CTs for SAH - Does Time Even Matter?

/

Vincent, A. et al. Sensitivity of modern multislice CT for subarachnoid haemorrhage at incremental timepoints after headache onset: a 10-year analysis. Emerg Med J 39, 810–817 (2022).

Background

Spontaneous subarachnoid hemorrhage (SAH) is a can’t miss diagnosis for patients presenting to the emergency department with a headache. The diagnosis is associated with a 30% mortality at 30 days, and approximately 30% of survivors may have long-term neurocognitive deficits (Rincon et al., 2013). The majority of spontaneous SAH are secondary to a ruptured arterial aneurysm (80%) while non-aneurysmal SAH are often due to low pressure venous bleeds, arteriovenous malformations, and other more rare causes.

The 2019 ACEP clinical policy states that a non-contrast head CT (NCCT) is sufficient for ruling out SAH when a patient presents <6 hours from headache onset (Godwin et al., 2019). For patients presenting >6 hours, further workup with either a lumbar puncture (LP) or CT angiography (CTA) is recommended due to the concern that the sensitivity of CT decreases over time as blood is degraded and re-circulated in the CSF. This 6-hour rule originated from a prospective multi-center study of 3132 emergency department patients that demonstrated CT had 100% sensitivity at detecting SAH <6 hours from symptom onset vs 87% at >6 hours (Perry et al., 2011). However, this 6 hour cutoff was an arbitrary number decided on by consensus review of prior data, and the study only evaluated a binary cutoff of <6 hours vs >6 hours without any intermediate time points. For instance, the >6 hour group ranged from 8 hours to 8 days from headache onset to CT, which may partly explain the significant difference between the two groups.

Trying to rule out the diagnosis of SAH with LP and/or CTA can cause its own headache with nearly 20% of LPs ending in a traumatic tap with no reliable way to tell the difference between this and a SAH (Godwin et al., 2019). Indiscriminate use of CTA for working up these patients is limited by the fact that 1-2% of the population has a cerebral aneurysm (Lawton and Vates, 2017). This is likely to be an incidental finding in most patients, potentially leading to unnecessary downstream testing and procedures. There is only one study comparing a CT/LP vs a CT/CTA approach for diagnosing SAH in emergency department patients which found no difference between the two strategies. The study was limited by a small sample size of 105 patients with only 5 diagnosed with SAH. This led to wide confidence intervals and limited statistical power (Carstairs et al., 2006).

This ongoing debate has led many providers to reconsider the test characteristics of noncontrast CT, especially given the limited data for time intervals >6 hours in the original Perry study. We also know that CT technology has only continued to improve since the early 2000s when the data for that study was collected (think thinner tissue slices, improved resolution, and better artifact reduction).

The Study

Vincent et al. (2022) hypothesized that there may be improved sensitivity for NCCT in detecting SAH at time intervals >6 hours hours from headache onset, potentially increasing the timeframe where negative CT may be considered sufficient for ruling out SAH.

Methods

This was a retrospective chart review evaluating the sensitivity of initial CT head in diagnosing SAH. All patients presenting to the Christchurch emergency department in New Zealand from 2008-2017 who had a CT head and were ultimately diagnosed with SAH based on ICD-10 codes were eligible. Christchurch is a tertiary care facility and the only acute major referral center in New Zealand with approximately 100,000 ED visits annually. Exclusion criteria included traumatic SAH, unknown day of headache onset, or transfer from an outside facility given difficulty obtaining full records retrospectively.

For patients with exact time of onset unknown (day of onset always known), the authors attempted to define a range of potential sensitivities at each time interval by performing 2 imputed analyses using the longest and shortest potential time intervals from headache onset to CT (00:01 on day of onset vs 30 minutes prior to CT). If a patient was assumed to have presented closer to CT than they actually did, this would potentially make it seem like CT was less sensitive at closer time intervals (maximizing early false negatives).

Imaging was obtained using modern 64- or 128-multislice CT interpreted by both general and neuroradiologists as a part of routine clinical practice. The final radiology report was used for analysis. Based on local practice patterns, if CT was negative, providers considered LP. If LP or NCCT was positive a CTA was performed. If there was ongoing clinical uncertainty an MRI was occasionally performed.

The primary outcome was the sensitivity of noncontrast head CT for detecting aneurysmal SAH at various time intervals from headache onset. The secondary outcome was the sensitivity of noncontrast head CT for anyspontaneous SAH at various time intervals from headache onset.

Point estimates and confidence intervals were calculated using 1000 bootstrapped samples (a technique that performs computer-generated re-sampling of the study population).

Results

347 patients with spontaneous SAH were included in the analysis, 260 with aneurysmal SAH. Exact time of onset was unknown in 64% of patients. When patients with an unknown exact time of onset were assumed to have presented 30 minutes prior to CT CT, the overall sensitivity for aneurysmal SAH at 24 hours from headache onset was 100% (95% CI 98.3-100%) and for all SAH was 99.3% (95% CI 97.5-99.9%). By assuming these same patients presented at longer time intervals from CT, the sensitivity at 24 hours for aneurysmal SAH was 100% (94.8-100) and 99.2% (97.2-99.9) for all SAH.

Table 1: Sensitivity of noncontrast head CT (± 95% confidence intervals) at various time intervals from headache onset in aneurysmal subarachnoid hemorrhage (aSAH) vs. all types of subarachnoid hemorrhage. Patients presenting at an unknown time from symptom onset were either assumed to have presented closer to CT (30 minutes prior, maximizing early false negatives) vs further (00:01 on the day of symptom onset, minimizing early false negatives).

15 patients had a negative CT head who were found to have SAH on LP or MRI. Only 2 of these patients presented within 24 hours (false negative or miss rate of 2/288 or 0.7%), neither of these patients had an aneurysmal SAH, required any specific surgical treatment, or had any symptoms on discharge. Of the remaining 13 patients, all presented >36 hours from headache onset. The authors note that given limited sample sizes for patients presenting >24 hours, these results may not be as reliable.

Discussion/Limitations

This study started with patients already diagnosed with subarachnoid hemorrhage to back-calculate the sensitivity of CT rather than directly evaluating the performance of CT prospectively in an undifferentiated population. While we can calculate sensitivities in this study (assuming we caught nearly all cases of SAH), we would need the whole population of patients originally evaluated for SAH to calculate full test characteristics such a negative predictive value – how well does having that negative CT result rule out SAH - which I think is the ultimate question we’re trying to answer here.

As a retrospective, single-center study, we also do not know if we are missing any systematic biases that could be affecting the results such as a clinical practice pattern that may differ in New Zealand. We also do not know the overall effect from all of the imputed data for the unknown times of onset. A prospective, multi-center study would be helpful to verify these results.  

Finally, many studies, this one included, are primarily focused on the sensitivity of NCCT for aneurysmal SAH under the assumption that non-aneurysmal SAH is a less time-sensitive diagnosis because there is rarely emergent intervention required. These other causes of SAH can still occasionally have important complications such as vasospasm or obstructive hydrocephalus, so it is likely not an appropriate diagnostic strategy to potentially ignore their presence altogether. The safety of this type of approach has not been directly studied. Regardless, the overall miss rate of NCCT for all SAH at 24 hours was low (~0.7% or ~1 in 142) and may be worth a shared decision-making approach based on a patient and provider’s individual risk tolerance.

Conclusion

In this single center retrospective analysis, over the 10-year period from 2008-2017, CT head was 100% sensitive for identifying all aneurysmal SAH when performed <24 hours from headache onset, and 99.3% sensitive for non-aneurysmal SAH (miss rate ~1 in 142). Prospective, multi-center study is needed to verify these results before widespread adoption into clinical practice. However, when evaluating a patient for SAH with a negative CT, this data can lead to more educated shared decision-making discussions about further workup.

References

  1. Rincon F, Rossenwasser RH, Dumont A. The epidemiology of admissions of nontraumatic subarachnoid hemorrhage in the United States. Neurosurgery. 2013 Aug;73(2):217-22; discussion 212-3. doi: 10.1227/01.neu.0000430290.93304.33. PMID: 23615089.

  2. Perry JJ, Stiell IG, Sivilotti ML, Bullard MJ, Emond M, Symington C, Sutherland J, Worster A, Hohl C, Lee JS, Eisenhauer MA, Mortensen M, Mackey D, Pauls M, Lesiuk H, Wells GA. Sensitivity of computed tomography performed within six hours of onset of headache for diagnosis of subarachnoid haemorrhage: prospective cohort study. BMJ. 2011 Jul 18;343:d4277. doi: 10.1136/bmj.d4277. PMID: 21768192; PMCID: PMC3138338.

  3. Lawton MT, Vates GE. Subarachnoid Hemorrhage. N Engl J Med. 2017 Jul 20;377(3):257-266. doi: 10.1056/NEJMcp1605827. PMID: 28723321.

  4. Carstairs SD, Tanen DA, Duncan TD, Nordling OB, Wanebo JE, Paluska TR, Theodore N, Riffenburgh RH. Computed tomographic angiography for the evaluation of aneurysmal subarachnoid hemorrhage. Acad Emerg Med. 2006 May;13(5):486-92. doi: 10.1197/j.aem.2005.12.010. Epub 2006 Mar 21. PMID: 16551778.

  5. Vincent A, Pearson S, Pickering JW, Weaver J, Toney L, Hamill L, Hurrell M, Than M. Sensitivity of modern multislice CT for subarachnoid haemorrhage at incremental timepoints after headache onset: a 10-year analysis. Emerg Med J. 2022 Nov;39(11):810-817. doi: 10.1136/emermed-2020-211068. Epub 2021 Nov 24. PMID: 34819306.

Authorship

  • Written by - Justine Milligan, MD, PGY-3, University of Cincinnati Department of Emergency Medicine

  • Recording and Audio Editing by: Jeffery Hill, MD MEd, Associate Professor, University of Cincinnati Department of Emergency Medicine

Cite As

Milligan, J. Hill, J. (February 2, 2023) CTs for SAH - Does Time Even Matter? TamingtheSRU. https://www.tamingthesru.com/blog/journal-club/cts-for-sah-does-time-even-matter