Date: October 9, 2013
Guest Skeptic: Dr. Jeff Perry. Senior Scientist, Clinical Epidemiology, Ottawa Hospital Research Institute. Associate Professor, Department of Emergency Medicine, Faculty of Medicine, University of Ottawa.
Case Scenario: A 34-year-old woman presents with acute onset of headache peaking in 30 minutes with no recent trauma, focal deficits and a normal neurologic examination.
Question: Can a clinical decision rule be used to rule out subarachnoid hemorrhage (SAH)?
Background: Headaches represent around 2% of all emergency department visits. Of these presentations 1-3% turn out to be SAH (Edlow, Vermeulen, Perry, Morgenstern). About 5% of SAH are misdiagnosed on the 1st ED assessment (Vermeulen). This is because 50% of SAH present with no neurologic deficit (Weir).
Traditional methods of working up a SAH has been non-contrast CT followed by a LP. The LP aspect has been questioned by our guest last week Dr. David Newman. He suggested LPs are not always needed. The NNT was 700. So are you part of the 700 Club.
Reference: J Perry et al. Clinical Decision Rules to Rule Out Subarachonoid Hemorrhage for Acute Headache. JAMA 2013
- Population: Adult patients with headache peaking within 1hr
- Intervention: Three different clinical decision rules (CDR)
- Control: None
- Outcome: SAH (blood on CT, xanthochromia in CSF, RBC in last LP tube and positive angiography)
Methods: Prospective multi center cohort study at 10 tertiary care hospitals. Consecutive adult patients presenting with non-traumatic headache that reached maximum intensity within 1 hour.
There were some fancy statistics you did in this study. One was multivariant recursive partitioning.This is a statistical method of making a decision tree that tries to correctly classify patients in the population based on a number of dichotomous dependent variables. There are some advantages and disadvantages to analyzing the data this way.
- Clinically a more intuitive models that do not require calculations.
- Can create a decision rule that is more sensitivity or specificity
- May be more accurate
- Continuous variables do not work well
- May over fit data.
You also did post hoc bootstrapping analysis of the data. This a statistical way of resampling the data to assign measures of accuracy to sample estimates. A great advantage of bootstrap is its simplicity while checking for the stability of the results. You can derive estimates of various complex parameters of a distribution. A weakness is that bootstrapping tends to be overly optimistic in its estimations.
Mean age 44 years, 60% women, 26% arrived EMS, 83% got CT, 39% got LP and 6.2% had final diagnosis of SAH
There were 605 (22%) patients who were deemed missed potentially eligible. These missed patients were similar to the enrolled patients (mean age 44, 57% women, 29% EMS, 83% CT, 38% LP and 5.5% SAH).
Of the 2131 patients in the cohort, only 35 (1.6%) without both normal CT and LPs could not be contacted. However, none were admitted to regional neuro surgical center or identified as dead by coroner.
Looking at each of the three rules they had sensitivity which ranged from 95.5-98.5% with specificity from 27.6-35.6%.
All three rules missed a few SAH. No SAH were missed by all three rules. Rule #1 identified 130 or 132 SAH. Only one of the two misses by Rule #1 was considered clinically significant.
The Ottawa SAH Rule consists of the four elements for Rule #1 (Age>40, neck pain/stiffness, witnessed LOC, and onset during exertion) and added two more elements. (thunderclap headache and limited neck flexion). This change increased the sensitivity to 100% (95%CI 97.2-100) but dropped specificity to 15.3% (95% CI 13.8-16.9).
Physicians were also asked about how comfortable they were using the rules and how accurate they were at using the rules. Physicians were comfortable (82%) using Rule #1 and correctly applied it 95% of the time. Misinterpretation of Rule #1 theoretically could have led to 1 missed SAH.
If Rule #1 was used it would have dropped the investigation rate down from 84% to 74%. However, the proposed Ottawa SAH rule would have an investigation rate of 86%
Authors’ Conclusion: “Among patients presenting to the emergency department with acute nontraumatic headache that reached maximal intensity within 1 hour and who had normal neurologic examination findings, the Ottawa SAH Rule was highly sensitive for identifying subarachnoid hemorrhage. These findings only apply to patients with these specific clinical characteristics and require additional evaluation in implementation studies before the rule is applied in routine emergency clinical care.”
BEEM Comments: This was a very well done large multi centered prospective validation study. The Ottawa SAH Rule is simple and contains only 6 variables. Applying this clinical decision tool could decrease the miss rate of SAH from about 5% down to almost 0% with only a slight increase in utilization. It remains to be seen whether the Ottawa SAH Rule would have the same impact in other health care systems with different practice environments. There are also some people that say SAH is too complicated a condition for a clinical decision tool to work. Regardless, we should always try and use EBM to increase patients choices using shared decision making. The Ottawa SAH Rule may turn out to be a good way to frame a conversation with patients presenting with a potentially life-threatening condition. We eagerly await the validation studies before we change our practice pattern.
BEEM Bottom Line: Ottawa SAH Tool is not ready for prime time to rule out low risk patients from investigations.
Case Resolution: You are clinically concerned and get a non-contrast CT head which is negative. You discuss the risks and benefits of an LP with the patient. A shared decision is made with the patient not to do an LP. She is discharged home with appropriate analgesia. She is to return to the emergency department if she developes focal neurologic symptoms, pain increases, LOC, seizure or is otherwise worried.
KEENER KONTEST: Last weeks’ winner was Colleen Sweeney a family medicine resident in Nova Scotia, Canada. She knew the NNT for ACLS drugs in cardiac arrest is zero or infinite, no patient benefitted with improved survival to hospital discharge.
Listen to the podcast for this weeks Keener Kontest Question. If you know the answer send and email to TheSGEM@gmail.com with Keener Kontest in the subject line.
This episode will be posted while attending ACEP 2013 in Seattle next week. Make sure you come up and say hello and I might give you a cool skeptical SGEM prize.
Remember to be skeptical of anything you learn, even if you heard it on The Skeptics’ Guide to Emergency Medicine.