Bioimpedance Monitoring: Better Than Chest X-Ray for Predicting Abnormal Pulmonary Fluid?

Noninvasive hemodynamic monitoring may be useful in the those clinical environments where invasive monitoring is impractical;¹ this includes the emergency department (ED),^1–6 inpatient stepdown areas, subacute care facilities, and outpatient clinics. Bioimpedance cardiography is a new monitoring strategy that employs the temporal measurement of thoracic electrical resistance changes to determine a number of actual and derived hemodynamic parameters.

Electrophysiologically, impedance decreases during cardiac systole when blood flow velocity, thoracic fluid volume, and red blood cell alignment are maximal.¹ By using cyclic changes in impedance, in relation to the ECG, the monitor is able to provide indices of systemic vascular resistance, cardiac output, and cardiac contractility. Measuring the average thoracic resistance provides estimates of total thoracic fluid content. An increase in total thoracic fluid content, from intravascular, intraalveolar, or interstitial compartments, corresponds to greater conductivity in the thorax. This is reported as a decrease in the base thoracic electrical impedance (Zo). Applied in the clinical scenario, this can assist in quantifying pulmonary fluid content.

In the outpatient and ED environment, assessment of intrathoracic pulmonary fluid is frequently performed using the chest x-ray (CXR). However, the CXR is known to have many limitations, including low sensitivity and delayed correlation with clinical status changes.

Bioimpedance monitoring provides a reasonably precise, repeatable, and digital method of thoracic fluid quantification. No previous study in ED patients has been reported examining the correspondence between bioimpedance measurement to CXR in the assessment of thoracic pulmonary fluid measurement. The objective of this study is to compare the bioimpedance measurement of base Zo to a cardiac radiologist's systematic physiological interpretation of a CXR in ED patients suffering from shortness of breath.

Methods

This study was a prospective convenient sample of ED patients with shortness of breath and suspected heart failure undergoing impedance monitoring. All patients gave informed consent. The Institutional Review Board of the Cleveland Clinic approved this study. Our facility is a 1000 bed, tertiary care, urban teaching hospital. The ED has an annual census of 40,000 and shortness of breath is a common presenting symptom.

Shortly after a patient arrived in the ED, a study coordinator collected and recorded one bioimpedance measurement prior to the initiation of any therapy anticipated to alter thoracic fluid volumes. Specifically, diuretics and hemodynamically active medication was administered after the CXR and baseline bioimpedance data was recorded. The CXR was obtained either immediately before or after impedance monitor placement. Initial bioimpedance and CXR data was used for all comparisons of pulmonary fluid content between techniques.

The Renaissance Technology, Inc., IQ® monitor was used to measure thoracic impedance. Normal thoracic impedance Zo values are 20–30 ohms in males and 25–35 ohms in females. Patients were excluded from the study if their dz/dt value (which reflects change in impedance/time) was <0.3 ohms/second. This was done because a low dz/dt value reflects poor left ventricular contraction force and poor signal strength, and may lead to inaccurate data.

A single study coordinator applied three impedance monitor electrographic leads (lead II configuration) and four bifunctional electrodes at the upper and lower limits of the thorax. Vital signs and bioimopedance hemodynamic values were collected while the patient was resting in a supine position.

A single cardiac radiologist, blinded to the clinical data, interpreted each CXR using a previously described^7–10 approach to the physiologic assessment of the plain radiograph and the chest for hemodynamic and fluid status. Each CXR was graded as per the definitions in Table I. CXRs were defined as having abnormal pulmonary fluid if they were graded as 1, 2, or 3.

Statistical analysis was performed by a general linear model comparing CXR and Zo. Post hoc Bon Ferroni pairwise comparisons were utilized to determine significant variables. Significance was defined as p<0.05.

Results

Of the patients, 139 met the entry criteria. Mean age was 67±14 years, with a range of 27–94 years. The sample was 63% male, 51% white, 47% African American, 1% Hispanic, and 1% Arabic. Mean vital signs were: heart rate 85 bpm (±18), systolic blood pressure 133 mm Hg (±30), diastolic blood pressure 75 mm Hg (±19), respiratory rate 22 bpm, and oxygen saturation 94% (±4%). The mean Zo was 18±5 ohms. There was no significant difference between groups in regard to age, oxygen saturation, or respiratory rate.

A significant difference (p<0.0002) was found between the groups with a CXR interpreted as Grade 0 (normal) when compared to all other grades (pulmonary venous hypertension with or without visualized edema) (Table II). Patients with a Grade 0 CXR had a higher mean Zo, an average of 23.4±5.4 ohms, compared to any other group. Those with a Grade 1, 2, or 3 CXR, indicating abnormal pulmonary fluid, had Zo values that were similar to each other, but significantly lower in value compared to the normal CXR (i.e., Grade 0) group. The mean Zo was 17.5±5.5 ohms for those with Grade 2 CXRs and 17.2±4.2 ohms when there was abnormal pulmonary (grade 3) fluid. There was no statistically significant difference between those with Grade 1 and those with Grade 2 or 3 CXRs (p<0.05)(Table I).

Discussion

Radiographs of the chest are frequently used for ED and outpatient clinic estimation of hemodynamic/fluid status. However, radiographs have a number of limitations: 1) the interpretation is a subjective observation; 2) reproducibility of the readings may be difficult when there are multiple practitioners from different specialties involved in a patient's care; 3) acute changes in thoracic fluid parameters are not reflected in the CXR for several hours; 4) radiographs are insensitive for the detection of pulmonary fluid in heart failure patients;^11–14 and 5) variations in technique can affect the interpretation (e.g., portable vs. posterior-anterior and lateral chest radiograph).^13,14

In patients with chronic congestive heart failure, Chakko et al¹¹ found that physical and radiographic signs of congestion had poor predictive value for identifying patients with high pulmonary artery occlusive pressures (PAOP). In fact, radiographic pulmonary congestion was absent in 53% of patients with mild to moderately elevated PAOPs (16–29 mm Hg) and in 39% of patients with markedly elevated PAOP (≥30 mm Hg). Since survival and quality of life is improved when the PAOP is normalized, these researchers concluded that therapy aimed at normalizing filling pressures may be superior to relieving clinical and radiographic signs of congestion.¹¹ An explanation for the disparity between radiographic evidence or pulmonary edema and PAOP is that pulmonary arterial hypertension, which commonly results from chronic pulmonary venous hypertension from congestive heart failure, becomes fixed and does not respond with to changes in pulmonary venous hypertension with therapy (e.g., diuretics).

The presence of pleural effusion is frequently associated with heart failure. However, it can be missed by the CXR. In intubated patients, chest radiographs are usually performed in the supine position. This degrades the diagnostic accuracy for the detection of pleural effusion. In 34 patients with pleural effusions proven by decubitus CXRs, the sensitivity, specificity, and accuracy of the supine chest radiograph was reported to be 67%, 70%, and 67%, respectively.¹³ Bioimpedance monitoring may be of help in the supine patient.

Portable technique CXRs are commonly used in ED patients who are unable to stand erect, or who are too hemodynamically unstable to be transported to the radiography suite. The associated findings of heart failure are, in descending order of frequency: dilated upper lobe vessels, cardiomegaly, interstitial edema, enlarged pulmonary artery, pleural effusion, alveolar edema, prominent superior vena cava, and Kerly B lines.¹⁴ However, the sensitivity for the radiographic findings associated with heart failure is poor. In 22 patients with mild clinical heart failure, only the finding of dilated upper lobe vessels was found in >60% of patients. These parameters improved with increasing severity of heart failure. If heart failure was graded as severe, radiographic findings occurred in at least two-thirds of patients, except for Kerly B lines (11%) and a prominent vena cava (44%).¹⁴

Impedance monitoring offers a digital, reproducible, numerical value that quantifies thoracic fluid volume content. There are no published studies validating its utility for determining thoracic fluid measurement in an ED or outpatient setting. Our data suggest that the measurement of mean Zo may quantify thoracic fluid volumes. Furthermore, a unique population of patients with abnormal thoracic fluid volumes can be identified statistically.

Recently, abstracts examining various aspects of thoracic bioimpedance hemodynamic monitoring in the ED setting were published. Milzman et al² conducted a study in the heart failure population to determine if there was a relationship between thoracic bioimpedance and chest radiograph changes. They reported that Zo accurately predicted the presence and severity of thoracic fluid in ED patients admitted for heart failure exacerbation.² Their data suggested a linear relationship between the grading of the severity of the CXR in relation to intrathoracic fluid volume, and the changes of mean thoracic Zo. We were unable to replicate those results. In our study, an abnormal Zo was associated with an abnormal CXR; however, there was no linear relationship between a worsening radiographic appearance and Zo. We suspect that this is due to the poor sensitivity of CXR for detecting pulmonary fluid in heart failure patients. The radiographs of chronic heart failure patients may not have “congestive” signs and may actually have relatively normal CXR results despite excess thoracic volume. In chronic heart failure patients admitted to an ED with shortness of breath, Zo determination may be superior to CXR interpretation in determining fluid state and planning an appropriate treatment plan.

Our study suggests that impedance monitoring may have the potential for clinical applicability. The variance between normal and abnormal Zo values are diverse and relevant to the patient's thoracic fluid state. An additional benefit is that the numerical values change immediately in response to therapeutic interventions. This is in contradistinction to the interpretation of CXR, which frequently lags behind the patient's clinical appearance of resolving thoracic fluid volume for several hours after the institution of therapies. Since there is no delay in the change of Zo, the physician may be guided in the appropriateness of his treatment.

Although thoracic computed tomography has been proposed, the gold standard for the quantification of thoracic fluid volumes is currently unclear. Computed tomography has been suggested as being able to detect pleural fluid not noted on routine CXRs. Twenty five patients undergoing pulmonary computed tomography, pulmonary artery catheterization, and CXRs, were found to have increased pulmonary fluid undetectable on CXR and before pulmonary capillary wedge pressure increased.¹² While thoracic computed tomography may offer a diagnostic alternative, it is difficult in the symptomatic heart failure patient.

Further research on thoracic bioimpedance monitoring is warranted. It is unclear how other pathologic conditions may impact the interpretation of the values obtained via thoracic impedance measurement (i.e., pneumonia, trauma, or pneumothorax).

Our study is limited by a small sample size. Future study should focus on outcomes and examine the inherent value of an objective numerical measurement quantifying the presence of abnormal thoracic fluid.

Conclusions

Noninvasive monitoring by bioimpedance technology is able to distinguish normal from abnormal thoracic fluid content. In addition, bioimpedance measurement may detect pulmonary fluid not apparent on routine chest radiography.

Note: This study was supported by a grant from Renaissance Technologies.