Pharmacokinetics of drugs in the epidural and subarachnoid space

This chapter is directly related to the learning objectives of Section B(viii) from the 2023 CICM Primary Syllabus which expects the exam candidate to "describe the pharmacokinetics of drugs in the epidural and subarachnoid space". Owing to the fact that future private billings are not going to depend on this practice, the college has previously only weighed these learning objectives as L2, meaning "important and relevant... A good understanding" etc. 

The fact that most ICU trainees will not be administering neuraxial anaesthesia does not change the fact that most neuraxially administered drugs are anaesthetics. Moreover, owing to the understandable anaesthetic fondness for the safety of routine, most neuraxially administered drugs are sodium channel blockers, fentanyl and morphine. In view of these truths, the chapter's skew is profoundly opio-centric. However, with some luck some trainees may find themselves in intensive care centres which care for non-elective patients. In some scenarios they may be confronted with non-opioid drugs being administered directly into the cerebrospinal sancto sanctorum. These drugs may include antibiotics, cytotoxic agents, immunosuppressants or diagnostic contrast media. A small amount of HTML will be squandered in discussion of these topics.

From an exam relevance standpoint, no SAQs have so far plumbed these depths. It is not inconceivable that at some stage they may. The time-poor trainee would probably be better off studying respiratory or cardiovascular physiology instead. However, if for some reason reading on this topic is called for, the best peer-reviewed literature would probably have to be this old article by A.G.L Burm (1989). Specifically for intratecal drug administration, the best concise resource is a 1992 review by Jeffrey Kroin, also paywalled by Springer. 

In summary:

Domain Epidural pharmacokinetics: Intrathecal pharmacokinetics:
Absorption into target site

Rate of diffusion into CSF is slower, and depends on Fick pronciple:

  • Volume (thus, surface area of available meninges)
  • Concentration
  • Lipophilicity (pH, pKa)
  • Protein binding and free fraction
  • CSF flow rate/turbulence

Rate of diffusion into target tissue is rapid:

  • High CSF concentration
  • Short diffusion distance (2-4mm)

Epidural spaces are irregular, segmental, and the injected material encircles the dural sack.

Depends on baricity (density of injectate relative to CSF)
Systemic distribution

Two-compartment model: Rapid early distribution (into epidural fat) and then slowly back out.

  • Slow absorption; increased half life
  • The more lipophilic the drug, the faster it is cleared from the CSF 
Metabolism and elimination Normal mechanisms prevail. However, because of the haemodynamic effects of spinal (anaesthetic) drugs, the perfusion of liver and kidneys may be decreased, which could delay clearance. 

Pharmacokinetics of epidural drug administration

The epidural space

Before more pharmacokinetic issues are discussed, it is important to regard the site of administration in some detail. According to Anatomy for Anaesthetists by Ellis et al (8th ed, 2004 - p.121) its extent is as follows:

  • Superiorly, the epidural space extends to the foramen magnum, where it terminates at the fusion of spinal and periosteal layers of dura mater. This has some relevance for awful accidents (i.e. the intracranial extension of epidurally infused drugs is made impossible by this fixed anatomical limit).
  • Inferiorly, the epidural space extends to the sacrococcygeal membrane.
  • Posteriorly, it is bounded by the vertebral laminae, capsules of the facet joints and the ligamentum flavum
  • Laterally, it is bounded by the pedicles of the vertebral arches and the intervertebral foraminae
  • Anteriorly, it is bounded by the vertebral bodies, intervertebral discs and posterior longitudinal ligament.
  • Its contents are fat, nerve roots, blood vessels,  lymphatics and various haphazard fibrous connections to the ligamentum flavum (which can have an unpredictable effect on the course of an epidural catheter
  • It communicates freely with the paravertebral space via intervertebral foraminae.
  • Its arterial supply arises from anterior and posterior spinal arterial arcades, arising from spinal arteries entering the space through every intervertebral foramen. These anastomose freely with the anterior spinal artery.
  • Its venous drainage is via a plexus of valveless veins (Bateson's plexus) which is in the anterior epidural space. There are also posterior veins, which are most prominent in the cervical epidural space. The venous plexus receives blood shunted from thoracic and pelvic veins, which means that straining and coughing can transiently engorge them.

The space is irregular and segmental. The epidurogram here is reproduced from Shim et al (2011) with hope that a picture tells a thousand words. 

Epidurogram from Shim et al (2011)

These are images of the L-spine of a 57 year old woman who has had some L3/4 disk herniation. The needle tip is seen in the L4/5 intervertebral foramen. The epidural space is easily seen; it is triangular and wedge-like, with the largest part at the caudal end of each vertebra.

The space is distensible and injection of fluid tends to distort its dimensions,  confusing anatomists accustomed to making measurements from sections of cadavers. Its capacity is substantial. A classic paper by Burn Guyer and Langdon (1973) saw up to 40 ml of material being injected into the L3/4 epidural space, "as rapidly as the patients, who were conscious, could tolerate: about 1-2 ml per sec".  That 40ml of injected volume was fairly uncomfortable to administer and was seen to spread all the way up to the upper thoracic and even cervical levels. This answers the sensible question of "how much stuff can I safely put up there".

Absorption from the epidural space

Absorption after the injection of something into the epidural space is fairly guaranteed- there can be no question of its loss. The discussion of absorption and distribution therefore becomes more an issue of "where does the dose go".

After injection, the drug solution coats the cylindrical dural sack, spreading up and down (as well as anteriorly to encircle the cord) in a fairly random fashion. Quinn Hogan (2002) demonstrated this by injecting ink into the epidural spaces of three dead men and one live baboon (to control for effects of death). "Injected ink ...showed spread as rivulets through numerous small channels rather than as a unified advancing front". One should not think of their epidural bolus as a discrete blob of drug sitting in the epidural space.

From the epidural space, drugs may go four ways:

  • Exit the intervertebral foramina to reach the paraspinous muscle space
  • Distribute into epidural fat
  • Diffuse into ligaments
  • Diffuse across the spinal meninges and into the CSF

Ligamentous effects of local anaesthetics being profoundly uninteresting, the last of these possible routes is the most important. Virtually the only mechanism for drugs to make their way into the CSF from the epidural space is by diffusion across the spinal meninges. Specifically, the arachnoid mater is the main meningeal barrier to diffusion. Apart from that, a small minority of drugs will penetrate the systemic circulation and then appear in the CSF after diffusing out of the spinal cord.

Diffusion being the main influence on drug absorption, one can guess what sorts of Fickian properties and factors will influence the penetration of an epidurally administered drug into the CSF. To list them:

  • Concentration
  • Surface area, which is largely determined by the volume of the infused drug
  • Lipid solubility, and therefore pKa of the drug and pH of the solution 
  • Protein binding, which determines the free fraction of the drug

The main influence on epidural drug behaviour seems to be distribution into epidural fat. Burm (1989) blamed this for the 5-10-fold dose difference between subarachnoid and epidural drug administration. The drug then forms a reservoir and redistributes gradually, creating a longer effect. The speed of onset is also mainly related to lipid solubility; DiFazio et al (1986) were able to demonstrate a significantly faster rate of onset for lignocaine which was adjusted to a pH of 7.15 (as opposed to the standard 4.6).

CSF flow and volume is a nontrivial influence on epidural pharmacokinetics, as approximately 500ml of this fluid may wash over the administration site over the course of a day. This changes the concentration gradient between the epidural and subarachnoid space. The availability of fresh CSF to the administration site results in an increased movement of the drug from its epidural fat reservoir; the flow of drug-rich CSF also improves the delivery of the substance to the target sites. This could be a good thing or a bad thing. Cephalad CSF flow carried enough morphine from a lumber epidural site for Gourlay et al (1985) to be concerned about it reaching the respiratory centre and causing apnoea. 

Systemic distribution 

It appears that the systemic absorption of drugs from the epidural space follows a two-compartment model. The graph below is modified from a study by Burm et al (1987), who injected patients with radiolabelled bupivacaine. There is a clear fast initial phase and a slow late phase. The latter is probably due to the redistribution into (and then out of) the epidural fat, a substance for which bupivacaine has significant fondness. A completely imaginary epidural fat curve has been overlaid on the otherwise factual Burm graph below.

bupivacaine pharmacokinetics from Burm et al (1987)

The systemic distribution of the relevant agents is obviously going to be governed by the physicochemical characteristics of the specific agents and needs not be laboured here. It suffices to say that there is systemic distribution which depends on all the usual factors (protein binding, pKa, pH, etc). Peculiarities of note could include the propensity of lung water to accumulate local anaesthetic (owing to the low pH thereof) and to act as an additional reservoir. 

Pharmacokinetics of subarachnoid drug administration

The subarachnoid space

The subarachnoid space is the space between arachnoid mater and pia mater. It contains the CSF, and the spinal subarachnoid space communicates with the CSF spaces in the head. To describe the shape of this space as complex would be a significant understatement. Sass et al (2017) had produced an excellent 3D model based on a mixture of MRI reconstructions and cadaveric measurements, which could be interesting to look at. 

At the level of the small vascular structures in the pia mater, the arachoid space extends along the vessel walls as they penetrate into the spinal cord. There, the arachnoid space interfaces with a series of miniscule pia-lined fluid pockets associated with at least all the arteriole and venules in the CNS (Lam et al, 2017), if not individual groups of neurons. These little fluid pockets are called perivascular spaces or Virchow–Robin spaces in the literature. Some authors claim that these spaces are continuous with the subarachnoid space (Burm, 1989) whereas others point out that the fluid in these spaces is chemically distinct from CSF (Kuchinsky, 1972). Either way, these spaces are implicated as having some importance to the penetration of drugs that are administered into the subarachnoid space. The diffusion distance is not very great - the high CSF concentration achieved by injecting directly into the CSF easily penetrates to the target site (substantia gelatonosa, lamina II - only 2mm deep). In fact, nowhere in the spinal cord are you ever further than 5mm from some CSF.

Absorption and penetration to the target site

High CSF concentration  is achieved rapidly. Absorption into spinal tissue is therefore very rapid. With no meninges in the way of diffusion, the dose required is much smaller, and the onset of maximal effect is more rapid. The smaller dose also means that systemic toxicity is lower. For example, to administer enough baclofen for the same therapeutic effect one needs to give 60mg orally, or 600 μcg intrathecally (Penn, 1990).

Bypassing the blood-brain barrier is made easier with administration directly into the CSF, and drugs that conventionally are to hydrophilic to cross (eg. methotrexate) can be administered in this fashion. The high concentration that they achieve in this manner improves the penetration of even the least lipid-soluble drugs.

Systemic absorption is slower than with epidural, and there is no rapid distribution phase. This is because the epidural space is rich with vascular structures which permits rapid uptake. The perfusion of the subarachnoid space is less extensive, and systemic distribution of the drug will be slower. This has implications for drug effects. Drugs that would otherwise be rapidly metabolised or distributed (e.g. fentanyl or morphine) have persisting effects because their absorption from the CSF will be slow, and there is nothing there to metabolise them.

Increased half-life, particularly for hydrophilic drugs, can be expected, largely because of the above. "A few hours" is mentioned by Kroin (1992). The more lipophilic the drug is, the more rapidly absorbed into the spinal capillaries and therefore the more rapidly cleared from the CSF. 

Distribution and systemic disposition

The distribution of drugs within the subarachnoid space is probably more interesting than the usual discussions of their Vd. Greene (1985) has published an excellent review of the subject.

Baricity needs to be discussed, as it is a matter which is almost unique to spinal anaesthesia. The term refers to the comparative density of the spinal anaesthetic solution in comparison to the CSF. Specifically, baricity is defined as the density of the solution divided by the density of the CSF, which is usually 1.003 g/ml. The baricity of solutions is by convention compared at  37°C, as temperature affects the density of solutions (density is inversely related to temperature; the hotter it gets the less dense the solution).

Some anaesthetic drug mixtures have their density intentionally interfered with, so that they produce a specific desired effect. Isobaric solutions have the same density as the CSF; and therefore presumably they stay put wherever they've been injected. Hyperbaric solutions are heavier than CSF and therefore sink, which means if the patient is sitting upright the local anaesthetic will really only affect their sacral nerve roots. Hypobaric solutions are lighter than CSF, and will float to the surface. The latter seems to be something virtually extinct in today's anaesthesia; the technique of using lighter-than-CSF solutions to create a high block had previously occupied a position that is now well-served by thoracic epidurals.

Metabolism and elimination of epidural and intrathecal drugs

Beyond the occasional surprise (eg. the presence of COMT) the meninges do not perform any major metabolic function in terms of drug biotransformation and most of the dose ultimately enters systemic circulation unchanged. Metabolism of epidural drug doses is therefore dependent on the same pathways as all other drug metabolisms, which usually means the liver does all the work. Minimal free opiate and free local anaesthetic is recoverable from the urine. The only additional thing that needs to be mentioned is the haemodynamic effect of the epidural, which may decrease perfusion of the clearance organs and thereby prolong its own weak systemic effect. This probably has greater relevance for long-term infusions, where the metabolites of the drug may accumulate over time.

Non-anaesthetic intrathecal injections

With such exotica, it is probably not even important that the primary candidate knows of them. 

  • Baclofen (via pump, for spasticity)
  • Clonidine (analgesia)
  • Analgesics (palliative care, cancer pain)
  • Antineoplastic drugs (eg. methotrexate for primary CNS lymphoma)
  • Antibiotics (for ventriculitis or post-neurosurgical infections)


Burm, AGL. "Clinical pharmacokinetics of epidural and spinal anaesthesia." Clinical pharmacokinetics 16.5 (1989): 283-311.

Bernards, Christopher M., et al. "Epidural, Cerebrospinal Fluid, and Plasma Pharmacokinetics of Epidural Opioids (Part 1) Differences among Opioids." Anesthesiology: The Journal of the American Society of Anesthesiologists 99.2 (2003): 455-465.

Bernards, Christopher M., and Harlan F. Hill. "Physical and chemical properties of drug molecules governing their diffusion through the spinal meninges." Anesthesiology 77.4 (1992): 750-756.

Pleuvry, Barbara J. "Factors affecting drug absorption and distribution." Anaesthesia & Intensive Care Medicine 6.4 (2005): 135-138.

Gala, Foram B., and Yashant Aswani. "Imaging in spinal posterior epidural space lesions: A pictorial essay.The Indian journal of radiology & imaging 26.3 (2016): 299.

Shim, Jae Chol. "Lumbar Ventral Epidural Approach." Hanyang Medical Reviews 31.2 (2011): 90-102.

James, M. B., P. B. Guyer, and L. Langdon. "The spread of solutions injected into the epidural space: a study using epidurograms in patients with the lumbosciatic syndrome." British Journal of Anaesthesia 45.4 (1973): 338-345.

Cousins, Michael J. "Intrathecal and epidural administration of opioids." Anesthesiology 61 (1984): 276-310.

DiFazio, Cosmo A., et al. "Comparison of pH-adjusted lidocaine solutions for epidural anesthesia." Anesthesia and analgesia 65.7 (1986): 760-764.

Higuchi, Hideyuki, Yushi Adachi, and Tomiei Kazama. "Factors affecting the spread and duration of epidural anesthesia with ropivacaine." Anesthesiology: The Journal of the American Society of Anesthesiologists 101.2 (2004): 451-460.

Hogan, Quinn. "Distribution of solution in the epidural space: examination by cryomicrotome section." Regional Anesthesia and Pain Medicine 27.2 (2002): 150-156.

Gourlay, Geoffrey K., David A. Cherry, and Michael J. Cousins. "Cephalad migration of morphine in CSF following lumbar epidural administration in patients with cancer pain." Pain 23.4 (1985): 317-326.

Kern, C., D. Mautz, and C. Bernards. "Epinephrine is metabolized by the spinal meninges of monkeys and pigs." Anesthesiology: The Journal of the American Society of Anesthesiologists 83.5 (1995): 1078-1081.

Burm, A. G. L., et al. "Pharmacokinetics of lignocaine and bupivacaine in surgical patients following epidural administration.Clinical pharmacokinetics 13.3 (1987): 191-203.

Sass, Lucas R., et al. "A 3D subject-specific model of the spinal subarachnoid space with anatomically realistic ventral and dorsal spinal cord nerve rootlets." Fluids and Barriers of the CNS 14.1 (2017): 36.

Lam, Magdalena A., et al. "The ultrastructure of spinal cord perivascular spaces: Implications for the circulation of cerebrospinal fluid." Scientific reports 7.1 (2017): 12924.

Kuschinsky, W., et al. "Perivascular potassium and pH as determinants of local pial arterial diameter in cats: a microapplication study." Circulation Research 31.2 (1972): 240-247.

Horlocker, Terese T., and Denise J. Wedel. "Density, specific gravity, and baricity of spinal anesthetic solutions at body temperature." Anesthesia and analgesia 76.5 (1993): 1015-1018.

Greene, Nicholas M. "Distribution of local anesthetic solutions within the subarachnoid space." Anesthesia & Analgesia 64.7 (1985): 715-730.

Hodgson, Peter S., et al. "The neurotoxicity of drugs given intrathecally (spinal)." Anesthesia & Analgesia 88.4 (1999): 797-809.

Kroin, Jeffrey S. "Intrathecal drug administration." Clinical pharmacokinetics 22.5 (1992): 319-326.

Penn, R. D., J. S. Kroin, and J. M. Magolan. "Intrathecal baclofen in the treatment of spinal spasticity." Clinical Neuropharmacology 33.Suppl 2 (1990): 400-401.