Author + information
- Received August 28, 2000
- Revision received May 4, 2001
- Accepted May 21, 2001
- Published online September 1, 2001.
- John E Madias, MD, FACCa,b,* (, )
- Raveen Bazaz, MDa,b,
- Himanshu Agarwal, MDa,b,
- Moethu Win, MDa,b and
- Lalitha Medepalli, MDa,b
- ↵*Reprint requests and correspondence: Dr. John E. Madias, Division of Cardiology, Elmhurst Hospital Center, 79-01 Broadway, Elmhurst, New York 11373
The relationship between the changes of weight (WT) and electrocardiogram (ECG) QRS amplitude in patients with anasarca (AN) was evaluated.
Attenuation of the ECG voltage occurs as the electrical current spreads from the epicardium to the body surface. The voltage registered is a function of the cardiac potentials, the electrical resistivities of the intervening tissues and the orientation of the ECG leads with respect to the direction of propagation of excitation. Lung congestion and pericardial and pleural effusions can cause attenuation in the ECG potentials; additionally, a similar change was recently observed in patients with AN.
A prospective study of this phenomenon in 28 patients with a critical illness was carried out. Electrocardiograms and patients’ WTs were recorded daily. Pericardial effusions were excluded by serial echocardiograms. The sums of the amplitude of QRS complexes from the 12 ECG leads (ΣQRS) were correlated with the corresponding WTs. Intracardiac ECGs, done in three patients, were correlated with surface ECGs.
Admission WT was 148.9 ± 37.8 lbs, and it peaked to 197.8 ± 52.3 lbs (p = 0.0005). Admission ΣQRS was 120.2 ± 41.6 mm and dropped to 54.8 ± 26.9 mm at time of peak WT (p = 0.0005). Regression of ΣQRS on WT revealed an r = 0.61 and a p = 0.0005. Subsequent WT loss in 13 patients (from 219.0 ± 40.7 lbs to 179.5 ± 41.7 lbs, p = 0.001) led to an increase of ΣQRS from 53.5 ± 24.5 mm to 86.8 ± 38.2 mm (p = 0.001). Intracardiac ECGs remained stable, while surface ECGs changed with perturbations of WT.
Attenuation of ECG voltage in patients with AN correlates with WT gain, and it can be attributed to a shunting of the cardiac potentials due to the low resistance of the AN fluid.
A low-voltage electrocardiogram (ECG) (LVE) (QRS complexes of <5 mm in the limb and <10 mm in the precordial leads) has been linked to obesity, pericardial and pleural effusions, constrictive pericardial disease, infiltrative heart disease, diffuse myocardial necrosis or fibrosis, emphysema, pulmonary infiltration or congestion and hypothyroidism; occasionally it represents a normal variant (1–4). Abrupt emergence of LVE occurs with pneumothorax (5–9), with resolution after its correction. Additionally, aspiration or gradual natural resorption of pericardial effusion restores ECG potentials.
Spurred by a patient with anasarca (AN) and a LVE, we prospectively investigated this phenomenon in a case-control study since ECG texts (1,2,10,11)and the literature did not refer to any etiologic link between LVE and AN.
Of the 493 patients admitted to our coronary care unit (CCU) in 1999, 28 (6.7%) patients with AN were studied.
Data on demographics, past and present illnesses, hemodynamics, clinical laboratory tests, chest radiographs, echocardiograms, respiratory parameters, duration of observation, management, complications and outcome were considered.
The patients’ weight (WT) on admission, at half WT gain (HF-WG), at peak value and, for the patients who subsequently lost WT, at the lowest point and the corresponding percentage of WT change were employed as variables.
ECG measurements and variables
Standard ECGs were recorded daily. Calibration was 1 mV = 10 mm. Measurements of the amplitude (highest positive + lowest negative deflections) of the QRS complexes were made by one of the authors (J.E.M.) to the nearest 0.5 mm using calipers and a magnifying glass. For ECGs with atrial fibrillation, the average of measurements of three consecutive beats was used.
Sums of the QRS amplitudes of all ECG leads (ΣQRS) from the day of admission, HF-WG, peak and lowest WTs were calculated for each patient. Since V5and V6revealed bigger changes than V1and V2with AN, the sums of the amplitudes from these two sets of leads (ΣV5V6and ΣV1V2) were used as variables. Changes in the ΣQRS, ΣV5V6and ΣV1V2were expressed as the percentage of the admission and peak WT values. The intraobserver variability of ΣQRS in 10 random ECGs was 0.41 ± 3.34%.
Controls were: 1) 28 patients (16 men) admitted to the CCU concurrently with the study patients for a variety of cardiovascular and other illnesses. Data included whether they had undergone endotracheal intubation, WT change, the duration of observation and the ECG variables similar to the study patients. 2) Of these 28 patients, 10 who were intubated, 17 who had pneumonia and 4 who had acute pulmonary edema were also separately analyzed to evaluate the above conditions as determinants of change in the ΣQRS by comparing the ECGs before and after extubation or recovery from their illness. 3) All 10 CCU patients who underwent hemodialysis (13 sessions) in 1999 had ECGs and WT recorded before and after this procedure.
Intracardiac ECG (IC-ECGs) recordings
Intracardiac ECGs were obtained in three patients, (nine sessions), recording via the Wilson central electrode, a needle and a saline-filled central venous catheter connection; IC-ECGs and surface ECGs were correlated.
The effects of changes in temperature and hematocrit, which are thought to minimally impact the amplitude of ECG complexes (12), were explored.
The influence of inspiration, expiration, endotracheal intubation and extubation and the application/change of positive end-expiratory, end-inspiratory and continuous positive airway pressures were studied in all patients in measurements on ECGs or by viewing the bedside ECG monitors before and after such actions.
Daily chest radiographs and frequent echocardiograms were obtained in all study patients to evaluate for evidence of pulmonary infiltrates, acute respiratory distress syndrome, pulmonary congestion, pleural effusions, pneumothorax, pneumomediastinum and pericardial effusions and to assess the cardiac structure and function.
Continuous data are reported as mean ± SD. Weights and ΣQRSs from admission, times of peak WT, of half weight gain (HF-WT) and subsequent lowest WT and ΣQRSs from control patients were analyzed by two-tailed paired ttests. The relationship between percentage of change from admission of peak WT gain or HF-WG and corresponding reduction of the ΣQRS, ΣV5V6and ΣV1V2were evaluated by regression analysis, considering the WT as the independent and the ECG as the dependent variables (13). Analyses of the HF-WG evaluated for a dose-response relationship between WTs and ΣQRS. The relationship between maximal loss of WT (expressed as the percentage of change from peak WT) and corresponding increase of the ΣQRS, ΣV5V6and ΣV1V2(expressed as a percentage change from the time of peak WT) was evaluated by regression analysis. The SPSS/PC+ 4.0.1 statistical package (14)was employed, and a p <0.05 was taken as statistically significant.
Demographic and clinical data are shown in Table 1. All patients underwent endotracheal intubation and received intravenous fluids and vasopressors or inotropic agents for hypotension or sepsis. Typical ECG changes are shown in Figures 1 to 3. ⇓⇓⇓Electrocardiograms changed gradually (Fig. 3), as could be shown by daily comparisons. Loss of the gained WT led to regeneration of the ECG potentials (Figs. 1 and 4). ⇓
Weights at various time points are depicted in Table 2. After peaking to a WT of 219.0 ± 40.7 lbs, 13 patients lost WT (179.5 ± 41.7 lbs; p = 0.001), reaching a WT not different from admission WT (161.1 ± 23.2 lbs, p = 0.143). Two patients gained/lost WT, repeatedly. ΣQRSs at various time points are shown in Table 2. Admission ΣV5V6was 23.0 ± 13.2 mm and dropped to 11.2 ± 6.07 mm at HF-WG (p = 0.0005) and 8.2 ± 6.0 mm at peak WT gain (p = 0.0005) (Figs. 1 to 3 and 5). ⇓Admission ΣV1V2was 24.0 ± 13.2 mm and dropped to 18.0 ± 9.9 mm at HF-WG (p = 0.0005) and 14.7 ± 8.1 mm at peak WT gain (p = 0.0005) (Figs. 1 to 3 and 5). ΣQRSs of the patients with subsequent WT loss are shown in Table 2. ΣQRS at the time of their lowest WT (86.8 ± 38.2 mm) was lower than the admission value (118.0 ± 38.1 mm, p = 0.041). Two patients who gained/lost WT repeatedly, showed reproducible ΣQRS and WT perturbations.
Regression of changes of ΣQRS, ΣV5V6and ΣV1V2on WTs revealed an r = 0.61, 0.65 and 0.22 and p = 0.0005, 0.0002 and 0.26, correspondingly; similarly, changes in ΣQRS, ΣV5V6and ΣV1V2and HF-WG revealed an r = 0.41, 0.27 and 0.20 and a p = 0.03, 0.16 and 0.30, correspondingly. Regression analysis of 21 pairs of WT and ΣQRS of one patient revealed an r = −0.68 and a p = 0.0007 (Fig. 3).
Regression of changes in ΣQRS, ΣV5V6and ΣV1V2on loss of WT showed an r = 0.56, 0.59 and 0.65 and a p = 0.0579, 0.04 and 0.02, correspondingly. The relationship between ΣQRS and WT in nine patients who gained and lost WT is depicted in Figure 4.
Controls aged 65.2 ± 19.0 years (20 to 99 years) had two ECGs recorded 9.1 ± 6.5 days (2 to 30 days) apart, had a change of −2.5 ± 5.3% (−17% to 13.6%) in their WT in the intervening time, and their ΣQRS for the two ECGs were 136.8 ± 46.6 mm and 134.0 ± 44.1 mm (p = 0.44). ΣQRSs of the controls before and after intubation, treatment of pneumonia and pulmonary edema were 150.5 ± 59 mm and 135.4 ± 53 mm (p = 0.1), 136.2 ± 62.6 mm and 141.1 ± 45.1 mm (p = 0.58) and 120.3 ± 26.4 mm and 117.9 ± 32.0 mm (p = 0.6), correspondingly. ΣQRSs before and after hemodialysis were 143.7 ± 20.2 mm and 144.9 ± 18.2 mm (4.9 ± 4.7%; p = 0.83), with a WT loss of 4.9 ± 4.7 lbs (0.9 to 12.1) (−3.2 ± 0.6%).
Temperature and hematocrit perturbations, alterations of respiratory parameters or the status of being intubated or extubated did not influence ΣQRSs. ΣQRS before and after thoracostomy in one patient with left-sided pneumothorax was 30.5 mm and 55 mm, respectively.
Anasarca of the torso and extremities was documented by daily inspection and palpation, and was found to be greater in the dorsal than in the ventral regions of the body, presumably due to the gravity effect in patients mostly maintained in a supine position.
Echocardiograms did not reveal pericardial effusions (Fig. 5). Only Patient 26 had a pericardial effusion early in his illness, but antituberculosis therapy led to its clearing (Fig. 6). This was one of three patients whose IC-ECGs remained stable, while the corresponding surface ECGs showed marked changes as WT changed (Fig. 6).
What is novel about our study is the linkage of LVE to AN. This is supported by: 1) the relationship between peak ΣQRS drop and peak WT rise; 2) the lower attenuation effect of the HF-WT than the peak WT on the ΣQRS; 3) the gradual decrease in ΣQRS while WT rose; 4) the reciprocal relationship of the absolute values of these two variables (Fig. 3); 5) the rise of ΣQRS in patients who lost WT; 6) the reproducible relationship between ΣQRS and WT in two patients who gained/lost WT more than once; 7) the marked effect that AN had on ΣV5V6, reflecting the region with major fluid accumulation; and 8) the attenuated influence of AN on ΣV1V2, corresponding to the area with less fluid accumulation.
Electrophysiology of current conduction
Body ECG potentials are a function of the currents generated by the heart, the transfer factors of the intervening tissues, organs, or air- and fluid-filled spaces (15–17)and the location and properties of the surface electrodes with the resulting lead axes. The conducting medium attenuates the heart’s potentials so that about 1/100 of their original value is recorded at the body’s surface (16). As per Ohm’s law, the potential difference between the two ends of a conductor is a function of current and resistance (11). When the resistance of the conducting medium enveloping the heart increases, surface ECG potentials are augmented, whereas decreasing resistance leads to an attenuation (shunting) of the heart’s potentials (1,2,15–17). Resistance in turn depends on the material of a conductor (its resistivity). The operational resistance is a composite of the different resistances, determined by the geometries and resistivities of the constituents (15–19). Bayley et al. (20)and Rudy et al. (21)have explored in theoretical models the effects of such inhomogeneity on the surface ECG. Voltages are also affected by the differences in resistivities at the boundary of adjacent tissues and the orientation of the activation front impacting these interfaces.
Geddes et al. (12)have summarized the work on the resistivity of various tissues, organs and body fluids in humans and animals. Accordingly, plasma has the lowest resistivity, whereas blood, lung, fat and bone have high resistivities. Additionally, increase in temperature, as in fever, decreases the resistivity of biologic tissues and fluids.
Previous work explored the influence of the transthoracic resistance on the amplitude of ECG potentials (22–24). Van Der Water et al. (25)used changes in the transthoracic resistance to monitor patients with a variety of heart and lung illnesses, but they did not employ ECGs. Also, an increase in ECG potentials after removal of 700 ml to 3,000 ml of fluid with hemodialysis was previously reported, but the loss of WT was small compared with the one noted in our study (26,27); the increase in the ECG potentials was small (26), and it was attributed to alleviation in lung congestion, correction of undocumented electrolyte abnormalities (27)and speculations about changes in the intracavitary heart volume (26). Our experience with comparable hemodialysis-induced WT loss did not corroborate these findings (26,27). Prior literature (21–26)has attributed the changes in resistance to the status of the heart and lungs or presence of air or effusions in body cavities, but nowhere was AN implicated. We feel that hemodialysis does not produce changes detectable by the ECG, due to the small WT change effected. This was corroborated by the absence of significant changes in the ECGs of our patients during the early phase of WT gain or loss. In contrast, hemofiltration with marked WT loss led to an impressive gain in the QRS amplitude (Fig. 6). Transthoracic or other body resistance measurements (not carried out in our study) would have provided another correlate of the ECG changes. Resistance changes have been previously traced to an increase in intrapulmonary water (25); however, our controls with pulmonary edema did not reveal an increase in their QRS amplitude after relief of congestion, indicating that the ECG changes in our study patients could not have occurred as a result of changes in the lung water content.
Potential confounding factors
Other factors that could have caused LVE in our patients can be easily excluded because their effect is known to be minor (i.e., anemia, fever) (12); they were present intermittently (lung congestion and infiltrates), while the QRS potential loss was gradual and constant, paralleling the WT gain. The influence of changing intracavitary blood mass on the ECG potentials (Brody effect) could not be studied; this would have required serial precise measurements of left ventricular dimensions. Such a mechanism in our clinically deteriorating patients would be expected to result in augmentation of the QRS amplitudes than the decrease observed (11,26,28). However, the Brody effect could also lead to a decrease in ECG potentials depending on activation orientation, and its influence may have been overrated (21). Although stability of IC-ECGs implicated the conducting medium for the LVE than the heart, an effect of AN on the IC-ECGs (not seen in our study) is theoretically possible.
Our findings are in accord with the experimental work of Green et al. (29)(Taccardi B, personal communication, 2000) who showed that low-resistivity Tyrode’s solution surrounding the heart results in a LVE, which was only noted over the portion of the heart immersed in the solution (29). A corollary of this in our study was that AN affected ΣV5V6(overlying larger fluid accumulation) more than ΣV1V2.
This study provides a diagnostic explanation for the frequently encountered LVE in patients with a large variety of critical illnesses.
We wish to thank Professor Bruno Taccardi of Utah University for reviewing the paper and for directing us to relevant work from his laboratory.
- coronary care unit
- half weight gain
- weight, at the point of half weight gain
- intracardiac ECG
- low-voltage ECG
- sum of the amplitudes of QRS complexes
- sum of the amplitudes of QRS in V1and V2
- sum of the amplitudes of QRS in V5and V6
- Received August 28, 2000.
- Revision received May 4, 2001.
- Accepted May 21, 2001.
- American College of Cardiology
- ↵Wagner GS. Marriott’s Practical Electrocardiography. 9th ed. Baltimore, MD: Williams & Wilkins, 1994:176–7, 179, 194.
- Goldberger AL, Goldberger E. Clinical Electrocardiography: A Simplified Approach. 5th ed. St. Louis, MO: Mosby, 1994:131, 133, 233.
- Armen R.N.,
- Frank T.V.
- Macfarlane P.W.,
- Vetch Lawrie T.D.
- ↵Mirvis DM. Electrocardiography: A Physiologic Approach. St. Louis, MO: Mosby, 1993:19, 76, 259.
- ↵Colton T. Statistics in Medicine. 1st ed. Boston, MA: Little, Brown and Company, 1974:17, 33, 120, 131, 189.
- ↵Norusis MJ. The SPSS Guide to Data Analysis for PDD/PC+. 2nd ed. Chicago, IL: SPSS, Inc., 1991;244,362,374,390.
- Rutkay-Nedecky I.
- Lepeschkin E.
- Rush S.,
- Nelson C.V.
- Rush S.,
- Abildskov J.A.,
- McFee R.
- Schwan H.P.,
- Kay C.F.
- Rudy Y.,
- Plonsey R.,
- Liebman J.
- Gamboa R.,
- Adair B.N.
- Kira S.,
- Hukushima Y.,
- Kitamura S.,
- Ito A.
- Brody D.A.
- Green L.S.,
- Taccardi B.,
- Ershler P.R.,
- Lux R.L.