Epidural And Paravertebral Space

Foundations and mechanisms

Conceptual anatomy: potential versus real spaces

The epidural space is a circumferential, extradural compartment extending from the foramen magnum, where dura is adherent to skull, to the sacral hiatus. It lies between the dura mater and the vertebral canal, bounded anteriorly by the posterior longitudinal ligament and vertebral bodies/discs, posteriorly by laminae and ligamenta flava, and laterally by pedicles and intervertebral foramina. It is not an empty space: it contains fat, lymphatics, segmental arteries, spinal nerve roots and the internal vertebral venous plexus, which is valveless and communicates with thoracoabdominal venous systems.

The thoracic paravertebral space is a wedge-shaped compartment lateral to the vertebral bodies containing spinal nerves, dorsal and ventral rami, sympathetic chain and intercostal vessels. It is bounded anterolaterally by parietal pleura, posteriorly by the superior costotransverse ligament, medially by vertebral body, disc and intervertebral foramen, and superiorly/inferiorly by rib heads. It communicates medially with the epidural space, laterally with intercostal spaces, and cranio-caudally across adjacent paravertebral levels.

Core compartments and contents

Feature	Epidural space	Thoracic paravertebral space
Principal neural targets	Spinal nerve roots, dorsal root ganglia, intradural roots by diffusion	Spinal nerve after exit, dorsal/ventral rami, sympathetic chain
Typical block pattern	Bilateral, segmental or extensive neuraxial block	Predominantly unilateral somatic and sympathetic block
Key vascular structure	Batson venous plexus; risk of intravascular catheterisation	Intercostal vessels; vascular absorption can be rapid
Major anatomical hazard	Dural puncture, intrathecal or subdural injection	Pleural puncture, pneumothorax, epidural spread

Mechanisms of neural blockade

Local anaesthetics block voltage-gated sodium channels from the intracellular side, stabilising the inactivated state and preventing action potential propagation. Differential blockade reflects fibre diameter, myelination, firing frequency and anatomical exposure. Preganglionic sympathetic B fibres are most susceptible, followed by C fibres and A-delta fibres; large A-alpha motor fibres are relatively resistant. Clinically, sympathetic block often extends 2–6 dermatomes above sensory loss, while motor block is usually less extensive.

In the epidural space, drug spreads longitudinally, circumferentially and through intervertebral foramina. Block height is influenced by dose, volume, site of injection, patient age, pregnancy, intra-abdominal pressure, vertebral canal compliance and epidural fat. A commonly cited adult approximation is that 1–2 ml of local anaesthetic per dermatome is required in the lumbar/thoracic epidural space, although this is imprecise and less reliable at extremes of age, pregnancy and spinal deformity. Thoracic epidural dosing is more segmental; lumbar dosing more readily spreads cranially with larger volumes.

In paravertebral blockade, injectate spreads cranio-caudally within the paravertebral gutter and laterally into intercostal spaces. Medial spread through the intervertebral foramen may produce epidural extension, explaining occasional bilateral or hypotensive blocks. Single-injection thoracic paravertebral block commonly covers approximately 3–5 dermatomes; continuous catheter techniques are used when reliable multilevel analgesia is required.

Pressures, dimensions and clinically relevant numbers

• The epidural space is narrowest in the cervical region and widest caudally; approximate posterior epidural depth is 1–1.5 mm cervical, 3–5 mm thoracic and 5–6 mm lumbar, with large interindividual variation.
• The adult spinal cord usually terminates at L1 but may extend to L2; dural sac typically ends at S2.
• Ligamentum flavum thickness is usually greatest in the lumbar region, commonly 3–5 mm, and may be calcified or discontinuous in the midline, particularly cervicothoracically.
• Normal epidural pressure is often subatmospheric in the thoracic region, but pressure becomes positive with coughing, straining, pregnancy, obesity or raised intra-abdominal pressure.
• Loss-of-resistance technique depends on transition from dense ligamentum flavum to a lower-resistance epidural compartment; false loss may occur in interspinous ligaments, paraspinal muscle or cystic degeneration.

Classification of neuraxial and paravertebral spread

Pattern	Mechanism	Clinical implication
Segmental epidural block	Limited longitudinal spread around catheter tip	Useful for thoracic surgery and rib fracture analgesia with less lower limb motor block
Extensive epidural block	High volume, reduced epidural capacity, cephalad spread	Hypotension, bradycardia, respiratory compromise if high thoracic/cervical
Subdural spread	Injection between dura and arachnoid	Delayed, patchy, disproportionately high sensory block with relative motor sparing
Intrathecal spread	Dural puncture or catheter migration into CSF	Rapid dense motor/sensory block; risk of total spinal
Paravertebral unilateral spread	Local anaesthetic within thoracic paravertebral space	Unilateral analgesia with less sympathetic block than epidural
Paravertebral epidural extension	Medial foraminal spread	Bilateral block and hypotension may occur

Pharmacological principles and dose constraints

Choice of local anaesthetic balances onset, density, duration and toxicity. Ropivacaine and levobupivacaine are commonly favoured for continuous epidural or paravertebral analgesia because they produce less cardiotoxicity and relatively less motor block than racemic bupivacaine. Vasoconstrictors may reduce systemic absorption and act as intravascular markers, but adrenaline-containing solutions are avoided where spinal cord blood supply is a concern or when interpretation of tachycardia is unreliable.

Drug	Typical neuraxial/paravertebral concentration	Common maximum single-shot dose	Key point
Bupivacaine	0.0625–0.25% epidural; 0.25–0.5% bolus	~2 mg kg-1; 150 mg often cited	Most cardiotoxic amide; avoid large inadvertent intravascular dose
Levobupivacaine	0.1–0.25% epidural/PVB	~2 mg kg-1; 150 mg often cited	Less cardiotoxic S-enantiomer
Ropivacaine	0.1–0.2% infusion; 0.375–0.75% bolus	~3 mg kg-1; 200–225 mg often cited	Less motor block; commonly used for thoracic analgesia
Lidocaine	1–2% for rapid onset epidural	3 mg kg-1 plain; 7 mg kg-1 with adrenaline	Rapid onset; useful for catheter testing or surgical extension

Systemic absorption is greatest from intercostal and paravertebral sites, then epidural, then brachial plexus/subcutaneous tissues. Toxicity risk is therefore determined not only by total dose but by vascularity, fractionation, aspiration reliability and patient factors. For examination purposes, lipid rescue should be associated with suspected local anaesthetic systemic toxicity: 20% lipid emulsion 1.5 ml kg^-1 bolus, followed by 15 ml kg^-1 h^-1, with repeat boluses and increased infusion according to national guidance.

Clinical assessment and investigations

Clinical presentation: what the anatomy predicts

Assessment of the epidural and paravertebral spaces is usually indirect, inferred from sensory, motor, autonomic and complication profiles after needle/catheter placement. Epidural spread is longitudinal, circumferential and segmental but variable because of septae, epidural fat, venous plexuses and foraminal escape. Thoracic paravertebral spread is typically unilateral, segmental, and may extend through intercostal spaces, epidural foramina or prevertebral planes; therefore bilateral or extensive sympathectomy after a paravertebral block suggests epidural or intrathecal spread.

Clinical finding	Likely anatomical/technical implication	Important differentials
Unilateral epidural block, patchy sacral sparing	Catheter lateralisation, foraminal exit, midline septum, inadequate volume	Wrong level, inadequate dose, rapid labour progression
Dense motor block with small epidural dose	Intrathecal or subdural placement; excessive neuraxial spread	Pre-existing neuropathy, local anaesthetic sensitivity, spinal cord pathology
High block: hypotension, bradycardia, dyspnoea, arm weakness	Cephalad neuraxial spread; blockade of T1–T4 cardioaccelerators and cervical roots	Local anaesthetic systemic toxicity, anaphylaxis, pulmonary embolism, haemorrhage
Pleuritic pain, hypoxia after paravertebral block	Pleural puncture/pneumothorax	Rib fracture, atelectasis, PE, myocardial ischaemia
Severe new back pain with motor/sphincter deficit	Epidural haematoma or abscess until proven otherwise	Disc prolapse, spinal infarction, cauda equina syndrome, functional neurological disorder

Bedside assessment and interpretation

Document baseline neurology before neuraxial or paravertebral procedures when risk is increased: anticoagulation, sepsis, trauma, spinal stenosis, previous surgery, neuropathy or obstetric neurological symptoms. After block, assess dermatomal sensory level with cold or pinprick bilaterally, motor power using a modified Bromage score, haemodynamics and respiratory function. A clinically effective thoracic epidural for laparotomy often requires bilateral sensory change spanning approximately T4–T12; excessive cephalad spread above T4 risks profound sympathectomy. Paravertebral block should be predominantly unilateral over 3–5 dermatomes after a single injection; multilevel or catheter techniques may cover wider fields.

Score	Modified Bromage motor assessment
0	Full flexion of hips, knees and ankles
1	Unable to raise extended leg; able to flex knees
2	Unable to flex knees; able to move feet
3	Unable to move hips, knees or feet

Aspirate before dosing but recognise poor sensitivity for intravascular or intrathecal catheter migration. A commonly used epidural test dose is lidocaine 1.5% with adrenaline 1:200,000, 3 mL, containing 45 mg lidocaine and 15 micrograms adrenaline. Intravascular injection is suggested by an increase in heart rate of ≥20 beats min⁻¹ or systolic arterial pressure ≥15 mmHg within 30–60 seconds; sensitivity is reduced by beta-blockade, labour pain, general anaesthesia and advanced age. Intrathecal injection may produce rapid lower limb warmth, sensory change or motor block within 3–5 minutes. A negative test dose does not exclude malposition.

Differential diagnosis of acute deterioration

High neuraxial block presents with ascending numbness, upper limb weakness, dysphonia, respiratory difficulty, hypotension and bradycardia. Total spinal anaesthesia adds loss of consciousness and apnoea from brainstem hypoperfusion or high cervical spread. Local anaesthetic systemic toxicity is distinguished by circumoral paraesthesia, tinnitus, metallic taste, agitation, seizures, QRS widening, ventricular arrhythmia or cardiovascular collapse; bupivacaine is particularly cardiotoxic because of slow dissociation from myocardial sodium channels. Anaphylaxis, vasovagal syncope, haemorrhage and myocardial ischaemia must be considered concurrently.

Investigations and thresholds

Routine imaging is not required for uncomplicated epidural or paravertebral blockade. Investigation is driven by neurological deficit, infection, suspected bleeding, respiratory compromise or uncertainty of catheter position.

• Coagulation before neuraxial procedures: seek platelet count, PT/INR, APTT and fibrinogen when clinically indicated rather than universally. Commonly accepted thresholds are platelets ≥80 × 10⁹ L⁻¹ for epidural placement/removal in non-obstetric high-risk practice, INR ≤1.4, and no unexplained coagulopathy. In obstetrics, stable gestational thrombocytopenia may be acceptable at 70–80 × 10⁹ L⁻¹ after senior risk assessment. Catheter removal carries similar bleeding risk to insertion.
• Anticoagulant timing: follow ASRA/ESAIC/AoA guidance. Typical minimum intervals include prophylactic LMWH 12 h, therapeutic LMWH 24 h before neuraxial puncture; unfractionated heparin infusion stopped 4–6 h with normal APTT; clopidogrel 5–7 days; ticagrelor 5 days; prasugrel 7 days; apixaban/rivaroxaban usually 72 h, longer with renal impairment or high bleeding risk.
• Suspected epidural haematoma: urgent MRI whole relevant spine is the investigation of choice; CT myelography is second-line if MRI unavailable. Red flags are severe back pain, radicular pain, progressive motor weakness, sensory level or bladder/bowel dysfunction. Neurosurgical decompression is time-critical, with best neurological recovery when performed within 8–12 h of symptom onset.
• Suspected epidural abscess: MRI with gadolinium is preferred. Fever, back pain and neurological deficit form the classical triad but occur in a minority. CRP and ESR are sensitive screening markers; normal inflammatory markers make abscess less likely but do not exclude early disease. Obtain blood cultures before antibiotics if this will not delay treatment.
• Paravertebral complications: pneumothorax is assessed by lung ultrasound or erect chest radiograph; ultrasound signs include absent lung sliding and lung point. CT chest is most sensitive when the diagnosis is uncertain or the patient is ventilated. Hypoxaemia, increasing airway pressures or haemodynamic compromise mandate urgent decompression if tension pneumothorax is suspected.
• Catheter position: contrast radiography, fluoroscopy or CT epidurography may demonstrate epidural, subdural, intrathecal, intravascular or paravertebral spread, but are rarely first-line in acute instability. Ultrasound is useful pre-procedurally to estimate depth, midline and interspace, especially in obesity or abnormal anatomy.

Management, pharmacology and procedures

Epidural analgesia and anaesthesia

Epidural techniques are used for perioperative analgesia, labour analgesia, thoracic/abdominal surgery, rib fractures and selected chronic pain interventions. Management begins with defining the intended neuraxial level: thoracic epidurals provide segmental analgesia with smaller doses, whereas lumbar epidurals rely more on cephalad spread. Spread is influenced by injected volume, age, pregnancy, obesity, epidural venous engorgement, spinal stenosis, prior surgery and catheter direction. A commonly examined approximation is 1–2 ml per dermatome in adults, but this is unreliable at extremes of age and in pregnancy.

Technique and confirmation

The epidural space is identified using loss of resistance to saline or air; saline is generally preferred to avoid patchy block, pneumocephalus and venous air embolism. Typical skin-to-epidural depth in adults is 4–6 cm, but may exceed 8 cm in obesity. Catheters are usually threaded 3–5 cm into the epidural space; excessive insertion increases unilateral block, intravascular migration and coiling. A test dose aims to detect intrathecal or intravascular placement, although it is imperfect. A common adult test dose is 3 ml lidocaine 1.5% with adrenaline 1:200 000 (45 mg lidocaine plus 15 micrograms adrenaline): intrathecal placement produces rapid dense motor block, while intravascular injection may produce tachycardia >20 beats min⁻¹ or systolic BP rise >15 mmHg, though sensitivity is reduced by beta-blockade, labour pain and general anaesthesia.

Indication	Typical epidural regimen	Key management point
Labour analgesia	Bupivacaine 0.0625–0.1% or levobupivacaine 0.0625–0.125% with fentanyl 2 micrograms ml−1; patient-controlled bolus 5–10 ml, lockout 10–20 min	Maintain maternal BP; monitor fetal heart rate after initiation and top-ups
Thoracic abdominal surgery	Bupivacaine/levobupivacaine 0.1–0.125% or ropivacaine 0.1–0.2% plus fentanyl 2 micrograms ml−1; infusion 4–12 ml h−1	Aim sensory block T4–T12 depending on incision; avoid excessive sympathetic block
Caesarean anaesthesia via epidural	Lidocaine 2% with adrenaline ± bicarbonate; chloroprocaine 3%; or bupivacaine 0.5%, incremental 5 ml boluses	Urgency-dependent; exclude high block before each bolus

Local anaesthetic choice reflects potency, onset and cardiotoxicity. Lidocaine has onset 5–10 min and duration 60–120 min; bupivacaine onset 15–20 min and duration 2–4 h, with greater cardiotoxicity due to avid myocardial sodium-channel binding. Ropivacaine and levobupivacaine are less cardiotoxic and produce relatively less motor block. Neuraxial opioids act at dorsal horn μ-receptors: fentanyl and sufentanil are lipophilic with rapid onset and limited rostral spread; morphine is hydrophilic with delayed respiratory depression risk up to 12–24 h. Preservative-free preparations are mandatory.

Paravertebral block and catheter techniques

Thoracic paravertebral block deposits local anaesthetic adjacent to the spinal nerve roots within the wedge-shaped paravertebral space, producing unilateral somatic and sympathetic blockade. Indications include thoracotomy, breast surgery, rib fractures and unilateral abdominal wall analgesia. Ultrasound-guided approaches identify the transverse process, superior costotransverse ligament, pleura and internal intercostal membrane; needle passage through the superior costotransverse ligament may produce pleural displacement. Single-shot volumes are commonly 15–25 ml at one level, or 3–5 ml per level for multi-level injections. Continuous catheters are infused with ropivacaine 0.2% at 5–10 ml h⁻¹ or intermittent programmed boluses. Compared with thoracic epidural analgesia, paravertebral analgesia usually causes less hypotension, urinary retention and bilateral motor block, with comparable analgesia for thoracotomy in meta-analyses, although catheter failure and pleural puncture remain relevant.

Anticoagulation, infection and follow-up

Neuraxial and deep paravertebral procedures must follow anticoagulation guidance because vertebral canal haematoma can cause irreversible cord injury. ASRA/ESAIC principles require assessment of renal function, traumatic puncture, concomitant antiplatelets and timing of catheter removal. Typical intervals include: prophylactic low-molecular-weight heparin ≥12 h before puncture and ≥4 h after catheter removal before next dose; therapeutic LMWH ≥24 h; unfractionated heparin i.v. stop 4–6 h and confirm normal APTT; clopidogrel stop 5–7 days; warfarin INR should be ≤1.4–1.5 for insertion/removal; DOACs usually require 48–72 h depending on renal function and bleeding risk.

Complications and immediate management

• Hypotension and bradycardia: due to sympathetic blockade and reduced venous return. Treat with left uterine displacement in pregnancy, crystalloid judiciously, vasopressors such as phenylephrine 50–100 micrograms i.v. boluses or metaraminol 0.5–1 mg; ephedrine 3–6 mg if bradycardic.
• High or total spinal: rapid ascending block, dyspnoea, hypotension, arm weakness, dysphonia or unconsciousness. Stop injection, call for help, secure airway, ventilate with 100% oxygen, support circulation with vasopressors/adrenaline, and manage as neuraxial emergency.
• Local anaesthetic systemic toxicity: circumoral tingling, tinnitus, seizures, arrhythmias. Stop injection, oxygenate, treat seizures with benzodiazepines, avoid large propofol doses in cardiovascular collapse, and give 20% lipid emulsion: 1.5 ml kg⁻¹ bolus, then 15 ml kg⁻¹ h⁻¹; repeat bolus for instability, maximum commonly 12 ml kg⁻¹.
• Dural puncture and PDPH: postural headache within 5 days, often with neck stiffness or auditory symptoms. Conservative measures include hydration, simple analgesia and caffeine. Epidural blood patch uses 15–20 ml autologous blood at or below puncture level; success after first patch is approximately 60–80%.
• Epidural haematoma/abscess: severe back pain, radicular pain, motor weakness, sphincter dysfunction or fever. Urgent MRI and neurosurgical referral; decompression within 8–12 h of motor deficit offers best neurological recovery.
• Paravertebral-specific complications: pleural puncture, pneumothorax, vascular puncture, hypotension from epidural spread, and inadvertent neuraxial injection.

Follow-up requires documented sensory level, motor power, pain scores, haemodynamics, catheter depth and infusion prescription. Motor block should be assessed serially, for example with the Bromage scale; unexpected dense or progressive block mandates stopping the infusion and urgent review. Catheters should be removed aseptically as soon as no longer required, with continued neurological observations for at least 24 h after removal in higher-risk patients.

Exam controversies and advanced synthesis

Conceptual controversies: the “space” is not a simple tube

The mature viva answer should challenge the simplistic model of a uniform epidural or paravertebral cavity. The epidural space is a discontinuous, fat-filled, venous plexus-containing potential space with posterior, lateral and anterior compartments; spread is governed by compliance, septations, foraminal escape, pressure gradients, age-related reduction in epidural fat compliance, pregnancy-related venous engorgement and injected volume. The plica mediana dorsalis is inconsistently demonstrated and should not be over-stated as a fixed midline septum. Similarly, “negative epidural pressure” is not a reliable anatomical property: much of the observed pressure change reflects tenting of the ligamentum flavum, needle advancement dynamics and measurement artefact. It is therefore unsafe to identify the epidural space using hanging-drop techniques alone.

The ligamentum flavum is also not invariably fused in the midline. Cadaveric and imaging series describe midline gaps particularly in the cervical region, reported in up to approximately 50–70% of cervical levels, less often in thoracic levels and least often in the lumbar region. This explains false loss of resistance, especially with paramedian approaches and in thin thoracic ligamentum flavum. In the Primary FRCA, a high-scoring answer links this to the smaller thoracic epidural depth, steeper spinous processes, proximity of cord above L1–L2 and the need for meticulous incremental advancement.

Loss of resistance, test dosing and drug spread: practical controversies

Issue	Exam-level synthesis
Air versus saline	No definitive superiority in all outcomes, but saline is commonly preferred. Air may cause patchy block, pneumocephalus after dural puncture, venous air embolism and compression of neural structures. Saline may dilute local anaesthetic or obscure CSF recognition. Use small volumes, typically <5 ml.
Epidural test dose	Traditional adult test dose: 3 ml lidocaine 1.5% with adrenaline 1:200,000, containing 45 mg lidocaine + 15 micrograms adrenaline. Intrathecal injection produces rapid dense motor/sensory block; intravascular injection may produce tachycardia. In obstetrics, sensitivity and specificity are reduced by contractions, anxiety, beta-blockade and haemodynamic variability; fractionated dosing and aspiration remain essential.
Segmental dosing	Lumbar epidural surgical analgesia often requires approximately 1–2 ml per dermatome; thoracic dosing is lower, commonly 0.7–1 ml per dermatome. Spread is greater in pregnancy and elderly patients because of reduced epidural compliance and venous engorgement.
Paravertebral dosing	Single-shot thoracic paravertebral injection commonly uses 0.3 ml kg-1 or 15–25 ml of long-acting local anaesthetic; continuous infusions often use ropivacaine or levobupivacaine 0.1–0.2% at approximately 5–10 ml h-1, adjusted for weight and toxicity risk.

Paravertebral space: anatomy versus block behaviour

The thoracic paravertebral space is classically wedge-shaped, bounded posteriorly by the superior costotransverse ligament, anterolaterally by pleura and medially by the vertebral body, disc and intervertebral foramen. However, block behaviour is better understood as injection into a fascial compartment variably subdivided by the endothoracic fascia. Spread may be ipsilateral segmental, intercostal, cranio-caudal, epidural through the intervertebral foramina, or occasionally contralateral. Thus, paravertebral block can produce sympathectomy and hypotension, but usually less than thoracic epidural because bilateral neuraxial spread is less common.

A common viva pitfall is to equate erector spinae plane block with paravertebral block. Erector spinae injection is more superficial, posterior to the transverse process, and relies on variable anterior spread through costotransverse tissues. It may provide useful analgesia, but it is not anatomically equivalent to direct paravertebral deposition adjacent to the spinal nerve and sympathetic chain.

Guidelines: neuraxial and deep block safety

Thoracic paravertebral block is generally treated as a deep, non-compressible peripheral block; anticoagulation precautions should therefore approximate neuraxial standards. Guideline details vary between ASRA and ESAIC/ESRA documents, but the following thresholds are commonly examined:

Drug or parameter	Typical neuraxial/deep block interval
Aspirin/NSAIDs	No mandatory delay when used alone.
Clopidogrel	Stop for 5–7 days before neuraxial or deep block.
Prophylactic LMWH	Wait 12 h before needle/catheter placement; delay next dose at least 4 h after catheter removal.
Therapeutic LMWH	Wait at least 24 h; longer if renal impairment or high bleeding risk.
IV unfractionated heparin	Stop 4–6 h and confirm normal coagulation before block or catheter removal.
Warfarin	Usually require INR ≤1.4–1.5 for placement/removal, with neurological surveillance after removal.
DOACs	Commonly 72 h for high-dose apixaban/rivaroxaban/dabigatran; extend for renal impairment, especially dabigatran.
Platelets	Obstetric consensus often accepts neuraxial techniques at ≥70 × 109 L-1 if stable and no DIC; <50 × 109 L-1 is generally avoided.

Trials and outcome controversies

Thoracic epidural analgesia provides excellent dynamic analgesia after open thoraco-abdominal surgery, but outcome benefits are less definitive than older teaching implied. The MASTER trial in high-risk major abdominal surgery did not show a major reduction in composite mortality or major morbidity, although analgesia and some respiratory outcomes improved. Modern ERAS pathways favour thoracic epidural for selected open abdominal surgery but not routinely for laparoscopic colorectal surgery, where hypotension, urinary catheterisation and impaired mobilisation may offset benefit.

For thoracotomy, meta-analyses comparing thoracic paravertebral block with thoracic epidural generally show similar analgesia with less hypotension, urinary retention, nausea and technical failure for paravertebral techniques. For breast surgery, paravertebral blockade reduces acute opioid consumption and postoperative nausea; evidence for prevention of chronic postsurgical pain is suggestive but heterogeneous. The examiner is looking for balanced judgement: anatomy predicts unilateral segmental analgesia and fewer bilateral sympathetic effects, but pleural puncture, pneumothorax, vascular puncture, epidural spread and local anaesthetic systemic toxicity remain real risks.

High-yield viva pitfalls

• Do not state that the epidural space extends into the cranial cavity; it terminates at the foramen magnum, where spinal dura fuses with periosteum.
• Do not forget the anterior epidural space is narrow and contains the posterior longitudinal ligament; clinically useful catheter spread is predominantly posterior/lateral.
• Thoracic epidural insertion above the conus carries cord injury risk; loss of resistance must be interpreted with anatomy, depth and patient feedback.
• Persistent motor block, severe back pain, radicular pain or sphincter disturbance after neuraxial block is an emergency: MRI and decompression for epidural haematoma are time-critical, ideally within 8–12 h of neurological deficit.