52nd Annual Meeting of the American Academy of Neurology
Randolph W. Evans, MD
Neurologists in the United States perform about 400,000 lumbar punctures per year. Although neurologists are not typically concerned about complications of lumbar puncture, our patients are inordinately worried and frightened. It seems as though almost everyone has a cousin or a friend who has been paralyzed by a lumbar puncture. Neurosurgeons appear to have less trouble obtaining consent for a craniotomy than we do for a lumbar puncture.
After the procedure, patients usually wonder what all the fuss was about. In a recent study, 46 patients underwent 3 procedures with needles: spinal anesthesia without local anesthetic, an intramuscular (IM) injection, and an intravenous injection (IV). Before the procedures, 83% believed that the lumbar puncture would be the most painful. After the procedure, only 11% perceived the lumbar puncture as the most painful, with 50% choosing the IM injection and 39% the IV.
On May 5, the seventh day of the AAN annual meeting, Dr. Stephen Silberstein from Philadelphia, Pennsylvania, Dr. Michael Strupp from Munich, Germany, and I presented a dinner seminar: "Complications of Lumbar Puncture and Low CSF Pressure Syndrome." My presentation is summarized. The interested reader may wish to look for the forthcoming AAN Technology and Therapeutics Subcommittee position paper that will be published in Neurology in a few months (Evans RW, Armon C, Frohman EM, Goodin DS. Prevention of Post-lumbar Puncture Headaches).
Possible Complications of Lumbar Puncture
Headache is the most common complication of lumbar puncture, usually lasting 1 week or less, and occurring in up to 40% of patients when using the traditional bevel tip or Quincke needle. Although the risk of headache can be dramatically reduced to about 5% by the use of atraumatic or pencil tip needles such as the Sprotte or Whitacre, most neurologists in the United States have never heard of the needles and only 2% use them. By contrast, anesthesiologists commonly use the atraumatic needles. I have been using the Sprotte needle for 3 years, especially in patients at high risk for headaches (eg, young women having lumbar puncture to evaluate the worst headache of their life or possible multiple sclerosis), and have seen a dramatic reduction in headaches.
Each year, 400,000 diagnostic lumbar punctures are performed in the United States. If we could reduce the incidence of post-lumbar puncture headaches from about 30% to 5%, this would mean a reduction from 120,000 patients with headache to 20,000! Clearly, neurologists need to become familiar with the use of these needles.
Other complications, including headaches lasting from 8 days to 1 year, cranial neuropathies, prolonged backache, nerve root injury, and meningitis, are rare, following perhaps 0.3% of lumbar punctures. Table 1 lists the complications of lumbar puncture.
Table 1. Complications of Lumbar Puncture
Even in the presence of a brain neoplasm, abscess, or hematoma, uncal or tonsillar herniation leading to neurologic deterioration or death is quite uncommon. When a complication occurs, it can be difficult to determine whether the lumbar puncture was responsible, particularly if deterioration is not immediate or if the procedure is performed on an obtunded or comatose patient who might have gotten worse or died in a short time anyway.
In the era before CT and MRI, lumbar punctures were often done on patients with brain masses. However, even when papilledema and elevated intracranial pressure were present, related complications occurred in only about 1% of patients. In children with bacterial meningitis, herniation can occur even with a normal CT. Rennick and colleagues recommended that in a child with suspected meningitis and decerebrate or decorticate posturing, focal neurologic signs, or no response to pain, antibiotics should be given but a lumbar puncture should not be performed even when a CT scan of the brain is normal. In a study of adults, clinical signs of herniation developed within several minutes to several hours after lumbar puncture in 5 of 493 episodes of bacterial meningitis. Durango and associates recommend obtaining a CT scan in patients with suspected meningitis and signs of intracranial pressure, or focal findings on neurologic exam before performing a lumbar puncture.
Subarachnoid Hemorrhage, Pseudotumor Cerebri, and Acquired Chiari I Malformation
In patients with suspected subarachnoid hemorrhage and a normal CT scan, lumbar puncture is certainly appropriate and safe. However, in cases of subarachnoid hemorrhage with hematoma and mass effect, lumbar puncture can result in immediate deterioration in about 13% of patients. Lumbar puncture is also safe in patients with pseudotumor cerebri. However, there is a single case report of respiratory arrest leading to death in a patient with pseudotumor cerebri who developed a Chiari I malformation. Reversible tonsillar descent, an acquired Chiari I, has been well documented as a complication of lumbar puncture, overdraining CSF shunts, and spontaneous intracranial hypotension.
Subarachnoid Block and Spinal Coning
Neurologic deterioration can occur when lumbar puncture is performed below the level of a complete spinal subarachnoid block. In one series of patients with rostral block due to neoplasms, 14% deteriorated after lumbar puncture.
Postdural Puncture Headache
Post-lumbar puncture or, more precisely termed, postdural puncture headache (PDPH), is the most common complication of lumbar puncture, occurring in up to 40% of patients. The headache begins within 48 hours in 80% and within 72 hours in 90%, although the onset can be immediately after the procedure or delayed for as long as 14 days. The duration of the headache is less than 5 days in about 80%, although the headache can persist for 12 months.
The headache is usually but not always bilateral and may be characterized by frontal, occipital, or generalized pressure or throbbing occurring when the patient is upright, and diminishing or resolving when supine. The headache worsens with head movement, coughing, straining, sneezing, and jugular venous compression. The longer the patient is upright, the longer the time before the headache subsides when supine. In one series, additional symptoms were present in the following percentages: neck stiffness, 43%; nausea, 66%; vomiting, 27%; cochlear symptoms, 15%; and ocular symptoms, 12%. In another series, nausea and vomiting were present in 22% and 2% of patients, respectively.
The cause of PDPH is not entirely certain. The best explanation is that the CSF pressure falls as the CSF leaks through a dural and arachnoid tear produced by the puncture, and the leakage exceeds the rate of CSF production. This CSF hypotension can produce headache and cranial nerve symptoms through a downward descent of the brain, stretching pain-sensitive structures including the dura, nerves (cranial nerves V, IX, and X, and the upper 3 cervical nerves), and bridging veins. Secondarily, intracranial venous dilatation and increased brain volume occur as the veins passively dilate in response to decreased extravascular pressure.
As a following section on risk factors demonstrates, even in the presence of low CSF pressure, there is individual susceptibility or resistance to headache. This is the case in many other pain disorders (including elevated intracranial pressure, neural compression, etc) where the presence of objective pathology unreliably correlates with pain complaints. There is an interesting hypothesis that may explain some of the individual susceptibility to PDPH. Substance P levels in plasma (and presumably CSF) rise after lumbar puncture. Patients with low CSF substance P levels are 3 times more likely to develop PDPH than those with higher levels. Thus, PDPH may be mediated by the release of substance P after lumbar puncture in those with low levels of substance P and receptor-mediated hypersensitivity.
Investigations are not commonly performed since PDPH can be easily diagnosed based upon the characteristic symptoms. A repeat lumbar puncture will usually demonstrate an opening pressure from 0-70 cm H2O, although the pressure can be in the normal range -- especially if the procedure is performed after a period of bed rest. As in the case of other low-pressure headaches, the CSF analysis may be normal or can demonstrate a moderate, primarily lymphocytic pleocytosis; the presence of red blood cells; and elevated protein, which can be even higher than 500 mg/dL.
An MRI of the brain may reveal diffuse meningeal enhancement with gadolinium and also, in some cases, subdural fluid collections, which return to normal with resolution of the headache. In a recent study, 8 out of 9 patients with PDPH showed meningeal enhancement, usually mild. However, 9 other patients who also underwent lumbar puncture but did not have headache had no change in the slight meningeal enhancement seen before the lumbar puncture. The diffuse meningeal enhancement on MRI may be explained by dural vasodilatation and a greater concentration of gadolinium in the dural microvasculature and dural interstitial fluid. In a single case report, MRI also showed abnormal dural venous sinus enhancement, a new finding in PDPH, suggesting compensatory venous expansion. The pleocytosis and elevated protein in the CSF and the subdural fluid collections are probably due to reduced CSF volume and hydrostatic pressure changes resulting in meningeal vasodilation and vascular leak. Lumbar MRI may also be abnormal following lumbar puncture. In one study of 11 patients, all had evidence of CSF leakage ranging from 1 to 460 mL.
Recently, color Doppler imaging of the superior ophthalmic vein has been demonstrated to be a sensitive indicator of intracranial hypotension in 13 spontaneous cases and 4 post-lumbar puncture cases. The Doppler study reveals a significantly higher mean maximum flow velocity and a substantially larger mean diameter of the vein, which may be due to a decline in CSF pressure transmitted to the cavernous sinus. The pressure gradient across the cavernous sinus probably causes the bilateral engorgement of the superior ophthalmic vein, whereby increasing amounts of venous blood are carried and drained. The superior ophthalmic vein changes were documented to be reversible after clinically successful epidural blood patches.
Patient demographics. Patient demographic risk factors include the following: female gender (twice as often); age (greatest in those 18-30 years of age and much less in children younger than 13 years and adults older than 60); lower body mass index; previous chronic or recurrent headache; and prior PDPH.
Quincke needle. With the conventional bevel tip or Quincke needle (named after Heinrich Quincke, who performed the first percutaneous lumbar puncture in December 1890 on a small child), the likelihood of PDPH rises with the use of a larger-diameter needle (20 to 22 gauge, 20% to 40%; 24 to 27 gauge, 5% to 12%). However, smaller-diameter needles are not practical for diagnostic lumbar puncture because of the slow flow of CSF (eg, 10.5 mL/hour with a 25-gauge Quincke needle, 40.6 mL/hour with a 22-gauge Quincke, and 133 mL/hour with a 20-gauge Quincke needle vs 64.7 mL/hour with a 22-gauge Sprotte needle and 206 mL/hour with a 20-gauge Sprotte needle) and the longer duration to 95.8% of CSF opening pressure (43 seconds with a 20-gauge Quincke, 225 seconds with a 22-gauge Quincke, and 336 seconds to 83.3% of pressure with a 25-gauge Quincke vs 79 seconds with a 22-gauge Sprotte).
The risk of PDPH is reduced by 50% when the bevel is parallel rather than perpendicular. Fewer fibers are cut with parallel entry because the dural fibers run parallel to the long axis of the spine. Parallel position means that a plane passing through the flat part of the bevel, going though both edges of the bevel, is parallel to the long or vertical axis of the spine. The face of the bevel and the notch in the hub should point in the direction of the patient's side, not toward the patient's head or feet.
Atraumatic needle. The most widely used atraumatic or pencil-point needles are the Whitacre and Sprotte (Figure 1). The tip is dull and the aperture or opening is just proximal to the tip. The dull tip spreads the dural fibers rather than cutting them and greatly reduces the risk of PDPH.[11,12] With the use of the Sprotte needle, the risk of headache is reduced from about 30% to 5%.
Lenaerts and colleagues conducted the only study showing no benefit of the Sprotte needle. This study is at odds with other studies[8,11-13] showing benefit of the atraumatic needle for reduction of headache after myelography and spinal anesthesiology, and may be the result of several factors. The 5.7% incidence of PDPH with the Quincke needle is much lower than that found in other studies; fewer women (43%) than men were in the cohort (women are 2 times more likely than men to have PLPH); and the study included patients with disorders such as hysteria, Guillain-Barré (were they on bed rest?), encephalitis, and viral meningitis, where the determination of PLPH may not have been accurate.
Because the tip of the needle is relatively dull, a sharp short introducer is provided with the Sprotte needle. The introducer should be inserted about two thirds of its length or even less in a small, thin person so as not to puncture the dura. The feel of the atraumatic needle during the procedure is a little different from that with the Quincke's sharp cutting edge, as you have to push a little harder during insertion. The cost is higher for the atraumatic needle than for the Quincke -- the Sprotte 20, 21, and 22 G 3 1/2-inch needles are about $12 each vs about $4 for the Quincke; longer Sprotte needles are also available.*
While atraumatic needles are commonly used by anesthesiologists, most neurologists have never heard of them and only 2% use them, despite the evidence demonstrating the greatly reduced risk of PDPH. In the United Kingdom, over 70% of neurologists and neurosurgeons are using the Quincke needle rather than the atraumatic cone. The added expense of the atraumatic needles is a potential barrier to use. However, a lumbar puncture tray with the Sprotte needle is available (Havel's Incorporated, Cincinnati, Ohio) at a cost comparable to that of a tray with a Quincke needle and may lead to greater use of this needle, especially in high-risk patients as described.
Reinserting the stylet before withdrawing the needle. Strupp and Brandt[15,16] recently evaluated the effect on PDPH of reinserting or not reinserting the stylet before withdrawing the needle. All the lumbar punctures were performed with a Sprotte 21-gauge needle. PDPH developed in 16% of the 300 patients without reinsertion and in only 5% of the 300 patients with reinsertion. They hypothesize, "The reason for this difference may be that a strand of arachnoid enters the needle with the outflowing cerebrospinal fluid during diagnostic lumbar puncture; when the needle is removed, the strand may then be threaded back through the dural defect and produce prolonged cerebrospinal fluid leakage along the arachnoid." Based on this study, it seems reasonable to replace the stylet when using the Sprotte needle and the Quincke needle (although similar evidence for replacing the stylet is not available for the Quincke needle).
Prevention of PDPH Postprocedure
Contrary to conventional wisdom, bed rest for up to 24 hours or various body positions such as prone or head down after the lumbar puncture do not reduce the incidence of PDPH as compared to immediate ambulation.[17-23] Intake of oral fluids postprocedure is also not preventive.
Treatment of PDPH
For patients who do not relish the prospect of a number of days at bed rest, there are other effective treatments. Oral caffeine 300 mg every 4-6 hours is worth trying initially, although the relief may be transient. Intravenous caffeine sodium benzoate given as a slow intravenous bolus of 500 mg may initially relieve headache in 75% but permanent relief in only 50%. Caution should be used in patients with coronary artery disease or seizure disorders. The epidural blood patch is the most effective treatment, with a success rate of 85% after 1 treatment and 98% after a second. To obtain maximal benefit, the patient should stay in the decubitus position for 2 hours after the procedure. The presumed mechanism is an immediate gelatinous tamponade of the dural hole. Side effects are usually mild and transient.
Dysfunction of cranial nerves III, IV, V, VI, VII, and VIII has been reported after lumbar puncture. Abducens paresis may occur in as many as 1 in 400 lumbar punctures and can be unilateral or bilateral and are typically resolved over 4 to 6 weeks.
Nerve Root Irritation, Low Back Pain, and Insertion of the Stylet
About 13% of patients have transient electric shocks or dysesthesias due to the needle contacting sensory roots. Permanent motor and sensory loss can occur rarely. Obviously, spinal cord injury can result from performing the puncture at the wrong level. Thirty five percent of patients complain of low back pain after the procedure and this can occasionally persist for months. Rarely, if the needle is pushed beyond the subarachnoid space, the annulus fibrosis can be damaged, resulting in disc herniation or discitis or vertebral collapse.
The stylet should always be used on insertion through the skin and subcutaneous tissue to prevent the rare complication of implanting a plug of skin that can grow into an intraspinal epidermoid tumor.
Lumbar puncture can cause infectious complications due to the following: a contaminated needle, eg, contamination due to respiratory droplets; disseminating skin flora without adequate disinfection of the skin; performing a lumbar puncture in an area of infection, eg, cellulitis, furunculosis, or epidural abscess; and the introduction of blood in the subarachnoid space in the presence of bacteremia. In patients with CSF leaks, lumbar puncture can reverse the flow gradient, produce retrograde spread of organisms from the nasopharynx through the dural leak, and result in meningitis. In addition to bacterial meningitis, discitis, lumbar epidural abscess, and spinal cord abscess can also rarely occur. Bacterial meningitis is a rare complication after diagnostic lumbar puncture and may occur after 0.2% of myelograms. The use of face masks during myelography has been recommended.
Intracranial subdural hematoma is a rare complication of lumbar puncture that can occur in healthy patients without bleeding disorders. A typical post-lumbar puncture headache is often but not always present initially. The age of reported patients ranges from 22 to 79. The subdural hematoma, which can be unilateral or bilateral, may be diagnosed after an interval of 3 days' to several months' duration. The mechanism may be low CSF pressure resulting in traction on the meninges and tearing of dural vessels. Similarly, traction on the basal blood vessels can, rarely, result in rupture of a saccular aneurysm and subarachnoid hemorrhage as a complication of lumbar puncture.
A traumatic lumbar puncture, needle-induced blood in the CSF, occurs in 72% of diagnostic lumbar punctures with the number of RBC/mm3 present in the following percentages of lumbar punctures: 1-5 RBC, 27%; 6-50 RBC, 21%; and more than 50 RBC, 24%. The cause is usually puncture of the radicular vessels that accompany each nerve root along the length of its surface and only rarely from the epidural veins. Occasionally, the radiculomedullary artery of Adamkiewicz and corresponding vein may be present in a lower than usual position and accompany the L3, L4, or L5 nerve roots where puncture could occur.
Thrombocytopenia, bleeding disorders, anticoagulation, and a difficult or bloody tap are all significant risk factors for spinal hemorrhages. The standard recommendation is to perform a lumbar puncture, when the platelet count is not rapidly falling, only with a platelet count of 50,000 or greater (although platelet counts of 20,000 or greater and 100,000 or greater have also been recommended). Most reports of spinal subarachnoid hematomas are in patients with platelet counts of less than 50,000, although there is a report of a hematoma in a patient with relapsing lymphoblastic leukemia, a platelet count of 63,000, and normal prothrombin (PT) and partial thromboplastin times (PTT).
In anticoagulated patients, adequate reversal (with protamine for those on heparin and vitamin K or fresh frozen plasma for those on warfarin) is mandatory before lumbar puncture. In patients with bleeding disorders, replacement therapy is indicated. For example, in one series, lumbar punctures were safely performed on 33 patients with hemophilia after pretreatment with clotting factor. In questionable cases, a platelet count, PT, PTT, and bleeding time should be obtained before lumbar puncture.
In one large study, preoperative antiplatelet therapy with aspirin or nonsteroidal anti-inflammatory medications and subcutaneous heparin on the operative day were not risk factors for spinal hematoma in patients undergoing spinal or epidural anesthesias. However, anticoagulation with intravenous heparin following the lumbar puncture is a risk factor for spinal hemorrhage. Based on a series of patients receiving heparin for cerebral ischemia after lumbar puncture, Ruff and Dougherty have recommended waiting at least 1 hour before anticoagulation. However, one report describes a patient with cerebral ischemia who was placed on heparin 26 hours after a traumatic lumbar puncture and developed a spinal subarachnoid hematoma.
Spinal subarachnoid hematoma, a rare complication of lumbar puncture, presents with severe low back and/or radicular pain followed within hours or days by progressive paraparesis, sensory loss, and sphincter disturbances. Rarely, the spinal subarachnoid hematoma may rupture through the dura, resulting in an additional subdural collection. Successful treatments include immediate decompressive laminectomy and hematoma evacuation as well as percutaneous hematoma aspiration. Epidural hematomas can also rarely occur after lumbar puncture with the same clinical presentation, risk factors, and treatment as spinal subarachnoid hematomas.
A variety of other complications can result from lumbar puncture. Vasovagal syncope and cardiac arrest can be associated with any medical procedure. Hyperventilation during the procedure can lead to multiple complaints, including chest pain, dizziness, and paresthesias. Seizures have been reported in association with PDPH. Although seizures have been reported with metrizamide myelography, this is a rare problem with iohexol and iopamidol. There is a single case report of a symptomatic spinal intradural arachnoid cyst which developed after a lumbar myelogram. Rarely, when a lumbar puncture is performed on a child with a ventriculo-peritoneal shunt, a low CSF pressure state can develop with ventricular enlargement despite an otherwise working shunt. If the opening pressure is measured when the patient is not relaxed or the legs not partially extended, the opening pressure may be incorrectly interpreted as elevated and thus result in wrong diagnoses such as pseudotumor cerebri (which we might call "pseudo-pseudotumor cerebri"). Finally, results of the various types of CSF analysis from the laboratory can be incorrect and result in harm to the patient.
Implications for Clinical Practice
Complications following lumbar puncture are numerous and can range from trivial to life-threatening.
Headache is the most common complication, occurring in up to 40% of patients with use of the standard Quincke needle.
Demographic risk factors for headache include female gender, age 18 to 30 years, lesser body mass index, and a history of headache.
When using a Quincke needle, bevel insertion parallel to the longitudinal dural fibers reduces the risk of headache by 50% compared with perpendicular insertion.
Use of atraumatic needles such as the Sprotte or Whitacre can greatly reduce the incidence of headache to about 5%. The use of atraumatic needles should be strongly considered, especially in patients with risk factors for headaches.
Replacing the stylet before removing the needle significantly reduces the risk of headache.
Postprocedure bed rest does not reduce the incidence of headache.
Although oral and IV caffeine are somewhat effective, the lumbar epidural
blood patch is successful in relieving postdural puncture headache in 85%