Clinical Anatomy 17:544 553 (2004)

REVIEW

Lumbar Puncture: Anatomical Review of a Clinical Skill

J.M. BOON,1* P.H. ABRAHAMS,2 J.H. MEIRING,1
1

AND

T. WELCH3

Department of Anatomy, Unit of Clinical Anatomy, School of Medicine, Faculty of Health Sciences, University of Pretoria, Pretoria, South Africa 2 Kigezi International School of Medicine, Cambridge, Girton College, Cambridge, United Kingdom, St. Georges University Grenada and St. Vincent, West Indies 3 Kigezi International School of Medicine, Cambridge, Queens College, Cambridge, United Kingdom

The safe and successful performance of a lumbar puncture demands a working and specic knowledge of anatomy. Misunderstanding of anatomy may result in failure or complications. This review attempts to aid understanding of the anatomical framework, pitfalls, and complications of lumbar puncture. It includes special reference to 3D relationships, functional and imaging anatomy, and normal variation. Lumbar puncture is carried out for diagnostic and therapeutic purposes. Epidural and spinal anesthesia, for example, are common in obstetric practice and involve the same technique as diagnostic lumbar puncture except that the needle tip is placed in the epidural space in the former. The procedure is by no means innocuous and anatomical pitfalls include inability to nd the correct entry site and lack of awareness of structures in relation to the advancing needle. Headache is the most common complication and it is important to avoid traumatic and dry taps, herniation syndromes, and injury to the conus medullaris. With a thorough knowledge of the contraindications, regional anatomy and rationale of the technique, and adequate prior skills practice, a lumbar puncture can be carried out safely and successfully. Clin. Anat. 17:544 553, 2004. 2004 Wiley-Liss, Inc. Key words: clinical procedures; cerebrospinal uid; subarachnoid space; spinal anatomy

INTRODUCTION
Many authors (Abrahams and Webb, 1975; McMinn et al., 1984; Beahrs et al., 1986; Crisp, 1989; Ger and Evans, 1993; American Association of Clinical Anatomists, 1999; Cottam, 1999; Boon et al., 2001) have highlighted the crucial role of sound anatomical understanding in the safe and successful performance of clinical procedures. Many medicolegal cases are based upon inadequate knowledge or misunderstanding of anatomy and the prolongation of any procedure due to lack of detailed knowledge leads to increased morbidity and mortality (Beahrs et al., 1986; Graney, 1996). Even for so-called minor procedures such as lumbar puncture, complications may result if carried out without a proper understanding of the anatomical implications (Ger, 1996). In the United Kingdom, the landmark paper Tomorrows Doctors (General Medical Council, 1993) focuses on the acquisition of practical skills. Similarly, the Association of American Medical Colleges (1998) states that, before graduation, a student should dem

onstrate the ability to perform routine technical procedures including the following minimum: venepuncture, inserting an intravenous catheter, arterial puncture, thoracocentesis, lumbar puncture, inserting a nasogastric tube, inserting a Foleys catheter, and suturing lacerations. The General Medical Council (2002) also stated that one of the duties of a registered doctor is to keep his/her professional knowledge and skills up to date. Kneebone (1999) pointed out that condence in performing a procedure comes from a knowledge base that knows what to expect. Similarly, Wigton (1992) mentioned that the most important elements of procedural competency are the cognitive
*Correspondence to: Dr. J.M. Boon, Department of Anatomy, Section of Clinical Anatomy, School of Medicine, Faculty of Health Sciences, University of Pretoria, Pretoria 0001, South Africa. E-mail: jmboon@medic.up.ac.za Received 2 December 2002; Revised 19 September 2003 Published online in Wiley InterScience (www.interscience.wiley. com). DOI 10.1002/ca.10250

2004 Wiley-Liss, Inc.

Lumbar Puncture

545

aspects. Anatomy plays a major role in this domain. This article reviews the clinical anatomy of lumbar puncture. It starts with a step-by-step description of the procedure and focuses on the anatomical pitfalls and complications. It is hoped that it will be useful to medical students and to qualied clinicians.

the L3/L4 or L4/L5 interspace (Abrahams and Webb, 1975). Both these spaces are below the termination of the spinal cord at L1/L2 in the majority of adults (Ellis and Feldman, 1997).
Step 3: Inltrate

PROCEDURE
Step 1: Position

Inltrate local anesthetic only subcutaneously. Deeper structures are less pain sensitive and increased volume may distort the tissues and make the procedure more difcult.
Step 4: Insert Needle

For spinal anesthesia, the patient may be in the sitting or the lateral recumbent position depending on the level of anesthesia required. For diagnostic lumbar puncture, the left lateral recumbent position (for right-handed physicians) is preferred. Ill patients are generally unable to sit up and the sitting position increases the risk of headaches post-puncture, probably because the CSF pressure and ow is higher than in the lateral recumbent position (Norris et al., 1994; Cook, 2000). Furthermore, if any pressure measurements are done (Van Dellen and Bill, 1978; Adams et al., 1997), the patient must be in the lateral recumbent position. Failure to enter the lumbar subarachnoid space may be overcome by doing the puncture with the patient in the sitting position and then moving the patient to the recumbent position for uid collection and, if required, pressure measurement. In preterm infants, Weisman et al. (1983) suggested the sitting or lateral recumbent without knees-to-chest position, which results in less hypoxemia than the lateral recumbent knees-to-chest position. The neck is best maintained in the neutral position. In the lateral recumbent position, the patient should be positioned with the back exed as far as possible. Ask the patient to try to touch the exed knees with his/her chin (Ellis and Feldman, 1997). This is to overcome the lumbar lordosis that narrows the interspace between adjacent spinous processes and laminae. The coronal plane of the trunk should be at right angles to the oor with one hip exactly above the other. The needle is passed horizontally, i.e., parallel to the oor. This ensures that the needle stays in the midline. In the sitting position, the patient is positioned with the neck and back fully exed. Flexion facilitates the course of the needle by widening the gap between adjacent lumbar spinous processes.
Step 2: Determine Site of Insertion

A line joining the most superior part of both iliac crests (Tufers line) will intersect the midline at the L4 spinous process or L4/L5 interspace (Ievins, 1991; Ellis and Feldman, 1997). The needle is inserted at

A pencil-point needle (2225 gauge) is indicated for spinal anesthesia. For diagnostic collection of cerebrospinal uid (CSF), a larger gauge needle (18, 20, and 21 standard gauge needles; 22 gauge, 3.5-cm long needle for neonates; 20 gauge 5-cm long needle for children) should be used. Insert the needle at the superior aspect of the spinous process that lies inferior to the space to be entered. Aim for the umbilicus (15 cephalad) if the L4/5 interspace is used (American Association of Clinical Anatomists, 1999). When the L3/4 or L5/S1 interspace is used the needle should also be directed toward the umbilicus. The L4/5 or L5/S1 interspace should be used in children as the spinal cord ends at L3. The bevel should be in the sagittal plane. This diminishes injury to the dura mater by separating its longitudinal bers rather than cutting through them, and reduces leakage of CSF. Pass the needle through the supraspinous ligament that connects the tips of spinous processes and the interspinous ligaments between adjacent borders of spinous processes. Pass the needle through the ligamentum avum. There may be a sudden yielding sensation or give as it is penetrated, often referred to by clinicians as a pop. After entering the ligamentum avum, remove the stylet at each 2 mm interval of needle advancement to check for ow of CSF. A second pop represents penetration of the needle through the dura mater into the subarachnoid space. If bone is encountered, withdraw the needle partially to the subcutaneous tissue. Re-palpate the back to make sure the needle is in the midline and try again. Clear uid will appear if the subarachnoid space is penetrated. If not, it is worth rotating the needle through 90 as the needle opening may be obstructed by a nerve root (Van Dellen and Bill, 1978). Cerebrospinal uid drips directly into the specimen tube. Never aspirate with a syringe for a small amount of negative pressure can cause subdural hemorrhage or herniation. The amount of CSF collected for diagnostic purposes should be restricted to the smallest volume necessary. For children this is typically 0.5 ml/

546

Boon et al.

tube and not more than 3 ml in total. Various analyses can be done on CSF, including bacteriologic and virological cultures, differential cell counts and cytology, protein, glucose, and immunoglobulins. Different specimen tubes are available for different tests. The three standard investigations are glucose, biochemistry, and microscopy, culture, and sensitivity. With the patient in the lateral recumbent position CSF pressure may be measured with a manometer. Normal pressure in the adult ranges from 100 180 mm water (8 14 mm Hg), and in children between 30 60 mm water. Pressure 200 mm water with a relaxed patient and straightened legs reects an increased intracranial pressure (Adams et al., 1997).

ANATOMICAL PITFALLS
Course of the Lumbar Puncture Needle

The lumbar puncture needle pierces in order: skin, subcutaneous tissue, supraspinous ligament, interspinous ligament, ligamentum avum, epidural space containing the internal vertebral venous plexus, dura, arachnoid, and nally the subarachnoid space (Figs. 1,2).
End of Spinal Cord

Adults. The vertebral level at which the spinal cord terminates varies widely from T12 to the L3/L4 intervertebral disc (Reimann and Anson, 1944). The spinal cord extends to the L1/L2 disc in 51% of people and to the L2/L3 disc or below in 12% (Ievins, 1991). Reimann and Ansons (1944) study of 129 cadavers became the standard reference for the vertebral level of the termination of the spinal cord with the mean level lying opposite the L1/L2 disc. In a recent magnetic resonance imaging (MRI) study of 136 adults, MacDonald et al. (1999) showed that the median level of termination of the spinal cord for both males and females was the middle one-third of L1 vertebra, a higher level than usually stated. This ranged from the middle one-third of T11 to the middle one-third of L3. Only 25% of cords ended below the L1/L2 disc, compared to the 49% of cords below this level reported by Reimann and Anson (1944) in cadavers. Similarly, Broadbent et al. (2000) reported an incidence of 19% of spinal cord terminations below L1 in a series of 100 patients undergoing spinal MRI scans. In another MRI study, Saifuddin et al. (1998) studied the position of the conus medullaris in 504 adult patients without spinal deformity. They found the mean level of the conus medullaris to be the lower third of L1 with no signicant differences between the sexes or with increasing age. The position ranged

Fig. 1. Sagittal section of lumbar vertebrae illustrating the course of the lumbar puncture needle through skin (1), subcutaneous tissue (2), supraspinous ligament (3), interspinous ligament (5) between the spinous processes (4), ligamentum avum (6), dura mater (8), into the subarachnoid space and between the nerve roots of the cauda equina (7). Lumbar vertebral bodies (9), intervertebral disc (10), and lumbar puncture needle (11).

from the middle third of T12 to the upper third of L3. Only 6 of 504 cords terminated lower than the L2/L3 disc. One source of error in determining the end of the spinal cord, may be the presence of lumbosacral transition vertebrae in 8 15% of subjects. Saifuddin et al. (1998) suggested lumbar puncture be carried out at the L2/L3 level (higher than conventional practice) in cases of spinal stenosis (most severe at L3-4 and L4-5) or where posterior spinal fusion had been carried out, for in their study a signicant proportion of the cords terminated higher than L2. Puncture is usually carried out at either the L3/L4 or L4/L5 interspace (Abrahams and Webb, 1975). Reynolds (2001) strongly advised not to insert a spinal needle above L3, for in a study on injury to the conus medullaris after spinal anesthesia, seven patients (ve

Lumbar Puncture

547

Fig. 2. Horizontal section of lumbar vertebra illustrating the course of the lumbar puncture needle through skin (1), subcutaneous tissue (2), between the spinous processes (3) and laminae (4), ligamentum avum (5), epidural space (6), dura mater (7), into the subarachnoid space and between the nerve roots of the cauda equina (8). Lumbar vertebral body (9) and lumbar puncture needle (10).

having cesarean section) had neurological damage when the needle was introduced at the L2/L3 interspace. The injury (uid collection seen on MRI, intramedullary haemorrhage, and small infarcts) was associated with neurological symptoms (foot drop, numbness, sphincter disturbance, weakness) involving more than one nerve root. Five of seven cases went to litigation. During insertion, the spinal needle is directed somewhat superiorly that, as Reynolds (2001) has illustrated convincingly, may be the reason for injuring the conus medullaris in 4 20% of people when using the L2/L3 interspace. Wall et al. (1990) demonstrated that a web of arachnoid membrane holds the nerve roots together at the level of the conus medullaris with the nerve roots forming a peripheral rim around the cord. Infants. The spinal cord ends at L3. Needle placement should therefore be at L4/5 or L5/S1. The differences between adults and children are due to differential longitudinal growth of the spinal canal and the cord. Fitzgerald (1978) and Moore and Persaud (1993) stated that the spinal cord, early in fetal life, stretches through the whole vertebral canal with the nerve roots leaving the intervertebral foramina in a horizontal fashion. Growth of the vertebral column causes the lower part of the spinal cord to ascend relative to the vertebrae as the upper end is attached to the brain. At 6 months of fetal life, the lowest limit

of the spinal cord lies at the level of S1 (Moore and Persaud, 1993). At birth the conus medullaris is mostly found at the level of L3 (Fitzgerald, 1978). Up to 30 mm crown-rump-length (CRL), the spinal cord and vertebral column grow at the same rate, whereas from 30 mm CRL onwards a disproportion in growth rate occurs. Embryologically, growth involves the mesoderm more than the neurectoderm with resultant relative displacement of the conus medullaris cranially and the vertebral column caudally (Vettivel, 1991). Hawass et al. (1987) carried out translumbar myelograms on 146 spontaneously aborted fetuses and showed that between 2533 weeks, the cord terminates at the level of L3 or above, but that signicant variation was noted before 25 weeks.
Epidural Space

The epidural space lies between the inner surface of the spinal canal and the outer surface of the dural sac. Spinal nerve roots emerge from the spinal cord and cross the epidural space, to their respective intervertebral foraminae. Parkin and Harrison (1985) stated in a cadaver-based study that the epidural space is a region containing fat, areolar tissue and the internal vertebral venous plexus (Fig. 2), which is articially enlarged when the dura is separated from the vertebral canal by solutions such as local anesthetics. Newell (1999) stated that the epidural space is a true

548

Boon et al.

potential space, lined by a uniform delicate translucent lining layer investing the epidural fat, perhaps a mesothelial layer, similar to pleura or peritoneum, which can be opened easily without tissue damage. In large areas the dura contacts the bone and ligaments of the spinal canal wall (Hogan, 1998) but is not adherent to the canal wall. MRI studies indicate the space is lled with fat. The epidural contents are contained in a series of circumferentially discontinuous compartments separated by zones where the dura contacts the wall of the vertebral canal (Hogan, 1998). Local anesthetic in the epidural space must spread through volume displacement. Hogan (1991, 1998) showed by means of cryomicrotome studies that the epidural fat increases proportionately at descending lumbar spine levels. At higher spinal levels the epidural space is empty in large areas where the dura contacts the spinal canal wall (Hogan, 1998). There is a denite fat-lled space, especially in the lower lumbar spine, which is on average 4.55.5 cm from the skin (Hogan, 1998). The distance a needle advances after entering the epidural space through the ligamentum avum before penetrating the dura, is about 7 mm (range 2 mm to 2.5 cm). This distance depends on where exactly the epidural space is traversed. Away from the midline the distance may be very small as the epidural space diminishes to the lateral side (Hogan, 1998).
Vertebral Venous Plexuses

cord, where the spinal needle can be safely introduced (American Association of Clinical Anatomists, 1999). The cauda equina lies within the cistern and is composed of the dorsal (mostly afferent axons) and ventral roots (efferent axons) of spinal nerves L2 to Co1, gathered around the lum terminale. Because the cord is shorter than the vertebral column, caudal spinal roots run varying distances to enter their corresponding intervertebral foramina. The ventral spinal roots contain axons of neurons in the ventral and lateral spinal grey columns and emerge from the cord as bundles of rootlets, whereas each dorsal spinal root has a spinal ganglion just proximal to its junction with a ventral root in the intervertebral foramen and attaches to the cord as a series of rootlets along the posterolateral sulcus. These roots cross the subarachnoid space and pierce the dura separately to unite in or close to the intervertebral foramen to form a mixed spinal nerve. The sacral spinal ganglia are found inside the vertebral canal (Williams et al., 1989). Beyond the intervertebral foramen, a ventral and dorsal ramus are found. The lumbar dorsal rami pass back dividing into medial and lateral branches and supply the erector spinae muscles and overlying skin. The lumbar and sacral ventral rami form the lumbar and sacral plexuses. The dorsal and ventral rami of S1S4 spinal nerves pass through the posterior and anterior sacral foramina, respectively.
Ligamentum Flavum

The internal vertebral venous plexus is located in the epidural space (Domisse, 1975; Parkin and Harrison, 1985; Brockstein et al., 1994) and may be involved in a bloody or traumatic tap (Mehl, 1986). The plexus consists of four interconnecting longitudinal vessels, two anterior and two posterior. It seems that this distribution is variable for the posterior internal vertebral veins may be rudimentary (Parkin and Harrison, 1985). The external vertebral plexus, in contrast, lies peripheral to the vertebrae. The external vertebral venous plexus (Williams et al., 1989) includes the anterior and posterior external vertebral plexuses situated anterior to the vertebral bodies and in relation to the laminae, spinous processes, transverse processes and articular processes respectively. These veins communicate with the segmental veins of the neck, the intercostal, azygos and lumbar veins. With the veins of bones of the vertebral column, the internal and external vertebral plexuses form Batsons plexus (Domisse, 1975).
Lumbar Cistern

The ligamentum avum is a strong yellow elastic ligament. It can be up to 1-cm thick in the lumbar region and spans the interlaminar space between adjacent vertebrae. Fibers are stretched in the exed position and can be more easily penetrated at lumbar puncture. If the needle is exactly in the midline, it may pass through the gap between the right and left ligamenta ava (Hogan, 1998). Practical experience and observations by CT (Hogan, 1998) show that the needle is usually not perfectly in the midline, and therefore passes through either the left or right ligamentum avum to a site in the lateral epidural space, before piercing the dura. In a exed spine, the extent of the ligamentum avum exposed is much greater than in the extended spine (Ellis and Feldman, 1997). In older patients the ligament may provide signicant resistance because it is often calcied. This resistance is felt at a depth of 4 7 cm.
Meninges

The lumbar cistern represents the expansive portion of the subarachnoid space inferior to the spinal

Dura mater. The dura mater lines the spinal canal to the level of S2. MacDonald et al. (1999) found in their MRI study that the median level of termination of the dural sac was the middle one-third of S2 (the

Lumbar Puncture

549

upper one-third in males and the middle one-third in females). The range extended from the upper border of S1 to the upper border of S4. Arachnoid mater. The arachnoid mater lines the dural sac to the level of the middle one-third of S2 (MacDonald et al., 1999). The subarachnoid space thus continues down to S2 level, the level of the posterior superior iliac spines or Dimples of Venus, seen especially distinctly in females. Pia mater. The pia mater leaves the spinal cord at the conus medullaris to form the lum terminale that traverses the subarachnoid space and terminates on the periosteum of the coccyx, after penetrating the dura and arachnoid at the level of S2.
Spinal Canal

Porter et al. (1980) showed in a diagnostic ultrasound-based study involving more than 700 people (from infancy to 65 years of age) that the lumbar spinal canal (vertebral canal) is fairly wide in children, reaches a maximum diameter in the late teens, and reduces slightly in later adulthood. On the level of L5, it was found that at 4 years of age the midsagittal diameter of the spinal canal was even larger than in the adult. The spinal canal thus seems of adequate dimension in both adults and children to allow a lumbar puncture, even if the needle is not exactly in the midline.

Infants have a total of 40 60 ml of CSF, young children about 60 100 ml, and adults 120 150 ml. Although less total CSF, children below 15 kg have approximately twice the volume of CSF per kilogram body weight (4 ml/kg) than adults (2 ml/kg). In an adult the removal of 10 ml of CSF is replaced in 30 min at the normal rate of CSF production of 0.3 ml/min. A child produces CSF at a rate of about 20 ml/hr (Dalens, 1990). Associated risk factors for headache are: female, lower body mass index, young age, large needle size, bevelled needle type compared to pencil-point needle of same size, bevel of needle cutting longitudinal dural bers (Brocker, 1958), and multiple punctures. The pencil-point needle separates, rather than cuts, through the dural bers, giving a signicantly lower incidence of post-spinal headaches (Lynch et al., 1991). The pencil-point needle (2225 gauge) is indicated for anesthesia, but not for diagnostic use, as it does not allow free ow of CSF with resultant difculty in obtaining sufcient CSF. The smallest possible atraumatic needle with a stylet should be used for spinal anesthesia and multiple punctures should be avoided. For diagnostic use, a larger gauge needle (18, 20, and 21 standard gauge needles with a short needle for children) should be used for collection of CSF (Van Dellen and Bill, 1978).
Traumatic Tap

COMPLICATIONS
Headache

Headache is the most common complication of dural puncture (Olsen, 1987), occurring in up to 36.5% of spinal taps (Kuntz et al., 1992). Usually it starts 48 hr after the procedure (Raskin, 1990) and may last 12 days or even 2 weeks. Sometimes it is accompanied by nausea, vomiting, vertigo, tinnitus, diminished hearing, and blurred vision. The headache is due to leakage of CSF through the dural puncture site into the epidural and paravertebral spaces faster than its production rate (Tourtellote et al., 1972). The incidence of headache after lumbar puncture is directly related to the size of the needle. Headache is commoner with a large needle because of a larger leakage of CSF through the rent made in the dura. Leakage leads to low CSF pressure, absolute reduction of CSF volume below the cisterna magna with resultant downward movement of the brain and traction on pain-sensitive structures in the cranial cavity, especially the basal dura (Raskin, 1990). Reduction of the brains supportive cushion and may also explain the headache. The CSF collected for diagnostic purposes should be restricted to the smallest volume necessary.

A traumatic tap (macroscopic blood in CSF) usually occurs due to the needle being placed too far laterally or advanced too far anteriorly (Mehl, 1986). The internal vertebral venous plexus in the epidural space may be involved in a bloody tap. A traumatic tap should be distinguished from a subarachnoid hemorrhage. Fluid generally clears after the rst and second tubes in a traumatic tap. The presence of a clot in one of the tubes strongly favours a traumatic tap. Clotting does not occur in a subarachnoid hemorrhage due to debrinated blood being present in the CSF. Entry to the internal vertebral venous plexus, poses a slight risk of neurological symptoms, as clots may compress the spinal nerve roots or nerves (American Association of Clinical Anatomists, 1999).
Dry Tap

A dry tap (American Association of Clinical Anatomists, 1999) is usually due to incorrect positioning of the patient and consequent misdirection of the needle, often on to bony structures. This is often due to inappropriate, usually too superior, direction of the needle, with obstruction by the lamina or spinous process of the superior or inferior vertebra. If the needle is directed too laterally, an inferior or superior

550

Boon et al.

articular process may be struck and may also injure the spinal nerve in the intervertebral foramen. The back may not be fully exed, with the gaps between the lumbar spinous processes not widened (Ellis and Feldman, 1997).
Difculty in Finding Landmarks

If only one iliac crest is used to locate L4, 30% of needles are misplaced at L2/L3. This high misplacement gure is diminished to 4% if Tufers line is used to determine L4, as determined by a cadaveric study (Ievins, 1991). Some authors, however, consider Tufers line unreliable in determining the lumbar interspaces. Anesthetists, for example, often select a space of insertion one or two segments higher than they estimated using Tufers line (Reynolds, 2000). The level indicated by Tufers line may vary from L3/L4 L5/S1. In a study (Broadbent et al., 2000) to determine the success of identication of lumbar interspaces by using Tufers line, correct identication was only seen in 29% of cases and the space identied was one space higher than expected in 51% of cases. Broadbent et al. (2000) concluded that one should not assume identication of the correct interspace when using Tufers line, although there is no better alternative (Reynolds, 2000). Broadbent et al. (2000), therefore, recommended insertion of the needle one space lower. Finding the landmarks may be difcult in obese patients. Broadbent et al. (2000) showed that the accurate identication of the correct lumbar interspace was signicantly impaired by obesity. In young patients the vertebral anatomy is well dened, consistent and amenable to easy localization of the epidural and subarachnoid space (Kopacz, 1996). In a study by Boon et al. (2003), measurements on antero-posterior lumbar spine radiographs indicated that the interlaminar area reduced in height and width with age. This may make lumbar puncture more difcult. Many conditions such as osteoarthritis, ankylosing spondylitis, kyphoscoliosis, previous spinal surgery, and degenerative disc disease with collapse of the intervertebral space may cause problems in needle access (Ellis and Feldman, 1997). Cousins and Bromage (1988) suggested a paramedian approach if needle access is difcult in the presence of such conditions.
Pain Referred to Lower Limb

procedure started again, although this may already have caused nerve damage (Reynolds, 2001). Samsoon and Grewal (2001) considered the spinal insertion technique important in avoiding nerve trauma. The risk of nerve injury increases if uncontrolled pressure is applied and the dura overshot. Possibly the best technique for avoiding uncontrolled plunging through the dura is that described by Bromage et al. (1993). It protects the underlying nervous structures by signaling the precise moment to stop the spinal needle. The needle is gripped with a gloved left hand (if the operator is right-handed) between the thumb and the entire st, with the metacarpal heads against the patients back. Highly controlled pushing of the needle is now possible with a gloved right hand. In this way the left hand, while the right hand pushes the needle in a controlled way, guides and stabilizes the needle in the correct direction. The left hand can halt the needle in a fraction of a millimetre at the very moment CSF is observed. Advancement of the needle until a dural pop is experienced should be discouraged and a periodic stop and check method for CSF whether a dural pop has been felt or not seems to be safer. With atraumatic pencil point needles, no pops may be experienced when passing through the different tissue layers. The plunger of a syringe should not be withdrawn if it is attached to the needle or when injecting anesthetic solution. The negative pressure may pull a spinal nerve root against the needle tip and produce paresthesia, pain, or injury (American Association of Clinical Anatomists, 1999). Disc herniation has been reported due to the needle passing beyond the subarachnoidal space (needle advanced too far) into the annulus brosus, with resultant herniation of the nucleus pulposus (Van Dellen and Bill, 1978). Multiple attempts may lead to paraspinal muscle spasm, presenting as backache.
Herniation Syndromes

If the patient complains of a shooting pain down a leg during the procedure, a nerve root may have been hit. The needle was probably angled away from the midline toward the side of the pain. If this happens, the needle should be withdrawn completely and the

Large pressure gradients occur between the cranial and lumbar compartments in supratentorial mass lesions. When the pressure in the spinal compartment is lowered by a lumbar puncture, transtentorial and foramen magnum herniation may occur. Duffy (1969) reported on 30 patients with post-lumbar puncture herniation syndromes of whom half lost consciousness immediately after the lumbar puncture. Pre-existing tentorial herniation is a contraindication to a lumbar puncture. This can be diagnosed by ndings of pupillary and oculomotor xation, quadriparesis, postural and respiratory changes.

Lumbar Puncture

551

Anatomy of Herniation

The decrease of pressure in the spinal canal precipitated by removal of CSF, will cause downward movement of the transtentorial and cerebellar tonsillar structures (Gower et al., 1987). Anyone is theoretically at risk of herniation during a lumbar puncture but the risk is greater with a pressure gradient between the intracerebral and intraspinal CSF (Gower et al., 1987). The following phenomena may occur in herniation: 1. Uncal herniation; expanding lesions in the middle cranial fossa shift the uncus, the medial part of the temporal lobe, medially over the notch of the tentorium and compress the midbrain, cerebral peduncle and third cranial nerve. The oculomotor nerve and posterior cerebral artery may be caught between the swollen uncus and the free edge of the tentorium. The earliest sign is a slightly dilated pupil unilaterally. Transtentorial herniation displaces the brain stem inferiorly, stretching the medial perforating branches of the basilar artery, which produces brainstem ischemia. 2. Transient unilateral or bilateral sixth nerve palsy may be caused by stretching of the abducens nerve as it crosses the superior border of the petrous temporal bone. 3. The anterior cerebral artery may be compressed against the falx and increase ischemia and edema. 4. Compression of the posterior cerebral artery at the tentorial notch can produce occipital lobe infarction. 5. Posterior midline displacement compresses the great cerebral vein. 6. Kinking of the cerebral aqueduct may interfere with CSF drainage. 7. Tonsillar herniation; the cerebellar tonsils compress the medulla at the foramen magnum. A CT scan should rst be carried out if there is any indication or history suggesting raised intracranial pressure (Gower et al., 1987). A careful neurological examination should always precede lumbar puncture.
Intraspinal Epidermoid Tumor

removed until the needle tip has passed the skin and unstyletted lumbar puncture needles should be avoided.
Retroperitoneal Abscess

One case of a retroperitoneal abscess has been reported, produced by dural laceration in a patient with meningitis (Levine et al., 1982). The patient developed a psoas abscess due to leakage of infected CSF into the retroperitoneal space.
Hypoxia and Ventilation-Perfusion Mismatches in Children

Gleason et al. (1983) demonstrated that lumbar puncture on an ill, premature baby using the traditional lateral recumbent position with neck exion, may result in signicant respiratory ventilation-perfusion imbalance leading to hypoxemia. Beware of too long a period of exion of the neck while positioning the child, for this may produce dangerous airway obstruction. Positioning is best accomplished by an assistant holding the child and maintaining the spine in a exed position by holding the child behind the shoulders and knees, but the neck in the neutral position in order not to compromise the airway. Weisman et al. (1983) compared the grade of hypoxemia in three different positions. The sitting and lateral position without knees-to-chest position gave less hypoxemia than the lateral knees-to-chest position.

CONCLUSION
In the absence of contraindications and with a thorough knowledge of the anatomy and technique as well as adequate prior skills practice, a lumbar puncture can be carried out without complications. Lumbar puncture remains a common procedure in various specialties, for diagnostic and therapeutic purposes, such as epidural and spinal anesthesia. The authors hope this review of the technique and its pitfalls and complications will be helpful to young clinicians. Lumbar puncture is an essential procedure in the armamentarium of clinicians, and usually is carried out safely and successfully.

This complication usually arises due to the failure to use a stylet (McDonald and Klump, 1986) and constitutes a mass of desquamated keratinized cells, arising from viable epithelial cells introduced into the spinal canal by the spinal needle. Skin tissue can easily be detached by a hollow needle and implanted into the subarachnoid space. The stylet should not be

REFERENCES
Abrahams PH, Webb P. 1975. Clinical anatomy of practical procedures. London: Pitman Medical. p 1315. Adams RD, Maurice V, Ropper AH. 1997. Principles of neurology. 6th Ed. New York: McGraw-Hill. p 1314. American Association of Clinical Anatomists, Educational Affairs Committee. 1999. The clinical anatomy of several invasive procedures. Clin Anat 12:4354.

552

Boon et al. Gower DJ, Baker AJ, Bell WO, Ball MR. 1987. Contraindications to lumbar puncture as dened by computed cranial tomography. J Neurol Neurosurg Psychiatry 50:1071 1074. Graney DO. 1996. Letter to the editor. Clin Anat 9:61. Hawass ND, el-Badawi MG, Fatani JA, Meshari AA, Abbas FS, Edrees YB, Jabbar FA, Banna M. 1987. Myelographic study of the spinal cord ascent during fetal development. Am J Neuroradiol 8:691 695. Hogan QH. 1991. Lumbar epidural anatomy: a new look by cryomicrotome section. Anesthesiology 75:767775. Hogan QH. 1998. Epidural anatomy: new observations. Can J Anaesth 45:R40 48. Ievins FA. 1991. Accuracy of placement of extradural needles in the L3/4 interspace: comparison of two methods of identifying L4. Br J Anaesth 66:381382. Kneebone RL. 1999. Twelve tips on teaching basic surgical skills using simulation and multimedia. Med Teach 21:571 575. Kopacz DJ, Neal JM, Pollock JE. 1996. The regional anesthesia learning curve: what is the minimum number of epidural and spinal blocks to reach consistency? Reg Anesth 21:182 190. Kuntz KM, Kokmen E, Stevens JC, Miller P, Offord KP, Ho MM. 1992. Post-lumbar puncture headaches: experience in 501 consecutive procedures. Neurology 42:1884 1887. Levine JF, Hiesiger EM, Whelan MA, Pollock AA, Simberkoff MS, Rahal JJ. 1982. Pneumococcal meningitis associated with retroperitoneal abscess. JAMA 248:2308 2309. Lynch J, Krings-Ernst I, Strick K, Topalidis K, Schaaf H, Fiebig M. 1991. Use of a 25 gauge Whitacre needle to reduce the incidence of post-dural puncture headache. Br J Anaesth 67:6923. MacDonald A, Chatrath P, Spector T, Ellis H. 1999. Level of termination of the spinal cord and the dural sac: a magnetic resonance study. Clin Anat 12:149 152. McDonald JV, Klump TE. 1986. Intraspinal epidermoid tumors caused by lumbar puncture. Arch Neurol 43:936 939. McMinn RMH, Hutchings RT, Logan BM. 1984. A colour atlas of applied anatomy. London: Wolfe Medical Publications. Mehl AL. 1986. Interpretation of traumatic lumbar puncture. Clin Pediatr 25:523526. Moore KL, Persaud TVN. 1993. The developing human: clinically oriented embryology. Philadelphia: W.B. Saunders Company. p 392394. Newell RL. 1999. The spinal epidural space. Clin Anat 12: 3759. Norris MC, Grieco WM, Borkowski M, Leighton BL, Arkoosh VA, Huffnagle HJ, Huffnagle S. 1994. Complications of labor analgesia: epidural versus combined spinal epidural techniques. Anesth Analg 79:529 537. Olsen KS. 1987. Epidural blood patch in the treatment of post-lumbar puncture headache. Pain 30:293301. Parkin IG, Harrison GR. 1985. The topographical anatomy of the lumbar epidural space. J Anat 141:211217. Porter RW, Hibbert C, Wellman P. 1980. Backache and the lumbar spinal canal. Spine 5:99 105. Raskin NH. 1990. Lumbar puncture headache: a review. Headache 30:197200. Reynolds F. 2000. Logic in the safe practice of spinal anesthesia. Anesthesia 55:10451046.

Association of American Medical Colleges. 1998. Learning objectives for medical student education. Guidelines for medical schools. Washington: AAMC. p 5 6. Beahrs OH, Chase RA, Ger R. 1986. Gross anatomy in medical education. Am Surg 52:227232. Boon JM, Prinsloo E, Raath RP. 2003. A paramedian approach for epidural block: an anatomic and radiologic description. Reg Anesth Pain Med 28:221227. Boon JM, Meiring JH, Richards PA, Jacobs CJ. 2001. Evaluation of clinical relevance of problem-oriented teaching in undergraduate anatomy at the University of Pretoria. Surg Radiol Anat 23:1 4. Broadbent CR, Maxwell WB, Ferrie R, Wilson DJ, GawneCain M, Russell R. 2000. Ability of anaesthetists to identify a marked lumbar interspace. Anaesthesia 55:11221126. Brocker RJ. 1958. A technique to avoid post spinal-tap headache. JAMA 168:261263. Brockstein B, Johns L, Gewertz BL. 1994. Blood supply to the spinal cord: anatomic and physiologic correlations. Ann Vasc Surg 8:394 399. Bromage PR, Van Zundert A, Van Steenberge A, Corke BC, Fagraeus L. 1993. A-loss-of-resistance-to-negative-pressure test for subarachnoid puncture with narrow-gauge needles. Reg Anesth 18:155161. Cook TM. 2000. Combined spinal-epidural techniques. Anaesthesia 55:42 64. Cottam WW. 1999. Adequacy of medical school gross anatomy education as perceived by certain postgraduate residency programs and anatomy course directors. Clin Anat 12:55 65. Cousins MJ, Bromage PR. 1988. Epidural neural blockade. In: Cousins MJ, Bridenbaugh PO. Neural blockade in clinical anesthesia and management of pain. 2nd Ed. Philadelphia: JB Lippincott Company. p 324 325. Crisp AH. 1989. The relevance of anatomy and morbid anatomy for medical practice and hence postgraduate and continuing medical education of doctors. Postgrad Med J 65: 221223. Dalens BJ. 1990. Pediatric regional anesthesia. Boca Raton, Florida: CRC Press. p 354 425. Domisse GF. 1975. The arteries and veins of the human spinal cord from birth. Edinburgh: Churchill Livingstone. p 8196. Duffy GP. 1969. Lumbar puncture in the presence of raised intracranial pressure. BMJ 1:407. Ellis H, Feldman S. 1997. Anatomy for anesthetists. 7th Ed. Oxford: Blackwell Science. p 108 112. Fitzgerald MJT. 1978. Human embryology: a regional approach. Hagerstown: Harper and Row Publishers. p 68 69. General Medical Council. 1993. Tomorrows doctors: recommendations on undergraduate medical education. London: GMC. p 128. General Medical Council. 2002. Good medical practice: guidance from the General Medical Council. London: GMC. p 3. Ger R, Evans JT. 1993 Tracheostomy: an anatomico-clinical review. Clin Anat 6:337341. Ger R. 1996. Basic surgical training 4: American and British scenes compared. Clin Anat 9:175182. Gleason CA, Martin RJ, Anderson JV, Waldemar AC, Sanniti KJ, Fanaroff AA. 1983. Optimal position for a spinal tap in preterm infants. Paediatrics 71:3135.

Lumbar Puncture Reynolds F. 2001. Damage to the conus medullaris following spinal anesthesia. Anaesthesia 56:235247. Reimann AF, Anson BJ. 1944. Vertebral level of termination of the spinal cord with report of a case of sacral cord. Anat Rec 88:127138. Saifuddin A, Burnett SJ, White J. 1998. The variation of the position of the conus medullaris in an adult population. A magnetic resonance imaging study. Spine 23:14521456. Samsoon G, Grewal K. 2001. Paresthesia during spinal needle placement [letter]. Anaesthesia 56:1024. Tourtellote WW, Henderson WG, Tucker RP, Gilland O, Walker JE, Kokman E. 1972. A randomized double-blind clinical trial comparing 22 versus the 26 gauge needle in the production of the post-lumbar puncture syndrome in normal individuals. Headache 12:7378.

553

Van Dellen JR, Bill PLA. 1978. Lumbar puncturean innocuous diagnostic procedure? S Afr Med J 53:666 668. Vettivel S. 1991. Vertebral level of the termination of the spinal cord in human fetuses. J Anat 179:149 161. Wall EJ, Cohen MS, Abitol JJ, Garn SR. 1990. Organization of intrathecal nerve roots at the level of the conus medullaris. J Bone Joint Surg Am 72:14951499. Weisman LE, Merenstein GB, Steenbarger JR. 1983. The effect of lumbar puncture position in sick neonates. Am J Dis Child 137:10771079. Wigton RS. 1992. Training internists in procedural skills. Ann Int Med 116:10911093. Williams PL, Warwick R, Dyson M, Bannister LH. 1989. Grays anatomy. 37th Ed. Edinburgh: Churchill Livingstone. p 11231143.