ebswoauorDfrbtaawtmCoftascssE

et

RA

The Journal of Emergency Medicine, Vol. 39, No. 1, pp. 13–16, 2010Copyright © 2010 Elsevier Inc.

Printed in the USA. All rights reserved0736-4679/$–see front matter

doi:10.1016/j.jemermed.2007.09.052

OriginalContributions

EVALUATING THE SENSITIVITY OF VISUAL XANTHOCHROMIA IN PATIENTSWITH SUBARACHNOID HEMORRHAGE

Sanjay Arora, MD, Stuart P. Swadron, MD, and Vinoo Dissanayake, MD

Department of Emergency Medicine, Keck/University of Southern California School of Medicine, Los Angeles, CaliforniaReprint Address: Sanjay Arora, MD, Department of Emergency Medicine, LAC � USC Medical Center, 1200 N. State Street, Room

1011, Los Angeles, CA 90033

HD(bpaepa

gbitcupnantdre

si

Abstract—Combined computed tomography and cere-rospinal fluid (CSF) analysis has been shown to be 100%ensitive for detecting subarachnoid hemorrhage (SAH)hen CSF is obtained between 12 h and 2 weeks from timef headache onset and spectrophotometry is used to evalu-te CSF for xanthochromia. Because most hospitals do notse spectrophotometry, we sought to evaluate the sensitivityf CSF analysis for xanthochromia by visual inspection. Weetrospectively identified all patients seen in the Emergencyepartment (ED) with an ED discharge diagnosis of SAH

rom June 1993 to November 2005. A structured charteview was performed on all patients with the additionalilled procedure charge for “lumbar puncture” or “spinalap.” Data collected included: CSF color, time from head-che onset to CSF collection, and confirmation of SAH bydvanced imaging. There were 1323 patients diagnosedith SAH, and 102 of these also had CSF collected. Of

hese, 81 charts were available for review. By predeter-ined protocol, 35 were excluded for lack of a report ofSF color, 1 was excluded because the time from headachenset to CSF collection was < 12 h, and 26 were excludedor lack of documentation of a definitive imaging study. Ofhe remaining 19, 9 were found to have xanthochromic CSFnd 10 were found to have colorless CSF, resulting in aensitivity for visual inspection of CSF of 47.3% (95%onfidence interval 24.4–71.1%). Visual inspection of CSFupernatant for xanthochromia lacks the sensitivity neces-ary to reliably exclude the diagnosis of SAH. © 2010lsevier Inc.

Keywords—subarachnoid hemorrhage; lumbar punc-ure; xanthochromia; diagnostic testing

ECEIVED: 19 April 2007; FINAL SUBMISSION RECEIVED: 12

CCEPTED: 16 September 2007

13

INTRODUCTION

eadache accounts for more than 5 million Emergencyepartment (ED) visits in the United States every year

1). In the vast majority of cases, the etiology proves to beenign in origin, however, approximately 30,000 of theseatients are eventually diagnosed with a non-traumatic sub-rachnoid hemorrhage (SAH) from a ruptured cerebral an-urysm (2). Because early intervention is necessary torevent long-term disability and mortality, prompt andccurate diagnosis is critical (3,4).

The diagnostic evaluation to exclude SAH often be-ins with non-contrast computed tomography (CT) of therain. The sensitivity of CT scan for SAH exceeds 90%f performed on the day of headache onset, but decreaseso 58% if performed on day 5 (5). As CT technologyontinues to evolve, its sensitivity will increase, but it isnlikely that CT alone will ever detect all cases (6). Inatients with a normal or equivocal cranial CT scan, theext step is to perform a lumbar puncture (LP). In thenalysis of the cerebrospinal fluid (CSF), the presence ofumerous, non-clearing red blood cells (RBCs) supportshe diagnosis of SAH. However, RBCs alone may not beiagnostic as they are often the result of a traumatic LP,esulting in a substantial high rate of falsely positivexaminations (7).

Xanthochromia is a yellow discoloration of the CSFupernatant that results from the breakdown of RBCsnto oxyhemoglobin, methemoglobin, and bilirubin. Be-

mber 2007;

Septe

cotiCw1sct

wshI9tapSTcdw

ItdNbRsstphatCtpcird

oiunr

eutorfvcisrr

InassrrcstFcicdah

Shhfc

ICcsutC

ss

14 S. Arora et al.

ause xanthochromia is the result of a process that onlyccurs in vivo, its presence is more specific for SAH thanhe presence of RBCs (8,9). Vermeulen et al. found thatn patients with SAH, xanthochromia is always present inSF supernatant if it is collected between 12 h and 2eeks from time of headache onset (i.e., sensitivity00%) (10). Basing their diagnostic algorithm on thistudy, many physicians consider the absence of xantho-hromia in CSF collected in the appropriate time intervalo be sufficient to exclude the diagnosis of SAH.

However, in the Vermeulen study, all CSF samplesere analyzed by spectrophotometry (10). A telephone

urvey performed in 2000 found that only 8 of 805ospitals used spectrophotometry to evaluate CSF (11).n 2001, a mail survey of 3500 hospitals reported that9.7% of laboratories that responded used visual inspec-ion, leaving very few that used spectrophotometry tonalyze CSF supernatant color (12). Hence, it seems thathysicians may be basing their approach to diagnosis ofAH on an examination that most hospitals do not have.o evaluate the sensitivity of CSF supernatant xantho-hromia as determined by visual inspection, we con-ucted a retrospective chart review of patients diagnosedith SAH over a 12-year period.

MATERIALS AND METHODS

n this retrospective study, a database was created con-aining all patients evaluated in the ED of a large aca-emic urban tertiary care hospital from June 1993 toovember 2005 with an ED discharge diagnosis of SAHased on International Classification of Diseases, Ninthevision codes from an internal hospital database. A

tructured chart review was then performed on the sub-et of charts with a procedure charge for “lumbar punc-ure” or for “spinal tap,” obtained by identifying theresence of the corresponding procedure codes from theospital billing records. Patients were excluded fromnalysis if the official laboratory report of CSF superna-ant color (determined by visual inspection of spun-downSF samples at our facility) was not available, if the time

o LP was � 12 h or � 2 weeks from headache onset aser current recommendations in clinical practice, or if aonfirmatory imaging study (specialist over-read of thenitial CT interpretation, CT angiogram, or magneticesonance angiography) demonstrating SAH was notocumented.

Data collected included: date and time of headachenset, date and time of LP, CSF color, and the results ofmaging studies performed. The nursing triage note wassed to determine the date and time of headache onset. Ifot documented on the triage note, this information was

ecorded from the physician note. If not documented in u

ither of these two notes, the date and time of triage wassed as a proxy for the time of headache onset. Date andime of LP as well as CSF color were recorded from thefficial printed laboratory results. If no official imagingeport could be found, imaging results were recordedrom the physician note. The first 15 charts were re-iewed by two authors (SA, VD), using the same dataollection form to ensure inter-rater reliability, availabil-ty of study data in the pre-defined locations, and con-istency of overall data collection before reviewing theemainder of the charts, which were reviewed by a singleeviewer (VD).

RESULTS

n total, 1323 patients were identified who had a diag-osis of SAH made in the ED, and 102 of these also hadbilled procedure for the collection of CSF. Of this

ubset, 81 paper charts were available for review. As pertudy protocol, 19 charts were excluded because CSFesults were not available, and 16 were excluded asesults were documented only on the physician note andould not be verified by an official laboratory report ofupernatant color. One patient was excluded because theime from headache onset to CSF collection was � 12 h.inally, 26 additional subjects were excluded because noonfirmatory imaging study was documented in the med-cal record to corroborate the ED diagnosis. Of the 15harts that were subjected to inter-rater reliability testing,ata collected for CSF results, time from headache onset,nd presence or absence of a confirmatory imaging studyad 100% agreement between both raters.

The final study set was comprised of 19 patients withAH who met all inclusion criteria. Of these, 47% (9/19)ad xanthochromic CSF supernatant, and 53% (11/19)ad colorless CSF supernatant, resulting in a sensitivityor CSF visual inspection in our sample of 47.3% (95%onfidence interval [CI] 24.4%–71.1%) (Figure 1).

DISCUSSION

n the current study, it does not seem that examination ofSF for xanthochromia via visual inspection is suffi-iently sensitive to rule out the diagnosis of SAH as atand-alone test. Because the vast majority of hospitalsse visual inspection without another confirmatory test,his calls into question the usefulness of a clear, colorlessSF supernatant in clinical decision-making.

Other authors have shown that the use of visual in-pection in lieu of spectrophotometry decreases the sen-itivity of CSF analysis for xanthochromia to potentially

nacceptable levels. Sidman et al. conducted a study

uif(9si5hmCibasr

ivmehbiseaboowce

wsnecs

(oCtdtspio

OtpdEwtC8cirnaptv

lfiMdcrcs

F gnosis

Xanthochromia by Visual Inspection in Subarachnoid Hemorrhage 15

sing artificially lysed RBCs in distilled water compar-ng visual and spectrophotometric xanthochromia andound the sensitivity of visual inspection to be 26.6%95% CI 16%–38%) vs. a sensitivity of 97.9% (95% CI5%–100%) for spectrophotometry (13). An in vivotudy by Perry et al. found the sensitivity for visualnspection of CSF supernatant in patients with SAH to be0% (95% CI 3.0%–81%) (14). It is important to note,owever, that in this study only 2 patients had SAH,aking the reported sensitivity number unreliable.ruickshank et al. published two case reports of patients

n whom initial visually inspected CSF was reported aseing colorless, and then subsequent spectrophotometricnalysis showed xanthochromia (15). In both cases, sub-equent cerebral angiography demonstrated ruptured ce-ebral aneurysms.

We do not know whether the CSF samples that werenterpreted in the current study as being colorless byisual inspection would have demonstrated xanthochro-ia using spectrophotometry. Although spectrophotom-

try alone has been found to be 100% sensitive, mostospitals in this country continue to use visual inspectionecause spectrophotometry is plagued by a lack of spec-ficity (10). Wood et al. recently reported on 253 con-ecutive patients undergoing both CT and LP for thevaluation of SAH using spectrophotometry for CSFnalysis (16). They found a sensitivity of CSF analysisy spectrophotometry to be 100%, but found a specificityf 75.2%. This results in a positive predictive value ofnly 3.3%. These values imply that a wholesale nation-ide switch to spectrophotometry as a screening tool

ould result in an increase in costly negative and unnec-ssary angiographic studies.

The final sample size obtained for review in our studyas small, but this is due mainly to the relatively high

ensitivity of CT scan alone, which often obviates theeed for LP. Future single-institution studies may haveven greater difficulty generating a significant number ofases as newer generation CT scanners are increasingly

igure 1. Chart review results of patients with discharge dia

ensitive, resulting in less false negatives that require LP s

6). In addition, many of the charts were excluded sec-ndary to a deficiency of official laboratory reports ofSF supernatant color and documentation of confirma-

ory imaging studies in the medical record. However,espite the small total numbers in our current study andhe large confidence intervals on the sensitivity, the re-ults do support a conclusion that, unlike with spectro-hotometry, visual inspection of CSF for xanthochromias not sufficiently sensitive to be used in isolation to ruleut SAH.

LIMITATIONS

ur study has several limitations. Due to the retrospec-ive study design, we do not know the total number ofatients evaluated for SAH in the ED, including thoseischarged home after negative CT and CSF analysis.ven if all the missing and otherwise excluded patientse identified in our study are assumed to be true posi-

ives, the best-case sensitivity for visual inspection ofSF supernatant in our sample is 90.2% (95% CI2.7%–95.5%). The calculated sensitivity for xantho-hromia by visual inspection in our sample does improvef all those patients were included, but still could nevereach 100%. Importantly, even the small number of falseegatives seen in our sample is enough to invalidate thepplicability of results from prior work using spectro-hotometry in this area, demonstrating a perfect sensi-ivity of xanthochromia in SAH to hospitals that rely onisual inspection (10).

Due to a priori exclusion criteria, many patientsacked the information necessary to be included in thenal data analysis, which greatly reduced our numbers.ost of these patients were excluded due to a lack of

ocumentation of an official laboratory report of CSFolor. Although the decision to exclude these chartsemoved over a third of our initial sample, this is ofritical importance as it is only the spun-down CSFupernatant from the laboratory that is relevant in this

of SAH.

tudy and we did not want to add in the subjectivity of a

ps

dpfCfbiewta

taCcwfpctuCd

hcistlma

TdatSslNwma

aCXswfrd

1

1

1

1

1

1

1

16 S. Arora et al.

hysician’s impression of CSF color when calculatingensitivity.

Another limitation imposed by the retrospective studyesign is that we cannot know which elements of theatients’ presentation prompted the treating physician tourther investigate those cases where both the CT andSF analysis for xanthochromia were negative. Upon

urther review, it seems that in some cases a large num-er of RBCs in the CSF was the inciting factor for furthermaging. In this same line of thought, it would be inter-sting to have follow-up information on all the patientsho were discharged after these two tests were negative

o see if any of them re-presented at another hospital withdocumented SAH.In addition, in the initial data collection we ensured

hat each patient included in the final analysis had eitherspecialist over-read of the initial CT interpretation, aT angiogram, or a magnetic resonance angiography toonfirm the diagnosis of SAH. Unfortunately, althoughe did record that one of these studies was done, we

ailed to record which specific study was done for eachatient. When we attempted to re-review the paperharts, many were unavailable, making the collection ofhese data essentially impossible. Due to this, we arenable to present a table listing each patient with theirSF findings and confirmatory study, which would un-oubtedly strengthen the conclusions of this study.

We chose to prioritize nursing documentation ofeadache onset time over physician documentation be-ause the nurse was the first health care team member tonterview the patient. This was done primarily for con-istency, as well as the feeling that the patient’s percep-ion of the headache duration may change and becomeess accurate with the passage of time. This decision wasade before the review and does not seem to have

ltered our results significantly.

CONCLUSION

he criteria for accurate interpretation of CSF in theiagnostic evaluation of SAH remain ill defined. As CTnd other imaging technologies evolve, it is probable thathe need for CSF analysis will decrease. Boesiger andhiber looked at the diagnostic potential of more recent CTcanners and found their sensitivity to be 100% in a popu-ation of 170 ED patients being worked up for SAH (6).onetheless, it is unlikely that the need for CSF analysisill disappear completely, as it represents the most directethod of detecting SAH. In the absence of universally

ccepted criteria for interpretation, clinicians should be

ware of the method by which their laboratory analyzesSF and the pitfalls in misinterpretation of these results.anthochromia as reported by visual inspection lacks the

ensitivity necessary to be used as a stand-alone test and,ithout spectrophotometry, physicians must take other

actors such as history, physical examination, and CSFed blood cell counts into account when making clinicalecisions in patients with suspected SAH.

REFERENCES

1. Vinson DR. Treatment patterns of isolated benign headache in USemergency departments. Ann Emerg Med 2002;39:215–22.

2. Mayberg MR, Batjer HH, Dacey R, et al. Guidelines for the manage-ment of aneurismal subarachnoid hemorrhage: a statement for health-care professionals from a special writing group of the stroke council,American Heart Association. Stroke 1994;25:2315–28.

3. Neil-Dwyer G, Lang D. ‘Brain attack’—aneurysmal subarachnoidhemorrhage: death due to delayed diagnosis. J R Coll PhysiciansLond 1997;31:49–52.

4. Fogelholm R. Subarachnoid hemorrhage in middle Finland: inci-dence, early prognosis, and indications for neurosurgical treatment.Stroke 1981;12:296–301.

5. Kassell NF, Torner JC, Haley EC Jr, Jane JA, Adams HP, Kong-able GL. The international cooperative study on the timing ofaneurysm surgery. Part 1: overall management results. J Neurosurg1990;73:18–36.

6. Boesiger BM, Shiber JR. Subarachnoid hemorrhage diagnosis bycomputed tomography and lumbar puncture: are fifth generationCT scanners better at identifying subarachnoid hemorrhage?J Emerg Med 2005;29:23–7.

7. Marton KI, Gean AD. The spinal tap: a new look at an old test.Ann Intern Med 1986;104:840–8.

8. Edlow JA, Caplan LR. Avoiding pitfalls in the diagnosis of sub-arachnoid hemorrhage [see comments]. N Engl J Med 200;342:29–36.

9. Fishman R. Composition of the cerebrospinal fluid. In: Cerebro-spinal fluid in diseases of the central nervous system, 2nd edn.Philadelphia, PA: WB Saunders; 1992:183–252.

0. Vermeulen M, Hasan D, Blijenberg BG, Hijdra A, van Gijn J.Xanthochromia after subarachnoid haemorrhage needs no revisi-tation. J Neurol Neurosurg Psychiatry 1989;52:826–8.

1. Judge B. Laboratory analysis of xanthochromia in patients withsuspected subarachnoid hemorrhage: a national survey [abstract].Presented at: American College of Emergency Physicians Scien-tific Assembly, Philadelphia, PA, October 23, 2000.

2. Edlow JA, Bruner KS, Horowitz GL. Xanthochromia: a survey oflaboratory methodology and its clinical implications. Arch PatholLab Med 2002;126:413–5.

3. Sidman R, Spitalnic S, Demelis M, Durfey N, Jay G. Xanthochro-mia? By what method? A comparison of visual and spectrophoto-metric xanthochromia. Ann Emerg Med 2005;7:51–5.

4. Perry JJ, Sivilotti ML, Stiell IG, et al. Should spectrophotometry beused to identify xanthochromia in the cerebrospinal fluid of alertpatients suspected of having subarachnoid hemorrhage? Stroke2006;37:2467–72.

5. Cruickshank A, Beetham R, Holbrook I, et al. Spectrophotometryof cerebrospinal fluid in suspected subarachnoid haemorrhage.BMJ 2005;330:138.

6. Wood MJ, Dimeski G, Nowitzke AM. CSF spectrophotometry in

the diagnosis and exclusion of spontaneous subarachnoid hemor-rhage. J Clin Neurosci 2005;12:142–6.