There's been a bit of a healthy debate regarding the sensitivity of a negative non-contrast head CT for the diagnosis of subarachnoid hemorrhage. At least one prominent Emergency Physician educator used some time at ACEP this year to describe why it wasn't time to rely solely on imaging, while a second prominent EP used his lecture time to say essentially the opposite.
This publication, in Annals of Emergency Medicine, tries to address this question and determine just how frequently a negative CT misses a subarachnoid hemorrhage from occult aneurysmal leakage. Over 11 years, in 21 Emergency Departments, they were able to identify 55 cases in which CT missed a subarachnoid hemorrhage – some of whom had negative CTs within six hours of headache onset.
This study suffers as studies tend to suffer from the retrospective nature of chart review, from missing and imputed data, and from a small sample size despite the extensive time frame reviewed. Overall, though, I would score it as a point for the "still needs LP" crowd – clearly, you will miss some SAH by foregoing LP. But, modern CT scanners may have better sensitivity than the ones included from the early part of this decade-long review – and the false-positives and harms from false-positive LPs grow ever-closer to the the false-negatives from CT.
Still not a perfect argument in either direction.
"Nontraumatic Subarachnoid Hemorrhage in the Setting of
Negative Cranial Computed Tomography Results: External
Validation of a Clinical and Imaging Prediction Rule"