Lumbar Puncture First? An old test and a new approach to lone acute sudden headaches
Michael J. Schull, MD
Clinical Epidemiology Unit and Emergency Department, Sunnybrook and Women’s College Health Sciences Centre, University of Toronto, Toronto, Ont.
On évoque souvent le diagnostic de l’hémorragie sous-arachnoïdienne chez les patients qui se présentent à l’urgence pour un mal de tête violent et soudain, mais dont l’examen physique est normal. La norme de soins de nos jours consiste à soumettre ces patients à une tomodensitométrie (TDM) sans substance de contraste suivie d’une ponction lombaire pour les TDM négatives. Cependant, comme la plupart des céphalées évaluées sont bénignes, la plupart des résultats de la TDM et de la ponction lombaire sont normaux.
Le présent article a pour but d’analyser l’impact d’un protocole diagnostique de rechange, où la ponction lombaire serait la première (et dans la plupart des cas, la seule) épreuve diagnostique effectuée chez les patients chez qui l’on soupçonne une hémorragie sous-arachnoïdienne et qui présentent ce symptôme isolé de mal de tête violent. Selon une hypothèse raisonnable, pour chaque 100 patients évalués, le protocole de «ponction lombaire en premier» nécessiterait 81 TDM de moins et seulement 9 ponctions lombaires additionnelles, comparativement à la stratégie diagnostique traditionnelle. Ces avantages demeurent stables même en variant les hypothèses à l’intérieur des limites cliniquement plausibles.
Nous croyons que ce protocole pourrait entraîner une utilisation beaucoup plus efficace des ressources en limitant le taux de morbidité additionnel, tout en offrant une précision diagnostique équivalente pour l’hémorragie sous-arachnoïdienne chez le patient à l’urgence qui répond au critère de mal de tête isolé violent et soudain.
Headaches are a frequent presenting complaint in the emergency department (ED), representing up to 4.5% of visits.1 While most patients have benign headache syndromes, some have dangerous secondary causes such as subarachnoid hemorrhage (SAH). This condition, which often occurs in young people, has a 50% mortality rate25 and leaves almost half its survivors with neurologic sequelae. ED physicians must maintain a high index of suspicion for SAH and aggressively investigate this potentially devastating disease. About 70% of ED patients with acute sudden onset (< 1 minute) headaches suggestive of SAH will have a normal neurologic exam, normal vital signs and temperature, normal level of consciousness, and no neck stiffness.2 Such patients are termed as having a lone acute sudden headache (LASH). Most LASH patients (8090%) have benign conditions, but they require further investigation, since 812% will prove to have SAH.2,6,7
The traditional diagnostic strategy is to begin with a CT (computed tomography) scan of the head. CT sensitivity at 24 hours varies from 81100% and decreases rapidly thereafter.815 CT sensitivity also varies depending on the patient's neurologic status (spectrum bias), being 99.4% sensitive in stuporous or comatose patients but only 86% sensitive in alert patients.5,10 Therefore, CT by itself cannot rule out SAH. Lumbar puncture is highly sensitive, and virtually 100% of SAH victims will develop spinal fluid (CSF) xanthochromia within 12 hours of the onset of bleeding. Consequently, in patients with suspected SAH, a normal CT must be followed by a lumbar puncture (LP). The CSF should be analyzed using a spectrophotometer, since direct visual identification of xanthochromia may be false negative in up to 50% of cases.5,8,9,12,1618
Recently, we posed the question of whether a CT need be done prior to LP.19 To answer this, we must distinguish between two subgroups of patients: those with normal versus those with abnormal clinical findings. CT scanning prior to LP is recommended in all non-LASH patients, but for patients who fulfil LASH criteria, we propose an alternate approach - the LP First model - in which LP is the first and, in most cases, the only diagnostic test performed. A more detailed description of the LP First model has been published previously.19
The LP First approach
Fig. 1. LP First: a new approach for an old test
Within the LP First model (Fig. 1), patients fulfiling LASH criteria would initially undergo lumbar puncture at least 12 hours after headache onset. Those with negative LPs would be discharged home with follow-up; those with positive LPs would undergo CT scanning and other appropriate tests (e.g., angiography); and those with indeterminate LPs (i.e., unable to obtain CSF, or elevated red blood cell [RBC] count without xanthochromia) would undergo CT scanning and further investigations as indicated.
What are the benefits of this approach? Since the majority of LASH patients do not have an SAH, the LP first model would shorten time to diagnosis by eliminating the intervening CT. This is especially true in centres without CT scanners who must transfer LASH patients, and in centres that lack 24-hour CT availability, who must call in technicians or hold LASH patients overnight. Given standard assumptions (CT sensitivity [90%], specificity [99.9%], SAH prevalence among LASH patients [10%], rate of indeterminate LPs [10%]), then for every 100 patients presenting with LASH, the traditional strategy would lead to 100 CTs, 91 LPs, and 9 patients requiring further investigation. The LP First strategy would lead to 100 LPs, 19 CTs, and 10 patients requiring further investigation, a reduction of 81 CTs (Fig. 2). Even if the above assumptions are varied over clinically plausible ranges, the benefits of this model remain stable. In fact, the lower a physician's threshold for investigating SAH (i.e., the lower the prevalence of SAH among investigated patients), the greater the benefit of the LP First model.19 Since LP is the gold standard test for SAH, both the LP First and the traditional strategies would have equal diagnostic accuracy for SAH.
The primary roadblock to implementing this model is the fear many physicians have that LP is unsafe without prior CT. There is a general consensus, and some evidence, that CT should be performed before LP in patients with abnormal neurologic findings or altered mentation. In one report of 283 patients with SAH, all 4 patients who suffered complications following LP were described as having neck stiffness and being in "HuntHess grades 1 to 3" (i.e., from normal neurological exam to major neurological deficit).15,20 In another report, all 7 patients who suffered complications after LP were classified as HuntHess grades 2 or 3.21 None of these patients would have met the LASH criteria defined above. In fact, evidence for LP risk is either anecdotal or extrapolated from patients with abnormal neurological exams.10,12,22 There is little evidence that LP is dangerous in LASH patients. Despite this, and the suggestions of several authors that it is unnecessary,15,2326 many centres require CT before LP in all patients.
Fig. 2. LP First vs. traditional diagnostic pathway
Limitations of the LP First model
Some patients with acute sudden headache have a secondary etiology that cannot be diagnosed by LP. In such cases, a normal tap may be falsely reassuring. In a recent study of 148 patients with acute sudden headache, 103 (70%) had normal neurological exams and 16 of these had secondary causes, including 12 with SAH and 1 with meningitis. Three patients (2.9%) had intracranial pathology not detectable by LP alone (one intracerebral hematoma, one giant aneurysm without bleed and one arterio-venous malformation with hematoma).2 This 2.9% "diagnostic miss rate" is concerning but is no higher than that seen in patients with slow-onset headaches and normal neurological exams, in whom the prevalence of intracranial pathology ranges from 3.0% to 6.9%.27 These LP First "misses" could be reduced further by excluding high-risk patients, including the elderly and those with known cancer or HIV, and arranging good patient follow-up to assure that patients with persistent symptoms undergo further investigation. This follow-up "safety net" is not unlike many other diagnostic models used in emergency medicine, such as those used to assess febrile infants.
The LP First strategy might, in some cases, cause diagnostic delays. To maximize sensitivity, LPs should be performed at least 12 hours after headache onset, unless meningitis is suspected. Since over half of LASH patients present later than 24 hours,28 most would experience no delay. Those who present early could be investigated in the traditional manner (immediate CT) or observed in the ED until the 12-hour mark. In some cases, this would mean delaying a diagnosis that could have been made earlier by CT. The impact of such a delay, however, is likely to be small, since surgical intervention is typically delayed for many hours or even overnight, pending cerebral angiography.
Performing more LPs might lead to more complications. The LP First model would result in 9 additional LPs per 100 patients investigated. This would cause from 0 to 3 additional post-LP headaches. Other LP complications are so rare that no discernible increase would be expected using the LP First model.19
Finally, the most important limitation of the LP First model is that it is untested. Before it can be advocated or widely used, prospective clinical trials are necessary to prove its safety. Younger patients without risk factors might be the ideal candidates for the LP First model, and prospective studies are needed to better characterize the patient group who will benefit most from this strategy.
The LP First model is intended for patients with LASH compatible with SAH, who have a normal neurologic exam, normal vital signs and temperature, normal level of consciousness, and no neck stiffness. Given that most patients with acute sudden headaches meet these criteria,2,6 the LP First model has the potential to expedite the investigation of headache patients and substantially reduce the need for CT scanning. Greater efficiency and improved utilization would result, but would not be expected to reduce diagnostic accuracy. Before the LP First model can be widely advocated, however, a prospective clinical evaluation is required to determine its safety, costs and benefits.
- Ramirez-Lassepas M, Espinosa CE, Cicero J, Johnston KL, Cipolle R, Barber D. Predictors of intracranial pathologic findings in patients who seek emergency care because of headache. Arch Neurol 1997;54:1506-9.
- Linn FH, Wijdicks EF, van der Graaf Y, Weerdesteyn-van Vliet F, Bartelds A, van Gijn J. Prospective study of sentinel headache in aneurysmal subarachnoid hemorrhage. Lancet 1994;344:590-3.
- Kassell NF, Torner JC, Haley EC, Jane JA, Adams HP, Kongable GL. The international cooperative study on the timing of aneurysm surgery. Part 1: overall management results. J Neurosurg 1990;73:18-36.
- Kassell NF, Torner JC. The international cooperative study on the timing of aneurysm surgery: an update. Stroke 1984;15(3):566-70.
- Joynt R. Clinical neurology. 3rd ed. Philadelphia: LippincottRaven Press; 1996.
- Lledo A, Calandra L, Martinez-Menendez B, Perez-Sempere A, Portera-Sanchez A. Acute headache of recent onset and subarachnoid hemorrhage: a prospective study. Headache 1994;34:172-4.
- Edmeads J. Acute headache and subarachnoid hemorrhage [editorial]. Headache 1994;34(7):447.
- Evans RW. Diagnostic testing for the evaluation of headaches. Neurol Clin N Am 1996;14(1):1-26.
- Schievink WI. Intracranial aneurysms. N Engl J Med 1997;336(1):28-40.
- Adams HP, Kassell NF, Torner JC. CT and clinical correlations in recent aneurysmal subarachnoid hemorrhage: a preliminary report of the cooperative aneurysm study. Neurol 1983;33:981-8.
- Sames TA, Storrow AB, Finkelstein JA, Magoon MR. Sensitivity of new generation computed tomography in subarachnoid hemorrhage. Acad Emerg Med 1996;3:16-20.
- Vermeulen M, van Gijn J. The diagnosis of subarachnoid hemorrhage. J Neurol Neurosurg Psych 1990;53:765-72.
- van Gijn J, van Dongen KJ. The time course of aneurysmal hemorrhage on computed tomograms. Neuroradiol 1982;23:153-6.
- van der Wee N, Rinkel GJ, Hasan D, van Gijn J. Detection of subarachnoid hemorrhage on early CT: Is lumbar puncture still needed after a negative scan? J Neurol Neurosurg Psych 1995;58:357-9.
- Wasserberg J, Barlow P. Lumbar puncture still has an important role in diagnosing subarachnoid hemorrhage. BMJ 1997;315:1598-9.
- Vermeulen M, Hasan D, Blijenberg BG, Hijdra A, van Gijn J. Xanthochromia after subarachnoid hemorrhage needs no revisitation. J Neurol Neurosurg Psych 1989;52:826-8.
- Xanthochromia [editorial]. Lancet 1989;2(8664):658-9.
- van Gijn J, Vermeulen M, Hasan D. Xanthochromia [letter]. Lancet 1989;2(8670):1036.
- Schull, MJ. Lumbar puncture first: an alternative model for the investigation of lone acute sudden headache. Acad Emerg Med 1999;6:131-6.
- Hillman J. Should computed tomography scanning replace lumbar puncture in the diagnostic process in suspected subarachnoid hemorrhage? Surg Neurol 1986;26:547-50.
- Duffy GP. Lumbar puncture in spontaneous subarachnoid hemorrhage. BMJ 1982;285:1163-4.
- Fontanorosa PB. Recognition of subarachnoid hemorrhage. Ann Emerg Med 1989;18:1199-205.
- Macdonald A, Mendelow AD. Xanthochromia revisited: a re-evaluation of lumbar puncture and CT scanning in the diagnosis of subarachnoid hemorrhage. J Neurol Neurosurg Psych 1988;51:342-4.
- Humphreys RP. Computed tomography and the early diagnostic lumbar puncture [letter]. CMAJ 1979;121:1442-3.
- Humphreys RP. Computed tomography and the early diagnostic lumbar puncture. CMAJ 1979;121:150-1.
- Patel MK, Clarke MA. Lumbar puncture and subarachnoid hemorrhage. Postgrad Med J 1986;62:1021-4.
- Frishbeg BM. The utility of neuroimaging in the evaluation of headache in patients with normal neurological examinations. Neurol 1994;44:1191-7.
- Morgenstern LB, Luna-Gonzales H, Huber JC, Wong SS, Uthman MO, Gurian JH, Castillio PR, et al. Worst headache and subarachnoid hemorrhage: prospective, modern computed tomography and spinal fluid analysis. Ann Emerg Med 1998;32:297-304.