CSF to Serum Glucose Ratio Interpreter
CSF to Serum Glucose Ratio Interpreter
A CSF glucose of 3.0 mmol/L is normal against a serum of 5 and markedly low against a serum of 12, and it is the absolute figure that gets reported and acted on. This page reads the ratio instead, in either unit, and asks the two questions the ratio depends on: were the samples actually paired, and was the patient hyperglycaemic when they were taken?
Is this CSF glucose actually low?
Paired glucose → ratio → interpretationA 58-year-old with type 2 diabetes presents with fever and headache. The CSF glucose is 3.0 mmol/L — a figure that would be entirely normal in most patients — and the blood glucose drawn at the same time is 12.0 mmol/L.
The ratio, and the two assumptions inside it
Assumption one: the samples are paired. CSF glucose reflects the blood glucose over the preceding several hours, not the value at the moment of the tap. A denominator drawn hours away, or after dextrose, is the wrong denominator.
Assumption two: the blood glucose is not extreme. Glucose enters CSF through a saturable transporter, so the expected ratio falls as the blood glucose rises. The same measured ratio means less at a serum of 20 than at a serum of 5.
Neither assumption is checked by the laboratory, and neither is printed on the report.
Worked example
A 58-year-old with type 2 diabetes presents with fever and headache. The CSF glucose is 3.0 mmol/L — a figure that would be entirely normal in most patients — and the blood glucose drawn at the same time is 12.0 mmol/L.
Ratio = 3.0 ÷ 12.0 = 0.25
That is below 0.4, the conventional markedly-reduced threshold, and below Tamune's optimal cut-off of 0.36
The absolute CSF glucose of 3.0 mmol/L is what the laboratory reported and what would have been glanced at. Against a serum of 5.0 it would have been 0.60 — normal. The same CSF number, two different answers
The samples were paired, so the denominator is the right one
One qualification the raw ratio does not carry: at a serum of 12 the expected ratio is itself lower than 0.6, because glucose crosses into CSF through a saturable transporter. So 0.25 here is abnormal, but less extraordinarily so than 0.25 at a serum of 5 would be
Action: treat this as a low ratio. Read it with the count, differential and protein in the CSF cell count interpreter, send the CSF lactate interpreter and a molecular meningitis panel, and start empirical antibiotics if the clinical picture supports it
The same CSF glucose against five different serum values
| CSF glucose | Blood glucose | Ratio | Reading |
|---|---|---|---|
| 3.0 mmol/L (54 mg/dL) | 4.5 mmol/L (81 mg/dL) | 0.67 | Normal |
| 3.0 mmol/L (54 mg/dL) | 5.0 mmol/L (90 mg/dL) | 0.60 | Normal |
| 3.0 mmol/L (54 mg/dL) | 6.5 mmol/L (117 mg/dL) | 0.46 | Reduced |
| 3.0 mmol/L (54 mg/dL) | 8.0 mmol/L (144 mg/dL) | 0.38 | Markedly reduced |
| 3.0 mmol/L (54 mg/dL) | 12.0 mmol/L (216 mg/dL) | 0.25 | Markedly reduced — and this is the value the laboratory would have reported as an unremarkable CSF glucose |
What each threshold on this page is, and where it comes from
| Figure | Meaning | Source and caveat |
|---|---|---|
| About 0.6 | The quoted normal ratio | A central value, not a lower reference limit. Leen’s age-specific study puts the 5th percentile at 0.41–0.53 for children and adults |
| Below 0.5 | Reduced | Conventional. The least informative band on the page — it contains normal results, partially treated bacterial meningitis and early tuberculous meningitis |
| Below 0.4 | Markedly reduced | The conventional threshold, and the one the CSF to serum glucose ratio calculator bands at |
| 0.36 | ROC-optimal cut-off | Tamune 2014: sensitivity and specificity 92.9%, AUC 0.97 — but from 15 bacterial against 129 aseptic cases in one emergency department. The direction is sound; the second decimal place is not |
| CSF glucose below 1.9 mmol/L (34 mg/dL) | 99% certainty of bacterial over viral | Spanos 1989, as a single finding. This is an absolute figure and the only one on the page that does not need a serum |
| Neonates | No cut-off is published here | Sarff’s neonatal means were 81% and 74% of blood glucose, above the adult figure; Leen’s infant 5th percentile of 0.36–0.44 is below the adult one. Both are real. Use the neonatal CSF reference interval interpreter instead |
Why the absolute CSF glucose is the number that gets acted on, and the one that means least
CSF glucose is derived from blood glucose, so a CSF value has no meaning until it is divided by the blood value it came from. That is not a subtle point — it is the difference between a normal result and a medical emergency. A CSF glucose of 3.0 mmol/L against a blood glucose of 5.0 is a ratio of 0.60 and entirely normal. The same 3.0 mmol/L against a blood glucose of 12.0 is a ratio of 0.25, well below every published threshold, and in a patient with fever and headache it is a reason to start antibiotics. The laboratory reports the 3.0. The ratio is left for someone to work out, and in a busy department with a diabetic patient it frequently is not.
Two assumptions sit inside the ratio and neither is printed on the report. The first is that the two samples are paired. CSF glucose does not track the blood glucose instantaneously: Tan and colleagues, who measured blood glucose hourly for six hours before lumbar puncture in 195 patients, found that CSF glucose correlated better with the average blood glucose across that window than with the value at any single time point, and that the correlation was strongest in patients with diabetes or impaired glucose tolerance. A blood sample taken three hours after the tap, or after a dextrose infusion has been started, is simply the wrong denominator, and the resulting error can run in either direction.
The second assumption is that the blood glucose is not extreme. Glucose crosses the blood-brain barrier through a saturable transporter rather than by free diffusion, so as the blood glucose climbs, the proportion that reaches the CSF falls. The expected ratio in a markedly hyperglycaemic patient is therefore lower than 0.6 to begin with, which cuts both ways: a modestly reduced ratio at a blood glucose of 15 may be physiology rather than infection, and a ratio that looks merely normal at that blood glucose is better news than it appears. Nobody has quantified that correction well enough to publish an adjusted threshold, which is why this page flags the situation rather than pretending to correct for it.
The neonatal case is a genuine refusal rather than an oversight. The classic neonatal series reported mean CSF glucose values of 81% and 74% of the blood glucose, comfortably above the adult figure of around 0.6, which would suggest a higher neonatal threshold. But the age-specific percentile data point the other way: the 5th percentile for the CSF to plasma glucose ratio in infants is 0.36 to 0.44, below the 0.41 to 0.53 seen in children and adults, because the neonatal reference interval is much wider at both ends. A higher mean and a lower limit cannot be reconciled into one cut-off, so this page publishes none and routes the question to the record that carries the age-banded intervals instead.
Frequently asked questions
What CSF to serum glucose ratio suggests bacterial meningitis?
Below 0.4 is the conventional threshold for a markedly reduced ratio and points at bacterial, tuberculous or fungal meningitis rather than viral. Tamune and colleagues found an ROC-optimal cut-off of 0.36 with a sensitivity and specificity of 92.9% each, though that came from a single-centre series of 15 bacterial against 129 aseptic cases, so the direction is better established than the exact figure. Around 0.6 is the quoted normal value — but it is a central value, not a lower reference limit: the 5th percentile in children and adults is 0.41 to 0.53.
Do both results have to be in the same unit?
Yes, and once they are the ratio is unit-free — 3.0 ÷ 12.0 mmol/L and 54 ÷ 216 mg/dL are both 0.25. That is one of the ratio’s practical advantages over the absolute figure, which needs a unit and a different threshold in each. If the two results have arrived in different units, convert one with the glucose unit converter before dividing.
How close together do the CSF and blood samples need to be?
Closer than they usually are. CSF glucose reflects the blood glucose over the preceding several hours rather than the value at the instant of the tap — in one study measuring blood glucose hourly for six hours beforehand, CSF glucose correlated better with the six-hour average than with any single time point. Within about an hour of the lumbar puncture is the practical target. A blood sample taken after a dextrose infusion, a meal or a seizure inflates the denominator and manufactures a falsely low ratio; one taken during a hypoglycaemic dip does the reverse.
Does hyperglycaemia change the interpretation of the ratio?
Yes, and in a direction that is easy to get backwards. Glucose enters the CSF through a saturable transporter, so the proportion crossing falls as the blood glucose rises and the expected ratio is itself lower in a hyperglycaemic patient. A modestly reduced ratio at a blood glucose of 15 mmol/L is therefore weaker evidence of meningitis than the same ratio at 5 mmol/L, and a normal-looking ratio at that blood glucose is stronger evidence against it. No corrected threshold has been published, so this is a reason to weigh the ratio less heavily against the rest of the profile rather than to adjust it arithmetically.
What is the normal ratio in a newborn?
There is no single figure this page will publish, and that is deliberate. The classic neonatal series reported mean CSF glucose values of 81% and 74% of the blood glucose in its two groups, above the adult figure of around 0.6. But the age-specific percentile data give a 5th percentile of 0.36 to 0.44 in infants, below the 0.41 to 0.53 seen in children and adults, because the neonatal interval is far wider. A higher mean with a lower limit does not collapse into one threshold. Use the neonatal CSF reference interval interpreter, which carries the age-banded intervals, and treat neonatal meningitis on suspicion rather than on a cut-off.
A normal ratio — can I stop worrying about bacterial meningitis?
No. A normal ratio argues against bacterial, tuberculous and fungal meningitis and is characteristic of viral meningitis, but three situations produce a normal ratio in genuine bacterial disease. Antibiotics given before the tap raise the CSF glucose, and that effect becomes more marked beyond twelve hours of pretreatment. Early bacterial meningitis can present before the glucose has fallen. And an immunocompromised patient can have a normal profile throughout. Read the ratio with the count and differential in the CSF cell count interpreter rather than on its own.
Related calculators
References
- Tamune H, Takeya H, Suzuki W, et al. Cerebrospinal fluid/blood glucose ratio as an indicator for bacterial meningitis. Am J Emerg Med. 2014;32(3):263–266.
- Spanos A, Harrell FE Jr, Durack DT. Differential diagnosis of acute meningitis. An analysis of the predictive value of initial observations. JAMA. 1989;262(19):2700–2707.
- Leen WG, Willemsen MA, Wevers RA, Verbeek MM. Cerebrospinal fluid glucose and lactate: age-specific reference values and implications for clinical practice. PLoS One. 2012;7(8):e42745.
- Tan QC, Xing XW, Zhang JT, et al. Correlation between blood glucose and cerebrospinal fluid glucose levels in patients with differences in glucose metabolism. Front Neurol. 2023;14:1103026.
- Sarff LD, Platt LH, McCracken GH Jr. Cerebrospinal fluid evaluation in neonates: comparison of high-risk infants with and without meningitis. J Pediatr. 1976;88(3):473–477.
- Nigrovic LE, Malley R, Macias CG, et al. Effect of antibiotic pretreatment on cerebrospinal fluid profiles of children with bacterial meningitis. Pediatrics. 2008;122(4):726–730.
Medical Disclaimer: The tools and content provided here are for educational and reference purposes only. They are not intended to substitute for professional medical advice, diagnosis, or treatment. Clinical decisions should always be based on the comprehensive assessment of a qualified healthcare professional.