Background

Patients presenting with severe, sudden onset headaches can present a challenge to Emergency Physicians. While most headaches are benign, a minority of them are  a symptom of aneurysmal subarachnoid headaches (aSAH); a rare but life-threatening emergency. The traditional workup for diagnosing an aSAH includes a noncontrast CT scan of the head (NCHCT), followed by a lumbar puncture (LP) if the initial NCHCT is negative. In fact, the most recent American College of Emergency Physician (ACEP) clinical guideline, from 2008, clearly states a level B recommendation to perform an LP if the initial NCHCT is negative in a patient being worked up for aSAH.

More recently, however, there is a body of evidence demonstrating that if a patient presents within 6 hours of the onset of headache symptoms, a NCHCT-alone approach is 100% sensitive in excluding an aSAH (Perry 2011, Backes 2012). This approach is promising as it would obviate the need for an LP, an uncomfortable and invasive procedure.

Subsequent research since these two papers, however, has failed to replicate these results (Mark 2013) suggesting that a CT-only approach may miss some aSAHs that are ultimately caught by either LP or MRI. Many questions remain about why followup studies have failed to replicate the high sensitivity of NCHCT for ruling out aSAH in these original papers. One opinion is that the lower sensitivity is due simply to misreading the NCHCT, i.e. a false negative NCHCT scans. In other words, NCHCT is a sufficiently sensitive modality, and that the rare scenario in which NCHCT misses an aSAH that presents within 6 hours of symptom onset is due to radiologist error — the CT shows signs of SAH, it is just being missed.

Clinical Question

Was the sensitivity of NCHCT for the detection of SAH < 100% due to misinterpretation of the NCHCT?

Population

Patients from the Kaiser Permanente Northern California health system with an ICD-9 code of SAH who were > 18 years of age and did not have a prior diagnosis of aSAH. Patients were then included only if they had a diagnosis of aSAH in a Kaiser Permanente Northern California emergency department, if the final NCHCT radiology report was read as negative for SAH, and if there was angiographic evidence of aneurysm that resulted in surgical treatment.

Ultimately, this left us with a patient population of 18 patients who had true aSAH in whom the final radiology interpretation of the NCHCT was negative for signs of SAH.

Outcome: The false negative rate for NCHCT performed within 6 hours of symptom onset in the evaluation of SAH in the Emergency Department (ED).

Design

Patients were identified by a retrospective chart review via ICD-9 codes.
The NCHCT images for the 18 patients with true aSAH missed on original NCHCT were compared with a control group of 7 patients with suspected SAH whose NCHCTs were read as negative and LPs were also negative.

These 25 image sets were presented to two blinded board certified neuroradiologists who were asked to interpret the images. The neuroradiologists were given the indication of the study (“headache”).

The two neuroradiologists were asked to classify the 25 NCHCTs as either: definite evidence of SAH (hyperattenuation (blood) in the basilar cisterns, refluxed blood in the fourth ventricle), probable evidence of SAH (isoattenuation in the basilar cisterns with or without hydrocephalus) or no SAH (hypoattenuation (CSF) in the basilar cisterns)

Excluded

Reported trauma within the previous 24 hours, patients < 18 years of age, prior diagnosis of aSAH, absence of any health plan for the 2 weeks preceding the diagnosis.

Primary Results

A summary of the primary results is presented in the table below. The distinction is made between NCHCT scans performed within 6 hours of the onset of symptoms and those performed outside of that time frame.

Case

Headache onset to CT (h)

Neuroradiologist 1

Neuroradiologist 2

1

<6

No

No

2

<6

Definite

No

3

<6

Definite

Definite

4

<6

Definite

Definite

5

<6

Probable

Definite

6

<6

Probable

Definite

7

<6

Probable

Probable

8

>72

No

No

9

>72

No

No

10

>72

No

No

11

>72

No

Probable

12

>72

No

Probable

13

>72

Definite

Definite

14

48-72

Definite

Definite

15

>72

Probable

No

16

48-72

Probable

Probable

17

48-72

Probable

No

18

48-72

Probable

Probable

Critical Results

  • False Negative NCHCT < 6 hours
    • 5/7 scans performed within 6 hours
    • Both Neuroradiologists agreed that these CTs were positive for SAH
    • In 1 additional scan (< 6 hours), there was blood in the basilar cisterns that one radiologist interpreted as definite SAH and the other felt it was a tentorial SDH
    • In the final early CT, there were aneurysm clips from prior SAH that obscured CT signs of SAH. 
  • True Negative (control NCHCTs)
    • All read as negative by the Neuroradiologists
    • Specificity of NCHCT high

Concordant Definite or Probable

Control (n=7)

0/7 (0%)

SAH CT < 6h (n=7)

5/7 (71%)

SAH CT > 6h (n-11)

4/11 (36%)

All SAH (n=18)

9/18 (50%)

Strengths

  • Two independent neuroradiologists reading CTs, leading to accurate reads of the scans
  • Radiologists were blinded to the aim of the study
  • Database of 21 hospitals over the course of 6 years provided a relatively large cohort of aSAH patients (despite such a clinically rare phenomenon)
  • The study included scans of actual patients with CT-negative SAH (this is the population we want to know where CT-only approach went wrong)

Limitations

  • Case collection and identification was retrospective and may have missed pertinent cases.
  • Raters were two board certified neuroradiologists, this may not reflect the real world where neuroradiologists are often not available in the ED
  • The study setting is different from a real world setting and may not be applicable to a real world situation. For example, the study setting does not take into account time-constraints, erroneous indications, and distractions that exist in the real world. Also, there are no consequences to “overcalling” a CT read in this setting i.e. the raters may have been more willing to call a NCHCT positive than they might have in the real world.
  • Only 18 cases (while this is a large number for this type of rare clinical phenomenon, it is still a small number of cases)

Author's Conclusions

“CT evidence of SAH was frequently present but unrecognized according to the final radiology report in cases of presumed CT-negative aSAH. This finding may help explain some of the discordance between prior studies examining the sensitivity of cranial CT for SAH.”

Our Conclusions

When sensitivities of NCHCT for diagnosing aSAH within 6 hours of symptom onset are reported to be low (i.e., 97% or lower), the problem may not be in the modality of NCHCT, but in misinterpretation of these NCHCTs themselves.

Potential Impact To Current Practice

While this study alone should not affect how one clinically manages the workup of aSAH, it strongly suggests that NCHCT is highly sensitive for this diagnosis. This further corroborates the conclusions of the original papers, that, if performed within 6 hours of symptom onset is, a CT-only approach in ruling out aSAH may be sufficient. However, the results emphasize the importance of radiologist reading accuracy since the imaging test is only as good as the person reading the images.

Bottom Line

This paper, in conjunction with the papers before it, should be used to reassess the ACEP clinical practice guidelines recommending that NCHCT must always be followed by LP in the workup of suspected aSAH presenting within 6 hours of symptom onset if the CT is read by a qualified radiologist and interpreted accurately.

Read More

Read More

REBEL EM: Does a Normal Head CT Within 6 Hours of Onset of Headache Rule Out SAH?

The SGEM: SGEM #48: Thunderstruck (Subarachnoid Hemorrhage)

References:

Backes D et al. Time-dependent test characteristics of head computed tomography in patients suspected of nontraumatic subarachnoid hemorrhage. Stroke. 2012; 43(8):2115-9. PMID: 22821609

Mark DG et al. Nontraumatic subarachnoid hemorrhage in the setting of negative cranial computed tomography results: external validation of a clinical and imaging prediction rule. Ann Emerg Med. 2013; 62(1):1-10.e1. PMID: 23026788

Perry JJ et al.  Sensitivity of computed tomography performed within six hours of onset of headache for diagnosis of subarachnoid haemorrhage: prospective cohort study. BMJ 2011; 343: d4277. PMID: 21768192