It may take 700 LPs to find a subarachnoid hemorrhage, but it’s a needle-in-a-haystack worth finding. Dr. Klauer challenges Dr. David Newman’s cost-effective approach in the SAH workup.
It may take 700 LPs to find a subarachnoid hemorrhage, but it’s a needle-in-a-haystack worth finding. Dr. Klauer challenges Dr. David Newman’s cost-effective approach in the SAH workup.
I’m proud to be a card-carrying member of the “700 Club,” as David Newman put it in last month’s issue. In his bi-monthly column, Newman dug into the numbers behind diagnosing subarachnoid hemorrhage (SAH), explaining that it takes 700 LPs to find one. In the end, he argued that the risks associated with these 700 LPs outweighed the benefits of finding the one needle in the haystack.
David is a smart man, and I do respect his thoughts and opinions. However, in this case, I believe the academic discussion of NNT could distract us from the reality that subarachnoid hemorrhage is a life-threatening disease and standard CT and LP is the best testing combination we have for ruling it out. In fact, a negative CT and negative LP provides an absolute answer that a subarachnoid hemorrhage is not present. One study reported a negative likelihood ratio of 0 when both tests are negative, making this a very effective testing combination1. Academic discussions focusing on statistical analyses, such as the NNT, are often difficult to translate to clinical practice. Is it time to quit performing LPs for patients with headaches suspicious for SAH? Probably not. It is a crucial element of the evaluation.
So, somebody please tell me why we should be surprised by an NNT of 700? Subarachnoid hemorrhage is a rare entity; it always has been. Based on that fact alone, a lot of people must be tested to find those that have a SAH. This is not a new concept. With approximately 136 million U.S. ED visits annually and 5% of ED patients presenting with headache as their chief complaint, finding the 30,0002 subarachnoid hemorrhages reported annually in the U.S. was, and will always be, selectively looking for the needle in the medical haystack.
The variability of the history patients provide and the wide array of headache presentations seen in the emergency department make accurate patient selection for LP very, very difficult. The test isn’t the problem. Risk stratification and patient selection are, as physicians are challenged in assessing the baseline risk for SAH among various populations. “Is this the worst headache of your life?” Are you kidding me? Is this our primary tool for deciding who needs a work up for SAH and who doesn’t? Until we come up with a better strategy for identifying those at greatest risk, and conversely, those not at risk, we should recruit more members to the “700 Club,” not less. The reality is that if we start doing fewer LPs, we will begin missing SAHs. Yes, admittedly, the number will be small. However, we should look at this disease from a purist’s perspective and not a medical economist’s perspective (e.g. avoiding misses as opposed to performing fewer tests). How many SAHs are you willing to miss so that you don’t have to do a bunch of LPs? I don’t know what others are comfortable with. My answer is 0! The likely outcome of a missed SAH is death or severe disability. Twenty-five percent of subarachnoid hemorrhages are misdiagnosed at the time of their initial evaluation and 48% deteriorate or re-bleed before they are able to return3. Omitting the LP will only increase the risk of delayed or misdiagnosis.
I challenge the assertion that significant dangers exist with the performance of a lumbar puncture when it is performed correctly, with the appropriate training, technique and equipment. The vast majority of LP complications are not life threatening, with back pain and post dural puncture headache being the most common, 25% and 22% respectively4.
Spinal headache is not uncommon. However, it must be recognized that with meta-analysis or systematic review of LP complications, it is impossible to avoid factoring in bad practice with the good. The data is clear. If you use an atraumatic needle, the spinal headache rate is substantially lowered. In fact, the postdural puncture headache rate using a 22 gauge Quincke (cutting needle) has been reported to be 15.2%, compared to 4.2% with the Whitacre (atraumatic needle)5. So, much of the concern raised by Dr. Newman about complications of LP, likely includes those not performing the procedure optimally with known techniques to avoid the most common complications.
I‘m “all in” on the NNT discussions on antibiotics and dog bites and rapid strep screening. Often, these items are, in aggregate, costly and offer little to no value to the care provided. Whether we perform them or not, the patient won’t be exposed to serious risks such as death or severe morbidity, like that of SAH. If we are going to use NNT to guide management, aren’t we better served applying this data to disease entities that can be missed without risk? For example, antibiotics are often prescribed for strep throat due to the concern for rheumatic fever. The NNT to avoid this complication is 4,0006. Treating 4,000 to benefit one is an unreasonable proposition.
It would be enlightening to calculate the NNT for obtaining an electrocardiogram for chest pain. The number of ECGs performed, compared to AMIs detected, must be astronomical and would likely produce an NNT much greater than 700. Although there is no risk associated with performing ECGs, LPs are also very safe. Both cast a wide net to look for life-threatening diseases. Testing many to look for the few is a reasonable approach when the outcomes of misdiagnosis are devastating.
We have to be careful with how we apply the NNT. Let’s apply this concept to the entire work up for SAH (CT and LP), and every other rare disease we can think of. The result could easily be, “Why bother.” In other words, in every uncommon disease, the absolute risk reduction will be low, resulting in a NNT that will be high for the diagnostic evaluation. Despite the low yield, I don’t think we are ready to abandon the search for these needles in the high-risk haystack.
As a risk manager, my goal is to improve patient safety while reducing provider liability. Miss a subarachnoid hemorrhage and I can almost guarantee that you are at serious risk of looking down the smoking barrel of a policy limits claim. In fact, in many of these cases, you may be lucky to resolve the claim without tendering your policy limits; plaintiff demands in such cases are often far north of $1 million.
Those sitting in the comfort of their armchair, ready to Monday morning quarterback, will likely say that these cases are defensible or that the demands can be negotiated down to policy limits or below. This may be true. However, that is an extremely cavalier position to take from the comfort of your Barcalounger. Ask anyone who has experienced the painful process of a lawsuit – even if it never reaches the courtroom. It takes months of the provider’s life. It consumes them emotionally and often destroys their confidence. I suspect that they’d be happy to perform a few, or a lot more, LPs to find the one patient that will benefit from it.
I caution those who choose to perform fewer LPs due to this NNT analysis. Perhaps the strongest rationale for doing fewer LPs due to the NNT of 700 is that in uncommon disease, you have luck on your side. With uncommon disease, good and bad practice are just as likely to result in negative workups. However, if you want to find a SAH, good practice, CT and LP, is still the best approach.
References
1. Perry JJ, et al. Is the combination of negative computed tomography result and negative lumbar puncture result sufficient to rule out subarachnoid hemorrhage? Ann Emerg Med. 2008 Jun;51(6):707-13.
2. Bederson et al. Guidelines for the Management of Aneurysmal SAH: Stroke 2009;40:994-1025
3. Mayer. Misdiagnosis of symptomatic cerebral aneurysm. Stroke 1996.
4. Ruff RL, Dougherty JH Jr Complications of lumbar puncture followed by anticoagulation. Stroke. 1981;12(6):879.
5. Hatfield, M.K., et al. Blood Patch Rates After Lumbar Puncture With Whitacre Versus Quincke 22- And 20-Gauge Spinal Needles, Am J Roent 190(6):1686, June 2008.
6. Worall GH. Acute Sore Throat. Canadian Family Physician; Vol 53:november 2007.
Kevin Klauer, DO, EJD is Editor-in-chief of Emergency Physicians Monthly, CMO of Emergency Medicine Physicians, Vice Speaker of the ACEP Council.
8 Comments
Here’s a suggestion: let the patient help make the decision. Describe the most likely diagnosis: headache NOT due to SAH. Describe the odds of finding a SAH on the LP: very very low. Describe the potential complications of the workup: discomfort, delay, more workups.
I think (in fact I know) that most patients will say: NO LP.
Wow, we’re never going to see a reasonable solution to our Med/Mal world of pain. If we can’t see our way to ending the search for an SAH at NNT of 700, we have no hope of ever reducing our own risk and the high cost of medicine for our country. Where do these assets spent on finding the one SAH come from? God? I don’t think so, they come from our treasure that should have been spent on feeding the poor and vaccinating the world.
It’s crazy, and we should just say, “I’m mad as hell and I’m not going to take it anymore”.
… is contained in the quote: “How many SAHs are you willing to miss so that you don’t have to do a bunch of LPs? I don’t know what others are comfortable with. My answer is 0!”
None of us wants to miss a potentially deadly diagnosis. But pretending that anything we do makes us immune to such a miss is missing the point. Physicians are not omniscient, and the implication that we should do everything in our power (regardless of cost, both financial, social and medical) to achieve a diagnosis is irresponsible. Given a negative head CT, the risks associated with LP may be small enough for any given individual with a bad headache, but must still be weighed against the benefit (to that individual) of potentially diagnosing a zebra. But the argument becomes even clearer, since as emergency physicians we must be in the patient advocacy business, both at the individual and societal levels.
The fact is we can’t afford a health care delivery system where no physician can accept even a single missed SAH, ever. Suggesting that we shouldn’t is simply adding fuel to the perception among much of the lay public that we should be perfect, and any bad outcomes should be grounds for malpractice.
Furthermore, I would argue that it is false logic to suggest Dr. Newman and others are willing to miss a SAH so that they “don’t have to do a bunch of LPs.” Any test we chose to do should be based on pretest probabilities and associated testing thresholds. Many factors are considered– including, but not limited to, the patient’s presentation, the provider’s medical knowledge, and a risk-benefit assessment of testing. The issue is not whether people are trying to get out of doing a lengthy and time-consuming procedure (although from a population health standpoint, and inasmuch as we are and should be guardians of our population’s health resources, we should be streamlining a lot of what we do in the ED). The issue is that this time-consuming procedure is not without risk and carries a significant false positive rate. Hence the appropriateness of a risk-benefit analysis.
I did not read Dr. Newman’s article as arguing against ever doing an LP on a patient in whom SAH was suspected, but rather whether a more rational and individualized approach might be in order.
There are two problems that complicate the issue and decrease the attractiveness of being in the 700 Club:
1. Traumatic LP results are uninterpretable much of the time, as partial clearing of RBCs cannot be used to rule out hemorrhage (so the gurus tell us).
2. Many Neurosurgeons in community hospitals are not impressed with a bloody LP and a negative CT and CTA of the head, recommending at most an overnight observation for the patient.
While these points do not obviate all of the benefits of doing an LP, they create doubt that the LP is the paragon of tests for this problem.
I never want to miss one–but that being said there are “sudden headaches” that I am worried about, and those I am not. As anything in medicine, each patient should be treated as an individual case. If I am really worried, with a patient with sudden headache, vomiting, looks like hell, then I’m doing the LP. If I am not, i.e. the patient having their worse headache ever that started suddenly and gives a great, story, that comes in playing Angry Birds on their cell phone, I am not…
Risk-stratification of the ED patient with a headache concerning enough to contemplate SAH is excellent fodder for the concept of “Overdiagnosis” as discussed at ACEP (see http://tinyurl.com/bs2vjj7). This exact scenario is painted in a new book entitled “When Doctors Don’t Listen: How To Avoid Misdiagnoses and Unnecessary Tests” (see http://tinyurl.com/alf5qoe).
I agree with all of the above perspectives. The objective is not test avoidance for the convenience of the physician. The motivation for this conversation is to minimize unnecessary, potentially harmful testing in favor of tests that offer the most good for the most patients.
I would offer these comments to Dr. Klauer
1) “If we are going to use NNT to guide management, aren’t we better served applying this data to disease entities that can be missed without risk?” No, we should immediately discard those tests for which there is no risk? That is the rationale for the Choosing Wisely campaign and we should not have to waste our time eliminating such obvious waste. More complex issues like SAH do merit our attention.
2) “I challenge the assertion that significant dangers exist with the performance of a lumbar puncture when it is performed correctly, with the appropriate training, technique and equipment.” Interesting that no data was provided on how often the LP is performed correctly or incorrectly, or with inadequate training. Even more interesting that Dr. Klauer did not include the caveat that LP should be performed on the appropriate patients — perhaps the most important detail of all — amongst these considerations. A test performed on the wrong patient is an all-risk/no-benefit proposition. How often does that occur in the angst over missing the rare SAH?
Obviously, the first steps are to understand the evidence (see a thorough review at http://tinyurl.com/6p5lrl5) and to effectively communicate these risks with patients to facilitate shared decision making (see some ideas at http://tinyurl.com/7he85w9).
Dr. Newman dedicated a chapter of Hippocrates Shadow to healthcare providers who fail to unlearn when new evidence becomes available (pseudoaxioms). Clinicians who fail to incorporate compelling new data into their clinical decision making under the guise of malpractice concerns are putting themselves ahead of the patient. Like Dr. Newman, I do not believe that there is no role for LP’s after a non-diagnostic CT in the setting where SAH is a concern. I simply believe that the role of LP is far more limited than in the era before modern generation CT’s and that a steadfast CT-then-LP approach in 2013 & beyond will harm more patients than we help.
First of all, I’d like to add in that LP’s do still have a place. BUT, the place for LP is shrinking. And, it should be made more public that there are circumstances beyond anyone’s control in which, despite the best that medicine offers, we cannot prevent every potential M&M event in the human condition.
Yet it seems that a zero miss rate policy is being promulgated (between the lines) in the context of SAH work-up–by a respected, nationally known speaker and author.
This is discouraging.
We are advisors on the best, most reasonable course of action for any given set of circumstances with which a patient presents. But we are not able to control the future.
So despite optimal medical care, patients can still have the MI after the million dollar work-up–even while driving out of the hospital parking lot. Ditto the patient with a low-risk community acquired pneumonia, discharged (appropriately) home, who becomes unexpectedly worse and despite returning to the hospital exactly as advised, still dies. Or the TIA, or the febrile toddler, or the low risk CHI, and so on…
Mainly I’d like to see more support provided, through nationally known educators, to the pit docs. Standard of care exists but it is dynamic. Help create an environment in medicine where the harsh realities of life’s morbidity and mortality are acknowledged in a more uniform way, so that medico-legal process will have more basis in reality when it comes to the standard of care question. This will not occur when our educators are putting forth opinions as discrepant as this. Particularly troubling is a summary comment like his, “However, if you want to find a SAH, good practice, CT and LP, is still the best approach.”.
He states that 1 in 40000 is excessive for treating strep, though rheumatic fever is not a walk in the park for those who get it. Is his NNT somewhere between 700 and 40,000?
National lecturers help disseminate to some degree standard practices, as other take up their recommendations in day to day clinical settings.
So the question for Dr Klauer is:
What is a reasonable NNT for LP’s in the pursuit of SAH, given Dr Newman’s discussion/context?
Yes all these LPs do harm, spinal headaches, pain etc, not to mention back up your department like in the usual community ER with 20 waiting and 1 or 2 doctors struggling to keep up. Remember, we’re not there to help people any more. we’re just there to dodge the lawyers…. Wish they told us that in premed. So… keep tapping maybe someday you’ll find one.