Objective: To investigate the association between ECG changes and the
presence of pericardial effusion. Background: The ECG changes associat
ed with pericardial effusion described in textbooks are based only on
small series of human cases and data from animals. These changes inclu
de low QRS voltage, electrical alternans, P wave changes, and T wave i
nversion. Methods: All patients who had undergone 2 temporally separat
e echocardiographic and ECG examinations, with 1 echocardiographic exa
mination indicating the presence and the other indicating the absence
of pericardial effusion were identified (n = 46). These patients were
age- and sex-matched to 46 patients without effusion (control subjects
). Pericardial effusion was classified echocardiographically as small
(n = 28), moderate (n = 13), and large (n = 5). The ECG variables were
independently measured by two investigators blinded to effusion statu
s. Results: When 2 temporally separate ECGs for 46 patients were obtai
ned in a repeated-measures fashion (1 obtained during the absence and
the other during the presence of effusion; median time interval, 1.24
months), only the mean heart rate in patients with sinus rhythm (98 be
ats per minute increasing to 106 beats per minute) and the percentage
of patients with QRS voltage of less than 0.5 mV (10 percent increasin
g to 22 percent) were associated with the development of effusion. A w
eak correlation (r = 0.296) was noted between QRS voltage and effusion
size. Electrical alternans occurred only in one of the five patients
with a large effusion but in no others. In addition, when the ECGs ind
icating effusion from the 46 patients were compared with the ECGs from
their age- and sex-matched control subjects, differences in heart rat
e (106 beats per minute vs 80 beats per minute, respectively) and smal
l changes in QRS voltage were associated with effusion status. No ECG
variable was sensitive for the detection of pericardial effusion. Conc
lusions: In both repeated-measures and case-control comparisons, ECG f
indings are too few, subtle, insensitive, and nonspecific to be useful
as indicators of the presence of pericardial effusion.