The study was carried out in the Department of Intensive Care Medicine of a University general hospital with 16 polyvalent beds that attends 100 post-heart surgery patients a year.

We included all patients over 18 years of age subjected to heart surgery and consecutively admitted to the Intensive Care Unit (ICU) between November 2014 and April 2016, with the physical presence of an intensivist with certified training in basic echocardiography.

The study protocol was approved by the local Clinical Research Ethics Committee. All patients gave written informed consent to participation in the study.

At the time of patient admission, we recorded the general demographic characteristics, the reason for admission, main comorbidities, severity scores (APACHE II, EuroSCORE) and hemodynamic and echocardiographic data prior to surgery, if available.

A clinical and hemodynamic evaluation of the patient was made by one of the 6 intensivists of the Department to establish an initial diagnosis, in accordance with routine clinical practice in post-heart surgery patients: hypovolemia, left ventricular dysfunction, right ventricular dysfunction, biventricular dysfunction, vasoplegia, massive mitral valve insufficiency or tamponade, and treatment was prescribed based on the diagnosis.

Following the initial clinical evaluation and within the first 8 h of admission, another intensivist trained in basic echocardiography performed basic TTE. At the time of the exploration, the use of mechanical ventilation was documented, as well as the administration of vasoactive medication, and the hemodynamic parameters (heart rate, blood pressure, central venous oxygen saturation, lactate concentration).

The echocardiographic study was repeated daily in the presence of signs of tissue hypoperfusion (arterial hypotension, evidence of poor peripheral perfusion, elevated lactate levels, oliguria). The professional performing the echocardiographic evaluation was blinded to the initial tentative diagnosis.

The echocardiographic analysis comprised systematic assessment of the long and short axes parasternal projections and four-chambers apical and four-chambers subcostal projections with visualization of the inferior vena cava. For each acoustic window, imaging difficulty was rated as either high or low. All explorations were made using the 1.5–4.6 MHz M5S-D echocardiographic probe of the Vivid-e (General Electrics) system, available 24 h in the ICU of our center.

The analyzed echocardiographic parameters were:

• 1)

  Left ventricle: visual size defined as normal or dilated, visual systolic function classified as normal (≥ 50%) or depressed (< 50%), and contraction pattern defined as either homogeneous or heterogeneous.

• 2)

  Right ventricle: visual size defined as normal or dilated, and visual systolic function (normal or depressed).

• 3)

  Valves: presence of massive mitral valve insufficiency.

• 4)

  Inferior vena cava: classified as normal, dilated or not visualized.

• 5)

  Pericardial fluid with or without cardiac tamponade, classified as either present or absent.

The echocardiographic patterns were defined as5,6,27:

• •

  Severe hypovolemia: small hyperdynamic ventricles, with obliteration of the left ventricular cavity (kissing walls), and small inferior vena cava showing great variation with respiratory movements (under mechanical ventilation, inferior vena cava < 1.5 cm with collapse > 50%; under spontaneous ventilation, inferior vena cava < 1 cm with collapse > 50%).

• •

  Left ventricular failure: reduction of global left ventricular motility, heterogeneous contractility pattern suggesting myocardial ischemia, or dilatation of the left ventricular cavity suggesting chronic heart disease.

• •

  Right ventricular failure: dilatation of the right ventricle and paradoxical motion of the interventricular septum - which in the critical patient is basically associated to acute respiratory distress syndrome (ARDS) or severe pulmonary thromboembolism. Isolated right ventricular dilatation is suggestive of right ventricular infarction. Other associated findings may be a dilated inferior vena cava without collapse.

• •

  Tamponade: presence of pericardial effusion with diastolic collapse of the right atrium and ventricle, and dilated inferior vena cava without respiratory collapse under spontaneous breathing conditions.

• •

  Acute massive mitral valve insufficiency: left ventricle of normal size (indicative of acute valve disease), left ventricle with normal or hyperdynamic systolic function secondary to volume overload of the left ventricle and massive reflow in the color Doppler exploration.

All the explorations were digitally recorded and later evaluated by an intensivist with advanced training in TTE (A.O.), rating the quality of the images as optimal (complete identification of the endocardial margins of the left ventricle), suboptimal (identification of > 50% of the endocardial margins of the left ventricle) or poor (identification of < 50% of the endocardial margins of the left ventricle, or visualization not possible). The analysis was based only on the optimal and suboptimal echocardiographic images. We determined the percentage of cases in which echocardiography confirmed the initial tentative or suspected diagnosis. If this diagnosis was not confirmed, we determined the percentage of cases in which echocardiography afforded an alternative diagnosis and the percentage of changes in treatment. Lastly, clinical improvement associated to the different therapeutic changes was analyzed.

In the event of doubt regarding the echocardiographic data, a second echocardiographic assessment was made by a cardiologist with expertise in advanced echocardiography.

A descriptive study was made, analyzing percentage clinical improvement secondary to treatment optimization. Continuous variables were reported as the median and standard deviation (SD), or as percentages.

The study sample consisted of 136 post-heart surgery patients consecutively enrolled in the period between November 2014 and April 2016, and representing a total of 203 echocardiographic studies. No patients were excluded during the study period. The demographic and clinical data of the sample are reported in Table 1.

With regard to the type of heart surgery, most patients underwent valve replacement either isolatedly (51 patients; 37.5%) or associated to other procedures (22 patients; 16.3%). A total of 64 coronary revascularization surgeries were performed (47.1%), either isolatedly or associated to other procedures. Seven patients underwent other types of surgery (5.1%).

With regard to the valve replacements, 49 (67.12%) comprised biological valves, 21 (28.77%) mechanical valves, and three (4.11%) corresponded to annuloplasties. There were 54 aortic valve replacements (39.7%), 17 mitral valve replacements (12.5%), four mitral annuloplasties (2.94%) and 15 tricuspid annuloplasties (11.03%).

All patients were subjected to preoperative TTE performed by a cardiologist during admission or on an ambulatory basis a few days before admission. This echocardiographic study showed that most patients had a left ventricle with no hypertrophy (72 patients; 52.94%) or dilatation (115 patients; 84.56%), and with a normal ejection fraction (Simpson’s rule 4C) (127 patients; 93.38%). The right ventricle was not dilated in most patients (131 patients; 96.32%), and exhibited normal function in all cases. Most of the patients had a non-dilated inferior vena cava (128 patients; 94.11%).

A total of 203 TTEs were performed, distributed as follows: 97 patients (71.3%) underwent a single echocardiographic study; 19 (14%) underwent two echocardiographic studies; 14 (10.3%) underwent three echocardiographic studies; four (2.9%) underwent four echocardiographic studies; and two patients (1.5%) underwent 5 echocardiographic studies. Table 2 shows the clinical data at the time of echocardiography. The main indication of the first echocardiographic study was hemodynamic instability (74 cases; 54.4%), and the main initially suspected diagnosis was hypovolemia (54 cases; 26.6%). The other 62 echocardiographic studies (45.6%) were performed with some other suspected diagnosis, or as controls. The echocardiographic studies performed on the following days were due to hemodynamic instability. With regard to the contemplated initial management based on the clinical data, in most cases (100 cases; 49.3%) no surgery was foreseeable. Table 3 reports the suspected diagnoses, the contemplated management before echocardiography and the echocardiographic diagnosis.

Most of the echocardiographic studies were made with the patient in supine decubitus (197 patients; 97%), and the mean duration of the exploration was 14.46 ± 2.16 min. Evaluation was made of imaging difficulty and imaging quality in each of the planes. In most cases, difficulty was low in all planes, and imaging quality was found to be optimal or suboptimal in most patients. The poorest imaging quality corresponded to the subcostal plane for visualizing the inferior vena cava. On analyzing imaging quality in each plane, we found the long axis parasternal plane to yield 85 echocardiographic studies (41.9%) of optimal quality, 81 (39.9%) of suboptimal quality and 37 (18.2%) of poor quality. The short axis parasternal plane in turn yielded 80 echocardiographic studies (39.4%) of optimal quality, 82 (40.4%) of suboptimal quality, and 41 (20.2%) of poor quality. The four-chambers apical plane yielded 135 echocardiographic studies (66.5%) of optimal quality, 42 (20.7%) of suboptimal quality, and 26 (12.8%) of poor quality. The subcostal plane yielded 122 echocardiographic studies (60.1%) of optimal quality, 29 (14.3%) of suboptimal quality, and 52 (25.6%) of poor quality. Lastly, with regard to visualization of the inferior vena cava, there were 116 echocardiographic studies (57.1%) of optimal quality, 19 (9.4%) of suboptimal quality, and 68 (33.5%) of poor quality. All the echocardiographic studies had some plane of optimal or suboptimal imaging quality, thus allowing echocardiographic assessment in all cases.

On analyzing the different echocardiographic parameters, we found the left ventricle to be of normal size (175 cases; 86.2%) and with normal ventricular function in most cases (151 cases; 74.4%), but with heterogeneous contractility in most patients (111 cases; 54.7%). Only 12 patients (5.9%) presented kissing walls. The right ventricle in turn was of normal size and function in most of the cases (184 cases; 90.6%). Sixty patients (29.6%) showed the presence of pericardial fluid, but only two (1.5%) presented signs of tamponade, and thus underwent surgery again.

Table 4 compares the suspected diagnoses based on the clinical findings versus the echocardiographic diagnosis. Overall, the echocardiographic diagnosis was consistent with the suspected clinical diagnosis in 102 cases (50.2%). In one-half of the cases (101 cases, 49.8%), echocardiography differed from the clinical diagnosis. In these 101 cases, echocardiography proved normal in 45 (44.6%); no alternative diagnosis therefore could be established. In 56 cases (55.4%), echocardiography offered a diagnosis different from the suspected clinical diagnosis, and treatment changes were decided in 30 of these cases (53.6%). The changes predominantly involved the start of inotropic treatment (14 cases, 25%). In relation to whether the changes were made on occasion of the first echocardiographic study or as a result of the subsequent echocardiographic explorations, we found that in a large percentage of cases in which a change in diagnosis occurred (34 cases, 60.71%), the change took place on occasion of the first echocardiographic study and implied a modification of the patient management approach in 22 cases (64.71%). In the other 22 cases with discrepancies between the echocardiographic diagnosis and the clinical diagnosis, the treatment changes occurred on occasion of the subsequent echocardiographic studies. All the changes made can be seen in Table 5.

In one case presenting diagnostic uncertainty, we contacted the Cardiology Department, which performed a new echocardiographic study that confirmed our own echocardiographic suspicion (severe pericardial effusion with signs of tamponade). We therefore consulted Heart Surgery, which decided repeat surgery, followed by clinical improvement.

Of the 30 cases in which treatment was modified on the basis of the echocardiographic diagnosis, 26 (86.67%) showed significant clinical improvement, with amelioration of the hemodynamic parameters (blood pressure, central venous oxygen saturation, increased diuresis, with a lesser need for inotropic and vasoactive drugs) over the following 30-60 min. On examining the treatment changes, we found that 14 patients (46.7%) started inotropic or vasoactive drug treatment, while another 14 patients (46.7%) underwent some other intervention such as the suspension of volume replacement measures or of inotropic medication. One patient (3.3%) received volume replacement measures, and another (3.3%) required repeat surgery because of tamponade. Four patients (13.33%) experienced no clear improvement as a result of the treatment adjustments. Two of them had started inotropic or vasoactive drug treatment (one with poor tolerance of dobutamine due to tachycardia, and another with scant hemodynamic improvement after dobutamine dosing). In another case, due to the presence of bradycardia, we decided to increase the heart rate by means of a pacemaker, though the tissue hypoperfusion parameters did not improve as a result. Lastly, in the fourth case, the clinical suspicion was hypovolemia with normal echocardiographic findings; we therefore decided not to administer volume replacement measures - though no significant clinical changes were observed.