15.3 CSF Analysis (02.H)
Key Takeaways
- Measure opening pressure in the lateral recumbent position before you send tubes; sitting punctures and flexed hips distort the number.
- For CT-negative suspected subarachnoid hemorrhage, time lumbar puncture at least 12 hours after onset and prefer spectrophotometry for bilirubin; visual inspection misses faint xanthochromia.
- Bacterial CSF is typically neutrophilic with high protein and low glucose; viral is lymphocytic with normal glucose; tuberculosis and fungal disease are lymphocytic with low glucose and high protein.
- Guillain-Barré albuminocytologic dissociation is high protein with few white cells, often absent in week one; more than about 50 WBC/µL argues against typical GBS.
- Do not lumbar-puncture through a mass with herniation risk, uncorrected coagulopathy, or skin infection at the site; ventricular EVD samples are not interchangeable with lumbar CSF for GBS protein or every meningitis pattern.
The CSF card: pressure, tubes, and blood
Quick Answer: After imaging has made lumbar puncture (LP) safe, measure opening pressure in the lateral recumbent position, collect four tubes, and always draw a paired serum glucose. Time xanthochromia >12 hours after thunderclap onset; spectrophotometry beats visual inspection for bilirubin. Bacterial meningitis: neutrophils, high protein, low glucose. Viral: lymphocytes, normal glucose. Fungal/TB: lymphocytes, low glucose, high protein. GBS: albuminocytologic dissociation (high protein, few white cells), often missing in week one. Tube-to-tube red-cell “clearing” does not exclude subarachnoid hemorrhage (SAH). Skip LP for mass with herniation risk, uncorrected coagulopathy, or infection at the puncture site. External ventricular drain (EVD) CSF is not a lumbar sample.
Independent OpenExamPrep teaching in this section covers CSF laboratory interpretation listed under Diagnostic studies in the ABPN Content Specifications. CSF physiology (production, hydrocephalus mechanics) is earlier in this guide. Here the job is the lab card: pressure, tubes, pigment, formulas, GBS protein, traumatic tap versus SAH, when the needle is contraindicated, and how ventricular fluid differs from lumbar fluid.
Opening pressure
Perform LP in the lateral recumbent position with legs extended once the needle is in (flexed hips falsely raise pressure). The manometer zero is at the spine. Normal adult opening pressure is often cited as about 6–25 cm H2O; many neuro-ICU protocols treat >25 cm H2O as elevated in a relaxed recumbent adult. Convert roughly: 1 mmHg ≈ 1.36 cm H2O, so 15 mmHg ≈ 20 cm H2O. A sitting LP does not yield a comparable opening pressure. High pressure with papilledema and no mass pushes you toward idiopathic intracranial hypertension or cerebral venous thrombosis. Low pressure after trauma or over-drainage suggests leak. If pressure is very high and meningitis is the indication, take the minimum diagnostic volume and do not attempt a “large-volume therapeutic tap” for a mass you have not imaged.
Tubes and the paired glucose
Laboratories differ on tube order. A practical neuro-ICU scheme is:
| Tube | Typical contents | Why |
|---|---|---|
| 1 | Cell count and differential (most contaminated if the tap is traumatic) | Compare with tube 4 |
| 2 | Gram stain, bacterial culture; add fungal/AFB, cryptococcal antigen, multiplex PCR as indicated | Microbiology wants volume |
| 3 | Protein and glucose | Chemistry |
| 4 | Repeat cell count; hold for cytology, extra PCR, oligoclonal bands, xanthochromia supernatant | Last-tube RBC and leftover assays |
Send more volume than you think when SAH versus infection versus autoimmune encephalitis are all live. Never spend the entire specimen on a single esoteric send-out. Draw serum glucose around the LP; CSF glucose is interpretable as a ratio. Normal CSF/serum glucose is about 0.6. Ratios ≤0.5 are concerning; ≤0.4 (or CSF glucose <40 mg/dL in a euglycemic patient) strongly favor bacterial, mycobacterial, fungal, chemical, or some carcinomatous meningitides. Hyperglycemia inflates serum glucose faster than CSF, so an unadjusted CSF glucose can look “low” in a patient on a dextrose infusion — use the ratio.
Xanthochromia: clock, eye, and spectrophotometer
After SAH, erythrocytes in CSF release hemoglobin that becomes oxyhemoglobin, then bilirubin via heme oxygenase in vivo. Bilirubin is the pigment you want. It is unreliable before about 12 hours after the bleed and can persist up to about two weeks. UK laboratory practice and multiple diagnostic-accuracy studies prefer spectrophotometry (net bilirubin absorbance, with oxyhemoglobin noted) over a human glance. Visual inspection of centrifuged supernatant against white paper misses faint bilirubin; oxyhemoglobin can make a sample look red rather than yellow and hide the yellow you are trying to see. Spectrophotometry is more sensitive and somewhat less specific (traumatic taps, in-vitro lysis, high serum bilirubin, and high CSF protein produce messy scans). Teach both facts: do not skip the 12-hour clock, and do not trust a “clear” visual read when the pretest probability of aneurysmal SAH is still meaningful and spectrophotometry is available.
Worked timing: thunderclap at 08:00, modern CT negative at 10:00. An LP at 11:00 can still lack bilirubin even if an aneurysm bled. The xanthochromia-timed LP is after 20:00. That clock does not replace vascular imaging when clinical suspicion stays high.
CT-negative SAH
Noncontrast CT is nearly 100% sensitive for aneurysmal SAH in the first 6 hours when the scanner is modern and the reader is expert. Sensitivity falls with time, small-volume bleeds, anemia, and less experienced reads. If presentation is delayed, the CT is imperfect, or the history is classic, LP still belongs in the algorithm. A negative early CT in a perfect setting may stop the workup in some emergency protocols; the neuro-ICU examination still expects you to know why the 12-hour LP exists.
SAH versus traumatic tap
A traumatic tap introduces blood from a radicular vein. Classic teaching — RBC count falls from tube 1 to tube 4 — is not reliable enough to exclude aneurysmal SAH. Perry and colleagues (BMJ 2015) found that last-tube RBC <2,000 × 10^6/L (<2,000/µL) plus no xanthochromia identified traumatic taps that did not harbor aneurysmal SAH in their cohort; isolated tube-to-tube decline, opening pressure, and D-dimer did not. Use the cluster: xanthochromia (especially spectrophotometric bilirubin), failure of RBCs to fall to a low last-tube count, crenated RBCs after a delay, and elevated opening pressure. A last tube still showing tens of thousands of RBCs without an explanation is not “just trauma.” Conversely, a few hundred RBCs that vanish by tube 4, no xanthochromia, and a delayed LP after a negative CT is the traumatic-tap picture.
Correcting WBC for blood contamination (roughly 1 WBC per 500–1,000 RBC, depending on peripheral counts) is a crude meningitis aid, not a bilirubin assay.
Infection formulas, GBS protein, and when not to needle
Textbook CSF patterns overlap. Use them as starting odds, then add Gram stain, multiplex PCR, cryptococcal antigen, mycobacterial PCR/culture, and cytology.
| Pattern | WBC (order of magnitude) | Differential | Protein | Glucose / ratio | Opening pressure | First add-on tests |
|---|---|---|---|---|---|---|
| Bacterial | Often >1,000/µL (can be lower after antibiotics or with Listeria) | Neutrophils | High, often >100–200 mg/dL | Low; ratio often <0.4 | High | Gram stain, culture, lactate (supportive) |
| Viral | 10–500/µL typical | Lymphocytes (neutrophils possible early) | Normal to mildly high | Normal | Normal or mild rise | HSV/VZV PCR if encephalitis |
| Fungal (e.g. Cryptococcus) | 10–500/µL | Lymphocytes | High | Low | Often very high in cryptococcal disease | Cryptococcal antigen, fungal culture |
| Tuberculosis | 50–300/µL typical | Lymphocytes (neutrophils early) | Very high, sometimes a clot | Low | High | Mycobacterial PCR and culture, adenosine deaminase where used |
| SAH (after hours) | RBC-driven; WBC rises secondarily | Neutrophils early | High | Normal | Often high | Xanthochromia, vascular imaging |
| GBS | Usually <5–10/µL, rarely up to 50 | Mononuclear if any | High after several days | Normal | Usually normal | None required to treat |
Spanos-like cutoffs sometimes quoted for bacterial disease (WBC >2,000/µL, protein >220 mg/dL, glucose <34 mg/dL) are highly specific when present and insensitive — treated pneumococcus can look viral. Listeria often has <1,000 WBC and may be lymphocytic. Early viral meningitis can be neutrophilic for the first hours. Do not withhold antimicrobials waiting for a perfect table row.
CSF lactate (often around ≥3.5 mmol/L or ≥35 mg/dL in meta-analyses) supports bacterial infection but also rises in SAH, stroke, and seizures. It is an adjunct.
Albuminocytologic dissociation in GBS
Albuminocytologic dissociation means elevated protein with a normal (or near-normal) WBC count. It reflects root inflammation leaking protein into lumbar CSF, not a ventricle-full of pus. Yield is time-dependent: many patients still have normal protein on day 1–3; sensitivity rises after the first week. A normal early protein does not exclude GBS. Brighton-criteria teaching and the EAN/PNS guideline treat WBC >50/µL as a red flag for mimics (HIV seroconversion, Lyme, sarcoid, polio-like enterovirus, West Nile, leptomeningeal cancer). Mild pleocytosis <50/µL occurs in a minority of true GBS. You treat GBS on the clinical diagnosis; CSF is supportive and is used to exclude infectious myelitis when the story is off.
When LP is contraindicated
Do not put a needle in the lumbar sac when:
- A mass lesion, obstructive hydrocephalus, or clinical herniation physiology (declining arousal plus a new blown pupil, posturing, Cushing physiology) makes spinal-to-cranial pressure diversion dangerous. Image first when there is papilledema, focal deficit, new seizure, known mass, immunocompromise with possible abscess, or GCS collapse. IDSA bacterial-meningitis guidance still uses that CT-before-LP list — and you give antimicrobials immediately, before the scanner, if meningitis is likely.
- Uncorrected coagulopathy or therapeutic anticoagulation: common practice thresholds are platelets <50 × 10^9/L, INR >1.5, or a heparin infusion still running. Spinal hematoma is the feared complication. Reverse what you can; if an EVD is already indicated for ICP, that may be the safer CSF source after a surgeon is involved.
- Infection of the overlying skin or suspected lumbar epidural abscess at the intended interspace — you can seed the thecal sac or miss the real lesion. Choose another level only if the skin is clean and imaging has excluded epidural pus; otherwise do not LP through cellulitis.
Respiratory failure from GBS is not an LP contraindication by itself; position and airway plans are. Papilledema is a reason to image, not a folkloric absolute if imaging is already clear and pressure is the disease (IIH).
EVD CSF sampling versus LP
Ventricular CSF is not lumbar CSF. Protein is normally lower in the ventricles than in the lumbar sac. A normal EVD protein does not exclude GBS, because GBS protein is a lumbar root phenomenon. Basal meningitis can look milder in ventricular fluid than in lumbar fluid, and the reverse can occur with an isolated ventriculitis around a catheter. EVD samples are appropriate for ventriculitis surveillance when the clinical question is catheter infection (CSF WBC, Gram stain, culture), with the caveat that colonization versus infection is a clinical diagnosis. Routine daily cultures on a quiet EVD increase contamination and infection risk — sample when fever, unexplained CSF pleocytosis, or catheter malfunction demands it. Early after a bloody EVD placement, RBCs and protein are contaminated; do not call that SAH xanthochromia clock from an intraoperative ventricle.
If you need lumbar xanthochromia, GBS protein, or myelographic-level cytology, you still need an LP (or lumbar drain) — provided it is safe. If you need ICP control and infection diagnosis in obstructive hydrocephalus, you need an EVD, not a lumbar needle.
Exam traps
LP at hour 3 for bilirubin. Declaring “traumatic tap” because tube 4 had fewer RBCs than tube 1 while xanthochromia was never measured. Calling GBS excluded because day-2 protein is 40 mg/dL. Calling GBS confirmed because protein is 90 mg/dL with 200 WBC/µL. Using EVD protein to “rule out” AIDP. Needling a posterior fossa mass because the emergency department wanted cells. Waiting on a Gram stain to give ceftriaxone, vancomycin, and ampicillin in suspected bacterial meningitis.
A patient has a thunderclap headache at 08:00. CT at 10:00 is negative. You plan lumbar puncture solely to look for xanthochromia. Which plan is most appropriate?
Which cerebrospinal-fluid pattern is most consistent with untreated acute bacterial meningitis in an adult?
On day 2 of typical areflexic GBS, lumbar CSF shows protein 38 mg/dL and WBC 2/µL. Which statement is correct?
A patient with suspected community meningitis has an uncorrected INR of 2.8, a large cerebellar mass with fourth-ventricle obstruction on CT, and cellulitis over the lumbar spine. An EVD is placed for hydrocephalus. Which CSF strategy is appropriate?