In reply
Article Outline
We thank Dr. Schwartz for his comments about our article. We agree that emergency physicians should use a Bayesian approach to risk assessment; thereby including the patient's pretest probability of disease and the test characteristics to determine the patient's posttest disease probability. This posttest probability allows physicians to make a risk-benefit assessment of further testing and/or therapy and to have a well-informed conversation with their patients.
There are, however, several problems with these suggestions. We have no scale to assess a patient's pretest probability of disease. Patients with “thunder-clap” headache have an approximately 12% risk of having a subarachnoid hemorrhage as a cause of their headache.1 However, we do not know how other historical or physical exam findings interact to change this probability. Kowalski et al found that we still miss a large of number of patients who are later admitted to neurosurgical intensive care units with subarachnoid hemorrhage and that these patients do poorly when compared to those who are correctly identified on their first encounter.2
Missed subarachnoid hemorrhage can result in neurologic devastation or death. The important risks of lumbar puncture are false positive results and the morbidity of subsequent cerebral angiography. The use of CT angiography lessens these risks.
Their presumption of the pretest probability of disease of 2% appears to be artificially low given that the literature suggests a 1% rate of subarachnoid hemorrhage among all headache patients presenting to the emergency department (ED).3 If we evaluated 25% of all ED patients presenting with headache, which we believe would be a high percentage, this would result in a 4% risk of subarachnoid hemorrhage. Using the suggested Bayesian analysis, one would miss 1 in 250 patients with subarachnoid hemorrhage by not doing the lumbar puncture. Possibly some physicians and patients would find this an acceptably low risk; we would not.
We agree with Dr. Pines that the accuracy of the CT interpretation is paramount when determining the sensitivity of CT scan in detecting spontaneous subarachnoid hemorrhage. We could have had a senior neuroradiologist review all of the cranial CTs; however, this would have reduced the generalizability of our results. What would the sensitivity results mean to a community emergency physician relying on a night-hawk read? In our study CTs were interpreted by community radiologists, radiology residents and academic radiology faculty.
In Perry's cited study,4 the sensitivity of head CT was 100% (95% confidence interval [CI] 94%-100%) but with a sensitivity analysis they were 98% (95% CI 91%-100%). The confidence intervals in our study overlap with these and, therefore, are not significantly different. Additionally, Perry had 60 patients lost to follow up. It is possible that some of the patients were lost to followup because they died from subarachnoid hemorrhage.
We agree with the comments concerning false positive lumbar puncture results as noted above. However, until we have a validated decision-rule for accurately predicting a patient's pretest probability of subarachnoid hemorrhage, which given the prevalence of disease would require a multicenter prospective derivation and then subsequent validation, we continue to perform and recommend lumbar punctures on patients with suspected subarachnoid hemorrhage with a negative non-contrast cranial CT.
References
- Worst headache and subarachnoid hemorrhage: prospective, modern computed tomography and spinal fluid analysis. Ann Emerg Med. 1998;32:297–304
- Initial misdiagnosis and outcome after subarachnoid hemorrhage. JAMA. 2004;291:866–869
- . Avoiding pitfalls in the diagnosis of subarachnoid hemorrage. N Engl J Med. 2000;342:29–36
- Clinical decisions rule to safety rule out subarachnoid hemorrhage in acute headache patients in the emergency department. Acad Emerg Med. 2007;13:S9
Funding and support: By Annals policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article, that might create any potential conflict of interest. The author has stated that no such relationships exist. See the Manuscript Submission Agreement in this issue for examples of specific conflicts covered by this statement.
PII: S0196-0644(08)01632-6
doi:10.1016/j.annemergmed.2008.08.013
© 2008 American College of Emergency Physicians. Published by Elsevier Inc. All rights reserved.
Refers to article:
- The Answer to Imperfect Computed Tomography Sensitivity for Subarachnoid Hemorrhage: Use Clinical Judgment
- Sensitivity Of Computed Tomography For Subarachnoid Hemorrhage
