INTRODUCTION — Quincke performed the first lumbar puncture (LP) in 1891 to relieve increased intracranial pressure in children with tuberculous meningitis [1]. This technique subsequently became important in the diagnosis of a variety of infectious and noninfectious neurologic conditions. However, its relative value for diagnosing central nervous system (CNS) conditions other than infection has diminished as new testing methods, especially imaging techniques, have appeared.
The indications, contraindications, and complications of LP will be reviewed here. With a few exceptions, the results of an LP are rarely diagnostic. As a result, analysis of the cerebrospinal fluid (CSF) has to be correlated with the history, physical findings, and other laboratory tests in order to be maximally useful. (See "Cerebrospinal fluid: Physiology and utility of an examination in disease states".)
TECHNIQUE — An LP can be performed with the patient in the lateral recumbent position or sitting upright. The lateral recumbent position is preferred because it allows accurate measurement of the opening pressure.
The choice of needle type (cutting versus atraumatic) and bore size can influence the risk of a post-LP headache, but also may increase the technical difficulty of the procedure. This is discussed in detail separately. (See "Post-lumbar puncture headache", section on 'Prevention'.)
The correct level of entry of the spinal needle is most easily determined with the patient sitting upright or standing. The highest points of the iliac crests should be identified visually and confirmed by palpation; a direct line joining these is a guide to the fourth lumbar vertebral body. The spinous processes of L3, L4, and L5, and the interspaces between can usually be directly identified by palpation. The spinal needle can be safely inserted into the subarachnoid space at the L3/4 or L4/5 interspace, since this is well below the termination of the spinal cord.
Correct patient positioning is an important determinant of success in obtaining CSF. The patient is instructed to remain in the fetal position with the neck, back, and limbs held in flexion. The lower lumbar spine should be flexed with the back perfectly perpendicular to the edge of a bed or examining table. The hips and legs should be parallel to each other and perpendicular to the table. Pillows placed under the head and between the knees may improve patient comfort.
The overlying skin should be cleaned with alcohol and a disinfectant such as povidone-iodine or chlorhexidine (0.5 percent in alcohol 70 percent); the antiseptic should be allowed to dry before the procedure is begun. Many product inserts of chlorhexidine-containing solutions warn against use of chlorhexidine prior to lumbar puncture because of a concern that it can cause arachnoiditis. The evidence that it does so is very limited, and many experts believe that chlorhexidine has an advantage over povidone-iodine because of its onset, efficacy, and potency [2-6]. Due to specific labeling prohibiting use, a formal institutional policy to support such use may be indicated. After the skin is cleaned and allowed to dry, a sterile drape with an opening over the lumbar spine is placed on the patient. Local anesthesia (eg, lidocaine) is infiltrated into the previously identified lumbar intervertebral space and a 20 or 22 gauge spinal needle containing a stylet is inserted into the lumbar intervertebral space.
The spinal needle may be advanced slowly, angling slightly toward the head, as if aiming towards the umbilicus. The flat surface of the bevel of the needle should be positioned to face the patient's flanks to allow the needle to spread rather than cut the dural sac (the fibers of which run parallel to the spinal axis). Many physicians choose to advance the needle incrementally, removing the stylet periodically to check for CSF flow, then reinserting the stylet until the subarachnoid space is entered [7]. However, others report a higher rate of successful LP when the stylet is removed, just after the skin is punctured and before it is passed into the subarachnoid space in order to better observe the flow of CSF upon entry of the subarachnoid space [8,9].
Once CSF appears and begins to flow through the needle, the patient should be instructed to slowly straighten or extend the legs to allow free flow of CSF within the subarachnoid space. A manometer should then be placed over the hub of the needle and the opening pressure should be measured (figure 1). Fluid is then serially collected in sterile plastic tubes. A total of 8 to 15 mL of CSF is typically removed during routine LP. However, when special studies are required, such as cytology or cultures for organisms that grow less readily (eg, fungi or mycobacteria), 40 mL of fluid can safely be removed. Aspiration of CSF should not be attempted as it may increase the risk of bleeding [7]. The stylet should be replaced before the spinal needle is removed.
No trials have shown that bed rest following LP significantly decreases the risk of post LP headache compared with immediate mobilization [10,11]. (See "Post-lumbar puncture headache", section on 'Bedrest'.)
The Queckenstedt maneuver can be used to demonstrate that there is free flow of fluid from the ventricles to the lumbar space. This maneuver is performed by measuring the CSF pressure and then observing the change in pressure after manual compression of both jugular veins. However, this test is rarely useful in modern practice, since newer techniques such as magnetic resonance imaging (MRI) and computed tomography (CT) readily identify most obstructing spinal or basilar lesions.
INDICATIONS — LP is essential or extremely useful in the diagnosis of bacterial, fungal, mycobacterial, and viral CNS infections and, in certain settings, for help in the diagnosis of subarachnoid hemorrhage, CNS malignancies, demyelinating diseases, and Guillain-Barré syndrome.
Urgent — The number of definite indications for LP has decreased with the advent of better neuroimaging procedures including CT scans and MRI, but urgent LP is still indicated to diagnose two serious conditions [12,13]:
The most common use of the LP is to diagnose or exclude meningitis in patients presenting with some combination of fever, altered mental status, headache, or meningeal signs. Examination of the CSF has a high sensitivity and specificity for determining the presence of bacterial and fungal meningitis.
The findings on CSF analysis also may help distinguish bacterial meningitis from viral infections of the central nervous system. However, there is often substantial overlap. (See "Viral encephalitis in adults", section on 'Cerebrospinal fluid findings'.)
Nonurgent — A nonurgent LP is indicated in the diagnosis of the following conditions. The findings are discussed in the appropriate topic reviews:
Conditions in which LP is rarely diagnostic but still useful include:
LP is also required as a therapeutic or diagnostic maneuver in the following situations [12,13,16,17]:
CONTRAINDICATIONS — Although there are no absolute contraindications to performing the procedure, caution should be used in patients with:
These are discussed in detail in relation to the complications with which they are associated. (See 'Complications' below.)
COMPLICATIONS — LP is a relatively safe procedure, but minor and major complications can occur even when standard infection control measures and good technique are used. These complications include:
The risk of complications was studied in a cohort of 376 patients who underwent LP for evaluation of acute cerebrovascular disease [18]. The following frequency of complications was noted: backache (25 percent), headache (22 percent), headache and backache (12 percent), severe radicular pain (15 percent), and paraparesis (1.5 percent). Severe pain or paraparesis occurred in 6.7 percent of patients receiving anticoagulants following the procedure and in none of the 34 patients who did not receive anticoagulants.
Post LP headache — Headache, which occurs in 10 to 30 percent of patients, is one of the most common complications following LP. Post-LP headache is caused by leakage of CSF from the dura and traction on pain-sensitive structures. Patients characteristically present with frontal or occipital headache within 24 to 48 hours of the procedure, which is exacerbated in an upright position and improved in the supine position. Associated symptoms may include nausea, vomiting, dizziness, tinnitus, and visual changes.
This risk factors, prevention, and treatment of post-LP headache are discussed separately. (See "Post-lumbar puncture headache".)
Infection
Meningitis — Meningitis is an uncommon complication of LP. In a review of 179 cases of post-LP meningitis reported in the medical literature between 1952 and 2005, half of all cases occurred after spinal anesthesia; only 9 percent occurred after diagnostic LP. The most commonly isolated causative organisms were streptococcus salivarius (30 percent), streptococcus viridans (29 percent), alpha-hemolytic strep (11 percent), staphylococcus aureus (9 percent), and pseudomonas aeruginosa (8 percent) [19].
While some cases of post-LP meningitis due to staphylococci, pseudomonas, and other gram-negative bacilli have been attributed to contaminated instruments or solutions or poor technique [20], other studies have suggested that post-LP meningitis could arise from aerosolized oropharyngeal secretions from personnel present during the procedure especially since many of the causative organisms are found in the mouth and upper airway [19,21-23].
Based upon these observations, some authors have recommended the routine use of face masks during LP and neuroradiologic imaging procedures involving LP [24-26]. Others have questioned the practicality and necessity of the use of face masks since there is no proof that face masks prevent such infections [22,27]. In 2005 the Healthcare Infection Control Advisory Committee recommended that face masks be used by individuals who place a catheter or inject material into the spinal canal, and in 2007 the CDC endorsed this recommendation [28]. These guidelines do not require use of a face mask for routine diagnostic LP. However, we believe a face mask can reasonably be used for diagnostic procedures especially if the procedure is likely to be prolonged or difficult, or if the person carrying out the procedure has an upper respiratory tract infection. (See "General principles of infection control".)
Because meningitis can be caused in animals by performing an LP after first inducing a bacteremia [29,30], several authors have speculated that an LP in a bacteremic patient without preexisting meningitis might actually cause meningitis [31]. However, this phenomenon is rare, if it occurs at all. In a retrospective study of 1089 bacteremic infants, the incidence of spontaneous meningitis in children who underwent LP and subsequently developed meningitis was not statistically different from those who did not undergo LP (2.1 versus 0.8 percent) [32]. We agree with other authors that theoretical concerns about inducing meningitis in patients with bacteremia should not be used as the basis to forego LP if meningitis is suspected [27].
An LP through a spinal epidural abscess can result in the spread of bacteria into the subarachnoid space. Because an LP is not needed for diagnosis, the procedure should NOT be performed in most patients with suspected epidural abscess in the lumbar region [33]. (See "Epidural abscess".)
Other infections — There are rare anecdotal case reports of discitis and vertebral osteomyelitis following LP. Most cases were due to normal skin flora such as Propionibacterium species and coagulase negative staphylococci [34-36]. These complications presumably result from direct inoculation of bacteria into the vertebral bone.
Bleeding — The CSF is normally acellular, although up to five red blood cells (RBCs) are considered normal after LP due to incidental trauma to a capillary or venule. A higher number of RBCs is seen in some patients in whom calculation of the white blood cell (WBC) to RBC ratio and the presence or absence of xanthochromia may differentiate LP-induced from true CNS bleeding. (See "Cerebrospinal fluid: Physiology and utility of an examination in disease states", section on 'Cells'.)
Serious bleeding that results in spinal cord compromise has occurred in up to 1 to 2 percent of patients in some case series [18,37]. The diagnosis of spinal hematoma is complicated by the concealed nature of the bleeding; thus, a high index of suspicion must be maintained. Patients who have persistent back pain or neurologic findings (eg, weakness, decreased sensation, or incontinence) after undergoing LP require urgent evaluation (usually spinal magnetic resonance imaging (MRI)) for possible spinal hematoma [38]. The appropriate treatment for patients with significant or progressing neurologic deficits is prompt surgical intervention, usually a laminectomy, and evacuation of the blood. Timely decompression of the hematoma is essential to avoid permanent loss of neurologic function [39,40]. Patients with mild symptoms or early signs of recovery may be managed conservatively with vigilant monitoring; dexamethasone may be administered to mitigate against neurologic injury [41,42]. (See "Disorders affecting the spinal cord", section on 'Spinal epidural hematoma'.)
Patients who have thrombocytopenia or other bleeding disorders or in those who received anticoagulant therapy prior to or immediately after undergoing LP have an increased risk of bleeding. In a series cited above, spinal hematoma developed in 7 of 342 patients (2 percent) who received anticoagulant therapy after undergoing LP; five of these patients developed paraparesis [18]. In one literature review, 47 percent of 21 published cases of spinal hematoma following lumbar puncture occurred in patients with a coagulopathy [41]. Thus, a high index of suspicion of spinal hematoma should be maintained in all patients who develop neurologic symptoms after a lumbar puncture, including those with no known coagulopathy. In rare cases, intraventricular, intracerebral, and subarachnoid hemorrhage have also been reported as complications of lumbar puncture [42,43].
We are unaware of any study that examined the risk of bleeding following LP based upon the degree of thrombocytopenia or clotting study abnormalities. Thus, at present the only guidepost is "clinical judgment." We generally advise NOT performing an LP in patients with coagulation defects who are actively bleeding, have severe thrombocytopenia (eg, platelet counts <50,000 to 80,000/µL), or an INR >1.4, without correcting the underlying abnormalities [44,45]. When an LP is considered urgent and essential in a patient with an abnormal INR or platelet count in whom the cause is not obvious, consultation with a hematologist may provide the best advice for safe correction of the coagulopathy prior to performing the LP. (See "Evaluation and management of thrombocytopenia by primary care physicians", section on 'Invasive procedures'.)
For elective procedures in a patient receiving systemic anticoagulation, observational studies and expert opinion have suggested stopping unfractionated intravenous heparin drips two to four hours, stopping low-molecular-weight heparin 12 to 24 hours, and stopping warfarin five to seven days before spinal anesthesia or LP [46,47]. This presumes that the underlying indications for anticoagulation therapy allow a temporary suspension of treatment. While the optimal timing of restarting anticoagulation after LP is not known, the incidence of spinal hematoma in the above-mentioned series was much lower when anticoagulation was started at least one hour after the LP [18]. Subcutaneous heparin administration is not believed to pose a substantial risk for bleeding after LP if the total daily dose is less than 10,000 units.
Aspirin has not been shown to increase the risk of serious bleeding following LP. In a prospective study of 924 patients who underwent orthopedic procedures with spinal or epidural anesthesia, 386 patients were taking antiplatelet therapy prior to surgery; 193 were taking aspirin [48]. Neither aspirin nor any other antiplatelet agents were associated with an increased risk of bleeding. However, none of these patients was taking clopidogrel, ticlopidine, or a GP IIa/IIIb receptor antagonist. Female gender, increased age, a history of excessive bruising/bleeding, hip surgery, continuous catheter anesthetic technique, large needle gauge, multiple needle passes, and moderate or difficult needle placement were all significant risk factors for minor bleeding at the site of catheter placement [48]. Given the unknown risk of bleeding with thienopyridine derivatives (clopidogrel, ticlopidine) it may be reasonable to suspend treatment with these agents, when possible, for one to two weeks prior to an elective LP, while pharmacologic data suggests that for GP IIa/IIIb receptor antagonists, a shorter period of treatment cessation (8 hours for tirofiban and eptifibatide and 24 to 48 hours for abciximab) may be indicated [47].
In all cases, the relative risk of performing an LP has to be weighed against the potential benefit (eg, diagnosing meningitis due to an unusual or difficult to treat pathogen). In cases in which LP is considered necessary but the risk of bleeding is considered to be high, it may be useful to perform the procedure under fluoroscopy to reduce the chance of accidental injury to small blood vessels.
Cerebral herniation — The most serious complication of LP is cerebral herniation. Suspected increased intracranial pressure (ICP) is a relative contraindication to performance of an LP and also requires independent assessment and treatment. The magnitude of the risk was evaluated in a report of 129 patients with increased ICP: 15 patients (12 percent) had an unfavorable outcome within 48 hours of LP [49]. Similar findings were noted in a series of 55 patients with subarachnoid hemorrhage: seven patients (13 percent) experienced neurologic deterioration during or soon after an LP, six of whom had evidence of cerebral dislocation [50]. Cardiorespiratory collapse, loss of consciousness, and death may follow. (See "Evaluation and management of elevated intracranial pressure in adults".)
A 1969 study of 30 patients with increased ICP who deteriorated after LP attempted to identify the clinical features of patients who were at greatest risk for this complication [51]. The following findings were noted: 73 percent had focal findings on neurologic examination (including dysphagia, hemiparesis, and cranial nerve palsies); 30 percent had documented papilledema prior to the LP; and 30 percent had evidence of increased ICP on plain skull films (erosion of the posterior clinoid processes). Deterioration occurred immediately in one-half of the patients, with the remainder declining within 12 hours.
The concern about this serious complication has resulted in routine CT scanning prior to LP being the standard of care in many emergency departments. At one institution, for example, 78 percent of patients with suspected meningitis underwent CT scanning before the LP was performed [52]. However, this practice, when applied to patients with suspected bacterial meningitis, delays the performance of LP, which in turn may delay treatment or limit the diagnostic power of CSF analysis when performed after antibiotic administration. Moreover, CT scanning is not necessary in all patients prior to LP and may not be adequate to exclude elevated ICP in others [53,54]. Some studies suggest that high-risk patients can be identified, allowing the majority of patients to safely undergo LP without screening CT [52,55]. This was best illustrated in a prospective study of 301 adults with suspected meningitis [52]. The following findings were noted:
Based upon these observations, we do NOT perform a CT scan before an LP in patients with suspected bacterial meningitis unless one or more risk factors is present:
Patients with these clinical risk factors should have a CT scan to identify possible mass lesion and other causes of increased ICP. Mass lesions causing elevated ICP are usually easily identified on CT scan. However, the CT scan should also be scrutinized for more subtle signs including diffuse brain swelling as manifest by loss of differentiation between gray and white matter and effacement of sulci, as well as ventricular enlargement and effacement of the basal cisterns [56].
Independent of the decision to perform LP, patients with possible elevated ICP based upon the above clinical features may require urgent life-saving interventions to lower ICP that may include head elevation, hyperventilation to a PCO2 of 26 to 30 mmHg, and intravenous mannitol (1 to 1.5 g/kg). When indicated, these should NOT await CT scan. The evaluation and management of patients with elevated ICP is discussed in detail separately. (See "Evaluation and management of elevated intracranial pressure in adults", section on 'Urgent situations'.)
When the LP is delayed or deferred in the setting of suspected bacterial meningitis, it is important to obtain blood cultures (which reveal the pathogen in more than half of patients) and promptly institute antibiotic therapy. Urgent evaluation and treatment of increased intracranial pressure, along with the administration of antibiotics and steroids, should be instituted promptly when this is suspected. Specific treatments are discussed separately. (See "Initial therapy and prognosis of bacterial meningitis in adults", section on 'Avoidance of delay'.)
Others
Epidermoid tumor — The formation of an epidermoid spinal cord tumor is a rare complication of LP that may become evident years after the procedure is performed [57-59]. Most reported cases are children ages 5 to 12 years who had a LP in infancy; however this has also been described in adults [60-62]. It may be caused by epidermoid tissue that is transplanted into the spinal canal during LP without a stylet, or with one that is poorly fitting. This complication probably can be avoided by using spinal needles with tight-fitting stylets during LP [63,64].
Abducens palsy — Both unilateral and bilateral abducens palsy are reported complications of LP [65-67]. This is believed to result from intracranial hypotension and is generally accompanied by other clinical features of post LP headache. Most patients recover completely within days to weeks. Other cranial nerve palsies are rarely reported [68].
Radicular symptoms and low back pain — It is not uncommon (13 percent in one series) for patients to experience transient electrical-type pain in one leg during the procedure [69]. However, more sustained radicular symptoms or radicular injury appear to be rare [70].
Up to one-third of patients complain of localized back pain after LP; this may persist for several days, but rarely beyond [69].
SUMMARY AND RECOMMENDATIONS — Lumbar puncture (LP) is essential or extremely useful in the diagnosis of bacterial, fungal, mycobacterial, and viral CNS infections and, in certain settings, for help in the diagnosis of subarachnoid hemorrhage, CNS malignancies, demyelinating diseases, and Guillain-Barré syndrome.
All topics are updated as new information becomes available. Our peer review process typically takes one to six weeks depending on the issue.