the diagnosis of subarachnoid hemorrhage rob hall pgy2 emergency medicine january 10th, 2002
TRANSCRIPT
The Diagnosis of
SUBARACHNOIDHEMORRHAGE
Rob Hall
PGY2 Emergency Medicine
January 10th, 2002
SubarachnoidHEMORRHAGE
OBJECTIVESWhat does the Literature Say?
• What is the sensitivity of CT?
• When should the LP be done?
• What is a positive LP?
• Should we be using spectophotometry?
• Does a -ve CT and -ve LP rule out SAH?
• Is lumbar puncture without CT safe?
The Diagnosis of SUBARACHNOID HEMORRHAGE
•WHY?• Misdiagnosis
• Mortality
• Angiography
• Literature
SENSIVITITY OF CT
L ite ra tu re R eview o f3 rd G en era tion C T scan n ers
S id m an 1 9 9 6 S am es 1 9 9 6
van d er W ee 1 9 9 5 M org en s te rn 1 9 9 8
K asse l 1 9 9 0
Reading Literature on CT and SAH
• Generation
• Technique
• Radiologist
• Time
• Prospective
• Spectrum Bias
SPECTRUM BIAS
• POSITIVE CT NEUROSURG
WARD
• POSITIVE CT EMERG PT (R/O SAH)
TIMING OF CT:Kassel 1990
Cooperative Aneurysm Study
0102030405060708090
100
day 0 day 1 day 2 day 3 day 4 day 5
Sensitivity of CT
Sensitivity of CT
• Sidman 1996– Retrospective
– Sensitivity 100% < 12hrs
• Sames 1996– Retrospective
– Sensitivity 93% < 24hrs
• SPECTRUM BIAS
• IGNORE RESULTS
van der Wee 1995Journal of Neurology, Neurosurgery, and Psychiatry
• Prospective, N=175
• All <12hrs with SUDDEN ONSET H/A
• 3rd generation CT done < 12hrs
• Neuroradiologist + 2 general radiologists
• Diagnosis of SAH in -ve CT – LP > 12 hrs– Xanthochromia by spectophotometry
Van der Wee 1995
• CT +ve in 117/175 ---->LP +ve in 2/58
• Sensitivity < 12 hrs 98% (94.0 - 99.8)
• Comments– probably the best study available– good diagnosis of SAH in CT -ve group– rate of SAH 68% (?spectrum bias)– note CI as low as 94%– note who read the films
Morgenstern 1998Annals of Emergency Medicine
• Prospective, N = 107
• Spectrum: “worst headache ever”
• Only 107 enrolled of eligible 170
• 3rd generation CT
• 2 Neuroradiologists
• Diagnosis of SAH in -ve CT questionable
Morgenstern 1998:Annals of Emergency Medicine
• Diagnosis of SAH in CT -ve patients:
• rbcs > 1000 in 1st tube and no decrease of 25% w.r.t. 3rd tube PLUS one of ……– visual xanthochromia– spec xanthocrhomia– positive d-dimer
Morgenstern 1998:Annals of Emergency Medicine
• Negative Predictive Value Overall
– 97.5% (CI 91.2 - 99.7%)
• Sensitivity < 24hrs
– 93% (no confidence interval)
Morgenstern 1998:Annals of Emergency Medicine
• Only 107/170 enrolled• 10% refused LP• 25% of LPs < 12hrs• Questionable
definition of +ve LP – 20 patients with +ve
spec defined as NO SAH, but no problems at 2yr follow up
• What if they missed even ONE patient < 24hrs
– 14/15 = 93.3%
– 13/15 = 86.7%
SUMMARY ON CT AND SAH
• TRUE sensitivity UNKNOWN
• EARLIER is BETTER• PGY2 versus NEURORAD• BEST estimate of sensitivity
– 0-24hrs 95%
– 24-48hrs 85%
– 48-72% 75%
• TO RULE OUT SAH, A LUMBAR PUNCTURE IS REQUIRED AFTER A NEGATIVE CT HEAD
LUMBAR PUNCTUREPathophysiology
• RBCs passively lysed and oxidized to OXYHEMOGLOBIN – detectible as early as 2 hrs (Barrows 1955)
• Oxyhemoblobin is actively converted to BILIRUBIN by hemoxidase enzyme found in choriod plexus and leptomeninges – bilirubin present by 6hrs (Barrows 1955)– max hemoxidase activity by 12hrs (Roost 1972)
What is Xanthochromia?
• The change in coloration of CSF
• Due to oxyhemoblobin, bilirubin, or methemoglobin
• BUT ----------> detected by VISION or SPECTOPHOTOMETRY?
What can cause a false +ve LP for Xanthochromia?
• Jaundice
• Rifampin
• Previous traumatic LP
• Traumatic tap -----> CSF sits in lab > few hours
TIMING OF LUMBAR PUNCTURE
• Barrows 1955: oxyHb in 2hrs, bili in 6hrs
• Roost 1972: hemoxidase max at 12hrs
• Where does the 12hr delay come from?– Vermeulen 1989– Walton 1956
TIMING OF LP:Vermeulen 1989
Journal of Neurology, Neurosurgery, and Psychiatry
• Retrospective review of 111 patients with SAH diagnosed by blood on CT
• ALL lumbar punctures done > 12hrs
• Xanthochromia detected by spectophotometry
• DOES NOT LOOK AT CT -ve PATIENTS
• DOES NOT LOOK AT LP < 12hrs
TIMING OF LP:Vermeulen 1989
Journal of Neurology, Neurosurgery, and Psychiatry
• TIMING SENSITIVITY– 12hrs - 2weeks 100%
– 2 - 3 weeks 91%
– 3 - 4 weeks 71%
TIMING OF LP:Walton 1956
Subarachnoid Hemorrhage. E & S Livingstone LTD.
• Retrospective look at 256 cases of SAH
• How was SAH diagnosed?
• Xanthochromia detected visually.
• Some results missing
TIMING OF XANTHOCHROMIA:Walton 1956
0102030405060708090
100
0-2hr 2-4hr 4-6hr 6-12hr 12-24hr
>24hr
XanthochromiaBlood
What is a positive LP?
• RED BLOOD CELL COUNT?– LEAK versus MAJOR HEMORRHAGE
– NO literature (Tourtelloote 1964 - none < 1000/mm3)
– How can you tell from traumatic tap?
SAH versus Traumatic LP
• Opinion, crenated rbcs, erythrophages, d-dimer unreliable
• Repeat LP only helpful if clear
• FOUR tube method UNRELIABLE and does not detect SAH + traumatic tap– Vandermeulen 1996– Buruma 1981
SAH versus Traumatic LP
• XANTHOCHROMIA is the only way to reliably distinguish between SAH and traumatic LP
So how should we detect Xanthochromia?
• Visual detection?– ? Poor
sensitivity
• Spectophoto-metry?– ? Poor
specificity
LITERATURE REVIEW:Visual versus spectophotometric
detection of Xanthochromia
X an th och rom ia
S od ers trom 1 9 7 7 M acD on a ld 1 9 8 8
V isu a l d e tec tionIN S E N S ITIV E
M org en s te rn 1 9 9 8 F oo t 2 0 0 1
S p ec top h o tom etryN O N S P E C IF IC
L ite ra tu re R eview
Visual detection of Xanthochromiais INSENSITIVE:
Soderstrom 1977:
• N=32
• 12 ICH, 12 SDH, 8 SAH
• Dx by CT + OR, angiogram, or autopsy
• Vision detected 16 of 32 cases of xanthochromia on spectophotometry– Sensitivity 50%– ?when spec done
Visual detection of Xanthochromiais INSENSITIVE:
MacDonald 1988
• Retrospective review of 61 patients with angiographically proven aneurysms who had LP done
• 28/61 had xanthochromia by vision for sensitivity of 46%
• 13 LPs were done < 24hrs (any < 12hrs?)– exlcude LP < 24hrs….28/48 ------> 67%
Spectophotometry is NONSPECIFIC: Morgenstern 1998
• 18 patients with +ve spectophotometry who didn’t meet their criteria for +ve LP
• Followed for 2 years with no problems
Spectophotometry is NONSPECIFIC: Foot 2001
• Retrospective study looking at the use/results of CT and LP in ED r/o SAH
• +ve Xanthochromia = > 0.02 absorbance units at 415nm
• 21 CT-ve, Xanthochromia +ve– 19 angiograms normal– 2 aneurysms
Xanthochromia
• Cruikshank 2001– “A prospective study of LP on CT -ve patients
undergoing r/o SAH to compare visual and spectophotometric detection of xanthochromia has never been done”.
UNPUBLISHED DATA:J. Croft, G. Sutherland, A. Gibb
• ALL CSF samples from calgary ED over a 14 months period
• R/O SAH in 110• Recorded
– rbcs count– visual xanthochromia– spectophotometry absorption peak– spectophotometry optical density criteria
LOCAL DATAN Blood
(<10)Visual Spec Peak Spec O.D.
CriteriaSAHdx
71 - - - 20+,51- 0
11 + - - 5+,6- 0
4 + + Hb/Bili 4+ 4
14 +/- 3+11-
Hb 12+,2- 0
9 +/- 1+8-
Bili 9+ 1
Optical Density Criteria
• SAH No SAH• OD+ve 5 45 • OD-ve 0 60
• Sensitivity– 5/5 = 100%
• Specificity– 60/105 = 57%
• NPV– 60/60 = 100%
• PPV– 5/50 = 10%
Visual Detection
• SAH NO SAH• Visual 5 4
xantho• No visual 0 101
xantho
• Sensitivity– 5/5 = 100%
• Specificity– 101/105 = 96%
• NPV– 101/101= 100%
• PPV– 5/9 = 56%
LOCAL DATA
• CONCLUSIONS– Visual Xanthochromia
did NOT miss any SAH
– Spectophotometry was not specific for SAH
• COMMENTS– NO gold standard for
SAH diagnosis
– NO long term f/u to prove that SAH wasn’t missed
– Small numbers
– One missed SAH ---> 5/5 to 5/6 ---> 83%
Summary on Lumbar Puncture
• LP isn’t perfect either
• LP should be done > 10 - 12 hrs (spectrum)
• Xanthochromia is only way to reliably distinguish SAH from traumatic tap
• Literature is unclear whether visual or spectophotometric detection of xanthochromia is superior
Does a negative CT and LP rule out SAH?
• Evidence this does NOT occur….– van Ginn1988: 71 patients
followed to 3 years, no problems
– Markus 1991: 16 patients followed to 20 months, no problems
– Harling 1994: 14 pts followed 18-30 mo, no problems
• Evidence that this DOES occur…..– Nine case reports
in literature
– Some had LP < 12hrs
– Some used visual, some used spec
PERSPECTIVE ON POSSIBLE MISSED DIAGNOSIS RATE
2 0 0 ,0 0 0 E D V is its in C a lg ary p er year
L P 5 0 % sen s it ive5 m issed S A H
5 /2 0 0 0 = 0 .2 5 %
L P 9 0 % sen s it ive1 m issed S A H
1 /2 0 0 0 = 0 .0 5 %
C T 9 5 % sen s it iveC T m isses 1 0
R /O S A H in 1 /1 0 H ead ach es2 0 0
H ead ach e 1 % o f E D vis its2 ,0 0 0
Does a -ve CT and LP rule out SAH?
• If LP done > 12hrs --------->
YES• Risk of angiogram > chance of
SAH with -ve CT and LP
DIAGNOSTIC ALGORITHM
R /O S u b arach n o id H em orrh ag e
D iag n os is = S A H
V isu a l X an th och rom ia
V ery s tron gc lin ica l su sp ic ion
con su lt N S xcon s id er an g io
R u led ou t S A HD /C h om e
N o N S x con su lt
N o V isu a l X an th och rom ia
C T h eadL P > 1 0 - 1 2 h rs
LP without CT
• Why wouldn’t you want to do this?– Risk of herniation– CT provides much additional information – How do you know this is a SAH?– How do you know there aren’t complications of
the SAH that increase the risk of herniation
LP without CT
• Herniation– Mass effect from hematoma or hydrocephalus
with a SAH, or a different dx (ICH, tumor, etc)
• Normal LOC and NO focal signs – Hillman 1986: 2.2% acute deterioration after
LP; 10% had hematomas associated with SAH– Duffy 1982: 12% with proven SAH (spectrum
bias) deteriorated while needle in back
LP without CT
• Additional information on CT– dx of non-aneurysmal causes: AVM, tumor,
ICH, hypertensive hem, perimesencephalic– look for acute complications: hydrocepahlus,
ICH, intraventricular blood requiring a drain– amount of bleeding is prognostic– bleeding on CT can help localize aneurysm and
identify multiple aneurysms
LP without CT
• How do you know that the acute H/A isn’t due to an intraparenchymal hemorrhage?– Van Gijn 1980: retrospective review of all
patients with initial dx as SAH – 15% had intraparenchymal hemorrhage– 8% were in posterior fossa
LP without CT
• Summary– There is NO literature supporting LP without
CT (Schull 1999: model only)– There is SOME literature documenting the risks – Risks and lack of additional information are not
justified in a tertiary care center– May be reasonable in periphery if no access to
CT although transfer in for CT is preferable