Dextrose-free local anesthetics and opioids, alone and in combinations, are being used increasingly to provide subarachnoid anesthesia and analgesia. These dextrose-free drugs have been described as hypobaric by some and isobaric by others. To accurately classify anesthetics with regard to baricity, the density of cerebrospinal fluid (CSF) must be known. The authors sought to determine the exact density of human CSF, and determine whether CSF density is altered by pregnancy.


Density measurements accurate to 0.00001 g/ml were made at 37.00 degrees C, using a mechanical oscillation resonance frequency density meter. Cerebrospinal fluid samples were obtained from 44 patients during spinal anesthesia. Five groups were studied: men, and premenopausal, postmenopausal, term pregnant, and postpartum women.


Mean CSF densities in men (1.00064 +/- 0.00012 g/ ml), postmenopausal women (1.00070 +/- 0.00018 g/ml), and nonpregnant premenopausal women (1.00049 +/- 0.00004 g/ ml) were significantly greater than in term pregnant (1.00030 +/- 0.00004 g/ml) and postpartum (1.00034 +/- 0.00005 g/ml) women. Cerebrospinal fluid density did not correlate with age.


Mean CSF density varies in different patient subpopulations. Pregnancy and the immediate postpartum period are associated with the lowest CSF densities. In addition, the cutoff values defining hypobaricity (mean CSF density minus three standard deviations) are greater than previously reported. Accurate CSF density values should be used when considering baricity as a mechanism for clinical observations of dextrose-free intrathecal local anesthetics and opioids. Gestational status also should be considered.

THE density difference between an intrathecal anesthetic solution and cerebrospinal fluid (CSF) is a major determinant of extent of spinal block. [1]Densities of nondextrose-containing intrathecal local anesthetics and opioids have been determined recently and are thought to be very close in value to the density of human CSF. [2,3]The values for the density of human CSF derive from two earlier reports. [4,5]These values have been useful for comparison with dextrose-containing intrathecal solutions that are clearly hyperbaric. However, accuracies of mass, volume, and temperature measurements were limited by the techniques used in both studies. The studies also reported different means, each with large standard deviations (1.00100+/-0.000304and 1.00030 +/-0.000305g/ml). The techniques used in earlier CSF density studies limit comparisons with current spinal anesthetic solution densities. As a result, dextrose-free solutions continue to be described as hypobaric by some and isobaric by others. [6,7]In addition, both previous reports restricted study subjects to particular population subsets, neither including pregnant women. Although CSF composition has been compared in pregnant and nonpregnant women, [8,9]CSF density in pregnancy is unknown. We obtained CSF from individuals in several patient subpopulations and measured density with greater accuracy than was reported previously (+/-0.00001 g/ml). We then compared densities among the subgroups, which included women during and immediately after pregnancy.

Materials and Methods

After approval from the human subjects review board and informed consent, 44 healthy ASA physical status 1 and 2 patients undergoing spinal anesthesia for various procedures were studied. Specific groups studied were men (n = 10), postmenopausal women (n = 8), nonpregnant premenopausal women (n = 6), women undergoing postpartum tubal ligation (n = 10), and parturients at term gestation undergoing elective Cesarean section (n = 10). Parturients with preeclampsia were excluded. Patient age was noted, as were gestational age and time since delivery, when applicable. Lumbar CSF was collected via a 25-G Whitacre needle at the time of induction of spinal anesthesia. After confirming free flow of clear CSF and before intrathecal injection of spinal anesthetic, 2 ml clear CSF was collected for immediate analysis.

The density of CSF at 37 degrees C was determined using a density meter (DA-310, Mettler-Toledo, Hightstown, NJ) that uses the mechanical oscillation resonance frequency technique. Samples are injected into a glass measuring cell. The cell is stimulated to determine the oscillation period generated at its natural oscillation frequency. The oscillation period (T) depends on the density, volume, and temperature of the sample. A cell factor (F) is calculated from the oscillation periods of two different standard substances (desiccated air and distilled water) at a specific temperature (37.00+/-0.01 degree C). The density (d) of an unknown sample is calculated when its oscillation period (T) is determined:Equation 1where dAis the density of desiccated air and TAis the oscillation period of desiccated air. Calibration was verified before and after each CSF density measurement, using desiccated air and distilled deionized water (dA= 0.99114 g/ml and dw= 0.99333 g/ml at 37 degrees C, respectively). After the calibration procedure recommended by the manufacturer (Mettler-Toledo), density measurements are accurate to+/-0.00001 g/ml in the range of 0–3 g/ml. An electronically controlled thermistor maintained temperature of the sample and the measuring cell at 37.00 degrees C, with a temperature control accuracy of+/-0.01 degree C.

Cerebrospinal fluid densities are expressed as mean (g/ml)+/-SD. The upper limit of hypobaricity was calculated for each group as defined previously (mean CSF density minus three standard deviations of the mean). [1,4]Comparisons between groups were made using one-way analysis of variance. Pair-wise comparisons were then made, using Tukey's post hoc test method of multiple comparisons, with a 5% overall error rate. The subgroups of women selected in the design of this study were likely to have significantly different mean ages (e.g., pregnant women are more likely to be much younger than postmenopausal women), making age a potential confounding variable. Therefore, we performed analysis of covariance controlling for group, then for age, to determine whether density differences are explained by age. Linear regression was performed to analyze the relation between density and time between delivery and CSF sampling. For all determinations, P < 0.05 determined significance.


Individual CSF density measurements are shown in Figure 1. Mean CSF densities, standard deviations of measurements, and upper limits of hypobaricity for each group are presented in Table 1. Analysis of variance revealed significant differences in CSF density between groups (P < 0.0001). Pair-wise comparisons revealed that CSF density during the peripartum period (term pregnant and postpartum groups) is significantly less than in men and nonpregnant pre- and postmenopausal women (Table 1). Means, standard deviations, and ranges of ages in each group are also shown in Table 1. Density differences are not explained by age, because analysis of covariance controlling for group revealed no effect of age on density. Controlling for age, analysis of covariance showed the groups to differ still (P = 0.02). There was no correlation between density and time from delivery to CSF sampling in postpartum women, which ranged from 9 to 70.5 h (mean 28.1+/-17.5 h).

Figure 1. Density measurements of cerebrospinal fluid from 44 patients.

Figure 1. Density measurements of cerebrospinal fluid from 44 patients.

Table 1. Cerebrospinal Fluid (CSF) Densities in Five Groups of Patients

Table 1. Cerebrospinal Fluid (CSF) Densities in Five Groups of Patients
Table 1. Cerebrospinal Fluid (CSF) Densities in Five Groups of Patients


The density of a spinal anesthetic solution relative to human CSF (i.e., baricity) is an important determinant of extent of subarachnoid block. [1,10,11]Stienstra and colleagues showed that small differences in density (0.00060 g/ml) influence intrathecal distribution of local anesthetics both clinically [12]and in an in vitro model. [13]Several authors recently proposed that small density differences between CSF and dextrose-free intrathecal agents could explain the rapid, high sensory block observed after administration of local anesthetics and/or lipid-soluble opioids for labor analgesia. [10,11,14–16]Baricity also was proposed as an explanation for postural differences in extent of block produced by dextrose-free intrathecal anesthetics. [6,17]Hypobaric intrathecal solutions have been defined as those with densities less than three standard deviations below mean human CSF density. [1,4]Previous CSF density studies suggest that this upper limit of hypobaricity is 0.99980 g/ml4or 0.99940 g/ml5(Figure 2). Most authors quote the latter. [2,3]Our data suggest that actual values are greater (1.00016–1.00037 g/ml), and vary for different subgroups of patients (Table 1, Figure 2). Therefore, density differences between dextrose-free intrathecal drugs and CSF are greater than previously reported. [2,3]These findings are consistent with hypobaricity as a mechanism to explain postural effects on extent of sensory block and side effects produced by dextrose-free intrathecal anesthetics. [10,11,14–16].

Figure 2. Comparison of human cerebrospinal fluid density measurements reported by Davis and King [4]and Levin et al., [5]and those measured in this study. Densities are presented as mean+/-three standard deviations (g/ml). Mean density minus three standard deviations has been defined as the upper limit of hypobaricity, [1,4]previously taken to be 0.99940 g/ml, [3,5]and determined by us to range from 1.00016 to 1.00037 g/ml.

Figure 2. Comparison of human cerebrospinal fluid density measurements reported by Davis and King [4]and Levin et al., [5]and those measured in this study. Densities are presented as mean+/-three standard deviations (g/ml). Mean density minus three standard deviations has been defined as the upper limit of hypobaricity, [1,4]previously taken to be 0.99940 g/ml, [3,5]and determined by us to range from 1.00016 to 1.00037 g/ml.

Two groups of investigators previously measured the density of human CSF at 37 degrees C, [4,5]each reporting different means (Figure 2). Davis and King [4]studied 9 patients (7 males, 2 females) undergoing pneumoencephalography for central neurologic disorders, and Levin et al. studied 9 healthy male volunteers. [5]Densities were calculated from mass and volume measured at 37 degrees Celsius. Both studies reported 1.7- to 7.5-fold larger standard deviations than we observed (Table 1, Figure 2). Reduced accuracy and unspecified precision of mass, volume, and temperature measurements associated with their techniques limit the findings of both. In addition, restriction of determinations to specific subpopulations (individuals with neurologic disorders [4]and male volunteers [5]) limits application of their results to all patients. These differences in technique and patient selection likely account for the differences between our results and those of the previous reports.

Our data indicate that CSF density is significantly lower in the peripartum period. The chemical basis of this finding remains to be elucidated. Two previous studies failed to demonstrate a significant difference in CSF total protein content between pregnant and nonpregnant patients, [8,9]although a wide range of values was reported in both groups. Progesterone may be a physiologic mediator of altered CSF densities during human pregnancy, because progesterone treatment of estrogen-primed nonpregnant rabbits significantly alters Sodium+, Potassium+-ATPase activity in isolated choroid plexi. [18]This enzymatic activity is the primary driving force for CSF production. [19].

We chose the mechanical oscillation technique to measure density because of its well-established accuracy (+/-0.00001 g/ml) and high level of precision. [20–23],* Pycnometry predated the mechanical oscillation technique for density determination, and despite its limitations, had been the method of choice for density determination. [22]However, soon after its introduction in 1966, the mechanical oscillation technique supplanted pycnometry as the preferred method for density determination for fluids (including gases, aqueous and nonaqueous liquids, organic and inorganic solutions, and colloidal suspensions) in a variety of industries, for both research and quality applications. [20],* This method has been compared to pycnometry, and is simpler, requires less time and smaller samples, and gives more reproducible results.*

Although all samples were inspected visually for gross blood contamination, we did not perform microscopic examination of the CSF samples. Although nonvisible amounts of blood may impact on density measurements, this source of error would be expected to confound results of previous studies as well. Such contamination also would be expected to influence density measurements in each of the five groups evaluated in this study, which does not appear to be the case, given the intergroup differences in both density means and standard deviations of the means. Because the density of human blood is significantly greater than that of CSF, [20,23]blood contamination of CSF samples would be expected to result in significantly greater density values. However, mean CSF density values of all five groups were found to be intermediate between the two previously reported values. The most striking finding is the exceedingly high reproducibility in each of the five groups studied.

In summary, we determined values for the density of human CSF. The ranges of values measured are much narrower than previously reported, likely because of the greater accuracy and precision of the measurement technique used. The values defining hypobaricity of intrathecal solutions are greater than previously proposed. In addition, we observed significant differences in mean CSF density among subgroups studied, with density significantly lower at term gestation and immediately after pregnancy as compared with men and nonpregnant women. Consideration of baricity as a mechanism for observed differences in extent of block using different dextrose-free intrathecal solutions should take into account differences in CSF densities between patient subgroups.

The authors thank Ronald A. Gabel, M.D., for assistance in the preparation of the manuscript.

*Strunk DH, Aicken JC, Hamman JW, Andreasen AA: Density meter determination of proof of ethanol-water solutions: Collaborative study. Journal of the Association of Official Analytical Chemists 1982; 65:218–23.*.


Greene NM: Distribution of local anesthetic solutions within the subarachnoid space. Anesth Analg 1985; 64:715-30.
Nicol ME, Holdcroft A: Density of intrathecal agents. Br J Anaesth 1992; 68:60-3.
Horlocker TT, Wedel DJ: Density, specific gravity, and baricity of spinal anesthetic solutions at body temperature. Anesth Analg 1993; 76:1015-8.
Davis H, King WR: Densities of cerebrospinal fluid of human beings. ANESTHESIOLOGY 1954; 15:666-72.
Levin E, Muravchick S, Gold MI: Density of normal human cerebrospinal fluid and tetracaine solutions. Anesth Analg 1981; 60:814-7.
Russell IF: Spinal anesthesia for cesarean delivery with dilute solutions of plain bupivacaine: The relationship between infused volume and spread. Reg Anesth 1991; 16:130-6.
Abouleish A, Abouleish E, Camann W: Combined spinal-epidural analgesia in advanced labour. Can J Anaesth 1994; 41:575-8.
Marx GF, Orkin LR: Cerebrospinal fluid proteins and spinal anesthesia in obstetrics. ANESTHESIOLOGY 1965; 26:340-3.
Davis LE: Normal laboratory values of CSF during pregnancy. Arch Neurol 1979; 36:443.
Cohen SE, Cherry CM, Holbrook RH Jr, El-Sayed YY, Gibson RN, Jaffe RA: Intrathecal sufentanil for labor analgesia—sensory changes, side effects, and fetal heart changes. Anesth Analg 1993; 77:1155-60.
Blomqvist H, Nilsson A: Is glucose-free bupivacaine isobaric or hypobaric? Reg Anesth 1989; 14:195-8.
Stienstra R, van Poorten JF: The temperature of bupivacaine 0.5% affects the sensory level of spinal anesthesia. Anesth Analg 1988; 67:272-6.
Stienstra R, Gielen M, Kroon JW, Van Poorten F: The influence of temperature and speed of injection on the distribution of a solution containing bupivacaine and methylene blue in a spinal canal model. Reg Anesth 1990; 15:6-11.
Campbell DC, Camann WR, Datta S: The addition of bupivacaine to intrathecal sufentanil for labor analgesia. Anesth Analg 1995; 81:305-9.
Hamilton CL, Cohen SE: High sensory block after intrathecal sufentanil for labor analgesia. ANESTHESIOLOGY 1995; 83:1118-21.
Richardson MG, Lee AC, Wissler RN: High spinal anesthesia after epidural test dose administration in five obstetric patients. Reg Anesth 1996; 21:119-23.
Carrie LES, O'Sullivan G: Subarachnoid bupivacaine 0.5% for caesarean section. Eur J Anaesthesiol 1984; 1:275-83.
Lindvall-Axelsson M, Owman C: Actions of sex steroids and corticosteroids on rabbit choroid plexus as shown by changes in transport capacity and rate of cerebrospinal fluid formation. Neurol Res 1990; 12:181-6.
Strange K: Regulation of solute and water balance and cell volume in the central nervous system. J Am Soc Nephrol 1992; 3:12-27.
Hinghofer-Szalkay H: Method of high-precision microsample blood and plasma densitometry. J Appl Physiol 1986; 60:1082-8.
Hinghofer-Szalkay H: Volume and density changes of biological fluids with temperature. J Appl Physiol 1985; 59:1686-9.
Kenner T, Leopold H, Hinghofer-Szalkay H: The continuous high-precision measurement of the density of flowing blood. Pflugers Arch 1977; 370:25-9.
Kenner T: The measurement of blood density and its meaning. Basic Res Cardiol 1989; 84:111-24.