Objectives: To determine the amount of spontaneous variability of card
iac output over time in critically ill patients, and to determine the
effect of mechanical ventilation on cardiac output variability over ti
me. Design: Case series. Setting: Medical intensive care unit in a Vet
erans Affairs Medical Center. Patients: Twenty-two patients with indwe
lling pulmonary artery notation catheters were studied. Two patients w
ere studied twice. Interventions: During a l-hr time period in which n
o interventions were required or made, thermodilution cardiac output w
as determined at baseline and then every 15 mins for 1 hr. At each tim
e point, five individual cardiac output measurements were made and a m
ean was computed. The covariables of heart rate, respiration rate, mea
n arterial pressure, mean pulmonary arterial pressure, pulmonary arter
y occlusion pressure, and temperature were also recorded at each time
point. Measurements and Main Results: The variability of the five card
iac output measurements made at each time point was expressed by calcu
lating for each patient a coefficient of variation of the measurements
. The overall mean coefficient of variation of the measurements was 5.
8%. The variability of the cardiac output measurements over time was e
xpressed by calculating for each patient a coefficient of variation ov
er time. The overall mean coefficient of variation over time was 7.7%.
A subgroup of 15 ''covariable stable'' patients (defined as those pat
ients with covariables within +/-5% of the mean covariable values duri
ng the hour) had a mean coefficient of variation over time of 6.4%, wh
ereas ''covariable unstable'' patients (with >+/-5% changes in any cov
ariable) had a mean coefficient of variation over time of 9.9% (p <.05
). Patients breathing spontaneously had a mean coefficient of variatio
n over time of 10.1%, whereas mechanically ventilated patients had a m
ean coefficient of variation over time of 6.3% (p <.05). Conclusions:
The spontaneous variability of cardiac output should be considered whe
n interpreting two cardiac output determinations made at separate time
s. Due to spontaneous variability alone, a patient with a baseline car
diac output of 10.0 L/min would be expected (95%) confidence interval)
to have a cardiac output range of 9.2 to 10.8 L/min if covariables we
re stable, and a range of at least 8.8 to 11.2 L/min if covariables we
re unstable. Patients who were mechanically ventilated displayed less
variability than patients who were breathing spontaneously.