investigate whether the protocol can be performed without encountering major environment- or patient-related limitations, detect occult anomalies, prompt urgent changes in therapy or further testing, and confirm or modify the admitting diagnosis.

Between March 2009 and January 2010, a total of 262 consecutive patients were admitted to the general ICU of Maria Vittoria Hospital, Torino, Italy. Excluded from the analysis were 121 patients who left the ICU alive within 48 h (usually postoperative or drug-poisoning patients); seven others because of lack of consent; and nine others because of patient-related conditions that strongly hampered ultrasound examination. The final data analysis was performed on 125 patients. The study design was approved by the ethical review board of our institution (Ethical Committee ASL TO2, Turin, Italy) and written informed consent was obtained. The ICU has six beds and the ICU staff consists of six physicians skilled in ultrasound.

The admitting diagnosis was made by the attending physician under the supervision of the senior physician and established on the basis of the diagnosis received from the department of referral, history taking, clinical examination, and laboratory and imaging findings. The admitting diagnosis was made without beside ultrasonographic evaluation.

Within 12 h of ICU admission, all patients underwent head-to-toe ultrasound examination performed by a different attending physician (not the same one as at admittance) under the “ICU-sound” protocol developed at our institution. Table 1reports the criteria for defining ultrasound-induced changes, confirmation, wrong evaluation, and lack of confirmation of the admitting diagnosis.

The decision to change the admitting diagnosis or medical therapy or to perform invasive procedures was taken by the senior physician jointly with the ICU director, considering or not a gold standard. Not included in the analysis were the data from a second and third examination performed by the attending physician according to changes in clinical conditions.

At the time of ICU admission, the severity of the patient's condition was graded according to the new Simplified Acute Physiology Score (SAPS II).7A report form completed for each examination at the patient's bedside immediately after sonography included items on environment- or patient-related limitations, ultrasound examination, and epidemiologic and clinical data on paper-based case report forms. The two ICU ultrasound systems were a CX50 (Philips Healthcare, Andover, MA) equipped with a 3–12 MHz linear probe, a 1–5 MHz sector probe, and a 1–5 MHz convex probe, and a Philips EnVisor C HD (Philips Healthcare) equipped a 5–8 MHz micro-convex probe, a 2–4 MHz sector probe, and a 2–5 MHz convex probe.

The ICU-sound protocol comprises the following ultrasound examinations:

Optic nerve. The optic nerve sheath diameter is measured 3 mm posterior to the papilla. This is done only in comatose or deeply sedated patients.

Chest. Six ultrasound areas are examined: anterior, lateral, and posterolateral views in the upper and lower thoracic walls on each side. The probe is moved ventral-to-dorsal on longitudinal and axial scans.

Heart. Basic views necessary to perform a goal-directed transthoracic echocardiography are used: parasternal long-axis and short-axis views, apical five-chamber, four-chamber and two-chamber views, and subcostal views. Transesophageal echocardiography is not used.

Abdomen. The probe is moved over six abdominal areas. The areas are epigastrium: longitudinal view; right hypocondrium: axial, longitudinal and coronal views; mesogastrial: axial and longitudinal views; left hypocondrium: coronal; hypogastrium: longitudinal; and right iliac fossa: axial view.

Venous system. Lower limb (right and left femoral and popliteal veins), upper limb (right and left basilica, cephalic and axillary veins), neck vessels (right and left jugular veins), mild compression maneuver.

On completion of each examination, the transducer is cleaned with a germicidal detergent. For the purposes of this study, specific diagnostic challenges and critical abnormalities were prospectively defined (table 2). Findings without a direct impact on diagnosis and/or therapy (renal or hepatic cysts, gallstones, prostatic hypertrophy, hepatic steatosis) were not taken into account.

The association between ICU mortality and number of pathologic findings was evaluated using logistic regression models. Patients were classified into three groups according to the number of pathologic findings: none, only 1, 2, or more. Crude and SAPS-adjusted odds ratios (ORs) were reported along with its 95% CI. In the estimation of SAPS-adjusted ORs, the SAPS score was included as continuous variable in the logistic model. Cohen's κ was applied to measure agreement between pairs of readers and averaged over all pairs to evaluate overall agreement.8Statistical analysis was carried out with R 2.15.0 (R Foundation for Statistical Computing, Vienna, Austria).

The study population was 125 patients. Table 3reports patient demographics; table 4reports the distribution of admitting diagnosis and SAPS II scores. In nine cases, patient-related conditions (obesity, subcutaneous emphysema, bandages, digestive gas) strongly hampered the examination in some areas, so that the protocol could not be performed entirely; these cases were excluded from the study. Environment-related limitations were defined as brightness of room illumination, space around the patient, noise, and isolation. Examinations were carried out under 0–1 limiting conditions resulted in “good or slightly impaired quality of the examination” in 101/125 (80.8%) cases, under 2 limiting conditions resulting in “moderately impaired quality” in 20/125 (16%) cases, and under 3–4 limiting conditions resulting in “impaired quality” but still possible in 4/125 (3.2%) cases. The time needed to complete scanning ranged from 17 to 54 min (median, 19.5 min).

To simplify data collection, the ultrasound findings were subdivided in three categories: new pathologic findings (previously unknown abnormalities); pathologic findings (already known abnormalities); and normal findings. All new pathologic findings were downloaded and stored on a server. This was done to allow us to later retrieve the findings and discuss difficult cases and images with colleagues.

Examinations showed “normal” findings in 18/125 patients (14.4%) and ultrasound abnormalities in 107/125 patients (85.6%). Table 5reports the distribution of the abnormal findings. In the group of 107 patients with ultrasound abnormalities, examination revealed 254 pathologic finding(s), 136 of which were classified as “new pathologic finding(s).” It is clear that the same patient could have had more than one pathologic finding. Table 6illustrates the number of pathologic findings observed and the related ICU mortality.

Ultrasound examination confirmed the admitting diagnosis in 73/125 cases (58.4%), modified it in 32/125 cases (25.6%), was ineffective in confirming or modifying it in 17/125 cases (13.6%), and missed it in 3/125 cases (2.4%).

The ultrasonographic findings prompted: 1) further testing in 23/125 patients (18.4%, 95% CI: 12.0–26.3%); 2) changes in medical therapy in 22/125 patients (17.6%, 95% CI: 11.4–25.4%; and 3) invasive procedures in 27/125 patients (21.6, 95% CI: 14.7–29.8%). The changes in medical therapy in the 22 patients were subdivided into: administration of a thrombolytic drug (n = 1); increase in dosage of low-molecular-weight heparin (n = 2); administration of nitric oxide (dilated right ventricle, right atrial with overload pattern, severe tricuspid regurgitation in patients with the adult respiratory distress syndrome) (n = 3); administration of an inotropic drug (n = 6); administration of a vasoactive drug (n = 4); diuretic therapy (n = 3); and antibiotic therapy (n = 3). Subsequent to ultrasound examination, 27 invasive procedures were performed: thoracic drainage (n = 23); pericardiocentesis (n = 1); paracentesis (n = 1); and emergency bronchoscopy (n = 2). The procedures were carried out under ultrasound guidance, and no procedure-related complications occurred.

In 61/125 cases (48.8%), the senior physician ordered a gold standard procedure (computed tomography, magnetic resonance imaging, transesophageal echocardiography) that confirmed the ultrasonographic findings. Ultrasound examination missed abnormalities in three cases: Pancreatitis (n = 1) and pneumonia (n = 1) were demonstrated by computed tomography, and transesophageal echocardiography revealed a severe aortic insufficiency (n = 1).

ICU mortality was higher, but not statistically significant, in patients with two or more pathologic findings respect to patients without any findings (SAPS-adjusted OR 2.49, 95% CI: 0.79–7.85) (table 6). The level of agreement between two sets of dichotomous findings was compared using Cohen's κ. We created a crosstabs table to compare the ultrasonographic findings recorded by the two physicians. The average κ coefficient of agreement in interpretation was 0.69 for all readers; the κ values ranged from 0.52 to 0.86, confirming good agreement among all observers.

Our study is unique in that it is a surveillance study of 125 consecutive patients whose evaluation entailed a global assessment encompassing clinical and ultrasound examination. Unlike previous works,2,–,6it was performed to evaluate a protocol for head-to-toe ultrasound examination performed by a team of ICU physicians experienced in ultrasonography. Real-time ultrasound evaluation (within 12 h of admission) was done by the attending physician. The approach is based on the assumption that intensive care physicians with enough expertise can interpret sonographic images.

Ultrasound training is not mandatory for ICU physicians in Italy, although proposals for graded competence have been advanced.9,–,11Since 2000, the Society for Academic Emergency Medicine guidelines for physician training in emergency ultrasonography12have informed the continuing medical education program of our ICU physicians. During their time in the course, physicians are required to perform 150 practice sections, including 60 heart, 10 chest, 60 abdomen, and 20 peripheral vein ultrasound examinations. After the course, the physicians receive 1 yr of tutored instruction.

Drawn up in 2008, the ICU-sound protocol was tested for 6 months before entering daily clinical practice. Our data suggest that the protocol is not excessively time-consuming: a mean duration of 19.5 min to complete the scan is acceptable considering the protocol's complexity. Patient- and environment-related ultrasonography limitations had little influence on the majority of the ultrasound examinations in this study.

We used optic nerve sonography as an additional noninvasive diagnostic tool to detect increased intracranial pressure preceding emergency computed tomography or the decision to start invasive monitoring of intracranial pressure.13,–,15Ultrasound examination revealed new findings of optic nerve sheath enlargement in six cases, in all of which computed tomography showed unknown intracranial hypertension, demonstrating the importance of this easy-to-perform ultrasound evaluation at admittance to a general ICU.

Two cases of unknown deep venous thrombosis were identified, one an internal jugular vein thrombosis (fig. 1). Upper extremity venous thrombosis is thought to be quite rare, but a recent report found that 18% of all cases involve upper extremity thrombi.16Patients in the ICU setting are especially vulnerable to developing upper extremity thrombosis, and the clinical symptoms associated with an upper-extremity clot are frequently absent because of the extensive collateral network.17,18A combined strategy using echocardiography and venous ultrasonography is a reliable method for diagnosing pulmonary embolism at the bedside.19,20 

Pleural effusion is often encountered in ICU patients; the diagnosis relies mostly on an anteroposterior chest radiograph obtained at the bedside. Pleural ultrasonography, an alternative imaging modality, has been validated against chest computed tomography, the accepted reference for diagnostic methods to identify pleural disease.21,22The prevalence of significant pleural effusions in a medical ICU varies widely from 8.4 to 62%.23,24 

In our study, effusions (small effusions less than 200 ml were not considered) were found in 35/125 patients (28%), 20 (57.1%) of whom underwent drainage and thoracenthesis. The estimated drainage volume was obtained using the formula proposed by Balik et al.  25It has been recently shown, however, that this formula can underestimate the volume,26whereas a multiplane approach can increase the accuracy of quantifying small and moderate pleural effusion. Although more time-consuming (10 min), an accurate evaluation of effusion volume is a critical element in deciding whether to perform thoracenthesis. And it becomes even more relevant when weaning is considered, because drainage may reduce the work of breathing and increase respiratory muscle efficiency.27 

The decision to drain was based on our clinical practice (reduced chest wall compliance, difficult weaning, refractory hypodynamic circulatory states) or when ultrasound suggested an infectious effusion (homogeneous echogenicity, septation, fibrin strands, nodular pleural changes). Pleural effusion can cause dissociation between effective preload and cardiac filling pressures. Drainage may be effective in patients with refractory hypodynamic circulatory states, particularly when there is evidence for diastolic chamber collapse.28 

Ultrasound examination was effective in evidencing the presence of anterior pneumothorax not detected by supine anteroposterior chest radiography (three cases). This finding holds clinical relevance, because during positive pressure ventilation a small pneumothorax may progress and cause hemodynamic instability. Ultrasound has proved to be more sensitive than anteroposterior chest radiography in the diagnosis of pneumothorax29,30and can decrease the need for computed tomography for the diagnosis of occult pneumothorax.31Lower lobe parenchymal consolidation without air bronchogram visualization can be difficult to distinguish from pleural effusion on an anteroposterior chest radiograph.32Ultrasonography showing consolidation (with or without pleural effusion) can help in avoiding a possible mistake because of a misread chest radiograph (fig. 2).

Lung sonography is a useful aid in differentiating cardiogenic respiratory failure from acute airflow limitation, consolidation, pleural effusion, or pulmonary embolism.33,–,35In our study, the diagnostic efficacy of lung ultrasound to differentiate dyspneic patients is well represented. Lung ultrasound pointed out 55 new findings, enabling us to differentiate the etiologic diagnosis in patients with a generic admitting diagnosis of acute respiratory insufficiency. Combining the data from lung sonography and echocardiography can enhance the diagnostic accuracy in differentiating respiratory insufficiency (fig. 3).

In a previous study, transthoracic echocardiography performed by intensivists provided new information and changed management in 37% of critical patients and added useful information to an additional 47% of patients, but did not alter immediate treatment.36In our study, echocardiography revealed 53 new findings, leading to changes in therapy in 11/125 patients. These data confirm the importance of transthoracic echocardiography in ICU patients. Abdominal examination evidenced 20 new pathologic findings and induced changes in therapy in 3/125 cases. The presence of the classic signs37of acute acalculous cholecystitis at ultrasound examination, together with clinical data, enabled us to identify the origin of sepsis of unknown etiology (fig. 4).

In our study, new ultrasound abnormalities were most often detected in patients with septic shock, followed by those with acute cardiac decompensation. This is not surprising, as these patients have multiple organ failure. ICU mortality was higher, but not statistically significant, in patients with two or more pathologic findings compared with patients without any findings. Mortality generally depends on the specific type of the most life-threatening abnormality, along with many other factors unrelated to ultrasound findings. Of note is that patients with no ultrasonographic pathologic findings had a better prognosis.

We plan to extend the ICU-sound protocol to include other diagnostic tools: transcranial Doppler and positive end-expiratory pressure-induced lung recruitment. Recent reports have shown, in fact, that lung reaeration can be assessed by ultrasound. It is rapid and repeatable at the bedside, does not require sedation and paralysis, and can be used to analyze dependent and not-dependent lung regions. Patients with acute lung injury, the adult respiratory distress syndrome, and pulmonary edema may benefit from ultrasound monitoring of positive end-expiratory pressure-induced reaeration;38,–,40however, a limitation is that lung hyperinflation cannot be assessed.

There is a potential for bias, as the operators in this study were not blinded to the patient's clinical picture, which is difficult to entirely eliminate in any use of ultrasound. The study relied only on transthoracic echocardiography; therefore, the true prevalence of cardiac ultrasound abnormalities may have been underestimated. Also, the prevalence of pleural effusion may have been underestimated with the use of a formula, which has been shown inadequate to quantify small and moderate effusions. An evaluation of the usefulness of a head-to-toe versus  a focused examination was beyond the scope of this study.

The major finding of this study is the discovery of a high prevalence of unsuspected clinical abnormalities in ICU patients. Ultrasound examination permitted us to modify the diagnosis and to improve the treatment, with prompt changes in therapeutic strategy. Patient- and environment-related sonography limitations had little influence on the majority of the examinations. The test can be performed by the attending ICU physician, with minimal risk of overuse or misinterpretation. Moreover, ultrasound use in the ICU could be optimized by making ultrasonography a routine part of intensive care training during residency.41In experienced hands, rapid global assessment of critical patients at ICU admittance under this ultrasound protocol holds potential for improving healthcare quality.