Response to Letter by Beetham and Lhatoo
We are not surprised at the comments expressed by Drs Beetham and Lhatoo, outspoken advocates of a diagnostic test which is rarely used in North America. It is disingenuous, however, to claim that any diagnostic test is “owed” to patients before a demonstrated improvement in clinical outcomes. Although we acknowledge that of the 4 definitions of xanthochromia we studied, the UKNEQAS definition performed best in terms of sensitivity and specificity, our results did not support the use of any of the spectrophotometric definitions of xanthochromia. As discussed in the article, by using 4 discrete absorbances rather than a continuous scan to estimate the baseline, our method would err on the side of overestimating the specificity of the UKNEQAS definition and therefore represents a best case scenario. We also demonstrated that freezing had only minimal impact on our study results by comparing fresh and frozen results on a subset of our samples.
The central issue from our article is the limited specificity of spectrophotometry for identifying CT-negative patients with aneurysmal leaks. The authors offer unpublished data suggesting that, in their experienced hands with meticulous sample processing and using an unspecified threshold, the test could be deemed negative in nearly 90% of patients. They further suggest that positive tests can be ignored in one third of the remaining patients with a corresponding specificity of up to 94%. We argue that most North American centers would in fact perform an angiogram on all patients with a positive result for xanthochromia, including the 9 patients characterized by Drs Beetham and Lhatoo as having a “viable” but otherwise unspecified “other diagnosis”, resulting in 81% of the angiograms performed be “unnecessary”, rather than the 25% suggested by their letter. Furthermore, we were unable to reproduce such a high specificity using any spectrophotometric definition of xanthochromia in a large, prospective cohort study of the intended target population for the test: patients in whom emergency physicians with ready access to CT scanning deemed a lumbar puncture was necessary to rule out subarachnoid hemorrhage.
Because the prevalence of aneurysmal hemorrhage in these patients approaches 1%,1–5 which is similar to the rate of detection of incidental aneurysms in the population at large, spectrophotometry would be expected to have a real-world positive predictive value of around 5%. In other words, the test would be falsely positive in 19 of 20 patients. One can add this to the litany of reasons why North American clinicians are reluctant to embrace spectrophotometry.6
Our group has presented data recently regarding the possibility of missed subarachnoid hemorrhage in the larger population of neurologically intact emergency patients with acute headache.7 We prospectively enrolled 601 acute headache emergency department patients over a 3-year period from 2 tertiary care academic emergency departments, including 64 positive subarachnoid hemorrhages. All subarachnoid hemorrhages were identified with our current strategy of CT then cerebrospinal fluid analysis for visual xanthochromia or erythrocyte count. We conducted long-term follow-up of the patients with negative CT/cerebrospinal fluid analysis to determine whether any of these patients had a subsequent subarachnoid hemorrhage which could be consistent with a rebleed. Our mean follow-up time was 477 (interquartile range, 193, 755) days after initial emergency department visit. Our sensitivity of this testing strategy for subarachnoid hemorrhage was 100% (95% CI, 94 to 100), specificity 66% (95% CI, 62 to 70), negative predictive value 100% (95% CI, 98 to 100), positive likelihood ratio 2.936, negative likelihood ratio 0.0015. These findings from an adequately powered prospective study argue that we can dispense with adding spectrophotometry to our current diagnostic strategy in this group.
We applaud the call for evidence-based guidelines to direct care of these patients. However, before incorporating a technically difficult diagnostic test involving an expensive machine which reports a result that is both labile and difficult to interpret, we owe it to our patients to weigh all available evidence in a balanced and objective manner. Advocates of spectrophotometry need to clearly demonstrate what benefits this test offers for the CT-negative patient with visually clear and colorless cerebrospinal fluid, and how that offsets the unnecessary angiography and inevitable identification of asymptomatic aneurysms in patients with otherwise benign headaches.
Perry JJ, Sivilotti MLA, Stiell IG, Wells GA, Raymond J, Mortensen M, Symington C. Should spectrophotometry be used to identify xanthochromia in the cerebrospinal fluid of alert patients suspected of having subarachnoid hemorrhage? Stroke. 2006; 37: 2467–2472.
Perry JJ, Stiell IG, Wells GA, Spacek AM. Historical Cohort Study “Use and Yield of Investigations for Alert Patients with Possible Subarachnoid Hemorrhage”. Can J Emerg Med. 2002; 4: 333–337.
Perry JJ, Spacek AM, Stiell IG, Forbes M, Wells GA, Mortensen M, Symington C, Fortin N. Is a negative CT scan of the head and a negative lumbar puncture sufficient to rule out a subarachnoid hemorrhage? Acad Emerg Med. 2005; 12 : 53.