Epidural Anesthesia, Hypotension, and Changes in Intravascular Volume

We studied 12 healthy volunteers, 3 women and 9 men, with a median age of 27.5 yr (range, 22–29 yr), a median height of 181.5 cm (range, 168–88 cm) and a median weight of 75.5 kg (range, 57.2–85.8 kg), in a prospective, randomized study. No subjects were medicated, and none showed abnormal findings in their medical history or objective examination. The women were included after a negative pregnancy test result on the morning of the study. The regional ethics committee (Hvidovre, Denmark) approved the study, and subjects gave written, informed consent before inclusion. The subjects fasted from midnight the day before the study. An epidural catheter was inserted in the T7–T10 interspace via  standard loss-of-resistance technique by an experienced anesthesiologist (C. L.). Appropriate catheter placement was tested with 3 ml lidocaine, 2%, with 1:200,000 epinephrine (t = 0). Then 5 + 5 ml bupivacaine, 0.5%, was injected (t = 5). The sensory upper level of the blockade (determined by pin prick) was aimed at T4. If the sensory blockade did not reach T4 after 20 min, an additional 5 ml bupivacaine, 0.5%, was administered. After 90 min, subjects were randomized by the closed-envelope method to receive either 0.2 mg/kg intravenous ephedrine or 7 ml/kg hydroxyethyl starch (HES; Voluven®, 6% HES, 130/0.4; Fresenius Kabi, Bad Homburg, Germany). No fluids or vasopressors were given up to that point. HES was infused over 5 min through a peripheral vein. Plasma volume, erythrocyte volume, hematocrit, and mean corpuscular volume (MCV) were measured before the epidural blockade, at t = 90, and at t = 130 (see following sections). The study was terminated 60 min after completion of the HES infusion (t = 155). Subjects were monitored with electrocardiography and blood pressure monitoring (arm cuff) every 5 min throughout the study by an anesthesiologist. The baseline values of systolic and diastolic blood pressure and heart rate were calculated as the mean of two preinfusion measurements. Sensory blockade was monitored (by pin prick) every 20 min. Subjects with a maximal decrease in systolic blood pressure of greater than 20% from the baseline value within 80 min after induction of epidural anesthesia were considered hypotensive, and subjects with a decrease in systolic blood pressure of 20% or less from the baseline value were considered normotensive.

Plasma and erythrocyte volumes were determined by standard tracer dilution technique 4after triple injection of 200 kBq ¹²⁵I-labeled human albumin and single injection of 2 MBq ⁵¹Cr sodium chromate in vivo -labeled erythrocytes, respectively. In detail, plasma volume was determined by injection of 200 kBq ¹²⁵I-labeled human albumin at each time point (subtracting any residual activity from the previous sample), obtaining blood samples after 10 min and determining plasma volume from these. 5Because the method of obtaining multiple blood samples and calculating plasma volume by regression to zero values requires unchanged capillary permeability, hematocrit, and plasma protein concentration (all of which may change after epidural anesthesia), we chose the single-sample technique, which furthermore has the advantage of minimizing blood loss due to sampling.

Body hematocrit was calculated as the ratio between erythrocyte and plasma volume measurements at t =−10, t = 90, and t = 130 min. At the same time points, peripheral hematocrit (sampling from a peripheral vein) and MCV (separate sampling from a peripheral vein [9 ml blood drawn per subject] after reinjection of the initially drawn 2 ml blood) were determined.

Samples for measuring blood hemoglobin were drawn from the venous cannula in the arm not used for infusion every 5 min throughout the study. The first sample (baseline) was drawn in triplicate, and the mean value was used in the calculation of the baseline value. Before each sample, 2 ml blood was drawn to clear the sampling line. This amount was reinjected after the sampling. A total of 14.0 ml blood was drawn for the analysis throughout the study (mean, 0.43 ml [14 ml/32] per sample) and replaced after each sampling by a total of 14.0 ml isotonic saline. The hemoglobin measurements were analyzed using an ABL510 Blood Gas Analyzer (Radiometer, Copenhagen, Denmark; coefficient of variation 1–3%, manufacturer’s data).

The distribution of the fluid given by infusion of HES was analyzed using a one-volume kinetic model. 6In this model, the fluid is given at a rate k  ⁱand is distributed in an expandable space having a volume (v ), which the body strives to maintain at a target volume (V ). Elimination occurs at a rate proportional by a constant (k  ^r) to the deviation from the target volume, V . In the current study, the base elimination constant (k  ^b), usually used to describe non-dilution-dependent elimination, was used by the analysis program to quantify the flow of fluid from more remotely located spaces back into the fluid space expanded by the infusion. The following differential equation describes the dilution changes in v :

Because plasma volume is a part of v , dilution of arterial plasma was used to indicate (v −V )/ V . Therefore:

Kinetic analysis was performed on all individual infusion experiments. The best estimates of the model parameters V , k  ^r, k  ^b, and their associated SEs were obtained by fitting the mathematical solutions to equation 2, which have been presented previously, 6to the experimental data by using a nonlinear least-squares regression routine programmed in MATLAB version 4.2 (MathWorks Inc., Natick, MA). The result was presented as a one-volume model with parameters V , k  ^r, and k  ^b.

Data were analyzed using nonparametric statistical methods. The Wilcoxon signed-rank test for paired observations was used to describe differences before versus  after epidural anesthesia. For comparing data between the groups (fluid vs.  ephedrine), the Mann-Whitney test was used. For correlations between two parameters, the Spearman ρ was applied. Categorical data were analyzed using the Fisher exact test. P < 0.05 was considered significant. Continuous data (systolic and diastolic blood pressure, heart rate, and hemoglobin values) were analyzed with the Friedman analysis of variance. Volume kinetic parameter estimates are given as medians and 25th–75th percentiles. These parameters are the results from nonlinear regression analyses and contain SEs.

The coefficient of variation of plasma volume measurements with the applied technique is known to be approximately 2%. We considered an increase in plasma volume of 150 ml after epidural anesthesia to be clinically relevant (initial plasma volume assumed to be 3,500 ml, estimated SD of difference 101). With a power to detect a minimal relevant difference of 80% and a level of significance of 0.05, eight subjects were needed. To account for variable subject responses, we included 12 subjects in the study.

All subjects completed the study and followed the study protocol. Two subjects received supplemental bupivacaine (25 mg) because of insufficient sensory blockade. A median of 500 ml HES (range, 500–625 ml) and a median of 16 mg ephedrine (range, 10–17 mg) were administered in the respective groups at t = 90.

Both systolic and diastolic blood pressure and heart rate decreased significantly after administration of epidural anesthesia from t = 0 to t = 90 (fig. 1). Systolic blood pressure increased significantly after administration of both HES and ephedrine (from t = 90 to t = 130) without a difference between the groups, whereas neither diastolic blood pressure nor heart rate changed significantly after administration of either HES or ephedrine (fig. 1).

Plasma volume did not change 90 min after epidural anesthesia compared with baseline (all subjects) or when subdivided into normotensive or hypotensive subjects (table 1and fig. 2). Changes in plasma volume did not correlate to the degree of hypotension (Spearman ρ= 0.26; P = 0.41). Forty minutes after administration of HES, plasma volume was significantly increased by a median of 324 ml, whereas no significant changes in plasma volume were observed after ephedrine administration (table 1).

Erythrocyte volume did not change significantly during the study (table 1and fig. 2).

Peripheral hematocrit decreased significantly (from 0.41 to 0.40) at t = 90 compared with baseline and decreased significantly after HES administration (from 0.40 to 0.38) (table 1). Body hematocrit did not change during the study (table 1).

Mean corpuscular volume data were only available from six subjects and did not change during the study (table 1).

Hemoglobin concentrations did not change significantly from t = 0 to t = 90 in either normotensive or hypotensive subjects. A significant decrease was seen after administration of HES (t = 90 to t = 130), whereas ephedrine administration did not change hemoglobin concentrations (data not shown).

The administration of epidural anesthesia created a small dilution, approximately 10%, followed by a dilution after HES administration (fig. 3). On average, HES expanded a central volume of 1,482 ml (range, 825–1,759 ml) and was eliminated from the system by a constant k  ^r, 56 ml/min (range, 36–72 ml/min) (table 2). k  ^b, in this model reflecting the rate of fluid recruited  into V , was estimated to be −3.7 ml/min (range, −5.3 to −1.6 ml/min).

This study was mainly undertaken to investigate the changes in blood volume after epidural anesthesia per se  and to capture the behavior of fluid given intravenously during epidural anesthesia in a situation in which volume loading is controversial. We found that plasma volume did not change per se  after thoracic epidural anesthesia despite a decrease in blood pressure. Plasma volume was increased with fluid administration but was unchanged with vasopressors, whereas both treatments had similar hemodynamic effects. Hemoglobin concentrations were not significantly altered by either epidural blockade or ephedrine administration but were significantly decreased after HES administration.

The observed decrease in systolic and diastolic blood pressure and heart rate after epidural anesthesia corresponds with previous observations in healthy volunteers, 7and the time frame of these circulatory effects are also well known. 7To allow for a possible redistribution of fluid from the interstitial phase to the intravascular space, we chose 90 min as the appropriate time to evaluate the intravascular volume parameters. Previous studies during experimental hypovolemia have found a capillary refill to occur within 5 min, 2,8and short-term studies (20–30 min) with and without concomitant fluid administration after lumbar epidural anesthesia with hypotension have also suggested a capillary refill to occur based on hemoglobin measurements. 1,9However, these suggestions have not been documented by actual blood/plasma volume measurements. In several series of patients undergoing lumbar epidural anesthesia, a larger percentage of an infused amount of fluid was retained intravascularly in hypotensive compared with normotensive patients (only indirectly measured by decrease in hemoglobin concentration, however). 1,10The reasons behind a possible intravascular fluid retention in hypotensive compared with normotensive subjects during experimental conditions have not been fully clarified, but mostly, they have been attributed to changing Starling effects. 10Transcapillary fluid flux is governed by differences in the colloid and hydrostatic pressures between the plasma and the interstitial space. Initially, when a fluid load is given, most of the infused fluid shifts from the circulation to the interstitial compartment, thereby increasing its hydrostatic pressure. When the arterial pressure decreases after 15–20 min, there is a fluid flux causing intravascular hemodilution. However, significant hemodilution 20 min after the onset of lumbar epidural anesthesia has only been observed when epidural anesthesia was accompanied by fluid administration. 1These findings were confirmed by this study because blood and plasma volumes were similar both at the beginning of the experiment and before interventions at t = 90 (table 1). However, when fluid was administered, there was a profound dilution and increased blood volume (table 1), which shows that epidural anesthesia per se  had no effect on blood volume. 1To more precisely describe the changes in intravascular volume, we measured hematocrit, erythrocyte volume, and MCV. Erythrocyte volume did not change significantly during the study but tended to increase with fluid administration (table 1). The reasons for this are unclear because no changes in individual erythrocyte cell volumes occurred (measured by MCVs). However, to be accurate, the indicator dilution techniques required uniform distribution of the tracer. Previous findings have suggested that this is in fact not true during epidural anesthesia because ⁹⁹Tc-labeled erythrocytes sequester within the denervated area after epidural anesthesia (assessed by gamma camera scan and plethysmography). 11However, these effects were studied only during the initial hemodynamic changes (until 25 min after epidural anesthesia). During the later measurements in the current study (t = 90 min) when all subjects were hemodynamically stable and systolic pressure was back at baseline value, we considered distribution of the tracer to be uniform. Blood and plasma volumes are traditionally estimated by indicator dilution and mass balance techniques, 12,13and in recent years, they have been supplemented by volume kinetic analyses, 14,15a pharmacokinetic tool used to study more closely the time course of fluid shifts in various settings. 6,16This method has many similarities to pharmacokinetics but is based on the dilution of the venous plasma caused by the infused fluid instead of measuring the concentration of a drug in the blood. To look more closely at the fluid behavior during infusion, indicator dilution techniques in this study were supplemented by volume kinetic calculations. Hemoglobin analysis showed that when epidural anesthesia was applied, there was a very small dilution (fig. 3). When HES was administered, volume kinetic analysis showed a dilution of the plasma volume (fig. 3) in accordance with the findings when the indicator dilution technique was used (table 1). The volume kinetic data delineate the time course of HES distribution, which cannot be assessed by tracer dilution technique. The administration of HES expanded a central volume, V , that seemed to be quite small, approximately 1.5 l (table 2), which is approximately half of the measured plasma volume (table 1). Therefore, the effect of HES may be seen as a rapid expansion of central volume with an increase in preload to prevent hypotension. The elimination rate constant k  ^rfurther indicated a rapid elimination (table 2), being on the average 56 ml/min, which is a value closer to what has been obtained for crystalloid fluid boluses. 6,15As a comparison, k  ^rfor dextran 70 in healthy normotensive volunteers was 10 ml/min. 6It could be hypothesized that the rapid reduction of the dilution after the end of infusion is due to redistribution between the small central fluid space (1.5 l) and the remainder of the plasma volume. This would indicate a step-by-step fluid distribution during infusion from centrally located spaces to more peripherally perfused parts during the administration of regional anesthesia. The preferential distribution of infused fluid to the central plasma volume, together with a slower transport of fluid to a more remote body fluid space, is a meaningful adaptation because infused fluid then restores cardiac preload more effectively. Another interesting finding is that the kinetic evaluation allowed quantification of the recruitment of fluid from the periphery to the central fluid space (in which hemoglobin is readily equilibrated) at a rate of approximately 4 ml/min (table 2). This quantifies the transcapillary flux that occurs during fluid loading under epidural anesthesia. On the contrary, ephedrine did not have these volume shifts.

In conclusion, we have demonstrated that epidural anesthesia per se  does not lead to any changes in intravascular volume and that both ephedrine and fluids have comparable hemodynamic effects when applied 90 min after establishing an epidural blockade. Furthermore, an infused fluid causes a profound dilution and seems to be more centrally located under epidural anesthesia. Because both ephedrine and fluids have similar hemodynamic effects, vasopressors may be preferred in the treatment of hypotension after epidural anesthesia, especially for patients with cardiopulmonary diseases in which perioperative fluid overload is undesirable. 3 

The authors thank Ingelise Siegumfeld and Bente Pedersen (Laboratory Technicians, Department of Clinical Physiology and Nuclear Medicine, Hvidovre University Hospital, Hvidovre, Denmark) for helpful assistance.