By Dean L. Winslow, MD, FACP, FIDSA

Professor of Medicine, Division of General Medical Disciplines, Division of Infectious Diseases and Geographic Medicine, Stanford University School of Medicine

Dr. Winslow reports no financial relationships relevant to this field of study.

SYNOPSIS: The investigators retrospectively evaluated ESCMIID, IDSA, and Swedish guidelines for neuroimaging in 815 adults with acute bacterial meningitis. Swedish guidelines omit altered mental status and immunosuppression as indications for imaging prior to lumbar puncture. Adherence to Swedish guidelines resulted in decreased mortality and more favorable outcomes.

SOURCE: Glimaker M, Sjolin J, Akesson S, Naucler P. Lumbar puncture performed promptly or after neuroimaging in acute bacterial meningitis in adults: A prospective national cohort study evaluating different guidelines. Clin Infect Dis 2018;66:321-328.

Researchers in Sweden prospectively followed a cohort of 815 adult acute bacterial meningitis (ABM) patients between 2008 and 2015. Primary and secondary endpoints were in-hospital mortality and favorable outcome at two to six months, respectively. Swedish guidelines recommend imaging prior to lumbar puncture (LP) in suspected ABM only in the presence of clear evidence of herniation, arm or leg drift, more than four days of neurological symptoms, or other symptoms atypical for ABM. European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines additionally recommend neuroimaging prior to LP in patients with a Glasgow Coma Scale (GCS) score < 10, arm or leg drift, or new-onset seizures. The Infectious Diseases Society of America (IDSA) guidelines recommend neuroimaging in suspected ABM patients with GCS < 15, arm or leg drift, abnormal ocular motility, visual field defect, dilated pupil, new-onset seizures, immunocompromise, suspected mass lesion, stroke, focal infection, or increased intracranial pressure with papilledema.

Of the 815 patients studied, 323 (46%) underwent LP without prior CT, and 378 (54%) had LP performed after CT. The patients overall were well matched by baseline characteristics. Mortality overall was 8%, and favorable outcome was observed in 50% of the cohort.

LP without previous CT was associated with a 4% mortality (14 of 323), while mortality was 10% (37 of 378) in patients who underwent CT prior to LP (P < 0.001). Lower mortality and increased favorable outcomes were seen in patients who underwent LP prior to imaging in all groups stratified by different indications for imaging. In patients who underwent CT before LP, 47% received antibiotics prior to CT imaging. However, antibiotics and corticosteroids were administered within two hours of admission in 41% of patients who underwent prompt LP and in only 30% of patients who received CT imaging prior to LP (odds ratio [OR], 2.12 for prompt administration of antibiotics and steroids after adjustment for baseline characteristics).


The Swedish, ESCMID, and IDSA guidelines differ widely in their recommendations regarding when to perform CT prior to LP. In this study, 7%, 32%, and fully 65% of patients, respectively, had one or more indications for CT imaging prior to LP. Following the Swedish guidelines resulted in a 50% lower mortality and increased favorable outcomes compared to ESCMID and IDSA guidelines. Much of this difference in outcome might be explained by shorter delay in administration of antibiotics and steroids in patients who underwent LP without prior CT.

By way of background, I began my medical training in the early 1970s just prior to the widespread availability of CT imaging, so we generally had no central nervous system (CNS) imaging available to rule out the presence of large mass lesions in patients with either suspected ABM or stroke.

The theoretical concern was that in the presence of a mass lesion, lumbar puncture could result in transtentorial herniation in certain circumstances when the cerebrospinal fluid (CSF) hydrodynamics were affected by LP. However, even 45+ years ago, more experienced clinicians often reassured us that herniation (even in the presence of mass lesions) was distinctly unusual with simple diagnostic LP when performed with a small #20-gauge needle and removal of only 4 cc of CSF for diagnostic studies. Apparently, most historical episodes of herniation precipitated by LP actually occurred when the procedure was performed with a larger gauge needle and when large amounts of CSF were removed during pneumoencephalography or myelography.

Most of my mentors back then gave us the advice that we should almost always perform a diagnostic LP if the pretest probability of ABM was high (fever, nuchal rigidity) and the patient did not have a clear large motor deficit. However, altered level of consciousness, seizures, cranial nerve palsies, or even increased intracranial pressure were not contraindications to LP. Having performed literally hundreds of diagnostic LPs over the years, I had only one patient who possibly deteriorated as a result of LP. (This was a patient I cared for prior to the advent of CT, and who, at autopsy, had a massive intracerebral bleed and had evidence of transtentorial herniation.)

This study is reassuring that our old mentors’ advice was sound. At least in patients in whom ABM has high pretest probability, the current IDSA guidelines (which recommend imaging prior to LP in all immunocompromised patients and patients with seizures, altered level of consciousness, or cranial nerve palsies) are overly conservative and not helpful. We should adopt the Swedish guidelines.