Usefulness of Kernig and Brudzinski Signs in Diagnosing Meningitis
Ah, the first Conference of the academic year. The new faces, with their bubbling enthisasm, suddenly confronted with the annual reading of the policies & procedures manual… why, it’s enough to give anyone a headache and photophobia. Which is why it’s so timely that we review meningitis, with a focus on some physical exam skills that can (but may not) help guide your diagnosis.
Tim reviewed a paper by Thomas et al out of Yale, published in Clinical Infectious Diseases (2002;Vol 35 July 1, pp46-52) called “The Diagnostic Accuracy of Kernig’s Sign, Bruzinski’s Sign, and Nuchal Rigidity in Adults with Suspected Meningitis.” Here’s his CAT sheet.
The researchers looked at anyone over 16 who presented to Yale-New Haven with “clinically suspected meningitis” (which they define as headache, fever, stiff neck, photophobia, nausea + vomiting) who were tapped. This was part of a larger cohort used in other studies. 301 were enrolled, 297 were tapped (four had CT’s that showed mass effect).
Physicians (28% interns, 55% residents, and 17% attendings) were required to document their history and exam findings before LP, but were NOT instructed on how to properly test Kernig’s and Brudzinski’s signs. Meningitis was considered to be present if the WBC count from the CSF was greater than 6 (regardless, as Scott and Ted pointed out, of whether the tap was traumatic or not).
Kernig’s sign (pain upon thigh flexion and knee extension) was shown to have a sensitivity of 5%, specificity of 95%, with a PPV of 27%, and NPV of 72%. Brudzinsky’s sign (flex the neck, see if the patient flexes thighs and knees in response) fared pretty much the same. Nuchal rigidity was 30% sensitive, 68% specific, with a PPV of 26% and NPV of 73%.
The group helpfully calculated likelihood ratios – but they were all around 1.0 (some subsets were higher, if they looked at severe meningeal inflammation, for example). An LR of around 1 means the pre-test probability is unaffected – your patient is just as likely to have the disease before you apply the test, as after. In other words, it’s useless. Furthermore, the sensitivities of Kernig’s and Brudzinski’s signs are poor – especially when you’re looking for a justification to tap (beyond, you know, your keen clinical suspicion). Also, these NPV’s are not good enough – we want a really high negative predictive value, because meningitis has such high mortality.
As we critically reviewed this paper (based on the format for evaluating diagnostic studies, see Jaeschke et al in JAMA 271(5):389-391 (Feb 2, 1994) and 271(9):703-707 (March 2, 1994) ) a few good points were made:
K+B’s oringal tests, 100 years ago, were based on a different population – one with mostly TB meningitis. Maybe that’s why they got 70% sensitivity. (Or, maybe they just knew how to do the test better than the Yale residents).
Ted pointed out something important: their criteria for meningitis was looser than our own: a WBC > 6 is meaningless in the presence of a traumatic tap, yet their methods don’t mention this possibility (maybe the Yale residents are really skilled?)
As for applicability, well Kerning’s and Brudzinski’s signs are benign tests, true, but this study suggests they don’t seem to be that useful. They don’t change the pretest probabilities of meningitis. Dr. Bais noted, wisely, this means we ought to study the signs properly applied – all Thomas’ paper shows is the signs aren’t useful as currently applied by Yalies.
An altogether more balanced approach to history and physical exam for meningitis was presented in Attia’s review, “Does This Adult Patient Have Acute Meningitis?” – part of JAMA’s rational clinical examination series (JAMA 281(2) ,175-181 July 14, 1999). This paper is chock full of useful tidbits (incidence is 3-11 per 100,000 person-years, 17% mortality in adults in the 80’s, even with therapy, 18% of survivors enduring longterm neuro deficits).
Their review of the literature shows that history alone cannot provide enough info for an accurate diagnosis of meningitis, but some aspects of physical exam could be useful: of the classic triad of fever, neck stiffness, and headache / altered MS, 95% of meningitis patients had at least 2 symptoms, and 99%-100% had at least one. Fever is the single most sensitie finding. Best yet, they report that jolt accentuation of headache has a sensitivity of 100% (who wouldn’t mind that maneuver?) and positive likelihood ratio of 2.2 for meningitis.
So, take home points from these papers? Fever, with headache (or changed MS) should prompt a meningitis workup, but no eponymous maneuvers should change your decision-making.